l J'

J.

,
I

i . ~'

: n~it'f. l,:qC.ttf:~AX.J
e r:~L
i .:
..

.11.

'd .

I,

;'.' .....
..

(;

'~

'I

if

IiIBUOH CA

Inll" FREDDY PUn.

CIV. 8620 Aviam ,

CONVERSION

. I

1/.

I'

.'

t,

I;

I
~

,~{

r:
*; IU

,;/

, I'.t
I

.I~

I.

,',If

'.v

'~'-1':' .'. ') 1.

. 'I' fl"

;-."

;.,

"'f,Hd

,/

,
,

:j

.,
/...

DAVID C. WHITE
Profeuor of Electrical Engineering

V"

HERBERT H. WOODSON
AniSlant Profeuor of Electrical Engineering

Deportment of Electricol Engineering

The Monachuse"s Institute of Technology

<l;

;,
',j

r l

..

ENERGY.

;:i;

'il

~'(l. ~IJ

. '.' I

ELECTROMECHA~CAL

, I,,)

jt/

I' ,

1:1J
NEW YORK JOHN WILEY & SONS, INC .

t,

London . Chapman & Hall, limited

it

,
~

t
I
/
1,/

" .1.1.....,1

.....--

h.,;;.. ...... ....., t

rl',I'lIAI'.

B1BLlOTECA

Oonado por

~.

~"I.c.u~-'"

I'

.~l /*>MJ.,..."

"~,..... ,"

;.0 ". .,,'~ \

'f;:":'I;

,(~ !'~'-

"

::/l'"

,
..

("

,
"

....

...

"\

TS; ", ;~!

~'f)."-

~;;j

t "('tiM!"
11,

J"4 -".,'~

t.t'

This book is one of several resulting from a recent reVISIon of the

Electrical Engineering Course at The Massachusetts Institute of Tech
nology. Thc books have the general format of texts and are being used
as such. However, they might well be described as reports on a research
program aimed at the evolution of an undergraduate core curriculum
in Electrical Engineering that will form a basis for a continuing career
in a field that is ever-changing.
The development of an educational program in Electrical Engineering
to keep pace with the changes in technology is not a new endeavor at
The Massachusetts Institute of Technology. In the early 1930's, the
Faculty of the Department undertook a major review and reassessment
of its program. By 1940, a series of new courses had been evolved,
and resulted in the publication of four rebted hooks,
The new technology that appeared during World War II brought great
change to the field of Electrical Engincering. In recognition of this
fact, the Faculty of the Department undertook another reassessment of
its program. By about 1952, a pattern for a curriculum had been
evolved and its implementation was initiated with a high degree of
enthusiasm and vigor.
The new curriculum subordinates option structures built around areas
of industrial practice in favor of a common core that provides a broad
base for the engineering applications of the sciences. This core structure
includes a newly developed laboratory program which stresses thl: role
of experimentation and its relation to theoretical model-making in the
solution of engineering problems. Faced with the time limitation of a
four-year program for the Bachelor's degree, the entire core curriculum
gives priority to basic principles and methods of analysis rather than
to the presentation of current technology.
J. A.
yii

STRATTON

.))'

~~'"

j.

"
~"

"';;;

~<'

'rEi

'~~

"

!ji J

')

,00

W l~ 'Ij ~i; ~ ,~;% ':~

fIe,)."

',j

.~",,>.I,

'HU:

.l;

'1.

.J:,'

t:

.,'

~n

:JPjr,'}

;;;~(r't
i

t~)

<.

,n .;

"nj":"

t
~
"

The increasing complexity and diversity of electrical technology, hoth

theoretical and applied, and the limits of time set by the conventional
four-year college program make it necessary to suhordinate the tktailcd
behavior and design of conkmporury devices to scientific concepts ;lIld
the techniques for their exploitation. This subordination of engineering
practice to scientific principles is today receiving strong stimulation from
the fact that advances in today's electrical technology often depend more
strongly on the exploitation of scientific fundamentals than on current
practice. Even so, examples of engineering practice still have a place
in the program but their function should be to provide a bridge between
scientific abstractions and the engineering world. Within these limits
we have made a critical review of the area of electric machines and
transducers and formulated our view of the broad principles of electro
mechanical energy conversion which are the foundations of contemporary
practice and which are important to today's electrical engineering.
In considering electromechanics as one cornerstone of electrical engi
neering, we felt that it should be placed on a firm and well-defined
theoretical base, one which shows clearly the interrelations between
electromechanics and other energy conversion systems. For this reason
we chose as a starting point the definition of state functions and state
variables of a given system and have used these via well-known principles
of analytical dynamics-Hamilton's principle or alternatively the conser
vation of energy and arbitrary displacement-to obtain the equations
of motion of a given system. We feel quite strongly that the analysis of
electromechanical devices built around steady-state concepts using equiv
alent circuits and the elementary ideas of network theory are not a

..

ix

...- -

-------~._---------~---------~."'-

'Jill'

J'

PREFACE

PREFACE

sound base on which to build an understanding of electromechanical

interactions. Although such an approach is useful for solving some
pre~cnt-day problems. it does not give adequate accent to the more
general problem of electromechanical interactions.
This book has been strongly influenced by three sources. The initial
impetus and major driving force was the formulation of a required, one
semester. senior course in electromechanical energy conversion that
would treat in some depth fundamental concepts and at the same time
treat specific transducers in sufficient generality that physical insight into
transducer dynamics could be obtained. It was felt that with the ever
increasing use of more accurate feedback control systems, the engineer
must be aware of the dynamic properties of the transducers that he is
substituting for human muscle. The second source of information was
research sponsored by the U. S. Air Force in which the dynamic behavior
of aircraft generating systems was studied. The third source was gradu
ate teaching and thesis research.
Since electromechanical energy conversion can occur only by the
interaction of electromagnetic fields and material bodies in motion, it is
reasonable to begin with a treatment of macroscopic electrodynamics.
We have experimented with this starting point in our senior course and
find that a student who masters this topic has excellent preparation for
proceeding to more advanced work. But we also have found that for
the average student the time required for an adequate coverage of elec
trodynamics leaves little time for a treatment of lumped-parameter sys
tems to the degree necessary if dynamic behavior is to be studied.
Furthermore, it is in general impractical to describe the general dynamic
behavior of energy conversion systems only in terms of fields. These
considerations, plus the presence in our curriculum of a junior course
in fields, energy, and forces which introduces electrodynamics, prompted
us to start with a lumped-parameter approach with some references to
field theory as the basis for determining when lumped parameters can
be used and in defining the system parameters.
Having chosen the lumped-parameter approach, we chose to make
the derivation of equations of motion in Chapter 1 quite general, the
principal limitation being that the coupling system (Le., energy storage
in fields) be conservative. Two approaches have been used: the method
of arbitrary displacement and conservation of energy, and Hamilton's
principle leading to Lagrange's equations. The first approach yields
only electromechanical coupling terms; consequently, the equations of
motion of a system must be written using force laws (Kirchhoff's laws
and d'Alembert's principle), and much bookkeeping must be done in

complicated problems. The Lagrangian formulation, on the other hand,

can be generalized to include the nonconservative parts or the system,
so all bookkeeping is done automatically. Some educators may feel that
the use of the Lagrangian formulation may tend to make crank-turners
out of the students; however, we feel that the Lagrangian has further
significance than just being a tool for solving problems. It introduces
students to variational principles which are in one sense as fundamental
as the conservation principles. The Lagrangian formulation also intro
duces generalized coordinates and lays a firm foundation for the mean
ing of independent variables. The Lagrangian state function and the
Legendre transformation provide an entry into state functions in general
and a tie with state functions commonly used in thermodynamics. We
feel that these advantages ot" the Lagrangian formulation make its intro
duction well worthwhile. This is in line with the desire to broaden the
scope of the present treatment and to lay a foundation which in the
future can be expanded to include systems other than those with only
electrical and mechanical in teractions.
Chapter 2 treats mathematical techniques for solving equations of
motion, starting with the straightforward classical solution of linear
differential equations with constant coefficients and ending with the use
of analog computers to solve nonlinear equations. As is obvious from
the content, the purpose of this chapter is not to teach mathematics but
to illustrate the use of known techniques in analyzing typical transducers.
No attempt has been made to include all the known methods for analyzing
linear equations. Notable omissions are electrical analogs, flow graphs,
and operational tcchniques other than Laplace transforms. We feel
that the techniques used are representative. No examples of the use
of digital computers are included. However, we recognize the increasing
importance of digital computation in system analysis and the general
equations derived for electromechanical energy convertors are directly
applicable to digital computation.
Chapters 3 and 4 represent an errort to integrate the earlier work

of Kron, Gibbs, and others into a unified treatment of the dynamics of

rotating machines. By defining a general physical model consisting

of concentric magnetic cylinders with current sheets on their surfaces,
a field solution is obtained, parameters are defined, and equations of
motion are derived. Starting from a single model, the complete dynamic
equations of most conventional machines and many unconventional ones
are obtained by selected constraints to the general model. Transforma
tions are introduced that mathematically describe the physical change
of variables made by a commutator and put the equations of motion in

t,,:

I' !;,

~ '~.'
~ 7,
'l)'
{,\

I~ .

'.

;~i;

I~ ~

~ ~

~,

;
., ;,..

I",;

_ -----oL

xi

(.

'...

xii

J.'.

I,ii,

,
'"

,'

':!i.

'.
~

~.;

",I
},~,

;:,

'i

".

PREFACE

more recognizahle and solvable forms. Thus one physical model with
one set of equations dcsnibes all of conventional steady-state machine
theory, the dynamic behavior of conventional machines, and the steady
state and dynamic characteristics of many unconventional machines.
The usc uf the generalized approach with the aid uf a Iahoratory
machine having the same generality reduces steady-state machine theory
to the solution of steady-state a-c and d-c circuit prohlems with which
our students arc fully familiar. Furthermore, the general field solution
allows interpret:1tion of energy conversion properties in terms of fields.
Since so little time is required for this aspect of machine theory, much
time remains for study of the dynamic properties which are so important
in the engineering of today and which will become more important in
the future.
There may he some question about the choice of a two-phase machine
as the deviL'e for teaching machine theory. It is shown in Chapter 10
that the energy conversion characteristics of any polyphase machine
with symmetrical impedances can be obtained from an equivalent two
[,hase machine for which the equations of motion arc ohtained by a
str:llghtforward change of variables. We feel that the mathematical
cUll1pkxily of L'ven a two phase systcm tends to ohscurc many of the
concepts contained in the treatment. The use of a three-phase o;ystem
as the analytical vehicle only introduces additional mathematical com
plexities and further detracts from the understanding of the nature of
energy conversion in rotating machines.
Chapter 5 gives a simple introduction to feedback control system
theory. We feel that such an inclusion is desirable because those students
who do not take such a course as an elective should nevertheless be
exposed to the essential ideas involved. In addition, this chapter illus
trates through an example how the dynamic behavior of an intercon
nected system of machines is affected by the characteristics of the
machines.
Chapters I through 5 contain the material prepared for a first-term
senior course. More recently, it is being taught as a second-term junior,;
course. The classroom work is augmented by a laboratory in which
transducers and the generalized machine are used to get the fundamental
concepts across. In addition, commercial machines and transducers are
used in experiments that stress dynamics, interconnected systems, and
feedback control theory.
Chapters 6 through 9 are detailed and specialized treatments of spe
cific devices. Their purpose is to illustrate techniques and to provide

PREFACE

xiii

information on applications of the general approach of the earlier char:

ters. This material is primarily intended for use in graduate courses
and for the use of practicing engineers. It has also been used as part
of a one-semester senior elective subject on electric machine systems.
Chapters 1() and 1 I are necessary for generality in the treatment.
Chapter I O.provides the analytical methods for reducing any n-m phase
machine to an equivalent two-phase machine for considerations of dy
namic energy c(lnversion properties and thus justifies the lise of the
two-phase model in the carlier treatment. Chapter 11 shows how the
results of the analysis of Chapter 3 with sinusoidal currcn t shcL:ts Can
be extended to analyze a machine with current sheets (windings) of any
arbitrary distrihution. It also contains a rigorous justification for the
concept of analyzing a commutator by a simple transformation of vari
ables. This one concept is of course vital to the establishment of the
unity among aIt machine types.

The topics in this book were first presented to a group of seniors at

MIT in the fall of 1954, and the material prepared at that time was a

group efTort. It h:1S sincL: gonc through several modifications, but lllany

of the basic ideas around which this text is formed were estahlished by

the initial group, which consisted of the authors plus Professors MahmOUd

Riaz, Robert M. Saunders, I-Jerman Koenig, and Richard H. Fnlzier.

In addition to the initial group effort, special recognition is due to

Professor Mahmoud Riaz for his contribution to Chapters 4, 6, 8, and

10; to Professors Leonard A. Gould and Robert M. Saunders for their

contribution to Chapter 5; and to David Bobroff for his contribution

to Chapter 1. Others who have made many contributions to the effort

have been Professors Richard B. Adler, Lan J. Chu, David J. Epstein, _

Robert M. Fano, Charles Kingsley, Jr., Alexander Kusko, Osman K.
Mawardi, Norman H. Meyers, and Karl L. Wildes. Special mention
should also be made of the very excellent works of Drs. Gabriel Kron
and W. J. Gibbs, whose very fine publications in this field were an invalu
able aid to the development of many of the concepts presented here.
The authors are also greatly indebted to Bernard Lovell, whose dili
gence in editing and checking derivations was invaluable. We wish to.
thank also the Misses Lucia Hunt, Ruth Coughlan, Lydia Bonazzoli, and
Evelyn Fraccastoro for their perseverance in typing the drafts of the
manuscript.
Special mention is due our wives Glorianna G. White and Blanche S.
Woodson for the invaluable assistance, both tangible and intangible,
given during the writing of this text.

j
I

Above all, the authors wish to express their appreciation for the

 .. , ifV"\1'

PREFACE

xiv

S. Brown,
dynami c leadersh ip and stimula tion provide d hy Dr. Gordon
would not

book
this
without whme foresight, perseve rance, and courage
have been possible.

C.

DAVID

WHITE

HERBER T H. WOODS ON

Cambridge, Mass.

Novemb er, 1958

n;
l.;

11~
,riU

Chapte r 1

The Dynami c Equatio ns of Motion of Electrom echanic al

Systems

2 Analytic al Techniq ues for Treating Electrom echanic al Equa

tions of Motion Includin g Typical Transdu cers as Example s

.
.n
../

f"

''l

"",,) . OU1; ~i

,.~~

'I,

,r,

(1

~';,

<.>1t1

III

.~1~:k}~~i

.~.~

);f

\rf!~r"'!

.j.

"'I',

. llj

'-:1:!

')".1:

.... !

170

4 Two-Ph ase Transfo rmation s and the General ized Machine

254

5 Fundam entals of System Dynamics

360

Dynamics of Transdu cers

395
421

Dynamics of Induction Machin es

477

Dynamics of Synchro nous Machin es

508

10

General ized Analysi s of the nm Winding Machine

545/

11

Space Harmon ic Analysi s in Machin es

603

-'

~.~

'~:;:'

2;

The General ized, Magnet ic Field Type, Rotating , Electro

mechan ical Energy Convert er

7 Dynamics of Commu tator Machin es

) ~~

,~~l:,

tt

+t

';of

I'

87

,Ii;

~i ~
. '~

Index

639

;,;;1

I:

'1'1 i
~.

. ~tt4

!:'.m
'

..

'JO"IIJ.b

Xy

"'--

The Dynamic Equations

of Motion

of Electromechanical Systems

,I
.:~:;

1.0

Ii1

Electromechanical energy conversion is the result of electrouynamic

interactions and a rigorous and thorough treatment requires a st1ldy of
moving, charge-carrying material bodies in electromagnetic fields. How
ever, for quasi-static (low-frequency) and low-velocity electromechanical
systems the dynamic equations of motion can be formulated to a high
degree of accuracy in terms of lumped electric circuit parameters that are
evaluated from static field solutions.'"
The subject of this book is the characterization of electromechanical
systems by lumped parameters. This involves three major problems: (I)
a physical description of the system, (2) derivation of the differential equa
tions of motion for the system, and (3) solution of the equations subject
to the specified operating conditions of interest. The physical description
of the system entails the establishment of an idealized model whose
character is determined by the physics of the problem and whose
completeness depends on the application. The equations of motion
are obtained from the idealized model in several ways, some of which
are covered in the present chapter. The solution of the equations of
motion is in general difficult because the equations for any energy con
version system are nonlinear. for some applications the equations can
be linearized (e.g., for small signals); in others a change of variables
leads to simplification. In a few cases it is necessary to solve the

* For a discussion of the relationship between circuits and fields for systems in
relative motion see R. M. Fano and L. J. Chu, Fields, Enel'!:';'. alld Forces. John Wiley.
New York, 1959.

'{-; '.

rf.
1

,W
. ~UJ

,;;
,~

,+

\
~i:.'
1
~
i

.. ~

.~'

,,'

,
I~

,:;
'-,';

fit
l~'

/,

.........

' ..'. .

Introduction

\
)-

-_.

:~.fIo...

...........j;

_~_

ELECTROMECHANICAL ENERGY CONVERSION

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

complete, nonlinear equations, and machine computation techniques may

bc requi red. The types of nonlinearities encountered and the simplification
possible depend upon the electromechanical systems and their applications.

1.1

The Various Approaches to the Study of the Dynamics of

Electromechanical Systems

The dynamic equations of motion of electromechanical systems can be .

determined from physical laws using either force density from electro
magnetic field theory or an arbitrary displacement and conservation of
energy to obtain the mechanical forces of electrical origin. Alternatively,
the equations can be obtained from variational principles applied to
selected energy functions. It is difficult to state if either of these approaches
is more basic, particularly since both will lead to the correct results if
properly applied. There is, however, a great difference between them as
to the formality of the analytical techniques employed.
The application of force laws to obtain the equations of motion is
probably the least formal analytically and requires a good deal of insight
and judgment when dealing with complicated systems containing many
variables. The electric coupling terms resulting from mechanical motion
and the equations of motion for the electrical and mechanical parts of the
system are obtained from known force laws, such as Faraday's law,
Coulomb's law, Kirchhoff's law, and d'Alembert's principle. The method
of arbitrary displacement and conservation of energy is then used to
obtain the mechanical forces of interaction between systems, because it
lends itself easily to the derivation of mechanical force equations expressed
in terms of electric circuit quantities. Alternatively, these coupling terms
can be obtained by integrating force densities obtained from electro
magnetic field theory.
The derivation of the equations of motion from variational principles
is significantly different from the previous method. First, one establishes
a common terminology for all types of systems, whether electrical,
mechanical, thermal, acoustical, etc., by defining state functions (energy
functions) in terms of sets of generalized variables. Then by the use of a
single fundamental postulate, e.g., Hamilton's principle, the equations of
motion for all systems including any coupling terms are determined. The
variational approach is quite formal analytically; and, as a result, insight
into physical processes can be lost in the mathematical procedures.
However, if the method is properly understood, and if adequate attention
is given to the selection of the state function, physical insight can be gained
because of the generality of the method. The variational method is one

of the most pnwcrful techniq lies of dynamics; and, .tltllout!h beginners

may tend to become "crank turners," the long-term value of variational
techniques precludes their dismissal merely on the ground of conceptual
difficulties or lack of physical insight when employed by the novice.

+
1.2 Fundamental Relationships in Electromechanics
The fundamental force relationship of statics is that for equilibrium
the summation of all forces acting on a body is zero. This basic concept
was used by d'Alembert who postulated that the sum of all forces equals
zero for dynamic equilibrium of mechanical systems, * For a multiloop
dynamic system d' Alembert's principle requires that at the kth mechanical
nodet

2: (A, I-I
<

where

AI

d~;, = ~ (mk,xk

fk)

= 0

(1-10)

= the ith inertial force on the kth node.

:''I.:~

fk j = the ith applied mechanical force, including any forces of

constraint at the kth node

:~~

In addition to satisfying d' Alembert's principle for dynamic equilibrium,

the continuity-of-space relationship must be satisfied. That is, around
any mechanical loop (kth loop) the summation of all displacements or
velocities must be zero.

~,~t; I'

2:

I~t

Xk l = 0

(l-lh)

where '\'k j = the ith velocity in the kth loop.

Similarly, for electrical systems, the dynamic equations of equilibrium
established by Kirchhoff are of the form of a summation of forces set
equal to zero and a continuity-of-charge equation. t For electric circuits

;zvt,
;'}.

-..i~"I\

.-'.'
... .'.

~.~"

. .f

r',;

of

For additional details, see H. Goldstein, Classical Mechanics, Addison-Wesley,

Cambridge, Mass., 1953.
t The terminology and definitions of this chapter are established assuming that a
physical system is replaced by its mechanical and electric equivalent circuits. Thus a
.. mechanical node" or .. mechanical loop" is a node or loop in the mechanical equiva
lent circuit and each mechanical node has only one degree of freedom and is described
by one coordinate, Xk .
t For additional details see E. A. Guillemin, Introductory CirCllir Theory, John Wiley,
New York, 1953.

it

..

-J"
i
-,t,

l~' .

~ /

..

,.

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSl EMS

ELECTROMECHANICAL ENERGY CONVERSION

"

=0

(I-Ie)

i--l

$'

where ek, = the ith voltage in the kth loop.

The continuity-of-charge relationship or continuity of current is
expressed by stating that the sum of all currents into a node (kth node)
must equal zero.

L
I-I

i kl = 0

1.1.1

,
'I
if,i,'",',,",

I
1

"--

Mechanical Forces of Electromechanical Coupling Derived by

an Arbitrary Displacement and Conservation of Energy

The first step in analyzing a complicated electromechanical system by

arbitrary displacement* and conservation of energy is to reduce the system
containing electromechanical coupling terms to a minimum. To do this,
separate out all purely electrical parts and all purely mechanical parts of
the system including losses, as shown in Fig. 1-1. This separation
procedure is carried out to the extent that each electrical terminal pair is
For treatments of the principle of virtual displacement, see Goldstein, loc. cit., or
E. T. Whittaker. Analytical Dynamics, Dover Publications, New York, 1944.

fiectromechamCOtI System

._

lQualllJlls of motton from

Kirchhoff's laws

__

t
"

ElectrICal

Input to
coupling fields

Electromechancal
Network

t: -

+ '-

i:t

Mechanical

mpul fo

system

Mechanical Network:
EQuatlolis of moflon from
d'AleflltHHI's prillclple

and Conllnl,Jlty of ipace

ji

(repfesentlng coupling
fre!dc,-electrlC .Illd IT1.I~rlr.ll()
!:.ljUatlOfls 01 Fllohon from
conSerVatIon 01 t'nt"I~Y and
,.HI arblltMy ulspl.lC\.'rlIent

I
I

(f,),

Mechanical
Input 10
coupling fields

JI

Simplification of electromechanical system for analysis by an arbitrary dis

placement and conservation of energy.

Fig, I-I.

1<

il.'

iJ

The separation procedure results in the general conservative electro

mechanical coupling network depicted in Fig. 1-2 in which there are n
electrical terminal pairs and m mechanical terminal pairs. By virtue of
the lumped-parameter approach, each electrical terminal pair will be
coupled to either a magnetic field energy storage or an electric field
energy storage. To fix ideas, assume that the electrical terminal pairs
I ~ i ~ I are coupled to electric field storage and terminal pairs,
I + I ~ i ~ n are coupled to magnetic field storage. The total stored'
energy W in the coupling network is given by
W = We

Wm "

(1-2)

where W, is energy stored in electric fields and W m is energy stored il)

magnetic fields.
The lumped-parameter representation of the coupling network of Fig.
1-2 requires the system variables to be functionally related. For example,

. __,
~

'.1_

flectucal Network'

(Ieetflcal
cnput 10
sY5tem

ail

~'

Compte-t~

r------------------------.

(1- Id)

where i Aj = the ith current Howing into the kth node.

D'Alembert's principle, Eq. I-la, the continuity-of-space relationship,
Eg. I-Ih, and Kirchhoff's laws, Egs. I-Ie and d, express the complete equa
tions of motion for electromechanical systems providing the mechanical
forces of electrical origin are included in d'Alembert's principle, Eq. I-la,
and the electric voltages and currents used in Kirchhotf's laws, Eqs. I-Ie
and d, include the effects of mechanical motion.
The fundamental laws needed to study the dynamics of electron'~~hanical
devices from a circuit viewpoint are now complete. Unfortunately, the
mechanical forces of electromechanical coupling, when expressed in a
form easily used for electromagnetic field problems, are not readily
adapted to an equivalent circuit treatment of connected electrical and
mechanical systems. The direct extension of the macroscopic force
equations in terrr.s of electromagnetic field quantities to force equations in
terms of electric circuit quantities is difficult by integration methods unless
the physical device is quite simple. It proves easier to derive these
coupling terms, using an arbitrary displacement and the conservation of
energy.

coupled to one energy storage, either electrical or magnetic. Any internal

interconnections between circuits that are coupkd to different energy
storages are included in the external electrical network. The Illechani,cal
variables represented by the mechanical terminal pairs arc those whi~h
affect energy storage in the ekctric and magnetic fields. Any purely
mechanical couplings between mechanical variables such as gear trains,
springs, etc.. are included in the external mechanical network.

the force equation is expressed as follows: the sum of all voltage drops
around a loop (kth loop) equals zero,
~ ek,

..=eM....

r.

ELECTROMECHANICAL ENERGY CONVERSION

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

..~

if the electromcchanical coupling network can be characterized by.

.,

electrically lincar inductanccs. the kth nux linkage I~ givcn hy

-"-...
+

.:\k =

2:"

i =,11 I

Lk;ii

(1-30)

("1

where tL.~ inductances L Ai are functions of the mechanical coordinates

,xm '

Xb

il;
+-...:....-~--Electrical

Inplll~ to

coupling

fields

L Ai == Lki(x" ... ,xm )

U.

Similarly, for a system of electrically linear capacitances there would occur

0-=---

--

qk

in

fj

Xm ,

~({')I

flux linkage and current of kth inductor:

.:\k

fk

g,~

~~

~q

~--- ({.Ik

"

,.!~

'1'''

\:

!~

- - ({.Im

%"1

,1

,
"'
1"

Fig. 1-2. Dcfinition of coupling system for analysis by an arbitrary displacement

and conservation of energy.

.~

(1-3d)

I~

xI> ... , x",)

( 1-4a)

ik = ik("I+I, ... , '\,,; xI> ... , XIII)

( 1-4h)

voltage and charge of kth capacitor:,~!

f10rJ ,,'

0-------

= ,\(i/+" ... , i,,;

or

.:tR

-...::~----

X",)

In general. the system may be nonlinear; thus a set of parameters Land

C cannot be defined. In such cases only general functional relationships
among the variables can be established, such as

XI

X.

(I-Jc)

Cki/l i

C kl = Cki(X" ... ,

Electromechanical Network:
(representing couplmg fields
-electric and magnetic)
All dissipative elements removed
to external circuits; therefore
system is conservative.

0-:;

2:I

Where the capacitances Cki are functions of the mechanical coordinates.

Xl' .'"

i,~

Un

Mechanical

mputs to
coupling

fields

( 1-3b)

;\.-""

~;

r"J.

",,'"
't~

Uk

= Uk(qlo . . . , ql; X Io .. , X",)

(l-4c)

qk

( 1-4d)

or
qk(l'lo .. ,

VI; XIo , X",)

In any case, regardless of whether the relationships between the varia hies
are linear or nonlinear, the relations are restricted to be single-valucd
functions because it is assumed that the energy stored in the electro
mechanical coupling fields can be described by state functions.
The assumption that the stored energy in the electromechanical coupling
network is a state function forces it to bc a single-valued function of the
system variables. independent of the derivatives and integrals of the
variables. Thus. the stored energy W may bc a function of the instan
taneous configuf<ltion of the system, but must be independent of the
dynamic slate and of past history. The requirement that energy he a stale

':..

ELECTROMECHANICAL ENERGY CONVERSION

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS . 9

function is a stronger one than the requirement that the expressions of

Eqs. 1-4a-d be si ngle-valued functions. Stored energy that is a state
function always yields singh~-valued internal constraints; hut single-valued
inl\:rnal constraints alone do not guarantee that the stored energy is a
state functillll (see Sec. 1.4.6).
In sUllllllary. the constraint of requiring the energy functions to be state
functions places the following limits on the electromechanical system:

can be only three indepcndent variables, the other lhn:!: variah1l.:s heing
required to satisfy the interna~constraint equations of the form
Vi - 11;(lJ" ... ,q,;

il

i/+1> . ,

i,,; x" ... , xm;fJ, .. ,fm)

(1-5a)

W =

o, ... ;0;0

+ i;(>\;r\>
+ f;(q;,

tj

I.;
"

;~/

j
l

q'q,:AJ! 1.,\n;Xtxm

0;0 ... 0

, A;,;

, q;;

'!.

'!!

L .L

i~

[v;(ql"'"

j~l

"

ql; Xl' , Xm ) dq,

xi, ... , x;,,) dA;

1.;+10' .. ,

A;,;

..~~.,

"

X;, ... , x;") dXiJ

~ ~i~

See Fano and Chu, loc. cit.

Xh . . . X,.,)

( I -flh)

( 1-6(')

W(ql"'" ql; 1.'+10 An; Xh . . . ,Xm )

( 1-6d)

Even though qi' \, and x j have been considered as the independent

variables above, the single-valued relations of Eqs. l-6a-c can be solved
to allow any three of the six variables to be treated as independent.
In gener,l!, in electromechanical systems it is possible to evaluate con
straint equations 1-6a and h by solving for the electric and magnetic static
field solutions determined by the physical configuration. The evaluation
of the constraint equation for the force fj is usually much more difficult.
Thus it is common practice to find the constraint relationships for the
electrical variables and to use these to evaluate the stored energy, utilizin'g
the fact that since the energy is a state function the correct energy function
can be obtained by keeping all electrical variables Ai and qi zero when
assembling the mechanical system and then holding all the displacements
x j constant when establishing the final values of the electrical variables.
For this path of assembly of the system the energy function becomes

W(qh ... ,ql; 1.1+ 10

. .t.~.

(1-5b)

where primes denote variables of integration. The energy function of

Eqs. I-Sa and h is evaluat!:d by performing the line integral over all the
system variables. The constraint of the proper single-valuedness of the
functional relationship indicated in Eq. J-5h to yield a state function allows
the evaluation of the line integral by choosing any convenient path of
integration. Also, because of the interdependencies of the variables in
Eq. 1-5b. it is clear that of the six variables 'Ii' Vi' ii' Ai' fj, and xj there

An;

W can be expressed as a function of three independent variables. which lor

the constraint equations I-oa-(' are 'Ii' ,'1/, and x j yielding :In energy
function

Ww~re,

can be evaluated as

1 , ,

(1-611)

In view of these constraint equations, it follows that the energy function

Restricting the energy functions which describe the electromechanical

coupling lields to b<: .,tate functions establishes that the etlergy stored in the
coupling lields can be determined by bringing all system variables to their
final values ill any arbitrary manner. Referring to Fig. 1-2, the euergy to
establish the electromechanical coupling llelds can be evaluated by
summing all the input energies at all the terminal pairs in bringing thOl
sy.,tem variables from zero to some final vahle. Thus the stored ener~A

= W(q" ... , q,; v" ... , VI; ,'1'+1> . , An;

i,(A/1

jj = jj(qh ... , 'II; A'+h' , An; Xh . . . , X",)

I. The lumped parameters evaluated from electromagnetic fields must

be derivable from static (zero-order) fields. *
::. The functional relationships between variables (Eqs. 1-4a-d) must
be single-valued.
3. Hysteresis cannot be included in expressing the functional dependence
of the variables in Eqs. 1-4a-d. This does not mean that the loss due to
hysteresis cannot be represented by a resistive element located outside the
coupling sy.,tem.

x",)

Xb . . . ,

An; x .. ... x m)

qlt .. lq,; ),,+l ... An;xll ...xm

0..... 0;0.....0;,l'1..........

+ i;(A;+1 ... , A~; XI>

L [v;(q~,
/_1

xm)

... , q;; Xh x m ) dq;

d,\;J

(1-7)

'jr

The energy functions given by Eq. 1-7 can then be used to find the force
constraints fj and hence the force of electromechanical coupling.
The forms of the stored energies and the nature of the network in Fig.
1-2 indicate that n electrical variables and m mechanical variables can be
specified independently, This can be interpreted as meaning that the
coupling network has (n + m) degrees of freedom. With the usual
interpretation that a degree of freedom represents one independent energy
storage element in a lossless system, the coupling network appears to

10

ELECTROMECHANICAL ENERGY CONVERSION

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

have (n + m) independent energy storages in spite of the fact that in

Eq. 1-7 energy storage was assigned only to each of the n electrical
terminal pairs. This apparent ambiguity is resolved by realizing that
mechanical energy supplied at the mechanical terminals goes into either
electric or magnetic field storage. Consequently, in a coupled electro
mechanical network the electric and magnetic fields can be interpreted
as storing both electrical and mechanical energy and hence can represent
both electrical and mechanical degrees of freedom.
With the electromechanical coupling network of Fig. 1-2 specified, the
terminal characteristics of the network can be found. According to the
discussion above, one variable at each of the (n + m) terminal pairs can
be specified independently. First, consider the I electrical terminal pairs
that are coupled to electric field storage. When the qj and Xj are specifie.
independently, the current in the ith terminal pair is t
.

I
I

dqj
=

dt

dXk occurs in the kth mechanical coordinate in time dt with all other
mechanical coordinates held fixed, i.e..

dXJ

I'

d\

di

for j 'I- k

The electrical variables change during the time ill. and these changes are
completely arbitrary except that the internal con~traints of Eqs, 1-6a and b
must be satisfied. This means that only one electrical variable at each
electrical terminal pair can be changed arbitrarily. During the arbitrary
displacement the conservation of energy must hold. The various energies
involved in the arbitrary displacement are:
energy supplied at electrical terminals =

..

(I-lOa)

= - (I.h-':k dt = - (le)k dX k
(I-lOb)

change in stored electrical and magnetic

energy of coupling field = dW

(I-JOe)

where W is the total stored electrical and magnetic energy of the coupling
fields (Eq. 1-5b)
energy lost in dissipation

(1-9)

and the current i l is given by the internal constraint of Eq. 1-6b. Thus,
the Yolt-ampere characteristics at the electrical terminals have been
determined so the effects of the electrical part of the coupling network
can be ineluded in the equations for the external electrical network
described in Fig. 1-1. It should be mentioned that instead of specifying
qj for the I electric field storages and A; for the (n - I) magnetic field
storages, the voltage Vi at the electric field storages and i j at the magnetic
ficld storages could have been considered as independent, in which case the
internal constraints of Eqs. 1-6a and b would be used to obtain qj and AI
for inclusion in Eqs. 1-8 and 1-9.
The next problem is to find the force due to the electromechanical
coupling. Since the m mechanical terminal pairs are characterized by
m independent variables, it is possible to consider each mechanical
terJl1inal pair individually to find the electromechanical force. Defining
the force (le)k shown in in Fig. 1-2 as the force applied to the kth
mechanical coordinate (node) by the electromechanical coupling network,
the force (le)k can be found by considering that an arbitrary displacement

2: v,i; dt
1=1

and the Yoltage Vi at the ith terminal pair is given by the internal constraint
of Eq. 1-6a. Next, consider the (n - I) electrical terminal pairs that are
coupled to magnetic field storage. When the Aj and x j are specified
independently, the voltage at the ith terminal pair is given by Faraday's
law
V, =

=0

energy supplied at mechanical terminals

(l-8)

II

.f

(I-IOd)

All lossy elements are removed from the network of Fig. 1-2.
The conservation of energy requires that the sum of the input energy
must equal the change in stored energy, thus, from Eqs. 1-10:

-(i')k dXk +

!,

.,it.

=0

V,il

;=1

dt = dW

(1-11)

From this expression the force applied to the kth mechanical node by the
electromechanical coupling field is:

(f,)k = d.i.... (
Xk

;=1

viii cit - dW)

( 1-12)

When the n independent electrical variables and the fit independent

mechanical coordinates are specified, Eq. 1-12 gives the force applied to
the kth mechanical node and the velocity of the kth mechanical node is
dxk/dt = Xk' Thus the force-velocity characteristics at the mechanical
terminal pairs of the electromechanical network of Fig. 1-2 have been
specified. The force of Eq. 1-12 can be included with d'Alemhert's

it

ELECT ROMEC HANIC AL ENERGY CONVE RSION

12

node aN
principle to write the equilibr ium equatio n for the kth mechan ical
(1-13)
ilk - Uk + Umh + (j,)d = 0

if

;1
~'

in

A system of current -carryin g coils is represe nted schema tically

g

couplin
agnetic
electrom
er,
first-ord
or
nt,
importa
Fig. 1-30. The only
for

and
statics'
quasi
for
k
networ
this
of
ts
elemen
field between various
e the
low-velocity mechan ical motion is the magnet ic field. To evaluat
AI
linkage
stored energy in the magnet ic field, assume that the final flux
are
Xl
coils
in each of the coils in Fig. 1-30 and the final position s of the
ints
obtaine d by any arbitrar y paths compat ible with the internal constra
their
and
of both the electrical and mechan ical variable s between zero
all non
final values. Assumi ng all mechan ical storage elemen ts and
to be
ts
elemen
coupled electrical storage clemen ts plus all dissipat ive
of J
ation
conserv
externa l to the electrom echanic al couplin g field, the

,:,f:}:~

Mechanical Force Due to MagnetiC Field Couplin g

1..,.

j-

;;

I.."

~.

as

W m = input electrical energy

j, 'l(lrct!

input mechan ical energy

(1-14a)

or

W",(AI>' .. A,,; x., ... , x",) =

, 1-"1>"

II

''"/II

0 ..... 0

A, ..... Ann

.~ i;(A;, . .. A:,; x;, . .. ,<..l tI,l;

J.

o.... 0

,. I

~~ f/Al, ... , A",

,

.,,:-

'.'

XI' .. ,

J .. I

I.'

xm)

d'\j

(1-14b)

of
A graphic al plot of the total energy W m' for the general case
function
alued
single-v
ar.
nonline
A;(x;, ... x;"; i;, ... , i;'), where A; is a

I
I

.~

~rlca,

sy~tem

---

.J:'\
'.

':.'

'

f'
Fig. 1-3a.

Coupled current-c arrying coils.

given
of the displac ements and current s, is shown in Fig. l-3b. For any
total
the
ii.
and
x
the
of
j
system, where A; is a single-v alued function
ters
parame
the
by
y
uniquel
ned
stored magnet ic energy W m is determi
these
of
values
final
the
upon
s
XI>' .. , x", and AI,' .. ,A" and only depend
. The
parame ters. The total stored magnet ic energy is a state function
al
electric
by
d
supplie
is
which
energy
ic
amoun t of the storeL! magnet
upon

depend
sources
ical
mechan
by
d
supplie
sources and that which is
energy

how the system is assemb led in reachin g its final state. The

...

j
..

Machines,
For alternate treatments see E. Fitzgerald and C. Kingsley, Electric
Force
ical
"Mechan
Park,
H.
R.
and
Doherty
E.
R.
1952;
York,
McGraw -Hili. New
between Electric Circuits, " Trans. AlEE, Vol. 45, 1926, pp. 240-252.

energy estahlis hes that the energy input from all sources
magnet ic field energy W",(A" ... , An; XI> . . , X n ,).

1.2.2

13

f..

force,

where Pk is the inertia force, fk the externa lly applied mechan ical
the
of
three
All
springs.
ical
mechan
and (t;"h the force applied hy
1-l.
Fig.
of
network
ical
mechan
the
in
d
include
be
to
terms arc conside red
were
In the derivati on of force in Eq. 1-12 the electrical varia hIes
of

ints
constra
internal
the
with
ible
compat
ily,
arbitrar
allowed to vary
of
ess
Eqs. 1-6a and h. Conseq uently, this force is the true force regardl
g

couplin
the
on
t!le externa l lermina l colIslraints that may he impose d
to

is
This
system.
ical
network hy the externa l electrical and mechan
satisfie d
emphas ize the point that only the inlernal constra ints had to be
in the arbitrar y displace ment.
s
The mechan ical force (f,h of electrom echanic al couplin g contain
and

(W,)
al
electric
fields,
storage
terms due to two types of energy
ical .

magnet ic (Wm ). At low electric al frequencies and low. mechan

:'"
lumped
by
nted
represe
be
can
system
al
electric
velocities such that the
and.

ances
capacit
in
be
will
(W,)
storage
parame ters, all electric energy
,he two:
all magnet ic energy storage (W m ) will be in inducta nces. Thus,
ely
separat
treated
be
can
g
couplin
field
ic
magnet
and
problem s of electric
l

termina
or
l
externa
proper
the
with
ed,
combin
be
and the results can
field

ic
magnet
and
electric
both
having
system
a
constra ints, to describe
of these

couplin g with a mechan ical system. The forces caused by each

.

sections
two
next
the
in
treated
be
will
two couplin g fields

SYS rEMS
EQUAT IONS OF MOTIO N OF ELECT ROMEC HANIC AL

ELECTROMECHANICAL ENERGY CONVERSION

14

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

",

j
'I

supplied from ekctrical sources and mechanical sources 1',)1' a particular

path of asst:J1lbly or Ihc system shown in Fig. 1-.111 i, drawn in hg. I-.~c.
From these ligures it should be clear thal it is possible to arbltral'lly
supply any portion of the total energy from either source by the prllper
choice of path used to assemble the system. For example, hold all !lux
linkages at zero and mechanically assemble the system, and then establish
the flux linkages with the mechanical coordinates held at their final values.
The energy stored in the magnetic field for the given values of flux linkage
will have LV be supplied by electrical sources. For this case, the stored
magnetic energy is

Wm(AI> ... , An;

ij

&1

x m)

XI . ,

"I .... '''" n

=
J 2 : i;(/,;, ... , A~;
0, .... 0

1
,~!

15

XI, . . . ,

,-I

x m ) dA; (1-15)

where Wm is evaluated as the integral of idA for any fixed spacing, i.e.,

all Xl are constants.

. 1t2

EXAMPLE lEI
X2(2:i..:c,;;;i{.
i-z

(Stored magnetic

ii

.i~)

i2

As an example of the method of evaluation of the integral of Eq. 1-14b.

consider the case in which

ii

energy~i""
i~d)"~
0

n=3

~~~~

Io

'1

i l = TllA l + T 12 AZ + T 13 A3
i2 = TZ1A I + Tn)o.z + T 23 A3
i 3 = T 31 A1 + TnAz + T 33 A3

Magnetic coenergy.

i"

'$' "

and the system is electrically linear

A'di'
n
A

-"I

..

, . f.

(lEI-I)
~'" /.
~ \1~)

with the T's functions of the x j only and

t 71

i~

Fig. I-lb. Path of operation to reach

final energy when mechanical coordi
nates are held at xi = XI' xi = x 2 , ,
X~ = x m while the electrical variables
are simultaneously brought to their final
values.

ill

i~

Fig. I-le. Path of operation to reach

final energy when mechanical coordi
nates are held at xi = XI' xi = xi,
x~ =
while the electrical linkages >'1
are brought to their final values, then the
mechanical coordinates are brought to
their final values.

x;.

T IZ = T 2b

:1

T 13

r 3 l>

T Z3 = Tn

f)J,

The meaning of Eq. 1-14b

Wm =

"I'''2'''3

0,0.0

2:

i=1

i;(A;, A;, Ai;

Xl> , X m )

dA;

I.'

(IEI-2)

is that with the xj held constant each flux linkage is brought to its final
value holding all other flux linkages fixed. The order in which the flux
linkages are brought to their final values is immaterial because the energy
is a state function. For the purposes of illustration assume that the flux

'"

--~

16

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

ELECTROMECHANICAL ENERGY CONVERSION

linkages are brought to their final values in the sequence AI> .1 2 , .13 ; then
Eq. IEI-2 can be written out as
Wm =

Jlr

A1 0 0
'

1;(,\;,0,0) dA;

0
A
l'A2'

0,0.0

APAz.A l

A1.AZ.0

Wm

I~(AJ> .1;.0) dA;

Al'O,O

ii(AI> Az, A;) dA;

(IEI-3)

TIlA; dA;

+ foAl (T3 \A 1 +

f:

(T21 A1

T 32 A2

+ T Z2 A;) dA;

Tn'\;) dA;

(11-4)

Recognizing that unprimed variables in the above integrations are held

fixed, there results
W m = !TllA~

+ r 21 A1A2 + !f'22A~+

1'31 A1 A3

+ T 32 A2 A3 + tr33Ai

L L

/~I j~

This expression establishes the dependence of Aon i and X which must be

maintained during the arbitrary displacement defined in obtaining
Eq. 1-16. This single-valued dependence results from the stored magnetic
energy
(1-17b)
W m= ~f/",(il> ... , in; XI> . X",)

!TijA/A]
I

dWm --

which is the conventional expression for energy storage in a linear

magnetic field device.

" via the stored

The interchange between electrical and mechanical energy

"

how).., and i; vary during the arbitrary displacement. It will be shown,

however, that th~/ force (.f..}k is independent 0/ the mriatiollS (If ,\, lind i;
tlurinK the arhitl'w')' displacel1lent pl'Ol'ided the changcs in '\; al/l/ i, ji,f!oll'
the /unctional relationships (the internal constraints) oj' thc I.I'\ICIII hcc
Fig. 1-3b and Eg. 1-7). In other words, the fact that only one of the two
variables A; and ii can be treated as indepcndent must be recogni1SJ as
an internal system constraint.
The arbitrary displacement is taken at the point defined (see Fig, 1-3b)
by
(1-l7a)
Ai = >';(11 ... ill; xI> ... , X",)

being a state function. The second term on the right of Eq. 1-16, dWm,
is a total differential (see Eq. 1- 11) and becomes

which can be written as

W m, =

energy in the magnetic field is a direct manifestation of the energy con

version process. The fact that the stored energy can be determined for
any configuration of the system, and that this stored energy is a state
function defined solely by the functional relationships between variables"
and by the final values of these variables, provides a powerful tool for
determining the coupling forces of electromechanics.
Now that the stored magnetic energy has been determined, the arbitrary
displacement and conservation of energy as expressed by Eg. 1-11 can be
used to evaluate the mechanical forces on the system of Fig. 1-3.
Assuming that the stored electric field energy W~ is zero, it follows from
Eq. 1-12 that

8Wmd
Xk
vXk

2:

II dA, - dWm

(1-16)

where dA; = v/ dt. To obtain Eq. 1-12 and hence Eq. 1-16, an arbitrary
displacement dx, of the kth mechanical node was assumed to take place.
At the samc time no explicit restrictions were placed on the changes in
Aj and I ; consequently, it may appear that the force (J;)k will depend on

~: 8W"'d'

L.. ~
i=1
vii

(1- I8)

I;

In obtaining this expression all the i's and x's are considered as inde
pendent. For instance, OWm!Oii means the derivative of Will with respect
to i j with all other currents and all displacements x j constant. Next,
evaluating dA; in Eq. 1-16 by using Eq. 1-l7a gives

'
~ 0\ d
d,,/....
L.. -0 x]
]=1

Xj

~ oA; d'
+ ,-1
L.. ~ I,
Vi,

(I -19)

For the arbitrary displacement under consideration it was assumed that

dXj

for j t:- k

Thus Eq. 1-19 becomes

, ' A>.; d
d"; = -,,----- X k
vX k

/=1

-<l-

C('.). dX k =

17

~
+ ,-1
L..

8>'; d'

~
Vi,

(1-20)

I,

Once again the partial derivatives are taken with all remaining i's and
x's held fixed. Now Eq. 1-20 can be used to evaluate the tlrst term on the
right ofEq. 1-16):
~"

_ Ii
i=l

;"

dAI -

n_

2:

;~l

I;

oA;,

0'. dXk

u'\k

+ Ln ./i
;=1

(~

0>'; .)

_ ~
"., 1 ul,

dl,

(I -21)

EQUA"I,JNS OF MOTION OF ELECTROMECHANICAL SYSTEMS

ELECTROMECHANICAL ENERGY CONVERSION

18

It is possible to put Eq. 1-27 in a simpler form by writing it in terms

of the magnetic coenergy
which is defi ned as:

Substituting Eqs. 1-18 and 1-21 into Eq. 1-16 yields

~, . all; d

- - 0Wmd
(/.)
, k dX" - - ' ) - - x"

+ 1=1 I;

L.

(;""

uX"

19

W:n

, l'l.. . ,/

x"

Wm =

~.(~fJA;d)
~fJWmd'
+ .L.
I, L. Y
I,
- . L . y- II
:1
r=l I,
Iz~1
I,

0 .. ,,0

n..

.~

.-1

,.,

0'

of

( 1-28)

A;(lp ... , I,,; xl> ... , Xm ) dl,

(1-22)

1 __

~:

Rearranging terms and interchanging indices in the last term change

Eq. 1-22 to the form

(Ie)" dx" =

di

f-

~j

- oW
aA') dx" +2:" (oW
"aA)
,2" i; ~
~ + 2: i, ~ di;
( ax" m + I-I
uX"
1=1
uI;
,=1
ull

-,,-.+ r=L l u
I, -:;;:u1i
I;

lill.. .

,-1

in n

0... ,0

~. oA,
-.,.+ r-I
L. 1,-:;;:01;
uti

- () Wm

~ . oA,

= -

L.

r= I

I,~ uti

(/.) = - oWm(i1> ... , in;

II;

OA,
+ r=L~.
l,~
1
uli

( 26)

i
;-1

I
I
I

I
,-,.
(," I' '''i'
,x +dxk ... .xm )
'\
, n' k

I
I
I

I
t

i ,~

ii

Fig. 1-4. Illustrating how Ai and ii can change during an arbitrary dis
placement dXk.

The relation between energy and coenergy has already been established
as a consequence of writing W m as in Eq. 1-25 thus:
n

XI> , X m)

Wm =

i;A; - W,:,

( 1-29)

;=1

oA;(;1> ... , in;

xl> .. ,

X m)

(1-27)

aX"

This is a perfectly general expression for (I,h which holds regardless of

how A; and i; are changing with time in the system.

If:..o..L

;"_~,1i\jt..~.f

ax"
i;

Change dunng
arbitrary displacement

Consequently, the force (Ie)" is always given by:

which is always satisfied.

"

";

(1-25)

Substitution of Eq. 1-25 into 1-24 and evaluation of the indicated

derivatives (keeping in mind the functional relationships of Eq. 1-17a)
yield

dX;

Change dUring
arbitrary displacement

(1-24)

A;U;, ... i~; xl>' .. , x m ) di;

K
i

It can be seen graphically how this condition makes (Ie)" independent of

the changes in i j and II; by referring to Fig. 1-4. For any di; there is a
corresponding dll; (two examples are shown in Fig. 1-4); thus, if (I,)" is
independent of di j , it is also independent of dA;.
To show that Eq. 1-24 always holds, the definition of stored magnetic
energy given by Eq. 1-15 will be used and integrated by parts
(f idA = iA - J A di) to obtain the alternative definition:
W m = ~ i,A, -

~
'i_

I
"1

Ai (i;, ,i~iXl',,,,XRt,,,xm)

In order that the force (f,)" be independent of the change in i; and AI

during the arbitrary displacement, the coefficient of di l in Eq. 1-23 must
be zero.
~. OA, _

di

dX,

( 1-23)

- oWm

r 'T

This relation is illustrated graphically for the ith CircUit In Fig. 1-5.
Substitu';')n of Eq. 1-29 into Eq. 1-27 and subsequent simplification lead to

(I.)"

aw'('
m I"

.
, In;

XI> . ,

aXk

xm)

( 1-30)

ELECTROMECHANICAL ENERGY CONVERSION

20

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

Equations 1-27 and 1-30 give the force U.)k when the displacements and
currents are used as the independent variables. If it is desired to express
the energy with the flux linkages ('\) and displacements (x) as independent
variables, Eqs. 1-27 and 1-30 must be modifted. For this new functional
dependence the stored magnetic energy is expressed as

With the stored magnetic energy given by Eq. 1-15, the last two terms of
Eq. 1-33 subtract to zero, giving for the electromechanical coupling force
U,oh applied to the kth mechanical node:

(n

- ()W m (,\\

and
xl> .. ,

An;

xm)

XI> . . .

(1-34)

(1-31)

W m = Wm(>'lo .. , , A,,; xI> ... , x m)

= ij(AI> ... , A,,;

OXk

, , I.

ij

21

(1-32)

x m)

This force can also bc evaluated in terms of the coenergy When

Eq. 1-29 is substituted into Eq. 1-34. remembering that the'" and ,Ij arc
the independent variables. there results:

U.h

An;

aw,;,p'l> .

Xl> . , , X",)

OXk

_ )- A, ai;(Al> . , . An;
;::;'1

xm)

Xl> ..

(1-35)

aXk

'

AI

Equation 1-35 is an equivalent way of expressing Eq. 1-34.

The several forms of the electromechanical coupling force ef..)k applied
to the kth mechanical node by a magnetic coupling field as found by an
TABLE I-I.

Mechanical Force Caused by Magnetic Coupling Field

Wm

Stored magnetic energy

t,,A. .-1i i;

d)..;

(1-15)

~ A; di;

(1-28)

0"".0

W~

Magnetic coenergy

o
.,,';.

0...0

Relation between energy and coenergy

W..

W~ =

I-I

L i,A ,

(1-29)

'-I

c
Fig. 1-5.

Since \

and

Conservation of energy during arbi

trary displacement dXt (Iossless)

-,

Ii

Graphical relation between magnetic energy and coenergy.

Xj

Independent
Variables

are the independent variables the differentiations of

Eq. 1-16 yield:

- aWm(AJ, ... , An;
U;')k dXk =
OXk

Xl> .. , x m )

~ aWm(A" ... , An;

i-I

Xl>""

+ L

;=,

i;(AI> ... , An;

Xl> , X m )

dA/

-oW..
([.)t -_
iJXk

([.) = -oWm

OXk

dW..

(f.)t dXt

(1-16)

Force Evaluated
from Coenergy

Force Evaluated
from Stored Energy

i. OA

1/

I-I

OXt

([.) =

iJW~
OXk

(f.)k = 0 W,~ _

OXk

I AI OXk
.oi
l

I_I

dA/

OA;

x,

Flux linkages A,
Coordinates x J

Xk

Xm)

Currents i,
Coordinates

L i, dA,
1-1

(1-33)

arbitrary displacement of the kth mechanical node XI. are summarized in

Table 1-1. The four expressions for U.)k given in Table I-I are equivalent
and will yield identically the same force, which is the true force, for a given

22

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

ELECTROMECHANICAL ENERGY CONVERSION

Another interpretation of this type can bc made by considering a system

excited by electrical constant-current sources and by inquiring about the
energy conversillll. From Tahle I-I the force is evaluatn! from the
coenergy lIsing i, and x, as independent coordinates; the force is mathe
matically given by the ratc of change of magnetic coenergy with the
currents held constant. In this case the mathematical restrictions in the
general case coincide with the electrical terminal constraints in the special
case. This leads to an interpretation of the coenergy W;n as a measure
of the converti bility of electrical energy from constant-current sources.
When a system is electrically linear some general statelT1l:nts can be
made a hnut energy conversion and about the relation between energy and
coenergy. By electrically linear it is meant that the flux linkages are
linear functions of the currents, thus

state of the system, i.e., for a given set of ii' Ai> and Xj' In order to find
the dynamic path of the system the force of Table (-I must be used with
d"Aklllbl:rt's principle and Kirchhoff's laws to establish the cquations of
dynamic equilibrium.
The results of Table I-I are complctely general and independent of
electrical source variations (assuming low electrical frequcncies and low
mechanical velocities such that a quasi-static solution is valid). It is
worthwhile to examine some of the results more closely.
For instance, the force obtained from the coenergy with i l and x) as
independent coordinates was given by Eq. 1-30
(f.)k = oW~(il> ... , in; Xl> ... , x m)
oX k

(1-30)

It has already been shown that this force is independent of the changes in
A; and ii which take place during the arbitrary displacement; consequently,
this expression is valid regardless of how>"; and ii vary, if the variation is
compatible with the internal constraints, and therefore it is a general
expression for the force. On the other hand, considering Eq. 1-30 from
a mathematical point of view, since il and XI are independent variables
the partial derivative is taken with respect to Xk, holding all other x's and
all i's constant. The holding of the i's constant is a mathematical restriction
imposed by the selection of independent coordinates and has nothing to do
with electrical terminal constraints. The mathematical restrictions are
often misinterpreted as electrical terminal constraints, and some confusion
about the generality of the force expressions results.
Statements similar to those just given about Eq. 1-30 can be made
about all the force equations in Table 1-1. These force equations are
general; the mathematical restrictions placed on the derivatives by the
choice of independent coordinates have nothing to do with electrical
terminal constraints in general.
On the other hand, the general expressions of Table 1-1 can be used to
interpret specialized electrical terminal constraints. For instance, if all
changes in flux linkages d>"; are constrained to zero, there can be no energy
flow between electrical sources and magnetic fields; conseq uently, energy
conversion must take place solely between the magnetic field and the
mechanical system. This is illustrated by noting that the electromechanical
coupling force, when evaluated from stored magnetic energy with>"; and
x) as the independent variables, is simply the negative rate of change of
stored magnetic energy with respect to mechanical displacement with the
flux linkages held constant. In this case the elcctrical tcrminal constraints
of the special case coincide with the mathematical restrictions of the
general case.

23

A; = L

(1-36)

[i,i;

r= I

where
[I,
"

=:

[1,(Xh .. X n ,)

( 1-37)

= I"

is a general single-valued function of the displacements. The use of

Eq. 1-36 in the definition of W m (Eq. 1-15) yields for the stored magnetic
energy
WIN =

;l .... IO

0 ..... 0

n (n

L i; r=1
L Ii' di;
;=1

) z:n

i-=d

LJ -tli,i;i,

(1-38)

r=,-"

From Eq. (I -29) the magnetic coenergy is

n

W~ =

i=1

i;A i

WIN

(1-39)

Substitution of Eqs. 1-36 and 1-38 into Eq. 1-39 yields the result:
fI

W;" = W m =

L L 1!,J,i,

( 1-40)

;=1 r=1

Thus in the electrically linear case the stored magnetic energy is equal to
the magnetic coenergy. This can be seen geometrically by considering
Fig. 1-5 with a linear relation between A; and i;.
The fact that the energy and coenergy are equal in the electrically
linear case has led to the use of the two state functions interchangeably.
Investigation of Table 1-1 shows that energy and coenergy must be
distinguished; otherwise in the electrically linear case the sign of the
mechanical force will be in error if the wrong state function is used.

"

.,
24

ELECTROMECHANICAL ENERGY CONVERSION

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

Equation 1-16 can be used to write the conservation of energy for an

arbitrary displacem'~nt c/.>;J.. as:

2:" ,

i; d>.; = dWm

'--.,--'

' __ ....... - . J

cl,,'rtrli.:.a1

Morcd
liclt.1
energ)'

Iliput
l'lll"rM)'

+ U"h dX k

(1-41)

~
ouqHlt
Cliel

the conservation of ener.gy the stored electrical energy W" must t:qu,il the
input energy from all sources,

W, = input electrical energy

W.(ql> ... , q,;

Ill",,,; 1\;ll11Gl I

xm)

XI' . . . ,

Next, the individual terms on the right-hand side of Eq. 1-41 can be
cV.lluated, assuming an electrically linear system. Using the force from
Table I-I and the coenergy with i j and x j as independent variables, the
energy converted from electrical to mechanical form in an electrically
linear system is

input mechanical energy

l "' . , ' '

=2:
n..... o ,.

~y

" " 1 of.

U,)k dXk = ;~I '~I 2" o_~: iii, dX k

.25

'!'\'m,~

n... :0

"

I';(q;, _. _,

(1-.44)

'I;; x;, ... , X;,,) !/l/

I.

..:.I Ij(q" - . - , q"

,I.'

XI' . . . \",) d\j

(1-45)

(1-42)

"

...

;;:.

f.".i"l

The change in stored field energy is found from Eq. 1-40 as:
dWm

_ ~.

,.

...

"

..

~ 1;,l j dl,
,

+ ~.
...

;--I ,

~~ ~ 01;, ..
~ ').
Ijl, dX k

(1-43)

I - (IXk

.,

According to Eq. 1-41 the sum of Eqs. 1-42 and 1-43 equals the electrical
input power. When all the electrical sources are constrained to be
constant-current sources,
di, = 0
and the electrical energy converted to mechanical form becomes equal to
the change in stored magnetic energy. Thus, when an electrically linear
system is excited by constant-current sources, the electrical input energy
is divided equally between stored field energy and converted energy.
1.2.3

Fig. l-'a.

Mechanical Force Due to Electric Field Coupling-

The mechanical forces produced by magnetic coupling fields in an

electromechanical system have been determined. A similar development
can be made for finding mechanical forces due to electric field coupling
in an electromechanical system. Consider the case where the only
sign iticant stored energy is electric flcld energy (sec Fig. l-6a). The
electrical stored energy W, can be found in a manner like that used to
find the magnetic stored energy in the previous section. Assume that all
purely electrical or mechanical energy storage elements plus all dissipative
elements have been removed from the system (see Sec. 1.2.1); then, from
For an alternate treatment

SL"C

Fitzgerald and Kingsley, [(/('. cit.

C41ed charge-earrying conductors.

The expression for stored elee.trical energy as given by Eq. 1-45 is plotted
in Fig. l-6b and c. The amount of the total stored electrical energy
supplied by electrical sources and that supplied by mechanical sources
arc dependent upon the manner in which the system is ,assemhled; how
ever, since the electrical stored energy is a statc function, YV. can be
expressed in the simple form:
W,(ql> ... ql;

Xl> . ,

xm)

"\ . . .
1

'11

n.... n

2: v;(q;, .. " q;; x", ..

j

XIII)

tlq;

(1-46)

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

ELECTROMECHANICAL ENERGY CONVERSION

26

II;
qj

. .

Stored electncal energy

jq, 1/; dq;

In this expression it is assumed that the system is assemblcd Illcch'lnically

before bringing the electrical system to its final state (i.e., the x/s are
constants in the integration). Equation 1-46 will be used in suhse
quent consiucrations of mechanical forces caused by electric coupling
fields.
For the electric field case, just as it was for the magnetic field case, it is
the interchange of energy among electrical and mechanical sources and
the stored electrical energy that is a manifestation of cnergy cOllver,ioll.
This, and the fact that the stored electrical energy is a state function which
is determined solely by the functional relationships between variables in
the system and by the final values of these variables, allows the use of the
stored -' ~ctrical <:nergy to find the mechanical forces. Therefore, an
arbitrary displacement plus conservation of energy as expressed by Eq. I-II
can be used to obtain the electromechanical coupling force Cr..). when
the total energy in the coupling field is electrical (We) and there is no
magnetic field storage. The resulting equation for the force of electrical
origin applied to the kth mechanical node is, from Eq. 1-12.

q'

= (}

f,\,<:<:",c::::<:,<::\(,<:::

l;II,,\.,C\\<~~'C'C\Y77'7777777:.P"'"1

,\\<,c;>'"1

v,

vi

t'l

27

v;

q2

Electrical coenergy.
u

fo

'

q~dlJ2

(f.)/, dXk =

2:

Vj

( 1-47)

dqj - dWe

i=1

t'-z

q~ +-

Slored electncal energy =

<1,.1("_,<:",\\\.,,,,::zs;:a==

lln

vi

l'~dq~

V2

where dqj = ii dt, and Cr..), is the force applied to the kth mechanical node
by electric field coupling when an arbitrary displacement dx/, compatible
with the internal constraints is made with all other mechanical coordinates
held fixed. Equation 1-47 is general in that it gives the correct force
regardless of how Vi and qi vary during the displacement prnviding the
variation is compatible with the internal constraints. The force (f.,)" can
be expressed in several simple forms in a manner exactly analogous to
the magnetic field case.
First, recognize that there can be only two independent variables in the
system. The characteristics of the system will impose a functional
restriction or internal constraint on the variables,

lit

q~

Electncal coenergy =:
u

1o

t--.' :rr.-- q~ (vi .... U~;%lo ... ,X m )

q'dv'

"~q~(Vi . ... U~;%l' ,x';:)

Un

l)~

Fig. l-6b. Path of operation to reach

lin;d stored electrical energy when
mechanical coordinates are held fixed
at x~

Xl;

Xl =

x 2'

x~ = x mt

while the electrical variables are simui

taneously brought to their final values.

Un

Fig. l-6c. Path of operation to reach

final energy when mechanical coordi
nates are held at x; = xi:: xi = xi:
x;" = x~, while the electric charges are
simultaneously brought to their final
values, then the mechanical co
ordinates brought to their final
values with qj held fixed.

Un

. :t.

qj = q;(v" ... ,

VI;

x" ... , x",)

(1-4~ )

Since Eq. 1-47 describes an arbitrary displacement for which Eq. 1-48
holds, the arbitrary displacement must be made with Eq. 1-48 as an
internal constraint. In writing the functional relationships of Eq. 1-48
the voltages I', and the displacements x j have been established as the
independent variahles; consequently, the stored electrical energy can be
expressed as;
We = We(VIo ... ,

VI; Xl> . , . ,

x m)

( 1-49)

If

:w

I:LEC IKOMECIIANICAL ENERGY CONVERSION

Now Eqs. 1-4l\ and \-49 lIsed with Eq. \-47 give the force
- i'IV
U~). d.\. = ( -----;-,-,-"
e\.

'
~

I~I

P,

?q,)
, (- i)W
~ tis. + ~ - i : !'+
(','.

Vi

j-I

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

U;.h

(1q)
tlv,
2 /', -i:!--~
'-1
Vj

'

as:

29

The use of Eqs. I-56 and I-57 with Eq. 1-47 yields the result

(I-50)

= _ oWr(q"

(f.)

.. , q,: Xl, ' ,X",)

,.

OXk

( I-58)

It can be shown quite readily that

- oW,
!.;.:. . V, 0:;oq,
-.-,+ ,-1
(,Vi
vV;

which is the correct force regardless of how Vi and qj vary during the
arbitrary displacement.
Substituting from Eq. ]-54 for W, into Eq. I-58 and recognizing that
the x j and lfi are the independent variables yield the force in terms of
the electrical coenergy as

(I-51).

Therefore the forcc (f,). is independent of how Vi and q, arc externally

constrained (providing Eg. 1-48 is satisfied) during the arbitrary dis
placement oecause the coef1icient of tlv, is zero in Eg. \-50. The
resulting expression for the force is:

Cf.h =

()W~(ql< . , .. q,: Xl' .. X m )

(lx.

(f..) = -OW,,(PI""; 1',: x, ... , x",)

k
cJx.

+ ')~
':-1

V
I

oll'(I'.
I' . X
X )
I
"',1'-1,",,"
OX.

(1-52)

W~ =

0 .... ,0

2 q;dv;
j.=,

Ol';(q,. ' ..

TABLE 1-2.

(I-53)

Stored electrical energy

(1-54)

L vjql

'-I

Substitution of Eq. 1-54 into Eq. I-52, making certain to keep

as the independent variables, leads to
(/')1(

= oW~(VI< ... , v,; XI< , , x",)

Vj

r.t ',t .
";,:~ ,

;~

(l-55)

Equations I-52 and I-55 give the force (f,). when the displacements xJ
and the voltages /'i are independent variables.
Expressions can be obtained for the force (fe)' that are equivalent to
Eqs. I-52 and I-55 but expressed in terms of the displacements X j and the
charges qj as independent variables. Assume that the voltages are
expressible as:
(I-56)
VI = V;(ql' ... ,q,; XI> .. x",)
Then the stored electrical energy can be written as:

W, = W~(qh"" q,;

Xh ,

x m)

Electrical coenergy

W'

Relation between energy and coenergy

W,

Conservation of energy during arbi

trary displacement dx_ (loss less)

2:

and xJ

'I" .. '"

.. V,

'-I

W; =

VI

"d'

L q, v,

0.....0

_'J'

(1-46)
(1-53)

'-1
I

L v,q,

'-I

dq, = dW,

+ ([.)_ dx_

(I-54)
( \-47)

'\I~

Force Evaluated from

Stored Energy

Independent
Variables

v,

Voltages
Coordinates

XI

Charges q,
Coordinates

XI

,(f.h

-oW'
+ ') v oq,

= --'

ax_

-oW
(1.)_ = --'
ax_

I~

ox_

Force Evaluated
from Coenergy

(1.)_ =

oW:
oX
1

aw', - 'ou,
L q, k = eXt
I_I I ox_

(f.)

~;

(I-57)

(I-59)

1-1

v' .....r l

.;1';

cXk

~ 'd'
q,

W, =

0.....0

W~ =

XI< .. X m )

"
ex.

Mechanical Force Caused by Electric Field Coupling

The relation between the electrical energy and coenergy is (see Fig. 1-6b)

W, +

,q,:

The various forms of the force U,)' which have been found for electric
field coupling are summarized in Table 1-2. The four expressions for

It is possible to put Eq. I-52 in a simpler form by writing it in terms of

the electrical coenergy W;.. which is defmcd as:
ul ..... v , '

.;.. qi

i-.I

of

force (f,h are equivalent and will yield exactly the same force, which is
the true force, for a given state of the system, i.e .. for a given set of qj,
Ilj. and Xj'
In order to find the dynamic behavior of a system the force
of Table 1-2 must be used with d'Alembert's principle and Kirchhoff's
laws to establish the equations of dynamic equilibrium.

38

The force expressions of Table 1-2 ~re;i completely general and the
mathematical restrictions imposed by the differentiations indicated must
not be confused with external electrical constraints imposed on the system.
Equation I-58, which yields the force for any general case, shows that in
the special case of a terminal constraint of constant charge, the energy is
converted between stored electrical energy and mechanical energy with
no input from the electrical source. The general expression for force of
Eq. I-55 also shows that when a special external electrical constraint of a
constant-voltage source is applied, the energy converted from electrical
to mechanical form is equal to the change in electrical coenergy. The
electrical coepergy can be considered as a measure of the convertibility of
energy from constant-voltage sources through electric field coupling, In
addition, it can be shown that when an electrically linear electric field
system is excited by constant-voltage sources, the electrical input energy
divides equally, half going to electric field storage and the other half to
mechanical energy. These results are clearly analogous to those derived
in the previous section for magnetic field coupling. They also indicate
that for electric field coupling a constant-voltage constraint and a con
stant charge constraint are analogous respectively to a constant-current
constraint and a constant-flux linkage constraint for magnetic field
coupling.

1.3

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

ELECTROMECHANICAL ENERGY CONVERSION

Hamilton's Principle-

usually considered to be the most fundamental is Hamilton's principle.

Hamilton's principle can be derived for mechanical systcms from
d' Alembert's principle and the principle of virtual work. However.
Hamilton's principle proves to be significant for other than just mechanical
systems, and as such overshadows d' Alembert's principle and is at least
a more general relationship if not a more fundamental one.
The usual statement of Hamilton's principle is that the variation of the
time integral of the Lagrangian L between fixed end points qi(tt) and
<1, (I)

:,,::'.

'Il '~,i
':1 .&

Path which makes OJ = 0

qi(I,)

/' Ii

"f~

I
;'1'r;~:

""

. to'

.; .:,~
,.,~

.)

'."~'
.

q(t )~--

."

',.\
J"~

"1' ":~t

:,l

;~' / /

,'?-,

. '"

;,~:

For a derivation of Hamilton's principle from the principle of virtual work, see
Goldstein, loc. elf. or Whittaker, loc. cit.

, (:'

In the previous sections the equations of motion of electromechanical

systems were obtained from force laws, These force relations were
determined from basic experiments or postulates of physics, and they
form a set of "differential principles," i.e., principles concerned with
incremental changes in the system.
It is also possible to develop an alternate approach to the problem of
describing the path of a dynamic system by postulating that the dynamic
path of the system is determined by finding the extremum of certain
integral functions. This alternate approach is called a variational method
and is based on a set of "integral principles," i.e., principles relating to
gross motion of the system.
When a system is completely described by one of the principles, the
other principle can be derived. This is just another way of saying that
one physical system is described by one dynamic path of operation regard
less of how the equations of motion are derived. The integral principle

jJ

,'.~tl':'

,~~~~.
:
. f'

e:'f"

;;~
'}1: ~~~:,

, ::t.

:~~.t:

.\I ,..

",\",\
'.:'' eff ;

;"lV,

:~.r:'
',1:/' '

, ~I; ,{.

.~\~

, -if"

':.t~\

';;!~'

,) ~ll '
,;~.

t't
,.i~

::~~
;~~r
;,~~~, '

;~t

'I

/
/
'/
/

'/

.
!
o

q,(I,)

Fig. 1-7.

I
I

I
I
I

1--,---------:

qi(t,)

'1,(t)

Illustration of Hamilton's principle.

Q,.(t2) must be zero. In order to state Hamilton's principle symbolically,

define a function I such that*

I =

l
2

L(qI. ... , qN;

q\t ' .. , qN;

t) dt

(1-60)

II

One way to state Hamilton's principle is "thc actual dynamic path of a

system described by the state function L is determined by f1l1ding an
extremumt of thc function I." That is, as thc system movcs bctwecn the
points t[ and t2. the truc path described by thc N coordinates ({,(') is such
that the function I is an extremum (usually a minimum). A graphical
The variables 'II and qi are the generalized coordinates and generalized velocities

of the system. They will be discussed in detail in Sec. 1.4.1.

t An extremum is analogous to a minimum. a maximum, or a point of inflection as

used in ordinary differential calculus. For most cases the extremum is (\ minimum.

.,

'''").~

32

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

ELECTROMECHANICAL ENERGY CONVERSION

make the integral I an extremum" and which will satisfy the end conditions
that
(1-63)
and q(12) - q2
q(ll) - q\

representation of the meaning of an extremum is shown in two dimensions

in Fig. 1-7.

,>,

The condition for an extremum of the function 1 is that the variation of

the function I equal zero,
M

S f.1 2L(ql(/), ... , qN(/); (Ml), " . ,QN(I); I) dt .. 0

subject to the end conditions that.

and oqj(l2)

=a

~l

for i = 1, 2, 3, ... , N

In the above equation the symbol 8 means the time-independent variation

as used in the calculus of variations and is analogous to a differential in
ordinary ditTen.:ntial calculus. (Sec Sec. 1.3.2.)
Equation 1-61 is Hamilton's principle. It is now necessary to see what
set of relationships must exist among the N coordinates, the qj, and their
N derivatives, the q" in order to satisfy Eq. 1-61. The reduction of
Hamilton's principle to a set of differential equations is best done using :f
the calculus of variations. Therefore, a short digression into the calculus
of variations is a worthwhile next step.
,}
1.3.1

where q\ and q2 are fixed end points.

As a first step in finding the extremum of I assume that qo(t) is the
actual function which makes I an extremum, and choose any continuously
differentiable function 7)(t) which vanishes at the end points q(t\) = ql
and q(lz) = qz. Then for any constant a, the function

(1-61)

'I

oqj(t\) = 0

q(t) = qo(l)

II

/:

~ \,

The calculus of variations is concerned chiefly with the determination of

maxima and minima (more exactly extrema) of expressions involving
unknown functions. It differs from differential calculus in that the
variables are known in ordinary calculus and a minimum or maximum of a .
function of these known variables is desired. In the calculus of variations,'1
the variables are unknown and it is desired that the relationships among {
the variables be found which will form an extremum (e.g., a maximum.
or minimum) of some integral containing these variables.
I
The essential techniques of the calculus of variations can be determined
by procedures analogous to those of finding maxima and minima of j
differential calculus. As an example, consider the function I defined as~
'";1
the integral between 11 and t 2 of another function L, where L is a function1i
of the unknown variables q(t) and !J(t) and of the independent variable t:$

' L(q(t); ,jet); 1) dt

J
2

"

I(a)

==

J".

L(qo(t)

(1-64)

7)(1)

The integral

+ a 1)(t); 40(1)

IX

>j(t); t) dl

( 1-65)

11

is obtained from Eq. 1-62 by setting

.',

"
~~

'1\'

q(t) = qo(t)

7)(t)

(1-64)

4(1)

+ IX 7j(/)

(1-66)

and
=

40(1)

IX

The function l(a), Eq. 1-65, is only a function of lX, once qo(t) and 1](t) are
assigned, and furthermore, the function I(a) is an extremum when a = a
because qo(l) was chosen to make it so. However, this is only possible if
.1

dljda

. J"
\ (If>:

.rJ'

'I;f.!,,

;':
. '".
~,

:}"

~.

=0

when ex = 0

(1-67)

Equation 1-67 is, therefore, the condition for an extremum and is a

defining relation which relates differential calculus to the calculus of
variations.
The condition for an extremum, Eq. 1-67, can now be applied to Eq.
1-6'5 to find the differential equation which will result. Since a, in
Eq. 1-65, is a constant in the integration. the differentiation wi.th respect
to a can be taken under the integral sign to yield

it

dl(a)
-=
da

II2
'I

[OL
()
8L,( )] d
0

-1]1+---;;-7)11=
oq

(I -68)

oq

Before investigating Eq. I-68 further, it is desirable to introduce the

variational notatilln of the calculus of variations. The only variation to
be studied here is the time-independent variation.

(1-62)

'I

A typical problem is to find the function q(t) and also (j(l) which will
h,r " Illor.: delaile,1 treatmcnt of Ihe I:alculus of v;lriations see, e.g., F. B.
Hildebrand, Methods oj Applied Mathematics, Prentice-Hall. New York, 1954, Chltp. 2.

+a

will satisfy the end conditions of Eq. 1-63.

Calculus of Variations

1=

33

To tix idcas <lSSUI11C that finding an extremum of 1 is mcrely finding tile poil1.s of
zero slope of I. and will rhus be determined by sClring some dilfercn\i:d of I eQUlI1
10 zero.

.. ..
~

34

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

ELECTR.OMECHANICAL ENERGY CONVERSION

Now return to the condition for an extremum, Eq. I-M!, and note that
from the definition of oL, Eq. 1-74,

1.3.2 Time-Independent Variations (0 Variations)

'I

Consider the case in which times

and 12 of the end points are held
fixed and in which the integrand (L in Eq. 1-65) is an explicit function of
the variables q, q, and'. Definc a variation of the function L in terms of
variations of q and q and not of the time. This is a time-independent

I'

~,~

Similarly, the variation oq of the velocity q from the true velocity

oq = q(t) - 40(1)

Sf = S

40 is

81 =

where q(l, a) and 4(/, <x) are given by Eqs. 1-64 and 1-66 respectively.
Expanding the first term in Eq. 1-71 in a Taylor series about qo(t) ancfaJ
40(1) and keeping only the lowest order terms yield
SL

[q(t, a) - go(t)]

~+

[q(t, <x) - 40(1)]

The use of Eqs. 1-64 and 1-66 with 1-72 gives

SL =

8L
;v,

"'1

a'1(t)

1(

,,"'(":'

8L.

+ F"q a'1(t)

:~

. ""

(1-72)

.~;

i~

1.3.3

9J
;

(t-73) .

'!~~

or the use of Eqs. 169 and 1-70 to introduce variational notation yields

8L

8L

= Oq

8L.
Sq + oq Sq

Note the similarity of Eq. 1-74 to a total differential.

defines the time-independent variation of L.

(1-76)

J/2 SL dt

= 0

(1-77)

'1

J' (8La
2

Sq

f:JL.)
+ 8"'
" 8q
q

(1-78)

dt = 0

The Euler-Lagrange Equation

T;ic relationship between the calculus of variations and differential

calculus has been established. To understand Hamilton's principle the
conditions imposed upon the function L by setting 'M = 0 must now be
investigated. To do this, define a function I as in Eq. 1-62, but now
consider L to be a function of 2N variables, N coordinates q,(/), and N
velocities q;(I).

1=

0-74)
Equation 1-74

Several forms of the condition for an extremum expressed in variational

notation are given by Eqs. 1-76, 1-77, and 1-78. An extremum of the
function f in the calculus of variations is found by setting the variation of
a function equal to zero (i.e., Sf = 0). This is analogous to finding a
maximum or minimum of a function F in ordinary differential calculus
by setting the differential of the function F equal to zero (i.e., dF = 0).

(1-71)

- L(qo(t); qo(t); t)

J'1

'I

is

= L(q(t, a); q(t, a); t)

( 1-75)

Expanding SL by Eq. 1-74, the condition for an extremum in variational

notation can be expressed as

The next step is to define the variation of a function, e.g., the variation
of the function L(q(/); .:j(/); I). To do this find the difference which results
from a small variation in q(/) from the true function qo(/). This difference
oL

oLdt

'2L dt

""!~,

(1-70)

= a 1j(t)

8L dt

'1

Since the vanatlon is time-independent, it can be taken outside the

integral sign to yield

.~:

(1-69)

= a 7)(t)

q(t) - qo(t)

J"

'1

true function qo(t) and obtain:

Since u. is a constant in the integration with respect to I, the expression of

Eq. 1-75 mu~t equal zero if Eq. 1-68 equals zero. Thus, the condition
for an extremum is

Now define the variation oq as the variation of the function q(/) from the

8q

a dl(a)

cia

(8) variation.
The function qo(t) has been defined as the true function which produces
an extremum, and any other function with the same end points was
defined by Eq. 1-64 as
(1-64)
q(t) = qo(t) + a 7)(/)

3S

i."
~
r.{'l

~;

f"

11

L(Ql(I), ... ,qN(/); ql(/), ... , qN(t); I) dt

(1-79)
~.,

The condition for an extremum of f between fixed end points at 11 and

t2 is that 81 = 0 subject to the end constraints 8q;(II) = 0 and oq;('2) = 0

"~f:;'~

F);'.'
r;

.j

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

ELECTROMECHANICAL ENERGY CONVERSION

36

for i = I, 2, 3, ... ,N. Since the variation 0 has been defined to be

independent of the time f, it can be taken under the integral sign as in
Eq. 1-77 to yield

oj = J' SL(qt(f), ... ,qN(/); eMf), ., ., tiNCt);

2

(1-80)

I) dl

.,1'\

'I

Expanding the

oL in Eq.
IlJ =
<>

1-80 gives

J/2!1:,-I (8L
aq/
.L.

8')
lJj 1 0
q,

8L
+Y

Il
<>qi

II

Cf

( 1-81)

if

01

(1-82)

oq

0lJ;(ll) = 0

~ 8L
5' .:- ','
2

111<1NI;

0'.
q, d t

= ~

L..

aL
c.

j-l(}lJ/

I) q/ \/

'I

I'2 j~l....~ [!!-dt (8L)]

8q, d
.,.
C'lJ;

oj =

( 1-83)

'I

Eq. 1-83 into Eq. /-81 gives

'2

'I

.,

8qj

dl

8q;

= 0

for k

I, 2, 3, ... , N

t!.. [OL - ~ (()~)] 8q; ell = 0

()q/

ell

h",

(1-84)

Euler-Lagrange equation of motion:

8L _ !!-dl (8L)
och

(1-84)

From Eq. 1-84 the conditions that L must satisfy for I to be an

extremum become evident. First, the extremum &1 = 0 must hold
regardless of the variation 8eh of any coordinate qk' Furthermore, if the
N coordinates qj are independent, i.e., if there are no equations of constraint existing among any of the N coordinates qj, the only way Eq. 1-84
can be satisfied is to set the term within the brackets in Eq. 1-84 equal to
zero for every value of i. Thus the condition for an extremum :.

8L _!!-dt (8L)
= 0
OCjk

f.'

which for all q; independent gives

it"

LN [8L
- - -d (8L)}
----;- oqj dl
;= 1

(1-61 )

for i = I, 2, 3, ... , N

and SlJj(t2) = 0

't j:-:"I

Since 'ilqj vanishes at 11 and 12, the first term vanishes and substitution of

oj =

Hamilto,n's principle reduced to ditrerentiul equution form:

and the second term of Eq. 1-81 can be integrated by parts to obtain
~,

'I

where

oq = ~

called Lagrange's equation) yields the set of dilferential equations which

defines the actual path of a dynamic system. It must be remembered.
however, that the Euler-Lagrange equation results only if all the
coordinates (the N qj) are independent. Should the coordinates not be
independent, Eq. 1-84 cannot be reduced to Eq. 1-85. Techniques for
treating systems with coordinates that are not independent will be discussed
later.
Hamilton's principle has been stated in Sec. 1.3. The Euler-Lagrange
equation which results from Hamilton's principle h:ls been dCh:rlllined.
In summary, these relationships are:
Hamilton's principle:
'2
01 =
L(ql> ... ,qN; ql> ... , itN; I) (/1

The

To simplify this expression, integrate the second term by parts.

o variation is independent of time; consequently,

37

(lqk

t.'j.

(1-85)

. ellJk

Equation 1-85 is the famous Euler-Lagrange equation and it yields the

set of differential equations which defll1es the conditions for an extremum
of the function I. This equation will prove to be invaluable in finding
the equations of motion of dynamic systems; because, as a consequence
of Hamilton's principle, Eq. 1-61, the Euler-Lagrange equation (often

'12

= 0

for k

1, 2, 3, ... , N

( 1-85)

These results, and in particular the Euler-Lagrange equation, determine

the dynamic path of conservative systems once the Lagrangian function L
is established for that system. Equation 1-85 proves particularly valuable
for the study of electromechanical systems since it yields the complete
equations of motion including the electromechanical coupling terms for a
conservative system. Thus the Euler-Lagrange equation plus the
Lagrangian state function are all that is needed to determine completely
the equations of motion of any conservative electromechanical system
which can be described by a set of independent coordinates. The use of
d'Alembert's principle and KirchhotT's laws becomes unnecessary except
where they may be needed to account for nonconservative forces and
dissipation, and for a large number of nonconservative systems it is
possible to modify the Euler-Lagrange equation to account for nonconservative forces. This will be' investigated in later sections after a
detailed examination of the Lagrangian function.

t~_~

38

IA

ELECTROMECHANICAL ENERGY CONYERSION

S 39
EQUAT IONS OF MOTIO N OF ELECTROMECHANICAL SYSTEM
l system
For exampl e, it means that a knowlc dge of onc type of physica
of an
system
l
physica
a
may prove helpful in gaining an insight into
work
whose
r,
enginee
an
that
entirely ditTerent nature. It also means
tely
comple
in
fecI
not
nced
zation,
takes him outside his field of speciali
strange territor y.
ns
Unfortu nately, the class of systems exactly describ able by state functio
be
must
tion
Dissipa
.
systems
al
echanic
electrom
all
docs not include
s. A
exclude d from systems if they are to be describ ed by state function
is.
hysteres
is
form of dissipat ion which proves particul arly trouble some
of
use
main
the
At first, this appears to be a severe limitati on; howeve r,

State Functi ons

was
In the previou s sections the Euler-L agrange equatio n of motion
gian
Lagran
develop ed from Hamilt on's principl e; and, in so doing, the
state
state function was introdu ced. The Lagran gian and the other
l
physica
of
n
erizatio
charact
function s are of central importa nce in the
thermo
l,
chemica
al,
echanic
systems (electrical, mechan ical, electrom
system
dynami c, etc.). The state function s include the total energy of the
These
ian.
Lagrang
the
as
such
s
function
and other closely associa ted
time,
of
instant
given
a
at
,
because
s
function
functio ns are called state
and
time,
of
instant
that
at
system
the
of
state
the
they depend solely on
long
a
for
zed
recogni
been
has
nce
importa
Their
not on past history.
ical
time in thermo dynami cs and in the statistic al and quantu m mechan
in
was
treatme nt of atomic systems , althoug h their first notewo rthy use
s
variable
advanc ed classical dynami cs. These state function s, and the
ly
explicit
t
describ ing them, are in many cases used by enginee rs withou
the
realizin g it. For exampl e, to describe a thermo dynami c system,
the
;
entropy
and
ture
heating enginee r will use such variable s as tempera
his
e
describ
to
ment
control system enginee r will use force and displace
about
talk
will
r
enginee
tical
connec ted mechan ical system; the aeronau
; the
the roll torque and roll angle in discussing the stability of an aircraft
electric
ing
describ
for
charge
and
electric al enginee r will use voltage
terms as
circuit behavio r; and the chemica l enginee r will employ such
in
works
rs
enginee
these
of
Each
.
number
chemical potentia l and mole
l
physica
t
differen
widely
these
about
talking
zation,
his field of speciali
these
y,
systems in terms which seem equally unrelate d. In actualit
physica l systems have much in commo n.
e by
For exampl e, in each of the above cases denote the first variabl
the
of
nature
the
of
tive
irrespec
11 and the second variable by qj; then,
write
to
system, it is always possible

dW =/; dql

(1-86)

I
I

i.'

+
v'
I

.,;" ~,,".
,~"

Lossy
electrical
system

Vi

(,
-+--

Loss less
electromechanical
system

x,

-:

..'."

'..

I: "\

,,..

Fig, 1-8.

Lossy electromechanical system divided into simpler compone nt parts.

l' ,i

ns of
state functio ns will be to obtain a general formula tion of the equatio
obtain
to
difficult
most
are
which
ies
quantit
The
motion of a system.
e.g., the
are the couplin g terms between differen t types of systems ,
tely,
electrom echanic al couplin g terms in eJectrom cchanic s. Fortuna
and
system
these terms arc determi ned by the conserv ative part of the
e
separat
to
l
are derivab le from state function s. Thus, it become s practica
part
ion
convers
energy
the problem into two parts, consisti ng of (I) an
For an
that is dissipat ionless and (2) other parts with dissipat ion.
system
al
electric
lossy
a
of
form
the
takes
electrom echanic al system, this
and a
system,
al
echanic
electrom
lossless
a
losses),
(includ ing hysteresis
way
this
in
possible
is
It
1-8.
Fig.
in
shown
as
system
ical
lossy mechan
ion
dissipat
to study the lossless electrom echanic al system and bring in
ance.
when conside ring the over-all system perform

,;}I;',

1.4.. 1 The Characterization of Physical Systems (Witho ut Hysteresis

-,;1

~.~

...

l..,~
,'[.r.,.

>'loR;'

.,

r'
11
.

,f;L

ed by a
where dW represe nts a differential change in energy produc
q, are
differen tial change dqj in the variable q;. The variable s f; and
in
relation
general ized vatiable s and their produc t describes an energy
usually
and
be,
each of the above systems. The energy functio,ns may
the
are, state function s and contain much valuabl e informa tion about
the
of
tation
manifes
a
is
1-86
system describe d by them. Actuall y, Eq.
have
systems
l
physica
all
natures
t
fact that in spite of their vastly differen
atical
a fundam ental similari ty and lend themselves to a commo n mathem
r.
enginee
the
for
ences
consequ
hing
far-reac
has
descrip tion. This fact

-+ ,
i.

-'--

t'~l' .

..

},'. ~,'

,:,::,-;

'.'

and Dissipa tion)

can
The constitu tion of a physica l system from a dynami c point of view
interto
subject
s,
particle
of
be regarde d as consisti ng of a number
ration
connec tion and constra ints of one kind or another . The configu
called
es
quantiti
of
terms
in
d
specifie
be
can
of a given system at any time
ates for
the coordin ates of the system. The choice of the set of coordin
ual
individ
each
general
in
but
y,
arbitrar
hat
somew
usually
a system is
For
ates.
energy storage elemen t of the system can have a set of coordin

'I

40

ELECTROMECHANICAL ENERGY CONVERSION

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

example, every discrete element of mass can have its position specified in
terms of three space coordinates, each inductance element can have its
nux linkage specified, or each capacitor can have its total charge specified.,
Examrk's of possible coordinates for several systems are shown in Fig.
1-9. When dealing with static systems (systems in static equilibrium) the
values of the coordinates completely specify the system. For a dynamic
system, however, the coordinates do not completely specify the system and.
an additional set of dynamic variables equal in number to the coordinates
must be used. These dynamic variables can be the first derivatives of
the coordinates, the velocities, or they can be a second set of variables,
e.g., the momenta. The velocities and the momenta are associated
variables and either set can be chosen as the dynamic variables.
So far, only the number of variables that can be ascribed to a particular
system have been discussed; however, in any given system all of these
variables may not be independent and hence they cannot all be specified
independently. The question of how many variables ;ire independent
is determined by the constraints of the system. The problem of handling
the constraints is one of the most diflicult single questions of dynamics.
Constraints are of two essential types-holonomic and nonholonomi~
constraints. The holonomic constraints are represented by sets of
relations among the coordinates or, if expressed as differentials, they can
be integrated to yield these relations. For example, if 11 coordinates can
be ascribed to a system and then 111 equations of the form

q I Xl

JiJ

q l %2

..

t) = 0

j = I, ... , m

Adiabatic
walls

T e

s
Gas

= displacement 01 mass

q, '" II = charge on C
q2
dt. !Otegral .Of

-J;'

q, .. S .. entropy

= displacement of sprlOg

qz

({II

Curren!

= V =volume

q,.

(1-87)

fit dt

q. ji2 dt

:Il

voltage on C
(e)
,

C2

integral of current in L 1

=integral of current in L 2

OR
q\ '" A, tOl.1 flux linking L \

can be written, it is possible to reduce the number of coordinates from n to

(n - 111) by using these rn constraint equations to eliminate m variables.
Holonomic constraints are always expressible in the form of Eq. 1-87;
furthermore. for a system which has only holonomic constraints it is
always possible to select a set of independent coordinates which does not
contain the constraint equations. Thus, if n is the number of coordinates
determined from all energy storage elements and m is the number of
holonomic constraints, then there are (n - m) independent cr~rdinates
and (n - Ill) velocities, or a total of 2(n - m) variables which can be
used to describe uniquely the dynamic motion of the system. The
minimum value of (n - m) that can be found is also the number of degrees
of freedom of a holonomic system. When a system is described, using
a selected set of coordinates which eliminates the various system constraints, it is accepted practice to caIl these coordinates the generalized
coordinah:s of the system. For a system of N = (n - m) degrees of
freedom there will always be 2N generalized variables needed to describe
the dynamic path of the system (i.e., N coordinates and N velocities).

q2 - .\.2 total flux linking L

q, ~

=. Integral of

:1'1

Q, Chafge on (.',

I.

c'OD~
q2:= Q2:=: Charge on

In

OR
'1 X I!ux I1nkmg L

(b)

q2 Z.,JF;dt

qJ

.~(q\> ... ,q,,:

41

q 1 X -= dIsplacement
q2" Q:o Charge on C

q 3 f;, dt Integral of current in L

fe, dl -_,ntegral of voltage .cross C\

(t)

q4 - fi2 dt integral of voltage across C

2
(d)
,.~

I'r.:

~! V

.~"l

~~

q 1 - :c 1 displacement of

q2 '" "'2 ~ displacement of spri",

Q3'" A" flux linking L

q .. j'. dl .. integral volt.ge on C

Iron
(f)

Fig. 1-9. Examples of physical systems and variables ql associated with each energy
storage, sources and dissipation omitted for simplicity.
~I"'

"

S 43
EQUATIONS OF MOTIO N OF ELECTROMECHANICAL SYSTEM
the q;(t)
require d and can be expressed as ql(/), qz(/). ... , qN(I) where
c
dynami
calkd
are
ed
describ
so
s
System
fes.
coordina
are the genl'l"u/i::cd
be
can
systems. For dynami c systems a second set of N quantit ies
/b,d
introdu ced such as the PI(/). P2(t), ... , PN(I). called the genera
is
lillie
of
inSlanl
moment a. The slale of a dynami c systcm at a Ricen
and
ates
coordin
ized
general
N
determi ned by the particu lar values or the
or a
the N generalized momen ta at that instant or time. Thus the statc
l
ensiona
2N-dim
a
in
point
a
as
dynami c system may be represe nted

ELECTROMECHANICAL ENERGY CONVERSION

42

there
The questio ns of independence of coordin ates which arise when
A
resolve.
to
ex.ist nonhol onomic constra ints are much morc dil1icult
not
do
int
constra
of
ns
nonhol onomic constra int is one in which equatio
1-87.
ex.ist among the coordin ates, i.e., constra ints not satisfying Eq.
gcncral
the
when
d
obtaine
is
int
constra
For example, a nonhol onomic
form of thc constra int equatio n is
(l-88)
i((/I> ... , 4n; t) = 0
exampl e
where the resulting differcntia1 cquatio n is not intcgrab1e. An
es
machin
of such a nonhol onomic constra int is the commu tator in electric
.
surface
rough
y
perfcctl
or the constra int imposc d on a rolling ball by a
of
type
ity
inequal
the
is
Anothe r type of nonhol onomic constra int
a gas
constra int such as that impose d by the wall of a contain er upon
int is
constra
onomic
nonhol
the
case
particle contain ed therein. In this
of the form
(1-89)
(ql)2 - b2 < 0

Generalized Coordin ates for Holonomic Systems

c.
There are two types or states of physical systems: static and dynami
rium
equilib
For the static state, only the description of the system in static
tely f
with its environ ment is given. The state of a static system is comple
in
system
a
specified by the values of its N generalizcd coordin ates. For
static
ing
discuss
static equilibr ium there can be no dissipat ion. In
al
systems it should be pointed out that the term steady state used in electric
term
The
t.
systems and the term static used here are quite differen
and
stcady state is used to denote a particu lar state of a dynami c system
ium.
does not denote a system that is in static equilibr
the
For the dynami c state of a physical system its configu ration (e.g.,
is
time
of
n
functio
a
as
point)
e
distanc e of all particles from a referenc

..1,
"

.;(""
"\

",
"

~.',

"X

... %

116

\,'>"i

J
~i.l 1

:.

1.4.2

,r
" I,

At'

ate of
where ql is the coordin ate of the gas particle and b is the coordin
the wall of the contain er.
find
When dealing with nonho10nomic constra ints it is not possible to
the
to
number
in
equal
ates
coordin
a set of generalized indepen dent
a
choose
to
ry
necessa
is
it
,
Instead
.
number of degrees of freedom
plus
freedom
of
degrecs
of
number
the
to
equal
number of coordin ates
type the
the number of nonho10nomic constra ints. For problem s of this
le
princip
on's
Hamilt
from
1.3.3
Sec.
in
derived
n
equatio
Euler-L agrange
,
general
In
cannot be used since the coordin ates are not indepen dent.
handle
to
difficult
any problem with nonhol onomic constra ints is very
mic
unless some trick can be devised to reduce it to an cquival ent holono
1.6.
Sec.
in
latcr
d
problem . A mcthod for doing this will be discusse

Fig. 1-10. Path in phase space for system of Fig. 1-9a

when p = 0 and x = ."0 at I = O.

"

ta
space, the 2N dimens ions being the N coordin ates ql and the N momen
c
ative
a conserv
PI' Thi ~pace is called phase space. Once the state of
is comple tely
space
phase
in
path
its
time
one
at
hed
establis
system is
is estabdetermi ned. This means that oncc a given set of ql(ll) and 1'1(11)
that
fact
lished, then ql(l) and p,(t) are uniquel y determi ned. It is this
s
analyse
the
in
gives state variable s and state functio ns their wide utility
the
of
'state
of dynami c systems. As the q, and PI change with time. the
ory. the
system traces out a path in phase space. If the system is oscillat
for the
space
phase
path will close on itself. Figure 1-10 shows a path in
e
distanc
a
d
displace
mechan ical system in Fig. 1-90, when it was initially
freely.
e
oscillat
Xo and allowed to

.,

.......,.,.....""',

..
EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS
44

ELECTROMECHANICAL ENERGY CONVERSION

In a dynamic system, the variables qb ... , qN' and PI' .... PN are
rct"erred to as the ... tate variabks. A pair qi and Pi is called canonically
conju~ate variables. The choice of the qt ... ,qN' and hence the
PI .. , .. PN' is not uniquc. That is to say, there llre alternative sets of
variables If; . ... , ((" and P;, ... , P'tv which can be uscd to specify the
state of the system. and means of transforming from one set of variables
to another can always be found. Much of advanced dynamics is concerned with transformation theory which has as its aim the expression
of the state of the system in the most useful way possible.

104.3

the study of electrical systems since Kirchhoff's equations follow from the
use of the Lagrangian in the Euler-Lagrange equation. The rl:lalcd
state functions may be obtained from each other by means of special
transformations called Legendre transformations to be disl;lIsscd in
Sec. 1.4.6.

1.404 The Number of Independent Generalized Coordinates of a

Connected, Electromechanical, Holonomic System
The problem of selecting the proper set of independent variables to
describe the dynamic behavior of any system always presents snme
dilliculty. This topic for electric circuits is treated in Chaps. I and 2 qf
Guillemin's Introductory Circuit Theory (lac. cit.). The discussion given
there for electric circuits is equally applicable to connected elcctro-

Associated Variables

Associated with every set of independcnt variables q, and Pi is a set of

dependent variablesj, and (j,. These two sets of variables

q" '12 .. , 'IN; p" P2 ... PN

and

4S

j1J2, ... IN; iII' (iz, ... ,cIN

(t~-l) ,

T1Ji

characterize the state of the system. The variablesji and (ji are functions
of the qi and Pi' so that once the 'I, and Pi are specified for a given state.
the.!; and (;, arc also detcrmincd. The variables iii and Pi are associated
variables; so are.l: and CJi' In describing the state of the system, either
0//1' of the pair of associated variables can be considered as heing the
independent variable. For example, the 2N independent variables which
specify the state of the system may be taken as the qi nnd (;, instead of
the conjugate varinbles Ili and Pi' The /; and Pi are then expressed as

functions of the qi and iIi'

The q; and Pi will be referred to as "extensive" variables. such as
volume and entropy which depend on the magnitude of the system, since
they are nlike in charncter to the extensive variables of thermodynamics;
the j, and iI, will be called" intensive" variables, such as pressure and
temperature which do not depend on the magnitude of the system, since
th,ey are analogous to the intensive variables of thermodynamics. The
extensive variables 'Ii and P, are associated respectively with the :.ltensive
variables!; and (; i' Il' the state of the system is to be given in terms of the
Pi and 11" the Hamiltonian H. which for conservative systems equals the
total energy. is the important state function. For most systems in
electromagnetics it is also the total energy of the system. If some other
combination of the state variables is to be designated as the 2N independent
state variables. the role of the total energy will be taken over by some
other state function. For example. if the 2N independent state variables
arc the q.. . , . , i/N nnd the iIt, ... , iIN' the state function is L-the
Lagrangian. This will prove to be the state function of most interest in

..
<",,,

, ),~J

l~

Fig. I-II.

Graph or connected system.

mechanical systems with holonomic constraints. Therefore, a detailed

treatment of the subject will not be given here but only a brief mention
will be made of those aspects which are most relevant to the present
considerations.
..
First, a connected system is represented by a graph, as illustrated in
Fig. I-II. The heavy dots represent the nodes, n, which are 7 in number.
The straight lines represe~t the branches, b. which are 12 in number.
Associated with each branch in ap electrical system is a voltage eb across
it and a current ib through it. Associated with each branch in a mechanical system is a velocity i b across it and a mechanical forcefb through it.
The branch currents (or forces) and branch voltages (or velocities) are
considered the primitive vffriab{es of the system. These are the basic
"unknown variables" oft~e system. Since the number of primitive
variables is twice the number of branches, the number of equations
relating these variables must be equal to twice the number of branches.
In the example, Fig. I-II, there must be 24 equations. In the absence of

..

.1,

....
46

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

ELECTROMECHANICAL ENERGY CONVERSION

largely a matter of preference. The fact to be remembered is that for

a dynamic system either 21 or 2(n - I) variables must be chosen, depending upon the choice of a loop or a node representation. It is also possible
to choose a mixed set of variables (i.e., both loop and node variables).
in which case the number of variables will lie between 21 and 2(11 - I).

mutual inductance, half these equations-i.e., the number of equations

equal to the number of branches-are of the form
Electrical System
eb = LeCib )
(1-90)

or
ib

= L;(eb)

Mechanical System
Xb = LAlb)
or
Ib = LJCXb)

(1-91 )

where Lc(ih)' L;(eb)' etc., arc differential operators which may be nonlinear.
With mutual inductance, a branch voltagc, for example, will dcpend on
other branch currents as well as on its own. However, the content of
the equations is essentially unchanged.
There still remains the number of equations equal to the number of
branches to be determined. For an electric circuit these are the (n - 1)
Kirchhoff current (node) equations and the I Kirchhoff voltage (loop)
equations. For a connected mechanical system these are the (n - 1)
Newton equations and the I continuity-of-space equations. The numbers
(n - I) and I must satisfy the relation
(1-92)
(n - 1) + 1= b

104.5 The Lagrangian State Function

.r
p

"'lIt'

"~!I.~f :
"

~~:~.

'J ~t~

f1l

where b is the number of branches.

In the example chosen n - I = 6, I = 6, b = 12.
The variables normally used to describe the system are not the primitive
variables given above. Instead, there is usually a set of independent
variables of lesser number than 2b. For example, choose:

"t

1. A set of loop currents (forces), which automatically satisfy the

Then I independent equations remain. In this
case the (n - 1) node equations are the equations of constraint which are
used to reduce the number of variables to an independent set.
, 2. A set of node-pair voltages (velocities), which automatically satisfy
the loop equations. Then (n - I) independent equations remain. In
this case the I loop equations are the equations of constraint which are
used to reduce the number of variables to an independent set. r

(n _ 1-) node equations.

One usually selects (at least in electric circuits) the set of variables
which gives the least number of equations to solve. If the example of
Fig. 1-11 is an electric circuit, it is possible to choose the loop currents
and identify the Iloop currents as generalized velocities qi or as generalized
momenta Pi' Conversely, the (n - I) node voltages could be selected
and the node voltages identified as generalized velocities qi or as
gegeralized forces I" Other possibilities exist, in fact, the numbers I and
(n _ I) only set the limits on the number of independent variables; the
characterization of specific variables as generalized qi, qi' Pi' or Ii becomes

47

':

't!":~, .J.11'
,~,

?",;,'

. ,j'
.1

""

if;

Once a choice of independent coordinates has been made or a set of

coordinates plus constraints determined, the equations of motion of the
system can be obtained. If a variational principle is to be used to find
the equations of motion, then the Lagrangian state function L(q, 4, t)
will generally be chosen to characterize the system. and the Lagrangian
will be used in the Euler-Lagrange equation to yield the equations of
motion. It is possible to use other state functions, e.g.. the Hamiltonian
R(p, q. t) plus Hamilton's equations of motion, to derive the system
equations. However, for electromechanics the Lagrangian. which is a
function of the q's and q's, is the state function usually preferred. There
are several reasons for this, but for electrical engineers one very compelling reason is that the Lagrangian state function will yield equations
of motion for the electrical portion of electromechanical networks
identical to those obtained from Kirchhoff's laws. The resulting identity
of form of the eq uations of motion enables one to use all the tools which
the electrical engineer has developed for electrical networks.
The Lagrangian state function is a function of the system coordinates
qt, the system velocities qi' and the time t. The Lagrangian was first
derived for use in elassical mechanics and was defined as the difference
between the kinetic energy T and the potential energy V of the system,
i.e., L = T - V. This definition of the Lagrangian is not sufficiently
general to allow it to be extended directly to nonlinear systems; thus a
more general definition of the Lagrangian must be given. The real
difficulty with the definition of the Lagrangian in terms of the kinetic
energy T and the potential energy V revolves around the definition of
independence of coordinates. For example, the Lagrangian L(q. (j. I) is
an explic:' function of q and q and not of the associated variables! and p.
Now for a linear system where f = Kq and p = Mq and the coefficients
K and M are constants, an interchange of! and Kq or p and Jvhj docs not
affect the original assumption of which coordinates were chosen as the
independent ones. It should always be clear. of course, that only two
of the four variables q, q, f, and p can be chosen independently. For
electromechanical systems the coefficients K and M of even linear systems
may be functions of the system variables and for nonlinear systems even

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

ELECTROMECHANICAL ENERGY CONVERSION

48

No restrictions of linearity were used in the derivation of Eq. 1-95 and

this definition of potential energy holds for nonlinear systems. Thl:rd'ore
a generalized force Ji. is defined as:

more complex interrelationships may exist. In these cases changing

from one set of variables to another is not possible without a change in
the definition of which of the variables are independent. This occurs
because the partial derivatives defined for one set of independent variables
are not independent of the variables chosen.
By definition from Hamilton'S principle the Lagrangian L(q, el, t) is a
function of the coordinate q. the velocity q, and the time t. It is possible
to decompose L into two functions which will satisfy the definition of
L = T - V for linear mechanics and which will also be valid for nonlinear
cases. If the Lagrangian is L(q\> ... q",; til> ... , tiN; I), the differential

1'(
)_ iJL(ql'.qN;O, ... ,O;t)
Jk qt. ... , qN, t "

uqk

... , 4N; t) = k~ldlfk

~ ?L dqk + k~lvlfk
~ ~~ d(ik +

~~
dt
vi

0 ..... 0

ql

qN;IiI ....IiN;'
qN;O .....O;1

(1-93)

.l;):h""rr

~ oL(qi, ., . , qN; 0, ... ,0; i) d '

L.
() ,
qk
qk
til .....IiN !!: oL(qJ, . .. , qN; q~, .... 4~; t) rI"

o.r

L.
k-l

n~

q"

ql .....qN

0..... 0

t[. 8L(q;, ... , q:.v; 0, ... ,0;

L.
k-I

i) d '
qk

Ifk

is a function of the coordinates ql> ... , qN and t and is completely

independent of the velocities of the system. For the linear case this is,
exactly the definition for the negative of the potential energy of a system.,

( 1-97)

8L(q\> ... , qN; q;, ... , (i~; t) d"

qk
f

dql.

.':':",

J~ "(x, i') di'

If the current i is chosen as a generalized velocity q, it follows that the
second term in Eq. 1-95 has the form of a coenergy, e.g. a kinetic
coenergy. Consequently, defining a generalized momentum Pk as
I'.

'(

". )_oL(q\o ... qN;q;, ... 'tlN;t)

__ ,

(1-95)

Therefore the particular choice of a path of integration for determining

L has separated the Lagrangian into two functions.
The first of these functions expressed as
~,

L.
1<,,,1

the kinetic coenergy T' becomes

k-I

0 .....0

... ,q':v;t)dq~

PI< q\o ... , qN, ql' ... , qN' t -

iql ..... qN
0.... 0

-f~(ql'

is a function of the final values of the coordinates q~:

qN and of the
velocities q], ... , ifN' For the linear case this is exactly the form of the
kinetic energy in mechanics or the stored magnetic energy in a set of
inductors if i = q. However, for the nonlinear case the magnetic
coenergy is defined by Eq. 1-28 as

Substituting from Eq. 1-93 L becomes

L(% ... , qN; ell> .. - , q",; i)

ti' ..... 4N ~

0 ... 0

.r<iI..... qN;O ... O;t dL(q~ ... ,q~;4~, ,4:.v; t) (l-94)

,)o .. o;o .. o;t

1<-1

The second term on the right-hand side of Eq. 1-95 can be identifi~d
next. The term

of integration:
fl

q! ..... qN N

V=

Since L is a state function, a path of integration can be chosen for determining the function L which holds all the eli constant for integration with
respect to the q; and which holds all the q; constant for integration with
respect to the iti' Furthermore. these integrations can be performed for
a specific value of time I. Such paths of integratio'1 are perfectly valid
because the Lagrangian L is determined uniquely by the final values of
the variables and not by the path of integration. Thus the correct
Lagrangian will be obtained. Consider L defined by the following paths

L =

( 1-96)

and the potential energy V is defined as

of Lis

dL(q\> ... ,qN; ell>

49

T =

iti1' ....4N N
0 ... 0

2:

8'r

qk

(1-98)

f'
w,.,.,

Pk(ql> ... , qN; ql' ... , qN; t) dqk

(1-99)

k-t

The meaning of the kinetic coenergy will be investigated for several

specific cases later, and it will be shown that for linear cases T' = T.
With the two terms on the right-hand side of Eq. 1-95 defined. the
Lagrangian becomes simply

L(qlo ... , qN;

410 ... , 4N; t)

= T'(q}, ... , qN;

<110 ... qN; t) -

or

V(ql> . " , qN; t)

L = T' - V

= kinetic coenergy -

potential energy

(1- J()O)

50

ELECTROMECHANICAL ENERGY CONVERSION

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

where T' is defined by Eqs. 1-98 and 1-99 and V is defined by Eqs. 1-96
and j-97.
This ddinition for the Lagrangian will prove to be particularly valuable
because it can be used for nonlinear problems, for problems of electromechanics. and fIJr prohlems of other fields; and it reduces to the more
restricted definition tlr the Lagrangian (L = T - V) that is used in
classical mechanics for linear ':ptell1s.
An expression has been derived for the Lagrangian in terms of two
energy functions, the potential energy and the kinetic coenergy. To help
clarify the general equations for these two energy functions, consider
the following example of a simple mechanical system.

and the displacements a and b are the equilibrium positions.

potential energy is

j'1,Xl _ [( _ K,x;

The

+ K1x;) dX; + (- K1x; + K1x;) dX;)

- K1x;

0.11

(I E2-1)

The above line integral can be evaluated by holding x; = 0 and displacing

from 0 to XI. then holding x; = XI and displacing x; from 0 to Xz.
This results in the integrals

x;
:,'"

",-1

'I;:

X1 ' 0

(K I

0,0

Kz)x; dX;

I'YI,Xl K1(x; -

Xl) dxi

x\.O

+ K2)X~ + -tK2~

= '!(K1

EXAMPLE lE2

K 2x 1X2

(I E2-2)

"'";:

1 "

A mechanical system of mass points interconnected by springs and

constrained to move in a line is shown in Fig. lE2-1. The potential

Of course, any other path of integration could have been chosen to

determine potential energy since it is a state function and is dependent
only upon the state of the system. The path chosen can therefore be
picked to give a simple integration as in Eq. 1E2-2.
The kinetic coenergy for the system of Fig. lE2-1 can now be derived
using Eq. 1-99, which is

,I
"i"

""""-""""\.'~~':-.."

Jt " \,
a
and bare
equilibrium
pOSItions

(1

'Ill

;",
~K

"
.....

T' =

','

,"
-),:1:"

. ,~~~",

BeY.2

"
~;"~

Fig. IE2-1. System of masses and springs.

0.....0

2: -

')

flql' ... ,qN;

To find the

d '

ql<

1<-1

f;(q;, ... , q~; t) = f;(x;, x2) = - K1x;

fi(q;, .. , q~; t)

Of

~'

", '.

':~'1

(I -99)

(IE2-3)

,.ol)

1X!,0

.,.,

MIX. dX I

/
I2
+ jX .. Mzx".,
dX

= 1M Ixi + -tMzi'~

X,.O

l?\

(1E2-4)

Equation IE2-4 gives the kinetic coenergy, and it is a state function in

which the velocities XI and Xz are the independent variables.

,'I,

1 "'.i~r,.'.or;:,.. ,

"

,~

(1-97)
.~

fl.X2[(MIX;) dx; + (Mzxi) dx;J

0.0

~ ". I

energy and kinetic coenergy of this system are to be found.

potential energy V Eqs. 1-96 and 1-97 are used, thus

V =

The line integral for this simple case is

:',

M2

"

2:

1<-1

0.0

;;,,;');

ql' ... ,qN N

0' .
.,
Pk(ql> ... , qN. ql' . , . , qN' t) dql<

41 ' . 4 2 N

For the system of Fig. lE2-l the moment;! MIXl and MZXl are independent of the coordinates XI and Xz; consequently, Eg. 1-99 becomes
.","!~

- --

,1

0..... 0

%1

1-

%2

%t

where

51

Kz(x; - x;)

fi(x;, xi) = - Kz(xi - xl)

'jJ

.~~:
ViP
,

:'i

~'i'

For linear mechanical systems the coenergies are not usually defined
because they are equal to the associated potential and kinetic energy
functions. It is interesting to note, however, that it is common practice
to determine the kinetic energy by integration with respect to the generalized velocities q, rather than the generalized momenta pj' Actually, such
an integration yields the kinetic coenergy rather than the kinetic energy.
However, for the linear case, the use of the N momenta PI as independent

'/

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

ELECTROMECHANICAL ENERGY CONVERSION

52

variables in finding the kinetic energy yields the same result as finding the
coenergy by using the N velocities qi as the independent variables.

Taking the total differential of H yields

dH = !0
L

I-I

1.4.6

Legendre Transformations and Other State Functions

;"

The Lagrangian can be used to formulate the equations of motion of

dynamic systems. However, for many dynamic problems, an alternate
energy function is more useful in formulating the equations of motion.
The Hamiltonian or total energy can be obtained from the Lagrangian
by a transformation of variables. The variable qi in the Lagrangian L
can be replaced by the variable Pi to yield the Hamiltonian H which is a A
function of qi and Pi- Similarly, it may be desirable to derive other state
functions by changing variables, e.g., changing from potential energy
(a function of qj) to potential coenergy (a function of j;), etc. The
Legendre transformation is valuabh: for such ends.
The Legendre transformation states that if a state Junction depends on
a particular variable (as ""'1.'11 as other variables), then a new state Junction,
depending on the associated l ariahle of this particular mriable (with the
other /'l1riables unchanged), can be obtained by subtracting .I"om the original
state junction the product of the two associated variables. The Legendre
transformation expressed in symbolic form is as follows:
Let the function
Ct;
F(x ... XN; Yh ... YN)
J

be a state function,

X,

XN:

.~),

' ,

P; -

F -

2:

(1-101)

y,.\'1

To show that the Legendre transformation will produce a new state

function in terms of a new variable, consider the Lagrangian function L

L(q" ... ,qN;

Ill> ... , qN)

(1-102)

Now define a new state function in terms of the variables qj and Pi rather
than q, and qi where the exact nature of the new variable Pi is to be
determined. Calling this function the Hamiltonian fI and using a
Legcndn: transformation to deflne 1I * gives

oL
= 0
uqi

(I-I03b)

(1-104)

-;;'i7'

Using Eq. 1-98 to define Pi establishes that dB in Eq. I-I03b is of the fonn
i

dH

"1

'"Y

N (
8L dql + qi dpi )
=.2
- -a
1.1
q/
.

(1-1050)

Thus dH is a total differential and is clearly a function of q; and Pi which

is the relationship demanded. In addition to proving that H, derived
by a Legendre transformation, is only a function of ql and P;, it also
establishes the auxiliary relationships that

8H
oqi = -

.~

8L 8 H .
oq;; 0Pi = q;

<:: I "N

for I

(I-105b, 1-105c)

Equations I-! 05h and c are particularly interesting and are Hamilton's
equations of motion. To show this, one needs only to refer to Eqs. 1-96
and 1-98, where it was shown that

at =. /';
oqi

.
f rom potentia
. 1s
torce

( I-96)

~~
(iif/

momentum from kinetic potentials

(1-98)

~.l(

'.'{

'It:\t'

.~ ,:t

= PI

The Hamiltonian, like the Lagrangian L, is a state function, and further

using Eqs. 1-96, 1-98, and 1-105b and c, a set of equations of motion
can be written using the Hamiltonian. In this case the equations of
motion using the Hamiltonian consist of a set of 2N first-order differential
equations. This is in contradistinction to the Lagrangian L which gives
a set of N second-order differential equations (see Eq. 1-85),
In addition to the Hamiltonian H it is also possible, using the Legendre
transformation, to define two additional state functions L' and H' as

L'(pl> ... , PN;f1o ... ,fN)

- H = - H(ql' ... , qN: Plo ... ,PN) = L - ~ Nil

;=1

The negative sign of H is used

sO

+ 2: qJ/
;=..-01

(1-106a)

H'(i/I, ... , C,N:/;' .. , ,fN) = L -

2: qJ;

(I-I06b)

;" I

Hamiltonian.

+ P; d'ql + ql. d!PI )

~,

XN)

j,.",l

8L. d'q;
-oL dq; - oq;
8q,

For H to be a total differential, where by definition H is a function of PI

and q/ only, the coefficient of the differential dql must be zero; i.e.,

.",.

and .\oj is obtained by the Legendre transformation

G

where

(-

A second state function

G(x\ .. , ,

having the variables

~t

;1

~l

53

(1-103a)

as to conform with the classical definition of the

These functions, Eqs. 1-IOod find ;h. have not becn used in the classical
literature on elcctromcchanics and are not named as arc the Lagrangian
L and Hamiltonian H. They could, however, bc used in place of L or

'...
';

-"-

!,#

ELECTROMECHANICAL ENERGY CONVERSION

'54

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

H in estahlishing the dynamic equations of motion of a given system.

The meaning of the energy function H' of Eq. 1-106b is particularly
interesting when it is examined using the concept of coenergy. The
Hamiltonian function H is readily shown to be the total energy

H= T+ V

Gihbs frCl' energy

( 1-109c)

( !I09d)

The interrelations between these various functions are of the general form
A(T, V)

(l-I07b.)*

This fact leads naturally to calling H', Eq. 1- I066, the co-Hamiltonian
and the function L', Eq. I-I06a, the co-Lagrangian, where

'.:b,,.~, T.- V'

ql = V

the volume
}
PI = not ddefifinedd
extensive variables
q2 = not e ne
P2 = S the entropy
11 = P the pressure " }"
. = not defined
...
qj
d fi d
mtenslve varIables
f 2 = not e ne
42 = T the teu;perature"

,.:~. ;~,

:.'t~

'i"

.' :"~i,. '

~,

,,;1

,;

j~"
~Li
.,

(1-108b).

ij

The fact that PI' Q2, tll, and /2 do not exist in thermodynamics is merely a .ft!,.
manifestation of the fact that thermodynamics is in reality thermostatics, ./,
. a fact that is recognized in the current literature. In terms of these l.
variables, the various well-known state functions of thermostatics take
the form
Total energy = U(S, V) (analogous to the Hamiltonian)
(I-I09a)

= Jqdp;

T'

= Jpdq;

=-

JJdq; V'

= -

(I-I lOb)

G(T, P)

(I-IIOc)

lI(S, P) - TS

The generalized coordinates and state functions have been defined in

the previous sections. It is now necessary to establish which variables of
both the electrical and mechanical systems can be chosen as the generalized
variables q, q, f, and p. For a mechanical system there is generally no
confusion about what is meant by a coordinate, velocity, force, or
momentum. For clarity a set of generalized variables will be chosen
which is consistent with the definitions usually established. Considering'
a mechanical system consisting of masses and springs, the generalized
variables of a mechanical system are
generalized
q. = mechanical
coordinates
generalized

i*

R
<

generalized
J; = mechanical
forces

, l.'

..1

mechanical
displacements

= x.

mechanical

= i

mechanical forces
upon
position only

= f,

i= 1,2,3, ... ,m

i = 1,2,3, ... m

= mechanical = velocities

velocities

= dependent

- K,x,

i = I, 2, 3, ... , m

-,

~-~

Helmholtz free energy = A(T, V)

(analogous to the Lagrangian)

(1-109b)
These equations assume that T. V, T', and V' are defined to have the general form '

H(S, P) = U(S, V) --- VI'

1.4.7 The Generalized Variables for Electromechanical Systems

,'I

...

(1-IIOa)

,',l"

\,

(1-108a)

U(S, V) - TS

All the various state functions of thermostatics and electromcchanics

completely describe a given physical system and the use of one in place
of the other for a particular problem is determined solely by the nature
of the system being investigatcd and the desired form of the equatIOns of
motion.

(l-107c)*

These definitions are, however, merely useful classifications and are not
widely used in the literature of electromechanics.
Although the functions H' and L' have not been used in electromechanics
in the study of thermodynamics these general functions have been used
in the literature and are known by such terms as the Helmholtz free
energy, enthalpy, Gibbs free energy, and total energy. For a thermodynamic system the variables.are usually defined as
j.
i"

(analogous to the co-Lagrangian)

..

(I-I07a)*

V'

(analogous to the l'(1-ll;ll1liltonian)

(i(T. P)

Enthalpy = If(S. P)

by using Eq. 1-103a. Similarly, it can be shown that H' defined by

Eq. 1-1066 is the total coenergy, i.e.,
H' = T'

55

~~'-

,1
,;

Jqdf

la

;~

".~

i.,;~~,:

PI =

mechanical momenta
lIsually dependent
generalized
lIpon wlocity only l>ut _
mechanical --._ can
be functions of - p,
momenta
coordinates and velocities

M,.~j

i = I, 2, 3, ... , m

S6

ELECTKOMECHANICAL ENERGY CONVERSION

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

With the generalized variables for mechanical systems determined, the

!1ext ';(ep is to relate the variables of electrical systems to the generalized
variables q. {j .I~ and p. Rclcrring to Eq. 1-46 for stored electrical energy,
an inspection of the variables, charge ij and voltage v, shows that it is
possible to define these dectrical coordinates so that stored electrical
energy is either potential energy (Eq. 1-97) or kinetic energy (Eq. 1-99).
That is, charge (7 defined as a generalized coordinate qj results in electrical
stored energy being defined as potential energy; and clectric charge ii
defined as a generalized momentum p; results in electrical stored energy
being defined as kinetic energy. A similar inspection of El). 1-15 shows
that magnetic stored energy can be defined also as either potential energy
or kinetic energy.
It is now necessary to answer one further question. Is it possible to
identify both the electrical and magnetic stored energy with potential
energy or both with kinetic energy? The answer is NO! If electrical
stored energy is taken as potential energy, then magnetic stored energy
must be taken as kinetic cnergy and vice versa. This is bccause of the
dynamic rdationships which exist among the variables I';. ij;, i,. and A"
e.g., dij,/dt = i,.

CASE 2.

I
o
"

W. Identified as Potential Energy, W m as Kinetic Energy

generalized
coordinates

electric

= charges

ii;

= I, 2, 3, ...- ,n

.~~J?,:~4-",,:,'

,~

tl"H,

,;01/'.,

generalized

iii

= cIcct rical

wlocities

"

Ii

generalized
electric
forces

electric

= currents

):"t(",~

i,

= 1,2,3, ... , n

negative

= electric

-Vi

i = 1,2, 3, ... , n

voltages

,t \

= flux
linkages

= AI

= 1, 2, 3, ... , n

= 1,2,3, ... , n

negative
electric
currents

= -i;

electric
charges

electric

I, 2, 3... ,

11

i = I, 2, 3, .... n

iii

= I, 2, 3, ....

These results are very illuminating. For example. they show why the
direct analogy between electric and mechanical circuits in which the
potentials are analogoLis to Corces and currents are analogous to velocities
works best for electromechanical systems with electric field coupling
whereas the mobility analogue in which ele.:tric currents are analogous
to forces and electric potentials are analogous to velocities works best for
electromechanical systems with magnetic field coupling.
Now that generalized variables have been defined and identified for
both electrical and mechanical systems and energy functions or state
functions for both systems have been derived, it is desirable to obtain the
total kinetic coenergy and total potential energy of a mixed electro
mechanical system. For generality. consider a system with n electrical
coordinates and 111 mechanical coordinates yielding a total of n + 111 = N
coordinates. Taking We as potential energy and W m as kinetic energy,
the generalized variables are
General
Variables

Pi
Pi

generalized
electric
forces

q/

magnetic

II

ifi
generalized
electrical
momenta

= voltages = Vi

momenta

~~ '

= Ai

generalized
electrical
velocities

generalized

,!

q/ = electrical

magnetic
flux
linkages

qt =

Pi = electrical

Thus, consider two possibilities.

CASE I.

W m Identified as Potential Energy. W. as Kinetic Energy

generalized
qi = electrical
coordinates

57

Ii

Electrical

iii, i = 1,2,3,
ii' i = I, 2, 3,
AI' i = I, 2, 3
i = 1,2,3,

-VI,

Mechanical

,~l /!;

,n
,n
,n
,n

i = 1,2,3,
,m
Xi, i = 1,2,3,
,m
p/ = M;:i,j, i = 1,2,3,
.f; = -Kix i, i= 1,2,3,
Xi'

,m
,m

Using these coordinates, tl\e potential and kinetic energies can be found.

fiI
j.

r
r

ELECTROMECHANICAL ENERGY CONVERSION

58

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

Substituting the above variables into Eq. 1-97 for V yields

V(ql> ... ,j,,;

Now consider a second, equally valid, representation for this electro

mechanical system. Choose W m as potential energy and W. as kinetic
energy, and define the generalized variables as

Xl' .. X",)

.' "-,
-f.'
,-I _ '
/j(IJI' .. 'qn'X ..... ,X",)'lcl i fi(XI, .. ,x",)dx i

'"
-_ JIrll), ... I;,,:xl .. .\',n N,-,::",~'
~
0 ..... (1;0 .... u

.f

"

General
Variables

(I-III)

The line integral in Eq. 1-111 can be taken using any convenient path.
The simplest path of integration is the one where electrical variables are
zero when the mechanical system is assemhled and where mechanical
variables are held constant when the electric charges are assembled. For
this path of integration Eq. I-III reduces to the simple form

.I.

rx

'

L.
I

Ji

-ii' i= 1,2,3,

,n

..

-,

(1-112)

)~,

xm ; it> ... , in)

i 11 . . . . / n

0..... 0

L \(/1' .. , In;

X" . . . , X",)

d"

:.1

I;

'"

,.,

L p;(X

j , ,

.,

X m)

0'
dX i

i-I

Equations 1-112 and 1-113 are the desired potential energy and kinetic
coenergy where the electrical energy W. is potential energy and the magnetic
energy W m is kinetic energy.

PI = M;:i'ir i = 1,2,3,
-Kjxi> i= 1,2,3,

AI .....An n

0 .....0

i-I

,m
,m

i;(A;, . .. , ,\~; x" ... , x m ) d,\;

rX1 .....Xm m

q'i'

'ir~

+ JI

(,...... (i,'H.:;ll~~

0.....0

L - f;(x;, ... , X;") dX;

(1-114)

i-I

and
T'(xl> . , x m ; Xl> ... , Xm ;

Vir , Vn )

_ J.VI..... '.n L.~

0.....0

+ JIr

;-1

X1 Xm

lJI,

i,

~."

0 ..... 0

-'( VI'
"
.
, V n , Xl> .. ,

q;
~

L..

i-I

,.,

p;Cx l ,

of'

, X m )

dx;

Xm

) dv;,
(1-1 15)

~.

".

,~Ji_

.. .
~~~,

:'~:.
,;."...

.~
, ~ .,.:i, .
'(,

"

" ;;1;']

\
}~l I .
~

(1-113)

,m
,m

;'

'

:~.~"

i-I

x1' ... ,tm

0 .....0

"~

'll

'-'f

fl.

= )(

I-I

The potential energy is a function of the final state of the system which is
defined by the unprimed variables; the primed quantities in Eq. 1-112 are
variables of integration. The unprimed mechanical variables in the
voltage function Vi of Eq. 1-112 indicate that these variables determine
the configuration of the system (i.e., the capacitance of a linear system)
and that these variables are held fixed during the integration.
The kinetic coenergy T' can be found using Eq. 1-99. To evaluate the
line integral in the kinetic coenergy, a path of integration is chosen in
which the current is zero for mechanical variations and the mechanical
var.iables are held constant at their final values during integration with
respect to electrical variables. This yields the kinetic coenergy as
.'n.

i = 1, 2, 3,
i = 1,2,3,

Ji=

-,

V(Al> ... , An; Xl> ... , x m )

f - f;(x~, ... , x~) dX;

X;,

x;,

With the variables chosen above and the same paths of integration used
to obtain Eqs. 1-112 and 1-113, the potential energy and kinetic coenergy

1;(1"

T'(x" ... , x m ; XIr ... ,

'~'rl

.~~,~<

1\
'1 '

:;v
~,~.~

... , ...':.'.."
~

.f'

'.,.

i:"

!"':"
.,'

,';1

I,

,[fit'

*'--

PI

,n
,n
,n

V;,

vi (f/I',qn;Xh""Xm dq;

xm

)0.....0

-,

Mechanical

.\, i = 1,2,3,
i = 1,2,3,
q;, i = I, 2, 3,

qj

are

j~

0..... 0

II

Jti! .....ti

Electrical

til

V(qj, ... ,q,,;Xh""Xm)=

59

In obtaining the energy expressions, Eqs. 1-112 through 1-115, for

electromechanical systems, a path of integration was chosen which caused
all of the energy stored in the electric and magnetic fields to come from
electrical sources and none from mechanical sources. In the resulting
expressions for the electrical energies, the primed mechanical coordinates
in the integrands were replaced by the unprimed mechanical coordinates,
and the paths of integration extended over the electrical coordinates only.
It is easy to show that the partial derivativc of the electrical or magnetic
energy with respect to a mechanical coordinate is equal to the partial
derivative with respect to this coordinate under the integral sign for these
special forms of the electrical and magnetic energy. This leads to a
desirable simplification in the forms of the energy expressions, but
introduces an assymetry between the electrical and mechanical coordinates
which leads to confusion if not perfectly understood. That is, in com
puting the stored electrical and magnetic energy for a particular electrical

"

60

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

ELECTROMECHANICAL ENERGY CONVERSION

61
(

and mechanical configuration of the system, it is usually assumed that

with the magnetic flux and the electric charge equal to zero the system is
brought to its final mechanical conliguration, and then with the mech
anical coordinates held constant the electrical configuration is brought to
its final form. The stored electrical and magnetic energy are then supplied
entirely by electrical sources. Except in purely electrical systems this
path will, of course, not be the path actually followed by the system, but
since the total energy is a state function, in the computation of the stored
energy a path of integration which is most convenient can be chosen.

The Euler-Lagrange equation (Eq. 1-85) in terms of the nonconservative

Lagrangian L o is of the same form, namely

8L a _ !!.. (aLa) = 0

,I

. t

; (P/c(qlt ... ,qN;

t'

1.5 Hamilton's Principle Extended to Nonconservative

Systems

V o =

0..... 0

2: -

Qi(l) dq; =

i~1

;-1

Jq.
I

L -

Q;(t) dq; =

<

Now define a nonconservative Lagrangian in terms of T'. V. and VQ as

In terms of the nonconservative Lagrangian

of Eqs. 1-96 and l-lJl\ that
'(
..) .
j J, If\t ... , IfN' 1 f

r) () _.
~J,

Vo)

(1-117)

La it follows as an extension

iJ!.Q(q" ... ,q!';lilt ... ,.liN;/)

and
Pk(ql, ... qN;q" ... qN;1) = oLo(ql ...

. -=--:....:...c...__

()lfle

,q~:q" ... qN;t)

0q/c

~'
"

.'~ I

.J

"}
~ ~.

.:~;

'," .-tp

Q;(/)qi

i=1

= T' - (V

"I'

.it

11-118)
\
(1-119)

,~,'
, :,'-:'

-" ~.!!.;
,.'

- f/c(q,, .. qN; t)

d(OL)
8L = Q/c
--

dt 81h

( 1-116)

Lo(ql, ... , qN; q" ... , qN; t)

q" .... qN; I))

Q/c(1)

(1-121)

Equation 1-121 immediately suggests that another way to express the'

Euler-Lagrange equation without defining a nonconservative Lagrangian
is simply to use Eq. 1-96 for fie and Eq. 1-98 for Pie and write

'~'I

"l .... "/N

(1-120)

8tlJc

and substitution of Eqs. 1-118 and 1-119 into Eq. 1-120 yields

.f"

Hamilton's principle as developed in Sec. 1.3 in terms of the Lagrangian

function L(qlt ... , qN; q" ... , qN; 1) defined in Sec. 1.4.5 was restricted
to conservative systems. All conservative forces of constraint were
eliminated from the Lagrangian by a proper choice of generalized
coordinates, and all systems with nonconservative forces of constraint
were excluded from consideration. For the class of systems in which
the nonconservati ve forces of constraint are independent of the generalized
coordinates and velocities and are constant or functions of time alone, it
is possible to define a nonconservative potential function and hence a
nOIl\;onservative Lagrangian. Thus the equations of motion including
the nonconservatiw forces can be derived from Hamilton's principle.
Define the nonconservative forces of constraint as Qi = Q;(t). Then.
lIsing the definition of a force and a potential given in Sec. 1.4.5, Eqs.
1-96 allll 1-97, dellne a nonconservalive potential V Q which results from
nonconservative forces of constraint as

dt

oq/c

~i"~~

(1-122)

oq/c

Thus all nonconservative fo~ces which are independent of the generalized

coordinates and velocities can be viewed simply as driving forces which
are applied to the conservative part of the system at the ter~inaJs where
these nonconservative constraining forces act. Typical exa~ples of such
nonconservative forces are the applied voltages at the terminals of an
electrical network, the torque applied to the shaft of a generator. etc.
In addition to modifying the Lagrangian so as to account for non
conservative forces which are independent of the generalized coordinates
and velocities, it is also possible to define a nonconservative kinetic
potential to account for dissipative forces in a system. To do this it is
advantageous to define a velocity-dependent function F, called the
Rayleigh dissipation function. as
:.)

(.

F =

L ;r;(4;)2

'0

-+

(l-123)

I-I

The Rayleigh dissipation function F has the dimensions of power. In

fact, 2F is the instantaneous power consumed by the dissipative forces.
The definition of one-half the power loss is conceptually unfortunate but
it is necessary in order to obtain the correct force relationships, and when

dealing with nonconservative systems the only check on the validity of a

given formulation is the force relationships because these must always

hold whereas the variational principles may not. Now use the' Rayleigh

dissipation function F and define a nonconservative kinetic coenergy

function as the time integral of ,,~ i.e.,

T~ =

Jor Flit = Jr' ~ ~

t

0 2;-1

;,.
~,

riliT dt

(1-124)

62

ELECTROMECHANICAL ENERGY CONVERSION

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

This nonconservative kinetic coenergy function is a function of the

generalized velocities qj' Therefore, define a new Lagrangian in terms of
this nonconservative kinetic coenergy T~ and also in terms of the non
conservative potential V Q' The resulting nonconservative Lagrangian
L FQ is
LFQ(q\> ... ,qN; q\> ... , qN; t) = (T'

T~) - (V

VQ)

for each special case, it is usually preferred to use the l.agrangian to

obtain the equations of motion of the conservative part of the system and
to equate these to all noneonservative forces acting on the system.
A simpler way to include nonconsl:rvative forces than the ol1e .iust
presented is to investigate the two terms of the Euler-Lagrange equation
and see if they suggest a way to treat nonconservative forcc'. The term
OL/oqk was defined as oL/8qh = .rh, a generalized force (see Eq. 1-96).
The term oL/alik was defined as (iL/alik = {J/." a generalized momentum
(see Eq. 1-98). Since Pie is a momentum, then ft, is a force, i.e.,

,j"'

(1-125)

Using this nonconservative Lagrangian L FQ , Eqs. 1-96 and 1-98 can be

extended to yield
fk(q\> . .. , qN; t)

oL

Qk(t) =

(
..)
FQ q\l ... ,qN; q\l' .. , qN; I (1-126)
Ogle

d (OL)
Oqh

\r ~

dt

and
Pk(q\l ... , qN;

el10 ... , qN;

t)

+ (' of(el\l .... , tiN) dt

Jo

. ... , qN;
. t)
oLFQ(q\l ... , qN; q\l
oqle

';r~'
"};:

(1-127)

The nonconservative Lagrangian L FQ also fits the conditions of Hamilton's

principle and thus, for all the qj independent, the Lagrangian L FQ must
satisfy the Euler-Lagrange equilibrium equation (Eq. 1-85); i.e.
oL FQ _ !!.... (OL FQ ) = 0
oqk
dt Oelk

~,

'~.:,

1.~"'1.,

(1-128)

F.

~,

oL _ !!.... ( O~)
Oqk
dt oqk
'

~.

-)

Similarly, if there is a noncopservative dissipative force of the form

of/oilk = 'kitk dependent on the kth velocity, it will be a negative force
applied to the system. In this case the equilibrium equation is

j"

'.,

; [Pk(q\l .. qN; q\l ... , tiN; t)]

~'

f
l
I,

'}
I~

~." .

t
~{

of(ti\l ... , tiN)

r(
)
Q (t )
;l'
-Jkq1o .. .qN;t =
k
qk

d (OL)

dt oit/( -

oL
Oqk

of
041e = Qk

(1-130)

A significant feature of Eqs. 1-122 and 1-130 is that they both have taken
the form of the Euler-Lagrange equation of the conservative system
expressed i'n terms of the conservative Lagrangian (L = T' - V) plus
additional ~~rms resulting from the nonconservative parts of the system,
Thus, rat~r than defining new nonconservative Lagrangians to account

l_

,;,
"

,~

..

.~

"

~'.

'.

'~

"

;,

:1',

"!I,,'

f;f.

+~1

. I.,~

~~;

"-r

j,

!',L"

I'" '

l't~!
i ,: .).:

. j

:.~

0:'...,'

Equation 1-129 can also be expressed in terms of the conservative

Lagrangian L by substituting Eqs. 1-96 and 1-98 for fk and Pie into Eq.
1-129 to obtain:

".'

'."
.;,

.~

1-:,

I
'
,"
;,:.

oL _ !!.... ( O~)
oqk
dt oqk

'.

~,

(1-129)

+ QIe = 0

Substitution from Eqs. 1-126 and 1-127 into Eq. 1-128 yields:

= dtd (Pk) = Pk.

is a generalized inertial force. Therefore, the Euler-Lagrange equation

(1-85) is seen to consist of two generalized force terms equated to zero
and the Euler-Lagrange equation merely states that for dynamic
equilibrium of a conservative system the sum of all forces acting on the
kth coordinate is zero.
Now if a nonconservative force Qk is considered to act on the kth
coordinate 'Ik along with the conservative forces of the system. then by
d'Alembert's principle dynamic equilibrium requires that all forces,
including the nonconservative forces, must equal zero. The extended
form of the Euler-Lagrange equation for nonconservative systems
becomes simply

'~I

"ir:\}~

oqle

63

.'.

_ Oqk
8: + Qk

This equation is exactly the same as Eq. 1-130 which was derived from a
nonconservative Lagrangian. Therefore. nonconservative systems can
be treated hy the simple technique of setting all applied nonconservntive
forces equal to the Euler-Lagrange equation of the conservative part of
the system, i.e.,

d~t (O~) \vqk

~)L

(qk

all nonconservative applied forces at kth terminal pair

(1-131)

The most significant value of the Lagrangian when used as in Eq. 1-131
for electromechanical systems is that it yields the electromechanical

64

ELECTROMECHANICAL ENERGY CONVERSION

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

coupling terms easily and directly. The external nonconscrvative forces

are then equated to the forces derived for the conservative part of the
system to yield the equilibrium equations. Thus, in essence, the con
servative Lagrangian plus Hamilton's principle is one way to find the
interaction forces of electromcchanics. These conservative forces of
interaction can then be used to obtain dynamic equilibrium equations
because the sum of all conservative and nonconservative forces equals
zero.
Using the conservative Lagrangian plus a summation-of-forces
approach. the steps leading to the dynamic equations of motion for any
system are as follows:

I. Choose a set of generalized variables qi' (ii, '/;, and Pi and determine
the associated nonconservative forces Qi that arise from sources and
of/ ('q, that arise from losses.
2. I n terms of the generalized variables, write the expressions for the
kinetic coenergy T' and the potential energy V for the conservative part
of the system. The dclinitions of potential energy and kinetic eoenergy
for electrical systems will depend upon the choice of coordinates, but for
any chosen set of coordin,ltes the potential energy and kinetic cocnergy
will he defined hy Eqs. 1-97 and 1-99.
3. Determine the nonconservative forces associated with dissipation
and write the Rayleigh dissiration function F.
4. Obtain the conservative Lagrangian as L = T' - V.
5. Substitute the L so obtained into Lagrange's equation

d(OL)
i:Jqk

di

8L

of

- vqk + V{ik = Qk

In terms of these variables the energy functions arc

T' = -!M.t 2 +

t-

EXAMPLE lE3a

to N.

/~.

k=2
electrical

qk

ilk
Pk

-jk
Qk

JtI

Mx

Kx

J(t)

e(/)

r'

til'
~.

~MI\
r

.,.
;!~
1l!.

;I:ft',

x) tli'

(IF1-3)

: I~

'" = mechanical friCl'"''

1'--'

. .coetf,clent

i.

il,\ xi
v

~~ ::'

,.,
,d' !~

.:i,

---]

K
'"

rltf'

(Oo1io'------t:;;

r4

',
I~;

EqUilibrium pOSition
for i = 0

.1

I:

Fig. IE3-I,

Electromagnet for Example IE3".

The graphical meaning of the term (i >"(i'. x) tli' is indicatt:d in Fig.

J"

1E3-2.

The loss function F for Fig. I 3-1 is

'-<'~-

= -!ri 2 + -!ax 2

(I E3-4)

"so

where r

resistance of coil

a = viscous frictional resistance of relay arm

iii

'r. ~ i."

To substitute Eqs. I E3-3 and I E3-4 in Lagrange's equation (1-130), the

following derivatives must be evaluated:

.'

:,gl$.'

(I E3-2)

= -tM.i"2 - tKx 2 + (I

~:.

;~;'-'

k=l

(I D-I)

-iKx 2

});'

.ll

mechanical

A'{i'. x) di'

L=T'-V
.1

The equations of motion are desired for the electromagnet of Fig.

I E3-1. Choose the magnetic energy to be kinetic energy; thus thef
generalized coordinates are

J:

The Lagrangian for this nonlinear device is

~',"1'

1..

':.<'"

and obtain the N dynamic equations of the system by setting k =

65

..

-.:"

J.\,.".
f:'

fl;f

, ~',,
';

~~.

.1.:/

d(VL)
(I) di oi/k
((\. ;

for k = I:
~~..f .~

fork

= 2:

f.....

(8L)

.
-d ----; = -d (Mx)
dt ox
dt

d .
dt (OL)
vi = dt[A(/,x)]

"d
.~

~'

,."

EQUA110NS OF MOTION OF ELECTROMECHANICAL SYSTEMS

ELECTROMECHANICAL ENERGY CONVERSION

66

Equations 1E3-5 and IE3-7 are the equations of force and Kirchhoff's
loop equation for the electromagnetic relay. The loop equation, instead
.of a node equation, is a direct consequence of choosing magnetic energy
to be kinetic energy. To see the effect of choosing magnetic energy to
be potential energy, the equations of motion with this selection will be
derived.

f!L
(2) i:)qk

I:

~~
=i:!.X

for k = 2:

oq = 0

for k

Kx

iJiJ

[Ji() >..'(i', x) di']

t
,!,

iJL

;:.
:

of

(3) Oqk

= 1:

for k

for k = 2:

(4) Qk

for k = 1:
fork = 2:

I.

of

'.:,~

8x = aX

of
8;

QI

EXAMPLE I E3b

~}},

Magnetic energy is chosen to be potential energy, and the generalized

coordinates are
k=l
k=2
mechanical
electrical

.'

.~~~",,:

...~1

~~';
i.'
"'.

= ri

"i~i

"

= f(t)

:",

1f

qk
qk

;.~

Q2 = e(t)

Pk
-fk
Qk

~.

i'

:1,.
ill),

Total area of rectangle

Kx

iI

f(t)

i(t)

=i),.

nr

t;tll

:;'.
I
.'~

> l

and for k = 2

or

;t [.A(i. x)] +
0>" di

ir = e(t)

a>.. dx

oi dt + ox dt + ir =

e(t)

coeffiCIent

u..(:

I. fi~JO~
f..

EQUilibrium POS,tIOIl"11i
for i = 0

:J't

Fig. I El-J.
energy.

Ii<: JC,

System for considering magnetic energy as potential

In terms of these variables, defined in Fig. IE3-3, the energy functions are
T' = !Mi 2

= f(t)

,) ~.\

.1

."

'i'

.~~

Substituting the above in Lagrange's equation yields, for k =

U~ >..'(i', x) di']

=mechanical fnction

I;~

".

Illustrating the relation between energy and

Kx - :x

..~,.

"

:r (Mx) + aX +

I ,

, " ..!

).'

).

i1
-~

i'"

tllrtltT

v
q

(:I

I '!~:

>..

i
Mi

.,

li~o.'.X)d)" =energy

'1

Fig. IE3-2.
coenergy.

67

(IE3-5)
V = -tKx 2

(IE3-8)

+ LA i;(,\', x) d>'"

(lE3-9)

The Lagrangian again can be obtained as

(IE3-6)

L = T' - V

= +-tMi2

(lE3-7)

~;.c

'J'~".

-tKx 2

f:

i;(A', x) d).'

(IE3-10)

68

ELECTROMECHANICAL ENERGY CONVERSION

The loss function is

where g

ta.x- + tgv
2

(IE3-11)

equivalent conductance of coil, assuming this resistance is in

shunt with the coil. This assumption is valid if the source
<'(I) and resistance r of Fig. I E3-1 is replaced by the current
source i(t) and conductance g as in Fig. I E3-3

II
;

To substitute in Lagrange's equation. the following derivatives must be

evaluated:

'.

'I:

}:

:!dt (OL)
c.x-

=~
dt (M')

d(OL)

fork = 2:

dr

(;Iu

,I;,;.

rL

(2) oC/k

for k = I:
for k

2:

~: =
aL
v,\

-Kx - :x

U:

1.6

'1('\, x)
',.,:

iJF

(3)

:~-~

och

"

7~:

for k = 1:
fork

aF

ox =
of

aX
'l~,

,,'(.,.

= 2: au = gu

(4) Qk

QI

fork = 2:

Qz

= i(t)

J(t)

~I (Mx) +

Xx

+ -$.-

ox

fJA i;('\', x) d,\']

l

and for k = 2
i l ('\, x)

i(l)

() W;"(i,, x)

ax

Bx

foAi;(,\'. x) d>.' = :x J:l ,\'(i;, x) d;

(11:::3-16)

(I E3-17)

An introduction to Hamilton's principle. the Lagrangian, generalized

coordinates, etc.. for a dynamic system has been given. One topic still

remai{\s which is important in the discussion of electromechanical systems

and that is a treatment of quasi coordinates. With any true coordinates

it is always possible to find a set of linear relations between one set of

true coordinates, say qt> qz, ... , qN. and a new set of true coordinates
q;, q~ ... , qN' The relationship between these two sets of coordinates
will be of the general form

+ .. , + a'NqN

q; = allql
q; = aZlql

+
+

Qlzqz
0nqz

qN = aNlq,

0NZqZ

+ ... +

aZNqN

+ ... + 0NNqN

or

aX

J(I)

q~

(lE3-12)

+ gu =

i!W/II (\1\, x)

1('

Substituting the above in Lagrange's equation yields, for k

ct

(IE3-15)

Quasi Coordinates and Ignorable Coordinates

"P,L

for k = I:

J>V(i;,X)di;

;~,;

i;(A', x) d,\']

W;PI> x)

Thus Eqs. IE3-5 and 1E3-12 are equivalent, i.e., the same force cquation

was obtained in each case. However, for the electrical equations. namely

Eqs. lE3-13 and IE3-7. note that Eq. IE3-13 is a node equation and

Eq. 1E3-7 is a loop equation. Thus. defining the magnetic energy as

potential energy instead of kinetic energy changes Kirchhoff's equation,

which results from the Euler-Lagrange equation, from a loop to a node set;

(I EJ-14)

_/ ~. = _

:1

- :x
I'.

i;(>.', x) d.\'

and the mechanical force of electric origin is

or

k=

f:

(ilL)
(I) dt r:cik

or

Wm ('\, x) =

I'

69

To see that Eqs. 1E3-12 and IE3-5 are equivalent, refer to Tahle I-I'
in which it is shown that

:,}

a = viscous friction resistance of relay arm

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

(1E3-13)

,.

,I','.

=I,"2:I 0k/q,

= 1"

.. , N

(1-132)

Now consider a case in which it is not possible to find an expression of

the form of Eq. 1-132 between two sets of coordinates, but instead it is

"w,~"'
Th',

of "",' ,ooro''''''

follo~ Wh""koc. ~ '_;'....._,

=_

S 71
EQUAT IONS OF MOTIO N OF ELECTROMECHANICAL SYSTEM
1, ... , N and the results are added to obtain:
is multipl ied by Pk, for k

ELECTROMECHANICAL ENERGY CONVERSION

70

ves of the
only possible to determi ne relation ships between the derivati
coordin ates in the general form
(i~

= 2:I hk // I k

(1.133)

= I, ... , N

~,

OL] _ t:': R Q

'!r,t,z,

If an arbitrar y

i .. ~

oqj)' The
where the coefficients bkl -:/- (oq~/ oqJ 0; (obkJ oqj) -:/- (obkJ!
le, and
integrab
not
are
1-133
Eqo
by
d
differen tial equatio ns indicate
of

form
the
of
ates
coordin
the
between
ns
equatio
hence a set of linear
q~ are
ates
coordin
the
ons
conditi
these
Under
exist.
Eqo 1-132 cannot
1-132).
defined as quasi coordinates (satisfy ing Eq. 1-133 but not Eq.
with
use
for
gian
Lagran
a
form
to
used
be
cannot
ates
coordin
These quasi
Euler
the
the Euler-L agrange equatio n; howeve r, it is possible to modify
of quasi
Lagran ge equatio n so that a Lagran gian express ed in terms
coordin ates can be used.
ates,
To develop a form of Lagran ge's equatio n valid for quasi coordin
ates
coordin
quasi
the
and
qN
,
...
qt.
the true coordin ates are defined as
of
ves
derivati
time
the
between
s
relation
are defined as gb ... ,gN' The
are
ates
coordin
true
the
of
ves
derivati
time
quasi coordin ates and the
N

2: a,ifle
= k-l

(1-134a)

r = 1, ... , N

:: R [d (OL)
k-::/"'k, dt oih - oqk - k<:-/'I., k
displ.tc ement is made. the term 2: QI. "''II.

L Pk, sg,

,=1

.)

where "

oa,i

(I-I 34b)

r = 1, ... , N

1
.:\~~~

become s

f"

'~:o

oL
oqk

{if
.~.

= 1, ... , N
k == 1, ... , N

tile

and the

(I-135a )

oL

dt oqk - Oqk

(I-135b)

'J
/~.

==

j,

Qk

~'. ~

0\1, {(

oqk

o<;s

1= 1

(1-139)

(qk

',;.~

::i\

{ fik
Ok-I

L L fikrask
k-I.-I

',.'
.II"

'"

-;'.,

,".

oa" q ]
[~( ~ oL a k) _ oL _ ~ ~ oL
O~.oqkl =

rdls_ IO~.s

oqk

'=JI-I

Z,

(1))

or

1;.....

i'
"

'"J
t,:':'

.-..: -'~i::
":"
,.;

....

L.
+.=1

= -

1.
,"

(1-138)
lell

t:': oL ~ oa" ql
L. -'10

oL

'j'U-. .

just as the true

.. The differentials of the true coordinates cannot be determined.
ed as a definition
interpret
be
must
1-134b
Eq.
e
Therefor
inate.
indeterm
are
tes
coordina
of the dilTerentials of the true coordinates.

~ oL a Ie
.-Iot

Substit uting Eqs. 1-138 and 1-139 into 1-137 yields:

Now Lagran ge's equatio n in the form of

d (OL)

.~I o~. oqk

oqk

oqk

1'?

~ oL o~.
.=Iof,oqk

oL = aL + ~ oL o~.

"

2: Pk,d~,
,-1

. ,

Similar ly,

By solving Eqs. 1-134, the i nverse relation ship between the

~, become s

dqk ==

(1-137)

Zr

gian
As the next step the variable s 410 ... , if N are replaced in the Lagran
, ~N'
~J,
es
velociti
te
oordina
quasi-c
the
by
I)
q",;
...
041'
qN;
,
L(q]> ...
~N; I). Then
To do this define a Lagran gian L as L(ql> ... , qN; ~J,
as
L
gian
Lagran
oL/oqk can be written in terms of the new

oQi

or

k=1

L Pk,~r
41r" ,-1

j-:/-r

OL]
~
-d1 (OL)
L."!. fikr [d
uqk
uqk - ~

:j

.;,fJ

i= O~,j

oqj

Sqkl~{I_O

arbitrar y
i.e., Pk, sgr is the displac ement of the kth true coordin ate if an
re
Therefo
.
allowed
displac ement of only the rth quasi coordin ate is
y
arbitrar
an
is
2:k Qkfikr S~, is the work done by externa l forces when there
l force Zr
displac ement of only the rth quasi coordin ate. The externa
Eq. 1-136
and
fikrQk
L
=
Z,
as
defined
is
ate
coordin
rth
acting on the
k

l.

2: a,k dqk
k_1

will he the work

I.

all the qk
done by all externa l forces for an arbitrar y displac ement of
coordin ates. Now the term

tials of
and the assump tion is made that relation ships betwee n the differen
ed
express
the quasi coordin ates and of the true coordin ates can also be
in terms of the same a,1e as

dg,

(1-136)

d'-I

(oL)
0
U<;s

oL dask]" LN.
fik'""fiT -dl - k-I

!;".

~ ~

oL
f3I., a'I.

qt
i Ph oL oa"
oqk

Ie-I .-1 I-I

o~.

Z, (1-140)

'\

ELECTROMECHANICAL ENERGY CONVERSION

Next, the following equalities are recognized:

"l) _y

(-

I,

{O

\I

( I)"
"(f3 A, <l "A )
,-I ,-I
Q

'1l

cc

'11

'I'll

~ = " :~
(qA = ~
.l' ,
' t

l'Ar '

(Iqk

I,

"qA ()\"

'1'11e d
' "Ilion
din

(1-141)

(1-142)

(}",

(,l,,,, -;-z
(J
Sr

y,...

()qA = -;;-z
(JL wou Id f 0 II ow d'lrect I
y ' If

[)L
-,-

(he ~, were (rue coordinates; ~jnce they arc not. (he ahliv('

'
('(lUll I11.1'

'
/.1'

I
1/('

I /'1IlIIIOII
,.
(('

aL
(.I~,

Ii /'

!!, ((jL)
dl

an d'It must be usc d

accordingly
N'
I
(.1) {""A
-ill = /-_,e",I,.
-',- li,
I (,q,

(1-143)

Using the equalities of Eqs. 1-141. 1-142, and 1-143, Eq, 1-140 can be

l.'

reduced to the following:

~~

(iL) + ~-

'(.
{I('s,
I

I,

~ )'-

S'

I ,

f3
I

I"

?L.

~t 4"

(va.

,I,

eq,

V\",

aa.'") _ ('!'
' "
({h
('L

(1-144)

_..'t .

Equation 1-144 is the equation of motion obtained from a Lagrangian

formed in terms of quasi coordinates. i,e" the Euler-Lagrange equilibrium
equation expressed in terms of quasi coordinates. It follows directly
that if the coordinates
were true coordinates, then

e,

Oct.. k

:\1

iht.'it

eql = oqA

I
I

and Eq. 1-144 would reduce to

~'i

!!- ((JL) _ oL = z
ell at
ae,
,
The term

(if

:Ii.

e, = true coordinates)

~.

~_

~.,

~,_ ~'f3
aL i (?Ja'A _ ca",)
_
1" ')t 7/"
"

,
1 , I' I
( S's
NI,
'{Ik

is the correction term resulting from using quasi coordinates instead of

true coordinates. Further, note that if cil is expressed as

4, = 2: I f31U~u
U

then a portion of this expression-,-namely

~ fL f3

1,- I , -I

A'

'u

=ALN.I ''Y
LI

f3d~/,,

(Oa"k
_ CaH)
aq
aqA

( 1-145)

Then Eq. 1-144 becomes

(I~r

73

is dependent only upon the connection between the true coordinates and
the quasi coordinates. and is independent of the motion of the dynamic
system. This quantity is denoted by y,su. which is called the Christolfel
symbol. i.e.

!..

l r
= r

.I'

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

~:{

(oa S oas')
i:{/k
k _

('(I,

====

?J~"

+ ~

')L

","

"

yT.\'U~U (Jt

aL

__ =

(J(,

z~,

( 1-146)

S'"

EquatioPiJ 1-146 is the Boltzmann-Hamel form of the Euler-Lagrange

equation and must be used when the system variables adopted are quasi
coordinates.
Future work will show that quasi coordinates are very useful in certain
problems in electro mechanics ; and. in particular, they are a necessity
when treating electric machines that have commutators. In general, for
problems in electromechanics, quasi coordinates must be introduced
when the system has nonholonomic constraints between the velocities and
simultaneously ignorable coordinates will usually occur. This is in
variably the case which arises in the study of magnetic field devices such
as commutator machines. An ignorable coordinate occurs whenever the
Lagrangian does not explicitly contain that coordinate. Thi.l; is of course
a common feature of magnetic field devices since the stored energy and
hence the Lagrangian are only functions of the currents and not of the
integrals of the currents, the charges. A problem with nonholonomic
constraints on the generalized velocities and with ignorable coordinates is
particularly dangerous because there is no need to solve the equations for
the coordinates and hence it is possible to overlook the fact that co~
ordinates are not determinable from the generalized velocities. Thus it is
possible to establish a Lagrangian in terms of a set of quasi velocities
without being aware that the velocities are quasi velocities and not true
velocities. Such a Lagrangian, expressed in terms of quasi velocities,
will yield erroneous results if used in the usual Euler-Lagrange equilibrium
equation (Eq. 1-85 or 1-131). The Boltzmann-Hamel form of the Euler
Lagrange equation (Eq. 1-146) must be used instead. As a general rule,
when working with problems in which ignorable coordinates exist and
when using variables that are not expressed in the same reference frame
as the actual variables in the physical device (i.e .. currents which flow in
physical coils), the generalized velocities which are chosen should be
investigated to see if a set of coordinates is obtainable from the relations
among the velocities. In other words, it is necessary to determine whether

.......

s=s

-~

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

ELECTROMECHANICAL ENERGY CONYERSION

74

or not the vdocilies <Ire integrable to yield a sct of coordinates.

merely rcqtlll'cs the establishment of whdher
(}rt..\./\

r)a... ,

i,y.\J,.

--~

or

~ql- ~= uqk

i)lJt

()u.",

-j; -

qk

--

Nc

~
J

1.'
,."

:i'

' ;.i~ .'

t

..

>"u

14

Nd

Vd

I.

v'f
T

v'

>,:

.,

.,

vb

.p

Z,

_-~

i!ou
ib

The equations relating the ~,.

J.p

tr' and the qk' q~ are of the general form

2
3
4
S

tik

'..,
.,
'.
"

Ib

4>

fie

Pk

Qk

>" a
>" &
>"

v'a

J~

>,;

(sin </i~

+ (cos </J)ib

where the various

"

all =

!.iJ.

a22

'.'f'
>~"

a33 =

~.;'i

a44

:<~,
~":' :

aS5

ark

are

al2

a2l =

cos </>
= cos </>
= I

(I E4-3)

a'k(/k

aD

= al4 =

alS

=0

a25

= 0

an = a24

=-

sin ~
= sin 4>

a34
a43

a51 = a52 = aS3

a31 = an

= a3S

a4t =

lX54

a42

a45 =

0
0

(lE4-4)

=0

tr

The new coordinates ~r and will now be investigated to see if they are
true coordinates. If the ~,and g, are true coordinates, then the coelficients
ark must satisfy the relationship

v:
v'a
v;

oa,; _ oa,)
oq) - oq,

fI :

oan

J'.:I:

Oqs

(lE4-1)

(I E4-2)

Designate these new coordinates and velocities by a set of variables er

and ~, respectively; thus the table of new coordinates is as given in
Table IE4,2.

(IE45)

It is obvious that a1l the coefficients of the first, second, and fifth
coordinates do satisfy Eq. 1E4-5, for all five values of i and j. This
should happen. of course, since ~1 = qlo '2 = Q2. and ~5 = Q5' However,
Eq. 14-5 applied to the third and fourth coordinates yields the following
results:

i;

i;

k~1

fe,....,
' I ".:,'

variables the generalized coordinates q" q2' q3' and q4 do not appear in
the system Lagrangian L.' In fact, L can be shown to be a function of
the' four currents, the mechanical displacement </>, and the mechanical
velocity .p; i.e., L = Lei;, iZ, I;. lb' </>, .p). Now when dealing with certain
classes of electric machines-- in particular, commutator machines-the
true coordinates and velocities on the rotor are not directly measurable
at electrical terminals on the rotor. Instead, the measurable rotor
currents are i~ and I;, where these currents (generalized velocities) are
related to the rotor currents and i b by the linear equations:

i;'''' (cos </J)i; - (sin '</ib

" =L

= i3(cos </
o</>

=-

.
SID

4>

(IE4-6)

oa35 = 0
oQ3

(I E4-7)

OIL 33
Ott35
#-.
oQ5
oQ3

(14-8)

Thus

,"c

I ,

lL

._.,--

.~,

TABLE IE4-1

1
2
3
4
S

oqk

p;(~ .. ~,)

n~,)

----

The equations of motion which are encountered in the study of electric

machines always include some variables where the coordinates are
ignorable. Choose a set of generalized coordinates and velocities for a
machine where the coordinates are true coordinates. These coordinates
and velocities will be designated as shown in Table 14-1. In this set of

.1,'

t,

~,

EXAMPLE IE4

TABLE IE4-2

This

Once this has been determined, the correct approach can always be chosen.
However, if true physical variables are chosen which exist in the reference
frame of the system being described, these variables will be a set of true
coordinates and true velocities.

75

:L,
"
or"~"~

76

ELECTROMECHANICAL ENERGY CONVERSION

EQUATIONS OF MOTION OF 'ELECTROMECHANICAL SYSTEMS

By a similar process it can be shown that

(',1 \4

oa ,~

C1qs

r)q4

8(.(44
cqs

Oa4S

- ' --"

(I E4-9)

(1 E4-1O)

a~43 - ar~45

(1 E4-11)

oq4

oqs

oq3

where a, b are constants. The winding resistance is R.

(u) Find the force applied to the plunger by the electrical system,
(I Write the electric,t1 and mechanical equations governing system dynamics
iptluding the force of gravity.
(/ 1-2. In the device shown in Fig. I P-2 fringing fields are neglected, allowing
the capacitance to be expressed as

d-x

~."x::.:<

>~;~~4 r

?~;~%;~I

d
I

tX

W';~?~'7~, ~~..,w:0..?;:v3'???:,;/p~";.,,/%,n*U,,W:*?

...&M"M;'\/; -?-.',.?"/.S:;;l/A? Stop

Fig. IP-I.

Wd /{;;-;Wffi;;;;/""Pi'-~/;P'W/;:W?rii/%'/'2x':d:'//<-~(-;:(J
;.

,)

+ 0 ~-

~Jx ----r
I

-cr

l!'d!0$J'..7dWlJ)#),J'Mii-WJ>b.&,wJ>..M,WffffbiWhffJ'Jd'ffd?aJ.

Zl

, F i g . I P-2.

1-3. Experimentaily it is found that the flux linkage in the system shown in
Fig. I P-3 is related to the location of the iron slug and to the current in the
coil by the relutionship;

>. = (IOOOi l)2 /l

x3 + 1
The resistance of the coil is R ohms.

bt\(x - d)2

C=~

I-I. I n the electromagnet shown in Fig. I P-I, the relation between current,
plunger position, and flux linkages has been experimentally found to be
a;V

f-:};;J

~ :.

PROBLEMS

i4'

dam~tng
tr

'~0;;X::

Thus the velocities ~J and f 4 are not true velocities since the equations
for these velocities are not integrable, The coordinates and velocities
~3' (, and t4' L are quasi coordinates and cannot be uscd to form a
Lagrangian which can be uscd in thc Euler-L.agrange equ,ltion to obtain
the equations of motion of the electric machine, In the study of electric
machines it often happens that quasi coordinates such as the above are
the preferred choice of variables with which to formulate a particular
problem, When this is the case the Boltzmann-Hamel form of tl'" Euler
Lagrange equation (1-146) must be used or the system equations must be
found by a ditferent techni4ue. This problem will be given additional
attention in latcr chapters. The signdlcant item to establish by this
example is that thc coordinates and quasi coordinates an: not easily
detectable by simple inspection when the ignorable coordinates arc the
quasi coordinates. Tables 1E4-1 and 1E4-2 certainly look similar but
Table I E4-1 has true coordinates whereas Table I E4-2 contains two
quasi velocities, i.e., 3 and

-c- - -r-

MeChanica"I

77

(a) Obtain an expression for the mechanical force of electrical origin.

'II,'

(b) Write the equations of motion for the system.

,0

"

where A is the plate area. When the ,Ipplied voltage' v = and the applied
mechanical force f = 0, the system is in equilibrium at x = O. Neglect any
mechanical friction but include the force of gravity.
Find the dynamic equations of motion for the device.

it'

;I-~

.'

_::>.

\~
I

'::;[J .
~

+_~_~~

~NO trict:on

~/~

Fig.IP-J.

f.

JI
78

1-4. For the singly excited magnetic field transducer shown in Fig. lP-4 it
has been determined experimentally that
W)I/2

,I

where a = 104 ,

q(l' .

'r

.I')

= q",(1 -

C'~"')

where f3 and qm are constants and v is the voltage across the cOllllcnser plates.
The equilibrium pPsllion or the sprll1g is .I' = xo.
Obtain the dynamiC equatlOn~ of motion for the device.

is measured in amperes, and x is measured in meters.

, ':.'

'.OLd

,I

1~

i"

"~:'

,,:.1

I I

1\

;~

. (.,.'

Fig. I P-4.

This representation is valid in the ranges 0 < i < 3 amp and 0 < x < 0.04
meter. Neglect the effects of gravity.
(a) Plot the transducer characteristics on the Aversus i plane within the given
limits for x = 0.01, x = 0.02, and x = 0.04 meter.
(b) On the plane of part (a), plot three convenient constant force contours.
This is to be done by calculation rather than by purely graphical means.
1-5. The lossless, massless system shown in Fig. 1P-5, with
q = v2

79

1-6. In the pressure pick-off in Fig. IP-6a the dielectric is descnbed by the
o nonlinear
curves of Fig. I pooh having the eljuation

A(volt-sec) = -1'--,
-- ax"

1<

II

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

ELECTROMECHANICAL ENERGY CONVERSION

[_A_]
B + Cx

where A, B, and C are constants, is subjected to the following sequence of

(1:1:
:" '1
operations:
(I) With x held zero, q is raised from zero to Qo
(2) The capacitor is open-Circuited.
(3) The slab center is moved to XI'
In terms of A, B, C, Qo, and XI express the energy converted to mechanical
form during this sequence.

;1
f

~,~ VISCOUS damping = It

';;

~Pistonof

,~"

mass

=M

'3\'"

~f~_t

,;~..:l

Ii; ~ ~

\~~:

'rJ,.

",#' '

~~~~~M

Nonlinear
fluid dielectric

,
I
..

');

Spring constant ..

(a)

';-."

:jr;'
If!'

I:t
:J;,

l{t,

qm

~~

0:-:
i

'To

I
I

I
I

I
I

':

. . !::

-;;;i -1

---

~1 + q
n
I

,~

.~,

)~~ :.,'

~'l"::l
j .'.0;
",

",

Note q and () are the

charge and voltage
respectively of the

If

%\< "2 <:r:1

condens~r

Massless dielectric on

ji

IX \

frictio"'"

ro"~

-I;

~j,~

-:.\"",

(b)

,'.I

Fig.1P-6.

('. ~I,

Fig.IP-5.

"

'f"

;;'
;-

":.j

1.;, ~("i'
~,,','

',,:1/.;.

k --

..

"

:.'
~

'I
,

------.

't

80

ELECTROMECHANICAL ENERGY CONVERSION

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

1-7. A vibrator driwn by an electromagnet is constructed as indicated in

Fig. I po?~ Con,ider the mechanical system to have the following lumped
mechanical parameters when referred to the magnetic axis of the coil:

The A-i magnetic characteristics arc ,Ipprox;mated by the following function:

A=

2A.

-.2

tan"

7T

mass = M kilograms
clastance = K newtons/meter
damping = a newtons!(llleter/sec)

v.'

f
'. I

.,']

~.
(/
Springs

- a, ;,
'< ,"

.-/

ti

~~:t;{.

,?"xN"X9/.'r.-.r"

Fig.IP-7.

<. .".

r'o

<0W';";/:i'~W'~~

1
Mass = M

gl

viI)

,+

II)!

'0'

Electromagnet

, '/,;//,,0,D:M;f'l:~;X0j/j;:r;;'l'//--,~;~:~''j~~~';'/

(x -

1-9. A chunk of materiul of mass M is suspended from a spring with a

stiffness of K newtons/meter, as in Fig. I P-9. The length of the ,pring is
when there is no electrical excitation. The Illaterial is diamagnctie (" .: 1'0)'
The ,\ wrsus i curve for the mechanically Iixcd coil may be assumed electrically
linear; hence ,\
L(x)i. 1\ Yoltilgc source /'., with internal resistance 1<, is

;;-;~0;

ci

I ,---_---.,."

where A" c, and d arc given constants. The coil has a resistance N.
(a) Find the electromagnetic force acting on the plunger in terms PI' ,\ anL! x.
(I Obtain the I1lcch~lnical equation 01' motion for this deVice, includll1g
gravity effect. (Neglect effects of damping forces.)
all;

81

it

JJ.<JJ.o

R.

".t

The coil has a resistance R and an inductance L that has been found to be

"

terminals

Fig. IP-9.

.~

attached to the coil terminals. The inductance may, in addition, be assumed

to be a linear function of position, L(x) = Lo Ax, where L o and A are
positive constants.
(a) Derive the equations of motion of this device.
(b) In which direction does the magnetic field of the coil exert a force on
the chunk of material?
(c) The coil current is slowly raised from zero to the large value IF. Find the
energy storage in (I) the magnetic field. (2) the spring, (3) the mass M.

t'

//() 1-8. A constant-current source supplies power to the exciting coil of an

electromagnetic plunger shown in Fig. 1poll. The plunger has a mass M.

.~

~.

Coil

henrys

where x is in meters, A is in weber-turns, and a, b are constants with dimensions.

The spring force is zero when x = Xo meters.
Find the dynamic equations of motion for the system.

.~

1-10. The winding of the singly excited. magnetic field energy converter
shown in Fig. I PI 0 has an experimentally determined inductance
L = Lo

+ L z cos 21> + L 6 cos 6

where L o [2. and L 6 are constants and 4> is the rotor position.
circuit is fed by a current source
i(t) = I sin w,t

The rotor is driven at constant speed by a mechanical source

Fig.IP-8.

,/.

~ =

w",;

rp

= w",t + 0

The electric

82

ELECTROMECHANICAL ENERGY CONVERSION

EQUATIONS OF MOTION OF ELECTROMECHANICAL SYSTEMS

(b) Find the values of mechanical speed W m for which average power is
converted between electrical and mechanical forms.
(6) At each of these speeds find the stator voltage l1(t).

--f7

83

(0) Find the equation of motion for the mechanical part of the system.
(6) The system is subjected to the following sequence of operations:
(I) With x held zero, q is raised from zero to Qo and ,\ is raised from zero to An.
(2) The coil is short-circuited and the capacitor is open-circuited.
(3) The slab center is moved to Xl'

)~~~

~;

1f,.
J~~~,

How much energy is converted to mechanical form?

1-12. A chemical balance has been adapted to "weigh" voltage against

current, as illustrated by Fig. I P-12. One side of the balance suspends a movable
plate of a parallel-plate capacitor, and the other side of the balance carries a
movable coil of a pair of coupled coils. With the capacitor and the coil
unenergized, the scales are perfectly balanced and the pointer rcads "0." To

'

,'I,

:$ff#$>-~~

V
,l;t

;i;';

Fixed coli

'--Fixed plate

I,

!~1

Fig. IP-Io.

Flg.IP-I2.

J!.

I-II. A slab of ferrite material of mass M rides on frictionless rollers as

shown in Fig. IP-I J. The slab is attracted both by a capacitor with mech
anically fixed plates and by a rigidly held coil. The location of the center of
the slab is defined by the coordinate x.
,

II

~
~.
t
o

W//.MW~.""
-q

:t
!:f.:
.~~
,~
>:~~~

1%1
I

r+j
:%1
, ,

:1:

'~:

I~:

41

'~L::J

;;W,M;.b.</A0Wff.WM.:....sJi.
:
Fnctlonless rollers

= v3(D

+ Bx2)

.; eJ

for coil
for condenser

capacitance of capacitor

"j

,
,.

i.

.'

: 1

li."i~t;
r"';"-':'~'

inductance of coils in series

X2 .

permittivity of air, O(36-rr) x 10- 9 mks unit

A = area of circular capacitor plate of diameter d

L = constant, B = constant

(0) Derive a literal expression for steady voltage V as a function of steady

current J, for the balance condition, and sketch a calibration curve for Vagainst
1, At the balance condition, XI = 0, X2 = b.
To calibrate the balance, the capacitor plates are short-circuited, and with
1.02 grams added to the movable plate, 100 rna are required to restore balance:
a = 100 em, b = 200 em, d = 10 em.
(b) If the current is 200 rna at balance (voltage applied to the capacitor, no
added weight), what is the voltage across the capacitor? What is the electro
magnetic pull of the coil?
in which

The characteristics of the coil and condenser have been experimentally found
to be:
.
P(A

EoA
XI

Fig.IP-II.

~ -

weigh a voltage. it is impressed across the capacitor, and balance is restored by

gradually increasing the current in the coils (connected in series) until balance
is achieved. as indicated by the "0" reading of the pointer of the scale. The
reading of the ammeter then is a measure of the voltage applied to the capacitor.
Flexible connecting leads may be regarded as having negligible effects on the
balance.
For small limited motion of thc balance between stops,

EO =

84

~
~

;:

ELECTROMECHANICAL ENERGY CONVE RSION

EQUAT IONS OF MOTIO N OF ELECTROMECHANICAL SYSTEM

S 85
(b) How many variable s are required for a complet e descript
ion of the system
in the dynamic case after M 1 is released ?
(c) Select a set of coorJin ates and moment a and express
the total energy

function , the Hamilto nian, In tt:rrns of them. For convcni cnce,

aSSlIllIC that

the total potentia l cncrgy of the system is zero when the system
hangs at rest.

You will find it conveni ent to use more th.ln thc minimu m
numbcr of co

ordinate s and moment a in exrressi ng the Hamilto nian.

(d) By means of a Legendr e transfor mation, obtain the
Lagrang ian state
function in terms of your chosen generalized coordin ates and velocitie
s. Does
the Lagrang ian thus obtained equal the difference between the
kinetic and
potentia l energies of the system'? If so, why? If not, why not'!
(e) Recogni zing that the system is conserv ative. i.e.,
the Hamilto nian is a
constan t, obtain a differential equatio n for the height of mass M
2 above its rest
position . Is this a linear equatio n?
I-IS. The cross section of a cylindri cal solenoid used to control
the valve
mechan ism of a hydraul ic servo system is shown in Fig. IP-15.
When the
currents in the two identical coils I and 2 are equal, the plunger
is centered
horizon tally as shown in the figure. When the coil currents are
unbalan ced,

1-/3. The doubly cxcitcd magncti c fidd tmnsduc er shown in Fig.

I P-13 is
dcflned by the following volt-am pere relation shirs at its elcctric.1
I termina l
pairs:

!".

v,

. di
2m, - l

t',

d
.
-J [h(x)/I]

dl

tI

+ -dld

.
[h(x)/z]
. di2

+ 2c/ 2 -d I

a and c arc positive real constan ts, and b(x) is indepen

dent of either current

and is single-valued at each x.

-i I

u1t

..

It

2-

--

i2

Contains movable
parts
A lossless device

u2

Fig. 1..-13.
(a) Is magneti c coencrg y a state function of iI, i , and x?
Find this coenerg y.
2
(b) Does stored energy equal stored coenerg y? Is
the device electrically

...

linear?
(c) Obtain an expressi on for force at the mechani cal termina
ls in terms of the

variable s ii, i l , and x.

1.14. The system of masses and springs shown in Fig. IP-14

is connect ed

together by rigid rods attached by means of frictionless pivots.

Initially the

system hangs at rest under (he influence of gravity. The springs

and masses

,Ire such that the upper rod" initi.dly make an angle of 45" with
the overhea d

support .
The mass M2 is then raiseJ vertically a short distance
and released.
(a) What is the minimu m number of indepen dent generali
zed coordin ates
required to completely specify the state of the system before M 2
is released ?
''',')J

.i/'

Fnctlonless pivot

~t

:1:

~~
Fig. IP-14.

Cyhnd"ca l iron shell

:<>
NonmagnetIc
sleeves

XX!]X
~l~

xxx
~2~L /~
xxxx

xxxx

X means current
away irorn observer

. ..
~

Coils

-;r-

means current
toward observer

Fig.IP-I S.

the plunger moves horizon tally a distance x. The nonmag netic

sleeves in the
two end gaps keep the plunger centered radially. Their thicknes
s is g, and
their mean radius is r, where g ~ r. The coefficient of friction
is (t, and
friction is assumed to be viscous. The mass of the rlunger is M.
The srring
constan t of each spring is K. The permeab ility of all magneti
c mataial s is
assumed infinite, the gaps formed by the nonmag netic sleeves have
permeab ility
j1.o, and leakage and fringing may be neglected. The resistanc
e of each coil is
R, and the number of turns is N. The currents in coils I and 2 are
i I and ;2
respectively, and time-var ying.
Derive the three differen tial equation s of motion of the system,
i.e., the
equation s for the voltages at the terminal s of the two coils and for
the mechan ical
motion of the plunger.
1.16. Figure IP-16 is a schemat ic rerresen tation of a position
ing device.
The means of damping , represen ted by torsiona l dampin g coefficie
nt a, and

86

ELECTROMECHANICAL ENERGY CONVERSION

E . R

the restraining spring represented by torsional elastance K, are not shown.

Thc angular ovcrlap 80 of each stator pole face with thc rotor pole face is for
the central position of the rotor shown, and thc spring torque is zero for that
position (8 = 0).

.~.

'
,,',1

,~

,0"

Side view of
magnetic structure
[0

=constant

Fig.IP-J6.

The permeability of the stator and rotor magnetic material is assumed infinite,
fringing at the gaps may be neglected. and leakage fluxes may be neglected.
In tcrms of the given symbols, and for 18 1 < 80 :
(a) Write the torque equation.
(b) Write the circuit equation for the control circuit (containing the rotor coil).

";

Analytical Techniques for Treating

Electromechanical Equations of Motion

Including Typical Transducers

as Examples

".....

tt

.r

2.0

The determination of the equations of motion for an electromechanical

system is an important first step in system analysis, but a derivation of
these equations never completes a system study. The objective of any
system analysis is to determine the behavior of the system under a specified
set of excitations, e.g., the response of a generator to a step change in
load, the frequency response of a loudspeaker, the closing time of a relay,
or the response of a shipboard gun director to pitch and roll of the ship.
In order to reach this objective, the equations of motion of the system
must be solved, or at least the important characteristics of the system must
be obtained. This chapter will deal with the analytical techniques that
are useful in studying the equations of motion of connected electro
mechanical systems.
The differential equations of motion of an electromechanical system
will fall into one of three classes of equations:
1. linear with constant coefficients,
2. linear with time-varying coefficients.
3. nonlinear.
There are particular analytical techniques that are applicable to the
solution of each of the above classes of equations. For instance. if a
system is described by linear eq uations with constant coetlicicnts. general
analytical methods exist that give the response of the system to any
driving function. On the other hand, there are no general analytical

"

""'II'

, '''i';'''lf ~:'l'"

,'i'!If ._-,
.i;,'

,'~.

,'1

,:,.

"r'

~(jrt '~"l.

')

&1 e~
;
"\'

lilL

!l

[t. '

..

.'J!'1

Introduction

';,\:1'..

87

"

ELECTROMECHANICAL ENERGY CONVERSION

88

ANAL YTICAL TECHNIQUES

techniques for handling nonlinear equations. Consequently, a solution

can be ohtained only for a specified driving function, and orten, to solve
the nonlinear equations, recourse must be made to machine computation
or numerical methods.
In many cases in which a system is described by nonlinear equations,
valuahle information about the system can be obtained by linearizing the
equations. Linearization of the equations will not yield solutions that
are as accurate as solutions ohtained with the original nonlinear equations.
Ilowever. approximate linear solutions arc easily obtained alld they usually
give indic;\ti"lls of system performance (".g .. relative stahility at a parti
cular operating point) thaI arc ditllcultlo ohtain wilh nonlinear e:ljll:.ltions.
The usc of linearizing techniques depends on the total information and
accuracy obtainahle from the linearized equations. The gain in the ease
of solution hy linearinltion must be weighed against the loss of "~curacy
and the dilliculties encountered in solving the nonlinear equations.
This treatment of analytical techniques will consider those methods

commonly used with e:ach of the three classes of equations listed above.

Techniques for simplifying the equations by linear approxim:ltions will

be given, and the analytical methods introduced will be illustrated by

analyzing electromechanical transducers.

2.1

Techniques for Studying Linear Differential Equations

with Constant Coefficients

Whenever a system is described by linear integrodifferential equations

with constant coefficients, the general behavior of the system can be
determined. One of the most important characteristics of linear integro
differential equations, the additive property which is common to linear
functions, allows the addition of functions which are solutions of integro
differential equations to obtain new solutions. This is commonly called
the principle 0/ superposition, and the ability to use superposition with
linear integrodifTerential equations yields many valuable and general
properties of linear equations.
The general techniques for treat'ing linear integrodifTerential equations
are well developed. There are four well-known approaches: classical
method, Cauchy-Heaviside operational method, Fourier transformation
method, and Laplace transformation method.* The classical method of
treating linear ditrcrential equations is the one most widely included in
standard mathematical texts on dilferential equations. The other three
for a discussion of all four methods, see M. F. Gardner and J. L. Barnes,

in Linear SyJl<'ms, Vol. 1, John Wilc:y, New York, 1'.142.

Transients

'~~r

'~

89

methods, Cauehy-Heaviside, Fourier, and Laplace, arc similar. hut it is

generally agreed that, of the operational methods, the Laplace Ira nsform
method is the most general and can be extcndetltto include th':.lIhcr !wo.
There is some question as to whether a sin81.e-sided or d~J hie-sided
Laplace transform should be used, but the authors feel thtl! such a
discussion is beyond the needs of this text. The review of sill utions of
linear integrodifferential equations presented here will be limited to
classical met hods und the wcll-k nown si ngle-sided I.a plllcc I r;insfor Ill.
In the study of linear systems there are many properties Ihat may be'
desired. and they can 1I11 be ddermined from the: system intcgrodilrcrcllti,t1
equations and initial conditions. For example, the desired characteri~tic
may be the transient response to various system inputs. If the inputs are
easily described (such as impulses, steps, and ramj;s), solutions can be
obtained by direct solution of the dilfcrcnt~t! equations or by the usc of
operational techniques like l.aplace transforms or I kllvisid\: npnalors.
If the information desired about the system is a steady-state frequency
response, this can also be obtained by direct solution of the differential
equations or by the use of operational techniques. Should the stability
of the system be of intcrcs,t, there are several methods available for
investigating this stability. One approach is to use the frequency response
of the system and a Nyquist diagram.
In all the above cases where a solution of linear differential equations
with constant coefficients is desired, the important item is really the
characteristic equation or behavior of the system. The method used to
determine the characteristic behavior is mostly a matter of choice, and
alternative methods entail approximately the same amount of work.
When the general characteristics of a system are studied, it is often
helpful to use a block diagram representatio~ of the system. The block
diagram has two advantages: first, it places in .evidence the severar.com
ponent parts of a system, and second, it can o~ reduced in simple steps to
yield a solution to a particulai problem. This. is especially helpful when
there are several inputs and outputs at different points in the system.
In the sections that follow, consideration will be given to the techniques.
mentioned above. Useful areas of application of eaeh method will be
discussed and examples of the application of each t~chnique given.

2.1.' Classical Solution

A linear differential equation with constant coefficients can always be
solved without recourse to operator techniques. For the general nth
order linear equation
d"x
dn-1x
-+al--+
dIn
d/ II - 1

... +

an-I -dx

+ a,.x = I( I)

d/ :~

(2-1)

!.

90

ELECTROMECHANICAL ENERGY CONVERSION

ANALYTICAL TECHNIQUES

with ai' . . . , an constants, the solution can be obtained in the following

three steps.* FIrst, the solution of the homogeneous equation
tiN.\"

lIN
I

+ 1

-,-I + ... +
dn-Ix
l

t"-

tlx
a"'_1 -d
I

+ a,,;\" = 0

...

~.,,\

sum of the orders of the several equations. Nonetheless, it is always

true that a solution of linear differential equations with constant co
efficients which satisfies both the equations and the boundary conditil'ns
is {he correct solution.

(2-2)

is found by assuming solutions of the form

EXAMPLE 2EI

X=~

O~

Substitution of Eq. 2-3 into Eq. 2-2 yields the characteristic equation
It

r"

+ al,,,-l + ... + a,,-I' + a"

= 0

:,

I:,,

(2-4)

for which the n roots 'I, .... r" are found. These roots arc then used
with Eq. 2-3 to form. the homogeneous (sometimes called complementary)
solutiont
XH(t) = cieri'

+ c2er~1 + ... +

c"er,,' =

2:"

clert'

xp(t) =

2:

cjert' + xp(t)

Consider the singly excited magnetic field transducer shown schematic

ally in Fig. 2EI-1. First obtain the lumped circuit parameters in an
approximate way, then derive the equations of motion. After linearizing
the equations of motion by assuming small displacements, find the
respom,:; of the transducer to a stcp of input voltage.
~

')

'-.

':' "'t'

,;.-!'.

The second step in the solution is to find a particular solution. xp(t);'

which satisfies Eq. 2-1. This solution for most practical driving functions
/(/) can be found by the method of undetermined eoefficients.~ Then
the general solution for Eq. 2-1 is

, '

(2-5)

I-I

x(t) = XH(t)

91

.'

II
I

~."...' . ".""
~'"
'

(2-6)

I-I

The third step in the solution is to specialize the constants CI to make the
solution fit the initial conditions of the problem.
The essential point to remember about the solution of lint:ar differential
eq'uations with constant coefficients is that a general solution which
satisfies the differential equation and which contains a number of arbitrary
constants equal to the degree of the equation is the most general solution.
As a result. the solution for any set of initial conditions can be obtained
by specializing these arbitrary constants.
In the solution of simultaneous linear dilferential equations with constant
cocfficicnJ.s, the comment abovc about uniqueness still holds. but the total
number of arbitrary constants in the solution may be smaller than the

- .. ~

~QUlllb.numposltlon.t
~ /':

:'.,'
::;r'"" .
-J

...

"'1''.

~y?;: .

Fig.2EI-I.

u , POSItion
where spring force
IS zero

plunRer,

Singly excited magnetic field transducer.

:1' ...
::,,>,

~Ij

;\~r
"':1"-.,,

".

i.:,':

See F. B. Hildebrand, Advanced Calculus for Engineers. Prentice-Hall. New York.

1949. Chap. I.
t If the characterislic equation has repeated rools, the homogeneous solution will
contain lerms of the form Ie ',' .. Ik-le'j'. where k is the number of times the root
rj is rcreated. See Hildebrand. 1oc. cil.
: Hildebrand. loc. cit.

~':DI~t

'l'i'

,'~-.,
"
_t.

::}:::r;;.'.1
:~ ,;:~

/:'

-,#

r.,::

,.''''

For the transducer of Fig. 2EI-l the resistance R includes the winding
resistance; thus. in accordance with the methods of Chap. I the lossy
part of the electrical system is separated from the coupling field. The
plunger (mass M) is constrained to move in the x direction only; the air
gaps d remain fixed. Thc motion of the plunger is restricted so that
o < x < t. The iron has a rectangular cross section. The coupling field
is magnetic; to find the coupling forces, the inductance L must be expressed
as a function of the independent variables of the system. Assume that
the iron has infinite permeability. Then the inductance L is independent

.,

,.

of current i.

The equilibrium equation for the mechanical system can be obtained

from d'Alembert's principle by including the mechanical force of electrical
origin {Eq. 2EI-4) as an additional applied force. Thus,

13y neglecting fringing at the air gaps d and x, the inductance.

becomes

N2A c fLO
d + x

L =

(2EI-I)

The next step in the study of the magnetic field transducer is to obtain
the equations of motion. To do this use Kirchhoff's voltage law for the
ekctric circuit and d'Alembert's principk for the mechanical system, and
to completc the equations evaluate the mechanical force of electrical origin
by the method of arbitrary displacement as given in Chap. 1. In writing
the loop equation for the electric circuit, the current i is the electrical
variable of interest; therefore, in computing the mechanical force of
electrical origin. select the current i as the independent variable. From
Table I-I or Chap. I, if ; is an independent variable, the for\...\... can be
evaluated from the magnetic coenergy W;" as:

[, =

oW' (.

I, X

ox

d 2x
I(t) = M dt2
where M

.
.

cis

d)"
dt

K(D - x)

-I.e

(2EI 7)

f"nctlOna I r,orce

According to Eq. 2EI-I, the inductance L is a function of x only, and

Eq. 2EI-S can be substituted into Eq. 2EI-6 to yield the electric circuit
equation:
. dL dx
L di
e()
(2EI-H)
t = R 1+ -+/-
dt
dx dt
Equation 2EI-4 can be substituted into Eq. 2EI-7 to yield the mechanical
equilibrium equation:

= Mddt-x2 +
2

1(1)

dx
dt

(X-

K(x - D) -

1. dL
2 dx

_/2

(2EI-9)

"f

ff
(2EI-4)

~l't,
':f~

tp:

Equations 2E 1-8 and 2E 1-9 are the equations of motion for the system.
Tn general the responses ;(1) and X(I) resulting from the application of the
driving functions e(/) and f{t) are the desired quantities. The equations
are nonlinear, and solutions may be difficult or impossible to obtain by
analytical methods. In many cases valuable information can be obtained
about a system by solving the linearized equations of motion.
Equations 2EI-8 and 2EI9 contain nonlinearities of the product type
[e.g., ;2, ;(dxjdl)] and in the variation of L with x (see Eq. 2EI-I). To
remove these nonlinearities, assume small changes about an operating
point, and rewrite the driving functions and variables as
e(t) = Eo + et(t)
f(t) = Fo + 11(t)

Now that A and L are defined, the Kirchhoff voltage equation for the
e(t) = ~;

= inertial force of mass

(2EI-3)

This gives the applied mechanical force developed by magnetic field

coupling. As shown in Chap. I, this force is valid regardless of the
trajectory followed by the system (i.e., valid for any variation of the current
i that does not violate the internal constraint).
To complete the equations of motion, the constraint between A and i is
needed. The nux linkage A of the electric circuit is given simply by
,. "tA
(2EI-5)
".
A = Li

electric circuit is

dx
dt

(X -

:;.":

/;,+

+1;2 dL

K(D - x) = spring force applied to mass M, where D is equilibrium

length of the spring
I. = magnetic field coupling force applied to mass M

(2EI-2)

The use of Eq. 2EI-3 in 2EI-2 leads to the result

f. ...

(/

({

coenergy is given simply bF.,. .

= tL;2

d12~

dx, =
a -

The transduccr was assumed to have no saturation in the magnetic

circuit; consequently, the system is electrically linear, and the magnetic

W;" = Wm

93

ANALYTICAL TECHNIQUES

ELECTROMECHANICAL ENERGY CONVERSION

92

;(t)

= 10 +

it(t)
x(t) = X o + Xt(t)

(2EI-6)

','i.

(2EI-IO)

*oiii,

For magnetic circuit calculations, see, e.g., Magnetic Circuits and Transform~rS,
MIT Electrical Engineering Staff, John Wiley, New York, 1943.

',;

where the capital letters indicate the operating points and the time
functions (ej, II> il> and XI) represent the small variations from the
"

94

ELECT ROMEC HANICA L ENERGY CONVE RSION

operati ng points.

In terms of x as given by Eqs. 2EI-I0, the inducta nce is

N2Ad.LO

=d+

Xo +

(2EI-II )

XI

which can be rewritten as

t,

Eo

!..--)

(2EI-12 )

+ d + Xo

N2A ciJ.o
L o = d + Xo

XI)2 L=Lo l - - Xl
- + ([
d + Xo
d + Xo

( -XI)3
-

X o

+ ...]

~
~\

(2EI-14 )

"',j

L = L o( 1 - d
Lo

/1 =

.'

!i'\

.~

i'

),:'
~,

XI

j ,

.~

I':I,~,i,i
~~;

:.~.t.
.~ ~ .~.

+ Xo

',) :?/'..
~

1-\

"

RIo + Ril(t) + L o dil(t) _ 10L o dXI(t)

dt
d + X o ' dt

,"

(2EI-16 )

+ /I(t)

= M

d 2xl(t)

~
1

dXl(t)
+ ex ----cit
+ K(Xo -

15Lo

+ 2d+

loLo.

+ d+

D)

., ~\",

"XI(t)

1.

t,'

1".

.~ .

i/",

+ KXI(t)

IJL o

XO'I(t) - ' J .

~)

(2EI-17 )

':1

:1",

~j~

~:;i .

i.

g;.

. ~hA\'
''P .,,~

d 2x I
M dt 2

dXI

+ ex di +

K - (d

I~LO]

)2 XI
O

/oL o .
+ X o I.

(2EI-21 )

Before Eqs. 2EI-20 and 2EI-21 can be applied to the study of

the
dynamics of the transdu cer, it must be determi ned that a steadystate
operati ng point exists. In other words, th~re must exist a value
of Xo
which lies in the range 0 < X o < / and which satisfies Eq.
2EI-19.
Whethe r or not such a value of X exists can be determi ned by
o
solving
Eqs. 2EI-18 and 2EI-19 for specified Eo and F . Exampl es of the
types
o
of conditi ons that can exist are shown in Fig. 2E 1-2 where it has
been
assume d that F o = 0, and therefo re X must satisfy the equatio n
o

"I~' .

!t .,..

Fo

(2EI-20 )

'

., .If'

(2EI-19 )

di l
/oL o dXI
= Ri j + L o dt - d + X dt

.:~.

:1 xJ

+ Xo -

1 15Lo

+ '2 d + X
o

,\

.*)

Substit uting into Eqs. 2EI-8 and 2EI-9 from 2EI-10 and 2EI-15
and
neglecting all product s of small variatio ns (such as i~ and ilxl),
the
linearized equatio ns of motion are

+ el(t)

(2EI-18 )

'.y'

(2El-1S )
= d

el

"

,(

With small variatio ns in X, xd(d + X o) ~ 1, all terms of second order

and
higher in Land oL/oxI can be neglected, yielding

Eo

RIo

In spite of the fact that these steady- state operati ng point relation
s were
obtaine d from the linearized equatio ns, they are exact and can
also be
derived from the origina l nonline ar equatio ns (see Eqs. 2EI-8 and
2EI-9).
The stea:iy-state relation s, Eqs. 2EI-18 and 2EI-19, can now be subtrac
ted
from Eqs. 2EI-16 and 2EI-17 to obtain the linear differential equatio
ns
with constan t coefficients that describ e the increme ntal behavio
r of the
system around the steady- state operati ng point:

(2EI-13 )

For small values of XI the induct~nce can be expressed in a Taylor

series
expansi on around X = d + X , or it can be expand ed by the binomi
o
al
theorem . Both processes yield the result

oL

Fo = K(Xo - D)

XI

where

OXI

95

The steady- state equatio ns describing the operati ng point can be obtaine
d
directly from Eqs. 2EI-16 and 2EI-17 by setting all the increme
ntal
variatio ns (el. II> Xl' and il) equal to zero. This yields

and

L~ L'(1
.

ANALY TICAL TECHN IQUES

K(D - X o) =

1 15Lo
X

2d +

(2EI-22)

The two terms in Eq. 2EI-22 are plotted in Fig. 2EI-2, and the points
of
intersection A, B, and C are possible equilibr ium points. The
fact that
point A is an unstabl e equilib rium point and Band C are
stable
equilibr ium points can be determi ned from Fig. 2EI-2 or from a
solution
of the increme ntal equatio ns of motion , Eqs. 2EI-20 and 2E 1-21.
Note
that in Fig. 2EI-2 the curve for /01 yields no equilibr ium point in the
range
o < X o < /; consequently, for this set of conditions and Fo = 0, the

.,

96

ELECTROMECHANICAL ENERGY CONVERSION

incremental equations of motion as well as the steady-state equations are

meaningless.
To continue the study of the dynamics of the transduc~r, assume that an
equilibrium point exists, and find the incremental behavior of the trans
ducer by the lise of Eqs. 2EI-20 and 2EI-21. Refore solving a transient

,I
~
:~I

then

L n( I

~I

1
,

Force
1

'li',I'11

,. i" +

l u,

'\0

> I n2 > 1m

97

ANAL YTICAL TECHNIQUES

tf,~
~:

;.!

"

+I 0.2 )

O.833L o

Lo( I - 0.2)

0.800L o

and

which gives an error of 4 per cent in the inductance.

Now lise Eqs. 2EI-20 and 2EI-21 to obtain a solution for the res[lonse
in mechanical displacement x to a step of applied voltage with no external
force applied (/ = 0). The step of voltage from En to /;'0 + 1:", oC':lIrs al
I = 0 with the condition that before the step is applied the system is in
equilibrium (il = 0, di,/dl = 0, Xl = 0, dXI/dl = 0, d 2 xddl 2 = 0).
Thus the problem is described by the incremental equations

~,

1 = Ri l

eli,

+ Lo cll

~ clXI

- cI

+ Xo

(2EI-24)

dl

"J

. J\~,

1'0= 0

iY'
,I,

;:

01.,-1

d2X,
O= M
dl 2

t~

Fig.2EI-2.

1 = _ M(d

u - --- I ----

Steady-state equilibrium points.

..

1 XI )
d + Xo

~ Lo(l

d: xJ
1

'f._:
':(.
"~\

(2EI-23)

and

;2

= (10 +

For instance, if the total swing in

XI

Xlmax

i l )2

~ I~( 1 + 2 fa)

is such that

+ Xo

+ [K -

(d

15LO ,x,
+ -/nLo
.
]
--I,

X u)

"

_ [RM(d

dl 3

Ra(d + X o)
loL o

+ Xo) + a(d + X o)] d2xl

loL o

10

dt 2

+ K(d + X o)] dXl _ R(d + XO)[K

10

(d

10Lo

dt

f~Lo

X O)2 XI

(2EI-26)

In the solution of Eq. 2El-26 a homogeneous solution is found by

setting 1 = 0 and assuming
(2El-27)
Xl = e"
The substitution of Eq. 2EI-27 into Eq. 2EI-26 (with 1 = 0) and
cancellation of the exponential factors lead to the characteristic equation:

M(d

+ Xo)

10
.";'f

(2EI-25)

+ Xo

1'1 .
"t4 '

~ 1.,( 1

dl

+ Xo) dJxl
10

problem, consider the accuracy of the linearizing approximations that

have been made. This accuracy can be determined from the expressions

dXI

1:(-

with all initial conuitions zcro. To find ,\"1(1) climinatc i l from till' two
equations and obtain an equation containing x, only

C --------

r +

[RM(d + Xo)
loLo

Ra(d + Xo)
10Lo

K(d

a(d + X o)] 2
10
r

+ X o)] r+ R(d + Xo)[x 10Lo

10

= 0.2

I~LO]
(d

+ X o)2

(2El-28)

</'.

98

ELECTROMECHANICAL ENERGY CONVERSION

ANALYTICAL TECHNIQUES

This is a third-order polynomial in " and it has three roots. One of the
roots must be real, and the other two can be complex conjugates. The
general homogeneous solution of Eq. 2El-28 is of the form

Simultaneous solution of Eqs. 2EI-32, 2EI-33, and 2EI-34 for the

constants yields

XIIf(I) =

Cle'll

+ C2

e'2 1

(2EI-29)

CJ(,'JI

C =

[('I - 'J)('I
'2 '3]
-

2
C =

[('2 - (3)('2
,I'3]
'1)

(2)

.,

....,~'r..:'..

R(d

+ X o)

R(d

10L oE I

'I 2 ]

R(d

+ X o) [ K -

(d

(2EI-30)

I~Lo]
+ X o)2

'~4

!,.~"~.:
".

~
':,- .ii,

\.

~'.

,.

;'"

Xl(t)

10LoE

+ X )[K _

I~Lo ]
(d + Xo)2

R(d

+ X o) [ K

x [

- (d

'2'3e'11
'1'3 er2'
'lf2er),]
- ('1 - f3)('1 - r2) - ('2 - (3)('2 - 'I) - ('3 - (2)('3 - rl)

(2EI-38)

~i':

~i'

When two of the roots ('1, '2, r3) of Eq. 2EI-28 are complex conjugates,

the two corresponding terms in Eqs. 21-38 can be combined to give an

exponentially damped sinusoid. In such a case the response might appear
as shown in Fig. 2EI-3, in which the total response x(t) = Xo + x\(t) is
given.

~~

.:;~.

Clerl l

+ C2 er 1 + C 3er )1
2

10LoEI
R(d

+ X o) [ K

- (d

."
)
(2EI-31

10L o
+X

)2
O

(2EI-37)

I~Lo]
+ X o)2

~~:'"

Xl ( / )

(2EI-36)

loL oE 1

': .1

Addition of Eqs. 2EI-29 and 2EI-30 yields the complete general solution

(2EI-35) ,

The substitution of Eqs. 2EI-35. 2EI-36, and 2EI-37 into Eq. 2E1-31
yields the complete solution

.."

)2

'5 o ]
+ X o) [
K - (d + LX )2
o

XJP

10LoE I

r
C _ [
l - ('l - (2)('3 - 'I) R(d

+ X o)2

in which case one of the roots ('1> 'l' or (3) will be positive, giving an
exponentially increasing and therefore unstable response. This is the
type of step response that would be obtained at the unstable equilibrium
point A in Fig. 2EI-2.
A particular solution of Eq. 2EI-26 can be found by inspection to be

K - (d

I~Lo
K < (d

[ '+~Lo

'oLoE I

!I,

where 1'1> '2' and I'J arc the roots of Eq. 2EI-28* and Ct> C l , and C 3 are
constants to be determined from initial conditions. From an inspection
of the last term in Eq. 2EI-28 it is evident that this term can be negative if

99 ,

hl
t.
11

)~

~~~

'i

4r

The constants can be evaluated from the initial conditions.

assumed to be zero: thus

Xl(O)

1
( dX )

dt

1-0

( dd/XI)
2 ,-0

=0
=

= C

0 =

_0

R(d

(2EI-32)

1.::
.~-j

2]

'oLa
o
X )[ K - (d + X
o

)2

'.

(2EI-33)

'~Cl + '~C2 + ,~CJ

(2EI-34)

CI

'2 C 2

Gardner and Barnes. op. <'il., pp. 164-165.

,,~

'.;

. ,~1

It

TOLOE]
RI5LO

I
RK(d + X o' - d+X~
...---~-----

)11',

:i'

+ '3C3

'1

+ Cl

I LoE I

'r,,..

These were

.~

;-~~;I

"
OI~O-

Fig.2EI-J. Typical response of transducer to step

change in voltage.

~~.
.~

~, ;~.

;j~"~ ..,'

~.

,:1

-----------"'"

'

100

.~,

:I
;

ELE CTROMECHANICAL ENERGY CONVERSION

The fact that the response of Eq. 2E 1-38 contains three exponential
terms results from the presence in the system of three independent energy
storage eJel)1cnts. namely the mass M, the spring K, and the magnetic
field. Insofar as the response is concerned, there is no mathematical
difference between mechanical energy storage and electrical energy storage.
This is simply a manifestation of the fact that it is possiblc to ddcrmine
mathematically difTerent physical systems in the same terms as discussed
in Chap. I.
An inspection of the characteristic equation Eq. 2EI-28 shows that the
codlicients of the equations. and therefore the roots. arc I"unctions of
the steady-statc opcrallng point; conse4uently, the shape of thc transient
response (Fig. 2El-:l) will depend on the operating point (Xo. 10 '
This
variation of the response with operating point is a result of the non
linearity of the system.
1f the current that flows during this transient response is desired. it can
be found by a simultaneous solution of Eqs. 2EI-24 and 2EI-25 using the
same method that was employed in finding the displacemcnt, or the result
of Eq. 2EI-38 can be substituted into Eq. 2EI-25 and the resulting equation
can be solved directly.

2.1.2

'1

Solution by Laplace Transforms

F(s) =

fooo J(t)r st dt

(I) 2'[f(t)

f:
~

-.:.'

'~ ,~

.,

2'[c /(/)

(3) 2'[/1(/)

==

J~ !(t)e-" til

c F(s), where c is

==

+ 12(/) ==

(4)

2'[(~~/)]

(5)

2'[tl2~?)]

'i'~

i~~

~'

'

(2)

= F(s)

Basic Laplace Transform Theorems

F1(s)

,I

constant

+ F 2(s)

= s F(s) - /(0 +), where/ (0+) is the value of /(1) at

I ==

"
~

. ~

S2 F(s) -

s/(O+) - /,(0+). where /,(0+) is the value of

df/til at I = 0+

(6) 2'[f f(t) til] =

F(s)

+ [f /(t)

dtll

1'-0+'

where [ff(/)d/),_o+ is the value of

ff(/) til at I = 0+
(7) limf(t) = lim s F(s)

[Final Value Theorem]

(8) lim/ (I) = lim s F(s)

(Initial Value Theorem]

l~<:o

1--0

$-.0

s__ co

(9) 2"(/1(/) 12{t)) ;f; F1(s) F 2(s)

Y,J

(2-7)

where .I' is a complex variable (.I' = (J + jw). Whether or not a transform

exists for a given time function can be determined in a straightforward
manner.* Equation 2-7 is valid if the integral exists, and for most
practi caJ time functions the integral does exist. A detailed examination
of functions transformable by the Laplace transform is left to other
references ....
The Laplace transform is a linear operator which converts time functions
to functions of a complex frequency variable. Taking the Laplace

* See. e.g., Hildebrand. op. cit., Chap. IT, or Gardner and Barnes, loc. cit.

transform of a linear differential equation transfers the problem from the

real time domain [0 the compli.:x frequency domain. It should be nOled
that the Laplace transform is delined as one-sided, Le., only time functions
which are zero for I < 0 can be complctcly n:prcsenled by a Laplace
transform. This limitation is not severe, since most of the problems of
interest are transient problems where the input functions exhibit dis
continuities at I = 0 and the effect of these discontinuities on the output
is sought.

TABLE 2-1.

,',

101

Work with Laplace transforms is considerably simplified by the usc of

general theorems. Some of the important theorems arc givcn in Tahle
2-1. As an example of the use of the first five theorems of Table 2-J

>.

Linear differential equations with constant coefficients can be solved

also by the usc of Laplace transforms. The Laplace transform approach
consists essentially of changing differential equations to algebraic equations
for solution and transforming the results back into time functions.
Before solving a set of equations by this method, a review of some of the
more important properties of Laplace transforms is in order.
The Laplace transform F(s) of a time function J(t) i~ defined as

ANALYTICAL TECHNIQUES

\.i
~~t:
,;- '~',

l
~

,~

r
~),
,

consider the equation

d 2x

di2 +

dx
dl

02 X

aJfx dt = f(t)

(2-8)

With the assumption of zero initial conditions the first five theorems of
Table 2-1 can be used to transform Eq. 2-8 to the algebraic eq uation
.1'2

':,

Ql

Xes)

ats X(s) ,..

a2

Xes)

X(s)

OJ -

.I'

"\
F(s}

(2-9)

When a solution of a diffen:ntial equation (Eq. 2-8, for example) is

desired for a particular driving fU,nction!(l), the transform of the solution

",'

~,;

102

'J:'
,,;',,"

ELECTROMECHANICAL ENERGY CONVERSION

is obtained by solving the' transformed equation.

solution of Eq. 2-9 for the unknown yields
.I'

X(s) = sJ

ANALYTICAL TECHNIQUES

For example, the

After this is accomplished, the second theorem of Table 2-1 can be used
with the first transform pair of Table 2-2 to obtain the desired time function
as a sum of exponential factors. This procedure will bc illuWated by
applying it to Eg. 2-10. Assume that the driving function l(t) in Eq. 2-8
is a unit impulse; Eg. 2-10 becomes

(2- I0)

F(.I')

+ al s2 + 02'\' + a3

":.

In order to complete the solution, the transform F(s) of the driving

function((t) must be substituted into Eq. 2-10, and the inverse transform
of the result must be found. The inverse transform can be found by
direct integration; however, the most useful method is a table of transform
pairs in whieh time functions and their transforms are tabulated. The use
of such a table entails finding a ti me function which corresponds to the
transform of the unknown function.
Several important transform pairs are given in Table 2-2. Actually,
pairs 2-6. in addition to many others not listed, can be derived from the
first pai r. Thus the first pair is extremely important.

X(s) _
The three roots

Function

e-<I'

a real or
complex

unit step

uo(t)

sin

cos wt

wI

unit impulse

+ Q\S2 + a2s +

.I'

.1'1)(.1' -

(.I' -

.1'1)(.1' -

.1'2)(.1' -

.1'3)

A =
:.~

j'l,

+ wI

(2-12)

=.I' - +.I' - +.I' - .1'1

- S2
Sj

(2-13)

:lj

r'\)
d

.:.1

,,:~,

.1'1

.,

,to

Sj)

= (.1'2

S2

.1'\

)(.1'2

.1'3

(2-14)

.~.

,-.::;
;",f".

.,"

S2)(.~t -

.
.H,

'~"<"

'I..

c=

s~
(.1'3 -

.1'\)(.1'3 -

.1'2)

When the first transform pair of Table 2-2 is used with Eq. 2-13, the
resulting time function is

s
$2

.1'3)

When the right-hand side of Eg. 2-13 is reduced to thl: saml: common
denominator as the left-hand side, the coerlicicnts of like powers of .1'
ean be equated on both sides of the equation to yield the coeflicientst

1
w
s2 + wI

.1'2)(.1' -

ABC

.I'

(.I' -

'J

(.1'1 -

(2-11 )

Now assume a partial fraction expansion of the form

s+a

SZ

OJ

of the cubic in the denominator are found* and

X(s)

Transform

unit ramp

(Sb .1'2 .1'3)

.1'3

Eq. 2-11 can be written in the form

$J

Transform Pairs

.I'

.~<'

TABLE 2-1.

103

'

investigation of Eqs. 2-8 and 2-9 shows that, when linear differential
equations with constant coefficients are transformed, the coefficients of
the transformed variables are polynomials in s. Furthermore, inspection',
of Table 2-2 shows that the usual driving functions have transforms that,
a(e: at worst, ratios of polynomials in s. Consequently, the transform
of an unknown will usually be a ratio of polynomials in s. When this
occurs, the denominator polynomial can be factored into first-order
factors, and a partial fraction expansion can be used to change the
transform into a sum of transforms, each having a first-order denominator.

1~iI:'

x(t) = AeJt'

(2-15)

where A, B, and C are given by Eqs. 2-14.

The solution of simultaneous differential equations with constant
coefficients is accomplished in the same manner as the solution of a single

,J
:~~;

For polynomials of higher than second order lrial-and-error l1Iethods of finding

the roots arc used. See, c.g., S, Lin, .. Method of Successive Approximations of
Evaluating thc Real and Complex Roots of Cubic and Higher Order Equations,"
Jour. Math. aod Phy.\ics, Vol. 20. No.3. Aug. 1941.
t For the expansion with repeated roots and with other complications see Gardner
and Barnes, op. cit., Chap. VI.

;~.

,
.\

.",

~, .\., I

.,,<~Jl

+ BeJl' + CeJ)'

. .'/<'.

"

-.,~."

104

ELECTROMECHANICAL ENERGY CONVERSION

ANAL niCAl TECHNIQUES

equa!ion. I n summary, this method consists of the following steps:

(J) all differential equations are transformed with due regard for initial
conditions; (2) the resulting algebraic equations are solved for the trans
form of the unknown function; (J) the transforms of all given (driving)
functions arc inserted in the result; (4) the unknown time function is
found by obtaining the inverse transform of the result, usually by first
expanding the transform of the unknown function in a partial fraction
expansion and then obtaining the time function as a sum of exponentials.

In Chap. I the mechanical rorce due to ekctric nell! coupling was derived
using the method of arbitrary di~placement with conservation of cncrgy,
The applicable expression for the force is given by Table 1-2 as
:{
:f~,

(, = ,

j,

ox

1 q2

and using Eq. 2E2-1 for the c~pacitance

This method of solution will be illustrated by considering the electric

field transducer of Fig. 2E2-1. The lower plate of the capacitor is fixed,

.~
!2'j

Mechanical
dampmg (J

t
I

I'

ApplIed l1I~cham(.1
force

"c

Plate area A p
i

~~

ira

Electric field transducer.

e(1)

~
,
.3
tl~~'~ "t" 1J:~

oA p
x

~"
.f'f

f~

...

and the upper plate is constrained to move in the x direction only, The
spring K has the equilibrium length D in the absence of electrical and
mechanical excitations.
The electromechanical coupling field in the transducer of Fig. 2E2-1 is
electric. and it can be characterized by a lumped capacitance C.
Negkcting fringing fields, the capacitance can be written as

c=

(21:2-4)

e gives
q2

2 oA p

(2E2-5)

=:!. = qx
("

~r
~'
"it
f~/

r
t

(2E2-1)

rhe charge q on the capacitor will be chosen as the independent

electrical variable, and the stored electrical energy expressed in terms of
q is
1 q2
(2E2-2)
We = 2C

(2F.2-6)

<oA,.

:t."

(;a

~
)))))))///7//// '.\

EQuIlibrium POSItion 01 plate.

t.t?, pOSition INhere spring

NjIJ"

(Ie

iJx

By using the KirchhotT voltage law with the current i ~- dq!dl, the

equation of motion ror the electric circuit is

. ;',

force IS zero

Fig, 2E2-I,

= -

dq
Rdl

,1

r- W..0//",n"nnn)))n
'\

"

The only othl'r informution necessary to complete the equations of

motion of the transducer'is the voltage I~'c across tht: capacitor, whieh is

//,

Ie

f(1/

(2E2-3)

Sub~titution of Eq. 2E2-2 into 2E23 yields

EXAMPLE 22

,~

W,(q, x)

,rr = + 2: C

105

/I'

+ qx

(2E2-7)

EoA p

The equilibrium equation for the mechanical system obtained from

d'Alcmberr's principle including (,(Eq, 2E2-5) as an additionalmcchanical ~
force is
j '( I )

2
d,
x +
M-

tll-

tl.r

(l -

til

, - D) + _
I q2
+ 1\(.\.
_

2 'oA p

(2E2-8)

Equations 2E2-7 and 2E2-8 are the complete nonlinear equations of

motion for the transducer of Fig. 2E2-1. The object of the present
investigation is to obtain the response to a small disturbance. This can
be accomplished hy linearizing Eqs. 2E2-7 and 2E2-8 about an operating
point and by applying Laplace transform methods, Before linearizing
Eqs. 2E2-7 and 2E2-ti a check must be made to bc sure that an equilibrium -:
point c!\ists. To do this assumc an equilibrium point deflm:d by the
constants Eo, Fo, Qo, and X o., If the wstem is in equilibrium, all
derivatives are zero, and Eqs. 2E2-7 and 2E2-8 reduce to

L" .

1''1'>""

QoXd
Eo - -oA
-.
p,
.
1 Q~
Fo = K(X'o - D) + 2- -A

'.'~'

t;
~'

It

EO

(2E2-9

(2E2-1O)

106

ELECTROMECHANICAL ENERGY CONVERSION

Solution of Eq. 2E2-'1 for Qo and substitution of the result in Eq. 2E2-1O
yield

FI)
Rearranging Ihis

= K(Xo _

e.~ pres~ion

D)

+ ~ EJ{oA"
2
2

gives"

KXU = F0

E2( A

1
+ KD --~
2 X2

(2E2-12)

From Eq. 2E2-12 it is evident that there are some values of the parameters
and variables for which no X o > 0 is possible. Whether or not an
equilibrium point exists can be determined by plotting the two sides of
the equation

Fo + K(D - X o) =

E5 (oA

2 ~p

(2E2-13)

as functions of X o. This is done for a typical set of constants in Fig.

2E2-2, in which the right-hand side of Eq. 2E2-13 is plotted for three
values of Eo. Note that for voltage E Ol there is no equilibrium point;

...

;~.

,?

:1',""",",\,'

.' "'' ', ." ~\

"r

e(t) = Eo

;~';j

~::iit.1

:.

(2E2-14)

Substituting the quantities from Eq. 2E2-14 into Eqs. 2E2-7 and 2E2-8
and negJecting products of small quantities (qlxl and q;) lead to the
following linearized equations of motion;
Eo

tr

+ el(t)

J(t) = F o + JI(t)
q(t) = Qo + qj(t)
x(t) = X o + Xl(t)

/'f;'

.)00.

> E02 > E03

'.~'

:j;, }'

Force

E Ol

107

for voltage E 02 there is one equilibrium point A; and for voltage Eo} there
are two equilibrium points Band C. For the voltage E 01 then. the
equiJibrium equations are meaningless, and linearized incremental
equations of motion arc not possible. For voltages E02 and Eu.\ the
equations of motion can be linearized since equilibrium points A. B. and
C do exist. It should be evident from Fig. 2E2-2 that since point A
OCcurs at a point of tangency it will be an unstable point. Likewise. the
reJative slopes of the curves at point B show that it also is an unstable
point. Point C. on the other hand, is stable. Whether or not an
equilibrium point is stable can be determined either from a graph such as
Fig. 2E2-2 or from the linearized incremental equations of motion.
In this problem the response of the system to small amplitude excitations
around an operating point is desired. Equations 2E2-7 and 2E2-8 contain
only product nonlinearities (q2 and qx). and linearization can be effected
by assuming the driving forces and variables to be composed of constants
pJus smaii time variations, thus:

(2E2-11)

Xo

ANALYTICAL TECHNIQUES

el(t) = R dqt(t)
dt

QoXo + X o ql(t)
(oA p
(oA p

Qo Xl(t)
(2E2-15)
(oA p
,,...,.

_Md2Xl(t)
dXI(t)
K(X
D\
F:0+ !()
If --+rt.--+
0
A
~2

+ KX l(r) + ~ Q~ +
2 (oA p

Qoql(t)
'foA p

(2E2-16)

Subtracting the steady-state equations (Eqs. 2E2-9 and 2E2-10) from

Eqs. 2E2-15 and 2E2- J6 gives the linearized incremental equations as
Fo+K(})-X,,)
('I
Fo~----"

'"

dql

dt +

o~__.

Fig. 2E2-2.

Xo

Equilibrium points in electric field transducer.

-'"

JI

= M

Co

= X

Co qj

d 2x 1
dXl
dt 2 + ex di

Where

(oA
p

-i&.. X

+ CoXo

'
KXj

I,
I

Qo
CoXo ql

(2E2-17)
(2E2-18)

..

t~,
108

:1

ELECTROMECHANICAL ENERGY CONVERSION

,.-.

(Qo + ql)2 :::: Q~( I + 2 Qo)

and

(Qo

+ qd(Xo +

Xl) ::::

QoXo( I +

ql
Qo

XI)
Xo

(2E2-20)

ql = 0.2

I
1

Then, from Eq. 2E2-19)

(Qo

+ ql)2 =

Q~( I + 2 ~J

1.44Q~

1.40Q~

;1.

;,J}\( .""

,.

H.
or approxi mately a 3 per ccnt error.
'
Assume that prior to t ~ 0 there is a constan t applied voltage
Eo and
a constan t applied force F o, and at time t = 0 a step in applied
force
from F to F + F occurs. Find how the charge q varies with
time,
1
o
o
For this set of excitati ons the lineariz ed increme ntal differen tial
equatio ns
(Eqs. 2E2-17 and 2E2-18) become

dql
o = R d1
+

FI

Co ql

Qo

+ CoXa XI

(2E2-P )

d 2xj
dx\
=M
- + a - + KXI +
dt o
d12

Qo
C X ql
o

(2E2.22

and all initial conditi ons are zero.

The first step in obtaini ng a solution by Laplace method s is to transfo
rm
Eqs. 2E2-21 and 2E2-22. With zero initial conditi ons the transfo
rmed'
equatio ns are

o=

Rs QI(S) +

F 1 = M S2 XI(s)
.I'

c1o QI(S)

(2E2-23 )

+ cQx XI (.I')
()
0

+ us XI(s) + K X,es) + cQxo QI(S)

00'

(2E2_24)ih

where Q\(s) is the transfor m of ql(I), XlI') is the transfo rm of

XI(t), and
Fds is the transfo rm of the step functio n F 1 U-I(t). Now, to
find the

129

time variatio n of the charge q\(/), elimina te Xes) from Eqs. 2E2-23
and
2E2-24, obtaini ng the transfo rm QI(.I')

Consid er, for exampl e, the case where

Qo

l~

"-.' r.l

An indicati on of the accurac y of the lineariz ing approx imation s

can be
obtaine d by investig ating the accurac y of the approx imation s
;, f.
(2E2-19 ),

:I

ANALY TICAL, TECHNIQUES

~,.:t."

!
"!

I
I

I
I

'.'''.

;lJ\

QI(S);:;' _ ....."....=~~_ _,....",....,,,........,.,F,...:..I

. [RCoX oM .1
.\. Q
S
0

(RCnXo('/.

Q0

'"

(RCoXoK
Qo

2
+ XoM)
Q0 .I'

"

+ Xoa)s + (XoK _ ~)]

Qo

Qo

'

:lll.i",

CoXo

(2E2-25)

To find the time respons e qt(t) of the system, the inverse transfo
rm of
Eq. 2E2-25 must be found. Since this is a linear system, the
respons e
must be a sum of expone ntial terms, and these can be obtaine
d by it
partial fraction expansi on of Eq. 2E1'.25. Before solving Eq.
2E2-25,
a great deal of informa tion can be obtaine d by examin ing the denomi
nator
polynom ial.
The last term in the denominator~of Eq. 2E2-25, [(XoK/
Qo)
(Qo/CoX o)], can in general be positive ; negative, or zero. If this
term
is zero, then an additio nal s can be factored out of the polynom
ial. This
indicate s that the charge ql(t) will contain a term which is the time
integral
of fl(r); thus, for a step in fl(t);'ql (t) will be unboun ded-i.e .,
unstabl e.

Anothe r way of seeing this is to realize that factorin g s out

of the

denomi nator polynom ial in Eq. 2E2-25 yields a denomi nator factor
s2 and;

from Table 22, the inverse transfo rm of l/s2 is a ramp 1 which

is the

integral of a step. This type of integrat ing behavio r occurs at a

tangent ial
intersec tion shown by point A in Fig. 2E2-2. It is evident
from the
nonline arity shown in the figure that the range over which this
lineariz ed
model holds is limited. When the last term in the denomi nator
poly
nomial of Eq. 2E2-25 is negativ e, one root of the cubic will be
positive
and give rise to an expone ntial enl with a positive expone nt
a. This
respons e is unboun ded, and hence unstabl e, and occurs at the
type of
equilib rium point shown by B in Fig. 2E2-2. When the last term
of the
denomi nator cubic in Eq. 2E2-25 is positive , all the roots have
negativ e
real parts, yielding expone ntials with negativ e expone nts. This
stable
type of respons e occurs at an equilibr ium point such as C in Fig.
2E2-2.
Now assume that all the roots of the cubic in Eq. 2E2-25 have negativ
e
real parts and that the roots (Sj, S2' S3) of the cubic have been
found by
some method . To find the respons e by inverse transfo rmation
, make a
partial fraction expansi on of Eq. 2E2-25 as

Q1(S)

QoFI
RCoXo M
= .1'(.1' -

$1)(.1' -

S2)(S -

.1'3)

; Ao
AI
~ S + -s---.1'-\ +

A2
S-+-S-2

A3

+ S---S3
(2E2-26 )

J~I0'

110

where
S

ANALY TICAL TECHNIQUES

ELECTROMECHANICAL ENERGY CONVERSION

Sl' S2'

and

S3

III
Since the coefficients of the cubic in Eq, 2E2-27 are real, comple
x roots
must occur in conjuga te pairs. The corresp onding complex terms
of
Eq. 2E2-3 I can be combin ed to yield a real, exponentially damped
sinu
soidal term. In such a case the response of Eq. 2E2-3 I might ,have
the
form shown in Fig. 2E2-3 where the total charge q(t) = Qu +
ql(t) is
sketched.

are the roots of the cubic equatio n

a + RCI)S 2 +
+ ( M

K+ RCoM
a) S + (K
RCoM -

Q~) RC~X~M - 0

'~-"'.

(2E2-27)
The solution of Eq. 2E2-26 for the constan ts A o, A I' A 2 , and A 3 yields
~ ~

A 0=

in

S-I""'"(S-I---S::..2)'"'(S"-I---S......
3)

QoF\
RCoXoM

A3 = -

l ,:.- ~

..

it: ','

'r;'

,If"

l'

~--~::.-_
S3(S3 - SI)(S3 - S2)

\"
,

r: ' .:; I .
si\,1

'~;'

0 \

RCoXoM

S\S2$3

,"1

If the time response of the displac ement

;'

'

_ ' . 't"21
S2($2 -

SI)(S2 -

From Eq. 2E2-27 the produc t

S3)

SIs2S3

eS)' r,'

'.

S3(S3 -

$1)($3

]
S2)

. (22-29)
.

can be obtaine d as

"4"~

'"..
"

~ I

.,

!I S2$3

Q~
= RC M - RC2X2"
K

(2~-3 0)

The use of Eq. 2E2-30 in Eq. 2E2-29 puts the time response in the
more
recogni zable form
ql(t) = _

[1 +
XoK _ ....ili....)
F1

( Qo

sZsJeS1
,(SI - S2)(SI -

S3)

CoXo

'.
l

StsJeSz
(S2 - SI)(S2 -

S3)

':....

XI

is desired in additio n to

2. 1.3 Transfe r Functio n of a Linear System

'v'

S3)

i'

qt(t), the solution can be obtaine d by solving Eqs. 2E2-23

and 2E2-24
for the transfo rm X(s) and finding the inverse transfo rm XI(t) in
just the
same fashion as that followed to determi ne qt(t).

"

SI(S\.- S2)(SI -

,!

;,."

eSl

_ , ,_

Fig. 2E2-J. Typical response of electric field transd ucer.

.1:

~.

the partial fraction expansi on of Eq. 2E2-26 to obtain the inverse

transform of Eq. 2E2-26 as
r " ,: ;
Q F

o
a

...

'tft ;

.
The first and second transfo rm pairs of Table 2-2 can now be used
with
,
ql(t) =

CoK

" ,

S2(S2 -"SI)(S2 -,'f3),'

,;
),.'

" r...",.

\ '

R~(

(2228)

QoFt

"

.~pi"

'Q;;- -

A =* " ' - , 'RCoXoM

Ij

'j

- - - - - - - - - - - - - - - - -- -_ .. _-~. XoK ,
F

QoF\
RCoXoM

= -

AI
'J

ql

QoF\

RCoXoM

_...::-~
S\S2 s 3

"'1

S
"SISZe 3'"
(S3 - SI)(S3 ... sz)

(2E2-31)

"-1'.,

An importa nt charact eristic of linear differential equatio ns is that

the
natural behavio r of the system is determi ned by the charact eristic equatio
n.
This fact was quite obvious in the two previou s examples, one based
on
:. classical method s of solution of linear differential equatio ns
and one
based on Laplace transfo rm method s of solution . In both of
these
examples, the system response was compos ed of a sum of expone
ntials,
and the expone nts giving the system' s natural behavio r were determi
ned
from the charact eristic equatio n of the homoge neous differential equatio
n
-i.e., the system equatio n with no excitation or initial conditio ns.
The natural behavio r of an unexcited system of linear equatio
ns is
sufficient to determi ne the essential charact er of any linear system
with
any excitation. The charact eristic time response and stability of
a linear
equatio n are intimat ely related to the natural behavio r or the
system.
For this reason comple te techniques have been developed to study
linear
systems by investigating the system charact eristic equatio n. One
suchH

.;

,..~

>,t

':I.~

...

'If'"','
112

',J1:' .;
;

,,1

t,

' f'f;t~,

ELECTR.OMECHANICAL ENER.GY CONVER.SION

= Io"" xj(t

~)h(() d~

xo(r)

-I

x j (l1)h(t - a) du

~.

I~ ,

1.\ 1! ~~;,

t:f. ,

113

Thus in Eq. 2-17, h(1 - a) is the response of the system to a unit impulse
applied at time I = a and h(1 - a) = 0 for I < a.
Equations 2-16 and 2-17 are completely general and show that the
impulse response of a linear system is as fuJI a description of the system as
the differential equations from which it can be derived. The time function
pf Eq. 2-/7 is the convolution integral which has it simple form when
transformed to the complex frequency domain. Using the definition of
the Laplace transform.given by Eq. 27, the Laplace transform of Eq.
216 is

I,

00

(! q

,5

(2-17)

Note that the impulse response must be zero when its argument is less
than zero. since a physical system cannot respond before it is excited.

- t)e-"

dl]hW de

, ";~!'; ) ,;

(2-19)

'.~. ~

Xo(s) ,:.JoOl).e~~t [fo"\{t,7, ~)e-.(t-t) d(1

-,~)] h(t) dt"

/'

(2-20)

In Eq. 2-20 the bracketed expression merely represents the Laplace

transform of the input which is a function of s and independent of ~ and
can therefore be factored out of the second integration, as
't'

':J~

Xo(s) - X;(s) IowhWe~~~

d,

'

(2-21)

',I

In Eq. 2-21 the remaining integral is just the Laplace transform of the,
impulse response of the system. The conclusion is that the transform of
the superposition integral yields a very simple relation for the response of
a linear system in the frequency domain

Xo(s)

I
I

-00

10"" [foOl) xj(t

Multiplying inside the brackets of Eq. 2-19 by elt and outside the
brackets bye-It and noting that for the first integration (inside the
brackets) dt =: d(1 yield'
,

u in.'

(2-18)

, =

xo(t)e-" dt =Jow [fow Xr(t' - )hW d ~] e- II dl

Xo(s) =

F~.

I -

10

:l..:

Reordering the integrals in Eq. 2-18 gives

(2.16)

where x,U) = input time function

xo(r) = output time function

h(~) = impulse response of system

I = time of observation of the output

Equation 2-16 can be written in another form by letting
the integration, giving

'1

ANALYTICAL TECHNIQUES

systematic method is the use of a system transfer function. The transfer

function will be shown to be the transform of the impulse response of the
system, which is simply the natural or unexcited behavior as expressed
by the characteristic equation of the system. Transfer functions can be
defined for parts of complex systems, and these parts can be interconnected
to give totaf system response. rn this way each significant part of a
system can be retained in evidence. For any linear differential equation
with constant codlicients the transfer function can always be unambigu
ously defined and can be considered to be the system equations of motion.
since it contains all the information present in the ditfen:ntial equations of
the system.
'j.
In the simplest situation.. a linear system can be thought 01" as a box with
one input and one output. Such a system will be described by a single'
linear equation which relates the output variable to the input variable.
Two basic properties of linear systems enable one to develop general
analytical methods which are extremely powerful in their application.'
First. the principle of superposition states that the response of a linear
system to an input consisting of a sum of components is equal to the
sum of the re~ponse~ of the system to each component of the input. A
second important principle states that the response of a linear system is
independent of the choice of time origin for the input and output variables.
A particular type of input that is useful in describing the response of a
linear system is the unit impulse. If the input to a linear system is the'
unit impulse, the output response is called the impulse response, h(/).
Since any input can be thought of as being composed of a sum of non unit'
impulses, the principle of superposition gives the response of a linear
system to any input in terms of the impulse response of the system. The
resultant expression is called the superposition integral

xo(l)

to,V,it'

I
i
I

. :1

\~

i'

.;.

= H(s)

XJCs)

(2-22)

Note that Eq. 2-22 relates to the response of a system that is initially at
rest because Xo(s) and Xj(s) are the correct transforms of xo(t) and xlI)
only when xo(t) and xl!) are both zero for I < O.
The Laplace transform of the impulse response of a linear system, H(s).
is called the transfer function of the system and is of fundamental import
ance in the analysis of complex interconnected systems. The transfer

ELECTROMECHANICAL ENERGY CONVERSION

114

ANALYTICAL TECHNIQUES

function can also be thought of as the ratio of the transform of the output
to the transform of the input
Xo(S) = H(s)
X;(s) ,

= transfer function'

A-'J

br:

(2-23)
;','

~(

i,;

A partial fraction expansion of Eq, 2E3-5 yields

'I,'.,:
,.

f:

EXAMPLE 2E3

11

;;

:f

'r;,

As an example of some of the preceding ideas, consider the series R-L

network shown in Fig. 2E3-1, which is initially at rest;!' At t = 0 a step
voltage of V volts is applied to the network. .The current is to be found.
The differential equation of the system is

,::-;.,!

e( t)

= 1+'

i(t)

+t

'\

L di
-d

"

;'!

<.

,;'

, ) ' l i '
"l

jl J

V'

,~

iimp(t)

:"

= .!..e-(R/L)tJ

. t~

q~~'

,'1 ~

(2E3~)

Transforming Eq. 2E3-2 yields for the transfer function of the system
I' I

"\

, ,-

R(s)

= L _,

ii}:~'J

(2E3-3)

IT, I "

The' excitation e(t) is a voltage step of V volts. Taking the Laplace

transform e(tkusing transform pair 2 of Table 2-2, gives
.;

'V
(s) = -

(2E3-4)

The transform of the system response found by using Eq.2-22 is

V
l(s) =-L'srs

"3'

I ~,j \

(2E3-5)

+ (RIL)]

'

Ie

:1, ~-'

~ ;";~;:;i~J~
"l:!

!R (1

r[(R/L)I])

(2E3-7)

In the preceding discussion the transfer function of a linear system was

defined, and one method for obtaining and using the transfer function was
given. The most difficult step in the approach outlined in Example 2E3
is to find the impulse response of the system, A more straightforward
technique is to solve the transformed differential equations for the transfer
function. Then, if the impulse response is desired, it can be found by
takingt~.:; inverse transform of the transfer function.

-t:;

:t

i(t) _

As an example of the method of obtaining the transfer function (or

'functions) of a system directly from the differential equations, consider

the linearized (incremental) operation of the singly excited magnetic field

transducer of Fig. 2401. This is the same configuration that was

analyzed in Example 2El except that in this example the iron character

istics will be assumed to be described by a nonlinear, single-valued,

magnetization curve.

In the transducer of Fig. 2E4-1 the resistance R includes the winding

resistance; thus, the lossy part of the electrical system is separated from
the coupling field, The flux linkage >. with the coil N is a function of
both current i and displacement x as shown by the two curves of Fig.
'2E4-2. The plunger (mass M) is constrained to move in the x direction
only, and the motion of the plunger is limited to the range 0 < x < I.
The equilibrium length of the spring K is D.
To analy,ze this transducer. first obtain the complete equations of

motion, and then linearize for incremental motion, Since this is a magnetic

field device. the current j is a logical choice for the independent variable.

With i as the independent variable the mechanical force of electrical

origin is given by Table 1-1 as

::';;'3./.,..

If e(t) is a unit impulse, the impulse response is readily found to be

'ft (' j,-IJ"

:::1

"I

The first term of Eq. 2E3-6 is the Laplace transform of a unit step (an
exponential with a = 0), and the second term is the Laplace transform of
an exponential (see Table 2-2). Therefore the time response of the system
can be written down immediately as:

E.XAMPLE. 2E.4

i~

Fig.2EJ-I. R-L network.

'...

(2E36)

1
"

'I

"

I]

s + (RIL)

L
.

rft 1::.t.

[I

RV S -

:'1

;'

e(l)

./ ..0,1:

:'fl

," . I

;;

(2E3-1)

': \ I,

I(s) =

~, ~, '

";..1

<,

R'

1/5

H:fj

,h",,tt.

j'

'.'.~\,

,~.#'

+ oW~~' x)

(2E4-1)

II

116

ELECTROMECHANICAL ENERGY CONVERSION

ANALYTICAL TECHNIQUES

The coenergy W;". like the energy W"'. is a state function; consequently,

W;" is a function of the tlnal values of i and x and is indercndcnt of the

path followed by i and x in reaching a given state. Thus. choose a path
of integration for W;" in which the mechanical variable is held fixed at the

'1'17

Using d'Alembcrt's principle for the mechanical system and Eq. 2E4-3
,\
gives
d 2x
d:c
(2E4-6)
f(t) = M -12 + a d- - K(D -;- x) - !.U, X)
((,

"

Equations 2E4-5 and 2E4-6 an~~~he equations of motion for the transducer
shown in Fig. 2E4-1.
.."
1

.~

j~,'gth

tot sprlna

'\

. ~ .'"

JlIl

r~

,':t
;J'

l, K
1000UO

I,

. t~

'.

\1.

'. ~~,~ 1

: ' ; t

) ~

Singly excited magnetic field transducer.

1.

;\

.:

f:

,::~

I" ~

~,

(2E4-2)

>,,'(i', x) di'

Substituting Eq. 2E4-2 into Eq. 2E4-1 gives, for the mechanical force
due to magnetic field coupling,

feU, x)

= (~>:

u:

>"'u', x) dt]

Now write the equations of motion.

the electric circuit is

I 8>..'~:.'

x) di'

(2E4-3)

The Kirchhoff loop equation for

e(r) = Ri

d;>,.U, x)
dr

e(r)

0>"(;, x) ~
oj
dt

'r

Ri

(2E4-4)

8>"U, x) dx
dt

+ -ax-

~.

(2E4-5)

1\-

j'

Fig. 2E....2. Saturation curve of magnetic

field transducer.
,\

.... ~~

.', ! .

}ID't

, :;

~~ J

"The 'next step in obtaining the line~r incremental t~ansfer functions of

this system is to linearize Eqs. 2E4-4 and 2E4-6 for incremental operation
around an operating point. To aC90mplish this, rewrite the 'driving
functions as
e(t)
Eo + el(lh,
'\ '
f(l) = Fo + fl(l)
(2E4-7)
1(1) = 10 + i 1(1) "
x(t) = X o + XI(t)

<1

where"Eo, Fo, 10 and 'Xo' are constants that define the equilibrium p~int'
and ej.fl> i lt andx\ are time functions that describe the small departures"
from the operating point.
,
The nonlinear functions >.. and [. can be expanded in Taylor series.
4
around the operating point, thus
."
.
,t)
'

or
=

.,',

: c\

~ "

final value while the current is brought to its final value (see Chap. 1),
and the coenergy may be expressed a s '

W;"

, :h

at';

Fig.2E4-I.

':

':)

'f',

plunger, , e, POSItIOn
Where sprmg torce
1$ zero

,I,."

"

It'"~

Equilibrium PosItion of

! ~

r:p,:,g

~
~//5a
,);;'/;/X/,7ffff#Wm:';;;;:,Y

",\,':

;'

.f

:\'
EqulhbriulY,

"

>,(j, x) ... >'(10, X o) + >'/(10, XO)i1 + >"",(/0, Xo)Xt

+PI/(Jo, Xo)i~ + 1';>""''''(/0' XQ)xi + >"/Jt(lo. XO)i1Xl

+ . ..

(2E4-8)

For a Taylor series expansion of a function of two or more variables, see Hildebrand.
op. cit. p. 353.

:~

,~
"

po

118

ELECTROMECHANICAL ENERGY CONVERSION

i~

where

oil.

oi

A;(lo. Xo) =
<

A;x(Io. Xo)

(10,

ANALYTICAL TECHNIQUES-

Xo)

<;

Assume that the variables (10 and Xo) satisfy these expressions, and
subtract Eqs. 2E4-13 and 2E4-14 from Eqs. 2E4-1O and 2E4-11 to obtain

"

,~

a2i1.

= 8i ax (lo;Xo)
;,

l."
~

Iii, x) == 1,(10. X o) + 1.;(/0 Xo)i l + le,,(Jo. XO)XI

+ t.f.iI(lO. Xo)i~ + t/m(lo, X o)x7 + 1./x(1o, XO)i1X l

,!.

I'
Jl

~{

+ ... (2E4-9)

<

,
ie!.

.;~,

dXI

+ a Clt + K(Xo - D) + Kx 1

- 1'</0, X o) "';"le/(1o, Xo)il.::"

.t~x(lo. Xo)~l~

(2E4-11)

The terms involving Ie can be put in terms of A by using Eqs. 2E4-3 and
2E4-8. Thus
f1

le(lo, X o) =

~l~'

II
0

'A~(/', X 0) di' ;1'"

fl.

Ir'

"II,.

',!,

le/(1o, X o)== Ax(Io, X o) \

lu(1o. X o) ==

,;g .

'W:

II"

~.

(2E4-12)

5:0 'A~x(i'. Xo) di'

: ~~t,!

';~' f

= RIo

Fo == K(Xo - D) - 1'<10, X o)
-'ft"
~ -, ,.

I"

il.x(lo, X o)

d;1

dXI

(2E4-15)

t2

1...,(/0, XO)XI

(2E4-16)

F 1(s)

(24017)
(2E4-18)

To cbtain a transfer function, the input variables and the output

variable must be defined. In the case of Eqs. 2E4-17 and 2E4-18 the two
quantities 1 and 1 are specified as given. and the equations can be
solved for either one of the remaining variables (II, Xl)' When more
than one input is specified, the usual process is to find the transfer function
from one input to an output variable with all other inputs constrained to
zero. This process is perfectly general, since superposition holds for the
linearized equations.
For example, find the transfer function from input voltage 1 to output
position XI with no incremental force applied (F1 = 0). That is, find
the function
'. ,
Xt(S)!
I(S) F1-0

Setting F 1 = 0 in Eq. 2E4-18 and eliminating 11 between Eqs. 2E4-17

and 2E4-l8 give the desired transfer function as

J;-

M - dI + a -d + KXI - 1./(10. XO)ll -

I',

However. for purposes of this, example where iI.(i, x) will not be explicitly
defined, the mechanical equation of motion inthe form of Eq. 2E4-11
will be used for further considerations.
';' .".
The equations that describe the equilibrium point (Eo. Fo, 10, Xo) are
obtained from Eqs. 2E4-l0 and 2E4-11 by setting all the .incremental
quantities (elt lit lit XI) and their derivatives equal to zero., 1,"hus

Eo

;1 +

= [R + iI.;(Io, Xo)s] II(s) + 'Ax(Io, Xo)s XI(s)

= [MS 2 + as + K - lu(/o, X o)] XI(s) - 1.;(10 , X o) ll(s)

l(S)

\ (1 X ) (
dil(t)
dxl(t) (
R10 + R/.I (t ) + II;
I t + >..,(10' X o) ~
2E4-10)
o
0

d2Xl
Po + II(t) == M dt 2

d 2x

iI.;(lo, X o)

Equations 2E4-1 5 and 24-16 are the linear incremental equations for the
transducer of Fig. 2E4-1.
The object of the present example is to illustrate how to obtain
incremental linear transfer functions for a physical transducer. Thus,
taking the Laplace transforms of Eqs. 2E4-15 and 2E4-16 assuming zero
initial conditions, there results

-~

l+

"

i;.

el (t)

Ri

el =

'::. < .

where the subscripts I and x mean"partial differentiation as illustrated for

A above.
Now substitute the expansions of Eqs. 2E4-8 and 2E4-9 into Eqs.
2E4-5 and 2E4-6 and neglect all terms of second order [e.g.
ilxl'
xI(diddt)] or higher in incremental quantities to obtain the linearized
equations of motion.

Eo +

~.

and

i'~t '!~.:"<

119

ti;>r~'

,- ~

Xt(S)1
1(S) F1-O

,I:

""
I.~

t,

+ 1.;(10, X o)
{MA;(Ia, X O)s3 + [M R + ail.;(Io, X ons 2 + [aR + KiI.;(Io, X o) \""
- lu(Io, Xo)A;(Io, X o) + il.x(lo, Xo)/.;(lo. Xo)]s + R[K - I.x(lo. X o)]}
(2E4-19)

",

(2E4-13)
(2E4-14)

There may exist steady-state excitations (Eo, Fo) for which Ihere is no equilibrium
point. For a discussion of this possibility. see Example 2EI.

;,::.14....,;:r;

It

.,.

,~.;

,", i .. }

120

ELECTROMECHANICAL ENERGY CONVERSION

"

Note that this tr<Jnsfer function is simply a constant divided by a polynomial

(cubic) in the compkx fn;:l\uency s. Note further that the coefllcienls of
the polynomial are functions of the equilibrium variables (10, Xo), and
therefore the transfer function varies with the operating point. This is a
manifestation of the nonlinearity of the transducer. The time response
XI(t) can be found for a given el(t) withfl(t) = 0 by using the techniques
of Sec. 2.1.2 with Eq. 2E4-19.
Three additional transfer functions

~1(.\')'

11(.\')1
1(.\') F,-O

I' tC~) E,-O

11(.\')
F I (.\') ":1_0

and

The response. say x1(t), to inputs Jl(t) and el(t) can be found from the
sum of the responses obtained using the two transfer functions

Xt(S)!

.~

EI(s)

X\(S)!

and

F\(s)

F1-0

1- 0

The response il(l) for driving functions e,(r) and fl(t) can be similarly
found from the other two transfer functions.
In the preceding discussion and examples the transfer function was
defined as a function of the complex frequency s. In the process of
deriving the transfer function from the differential equations, all initial
conditions were specified to be zero; consequently, dldt was replaced by
sand f dt was replaced by I Is. It is just as easy to define a transfer
function in the time domain by letting the differential operator p = dldt
and lip = f dt and solving the resulting eq uations for the ratio of the
variables of interest.
For instance, in a series R-L circuit the differential equation is

"

+ L diet)
dt

(2~24)

Now replace dldt with p, and solve the resulting algebraic equation for
the ratio of current to voltage to obtain

i'il,,4;"

<

1:

'"
'I

i.

'.-'

, ~ ';'

i(t)
e(t) = R

Lp

(2~~5)

This is the same form as Eq. 2E3-3 of Example 2E3 except that p replaces
s and the variables are time functions.
The time responses from Eq. 2-25 require the solving of a differential
equation (Eq. 2-24); thus, it would seem to be of no advantage to place

'. 1:1 '"

,',,;

jj1(,\

'I'

2.1.4 Frequency Response

.

j.
','n

,I
t

Whenever a system is describable by linear differential equations with

constant coefficients. the steady-state frequency response~ which can be
calculated quite easily, contains all the information necessary for the
description of the system under any dynamic conditions. Since a steady
state frequency response is relatively easy to measure in the laboratory,
it provides a practical method for obtaining the constants of a system. :
To obtain a definition of the frequency response, consider again Eq. 2-1
,

dnx

I
, '

dIn

hI;. Hti'!l: tf'd,

,'/

t,

+ al

dn-Ix"
dt n- I , + ..

:,:::, all-I

,dx
dl

"

+ a"x = J(t)

(2-1)

,:

~J.HJ

where al> ' .. , a" are constants. Assume that this expression describes
a system in which f(t) is the input variable (force, voltage, current, dis
placement, etc.) andiC(t) is the output variable. Now determine the
steady-state response in x(t) to a sinusoidally varying/(t). It was shown
in Sec. 2.1.1 that the complete solution of Eq. 2-1 can be written as
'" ;'.r:t':",<. ,; .... -.

.
'
where the
equation

C,

"

.,~ ;:';.:
C'

\(

!"

L:

, x(t) =

<

Cierl'

+ xp(t)

;!,.;)

'

J~!

1.. \

(2-6)

'

:!
are constants and the
are the roots of the characteristiC?
; .," \
+ al",-l + ... + a,,-I' + a,,= 0 .'1,
,,' (2-4)

'1

'II

If

",j,

(21

the equation in the form of Eq. 2-25. However. Eq. 2,25 is usci'u! when
studying the steady-state frequency response of a system, because it gives
the frequency response characteristic without first Laplace-transforming
t,he system equations. This pOint will become evident in the next section.
., A further and more significant advantage of writing ditTerential equations
and transfer functions in terms of the operator p is the ready applicability
ofsuch equations to the use of analog computers. The solution of system
equations by the use of analog computers is accomplished in the time
domain; consequently, El\. 2-25 is in the proper form for lhis usc.

,,'

e(t) = R i(t)

ANAL YTICAL TECHNIQUES '

can be found from Eqs. 2E4-17 and 2E4-18 in the same manner in which
Eq. 2E4-19 was obtained.

~ f)'

Only in the case when the system is stable can a steady-state frequency
/I

response be defined, and in this case the term

L Cieri'

will approach

i~1

zero as time t progresses because all the '1 will have negative real parts.
Thus the steady-state solution of Eq. 21 is given by

xss(t) - xp(t)

'"

(2.26)

122

ELECTROMECHANICAL ENERGY CONVERSION

I

"tl>, ,.;:.

'Sf:

i,'

..,.

The particular solution xp is 'foli~d'p~:determining a solution for Eq. 2-1

for the driving function /(1).
" ~\
Assume thatf(t) is described by
.

"Xp(t) = Re (Xel",l)

In order to obtain the mutual inductance between stator and rotor, assume

Ixil

(2-28)

t--~

F ;

.'.;

Ixll

~'.

~'Ifi>,

~"l" >"
.~"

,~.;

(jw)n

+ a,{jw)n-' + ... + an-,(jw) + an

!.,.' \- ,"':

(2-29)
Equilibrium poSItion
of uneXCIted system
'I <p. 0

"rH,:

.~, ,i

I iI',

[if ; ' , ' "

'.I''1m!}:'

'";

t~iiJ

:;'t,

...-t. ': I

~it.,

Hildebrand, op.

Chap. I. , I f ; i '

f I ~

'},t). ,

I .,

",

hH ~;:1 ~

In

'.

;~b!.

'r '

.'(1)

("

: 'i I!

R'

I',' ."

", j:1

~.

j'

"(lFr~',"

Fig.2E5-I. Doubly excited magnetic field transducer.

that the magnitude of the flux density is constant around a circumference.

Then by a simple integration the mutual inductance is given as

1('

mutual inductance between stator and rotor = L" = M cos .p

...t

Next assume that the resistances R' and R' include the coil resistances,
and the mechanical parameters are as defined in the figure:
rotor moment of inertia "" J

rotor damping factor = IX

torsional spring constant = K

, "~) ':f'1 .,

;J'
, .1

(\)

)t i t

.,1>(

As an example of the application of frequency response techniques

consider the operation of the doubly excited magnetic field transducer of
Fig. 2E5-1. The first step in finding the frequency response characteristic
of this tranducer is to formulate the 'equations of motion. This will be
done using the Lagrangian method developed in Chap. 1. The choice of
'a Lagrangian formulation is arbitrary and merely illustrates a different
approach for finding the system equations of motion.
In the transducer of Fig. 2E5-l assume that the' iron has 'infinite
permeability. thus yielding electrically linear inductance parameters that
describe. -"ithe
magnetic
field energy storage and coupling.
Since the . air
;
: to.: ."
- .

~"".! ,I

Rotor mlgnet"

h-...

Equation 2-29 is the frequency response of the system described by Eq.

2-1; i.e., it gives the ratio of the con\plex amplitude of the output X to
the complex amplitude of the input F.
d '
Note that the frequency response of Eq. 2-29 results from replacing the
variables by their complex amplitudes and replacing d/dt by (jw). Thus
the frequency response can be obtained directly from the differential
,equation as above; it can be obtained from the transfer function in terms
of time functions and the operator p; or it can be obtained from the
transfer function in terms of Laplace transforms by replacing the complex
, frequency s by jw. < These methods will be illustrated in Example 2E5.
EXAMPLE'lE5.'!

5tllOr mlgnetlc

{1?

where X is the complex amplitude and is in general a function of the

frequency (Jw). Now substitute Eqs. 2-27 and 2-28 into Eq. 2-1, drop
the Re from the equation, and cancel out the elwi factors to obtain

~ (jw)

self-inductance of stator == U == constant

self-inductance of rotor = L' = constant

!l'r:.

(2-27)

where Re denotes ., real part of," and F= Fel O It is known from the
method of solution with undetermined coefficients that the solution
xp{t) will contain only sin wt and cos wt; consequently, assume' a solution
'
,
,
for Eq. 2-1 of the form

W(jw) =

gap is smooth, the self-inductances of the rotor coil and of the stator
coil will be constants; i.e.,

"~I

", ' '

f(t) = F cos wi "'" Re (Fel "")

123

ANALYTICAL TECHNIQUES

..<--

-. \

. "..

lr

.:l:i:r

. '

"

'.

12<4

I
,

ELECTROMECHANICAL ENERGY CONVERSION

I!

To determine the equations of motion the independent coordinates must.

be chosen. Select magnetic energy as kinetic energy; thus the independent
coordinates are as given in Table 2E5-1. The Lagrangian in terms of
these variables is given by (see Chap. 1)
L

T' - V

4 ~; -

Mi' (sin

1: . .

e'(t)

(2~-7)

L' ~;

.+ R'i' + M(cos 4 ",~: -

Mi' (sin

~)dj; =

e'(t)

(25-8)

= T(I)

. (2E5-9)

. .

, I

Coordinate q
Velocity Ii
Momentump
Force!

q'

q'

j'

j'

A~v,

)..,

~J</>
-K4>

~.

Since this is a magnetic field device, it is reasonable to neglect any

capacitance in the transducer of Fig. 2E5-1. Thus the conservative
Lagrangian becomes

L = ;L'i'z + iL'i'z + M(cos 4i'/'

+ tJ</>2

- tK!fi2

M(cos

. J dd2t

ex

..

'

ddf~
t

K4> +

Mi':ir~Sin 4

Equations 2E5-7, 2E5-8, and 2E5-9, are the complet~ eq~atio~s 'of
motion for the transducer of Fig. 2E5-1. Note that these equations are
nonlinear., Assume incremental operation around a steady-state opera
ting point, and, as before, let the variables with subscript 0 denote the
operating point and subscript I denote the incremental variation: Define
these new variables as
e'(t)... E~ + eHt)
.' .
,i'(t)' ~ J~ + if(t)

'.'. "

Mechanical

R'i'

Electrical
Rotor

preceding~quations

L' ~:

TABLE 21:5-1

Stator

125

Performing the indicated differentiations in the three

gives the more conventional equations:

(2E5-1)

'

ANALYTICAL TECHNIQUES

e'(t) =: E~ + el(l),:.
i'(t) .=: J~ + ir(t)
T(t) = To + Tt(t)

'~q:~
.-~

,,,"

(2E5-2)

\~

Ll

".

4>(t)

=:.

. ,~~ '/

"L'"" _

"

' ,"

,,,' .., ,' , "(2E5-10)'

,t.:

CPo + 4>1(1),

~:;tH~,<;; ..\ ;";'~_h.t

~'fli,;

Substitution of Eqs. 2E5-10 into Eqs. 2E5-7, 2E5-8, and 2E5-9 yields
The Lagrangian of Eq. 2E5-2 can be substituted into the Euler-Lagrange
equation
.)

d(OL) oL

d~

L'

~; +

R'(Jf

(2E5-3)

dt O(ik - Oq/f "'" Qk

f _

di'
L' d; + Rru~ +

where Qk includes all nonconservative forces (dissipative and sources).

Making this substitution results in the three equations of motion:
Stator circuit:

'>\

!!.... [Ii'
dt

M(cos

,.

R'i'

+ e'(t)

(2E5-4)

d~

+ M cos (CPo + 4>1) ~:

M(lo +' iD sin (CPo +!fil) ~t

iD +

+ ef(;j"(2E5~11) ::..
'",H, :,', ; :,.,'.
'k

'
di"
M cos (CPo + 4>d d;

d~;l + ~I + K(CPo + +Ir~

(X

M(Jo

:;r;

:"E~ +e~(;) (2E5-':1~)

+ i:>(/o + iD sin (CPo + !fit) :~

, =

Rotor circuit:

~ [L'i' +

M(cos

4i']

= - Ki'

+ e'(t)

~ (J</ + K!fi

~ [M(cos !fi)iSi') =

-ex</>

:f
T(t)

To

+ T\(t)

(2E5-13)

ir.

(2E5-5)

Mechanical system:

lEo

,C'L,

- MUo+ if)'sin (CPo + 4>\) d:/

...

4;'] ~ -

,
if)

(2E5-6)

By setting all incremental variables (ef, 'e~. if.

T" 4>,) equal to zero, the
equations describing the steady-state operating point are obtained from
Eqs, 2E5-11, 2E5-12, and 2E5-13 as
(2E5-14)
R'lo = ~
~2E5-15)
R'/~ = E~
.'(2E5-16)
KCPo + M Jo/~ si n CPo = To

.~
'I

'.

."',
,

r.

~.\" ~

ELECTROMECHANICAL ENERGY CONVERSION

126

(",'

&S
L' ---!
dt,+ RSi'I

+ M cos (~. 0 + '1'1

-I. )
_ M(I

, . ' :1:.,

di'
L'.......!
dt

J ddt4>1
2

'+

IX

d4>1
dt

+ K~'1'1 +

to a sinusoidal driving force of any other variable.

and driving functions as

J.

......!
dt

= er(t)

(jwL'

_I

M(l0l "+ i ....

(J'+ i""
sin
II 0
II

1 ==

e~(t)

[(jw)2J

(2E5-18)

(2E5~19)

These equations are still nonlinear. To linearize them, approximate the

sines and cosines by Taylor series 'expansions around <%>0 to obtain
,

+ 4>1 cos rtJ o

-1

sin rtJ o ,

'I.

diS

di' '>

d"ll
L' d +\R'i[
t

d 2</>"
J

dt 21

dt\,': e1(t)

d'S
d4>
ll
M(cos <Z>o) d - MIa (sin <Z>o) -d1
t ,
t

d4> ", "

ex d/

de/>

=e~(t)

(2E5-26)

MI~/o cos <Z>O~~l

Mlo(sin <Z>o)If

+ MIMsin <Z>o)Ir

== T 1 (2E5-27)

;.!

= ~ll (jw)

I",

'I'm

I L'

1\

(2E5-28)

(2E5-23)

=~

cZ I (. ) JW

M/~/o

cos

~o

[1 - ()2]
:n + 2j'

(2E5-29)

Wit

where

+ MIMo (cos < % > 0 ) ] 4 > 1 ,

+Mlo(sin ~o)if

~0)cZ1 = E~

(2E5-25)

1. Stator excited by a constant-current source I~ so that Ii' "" O.

2. Rotor excited by a constant-current source Joso that Ir "" O.

W(jw)

[K

Re (I~eiw,)
Re (w 1eiw')

This is the frequency response of a second-order system (the highest

power of w in the denominator is w 2). Putting Eq. 2E5-28 in the
"I-'-'...\i
1
"it ~I,',l
normalized form gives

(2E5-22)

'

+ jwM(cos ~o)I: - jwMlo(sin

+ jwex + K +

W(jw)

'"

,L~ ~~~t R~if + M\cos ~o) d;:~. -: MI~ (sin rtJ o)

;
Find the frequency response in the angle 41 when the torque T. is
," perturbed sinusoidally at the frequency w.
To solve this problem use Eq. 2E5-27 from which the desired frequency
response is obtained as

(2E5-20)

Now substitute Eqs. 2E5-20 and 2E5-21 into Eqs. 2E5-17, 2E5-18, and
2E5-19, and keep only first-order terms in the incremental quantities to
obtain the linearized incremental equations of motion
.

i~(t)

</>1(1)

" These three equations can now be used to find the frequency response in
, any variable for the sinusoidal perturbation of any input or combination
.' of inputs at the single frequency w.
~ Consider the frequency response under the following conditions:

R')I~

,. +

(~0 +.J.)
'1'1 "

- Mltlo sin <%>0 - T 1(t)

if(t) = Re (Ire)"')

+ R"}If + jwM(cos ~o)I~ - jwMlo(sin <t>O)~1 = Ef

(jwL' +

dt

1t

Write all the variables

Substitute these variables into Eqs. 2E5-22, 2E5-23, and 2E5-24, drop the
Re, and cancel the e)w' factors to yield

(2E5-17)

diS

M(Io + if) sin (~o + 4>1)

sin (rtJ o + 4>1) == sin <%>0

e;(t) == Re (Efe)"')
e~(t) = Re (E~eiwl)
TI(t) = Re (Tle iw')

a +iD sin (~o + 4>1) d;/

+ R'j'I + M cos (~
+ '1'1
-I. )
. 0

&'

127

ANALYTICAL TECHNIQUES

Assume that the driving forces (E~, E~. and To) are such that these equations
can be satisfied, and subtract Eqs. 2E5-14, 2E5-l5. and 2E5-16 from
Eqs. 2E5-11, 2E5-12, and 2E5-13 to obtain the following incremental
equations of motion:

+ Ml o(sin rtJo)i[ == T 1(r)

wn =

ill' .

,+

'(2E5-24)

, =

Now Eqs. 2E5-22, 2E5-23, and 2E5-24 can be put into a general form,
and an expression can be found for the frequency response of any variable

The factor

,'./.

"~-'

Wit

JK

+ M/~/o cos rtJo

'!w~'tb 1, 'f',~

ex

2 V](K + Mltiocos ~o)

is the natural frequency, and' is the damping ratio.

.- LlltVV,

~Y
\

);i J~

128

ANALYTICAL TECHNIQUES

ELECTROMECHANICAL ENER.GY CONVER.SION

.

,j

SL----

W (0) -

[i - (iJ,d]+

2jt

... ~ .... L

','

JJ'

""\

(4) Since at wlw" = I the magnitude is proportionaJ to Ig. the height of

the resonant peak depends strongly on ,. Of course, for , > I no
resonant peak occurs.
Another lIseful way of plotting a frequency response is on a polar plot
or Nyquist diagram. This diagram is obtained by writing Eq. 2E5-29
as either

W(O)

wliw)

.".;.

LIi"

..
i

i.ll

20 log!o\ W(jw)!'"" 2010&101 W(O)I - 40 logl~ ~,"'Ji

.',

\W(M\

I) the frequency

essentially constant; (3) for very high frequencies (w/"'"

response is asymptotic tp the linear equation

Next, one common technique for plotting frequency response curves

will be i1lu~trated with the response of Eq. 25-29. The Bode diagram
for plotting frequency response consists of two graphs: one graph contains
a plot of 20 10gIOI W(jw) I in decibels (db) as a function of IOglO w; the

2010g l0

129

~n

+j

W(jw) = Re [W{Jw)]

r .0.2

(2E5-30)

1m [W(jw)}

or

(db) 0

Slope
-1

/ W(jw )f~d""" '

(2E5-31)

Using either Eq. 25-30 'or 2E531. draw a phasor in the complex plane
for any particular value of w. e.g.>. the phasor shown in Fig. 2E5-3. Now

=40

db/decade

-2

IW(jw)!

W(jw) =

,./'~

('

10&10.

~n

,'(.'

1, ,,;.

~-

."

Ill.,,'!

,"'",l!! ,iV'\

Imaginary aXI,

. et

t
'J

IwU w )

>',

'!

10gl0 : .

,,'

.... J

.i1"

"';r'

'

1~

"

.'

;\ 'ih

'in

~t4;':' :}

-90

';J,~

i'

.."flat;

~,'t

" '~,

"..--- -----

_lS0L----------------~

, : ,1

> ,

")

Fig. 2E5-2.

'\
" ..

1"(

fl'

\<>d'l
! .l

Bode diagram of second-order factor.

,'"

Fig. 2ES-3.

"!

'I

Frequency response phasor plotted in complex

plane.

other graph contains a plot of the angle /W(jw) as a function of log10 w.

7;

The expression of Eq. 2E5-29 is plotted in Fig. 2E5-2 for a value of

{ = 0.2. Several features of this set of graphs are: (1) at w = w" the
real part of the denominator of Eq. 2E5-29 is zero; thus the angle is
_ 900 and the magnitude is given by the imaginary part; (2) for
w/w" ~ 1, i.e.. for very low frequencies, the frequency response is

1",

, ,,,~

f',

a plot of the locus of the tip of the phasor W(jw) as w is varied from zero
to 00 gives the polar plot or Nyquist diagram of the frequency response
shown in Fig. 2E5-4.
In the preceding discussion the frequency response of the transducer of
Fig. 2E51 has been consi~~\ri~; ~ith excitations of direct current sources
I'

'I

.. ;Jl;~ ..

~ ~l

.f

....

'

'l',

!'

i,

1r.-

,I.'

.~

'" ~!~r~t-)

Imaginary axis"

. t)

,\'

','
'i,'

3. A constan t torque To applied to the rotor so that T J = O.

, 1
;: ; '
~r
j

The frequency response of Eq. 2E5-32 is fourth-order,

can be put in the more recognizable form

i
.oc

~.

ANALY TICAL TECHNIQUES

",1

1-,

lJ

'

, \,1.".. . '".',

. : '1

K+M/~t~ cos~~

----7 "iiJ

k ",.0

I:

" :;' j

ii:

.:

Real axis

~I-~ _~

v-'

'.I

~,I

:nr" "',. ,(

;H

'r)

"("'.

f1_ (~)2L..2jr ~

!~

,')t>.1

;.

Uol"lj

W II

rr'* 0.2

~.

J j : ;",

Now find the frequency response in the angle <PI for the driving voltage

To find this place the foregoing restrictions on Eqso 2E5-25, 2E5-26,
and 2E5-27 and eliminate the twO currents from the resulting equatio
ns
to obtain the frequency response
ej.

_Rs/~

+ jw(Ml

cos ~o - L'l~)

0
1 (Jw) ~~A(jw)4 + B(jw)3 + C(jw)2
+ D(jw) + F
0

A = J...;(,-L_'L_s_-_M_2...,C_O_S2_~....::;.O)

k':~
!t~ .

'''where

(2E5-32)

o'

M sin epo

a(L'L' - M2 cos 2

==
C

s.~
F

+ J(R'L' +

RSL')

M sin ~o

(X + Iv!/ l 0cos <Po)(L'LS - M2 CO~2 <Po) + J~R' + a(R'U

o '. ,.
M Sin ~o
I. ,;
;~.;l
+ M sin ~o(LS 1~2 + L'l02 - 2M1010cos 4 ' > 0 ) ' ; \ '

_ (X

4'>0)

(K +

Iv!10/0cos ~0)(R'LS + R S L r )
Iv! sin 4'>0
Iv!JSI'cos~

M
1,'

O?

Sin

",,;

epo

+ aR R';

M' <P (f S2 R S

+ .~~;n

0, 0

+ RSL')

J'2R')
0

.,

'l~

)RSR'

,I

JW

==

R'/~

+ jw(M/~cos <Po-U/~)
+ Aw 4 ) + jw(D - B( 2)

(F - Cw 2

(2E5-33)

In the study of electric circuits it is accepted practice to call the externa

l
points of entry into a network its external terminal pairs. A networ
k
characterized by n external terminal pairs, chosen on any arbitrar
y
basis, ;<; called an n termina l-pair network. Associated with each termina
l
pair of the network is a pair of variables, one independent and
one
dependent, such as a voltage and a current . The concept of termina
l
pairs, from electric circuit theory, is useful in the study of electro
mechanical systems. In an electromechanical system an electric
al
terminal pair is characterized as before by two variables, a voltage
and a
current. A mechanical terminal pair 'similarly is characterized by
two
variables, e.g., a force and a displacement or a torque and an angular
displacement. In electromechanical systems, as in electrical network
s, it
is often desirable to obtain a transfer relationship between a variable
at.
one terminal pair and a second variable at another terminal pair.
In a
simple two termina l-pair electromechanical system the desired transfe
r
relationship might be the ratio of the mechanical torque to the applied
voltage. In general, these transfer relationships are more comple
x in
electromechanics than in electric circuits because any electromechanic
al
system involves energy conversion and hence is inherently nonline
ar.
A useful way of representing the equations that describe a system is
by
block diagrams. This technique consists of pictoriaIly representing
the
equations with all pertine nt inputs and outputs placed in evidenc
e. A
block diagra m" algebra " exists for solving the system equations for
any

Fig. 2E5-4. Polar plot of second-order frequency response.

41

Er

2.1.5 Block Diagrams

1\ + M/~ T{, cos ~o

W(jw)

~ttl, .I,!

4>1 (. )

This expression

The frequency response (Eq. 2E533) can be plotted on a Bode diagram

(Fig. 2E5-2) or on a polar chart (Fig. 2E5-4) or in any other
way that is
desired. The plotting of Eq. 2E5-33 proceeds in exactly the same
way
as for the simpler reponse of Eq. 2E5-29.
From the foregoing treatme nt it is evident that a multiply-excite
d
system does not have a single frequency response; instead, it
has as
many frequency responses as there are types of excitation. The various
types of frequency response for a system will become more apparen
t
when a block diagram is drawn for the system in the next section.

~,

In:

131

ELECTROMECHANICAL ENER.GY CONVERSION

130
on the stator and rotor. Now change the excitations to the followi
ng:
I. Stator excited by a direct voltage source E~ so that ~ = O.
2. Rotor excited by direct voltage source E~ and in additio n by
a
small sinusoidally varying voltage of frequency wand complex amplitu
de

:!

. .. E. A. Guillemin, Introductory Circuit Theory, John Wiley, New

York, 1953,
pp. 153-161.

;!.i
'1

;~., ~'

,..

:-

. .

,,~

132

ELECTROMECHANICAL ENERGY CONVERSION

)o~

Y)o Y

TECHNIQUES , " ,

"

t .

Ri + Lpi = e '

."

"

"

i,

"t'~ t ;

I ""

rpI (e -. iR)

.'

(2E6-3)

". ; : .

"

j.

...{

'

."

<~

j~t

k"

R
~

+0

..

(3)

~ X

',1

;~.

l _

.,;t. ,

"~~ ;

z. z=x- Y(Summation)
a

An alternative and equally valid block diagram representation of the R.L

=AX (Multiplication by a constant

(2E6-2)

Then, by using the three basic symbols of Fig. 2-1, dra w the block diagram
of Fig, 2E6-2a. This hlock diagram essentially represents the cljuation

or by a Ilnear operator)

-;0

,i;

"

(2)

1,33

First use the notation p ="d/dt and rewrite Eq. 2E6-1 as

.....

.'....

(1)

,ANA~l)TICAl

.,t

specified set of conditions. In many cases a system block diagram can be

used to obtain useful information without making a complete solution of
the system equations.
'
The lirst LIse of block diagrams will be for those systems that are
describable by linear differential equations with constant coefficients.
Later, techniques will be introduced for treating two types of nonlincarities,
including methods of linearizing for incremental operation arc'od an
equilibrium point.

In order to represent linear equations, the three symbols shown in

Fig, 2-1 are needed. The first symbol represents multiplication of a
variable by a constant or by a linear operator. If the equations are in

'""""1"'"

r'""'
Fig.2E6-I. Series R-L circuit.

._!

=X (Identity)

Fig. 2-1. Basic symbols used in linear block diagram represen

tation,
~l~

:f!.J,

,t',;:,h 4

j;,

the time domain, A will be a differential operator; if the equations are in

the complex frequency domain (Laplace-transformed), A will be a
function of the complex frequency s. The second symbol represents
summation, i.e., addition andior subtraction of variables. Even though
only two inputs are shown to be summed, the number of inputs can be
greater than two. The sign associated with each input indicates addition
or subtraction. The third symbol in Fig. 2-1 merely means an identity
and is often called a splitting point.

EXAMPLE 2E6

I
I
t,

As an elementary example of linear block diagram representations

consider the R-L circuit of Fig. 2E6-1. The differential equation of this
circuit is
di

+ L di = e

, " : '.

(a)

l)~

iJ< 1 f.

I-

')!'f'Ni'I~:vrh'tY"lt.(, .... !.

""....

'{,.l.l.l'01:I i",I:. Flg.2E6-2.

i.:: _

},}f,

:'~l:;."t~,

(6)

Block diagrams in the time domain'~ II; n.,

~ ",: .d ! ri li ~

circuit of Fig. 2E61 is shown in Fig. 2E6-2b.

from writingEq. 2E62 in the form

~,

oJ

'

'.

This block diagram resufts

).1 ;

'..~ t
-;)

.1,,'

I,

.;

:C'j

:.
, ..

;,../ =- 'I ( e -" L.pI.)

'., \ f
.J

"i,

" (2E6-4)

It is seldom necessary to rewrite the original equation in a form like Eq,

2E63 or 2E6-4 because with a little practice the block diagram can be
obtained directly by inspection from the original equation (or equations).
The block diagram derived above was in the time domain using the
operator p. The block diagram of the R-L circuit in the' c9 m plex fre
quency domain is also useful. In order to obtain it, Laplace-transform
Eq. 2E6-1. To maintain generality, the initial current in the circuit

"

Ri

t.

n~"

(26-1)

H
~":

134

ANALYTICAL TECHNIQUES

ELECTROMECHANICAL ENERGY CONVERSION

must be taken into account. By using Theorem 4 in Table 2-1 of Sec.

2.1.2, Eq. 2E6-1 can be transformed to

(2E6-5)

(R + Ls) /(s) - L i(O+) - (s)

where i(O+) is the initial current at t

0 +.

Now rewrite Eq. 2E6-5 as

(R + Ls) I(s) = (s) + L i(O+)

(2E6-6)

of a linear system are being studied in the frequency domain, the initial
conditions are normally assumed to be zero because system properties
such as relative stability, speed of response, etc., are not affected by the
presence of initial conditions.

"f:

~1.,

t'
:

In the examples of electromechanical energy converters that have been

treated previously it has been shown that, although the equations of

motion are nonlinear, much valuable information about the transducers

can be obtained by considering incremental operation around a steady-

...

in which L i(O+) is a constant and is considered as an input in addition

to (s). Draw a block diagram in the frequency domain by the same
techniques used in obtaining the block diagrams of Fig. 2E6-2a and b.

v.l;

't~

i(O +)

i(O +)

135

.1,,\

(a)

f
"~

XoYo

,;~.

I'

f---,--+ 1(8)

lis) ,:'*'11<

I,,"4:

-:'.1 ~", '.',

r---z'" Zo+ Zl

'l'j.

~,

(a)

.,.. ..

_
....

'" XoYo + }QxI+XOYI

Zo'" XoYo
zi

(b)

;f;
i.
,

Flg.2E6-3. Block diagrams in the frequency domain.

=Yo xI + XOYI
'I

YI
~"1I~'

(/)

-~,

The analogous two-block diagram representations in the frequency domain

are shown in Fig. 2E6-3a and b. ' Note that with zero initial conditions
" [i(O+) = OJ it is possible to obtain Fig. 2E6-3a from 2E6-2a and 2E6-3b
from 2E6-2b by replacing the operator p, by the complex frequency s.

Fig. 2-2. Linearizing product nonlinearity.

nonlinearity. (b) Linearized representation.

(0) Block diagram symbol for product

state operating point and by linearizing the equations of motion accord~

ingly. The Iincarization tcchniques can be represented pictorially in
block diagram form.

From the preceding discussion and example it is evident that for a

particular system there is no unique block diagram; rather, there are as
many different ways of drawing a block diagram as there are ways of
arranging the equations. When block diagrams are drawn in the time
domain, initial conditions are implicitly included and must be taken into
account in the same way that they would be in solving a set of differential
equations. On the other hand, when differential equations are trans
formed to the frequency domain, the initial conditions may be explicitly
il'lcluded in the transformed equations. Thus, a block diagram in the
frequency domain can, and in most cases should, include explicitly the
initial conditions which can always be represented as additional inputs as
shown in Fig. 2E6-3a and b of Example 2E6: When the general properties

The tirst, and possibly the most common, type of non-linearity

encountered in thc study of energy converters is the product type. This
consists simply of the product of two or more variables and is represented
pictorially as shown in Fig. 2-2a for the product of two variables x and
y. When it is assumed that the operation can be described by small
variations from a steady-state operating point, the three variables of
Fig. 2-2a tan be expressed as

,,

; .. ,i

x
Y
Z

= X o + Xl
= Yo + YI
= 2 0 + ZJ

(2-30)

",:,

136

",I

ELECTROMECHANICAL ENERGY CONVERSION

')~ J~

ANALYTICAL TECHNIQUES:_ .. ,

137

x\ . .. ,X" as shown in Fig. 2-30. For small variations aboiit a'n orerating

where X , Yo, and Zo are constl.'.nts and describe the steady-state (or
o
quiescent) operating point. and Xl. Y1' and ZI represent small variations
from the operating point. With the assumption that x" YI. and Zl are
small quantities, then to a first approximation the products of small
quantities can be neglected to obtain the product

point each variable can bc expressed as

= X~ + x~

xl<

).

(2':J4)

where k denotes the kth variable (it is not an exponent). By neglecting

, all products of two or more small variables, the product of all k variables
can be linearized by the approximate product

(2.31)

xy = (Xo + XI)(YO + YI);:::; XoYo + foX 1 + XOYI

. ,;;r, I' ... j1!

'

Since from Fig. 2-2a

xy

Xl
I

k_1

II

"

1<=1

1<-1

(2-35)

,X o

This linearized product is shown in symbolic form in Fig.2-3h.

r..:..... ,

\I

"

1<-1

nxk~nX~+ L~~

(2-32)

=%

n X;

Xl)

fl Xo. Yo>

Xl

,,2

z ={(lC.y)

lC

, ....

.J:~
,

'l

z .. n

"k

,:1:

10.,

(a)

X"

fI
X'

la)

k'
"I

!~r

---..---

Linearization of general product nonlinearity. (0) Product of n variables.

it follows that,

Zo = XoYo
Zl ~ YoXl + XOYI

M7

"'\"'!

,\

The block diagram symbol for the functional nonlinearity is shown in

Fig.2-4a.
For incremental operation XI> Y" Zt around a quiescent operating point
Xo, Yo, 20 a linearization of Eq. 2-36 can be made by taking a Taylor
series expansion around the point Xo, Yo and keeping only first-order
terms in Xl and YI' By so doing, the linearized functional nonlinearity
of Eq. 236 becomes

. L~).1.

Nonlinear

(2-36)

z = j(x,y)

= %1>

(o)

The second type of nonlinearity encountered

the study of electro
mechanical energy converters is the functional nonlinearity. The
functional nonlinearity with two independent variables is usually expressed
as

(b) Linearized representation.

and the linearized z is defined by Eq. 2-30 as z

y\

in

,,'

Ib)

k;

'.
Fig. 2-4. Linearization of nonlinear function of two variables.
function of two variables. (b) Linearized representation.

Xi

-----xr

It\

L
(b>

i_l

',r
Fig. 2-3.

%1

(2-33)

The linearized approximate product xy of Eq. 2-31 is represented in Fig.

2-2b. For small signal studies the nonlinear product of Fig. 2-2a can be
replaced by the linearized product of Fig. 2-2b.
In Fig. 2-2 a product of two variables is linearized for small variations.
This technique is equally applicable to the product of any number of
variables. For instance. consider the product of n variables Xl, Xl,

~ [(Xo Yo)

of
+~
(Xo YO)XI +
uX

Of

~ (Xo Yo)Yt
uy

'I
'~.'-

(2-37)

'

138

ELECTROMECHANICAL

~:rr

~~.

EN~RG~' CONVERSION

where [(of/oy)(Xo, Yo)) indicates the partial derivative of f(x, y)

evaluated at X , Yo. The linearized functional nonlinearity, Eq. 2-37,
o
is represented in block diagram form in Fig. 2-4b.
In a similar manner it is possible to generalize the linearization tech
nique of Eq. 2-37 to a function of n variables. The general nonlinear
function of the n variables Xl, x 2 , x3, ... , x n can be written as
(2-38)
z = f(x 1, x 2 , Xl, , x'1)

'}~

~.

i..i.i.;;;.:
,).1

~'~

..

:i)I;'~'

"~

"

xl

T __ oW mP'I' A2' 4

x3

+I

I
I
I
I
1
I

t+

\.._~:>.--

xl

--";;.;..--
.1

:t

".

' ,~

f(XI

O'

x 0'2 x O'3 ... , X"I

0'

x"
x 21

,.if

~. " ~

.l,... ,~ . '."'S

Fig. 2-5. Linearization of general

nonlinear function of n variables.
(a) General nonlinear function of n
,.variables. (b) Linearized representa
-tion.
" ~:

;I~:I

"t;)

: ...

,.

,.:1':

x~

II";

..

,~

l j")~

0;

,l;q,'~~' 1-};il,~:1t

. i:d

Kk =

ar

(Xl

O%1c

0'

X;'X 3
0'

,E:;1

1. Summation points

%11

/'>,
.

0'

'L]

In addition to these three rules there are several theorems which are
illustrated in Fig. 2-6. It is well to note that in all the theorems of

.,
j

gives the linearized equation

Xli)

+ k_IVX
i ~Z (X~, X5, xg, ... ,.Xo)~

See Hildebrand, op.'cit., p. 353.

!t,.~

~ ..

; \,
:

(-+ep-+) cannot be moved past splitting points

2. Summation points can be moved past summation pointSj).U

3. Splitting points can be moved past splitting points.

and its block diagram symbol is shown' in Fig.2~5a. Assume incremental

operation (xl, xi, x~, ... , x';) around a quiescent operating point
(X6, X5, XJ, ... ,Xo)' The general Taylor' series ;expansion for n
.variables, * keeping only first-order terms in the incremental variables,

z ~ !(X6, X~, X~, ... ,

(2-40)

(~-+)in the manipulation and reduction of block diagrams.

X")

(b)

:'\

84>

The stored energy W m is nonlinear, and the torque is nonlinear. In order

to ensure that no essential terms are omitted in the equations of motion,
the differentiations of the type of Eq. 2-40 which evolve to a given variable
must be carried out before the linearization process is applied. In other
words, if Wm is linearized before the differentiation of Eq. 2-40 is carried
out, the resulting expression for torque Twill not contain all the necessary
terms. Linearization of T, after it is evaluated by Eq. 2-40, is carried out
by the techniques given.
Whenever a linearized block diagram representation is obtained it can
be reduced to simpler forms by the use of a block diagram "algebra"
which consists of pictorial techniques for solving algebraic equations.
Three simple rules to observe in the manipulation and reduction of block
diagrams are:

(a.)

~ (

--

'.\

z =f(x l .x 2.r 3 ... .x",

f(xl.x2.x3. .x n )

"

'.'c

"i

139

The block diagram symbol for the linearized nth order functional non
linearity is given in Fig. 2-5b.
It should be apparent that the product nonlinearity can be treated by
the method used for the general nonlinear function; however, the product
nonlinearity occurs so frequently that it is designated by a special
symbol.
A word of caution is in order about the linearization techniques just
described. In the treatment of electromechanical energy converters the
coupling forces are determined by differentiation of some function with
respect to a variable. For instance, in the rotational device of Example
2E5 the torque is

'">'

x2

ANALYTICAL TECHNIQUES

l~~;

(2-39)

~t-1

~Ij
\ ~J

This rule limits the allowable complexity of block diagrams that can be reduced by
this technique. For more complex block diagrams reduction can be effected by insuring
that all transmissions between variables retained remain the same. For a systematic

method of reduction using this technique, see S. J. Mason, "Feedback Theory-Some

Properties of Signal Flow Graphs," Proc. IRE, vol. 41,1953, pp. 1144-1156.

'"

t~i

I"!O

(J"",

ELECTROMECHANICAL ENERGY CONVERSION

= A(X

~0

(1)

.,0

,~

",

"

>

-:. ...>

'j

~:-;

"~.F-<i .'iI';

,r";',l

l!
~

If. t,

~"

z ,.
-

ffi=t

~~

(2)

~z)

1"11

As an example of the appli,yation of the preceding ideas, consider the

problem of reducing the block diagram of Fig. 2.7 to, i,ts simplest form
and finding the ratio of output to input. The successive steps in the
reduction are shown in Fig. 2-8.

Fig. 2-6 the transmission between two variables is always kept the same;
for instance, in Fig. 2-6-(6) the theorem illustrates the equality
Y = AX Z

,~rt~ "jJ'?

'

ANALYTiCAL: TECHNIQUES

AH

Fig. 1-7. Block diagram.

"j

(3)

:B

Ii

{~]

(4)

(5)

~y

'*"

(6)

y
(a) ,

~=

x~

=~

(bJ

~..I

\\

. .c.

;~ IC'
,,-(4 ~'

y~-

'~1~

'"ii'",

~;i

I,

(dJ

x:~

XE~
+

\.11"

,t'f~

(7)

y
_

:~
-,

.,

Y -

-~

++

~"

I~b"

'In'
.'

Y.,

(e)

Y
1+ HC + A(l + H) I----"
A(l

+ 8)

'.'.

'

'."~

"

J)'
.1.1<~,

Fig. 1-6. Theorems for the reduction of block diagrams.

l+HC

Fig. 1-1.

Reduction of block diagram.

.\

"

,
'I"

142

."

EXAMPLE lE.7

'1

a
Nonmagnetic
sleeves

.J

away from observer

rxxx

Ab , which are the flux linkages with coils a and b

The flux linkages can be expressed as

'~~ . ".'~
'"

x means current

Cylindrical iron shell

-x-x-x"'"

1\ and

in terms of
respectively.

As an example of both the linearization and manipulation techniques

for block diagrams that have been presented, consider the doubly excited
magnetic field transducer shown in Fig. 27-1. This type of transducer
is often used to control valve mechanisms in hydraulic control systems.

'-1

143

ANALYTICAL TECHNIQUES

ELECTROMECHANICAL ENERGY CONVERSION

means current

Aa == Aa(ia, i b , x)

(2E7-1)

Ab = Ab(ia, ib , x)

(2E7-2)

The assumptions, approximations, and definitions in the preceding

paragraph completely define the transducer of Fig. 2E7-1. The first step
in the study of the dynamic operation of a transducer is the formulation
of the equations of motion. In this example, Kirchhoff's voltage law

will be used to write the equations for the two electric circuits, and

d'Alembert's principle will be used to write the equation for the mechanical

, circuit. The method of arbitrary displacement derived in Chap. I will

\; give the mechanical force of electrical origin.
In Eqs. 2E7-1 and 2E7-2 the flux linkages are specified as functions of
the currents. In writing the voltage equations choose the currents as
independent variables; thus, the mechanical force f~ of electrical origin
is most easily found from Table I-I of Chap. I as

" toward observer

'i-~.-"

/.

rO
W 'mIa'
(' ib
J.
_

.......

Flg.2E7-1. Doubly excited magnetic field transducer.

[
R

when>

~ .'

, the point ia , i b , x.

/ta(l)

ih

Since the magnetic coenergy is a

Choose the path indicated in the integral

~.

l"

tic

is the magnetic coenergy.

(2E7-3)

r state function, any path of integration may be used to evaluate W;" at

(:~~o_

:>....

W~

x)

AX

W:.,

f: ~a(i~,

0, x)

di~ +

J:.

Ab(i.,

ii"

(2E7-4)

x) dii,

...'

,J

This path of integration corresponds to first assembling the mechanical

system, bringing i a to its final value, and then bringing ib to its final value.
The currents iu and ib in Eq. 2E7-4 are independent of the mechanical
displacement x, Substituting Eq. 2E7-4 into Eq. 2E7-3 gives

"

Ilb(t)

Flg.2E7-1. Schematic representation of transduee'~.

J. - +

For the purposes of writing the equations of motion a schematic

diagram of the device is shown in Fig. 27-2. The necessary details on
the device are the following: The plunger (mass M) is constrained to
move in the range - d < x < d. The equilibrium spring position with
no electrical or mechanical excitation is x = O. The two coils a and b
. are identical and have the same resistance, R which is shown in the external
circuit of Fig. 2E7~2. The cylindrical iron 'shell and the iron plunger are
subject to magnetic saturation; thus, the magnetic fields must be described

flU O>'a(i~. 0, x) d"

o

"

uX

'a +

fl~ OAb(ia, ii" x) d"

0

aX

(2E7-5)

'b

0)

Equation 2E7-5 will be represented by the general functional notation

(2E7-6)

f. = f.(ia, ib , x)

The force in Eq. 2E7-5 or 2E7-6 is valid for all operating conditions;
hence it can be used in the mechanical equation of motion.

(:.... 'I;

"1

'~,.'t. ':
+- "

',~:

r;"."
"

...

~.:

:'{

~~,.
.~:1"

:\

The equations of motion for the system of Fig. 2E7-2 can be written,
using Kirchhoff's laws and d' AJembert's principle. as
t,,(t)

= i"R" + dAaUa'dlib, x)

(2E7-1)

dAb(i", ib, x)

(2E7-8)

ebU) = ibR

(I
I)

1',,(1)

';\

dl

eh(t)

.,

d2X
dx
K
r (. . )
= M -d + a-I + X -J. ,,,,lb' X
I2
( I

145

ANALYTICAL TECHNIQUES

ELECTROMECHANICAL ENERGY CONVERSION

144

) ... 't

'"

:t
'b

(2E7-9)

II

With the introduction of operator notation (p = dldt) Eqs. 2E7-7

2E7-9 become

e,,(t)

eb(t) = ibR/)

l(t)

(2E7-10)

= i"Ra + pA"Ci". i b, x)

+ PAb(ia, i b, x)

M p 2X +

apX

:'-'';',~".,

(2E7-11)

+ Kx - [,U", i b , x)

(2E7-12)

---fit) +

By using Eqs. 2E7-10-2E7-l2, a general nonlinear block diagram repre

senting the dynamic behavior of the transducer can be drawn. One such

j:

r(~~13

\:~

fl.

eb(t) = Ebo

f(/)

ib(t)

+ ebl(/)

X(/)

= F o + !t(/)

Eao

(2E7-13)

Eao

.~, ~,'

<,;.'

.. \

~"'I(~) =
;

.'.

EbO

,:';;.

'

-h

~rntJl .~

;:)";,

Fo + II(t)

M p 2X1

1.0 , XO)pib1 + T (l"o. J"o, XO)PX 1

tlX

ibJ)R

+ ~~b (l.o,

a:PXJ
~~

1'

(J>'b

+ -8 (1,,0, lbO, XO)PXl

.'~

(2E7-17)

(2E7-18)

K(Xo +

8/r (1

- -,;:- ,,0'
. 'vI",

i"

.!

lbO, XO)pial

'.

,:''&

(jl"

.
lbO, XO)Plbl

0/'(1QO' IbO, X)'-'

- -8
0 XI
X

Now substitute the variables of Eq, 2E713 into Eqs. 2E7-10-2E7-12.

make Taylor series expansions of all nonlinearities around the quiescent

tllb

tllb

(2E7-16)

'xo)pj"J

0>":

OAb

Ebo = IboR

Fo = KX o - f.(lao, lbo. Xo)

oA

~ (lQo.

+ ~ (laO'

(2E7-15)

I.:

+ lal)R +: ~;: (l.O"lbO.

+ ebl(t) = (If)O +

(2E7-14)

= laOR

10

(lQo

where all the capital letters are constants and are related by the equations
:,1

'.;

operating point. and neglect all second-order small (incremental)

quantities. The resulting linear equations are
, ,

i,,(t) = laO

= Eao + ea1(/)

'

.
' .:
>

.'
.t'l

Assume the variables to be described by

eit)

~ x

l.

)0

FI,. 2E7-3. 'General nonlinear block diagram,

block diagram is shown in Fig. 2E7-3.

In accordance with previous discussions these equations can be
linearized for small variations about a quiescent operating point.

+ ia1 (t)
= IbO + ib1 (/)
= X o + Xl(t)

XI) -

/'.(1,,0, lbO' X O)

IbO, X)'
af~(1 I X)'
0 1,,1 - -,.-, .0, bOo
Olbl
(II"

(2E7-19)

146

&:.>

r~.. %

:,f'- I,,"

ANALYTICAL TECHNIQUES

~'

ELECTROMECHANICAL ENERGY CONVERSION

Subtraction of the steady-state equations (Eqs. 2E7-l4-2E7-16) from

the complete linearized equations yields the incremental linear equations,

~./i<~

'.I
'1

,;j

eal(/) = ialR

\",
{,(

.]

;'

rr

.....,

! .~

+ 8a~a
(laO.
lb

p.

>0::

ebl(t) = iblR

lbO' XO)pial

+ ~-\,
(laO. lbO,
uX

IbO. XO)pib\

+ ~~b (laO. lbO,

a

..J

aA

'''s

0\"
~'"

~~
..;

><,0

oJ
g

~~1'

f"Ott)

'6

><

J
8

ho.,j

1~

'";!r

to
...

: Z(')j.t~T
.'

lot

Oi

E
...

'u

~H

.5
1;;
.St-l

J'

".~

I ~~

1';

;;

'~'1li

i.;

= Mp 2XI + apx 1 + KXl -

- 0;. (laO' lbO, Xo)ial

Via

XO)pX\

(2E7-2l)

i:

(laO. IbO XO)XI

aa!'Ib (laO. I bO XO)ibl

(2E7-22)

.These linear incremental equations are represented in block diagram

form in Fig. 2E7-4.
The linear incremental block diagram of Fig. 2E7-4 could just as easily
have been obtained by applying the linearization technique illustrated in
Fig. 2-5 to the nonlinear block diagram of Fig. 2E7-3.
The block diagram algebra described earlier can be used with Fig.
2E7-4 to obtain information about the incremental dynamic behavior of
the trar;o,]ucer of Fig. 2E7-1. An example of useful information might be
the response of the mechanical system to a small step in one of the applied
voltages. By applying the block diagram reduction techniques illustrated
in Fig. 2-8, the transfer function of interest can be found and from it the
desired response can be obtained. By noting changes in the transfer
function caused by changes in parameter values and quiescent operating
. point, a study of the small signal behavior over a wide range of conditions
can be made.

:0

... .... ;;

.I

,;

(2E7-20)

XO)pial

0\
+ -;(laO. lbO,
vX

XO)P1bl

XO)PXI

..>C

~I'"

11(/)

~'b

b
+~
vlb (laO.lbO.

Ii...

...e

J
;:I.!
;

+ ~~a
(lao,
Ul
a

:z,

147

.;.-'

I'J~I

'1
"'UJM"
00

Ii:

.+
t

.Ii-i'

.,...

+.

H\

,~.

'

2.2

Techniques for Studying linear Differential Equations

'~i

with Time-Varying Coefficients

'I

.I

.,.
,

. In the preceding sections some of the general analytical methods for

studying linear differential equations with constant coefficients have been

., ' discussed, and ways of linearizing nonlinear equations have been presented.

Unfortunately, such general techniques do not exist for studying more

,.:~

.--:-;

.,

1..''

;'-.",.

.. \'

;.

;,~.

',' .
.I
.' .c.;.

'" !-;" . .

::~,

;;
','i
'';

ELECTROMECHANICAL ENERGY CONVERSION

148

dnx

-,+
( I"

ANALYTICAL TECHNIQUES

complicakd differential equations. In some cases only restricted sets of

equations can be solved: in other cases solutions can only be obtained
hv machinc computation nr numcrical mclllllus.
In this section linear ditferential equations with time-varying coelllcients
will be considered. A general nth order equation of this type can be
written as

dn"'x

01(1) -,-I
I

til"

d n- 2 x

0,(1) -'-2

In-'

+ . , . + (/"

dx

1(1) -,
( 1

dnxu
-dt n

d"- I xu

(/1(/) - - -

d"-2'(,,
0,(1) - -

dl"--!'

In other words, any driving function /(t) can be decomposed into its
several additive comronents. and the solution for each component can
be found. Then all the components of the solution can be addect" to
obtain the complete solution for the driving function f(t). This is a
manifestation of the linearity of Eq. 2-41, which means that superposition
holds for linear differential equations with time-varying cocllicient~.
It should be evident that, if the driving function/(t) is made an impulse,
the response xU) will depend on the point in t at which the impulse is
applied because the coefficients (11(1). 02(1) ... 0,,(/) are varying with t,
Thus. although superposition docs apply to equations with tim.:-varying
coetncients. the superposition integral (Eq. 2-16 in Sec. 2.1.3) does not
apply to solutions of linear differential equations with time-varying
coefficients.
In the case of a first-order equation with time-varying coetncients

a,,(/)x = 1(/) (2-41)

.

The essential features of this equation are: it is linear because no products

of x and/or ib derivatives occur. and the coeflicients (J\(t). (J2(/) .. , o/t)
are functions of only the independent variable I.
That Eq. 2-41 is a linear equation can be verified as follows: Assume that
for a driving function j~(/) a solution Xl/(I) can be found. Then j~(I) and
Xl/(I) must be related by Eq. 2-41. Thus

,';

"~

5\

,,."

dx
dt

+ ...

dl,,-2

dtn

dn- 2Xb

dn-IXb

+ 1(1) -;Ji;;l +

2(1) dt,,2

o"(I)x,, =r,(t)

(2-42)

pet) = exp [J 0l(t) dt]

0n-!(I) -d
I

>

a/t)xb

= !b(t)

(2-43)

+ Xb) +
a, (t ) dn-I(x"
(h"-!

O,,-I(t)

d(xu

Xb)""

.J.

an(t)(xu

Thus a new input h.(t) can be defined as

",'

I" "

I.
.

( ) dn- 2(x" + Xb)

02 t
dt n 2

" 'r

~.

+ ...

Xb) = .!a(l)

+ fbV)

(2-44)
r

'I':;'"

'if)

'\ n

fc(t)

= fu(t) + fb(t)

. -r:'

(2-45)

which will yield the solution

xJt)

xu(t)

Xb(/)

(2-48)

In terms of this integrating factor the complete solution of Eq. 2-47 is

dXb

Equations 2-42 and 2-43 can be added to yield

dn(x u + Xb)
dtn

(2-47)

Such a solution is

obtained by the use of an integrating factor* p(1) defined as

I
X(t) = pet)

+
,

+ G,(t)x = f(t)

A solution can always be obtained, at least in theory.

dxU
0,,-1(1) dt

A different driving function fb(l) will have a different solution Xb(t) given
by Eq. 241 as

dnXb

149

p(t) !(t) dt

C
pet)

{~

(2-49)

where C is a constant to be determined from initial conditions. The

only ditnculty that arises in the use of this solution occurs when the
integrals of Eqs. 2-48 and/or 2-49 are ditncult to evaluate.
For linear differential equations with time-varying coefficients which
are of higher order than the first. no such general solution as Eq. 2-49
exists. If some of the homogeneous solutions [f(1) = OJ are known. the
order of the equation can be reduced. and ifall the homogeneous solutions
are known. the remainder of the complete solution can be obtained by the
method of the variation of parameters. t In the cases where f(t) can be
expanded in a series in powers of t. a series solution for x(t) can often be
foundt
In the following example the type of electrl'Jmechanical problems in
which linear equations with time-varying coeflicients occur will be
investigated and solutions obtained for some special cases.
Iliidchrand. Ofl. cil . Chap. I. Art. 1.4.
t HilJ.:b.. and. Ofl. C;I., Chilp. J. Art. 1.\1.
t Ihid. Chilp. IV,

(2-46)

t,

~.

...

..

..

151

ANALYTICAL TECHNIQUES
150

ELECTROMECHANICAL ENERGY CONVERSION

The electrical equation of motion can be written with the use of Eq.
2E8-1 as

EXAMPLE 2E8*
Consider the singly excited transducer of Fig. 2E8-1. The iron plunger
(mass M) is constrained to move in the range 0 < x < d. The equilibrium
position of the spring K is x = O. The resistance of the electrical winding
is included in the resistance R.

AI);":
r .,

"

>l~"

e(t) = iR

+ d>'

e(t) = iR

I'

L'(x) di
dt

+ ; dL'(x) dx
dx

(2ER-5)

dt

When the mechanical system is driven by a position source such that

the displacement x is an independently specified function of time x(t),
then the inductance L' becomes an independent function of time L(t) and
the electrical equation (Eq. 2E8-5) becomes

Iron

e(t)

R
'i

dt

;..

[R + dL(t)]
i + L(t} '!!.
dt
dt

(2E8-6)

Equation 2E8-6 is a linear differential equation with time-varying co

efficients. This equation will be subjected to some interpretation before
any solutions are attempted.
When Eq. 2E8-6 is multiplied by the current i(t) there results

Nonmagnetic sleelll

+ d~;t)] i 2(t) + i(t) L(t) d~~)

e(t) i(t) = [R

(2E8-7)

N:",

Fig. lEa-I. Singly excited magnetic field transducer.

The assumption is made that there is no saturation in the iron parts of

the transducer; consequently, the flux linkages with the coil will be linear
with current, and the inductance of the coil will be a function of x only.
Thus
(2E8-1)
A = L'(x)i
. With such a linear variation with current the magnetic coenergy w~ can
. oe readily found as
(2E8-2)
w:., = tL'(x)i 2
From Table 1-1 of Chap. 1 the mechanical force of electrical origin
for this transducer is

f. ... +
,

aW;"(i, x)

ax

= + !;2 dL'(x)
2

(2E8-3)

dx

The left-hand side of this expression is the electrical power input to the
system. The term Rj2 on the right-hand side is the power dissipated in
the resistance. The term (i2(dLfdt)] may appear to represent the energy
converted; however, it is not all converted. To show this write the
. ~: stored magnetic energy as
..:Ii,
~.
.of

Wm

iL(t) ;2(t)

(2E8-8)

from which the rate of change of stored energy is

;\

--=-1

dW m
dt

:;1 ~.

jl~'J,

I '2()
dL(t)
) .(
diet)
t- + L(t
l)t

dt
dt

(2E8-9)

Equation 2E8-9 can be used to write Eq. 2E8-7 in the form

.~.l

"

~,
~,

,-I

.1."

." "~.

e( t ) "'(" J = I'2( t )R

dLet)
dL(t)
di(t)
+ 2-1 1'2()t + -1 I'2( t ) + L( t ) 1.( f )
dt
2
df
dt

'--v-'

"----v----'

'----v----"

electrical
power
input

power
converted
to heat

power converted
to m~chanical
form

...----

(2E8-10)

rate of change of
stored magnetic
energy

Thus the mechanical equation of motion can be written as

f(t)

d 2x
M -2
dt

dx
adt

+ Kx

1 .2 dL'(x)

-1

--

dx

(2E8-4)

This example follows closely the treatment of N. H. Meyers, "The Singly Excited
Transducer," &.0. Thesis, M.I.T., Cambridge, June 1957. Chap. 5. Sec. 5.16.

The delineation of terms given in Eq. 2E8-10 leads to the interpretation

of the system in terms of the equivalent electric circuit of Fig. 2E8-2.
In this figure the flow of power among the various parts of the system is
made evident. Thus the time-varying resistance R c (which can be

'1

152

ELECTROMECHANICAL ENERGY CONVERSION

ANALYTICAL TECHNIQUES

nl:gative) can be used to describe the electromechanical conversion of

power. That the resistance Rc truly represents the power converted to
mechanical form can be proved by realizing that converted power is
given by
dx
+.r.,-,
(I

d to mec h
' I'"lorm
power converte
anlca
.

The general solution of Eq. 2E8-13 is found by substituting the right-hand

side of Eq. 2E8-13 as/(/) and Eq. 2E8-14 for p(t) into Eq. 2-49 to obtain
. = per)
C
l(t)

(2E8-11)

dx
dl

! i2 dL'(x) dx
2

dx

= !;2 dL(t) = i 2 R
2

dl

dl

R dl ] d/
L(r)

(2E8-15)

i(O) = p(O)

(2E8-16)

C = p(O) i(O)

(2E8-17)

from which
"~'.

Substitution of Eqs. 2E8-17 and 2E8-14 into Eq. 2E8-15 yields the
complete solution for i{t)

I d/.lt)

2 Jt-

/,(1)

(2E8-12)

"

Thus an interpn:tation of the electrical equation of motion, Eq. 2E8-6,

has been given and the equivalent electric circuit of Fig. 2E8-2 developed.

r e(/) exp [J

I Jo
+ p(t)

By prescribing the definite integral to start at I = 0 the constant C depends

on the initial current; thus, setting I = 0 in Eq. 2E8-15 gives

Using Eq. 2E8-3 for the force gives

+/.

153

/(,.=

..(I)

I d/il)
L' cJf'

.~

I,

i(t) =

1,(1) exp

ti'
,-~

I_{t)

dl

1 {P(O) i(O) + r' e(/) exp[fL

Jo

(1)

til] til}

(2E8-18)

,. ,(I,.

Electncal I Lossy part I

source I of electncal I
I system
I

Conservative
coupling field

~.'),

Fig.2E8-2.

As a next step consider the general solution just derived and use it to
study a particular problem. Let

Mechanical
system

L(/) = L(O)

Equivalent electric circuit.

e(t)

The next step is to solve the equation in the most general manner
possible. With the inductance a prescribed function of time L(t), Eq.
2E8-6 is a linear equation with time-varying coelficients. Division of
both sides of Eq. 2E8-6 by L(I) gives
iff

R + dL(t)]

di

di + [

L(t)

d l . _ e(t)
I - L(t)

(71:8-13)

Equation 2E8-13 has the same form as Eq. 2-47; thus, assume that e(t)
is a specified driving function, and find a solution for the current by
first finding the integrating factor (Eq. 2-48)

p(/) = exp

[f(

dL(t)

L(;~r

]
dl

= L(/) exp

[f r'&> dl]

bl

where b is a constant

(2E8-19)

and

EuoU - D)

..
Sf'

In this case

11' ,

Hence

.~

JL~/)
p(t)

Or,

dr

= L(/) exp

R,

[JL~t)

dl]

T If\\

e(t) exp

= In [L(O) + bl]R/b

dt

= [L(O) + blJ[L(O) + b/)~/b

=

[L(O)](Rlbl+l.

[J L~t) til] dl = J: EUo(1 =

(2E8-14)

I..

From Eq. 2E8-22 it is clear that p(O)

J:

I =

(2E8-20)

9;".

:~

f~.;.'. ~
\::r!

an impulse of value Eoccurringat

E[L(O)

D)[L(O)

(2E8-21)

(2E8-22)

Furthermore
hr]RI. dr

bD]Rlb u,{t - D)

(2E8-23)

The integral of Eq. 2E8-23 is readily e~aluated merely by noting the fact
that the impulse has a value only at I = D.

".,.....

ANALYTICAL TECHNIQUES

ELECTROMECHANICAL ENERGY CONVERSION

154

As the time t increases, the term L(O)/[L(O)

and the final value of the current i(/) is given by

Substituting expressions 2E8-22 and 2E8-23 into the general solution

Eq. 2E8-18 givcs the result
.
1(/)

i(O) L(O)(R/h/I!
rrfl"\\

L.li"Ih\L'

[L(O) ., hD]R/b u 1(1 [L(O) + bl](R/bl+l

D)

(2E8-24)

particular

homogeneous

The particular solution part of Eq. 2E8-24 is illustrated in Fig. 2E8-3

for two values of D. Note that the response to an impulse at t = D is
not the response to an impulse at I = 0 merely delayed D seconds, even

"

".~

."

.;!"

"
I,'

:,-'

- - - - L(D)

i(l)

J"

~.

! ~~

..

40)+bD-----------

.. '.

~ !"
'''\ ~. '

__ 1

Fig. 2E8-J. Effect of delay on impulse response of linear

time-varying system.

1t

[R

:I ,~r
/

+ h

(2E8-27)

which is a constant. An inspection of the parameter values of Fig. 2E8-2

will verify that this is so because the inductance L varies linearly with
time t and the time derivative of L is a constant. The changing inductance
reflects a constant resistance ofvaJue b into the circuit, yielding Eq. 2E8-27.
The derivative of the response to a step applied at I = 0 is not the
response to an impulse applied at t = O. This can readily be verified by
taking the time derivative of Eq. 2E8-26 and comparing the result with
Eq. 2E8-24 with D = O. Only in a system described by linear differential

equations with constant coefficients is the impulse response given by the

derivative of the step response.

It was pointed out earlier that linear equations with varying coefficicnts

of higher than first order, for which an integrating factor cannot in general
be found, can frequently be handled by assuming a series solution. As
a simple example of the power series approach, consider the problem of

finding the homogeneous solution for the current in the problem just
discussed. The equation to be solved is obtained by setting the right-hand

side of Eq. 2E8-13 to zero

.~

though the system is linear with L(t) prescribed. This illustrates the
earlier statement that convolution in the time domain, as it is usually
applied in the superposition integral to linear differential equations with
constant coefficients, cannot be used with equations having time-varying
co.efficients. The superposition principle can, however, always be applied
in a Iineor systcm whether the parameters are time-varying or not. Also
note that the impulse response is not an exponential as it would be if the
equation had constant coefficients.
The stcp response of the system is also interesting. The homogeneous
solution is of course unchanged. To obtain the particular solution, let

+ hi] becomes smaller,

t--+CX>

'ifi

solution

solution

lim i(/) = R

)i

ISS

'......

where L(t) = L(O)

written as

bl.

+ d~~t)] i + L(/) ~

(2E8-28)

Thus [dL(/)/dl = b, and Eq. 2E8-28 can be

~,

o = (R + b)i + [L(O) + bl] ~

'(

(2E8-29)

,h

The procedure for finding a solution consists of assuming that i(/) may be

expressed as an infinite series in powers of I and that the series converges

in some useful time interval. Hence, assume

\~,

e(l) = EU-J(t)

or..

(2E8-25)

i(t) =

2:

k-O

a step of height E occurring at t = O. Substitution of this expression into

Eq. 2E8-1 S and evaluation of the integral yield for the particular solution
.
1(1)

+b 1-

[L(O) ](Rlb)+l}
L(O) + bl
U_I(/)

(2E8-26)

t:tk lk

(2E8-30)

;\

i;11

where the t:t/s are constants to be evaluated by substituting this expression

for i(/) into Eq. 21::X-29. The derivative of i may als,) be written as a
series, since the series for i(/) can be differentiated term by term over the
interval for which Eq. 2E8-30 is absolutely convergent.

156

ELECTROMECHANICAL ENERGY CONVERSION

Thus
di

00

Ji = l:

k=O

The term for k

= 0 vanishes so

ukktk - I

There is one arbitrary constant, ao, in the solution as there should be

for a first-order linear equation. This arbitrary constant au is clearly
the initial current i(O). Thus

(2E8-31)

4'

:;'

that

di
-dl --

"

.( ) _ '(0) [I

I t

".
00

k 1
<:- ukkt
k-I

= uti Uk+l(k +

(2E8-32)

00

(lk+l(k

k~O

I)tkl-l-I

(2E8-33)

1)lk

}l..

Substituting Eqs. 2E8-30, 2E8-31, and 2E8-33 back into Eq. 2E8-29 gives

,1,

,:,l~

r.

{I'

00

o = k=O
L {uk(R + b + kb) + Uk+l[L(O)(k +

I)]}t k

.,

(2E8-34)

Each term within the brackets, i.e., each coefficient of a power of t,

must equal lew independently if the solution is to be valid over all time
from I = 0 to I = 00. Thus

o = ak(R + b +

kb)

Hence, a recursion formula for the

ak+1

-Uk

Uk'S

(R + b
L(O)(k

aHI

[L(O)(k

I)}

(2E8-35)

has been obtained

+ kb)
+ I)

for k

(2E8-36)

~..~~

,:l',V

"1>t

R+h

= -uoL(O).1

(X2

= -at L(O) .

,nt.1i
R

+ 2b

.. +fl',fo.,

!\J
~::~

i';.".\~ .:~\,:.:j.

~;.

:\~;

2b

+b

ao L(O). 2 . L(O) 1

~.t(J~,r\

''\ t)

R
a3

= -

(X2

3b
L(O) . 3 = -

3b R + 2b R + b
ao L(O) . 3 . L(O) . 2 . L(O) . 1
R

I,

L(O). I t

ak+2

,
,
"

Writing out a few terms as indicated by Eg. 2E8-36

ell

26

+b

L(O). 2 . L(O) . I I

.. ,

(2E8-37)

By the integrating factor approach, the hOl'llogeneous solution for the

same problem was obtained in Eg. 2E8-24, Expansion of the homo
geneous part of Eq. 2E8-24 by the binomial theon:m checks the result of
Eq.2E8-37.
The power series approach to the problem can also be used to obtain
the particular solution as long as the given' voltage e(l) can be expanded
in a power series about t = O. The impulse and step are not expressible
in this manner since they do not possess finite derivatives at t = O. How
ever, by rounding the corner of the step, a power series approximation to
it can be found, and by approximating the impulse by a finite pulse with
continuous derivatives a power series approximation to it can be found.
Although the use of power series solutions has been illustrated with a
first-order equation, the method is actually more useful with higher order
equations for which no integrating factor can be found. The principal
difference between the solution of the first-order equation and the series
solution of a higher order equation will be in the number of arbitrary
constants and therefore in the form of the recursion formula, Eq. 2E8-36.
It is known that the complete solution for a linear differential equation
of order n must contain n arbitrary constants; therefore, the recursion,
formula for a series solution must relate (n + 1) coefficients. For instance,
for a second-order equation the recursion formula will contain three
coefficients

00

= L

11;,

Sinee k is merely an index of summation. the value of the sum expressed

in Eq. 2E8-32 remains unchanged if k is replaced by (k + I). Hence
di
dl

(51

ANALYTICAL TECHNIQUES

'IQ

.,;"

,1

i'~.;:'

...
). . 1
..
'

"~

<)1 '

:tY

cxk+d

in which ao and al are the arbitrary constants. The arbitrary constants

are always evaluated from initial conditions in a specific problem. In
some series solutions more than one recursion formula may result;
however, the total number of arbitrary constants is unchanged. For
example, in the solution of a second-order equation two recursion
formulas might result
f( ak)

k even

ak+2 = g(ak}

k odd

CXk+2 =

;~;,

== f(ak>

In this case there are still two arbitrary coefficients ao and al to be

determined from initial conditions.

:~

,:

/58

2.3

ELECTROMECHANICAL ENERGY CONVERSION

ANALYTICAL TECHNIQUES

~;

of Fig. 2-9 bear a strong resemblance to the block diagram symbols in

Figs. 2-1,2-2, and 2-4. Of course, this is not accidental because both the
block diagram technique of Sec. 2.1.5 and the analog computer symbols
of Fig. 2-9 are used to pictorially represent differential equations. The

Techniques for Studying Nonlinear Differential Equations

From all the preceding examples it should be apparent that electro

mechanical energy conversion devices are in general described by nonlinear
differential equations. In those cases in which information obtained from
incremental linear equations is insufficient, the nonlinear equations must
be solved. No general analytical methods exist for solving nonlinear
equations; consequently, a scparate solution is required for each set of
excitation conditions of interest. For instance, if the response of a
nonlinear system is known for a unit step input and the response is
desired for a step input of a different value, the response for a unit step
cannot be multiplied by an appropriate constant to obtain the required \, I~
solution. Instead, the complete problem must be solved again for the'
new stcp input.
The study of electromechanical devices and systems in which nonlinear
equations must be solved requires the separate solution of the equations "'
for all the inputs of interest. The solution of nonlinear equations can be
accomplished by numerical means; however, this method is often time
consuming. Thus, for the solution of nonlinear problems use is sometimes
made of digital computers which perform electronically all the manipula- '
tions of the numerical methods or of analog computers which simulate "
the actual system with electrical or electromechanical components.
Other methodst of studying the dynamic properties of restricted classes
of nonlinear problems are phase-plane techniques, describing functions,
quasi-linearizations, and statistical methods. Most of these methods are
chiefly concerned with the consideration of stability in nonlinear systems
and are beyond the scope of the present discussion.
Analog computers are very often employed in the study of electro
.mechanical devices and systems, and a short discussion of the use of an
analog computer to study how a simple electromechanical device can be
simulated is in order. The analog computer consists of components of
various types that can be interconnected to simulate the operation of a
physical device in the time domain. The symbols for the various building
blocks inc shown in Fig. 2-9, and the function which each component
performs is illustrated clearly. It is interesting to note that the symbols
Hildebrand, op. cit., Chap. Ill.
t A. A. Andronow and C. E. Chaikin, Theory 0/ Oscillations, Princeton University

Press, Princeton, 1949. J. G. Truxal, Automatic Fudback Control System Synthesis,

McGraw-Hill, New York, 1955. R. J. Kochcnburger, "Limiting in Feedback Control

Systems," Trans. AlEE, Vol. 72, Part II, Application and Industry, 1953, p. 186.

Kan Chc!l, "Quasi-Linearization Techniques," Sc.D. Thesis, M.I.T., Cambridge, 1953.

159

"
. , '.~"",

, ";'~'.i:"

x(1) ,

Adder

,vlt)

>.;

+1

"'

+ y(l)

[x(1)

- z(tlj

z(1) (

I~

Inpuls

g,

0,,11'''1

,
"of

~i.'."
','

~'

:{ J
.

: : : [;> '"

Multiplier

-;..

~;'r.

if

..

,~

Inputs

J,ro

Output

;.,
I,',

-1'"

<I:,

,(')0

Integrator

y(t)

Qoo----V

{[,ro: ,)jd,

Inputs
~v

,'/ ,;

Function generator

,j.

....

{,

x(I)O>------I

Input

l~

r>

Output

,1

Coefficient unit

'x

~(x)

x(t) 0 > - - - - - f)

Input

('\

Output

Fig. 2-9.

Kx(l)

Output

Analog computer symbols.

difference in nomenclature, which resulted from two different working

groups attacking the same problem, is unfortunate.
Note that among the symbols listed in Fig. 2-9 there is an integrator
(lIp) but no dilTerentiator (p). This Occurs because in practice it is
difficult to build a good differentiator as differentiation enhances stray

,~~~

,.r

l!'

160

1.

ELECTROMECHANICAL ENERGY CONVERSION

j~~",

ANAL YTICAl TECHNIQUES

'~"

';11'

noise. The lack of a dilrercntiator docs not limit the usel"ulness of analog
computers because any differential equation can be changed to an integral
equation.
To show how an analog computer can be used, consider a simple
second-order, linear diffen.:ntial equation with constant coeflicients.

161

. I

xIII"
.~-"

d"\

-;I-~

~;J

i,

tl 2y
-,
2

(,1

dy

A -dI

By = X(f)

(2-50)

dv(ll
-dr"

This particular equation can be solved without resorting to a computer

since it is linear. but it will demonstrate how the computer operates.
Suppose for the moment that .1'(1) and dy(/)jdf arc known quantities.
X(/) is a given quantity; therefore, d ZY(I)JdI 2 can be found.
This is seen
by rewriting Eq. 2-50 in the form
(j2y
dy
- 2 = X(/) - A - - By
dl
df

la'

xltl c

(2-51)

(hi

x(tlc

'':

J'f

M tl 2x
-dl 2

dx

(X-

til

dL'(x)
x - -1I'2 2
dx

dl

~y(ti

.I

Solution

(e)

:i., '

'-,:
"

(2E9-1)
(2E9-2)

.t!l.

dt1.

'~'~",,<'

':(,

1:( )

d 2y

function

,~~~

In this example the analog computer representation for the singly

excited magndil: f1dd tran~dul:cr of Example lEg will be obtained. With
the assumption of no magnetic saturation in the transducer of Fig. 2E8-1,
the inductance is a function of x only; L'(x) and the equations of motion
for the dcvice were derived in Example 2ES as Eqs. 2ES-4 and 2E8-5.
These equations are

. I

Orlvlng

EXAMPLE 29

+dLl'(X)dX].
-dx- - cit
I + Jl,()di
Xtil

. I
)'--.,r--o'!l.
dt

Equation 2-51 is set up in computer symbols in Fig. 2-IOa. If d 2yJdl 2 is

next integrated, dy/clf is obtained. tly/tli can then be fed back to provide
part of the information which was originally assumed to be known.
This step is shown in Fig. 2-IOb. As the final step in the solution, dy/dl
can be integrated and the resulting y(r) signal fed back to provide the
remaining information that was initially assumed known. This step is
shown in rig. 2-10('. and the computer representation for solving Eq. 2-50
is complete. The driving fUllction x(t) is fed in, and the solution y(l)
is ta ken out after the second integration, as shown in Fig. 2-lOe.
The procedure just explained for solving a simple, linear differential
equation can readily be extended to more complicated cases. In nonlinear
problems. additional components such as multipliers and function
generators are needed, but the basic procedure is essentially unchanged.

()
et=
{R

.v(t'

Fig. 2-10.

Solving a linear equation with an analog computer.

The analog computer representations for these equations are shown

in Figs. 2E9-1 and 2. These rqm:sentations wen: obtained in a manner
similar to that used with the linear system of Eq. 2-50 and Fig. 2-10, with
the addition of the nonlinear components-the multipliers and the function
generators. When the two parts of the computer representation are
connected as indicated, any driving functions e(1) andl(t) can be applied
with any initial conditions desired, and the resulting response in i(r),
x(r), or any of the auxiliary variables can be mca~ured.
The use of an analog computer is essentially an electronic or electro
mechanical simulation of the actual device to be studied. It is evident

ell) = JR
L + 'i.~.
dx

ell)

,~

ELECTROMECHANICAL ENERGY CONVERSION

162

+ I

L'(x)

ANALYTICAL TECHNIQUES

tl!.]+
1:(%) !i.i.
dl
dl

from Fig. 2E9-1 that the several parameters governing the operation of
the device retain their identities. Consequently. before a device is ever
constructed, its operation can be simulated, and the effects of the various
parameters on dynamic performance can be studied. The analog com
puter is a useful device in optimizing the design of both linear and nonlinear
systems.

Jor"

di

dt

'i

ill)

163

To Fig. 2E9-2

i
2.4 Summary
,[

, R

~(

dL'(x) dxl

+ ---;IX TtJ

This chapter has presented many of the analytical techniques that are
used in studying the dynamic behavior of electromechanical devices and
systems and has illustrated them with transducer examples.
Although practically all electromechanical devices are described in
general by nonlinear differential equations, important and useful in
formation can often be obtained by linearizing the equations of motion
with the assumption of small signals. Thus, several techniques for
handling linear differential equations with constant coefficients have been
given.
Under certain types of physical constraints an electromechanical device
may be described by linear differential equations with time-varying
coefficients. Consequently, the techniques available for study of this
type of equation have been presented.
When the large signal behavior of an electromechanical device is of
interest, the original nonlinear equations must be retained. In this case
no general analytical method exists for studying the equations, and
recourse is usually made to machine computation. The most common
computing machine employed for electromechanical devices is the analog
computer, and a short description of its use has been given.

I
L'(x)

"I

From Fig, 2E9-2

dU(.) dx

dx

dl

From Fig. 2E9-2

'';I'"

Fig.2E9-I. Analog computer representation of Eq. 2E9-1.

,"

,)

"
'It)

I'

+ '2'
I ,2

dL'(.)

dx

=M (jiT
d-'x + K:c + a dx
Tt
..

I(t)

~\~

ill)

"~I'

From

Fig,2E9-1

tlL'(s)
dx

PROBLEMS

tlL'(x)
dL'(x) dx
~'di

---;IX

1'1

Ii ,

2-1. The pressure pick-off shown in Fig. 2P-Ia has a nonlinear dielectric

with the characteristics shown plotted in Fig. 2P-Ib. The curves are described

analyt.cally by:

"

To Fig. 2E9-1
1

171ii
To Fig. 2E9-1

i '

where f3 and qm are constants and v is the voltage across the capacitor plates.

%(t)
I

Fig. 2E9-2. Analog computer representation of Eq. 2E9-2.

.'
,'~.

"j":"'"

.~"

q(v, x) = qm(1 - e-/Ju)

J',.,i'

",-<;:--,'::' . . "

1
I

(a) Find the equations of motion for the system.

(b) Linearize these equations for small signals about an operating point.
(c) With the supply voltage v held constant, find the response of x and
a step in the applied force f. Assume that the system is underdamped.

ito

,
,,'
'<I,'

,'.

2.2. A chunk of material of mass M is suspended from a spring with a

stiffness of K newtons/meter, as in Fig. 2P-2. The length of the spring is 10
when there is no electrical excitation. The material is diamagnetic (iJo < iJoo)
The" versus i curve for the mechanically fixed coil may be assumed electrically

': . .
____

II v(l)

--i.-

J+

~~~AIJ<

'" 'I,'

It

g~

~' '\

...W
~,;~""
"
Viscous damping

;~ Plst?n

165

ANALYTICAL TECHNIQUES

''': 1

ELECTROMECHANICAL ENERGY CONVERSION

164

of mass

L;

~~

Mass
IJ.

=M

<""0

.+

;~'.

Ci

=M
:~

..'
~

"

'1,:Jk
;:
'"t
Nonlmear~

~'~, '!1

I(t)

flUid dlelectnc

jtt'.

Spring constant

=K

ifI' ,

t:i', ',t';~"",
l:i
I

'

J4~

J';r
f

(a)

.Ii

."
.;.

.:I"~r~'j

;~,.

;;:..

<It

~<:.~>

,l,

,~

Flg.2P-2.

linear, hence ,\ = L(x)i. A current source i is attached to the coil terminals.

The inductance may, in addition, be assumed to be a linear function of position,
L(x) = L o Ax, where L o and A are positive constants.
(0) Derive the equations of motion of this device.
(b) Linearize the equati~ns of motion for small flignals about an operating
point.
,'
(c) find the response of'x to a small step of current applied at I = O.
2-3. A vibrator driven by an electromagnet is to be constructed as indicated
in Fig. 2P-3. The light rigid beam plus load are equivalent to a mass of 10 kg
at the magnetic axis. The beam is supported at one end by a pair of springs

.,~

_,m

'4.~

J~
.~ ";~

~
""

Note '/ and II are the

charge and voltage
respectively of the
condenser

Xl

< x2<x3 . ;.; l: f!

~"

~:'

,.,

..

'; ~,

"

,:
II

(b)

Load

:~;'(~'

~~

,.

~f

x
~
I

.~ '~f:
i:

. .;d.a
1,\ ':W'Htw

,~~Jl"

\',

Fig.2P-I.

Fig.2P-J.
"

...

ELECTROMECHANICAL ENERGY CONVERSION

166

ANALYTICAL TECHNIQUES

that have a total elastance equivalent to 40 x 10) newtons/meter, and the

mechanical damping factor is equivalent to 50 newtons/(meter/sec), both
referred to the magnetic axis. The coil has 5000 ohms resistance and its
inductance is
L, = 0.050/x henrys
when x. the gap opening, is expressed in meters. Leakage flux may be ignored.
Spring force is zero when x = 6 X 10-) meter.
(a) Write the equations of motion of the device.
(b) Linearize these equations for small signals near a quiescent operating
point.
(c) Obtain the transfer function from applied voltage v to position x.
(d) Considering v as the input variable and x as the output variable. sketch
the frequency response of the system to scale.
2-4. The cross section of a cylindrical solenoid used to control the valve
mechanism of a hydraulic servo system is shown in Fig. 2P-4. When the

(a) Find the equations of motion.

(b) Assuming that Ai ~ 10 and the displacement

x is small, linearize the

equations of motion.
(c) Is the system stable for all values of l o ?
(d) Draw a block diagram for the system. assuming Lli as input and x as
output.
(e) Find the response in x to a small impulse in Lli.
(f) Obtain the frequency response for the device. Sketch and label Lhe
response assuming the system to be underdamped.

.my

25.

In the electromagnetic balance shown in Fig. 2P-5 the capacitance is

CO

C( X 2) = d -

"

.t:~

":",

Cylindrical Iron shell

r--

)f

means current
... 1.

away from observer

= Lo +

AXl

1
2

Li'
, ,

Non magnetic
sleeves

X2

and the inductance is

L(xl)

---,

167

,I.

..... .....

,.

"",

L.r6'~66"~r-J._--"

rs~ringK

Iron plunger

~%l

r606"6'~
SpnngU~
""/kM""W""*fm""~",,tM~%

vi

;j',

tffl.$!

means current
toward observer

I"'~.':

C(Xt)

~d

"'-MechaniCally
fixed

Mechanically
fixed ------

"

Fig.2P-4.

currents in the two identical coils 1 and 2 are equal, the plunger is centered
horizontally as shown in the figure. When the coil currents are unbalanced,
the plunger moves horizontally a distance x. The nonmagnetic sleeves in the
two end gaps keep the plunger centered radially. Their thickness is g, and their
mean radius is r, where g ~ r. The coefficient of friction is a, and friction is
assumed to be viscous. The mass of the plunger is M. The spring constant of
each spring is K. The permeability of aU magnetic materials is assumed infinite,
the permeability of the gaps formed by the nonmagnetic sleeves is p.o. and
leakage and fringing may be neglected. The resistance of each coil is R, and
the number of turns is N. Coils 1 and 2 are assumed to be excited by current
sources
i l = 10 + Ai
i 2 = 10 - Ai.....
where 10 is constant and Ai is time-varying.

.,...L

"

"

'.1

'.i'.' .
,
:,,1:

~,
L(x )

NW

L.L

Fig.2P-5.

,
i

The uniform beam has total mass M and the spring forces are zero when
The motion is limited so that
XI ~

r'

2;

X2

I.'~

'ill"
~~
;1'

' ':

balance is possible with 8 = O.

this a stable equilibrium?

O.

(a) Find the equations of motion for the device. \

(b) With a sinusoidal applied voltage v = V sin

whic~ a
(c) Is

e=

:~ '.

;;':'

/.;-=

W'.

, '~
j

find the frequency

Iv

at

2-6. The V0!t<1ge indic<1ted in Fig. 2P-f,a is arr1ied to thc series R-t circuit
shown in Fig, 21'-6/1 having a time-varying inductance given by LU) = L o + Kr
where L o = mductance at I = 0
K = constant (henrys/sec)
r = time (sec)

'.

t,r..'......,,.
168

ELECTROMECHANICAL ENERGY CONVERSION

(a) Obtain the solution for ;(/) in closed form.

(Hinl: Apply superposition).

(h) Check the solution for the ramp component of part (u) by a power series

,
~

.~':'.}

~",

~":"

.'
e(t)

"

and a constant velocity source of value v meters/sec is applied to the mechanical

terminal pair.
Find at each instant of time in terms of E, v, and I:
(I) The instantaneous electrical power supplied to the transducer.
(2) The instantaneous power converted 10 mechanical form .
(3) The instantaneous rate of increase of field energy.
2-8. For the singly excited magnetic field transducer shown in Fig. 2P-ll it
is experimentally found that

..
1':

solution.

169

ANALYTICAL TECHNIQUES

"lot""

l( ')1/2

A =3

,,";Pi

webers

where a = 102 meters' 2 This representation is v;lIid for 0 < i -: ) amp and
x < 0.04 meter. The spring constant is K = IO~ newtons/meter and the

o<

..

1=0

t=2D

=])

'--

KIJ:x

...R

,.

F..

(a)

e(l)

0{

J'i,J .

Lit)
:',

.~
l

(h)

.,

Fig.2P-6.
Fig.2P-8.

2-7. The internal behavior of the singly excited magnetic field energy
converter shown in Fig. 2P-7 is characterized by the equation
i

= (AX)2

.~

...

The resistance of the coil is negligible.

(a) Find the mechanical force exerted in the x direction by the transducer
as a function of A and x.
(b) Before 1 = O. both A and x are equal to zero. At 1 = 0, a constant
voltage source of value E volts [e(/) = E] is applied across the electrical terminals

c"

,'\:

r----- ---I
I

I
I
I
I

~x

,L...

Magnetic field
transducer

Fig.2P-7.

spring force is zero at x = O. The mass of the armature and the resistance of
the coil are both negligible.
Prior to t = 0 both x and i are zero. At t = 0 a voltage source

:
-.II

'
J
\'
. . .,
~

','

;-.'.

Vsinwt

where V = 75 volts and w = 21110 rad/see. is applied to the terminals.

(a) Find A(/), X(/). and i(l).
(b) Sketch and dimension the waveforms of A, x, and i for the first two
cycles of operation.
2-9. Using the results of part (a) of Prob. 2-3 drawn an analog computer
diagram for the vibrator of Fig. 2P-3.
2-10. Using the results of part (a) of Prob. 2-5, draw an analog computer
diagram for the electromagnetic balance of Fig. 2P-5.

",.,...

~-

THE GENERALIZED ROTATING MACHINE

.,

.;: ;' I ' .

'~~.; .

'I

..

I(

The Generalized,
Magnetic Field Type, Rotating,
Electromechanical' Energy Converter

/i
':j~

,.
"
;~)

f:
{:;
3.0

Introduction

Gross motion types of. electromechanical energy converters occur in

innumerable configurations and have almost unlimited applications, as
electrical power sources, ~echanical drives, control devices, and signal
transmitters and receivers. The physical configurations which have
proved to be the most versatile are those which produce rotary motion.
These are usually called rotating electric machines, and are industrially
the most important of the electromechanical energy converters. Electric
machines are the source of nearly all commercially available electrical
power. In addition, the electric machine as a controllable source of
mechanical power is one of the most useful tools available to modern
industry or to man in his'everyday living. The availability of any desired
size of electric machine, the relative ease with which it can be controlled,
the wide range of performance characteristics which can be obtained, and
the relative abundance of electrical power, plus the simplicity of transmission to the point of lftilization, are all factors which have made the
electric machine one of the most important devices in the modern world.
Electric machines range in size from very small, 1/300 horsepower or
less for indicating instruments, to 1/20 horsepower or above for small tools
and home appliances, to 1 horsepower or over for miscellaneous industrial
tasks, to 3,000 horsepower or above for blooming mill drives or electric
locomotives, and over 350,OOO-horsepower electric generators for large
turbogenerating stations. The range of horsepower sizes in commercial1y
available electric machines exceeds a factor of 10 8 , and within the range
of sizes there are various commutator and noncommutator machines

t't

I
.. lI'1

~~;

;'.

.f

"': ';t

I.:Y\'

'. J.r ~: '"

I'

..

'.'',: ....." i

:;~..

170

....

.;7.-.
: .....
""of

'".

171

of both synchronous and asynchronous types. It is indeed interesting

that all these machines operate on the same principle; namely, the tendency
of two magnets to align themselves. Thus, an analysis of an idealized
structure of one electric machine will provide the essential concepts
necessary to understand the operation of most practical electric machines.
Only two-phase machines will be considered in this treatment. The
interest is primarily in energy conversion and, therefore, in torque production in machines. It is shown in Chap. 10 that, regardless of the
number of phases on stator and rotor. any machine can be reduced to an
equivalent two-phase machine for the consideration of torque production. *
Thus, the simplest physical device, a two-phase machine, is used to
illustrate fundamentals; but the treatment is completely general when
used with the transformation techniques of Chap. 10.
In general, electric machines consist of two sets of windings or coils
in which one set of coils can rotate with respect to the other. The
mechanical motion between these two sets of coils is generally restricted to
one degree of freedom, which, in most instances, is rotary motion. Consequently, a rather general type of rotary motion machine, which will be
called the generalized magnetic field type of rotating electromechanical
energy converter, or simply the generalized machine, will be considered.
In essence, this machine consists of a pair of concentric magnetic cylinders
with magnetic saliency on one member and with two sinusoidal current
sheets located in electrical space quadrature on each of the magnetic
surfaces facing the common air gap. This physical configuration is
representative of most electric machines; therefore, this model will be
used to develop the dynamic performance characteristics of electric
machines.
The study of electric machines in this chapter is based on the fundamental consideration that machines can be viewed in terms of sets of
linear lumped circuits in relative motion. The existence of relative motion
between the two magnetic members of the electric machine (statorstationary member; rotor-rotating member) raises the question of which
of several coordinate systems (reference systems) should be used to express
the equations of motion. There is considerable latitude in the choice of
a coordinate system for studying any particular device and the specific
selection is usually governed by the individual problem and the desired
form of the equations of motion. If a Lagrangian function is to be used
in formulating the equations of motion, it is necessary to .choose true
The reduction of any n-m phase machine to an equivalent two-phase machine for
considerations of torque production is only possible if the imt>edances of all stator
phases and all rotor phases are equal on a per-unit basis. The excitations for such a
reduction need not be balanced.
.~"tt -atlli

172

,
~

THE GENERALIZED ROTATING MACHINE

ELECTROMECHANICAL ENERGY CONVERSION

coordinates. The equations of motion of the generali7ed machine will

be determined using trtll: coordinates, after which these equations will be
expressed in krms of alternate sets of true coordinates and also sets of
quasi coordinates. The several coordinate systems used in this chapter
(and in subsequL'nt chapters) will be shown to be valuable for many
specific studies of various types of electric machines.

I pole

,.

~
!

t.

.j
~

~ .. ~.

ir~

</>.12:

f'

3.1

173

The Generalized Machine

Electromechanical energy conwrsion occurs in a magnetic field type of

converter whenever the stored magnetic energy changes with mechanical
position. In a Singly excited system (i.e., one current-carrying coil) this.
change in stored energy is caused by a change in the configuration of
magnetic boundaries with mechanical position. When no change ill',t
magnetic configuration occurs with a change in mechanicai position,
there must be a change in the relative positions of two or more currentcarrying coils. Thus, it is possible to build electromechanical energytlVtUf,.
converters with one or lTIany current-carrying coils which are excited from
one or many independent sources.
Practical machines are constructed in a variety of ways. One method
N:i
of construction is illustrated by the developed axial view shown in Fig. 3-1.
fall
This structure contains four sets of concentrated coils (only one coil per:;;:
set is shown), commonly called phase belts. Two of these phase belts of;"
arc located in slots in the stationary magnetic structure called the stator,
and the other two in slots in the rotatable magnetic structure called the
rotor. For purposes of analysis the magnetic boundaries and phase belt
winding concentrations of Fig. 3-1 will be smoothed to the continuous
structure of Fig. 3-2 in which the phase belts are sinusoidally-distributed
conductor distributions and the polar projections of the stator are
accounted for by an angular-dependent radial permeability.
Even though the physical model of Fig. 3-2 may seem to be a drastic
approximation of the structure of Fig. 3-1, it will develop that the structure
of Fig. 3-2 is entirely sulfJcient for studying the fundamental dynamic
behavior of most rotating machines.
For the generalized physical model of Fig. 3-2 the physical magnetic
structure and the arrangement of the phase belts are assumed to satisfy
the following conditions:
'J<:j
t.J('~
t!
I. Electrical space angle ef;. and mechanical space angle ef; are related
by the equation '/',. = (PI2)</> = lief;, where </>. = electrical space radians, Hlil
'" = mechanical space radians, P = number of poles in the machine, and

2 poles. 2'11' elect. radians

=~~ mech. radians

Fig. 3-1. P-po!e, two-axis, salient-pole machine. Note: </>' =

Current sheet phase a

KQ' (</>' )

:~.'

l t

j"

~. 't."
t

</>; P/2 = n.

4>J -

Current sheet phase b

Kt(</>) =ibZ~COSn(<f>-f.)

=i:Z~ cos n</>

</>'.0
f

l.

.. ~;Ai
K ;(</>') = i;Z; cos n(</>' -f,a

~.:1
'~ I~
'.

;4~

R
j

.;:

Fig. 3-2. Current sheets produced by sinusoidal distribution of coils in generalized

machine. Note:.p' = </>. - </>; j.radl.l = p. - 1'-2 cos 2,,</>'.

...

-.
j

r'

174

= number of cycles in a current sheet. A pole can be defined as the

region in which there exists a unidirectional radial magnetic flux, or as ;
half cycle of a sinusoidal current sheet. A fundamental pole span equalSthe span of a phase belt, i.e., one pole span is 17 electrical space radians.
Poles must appear in even numbers, 2, 4, 6, etc., and a pair of poles
consists of one north pole and one south pole.
2. There are two phase belts on the stator and two on the rotor per
pole. A phase belt is assumed to produce a sinusoidal space distribution
of current density. A sinusoidal current sheet results when the turns in
a phase belt are sinusoidally distributed in ~pace and span 17 electrical
space radians. The generalized model is assumed to have its turns
sinusoidally distributed in space, so that a current i into the terminals of ..
a phase belt produces a surface current density K(ef = iZ cos nc/>.
3. The phase belts on a given magnetic structure, either stator or rotor,
are located in electrical space quadrature; i.e., the magnetic axes of the
phase belts are displaced 17/2 electrical radians or 17/2n mechanical radians.
4. The air gap is the region between the stator magnetic structure and
the rotor magnetic structure. One side of the air gap is assumed to be
magnetically .. smooth"; the other side of the air gap has magnetic
variations (polar projections), but these variations exhibit magnetic
symmetry about 17 electrical space radians or 17/n mechanical radians.
5. As a consequence of condition 4, all magnetic variations in the air
gap produced by slots used in practical machines for placing windings
in the magnetic structure are neglected. This, of course, does not exclude
magnetic variations on only one side of the air gap which repeat every 11
electrical radians, since this variation, which, typifies a salient-pole
machine, was allowed by condition 4.
.
6. For salient-pole machines, the field intensity will be derived b y , 1
.
assuming a uniform air-gap length g equal to the average gap length of
the salient structure. The nonuniform air gap ot:the salient-pole structure
will then be accounted for in the idealized moJel by defining a spacedependent radial permeability, fLradial = fL- fL2 cos 2nef>-.
7. The phase belts are not connected to commutators or slip rings and
all parameters and excitations will be defined for each phase belt at a set
of terminals that is stationary with respect to its magnetic structure. That
is, the terminals of a phase belt on the stator are stationary in space and
the terminals of a phase belt on the mtor are stationary with respect to
the rotor, but rotate in space with respect to the fixed reference frame of
the stator.
8. There is an equal number of turns in each of the pair of phase belts
located on the stator. Both rotor phase belts have the same number of
turns.

ELECTROMECHANICAL ENERGY CONVERSION

THE GENERALIZED ROTATING MACHINE

9: The magnetic material of both rotor and stator is described

magnetically by a linear B-H characteristic with a very high relative
permeability that docs not exhibit saturation. This assumption makes all
the lumped electrical parameters independent of electrical variables;
however, these coefTicients may be functions of mechanical variables.
10. The stored electromagnetic energy used to descrihe the machine is
taken as the zero order or static magnetic field energy only. and the
electrostatic field energy is neglected. The elimination of stored electric
field energy neglects the effect of intrawlnding and intcrwinding
capacitance. Any electric fields produced by time-varying magnetic
fields or by relative motion in a magnetic field are not included in the
energy function which describes the system. These electric fields will be
included as a result of the formulation of the equations of motion from
the system energy function containing the magnetic field energy.

',.

i~~~"

,[;
"

'~"

1:

3.1.1

~..i.'
,."

..

j",<
~,;:

Dynamic Variables for the Generalized Machine

The physical description of the generalized machine must be reduced to

equivalent parameters such as resistances and inductances as seen from
the four electrical terminal pairs, and viscous damping, stilTness, and
inertia* as seen from the mechanical terminal pair. To estabhsh these
parameters one must clearly define the variables which will be used to
describe the motion of the generalized machine. Choosing the equivalent
circuit representation of the generalized machine shown in Fig. 3-3, which
contains the four coil currents as electrical variables and the shaft position
as a mechanical variable, one can define all the system coordinates and
velocities. The generalized coordinates and velocities for the generalized
machine, which has five terminal pairs, four electrical and one mechanical,
are shown in Table 3-1.
The variables defined in Table 3-1 include the generalized coordinates,
velocities, forces, and momenta. Only two of these four sets of variables
(which are the five q's, its, p's, and !,s) can be chosen as the independent
variables used to describe the generalized machine. The generalized
coordinates qk'S and generalized velocities ilk'S are chosen as independent
dynamic variables in this treatment. This choice is advantageous for
use with a Lagrangian formulation to find the equations of motion.
Once the independent variables are selected, the dependent variables (in
this case the generalized forces fk'S and generalized momenta flk 's) must be
expressed in terms of the independent varinhles. For the gencrnli/cd
machinc this means that the four electrical nux linkages ,\~. ,\~. ,\:. ,\~, the

<'S,*'4'iJ

"
i.~;: ~

l.:f ~.'.~.
~

',il

.~,.

It,

I:t f
j

"~',
";

I.

l .. ~ __

175

'f'

The purely mechanical part of the system is assumed to be linear.

is not necessary in the lreatment but it is convenient.

.,

This restriction

"
_....J

. ,f;~"f'

THE GENERALIZED ROTATING MACHINE

ELECTROMECHANICAL ENERGY CONVERSION

176

mechanical momentum p"" and the mechanical force!", must be defined

in terms of the independent variables i~, ii:. i;. i;'. <I,. and <{, The
mechanical momentum p", and force!", are functions only of the mechanical coordinate 1> and mechanical velocity q,. Defining the moment
of inertia of the rotor as J and the stiffness of an external spring as
K allows the mechanical momentum P", and force!", to be defined as:

\ a'-axls

1Magnehc

I"

aXIS

+0
(
(
(

U"u

Mechanical

P..

!. =

'q+.,

A..
'~

'(r

<f>-?

~t;"

'"
:J

....-

'"" "'"
-~~

Vb

l'
~<-,

.;,c...1) ~'"

<::-~"

~'b'JC

Fig. 3-3.

(,;0

TABLE 3-1

Mechanical

Electrical

Rotor

Rotor

Stator

-_ .. ---~---'k=5
b=lk=2k=3k=4
-------------

---"-

Conservative variables

"

If,

tik
fJ.
j~

"

Viscous friction

v'b

v.

R"

R"

Nonconscrvatlve variables

Qk

j~

ib

A'b

'

..

R"u

ib

A;

P.

f.
vt
RU

T
III

(3-2)

Electrical
A~(i~, i~, i~. i~, cjJ) = L~~i~

(3-3a)

i"

X;'(i~,

it"

i~. i~,

(3-3b)

A~(i~.

ii,.
it,.

i~, i~. cjJ) =

"I

'f~'~f .

f;
I

:"'X

K~

1\',

:~:;~

Xl

Schematic representation of a two-axis. two-pole machine.

(3-1)

The dependence of the electrical momenta (i.e., the flux linkages ;\~, ,\~, ""
A~, and Ai;) on the generalized coordinates and velocities qk and ilk can be
expressed by writing these flux linkages as functions of the currents and
the inductances. The inductances will be shown to be functions of the
mechanical coordinate 4>. In terms of the various self- and mutual
inductances, measured at the four electrical terminal pairs. the expressions
for the flux linkage at each electrical terminal pair are:

<t+,.,j-

O<l

Mechantca l
,halt

= Jq,

and

Q~

177

AW~.

i~. i~.

q,)

q,)

+ L~~it + L':.i; + L':t,i;'

L~~i~ + L~~ib + L~~i~ + L~~i;'
L~~i~ + L~~ib + L~~i: + L';;j,i;'
L~~i~ + L~it + L'':;,i~ + LbbiJ,

(3-3c)
(3-3d)

The various inductances (L::;', L~, ...) are not constants, and the
dependence of these inductances on the independent variables, in this
case the mechanical position ~. must be determined before the dependent
variables are completely specified. The large number of variables needed
to characterize the generalized machine leads to long and cumbersome
analytical expressions. To partially alleviate this problem, a matrix
notation will be adopted. All variables will be expressed in matrix form
before the variations of the inductances with q, are derived.
In this treatment the following matrix notation is used. A complete
matrix for the whole machine is denoted by a script letter. A matrix
for part of the machine, e.g., the electrical part, is written as a script
letter with appropriate subscripts and superscripts (31~~.Qb)' A sub-matrix of the system is sometimes written as a lower case letter. with
subscripts and superscripts, enclosed in brackets [~b]' Individual
In this treatment a knowledge of simple matrix algebra is assumed. See E. A.
Guillemin. The Mil/hematic", 0/ Circuil Allalysis, John Wiley, New York. 1949.

~ . . . ._

-~

tti.""k

178

ELECTROMECHANICAL ENERGY CONVERSION

THE GENERALIZED ROTATING MACHINE

~" L'~

elements in a parameter matrix are indif,ted by capital letters and variables by lower case letters, both with subscripts and superscripts. On a
variable the superscript means the member on which the variable exists,
r for rotor and s for stator, and the subscript a or b means the a or b
winding on the member. For instance, the current i; is the current
flowing in the a coil on the rotor. The subscripts and superscripts on a
parameter element have essentially the same type of meaning; for instance,
the mutual inductance L~b gives the voltage induced in the a coil of the
rotor by a rate of change of current in the b coil of the stator. In the
case of a parameter matrix double subscripts are used (e.g., Sf~b,"b)' In
this notation the ab directly under s indicates that stator quantities are
written in terms of coil currents and similarly for rotor quantities. Later
on, the coordinate systems will be changed; for instance, 'the rotor
quantities may be defined in terms of dq variables, in which case the
mutual inductance matrix is written .!l'~~.dq'

(10

"It

Lb~

L/,/,

aa

ab

L~~

L~;,

".

'-\;

0
0
9"=10
0
0

"

4:

The Independent Generalized Coordinates and Velocities

;,

q~

la

; =- q:qa

,. i~

0
0
0
0
0

~=

(3-4a)

/.,.J.

\J~"

'S

itb

~ = i
eP

4.

eP

" t;~':

ab.nb

~ oo

"D

91=

o
o

ol

RblJ
oo

ol
~"
0
R~
aa

'r

0
IX

= ~

I.'

'f'" '

~~b] ~

ab.ab

alJ.ab

ab

--l

0
0
0
0
0

0
0
0
0
0

0
0
01
0
K

(3-5c)

..

fb

1:1 =-1

::i~ .~~~

(3-6a)

K~,

p'la

ASa

p~

ASb

',-,';,

" )", r;(j,

.1"#

,\;j',

Pbl

Ab

P~I

JeP

t'l

~
_is''
_
S,"

sr

(3-6b)

'1'

I;

The Nonconservative Forces

'.

Q~

FI ~ ..,
ob.ab

I~I

t?IJ = p~

Ib

[[r~J 0 OJ

~[/~
][I~]) 0~ = ~"
~
[I'.S)[I'
0
J

, ~ A: ~ ~~~ ~ 2;ool~

(3-4b)

;.f

0
0

fb

t'. "1,

1'.

The Parameter Matrices

o
o

I:

~1 ~ I,~ ~~.

SS

~"
L'ba L'
LbbJ
ab

The Dependent Generalized Forces and Momenta

qi,
q.

bb

QQ

=-

aU

(3-5b)

qi,

= q~

L'L"J ol

-,

q~

J'

QQ

~ I~" L'~

~' ,

"~\ ,

~n
L"bn

179

I
I.

"

Q~

'.

,",[

T'

\' .,

-; (3-5a)

0<,

,.,

,'t

'.

i'

,I.

"

~.

,j,

"

.,

.,'\

. .~ 1
;h~i';)'- '. '
"

~ = Q~

Q~

Q.
;\

vS
"
=

'II

", = ~
v~

.\,

T"b."b

(3-7)

v'b

-T

180
3./.2

,
)

r
:t

ELECTROMECHANICAL ENERGY CONVERSION

Evaluation of Parameters for the Generalized Machine

J.

-!-H . B du,

Stator Surface (r = b)

and by equating this expression for stored energy

to the stored energy expressed in terms of circuit parameters, as

Rotor Surface (r

Wm -_l""L"
2- .(...,.. ijljlj.
I

:.-

.-

= 3 z i;7;: cos ncf/

(l-X(I)

k~(,p") = azi~Ztcosn(,ps - ~)

(3-8b)

k~(c/>S)

volume

The first step in obtaining the machine inductances is to determine the

flux density at all points in the machine. In Sec. 3.1, items 5 and 9, the'
magnetic material of the stator and rotor was assumed to have very high
permeability and the effects of winding slots were assumed negligible.
As a consequence of these assumptions, the only significant stored magnetic
energy is in the air gap. Therefore the inductances can be derived from
the air-gap field quantities Band H that are obtained from a static field
solution for the machine in accordance with item 10 of Sec. 3.1. It should
be mentioned that in practical machines additional magnetic energy is
stored in the winding slots and around the end turns of the wir.dings.
These terms lead to additional inductances, which for practical machines
must be included in the self-inductance of each winding. These in-:tt
ductances will not be included in the ideal model; first, because these slot
and end-turn leakages do not affect the energy conversion properties of
the machine, and second, because these inductances can be accounted for
later by adding them in at the terminals of the ideal model. Therefore,
for the generalized machine it is only necessary to include the effects of
magnetic fields that exist in the air gap to establish a model that is ~
valid characterization of the energy conversion propertics of practicaf"
machines.
As pointed out in item 6 of Sec. 3.1, the saliency is accounted for by Ii
space-dependent permeability; consequently, saliency only appears in the
flux density B. The magnetic field intensity H is found by assuming no
saliency. To find the magnetic field H consider a smooth-air-gap structure which is described by cylindrical coordinates r, </>s,::. The rotor
surface is at I' = a, the stator surface is at I' = b, the radial air-gap length
is g = b - a, and the axial air-gap length is I. The four current sheets
set up four distributions of surface current density, two on the rotor and
two on the stator. These surface current densities flow in the axial (z)
direction.
With reference to Fig. 3-2, but remembering that the stator surface is

.181

assumed smooth in the solution for II, the four current sheets sci lip the
following surface current densities:

The inductances in the generalized machine can be evaluated and the

dependence on mechanical variables can be determined by finding the
stored magnetic energy in tcrms of B, H, and thc dimensions, as

Wm

..

THE GENERALIZED ROTATING MACHINE

,,~

;i
:1~
I.,i'

t:;~

''S

a)

k~W)

= azi~Z~ cos "tll =

k;W)

azi~Z~ cos

a,ibZ~cosn(,p' -

2:)

n(,pS - ,p)

azibz~cosn(,ps -" -

(3-9a)

2:)

(3-9b)

These surface current densities will provide boundary conditions for the
air-gap magnetic field H.
The device being considered is linear; consequently, superposition holds,
and it is convenient to find the field due to one surface current density at
a time and superimpose the results to obtain the total field. Consider the
case when only the a winding on the stator is excited and the other three
currents are zero. Furthermore, only the static magnetic field is of
interest so the angular displacement ,p and the current i; are assumed to
be constant. The static magnetic field set up by i; is designated H~.
Inside the air gap (a < r < b) where it is desired to determine the static
magnetic field H~, there is no current density and no magnetic material;
therefore, the curl of H~ and the divergence of H~ are both zero. Since
V x H~ is zero, H~ must be the negative gradient of a scalar potential
.p~, and with the additional restriction that V' fJ-oH~ be zero, the scalar
potential .p~ must satisfy Laplace's equation. In equation form these
defining relations are:
VZI/;~ =

H~

and

...

(3-lOa)
(3-IOb)

~ V.p~

For the physical configuration of Fig. 3-2 there are no variations in the
axial (z) direction (fringing at the ends is neglected). Thus, 1/;:' and H~
expressed in two dimensions (1', q,S) will yield the desired field distribution.
In cylindrical coordinates, the Laplacian V21/;~ = 0 in two dimensions r
and cPs is
~J~ .

"t

V21/;Q

821/;~

8r 2

1 01/;:'

oZI/;~

+ r ar + f2 8eps2 = 0

(3-11)

For a discussion of the approximations involved in using the static or zero-order

field solutions. see R. M. Fano and L. J. Chu, Fields, Eller/f)', and Forces. John Wiley,
New York. 195<).

182

Now
thus,

The general solution for Eq. 3-11 is well known and is of the following
form
if/~ =

Bler-k)(CIe cos kg,'

+ Die sin kc/>')

Bolnr)(D o)

1
r
k=l

+ 2:

(3-12)

8rjJ~

= - Vif/. = - a, ar -

a",

o,p;

r 04>1

Cn = 0

Bkr- k)( -kCk sin k4>1

D n = n(AiJIJ-1

Do = Dk = Ck

+ kD k cos k>') (3-16)

(H~)q1 = - (A n",-l

I
!

(H')qI = _ [('a_)n-l _
Q

'.I

,!

for k = n

+ Bnb n-I)

for k '" n > 0

Bnr-n- 1){[(A lJn- 1i"Z'

~
b- n - 1)]} cos n4>1
n
n

(3-17)

Using the boundary condition at, '"' a, Eq. 3-15 gives En = -A,p2n.
Substituting this into Eq. 3-17 and rearranging yield

8c{t'.

i:Z~

Therefore,

(3-13)

radial component

.:tc .

or

a:"

-(Akr k

Using the boundary conditions at r = b (Eq, 3-14) gives

To find the scalar potential t/J: for the physical configuration of Fig. 3-2,
the constants A k , Bk , Ck , Die of Eq. 3-12 must be evaluated from boundary
conditions. The boundary conditions are imposed on the magnetic field
intensity H~ and the magnetic field intensity is defined as the negative
gradient of the scalar potential .p~. The magnetic field H~ has a radial
and an angular component, i.e.,

(H~)r .. -

can be written in general terms by using Eqs. 3-12 and 3-13;

>

(Alerle

k-I

H~

(H~)..

(H~). = ~(Ao
+
r

(A o + Bo In r)(C o + Do"")

+ 2:

183

THE GENERALIZED ROTATING MACHINE

ELECTROMECHANICAL ENERGY CONVERSION

(ra_ )-n-l]

r n.\ __ I

In
"Z'a

IL\

,~

cos nl

(3-18)

1 BtlI'

(H~).,. ... - ;: 8r1>~

..

,:.1

angular or tangential component

Equation 3-18 satisfies the boundary conditions at , = a and r = b. and

in evaluating Eq. 3-18 it has been shown that AIJ' Bn and Dn are related as ~

)C

The boundary condition that the tangential component (H~)", must

satisfy at the stator surface (r "" b) is
ttl

Bn = -A,p2n

. . - (k~)t = - i~Z~ cos nl

,.,

'

(H~)q1I'_Q =

(3-14)"
.~

wh.ere (k~)1 is the z-component of the surface current density at the stator .
surface due to i~ and is given by Eq. 3-8a.
The boundary condition that the tangential component of the magnetic
field must satisfy at the rotor surface (r = a) is
l~

The scalar potential t/J~ can now be found by substituting Eq. 3-19 into
Eq. 3-12:

<#:

,:~

= (;, -

0'::,) (n [(~r/;~:m"

'J)

,'n

n'"

C,(Ao+ B, In,)

!~,)"

(3-20)

The radial field (H~)r can be found from Eq. 3-20 by taking

oo/;:/or, yielding

",'i

(3-15)

'<-i'

',.'f.'i,~

";:1$
because the rotor windings are not excited.
Equations 3-14 and 3-15 are the boundary conditions that the static
magnetic field H~ must satisfy.

See any text on advanced calculus. e.g., Advanced Calculus for

Hildebrand. Prentice-Hall. New York, 1949, pp. 423-424.

Enginee,~,

F. B.

(H~), = _ O';/~

:,h.

or

"~;"

,~

,.

""I"

",.;,~1 '
j

\.

,., .j

--,.-

(3-19)

"ZI
D =
In
n n(Anbn-1 + B"iJ-n-l)

.ttl

(H~)q1 !r-b

h
!

,~

= -

,:j)'(

':,'(J" t

~~

r"

'} i,;,;'

r(,)"-l (,)-11-1]
jSz.'
C B
a + a [(~r-l ~ (~rn-l] si.n n~' -7
l

(3-21 )

184

ELECTROMECHANICAL ENERGY CONVERSION

~~,'

The constant CoB o is independent of the stator current. When the

current is zero, the field must be zero. Consequently, the constant C 0 8 0
must necessarily be zero.
The static magnetic field strength H~ produced in the air gap by a
sinusoidal current sheet due to i~ on the stator side of a pair of concentric
cylindrical magnetic boundaries of infinite permeability is, from Eq. 3-18,
Eq. 3-21 and the condition C 0 8 0 equals zero,

_Jl(;r(;rn-'
l
(~r-I _(~)
i~Z~

THE GENERALIZED ROTATING MACHINE

~?

higher when the field is squ;lred and integrated over the volume. Thu
for the purpose of finding stored energy (and, also, circuit parametl'
values) Eq. 3-24 can be simplified to

.;>":

It,
'.:~'l

H"u c::
- a , I1g
!!.... i"Z"U sin l1</>s
-

"

f!i.

Equation 3-25 is the appropriate field expression to use in subsequen I

considerations of the characteristics of electric machines with sinusoid:1
current sheets on both rotor and stator when both surfaces are smootl
concentric cylinders and is the approximate field expression to be used
with the space-dependent permeability in the treatment of salienc)
This approximate field equation consists of only the radial magnetil
field, a fact which has led to the use of air-gap magnetomotive force (mmf
rather than magnetic field intensity H in the classical analysis of electriv
machines.
The approximate magnetic field intensity produced in the air gap b;.
the four sinusoidal current sheets shown in Fig. 3-2 and expressed b)
Eqs. 3-8 and 3-9 can be obtained by superposition of the fields produced by
each of the current sheets. The field due to each additional current sheet
can be found from Eq. 3-25 by replacing i~Z~ by the appropriate product
from Eqs. 3-8 or 3-9 and by replacing the angle </>" by the angh: given in
the appropriate expression of Eq. 3-8 or 3-9. This process yields the
four components of the air-gap magnetic field

'~i

,j

a~ (~r-l

,.

_(~r"-l

sin n</>"
L;~

'Ji;i

1
r

i~

i~Z~ cos ncp"

(3-22)

r ",

...

T'
"

Equation 3-22 is the total magnetic field in the air gap of the machine due
to i~. i,e., for a <: r ..; h, for 0 .;;;: </>' <:: 27T, and for 0 " z -;;; I.
The other three components of the magnetic field H~, H:'. and HI; due
respectively to currents it, i:', and i;. could be written down immediately
by specializing Eq. 3-22. However, before this is done, Eq. 3-22 will be
simplified by making the assumption of a small air gap.
The magnetic field intensity given by Eq. 3-22 can be greatly simplified
by recognizing that for most electric machines the air gap is sm:lll compared
to the radius of the rotor and therefore b ~ a. Thus the radius r can
be replaced in the air gap by a new variable defined by:

i,";,

..

-n-l

_ r(~r-l -(~rn-l1

Ii

where

. ;:~\

~tj

rdr

r=a+p
b c:: a

~g

adp

and

b=a+g
O<;,p<g

(3-2)

lJ

H' =

18

';~f

~~
.,'

. .

it,.,

\S;;c

1,.,

;t"
":

(326a)

Wb = - a, ng
a I"Z"
b b SIn n

..

(3-23)

= -a !!..
j"Z' sin n4>'
ng a

HS

(4), - 2n1T)

H'U = - a , !!ng i'a Z'a sin n(~'" -

~Jt

Wb

Using these variables and neglecting all terms in pia or gla of higher than
first order gives from Eq. 3-22 the simplified result:

4"!S.
~.

C3,-26b)

(3-26c)

7T )
= - a'ng
-a I.b'Z'
S ill n (4), - 4> - b
2n

(3-26d)

The total air-gap field is obtained by superposition. and is

e.

H"u c::
-8 a - p iSZ" sin n</>' - a
i'Z' cos n</>'
,
ng a u
4> g a a

(3-24)

H = H~

+ Ht +

H~

HI;

(3-270)

Substitution of Eqs. 3-26a-d into 3-27a gives the total field intensity;

The approximate magnetic field given by Eq. 3-24 is correct to the first
order in pia and gla; however, when evaluating the energy stored in the
air gap to the first order in pia and gla by integrating the energy density
!(J-! off2) throughout the volume, Eq. 3-24 can be further simplified. Terms
that arc first order in pia in the field equation become second order or

H ... a,(H),
= -a, -a [.1'Z.
"a sm
JIg 0

c.

.....

.I.s

n.,,~

_-------

(3-27b)

. 'z'
.I. .
- I"
"h cos n."s

i~Z~ sin n(</>' - </ -

i;Z~ cos n(</>" - </]

186

ELECTROMECHANICAL ENERGY CONVERSION

The flux density B is obtained from H by using the space-dependent

permeability defined in item 6, Sec. 3.1. Thus,
B = alfL - fL2 cos 2n</>')(H),

Z; = Zi, = Z'

(I (g (2"

Jo Jo Jo

L S = (zs)2[7TfL ~

.i~

I..

L' = (zs)2[7TfL2 a

fo

i;Z; sin n(</>s - </ - i;Zi, cos n(</>' - </]2}(0 dp d</>s dz)

n7T

t(i;,)2[ [ (Z;,)2 2

.s -r [
(lala)[

O'

"

'4U

'H
'

+
+
+ (I;jF,)( - L::
+ (ibii,)(L:s -

+ (itiJ/[ZbZ~;(~ J10

- ;; JAo2) sin n</>]

:t;:

'r~1

" ,v

"'~..-', ,';~'

- ;;J102) sinn4>]

(ibii,)[[ZtZi,;(~J10 -

;;J102)cosn</>]

!~; "
,t,

(3-29c)

L~;) sin n</> + (i;/~)(L::

L~;) sin n</>

L~;) cos n</>

Wm

::i

.,dT

"I

.
(3-37)

"~'

~,:

hl~(J]

t(it)2(L~ - L~2)

The stored energy given by Eq. 3-37 can be equated to the slored energy
defined in lerms of circuit parameters, i.e.,

;i:,',,';

,"

+ L~) +

(i;i;')(O)
-t(i~)2(L: + L: 2 cos 2m/ + t(i;Y(L: - L:2 cos 2n</
sr
(iarr)(
+ (jsjr)(L"
b - L'1'2 sin 2nA-)
'I'
a a I' + L 1'2 ) cos nA'f'

r 7T

+ (i; ii,) I [Z;Zi, :2 (- ~ JAo

,+

~,

3 a3 )
ZaZa n2 (0g JAo + 2g
JAo2 cos n</> ]
.,

Using the quantities from Eqs. 3-31-3-36, the stored energy, Eq. 3-29c,
can be written as:

Wm = W~)2(L~

fL2 )]

(3-36)

n2 2g

1'2
I>Ji

(3-35)

= (z,zr)[71'fL2 ~3

L Sf

2g !.l2 cos 2n</>

3
'r -r
[
r
r 7T (a
).
.L] - , ,.,
+ (lalb)1
-ZaZb n 2 2g JAo2 sm 2n'f'

(3-34)

7T

g!.l -

2gff-'

L Sf = (z szr)[7TfL ~
I'
n2 g

3 a3
+ t(i~it)/(OJ + tu;)21 (Z;)2 n2 (0g JAo + 2g
JAo2 cos 2n</> )]
(0 3 a 3
)J' ~H
,[

//2

I'z

(3-29b)

3
02g

(3-32)
(3-33)

= (Z')2[7TfL2 0

L'

7T
= t(i~)21 [ (Z~)2 n7T2 (ag!.l + 02g !.l2 )] + -WrY [(z;,)2
n2 (ag!.l "

.H(JAo - U2 cos 2n</>s)(Hm(a dp d</>s dz)

Integrating Eq. 3-29b with respect to z, p, and </>s and then collecting all
the coefficients for each product of currents yield for the total energy
Wm

= (zr)2[7TfL2 ~

I'

'II

Lr

Substitution for (H)r from Eq. 3-27b yields:

l
2
Wm =
fog " t{(JAo - JAo2 cos 2n</>,) (n~r[i~Z~ sin n</>s - itZt cos n</>, l'

n2 2g

I'Z

(3-29a)

(3-31)

n2 g

I'

,:tl
:

(3-30b)

and the following symbols are defined:

IJg JI(2" tB' H(a dp d</>s dz)

Jo

187

Equation 3-29c can be simplified by introducing the fact that the

generalize 1 energy converter has two identical phase belts on the stator and
also two identical phase belts on the rotor. Thus, the conductor densities
can be written as:
Z~ = Zt = Zs
(3-30a)

(3-28)

where (H), is given by Eq. 3-27b.

The air-gap fields, Eqs. 3-27b and 3-28, can now be used to find the
stored energy by taking a volume integral throughout the air gap. The
air gap of the idealized model of the salient-pole machine is defined by the
axial variable z from 0 to I, the angle </>s from 0 to 27T, and the distance
into the gap p from 0 to g; g is the average gap length of the salient-pole
machine, and becomes the gap length of an idealized cylindrical model.
The total air-gap magnetic energy is
Wm

THE GENERALIZED ROTATING MACHINE

rI', <"
f~;.

~"

'l

I , ' L ..
-2";''';' ijlIlj
I

'HI '>')1,-1,,', .

(3-38)

which for the four-coil configuration of Figs. 3-2 and 3-3 with the
inductances defined by Eq. 3-Sh.. is
.' ."

",.

, ;-1'

""

Wm

= "ZlL"'(j")2
+ .1L'S(r)2
+ lL"'j"jS
lL"'j"is
""
aa a
2 hh b
"Z ab a b + "Z
ha h a

+ !L;;;,(i;)2 + -tLbi,(ii,)2 + tL~j~ji, .+ !L';"ji,i~

+ jL~:j;i~ + tL;;/~j~ + !L~i~j;' + tL;':/bi~
+ !Lbaiti~ + !L:;'i~ib + tLJ,'i,i!/;' + tL/;;,i;'it

(3-39)

. ,~l

'"i~

..

"

:. ~\
,,'r

L __

.' <:,;'"

-,~!

....

L
t'
THE GENERALIZED ROTATING MACHINE

Equating Eq. 3-39 to Eq. 3-37 and noting that the mutual inductance Lij
must equal Ljj in this assumed linear system, the inductances become:

zero in a nonsalient machine. It is evident then that for a smooth-air-gap

machine all self-inductances are constants and all mutual inductances
are sine or cosi ne functions of the electrical space angle (m/ hetween a
coil on the rotor and a coil on the stator. This type of spatial variation
of the mutual inductance is exactly what one would expect between
sinusoidally distributed coils that rotate relative to each other. The terms
due to saliency are similar to the nonsalient terms except for the rotor
self-inductances. The stator self-inductance tcrms are still constant and
the stator to rotor mutual inductances vary as the sine or cosine of thl'
electrical space angle n4>. This result follows from the fact that as seen
from the stator the rotor is magnetically smooth; thus the stator selfinductances should be independent of rotor position and the stator to
rotor mutual inductances should be similar to those of smooth-air-gap
machines. Conservation of energy, which demands that stator to rotor
mutual inductances must equal rotor to stator mutual inductances, establishes that all mutual inductances be similar to those of a smooth-air-gap
machine.
The rotor self-inductance due to saliency is, however, seen to be a
function of rotor position and varies as the sine or cosine of 2nrP which i.,
twice the electrical space angle nc/>o This result follows from the fact
that the air-gap length completes one cycle of variation in an angular
distance equal to one-half the span of a phase belt. Thus the rotor
inductance variation due to stator saliency_completes 2n cycles in each
revolution of the mechanical angle 4>.
The inductances have now been expressed as functions of the independent coordinate 4>. The mechanical inertia and stiffness were
previously defined as the constants] and K respectively. These parameters
are constants independent of any of the independent variables.
The only parameters that remain to be defined are the dissipation terms.
By virtue of assumption 8, Sec. 3.1, the resistances of both stator coils are
equal and the resistances of both rotor coils are equal; thus, these
resistances are defined as

=L +L
L~~ = L~ - L~2
L~~ = Li:. = 0
L~~, = L~ + L~2 cos 2nt/>
L~b = L: - L: 2cos 2nt/>
I~b = I':a = - I~2 sin 2nq,
L~: = L':a = (L:: + L:~) cos nt/>

(3-40)

= (- L::

- L::Z) sin n~

(3-47)

L:~) sin

nt/>

(3-48)

nq,

(3-49)

US

aa.

jJ.

jJ.2

L::b = L~
LS;"

= L~1 = (L:: -

I;:', =

L~b

= (L::

L,:~)

cos

(3-41)
(3-42)
(3-43)
(3-44)
(3-45)
(3-46)

When Eqs. 3-40-3-49 are substituted into Eq. 3-Sh the inductance
submatrices for the generalized machine become:

[/;b] =

L~ [~ ~] + L~2 [~ _~]

(3-50)
~'1 ~

0]

, -- L'jJ. 0
[I"b]
[

[1''' ]
"b.ab

L' [
"2

cos
-sin

2nc/> I -sin 2nc/>]

2nc/> I -cos 2nt/>

L" [COS nc/> I - sin nc/>] + L" [

sin nq, I cos nc/>
jJ.

"2

[/~b:"bl = [/;b~abll

.'

18')

ELECTROMECHANICAL ENERGY CONVERSION

188

cos nc/> I
sin n4> I

i.~

(3-51)

- sin n4>]
- cos n4>

(3-52)
(3-53)

Equations 3-50-3-53 define all the self- and mutual inductances as seen
from the four electrical terminal pairs. The character of these inductances
is quite significant. Note that some of the coefficients have a subscript
I', such as I~ or
and the rest of the inductance coefficients have the
subscript fL2, such as 2 or L~;. Reference to Eqs. 3-31-3-36, where these
coefficients are defined, indicates that the subscript fL corresponds to the
inductance heing a function of the permeability fL and the subscript !J-z
corresponds to the inductance being a function of the permeability !J-z,
From the way the permeability was defined (Eq. 3-28), the conclusion
results that all terms with a suhscript I' will occur in any smooth air-gap
machine, whereas the terms with a subscript !J-2 are the additional terms
that result from saliency, i.e., a nonuniform air gap, and these terms are

I;:,

R~~

= Rw,

oW

(3-54)

Rr

(3-55)

and

L:'

R~: =

~:,tfi,j /', ':

,,:

~.,

\' .
;~
o

r'
l"'~,'
r~Ii'

1'1

RbI,

The only remallllllg dissipation term is the viscous damping in the

mechanical system, and this is assumed to be a constant independent of
the velocity. The symbol Cl is used for the mechanical loss parameter.
The necessary parameters to characterize the generalized machine have
now been evaluated. These parameters arc summarized on p. 190 in
matrix form in terms of'the quantities defim:d previously by Eqs. 3-5a-c.

190

ELECTROMECHANICAL ENERGY CONVERSION

-.0
I

<"I

-e-

'-'

---

::; i'
"-l

~"

0
()

...,J

'-'
I

"-l
'-'

t:;"

-e.

"'9-

-e-

<h

Cl

II::

'm

... ..

::' ,"

"'- ,. . ,-,
. . ,. .
10000

>oJ,

oooi:c:o

.....

~...

~"

'-'

'-'

""
~

~s::

II
',.

""
-::""
...

~..
I

'"0

..

:' ,'"

,.

-.
+
.~. "

""
~...

'C>

.c'"
..
+

" .....
+

V)

II
,l,i,

'.'~1'

",vt,J
l;~~ ',_'l J ;\ lj'":"

0,

'-"

'-'

v~;'ab = gp~r,.ab/~r,.ab

. <:;t!'t:':;~

,~

\.

~'".

't'

~
II

I'"~

v~r,.ab

v"a

la

,\,'

lb

x'b

e"."

ab,ab

.. ~~

1<':

, ' '-'1

'*

(3-60)

..

J".<\'.

.,

v;;

>..~r"ab =

i;
ib

>'"

r
l"'"
ab,ab
'ab,ab

{OS

..;r;

II

>'" b
"

.~:/

"t~<

t;: r~ '. :.~. . , , 'i~

... ,,' "

,,,~ l ",:", ~i"f~~

I;f

t<:

gp~;'.ub =

Eq. 3-56

'-~ ~~

Q;

I,.

Eq. 3-57

..1 v'
.li~;."

.
')11.1

;!W1

Equation 3-60 can be written out explicitly in several ways, This same
set of volt-ampere equations is written in detail in matrix form later in

,I,
"
I?

::e~;',Qb

i",
,'!

jj:; ~

.J"
. ",k';
",:o~::--.,--' _'

+ p(::e;r,.abe'~~ab)

If I.

,~

If J ~

Va

fr':"-t,,:'iR

~~>,,;:

1o~1""11\

.,

j'

.'

;_ ~r
',,~d_i'1

(3-59)

where

"it.','
,

+ PA~
vb = it R' + pAl,
v~ = i;R" + PA~
v~ = ibR" + PA~
v~ = i;R'

The four flux linkages >..~, >"1" >,,~, >../;, are defined by Eqs. 3-3a-d. Rather
than substitute directly for these flux linkages it is advantageous to put
the four vOlt-ampere equations of Eq. 3-59 into matrix form. The use
of the matrix quantities defined by Eqs. 3-4a-3-7 yields

:.:'

"-l

"l

:;.~

t:;"

f,

</)

~'"

"l.""..o
~

, ".

,'I

;.. '"

'\~;

'I""~

"1\

lP.,,;

.S

, \\1..,(

,:!'t,

I~

('...

()

s::

'l.tI,<

\000001

~"

-e-

~ ~

00000

'-"

l'

C.)

..,

00000

The Volt-Ampere Equations

First, using Kirchhoff's voltage law and Faraday's law of induction,

one can write a loop equation for each of the four electrical terminal pairs,
thus:

000001-

~"

J ' ~

foooo:<1

~' ~;

'"

I
~..
~"I

OJ

. 3.2.1

l::

'-"

The coordinates for thc gcncralizcd machine were chosen in Tahlc 3-1.
and the other parameters necessary to complete the dc~cription of the
model were derived in Sec. 3.1.2, and summarized by Eqs. 3-56, 3-57,
and 3-58.
The next step in the study is to find the equations of motion of this
machine, It is advantageous to derive these equations in two ditl'erent ways to show explicitly the similarities and differences between the
use of Kirchhoff's laws, Newton's laws, and an arbitrary displacement
plus conservation of energy. and the use of a Lagrangian formulation.

'I

""9-

>oJ,

tl I

I~

".,

<"l

... ~"Vi

</)

~
~

s::

Cl

'~.'

""9-

<"l

s::

10

>oJ,

""9-

I~ooool

I
.....

.. i','

o~ooo

.S

"-l

8 0-Q
::,rIi." . .'"i'
+ >oJ,

oo~oo

"

s::

l::

,~H

'~;

</)

'" . .(,)!i'

191

Using Basic Force Laws

<"l

:-N.S
~"I
"-l
... ~
I

3.2 The Equations of Motion of the Generalized Machine

---

l'

-e~

'"

"

<"I
'-'

<"I

s::1'

c::

THE GENERALIZED ROTATING MACHINE

00
Vi

';ji

<u

l~

1 0 1 0 1 0 1 0 t ....

Vi

,~

f
~

ELECTROMECHANICAL ENERGY CONVERSION

192

THE GENERALIZED ROTATING MACHINE

the chapter (see Eqs. 3-95). The use of Eqs. 3-56 and 3-57 with Eq. 3-60
yields the four vOlt-ampere equations:
v~

J;"

V~ = RSi~
v~ =

Vb

The change in stored mechanical energy is evaluated from Newton's law:

as (N) de/> + (Ke/ dep; thus Eq. 3-66 becomes

= R"i; + p(L~ + L~l)i: + p(L'': + L~)(cos "e/i~ - peL'; + L~)(sin ne/i~

(3-61)

+ p(L~ -

L~l)i~

+ peL'; -

L~~)(sin

neW;

+ peL'; + L':;)(cos ne/i~ + peL'; - L~)(sin ne/i~

+ p(L~ + L~'2 cos 2ne/i; - pL~2(sin 2f/(p)i~

(3-62)

(3-63)

\i

R'i;

R'ib - peL'; + L,:)(sin f/(/i~

- pL~/sin 2I1cP)i; + p(L~ -

+ peL': - L':)(cos ne/i!

L~2

J,

+ a~ + Kep + dcP ( dW", -

T - Jf>

'~

+ peL': - L';)(cos ne/i;

193

(3-67).

iJ d>V)

The independent generalized coordinates and velocities have been chosen

as the mechanical position, the mechanical velocity, and the electric
currents. With the stored magnetic energy expressed in terms of these
coordinates as Wm(t/> , i~, it,
i;) and similarly with the flux linkages
expressed as Aj(e/>, i~, it, i;, i;), the term in parenthesis in Eq. 3-67 is
evaluated as:

i;,

(3-64)

cos 2n4i;'

dW'"

These four equations are the equations of motion for the four electrical.
terminal pairs.

~
'd'
L. I j IIj

j-I

oWmJ-J.'
~ oW"'d'
U'f' + L. ~ I,

= ~

,-I

v't'

vI,

~L. I.(OAJJ-J.
~ o>v dI,)
':I-J. U'f' + L. 7
j

J-I

,-1

V'f'

Vi,

(3-68)

! 1

: ;I

3.2.2

Rearranging this expression and collecting the coefficients of dep and di,
separately yield:

The Torque Equation

The generalized machine forms a coupled electromechanical system;

thus mechanical motion affects the electrical system and electrical motion
affects the mechanical system. The effect of mechanical motion on the
electrical system is automatically included in the electrical equations of
motion by Faraday's law, v = pA. hecause A is expressed as a function of
the mechanical coordinate e/>. Thus pA contains the voltages thu< result
from mechanical motion. The effect of electrical motion (current flow)
on the mechanical motion is not included in Newton's laws; thus some' "
auxiliary technique must be used to find the mechanical torque of electrical
origin.
Several techniques for finding the electromagnetic torque are available.
The method to be used here is the conservation of energy and an arbitrary
displacement. When an arbitrary displacement dc/> is applied to the shaft
in time dr, the conservation of energy yields the result

,..

f'

mechanical
energy
supplied

electrical
change in stored
energy =
mechanical
supplied
energy

change in stored
magnetic
energy

+.L ijdAj
j=1

dWmech

+ dWm +

a.(~)2dr

~"

L. IJ

j='

d'".J

(OW",
~
(/'f'

~.
OAj ) d'-J.'f' + L.
~ (OW",
L. IJ ':1-1.,
v'f'
,-I --;;-:vi,

j-I

l'

~. OAj ) d'I,

L. IJ ~

j_1

Vi,

(3-69)

The arbitrary displacement must be compatible with the internal con'!'

straints; thus the coefficient of di, is always zero (see Sec. 1.2.2, Eqs.
1-23-1-26). This demonstrates that the electromagnetic torque is.
determined solely by the magnitudes of the currents and not by the rate:
of change of current. Consequently, only the term in dep remains in Eq.
3-69, and substitution of this expression into Eq. 3-67 yields:

'!J{(

= J~ + a~ + Kep + o~'"
CJ't'
,

i ~
ot/>

J-I

(3-70)

W", can be expressed in terms of the coenergy W';' as

losses
,,"

or

Tile/>

dW",

Equation 3-70 can be expressed in a simpler form by recognizing that

(3-65)
!';'tJ,

(3-66)

In this developmcnt the mechanical potential and inertial forces and mechanical
losses are not removcd from the coupling network. Thus Eqs. 3-65 and 3-66 will give
the forces due to magnetic field coupling plus all other mechanical forces acting on dif;.

Wm

= L

j-I

ijAJ - W';'

(3-71)

The independent coordinates are the four currents ij and the angle ~;
thus, differentiation of Eq. 3-71 yields

oWm

~ =

.IJ OAoep j

j_1

OW~

7if"

(3-72)

ELECTR.OMECHANICAL ENERGY CONVERSION

19<4

and substitution of this result into Eq. 3-70 gives the torque as

T = J~
where

W;'

+ K.J.. _ 0 W;"

a.p

or/>

'f'

s 's 'r:r

(3-73)

1.,lb,I.,'. L A di

0 ... ,0

)-1

The inductances were assumed electrically linear (assumption 9, Sec. 3.1);

hence the inductances are not functions of the currents. The magnetic
coenergy W;" is integrable and equal to the energy Wm' Thus W;' is
4

W;, =

tL

AjiJ

j-I

t i~b,obl

(3-74)
2~t,ob i~b,"b

The torque equation can be written as

= N + a.L'f', + Kc/>

J.r
ob,Db,

o!l!s,r

sr
~

84> i ob.ob

(3-75)

Equation 3-75 is the mechanical equation of motion for the generalizet:

machine. When the elements of l~~.ab and !l'~;.lIb from Eqs. 3-4b and 3-57
respectively are substituted into Eq. 3-75, and when the differentiation
with respect to </> is performed, the mechanical equation of motion
becomes:

T =

J~ + a.p + Kc/> - n{[i;it(L:; ..:. L~) - i~i;(L:: + L~)] cos nc/>"

Ii
- [ibit(L:: - L~) + i;i;;(L:J + L:~)] sin n</>
+ L~zl(i~)2 - (i;)2] sin 211</> - 2L:/;i; cos 2nrf>}
(3-76)

The Lagrangian state function for the generalized machine will be a

function of the mechanical coordinate and velocity, 4> and .p, and the
four e1cctril; currents i;, i~, i~, iF,. These four currents are the generalized
electrical velocities; the four corresponding electrical coordinates do not
appear in the Lagrangian state function. Thus, in this system the four
generalized electrical coordinates are ignorable. The occurrence of
ignorable coordinates indicates that particular care must be taken to
make sure that the currents chosen for generalized velocities yield true
coordinates and not quasi coordinates, i.e., the currents must be
integrable. The coordinates chosen in Table 3-1 do meet this condition,
because they are the physical coil currents and as such are integrable to
yield the charge through a coil in a given time. Furthermore, the coils
are not, as yet, interconnected, so these currents will be independent of
each other and also will be independent of the mechanical coordinates.
This follows from the fact that the final value of each of these variables
can be adjusted independently of all others. Since the electromechanical
system is expressed in terms of true coordinates and velocities that are
independent, a Lagrangian function can be written which, used with the
Euler-Lagrange equations, will yield the equations of motion.

"

3.3.1

Equations 3-61-3-64 and 3-76 completely characterize the generaliuG

rotating electromechanical energy converter from its four electric;,
terminal pairs and its mechanical terminal pair. These equations wen:
derived using the basic force laws for electromechanical systems.

The Equations of Motion of the Generalized Machine by

Lagrangian Methods ~,' ,_

The Lagrangian of the system can be derived from Eq. 1-100 as:

L = T' - V == kinetic coenergy - potential energy

(3-77)

where

T' =

N.

ril ..... f/N

.,.,.,

(3-78)

p;(qJ, ... ,qN; ql' ... ,qN; ()dql

;=1

"and
q1

= f0.....0
.'

:..1 '.:.'.

~~
~;,t

-,

"!.

.... ,qN

L-

I-I

I'l(ql"'"
I

).J'

(3-79)

qN' t uql

,VI

._The potential energy V is quite simple, since 11 = Iz = IJ

the only force is the spring force of the mechanical system.
positive spring constant,
.. "~" . ,. ,;,~,

V=
In the formulation of the equations of motion by Lagrangian method
the coordinates defined by Table 3-1 and the parameters defined in ten
of these variables in Eqs. 3-56, 3-57, and 3-58 will again be used. T
simplify the formulation of the Lagrangian function and the equations of
motion, the matrix notation defined by Eqs. 3-4a-3-7 will be used.

(($

The System Lagrangian

0.....0

3.0

195

THE GENERALIZED ROTATING MACHINE

f'

Kef> drf> =

Thus, with a

l:

-tK4>2

'i

= /4 = 0, and
.j

l'iII

'('

(3-80) f:l(;

'The kinetic coenergy can be simplified by subdividing it into two integrals

as
T' =

'I :

{~
'j'ii.'6'j~'

rj~,i:,j~,j;

Jo,....o

i
j-I

A;(c/>,

il> ;2' ;3' ;4) diJ + f~ (J~') d.p'

(3-81) ,

. lHll

OJ

.il,

" t:w,
;.

ELECTROMECHANICAL ENERGY CONVERSION

THE GENERALIZED ROTATING MACHINE

The first term on the right-hand side of Eq. 3-81 is recognized as the
magnetic coenergy which, in this electrically linear case, is equal to the
stored magnetic energy, thus:

differentiation of the result as indicated, and the use of the Qk from

Eq. 3-7 yield:

196

T' == W;"

The stored magnetic energy

+ tJcf,2 ==

Wm

+ tJcf,2

is a quadratic form,

Wm

~
s,r

ab,ob

---

(3-82)

~ L;.J

and

as

== - L L
2 ;=1

'\~ib

)~I

A~i;

Abi~)

'J:'~

'.
~

(3-83)

'S

'r 'r

1) _

'b' la' 'b' 0/

~ e's,r
ros,r
s.r
2 ab,ob t .L ab.ab e ub.ab

~J12
2- Cf -

lK.l.2
-1 't'

(3-84)

l.,..~

;Z~1'

;~~
~~P

"J.

j;.

t...

F =

I~

tr;(lU 2

= tRs(i~)2 +

tRS(ii,)2

il

+ tRr(i~)2 +

or in matrix form

F ==

lis,r
!Jis,r
is,r
"Z ab,ub,
ub,ab ub,ab

tRr(i~)2

+ t a (cf,)2

1 1
+ t <parp

(3-86)

(3-87)
~~l

3.3.2

(3-85)

For the generalized machine this becomes

".

",,'

The Equations of Motion.

The equations of motion of the generalized machine can be found from

the Euler-Lagrange equation, In terms of Land F, it is defined by Eq,
1-130 as:

d(()L)

Jr

(14k

iJL

- Oqk

of'

+ 8ilk

Qk

8:,s,r
~is,r
~ e~,r
~ ab,ub,
8</>
ab,ub

(3-89)
.:,:;

+ p(:,s,r
is.r )
ab,ub ub,ub

(3-90a)

Mechanical

To complete the formulation of the equations of motion, dissipation

forces must be accounted for. This is accomplished by the use of the
Rayleigh dissipation function F defined by Eq, 1-123 as

+ J.1<p + KtL'f' -

vs,r
_ :!is,r
is,r
"b,ub -;
ab,ub ub,ub

In terms of Eqs. 3-77, 3-80, 3-82, and 3-83, the Lagrangian becomes
s
L( ,'.'
~, '",

a..l
<p

== ti~:'abl 2'~~,ab i~:'ab

,l

"'"

+ p(:,s,r
t-'s,r )
ub,ub ub,ub

Electrical

,~
.;.'"f
f~.l

..
Lijl;l)

r
i",r
a'b,ab ub,ub

This matrix of the equations of motion can be decomposed into an

electrical set and a mechanical set:

such can be expressed in either the form of a double summation or as a

matrix product. In terms of variables previously defined, W m can be
written as follows:

Wm == t('\~i~

fJtS

Liji;i),

197

(3-88)

Substitution of L from Eq. 3-84 and F from Eq, 3-87 into Eq. 3-88,
...,

'

== a</> + J<p: +

K'I' -

1 's r

'[t-ub,ub,

82's,r
"b,ub is.r
ab,ab

(3-90b)

Equations 3-90a and 3-90h are the complete equations of motion of the
generalized machine shown in Figs. 3-2 and 3-3.
At this point a comparison can be made between the two methods of
formulating the equations of motion for the generalized machine. Of
course, since the same physical model was assumed for both methods, the
resulting equations of motion are identical. This is verified by comparing
Eq. 3-60 with Eq. 3-90a and Eq. 3-75 with Eq. 3-90b. Since the starting
point (physical model, variables, and parameters) was the same for both
methods, the detailed manipulations are the same. However, in the use
of the basic force laws these manipulations were carried out in a series of
three apparently unrelated steps. In the Lagrangian formulation, on
the other hand, all the manipulations were carried out in a logical sequence
using information about the entire system. Thus, the Lagrangian method
is a simpler method from the viewpoint of formality, but the use of force
laws tends to give more insight into system interactions. In terms of
actual manipulations the two methods are equally long; therefore, which
method is chosen for a specific problem is a matter of taste. In more
complicated problems in which insight usually fails, the formalism of the
Lagrangian method leads to a more systematic (and presumably less
error-prone) formulation.
In later sections Eqs. 3-90a and 3-90b will be used to analyze the enetgy
conversion properties of the generalized machine. For this purpos!.the
equations of motion in terms of the variables and parameters will D~
needed. Thus, from Eqs. 3-56 and 3-57 the electrical parameters are as .
given in Eqs. 3-91 and 3-92:

L~

0
=

fES"

ob,ob

(LS; + L~) cos n4>

-(LpoJr

(LS; + LS;) cos n4>

- (LS; + LS;) sin n4>

L~ - L~l

(LS; - L::Z) sin n4>

(LS; - L',;/) cos n4>

+ L~l

+L

Jr

sin n,p

J.l2 )

(LS; -

V:z) sin n4>

(LS; -

L~)

+ L~2 cos 2n4>

L~

L~2

cos n4>

....m

- L:2sin 2n4>

L: - L:2 cos 2/l,p _

sin 2n,p

(3-91 )

.....

:r:

:>

R:

"

(3-92)

9t',;,ob

,~

K
:. ,'\.........-~ .""io." ~::..
:-c,
-<;-

..

R'

--;

'.

'.
.-

,- .,

." ,1"

,_,;sl

::;

.'

""~

(3-93)

-'- Vb

~
V"

(3-94)

(5
Z

vb

"

'b

.'.

<

v.ib..rb - v~

.and

.:.

/I

o
z
S
VII
S

'S

Jbl

is''
~
/lb,ob - i'

-<

~,

'/I

,-"

(;)
()

and from Eqs, 3-4b and 3-7 the electrical variables are:
''':

L.--

'~

:>
....

~ ~~~

3:

()

R:

Substitution of Eqs. 3-91-3-94 into Eqs. 3-9Oa and 3-90b yields the complete, detailed equations of motion for
the generalized machine, Eqs. 3-95a-d:-. .:
J

.";:,

;;,...

,"""

tt',.~

~.

~., ..... ~

~,~.#li;ii';~:fi:'<,

.~

,"-'''f~f:2'

.::-:,.

-,.
.\;-

.:~

-,'

A,
R S + p(L~ + L~z)

v~

Vb

~~<'

.~-

?-

;;;...

.;.."

Vo

["

'f ?

;~

.,--,

Electrical
S

' ;;.

J-

1;.
,
.:i4

o
R

+ L':z) cos n,p

peLS; + L:~)( - sin 114
peLS;

peLS; + LS;z) cos n4>

+ p(L~ -

peLS; -

b4 iyfi i

.. ,~}4

:"'"

JIi'J
~

..).

ll<;

<

_~..

,', _ill'l<f~~5''':;~~~':"i.,

L~;)

L~z)

sin n,p

peLS; +

pel',; - L',;) sin n4>

R'

+ p(L: + L: z cos 2n,p}

p(LS; - L::Z) cos n4>

- pL:z sin 2nrJ>

-f.

R'

L~;)( -

sin n'p)

1/1

's

Ib

- pL;.z sin 2n9

1/1

ii,

(3-95a)

....
N

"

+ p(L: - L:2 cos 2nr/>}

Krj> - n/2 x

.--

-(LS; + L':z) sin n4>

- (LS; + L::Z) cos n4>

(LS; - L::Z) cos n,p

- (LS; - L',;.J sin n</>

- (LS; + L',;) sin nrJ>

(L:; - L':z) cos n4>

- 2L:z sin 2nrJ>

- (L:; + L::Z) cos nrJ>

- (LS; - L::Z) sin n,p

,--.

2L~

- 2L;,

cos 2n4>

$~:

~~.

.~,

'S

1/1
'S

Ib

.,

~
~

C)

3:

:>

()

:r:

z
m

(3-95&)

" ,~

'-'

co~ 2n.~

2L:z sin 2n4>

.'

"

...;.

:>

;xl

l ....

j
~

;xl

+ ac/> +

<X.
I

C)

Mechanical

T= J~

-l

:r:
m

p( L',; - L',;) cos n,p

..

--..
~

...,~

':ri '

;;

:s

.""
200

THE GENERAL1ZED ROTATING MACHINE

ELECTROMECHANICAL ENERGY CONVERSION

T= (}1
If'

aJ..
'f'

K,~)
- nl[i'r'(l'"
'f~
t u b -'IA -I")
-'jJl - i'i"(L"
b
I~
(J

L")]cos
1-J2
~ n -l..
~

+ i:'i;;(L'; + L';)J sin n,p

- U:')2] sin 2n,p - 2L~/:'ib cos 2n,p}

" "Ii

deal about the important parameters for energy conversion in the generalized machine can be learned by investigating the general form of these
VOlt-ampere equations which were written in matrix form in Eq. 3-900 as

[if,i/:(L;: - L::)

+
T

N + ae/> +

L;,JU~)2

K,p - T,

(3-95c)

v~t,Qb = ~~;'Qb e"~:'ub

(3-95d)
;~~
:ii'

3.4

.l
~

Discussion of the Equations of Motion

The five equations of motion of the generalized machine are the four
volt-ampere equations (Eq. 3950) and the torque equation (Eq. 3-95b).
These live equations form a rather complex set of nonlinear differential
equations, the solution of which cannot be found in closed form for any
arbitrary set of conditions. It is necessary, therefore, to consider a
number of special cases for which solutions of the equations are possible.
One characteristic of the equations which is immediately apparent is that
the nonlinearities are the result of products of the independent variables
or functions of the independent variables. In the volt-ampere equation
the products are of the form of a current times some function of the angl~F'
,p, Thus. when the shart is constrained by a speed source such that <P is
explicitly expressed as a function of time, the volt-ampe"~ equations
become linear differential equations with time-varying coefficients.
Under such a constraint the volt-ampere equations are solvable, and there
are several changes of variables which greatly simplify the solution of
these equations.
Similarly the torque equation is simplified when the four currents in
the machine are constrained by current sources. The torque-speed
characteristics of the machine can thus be easily obtained for various
current constraints.
The technique of studying the generalized machine under different
constraints is particularly useful in determining many of the important
characteristics of this machine. Some of the important torque characteristics of the machine under current constraints and some of the important
VOlt-ampere characteristics under speed constraints will be studied in
detail. However, before this is done the important concepts of energy
conversion can be emphasized by a review of the equations of motion in
their general form.
3.4.1

,,~

~.

,
'.:!.',o.

"

......"

" .~ .

;..

;1'
,~;

'"

+ p(.fiJ:t,ab e"~:'Qb)

(3-900)

where the various matrix quantities are defined by Eqs. 3-4b, 3-7, 3-91,
and 3-92.
In this equation the time derivative p of the product .!I!~i:.(/I> <:i~.(/b can be
subdivided into two terms by recognizing that 2>~~,Db is a function of time
through its dependence on the angle,p. Thus the last term on the right
of Eq. 3-900 can be written as:
,I
",H I !
"

,
p(!lIS"ab,Db e""
ab,Qb) = !f.""
1Ib.1Ib pi'"
ab,flb

8!f.'S"
)
+ (~
1 i'"
8<jJ 'f'.
' ub,lIb

(3-96)

The term (8!f.'~Lb)/8,p, which is a result of the differentiation of the

product of the inductance and the current, has previously occurred 'in
Eq. 3-90b premultiplied and postmultiplied by the current to yield the
electromagnetic torque, and it also occurs in the volt-ampere equation
multiplied by a speed e/> and by a current to yield a voltage. This voltage
is due to mechanical motion and in fact is, in part, the energy conversion
term which appears in the volt-ampere equations to account for the
interaction of the electrical and mechanical systems. Thus the voltage
If~~.flb which accounts for energy conversion is defined as:

...

IfS"

=<

ab,ub

.j)

O!f.'s,r
o,p

~ 1

'P

is''
ab,ab

(3-970)

It is possible to relate the energy absorbed by this voltage from the

electrical source to the mechanical energy which is a result of conversion.
MUltiplication of Eq. 3900 by i~~Qb, yields

+ i~:'Qbl !f.'~~,Qb pi~;'ab +

i~;'abl v~t,1Ib = i~'["abl 9t~;'Qb i~:'Qb

'---v-------J'

electrical
input power .~'" fA 'h-J

v--i2R losses

i~:'Qb, 1f~~,Qb

power into coupling

field and conversion

.0';;

(3-97b)

The mechanical power supplied by electrical sources is obtained by

multiplying the electromagnetic torque T~, found from Eq. 3-90b as

'~.,t;
' '.' '.:
"if
~~r

A New Formulation of the Equations of Motion

The volt-ampere equations of the generalized machine, Eq. 3-95a.

consist of four simultaneous, nonlinear differential equations. A great

201

~".,
~'"

"I

afi>S"

T = .ie's,r
~

ab,ab,

o,p

is,r
ab,ab

(3-98)

by the mechanical speed e/> to obtain

T

1 -

,'I' -

.1 t 's,r

Z ab,ab,

Ifs"
ab,ab

(3-99)

~.

~t',

..
I

:j,

202

ELECTROMECHANICAL ENERGY CONVERSION

THE GENERALIZED ROTATING MACHINE

Thus half of the power e;b.ab, (f~~.ab absorbed by the voltage of energy
conversion is converted to mechanical power. The remainder of this
power goes into field storage.
Equation 3-97 b can now be put in the more meaningful form:

be called the energy conversion coefficient or torque coefficient, and it will

be designated by the symbol:

",
I

-,

'----~

electrical

"

/2R losses

e'~-;.ob, !f'~b.ab pe'~'b,ah

v

,
'L

:.'

Je'~'b.ab, (f~'b,ab
,

'f-

'I

power into magnetic field

(3-100)

mechanical form

,..

(e'~b.abl !t'~b.ab pe'~'b.ab

v s.r - ;Jf".r i,r

ab.ab ab,ab ab.ab

power converted to

This expression illustrates where all of the input electrical power goes.
An interesting interpretation of Eq. 3-100 can be made. Whenever
continuous energy conversion occurs, the term !e'~-;,ab, (f~:.ab must have an
average value; consequently, there appears to be an input power of equal
amount into the magnetic field, The amount of energy that can be stored
in the air gap of a machine is limited by saturation of magnetic materials
and by mechanical stresses; consequently, whenever continuous energy
conversion occurs the average power: flow into the magnetic field must be
zero:

+ te'~'b.llbl

6~b,ab)avl

== 0

(3-101)

Thus the power input to the field !e'~'b.ab, (f~-;,ah by the voltage of energy .
conversion is exactly balanced out by the first term in Eq. 3-101 when
continuous energy conversion occurs.' When the currents are constrained
to be constant the first term of Eq, 3-101 is zero because pe' = 0; thus it
is jmpossible to obtain continuous energy conversion by constraining'all
the coil currents to be constants. On the other hand, when the currents
are alternating currents, Eq. 3-101 can.-be satisfied with a nonzero average
value of the converted power ie';-;,ab, (f:;;,ab' Herein lies one version of
the whole problem of continuous energy conversion with rotating machines.
The alternating currents that are necessary for continuous energy conversion can be obtained from a-c sources, combiootions of a-c and doc
sources, or from a-c or doc sources fed through. n'lechanical frequency
changers (commutators). One of the objects of t~\ later sections of this
chapter is to demonstrate the current constraintt.\;hat are necessary for
continuous energy conversion with the generalize'ti Iflachine.
From the foregoing discussion it is apparent that the factor (8!t'~b,ab)/8q,
has special significance in the energy conversion terms in both the
electrical and mechanical equations. Thus, this parameter matrix will

(3-102)

The volt-ampere equations and the torque equations can now be

rewritten in terms of the energy conversion coefficients (Eq. 3-102) and
also in terms of the energy conversion voltage ff:,'b.ab (Eq. 3-97). Thus the
volt-ampere equation can be expressed in terms of .'T~'b,ob as:

~ i'

input power

+ !e:~b.ab, (f~-;.ab

or/>

"

e'~b.oh, v~'b.ah - e'~-;.ab, g,~'b.ob (;~-;,ab

frs,r
o!f'S"
ab,ab= ~

.
....

203

2 s., piS.'
ab,ab
ab,ab

+ frs,r
1 is,r
ab,ab'P ab,ab

(3-103)

or in terms of ~~b,ab as:

'..

v~b,ab - (f~b.nb = 9t~'b,ab e~'b.ab

+ 2~'b,abpe'~~ab

(3-104)

Similarly, the torque equation can be written as

~

ii"

T = J<j;

.'

+ ac/> + Kr/>

- T~

(3-1OS).w

where

11'")

Til

(3-106)

= ti~b':abl .:r~'b,abe'~b.ab

(3-107)

'Sf

or

i:

,'t.
:'.!~Ii>:..t.'

T.

-'::,iI:';';

~.'~;:.
--'!c."

The various matrix quantities in Eqs. 3-103-3-107 can be written in

terms of the coil variables using Eqs. 3-4b, 3-7, 3-91, 3-92, 3-970. and
3-102 as given in Eqs. 3-109-3-113.
\t'"

t;
"

:2 ~ e ab.ab, (fSf
ab,ab

~ =

-,1..

J i"

,*t'

"

. ~ ..

lV~:'

v~

. 1i.sO< -..

:fi~~
;;

~,):-

~I:'i

'if

~i

t'lOfd.'

vs,r
=
ab.ab

b
r
va

(3-108)

t:
_

. ab,ab -

[0 0~J ~""
0
0

R
0,

o (}

0
R'

(}

R'

V
(3-109)

II

. _.~ij

l S "

J
,$

is,r _ i"b
ab.ab - 'r
la

Vb

hi

s
la

,;1.1

(3-11(}}
Jh

'J
"."

f!

';'J

Equations 3-103 through 3-107 are particularly interesting in that they

show that the generalized machine can be represented by a network of
resistances [j~;'lIb and inductances !f'~b':ab with a driving voltage given by

~.;

',if.;!'':!:

'.

t!'
2<M

ELECTROMECHANICAL ENERGY CONVERSION

I
-es:::
C

'Vi
~

1=, I
~

"!
,"
0

-0
r~

~;: ..., .:t

-..J
+ I
~=l :;"
"-'
-..J

<5
U

';:;;

,-..J'"
I

'"U0

:::

":;:;

t'l

Vl

~~ ...o...l'"

+
"

-..J
I

:;-..J

"

~-..J

"-'

-e-

...

-e:::

"-'

~
I

0
~ '"
-..J

:::",

-..J

--N

c::;-

tA
"-'

--,

~o()

'::-..Q

''Q

-G

';:;;

'"0

s:::

'"00

.'

-e-

"

:;-..J

"-'

c::
'Vi

""9-

Vl

Vl

()

:;"
-..J

"

~.

s:::

Vl

Vl

s:::

-e---s:::
Vl

!II

""9-

s:::

:::

Vl

0
0
~:;: 'Vi
'
---I ...-..J'" t;;:'
-..J
I
-..J'" I

..

...
---~"

-..J'"

:;

:;"

""9-

....

'u;

o...l

~~

--~~

+
~"

"-'
-e-

s:::
I

~"
" "-'
Ii::

s:::

-e----

-e:::

'"00

"-'

s:::

0()

1>

:::
I

1';~.

"-'

..
.... ""
-ti
.....
t~

-..J"
~
>:::

-e-

Vl

,=

()

"-'

-..J
I

I
:;"
o...l

:; :I.

o...l

.::s:s:::
I

-e---s:::

---o...l"... ----...-..J"
+
,

'-'
~

...

:;o...l"

-..J'"

--

:;

'-e-

I ,,'"1

~""

~:I.

'f
I

s:::

s:::
I

,'J

~
s:::
!II

Vl

;'.1

~o

"-'

()

~;:

+
-e--->:::
c

Vl

"-'
..-..

"-'
-e-

~o

~a

Vl

"-'
-e-

:::

.~.

>:::

---~r

s:::
I

:;

+ +

---.., "
o...l
+
~

:;" :;"
-..J o...l
"-'

II

'-'

-------...
.."
o...l .o...l"

...

""
........ ..Q

Vl

..

+ ..+'"
...-..J" -..J
....
,

c:

'Vi

Vl

""9-

Ii::

.. ... "..
...-..J'" o...l

~:;:

+
..,o...l"

:::
c

""9-

>:::

'Vi

-e-

~'3

I
':"-....a
..-..

>:::

-e:::
c

'Vi

~~

,....."

"-'

-e-

':--~

Ii::

c:

:;-..J

-..J

'

s:::

..,""

~.()

:::

""9-

-e-

s:::

..

':4:1.

-..J'"
~

:::

"-'

~.
~

()

Vl

'Vi

-e~

!II

"-'

~Q
..-..

N
c

""9-

:;'"
-..J :; "
"-'
::: -..J

'-..J

"-'

s:::

filL

."1;M:t-

3.5 Torque Production and Energy Conversion

I
~Q

~
o...l
I

'-e~

-e-

:I.

205

-..J

~
>:::

~
..
---...-..J" . ,=
... '" .
.
+ -..J o...l" ':4"

s:::

.. 'Vic::.
"
-..J
"

~;:' ...-..J"
+ I

l"

:::
c::

'"

-e-

3...

s:::
N

'i

N
C

3-

':4'"

"

j~

'-'

>:::

...

..'

.-..

---

~,

c::

THE GENERALIZED ROTATING MACHINE

(v,r
The voltages v,,r
ab.tJb - $",r)
lib, lib
llb,Db are from sources , and the voitages
~~~.Ub are dependent voltages resulting from energy conversion, i.e"
dependent voltages resulting from currents in stator and rotor and motion
of the shaft. In addition. when using coil quantities the rower converted
to mechanical form is simply one-half of the product of the dependent
voltages of energy conversion multiplied by the current. This latter
relationship is often useful for determining the torque.

.,

In general. an investigation of the dynamic behavior of the generalized

machine, and in fact of any electromechanical energy converter, requires
the use of the complete set of differential equations which describe the
device. These differential equations are always nonlinear, containing at
least product nonlinearities as a result of the energy conversion, If,
however, the interest in a device (in this case the generalized machine) is in
whether or not the machine will continuously convert energy, then a
treatment of steady-state operation is in most cases sufficient. For example.
if a machine produces an average torque for a given excitation and
speed, then the machine can run at this speed driving a load equal to the
torque developed by the machine. Under these conditions energy is.
converted continuously. Thus the necessary conditions for energy conversion can be obtained by considering various excitations and
determining, when an average torque is obtained under conditions of
continuous rotation of the shaft. For this purpose the torque equation
of the generalized machine (Eq. 3.95b) will be studied when the machine
is subjected to various current constraints placed at the four electrical
terminal pairs.
The torque equation which describes, a rotating electromechanical
'."
energy converter, (Eq. 3-95d), has the form ""'~

,....."

~r

T=

~"
"-'

J> + a.f> + K</> - T.

.dJ

,1

(3-95d)

"

where T is the torque applied externally to the shaft and T. is the torque
developed internally and applied to the shaft by the electrical syste,m.
The mechanical power flowing into the machine from external mechanical
systems attached to the shaft is.

'-G-

Ii

I \." I \."" I

_iii

II

:;}

.~

"'+1~_: ~.t)

'~'J)f-:'~'l)(i~J

Pm = T.{>

and the power converted from electrical to mechanical form is ~

P.", ... T .f>

,It.'

(3-114)

P-115)

206

ELECTROMECHANICAL ENERGY CONVERSION

THE GENERALIZED ROTATING MACHINE

Motor action is defined by a conversion of electrical power to mechanical power (P.m > 0) and generator action is defined by a conversion
from mechanical power to electrical power (P.m < 0). By examining
Eqs. 3-114 and 3-115 the several possibilities given in Table 3-2 can be
distinguished. In all subsequent considerations of torque production
only positive rotation (~ > 0) will be considered. This restriction is
imposed in the interest of clarity and does not limit the generality of the
treatment. The extension of the results to negative rotation (,p < 0)
should be apparent.

Eq. 3-117, will be treated in detail. In deriving these current constraints

it is advantageous to recogni,.e thaI the torque produced by saliency in a
machine is somewhat difTcrcnt in character from the nonsalicnt or smoothair-gap torclue. Consequently, the current constraints will be comidcred
for two cases-a machine with a smooth air gap and a machine with
saliency.

TABLE 3-2.

.'

I~
Torque

T(.

Motor and Generator Operation

~>o

~<O

Motor
operation

T. > 0

T.< 0

Generator
operation

T. < 0

\}

~ ~

~> 0

T;> 0

< ejJ < 2'7T

------+------

(T. ).v8

I
I
r
r
1

. for

(3-116)

~ ,=?' 0."

(3-117)

A machine with the torque versus angld characteristic of Fig. 3-4 will run
successfully.
Energy conversion implies rotation in machines because power is the
product of a torque and a speed, and energy is converted only when the
speed is nonzero. Thus the current constraints for successful running,

,,~

~i

0'

:;

2'7r

~,.,

Angle cJ>

Stable point for a static

system Or for a stalled system

{v ....

,of'<'

(L f" T. dejJ) > 0

.~

Fig. 3-4. Torque variation with angle that is suitable for running but unsuitable for
starting.

When Eq. 3-116 is not satisfied the machine may come to rest at the stable
point shown in Fig. 3-4. When a machine has no rotor inertia the"
condition for successful running is also given by Eq. 3-116. All physical
machines have rotor inertia; consequently, the condition for the successful running of a machine is less restrictive than Eq. 3-116. Since kinetic'
energy is stored in the rotating inertia of the rotor, the condition for
successful running is that the average torque be greater than zero over a
revolution. In analytical form this condition is:

"

~....

I
I

r$f

',I

I
I
to',

for 0

)"

For consideration of torque production in a motor, two cases-starting

torque and running torque-must be distinguished. When a machine
has inertia in which to store kinetic energy, the conditions for successful
starting and running will be different. The condition for starting in the.
positive direction with no externally applied shaft torque T and with the'
spring constant K ... 0 is:
..
.

(T.)ava =

207

3.5.1

J::..';

Energy Conversion in a Machine with Uniform Air Gap

The torque equation of a salient-pole, two-phase, generalized machine

has been derived in a previous section and is given by Eq. 3-95c. The
electromagnetic torque T. consists of terms proportional to
and L:~.
The L:; terms are characteristic of a smooth-air-gap machine and the
L~ terms are those which arise from saliency. The electromagnetic
torque will therefore be divided into a uniform air-gap torque T,,, and a
saliency torque T'I'2 as

L::

" . .:~!

\it. '~ ,

;~:

T. = T."

where
and
I'
,

l
t
;p,iljN
P.i

+ T'I'2

.,

T.I/o = nL:;[U;ig - i;i:) cos nejJ - (ij;ig + i;i:) sin nejJ 1

(3-118)

(3-119)

T.I/oz = nL;~[ - U;ig + i~i:) cos n,p + (ibi~ -. i;i:) sin n.p1

+ nL;'z{[(i;Y

- (iy] sin 2nefJ - 2i:'i;' cos 2nq,}

(3-120)

..

ELECTROMECHANICAL ENERGY CONVERSION

THE GENERALIZED ROTATING MACHINE

To tlnd the torque production in a uniform air-gap machine, only the term
T". must be con~ilh:red because L;'~ = L~'2 = 0 for a uniform air-gap
machine.
It was stated previously that for consideration of whether or not a
machine will succes~fully convcrt encrgy it was only necessary to treat
sll:ady-state operation. Therefore, assume the speed eP to be a constant
and cqual to w""
eP
w", = constant
(3-121)
Hence the anglc 4> is
(3-122)
4) = wn.t + Ii

equal to n times the mechanical speed, where n is the number of pairs of

poles. The term nW m is the speed with which the' rotor moves relative to '
the stator in electrical space radians per second, since in travelling 21T
mechanical radians the rotor passes n pairs of stator poles, Using the
conventional definition of 21T electrical space radians per pair or poles,
Eq. 3-125 requires that for the production of an average torque the
stator and rotor frequencies must differ or add to an amount equal to the
relative speed between stator and rotor in electrical space radians per
second.
Some simple examples will now be considered. If the stator current is
constrained to be:
is = JS sin nw",1
(3-126)

208

For constant speed operation, the instantaneous torque Te ,. becomes:

T,,.

= nL;:[(i;ib -

ii/~)

cos n(wml

Ii) - (iEib + i~i~) sin n(w n ,!

-."I,. ~

or

This expression for torque contains four terms that are triple products;
these are:
I~
(
(i~ii:) cos n( wml + 8)
-(i~i~)cosn(Wn'! + ill
-(i~i,:)

sin n(w",1

-(if,i,~)

0)

sin n(w n,!

B sin (nwmt)

il)

+ ...

(3-127)

(3-128)

(3-129)

wI

then for average torque production the rotor current must l;>e given by:
i' = l' sin ( W nwm)t

(3-130)
\

(3-124)

This can be shown to be the induction motor or a synchronous-induction

machine, depending upon how the rotor excitation is realized. In the
next chapter the types of machines that can be produced by specific
excitations and constraints will be examined in detail.
In summary, the conditions for the production of average torque, Eq.
3-125, can be stated as follows:
,
Whenever a current of frequency w (w may be a function of wm ) flows
in a winding of one member (rotor or stator) due to impressed or induced
voltages or a combination of the two, then the only current in the other
member with which the original current will interact to produce an average
torque must have the frequency w nw m , where n is the number of
pairs of poles in the machine.
Consequently, with fixed frequency currents on rotor and stator, the
machine will run only at "synchronous" speed. In order to have torque
over a continuous range of speeds, the currents on one member must have
a varying frequency that follows the speed variations. This varying

"!,,,

'~

r'"

is = JS sin

A necessary condition for Eq. 3-124 to be satisfied is that nW m be the

sum or the difference in frequency of a rotor current and a stator current.
Thus for the production of an average torque under conditions of constant
speed operation the stator frequency W S and the rotor frequency w'
must be related by:
:~ r
(3-125)
W' = ws nW m

i' = l' sin 2nw m l

These constraints produce a conventional synchro~~us machine for

i' = constant, and a synchronous-induction machine for i' = l' sin 2nw m t.
If, on the other hand, the stator current is constrained to be

.J-{,

= A cos (nwmt)

,,;n

i' = l' ... constant

Each of these four terms has the form of a product of a rotor current, a
stator current. and a sine or cosine function of n(wml + 0). It is readily
apparent from the orthogonality condition 01' sine and cosine functions
that these four terms can produce an average value only if the product of
a rotor and a stator current produces a term which is a sine or cosine
function of nw",l. Thus a necessary condition for production of an
aL'crage torque is that the product of one rotor current irolor and one
stator current iSlator have the form:
(irolOr)(islalOr)

209

then there are two possibilities for i':

0)]

(3-123)
~

If the stator and rotor currents contain several frequency components,

then an average torque can only result from the interaction of those
components of the currents whose frequencies satisfy Eq. 3-125. The
frequencies of stator and rotor, therefore, must have a sum or difference

t
''..;

210

I
~

ELECTROMECHANICAL ENERGY CONVERSION

THE GENERALIZED ROTATING MACHINE

"":~?i;:'.,'.

frequency is automatically obtained if currents are induced electromagnetically in one member from the other member. The currents may also be
obtained from controllable variable frequency sources, or they may be
obtained by physical constraints (a commutator, for instance) so that the
nW m ji'cquency dU]'rence is introduced mechanically. These various
possibilities will be investigated in detail later.

frequency nw m.
torque is

~~,
"J" .
~t
""

Thus one condition for the production of an average

irotor =

A sin nwmt

B cos nwmt

+ ...

(3-133)

A second possibility for the production of an average torque arises from

the term proportional to i~ib' It is readily apparent that an average torque
can resL:' jf one of the rotor coil currents is a direct current and the other
coil current has a frequency of 2nw",. If w~ is defined as the frequency of
i; and w~ as the frequency of ib. an average torque can result for any of the
following three conditions:

3.5.2 Energy Conversion in a Machine with Saliency

211

'"~ii"

In a salient structure the electromagnetic torque consists of two parts,

T,,,, and T"<2' The term T,,,, has already been treated in the preceding
section, so only the term T.I'z must be studied.
given in terms of coil variables as:

From Eq. 3-120, T''''2 is

T''''2 = nL:.';[ -(i~i~ + ii,i~) cos n4> + (i;i~ - i~i~) sin n4>]
+nL:z{[Ub)2 - U;)2] sin 2n4> - 2i;ibcos 2n4>}

'!~\ l;:

'j'

'~"~."

r;

(3-131)

).

...."',

,.::

}, ':~; ,:
....

21~ib

(3-132)

cos 2m/>}

:,'.

"~ ". :
;':.

~':'

L:2{[U;)2 - (iJ2] sin 2n4> -

-,

To ;,;:

,',

k, .

'. -j'.'

,
~:. ~}

These terms produce the reluctance torque-a torque resulting from the
rotor currents and the stator saliency.
Now consider, as in Sec. 3.5.1, that the mechanical speed is constant and
the angle 4> is given by Eq. 3-122: ,.dJ 'Ii,) 1 , , '.;
;,;J
'1;
4> = wmt

+ S

';oll:? Ml

.1WI~

''I;'

(3-1,22)

"f:

The reluctance torque (Eq. 3-132) consists of three terms; these are:

(ib)2 sin 2n(w",t

+ S)

-U;)2 sin 2n (w",t + S)

-2i;ibcos 2n(w mt

':ol~,~.

-',

nW m
nW m

2nw..

2nw m

Case 3

~\

The production of an average value of the reluctance torque requires

rotor currents whose frequencies are equal to the mechanical speed
expressed in electrical space radians per second as nw"" or whose
frequencies are such that
= w~ 2nw m In most practical machines
currents of two different frequencies are not found on the rotor; thus the
only practical condition for the production of an average reluctance
torque is to have rotor currents which have a frequency equal to the
mechanical speed expressed in electrical space radians per second. The
reluctance torque is seen to have a nonzero average value in a synchronoustype machine.
In general, there are three types of torque terms for a machine with
saliency on the stator and excitation on both stator and rotor:

w:

1. the interaction of a stator current with a rotor current for which the
frequency condition is given in Eq. 3-l25

w' = w$

nWm

(3-134)

2, the interaction of a rotor current with another rotor current, for

which the frequency condition is

-';-

w~ =

+ S)
.,

The currents in these three expressions are rotor currents and, furthermore,
two of the terms contain the square of a rotor current. The two terms
proportional to the square of a rotor current can produce an average
value only ,if the rotor currents contain a component which has a
~:-.~ .
"'~l;' \
(>"

'

-,::!{:

It

Case 2

'

...

The coefficients of cos n4> and sin n4> in Eq. 3-131 have the same properties
and, in fact, are identical in form to T.", for the smooth-air-gap structure
that was treated in the preceding section. Therefore, from the condItions
expressed by Eq. 3-125 it is apparent that these terms can produce an
average torque only if the rotor current and stator current frequencies
differ by or add to the speed nw"" where .p = W m The only additional
conditions which will be introduced by saliency will come from the last
' .
two terms of Eq. 3-131:

w~

w;

Case I

..'...'1

wb

2nwm

(3-135)

3. the interaction of a rotor current with itself for which the frequency
condition is
,~

,A
.r. comparison
1

w; = nWm

or

w; = nWm

(3-136)

of these three frequency conditions shows that all three

212

ELECTROMECHANICAL ENERGY CONVERSION

w' = nW m

and

or in the case where

and

ul = 2nwm

W'

nW m

The conditions which the frequencies of stator and rotor currents must
satisfy in order to produce an average torque and hence give successful
energy conversion arc summarized in Eqs. 3-134, 3-135, and 3-136.
When a machine has a smooth rotor, Eq. 3-134 applies; for a salient-pole
machine, all three expressions may lead to an average torque.

Phase a-Stator
~

v:; I',

~ {:,

(3-137)

z;~

Vb

L~)[cos n(w m'

-(Ls,:

+ o)]i;

o)]pi~

+ L~)[sin n(wmt +

8)Jpi; (3-139)

',I

nwm(L'; - L~;)[cos n(wmt + 8)]i; + nwm(L'; - L~)[sin n(w m' + S)]i;

... (R S + (L~ - L:2)pJii, + (Ls,: - L~)(sin n(wmt + 8)]pi;"
+ (L'; - L~)(cos n(wmt + 8)]pi; (3-140)
J"

+ nwm(Ls,: + L::;)(sin n(wmt +

8)]i; - nwm(L'; - Ls,:2)[COS n(wmt + S)]ib

+2nwmL:2(sin 2n(w mt + 8)Ji; + 2nwmL~2(cos 2n(w m t + 8))i~
= {R' + (L: + L:2 cos 2n(wmt + 8))p}i;
[sin 2n(w mt + 8)]pi;
2
+ (L'; + L::;)[cos n(wmt + 8)]pi~ + (L'; - L:.';)[sin n(wmt + S)]pi:
(3-141)

Phase b-Rotor
+ nwm(L'; + L::;)(cos n(wmt + 8))i~ + nwm(L'; - L:;)(sinn(wmt + 8))i:
+ 2nwmL:2(cos 2n(w mt + S)]i~ - 2nw mL:2(sin 2n(w mt + 8)]ib
'.. {R' + [L: - L:2 cos 2n(wmt + 8)]p}i; [sin 2n(w mt + 8)Jpi;
2
- (Ls,: + L:.';)(sin ~(wmt + S)Jpi; + (L'; - L::2)(cos n(wml + 8)]pit

L:

'/.,'

'j.).

;n

(3-138) "

A ~onstanl specd constraint inherently assumes that the spring constant K in thClJ
torque relationship, Eq. )-95<1. is zero.

+ nwm(Ls,: +

w"',

then the volt-ampere equations reduce to linear differential equations with

time-varying coefficients. This latter case of constant speed operation
proves to be a very practical case which can be used to find volt-ampere
chMacteristics of many practical machine types. In Sec. 3.5.1 it was
shown that for currents of fixed frequency w' on the stator and w' on the
rotor only the speeds given by W m = (W S w')/n will produce an average
torque by interaction of a rotor current and a stator current. For
steady-state operation, then, a speed constraint should allow the determination of the volt-ampere characteristics for any given speed of any
machine type represented by the generalized machine.

'

S)]i~

-L:

'0

and the instantaneous position of the shaft is

wmt

(,

Phase a-Rotor
it-

,p

'l,

Phase b-Stator

The volt-ampere equations contain the shaft posltlOn,p. When the

motion of the shaft is not known explicitly as a function of time and
must be determined from the mechanical equation of motion (Eq. 3-95b),
it is impossible to find general solutions for the voltages and currenL in
closed form. The volt-ampere equations become linear differential
equations whenever the shaft position ,p is known. For example, if the
position of the shaft is held fixed, the volt-ampere equations are linear
differential equations with constant coefficients. Similarly, if the speed
of the shaft is maintained constant by a speed source such that*

,p

The Volt-Ampere Equations for Constant-Speed

Constraints

WIll

L~~)[sin n(wmt

!O

,p =

+ nwm(L:: +

= (R' + (L: + L~2)P]i~ + (Ls,: + L~)(cos n(wmt +

.(

3.6

213

For ,p and .p given by Eqs. 3-137 and 3-138, the four, volt-ampere
equations, defined in matrix notation by Eq. 3-104, are:, ,'"

torque terms give nonzero average values only in a conventional synchronous machine where
w'

THE GENERALIZED ROTATING MACHINE

'"

'

(3-142)

The four equations (3-139-3-142) are written with the source voltages
minus the voltages of energy conversion on the left of the equal sign and
the resistance and inductance terms on the right of the equal sign. This
is done so that in later examples the energy conversion voltages will be
explicitly defined and their effects on the performance of the machine can
be observed.
Several specific examples which are chosen to show the essential voltampere characteristics obtainable with the generalized machine will be
treated.
3.6.1

The Single-Phase Synchronous Machine

A single-p"ftase synchronous machine has one winding on the stator and

one on the rotor. The field winding is usually OIl the salient member
and is excited by direct current. The generalized machine d.cscribes a

..

~t'~;

214

215

ELECTROMECHANICAL ENERGY CONVERSION

THE GENERALIZED ROTATING MACHINE

single-phase synchronous machine when the two b phases, one on the

rotor and one on the stator, are removed. Thus the constraints necessary
to reduce the generalized machine to a single-phase synchronous machine
are:

electrical radians per second, the amplitudes of the mutual inductances

between stator and rotor, and the magnitude of the lIe1d current j' which
under these conditions is given by Eq. 3-145 as j' = - V'/R". The frequency and relative phase of the open-circuit terminal voltage is
determined by the position r/> of the rotor, which gives a frequency of
nw", and a phase angle of nS.
When the single-phase synchronous machine delivers a load, i.e., when
i~ i' 0, the simple character of the machine is altered. This occurs because
the rotor current induces an additional alternating voltage in the field
circuit such that i' is no longer a constant direct current. This interaction between stator and rotor leads to an inllnite sequence of even
harmonics of nw", in the stator and an infinite sequence of odd harmonics
of nW m in the rotor. Consequently, the single-phase synchronous machine
under load becomes too complicated for further investigation of its energy
conversion characteristics at this time.

i~ =

i'

v~ =

it, and

V' direct voltage source

v~ do not exist (b coil removed from stator)

i; and v~ do not exist (b coil removed from rotor)

~ = w",
cP "" w",t

.,.~j'l'

The choice of v~ = - V' is arbitrary and is made so that the resulting

expressions will occur in a form which is often used in studying synchronous machines. Obviously it is not a required condition. Using
these constraints, the two volt-ampere equations of the single-phase
synchronous machine become:
, '1
\"

- V'

+ nw",(L:: + L~)[sin n(w",t + 8)]i;

... [R' + (L~ + L~l)P]i' + (L:: + L~) cos n(w",t +

8)]pi; (3-143)

3.6.2 The Smooth-Air-Gap Two-Phase Synchronous Machine

~.
~ ..

t~' ,

v; + nw",(L:: + L~)[sin n(w",t + 8)]i' + 2nw",L~ sin 2n(w",t + 8)1;

= {R

(L~. + L~3 cos 2n(w",t

S)]p}i;

+ (L;;' + L~) cos n(w",t +

8)]pi'

(3-144)
First the synchronous machine is assumed to be a generator and the
rotor circuit is open. Then i; = 0 and, under a steady-state condition,
pi' "" 0 since the voltage ... V' is a direct voltage. Thus. there results
;,

"I

.'
., R'l' \~.l

Tn",

and

rr,f

'.j'

..

'.1-.

~.,

-,,-

\ I

"

,:;W;j':\o'th.
~'~'q

var -

nw Y6

----!!!.R'

(L" 1
+ 'L2 )
Ir

Ir

! ~i '
.,

(3-145)

:l.:: J'lt

' \ ' .'

sm new", t + 8) "" 0 II .,

(3-146)

Under these conditions, the open-circuit rotor voltage is a sinusoidal

voltage of frequency nw", and from Eq. 3-146 the magnitude of this
: open-circuit voltage is

Iv;lopen c:lrcult

= nw",(L::

The analysis of the single-phase synchronous machine becomes very

complicated under load conditions because of the interaction between
stator and rotor currents. A second example, which is much more
enlightening, is the two-phase synchronous machine, with a smooth air
gap. Under steady-state conditions with balanced loads on the two
phases, the interaction between stator and rotor is eliminated. For these
conditions of operation a useful steady-state equivalent circuit for the
synchronous machine can be derived, and its steady-state torque
characteristics can be determined. The two-phase synchronous machine
has a single winding on one magnetic member and a two-phase winding
on the other magnetic member. Considering a smooth-air-gap machine,
the volt-ampere equatIons for the generalized machine can be reduced to
those of a two-phase synchronous machine by the following constraints:

t"
'!'.
.

i~

i$
it, and v~ do not exist (6 winding removed)
v~ = - V' = constant

j~'"
(L

L' = L" = Lr =
J,!;Z

~2.

JlZ

.l=wm
'f

V'

+ ,L:;') W'!1'!o1!'Jlllf1 .,rttrl,

"\1

r/> = w",t

t' ,,}.

.,,,

" ,,~,),.J

~~b,;.~

rt';~f
,~~

..
rt...

',f'

':J
i

Vi..

8
.1

Thus, the single-phase synchronous generator has an open-circuit terminal

voltage whose magnitude is determined by the mechanical speed nw", in

Using these constraints and again recognizing that the choice of the;'t
stator voltage
to be - V' is to obtain a conventional representation for

v;,

,)t

,~~~~\

216

no

the machine in which

Eqs.3-139-3-142,
- V'

THE GENERALIZED ROTATING MACHINE

ELECTROMECHANICAL ENERGY CONVERSION

< 17, the volt-ampere equations are, from

+ nw",L;:[sin n(w",' + &)]i~ + nw",L;:[cos n(w m ' + S)Jih

= (R' + L;,p)i"'+ I_;;[cos n(w,.,1 + 8)]pi~ - L;;[sin n(w.,,1 +

S)]pi~

(3-147)
v:

+ nw,.,L;:[sin n(w n,!

I-

SW' = (R'

+ L~,p)i~ +

[L;: cos n(w,.,1

v~

+ Ilw",L;:[cos n(w,.,1

+ S)]i' = (R

+ S)]pi'

lJ(I

'J

- [L:: sin n(wn ,! + o)]pi'

(3-149)

vIIr

Vb

(3150)

R'i'

,r
= +nw,., L "

V' . (
R' Sin n wml

V'

+ 0<')

'i,t) J&

= +nwmL:: R' cos n(wmt

V(~

0)

":Ufi

(3-152)

(3-153)

In these equations for the phase currents the angle 0' is as yet undefined
since the zero time reference has not been specified. The three time
functions which must be considered in choosing the zero time reference
arc the current, voltage, and position </> = w",' + o. In the treatment of
A stahle operating point will result when the stator mmf oppOSO::S tho:: rotor mmr
(i.e. Len/'s law), This cc>ndition will occur when /IS excl'eds rr for the direction of
currents assumed in Fig. 3-1. To keep no < rr a negative stator or rotor current can
be chosen. In this development the stator (field) current is reversed.

',\ ,
'

..

eo
=
.

'4J~

"

dt

;,

'J \

0['"h n ',f
dA btl
~u. ,;,," = -dl
V'Y~
~".

<

(3-151)

For open-circuit, steady-state conditions, the two-phase synchronous

generator is similar to the single-phase machine treated in Sec. 3.6.1.
except that the two-phase rotor winding has a balanced two-phase
generated voltage at the terminals of the a and b phases on the rotor.
As long as the field current is is nontime-varying, it is evident from Eqs.
3-148 and 3-149 that the generated voltage of the machine is equal to the
open-circuit voltage of Eqs. 3-151 and 3-152 even under conditions of load.
Assuming then that the field current is is constrained by a constant doc
source to be - V'/R', it is apparent from Eqs. 3-148 and 3-[49 that for
any balanced passive load the stator and rotor currents must, under
steady-state conditions, always be of the following general form:
i~ = l' sin (nwml - 8')
ib = l' cos (nw m ' - 0')

8</>

=dA
- au
-

i;-COnJlnnt

'

(3-154)

'J

'a-conslunt

",,1)... ,.,. ,:. . "k'

'.

Referrmg to Eqs. 3-46 and 3-47 where L~~ and L'ta are defined, it follows
from Eqs. 3- 154 that e~ and e~ depend on time through their dependence
on the electrical angular posi tion n,p = nWml + no.
The generated voltages of Eqs. 3- 154 are also given by Eqs. 3-151 and
3152 since at zero load v~ =
and v~ = eb. For the condition of zero
load U; = i~ = 0) the torque angle
= O. Thus by choosing the termi"nal voltages u~ and v~ as the time reference with zero phase, for any
condition of load, the relative angle between v~ and e~ and similarly
between Vb and e~ will be the electrical torque angle no. This choice of
the terminal voltage as the time reference means that the angle 0' defined
by Eqs. 3-153 is the phase angle between the terminal voltage and
terminal current of either rotor phase a or rotor phase b.
l~"
'- With the gene;al form of i; and ibfor a balance<;1load under steady-state
conditions established, the field circuit equation (Eq. 3147) can be
investigated. Substituting i; and ibfrom Eqs. 3-153 into Eq. 3-147 yields

For open-circuit conditions on the rotor the volt-ampere equations

3-147,3-148, and 3-149 reduce to the following:

"

rs

ls
<pI

rs

_ V'

+ L~p)i~

rs

8Le r ... -

(3-148)
r

2/7

synchronous machines, it is common practice to choose the. torqut: ,lOgle

(no) as the relative angle net ween the generated voltage and the terminal
voltage. DeJining the generated voltage as in Eqs. 3-')7(J and 3-113
gives the generated voltages in phase a and phase b on the rotor due to a
steady direct current in the a phase of the stator as

.",1.

no

"&.

- VI

+ nwmL:;J'[sin n(w",t +

= (~I + L~p)i' +

The terms in brackets on the right and left sides of Eq. 3-155 are equal;
thus the field circuit equation reduces to simply:

l
I1,'.

"

'.;~

,I

0) sin (nwml - 8')

+ cos n(wml + 0) cos (nw",t - 8')]
nwmL::1'[cos n(w m ' + &) cos (nwml - 0')
+ sin n(wml + 0) sin (nw",t - 6')] (3155)

"'~

;! ;!
,~
:~

'I~

- V' = (R'

(3156)

Consequently, for a balanced load on the rotor and steady-state conditions,

the speed voltage induced by rotation into the field circuit by the a phase
rotor current i~ is canceled by the transformer voltage induced into the
field circuit by the time rate of change of the b phase current ib. Similarly,
in the field circuit the spee<;l v.9Jtage resulting from ib is canceled by the
transformer voltage due to
Thus, the field circuit under balanced

i;.

~
',I

'oj

+ L~p)iS

~,:
,-".,~.i

ELECTROMECHANICAL ENERGY CONVERSION

218

(Eqs. 3-160-3-161) as excitations, and recognlzmg that w'

vOlt-ampere equations (Eqs. 3-158 and 3-159) become

load and steady-state operation is independent of the load circuit. This

leads to the simple result that for balanced loads and steady-state conditions the field current is simply - Y'/ K, a constant; hence the three
volt-ampere equations become:

_ Y' = Ki'

Y' sin w't - w'L:;

;~ .

Y'

+ 0~)

= (R'

+ L')"
IlP 'a

(3-158)

nwmL:'; ~ cos n(wmt

+ 8)

= (R'

+ L~p)ib

(3-159)

Y' cos w't - w'L:'; Rs cos (w't

:~:
'!Il<\",
it

f~'"
"

The steady-state volt-ampere equations given by Eqs. 3-157-3-159 were

derived by considering the two-phase synchronous machine as a balanced
generator. They are equally valid, however, for motor action, providing
the rotor excitations v: and vb' which have a frequency w' = nWml, are
balanced (the same amplitude) and have the relative phase relationship
indicated by Eqs. 3-151 and 3-152. Maintaining the terminal voltages
as the zero time reference so that the relative displacement between the
terminal voltage and generated voltage is no, the general form of tll-e
terminal voltage for either motor or generator action will be

v;
vb

-= Y' sin w't

(3-160)
,~

Y' cos w't

~: sin (w't + no)

'\

= nwm ,

the

L~p)j~

(3-162)

+ L;p)i~

(3-163)

= (R' +

(3-157)

Y-. (
va, - nw", L"
Il R' SID n wmt

Vb -

219

THE GENERALIZED ROTATING MACHINE

;i,

'i

....

'

'

, (3-161)

where w' = nw m. With the terminal voltages defined by Eqs. 3-160 and
3-161 and steady-state operating conditions, the volt-ampere relationships
for either motor or generator operation are given by Eqs. 3-157-3-159.
Furthermore, from the investigation of torque in Sec. 3.5.1, it is known

nS)

(R'

It is quite evident from Eqs. 3-162 and 3-163 that the rotor currents i;
and i b for steady-state conditions must have a frequency of w'. This of
course is essential since the previous result that is was independent of the
rotor currents under a steady-state condition was based on Eqs. 3-154 and
3-155, which contained the assumption that the rotor currents have a
frequency w' = nw",. Caution must be exercised in the use of Eqs. 3-162
and 3-163. They contain the time derivative p, but they cannot be
used for transient studies because they were derived assuming steady-state
operation in which is was not time-varying. Only steady-state values of
i~ and ib may be obtained from Eqs. 3-162 and 3-163 unless the current
. i' is constrained by a current source, which, for this example, is not the
"~ '.
case.
To avoid mistakes it is advantageous to express Eqs. 3-162 and 3-163
Vit, in terms of steady-state complex magnitudes instead of in the form of
instantaneous quantities. The adopted convention for steady-state
complex magnitudes (phasors) is to define the current as:
i 1 = 11 cos (wt

+ 8)

= Re {[/1e19 ]e}wl}

= Re {/ 1e1w1 }

that for an average torque to exist the frequencies of stator and rotor
"currents must be related so that w = w' nw m. In the synchronous
machine w' = 0, doc excitation; thus w' must equal nW m for an average
torque, but this is exactly the frequency required to yield the simple
steady-state volt-ampere equations which have been derived.
The steady-state volt-ampere equations have been found, and it has
been shown that an average torque exists for the assumed conditions of
operation, namely mechanical speed ~ = w"', rotor frequency w' = nw m ,
stator frequency W S = O. Next, the torque characteristic of the synchronous machine will be found in terms of its excitation, Eqs. 3-160-3-161.
",' The terminal voltage of the rotor has the same frequency and phase
sequence as the induced voltages due to the stator current. The phase
angle of the rotor terminal voltages has been chosen as zero; thus these
voltages are taken as the time reference. With the rotor terminal voltages

.c,

"j.

':~

,t i 2

= 12 sin (wI

+ 8) =
=

(3-164)

Re {[ -jJze18]elwl} = Re {[Ize J(h,/2)]e iwl }

Re {/2e1w1 }
(3-165)

~~;

>

,f...,. ".'"
~;

~;

".

"...

,I';

Thus a complex amplitude, often called a phasor, is defined as I = lei 8

and the cosine function Re [e Jw1 ] is taken as the time reference. Using
the complex notation of Eqs. 3-164 and 3-165, the rotor volt-ampere
equations, Eqs. 3-162 and 3-163, may be converted to complex form,
recognizing that p = jw', thus:
-jV'eiO

+ jw'L:; ;: eM

= (R'

+ jw'L~)I:

'UJ

1':\,rt

(3-166)

i;~t. f~

"

V'eiO _ w'Ls , Y'

'" R'

J{J

.j

;"i~'~

'le

""

'..

',."

,i

eln~ = (R' + J'w'L')/,

'"

(3-167)

"

,
+

,,~,

"

220

ELECTROMECHANICAL ENERGY CONVERSION

It is quite clear that Eqs. l-l66 and 3-167 are identical in form and that
I; and I;' arc equal in magnitude and also satisfy the relationship that

I; =

(3-168)

-j/~

/'
'11'

~)
,.

,.,',,'

41>
~

'l'

where

i;.

E" ~ :'

ita}

y' =
Er

Vb = jY; = Y'e
Eb' = J'E' = w'Lsr R
y
jO

I'

I' = I; = jl; = l'rje'

zr = (R' + jwrL~) = (R'
=

J
J

e inlJ

~~

.........-......

I'

.....,.

...

'(J'r~
(

- 1I'}1
I
I
(11'/2 _I/Ir) I
I
I
I"

IJ.. '

+ jX') =

V(R')2:t(w'L~)2eW

.;1

\!,,"

hA:<

zreN'

.. ,
__)...

____________ J

w'L'

The volt-ampere equation for each phase of a two-phase synchronous oJ

machine which is operating under balanced, steady-state conditions can
be represented as a generator with an internal generated voltage of E',
with an internal impedance of zr, and with an applied phase voltage of
yr. An equivalent circuit and phasor diagram for the volt-ampere
relationship of Eq. 3-169 is drawn in Fig. 3-5. In this representation of
the synchronous machine it is common practice to call the reactance'
X r = w'L;, the synchronous reactance of the machine. In this equivalent
circuit (Fig. 3-5a) it should also be recognized that the voltages v~ and vb
were defined in the original formulation of the equations of motion as the
applied voltage to each phase and that the currents were positive into each
phase (see Fig. 3-3). Thus the voltage drop across a load Zj connected
to the terminals is - i;Zj or - i;'Zj for phase a or b respectively. The
equivalent circuit of Fig. 3-5 is equally valid for either motor or generator
action as long as balanced, steady-state conditions are maintained. The
simple equivalent circuit of Fig. 3-5 is often used for the study of synchronous machines under other than steady-state balanced conditions.
This is sometimes permissible, but it is an approx.imation, and one must
remember that for the general case Eqs. 3-147,3-148, and 3-149 are the
only valid equations for studying electric transients at constant speed.

E'
I
I

== Ere jlllJ

if/ = tan-\ ----f

R;'~

.'

221

and if speed variations exist the general equations of motion given by

Eqs. 3-95a and 3-95b may be required.
The current I' in Eq. 3-169 and Fig. 3-5 was defined to have a magnitude
r and phase 0', and these quantities are consistent with the I' and or
defined in Eq. 3-153 for the rotor currents and
The phasor diagram

i;

Equations 3-166 and 3-167 can be represented by one equation by using

the relationship between I; and I; defined by Eq. 3-168. Furthermore,
the quantity w r L::( V RJ)ej,,~ is the complex. magnitude of the voltage of
energy conversion defmed by Eq. 3-168, and it results from rotation of
the rotor and current in the stator. This quantity is the internal generated
r
voltage of the synchronout; machine, and it will be designated by E
The two volt-ampere relations of Eqs. 3-166 and 3-167 can he written as
(3-169)
yr _ E r = Zrl'

THE GENERALIZED ROTATING MACHINE

;ilo

.~

~J

...."

.~

......._<.,;,..~~, ...._~'\.

_I'
(tt~.(}

,l_._.

'

II

(b)

".
i"' ....
~"
90':"--"'--.I
......

">~.
,

., "

Fig. 3-5. Equivalent circuit and phasor diagram. for two-phase, synchronous machino
with balanced load or balanced excitation.

of Fig. 3-5 can therefore be used to solve for rand 0' in terms of the
excitations and parameters of the synchronous machine. To do this it
is advantageous to draw another p,hasor diagram in which r is decomposed
into two currents, I~ resultingJrom the terminal voltage and IE resulting
from the generated voltage';E": Jhe phasor diagram in terms of 1', I~,
and IE is drawn in Fig. 3-6. The current I', as given by Eq. 3-169, is

l' = I' e-/s' = I'v - 1;'

..

V' i
e- .'
Z'

E'
e/(n8-",')
Z'

- -

"= -

(3-170)

y'
:.

.." . .

V(R')2

w'L ' V'S

S

e!l':Jo/J'

+ (~;'L~)2"

... R

ei(nlJ-.')

.
,

I,

.t,-, "

~~l1l'<~';.

or

.~

'

,-----'

222

t
,~

ELECTROMECHANICAL ENERGY CONVERSION

iJ;,

~'

'~

(
~ -~:ti'~

; ),lit

+i

"",,:",

223

THE GENERALIZED ROTATING MACHINE

,.

'.....~l,;
'. .

expressions for r which will prove useful later for deriving the torque
equation; these are

[' =

E' cos n8 - V'

zr sin (8' - .pr - 'IT/2)

= vr

E cos nt>

(3-174)

zr cos (8' _ .p')

and
[' =

'~" '

ltl;

;i

.:~~.

l;

"~

,/'

FIg. 3-6. Phasor diagram for balanced two-phase synchronous machine of Fig. 3-5
with l' decomposed into two components.
.

_.:.--.-......

'\1'

The magnitude of l' is obtained from the law of cosines as

1'-

j(

V r)2

-~,:~,

(o/L" VS)2
_ (2V.rwrL" VS)
cos n8
"'R s
"'R r
r
(R )2 + (wrl~)2

The phase angle or is obtained from the components of /' parallel and
perpendicular to r as

T~",

,! . "

,\

, 11' ",

1'sin (or - 'IT)

tan (0' - 'IT) = rcos (or _ '11')

l',i~'i

~.,

(3-172)

.
and the right-hand side of Eq. 3-172 can :be expressed in ,terms oC,lv and
.

IE as

.....

'
or - 'IT

= tan- 1

.. ',

y
D::l.
yr sin ifl - w rL:: RS
sins
(if/ ,
-' n8) ] ",.
ys
(3-173)
W cos ifl - wrL:: R S cos (ifl - n8)

This establishes the magnitude l' and phase or of the rotor currents in
terms of the known excitations and parameters. There are two other

'i".

T~", =

= nL';[U;it - iF):J cos ntl> - U;it +

"~'-

.'Y,
.;..'

i;i~) sin

ys

+nL';{ + R" ['[cos (w't - or) cos n(wmt + 8)

'I/'i>.:,

ii; ;,:'

(3-175)

ntl

(3-119)

Substitution of the rotor currents given by Eqs. 3-153 and the stator
currents it, = 0, i~ = jS = - V'/ W yields

"

of

E' sin n8
Z' sin (0' _ .pr)

Equations 3-174 and 3-175 are derived directly from the phasor diagram
of Fig. 3-6 by projecting Er and rzr onto the real and imaginary axes of
the diagram.
Now that the steady-state values of all currents in the stator and rotor
have been found in terms of the voltages and the parameters of the
synchronous machine, the steady-state electromagnetic torque can be
evaluated. The torque equation for a smooth-air-gap machine was given
by Eq. 3-119 in terms of the instantaneous stator and rotor currents. The
currents have been found as steady-state complex amplitudes (phasors)
but these cannot be substituted into the torque equation because the
torque equation contains the products of pairs of currents and functions
of the mechanical position. The torque equation is nonlinear, hence the
instantaneous currents cannot be replaced by phasors to give the steadystate torque. The torque equation can be reduced to the desired form by
using Egs. 3-153 for the instantaneous rotor currents defined in terms of
l' and (Jr. The electromagnetic torque T~", was derived previously and
from Eq. 3-119 is

(3-171)

L'

E' sin n8
zr cos (0' - .pr - 'IT/2)

Simplifying using w r

= nWm
T~,.

sin (wrt - 0') sin n(wmt

yields

+nL'''VSI'
~',

=:

n''''

"II" f '.J',

+ &)l} (3-176)

'to

cos (n8

+ 8')

(3-177)

;,. . Equation 3-177 is the instantaneous steady-state torque, because it was

:''11: derived using instantaneous steady-state currents: The torque given by
Eq. 3-177 is a constant value for any fixed operating conditions. This
"' r

'1'.'4
..'.:'. 4

::"

:~:'
:~.!;
,;

'';:~'.1~'f;

224

ELECTROMECHANICAL ENERGY CONVERSION

THE GENERALIZED ROTATING MACHINE

shows that under steady-state conditions the instantancous electromagnetic torlJuc in a balanced two-phase synchronous machine is a
nontimc-varying quantity. The electromagnetic torque 7~" can be
expressed in a form which is more informative than Eq. 3-177 by
eliminating I' and IJ'. To do this write cos (no + D') in the torque expression as cos [(0' - ,V) +- (nil + ,V)) and expand to obtain

torque possible occurs at an electrical angle of no = 'TT/2. The torqueangle characteristic of a synchronous machi!1e with constant field voltage
and constant rotor terminal voltage is a sinusoidal function of nt.. This
torque-angle characteristic is plotted in Fig. 3-7. Whether the machine
operates as a motor or as a generator depends only on the sign of S;
generator operation occurs when (\ is positive, because T e.. is negative and
r/> is positive, and motor operation occurs when 0 is a negative angle,
because T... is positive and r/> is positive.

Tr .. =

nl.fTV'

-- {(l' cos (0' -

.'

<1/) cos (no + t/J')

- (l' sin (0' -

t/J')

sin (nS

IV)}

225

(3-1 ?8)
,~;

Substitution (If Eq. 3-174 for I' cos (fI' - <II), Eq. 3-175 for I' sin
and E' = (w'L;:V')/R' into Eq. 3-ln yields

CO' - t/J'),

Tep.

I/Mot"'----

,~~,

'.

r..

nE'
co'

[< V' -

+ t/J') _ E' sin no sin (no +t/J')]

E' cos no) cos (no

Z,

<'"

,,,>

Z'

(3-179)

Simplifying Eq. 3-179 using the identity that

X' = Z' sin ,V gives for the electromagnetic torque

-no

R' = Z' cos.p' and

- r = -=..!.[(E')2R' - E' V'R' cos nS + E' V' X' sin ns]

T ..
W m
(R')2 + (X')2

~.

.. """ _

E'V' .
W

X' sm

no
~

E'V'
-T.p.

Fig 3-1, Approximate torque-angle characteristic of a balanced two-phase synchronous machine.

The present discussion of the synchronous machine has been limited to

smooth-air-gap machines. For synchronous machines with saliency one
often finds that the simple equivalent circuit of Fig. 3-5 and the torque
equation of Eq. 3-181 are used. This is an approximation even for
balanced. steady-state operation. A more detailed discussion of the
synchronous machine, including saliency, will be deferred until the dq
variables are introduced in the next chapter.

(3-181)

The torque expression of Eq. 3-181 is particularly interesting in that it

establishes that the torque is proportional to the terminal voltage V'
divided by the synchronous reactance X', proportional to the generated
voltage E' divided by the speed w m ' and proportional to the sine of the
allgle no which is the displacement between V' and E' in electrical radians
as shown in the phasor diagram of Fig. 3-5. The angle nS has further
meaning in that it is the space displacement between stator phase a and
rotor phase a in electrical radians at time I = 0, i.e., <p = 0 at I = O.
The angle;) is called the torque angle of the synchronous machine, and.
for a synchronous machine to develop more electromagnetic torque with
constant excitation, the torque angle 0 must increase. The maximum

.,

wmX'

(3.180)

'J'i~,iH'

no

-,..
f.

'tit'

The torque equation in its general form, Eq. 3-180, is somewhat complex.
One approximation which is valid for most conventional synchronous
machines is that the rotor resistance R' is very small compared to the
synchronous reactance X' = w'L;'. Also, the generated voltage E' and
the terminal voltage are of the same order of magnitude; thus for
R' ~ X' the torque equation reduces to simply

3.6.3

J.

;,~y:,
)'.:,>,

;;::

Jii:.
;3:t

';' ~ . . tf:'

The Two-Phase. Smooth-R.otor. Synchronous-Induction

Machine

In the two previous sections the volt-ampere and torque relationships

of a conventional synchronous machine (d-c field excitation and mechanical speed in electrical radians per second equal to the frequency of the
armature excitation) have been studied. A more general case of a
machine in which two-phase excitations of different frequencies are applied
to both stator and rotor will now be considered. The VOlt-ampere
equations are once again linearized by assuming that the shaft is constrained by a constant speed source W m Then by virtue of Sec. 3.5.1 an

226

~~~
IJ

ELECTROMECHANICAL ENERGY
CONVERSION
t .

THE GENERALIZED ROTATING MACHINE

Phase b-Rotor
vb + nwmL:.'I'[cos n(wmt + 0) cos (w't - 8') + sin n(wmt + 8) sin (w't - 8')]
= (R' + L~p)i/, + w'L:.'r[sin n(wmt + 8) sin (w't - 8')
"
+ cos n(wmt + 0) cos (w't - 8') (3-189)

average torque will be produced when the stator frequency w' and the
rotor frequency w' are related such that w' = w' nwm. To determine
what the voltage excitation on stator and' rotor must be to produce an
average torque, the volt-ampere equations for a smooth rotor machine
are written with the voltage excitation undefined. The volt-ampere
equations of the generalized machine running at constant speed with

eP

and

Wm

.p

= wml

L~2 = L~2

,fJ.,.:,
'.'~,+
'k'

+ nwmL;[sin n(wml + 8)]i~ + nwmL;[cos n(wmt + 8)]iJ,

= R'i~ + L~pi; + L;[cos n(wmt + 8)]pi~ - L;[sin n(wml + o)]pib (3-182)

v~ - nwmL;[cos n(wmt + 8)]i~ + nwmL;[sin n(wmt + 8)Jir,

= R'it, + L~pit, + L;[sin n(wml + 8)]pi~ + L::'[cos n(wmt

.~

~~f~

~;y.

~:,Jl
;~

(3-183)

,
!~.. '

+ nwmL:.'[sin n(wmt +
= R'i~

8)Ji~ - nwmL:;[cos n(wmt

+ L~pi; + L;[cos n(wmt +

8)]pi;

8)]it,

+ L:.'[sin n(wmt + o)]pil,

vb + nwmL:.'[cos n(wmt + 0)Ji; + nwmL:.'[sin n(wmt + oW!'

= R'i; + L~pir, - L:.'[sin n(wmt + o)]pi; + L:.'[cos n(wmt + 8)]pit

(3-184)

i=
it

= I' cos (w'l - 8')

=

sin (w'" - 0')

(3-186)
(3-187)

Substitution of these stator currents into the two rotor volt-ampere

equations yields:
Hl

v:

1i

11:,::

These four volt-ampere equations of the generalized synchronous'induction machine are, in their present form, very difficult to interpret.
To facilitate the study of these equations, the stator currents are constrained to be a balanced set of steady-state, two-phase currents:
:,<l!liOn,p

(.}

"
:J'

(3-185)

Phase a-Rotor
~,"
+ nwmL:.'I'[sinn(wmt + 0) cos (w't - 8') - cosn(wmt + 0) sin (w't -,~']
= (R' + L~p)i; + u/L:.'I'[ -cos n(wmt + 0) sin (w'l - 8')
+ sin n(wmt + 8) cos (w'l - 8')] (3-188)

'.

Phase b-Rotor

(J$ -

n8]

(R'

+ L~p)i~

(3-191)

'1

';"".':.;" .

v~

(3-190)

The use of the trigonometric identities given by Eqs, 3-190 allows

simplification of the two rotor volt-ampere equations (Eqs. 3-188 and
3-189) to:
" 1"',

Phase a-Rotor
S
v~+ (W - nwm)L:.'r sin [(w' - nwm)t -

'~

""

+ 8)JpiJ,

sin a sin # = 1- cos (ex - #) - t cos (ex + f3)

cos a cos # = 1- cos (ex - (3) + t cos (a + (3)
sin ex cos f3 = t sin (ex - (3) + t sin (u + (3)

,~

lJ!j'"

= L::Z = 0

Thus the four volt-ampere equations for the generalized, smooth-air-gap

machine become

v;

These two equations can be greatly simplified by applying the following

three trigonometric identities for the products of sine and cosine functions:

+ /)

are given by Eqs. 3-139-3-142. The equations for a smooth-air-gap

machine can be obtained from these equations by setting

227

Vb - (w' - nwm)L';l' cos [(w' - nwm)t - 8' - no] = (R'

+ L~p)i;

(3-192)

These two volt-ampere equations are both of the form of an externally

applied exciting voltage due to a source (v~ or vi;) plus an apparent voltage
source in each rotor circuit, which is produced by a combination of the
stator current and rotation (the voltage of energy conversion equation,
Eq. 3-97a), in series with an R-L circuit consisting of the phase resistance
and self-inductance. Treating the induced sources due to the stator
current as apparent voltage sources yields the result that the steady-state
volt-ampere equation of rotor phase a is independent of the volt-ampere
equation of rotor phase b. If the rotor voltages v~ and vI, are now constrained to have the same frequency as the induced voltage due to stator
currents, i.e., a frequency w' - nw m , then i; and i; will have a frequency
of w - nw m This is exactly the frequency which rotor currents must
have for an average torque to be produced in the machine.
Define for the rotor excitation a balanced two-phase set of voltages with
a rotor frequency w' = w - nw m , and a phase angle y'.

,b',
S

v~

= V' cos (w't + y')

(3-193)

vb

v' sin (w't +

(3-194)

y')

where w' == W - nw m , The phase angle y' is necessary, because only

one quantity can be picked as a time reference, and for this machine it is
preferable to choose the stator voltage as the reference quantity. These

.,

228

ELECTROMECHANICAL ENERGY CONVERSION

&..

1\

two applied rotor voltages are chosen to have the same phase sequence as
the induced voltages due to stator currents, as well as the same phase
sequence as the stator currents (Eqs, 3-1!\6 and 3-187), This is easily
seen by noting that cos (a - 7T/2) = sin ex and sin (ex - 7T/2) = -cos a.
The rotor volt-ampere equations are now rewritten, substituting
w' = w' - nw", and using the excitation of Eqs. 3-193 and 3-194.

V'cos(w't

+ y') + w'L:;rsin(w'/-

Os - no) = (R'

V' sin (w't

+ y')

(f -

- w'L:;F cos (w'/ -

nS)

+ L:p)i~

(3-195)

= (R' + L:p)i'i,

(3-196)

,.

The steady-state instantaneous currents

Eq. 3-199: I "
i~ = l' cos (w'/ -

(I,

+ jw'L~)I~

(3-197)

-jV'ely' - w'L';I'e-J(8'+n4)

= (R' + jw'L:)I[,

(3-198)

Z'

= (R' + jw'L:) = Z'eN' =

'=

V(R')2

'\

-_I'

(3-203)

..

jw'L~

+
y'

e "
',~.

(a)

..

'::-..10',.

: ''',I'

E'
fj' -

""

\
\

....

<."

+ (w'L~)2 eN'

./,' = tan- 1 w'L'

R'
= l~ = jIb

+ y')

= I' sin (w't - 8'

,~

, + \"

-",I

_.\
..,,-'.....

~;tf~,

.1Yo

'f'

I'

(3-202)

(3-199)

V; = jV[, = + V'e ly'

E~ = jE = jE'eJ(-8S-n4) = w'L';lse1<-8'-n4+_/2l
b

using

8' + y')

/r-

where

V'
E'

i; and if, can now be written

Rr

It is apparent from Eqs. 3-197 and 3-198 that I; = jIb; therefore, one
volt-ampere equation which will be valid for either phase a or b on the
rotor can be written. The resulting volt-ampere equation is
:j

V' - E' = Z'l'

229

Equations 3-202 and 3-203 are the instantaneous steady-state values of the

The volt-ampere relationships of Eqs. 3-195 and 3-196 are vnlid only under
steady-state conditions; thus these instantaneous expressions can be
rewritten in terms of phasors using Eqs. 3-164 and 3-165 without any loss
of information. Recognizing that p = jw', Eqs. 3-195 and 3-196 become

V'eiY' - jw'L';I'e-}(8'+n4) = (R'

THE GENERALIZED ROTATING MACHINE,.

l' el (-6'+Y')

.....c 1

'-'---tl---

8' = angle by which I'lags V' (to be evaluated later)

The volt-ampere relationship for the rotor given by Eq, 3-199 is expressed
by an equivalent circuit and a phasor diagram in Fig, 3-8. From the
phasor diagram it is seen that 0' is the angle by which the current I'lags
the voltage V' and l' is the magnitude of the rotor current. This phasor
diagram can be used to solve for l' and 0' to obtain

, _ JlV')2'+-(;JL::r)~~WJ[::l'COS(OS
I (R'F + (w'L~)2
{}' -

71

tan-- I [

no

+ y' -

7Tj2)

(b)

Fig. 3-8. Equivalent circuit and phasor diagrlUTl for the rotor circuit of the balanced,
two-phase, synchronous-induction machine.

rotor currents, and the magnitude l' and phase f)' are given by Eqs. 3-200
and 3-201.
These rotor currents are for a constant speed of rotation w m , balanced
two-phase currents of frequency w" in the stator, and balanced two-phase
rotor voltages of frequency w'. This information can now be used to
investigate the stator VOlt-ampere equations and see what stator voltages

_
(3 2(0)

V' sin ~l - /L':" sin (ifl - (f - nil - y' + 7T / 2)]

.
3-201
V'cos,Jl - w'L~;IJcos(.p' - 0' - no - y' + 7Tll) (
)

".~
'I;"'W".

-'
.

230

~.
'~l:<

ELECTROMECHANICAL ENERGY CONVERSION

THE GENERALIZED ROTATING MACHINE

are required to maintain the assumed stator currents (Eqs. 3-186 and 3-187)
in the steady state. Substituting for i~ and i b from Eqs. 3-202 and 3-203
respectively into Eqs. 3-182 and 3-183, the vOlt-ampere equations for the
stator become

previously, the stator voltage can be chosen as the reference, i.e., with
zero phase angle. Thus the stator voltages v~ and Vb are defined as:
v~ =

vt

Phase a-Stator
v~

+ nwmL:;r[sin n(wmt +

= (R' + L~p)i~

~)

cos (w't - 0'

+ cos ~(wmt + ~) sin (w't - 0' + y')]

+ w'L:;I'[ -cos n(wmt + ~) sin (w't - 8' + y')
- sin n(wmt + ~) cos (w't - 0' + y')] (3-204)

\;;'
, ";

= (R' + L~p)ib

1"

+ y')
+ sin n(wmt + ~) sin (w't - ()' + y')]
+ w'L:;r[ -sin n(wmt + ~) sin (w't - 0' + y')
+ cos n(wmt + ~) cos (w't - 8' + ')I')] (3-205)
- 0'

~':

y'

V'sinw't - w'L:;rcos(w't - 8'

y'

V'e JO -

'If...
~

+ y' +

n~)

= (R' + L~p)i~

'~.

\.'(

Jl

,:l-

+ Y + n8) =

(R'

+ L~p)ib

'l ' ~

= (R'

+
+

L:.p)i~

(3-210)

L:.p)it,

(3-211)

(3-212)
(3-213)

(3-214)
\"'4\

V' = V;

'l

E'

~;,.

.I.
'I'

'I,

'j

l'

jvt

V'e JO

'ow,

;~'~

,""'~

"a',

= jE; = jE'e}(-o'+Y'+n8) = w"L::re j (-o'+Y'+nH1r/2)

E~

Z' = (R'

)',;

?;:

(R'

V' - E' = Z'I'

]"

Equations 3-206 and 3-207 show that the stator volt-ampere equation for
phase a is independent of that for phase b. Further, the rotor currents
react back on the stator circuit for both phases a and b to produce a
voltage of frequency w' whose amplitude is proportional to the magnitude
I' of the rotor current. This voltage has a frequency w" equal to the
frequency of the stator currents defined previously by Eqs. 3-186 and 3-187,
and the phase sequence of the voltages induced in phases a and b by rotor
currents has the same phase sequence as the stator currents. Consequently, when the stator is excited with a two-phase voltage source which
has a frequency w' and a phase sequence the same as that assumed for the,
currents, the balanced stator currents assumed in Eqs. 3-186 and 3-187
will be the true currents for steady-state operation. Furthermore, since
arbitrary phase angles have been maintained for all quantities defined

= (R' + jw'L:)I~
= (R' + jw'L:)lt

where

J'

(3-207)

+ n~)
+ nS)

For the assumed steady-state condition, it is evident that Eqs. 3-212 and
3-213 are independent, and recognizing that I; = jlt" one of these equations
describes both stator volt-ampere relationships. The single stator voltampere equation can be written as:

,~~

(3-206)

Phase b-Stator
vt - w'L:;r cos (W't - 8'

jw"L;:reJ(-o'-t-y'+n~)

-jV'eJo - w'L;:reJ(-o'-t-y'+nll)

'

'if') ..

Phase a-Stator
w'L:;r sin (W't - ()'

Using the three trigonometric identities of Eqs. 3-190 and recognizing

that w' = w' - nw m, the stator volt-ampere equations can be written in
the following simple form:

(3-209)

~~

v~

= V' sin w't

These two instantaneous equations are valid only for steady-state operation; therefore, they can be written in terms of phasors with no loss of
information. Recognizing that i~ and it, have a frequency w" and that
p = jw', Eqs. 3-210 and 3-211 become

',',

~)cos(w't

(3-208)

w'L:;r sin (w't - 8'

V' cos w't

'~{;,

+ nwmL:;r[-cosn(wmt +

V' cos w't

The stator volt-ampere relationships, Eqs. 3-206 and 3-207, can now
be written using v~ and v~ as defined by Eqs. 3-208 and 3-209:

+ y')

Phase b-Stator
v~

231

+ jw'L~) = v(R'F + (w'L:F eN' =

w'L"
tan-! --"

J~

Z'eN'

R'

= jl; = l'e-jo'

The stator VOlt-ampere relation of Eq. 3-214 can be expressed by the

equivalent circuit and the phasor diagram of Fig. 3-9. This phasor
diagram can be used to write the current I' and its phase ft' as

(V'F

, 1'..
I. ,',

8' - ."

+ (w'L;:rF - 2V5w"L::rcos(nS + /-(R')2

tan- 1 [

2~)

+ .,,/2)
(3-215)

V' sin .p' - w"L"rsin (.p' - nS - '

"
y
V' cos.p' - w"L;: r cos (.p" - n~ - y'

'\

(I'

(w"L")2
,..

'))
~'"_.

"

+ (I' + (I' -

.,,/2)]
3-216
.,,/2) (
)

'::,,'
I

ELECTROMECHANICAL ENERGY CONVERSION

232

ii,

ft')

= l' sin (ult - ff)

(3-187)

J"

~f:.

E'

'
I

(3-219)

B"

';'

',~, ,

+ " .....

"-1
0" -11'
I
I

("'t/v

(R' + jwSL~W

(3-221)

J~1

V' - (w'L';e)(-nH,,/2JS = (R' + jw'L~)r

(3-222)

'I.

An interesting equivalent circuit can be derived for Eqs. 3-221 and 3-222
by defining the slip s which is the ratio of rotor to stator frequencies

I
I
I

I'

.,'3

I
...J

...,;~

\ IJ"

. 1~1

1~ .

!.'-

:.(
~ ~i

-I"

,~

,'j""

!i
~

Fig. 3-9. Equivalent circuit and rhasor diagram for the stator of the balanced.
two-phase synchronous-induction machine.

"~

!~

w'

s =WI
- ...

I~

w' L,~U'e-jOS)e j (-nH,,/2)

(3-217)

WS

(3-223)

= rein'

(3-224)

Stalor

V' -

t .
l~

w - nWm

Also, it is desirable to eliminate the phase shifts ei(nH,,/2) and e j (-nH,,/2)

in the two equations. This is .not possible; however, a worthwhile
modification results when the phase is shifted so that it is absorbed in
the rotor voltage.
To shift the angle no to the rotor voltage, define a new rotor current

f':..'.;I
....

but these currents are exactly of the form of the stator currents assumed
initially in Eqs. 3-186 and 3-187. Therefore, a consistent steady-state
solution for all rotor and stator quantities has been deriwd. The
equations describing the rotor and stator circuits given by Eqs. 3-199 and
3-214 and their respective equivalent circuits in Figs. 3-8 and 3-9 can be '.
combined into one single equivalent circuit.
"I
To accomplish this, Eq. 3-199 is used to obtain

(w'L~e)(nH,,/2I'=

Using Eq. 3-224 plus the slip s defined by Eq. 3-223, the volt-ampere
equations for the stator and rotor become

(b)
:i

V' Rolor

'1S-'
~.

(3-220)

Stator

/.

. l;v"
I

= W SL~(l'e)(-"+Y'ej(n~+,,/2)

Combining Eqs. 3-214 and 3-220 for the stator and Eqs. 3-199 and 3-218
for the rotor gives:

" '~.' :
"''"'''''''.';j" ..\..

(a)

E'

'.. ' (3-218)

Similarly, from Eq. 3-214,

..; ,.:,

-\

'

'I(

L 3,.' l'ei(-;'~+"/2)

E' = w SL';l'e)(nHfT/2)

<1k.i~\ :'0.:.\

W'

which, from Eq. 3-199 for l' becomes

v"

..

E'IIO:

jw"L~

R'

133

but, by Eq.~214 for /', this becomes

Using Eqs. 3-214, 3-215, and 3-216, the stator currents i~ and i~ written
as steady-state instantaneous quantities are

i; = /s cos (w"l

THE GENERALIZED ROTATING MACHINE

' ','':1'"

(3-186)

~ "'"

i~

i~

$f'\

.,

Rotor

jwsL::l~ =

(R' + jw'L~W

(3-225)

.,;'"

V'~jn~ _ jW3~';P = (~' + jW3L:)I~

(3-226)

where the equality j = ei ,,/2 is used. Equations 3-225 and 3-226 are
analogous to the equations of a pair of mutually coupled circuits or of a
transformer which has a secondary resistance of R'js and a secondary
excitation of V'ein'/s. Equivalent circuits for these equations are drawn

' '.'.;
..

...

...

.....:.

..

.,

.------0<

ELECTROMECHANICAL ENERGY CONVERSION

2304

w r,

in Fig. 3-10. One interesting characteristic of these equivalent circuits is

r
that the rotor circuit which physically has an excitation frequency w is
converted tq a.circuit which has an apparent frequency of w', In so doing,
however, the rotor resistances become R'ls. This change in rotor resistance is a manifestation of energy conversion. The equivalent circuits of
Fig. 3-10 are passive circuits but the machine certainly is not; thus a

R"

jw'(L~-L:;:)

Rr

jw'(L~-L::)

---

~~

..

'~".

:t

in8

v' e
--s--

y'

will have a torque characteristic dependent upon S.

frequency w' and
In addition, from the equivalent circuit of Fig. 3-10 it is evident that for
fixed stator and rotor excitation the only way the currents can change with
load, which they must do, is to have S change. One should also expect
to find that torque for a synchronous-induction machine is a function of
the angle S. To show that this is indeed the case, the torque will now be
determined.
The instantaneous steady-state torque can be evaluated, since the
currents have been found in terms of the various excitations. Using Eq.
3-119 for T,,, and Eqs. 3-186, 3-187,3-202, and 3-203 for i~, it, i;. and i b
respectively, the electromagnetic torque T e" becomes

i'~

IT

/'

T.,. = nL:;{IrI'[cos (wrt - (Jr

- sin (wrt - (Jr

(a)

R'
s

:......'->

-/i

L;

y'

.'\

. i

-s-

f:

...J~'
t

'14',.1:'"
,
';'I."

'5

r"', .f ~

Frequency = w'

{l ).r

... \

. "lin

(b)

Fig. 3-10. Equivalent circuits for the balanced, two

phase synchronous-induction machine. (a) Equivalent
circuit as two coupled circuits or as a transformer.
(b) Equivalent circuit as two coupled coils.

.,

..:!;'....

,."

l'J..~
i

source or sink of energy must appear to account for the energy converted
by the machine. This will be treated in more detail later when the
conventional induction machine is studied. Another interesting characteristic of these equivalent circuits is the occurrence of the angle S as eJn'.
The angle S is the physical angle between the a phase on the rotor and the
a phase on the stator at time t = 0 and in the study of the conventional
synchronous machine in Sec. 3.6.2, S varied with load. It is also reason
able to expect that the synchronous-induction machine, which is a constant
speed machine for constant stator and rotor voltage excitation with fixed

I*
'~.

+ yr) sin (w't

+ yr) cos (w't

0)

o)} (3-227)

- IJ')

- (J')] sin n(wmt

Tel' = - nL:;l'F sin (nS + ()' + yr - ff)

;.:

Vre inl

yr) sin (w't - IJ')

Simplification by the use of the trigonometric identities of Eqs. 3-190 and

the relationship wr = w' - nw m, the electromagnetic torque becomes

~
Xt

,~

\.

cos (wrt - (Jr

yr) cos (w't - (J') cos n(wmt

- l'P[sin (wrt - ff

RS

235

THE GENERALIZED ROTATING MACHINE

':rr

'It! .
~:'

The electromagnetic torque for the two-phase synchronous-induction

machine, Eq. 3-228, is similar in form to that derived for a conventional
two-phase synchronous machine in Eq. 3-165. It is a function of the
stator and rotor currents, their time phases (J' and (Jr, the time phase yr
of the rotor voltage, and the steady-state rotor displacement nS in electrical
space radians. In this case where both the stator current and rotor
current phase angles are dependent upon their respective voltages, it is
difficult to derive a simple expression for the torque as a function of the
displacement {) between the stator and rotor at time t = O. However,
expressions can be obtained for the electromagnetic torque which are
convenient to use in conjunction with the equivalent circuit of Fig. 3-10 :,
for calculation of the electrical and mechanical performance characteristics of the synchronous-induction machine. One simple form for the
torque expression can be obtained by using the phasor diagram in Fig.
' .,' ,.
3-9 to show that

E' cos (()'

~~..

..:1'
'

*'

,\,
}'

;.~;

(3-228)

Substituting

E'

nS

+ ;I' -

ff - 'Tr/2) = V' cos (()' - 'Tr)

w'L:;r and

w'V':l'sin(J'

nS

+ FR'

(3-229)

simplifying, this equation becomes

+ yr

- (Jr) = -V'cosfJ'

PR'

(3-230)

'

..

236

ELECTROMECHANICAL ENERGY CONVERSION

THE GENERALIZED ROTATING MACHINE

Equation 3-230, substituted into the electromagnetic torque relationship

(Eq. 3-228) along with the identity nwm/w s = (I - s), yields
.

SUbstituting for
"1 1 '

T.,; = (I w- s) [vsI' cos 8' - (lsFR']

(3-231)

T,,.

237

from Eq. 3-231; there results

[V'l' cos 8' _(lS)2 R S)

,

= i[
S

V'1' cos 8'

(1')2 R']

(3-237)

Equation 3-231 for the torque is most interesting in that it shoUl~ that the
mechanical power T."w m is given by the stator air-gap power, multiplied
by (I - s) which is the ratio of mechanical speed in electrical radians
per second to the electrical speed of rotation of the stator magnetic field
with respect to the stator. The air-gap power is the power into the
minus the copper loss on the stator, (I')2R'. Thus, the
stator, V'I' cos
electromagnetic torque can be determined simply from a knowledge of the
stator input power and the stator copper loss.
A second expression for the electromagnetic torque in terms of rotor
quantities can also be calculated. Using the phasor diagram of Fig. 3-8
gives

Equation 3-237 proves that Eqs. 3-231 and 3-234 are equal and, furthermore, it shows that the conservation-of-power relationship of Eq. 3-236
can be used to calculate the electromagnetic torque as

T,,. = _1[V-Pcos
OS
.
W
m

8"

E' cos ( - OS - n8 - y' - n/2

+ 8')

= V' cos (8' - n)

+ l'R'

w'L'; I' sin (8 S

+ no + y' - 8') =

V' cos 8' - 1'R'

,,-~'ti

(3-233)

;<

Substituting Eq. 3-233 into the torque expression of Eq. 3-228 yields

T~,. = (I -

W",s

..

"

s) [ _ V'I' cos 8'

+ (1')2 R'J

,-I',

Equation 3-234 for the electromagnetic torque shows that the torque can
bl: determined from the rotor air-gap power, defined as the power consumed in rotor copper losses minus the power supplied to the rotor (i.e", .
the power flowing out of the rotor plus the rotor copper loss). The rotor
air-gap power is used to find the mechanical torque by multiplying by
(I - s)/s which is the ratio of the mechanical speed in electrical radians
per second to the electrical speed of the rotor magnetic field with respect
to the rotor. Now both Eq. 3-234, which gives the torque in terms of the
rotor air-gap power, and Eq. 3-231, which gives the torque in terms of
the stator air-gap power, must be equal since they give the same torque. "
and were derived from Eq. 3-228. To show that this is indeed the case"
a steady-state power balance for the synchronous-induction machine is
written as:
Total power supplied to
mechanical power
=.
stator and rotor
delivered
V' /' cos 0' + V' r cos 8' = T,,,w m

(Is)2 R'

osses

V')2 R'

stator power,:"

(3-235)

(3-236)

1:;1

>-

'.;

-'

(l'FR')

ul
T,,. n
= eV-/, cos~,-

(IS)2R S ]

mechanical power = T,,.wm ~ (1 - s)[Vs]' cos 8' - (I,)2R sJ

rotor power=

(3-234)

v'rcos 8' - (I')2R s

(3-238)

The three torque expressions, Eqs. 3-231, 3-234, and 3-238, are
particularly interesting in that they show that the torque of a synchronousinduction machine is like the torque of a fluid coupling and that this
torque times an apparent mechanical speed will give the air"gap power
as seen by a system moving at that speed. l,Jsing the three torque
equations, it can be shown that

(3-232)

Substituting E' = w'L:;I' and simplifying, this equation reduces to

w'

T,,. n

= s[Vsl' cos 8' - (IS)2R sL.

(3-239)

(3.240)

<

(3-241)

Thus, the synchronous-induction machine can be considered as a f1uidtype coupling which develops a fluid torque or air-gap torque that is a
constant value for any system moving in the fluid field. The concept of
a fluid coupling is conceptually not too satisfactory with the synchronousinduction machine, which is constrained to operate at a constant speed.
These concepts will, however, be useful in the next section Where the
nonsynchronous types of induction machines will be studied.
One further study of the synchronous-induction machine which is
interesting is its torque-speed or torque-angle characteristic. The major
diffiCUlty with such a study is that the synchronous-induction machine is
a constant speed machine, and the speed torque curve is a straight line at
a given speed. The torque-angle characteristic is derivable from the
torque expressions of Eqs. 3-228, 3-231, 3-234, or 3-238 and the equivalent
circuit given in Fig, 3-10. The resulting characteristic is, however,
dependent upon the rotor and stator voltages; thus the torque-angle
characteristic is meaningful only for a given set of operating conditions.
The synchronous-induction machine, sometimes called the doubly
fed induction machine when it is operating at a constant speed
W m = (w' - w')/n, has been treated in some detail.
A more general

238

.,,';

i'

ELECTROMECHANICAL ENERGY CONVERSION

study of the doubly fed induction machine operating at a non synchronous

speed in terms of the coil variables is too complicated for study at this
time. However, there is one special configuration, in which the rotor is
not excited from external sources, that is extremely important. The
configuration of the synchronous-induction machine, in which the rotor
is short-circuited or connected to a passive load, is conventionally called
the induction machine. The terminology .. squirrel-cage induction
motor" is used for such machines when the rotor circuits are shortcircuited in,ternally with no external rotor connections, and" wound-rotor
induction machine" is used to designate those configurations when
connections to the rotor circuits are brought out external to the machine.
The squirrel-cage induction motor is the configuration which has, by far,
the largest commercial application of any standard motor.

THE GENERALIZED ROTATING MACHINE

:~':;1,
, ,"

,", t,"~'>;
1 ,'.

.;r\"

{',:

and

y'

0-242)

By setting y' = 0 the phase of the rotor current (J' is thus measured
relative to VS, the stator excitation. Using these equations, the stator~;f)
volt-ampere relationships are found from Eqs. 3-210 and 3-211 and the
'\\ rotor volt-ampere relationships are found from Eqs. 3-195 and 3-196.

"

Stator

,r~

V' cos w't

M'L::1' sin (w't

V'sin w't - w'L::1' cos (w't

+ no
+ no

- 8') = (R'
-

(J')

= (R'

+ L~p)i;
+ L~p)ib

(3-243)
(3-244),Je

Rotor
w'L"l' sin (w't ~

(J' -

no) = (R'

-w'U;l' cos (w't - (J' - no) = (R'

+ L'~ p)i'a
+ L:p)ib

't":~

(3-245)
(3-246)

It is clear from the two rotor volt-ampere relationships. Eqs. 3-245 and
3-246, that the rotor currents are produced by the stator current I'
inducing voltages in the rotor circuits. The frequency of these induced
voltages is w' = w' - nw m , i.e., the stator frequency minus the mechanical
speed in electrical radians per second. The currents in the rotor thus have JU
, a frequency w' which always differs from (J! by nW m regardless of M' or .
nwm Since this is the necessary condition for the development of an ,
average torque as derived in Sec. 3.5.1, the induction machine is capable"
of developing an average torque at any speed.
r~,t
The volt-ampere relationships for the induction machine given by Eqs.I'J
3-243 through 3-246 are valid only for steady-state operation. Conse- tJ
" quently, they can be written in terms of phasors with no loss of information. ,'*
Also, it will be shown, as it was for the doubly excited induction machine, .,
that the phase a and h currents are so related that one volt-ampere
equation for the stator and one for the rotor will completely describe the
machine. These two equations can be derived from Eqs. 3-243 through

;.. } 'I',.'

s;

V' = 0

'.;,'

The conventional, or singly excited, induction machine is an electromechanical energy converter which is particularly interesting and which can
be studied using the generalized machine constrained for constant-speed
operation. The use of a constant speed constraint to study the induction
machine seems. at first thought, to be overly restrictive, because the
singly-excited induction machine is not a constant-speed machine, whereas
the synchronous machines and synchronous-induction machine treated in
the previous examples were constant speed machines. However, for
most practical induction machines, the mechanical speed of response (i.e.,
its rate of acceleration) is much slower than the electrical speed of response.
Thus the volt-ampere equations with the speed constrained to be constant
will very closely describe the electrical performance of most induction
machines even under conditions of dynamiC operation.,;
, . Since a constant-speed constraint is satisfactory for treating the induction
machine, the VOlt-ampere equations can be obtained as a special case of
the two-phase synchronous-induction machine treated in the previous
sections. The essential character of the induction machine is that the
rotor is completely passive; i.e., no external sources are connected to the
rotor circuit. The most common induction machine is the squirrel-cage
induction machine which has its rotor circuits short-circuited. There are
also wound-rotor machines in which external impedances are connected
in the rotor circuits. The most general treatment consists of considering
a wound-rotor machine with v~ = - i;zr and vb = - i;'Z'. However, to
minimize the changes in the equations already derived, it is preferable to
set
= 0
and
vI; = 0
#'

v;

!'

3.6.4 The Two-Phase Induction Machine

'"

which exactly describes the squirrel-cage induction machine. To include

in this analysis the wound-rotor induction machine which has balanced
external impedances in the rotor circuits, the rotor circuit constants R'
and
can be changed. Thus, a perfectly general approach which is
valid for both squirrel-cage and wound-rotor machines operating under
balanced conditions is obtained; and. furthermore, all quantities defined
for the synchronous-induction machine in the previous section remain
unchanged. The equations which describe the induction machine can
be obtained from Sec. 3.6.3 by recognizing that v~ and v~ are zero.
Setting the rotor voltages equal to zero means setting

L:

, i~

t'"

239

J' {"

'("\J -

",'t~
"".."
"

,i

..

240

ELECTROMECHANICAL ENERGY CONVERSION

THE GENERALIZED ROTATING MACHINE

3-246, but they will be merely Eys. 3-199 and 3-214 which were derived
previously. modifkd by the conditions of Eqs. 3-242, thus

The equivalent circuit for these two equations is drawn in Fig. 3-11 and
the phasor diagram is constructed in Fig. 3-12. The balanced, singly
excited induction machine can be represented by a transformer or coupled- .
circuit equivalent circuit in which the rotor is short-circuited through a
resistance R'(l - s)js which is the apparent power sink which accounts

VS _ ES
where

= V;; = jV;' =
E = E~ = jE~ =
V'

(3-247)

V'ejO
E'rJ(n~-O'+.r/2)

= wSL:/'ej(n~-o'+.,/2)

+ jw'L: = V(RsF + (wsL:)iei"" = ZSeN'

Z'

./.S
'f'

= tan- 1 w'L
__
Rs

= I'Zs

24'

.\

jw'(L: -L~')

R'

jW'(L~-L::)

R'

I'

/' =

I~ = j/~

(J' =

angle by which IS lags VS

PriBS

and
.\ I

= /,Z'

-E'

;f~

(a)

(3-248)

where

E' =

= jEi,

E~

Z' = (R'
./.' =
'f'

+ jw' L~)

tan - I

/' = I~

= E' e i (-O'-nH,,/2)
=

V (R')2 +

= w'L: J'ei(-O'-n~+,,/2)

R'

(w' L~)2eN' = Z'eN'

w'L'

u;-~

t._,

-_I'

R'

= jib

'.ll'.'P':'

.;.(

= J'e-jO'

1""

8' = angle by whIch I' lags V' .

~,

I;

w'
= w'

w' - nw",

Yi

(3-224)

l'ein~

+--

I'

'_r, ' ;

I '6

L'1',

l 'I'

R'(l - s)

-~s-

(b)\','

FIg. 3-11. Equivalent' circuits for the balanced, singly

,excited, two-phase induction machine. (a) Inductively
coupled circuit. (b) Coupled coils.

Hn,
~;

(3223YH

--w-"-"-'

"tIJ-(

R'

----+-

Frequency = w

~.i

The twe volt-ampere relationships, Eqs. 3-247 and 3-248, can be rewritten
in such a way that an equivalent circuit diagram of two coupled circuits
can be drawn. Equations 3-223 and 3-224 which are

':'

~
'At

:l-I.:

.~

"1

At'<:

-J"R'

.,...... v

.'

will be used for this purpose. Then, the volt-ampere equations of the
singly excited induction machine become
V' - jwsU:/~ = (R'

t'

I'

+ jwSL~W

(3-249)

. sL-"/s
. SL')/'
-.Jw
.. -_ (R' + .JW
I'
a
where

JlS

V'riO

r = l'e-iO '
I; = /' e-j(O'-nM
OS and 8' are measured relative to JI'

(3-250)

I{
eft

;;;~r

)n.

.j

~~.4~

Fig.3-I2. Phasor diagram for the balanced, two-phase, singly excited induction motor.

Cl..c\..II'..UI"IC\..HANILAL t:Nt:K<-3'Y

4"tL

CONVERSION

THE GENERALIZED ROTATING MACHINE

for the energy converted between the electrical and mechanical systems.
To verify this fact, the electromagnetic torque will be found. The
torque could. be derived from Eq. 3-119 and the stator and rotor currents;
however, it is unnecessary to rederive the torque since Eq. 3-228, derived
for the synchronous-induction machine, applies by merely setting y' = O.
Thus the ~lectromagnetic torque for the singly excited induction machine
is

'

T,,, = -nL::I'f'sin (no

(J' -

fY)

"I."

'.j

i_,T

(3-251)

:~

The torque equation can be put in a more understandable and useful form
by using the relationship that (l - s) = 1 - w'/ws = nwm/w'

T." = (I - s) l'[w'U;J' sin (0' - 0"' - n8)]

(3-252)

Wm

R'

but from Eq. 3-224

power converted

(3-257)

(3-258)
\j.

I', so

'.\

r -R'
= "
w'L"I' sin (0'
s

..

- n8 _

rotor loss (l')2R' = s(air-gap power)

= s[V'1' cos 6' - (IsFR']

------.. .....

IV\

V)

(3-253)
stator loss

Substitution of Eq. 3-253 into Eq. 3-252 for the electromagnetic torque
T.,. yields
(1')2 (I - s) R'
(3-254)
=W
S

T.,.

(3-255)

..it

~'j

s) [V'I" cos 8' - (l')2R'J

i:

(3-256)
1

(3-260)

(3-261)

which is the required power balance, Also, using Eq. 3-258 for the
power converted and equating this to the mechanical power T."w m , the
torque expression of Eq. 3-256 is obtained.
Several expressions have now been derived for the electromagnetic
torque. Equation 3-254 gives the torque from the power dissipated in
the resistance of energy conversion R'(1 - .1')/.1', and Eq. 3-246 or 3-258
gives the torque in terms of the air-gap power defined as the power into
the stator minus the stator copper loss. These relationships can be used
to treat the balanced, two-phase, singly excited induction machine for
steady-state operalion in terms of the equivalent circuit of Fig. 3-11,
obtaining the electromagnetic torque from the conservation of power as
represented by either Eq. 3-254 or 3-256. Thus. the study of the balanced,
two-phase induction motor has been reduced to a simple circuit prohlem.
One additional comment is in order about the power expression and the
equivalent circuit of Fig. 3-11. All variables
1', I' have been defined
as peak values; thus power per phase is } V'I' cos 0" 1(1')2R', etc.. hut
since this is a two-phase machine, the total power is twice the power in

V"

m
.'lo(tm;!,1:'

h'l

(3-259)

Substitution of Eq. 3-255 into Eq. 3-252 for the electromagnetic torque
T. IJ leads to

= (1 W-

power into stator = V'/' cos 6'

V"cos (J' - FR' = w'L::1' cos W - n8 - 1T/2 - 0')

.'
= w'L::r sin (0' - no - 6')

= (lJF R'

'"So.

Subst!:uting Eqs. 3-258, 3-259, and 3-260 into 3-257 yields

Equation 3-254 for the torque confirms the interpretation given previously
that the term [(1 - s)/s]R' is an apparent sink of energy to account for
energy conversion. From the equivalent circuit of Fig. 3-11 it is evident
that (1')2 [(1 - s)/s]R' is the power dissipated in the fictitious resistance
of energy conversion. Since the mechanical power is T,... w m , the torque
expression of Eq. 3-254 can be obtained by conservation of power directly
from the equivalent circuit for the induction machine shown in Fig. 3-11.
Another interesting expression for the electromagnetic torque can be
derived from the phasor diagram of Fig. 3-12 by noting that

T.IJ

rotor loss

power converted = (I - s)(air-gap power)

= (l - s)[V'1' cos 6' - (I')2R J ]

"

I~ =

From this discussion it is evident that

w'Lsr l' cos (0' - n8 - 0' - 1T/2)

Equation 3-256 shows that the converted power T,,,w m is the power into
the stator, V'I" cos 6', minus the power loss in the stator, (/'FR', multiplied
by (1 - .1'), the relative difference in frequency between stator and rotor;
(1 - .1') = (Ill" - (,!)/w' is also the ratio of the mechanical speed in
electrical radians per second to the speed of stator magnetic field around
the air gap relative to the stator. An air-gap power defined as
V'l' cos 8' - (/')2R', which is the input power minus stator losses, will
divide in such a way that the mechanical power is the air-gap power
multiplied by n/('m/w" = (I - .1'), and the power transferred to the rotor
is the air-gap power multiplied by w'/w" = s. These relationships must
satisfy conservation of power, i.e.,
power into stator = stator loss

Next, the phasor diagram of Fig. 3-12 yields

l~

243

"'~

J1

2+4

ELECTROMECHANICAL ENERGY CONVERSION

..

'.

Next the torque at the value of slip given by Eq. 3-266 is found by
substituting Eq. 3-266 into Eq. 3-264, yielding
~l

(T.,,)l1\U = V[(L~R")2

+ (w')2(L~L~

(n/2)LS;2 V2
_ L:,'2)2][(R')2 +

(w'L~j2f+ -isw'LS;z
(3-267)

From Eqs. 3-266 and 3-267 it is evident that the slip at which maximum
torque occurs is linear with rotor resistance R'. Since these two variables

7',,,

sr

I; =

245

THE GENERALIZED ROTATING MACHINE

one phase, yielding the n:sult given previously, Later, when machines
which have three or more phases are studied, the equivalent circuit of Fig.
3-11 will again be obtained, but for an m phase machine the total power
will be given by In times the phase power or ml2 times the power obtained
from the peak quantities defined in Fig. 3-11.
One further characteristic of the induction machine which is important
is its speed-torque curve. To evaluate the speed-torque relationship
Eq. 3-254 is used and I' is evaluated from the equivalent circuit of Fig.
3-11 or from the two volt-ampere relationships given hy Eqs, 3-249 and
3-250. Solution of these equations for J; yields

jVSw"L
"+ .
.lw ,s
-

R rsR'

,(L~R' + l'R )

(w s )2(L'L' _ U'2)]
......
...

(3-262)

-JJ~

~'

"

from which the magnitude of I; is

i
1

I;

=r

J[R'~R"~(~')2(L;:~:

f!

'"1

';:Ji

VSwsL sr
_ L:,'2)

f" + (wS)2C~;,,~R'r-; L~,

RS

(T.,,) mo.

r
(3-263)

Substituting
torque gives

{jtYl~ ~"~' flO

'"

I- W '

from Eq. 3-263 into Eq. 3-254 for the electromagnetic

nw L;;2( R'Is)( V')2

[R r R S
] 2
_ _ (w s )2(L'L' - L<r 2) + (w s
s
"""

/LW m

j"

'lll

T~ ...
;i,

,,;

~'

L' R'
~
(3-264)
)2(-"'+ L' W}::J?r "". ,',.,
s"

Equation 3-264 shows that the electromagnetic torque varies as the

square of the stator voltage. The effects of the parameter values on the
torque is also important. To illustrate the significance of the various
parameters, the starting torque, maximum torque, and speed at which the
maximum torque occurs will be investigated. Consider the speed, or
equivalently the slip s, at which maximum torque occurs. To find this
the partial derivative of Eq. 3-264 with respect to slip s is equated to zero
and the result is solved for .I max ' thus:
,;;,"

aT,,,,
as

(3-265)

'-"m.x

yields
Small =

+ R'J
-

(Rs)2 + (w L~)2
(w'L:R')2 + (w')'(L:L~ - L;:2)2

~~,;)uarrr

Fig. 3-13. Torque-speed curve of a balanced. two-phase induction motor.

-.,-'"
J.!

occur in Eq. 3-264 only as a ratio, the maximum torque is independent

of rotor resistance. Therefore, although the shape of the induction
machine torque-speed curve can be controlled by rotor resistance, its
maximum torque cannot.
A typical torque-speed curve for an induction motor is shown in Fig.
3-13. The region of speeds over which motor operation is achievable is
evident from the figure. In the region 0 < s <; I

T." > 0

and

> 0

(3-266)

Therefore electrical energy is being converted to mechanical energy.

Other values of slip yield generator and braking regions where the torque

246

opposcs the rotation and mechanical energy is converted to electrical

energy. The essential difference between these two regions is that in the
generator region the converted energy is supplied to the stator electrical
source, whereas in the braking region the converted energy is dissipated
in the rotor circuits as ilR losses.
The starting torque, which is given by s = I (or W m = 0) in Fig. 3-13,
can have any value bctween zero and thc maximum torque given by Eq.
3-267. The starting torque can be varied between these limits by changing
the rotor resistance R'; however, an increase in R' to raise the starting
torque will lower the slope of the torque-speed curve between s = 0 and
s = Sm... which is the region of normal operation. Such a reduction in
slope will make the motor speed more sensitive to load torque changes.
Two-phase induction motors with high rotor resistances are called
servomotors and are widely used in position control systems. The rotor
i2R losses associated with these motors limit their use to instrument
systems (25 watts or less) although servomotors in the range of -t horsepower are built by incorporating auxiliary cooling. The principal
characteristic of two-phase servomotors is that the rotor resistance R' is
made sufficiently high so that the maximum torque occurs at a slip of
around s = 1.5. This insures that the torque-speed curve will have a
negative slope, and therefore a stable operating region, around zero
speed (s = 1).
The principal characteristics of the balanced, two-phase, singly excited
induction machine have been determined including its equivalent circuit,
speed-torque characteristics, etc. All the characteristics were derived
for steady-state conditions and, strictly speaking, are only valid for
steady-state problems. It is often allowable to solve dynamic problems
using the equivalent circuits and speed-torque characteristics developed.
The allowable limits of such an extension, however, cannot be determined
by'the analytical techniques that have been used up to now. It is essential,
then, that other techniques be considered for finding the performance
characteristics of the induction machine as well as of all the machines
studied previously so that unbalanced problems as well as dynamic
problems can be considered.
..1:.
u.
iU,
(It;, I J'l;y' .~ ~I hv

"~'.l'

nonlinear differential equations. By considering various constraints for

the currents and constant speed con,traints ror the rotor. the equations
of motion we~e t'implified. From these lincarizcd equations, the necessary
conditions for the production of an average torque were derived.
Similarly, the steady-state, volt-ampere characteristics and torque-angle
or torque-speed characteristics were determined for a synchronous
machine, a doubly excited synchronous-induction machine, and a singly
excited induction machine.
The general equations of motion derived in this chapter are valid ror any
conditions or operation, but the examples treated were mostly limited to '
balanced, steady-state operation. It is essential that the range of
application be extended and techniques developed by which unbalanced
operation as well as dynamic operation can be considered. Changes from
the coil variables used in this chapter to new variables will greatly broaden
the range of operating conditions of machines which can be handled
effectively. The next chapter will, therefore, introduce several sets of
variables which arc userul for the study of machines under unbalan-.:ed
excitations and under dynamic operating conditions.

PROBLEMS--'~''': -

'1"
;~if:':

Resume

~'

'-~.-

3-1. Figure 3P-I shows the cross section of a machine with two identical
rotor wi;-,,~ings in space quadrature. The stator has four symmetrical projections or pole faces. </> is the angle between the plane of rotor coil ad and a
fixed point on the stator.
It is experimentally found that the self-inductance of coil QQ' may be
approximated by the first three terms of a Fourier series:
L~, = L~o + L~I cos 4</> + L~3 cos 12</>
"
"t
Mutual effects between the two rotor coils may be neglected.
, ,;~.

,~,~'l'

'ff>Y

"wo4;.;

1flJr

1c,
>,:til

J;i)~.;

i'l,

VJ

"'1): .

3.7

2...7

THE GENERALIZED ROTATING MACHINE

ELECTROMECHANICAL ENERGY CONVERSION

'*t11

,r

, ~~;!i'il

In this chapter a physical model of a rotating electromechanical energy

converter called the "generalized machine" has been postulated. The
parameters necessary to describe the generalized machine were derived
and, using the techniques of Chap. 1, the equations of motion were
derived. These equations of motion proved to be five simultaneous

Flg,3P-I.
t,!

J:,

.T~

148

33. Two sinusoidally distributed currcnt shcets (II = I) arc placed one on
the stator and the othcr on the rotor of an idealized rotating elc(;tric milchine
.
with smooth air gar. The pUl'4lmcters of the machine arc:

(a) Express the self-inductance of coil bb' in terms of the known constants
L~o. L~lo L~l' and the angle
(hl Derive an expression for torque on the rotor in the + 4> direction in terms
of the constants of part (a) and the two coil currents i~ and it

cp.

Can the machine develop an average torque and continue to run with
S
balanco.:d two-phase excitation; i.e. with i~ = 1 sin wSr and i b = 1 cos W ,?
(d) Can the machine develop an average torque with single-phase excitation;
i.e., with i;' = I sin W S , and i~ = 1 sin w'r?
(e) I I' the machine develops an average torque in either part (c) or part (d),
at what speeds can it run? Let 4, = kw'r + 13 k where k is a number to be
determined and Sk is constant corresponding to each value of k. (Hint: Let
4> = kw s , + ti k when: k is a number to be determined and Sk is a constant corre
sponding to each value of k.)
3.2. Figure 3P-2 shows a synchronous motor operating on reluctance
torque. The rotor has k teeth. (The sketch is drawn for k = 12.) The
stl\tor has a single winding shown in cross section by the dots and crosses.

stator self-inductance L' - 1.0 henry

rotor selfinductanccJ L' = 0.3 henry
L" = 0.5 cos </> henry
. mutual inductance

(c)

_,I

"

where cp is the relative angle between the current sheets. Stator lind rotor
resistances are considered to be negligible.
'
The stator is excited by a sinusoidal current is = I cos wt amp while the
rotor is short-circuited.
(a) With the machine at standstill. compute the instantaneous torque developed
as a function of 4>.
(h) If the rotor is allowed to rotate, at what angle will it come to rest?

,I

249

THE GENERALIZED ROTATING MACHINE

ELECTROMECHANICAL ENERGY CONVERSION

3-4. The machine depicted schematically in Fig. 3P-4 has a smooth air gap.
The parameter values are:

..

"

R' = 3.5 ohms

R' = 4.0 ohms

L~
L~

= 0.04 henry
.., 0.125 henry

'f
f

: 1'1J..li

Axis fixed to stator

V -=-

~--

i~

L:: = 0.07 henry

I

11"~-- fZ'li"d","~:'
-------.:--7.11 1

=L" cos</>
I'

LI

----

Oro

.
Axis fixed to rotor

..

ilKj

fI,

;i$'

,4

.vh

ru~

,,~.

''lfJi

""it'

,if"

Fig.3P-2.

Fig.

The lotor has no winding. Assume that the inductance of the stator winding
as a function of the angular position cp of the rotor is

,,'

L' =

wherc L o and Lk are constants.

Lo +

L k cos

kcp

Also assume that the stator current is

i = 1 cos wt

(a) Derive an expression for the speed in radians per second at which the
motor develops average torque. Express the speed in terms of the angular
frequency w of the current and the number of rotor teeth k.
(b) Also derive an expression for the average torque at this synchronous
speed in terms of the current I. the inductance coefficients L o and/or L k and
any uther variables which you f11.IY need.

3P~.

The value of the voltage source is V = 10 volts doc. The rotor is locked at
= 7T/2. With the currents initially equal to zero the switch S is closed at
time t = O.
(a) Find numerical expressions for
and i~ as functions of time.
(h) Find a numerical expression for the electromagnetic torque produced as
a function of time.

4>

i:

)-5. By applying constraints to the equations for the generalized electromechanical energy converter, the equations of motion for many ,pecific
machines can be obtained. Determine the constraints necessary for the following configurations and write the resulting electrical and me(;hanical equations.

250

'~

ELECTROMECHANICAL ENERGY CONVERSION

Ii
,

(a) A two-pole machine with a salient-pole stator and two windings at right
angles to each either on the rotor excited by semi four-phase (two-phase)
voltages. The load is characterized by inertia and viscous damping.
(h) The same machine with only one winding on the rotor excited by
v~ = Von sin wt and the load characterized by viscous damping and a spring.
Determine the steady state current i~. (Hint: In the steady state assume c/> is
const<lnt.)
(e) II' the damping ex is large, negligible time variations will appear in c/> for the
result of part (h). For this case, write the transcendental equation in terms of
Vm and the parameters from which c/> may be obtained and indicate methods for
solving this equation.
3-6. An expression is developed in the text for the instantaneous torque of
a two-pole rotating energy converter having one winding on the stator and
one on the rotor. This is

T = K{cos [(w'

w')/+ y]

+ cos

[(w' - w')1

THE GENERALIZED ROTATING MACHINE

... '

at a constant speed by an air turbine drive motor. The self- and mutual
inductances and resistances of stator and rotor are:
stator self-inductance = 2,5 henrys
stator to rotor mutual inductance = 1 cos c/> henry
rotor self-inductance = 0.5 henry
stator resistance = 100 ohms
rotor resistance ~ 0
The generator is driven at a constant speed of 3000/2" rpm, and the stator
winding is excited by a I-amp constant-current source. That is,
stator current = 1 amp doc (from a current source)
rotor speed = 3000/2" rpm
An inductance load of 0.5 henry is connected to the rotor terminals. Assume
that the current in the inductance load has reached steady-state conditions;
then evaluate the following:

~,.

y)} sin'"

(a) The instantaneous steady-state rotor current.

in which K = constant
W = angular frequency of stator current
w' = angular frequency of rotor current
y = time-phase angle
c/> = space angle of rotor structure with respect to stator structure

(b) The instantaneous steady-state stator voltage.

(c) The instantaneous steady-state electromagnetic torque.

(d) The average steady-state electromagnetic torque.

(a) Adapt this expression to the single-phase synchronous motor in which

the stator coil carries direct cu'r:rent and the rotor coil sinusoidal alternating
current of angular frequency w. f[he instantaneous torque will of course involve
the rotational speed W m of the rotor.
(b What is the instantaneous torque when W m = w? When Will ... -w?
At standstill ?
(e) What is the average torque under each of the three conditions of part (b)?
3-7. A single-phase, cylindrical stator and rotor, two-pole, synchronous
generator, as shown schematically in Fig. 3P-7, has a single winding on the
stator and a single winding on the rotor and each winding is sinusoidally
distributed in space around the periphery of the air gap. The machine is driven

J
I

i.= 1 amp d-c

+

r-

,jp'.
':_~~ ,
;'il

~'eJi\:\l \

o
~,

~f~!~
-~

o,._.,._.!

c/>=wt+8
rotor windings (i' and b' is connected through slip rings to a resistance load
of 60 ohms.
(e) What is the instantaneous electromagnetic torque Tt in newton-meters
under the operating conditions of part (b)?

-::

','.

'fl
t,"

---

~'

ir

".

I"

Ir,'1

:I:'~

j~~'
~.~

Inductive
load of
0.5 henry

"q

;-.'

,.f.'
'~;"

"F

t,

Flg.3P-7.
iL'

,1

:;p

,IJ(';

, "",
,~

"J"'~a.!

3-8. A two-pole, two-phase, 60 cps, nonsalient synchronous generator has

inductances L~ = 0.06 henry and L:; = 0,15 henry. The rotor resistance is
1 ohm/phase. The a' winding carries the field current of 200 amp doc and
there is no b' winding. The machine is driven at 3600' rpm.
(a) Write the expression for the instantaneous voltage v~ across the a' winding
in terms of i; = I', L~, R', S, and /' if
i~ = I' cos wI
" ...j~
i~ = -I' sin wI
(b) Find the steady-state magnitude /' of the rotor currents when each of the

-V

I
I

,"

C'r'

\,,',

<t> ... wi

251

3-9. A two-pole synchronous generator with cylindrical stator and rotor

surfaces has two-phase windings on the rotor and a single field winding on the
stator. The windings are distributed sinusoidally in space around the periphery
of the air gap. The resistances and inductances of stator and rotor are:
stator resistance = 100 ohms
stator self-inductance = 25 henrys
rotor resistance = 0
rotor self-inductance (each phase) = 0.10 henry
The stator-to-rotor mutual inductance varies as a sinusoidal function of
angular position of the rotor and has a maximum value of ;/2 henrys.
The generator is driven at a constant speed of 12.000/" rpm and the field
winding (stator) is connected to a doc source of 1 amp.
(a) Find the rms value of the voltage at the terminals of a rotor phase when
both rotor phases are open-circuited.

'"

ELECTROMECHANICAL ENERGY CONVERSION

THE GENERALIZED ROTATING MACHINE

(h) Find the rlll.v value of the steady-state current in a rotor phase when tho
terminals of h"th rolt)J' philses are short-cil'l:uilcJ.
(c) Find the rillS value of the steady-st;lte current in a rotor phase if a 40-ohm
resistor is connected across rotor phase a and a 40-ohm resisll1l' is connected
across rotor phase b.
(d) Find the average power converted from mechanical to electrical form in
pilrt (c).

machine tool is rateJ Ht 2 kw, 100 volts rms line-to-neutral, O.S power factor
lag. An inJuction nHlchine with the following parameters is available:

252

= O.I.Q

R' = O.I.Q

#'

L~

= 57.4

x 1O- 3 henry
L~. = 57.4 X 10- 3 henry
~

L:: =

56.4 x 10- 3 henry

,(
'~
,i

"

(0) It is found that if the angular position of the rotor is varied the magnitude
of the open-circuit rotor linc-to-neutral voltagcs does not change. However,
the phase of the rotor voltages with respect to corresponding stator voltages
changes linearly with the angular position of the rotor. Explain this
phenomenon.
(b) How far must the rotor bc rotated in mechanical degrees to change the
phase of the rotor voltages by 45 electrical degrees?
(c) The rotor of the induction machine is to be attached to another 230-volt
rms line-to-ncutral, 60-cps, two-phase voltage source. It is found that before
the connection is made the line-ta-neutral rotor voltages lead the corresponding
line-to-neutral voltages of the new source by 15 electrical Jegrces. When the
connection is made, what is the steady-state real and reactive power delivered
to the source attached to the rotor? Make use of Thevenin's theorem.
(d) How could one reduce the power flow calculated in part (c)?

312. A certain high-speed machine tool requires two-phase power at

100 cps. Howcver, there is only two-phase power at 60 cps available, It is
suggesteJ that the' IOO-cps powcr be supplied by driving an indlll.;tion machine
at a slip greatcr than I, so that the rotor frequency will be 100 cps. The

= 56.3 x I() .\ henry

3-13. Use the equivalent circuit of . lin induction motor under balanced
operating conditions given in Fig. )-11 of the text to answer the following:
(a) Write an expression for the torque of the motor neglecting the exciting
impedance jw'L',:.
(b) Show that the torque is the power transferred across the air gap divided
by the synchronous speed.
(c) Show that if the slip of an induction motor is small, then its torque
varies approximately linearly with the slip.
.
(d) Show that the maximum torque that an induction motor can deliver is
independent of rotor resistance.
(e) Show that the slip at which the maxi~um torque occurs varies directly
with the rotor resistance.
(I) Find the ratio of rotor copper loss to output power in terms of slip alone.

3-11. A six-pole, two-phase, wound-rotor induction machine has its rotor

blocked so that it cannot rotate. However, provision is made so that its
angular position may be changed. The stator of the machine is excited with a
230-volt rms line-to-neutral, 60-cps, two-phase voltage source. The parameters
of the machine are as follows:

R'

I.:.'

"

(a) At what speed is the synchronous machine running in rpm?

(b) What is the peak mIlle of the excitation voltage E'?
(c) rind the phase angle between the excitation voltage E' and the line-toneutral voltage V'.
(d) What is the rms line current?
(e) What is the minimum field current at which the machine can possibly
supply 2.64 kw to the voltage source?
(f) What happens if the field current is reduced below the value determined
in part (e)?
,f;,

'f,

57.4 X 10- 3 henry

L: = 57.3 x 10-' henry

(0) At what slip must the induction machine he driven to supply the correct
frequency to thc machine tool?
(b) If the machine tool is to operate under rated conditions, what must be
the stator supply voltage to the induction machine?
(c) How much power is supplied by the 60-cps line?
(d) What is the mech,lnical power input to the induction m,lchine?
(e) What is the power loss of the induction machine?

310. A six-pole, uniform air-gap, synchronous machine has its stator

connected to a 60-cps, DO-volt rms line-to-neutral, two-phase voltage source.
The synchronous reactance of the machine, w'L;., is 20 ohms and the mutual
reactance between rotor and stator, ",'I.;:, is 60 ohms. The synchronous
machine is driven mechanically by another machine so that the synchronous
machine supplies 2,64 kw to the two-phase voltage source. The field current
IS 3.84 amp doc.

!I'

I.;. =

R' = 0.106.0
R' = 0.095.!J

253

..... '"

II
I

~'
i

:f

"'".

~~W
t;

/It

'f;.:J

:('.

M~( 'j'
"

~'''I.

,~'
';'0

,Ji;,

.,;J :';;;i

,'W

WIlT
';j: ~ '

.'i--;t

t:; :"111'

~--

TWO-PHASE TRANSFORMATIONS

Two-Phase Transformations
and

:i
..

The Generalized Machine*

4.0

Introduction

An idealized model of a general-purpose electromechanical energy

converter has been established and the equations of motion derived. The
mOdel adopted has a two-phase stator and a two-phase rotor, and the
independent variables used to obtain the equations of motion were the
four coil currents and the angular position of the rotor shaft. These
variables were shown to be useful for the study of the steady-state
characteristics of the generalized machine with balanced, two-phase
excitations or loads. For unbalanced excitations, unbalanced loads, or
general dynamic operation, the use of the coil variables led to complex,
sometimes nonlinear, differential equations which were difficult to solve.
It has been found that for many of these cases the equations of motion
can be greatly simplified by choosing new sets of variables and eliminating
the coil variables from the equations of motion. It will be shown in this
chapter that for many interesting and important modes of operation of the
generalized machine considerable simplification can be achieved by
effecting a change of variables in the equations of motion. The new
variables (currents and voltages) will be defined in terms of the coil
variables and subsequent substitution of the new variables into the
equations of motion will provide the simplification.
The techniques introduced in this chapter are similar to those used by Dr. Kron,
Dr. Gibbs, and others; e.g., see G. Kron, "Non-Riemannian Dynamics of Rotating
Electrical Machinery," Jour. Math. and Physics, Vol. 13, 1934, pp. 103-194, W. J.
Gibbs, "Algebra of Electric Machine Analysis," Technical Monograph T.M.S.
757, British Thomson-Houston Co., Rugby, England.
254

;Jffi,

,
:'~~

t;l

',

""

"J

'~!,' ~'I,

r
",

'c~j.','

'.;:

J
l

255

In somc electromechanical energy converters the coil variahles are

different from the tcrminal variables. Such a difference exists in the
rotor of a commutator machine. The commutator makes a physical
change of variables. The equations of motion, as derivcd previously, are
written in terms of coil variables, yet any external measurement on a
commutator machine will only yield terminal variables. Thus, in order
\ to write the equations of motion for a commutator machine in terms of
terminal variables, the physical change of variables achieved by the
commutator must be descri bed mathematically. The eq uations of motion
derived in thc previous chapter cannot be used to describc a commutator
machine until the relationships between coil variables and terminal
variables have been established. One of the transformations derived in
this chapter will establish this relationship and thus allow the use of the
equations of motion of the generalized machine in the study of commutator as well as noncommutator machines. The changes of variables
(or transformations) that will be considered are both linear and nonlinear.
The coefficients in the transformation equations will be real or complex,
and will be constant, time-varying, or functions of thc independent
variables. Because of the complexity of the equations of motion and of
the transformations, the techniques of matrix algebra will help in the
development and analytical treatment of the transformations. A brief
review of the matrix algebra that is useful in the analysis of electromechanical energy converters is given in Guillemin. *
In the study of rotating electromechanical energy converters the twophase s) _:;:m is the most fundamental. The importance of the two-phase
system lies in the fact that, regardless of the number of phases that an
energy converter may have on its rotor and on its stator, the equations of
motion can always be reduced to equivalent two-phase systems on rotor
and stator for considerations of torque production. Thus a thorough
understanding of a two-phase energy converter is a useful starting point
for the study of all rotating electromechanical encrgy converters. The
analysis of a converter with any number of phases is accomplished by
reducing the equations of motion to an equivalent two-phase system.
The transformations for reducing both a three-phase machine and an
nom phase machine to an equivalent two-phase machine are given in
Chap. 10. It must be kept in mind that the equations of motion used in
this chapter can be those of an actual two-phase machine or those of an
equivalent two-phase system of a more complicated machine.
As each of the transformations is introduced, its use will be illustrated
using the equations of motion of the two-phase energy converter that

..

E. A. Guillemin, Introductory Circuit Theory, John Wiley, New York, 1953, Chap.
10.
'.'

"

~i

'(I

~1'

t"~;,

",':1,"

256

'i

ELECTROMECHANICAL ENERGY CONVERSION

TWO-PHASE TRANSFORMATIONS

were derived in the previous chapter. A summary of all the transformations and of the transformed equations of motion appears at the end of
this chapter.

Defining the transformed impedance matrix !Z" as

f!l"

VI =

.'..

:J

in which the z's are independent of the currents but may be functions of
time or may be time operators (derivatives and integrals). These equations
are linear and can be written in matrix form ast

..1.'.

1'3

Z2l i l

Ii

'.

J:.

+ ZI2 / 2 + z\3i 3
+ Zn i 2 1- ::23i3
z 3J i J + Z32;2 + Z33;3

ZIIi 1

(4-1)

1'2

where

"

= V2

V3

i~]

:Y

/3

~'

zill

::21

Z22

Z23

Z31

Z32

Z33

"

p=

(4-3a)

.A e. " .;i
.
:=. ..w

(4-3b)

v2
Vl

i' =

i;

i;

'Wl't
ll

i)

~"

IiI

ill

i2

p
, ~-~

,~

~
I

V2

== ~'tV

==

e;.,'

(4-8b)

(4-9)

On the other hand, by restricting the transformation matrices in the

proper way, the power will always be given by an expression of the form
of Eq. 4-9. This restriction, called the constraint of power invariancy,
is easily found by substituting Eqs. 4-3a and b into Eq. 4-8b to obtain
(4-10)

p = i;.s#,.'4v'

The transformation matrix d is, of course, a 3 x 3 matrix which is

assumed to be nonsingular, real, and independent of all voltages and
currents. When Eqs. 4-3a and b are substituted into Eq. 4-2, there
results
.'4v' = 2:s#i'
(4-4)
Premultiplication of this expression by d- 1, which is the inverse of
matrix sl, yields
v' = d-I!!r.s#i'
(4-5)

(4-8a)

,?,

l'

f'

+ ;3 Vl

Since both the original (unprimed) and the new (primed) variables
describe the same physical system, the power into the system will be the
same regardless of the variables used to describe the system. However,
unless care is taken in selecting the new variables, the power will not
necessarily be expressible as

o
"'fli~J

;2 V2

V3

,.

= di".

where
v' =

This expression can be written in the form

,j.

vi

;IVI

".f,. ,.
Z12

A change of variables is defined by

v
e

P =

(4-2)

2:i

(4-7)

This expression gives the volt-ampere equations that result when the
variables in Eqs. 4-1 are changcd to primed variables. Note that the
form of the expression has not changed. This is natural because a set
of linear equations has been subjected to a linear change of variables.
Consider next the instantaneous power flowing into the system described
by Eqs. 4-1. The power is givcn by

Consider the set of volt-ampere equations

,.

(4-6)

v' = 2:'","

t~

Li near Transformations

= s(- I !!r.fII

Eq. 4-5 becomes

:.1.

4.1

257

But in order for Eqs. 4-9 and 4-10 to be identical, the transformation
matrix must satisfy the condition
",

~;

d,d
o~r

i!!J 10'.1

>

~},

~,

0\:;

,;,

>

IJIt = the unit matrix

= d- 1

(4-11)

".I
d;1 = d

.-'1

t In the treatment 10 follow, a knowledge of matrix algebra is assumed.

of matrix theory is given in Guillemin, Joe. cit.

Some of the transformations that will be used are complex; i.e., the
matrix d of Eqs. 4-3a and b will contain complex elements. As a result,

A review

,.

:ji
.,

258

with all variables and parameters real.

~ was defined in Eq. 3-97a as

the new variables as well as the new impedance matrix will be complex t
The complex transformations to be used in this chapter have been arrived..
at by finding complex transformations that yield desired simplifications'
in the impedance matrix. Consequently, the original selection of these
matrices was a trial and error procedure. Nonetheless, all of these
complex transformations satisfy the relation

"iI';"

T.

or

~/\

$;1'" - d

'il

,,

Notice that the conditions expressed by Eqs. 4-12 reduce to those of Eqs.
4-11 when the transformation matrix $ is real.
Transformation matrices that satisfy Eqs. 4-12 are known as unitary,
or Hermitian orthogonal, matrices. t When the complex transformations
are of this 'type, the power (a quadratic form) will remain invariant in
form and magnitude when the power is defined as

,:116;-

(4-13)

! t.:'" ..

;.~

'I~J,

." !.

4.2

(4-14a)

The use of Eqs. 4-12 reduces this expression to

ti

(4-14b)

Thus, complex transformations that satisfy Eqs. 4-12 keep the power
expressed as in Eq. 4-13 invariant in form.
The electromagnetic torque as derived in the preceding chapter has the
form

(4-18)
Ii

+t,

~"'\~. ~~

'i...

ttl&fl
of/!

(4-17)

When the variables are real, Eq. 4-18 becomes the same as Eg. 4-17.

,<;.~~~:

i l

p = iiv = ($i')idv' = i'ididv '

T. =

H~)i(~

~:~j1:'

When the variables are real, Eq. 4-13 is the same as Eg. 4-8b because the
conjugate of a real variable is the variable unchanged.
To show that transformation matrices satisfying Eqs. 4-12 yield the
invariant form of the power given in Eq. 4-13, assume that the transformations of Eqs. 4-3a and b are complex and satisfy Eqs. 4-12.
Substitution of the new variables into Eq. 4-13 yields

f-"i v '

.",,,
T -_ "21(1)
~ f- I <9

"I':
.

p = iiv =

(4-16)

It is quite easy to show that the voltage of energy conversion If transforms

in the same way as voltage in Eg. 4-3a; consequently, since the form of
Eq. 4-13 maintains power invariant when the transformations satisfy
Eqs. 4-12, the torque will also remain invariant to transformations
satisfying Eqs. 4-12 when it is written as

'd

i~:

f-';v

aa~ ~i

(4-12)

$-1

or

p =

The voltage of energy conversion

Substitution of Eq. 4-16 into Eq. 4-15 establishes the torque as

.9Ii.91 = q,
$i =

259

TWOPHASE TRANSFORMATIONS

ELECTROMECHANICAL ENERGY CONVERSION

.~:;;

,
,,'

.~

......
.,{

(4-15)

.~~

t F. B. Hildebrand, Methotb 0/ Applied Mathematics. Prentice-Hall. New York.

~f~

1952. pp. 45-46.

t,

Equations of Motion for the Generalized Machine in a{J

Variables

The equations of motion for the generalized machine were derived in

Chap. 3, Eqs. 3-95a-d. These equations were expressed in terms of the I
two-phase coil variables ab on the stator and ab on the rotor. These
two-phase equations of motion have much broader application in the
study of multiphase machines than one would at first anticipate. In
Chap. lO it is shown that for the study of energy conversion or torque
production any multiphase machine which has balanced symmetrical
impedances internally and externaIly, operating with any arbitrary
unbalanced polyphase excitation, can be reduced to an equivalent twophase machine. This result is not too surprising because any set of'
multiphase current sheets can always be reduced to two equivalent
current sheets of constant amplitude rotating in opposite directions or to
two current sheets of varying amplitude fixed in space in electrical space
quadrature. This result is of course equivalent to a two-phase representation of the multiphase machine. In this chapter the phase variables will
carry the subscripts a and jj in place of the subscripts a and b to indicate
that the equations of motion and all transformations are valid for and

.~

ELECTROMECHANICAL ENERGY CONVERSION

260

TWO-PHASE TRANSFORMATIONS

applicable to all multi phase machines expressed as equivalent two-phase

machint:s. t
In Sees. 3.1 through 3.3. the equations of motion for the generalized
machine were derived. In these equations the inductance coenicients had
either a subscript Ii' or IL2' The subscripts indicated whether the
inductance coetlicients defined by Eqs. 3-31-3-36 contained the permeability 'L or IL2' In accordance with the definition of an anisotropic
permeability (Sec. 3.1), the inductances proportional to fL were those of a
smooth-air-gap machine. and the inductances proportional to fL2 were
those which resulted from saliency.
The equations of motion for the generalized salient-pole machine may
be divided into two parts: (I) All resistive terms and all inductive terms
proportional to IL; these are the equations of motion of a smooth-air-gap
machine. (2) The remaining inductive terms proportional to fL2; these
are the saliency effects. The following equations of motion, derived from
Eqs. 3-90a and band 3-95a-d, show this division into nonsalient and
salient parts.

"

iJ

>.j

+ p![,. ~~.al3)t':';'a/3

(4-190)

::'1

(4-19b)

,f.;J ~ ~..J + .[/. :'11 .(1.

t:;l
~ [P(/,. ::'aIll. I rr./3] + p(l,. :~ ]

~
a

VP

"

~
~

fR , + pL~ I

r--

I R + p.L::1

--="

cos ne/>
sin nE
pLS< f--------+--,. sin ne/>
cos ne/>

pL';

I sin n4:l 13' + pL: I o

Esin ne/> Icos n'tJ [0 I R' + pL~

[ cos ne/>

-il~l~lill

10
I~

1- si n n</>
~
~

'~~J

'-~:~

~
]i;

,f"1'>'~(

i~

'{

.~',.

ili i
i~

- nL::'[(i~i; - iIi;) cos ne/> - (i~i: + i~i~) sin nt/>]

~
~

(4-21b)

lc

.:1

'~'<f

"f

';~l

_.);'

,~

"

:!r" .'

~A:+ ~:r,j 't-

4\. "'":~

...
~I

pLJT

'so

!(Ul

PL~2~
~

la/3

I' 1

Eq. 4-19c I +

~'"

::,'

L'" J
..

~;,;

by Edith Clarke in Circuit Analysis of A-C Power SY,'lems. Vol. I, John Wiley, New
York, 1943,

(4-21a)

~I
p(/"'2 ~.aIl] I p[I"2 ~l
-~
P[l"2~] IP[l"2 ~~,a/3J

r-

t The equivalent a{3 two-phase representation of three-phase networks was introduced

~JEq.~J+
t; t

~
~

la'

(4-20a)

i',.

"2

I~
~ cos ne/> 1- sin nj;
-sin ne/> I-cos

n'L

JT
pL"2

pL~2

co,..j - s;n

-sin nt/>

~cos

n'

I -cos ne/>

20.1 - s;n 2o~

~sin 2ne/> I-cos 2ne/>

~
~

iJ
-=-1
i~

','

(4-20c)

= (~~.aIl + p.?", ::;.a/3)e'::;.aIJ + (p2,.. ~~.njl)(.'~~a/3

~--I
~~

Electrical
v ~~.aIl

>

Te ,.

1- cos n~/>

(4-20d)

la

Mechanical
-

sinne/>
cos ne/>

= (J4> + a~ + Kt/

ilL"
,.

sin ne/> I cos n~

,. - cos nc/> [- sin fie/>

nLJT

(4-19c)

T = Tm

I 01
@ I ~

fO

j .. ~

1-

Lt,. ~.lIjll , O-=.I 5!J,

(4-20b)

~...!

,:~~

~~I

2. Equations of motion for a salient-pole machine

Electrical

l;t.~...,

;;)

[0 I(T/. ::~"f1n

Ke/ - 1

T = (J4> + a.~ + Ke/

I, Equations of motion for a smooth-air-gap machine

')~;.aIl = (!:j/~~a/3

(N + a~ +

T=

26/

(4-21c)

TWO-PHASE TRANSFORMATIONS

ELECTROMECHANICAL ENERGY CONVERSION

262

Mechanical
T = (Tm

T t ",) - T.",~ = (Eq. 4-20a) - Tt,,~

I [1""'2 ~;,a~ ~

;:;-:l

::,n~] I

[1""'2

4.2. J

(4-22a)

[1""2

:Il~

(4-22b)

'.~
-~:;

....;l"..

"

nL"'2
-

I
I

~ sin n'; I- cos nc/>

nLsr
"'~

'os n~
sin n.p

I
~ sin nc/> - cos n.p I

-cos nc/>

s
la

~
~p .
i~

sin n.p
-

'~ sin 2n'; I- 'os 2n+

2nL" 2

cos 2n.p

r
la

sin 2nc/>

The transformed equations of motion can be obtained from the equations

of motion expressed in terms of a{3 variables. In writing the torque
equation, the torque expression derived for complex variables (Eq. 4-18)
must be used, since the transformation d may be complex. Thus the
equations of motion become

-----"

(4-22c)

(Eq. 4-20d)

+ nL::Z{[U;i; +

i~i:J cos n.p + U;i; - i~i;) sin n.p]}

- nL~2{[(i~)2 - (i~)2] sin 2n.p - 2i;i; cos 2nc/>} (4-22d)

vs,r =

"\,

The four sets of equations (Eqs. 4-19 through 4-22) express the equations
of motion for a smooth-air-gap machine and for a salient-pole machine.
These equations as written also serve to define the various submatrices
[I", ~], [I" ~~,aIl]' [1"", ~~,all]' etc. One further formulation of the equations
of motion which is particularly useful occurs with the term p2i in the
volt-ampere equations written explicitly in terms of the voltages of
inductances 2[1(-' and the voltages of energy conversion,r ~(-' = (a2Ia.p)~i.
Thus, following the definition established by Eqs. 3-102, 3-103, and
3-"104, the volt-ampere and torque equations for the salient-pole,
generalized machine are
S r
(/1lJS r
+ 2s,r + ,rs,r 1)(-'s"
(4-23a)
v <4J.aIl =

all,aSP

on <4J,all

J1
'/'

+ a,/, + fi.'I'
V"/,,

as."'. [t-.

all,aIl'/'

.1~s,r

7:" aB,allt

'j\\).I'

,>~':~

ofJ,ajJ"'"

.,.

all,all

'\

"

[I" ~S] + [1"'2 ~Ill I [I" ~:,all] + [1"'2 ~';,afj]

I
= I [I'" r']
[I rs ] I [I']
[I' ]
a~,aS + "'2 a~,ofJ
ofJ + "'2 ."

I, ,

.r ','
I
II/J,II/J =

[0]

[1"I"

:~,aIl] + [1"1"2 ~.a/l]

I [1"", ~~ ofJ]

'

[1""'2

[1"

~]

"-

s,'
all,all

]j

i . , (4-23d)

;,''.f;'

~'.

,;~:,

I'

-:;

"

",1

, ... j'
j.

':\~
'..

.
.',

'.,'.

~.'

"ii'"

.k,
:.~:

a;,all

&lis,r

(4-25b)

all,ajJ"'" .

T = N + a~ + K.p - :'d-I,r~;,a~e:'s.r

a/l,ajJ"'"

(4-26a)
(4-26b)

It is easily seen from Eqs, 4-26a and b that the parameter matrices !Jf;;,all'
2~~a{J' and ,'T~i;,n{J are premultiplied by ,(;)1'-1 and postmultiplied by .<;1/,
New transformed parameter matrices can be defined as

(4-23c)

I"

and

ti;,r. &11

+ &I-12s.r
d.w is''
all,all dt

':;,

where
2"r

+ K</> -

(4-25a)

82's,r

'"~

""l,,~'

a~

+ p2~~,a/l)di'"

Th!':se equations can be rewritten by expanding the differential of the

triple pL,juct pC!'&Ie:') into three terms, Using the definition established
previously that ,r~;,aS = a2~;all/a</> and further that the transformation
matrix &I is power invariant so that &11 = d- 1 (Eqs. 4-12), the equations
of motion can be written as
,.,....
v s,' = d-I~s,r rYle:,.r + &I-12"r rYlpi'" + d-I,r"r rYI~i',r

(4-23b)

all,a,,- all.'"

&I-l(&f~~.a/3

T= N +

IJ

,rs,r .,L,'s,'

General Transformation of the Equations of Motion

The equations of motion were derived in Chap. 3 in terms of coil

variables and were generalized in the previous section to "r~ variables on
both stator and rotor. It is advantageous in further studies to choose
new variahles which may be either linear or nonlinear com hi nations of the
af3 variables. Consider such a general transformation ,9/ which may be
either real or complex, but maintaining power invariant (Eqs. 4-11 or
4-12). Thc clements of this matrix may be constant, functions of time,
or functions of the independent variables. This transformation matrix
&I is defined as
(4-240)
i~~.a~ = &Iis,r
and
(4-24b)
v~;,1I/J = &I t - 1v'' = dvs,r

''I'>.

;:;:::l

263

'1;,..'

).

/.

:}f"r .... c1-l,Jf~;,a/l'W

(4-27)

.!f'" = &I-12;;,~

(4-28)

,r"r

&I-I9"'.,.r
.",
all.ajJ'''"

(4-29)

'.

"""

')
I'

264

TWO-PHASE TRANSFORMATIONS

The new volt-ampere equations, expressed in terms of these parameters,

become

any single component of speed voltage to calculate a partial torque.

More attention will be given to this in the discussion of the d</
transformation.
It is now worthwhi Ie to consider several restricted transformations
which yield equations of motion that are more solvable for unbalanced
or dynamic operation of electric machines than the equations in terms of
a{3 variables. The first of these is the dq transformation. which will be
shown to have a physical significance beyond merely a change of variables
and which can be used to analyze commutator machines as well as noncommutator machines.

,v,r = ('fl'"
,
T =

+ YS"p + :7",r ~.
l)"""r + 91-1!lJ ",r
del i"r (4-300)
~~~

J;r, + a~ +

Kef> - ~/:-,*.:rs,r(.'s"

(4-30b)

The transformation matrix Sy which eliminated the a{3 variables has not
affectcd the general form of the torquc relationship (Eq. 4-30b), but it has
produced a signifIcant change in the volt-ampere relationship (Eq. 4-300).
In Eq. 4-300 one observcs that thc three terms in parenthesis on the righthand sidc of the eq uation arc the usual VOlt-ampere eq uations, but in
addition there is a term proportional to ds.Y/dt. It follows from Eq.
4-300 that any transformation sI which has only constant elements will
not modify the form of the equations of motion, since the time derivative
of a constant is zero. Any transformation sI having elements which are
time-dependent or functions of the independent variables, and thus
implicitly functions of time, will change the general form of the voltampere equations.
In the study of the generalized machine there are two transformations,
........ -.....
dq and lb, which are particularly important. The dq or fb variables are
obtained from the a{3 variables by the use of a transformation which is a
function of ef>. i.e., sI(qJ). For these restricted cases the volt-ampere
equations take on an important and useful form.
Consider the restricted case when ,9/ is .e/(c/; then

1(1

f'

d.e/( c/ _ od( t/ ~

4.3

The equations of motion expressed in terms of the coil variables a{3 on

the stator and a{3 on the rotor are a set of five nonlinear equations which
are functions of the four coil currents and the position.p. A new set of
equations of motion which are functions of four new currents and the
velocity can be derived by a transformation, called the dq transformation,
which is a function of the mechanical position ep. The desired transformation does not affect the stator variables i~,BJ and v~,BJ but does
change the rotor variables i~/lJ and v:,eJ to a new set defined as i~qJ and
V~qJ. The new set of real variables, the dq (direct and quadrature)
variables, are written as
.!iil :.!-fl': H'" ~ ~ ,)' i.~>i,

~ =~

and the term in Eq. 4-300 which is proportional to dsl/dt becomes

proportional to~. Using Eq. 4-31 it is possible to collect all terms in
Eq. 4-300 proportional to <p and to define a speed voltage matrix or f
matrix as
f'" = .:rs,r + ,9/-1(.I..)2"r
osl(ep)
(4-32)
'f'

uP,aII

+ !f",rp + f"r vs,r

= N + a~ + Kep - !e;.;,r*.:rs"is,r

'., :~..~. .f:'.~:'"

~,

(4-33b)
.(',

The most significant item in Eqs. 4-33a and b is that 's"~:Y',.r, i.e.,
the matrix of speed cod]icients does not equal the matrix of tOI'1 Je
codlicients. This of course does not lead to violation of the conservation
of energy. but it does mean that it is not possible to indiscriminately use

(4-34)

'r

lq

t:..

~=~~J~

(4-35)

where the matrix ~ represents either voltages or currents.

formation matrix

(4-330)
?

jr"l _ ~'d

.:!!!J-

'"'~'"

,;:;)J::f)"~ .~!"~

o.p ",

v,r = (:Jt',r

and

.~

These new variables are defined in terms of the a{3 components as

The equations of motion can be written in terms of this f."r matrix as

The dq Rotating Real Transformation-

(4-31 )

~-~

;;

t'

265

ELECTROMECHANICAL ENERGY CONVERSION

[a~qJ

is given

The trans-

bYj'~

[~J:: ~ cs~~ :~

sin
cos

nil
n!J

~tf~

(4-36)

The dq variables were first introduced in R. H. Park, "Two-Reaction Theory of

Synchronous Machines," Part I, TrUll.\', AlEE. Vol. 48. 1929, pp. 716-727; Part II,
Trans. AlEE, Vol. 52, 1933, pp. 352-355.

liil'l,

266

ELECTROMECHANICAL ENERGY CONVERSION

i~

The inverse transformation is

X~ql

[a~J -1 ~

and

idand

(4-37)

i; attached to the rotor can be translated to a set of rotor currents

~

I- sin n~ ~
~ Gin nt/> I cos niJ ~I

[adq]-l

= IC?S n~
~1ll n~

'.'

nil

- sin
cos n!J

.!fI/'"

i; that are stationary relative to the stator by the transformation

t.

with

267

TWO-PHASE TRANSFORMATIONS

(4-38)

= )c"os

n~

,Jf'

(4-42)

Im.~t

Equation 4-42 is simply the transformation given by Eqs. 4-37 and 4-38.
.When the transformation of Eq. 4-35 is written out for a set of twophase voltages, the two resulting equations are:
!

Quadrature (q) axis

"

V~ = v~ cos n~ + v; sin n~
p = - sin n~ + v~ cos

v;

(4-39)

n,p

These relations can be interpreted geometrically in terms of the two

coordinate systems shown in Fig. 4-1. It is assumed that the coordinate
system formed by the d and q axes is fixed and as a result the coordinate
system containing the (X and /3 axes is rotated through the angle ~. Hence,
when ,p is a function of time there is relative rotation between the two
coordinate systems. From examination of Fig. 4-1 it is apparent that
the transformation of Eq. 4-35 refers two perpendicular vectors from a
stationary coordinate system to one which is rotated through the angle 4>
about the origin.
The a{1-dq transformation is particularly useful in the analysis of doc
machines because it can be used to describe mathematically the transformation imposed physically on the rotor current and voltage variables
by a commutator. To illustrate this physical transformation, use Fig. 4-2
(the diagrams are drawn. assuming n = 1) where it is shown that by
locating two current sheets in electrical space quadrature on the rotor
but fixed in space relative to the stator, and by exciting these two windings
with currents proportional to cos n~ and sin n,p, an equivalent current
sheet is produced on the rotor that remains fixed in space relative to the
rotor. If the rotor currents i; and i; in Fig. 4-2 are related to the rotor
phase current i; as
l~;l',

.41ft

id = i; cos n,p

i; =

i; sin n,p

.1

r;

8-""

\1
",'"

..",..",......

...... ......
"

",'"

... '"

........... "......

'11.;'11 (,l.",d ...

vqCosncf>

\~,

\
\

)..~ v~ sin ~ncf>... '" ....... '"

\"'t-

<co''''''

,u

~:.

Direct (d) axis

\d

.pt>

'J ",.

'H]<.,,;;

\
2\

'f""'"

,+

'1t~

I bt~,i'

,,qO"l<.?

'k
s'

(4-40)

(4-41)

then the two rotor currents on the rotor surface which remain fixed in
space relative to the stator produce an equivalent current sheet on the
rotor identical to the current sheet that is fixed to the moving rotor. In.
view of this equality, it follows that a set of two-phase rotor currents

Fig.

~I.

Geometrical interpretation of the afJ-dq transformation.

.:]

A set of current sheets on the rotor which remains fixed in space relative
to the stator can be established by injecting currents into the rotor through
brushes that are fixed in space relative to the stator and that slide on the, ~In
rotor surface. This method of establishing current sheets on the rotor
is shown schematically in Fig. 4-3. These current sheets established by
stationary brushes produce the same air-gap magnetic fields as do the
current sheets which are fixed to the rotor and rotate with it. provided
.
the angular dependence expressed by Eq. 4-42 is included in the coil ,..fJ
currents. The physical arrangement of Fig. 4-3 shows sinusoidal brushes, ,~

",'

/- \

/\

Y
\

..... _

vr cos ncf>

\\

...

~ \'

,'l!.

1'1'% )..........

t..

~,.
,

t~:

\ Va'

G~~

1\

't>l

V{J

//

ncf>
)... a-axis

II

r It

,...,",j'

"ll"l

vr

';.'

.0'

"

!~~'\
J._urw.~~ ~ \~~\
~

x.(; ~"X.. :,/~io.y

-f)

t: ~

~r-

-j

.3~
~~
o:l ;>-.
...
"'0 ";;;

\~~ ~ ) 8

' ; ) . ' . r " , > t ,;_

but in a practical machine brushes inject current usually at a pair of points

(commutator bars) on a continuous rotor winding. 1n the practical
machine with concentrated brushes, the actual space distribution of rotor
current density approaches that or a square wave. The sinusoidal
brushes and the resulting transformation consider only the spacefundamental component of the square wave. A more complete analysis

::1\ .~~ ,.
'" e

269

TWOPHASE TRANSFORMATIONS

ELECTROMECHANICAL ENERGY CONVERSION

268

,I

... ."

~r"" J<) ~ ~

;>-. c::
~
<l)
>;j) "'0

~ i:1

-s

~t: ...~

'-'cc:

VI

-'~~
..\-

,
'

'."_C':'~'

8~

8"

Infinitely distributed conducting grid into

which a sinusoidally shaped lHush inJects
a slnusOlddl current denSity t!lat remainS
fixed In space relative to the brush

Sinusoidally shaped
/1 brushes

dr,

~_d

<3

<t::

0.6

~2

.::: 0
oj

...

~irl
v,

<l)

..

-5

oS

8 .~
:::J

~,

"81i-~
5.~

.::

.J:: '"
.~
.J:: '"
~

,8

t.:,J

];' I'i--

"

'"-s -s

':1

3(.)

; . , .....:s

.~

;g S
<l)

'0

of commutator machines, taking into account all harmonics of current

density, is developed in Chap. II, and the results obtained there justify
the consideration of the fundamental component of current density at
this time.
An interesting property of the dq transformation can be illustrated by
transforming the set of unbalanced, two-phase voltages defined as

'"

6gi

- t::

~ G~
"oS
rj 2~
"""~~
.io~
]
u. '-' <=:

V~ =

V; sin (WI + y) .

p = Vpsin (WI + Y ~ ~)

,f':
.a

Gi!,-~

Fig. 4-3. Idealized rotor structure for producing a sinusoidally distributed current
sheet on a moving rotor where the current sheet is maintained fixed relative to the
brush axis. (dq brushes drawn for n = 1.)

:E .g
;g:a
-8 ~
c
~

Note; This rotor replaces the rotor

in Fig. 4-2 when analyzing
commutator machine5 or
when using dq variables

v; cos (WI + y)

(4-43)

II

!'

,~

but in a practical machine brushes inject current usually at a pair of points

(commutator bars) on a continuous rotor winding. 1n the practical
machine with concentrated brushes, the actual space distribution of rotor
current density approaches that of a square wave. The sinusoidal
brushes and the resulting transformation consider only the spacefundamental component of the square wave. A more complctc analysis

-;--- .~'"
'C

l::

.2~
~~

oj

;.
"
~
u

'"

.~ ~

~ dOJ

"; ,~

Infinitely distributed conducting grid into

which a SinUSOidally shaped lHush Inlects
a SinUSOidal current denSity lilat remains
fixed In space relative to the brush

"0

-QOJ t::<>::J
t:: ...

dr

::J

,-,0

p:;

vi .......

~_d

\I

v>

u-

,
.....

.~

;... c::

"

l,

;...

-0

'-"
p

269

TWOPHASE TRANSFORMATIONS

ELECTROMECHANICAL ENERGY CONVERSION

268

:2$
OJ
<l';

--

o.~

'<>1"

<11_

...

""0
~

0; .8
~ i{
~ is.

OJ ~
-5..::
8 .~

J:t

Quadrature-axis brushes

;:3

1]
:a ';j
~

VI

,8
~3

II
-I'"

-2.

.~

.-

.:
~

........

./

~n~I'"

I I 1_

I'
"A.. '

;...l$

c .~ 3
:;

\.

Fig. 4-3. Idealized rotor structure for producing a sinusoidally distributed current
sheet on a moving rotor where the current sheet is maintained fixed relative to the
brush axis. (dq brushes drawn for n = 1.)

<.)

~:a

.g

~ ~

0
't'

in Fig, 4-2 when analyzing

commutator machlnt!s or
when using dq variables

-B r:!

Note: This rotor replaces the rotor

OJ

<II

of commutator machines, taking into account all harmonics of current

density, is developed in Chap. II, and the results obtained there justify
the consideration of the fundamental component of current density at
this time.
An interesting property of the dq transformation can be illustrated by
transforming the set of unbalanced, two-phase voltages defined as

~ ~;

2 ~ &

o ;:3 >'1
P:: <J ..

"'oS

,..l 2 .~ 0 ~

IlO p:; ~
it :; ~

V~ =

"Ii'~J'~

,
I

I
J,.~

,r'

p=

V; sin (WI +

V; sin (WI

y) .

+ Y~

~)

V; cos (WI

+ y)

(4-43)

'\'.

this transformation may bc thought of as a mechanical frequency changer,

producing a change of n(J)",.
The dq transformation, in addition to being valuable for the study of
commutator machines. is also useful in the analysis of synchronous
machines, For a ~ynchronous machine the rotor frequency 'u must eljual
to nW m fbr' steady-state operation. A synchronous machine operating
with balanced excitation and at constant electrical speed nW m equal to
the exciting angular frequency will have dq volt;lges defined by Eqs. 4-46
that are direct or nontime-varying voltages. Thus for constant-speed
balanced operation the dq transformation can be used to reduce the a-c
transient analysis of a synchronous machine to a simpler doc transient
problem. Both steady-state and transient analyses of the synchronous
machine operation at constant speed become quite simple when dq
variables arc used. The transient analysis of the unbalanced synchronous
machine at constant speed cannot be reduced to a doc problem. This is
shown by Eqs. 4-46, where for unbalanced operation with w = nW m the
dq voltages consist of a direct voltage and a double-frequency voltage.

When the transformation of Eqs. 4-37 and 4-38 is performed on these

voltages, there results for the dq variables

Vd

V; cos nc/> sin (wt

V; sin nc/> sin (wt

v;

V; sin nc/> sin (wt + y) + V; cos n</> sin ( wt +

+ y) -

i)
+ i)

+ Y+
Y

(4-44)

,J,

The use of trigonometric identities reduces Eqs. 4-44 to

.'

vd =
v;

(V r + V~
a...

f3

sin (wt _ n.p + y)

(V' - V~
a...

f3

sin (wt + n.p

(V

+2

V;)
f3

cos (wt - n.p

+ y) +

(V'

f3;

V')
a

cos (wt

vd=

(V'+
a 2

V~
f3

..'."
:.~.

w",t

sin [(w - nw",)t - nS

+
v; =

+ y]
(Y' _ V')

."!tg

:-;

+ n</> + y)

When the rotor is turning at a constant speed, ~ = w",; then.p

and Eqs. 4-45 become

"

~ -"

+ y)
(4-45)

271

TWO-PHASE TRANSFORMATIONS

ELECTROMECHANICAL ENERGY CONVERSION

270

(Y' + V;)
2 f3 cos [(w - nw",)t - n8
II

"

""

f3

sin [Cw

nw",)t

+ n8 + y]
(4-46) ~

+ y]

+ (Y'f3

=-

Y~

cos [(w

+ nw",)t + nS + y]
I

v;

y'sin [(w - nw",)t - nS + y]

v; = V' cos [(w - nw",)t - nS + y]

.'

..w~~.dq

//

ITu]

19

01

[adq~

(4-48)

where

(4-47)

(u] = unit matrIX =

~'0'

which is a single-frequency two-phase set. The new frequency for the

balanced dq components is (w - nw",) which is decreased from w by an
amount equal to the mechanical speed in electrical radians per second.
This frequency change due to the transformation is identical in magnitude
to the differences in frequency between stator and rotor found in Sec. 3.5
to be required for the production of an average torque. Consequently,
if the stator a{3 currents and rotor dq currents are of the same frequency,
an average torque can result. The commutator which is describable.by
~

The dq Transformation of the Volt-Ampere Equations of the

Generalized Machine

The dq variables can be used to transform the equations of motion of

the generalized machine and to produce a set of equations of motion that
is simpler in form. The dq transformation for the rotor variables [(l~qJ
was defined by Eqs. 4-36. The total transformation ..w:-:.dq for both
stator and rotor variables which leaves the stator variables unchanged is
simply

The dq voltages of Eqs. 4-46 consist of two different frequencies (~ - nw",)

and (w + nw",). Each of the voltages corresponding to these two
frequencies forms a two-phase set; thus the voltages v~ and
form two
two-phase sets of voltages. When the two-phase voltages v: and v~
form a balanced set, Eqs. 4-46 can be greatly simplified. Setting
V~ = V; = V', Eqs. 4-46 reduce to

v; =

4.3.1

i':l

~.
,~t"

[a'

J=

c?s n</>
- sm n</>

d'l

l?\I OlU

sin nc/>l
cos n~

,-J

..

The afl-dq volt-ampere and torque equations can be written in terms of

the, rxfl-dq variables and the transformation of Eq. 4-48 as
;l;~.

t~

eI..'
- [9/s"
- I ('j/s"
+ p2?'"
)..ws., ]i s"
P.dq , afl.",, af3.af3
' afl.a/3
afl.d" afl,dq

I:"

,;~,","

~k~
',.

;{, +
J 'I'

.1

a<p

+ L\.'f'
V.l..

1 ....'
(.SI1...afI,d",
, yS"afJ.dq9/""a/3,dq
"If. af3,dq,

(4-49a)
r
)e:...afI.dq

(4-49b)

lLtL I KUl"lI:LHANILAL tNEKGY CONVEK~ION

loll

= (;j.f-'"
aIi,dq

Iff..

ali,dq

!f-'.'
1)
ali,dq

= J> + Hi> +

f
'!I'"ali,dq'f'
J)e:-a/J,d,
,.

=r~'

(4-50b)

J~~IJ]0

[a~q]-I

~~

o ~~U]0

[r:.e]

[(~~

(/" ~IJ] + (/"2 ~/J]

[ll~q]-~

[11]-1

(I,. :~,alJ]

ITI,. :11] +

+ L:'2

0
L~ - L~2

V; + L::

[S'
~"

,-

/
I

.'
I

\"
)

LS;

l:

(1"2 ~IJD [U]

IS(
-'142

L:~

L11S

IS(

L~

rs
Lqli

L~

L0

ITI" :IJ] +

+ L::2
+
0

"

~liq

.,

L sr )

"

"2

-'-aq

[a~q~

sr
'-lJd

,-~

f,.r

- .r,.r

oP,dq -

ap,dq

Lo
. ,~

L: - L:2

;
'.)

.r'.:;,dq

dq

-T

-2nL:2

-2nL:2

(4-50e)

...

ods,r
+ d-1 oP.dq
S,' !'f"r ~ _
o#,olJ 01> -

oI[U]-I ITI" '.:;,oll] +

oI [~q]-l ITI,,~] +
.~i-

[1"2
[/"2

.rs,r

oP,dq

+ ._

I[I. :;'""J + II", :;>,

[U"Jj ~ :;(...J + [I" :;(..J

'If

neLS; - L':z)

~.

o ~ fl.:"l + [1",:"1

.ff

sr )
+ L "2

, ~uJ-'

"2

-n(Lsr

sr

--r;d
)'

:/JTI [a~qD

(4-50d)

0:'

-,-~

L sr

['-"2

[O;;'J

0
n(Lsr _

"

0
L sr _

L:2

[1"2

[U],

0',

sr )
+ L "2

Ua

Lrsda

[a'dq]-l

-n(Lsr

u]

~.

"

I[U]-I ITI,. ~~,alJ] + (1"2 :~.aflD [a~~

0
L~

Jto [ff~

('-"2~]

:-;'a,8] 1_

~~!l.,

'J

(1"2 :t,alJ] I

~~Iq]-I [II,. :t,alJ] + [/"2 ~J,a/JTI [U] I

L:.

(4-50c)

+ [/"2 ;;~.a/J]
[I,. :IJ] + (1"2 :/J]

('-"2

0
\ ,,; ,

I [I,. :;,a8]

:-;,oP) +

RS 0 0
OR'O
o 0 R'

2's,.
- sis"ali,dq
-12'S"ali,u./!'w'"afI,dq
a/J,dq _ ru]-I

('-"

:~,alil + ['-"2 ~~alJ~II-[uJ

r ['-"

[U)

0 0 OJ

IR'

[a~q~

ali,arntis"afl,dq

= ~Ub-I

OJ

I IT,-" ~~OIJ] + ~OflD [a~~

- [ra;,.l, ITT. ::'""J + [T., ~.;:,.,B I ' I"'.l, [T., :,J
J

where the transformed parameter matrices are

gf'"
_ #_<,.-1 ges"
, ali,dq - . ali,dq

273

.'Y"J,dq ... ~o/~~.""::r::,.,,,.cV~;"lq

(4-50a)

~l~~dql.r~;,dq<:'dq

Kq, -

TWOPHASE TRANSFORMATIONS

The transformation matrix .c/:~;~.d" is a function of ,I,; thus the equations of

mution. I:qs. 4-4'Jll and b, can be expressed in u form similar to Eqs,
4-JJll and " with ptlrameters detincd similar to Eqs. 4-27, 4-2~, 4-29,
and 4-32 as

[1.:"1 + [/,,:"l

Jt ~;.~

'Ii'

'::l O~qlJ!

~~,afi~

tat~J1I

:IJ~.1Ir8a~~u
[01> j

274

ELECTROMECHANICAL ENERGY CONVERSION

ads,r

_ fTs,r

ofJ,dq -

<lfJ,dq

~:-

+ sI-1 s,r
fe"r
~
<lfJ,dq
ofJ,ofJ
Oep

275

'

"

fs"

TWO-PHASE TRANSFORMATIONS

k
l'

--.

7" ...

<::s

'I.'"

V)
I

'V

"

=fTs,r

ofJ,dq

neLs,:

+ I0

-neLS; - L~)

-n(L~ - L:z)

+ L':z)

'-'

'l.

0
n(L:

+ L:z)

\~-

I.,'-"

-'--l--.. ..-......

J
:t~
r~~

neLS; - L~)

n(L~ - L:Z>

-neLS;

+ L~)
0

=1

-n(~

Grs
dfJ

-G;'

-G;'d

'

t; ..

<-.,{~

",t",

....
-...l

..
'--l

...

~ , .....

'--l
~

1(

+
~
--.- -

~~ --.
~~

I
l

;~

+ exl.
, 0/ + K~
'f' - nL'''(iSi'
fJ d - i'i"
qJ + nLsr(i'jr
fJ d +
<I

1-'2

i'i"
ql
<I

'-'

t; ..

t:l..

+ nL' (2irjl'\'~~
d ql

'if:, .

.,,',

I-'Z

~"I ~".
~

.,

~..

'-'
t:l..

'

.. ,/,':

>~\; ;'

.r: '

J ~.

>, ;..

: '~

'--l

.;.J;

I
....
-...l
~

+
C<

':.l
....
'-4

~
r::
I

,.'
I!
. t'llt i

-..
,

(}

::.

.....

'--l

'-'

t:l..

(.\

lS::

.,

r.:
I

I
.,J

II

Q1

"j

"'~"f:J ~r:,.l

t).;-1

('
:1, ,

(\'

i+

'--l"""$-

I~

I~tl

""

t; ...
~::s..::,.

, \

"-l

'
N
--.
""::1
~ '--l

-...l

;',,::;

,:

:.:

~ ;'

An important feature of the dq transformation is shown by Eq. 4-51a

where it is evident that all angular dependence of the coefficient matrix
has been removed. The angular speed ~ does remain; consequently,
Eq. 4-51a contains a product nonlinearity of j and~, When the speed~ is
constant, Eq. 4-51a describes a set of linear differential equations with
constant coefficients which can be readily solved. In many problems,
especially with systems in which the mechanical time constant is much
longer than electrical time constants, electrical transients can be
considered to occur with speed ~ constant,
There is one further relationship between the exf3-dq variables which is
particularly useful. It has been shown that ~~;,dq .,e :r',;;,dq' but from

-...l

"'l.:t

'~\ l.'

(4-51b)

'-

- .. I
....

+
' .. +

t; ..

\.tr'

, .. I

----,.. ,-.

..-..
.,' N

(4-50f)

(I

0-.1

..

'--l ''--l
-'
"""$~ :.:

The equations of motion given by Eqs, 4-50a and b can be written in

explicit form by using the parameter matrices fJl~:'dq, fe~:'dq, :r~dq and
~:;;,dq derived in Eqs. 4-50c-f Equations 4-51a and b result.

\I'

.....1
;';1 \

.>...~~
~--

G"dq

I-'

-...l

. ~h

T = p:
0/

0
0

+ L:2)

.,..

1'1

~_"':J

~<X:t

,~

'., a-"

276

ELECTROMECHANICAL ENERGY CONVERSION

Sec. 3.4.1 one might expect that one-half Of the speed-dependent power,
obtained by multiplying all voltages proportional to </> by the appropriate
currents, would be equal to the electromechanical power T,.</>. That this
is not true (or dq variables can be shown by considering the power due
to speed voltages pI( where

f"
_ (i"UP.llt,,
(lfs., is'' ) 1
!J4 ufj,dq nfl,dq r.p

4.3.2

"

'I
:~.
jti<
;'
,~

(4-52)

and the electromechanical power Pm = T,</> where

11'

Pm =

1 ( 's,
:rs' ,
's, ) j
2 t-,,~.tlql' ~,tli/.~,tlq 'I'

(4-53)

.~

,~

A compari~on of the values of Pti and Pm' found by suhstituting for i,

f , and ~f} from Eqs. 4-50 into Eqs, 4-52 and 4-53, shows that P8 = Pm or
1<;,tlq,:r~;,tI/~;,tlq

= i~:'Jq, '.~~,tii~;.tlq

(4-54)

The right side of Eq, 4-54 is not an identity term by term with the left
side, There arc terms in the power due lo speed voltages {J,r which are
not energy conversion terms, but instcad are speed voltages where
electrical energy in one axis is transformed to electrical energy in the other
axis of the machine. These speed-dependent transformer-type voltages
will be investigated later in the study of the two-axis d-c machine operating
as a d-c transformer. Another interesting observation is that, in terms of
af3 variables, one-half of the speed-dependent power Ps goes into mechanical power and one-half into energy stored in the field. For the dq
variables all the speed-dependent power goes into mcchanical power, and
none goes into stored energy in the field, The fact that Eq. 4-54 is an
identity can be used to write the electromagnetic torque T. either in terms
of .:T;;,tiq or <;;,tlq' Thus with no loss of generality the torque can be
expressed as
s"
T - 1'Z~'s','
'S,'
(4-55a)
"fJ.ilq,
"fJ,tlq~ ufJ,dq

~!
"

~.

J/;

1
~I
:.':'1;-

;,

e'

is''
<s"
~'s"
fJ,tiq, ufJ,dq ufJ.dq

(4-55b)

v'd

F
.~."
"

,~'

'I.

:i~

,I

.. '7

;il

=L :PL:

v'q

"j

VfJ

,."

The equations of motion have been transformed to a()-dq _Jordinates

(Fqs. 4-51a and h) and further it has been shown that the electromagnetic
torque can be written in terms of either :Y~;,e/q or '!J~;."q (Eqs. 4-55a and b),
The significance and versatility of these equations are best shown by
studying several examples of d-c commutator machines and a-c
synchronous machines.

Te --

The D-C Commutator Machine

The dq transformation was shown In Sec, 4.3 to describe the generalized

machine in terms ofvariabJes which were the brush currents and voltages of
an idealized commutator composed of sinusoidal brushes and an infinitely
distributed conducting grid for a rotor winding. A conventional commutator machine usually has a uniform re-entrant type of winding which
forms a completely closed circuit. but which is tapped at equal intervals
to conducting segments called commutator bars. The commutator bars
are insulated from each other, and concentrated conducting brushes slide
on the commutator bars and make elcctrical contact. The brush currents
and voltages of a conventional commutator machine and the dq variables
of an idealized commutator machine are intimately reljted. The exact
relationships are deri~d in detail in Chap. II, where it i~'proved that the
only differences bet~ye,en the equations of motion in term's of dq idealized
variables and thos~':or a 14onventional physical commutat,or machine are
changes in the relative magnitudes of the inductances ~nd speed coefficients. Therefore. it is permissible to use Eqs. 4-500. 4-50c-f. and
4-55b to write the equations of motion of commutator machines valid for
idealized dq variables. With only slight modifications. these same
equations are also applicable to the physical brush variables of a conventional commutator machine. These equations of motion for both
idealized dq variables and physical brush variables are
V

.:r

or

277

TWO-PHASE TRANSFORMATIONS

= J~

R~

+ pL'p

pL~~

,pG'S
dP

-</>G';.

pL~p

+ ex </> +

KeP

[Grs's"
tlfJlfJl d

pL~~

{)

"

R'd

odqrr
</>G

+ pL'd

-</>G~d
-

Grs'$'T
qa1alq

J'
'a

pL~d

(4-56a)

j'd

+ pL'q...i..J
j'

R'q

+ (G"dq

i; I

rr )',"]
Gqti
Ie/lq

i'.

(4-56 b)

where the symbols are as defined in Ta.~le 4-1. The equations of motion
given by Eqs. 4-560 and b are co~ktely general for any two-axis
commutator machine. The most impoi'tftnt condition to observe is that
for a conventional commutator machi~e':the inductance coefficients and
speed coefTlcients are not" equal. That' is. nL~fu ,9'= - G~~, I1L~~
G'jp,
nL~ 9'= - G~~/' and nL~
G';q for a conventional physical commutat(,)r,
whereas for the idealized machine the incqualities are equalities,
A most interesting effect can be observed by considering these inductances
and speed coefflcients as equalities. Observe that the speed voltages due
to a given current are always produced in an axis displaced - 7T/2 electrical

i_

278

ELECTROMECHANICAL ENERGY CONVERSION

TABLE 4-1

dq (Idealized)

Parameter

R'a
R'$
Rd
Rrq

Resistances

R'

NaR'

R'

N~R'

R'
R'

R'
R'

L'a

L~

L~1

KaL~

L'/l
La
Lrq

L~

- L~2

K/lL~

L~1)

Kq(L~

L~1)

L~~ = L;~

L': + L::1
L:: - L::1

Kad(L',; + L::2)
K$d(L:: - L':1)

Gro
qa
Grs
4/l
Grr
q4
Grr
<Iq

n(L',J. + L::)
neLs,: - L:~)
n(L~ + L:)
n(L,..r - L',..2)

nMqa(Ls,: + L::1)
nMd/l(L'; - L~~)
nMr(L~ + L:2)
nMr(L~ - L:2)

= Ld~

one stator circuit and one rotor circuit. It is apparent from the torque
equation (Eq. 4-56h) that i~ and i; or i; and i~ must be eliminated.
Removal of any other pair of currents would destroy the energy conversion
properties of the machine. This can be seen physically with reference to
Fig. 4-4, which is drawn for a two-pole (n = 1) machine. A rotor and
a stator field 'can interact to produce a torque only when there is an angular

I
i..
r
I

Stator C<- field magnetic axis

Rotor d-field magnetic axis

:>.
.8

4.3.3 The SIngle-AXis D-C MachIne

~1

.J,

i I'
Stator
!:I-field
magnetic axis
Rotor
q-field
magnetic axis

a";j,

\.~tt

....

1.1t.

;rw

~ ,:.~:.;n ~f

a;i

~l '.:'1' ".'

~,~
*f!\'1 x ,,;

. 6

1",= If

....".

radians from the axis of the inductance voltage. The minus sign means
that the axis is shifted opposite to the direction of rotation. Consequently.
an a current in the stator produces a negative speed voltage in the q axis
and an inductance or transformer voltage in the d axis. Similar results
are obtained for the other terms.

:JiJ

<P'~<p

. ,

1",= I f

The factors N, K, and M are constants determined by the winding and brush configurations in the physical machine. See Chap. 11.

279

".\.'

Kd(L~

L~~

TWO-PHASE TRANSFORMATIONS

..

L~ + L~2
L~ - L~1

Inductances

Speed coefficients

dq (Physical)

1-

')";

.,
"

"

lh!ot:

iI .

, '.

1: H

i~lj

A conventional single-axis doc machine (motor or generator) has a

single magnetic axis on the stator excited by a direct current. The rotor
has a commutator, and rotor excitation is supplied as direct current
through brushes which make electrical contact with the commutator.
The brushes are fixed with respect to the stator and are placed in the
angular position which gives the greatest torque, Le., where the rotor
magnetic axis is displaced 77/2 electrical radians from the stator magnetic
axis.
The brush currents and voltages are described by dq components, and
in general the dq equations apply to a two-axis commutator machine.
When only one axis' on each member is excited, as in a conventional
single-axis d-c machine, the equations must be specialized by eliminating

Fig. 4-4.

Field directions in two-axis commutator machine.

displacement between the fields, This is borne out by the fact that there
are no i~i~ and i~i; products in the torque equation.
With reference to the above discussion and to Fig. 4-4, Eqs. 4-56a and h
will be simplified for a single-axis machine so that positive currents
produce a positive electromagnetic torque; and a positive speed ~ plus a
positive electromagnetic torque produce motor action. This result can
be obtained by choosing the ~d axes with positive ~ or the aq axes with
negative ~ and negative T e The simplest choice is the (3d axes, but the
physical model of the generalized machine was chosen with the a stator

~,;;\"

~-~--~-

'.

280

ELECTROMECHANICAL ENERGY CONVERSION

l'nil hllvlng Ih nla~nl'lil' ;I.\:is altlng the s:ilil'nt Illl~mOer, The /tlf choice
makes thc physl(;al mudd match more c10scly a convcntional single-axis
d-c machine. Tl) eliminate any confusion it is advantagl:olls to show the
neCl:s,ary constrainb to the general equations of motion 01' El.js, 4-5(,(J and b
so that both the (3d choice of axes and the rxq choice of axes will lead to
the same equations of motion.
The essential ditference in the constraints of Table 4-2 other than the
choice of axes is that for the aq choice both the speed 4, and torque
T, must he negative to obtain motor action, Reversing the sign on both
quantities assures a positive electromagnetic torque plus a positive speed
TABLE 4-2.

by cupit.. 1 letters. and the rotor srced cp ..

will he II comlant. Witll
thcse constraints, the volt-ampere relationships of Eq. ~-57{/ rcduc:;c to

,>t'i

IJ)",

:-:~ ~

. I

Vf = Rflf
Vu = wmGuflf

(4-5l-i\

- ,,;

~i:

-Iu

tI

rxq (Primed Axes)

Rala

(4-59 )

First consider the steady-state energy conversion pr,opcrties of u

shunt-connected machine. The name comes from the ,f~lct that the stator
(field) circuit is connected in shunt across the rotor (armature) circuit, as

Constraints

(3d (Unprimed Axes)

281

TWO-PHASE TRANSFORMATIONS

'1

i~ =
i~ =

0
0

is = if
'r

'

G'.Jfl

1.1 =. 10

~ = ~

i~ = 0

Rfi = Rf
q = Lf
R'.J = R"

,f,

iJ =
".'Ii

I"
'r

= G ul

I.

t/>
4>

T= +T

R~ = R;
L~ = L;
R; = R~

0
0'

= If
= I.0'

L~ = L~

= -,p'
=.-cP'

G;~ = G~f

T=-T'
.,~""""

!k

to obtain motor action for either the f3d or rxq choice of axes, These
constraints used to simplify Eqs. 4-56a and b yield equations of motion
which are identical for both unprimed and primed variables corresponding
to a choice of the f3d and rxq axes respectively. Thus, dropping the primes
yields

rf +

~I

Ljp

[G:AlR" + La~

i7l
~

(4-57a)

Va

VI

.-~,~1

w",

tcr.

Fig. 4-5.

'~

t
,!

shown in Fig. 4-5. It is evident from Fig. 4-5 that this type of connection
places the constraint

VI = V"

,l'

"r;,
1,1,<

= if> + ct.,p +

Kc/> - T e

= J~ + ct.,p + Kc/>

- i.JfGuf

(4-60)

VI
If = R

(4-61)

and

.....

r-""

"1-'.,

(4-57b)

= V,

on Eqs, 4-58 and 4-59. Substitution from Eq. 4-60 into Eqs. 4-58 and
4-59 and solution for the current yields

Shunt-connected d-c machine.

:.

1
a

(1 _
R"
VI

-''''
w m G4

RI

(4-62)

Substitution of these currents into the torque equation gives

}
~'

Note that the electromagnetic torque is positive for positive currents,

which is simply a mani fcstation of the fact that the rotor a-magnetic
field tries to move with a positive ,p to line up with the stator j~magnetic
field.
By considering the energy conversion properties of a d-c machine, it is
possible to work with only steady-state solutions of Eqs. 4-57a and b.
Thus, all voltages and currents will be constants which will be designated

T~

= Ga/V;
RaR f

(1 _

wmGaf )
Rf

(4-63)

Note that the torque varies as the square of the terminal voltage VI and
linearly with speed w m Sketches of the torque-speed curves for two values
of VI are shown in Fig. 4-6. Note that the slope of the torque-speed

.,

J.oJ.

t:LI:C I KUMt:CHANICAL ENERGY CONVERSION

>~:'t

curve varies inversely with the rotor (armature) resistance Ra , and if R a

is made very ~mall the shunt-connected doc machine can be made to have
nearly a constant speed characteristic over a limited range of torque.

Substitution of these constraints into Eqs. 4-58 and 4-59 and solution for
the terminal current yields
I _
V,
, - wmG af + Ra + Rf

,.
'"

Electromagnetic
torque

283

TWO-PHASE TRANSFORMATIONS

(4-65)

Substitution of this current into the torque equation yields

Te

Te = (
Slope

=_G,~,V,2
. [{JR.

G
Wm

af

Gaf~;
+ a+

R )2

(4-66)

The torque-speed curve is sketched from Eq. 4-66 in Fig. 4-8, One of
the practical advantages of the series-connected machine is its high starting torque which is obtained by making the resistances R a and Rf small.
~1

Electromagnetic
torque. T.

Brake

I"

i<

Motor

G., vf

Genehtor

(R,

Fig. 4-6. Torque-speed curves of shunt-connected d-c machine.

+ R.J i

.iJ
li;~',

Next, consider a series-connected doc machine. This name comes from

the connection of the stator (field) circuit in series with the rotor (armature)
circuit as shown in Fig. 4-7. For this connection the ~onstraints are ,J
If = la =

1,1

and

."
V,

'--i.}

+0

= Vf

\""1,

I,
--

+
V,

,.

+ Ya
,0"

Mechanical
speed.

01
,~i'

ttt

(4-64)

,1

'~~

Wm

Fig. 4-8. Torque-speed curve of series-connected d-c machine.

11:'

Next consider the energy balance of a doc machine. There are two
electrical terminal pairs and one mechanical terminal pair; and, since
energy is converted, it is of interest to look at the power /low among the
'I',
\,'i
three terminal pairs. Using the steady-state equations and Fig. 4-9,
j'
,:~?' the power flow can be established. In Fig. 4-9 the power is shown as
'~
input power to be consistent with earlier treatments,
;;1~::
First, look at the field input power, which is
.t .:trt1

11 \

~.

-In

v.., . . "
'~H

.1,0',\"

"

, V}

Pf = Vllf = -

Rf

_'

l'

" ,".

f~ , . I"

(4-67)

This is simply the ijRf loss in the stator winding'; consequently, there is
no electromechanical energy conversion between the field circuit and the

Fig. 4-7. Series-connected d-c machine. '

ir,

,:.{

',~
'{

.(~<

.,

.fl.

"-*11-:
284

ELECTROMECHANICAL ENERGY CONVERSION

mechanical system. That is to say, none of the electrical energy put into
the lIeld is converted to mechanical form, and none of the mechanical
power input can be converted to electrical form ill 'hI' field (stator).
Next, consider the armature input power. From Eq. 4-59 this power is

Pu

VJu= w",Guflf1a

I~Ra

Assuming the
coil currents,

Fig. 4-9.

Mechanical power

Power flow in a doc machine.

mechanical power input. The torque T e given by Eq. 4-57b is the torque
applied to the mechanical system by the electrical system, and the mechanical illput power is given by ..
~l4t

Pm

."

-Tew m = -wmGuflfla

'

IC~s~
tsm~

sinl
cosiJ

iJ

Two-Axis Four-Brush D-C Commutator Machine

v: =
V$

<":;0'

(4-69)

which is the negative of the first term on the right of Eq. 4-68. A com
parison of Eqs. 4-67, 4-68, and 4-69 shows that energy conversion takes
place between the armature (rotor) electric circuit and the mechanical
system, and the amount of energy converted depends on the field (stator)
current.
It has been shown in this section that a doc machine will successfully
cunvert energy with direct current flowing in the field (stator) and armature
(rotor) brush circuits. In Sec. 3.5.2, Eq. 3-134, it was found that for
rotor and stator coil currents to interact their frequencies must differ by,
or sum to, the mechanical speed W m (for a two-pole machine n = I). In
the doc machine the stator current is direct current; thus th, rotor coil
currents must have the frequency W m when the machine has two poles
(n = I). The relation between brush and coil currents for such a machine
is the af3-dq transformation for n = I:

(4-71)

The conventional two-axis four-brush doc commutator machine is

described by the general equations of motion, Eqs. 4-56a and b. The
dynamics of the four-brush machine will be treated in a later chapter,
but the energy conversion possibilities can be considered by a study of
simple steady-state operation. The doc steady-state equations of motion
are obtained from Eqs. 4-56a and b by setting p = 0 and using capital
letters to indicate steady-state values, which yields

Pm

I'

... la cos wmt'

-10 sin wmt

The coil currents have the frequency W m , and the frequency condition for
interaction of currents to produce torque is satisfied in the doc machine,
the frequency conversion between direct current at the brushes and the
frequency W m in the coils being produced by the commutator.
4.3.~

D-c machine

"'" I", i~ "-' 0, and

(4-68)

r
)I

= w"" i~ "" i a

/ld choice of axes gives q,

i~

Field or stator powel

Po

285

i; =

The term I~Ru is lost as heat in the armature copper, but the other term
on the right is converted to mechanical form. To check this, evaluate the

Armature or rotor power

TWO-PHASE TR.ANSFORMATIONS

(4-70)

R~/:

= R~l;

rl
rr l' + R'l'
Vdr = 'Pletil3fJ
IS + 'PlGdqq
tid
rr l' + R'T'
V'q = - 'PJG's
IS
_
JG
q....
'P qd d
"q

L
,-:,4;~"

(4-72)

T= a~ - T.
T. = G'.#Jl:l; - G;~l:J;

'lA,

.:~
I~

,,-...

(G; -

G;;d)/dI;

(4-73)

"r:' ',;

The two stator volt-ampere equations do not contribute to energy conversion because any steady-state stator power is dissipated in heat in the
stator windings. Energy conversion must take place between the power
delivered to the rotor and the mechanical shaft. To show that this is so,
write the total power into the rotor as
PrOlor =
'J;

.,~ ','

rJ

V~l~

+ V;l;

(4-74)

~,

and from Eqs. 4-72 and 4-73 this becomes

~.~

PrOlor

= Vdld +

v;t; = R~Ud)2 + R;U;,'P

q,[G~/;I; - G;::/;I:

+ (C~

- G;'~)ldl;]

= R~(Jd)2 + R;O;J2 + q,Te (4-75)

where T. is given by Eq. 4-73 and all currents have constant d-c values.

286

,.-l'

ELECTROMECHANICAL ENERGY CONVERSION

Equation 4-75 is an expression of the power balance in the rotor, and can
be interpreted physically by distinguishing three cases:
1. The torque T, = O. This case corresponds normally to the reversal
of power flow at one of the two rotor axes, and means that all of the
electrical power input is either consumed as losses in the armature or
delivered to an electrical load. Zero torque is often known as defining
transformer action, the connotation with direct current being that power
exhibiting a given current-voltage characteristic is transferred from one
axis to the other with a different current-voltage characteristic.
2. The torque T, > O. If all the electrical input power, apart from
armature losses, is transformed into mechanical form, one speaks of pure
motor action. When part of the electrical power supplied at one rotor
axis is also transferred electrically to a load across the other axis, then there
is a combined motor and transformer action.
3. The torque T, < O. In this case generator action results which
transforms power from mechanical to electrical form. This may, or may
not, be accompanied by transformer action, depending upon whether
electrical power is, or is not, being furnished to the rotor.
.
These various processes of steady-state power transfer are depicted in
Fig. 4-10.
The transformer action that occurs in the two-axis doc machine is the
result of rotation, but it does not involve energy conversion. To examine
further the transformer action, consider a smooth-air-gap, two-axis,
four-brush doc machine with the stator unexcited. The' d axis of the
rotor is excited with a voltage source Vd , and a resistive load of R L ohms
is connected across the. q winding terminals. Then for steady-state
operation, Eqs. 4-72 and .4-73 can be used by setting

Mechanical power

~Ii;~i
....'.'..

Vd
V'q

Vd
,,*-'!J \).;'-'I"l" itl" .~\ ~<'l'"'f'
I'R'
,
- q L "n'
"
\i:.ftll I' ''f~fI

which gives the equations of m o t i o n !

,
Vd = ~Grr I;' +.k~Id
- RLI; = - ~GrrId + R;I;
T, = 0

.'
i

7- - 0

"

r;/

~.

-.

~)

:.~,

rd

(~Grr)2 + R'

R'q

RL

11';

T e >0

:'1

..

tI

Electrical power

~z::::~./

Mechanical power

.....

_i .
u

--

. ' ,

Electrical power

/1

T e >0

~--"

,
I

-, ,$'\\'c"1'

i -.'

(2a) Pure motor action

(2b) Motor and transformer action

Ii
'~',.i

Mechanical power

Electrical power

Mechanical power

::.;..:.

i( .

Electrical power

".,

T,<O

-~t(.
:'1.1' ,Jti1 ':.

'H

'W,t:j,;J..;

rf~l"

>

I
I

-"
,

(4-76)

I
I

~".
~'! ~

it:!fI"

- .... "

.-",,'
\

T,

(3a) Pure generator action

Fig. 4-10.

<0

LJ..=:j--""
"

I
\I

~-

(3b) Generator and transformer action

Power transfer in the four-brush doc machine.

,.1

The power dissipated in this apparent resistance in the d axis is just the
power delivered from the d axis to the q axis and consumed in the com.
bined resistance of R~ + R L That is,

(4-77)

Solving Eqs. 4-77, the apparent resistance of the d axis is

Vd =

Mechanical power

'd

,L;-,

Lf--1

(1) Pure transformer action

,~In

l'~i

Electrical power

Armature losses

:~':'~~

l~ =

I; = 0
Grr
==
G"
- G"
d~
qd-

287

TWO-PHASE TRANSFORMATIONS

jN~"

(4-78)

:'.::

U;)2(R; + R

'.

L)

The term (~Grr)2/(R; + R L ) is the apparent resistance reflected from the

q axis into the d axis by a pseudo-transformer action as a result of rotation.

.. :
;,,;

,~,

1~

.~<.~;

\, 1.1'>

,;!'

(A~'~J2(R; + Rd
(~Grr)2 (1')2

R'q

RI.

~<

:
(4-79)

-J

288

ELECTROMECHANICAL ENERGY CONVERSION

TWO-PHASE TRANSFORMATIONS

magnetic axes us the stator windings. In practical cases. however. some

of these windings may not exist; furthermore. additional s~ator windings
may be introduced along each axis to produce specific characteristics.
The equations of motion are still of the form given by Eq~. 4-56a and h.
However, in expanded form, they may necessitate the additi,on or removal
of the respective rows and columns in those cases where other windings
are present or absent. Specific constraints may be imf:losed upon the
currents and voltages by the particular interconnections of all these
stator and rotor windings.

Equation 4-79 shows that electrical power is transformed from the d axis
to the q axis. but no energy is converted from electrical to mechanical
form. The ability of a two-axis doc machine to act as a doc transformer
and change the current level between the two systems can be seen by
solving for both I~ and I;:

I'd -

I'
q

V,iR~

(4>G")2

+ RL )

R~(R~

VA~Grr)

=
(4>G")2

R'd(R;

(4-80)

RJ]

RL )]

(4-81)
4.3.5

For practical machines the rotor resistances R~ and R; are quite sman.
and it is easy to drive the rotor at a speed 4, such that
(~G")2 ~ R~(R;

RL )

(4-82)

For the conditions which satisfy Eq. 4-82 the quadrature axis appears to
be a current source of magnitude
1'''''''' Vd
q = 4>G"

f"

-~:

289

(4-83)

The quadrature-axis current is nearly independent of the load on the q

aXIs. The doc transformer action of a two-axis commutator machine
converts a voltage source applied to one axis to a current source in the
other axis. Similarly, a current source could be changed into an
approximate voltage source. This connection of doc two-axis machines
effectively produces d-c source transformations.
Three features of electromechanical energy conversion are readily
observed with doc commutator machines-namely, transformability,
reversibility. and interchangeability of the forms of energy. The control
of these energies is dependent upon the changes brought about in the
electrical excitation levels which produce specific sets of stator or rotor
currents in a given machine configuration (interconnection of windings,
supply sources, and loads).
When going from one steady-state condition to another, via the transient
state. the process of storing energy in the electromagnetic field comes into
play in what may be termed the transient modulation process, These
aspects will be discussed in detail when more specific configurations of
machines are being dealt with. In this respect. it must be remarked that
the Cour-hrush commutator machine should be considered as essentially
constituting a generic configuration for various types of commutator
machines. As shown, it is composed of two equal stator windings in
space quadrature and two equivalent rotor windings having the same

.";,'

,i
~

,:,'

The Synchronous Machine

'0 l

The previous examples of the use of dq component~have been for

commutator machines. The equations of motion QJ commutator
machines must be expressed in terms of dq components si~ce the terminal
variables are brush currents and voltages which were shown to be physically.
equivalent to the dq variables. Any constraints placed on the terminals
of a commutator machine are therefore constraints on the dq variables
and not constraints directly applied to the rotor coil variables.
The use of dq variables for the study of noncommutator machines has
also proved fruitful. In this case, the dq variables are merely new nonphysical variables which aid in the analytic treatment of a given problem.
All physical constraints must be placed directly on the coil variables and
not on the dq variables. Any such constraints can, of course, always be
transformed to equivalent constraints in terms of dq variables, but th~
must be recognized as a mathematical fiction yielding convenient analytic
results.
The noncommutator machines which are most profitably studied in terms
of dq variables are the polyphase synchronous machines. The simplifications resulting from the use of dq variables are twofold: (1) The balanced
synchronous machine is reduced to an equivalent doc machine. (2) The
d axis and q axis are located along the two axes of magnetic symmetry,
and thus lead to inductances which have significant physical meaning,
namely, inductances along an axis of minimum reluctance (d axis) and
inductances along an axis of maximum reluctance (q axis).
In previous discussions in Chap. 3, a synchronous machine was defined
as one in which an average torque is produced at one speed when singlefrequency polyphase sources are applied to rotor and stator. Conventionally, synchronous machines have one member (usually the rotor)
excited by direct current and the other member (usually the stator)
excited by polyphase alternating current. Because the model of the
generalized machine has, saliency on the stator,
it is necessary to apply
.

"

..;

"~

--I
j

'

291

ELECTROMECHANICAL ENERGY CONVERSION

TWO-PHASE TRANSFORMATIONS

direct current to the stator and balanced two-phase alternating current to

the rotor to obtain conventional results.
The equations of motion which describe the generalized machine with
a{3-dq variables arc given by Eqs. 4-50 and 4-51. In this synchronous
machine example (as in practical synchronous machines) only one axis of

associated with the f3 winding yield the following equations of motion

in terms of dq variables from Eqs. 4-51a and b:

290

j
I"

~Vd
vq

+pL,
pL/
d
-nL/d

OJ]

pL/d
R, + pLd

nLq

i"

-nL"

R, + pLq

iq

(4-840)

'

+ Ct. +

K4> - n[ -L/di/iq - (Ld - Lq)i"iq] (4-84b)

where all superscripts are considered superfluous and are dropped from
the voltages, currents, and parameters, and the following new symbols
are defined:
R/ = R'
L d = L~ + L~l
(4-84c)
L q = L~ - L~2
L/ = L~ + L~l
R, = R'
L/d = L:'; + L:Z

.f

'{

n~~

.'+'

Stator
magnetic
~-.......+ {3-axis
Rotor ,
magnetic
q-axis

,f

.,

.....

d;:k
; illH-

.>,
1

I'
, ,

',"

',~.,
~'l;,1'

'r"

Fig. 4-11. Simplified' schematic of synchronous machine.

n

= J~

(Machine drawn fot,

1.)

the stator will be excited with direct current. This eliminates one stator
axis from the equations of motion and one stator coil from the machine
drawn in the simplified diagram, Fig. 4-11, with concentrated coils. The
reduced equations of motion for the synchronous machine can be
obtained from the general Ct.{3-dq equations of motion given by Eqs. 4-5la
and b. To treat a salient-pole synchronous machine the IX winding on the
stator must be kept for the doc field winding. Eliminating the f3 winding
by setting vp = i p = 0 and also eliminating all self- and mutual parameters

."

,;',.

The equations of motion in terms of dq variables given by Eqs. 4-84a and b,

are completely general for any two-phase synchronous machine. In Fig.
4-12 the VOlt-ampere equations are expressed in the form of an equivalent
circuit which clearly shows a transformer coupling between the f and d
axes and speed couplings between f and q, d and g, as well as q and d.
Thefg speed voltage is the generated voltage produced by the field current
and mechanical speed.
As an example of the use of dq variables, consider the problem of
evaluating the principal performance characteristics of a synchronous
machine with unbalanced polyphase excitation. To determine the
essential character of the necessary excitation, refer to Fig. 4-11 and note
that a direct current applied to the stator winding (i~ = if) produces a
stationary stator field in both time and space. Similarly, note that
balanced two-phase currents of frequency w applied to the rotor coils
produce a rotor field which rotates at a speed win with respect to the
rotor. Thus if the rotor is turning with speed win in the opposite
direction to the rotor field, the rotor field will be stationary with respect
to the stator field and an average torque will result. For the sake of
!iefiniteness, assume that an average torque is produced when the rotor is
turning in the positive direction with a speed w electrical space radians
per second. In this case the rotor field must rotate in the negative 4>
direction with respect to the rotor; consequently, the phase sequence of
the rotor currents (or voltages) must be fJ to Ct.. In other words, the rotor
phase voltages (not the dq voltages) must be constrained to be

,',
'I

'.. . 1

,
'}i,"

V~

= Va sin wI

v;

Vpcos

wI

(4-85)
1",:

f'

(4-86)

';~

-=:=J
+

Rf

R,

Equations 4-85 through 4-90 establish the conditions of operation of

the synchronous machine. These operating conditions or constraints
are expressed in terms of the coil variables (af3) since these are the true
physical variables of a synchronous machine. To use the equations or
motion derived in terms of dq variables in Egs. 4-840 and h, these
constraints must be expressed in terms of dq variables. The rlq transformation was defined in Sec. 4.3 as

"

!.

~LU

\f\..

v;

lid

~os
. n~
sm n~

- sin

n~ ~~

cos n~

(4-91)

v~

Using this transformation to change the af3 voltages of Eqs. 4-85 and 4-86
to dq voltages, recognizing that n~ = nWml + no = wI + no, and dropping
for simplicity the superscript r from the symbols v~ and v; yield

;>""

R,

293

TWO-PHASE TRANSFORMATIONS

ELECTROMECHANICAL ENERGY CONVERSION

292

Lq

Vd = - (Va; Va) sin nS _ (Va; Va) sin (2wt

+~

vq = ( V+V)
a 2
a cos no + (V-V)
a 2 a cos (2wt

no)

+ no)

(4-92)
(4-93)

IIq
~

.~

.1;

~f.

nv,
"'l

~.'

,1,,-

~:
Fig.4-I2. The general dynamic dq equivalent
circuit of a two-phase synchronous machine.

:i

,,,;.Sit

The excitation for the stator or field must also be established. To

simplify the initial discussion of the unbalanced problem, assume that the
field is excited by a current source, i.e.,

:.'

'a = 'I = - I I
'$'

,fj

?'

= constant

(4-87)

and of course from previous discussion the f3 coil on the stator has been
removed, which requires that

i!J

(4-88)

The final constraint which must be considered is the speed. Since this
example will be limited to steady-state performance, the speed can be
constrained to be a constant and equal in electrical radians per second to
the excitation frequency w, i.e.,

~
and

= w",

~ =

wml

=
+

~
0

,.J.},

(vq)s

(IM~)

(4-89)
(4-90)

For balanced excitation (i.e., Va = Va) the dq variables are constants,

i.e., equivalent direct voltages. The effect of the unbalance in the phase
voltages is to introduce a double-frequency component of voltage
proportional to .1< Vjl - Va) into the d and q excitations. The direct voltage
and. double-frequency voltage in Eqs. 4-92 and 4-93 can be separated
into one term proportional to the sum (V~ + .Va), which is a direct voltage,
and a second term proportional to the difference (Va - Va), which is a
sinusoidally time-varying voltage of frequency 2w. With the speed held
constant, the volt-ampere equations are linear with constant coefficients,
which means that superposition holds. The unbalanced steady-state
problem can be treated in two steps: (1) the currents and torques due to
the sum components which are the equivalent of a balanced excitation
can be found; and (2) -the effects of the difference component-s which are
alternating currents of twice frequency can be found. The following
quantities in which the subscript S means sum and D means difference
are defined.
(4-94)
(Vd)S = --!(Va + Va) sin nS

t~;e

.li<'

. ,r'

: ~,

= t(Vo + V,a) cos n8

(Vd\~' =:

(vq)jj

(4-95)

-tWa - Va) sin (2wl + no)

= .,( V,a

- Va) cos (2wI

+ no)

(4-96)
(4-97)

294

",
I:(
,

ELECTROMECHANICAL ENERGY CONVERSION

~,."

As a first step, the steady-state doc problem resulting from the

excitations (Vd)s and (vq)s will be solved. Using the constraints given by
Eqs. 4-87 through 4-89 and the voltages defined by Eqs. ,4-94 and 4-95,
and remembering that for steady-state conditions with only direct voltages
p = 0, the equations of motion given by Eqs. 4-84a and b reduce to

--tev" +

V/l) sin n8

= Rr(id)s + wLq(iq)s

V/l) cos n8

t(V..
=

a~
n

1
'i~
"

, 't

+ Rr(iq)s

: t
"" '1

(4-99)
(4-100)

n[LrdIr(iq)s - (Ld - Lq)(id)s(iq)sJ

.~

,:
'

",'

~~

(4-98)

wLrdIr - wLd(id)s

TWO-PHASE TRANSFORMATIONS

..',

The currents given by Eqs. 4-101 and 4-102 are constant or direct currents;
therefore, substituting them into Eq. 4-100 will establish the instantaneous
torque as a constant, i.e., the average torque equals the instantaneous
torque. The direet ~ubstitution of Eqs. 4-101 and 4-102 into the torque
equation will produce a rather complicated expres"ion whieh is diflicult
to interpret. Approximate expressions for the currents and the torque.
whieh are quite useful for most synchronous machines, can be obtained
by recognizing that

-;/

R,
in most practical machines.
approximation by

The equivalent circuit of the two volt-ampere relationships expressed by

Eqs. 4-98 and 4-99, which is drawn in Fig. 4-13, is composed of two

(id)s "-'

Rr

---

wLq(iq)s

A
:~

'0

Constraints:

Rr

nip-w

t)'JI!U!'
:t"

)'',i'

:).

Fig .... 13. The steady-state dq equivalent circuit of a

two-phase synchronous machine, for balanced excitation or sum c o m p o n e n t s . '

>.,

:<,

't:~
"

resistive circuits coupled by dependent voltage sources. The simultaneous

solution of Eqs. 4-98 and 4-99 for (id)s and (iq)s yields

ld s

Vp)R r sin n8 - -t(V" + V/l)wLq cos n8

(R,F + w2LdLq
,
l,

(i) =
qs .

teva +

..

,.
i~.,'

+ wZLqLrdI(
(4-101)

,,

~~~'

V/l)wLd sin n8 + t(Va + V/3)R r cos n8 - wLfdRrlJ

(R,F + wZLdLq
, .::
(4-102)

"""""
'~$jll';.'~~.'
~~~:;; .'
,."

(4-103)

VII) sin
wL q

) LldI
v/3 wL I
d

.
SIO

no - g(V,.

n8

(4-104)

V/l)

z(Ld - L q) .
. ,71 1 SIO 2n8

(4-105)

When the rotor (armature) supply voltage amplitudes Va and V/3 and the
stator (fit.J) current If are fixed, a change in torque is accompanied by a
change in torque angle along the sinusoidal curve shown in Fig. 4-14 with
",~~t '1
t~'~

(T~/lo)m.x

(n/2)eva + V/l)IfL fd
wLd

(4-107)

Since the rotor is turning in the positive </> direction. a positive torque
indicates mechanical power output and therefore motor action. Conversely, a negative torque indicates mechanical power input and hence
generator action.
~~""

~J

c2 no + wLJdIJ

1"i;"

(') = -t(V"

V/l)

This expression gives the instantaneous torque (a constant and therefore

equal to the average torque) which includes the effects of saliency but
neglects the rotor resistance.
To study some of the steady-state charactcristics of thc synchronous
machine with balanced excitation, consider a machine without saliency
(Ld = Lq = L~ and L~z = L::; = 0) in which case the torque (Eq. 4-105)
reduces to
T~/lo = - (n/2)( V" +w~;)IJLrd sin n8
(4-106)

'1l,';

';,.

(lI q )s

T.~ -leVa

ift

if'" -If

'\1\1\/'

-tWa +

These two currents substituted into the torque expression yield

.

'."':

~ :

The currents are thcn given to a vcry close

"-' - t(Va
(I.)
q S =

(i d )1I

(lId)1l

WL q
wL d
{ wL
Jd

NVv

~~,:,

I,

295

,11:.

i;

~:

296

'

ELECTROMECHANICAL ENERGY CONVERSION

TWO-PHASE TRANSFORMATIONS

As indicated by hg. 4-14, the machine can supply torque only up to

(T.,Jmil" If a load torqUl: greater than (T,.Jm~. is applied to the shaft,
the machine will break out of synchronism, rloln at a speed other than
synchronous, and produce no average torque.
,
In the region of very small torque angles (where sin n& '" n&) the
torque-angk curve is linear, and the machine exhibits the characteristic
of a linear torsion spring. That is, a torque is produced which tends to
return the torque angle to zero, and the torque is linear with angle. The
slope of the torque-angle curve at 0 = 0 can be considered as a measure
of tht: .. sti tfness" of the synchronous machine. For a smooth rotor
machine the torque-angle curve is always a sinusoid, and the "stiffness"
is purely a function of (T.,J01ax'

:{

for these difference components can be obtained from EllS. 4-X4a and 11
using the constraints defined by Eqs. 4-87 through 4-90. In writing the
volt-ampere equations only those terms which are "not doc terms must be
retained. The direct voltage n~L/dlr does not contribute to the voltampere equation for the difference componentS: The difference voltampere equations ar~

-HVjl -

Va) sin (2wt

t(V/l - Va) cos (2wt

Motor

+ nil) = (R r + pLtI)(ld)D + wLq(iq)D

+ no) = (R r + pLq)(iq)D - wLAid)D

I"

action

Generator action

Electromag',lelic torque

k"""- Motor action-+Generator aclion-l

Electromagnetic torque

~I(T'/J)m..
I

_~
-90
0

I
-180

~O

"I

.;;

I'

,1

l~O Torque angle tlO

I
-<'

-180'\

7_Qno

,,~

_nn

)'Torque angle 116

TMO...

180'

-(T'/J)m..

Fig. 4-14.

Output torque as a function of torque angle.

,..

tic"

Next consider a machine with saliency, and write the torque from Eq.
4-105 in the form

T.

-(T.,J01ax sin no - (T.,.z)01ax sin 2no

(4-108).

If Va' Vjl, and Jr are fixed, then (1'.,.)010. and (1"/J)01a. are constants.

The

factors in Eq. 4-108 are plotted and combined in Fig. 4-15. 80te that.
the second harmonic term (T'I'z)01a. sin 2n& which is caused by saliency'
has the effect of increasing the slope of the torque-angle curve at the origin.
Con seq llently, saliency can be used to improve the" stitfness" of a synchronous machine.
The previous treatment of the synchronous machine in terms of dq
components gives the currents and torque for balanced excitation or for
the sum components of an unbalanced excitation. To complete the
unhalanced problem the currents due to the difference components
defll1ed by Eqs. 4-96 and 4-97 must be found. The volt-ampere equations

1.

4~.ii ~

.:.:.

1:"

w..-

,;

(4-110)

1'.

I
I

(4-109)

.' A

I
I

297

,P.i'

Fig. 4-1 S.
machine.

Illustrating how saliency affects the torque-angle curve of a synchronous

The equivalent circuit for Eqs. 4-109 and 4-110 is drawn in Fig. 4-16 and
is merely two R-L circuits coupled by dependent speed voltages. The
exact expressions for (id)D and (iq)D can be obtained by solVing Eqs. 4-109
and 4-110 simultaneously; however, considerable simplification can be
effected by recognizing that the rotor resistance is small, such that
R r ~ 2wLd and 2wLq Neglecting the resistance R" the two currents
are easily shown to be
.
(V - V)
(1t1)D '"
~WLd u. cos (2wt

no)

(4-111)

- Va) . (2
( lq. )D ~ (Vjl2wL
sm wt

+ no~)

(4-112)

..

..

298

~t,

ELECTROMECHANICAL .ENERGY CONVERSION

-,

Rr

Ld

--

I,

i,,;:'
I:~"

~!:
.1;;.,\,\:~

Lq

Rr

.bnstraints:
if = -I,
nit>=w

+
(Uq)D

,t
Flg.4-16. The steady-state dq equivalent circuit of a

two-phase synchronous machine with constant field

current for difference components due to unbalanced

excitation.
These two currents of frequency 2w can now be combined with the
direct currents of Eqs. 4-103 and 4-104 to give the total d and q currents
(assuming R r negligible) as

id = (id)s

(id)D
f(Vcr

'" wLfdlf -

VII) cos no

=
iq =

wLd

(iq)s + (iq)o
,...., -f(V. + Vp) sin no

+ -t(Vp -

f(V/J - Vcr) cos (2wt

+ nS)

(4-113)

.,/

V.) sin (2wt

+ no)

(4-114)

wLq

4-84b.

~ Vcr ~

Vp) [ ::~f sin nS

1 '". ~

+ (Ld - 4~lrLq+

Vp) sin 2no

I

'Lldl

- i(VIJ

.,.

'~

- ,-

- Vcr) ~f sin (2wt + nS)

Ld
n
(Ld - L)
-4( V~ - V;)
2L L q sin (2wt + 2nS)

+ iVp

. -,
0'''1 A;~,

'i\wdq

V..)2 (L~2-;'
d

It is immediately evident by comparing Eq. 4-115 with 4-105 that the.

difference components which result from unbalanced excitatIOn do not
. change the average torque. hence do not afTect the ability of the
synchronous machine to convert a given average power. The unbalanced
excitation docs. however, re5ult in time-varying torques of twice and four
times line frequency. One of the major advantages of a balanced polyphase machine, which is to supply or demand a nontime-varying torque.
is destroyed by unbalanced excitation. If the unbalances are large
enough, the time-varying torque may, depending upon the dynamic
characteristics of the load, cause a loss of synchronous operation of the
machine.
Before leaving this example it should be pointed out that for simplicity
the field was constrained with a doc source. This constraint for balanced
steady-state operation does not affect the performance of the machine.
For unbalanced operation, however. the field is coupled to the direct
axis (see Fig. 4-12) even in steady-state operation. If the field is excited
with a constant-voltage source rather than with a current source, the
field current will contain current of the same frequencies as the direct axis
currents. This sets off a chain of effects which results in all possible
even harmonics appearing in the currents if' id , and iq . The simple result
obtained by imposing a current constraint on the field is only a partial
solution in a practical synchronous machine with voltage excitation on the
field. The sequence of even harmonics in the currents if. i d and i q ,
produced by unbalanced excitation, results in the two-phase currents i a
and ilJ being composed of all odd harmonics, whereas the field current
consists of direct current and all even harmonics. That the a{3 currents
are
odd harmonics if the dq are even harmonics results because the
1if .
transformation from dq to a{3 changes the frequency of each harmonic
~} by w.
;r
The previous examples of the synchronous machine with unbalanced
.~

;:~
excitation show clearly how a change of variables, in this case from a{3
to dq variables, can lead to simplification of the equations of motion.
In Chap. 3 it was quite difficult to treat the synchronous machine with
balanced l'xcitation and a smooth air gap. By introducing dq variables,
the salient-pole machine becomes quite simple, and even unbalanced
excitations are relatively easy to handle. Furthermore. this change to
dq variables -greatly simplifies the problem of machine dynamics. The
general dynamic problem of synchronous machines will, however, be
delayed until later and treated in Chap. 9.
.' The use of dq variables to treat doc commutator machines is of course
essential. Their use for the study of synchronous machines leads to
considerable simplification; they can also be employed to study other

.!.

.'j

These two expressions for the currents which contain direct and doublefrequency components can be used to find the instantaneous torque by
substituting them into Eq.
The total instantaneous torque is

T. - -

299

ri., .

(id)D

(Ud)n

TWO-PHASE TRANSFORMATIONS

fq) sin (4wt + 2nS)

.)'.-

. ,'1'

q. :

"
'J"

{ I~

,Jt.~'i;"

4:

"

(4-115)

'~

, ~ "'~

jll".

,:.o.
~.,J,&c-~{'
o~J~ii

';

L
300
;

ELECTROMECHANICAL ENER.GY CONVERSION

TWOPHASE TRANSFORMATIONS,

classes of machines, e.g., the induction machine. In the study of induction

machines the dq variables yield equations of motion in a form which is
classically called the cross-field theory of induction machines, but since
another example utilizing dq variables could contribute little in new
concepts, it is not necessary to treat the induction machine using the
cross-field theory, Instead. it is more advantageous to proceed to new
sets of variables which are also serviceable in the study of machines.

The transformation of Eq. 4-117, when written out for the v~liages,
appears as

4.4

I
I

vI'

~
,r

','

and

IJ~

--

~
Iy

j~

(4-116)

[a~6J ~

(4-117)

where the transformation matrix [a~aJ is given by

r(J~,~]

- ~J)
n(eP - ifJ)

C?S n(eP

= [
-

Sin

~
with

[(J~,~l-l

\
=

fc'"osn(eP -.fJ)
I_~n n(c/> -- ~J)

- sin n(c/> cos n( c/> -

(4-119)

,;j;)l

ill

f3- ax

iS!....,

. !.. i

I'

"

~fJ

r',
",~

IiJ

i.'

: t,~,

/'/

//

l~
I

(4-120)

A comparison of Eqs. 4-118 and 4-120 with Eqs. 4-36 and 4-38 shows that
when the angle ifJ is zero. the dq transformation results.

~.

''j

.$

[(J~a)-I~

'-...

1;,+
";f
C.~ '

/'/'/
,

.,//'

I
Fig. 4-17.

/'

I'
,

/'f

(4-118)

la-aXiS

"(0

'. "~,.

The inverse of Eq. 4-117 is

"dSil/~

.~

sin n(eP - ~J)]

cos lIe eP - ~J)

(4-121)

Fixed
axis

These new variables are defined in terms of the a{3 variables as

x afJ I =

.fJ) + v6 sin n(,p - I/J)

.fJ)' + Va cos n(,p - .fJ)'

sin n( -

a-axis

= vy cos n(c/> -

== -vy

The use of these equations with Fig. 4-17 establishes a geometrical

interpretation of the yo transformation. The angles c/> and .fJ are con-

The Generalized Rotating Real Transformation (y5)

The yS transformation is a rotating reul transformation and is a

generalization of the dq transformation given in the preceding section.
The equations of motion in terms of yO variables on the rotor and afJ
variables on the stator are useful in the analysis of repulsion motors and
other commutator machines in which the brush axes are not aligned with
the stator axes. The brushes can be either fixed to the stator reference
frame, as in a repulsion motor. or fixed to an arbitrary reference frame
which rotates with respect to both stator and rotor.
The yO variables are written as

f'

Va
!

, ,301

J"

.::l'

",. c

Os "{ '(to)

....... ,........

.........' ............

,..(.

Geometrical interpretation of the ,.8 transformation.

sidered to be measured from the same fixed reference. When both angles
are functions of time, Fig. 4-17 shows two coordinate systems (a{3 and yo)
in relative motion with respect to a fixed set of axes. The fixed axes are
the a{3 axes of the stator; the a{3 axes shown in the figure are those of the
rotor and they describe true rotor coil variables. The yO axes are the
positions of the brushes. With such an interpretation the afJ-y'8 transformation describes the physical transformation which is performed by a
commutator. If the brushes drawn for the dq transformation in Fig. 4-3

302

ELECTROMECHANICAL ENERGY CONVERSION

TWO-PHASE TRANSFORMATIONS

are rotated to some arbitrary angle ljJ with respect to the dq axes, then the

The transformed volt-ampere equations in Eq. 4-128a arc of the general

form of those discussed in Sec. 4.2.1. The volt-ampere equations contain
the time derivative fI of the product Y>,0/,,' and since ..#~;;",~ is a function of
time, by virtue of its dependence on 1> and </1, a term proportional to
dsl:;;'ya/dt is obtained. From Eq. 4-26a and 4-30a the additional voltage
matrix due to the time dependence of .w;;.ya is of the form

yS transformation relates coil variables to brush variables.

Now consider how the yS transformation affects a set of balanced two-

phase voltages.
Va

= Vcos (wt

V sin (wt

v/J -

"I

,:}.',

{,"

....

y)

+ 1')

= V cos ( wt

+ I'

1T)

(4-122)

2"

When the transformation of Eq. 4-119 is performed on these voltages,

there results
Vy

= V cos n(e/> - ljJ) cos (wt + 1') - V sin n(e/> - ljJ) sin (wt + 1')

Va =

V sin n(ep - .jJ) cos (wt

+ 1') + V cos n(e/> -

rfl) sin (wt + 1')

but sI",;i,ya

(4-124)

These two voltages still form a balanced two-phase system. When the
angular speeds of the two reference frames (see Fig. 4-17) are
W

.jJ= Wb

(4-125)

constant

'~~
.~

= constant

~"

Eqs. 4-124 become

Vy

Va

= V cos [(w
= V sin [(w

+ nWlll + nWm -

nWb)t + y]
nWb)t + 1']

,r,

(4-126)

[a~J

"

v''
~,ya

:c:

+ p.fl afJ,afJ)sI'"af3,y8 ]i'"ai3,y8 ,

+ Ke/> - , -te''
_ aIJ,y8, (.91'"
<rI3,ya, ,r','
aIJ,aIJ"'"fS.'
# aIJ.ya )e""
a/J,ya

[sI,r"-lcPls,,
afJ,ya afJ,aIJ

T = Jf>

1
a 'I'

,'

(4-128a)
(4-128b)

:~
!f's"

.: aIJ,ya -

,r.,r _

'!l.r"

AS,!"'

(l.W""

' I;:~ ,;'

",

::'

!./'"
"
L

"

it
:,~:

:f

~:.

(4-129b)

UJ-I

_0

'[U]-l

<rp.ya

Ul-. 1

a8.~

,:

" all

[a~81-1 [/" :;{a/l)

J~

l
0

,r,'aJl ffu]
afJ. ~_lll

[/" :fJ1-.J

La

OJ

: .;
Tl;J>

(4-131)

[a~)

oJ
fa~8]

0
[or" :;;'alJn
_Ill
Iful
[orI' ::'a/J)
a J 0

~/. ;':."J

[I"

[II' :;,apl

[II'

[I"

:8.a~J ro

[I. :'J

~ [a;j'~

''- ":';'10,0.

"

(4-133)

0 [ at/J]

......1

I ,'"

2a~

~81

,"
'/ ,'

(4-132) ..

J[f/" ~/lJ ~;'afJJ~

[a~]-l

l r[/" ~l

1a',.,J]

[II'

(4-130)

0-J ~[I ' ]

+ tU1-1

,tfOl

"I ,

- y . r

aIJ,ya -

81"r

ol ~u]
l
JL~r:8J [r:/l~
L [a~tl

[a~]-I

[a'~l-I

,~.,.j
.,,:.,

~U]-l

LO

aIJ.y4 -

(4-127)

and the equations of motion transform as

JS

Pl,.r
'aP.~

t6

[[Ul

(4-129a)!!.

aIJ;>'f

From Eq. 4-129b the yS volt-ampere equation should contain the usual
parameter matrices M:;'y~, !~;'y~. and :Y:;'y~ plus two additional matrices,
one proportional to ~ which can be combined with Y~;,ya to define a
matrix <~;'y~ and a second matrix proportional to ~I which will be defined
as ~~;,Y8' The matrix manipulation can be carried out in detail, but the
method was clearly established in Sec. 4.3.1 for the dq transformation so
a detailed repetition is not necessary. Furthermore, since the yS transformations are only used for a few special cases, considerable algebraic
detail can be eliminated by writing only the equations for a smooth-airgap machine. Considering only a smooth air gap and using Eqs. 4-128a, b
and 4-1 29b, define
"1

.f.

In this case the angular frequency has been changed from w

(w + nW lII - nWb) by the transformation, The change in frequency
n(w lll - Wb) is the relative angular speed in electrical radians per second
between the two coordinate ~ystems (aj3 and yS) between which the
transformation was performed.Y.,"
Next, consider how the yS transformation affects the equations of
motion of the two-phase machine treated earlier. It is necessary to
transform only rotor variables from af3 to yO components; hence the
transformation is

sI'"afJy8 -_
,

<

that

de:"!""

Va

~ =

d9lr.,
a!J,,,8 ) e',,r

aIJ.ap ~

ai3.y8

~ = ~~ + ~t/J
dl
0e/>
oif;

(4-123)

The use of trigonometric identities reduces these expressions to

vy = V cos (wt + ne/> - nif; + 1')
= V sin (wt + ne/> - nif; + 1')

= /(e/>, ljJ) so

..#"",-1 .!E",r

303

(4-134)

Ii,

304

ELECTROMECHANICAL ENERGY CONVERSION

TWO.PHASE TRANSFORMATIONS

In terms of these new variables the equations of motion are

ojj,...a = (ji""alJ...a

VI.,

T= N +

+2

alt....a P
5

.'

+ ':1ajj ...a q, + ffttajj ...a t/J)lS"ajj.)'d

1

"

.,

aq, + Kr/>-1i:';'~/!T:;;'~i~;'""

r")

(4-1350)

(4-135b)

Solving for each of the quantities defined by Eqs. 4-130 through 4-134,
using the tran~formation of Eq. 4-127 and the parameters defined by
Eqs. 4-19b, 4-19c, 4-20b, and 4-20c, the equations of motion are obtained
for the af3-y'jj variables of a smooth-air-gap machine as Eqs. 4-136a and

Vl

'--l

l:l,

~.!!!:~
::: -s.
l:::
c
l:::
~:I.

._

..eI '-'
8

4-136b.

~t""

~:I. ~:I.

Notice from the equations of motion that all dependence of coefficients

on the angle 4, has been removed, but a dependence on ~ still remains in
the voltage equations. Notice further that the yo transformation has
introduced a new dependence on the angle If into both voltage and torque
equations as well as a dependence on t/J in the voltage equations. Whenever the angle VI is constant (and as a result .p = 0), these equations of
motion have the same types and degrees of nonlinearities that were
discussed in the preceding section with dq components, and in fact it is
easily seen that the dq transformation is a special case of the yo transformation. For example, make the restriction <jJ == O. Then Eqs. 4- J 360
and 4-1 J6b rr.:d Ul;e to Eqs. 4-51 a and 4-51 b, respectively, for a smooth-airgap machine.

~,~
l::

~.::s.
l:::

c:

l:::

';;j

~~
l::

l::

:.,

8
....,

c:

Q.

Vl

~tJ

..-,

q,--s.

~:I.

.:s:q,, + '--l+

';::( _.. .-,'

f'

The Repulsion Machine

The repulsion machine is an a-c operated commutator machine normally

used as a motor. With reference to Fig. 4-18, the construction and
operation are as follows. The stator circuit's excited by an a-c source.
The brushes are short-circuited together, and the brush position is at
some arbitrary angle.p. Transformer (i1!J-oH /et) voltages are induced in
the rotor conductors, and currents flow through the short-circuited
brushes. The spatial position of the field due to the rotor currents is
determined by the brush position; thus the rotor fteld is not aligned with
the stator fteld, and the two fields interact to produce a torque.
With this brief description of the repulsion machine, equations to
describe the device can be determined. First, this is a commutator
machine with the brushes making an arbitrary angle ~I with respect to the
stator axis. The (tf3- y 'ii components must be used to describe the machine
in terms of its terminal variables. There is only one axis excited on the
stator and one on the rotor, and the brush position angle if) is fixed

!'"....'O

:"..,.tQ.
'-'

;~H,i'

~ ?< ~

s::

I',,"!/."
') ",~

~
~:.

""

'\

~~ '--l
~

C)

'\

---

~'O

':"_tf

-=$.-=$.
l::

l:::

l::

l::

~'"

:..~

...

~~

4.4.1

305

1
1:1

t1

K\.\\

q, ~ .S
",::t
C)
r.n
'--l ...., '-'

. Ii

... ~~I,,~IJ<;~

t, ~ R

+ +

'--l
l::

I AU

j\(,'

;::,...!!!:~

'iV'

...... -s.-s.
l::
l::
c:
._

0Vl

~~

(41:. j ./c

~=1 ~::t

':t

-=$.~
:0::
::::
~~

.... ;.

:::

c:

='+

,'t

Cl::

vo

l:::

.!!!:

~
:: Ec

'-

:'/

~I

~
~I

~:;)~ ~Ol.
~~'" ~ool
:;,;,
;;::;.

II

f-"

Vl

.8
~'-'
Q

~;l ..."'

""t

l~l.J

hi

~.....,~.

~.Ii;;:

I~ (

With these constraints the volt-ampere equations are, from Eq. 4-137a,

(~ = 0). Eq uations 4-136a and b with the stator (3 and rotor I) varia bles
eliminated will give the general equations for the repulsion machine as

"a

R' + L'"I'n
L"(
cos nl/1B
~.,
II
1'"
- (sin mp)n</> J
R' + L:p

=I

[!~ b.;,: [(cos /I~I)[1

T.1Io =

V' cos

-11

i~

u;i~j; si n nl/1

(4-1 37a)

wI

R" i"
"

di"

di'
L" cos nd, ----!
..
' dt

+ L"" ~
+
dt

(4-141)

di"
0= L:cosm/'d" - nw",L;:sinl/l/,i;:

R'i~ !

di'
l.:-,...'!
((

(4-142)

Equations 4-141 and 4-142 are linear differential equations with constant
coefficient!!; therefore. they can be solved easily by using the complex
value (phasor) of v~ given in Eq. 4-138 and by assuming the currents to
be given by
i~ = Re U~ejwr)
(4-143)
jw
(4-144)
i~ = Re U;e ,)

(4-137b)

Substituting Eqs. 4-143 and 4-144 into 4-141 and 4-142. dropping the
Re, and dividing out the exponential factors yield

VI = (R' +

o=

.'.\

jwL~)/:

+ jwL'; cos n1/11;

(jwL'; cos nl/1 - nwmL'; sin

(4-145)

n~J)I~

(R'

+ jwL:)I;

(4-146)

Solution of Eqs. 4-145 and 4-146 for the phasor currents I~ and I; gives
I

I,~

I =

(R' + jwL:)(R" + jwL~) +

(R' + jwL~) V!
w 2 L:;l cos 2 n.jJ

+ jwnw,,.L';l sin nl/1 cos nl/1

(4-147)

,,~:,.,~

:1"". '
:

'i

:i

In order to study the energy conversion properties of the repulsion

machine, assume a steady-state problem with the following constraints:
1. The stator is excited by a sinusoidal a-c source
v~ = VI cos wI = Re (Vse

2. The brushes are short-circuited together v;=o' .~.

~'

"I'Vi
'~:t.,

(R'

+ jwL:)(R' + jwL~) + w 2L:: 2 cos 2 nif + jwnw mL:: 2 sin nl/1 cos nl/1
(4-148)

The currents must be expressed in the time domain before the torque can
be evaluated; in order to do this, Eqs. 4-147 and 4-148 must be put in
polar form, as

~ '~~

1)

t.'6

(4-138)

(nwmL'; sin nl/1 - jwL;: cos nl/1) V'

1; =

'\

Fig. 4-18. The repulsion motor.

i;
i ~ p

Jwt

':-""

"".~,',"

Wm

= constant

"f'

J(R'R' -= ~;2(L:L~
+

8 = -tan- 1

(4-140)

,.

p"

l~ej9y

(4-150)

I; =

3. The rotor is turning at constant mechanical speed

eP

(4-149)

where

1,. =

(4-139)

I~ = l~eJO,.

w(L:W

VR,2 +
_L,;2

[w(L:R s

(wL~Y V'

~~s2n-J,)]i

+ L~R' +

(4-151)

'--

nw mL';2 sin nlji cos

nl/1)J2

+ L:'R' + nw m L:;:2 sin nl/1 cos nl/1)

wL:
+ tan- 1 - R
"(L'LS
L"srl cos 2 II<I)
- w~ " " -

nr n '

1.

(4-152)

:'AJ

:-m

'f

'~

308

ELECTROMECHANICAL ENERGY CONVERSION

I(

nW m

I-"s rsinn,/,
'
.1.),
-

v'R

8 = -tan-I

COS

nW m

r1

(L sr

"

COS

.1.)2

n'l'

(wL:F

TWO-PHASE TRANSFORMATIONS

In the case of the repulsion motor it can easily be shown that the rotor
coil currents and the stator coil currents differ in frequency by the
mechanical speed by applying the a(3-yo transformation to the brush
The frequency condition for torque production is satisfied in
current
the repulsion machine.

(4-153)

n.jJ
wLr
- tan--\ - "
n.jJ
Rr

SIO

8
"

i;.

(4-154)

Using the magnitudes and angles defined by Eqs. 4-151 through 4-154,
Eqs. 4-149 and 4-150 can be combined with Eqs. 4-143 and 4-144 to
obtain
(wt

0a)

(4-155)

I; cos JJr

8y )

(4-156)

i~ = I~cos

i;'

309

(1',,).1'...
V' = constant

Substitution of Eqs. 4-155 and 4-156 into 4-137h yields for the
instantaneous torq ue

..,.'.1

./

T,,, = -nL: sin

n.jJ[/~/;'

cos (wt

+ OJ cos (wI +

8y)]

(4-157)

Using the trigonomctric idcntity ror the product or two cosincs gives

',,1,4\tii.4;A,.i.,1'

T,,.

= -; L: 1;;1; sin nift[cos (0"

- Oy)

+ cos (2wI + 8" + By)]

(4-158)

The second term in brackets has no time-average value and does not
contribute to the energy conversion process; thus the averagc torq ue is
/
given by
(T".)avg

= -~ L;: I~/; sin n.jJ cos (0"

(4-159)

- oy)

Substitution from Eqs. 4-151 through 4-154 into Eq. 4-159 and subsequent
simplification lead to
'l'

"':'.

s2

(T)
=
(nj2)L:;2 sin n.jJ(w2L~ cos n.jJ - nwmR sin n.jJ) v
r,. avy
[R rR' _ w2(L:L~ _ L:;2 cos 2 n.jJ)]2
+ [w(L:R s + L~R' + nw L:;2 sin mil cos nifl)]2

(4-160)

f'

A typical set of torque-speed curves is shown sketched in Fig, 4-19.

Note from Eq. 4-160 or Fig. 4-19 that, except for the appearance of
W m in the denominator, the average torque varies linearly with speed with
a negative slope. This indicates that the repulsion motor could be
applied successfully to a feedback control system. Before two-phase
servomotors wcre developed, the repulsion motor was used in such
applications. One disadvantage of the repulsion motor is the brush
friction which reduces the static positional accuracy obtainable in a
position servo.

"'m

,,'.

....

;'\~""

'f~

v,':~,li~"

,~:;~; V~Lt.-

IJ;.,(
o"t'"
'.;~. -

,r:!'

~~\l
,?,J..
ii

~."'
'I

. 1-1

I
I~,
.~,

Ib=f
,\>,

(cr.l

Fig. 4-19. Torque-speed curves of repulsion motor.

!'

i:,t

4.4.2

.:j

The D-C Induction Machine

In previous examples constraints were placed on the equations of

motion of the generalized machine to obtain energy conversion in conventional machine types. It is also possible to consider unconventional
modes of operation of a commutator machine by using the general
equations of motion. Consider a commutator machine with a brush
carriage that can be driven in continuous rotation with respect to the
stator. The general equations which describe this type of machine in
terms of its terminal variables are the equations in terms of a{3-yo components, given by Eqs. 4-136a and h.
Now apply the following constraints to the machine:

',':,

f ...

',,,,>

1. Inject a constant current into one pair of brushes

i; = I

constant

(4-161)

...01

.) I V

r:L.r:\..1 t\VIYIC\..HANILAL tNtKv T

('UNVI::K~I()N

This instantaneous torque is a constant and is the averagc torque. Since

the m;lchinc doc~ produce an avcrage torque. it will convcrt energy.
Notc lhat the torquc is indcpendent or rotor ,peed "'",. This is a consequence or having assumed a current ~ollrce in!,'llIt.
The electromagnetic torque as a function of brush carriage speed is
illustrated in Fig. 4-20. The torque i~ always in a direction opposite to
that in which the brush carriage is turning.

2. Open-circuit the other pair of brushes

i~

== 0

(4-162)

3. Short-circuit the stator circuits

v~ =

(4-103)

v~ =

(4-164)

4. Drive the brush carriage at a constant speed

= Wb

'" =

wb t

and inquire jf the machine will successfully convert energy. To answer

this inquiry, it is only necessary to solve a steady-state problem; therefore,
the rotor speed can be constant

4>

= W m = constant

LS

I'

dla.s

~';

t ~.

(4-166)

Substituting the constraints of Eqs. 4-161 through 4-166 into Eqs. 4-136a
and b yields

O :a: R S"I..

Electromagnetic torque
. Tep.

(4-165)

.#

sr .
dt - nWb L" ! S10
nWb t

(4-167)

n L:;t

~
l

2[;8[2

,j "

"

= 1'1L "sr( IfJs/ cos nWbt -

'\I

-nwb

K:

Brush carriage speed

i;Wb

(4-168)

(::
Tel'

p.

d' S

o = RSi~ + L~ ;; + nWbU:! cos nWbt

=constant

s/ sm
. nWbt)

....1

_!! _12.,- (~f-----~

(4-169)

la

,.

L ,;rt

to;

2 L~

,,'

Solution of Eqs. 4-167 and 4-168 for the currents gives

L :; J
" i; = V R s1- nWb
+ (nw~~)2 cos (nwb t + 8)

(4-170)

I~ =

with

-nwhLsrJ
R s2 + ( " b :J2 sin (nwbt + 8)
nw L
""~i . , ..; ,j.lt,;l <I,"
,,
,.,'_ .. ,.....
t ;t~ l"~

(4-171)

'j

~"

)~i

';~ J.J'''~'~

"H.a:(I,l:i~:;I.

~~.:'}:l'~ '~:h ~.~)

;1:

. L,~,Ut!

Note that these are a balanced pair of two-phase currents of frequency

nWb and phase sequence ex to {3.
Substitution of Eqs. 4-170 and 4-171 into the torque expression (Eq.
4- I 69) and subsequent simplification lead to

j
~

i.
; ~

n2w RSLsr2
b
"
/2
s2
Te/lo - R + (nwbL:J2

,I
J,

'(4-172)
~\'N

.!

I:

Fig. 4-20.

Torque versus brush-speed curve of machine of Sec. 4.4.2.

S
1

'i~

8 - tan- 1
. JS
nwtr-'/Io

r
'.

.~i;jj

,.i

t~
~

4;

Now consider qualitatively what makes the machine produce torque in

terms of fields. Whenever a eommutated winding is fed through brushes,
the spatial position of thc resulting Jield depends only on the brush
position; consequently, when the brushes arc fed with a direct current
the rolor field is of constant amplitude and its position is fixed by the
brush position. Since the brush carriage is turning at a constant speed
Wb with respect to the stator, the constant amplitude field is rotating at
spced Wb with respcct to the slator. With thc assumed sinusoidal distributions ofiiclds, each stator circuit will have a sinusoidal voltuge of frequency

.t,

-'0

i;".l1~

'.

312

ELECTROMECHANICAL ENERGY CONVERSION

induced in it by the traveling rotor field. The two stator circuits

are displaced 7T/2 electrical radians in space, and these induced voltages
will be 7T/2 electrical radians out of time phase. The induced voltages will
cause two-phase currents to flow in the stator and the two-phase currents
of frequency nWb will set up a constant amplitude field rotating in the air
gap with angular speed Wb' This field due to the stator will be fixed with
respect to the rotor field, but will be displaced in the negative r/> direction,
because the currents lag the induced voltages in the inductive stator
circuits. Thus the rotor field is pulled in the negative c/> direction by the
stator field, and the torque on the rotor is in the direction opposite to the
direction of brush carriage rotation. From the foregoing discussion it
is apparent that this machine has some of the characteristics of both a
d-c machine and an induction machine.
nWb

TWO-PHASE TRANSFORMATIONS

313

Ce JO )

or
the complex transformations yield new variables which do have
physical significance. The two-phase, steady-state, symmetrical component transformation (identical to the + - instantaneous transformation)
is widely used to reduce unbalanced two-phase networks to two equivalent
balanced networks. Similarly, there are three-phase. or for that matter
all orders of multiphase, steady-state symmetrical components which can
be used tb study unbalanced steady-state networks. These symmetrical
component transformations are identical to the instantaneous complex
transformations with the one difference that the instantaneous symmetrical
components are obtained by transforming instantaneous variables where
conventional steady-state phasor symmetrical components are ohtained
from the transformations applied to steady-state phasors.

4.5. I Instantaneous Symmetrical Component Transformation (+ _)*

4.5

"

..

"

The two-phase, instantaneous, symmetrical components or

ponents for both stator and rotor are written as

The Complex Transformation

The equations of motion of the generalized machine have been:.\pressed

in terms of three different sets of real variables af3, dq, and yS. Each of
these sets of real variables was shown to have true physical signilkance.
The dq variables were the equivalent of brush quantities of a commutator
machine with fixed brushes lined up with the stator axes, and y8 variables
were the equivalent of brush quantities of a commutator machine with
movable brush structure. The real variables are sufficient for the treatment of most electric machinery problems. There are, however, other real
transformations which may be used for special cases, and there are also
complex transformations which are particularly useful.
The complex transformations, which will be introduced, yield the
complex counterparts of the af3 and dq real variables. The complex
counterpart of the exf3 variables is called the + - variables. and the
complex counterpart of the dq variables is called the fh variahles.
The complex variables + - and fb when expressed as instantaneous
quantities cannot have true physical significance since any physical
quantity is of necessity a real quantity. These instantaneous complex
variables do, however, have many analytical advantages that will soon be
shown.
The complex transformations were first introduced to electrical engi neering by Fortescuc* in the study of multiphase networks operating under
unbalanced steady-state conditions. When used with quantities which
are steady-state phasors (i.e.. currents and voltages expressed as A + jS
C. L. Forl<:scue, .. Method of Symmetrical Coordinates Applied to the Solution of
Polyphase Networks," Trans. AlEE, Vol. .n. PI. II, pp. 1027-1140.

~=~
~

.JI~tJ')I~; 1, ~,'"

1J<

t .~

I"~

~<--,\I,

I'Hili

Tl-~'~is..

com-

~=~

(4-1730)

~=~

(4-1 73b)

.,...

~-

~,~~'

The transformation which relates these components to the afJ components is

L,.:.",

';"
1

,.

+-

XaJll

= [o+-J

x+_1

(4-174)

. _' 1

~.;

1..~:1

i
j

The matrix x] represents the voltages or currents on rotor or stator. and

the transformation matrix [a+-] is defined as
,.~

o'

~.

[0+_]
where j

I'I

= Y=1.

1 [ _j1 jI]
v2

(4-175)

The inverse of this matrix is

[0+_]-1 =

I
v'2

[1I -~]

(4-176)

Instantaneous symmetrical components were introduced hy Professor Lyon and

appear in W. V. Lyon, TrUllsient Allulysis of Alternatillg-Currellt Machinery. Technology Press, Cambridge, and John Wiley, New York, 1954.

t._

I.

31'1

and this inverse transformation is used to yield the relationship that

x+-l = [a+_]-1 ~

(4-177)

The complex amplitudes of e}wl and e-}WI can be defined as steady-state

phasors, i.e.,

v; =
V;

Notice from Eqs. 4-175 and 4-176 that

Thus the symmetrical component transformation satisfies the conditions

of Eqs. 4-12 for invariancy of power.
An important characteristic of the + - components can be seen by
considering a special case. Consider the following two-phase stator
voltages which arc unbalanced in both magnitude and phase:
v~

...

v~ =

V~

cos (wt

v'+

(4-180)

- = V2 (v~ -

jv~) '. ..

" ,

+-

components which is immediately apparent

(v'+)

(v'_)* "_'-

(4-181)

The + - components are always complex conjugates. Furthermore, the

instantaneous + - components arc the complex sum and difference
formed with the a and jf3 components. Since the + - components are
complex, a true physical interpretation will be impossible, but to develop
some appreciation for these variables consider the Euler equation for sine
and cosine functions and rewrite Eqs. 4-179 in terms of complex
exponentials:
'
v~ =

v~

V~

t
,f

lt~,

V;* e-jWI)

= 1( Vflel~" +

V~*e-jwl)

(4-185)
f~

21[(

ysa.,;{
+ 'YS) e}wl + (y.*
+ .y,*) e-jwl ]
a.,;;
fJ.

IJ

(4-186) .~

* - 'V'*) e'-}wr]
"";2iV') e}wl + (V'a.,;{
{J

(J

Equations 4-186 for the instantaneous + - components are particularly,

interesting in that /"+ and I"~ are seen to be composed of the steady-state:J
symmetrical components and their conjugates multiplied by e1w1 and
e-}wl. That is. by transforming the steady-st:lte phasors of Eqs. 4-11l) "
to steady-state + - components using Eq. 4-177, the result is obtained

fuu

v: = ~ (V~ + jVfJ

'-,'

(4-187)

V~

~2 (V; -

.'~

jVfJ

..10

The voltages in Eqs. 4-187 are all steady-Slate phasors, and they can be
plotted on a phasor diagram as in Fig. 4-2 I. These steady-state phasor
+ - symmetrical components defined by Eqs. 4-187 can be used to write
the instantaneous + - symmetrical components of Eqs. 4-186 as

i:..

'.'

i~~'

(4-182)
. ~. ~<t

1( V~e}wl +

v~ = 2:J [( v' -

,t,-\ '"

v~

~ "

2 (e}(wt+6 a) + e-i (",,+6a

-,
v; 2}1. (e}(WI+6~)

(4-184)

Ji

'~~,

One characteristic of the

from Eqs. 4-180 is

= j VJrjOfJ

Now substitute IJ; and ()~ as defined by Eqs. 4-185 into Eqs. 4-180 and b
solve for the + - voltages
II

(4-179)

v~

"

= -V12 (if- + jv~)

v'

VJ*

v~ =

9a )

The transformation of these voltages to instantaneous symmetrical

components can be obtained using Eq. 4177, which gives

V;e- j6"

Using these phasors, the instantaneous a{3 voltages of Eqs. 4-182 become

i,

V; sin (wt + 8~)

(4-183)

= - jV~ej6fJ

V~*

(4-178)

[0+-];" = (0+-]-1

V;e}6 a

v~ = t(V:e}wl

~!'l;

- e':"}(wl+9/1

v~

-HV~e-}WI

.'.

V~*e-}wl)

V~e}wt)

(4-188)

;
;Vt~

'r
.~

I'

,,",

po;,.

11:,

;' (tf

'n

1 "
"U

.~

316

;~:.:'

,",h~
"~'14'
j,

The: instantalH:ous + - symmetric'll compol1<:nts ure see:n to be formed by

taking the phasor symmetrical components or their conjugates and
multiplying by either ej",t or e-jwt .
For balunccd cxcil:ition V~ = V; = V and Ba = OtJ; thus by Eqs. 4-182
and 4-1 H7 the: phasor symmetrical compone:nts for bulunced excitation are

f ..

,,..,

"

. ..,...'

.,1
i.!t

V~

It

e'

,. ..

.. _""

;~-",..

~I,

~V'elq",
V2

= 0

balanced voltages:
S
V~ = V~ = V

1
,

(4-189a)

(J",=(JjJ

Equations 4-189a show that for balanced two-phase excitation the steadystate positive sequence component is proportional to V'. As a consequence

-Ji

\j

I'

317

TWO-PHASE TRANSFORMATIONS

4-188 holds. There is one important difference between the instantaneous symmetrical components and the phasor symmetrical components
which must be observed. For balanced excitation the phasor (-)
component V~ 0, but the instantaneous ('-) component v~ is not zero
and in fact is merely the conj ugate of v~. The difference often leads to
considerable confusion unless one merely accepts the fact that the
instantaneous symmetrical components are representative of a set of timevarying quantities and the phasor symmetrical components are representative of a set of voltages expressed as complex magnitudes or phasors.
The interrelation between the instantaneous and phasor components has
no particular significance beyond the analytical relationships defined by
Eqs.4-188.
It would be helpful if a physical interpretation of the + - components
could be developed in a manner similar to that for the a{3 componen,ts.
Unfortunately, it is not possible to define coils which carry complex,
time-varying currents, and when using instantaneous symmetrical com
ponents no physical models of the generalized machine will be assumed.
It will be found in later studies that for steady-state conditions the phasor
symmetrical components do have a physical significance. The connection
between phasor symmetrical components and physical models will be
discussed in the treatment of examples later in the chapter.

";. ~', ,,,,, 1>

"

~,

ELECTROMECHANICAL ENERGY CONVERSION

V~

\"'-1'

li.

1.'.~.

4.5.2

,'t~.

Instantaneous Symmetrical Component (+ -) Transformation

of the Equations of Motion of the Generalized Machine

'~

-jV;
,I
~,.1

"I
I

!I ",

~t ,-,'
-"[1'>'
':f?~I.

Fig. 4-21. Phasor diagram showing two-phase voltages

as steady-state phasors and the steady-state + - symmetrical
components as phasors.

'<P;; ~

I '/
! '

f'i
I

the instantaneous symmetrical components for balanced excitation

b~comc simply
US

v:'

jw
V,.
'
V; e*e'-jwt

for balanced voltages

The symmetrical component transformation defined by Eqs. 4-175 and

4-176 consists of constants. From the previous treatment of transformations in Sec. 4.2.1 it is known that a nontime-varying transformation
, will result in new impedance matrices &I, 2, andY but will not introduce
any new matrix quantities such as the C or !J/ of the dq and yS transformations.
The symmetrical component transformation will be used to introduce
new variables for all stator and all rotor quantities by defining the new
+ - variables as

(4-1 89b)

M):l,

The instantaneous + - components for balanced excitation have a time

dependence which consists of a positive exponential e jwt for the (+)
component and a negative exponential c- jw1 for the (-) component.
For unbalanced excitation the more general time depwdencc of Eqs.

~
~

1-N

+-.+-

';.

~
= ~-

;;I
:....:..:J

~'

tn
~
S.'

e+_.+_

"

318
I

It
, i

iJI,'
.;z;

By using Eqs. 4-175 and 4-176 the a.{3 voltages and currents are related to
the + - voltages and currents as
V'"

ail,of!

is,r

aiJ.a,g

+ -. + -

where

.s;1';_.+_
[a+_]

=!.

[[0+-1

+ -, + ,'f;

0]
[a+-]

"

1]
j

hl
i

(t;4 .

r:~

Using Eqs. 4-191 the equations of motion given by Eqs. 4-19 through 4-22
in terms of a{3 variables can be transformed to + - variables. Before,..
writing these transformed equations, it should be observed that the
transformation d'':; _.+ _ is complex; thus the equations for power and.
torque must take on the form defined by Eqs. 4-13 and Eq. 4-18
respectively. The transformed equations of motion become
(?

tJ'.;_.+_
T = J <p
I

-I

= .s;1':;_.+_{8i'.;;... + pfi''.;;.aiJ).s;1'::_.+j';.'_.+_

+ a tp1 + K ~A.

!ti.

:rs+.,-.+ -

(4-192a)

T =

jJ
'f'

1
'I'

KL
'I'

fitS.'
+ _. + _

('

- odS,'
o~,~ + -, + -

~;,.

/to.;.'r,,
~

t..,

[0+-]

LO

~i'

~<

tt..
t,

[a+~

L"e-J~
"2
,

L'

L'"2

L':eJ".

L~z

LS

L~eJ~

1
L':e- ".
1
L':ze ".

L':ze-1".

L'SA

L~ze-l2Jt.

L':el~

L~le12".

L'SA

"

= .s;1s,,-1

+ -.+ -

ge-J"J

(4-1 lJ3(;,'

,-

,i_ _

y'". od','

af3,~

[a+- J:]

- jnL::e-j".
sr i
. 'inL e ".

= sl'"+ _.- I+ _ [J/S."

[:+-1 o~

.(,

+ [1"1,:] r [I" ~~.:] + [1"1 ~,aJl

+ [I"l ail,aIl] I [/.. all) + [/..z afl]

+ -.+-

[.~

'--

~2

c,.....'

0"

[ESA:t,ail] + h1~:{a/l]

..,~

r[~~ i-;,~] :'[""1 ~,~~

[""2 :/l]

~1.

O~

,..

[:+_]-1
0 Jf'~l0 (r:,g]OJ
o
[0+-]-1

i,
~

(4-193b),

where

~+-]

~ a~.a/l]

l '~
[a~2Ji
~.
Lo

(4-193a~:

_~~'v.
:7'''.'
+-.+-1
+-.+- i""+-,+-

'1Ic'

::. ..;]

~+-J
X

s
+ !f'J.'
l)e"S"+-,+"
+_.+_ P + :r +_.+_Cf

[a+-1j

-L+t

i,

-te"s".
.Ids".
cds.,
e"s"
(4-1 92b)' '
.. +-.+_,
+_.+_, ys.'.
\J13,~ +-.+- +-.+-

= (dS,'
+_,+_

'.

jnIJ;e1".
jnL':zein"".
,

- jnL~~e-in</>
_ jnLsre-in</> !

- jnL':ze-N

-2inL:l-j~

jnL::e jn",

2jnL:2el~

(4-19.1,

.
.....
'.......

.......,...-,,
'~~

,._

~~,

-1

,1;;":

I:

'i,e,

(4-193c)

o~

'7

/f'

OR'O
0 0 0
o 0 R' 0
o 0 0 R~

~t
,..",.

v,,'

The transformation matrix .s;1':; _.+ _ is a constant, and the derivative p of

the product !I!de" can be expanded i,nto an !I! and a:r matrix. Noting"
also from Eq. 4-178 that s1':;~'+-1 = .IdJ:=.I+_, the equations o~
motion may be written as
i~t!f
'VS,'
+_.+_

0]

..

:.;f'

31'9

iJ S." _ rd S,' .z+-,+~ _c4J,ot:r""" +~,+-

~a+_]-t

'.;

SlJIs., - I

1
f

(4-191)

is''

1 [ 1
= Vi
-J

=:I

+ ":""2--"'!:'-=----

...

= .fils"+ -.+ - vi"+ -.+= ds,r

TWO-PHASE TRANSFORMATIONS

ELECTROMECHANICAL ENERGY CONVERSION

,.

:\-1,

~,~L

Taking the various parameters derived in Eqs. 4-193c, d. and e and

substituting into Eqs. 4-J93a and b, the equations of motion can bl.'
written in the explicit form of Eqs, 4-194a and b.
The equations of motion expressed in terms of + - components 0
both stator and rotor are given by Eqs. 4-1940 and b. Note that all tl,
sine and cosine functions of </> have been replaced by compkx exponentiaL
of <!>; also from Eqs. 4-188 note that the + - excitations are complex

"".~

_.
.

~,

"

f.~
~,'

,.

~--~----

'" I

til

it .. , .

"It', '''~'''''~'''il'''~''..

.., I

T'

'll'

I.'. "'-1-.'" 'PI tI'.'.,'" H.,"...... fth,14"H'U'"

,u; I

J .",... _t1

iJ;

IF. tn.

to; UA-FTtd,J!tL' ,.H

f'

'~

...

?olO

,.....

0f

..j.

......
'<t

k+

~I

......

'- I

":..+

,.....
I,..... ,..... -e-e- -e-

.s, .s,
I

.5
"E>

i'

.....
~

.!;

~"

,.......

,.....

1~

--,.....
-e-

.s, .s,
+ +

.~

C'l

~"

~
......
+
"E>
.....
Q:; .....
"
~
:;'
~

i'

'-'
"E>

~"

~r

...,;j

--,.....

-e- ~
.~

... ~"

~"

+ ~
.!;

Q::

.~'

+
~

;-."

-e- ~
.IS,

~"

'"
+ .,,,
...,;j

'-"E>
.~

a.::

1:;"

"'-l

.,+
:::.

.~

.~

......
."E>~
~

:.f I

;;,

'-

"
']

'-

,.

::=t
~r

+
0:;;:"
.~

\\i .

~+

'-

.,+

I'll

'I

, !

1\

!,Y

~+':;""

t+~
'-' I

::-'~

~"~

'--' ~ +
.~~

."'"
I!

~.!.....

+ N~
,.....

-e- ~ +

.\

~ ~l

..... +

:::.

;;,

The Complex Rotating Transformation (fb)

")'

'-&. "'-l
~+

4.6

,t(,

':-,..+
.. -t-"

exponentials if the a{3 phase variables are sine or cosine functions of wt.
The ability to express the vOlt-ampere equations in terms of complex
exponentials is a significant gain when trying to find an analytic solution of
the equations of motion .
The instantaneous + - equations of motion are still functions of </> just
as are the ap components, and indeed they should be, since the +components are the complex counterparts of the a,S components. For
most analytical work there is a significant gain in the ease of handling by /,II
eliminating the ,I, dependence from the equations of motion. This can be
done hy introducing the complex counterpart of the dq transformation,
which ;e salied the fb transformation. Before working an example with
the complex components, thefb components will be introduced.

q,:

+d'

II
.,1

:~

,.
:r ~~

'"
~" ~~
~

~+

......

-e- -e-

(.y.,

i"~ Q:;
~" ~r

"'-l

f
~

::;r

.~
N

'-'

t:3
'<t

PJ.:f;i

I'

,>;, ~f"

II
Fo-.

The vOlt-ampere equations in terms of + - components exhibit nonlinear terms of the form T(</~i and L(</(dijdt). The torque equation'
exhibits a nonlinearity of the form T(</iSi'. It is apparent that in the
simple case of rotation at constant speed ~, the voltage equations are
linear with time-varying coefficients, and the torque equations are nonlinear.
To partially relieve this difficulty of handling the VOlt-ampere equations
with time-varying coefficients, a change of variables will be introduced
whose transformation matrices are functions of the angle </>. These
transformations, called rotating because of the angular dependence, will
remove the angular dependence from both voltage and torque equations
but will leave the voltage equations with speed terms. Hence the voltampere equations wiII contain nonlinearities of the type ~i, and the
torque equation wiII contain nonlinearities of the type iSi'. Even though
the nonlinearities are not completely removed, the equations are simplified.
For instance, with constant speed of rotation ~, the voltage equations
become linear with constant coefficients.
The rotating transformations for real variables have already been
introduced and are called the dq and yS transformations. The complex
rotating transformation called fb is applied to the equations of motion
in terms of complex ( + -) variables.

l'tDW., ,

~"t~
~\'<1

'V'H':l'~

';ttt

j'''' I
,.,';qlol!

.........."

_.~J

;,:=;'1
t ..
,!

'

I.,
"I,
I!
I"

...... ,I

,I.

-~,.~:

:'lI!

I ...... '
\.
.Jl"~;':_~

"
~

"-f,

t{~

l';

Il'

Thcfb transformation was introduced by Professor Ku. A recent publication

utilizing fb components and having an excellent bibliography on the subject is:
Y. H. Ku. "Transient Analysis of Rotating Machines and Stationary Networks by
Means of Rotating Reference Frames," Trans. AlEE, Vol. 70, Pt. 1, 1951, pp. 943-957.

.....

,."

. '"
~

.,p6t

3n

ELECTROMECHANICAL ENERGY CONVERSION

TWO-PHASE TRANSFORMATIONS

The new set of symmetrical components, the fb (forward-backward)

components, are written as

~=~f
v'

'
t;.:l
~ =~
.:
.,

and

(4-195)

2J

These new variables are defined in terms of the

transformation

x~_1

[alb]

+-

derived in the preceding section.

derived from Eqs. 4-188 as

v:. = -!(V:eJwt + V:*e-Jwt)

v:' = 1(V:*e-}wt + V:eJwt)

vf = H V:e}(wt+n</ +
v~

i:

(4-196)

= -HV:*e-Jlwt+n</

where the matrix x] represents either voltages or currents and the transformation matrix [alb] is given by

[alb]

~}n</>

LO

These voltages for the rotor can be

eg

(4-201)

When the transformation of Eq. 4-198 is performed on these voltages,

there results
"

components by the

13.23

V:*rJ("I t - n</)
V:eJ(wt-Il</)

(4-202)

Imaginary axis

(4-197)

The inverse transformation is

Vb

~
[']-1 ~
~ = arb
~

. ,1

(4-198)

and the inverse transformation matrix is

>

[a/b]-l ==

Rn<l>
~

e-J~

'0

v~ =

rJn<l>vj

v':,. =

ejn</>v~

t.> ..

~'J~t

tl"'~

l;~'

,( t,.

:)~~.

}'
I ' ~,>

"0

":~'

(4-199
.rt.l !C~

When the transformations of Eqs. 4-196 are written out for voltages,
there results

..

t \

~"

,1"1

'd.

}~";! ~f.~')

Real axis' , ....

,-.

. . . . .1

i?

, ~,

(4-200)

If the angle t/> is a function of time, Eqs. 4-200 indicate that there is relative
angular motion hetwcen the + - components and the Ih components in
the complex rlanc. Sincc this motion is n:lativc, it can be assul11ed that
the fb components are fixed, and this will yield thc geometrical interpretation shown in Fig. 4-22. The vector v'+ has the same magnitude as
vf and rotates in a negative direction with angular ekctrical speed nq,.
The vector v'_ has the same magnitude as v~ and rotates in the positive
direction with electrical angular speed nq,.
An important etfect of the relative motion in the + -fb transformation
can be illustrated by performing thc inverse transrormation or Eq. 4-198
on the + - components or the unhalanced two-phase voltages th,lt were

Fig. 4-22. Interpretation of the rotating symmetrical

component transformation.

(-

If the rotor is turning at a constant speed

q,

Wm

and

ep

= wmt + 8

(4-203)

Eqs. 4-202 can he written as

1[ v: ej[(w+nwm)t+n~l + V~*e-}[(w-nwmll-n8J]
= ~[V:*e-jl(w+n.U"I),+n"J + V:e}[(lir-murn)/-m'lIJ

pi =
l:~

(4-204)

A comparison of Eqs. 4-201 and 4-204 shows that the rotating (fb)
transformation has changed the frequency of the voltage, from w to
w nw m
The change in frequency nWm is simply the relative electrical

ELECTROMECHANICAL ENERGY CONVERSION

324

TWO-PHASE TRANSFORMATIONS

and the equations of motion are

angular speed 114> introduced by the transformation. This is a manifestation of the fact that a voltage source of a given frequency will appear to
have a different frequency when viewed from a reference frame or coordinate system that is in relative motion with re~pect to the source.
Such a result is analogous to the Doppler effect in acoustics.
A comparison of the transformation matrices, Eqs. 4-197 through 4-199.
shows that
[arb]~ =

[arbJ-l

,";-./b =

= v~

= vb

-I

+ K.-L'P -

(1.1

'P

.w,.r.
/7'",.r
d,.r. i".r
_ +-,fb,
+-.fb, +-,+- +-.Ib +-Jb

(4-208a)
(4-208b)

LO ,~J

p-jn~

and

[arbJ =

Since the transformation ",;";_ Jb is a function of 1>, the derivative p of

the product Y.w i can be expanded into an ! and a C!J matrix. The equa- ~
tions of ;"otion can be written in terms of + - /b components as
'

(4-181)

l'

_',r

+ -,fb

The rotating symmetrical component transformation is useful in the

analysis of induction motors. In an induction motor the rotor coils are
short-circuited; therefore. the frequency of rotor currents and voltages
which are the result of induction differs from the stator frequency by the
mechanical speed. A + - tolb transformation of rotor variables removes
the frequency difference and makes all currents and voltages have a
single frequency when viewed from the stator reference frame. The
resulting simplification can best be illustrated by performing a transformation of rotor variables on the equations of motion given in Eqs.
4-194a and b.

p.!F~r_.+_)$/'+'_Jht'~r_Jh

~e',.t.

""-./0 ~ ~J [;~

(4-205)

(4-206)

,w,~r_Jb(grs~'_.+_

where

Consequently, it follows from Eqs. 4-200 that

vI

T = J.i.
'I'

which satisfies the conditions for invariancy of power given by Eqs. 4-12.
In the preceding section Eq. 4-181 established that
v+

325

'">'

= (Ms.r
'+ -Jb

ro'.'

.z,

= J1'P + a.1'P +

+ -JbP

,,.r

.1),;,.r
+ -,[b

+ -Jb'P

K,J..'f' - leo,r.
:y,.,
e:,.r
2 + -,fbi
+ -.fb + -,fb

(4-209a)
(4-209b)

where
f:jf""

:ii

+ -.[b -

~J

",;'",-1 9f'"

+ -,fb

+ -.+ -

~J-l

LO
10

r-

OJ
OJJ G [a~~
0R' U

[arbJ-1

~'OR"

11.\

",;'s,r
+ -,/b

0
0

0
0
R'

[r:,

I.

(4-209c)

:1""+-,fb - .~"r-I
.!f""+-.+- $/'"+-.fb
+-.[b
4.6.1

The (b Transformation of the Equations of Motion of the

Generalized Machine

The[b transformation, Eq. 4-197, is a function of,p. which means that

the fb equations of motion will require the introduction of a ' matrix
defined as in Eq. 4-32. Sec. 4.2.1. Thefb transformation is only applied
to, the rotor variables; thus the +- fb components are defined as

= ~]-1

,j

'..
-!

, LO
x

~J
~"

OJ [
[O/bJ-l

[I.. ~-J

+ [1"2 ~-]
I [I..~-,+-] + [1"2 ~-.+-ll
~ ~'-.+_J + [1..2 ~r_.+_] I [I.. ~-J + [1"2 ~_J ]

;
J
I

0-'

[a;~

ELECTROMECHANICAL ENERGY CONVERSION

326

::rs.,
+ -.fb

o ~
[aJo]~ L.!:"I' '.;.$-.+ -]

=t~'
o
o
'nl

[7"1'/+ - ]

jnL':

- jnL':z

jnLs;.z

-jnL::

s,

-2jnL;z

2jnL~z

jllLs;.

~I'z

==

t:-' [a;brJ ~I'

~

[/,,~s+_]

'/- ,+ -]

+ [11'/-;$_.+ -]

1[1,,'+'_.+_] +

[1..'+ -]

I1

10

-jIlU:Z jnL':

T =

-jnL::z

-jnL:

-jnL:z

jnL':

jnL: z

... 1

jnL:~

jnL~

L':zp

U';'(p - jn~) R'

+ jn~) L::(p + jn~)

L:/ Ii

L~(p - jn~)

L~z(p

i'.
is..

L':zp
C:p

+ Jn~)

R'

ij.

- jnt/;)

L~.(p + jn~ I~

ff.<P +

J
a<p

+ K.L' ..:... J'n lr L"(is*i'

I' + I -

;S+ i'*)
J

+ L"(?
i~ 1'2 + j

is*i~*)
+ j

,"

(4-210b)

The equations of motion of the generalized machine expressed in terms

of +-/b variables (Eqs. 4-210a and b) are functions of the speed and
all explicit dependence on the position cP is eliminated. This result is
exactly analogous to that produced by the real dq transformation. Tl)e
elimination of the rp dependence simplifies the equations of motion. arid
when t/; is a constant the volt-ampere relationships (Eq. 4-2100) reduce to
linear differential equations with constant coefficients. To show the
versatility of the + - fh variables, the characteristics of an induction motor
with unbalanced excitation will be studied.

. . t;.'..'

. ,"',
~;

4.6.2

-i,S
-J~"

'{.

R' + L;:p

+ L;'J(i;)2 - (if?])

."~l'.;~J l

- jnL: z jilL:

L':p

(4-21 Da)

it

+ I10
0 -jilL: jnL:)
I

L~zp

,j

; '!'

-jnL':

c.~( p

ll~.-/'

~o

~~

L:;(p - jn,p)

-jnL': jnL~;

+ L~p
L:,zp

Darb]
[ i'q,

s ,'
+-.fb

R'

[l1'2s.(_.+_~

+ [/"z -]

v~

,i
..

r0_ ::r
-

iJ'"

[ll's",,_]

v'I

8,s;1S:; _ ,fb

.I b

o -I~

v~

::r s+.' -,fb + ,>1$,,-1.

2""
+ - ,Jb + -. + :!T"~r

it.

t.'

II

<s.,
~
+ -,/b -

I
(4-20ge)

ti
!

-jIlU:z -' 0

-jilL':
.I

.+ [or1'/.;.$-. + -] I

[;~

~J

I[T. ':-., -1 + [T::-,-21

327

Taking the various parameters defined by Eqs. 4-209c-fand substituting

them into Eqs. 4-209a and b give the equations of motion in terms, of
+ -, fb components as

+ -,fb

::rs+-.+
.'
,s;I$"
- + -./b

.9i/,.,-1

TWOPHASE TRANSFORMATIONS

(4-209f)

:.i:.:.

'

'f'
t

J.'. .~_.
.

The Induction Motor

-w

\. -

A squirrel-cage induction motor usually has a uniform air gap, ac

windings on the stator, and short-circuited windings on the rotor. With
the rotor circuits short-circuited. the rotor is excited by electromagnetic
induction from the stator. When the rotor is turning, the voltages induced
in the rotor will depend both on stator frequency and rotor speed, and
for a given speed W m the stator and rotor frequency will differ by !1W m
The proper frequency difference nW rn exists regardless of rotor speed. and
qualitative reasoning indicates that an induction motor will convert

328

ELECTROMECHANICAL ENERGY CONVERSION

TWO-PHASE TRANSFORMATIONS

energy at all speeds. To verify this and to determine other performance

characteristics, an analysis of the induction motor will be performed
using + - jb variables.
The machine to be assumed in the analysis is a smooth-air-gap, 2n-pole,
two-phase machine. The equations of motion of the smooth-air-gap
generalized machine derived in the preceding section may be used. The
constraints placed on coil variables by the operation as an induction
motor are
l'~ = V~ cos (wI + ea )
(4-211)
= sin (wI + elJ)

v;

v~

I~+
I

.,

~" ~-QI

'='-.1

~I
E
3
.~

~~

= Wm

N,
:!

,;
:'t'.

To study the energy conversion properties of the machine, consider a

steady-state problem with the rotor turning at a constant angular speed

t/J

(4-212)

= 0\

'""'
~

'r)

V;

v: = 0

</> = wmt

f1

~= V2~

.....::

(4-213)

(4-214)

.)

~~

-~ ~

t:>..
0
"4

c:..

,i ~ :,! Jf

'-'

'~"'fll. 'I- /'\.

~'1

'~i

..::;1

~
.::;

t:>..

.,,,
~

,.-,

.....::

*~'"

'-

c:..

~'1

.::;

!.'+

:=-

I:l:::;

::; '1

~'1

II

I., +
;;;,

.,;;,I

,I,

~Ioo...,

I 0
t:>..

.,

"

.,+

;.. :,;',.

""'E

c:..

.,"'-l"
~

;
t

4\ .

---

,.-... .
0

-..J

.::;
II

;;,"" ."'[
;;,

",'

~
f-.,

,~"'l"-

(1

r,

<~

(4-216)

0.

.,' H

,-,

I
,

.~

~d~

~.,

::; '1

,I '

.~

If"~

t:>..

::.-i

'The analysis of induction machines is simplified and useful equivalent

circuits are obtained if the + - fb variables derived in Sec. 4.6.1, Eqs.
4-210a and b, are used. When the constant speed of Eq. 4-213 is
substituted into the volt-ampere relations of Eq. 4-210a and the constraint
ofa smooth air gap which forces L~2 = L~l = L~ = 0 is used, the equations
of motion, Eqs. 4-210a and b, become Eqs. 4-215a and b. With the
speed w'" constant, Eqs. 4-215 represent a set of linear di fferential equations
with conslant coefficients. Because these equations are linear, they may
be left in the time domain and solved, or changed to the frequency
domain and solved. The currents found by a solution of Eq. 4-215a
must be expressed in the time domain before they are substituted into
Eg. 4-215b for the torque, because this expression contains product
norlinearities, and as written is expressed in terms of instantaneous
quantities.
The constraints of Eqs. 4-211 and 4-212 must be expressed in terms of
the + - fb variables. The transformation from real (af3) components to
symmetrical (+ -) components is given by Eqs: 4-176 and 4-177 as

v-=l

c:..
::;'10

with an instantaneous rotor position

I

329

:~

(~
\ 'i-'~~, -.

\1t~'~

ll'h,';ji..t

,,,"'",

..

..

330

ELECTROMECHANICAL ENERGY CONVERSION

TWO-PHASE TRANSFORMATIONS

When the transfo rmation of Eq, 4-216 is applied to the unhalan

ced twophase voltages of Eqs. 4-211 and the results are put in expone ntial
form
using Eqs. 4-179, 4-182, 4-183, and 4-184, the instanta neous symme
trical
compon ent voltages for the stator are obtaine d as

I'

11"

current s which result from each driving voltage . To do this,

write the
volt-am pere equatio ns using Eqs. 4-215a, 4-220a, and 4-220b as
'
,

[r

(4-217)
V"

~2 [_1_
Vz (V''''

- J'V'*)e-jw1
iJ

_1_
V2 (V

J' V') 'Jw']

iJt

These are the + - stator" voltages applied to Eq. 4-21 Sa.

The transfo rmation froll1 symmet rical (+ -) compon ents to forward
backwa rd (jb) compon ents is given by Eqs. 4-198 and 4-199 as

ol~

1 \-;In<l>

, 01. =

..
t

,<'

La rjn~ ~

Eqs. 4-221b and d; consequ ently, by conside ring only steadystate

conditi ons and conside ring that each of the four excitati on voltage
s acts
indepen dently (superp osition) , the current s become
steady state i~ "'" tI~eJwt + tls..*e-iwl
steady state is.. = Y~*e-jwl + tl~eJ""
steady state 1/ V;e}w, + tI~*e-jwt

, f',

,'f

(4-218)

/\,

'"'

The use of Eqs. 4-216 and 4-21 g with the rotor voltage constra
ints of
Eqs. 4-212 leads to the results

steady state

v~ =

and

These are the rotor voltage constra ints in terms of jb variable

s to be
placed on Eq. 4-215a.
The stator + - voltages can be simplified by introdu cing the
phasor
symmet rical compon ents defined by Eqs. 4-187:

r'*' , J

..

V"

v2

= - . (V'

'</'

,_,

t'

V -s

I
v2
( VSa

Using Eqs. 4-187 the instanta neous

written as
!'~ = !( V~ejwl)

+-

+ ,'V')
0
.

\',

(4-1 !l7)

voltage s of Eqs. 4-217 can be

(4-220a)

(4-220b)

These instanta neous voltages are the only excitati ons applied to
I he voltampere relation s of Eq. 4-215a. and since for constan t ~ = "'", Fl!'
4-215a
is linear with constan t coemci ents the prohlem can be solvnl
hy super.:
position . There arc. therefor e, four indepen dent driving volt:l~e
s (two
pairs of compk .\ conjuga te voltages). and it is possible to dcline
the

\,

tl~e}wl

II+ = if*
-

;1

l~

'j"

'V')
J iJ

+ 1(V:eJw ,)

4,\,-.

,I~l)i'

-'

,,"'-

+ H V~"'ejw()

v~ = 1(V~*e-J"")

'\

tJ;*e- j"" +

Equatio ns 4-222 give the instanta neous steady- state current s, and
these
may be substitu ted into Eqs. 4-221 a-d and simplified. Howeve
r, before
doing this, conside rable simplifi cation can be achieved by recogni
zing
that the instanta neous + - andfb variable s always satisfy the conditi
ons
(see Eqs. 4-180 and 4-206):

(4-219)

i~ =

(4-2~1l
"
I

,., "

VI =

+ t<V:,*e-jw1 ) "" (R' + L~pW.1o '+ LS;pi; (4-22Ia )

,
!(V~*e-j(~I) + !W:eJ'./) =(R + L~p)i: + L";pi;" (4-22Ib )
o = L::(p - jnwm)i~ +
+ L~(p - jnwm)]i;
(4-221'c)
o = L";(p + jnwmW..,. + [R' '+ L:(p + jnwm)]i; (4-22Id )
Note that i"-r and i; occur in Eqs. 4-221a and C, wherea s is.. and
i~ occur in
.,(V~ejwl)

= ],[_1_ (V' + jV')eJ",' + _1_ (V'''' + jVf/k-J u,,]

2 \/2 u
J.J
v2

331

v/ "" v~*

(4-180)

= t:,*

(4-206)

i; = i~*

:~,

Note that the voltage s of Eqs. 4-220a and b satisfy Eqs. 4-180,
and the
current s of Eqs. 4-222 satisfy Eqs. 4-180 and 4-206, which they
must by
virtue of the definiti on of the + - lb transfo rmation s. This
comple x
conjuga te relation ship between the + - variable s and the fb
variables,:
coupled with the fact that only (+) and I compon ents occur in
a single:
volt-am pere equatio n 4-221 a or 4-221 c or that only (- ) and b compon
ents'
occur togethe r, as in Eqs. 4-22lb and 4-22Id, leads to the elimina
tion of.
two volt-am pere equatio ns, because they are superfluous, Since
Eq.
4-221h is the conjuga te of 4-221a and 4-221d is the conjuga te
of 4-221(',
it is sufficient to keep only Eqs. 4-22la and 4-22lc. Suhstit
uting for
and i; in Eqs, 4-.?.?la and 4-221(' their steady- state values from
Eqs.
4-222 yields for the remaini ng two volt-ampere: equatio ns:

i:.

V~ejwl

V~*e-Jwl

= (R'

L~p)(l:ej""

"

+ l:*e- jw' )

+ LS;p(lje j""

l~*rjw')

(4-223a)

o=

L';(p - jIlW",)(I: ejbJt

Equations 4-227 and 4-228 can be used to rewrite Eqs. 4-224a and b
and 4-2250 and b. In rewriting Eqs. 4-225a and b the conjugate values
can be eliminated by conjugating both sides of the equation. The main
advantage of doing this is that the reactances in terms of unconjugatcd
voltages and currents will be inductances, whereas as now written. they
appea r to be capacitances. i.e., reactivc tcrms arc - j X. Rewriting Eqs.
4-224u and band 4-225a and b gives

]:"'*e-J",t)
[R' + L~(p - jnwm)](l;ej",t + Ib*e-j"'t)

(4-223b)

Since Eqs, 4-223a and b are linear, superposition can be used to solve
first for coefficients of the eiwt terms and then for the codflcients of the
e-Jwt terms. Subdividing Eqs, 4-2230 and b into two sets of equations,
performing the indicated differentiations, and canceling out the
e~ponential terms yield

Volt-Ampere Equations lor Excitation V:

Volt-Ampere Equations for Excitation V:e iwt

= (RS + .I'w L')]S

+ J'wL"],
I
" +
" f
o j(w - nwm)L';I+ + (R' + j(w

V: = (K

0:=

- nw",)L~]l;

o=

(4-224b)

= (R

nwm)L';I:...

+ [R' -

j(w

nwm)L~]Ib

V:'

(4-225b)

The slip s is

s=~m

+ nW m in Eq. 4-225b is
w + nw", = (2 - s)w

jwL::/'_

jw(L~- L';:)

R'

(2

~ s + jwL~ )If,
jw(L;- L";J

+
I~

j");

<

v~

.~

(4-230b)

.-i';~'

", $

,,.,(

R'

(1-,) R'

i,j

v~

~jWL":

Iy

_(I-,)
(2 -,)

R'

R'

:~

Fig. 4-23.
ine.

\111

jW(L~- L";)

jw(L;- L';:J

. '".

"'f

t,
J

(4-227) ,
'\

t-o
(4-228)>1'

R'

Equivalent circuit of an unbalanced two-phase induction mach-

i;,:

~(

, i

and in terms of the slip the frequency w - nW m in Eq. 4-224b is

-I

(4-230a)

.~

(4-226),;:

and the frequency w

(4-229b)

'~

w - nW m = sw

+ (R'
+ J'wL')]'
S
I'
I

= (R + jwL~)I:'" + jwL';If,

o=

T;he two sets of vOlt-ampere equations given by Eqs. 4-224a and b and by
4-225a and b are independent. This ability to divide the vOlt-ampere
equations of an unbalanced induction machine into two sets of equations
representing two apparently different machines is most interesting and
useful. It should be realized, of course, that the ability to perform this
subdivision into two independent sets of volt-ampere equations is dependent upon linearity, which requires c/>
W m to be a constant.
Induction
machines are not constant speed machines, and it is an approximation
to use these equations for other than steady-state operation. However,'
for most conventional induction machines, the mechanical acceleration
is so slow compared to electrical transients that dynamic operation can
be calculated by a sequence of steady-state calculations for different
speeds, using Eqs. 4-224a and band 4-225a and b.
, To understand the full meaning of the two sets of volt-ampere equations
of Eqs, 4-224a and b, 4-225a and b, it is advantageous to define the slip'
s in the same way that it was defined in Chap. 3 when studying induction) ,
machines.

J'w LrrI"
" +

Voll-Ampere Equations for Excitation V:'"

(4-225a)

jwL~)]:"'* - jwL::lb*

0= -j(w

(4-2290)

lit

Volt-Ampere Equations for Excitation V:"'*e- jbJ'

V:...*

+ jwL~)I: + jwL';I;

(4-224a)

VS

333

TWO-PHASE TRANSFORMATIONS

ELECTROMECHANICAL ENERGY CONVERSION

332

.
The equivalent circuit corresponding to the set of volt-ampere relationships
given by Eqs. 4-229a and b, 4-230a and b is drawn in Fig. 4-23. There
are two characteristics of this equivalent circuit which are particularly
significant. First, note that the apparent rotor resistance for the excitation
V~ is R' /s.
The equivalent circuit corresponding to V: is identical to
that derived in Fig. 3-11 for a balanced two-phase induction machine.

33<4

ELECTROMECHANICAL ENERGY CONVERSION

TWO-PHASE TRANSFORMATIONS

This result is consistent with the present analysis also, since from Eqs.
4-187 V: and V~ are defined as
V:

J2 (V: +

jVJ)

(4- 187)
I
- - V2
(V: - j VJ)

V' _

For balanced two-phase voltages V: = jV;, and the phasor negative

sequence voltage
is zero, giving the equivalent circuit derived in Chap.
3. For two-phase voltages v~ and v~ which have unbalanced phase or
magnitude, there will be a phasor negative sequence voltage V:, and the
complete equivalent circuit of Fig. 4-23 results. The equivalent circuit
corresponding to V'_ has an apparent rotor resistance of C,<2 - s),
which was derived from Eq. 4-225b where the apparent frequency in the
rotor circuit was (2 - s)w = W + nw m. The equivalent circuit corresponding to the excitation V: has an apparent frequency w + nw",. and
the equivalent circuit corresponding to the excitation V~ has an apparent
rotor frequency of UJ - I1W",. This result will be discussed later from Lhe
point of view of the fields in the machine.
As the next step in the analysis of the induction machine, Eqs. 4-229a
and b, 4-230a and b will be solved for the currents. Solving first for
I: and Ijfrom Eqs. 4-229a and b gives

V:

'If::
l((

..

T .,. = jnL',;
[(1'*1'
- I'+ 1'*)
4
+ f
I

+...(/3-r[ - r

l'+

Ij

R
(s

(Tr/Jav~ =

., .-

(~r + jwL~ )(R' + jwL~) +

Next, evaluating I: and

.
.,:

I:
l' _

.
"

I~

(wL;:)2

+ jwL:) V:

(2 ~ s + jwL: )(R' + jwL~) + (wL;:)2

b-(

. LsrV'
-JW
I'
R'
) s
2 _ s + jwL:, (R + jwL~) + (wL';)2

, ,

(4-233)

R 'R

2(L'LS

""

L,,2)
,.

R'V.2

S
12

w2

L'Rs

I'

LS
,.

R')2

"

-tnwL'; 2 -R'
- V z
2 - s -

"

,'"

j2 + (I."R. + L" 2 R'_ s)2' (4-234)

R
2' s
sr2
rL\T>S
2 - s - w (LI'L" - LI')

.r

'"

Before discussing the torque-speed characteristic of the induction

motor, a review of the equations for the currents is in order to see how they
meet the frequency condition that for an average torque stator frequencies
must differ by nW m from rotor frequencies for a smooth-air-gap machine.
These frequency conditions were found for the real (a{3) variables. The
a{J + - transformation is given by Eqs. 4-173 and 4-174 as

(4-23lb)

from Eqs. 4-230a and b gives

Rr
(,2 _ s

+ (p*
r*b
- _
p*[
l' *)e-/ 2wt ]
+

(wL;:)2

- jwL;:V:

]')e/ 2""
+b

J.
1",2
'lIlW -"

-S-

(4-23Ia)

.
+ jwL~. )(R'
+ jlvL;J. +

_ l'*/')
_ b

. L') V'+
+ JW"

+ (I'r*
~ .b

It is evident that the time variation in the last two terms proportional to
ej2wl makes these terms yield no average torque.
Consequently, by
substituting from Eqs. 4-231a and b, 4-232a and b into the first two,
nontime-varying, terms of Eq. 4-233, the average torque is obtained.

:J

R'
(5

;335

The four steady-state currents defined by Eqs. 4-231 a and band 4-232a
and b completely determine the VOlt-ampere relationships of an unbalanced induction machine. These currents can be used to find the
electromagnetic torque, but before substituting into the equation Jor
instantaneous torque the currents must be expressed as functions of time.
By using Eqs. 4-222, where the steady-state currents I~, I:, I;, I~ and their
conjugates were originally defined, and Eq. 4-2 I 5b, the instantaneous
electromagnetic torque can be written as

.~.~

(4-232a)

':' r> ,

~=V2L-j

Il"Z}

jJ=J

(4-235)

The transformation matrix contains only constant elements; therefore,

the frequencies contained in the + - variables will be the same as the
frequencies in the a{3 variables. Thus from Eqs. 4-222 the stator Currents
contain only the electrical frequency w. In order to see which rotor

(4-232b)

.~"::

-,

336

ELECTROMECHANICAL ENERGY CONVERSION

TWO-PHASE TRANSFORMATIONS

currents contain the proper frequencies for energy conversion, the rotor
currents of Eqs. 4-222 are transformed to + - variables using Eqs. 4-196
and 4-197 to yield

to the stator field labeled 1 in Fig. 4-24a. Such a result gives rise to an
average torque. The field labeled 2 in Fig. 4-24b travels at the samt:
speed w in the negative 4> direction with respect to the stator and is fixed

i:

= V.re}[(w-nwmll-naJ

V;;*e-j[(w+nwm)l+na)

(4-236)

i'-

= t1j*e-J[(w-nwml,-naJ + tI;;eJI(w+nwml/+na)

(4-237)

'F\

CD

~J~

In the light of the discussion above concerning the + - to a{3 transformation, it follows that the frequencies contained in Eqs. 4-236 and 4-237
are the same as in the exf3 variables on the rotor.
According to results obtained in Chap. 3, when the stator currents are
constrained to a single frequency w the.! rotor currents must contain the
frequencies (w + nw m ) and/or (w - nw m ) if there is to be an average
torque and therefore successful energy conversion. Examination of
Eqs. 4-236 and 4-237 reveals that this condition is met, and it is verified
that all the rotor currents produce average torque: , This fact is, of course,
verified by Eqs. 4-233 and 4-234.
It is often conceptually helpful to think of the energy conversion process
in terms .of the interactions of fields. It is to be emphasized that a
complete solution for all the currents is not necessary before considering
fields; it would have been possible to start with a qualitative discussion of
fields. For steady-state operation it is possible to consider unbalanced
two-phase excitation on the stator of a smooth-air-gap machine in terms
of fields which are composed of two traveling waves. For the ex to f3
phase sequence that has been chosen, one wave whose amplitude is
proportional to the sum of the ex and f3 excitations travels in the positive
~ direction. The other wave has an amplitude proportional to the
difference between the ex and f3 excitations, and travels in the negative 4>
direction. These waves are depicted vectorially in Fig. 4-24a.
An examination of Eqs. 4-236 and 4-237 shows that the rotor has
unbalanced excitations in the ex and f3 windings with frequencies (w + nw m )
and (w - nwm ); therefore, there will be four traveling waves as a result
of rotor excitation. First, there will be two waves traveling in a positive
<p direction with respect to the rotor, one wave traveling with the speed
(w + nw m ) and the other with speed (w - nw m ). Second, there will be
two waves traveling in a negative'" direction with respect to the rQtor
with the same two speeds. These rotor traveling waves are depicted
vectorially in Fig. 4-24b.
Now superimpose the rotation of the rotor by making the whole picture
of Fig. 4-24b rotate with speed W m in the positive <p direction, and note
that the field labeled 1 in Fig. 4-24b travels at the speed w in the positive
'" direction with respect to the stator and is therefore fixed with respect

337

c/>

t
Axes fixed to stator

'.
,?
(0)

Axes fixed to rotor

(b)

,'0,

f'
-;

-~

~lt.
~~.

~
'';-

",f."

Fig. 4-24. Rotating fields in an unbalanced

induction machine. (a) Stator rotating fields.
(b) Rotor rotating fields.

in space relative to the stator field labeled 2 in Fig. 4-24a. These fields
also yield an average torque. Since the other two rotor fields do not
remain fixed with respect to any stator fields, these rotor fields interact
with stator fields to produce pulsating torques, but no average torque.

."

ELECTROMECHANICAL ENERGY CONVERSION

TWO-PHASE TRANSFORMATIONS

It is not apparent fron' this field interpretation, but it can be shown that
the positively traveling fields, labeled I in Fig. 4-24, give rise to the first
term (proportional to +, /) in Eq. 4-233. and the negatively traveling
l1c1ds, labeled :2 in Fig. 4-24, give rise to the second term (proportional
to -, b) in Eq. 4-233.

balanced machine. Observe that this equation agrees with Eq. 3-264
derived in a dilferent manner for a balanced induction machine.
In the analysis of machines, several characteristics such as starting
torque, maximum torque, speed at which maximum torque occurs, etc.,
an~ of interest.
In addition, a knowledge of how these characteristics
vary with parameter values is valuable.
,
First consider the speed. or equivalently the slip s, at which the
maximum balanced torque occurs. To find this the partial derivative of
Eq. 4-241 with respect to slip s is equated to zero and the result is solved
for sma.; thus

338

4.6.3

The Unbalanced Two-Phase Induction Motor

Most large induction motors are operated from balanced voltage sources,
yet unbalanced operation can occur as a result of unequal line impedances
or other system malfunctions. To see the effect of such unbalance on
the performance of a two-phase induction motor, consider a two-phase
supply which has the unbalanced voltages

..

U~ =

Up
II'

(Vb

= (Vb -

VJ cos

wI

Vu ) sin

wt

8( TeIJ.aVi). = 0

(4-238)

(smax\ =

where Vb signifies the balanced portion of the voltages and V u the

unbalanced portion of the voltages. These unbalanced voltages may be
represented as complex phasors using Eqs. 4-183:
V~

= (Vb +

V p = -j(Vb

VJeiO
-

Vb

Vu

Vu)e jO = -j(Vb

(4-2400)

ys _

(4-240b)

1 (V s
_ - V2
-)'YS\
SJ

,"

_M
= v 2 Y"

As a first step it is useful to consider only Eq. 2400 and solve for the
balanced torque, i.e., the torque which will result if V" = O. Using
Eqs. 4-234 and 4-2400 for this purpose gives

sr2 Rr V 2
nw L ,.. S b

:ft..
,

(T'lJ.avi\ =

I,

rR s

[ s

w2(L~L~.-

zJ2
L:;

( r DS
2
w LjSft

JsRr)2

(4-241)

+ -,.. S

Equation 4-241 is the average torque produced by the balanced portion

of the supply voltages and would be the total instantaneous torque of a

R'

(RS)Z + (wL~)2'-
(wL:W)2 + w4(L:L~ _ L:;2)2

(4-243)

~ L sr2 V Z '

[(Tel' ) ]b
'Vi ma.

= + V' .
-

[(L:RSF

_. - .

+ w2(L~L~ _

2 ..

. S._.

L:;l)2][(R )2

..
+ (wL~)2] +

R'wL:: 2
(4-2~)

The phasor + - symmetrical components can be calculated from Eqs.

4-187 and 4-239 as

Y~ = ~ (V: + jYp) = V2Vb

Next the torque at the value of slip given by Eq. 4-243 is found by
substituting into Eq. 4-241, thus:

(4-239)

Vu)

(4-242)

as

yields

339

'0,

~,

":j.

Equations 4-243 and 4-244 show that the slip at which maximum balanced
torque occurs is linear with rotor resistance R'. Since these two variables
occur in Eq. 4-241 only as a ratio, the maximum balanced torque is
independent of rotor resistance. The shape of the balanced torque-speed
curve, but not the maximum torque, can be controlled by rotor resistance.
A typical torque-speed curve for a balanced induction motor is shown
in Fig. 4-25. The region of speeds over which motor operation is
achievable is evident from the figure. In the region

(T,IJ.aVi). > 0

and

Wm

> 0

electrical energy is being converted to mechanical energy. In both the

generator and braking regions the torque opposes the rotation, and
mechanical energy is converted to electrical energy. The essential
difference between these two regions is. that in the generator region the
converted energy is flowing back into the electrical sources, whereas in
the braking region the converted energy is dissipated in the rotor circuits
as j2 R losses.

.
~

ELECTROMECHANICAL ENERGY CONVERSION

340

TWO-PHASE TRANSFORMATIONS

The starting torque for balanced excitation is given by s == 1 (or

W m = 0) in Fig. 4-25 and Eq. 4-241 and can have any value between zero
and the maximum torque give!) by Eq. 4-244. The balanced starting
torque can be varied between the~e limits by changing the rotor resistance
R'; however, an increase in R' to raise the balanced starting torque will
lower the slope of the torque-speed curve between s = 0 and S = smax'
which is the region of normal operation. Such a reduction in slope will
make the motor speed more sensitive to load torque changes.

magnitude reduced in the ratio of (V,j Vh )2 and shifted 2 units along the
slip axis. It is possible to use both Eqs. 4-243 and 4-244 for maximum
slip and maximum torque by setting

(Smu)u = (Smax)b
[(T'J1.ava)max]u =

'1,
j,~

Balanced speed-torque
1 curve due to
h

Resultant speed-torque
curve due to V/> + v~

I
I

"

-w!

'\:

...

0,

1',

." . . . . . . 1

~w
~
A.

II

2",1

nw

(~:)

+2

(4-246)

[(T'J1.ava)maJb

(4-247)

The torque-speed curve for the unbalanced excitation Vu is illustrated in

Fig. 4-25. and it has the same shape as the curve due to Vb, but it is
reduced in magnitude and shifted 2 units to the left along the slip axis.
The resultant average torque is the sum of the two torques, and since
over the normal motoring region they subtract there is an effective
deterioration in performance. Note particularly that there is both a
loss in sl(~pe around S = 0 and also some shirting away from s = 0
resulting in considerable loss in speed for a given value of torque. Also
notice that if Vu > VJ,. which corresponds to a reversal of phase sequence.
motor action in the reverse direction can occur between s = I and s = 2
which is normally a braking region. Tn addition to these obvious effects,
it must be remembered that the unbalanced voltages introduce pulsating
torque, Eq. 4-233, and also introduce additional losses. Thus the
detrimental effects of unbalanced excitation can be prohibitive in
marginally operated machines.

,fl
c
,1

(T,p. ).vg

HI

I s

4.6.4 The Two-Phase Servomotor

tUff

I--- (Braking)b

(Motor
operation)/>

Fig. 4-25. Torque-speed curve of an unbalanced two-phase induction motor.

":;

The average effect that unbalanced excitation has on the operation of

the induction motor can be determined by evaluating the second term in
the average torque expression of Eq. 4-234. Using the negative sequence
voltage V~ defined by Eq. 4-240b, the average unbalanced torque becomes

R' V 2u
-nwL"/A 2-(T'/Aava). =

r-

~~: - w2(L~L: -

J2 -s

L:;J

w2(L~RS + L~ 2 ~

sf

(4-245)

A comparison of Eq. 4-245 for the unbalanced torque with Eq. 4-241 for
the balanced torque shows that it will have the same shape but a

fi:'
.j,

~i;<

;!. .

"l~~: "

~l
Ii

III.

Two-phase induction 111otors are widely used in POSItion control

systems. The rotor i 2 R losses associated with these motors limit their use
to instrument systcms (25 watts or less). aJlhough with special cooling
arrangements larger sizes (250 watts) are built. Motors designed for
such purposes are called servomotors. The principal characteristic of
two-phase servomotors is that the rotor resistance R' is made sufllciently
high for the maximum torque to occur at a slip of s = 1.5. This ensures
that the torque-speed curve wil1 have a negative slope, and therefore the
motor will have a stable operating region around zero speed (s = 1).
. The method of operation of a servomotor is as follows. One phase of
the stator. designated the reference phase, is connected to a sinusoidal,
constant-I"requency, voltage source called the reference source. The
second rhase is excited by a voltage which has the same frequency as the
reference voltage, but is 90 degrees out of time phase with respect to the
reference voltage. The amplitude of this second voltage is controlled to
controllhe torque or speed of the motor. This second voltage is designated
the control voltage.
lSi. .
/{.'

;1

~~

"":;f4jf'"

'0

;,

r.;

3~2

TWO-PHASE TRANSFORMATIONS

ELECTROMECHANICAL ENERGY CONVERSION

3~3

of
spaced lines which best fit the actual torque- speed curves in the region
in
zero speed. Such a set of linearized torque- speed curves is shown
Fig. 4-27.
The linear approx imation is widely used in feedback control system
approx ianalysis. Starting with the torque expression of Eq. 4-234, the
will
mation snecess ary to obtain a linearized set of torque- speed curves

much
In two-phase servom otors the electric transien ts usually die out
steadyof
set
a
ently,
consequ
ts;
transien
ical
mechan
more quickly than
ter will
state torque- speed curves with the control voltage as the parame
of the
r
behavio
c
dynami
the
to
imation
give a good first-order approx
the
where
4-26,
Fig.
in
shown
is
curves
speed
motor. Such a set of torquevoltage
control
the
and
t
constan
is
reference voltage amplitu de VR
s a
amplitu de Vc is the parame ter. A negative control voltage indicate
phase reversal.

u,.H.~'

(T./Jo )aYI

(Tel' ).VIl

VH

,:;.

~,

'"

constant

V H = constant
\.

It

"'"

...........

=---::I- ........

J:'=--,.
.~

nw m

........."

{t.;' ,

"'-k
-

'-'<

=::"'1

. . . . .. l

.'.. .. nW m

'.1

':

'i

(1:

I.

J ..

1\

Fig. 4-26. Torque-speed curves of two-phase servomotor.

t'

ced
In general, the two-ph ase servom otor is operate d with unbalan
The
d.
produce
be
will
torque
g
pulsatin
a
and
s,
two-ph ase voltage
system
mechanical time constan t is normal ly so large that the mechanical
r
conside
to
ry
does not respond to the pllls~lting torqUt:, so it is necessa
4-234.
only the time :lVcrage torque as given by Eq.
The usc of the torque expression of Eq. 4-234 in the analysis of a system
because
which contain s a two-phase servom otor would be rather difficult
ed by
simplifi
is
system
a
of
analysis
of the nonline arities involved. Thc
al
unalytic
ed
lineariz
gh
Althou
ents.
linearizing the Lkscriptions of compon
the
make
do
they
results,
e
accurat
ely
descrip tions do not yield absolut
results.
problem s concep tually simpler while yielding reasona bly accurat e
practice
usual
the
speed,
zero
near
mostly
s
operate
Since the scrvom otor
, equally
is to linearizc the torque- speed curves by drawing a set of parallel

1"

l:

~~!

\1'

p',r

"i.J!,i I.

~i ..,~

n'if
.. 'j"':'!'<'rH "

r~;'
,I

~,

~t,

t'!

n';

1
(f

>1:

Fig. 4-27. Linearized torque-speed curves for a two-phase

servomo tor. '.f., '.,

"{I
,.

is the
be determi ned. First assume that the ex winding on the stator
that
so
,
winding
e
referenc
t.riP; ~''',.
(4-248)
V~ = VR cos wI
control
where VR = constan t and w = constan t. Next assume that the
angle
phase
the
that
and
stator
the
on
winding
f3
voltage is applied to the
vary
may
de
magnitu
the
h
althoug
t
constan
is maintai ned
''';M

where Vc = variable.

p = Vc sin wI

(4-249)

Then from Eqs. 4-182, 4-183, and 4-187 these

3#

ELECTROMECHANICAL ENERGY CONVERSION

TWO-PHASE TRANSFORMATIONS

phase voltages can be expressed in steady-state symmetrical component

form as

V~

and

v:.

=-

:2(VR + Ve)

:2

(4-250)

:-.

T
..Java

= ~ wL"ZR'
{ [R'RS 2
"

- [R'R S - (2 -

sVs

sw2(L~L~

s)w2(L:L~
,.....

( T)

+ w2(sL~RS + L~K)2

(2 - S)V~2
/'

~,~

w 2 [(2

= ~ L,,2 R ,[S(VR +

s)L~RS + L~R'J2

.Nt"'"

"

c.

(4-253)

'" "

+ 2VRVc )

(R s)2

VeP]

(2 - s)(V~ - 2VR VC ) ]

+ (wC.v

(4-254)

(4-255)

VRnw m)

(4-256)

::;...-

2w V c

(<)mIT-O

=n

"c

nwL:;VRVc
(T,,,)avg!wm-o = R'(R s )2 + (wL:)2J

,~~

(4-257)
(4-258)

..

and with the slope

,,(I

..\.

(n 2/2)V;2 V-;

o(T.,,)ava

~\

aU)m

In the region around zero speed (s = I) the factor (2 - s) is very nearly

unity. This makes the first two conditions of Eqs. 4-253 equivalent and.
the last two equivalent. How well the first two inequalities of Eqs. 4-253
are satisfied depends to a large extent on the coefficient of coupling k
which is given by

kVL~L~

Equation 4-256 describes a family of straight lines as shown in Fig. 4-27.

For instance, for a particular control voltage Ve , the torque-speed curve
is described by a straight line with the intercepts

(2 - s )L~R' ~ L~R' ,

~1.'HU,,:

(2 - s)(VR

+ (wL~R')2

(n/2)L::VR
(T.Java = K[(R S )2 + (wL:)2J (2wVc

sw2(L:L~ - L~') ~ R'Rs

.,',1

v,y -

(R'Rs)2

Equation 4-255 is now linear with respect to mechanical speed and with
respect to control voltage, By substituting for the slip s = (OJ - nwm)/w
and by rearranging thc result, the more recognizable form is obtained:

i~~

The slip s is linear with mechanical speed; thus, to linearize Eg. 4-252 with
respect to mechanical speed the terms in the denominator which contain
the slip s must be negligible. This requires that .. ~

(2 - s)w2(L~L~ - L:;) ~ R'R'

sL'Rs -f L S R r y'-

4w "

_ ~ WL;;[S(V~
(T,,,).v& - 4 R'

>.~-

(4-252)

.r'

Now assume operation with small control voltages Vc so that second-order'

tcrms in V r can be discarded. Then

- L:;2)]2

- L:;)]2

:,

(4-251)

Vc)

To find the linearized torque equation, first multiply the numerator and
denominator of the first term in Eq. 4-234 by S2 and of the second term
by (2 - s)2 to obtain

Using the inequalities of Eqs. 4-253 and substituting from Eqs. 4-250
and 4-251 into 4-252 gives

"

' " Il

(VR

345

R'(R S )2

(wL~)2]

! ;

(4-259)

Ii

The linearized torque-speed characteristic of Eq. 4-256 can be useful

in system analysis and design; however, the accuracy of the results is
limited by the assumptions that were necessary in the linearization process.
For any particular case the accuracy of the approximations can be
checked to obtain a measure of the accuracy of Eq. 4-256.

= L:;

The nearer k is to unity, the better satisfied are the first two inequalities.
The last two conditions of Eqs. 4-253 depend on parameter values which
can be adjusted if linearity is desired. In all of the inequalities of Eqs.
4-253 the rotor resistance R' appears on thc right-hand side. It was stated
earlier that servomotors have relatively high rotor resistance; thus the
design tends to satisfy the inequalities of Egs. 4-253.

~~

'.""
';'-:

"

4.7

Summary of Transformations

Five sets of variables that are useful for describing the generalized
machine have been derived in this chapter. Three sets of variables, a{J,
dq, and
were real; two sets, + - and lb, were complex. Each set of

yo.

'-iI.'

""
"

.J

346

ELECTROMECHANICAL ENER.GY CONVERSION

TWOPHASE TRANSFORMATIONS

variables was shown to have certain advantages which allowed simplification of the analysis of a particular confIguration of the generalized
machine. The u13-dq variables were shown to be manciatory for treating
conventional commutator machines and useful for studying synchronous
machines. The ufJ-yr. components were needed to study repulsion
machines, and the + - and fb variables were useful in the analysis of
induction machines. Each of the transformations can be used for many
cases not treated in this chapter. It is also possible to define new transformations which will help in the solution of specific problems. The five
sets of variables selected for detailed treatment are the principal ones used
in the analysis of electric machines, and they will have repeated u,efulness
in many ditll:rent applications. These transformations are summarized
in Fig. 4-2!l and Table 4-3 (pages 34~-349).

V
I:~

r-

"'f-+-

t 0
~

10

't

'"

~
I~~ I

'j'
~

,--------,

....

,---,

-I~

'~IL>

-I~

-i~

1"

I~~II

.....

~I
" -eS' ~

Jf4

r+ -e-I

.'"

<II

..c:

l I

.. ..
'"

1"

:'"2
I

:.I

~.

-+,

,,,,.'
'1'

~,~
1.5
01'"

'0

.: i ~

,itJ,

.:, ~o

~-"

1')"

't'

VI

...;

;;,.,

j;i

.~

J~

-~

.~

..

~I-I ,~

-0

-e- I I
..... I-e~

i's:!

'.,.

.~ i ~

9-

r:;;-;l

1..:' .. 1

....

,i,

E
....
.2
'"
c:
~

-e-I-e~I;

.~

/-+

i)

~ I -=
8 i'~

r-----l

,J:i

- ...,....

,.
"t... ,',

>,;~

347

-1-1:

'.!:

""'1"
I

,-

I-

:;; !

V'I I

o I .S'

~I

"'lI'

..

'

I~
I
I

-0

'-"'w

-~,~

-0- i

,~

~~,

i)

I ';;"i~
i=
'/1
0,,-,,

~
Y

.-.

~I:

S'~ ~
l-J-~-=
"I,

-@
"

.. ..
rm

~I"B;
<: 1<:

.." ..
....
t

~,

~i~

)(

.-

-t ,~

II

"~'>

>:.:!
w

""

.."

[
E
:;l
til

cO

t>ll
ii:

~'

"-TABLE 4-3.

Summary of Equations of Motion for Generalized Machines

(a) Statoicifl, iotor-{i{3

[R'

IT,.

vB
v'a
V

L~2)
R'

+ L")
cos n-l1J2
'f'

p(L"
~

p(L~ + L::) cos n1>

T =

+ p(L~ +

+ p(L~ -

L~)

p(L:: - L::2) sin n1>

R' + p(L'1 J
+"L'
cos 2n't'
-l-)
"2

p(L"
- L")
sin n-lJ,J.
J.l2
'f'

- p(L;: + L::2) sin n1>

- pL~2 sin

p(L:: - L::2) cos n1>

(J{> + a~ + K1 - nL::[(I:i8 - i~i~) cos ne/> - (i~i~ + i~iP sin ntPJ

nL::2[(i~ip

i~i~) cos ne/>

- p(L:: + L::2) sin n</>

p(L;: - L::) cos n1>

(i~i~ -

'0'osa

'fJ

.,

- pL~2 sin 2ne/>

2n.p

R'

+ p(L~ -

L~ cos

ct

'a

m
m

(4-21 c)

n- j

i~

2114

:;%J

o
iJiP sin n</>] - nL~2{[(i~)Z - (i;)2] sin 2114> - 2i~ip cos 2n</>}

3:

rn

(4-22d)

(b) Stator afl, rotor dq

': [R' + p(L; + L;,)

vp

vS

v~
v~

I
I

a~

L~)

P(L;-

0
K

+ p(L~ +

- n~(L~

p(L:: - L::2) "

+ K1> - nL"(i'i'
fJ 4 -

is;')
a

I'

L~2)

+ L~2)

L~,l

+ p(L~ -

'a

n~(L~ - L~2)
R'

os

)-

ip
id

L~2)

(4-5Ia)

:;%J

j~

C)

-<

+ nL""2(isi'
tJ 4 + iSi')
a + 'nL'1'2 (2i'l"')
4

(4-5Ib)

o
z

ap. rotor yS--.-Smooth-Air-Gap Machine Only

(C) Stator
"""'.'

<

.~

R'

+ LtJ1

nf)n~J
T =

""~-:"'~

..

,'

J1> + "a +

L:;[ - (sin mp)p - (cos

+ (cos nf)n~J
R' + L;.p
- L;.n(4> - '1)

Kcj) - nL::[(i8i; - i~iD cos nifJ - (i~i~

L~n(~ -

R'

:;%J

nf)~I~J i~

(5

i~

L::[(cos nf)p - (sin nf)nf]

L::[(sin nf)p

L;:[(cos nf)p - (sin !ll/J)n~]

L;:[-(sinnf)p - (cosn"')n~]

"'"",

L::[(cos nf)p - (sin !llp)n~]

o
R' + L~
L::[(sin nf)p + (cos n.p)n~]

,'-',"

0
L::[(cosn')p - (sin

T = J~

L~)

n~(L"
- L")
I'
1'2

+ L;:)
-n~(L"I' + L")
1'2

v'f

p(L::

+ p(L~ -

R'

p(L::

v,j

u:.

)-

VI

,;

+ L~p

i~

(4-1300)
(4-136b)

+ i~i;) sin nf]

"'~ ~' ..'"'jiG'rt.., ~rj.::t'~~AJi~:'-i'7:'~i~:'

..... ~~"".-

.',: d~" v1tr1e'

'

TABLE 4-3 (Continued)

+ -, rotor +-

(d) Stator

vs,.

R' + L;.p

v~

L~lP

, v'y -v~

= J~ +

a~

R' +

L~p

L"e-Jnc'(p - jn)

L::r1n(p - jn)

;::2e1n,(p + jn)

L:: ein(p + jn~)

+ Ktj>

L::e,n4>(p + jn~)

L~2P

L::2e;n(p

+ L~p
L',-2 e;~n(p + j2n)
R'

[R' + L:,

if,
v~

L~2P

v.

L::r;nl>(p - jn'hl
L::e-;n(p - jn~)
L~2e-i2n4>(p - j2.n~)

R'

+ L~p

i~

L::(p - jn~)

!:':l(P

T = J{>

+ jn~)

i!..
i'+

(4-194a)

o
I

'J

I
)-

i'_

- jn{L::[(i'+*i'r)ein~ - (i'ri';.*)e-;n] + L::2[(isri't)ein</> - (i,;!,i'+*)e- i"] + L~2[(i't)2el2no/> - (i'+)2e-12~J)

(e) Stator

v(

+ jn~)

t/)

(4-1 94b)

-j
:;%J

)-

+ -. rotor jb

V>

-n

L~2P

R'

L~p

L:;Z<p - jnrf,)
L::(p

+ jfl~)

L~" J'~

L::p

L"p

L~2P

R'

L~Z<p + jn~)

~I

+ L~(p + jn~2..

+ a~ + K1> - jn{L:;(i';!'lf - i'ri(*) + L::/i~lf - ''':'f*) + L~2[(i/)2 -

3:

is

L~Z<p ~ jn~)

L~(p - jn~)

o:;%J
(4-21Oa)

i'

(i(*)l])

(5
Z
en

(4-2 lOb)

""
$

C'

\'

350

ELECTROMECHANICAL ENERGY CONVERSION

PROBLEMS
4-1.

(a) Show that any 2 x 2, real, power-invariant transformation must ~

of the form

.<:1

r
c~sB
~ smB

+ si~

co~

= :!....Yt'
dt

e
where

e=
=

~
II
1

~'l

cos wl\
I - 2 sin wI cos ~

+ 2 sin wt cos wI

sin2

~n2 wI - cos 2 wt

wI -

where

e,

s1

= ~'l
e~

e",

= EswI
~wI

to obtain the new equation

e'

=t
'

~
11

i;

-sin

:tl

cos~

f:::

j'

3si,

with your previously assumed conventions) to establish a magnetic field in the

gap identical with the gap field that exists in part (a).

= J cos wI
= I sin wI

With the rotor circuits unexcited, the resulting air-gap flux wave is a constantamplitude sinusoidally distributed space wave which rotates around the air
gap in the positive <p direction with the angular speed w.
Now, assume that the rotor is turning at an angular velocity W m (<!> = wml + S).
and that the stator circuits are open.
Find the rotor current excitation through slip rings required to give exactly
the same air-gap flux wave as that described above for the given stator excitation
(identical in magnitude, time, and space).

make the real, power-invariant transformation

s1e'
s1l'

TWOPHASE TRANSFORMATIONS

i~

12

i~

f. = .

e2

,' .

4-3. A two-pole, two-phase, smooth-rotor machine with sinusoidally

distributed windings can be described in terms of coil variables by rx{3 components
on both rotor and stator. The machine in this problem has an equal number
of turns on rotor and stator windings, and, in addition, L~ "" L~. The stator
circuits are excited by balanced two-phase currents:

or its transpose. Your answer must consist of two parts: f1rst, showing that
the above:: transformation has the required property; and second, showing that
no other transformation does.
(b) In the special case of the matrix equation

"

Jr(

: _,..~< .

\i

4-4. A smooth air-gap machine with symmetrical windings is run as an

induction motor with balanced sinusoidal excitation on the stator.
(a) Place suitable constraints on the dq impedance matrix to convert the
general equations to the special case of an induction machine.
(b) Assume that the stator windings are excited by balanced two-phase
Current sources and that operation is at constant speed. Derive an expressiol)
for torque as a function of speed, stator current amplitude, and circuit
parameters.
(c) Determine the value of rotor resistance which maximizes the torque with
current amplitude held fixed. What is the maximum torque possible under
these conditions? Will an induction motor run with zero rotor resistance?
4-5. The generalized nonsalient machine is described in the a'{3s, d'q' frame
of reference through the equations:
.
!" ,

= Y' :!.... (., + ~'t"

v:J.dt

T.

dt

Carry out all the matrix multiplications and differentiations to determine

2" and $'.

4-2. The rotor of a two-phase machine (smooth stator and rotor) is excited
by two sets of sinusoidal brushes located 90" apart. The stator terminals are
open-circuited. One brush terminal pair is connected to a sinusoidal current
source
i = I cos wI

' ;l>~ i I

in which
i~

v'a

I'j

v~t.J4 =

V~

V:,

p.,

a[J,4f =

j'

,-'!~i t ~<,j ;vr-, r; !.

[J

i'
i'4

V'4

R' + L'"p

(a) Find the open-circuit voltage appearing at the other brush terminal pair,

when the shaft is turning at an angular velocity W rn Draw a jigurc and define
polarities of ['oltages and C/lrrents and direction of rotation assumed hy you.
(h) C\lOsider a two-phase machine identical with the above but containing
sinUSOidally distributed rotor phase wil1lhngs that are exclll::d fr0lTI slip rings.
Find the currents that must be supplied to phases a and f3 (def1ned to conform

= ;Z~~.d4 r'~iJ.4f
= L:;( - i~j; + iAm

;Z~~.d. ""

L::p
R' +

L::p
f-------------

-L"t/>
I'

L~p

L"~
,.

~::-;-l

L':p
R'

L~p

-L:~

L~~
R' +

L~p

352

ELECTROMECHANICAL ENERGY CONVERSION

TWO-PHASE TRANSFORMATIONS

We now wish to go to a new reference frame designated as 1'2', ['b' which is

related to the a'f3', d'q' frame through the transformations

4-7. The doc series motor shown schematically in Fig. 4P-7 drives a fan.
When the applied voltage V, = 400 volts and the armature current I. == 15 amps,
the fan runs at a speed of 450 rpm. Assuming that the torque required to drive
the fan varies as the square of the speed, determine the voltage and current
needed to drive the fan at 600 rpm.
The internal resistance R. + R r of the motor is equal to I ohm.

v~t,df = dvi'{,,.

1 e:J,df

= de~'I,fb

in which

'it;

V2

,~~~;

j
l----1-1

1 I -j

d =

353

..

+
.~Il'

- j

v~

v,

"

~li,fb

'S

'2

.,

If
Ib

If the voltages and currents of the two reference frames are related as
expressed above, find through matrix operations the elements of the matrix
.2'i'2,fb of the equation
vtf,fb = .2'i2.f~ff,fb
(0)

lnterms of L~, L;" L;;, R', and R',

(b) Find the torque T( in terms of the new currents if, il,

if, and

.""",.

'~(~

\~ .

;;- h

i$

= If cos wt
= l' sin w/

i~

= 0 (coil open)

1-:'

:<1..
.:; h '

_,Li.

'tJ)

'"'idw
Fig.4P-7.

4-8. A d-c motor M has its armature permanently connected to a source S

as shown in Fig. 4P-8a, The volt-ampere characteristic of the source is shown
in Fig. 4P-8b. The motor field f is separately excited from a voltage source E f
as shown.
Motor armature resistance R. = 0.5 ohm
Motor armature Inductance negligible
Motor field resistance R r = 50 ohms
Motor field inductance L f = 50 henrys
Motor torque at 5 amp field current and 100 amp
armature current = 200 newton-meters
Motor load-pure inertia. Total moment of inertia J of
load and armature = 10 kg-meters 2

't1,'
~

~. ,.~,~:_;.r

, i.~,

i-y_

I;

v;,

i,.

i 4-6.
A two-phase. nonsalient, two-pole, synchronous energy converter has
(n'its four windings the currents
i

b--------:-,

v,

.,

v;

la",

.,
11

vi
vti'.,. =

~',r

Neglect magnetic saturation effects in the motor.

is closed at t = O.

The machine is running in the positive direction so that the position angle of
the rotor is
t/> = wi + 8

"

100 amp ----~

(a) Express the currents in the ct.'f3', d'q' frame of reference.

(b) The .2' matrix for a nonsalient machine is given in Prob. 4-5. Write out
the expression for the instantaneous terminal voltage across the ct.' coil in
terms of w, IS, 1', and the resistances and inductances of the !l'matrix.
(c) Write an equation in complex rms voltages and currents which is
equivalent to the result of part (b). Draw the corresponding phasor diagram
for 0 = -45 0 (motor action).
(d) Realizing that P av = Re (/::",V,ms), find the Qverage power delivered to
the ct.' winding.
(e) Find the average power delivered to the ct.' winding.

".;
1-,

;:~;

"~~~
,,~.,

i:it

I
'.c,;.'

~,t' "

:\

L-

The motor field switch K

"161

e,,,

Q.

i.

(a)

Er =

250 volts

Flg.4P-8.

..

~;,

"~'-'

I~...

~l!.

500 volts
(b)

354

ELECTROMECHANICAL ENERGY CONVERSION

TWO-PHASE TRANSFORMATIONS

Derive an expression, with numerical values, for the speed in radians per
second as a function of time in seconds. Sketch this curve roughly to scale.
Indicate on it the final steady-stale speed, and lhe speed and time at which
the break point in the source characteristic is reached. Compute the time
required for the motor to reach approximately 96% of its final speed.
4-9. The machine shown schematically in Fig. 4P-9 has a sinusoidally
distributed stator winding and sinusoidally shaped brushes in contact with the
rotor conductors. The brushes are fixed with respect to the stator so the
magnetic axes have the relative orientation shown in the figure.

: Rotor magnetic axis

I

Stator-magnetic
axis

~
__

1+
I

----

V,

u.-::

are connected to a resistive load Rr...

The generator parumeters arc:

The system is operating in the steady slale.

Resistance of armature R ... 0

Self-inductance of armature L. = 0
Speed constant = Ga(
Resistance of field = R,
Self-inductance of field ... L,

I
j

./.

(a)
(b)
(c)
(d)

Find the instantaneous current 'r.. flowing in the load.

Find the instantaneous voltage V( across the field terminals.
Find the instantaneous electromagnetic torque produc<:d by the generator.
Find the instantaneous power converted from mechanical to electrical

form.

(e)

Find the average power converted from mechanical to electrical form.

.,

s~

Direction of flux
3 positive If

for

in:
--

: V,

Fig.4P-9.

If

+.

uf

8<I>

L~

+
VL

::-

R I

The machine is describable by the usual circuit parameters:

resistance R'
Stator {
.
,self-mductance

355

Direction of speed voltage

due to POSitive If
and positive <I>

\:-" .
Flg.4P-IO.

V'"

resistance Rr
Rotor

t'

self-inductance L~
speed coefficient L;;n~

4-11. A two-pole, four-brush, smooth-air-gap machine is to be operated in

the following manner. The rotor is excited through the brushes as shown in
Fig.4P-II. The stator circuits are unexcited (open-circuited).
(a) What is the electromagnetic torque developed in this machine configuration'! Write the differential equations which determine the volt-ampere
relations in the machine using the dq system of variables.
(b) With constant rotor speed </> = W m, a set of balanced two-phase voltages
is applied to the brushes:

The rotor is driven at constant speed W m by a mechanical speed source.

(a) With switch 52 open, switch 51 is closed at I = O. Find the voltage u,
as a function of time. Sketch the result.
(h) With switch .s\ closed and the system operating in the steady state,
switch S2 is closed at t = O. Find the voltage VI as a function of time. Sketch
the result.
(c) For the conditions of part (b) find the electromagnetic torque produced
by the machine as a function of time. Sketch the result.

VJ
"",_

.)

Do

v~'"

Vcoswl

V sin wI

Determine the magnitudes I;, I: of the steady-state currents i J and i;. Explain
in one or two sentences the characteristics of this mode of opl:ration whcn:

4-10. The doc generator shown in Fig. 4P-1O is driven by a speed source
whose value varies sinusoidally
</> =

(1)

+ D 1 sin wt

= w ...

(2) w < w.

The field of the generator is excited by a doc source I( and the armature terminals

(3) w > w,..

(0) rf, for rated full-load operation, the rms current in the motor is I = 1.63
amp, calculate the speed mutual coefficient

(c) Consider next constant speed (</> = w m ), steady-state doc operation when
= V is applied to the quadrature brushes and a variable
a constant voltage

v;

L~}

load is placed across the direct-axis brushes. Neglecting the rotor resistive
drops compared with the speed voltages, find the load current i~. Briefly
descri~ thc characteristics of this doc mode of operation.

.,

.,
q

11;

6.

4-13. A generalized four-brush, two-pole. smooth-air-gap machine is to be

operated with the brush carriage driven by a mechanical source. The a{3-yS
equations of Eqs. 4-1360 and h descrihe this mode of opcrat,,'n. Impose the
following constraints on this set of equations:

v~=
v~

L. = 3.12 henrys
L c = 3.12 henrys
L r = 0.Q7 henry

;
i

'1'

-.t

-:.;'~~-<

,I' "t-,.
~9i,1

:Jlllm
. 'I

,~,'

wbf, Wb

.r

ld

!i;~.~\':.'~ l.;jt!~ 'A/~(t;.

S~ries

field
Winding f

= 115 volts

'

h'_

\' I ~

t,~~' "~~

;,~;...;.. ~~,J,

;.'r~,;'~l~"

(60-cycle)

constant
constant

wmf, W m

.p

I,l..

j>-' ",,; .,-\,

"'1

4-14. A wound rotor equipped with a commutated winding is placed in a

cylindrical unwound stator which merely serves to compkte the magnetic
circuit. The commutator has two sets of brushes in quadrature. A two-pole
equivalent rotor illustrated in Fig. 4P-I4 is defined by the resistance R r and
inductance L' between opposite brushes.
(0) Write the volt-ampere relations governing the transient behavior of this
machine using the dq system of real variables, and then show in detail how to

LtU; "" O.99L.

:-:I.~ ;

I' = constant

==

'" =

Assume steady-state operation.

(0) Find the instantaneous currents i; and iJ. Are the frequencies of these
currents functions of rotor speed w m ? of brush-carriage speed Wb? Are these'
currer! amplitudes functions of rotor speed w m ? Of brush-carriage speed w.?
(b) For fixed RJ, what brush-carriage speed produces maximum torque?
(c) Evaluate this maximum torque. Is it a function of R I? of Wb? of W m ?
(d) Can this machine run successfully as a motor? If so, what will be the
sign of W m if Wb is positive?

.. b +

4-12. A universal motor is a series-excited commutator machine (shown

schematically in Fig. 4P-12) which can run on either direct or alternating current.
Determine for a-c excitation the steady-state characteristics of a t-hp universal
motor having the following rating: two-pole, lIS-volt, 60-cycle; full-load
speed = 5400 rpm. The following constants are also known
R c = 3 ohms
R r = 9 ohms'

i~

-R\i:

= -R t i 8

i; =

Flg.4P-II.

R. = 3 ohms

= Gar

(Assume sinusoidal air-gap flux distribution.) Also for full-load steady-state

conditions, find the electrical power supplied to the motor, the electromagnetic
torque developed, and the rower factor.
(hl Sketch curves of electrical power, torque, and power factor versus speed.
Brietly explain the effects of the compensating winding on these steady-state
characteristics.

ur

357

TWO-PHASE TRANSFORMATIONS

ELECTROMECHANICAL ENERGY CONVERSION

356

r
u'

,;Ii~

(supply)
. r

Lq

"1',
"
!

lit,

~v~~6+

, I

f ..

Fig.4P-14.

Flg.4P-I1.
?'

'i '('

~n

359

ELECTROMECHANICAL ENERGY CONVERSiON

TWOPHASE TRANSFORMATIONS

transform these relations to an equivalent set expressed in fh complex variables.

What is the electromagnetic torque developed in this machine?
(b) Consider constant speed, steady-state doc operation. A constant voltage
v~ = Vm is applied to the quadrature brushes while the direct brushes are
connected to a variable load. Neglecting the resistive drops compared with
the speed voltages, find the load current id Describe in one sentence the
characteristics of this type of operation. (Use dq variables.)
(c) Consider constant speed, steady-state a-c operation. A balanced set of
two-phase voltages are applied to the brushes:

require that the stator voltage on the open-circuited phase be zero. It should
also be noted that the equivalent circuits will be the same for a-c steady-state
quantities as for instantaneous.)

358

v~ =

v~

V.. cos wI

4-16. When a two-phase servomotor is driven from some external source of

mechanical energy and one phase is excited at rated voltage and frequency, the
voltage developed across the unexcited terminals is proportional to speed under
some conditions. For a two-phase servomotor find the voltage-speed relationship. The load on the unexcited phase may be considered an intinitc
imred,tnce. What two sets of conditions for rotor parameters will yield linear
voltage-speed characteristics?'

Vm sinwt

Derive the steady-state rotor volt-ampere relation expressed in the conventional

phasor notation of sinusoidal quantities. Describe in one sentence the
characteristics of this mode of operation when tJ. > w. (Use fb variables.)

4-17. A two-ph,lse, cylindrical-rotor, induction motor (two-phase, sinusoidally distributed windings) can be used as a dynamic brake by putting the
following constraints.
(1) Stator phase a supplied by a d-c current source of magnitude [I.
(2) Stator phase f3 open-circuited.
,
(3) Rotor phases a and fJ short-circuited by a resistance of Rohms 'at their
terminals.
".
Assuming that the parameters of the machine are:
Rotor resistance = 0
"
.." n'
Rotor self-inductance = L'
" ,
I,

4-15. The two-phase, two-pole servomotor shown schematically in Fig.

4P-15 is to be studied. The constants are:
R'
R'

= 60 ohms
= 450 ohms

w(L~ - L~)

230 ohms

= 385 ohms
= 21T400 rad/sec

wL:J

w(L~ - L~) = 85 ohms

(a) Compute the steady-state average output torque in newton-meters for a

speed of 6000 rpm and an applied voltage of 230 volts rms on the reference
phase and 115 volts rms on the control phase, the voltages being in time
quadrature.
l'I:Qi:,

~
:, .. '1;,'

0
iRtO
0

->dJ tv

"i

'V

,:),~

.j:. '

,C;"

0
0

1$

:J>1"

; '" ", '

4-18. A two-phase servomotor can be used as an accelerometer by applying

direct current to one stator phase and taking the signal off the other stator phase.
Derive the input-output characteristic of the accelerometer. In your analysis
assume that purely electric transients die out much faster than electrlr
mechanical transients.
'~1 t.

: iT >,

Fig.4P-IS.

"ll

La

flU

f'

("
Maximum value of rotor to stator mutual inductance = LI'
and that the velocity of the rotor is wand is constant:
(0) Derive an equivalent circuit for the rotor currents.
,
(b) Find the average power dissipated in the resistance R.
"',
,
(c) Find the average torque as a function of w (sketch it). Compare your
result with answer (b).
(d) Find the value of the maximum average torque and the speed at which
it occurs.

:,. .... ;,:.

(b) Servomotors must not "single-phase" when the control signal is removed.
This says that the ll1otor must develop torque oppllsite in sign to that developed
when a .:on{w! signal is applied. For a speed of 6000 rpm find the average
steady-state torque developed when the control is opened (by opening the
switch in the figure) ,lOd the reference phase still has 230 volts applied. (Note
that opening the switch only constrains one stator current to be zero; it does not

'~(1"3

;1 ....

I;t:.t~

';.t:'l'l

.1>\1:1'1<'

'. !J<j

~l1j

;:J

. 0 c'

FUNDAMENTALS OF SYSTEM DYNAMICS

"':1
i

Fundamentals

:(

};Jt

of

<1

"'ol'

~i

System Dynamics
, .,".

, i

,.,' "t"

,J.'i.&,,~

;,!

5.0 ' Introduction

Electromechanical energy converters are often used in interconnected
systems. The successful operation of an interconnected system depends
strongly on its dynamic behavior, which is affected by the interaction of the
various components. Examples of systems containing electromechanical
energy converters are widespread. Interconnected synchronous alter~
nators and their associated passive and active loads coupled through power
transmission lines to the alternators make up power" grids" which supply
the nation's electrical power. These" grids" must be able to maintain
essentially unvarying line voltage and frequency under widely varying
load conditions; hence regulators are used. The design and interconnection ofthc machine systems and their regulators require a knowledge
of the dynamic behavior of the components that make up the systems.
In aircraft. electrical power requirements are met by generators connected
to the engines which are variable speed power sources. Requirements of
unvarying line voltage and frequency must be met--in this case under
severe weight and space limitations. In military fire control systems
machines are used to position guns and radar to extraordinary degrees of
accuracy in order to track and hit rapidly moving targets. In industrial
processes, constant speed or constant torque electric drives are used in
processing sheet steel or cloth or paper. In the high-fidelity reproduction
of sound the microphone in recording and the speaker in playing back
must, in conjunction with associated circuitry, have extraordinary dynamic
behavior over a wide range of frequencies and power levels. In' all these
examples dynamic behavior limits the accuracies that can be achieved

'~

J.
-:;,

'r.

",'

'j'.

ll1;"

:l~\'-'

.;~- ~;;'

');

"

,go
,J~

360
,

'if:'.~'.'
L

and indicates the properties and parameters that affect the stability of
these complex systems.
It was illustrated in Chaps. 1 and 2 that electromechanical energy
converters are described in general by nonlinear differential equations;
consequently, any system containing such a device will be nonlinear. The
solution .of nonlinear differential equations is difficult and often requires
the use 'of computcrs as indicated in Chap. 2. Whenever the equations
of a system are linear with constant coefficients, either because of special
constraints or because of reasonable approximations, the powerful
analytic tools of linear feedback control theory can be used to obtain
valuable information about system dynamics, stability, accuracy, etc. In
this chapter the fundamentals of linear feedback control theory will be
covered, along with other considerations that are essential in the study
of systems containing interconnected electromechanical energy converters.
Systems in which nonlinearities cannot be ignored will be treated as
examples in later chapters.
Feedback control systems are often thought of as information-handling
systems alone. However, this is only part of the picture because powerhandling capabilities are essential too. For instance, suppose a mass M
is to G" accelerated at a rate dvjdt from an initial velocity v. The power
required to provide the acceleration is Mv(dv/dt). Now suppose it is
desired to double the acceleration rate. This means that the instantaneous
power into the moving mass must be doubled. The device which supplies
the power must be capable of supplying the extra power for the increased
acceleration. If this power requircment is ignored. a system design on
the basis of information handling alone may lead to a system that will
not meet performance specifications.

5.1

t:

,:f:i.f

361

"

The Feedback Principle

In this and the following sections, attention will be focused on the

behavior and simple design principles of feedback control systems.
Complex power systems, industrial process systems, or military fire control
systems all have the same fundamental objective, namely, a specified task
is to be accomplished which may involve the utilization of large quantities
of energy, but the task is to be accomplished under controlled conditions.
In all these systems, feedback is used to monitor and control performance.
The term" feedback," when used in connection with automatic control
systems such as servomechanisms, implies the transmission of information
regarding the behavior of one part of a system (usually ncar the output
and at a relatively high power level) to another part of the system (usually

'V

"'.t.

362

ELECTROMECHANICAL ENERGY CONVERSION

near the input and at a lower power level) in order to alter the performance
of the entire system. In the simplest case, feedback implies comparing
the output of a system with its input to ascertain how \'tell the system is
doing its job. This basic aspect of the feedback system distinguishes 'a
closed-loop syskm from an open-loop or calibrated system. Rut feed
back performs other functions besides comparing an actual quantity and
a desired quantity. The presence of feedback in a system alters both
the static and dynamic response; it may make the system more accurate,
faster, or smoother, or introduce instabilities that may make the system
unstable.
In more complicated systems several feedback paths can be found, each
receiving an indication of the behavior of some portion of the system,
and each introducing its own information at other points in the system.
Although such systems appear ex.tremely complicated when put into block
diagram form, they can usually be reduced to a single feedback loop,
with one or two blocks, the method of reducing networks to this state
, being shown subsequently. Such a procedure may not be obvious from
the original block diagram, but it is possible to manipulate the blocks
to form one equation which describes the complicated system from one
end to the other.
The elements of a feedback control system are demonstrated by the
simple system shown in Fig. 5-1 in which the block diagram notation

.'!t.)~;~'

\:)

l.t;:

Fig. 5-1. Elementary closed

loop system.

. "'"c'
(

'.~

" .

. ;1' l'

FUNDAMENTALS OF SYSTEM DYNAMICS

. 363

changes in the controlled: variable and corrective action will occur quite
rapidly.

or

A more sophisticated. example is shown in Fig. 5-2. The syste'm

Fig. 5-2 differs from that of Fig. 5-1 in that an uncontrolled variable
disturbs the system. Here the performance requirement would be to
i

~
R+_"~C

Fig. 5-2. Closed-loop system with disturb

ance.

maintain a specified relation between C and R while keeping C insensitive

to the disturbance U. Thus, if the system of Fig. 5-2 could be designed
so that the combined 'amplification of the components 0 1 and G2 was
large while the amplification of the component G2 was small, then C
would be sensitive to errors but insensitive to disturbances.
.
At this point the advantages of feedback or closed-loop control as
opposed to open-loop control should be examined on a static basis.
Open-loop systems are shown in Figs. 5-3 and 5-4. Consider the systems
. )

'J t"?:t

ofttff-

J~ ",,;;'~;"';j

:~

J,

~
'4

c.

.(,10

~ r::l..- ',,,
Fig. 5-3. Open-loop system.
~".,'YI'!U".>1I",

.....-,__..,.

__hJ

.:>.'l/I~

'ft'"

"

developed in Chap. 2 is used. In Fig. 5-1, C represents a controlled

variable which is required to maintain a specified relation to a reference
variable R. The system operates by measuring the controlled variable and
feeding back the measurement to the reference. A comparison between
the reference and the controlled variable indicates whether an error E
exists. The error E is used to correct the controlled variable so as to
reduce the error to as small a value as is practicable. The performance
of the system shown in Fig. 5-1 is completely dependent upon the
properties of the component n:lating the controlled variable C to the error
E. The sensitivity of the system to error can be increased by inserting
amplification into the component G. Thus, small errors will produce large

Fig. 5-4. Open-loop system with disturb,:

ance.

1& )."

'\

!'t.,

of Figs. 5-1, 5-2, 5-3, and 5-4, in the case where all the indicated transfer
functions are pure gain or amplification with no dynamic effects present.
The desired relation between C and R is that C = R at all times. Firse"';.
consider what happens in Figs. 5-1 and 5-3 if, for some reason, the gain '.'
changes. In the closed.loop system of Fig. 5-1 the relation between C
and R is

R=I+O

(5-1)',

ELECTROMECHANICAL ENERGY CONVERSION

FUNDAMENTALS OF SYSTEM DYNAMICS

To meet the specification C = R, G must be very large. If G is subject

to change, then the fractional change in the ratio CfR is found to be:

Of course, closed-loop systems are not limited to the simple types

shown in Figs. 5-1 and 5-2. There can be transfer functions in the feed
back path, multiple feedback paths. and multiple feed-forward paths.
The exact configuration depends on the equations describing the system.
No matter how complex a system is, however, its block diagram can always
be reduced to the simple forms of Fig. 5-1 or 5-2 by the techniques given
,c.,
in Sec. 2.1.5.

364

d(CjR) = _1_
(CjR)
I + G

(dG)

(5-2)

The quantity 1/(1 + G) is small and the fractional change in Cj R is

much less than the fractional change in G. In the open-loop system of
Fig. 5-3, the requirement that C = R forces G
1. If G is subject to
change, then the fractional change in the ratio CjR is found to be:

d( CfR) =
(CfR)

dG

IJS
It

5.2

(5-3)

U=

Graphical Methods of Representing Transfer Functions

Comparison of Eqs. 5-2 and 5-3 shows that a feedback system is much
more insensitive to changes in system parameters than an open-loop
system.
The effect of disturbances in the two types of systems (closed-loop and
open-loop) can be demonstrated by examining Figs. 5-2 and 5-4. In both
figures the error will be defined as E = R - C. The specification that
C = R requires that must be as small as possible. In Fig. 5-2, super
position can be used to ascertain the effect of the disturbance U alone.
For the closed-loop system of Fig. 5-2, the ratio 1 U is found to be
(witn R = 0)
,"0

365

Go

GGz

Only systems that are describable by linear differential equations with

constant coefficients are being considered here. In [hc analysis and
synthesis of such systems the equations can be used in the time domain or
they can bc transformed to the frequency domain. The equations retain
the same amount of information whether they are in the time domain or
freq uency domain; con seq llenlly. si nce differential eq uations become
algebraic equations in the frequency domain. closed-loop system analysis
and syn thesis is usually performed in the frequency domain. In all of the
considnations to follow, each transfer function (e.g., G in Fig. 5-1) is
assumed to be the Laplace transform of the impulse response of the
component represented (see Sec. 2. I.3).
Not only is the frequency domain useful for studying absolute and
relative system stability, but also the steady-state sinusoidal frequency
response of a component is relatively easy to measure. Thus system
functions G(s) are often measured for s = jw, to give G(jw). When
transfer functions are specified in this way graphical techniques are often
useful in system studies. There are several graphical methods for plotting
system functions. Only two of these methods will be described and used
here.

~t

.j

'.

'\1.1..

(5-4)
'1"

The requirement that C = R makes G 1 Gz large and the requirement that

EIU be small makes Gz small; thus this is a small quantity. In Fig. 5-4,

~
R~"E

,"

,
j

One of the simplest and most useful methods for plotting transfer
functions is the Jog-modulus and angle versus log-frequency plots. The
log-modulus is defined as

Fig. 5-5. Open-loop system with disturbance

and error indication.

I:

does not appear directly but E can be indicated by redrawing the

diagram as in Fig. 5-5, from which the ratio 1 U is found to be (with
R = 0)
\\ill
,.....
"
.;:;Pj!
E "
.
".
..:. = G z
,id,..f
(5-5)

i... J~:/,.
.~

Comparison of Eqs. 5-4 and 5-5 shows that a feedback system will be
much less sensitive. to disturbances than an open-loop system.
~~,

I
-I

,i~,

Lm G(jw) = 20 lOgIOIG(jw)1 :;:

and is measured in decibels.
designated Ang G(jw).

li I

(5-6)

The angle of the transfer function is

An example of a plot of the transfer function

:t}l'j.

[<'.

(5-~

G(jw) = _,_1.

+ JWT

is shown in Fig. 5-6 by the solid curves.

'~'\.

'~~.

~':i:'?

,I!'

-to
fq

:1

."

'~

L.
,
366

ELECTROMECHANICAL ENERGY CONVERSION

1,

The log-modulus and angle versus log-frequency plot is useful when

cascaded elements are to be treated. The usefulness stems from the fact
that cascading (multiplication) of transfer functions means addition of
log amplitudes and of angles. Thus the transfer function GI(Jw) Gz(jw)
is represented by
Lm [GI(}w) GzUw) = 10 loglo GIUw) + 20 loglo Gz(jw)
= Lm G1Uw) + Lm GzUw)
Ang [GIUw) G2C}w)] = Ang GI(jw)

+ Ang Gz(jw)

FUNDAMENTALS OF SYSTEM DYNAMICS

For low frequencies (w -+ 0) Eq. 5-10 becomes

; ;

and for high frequencies (w -+

10

-0

:;

;::;,

Phase angle

-45

I
I

f'f,

1- 90

I
T

l
T

lQ'

'"'"'"tio
"0
'"
~

. ,iLl

.i',t

',m
f>'ll

Fig. 5-6. Actual and approximate plots of the transfer function

G(jw) = 1/(1 + jWT).

f'

Eqs. 5-8 and 5-9. This technique can be extended to any number oC~
cascaded transfer functions.
Another usd"ul aspect of the log-modulus versus log-frcquem:y plot is
that the log-modulus characteristic of a linear system, which is always a
ratio of polynomials in (jw), can be represented approximately by straight
line asymptotes. For instance, for the transfer function of Eq. 5-7

1
. .:.

1
"

~I

i:

...
~

G(jw) = _1_.

Lm G(jw) = 20 loglo

vi-=+-

W ZT

(5-13)

,. G(jw) ==

I.
1

The log-modulus

,;

(5-10)

which is termed the corner frequency or break frequency of the first-order

transfer function. Note further that the maximum difference between the
actual log-modulus curve and the asymptote in Fig. 5-6 is 3 decibels and
~:;'; OCcurs at the break frequency.
r:
Unfortunately, there is no method for approximating the Ang G(jw)
versus log-frequency plot by a series of straight lines. However, for a
first-order transfer function the angle always has the same value at each
,.
multiple of w = l/T. For instance, at w == liT the angle is -45 degrees.
At w = 1/2.,. and 21T the angle is - 26.6 degrees and - 63.4 degrees
respectively. As w -+ 0 the angle approaches zero and as w -+ 00 the
angle approaches - 90 degrees.
A development similar to that given above for the first-order transfer
function of Eq. 5-7 can be made for a second-order transfer function of
.the type
.

:';,

the asymptotes can be derived in the following manner.

is given by

= -20

~~';f'~

(5-7)

+ JWT

, 1"
(>~2)

w=
(5-14)
.,.

'&

-r

Frequency. radians/second

== -20 10g10 W7'

Equation 5-13 indicates that for every change of unity in loglo w, which
means a change by a factor of 10 (a decade) in w, the Lm G(jw) asymptote
changes by 20 decibels. Thus the straight-line asymptote for large
frequencies has a slope of - 20 decibels per decade.
The asym ptotes of Eqs. 5-11 and 5-12 are shown as dashed lines in
Fig. 5-6. Note that the two asymptotes intersect at

-20 f-------j----+---+-

(5-lf)

Eq. 5-10 becomes

d[Lm GUw)JI
d[log lO w] ,.,....,00

;'

aD

E
-'

I = 0

The slope of the asymptote of Eq. 5-12 is

"

10gIO

,.,....,00

(5-9)

VI

~ -10\

00)

Lm GUw) I

(5-8)

'j.

"I

= 20

Lm GUw)!
,.,....,0

When a log-modulus plot of each transfer function is available, the

cascaded transfer function is found by the simple additions indicated in'

--~-",_c.::--,...---

367

~
.~
1

+ 2R!::!.. +
W"

(Jw)

(5-15)

(un

For low frequencies the asymptote for the Jog-modulus curve is 0; for
large frequencies the asymptote has a slope of -40 decibels per decade
and intersects 0 db at w = W n . The deviation of the actual curve from
the asymptote depends on the damping ratio ,; and this difference can

'~: t,

.~ ,;,

"~"

,I.'

Jow.;,

':.H

'l'~"''''

-~lC'_'''~''''

~J'I,"'-'
~ ~,'t". "

h'
~'l \

I II ~

ii' : i:; ,
:

~n

1'::.

;.'

GUw)'

s.*

jW;

IL '.

~"lIlj ,

1\:\1 \

I- 1),0\

~~,~~

':l)

"

I" I':'

I'"i',i'>' I:

it"
;i'

~1,
~
:';;"'1 ,j
,

Ii'

'. J!
,~ 1

J.;,

~~~!:I

,'I'

~~~l\;1
'~ I

In

C(jw) = .;..
JW

~ , 11.1t t

J~1i

~I5i,

~;;;; I
til;,r-IJ

'fJ ~~.

ial in
can be used to represe nt .my linear system bect\use any polynom
given.
s
function
of
types
three
the
into
(jw) can be factored
cy
A second, and very commo n, way of display ing system frequen
this
In
.
respons e data is by means of the polar or Nyquis t diagram
arithme tic
plot the frequen cy is kept as the parame tric variable , and the
on polar
values of the amplitu de and phase angle are plotted directly
on
directly
coordin ates. The frequen cy is usually indicate d by writing
e
absolut
ng.
the curve. These diagram s are used extensiv ely in predicti
and relative stability .
polar or

The differen tial equatio ns of a system and consequ ently its

of
number
Nyquis t diagram are determi ned in a large measur e by the
system
x
comple
a
energy storage s. In gCl\.eral, the transfer function of
storage
can be: factored into terms corresp onding to either single energy
of the
terms
in
ed
Express
s.
element s or pairs of energy storage element
form
the
variable jw there result terms of

(3) [(jWT)2
,'1..

f:', .
N ~\'

'~l

~\

I~,

:"

I : ; ,,:'1
~A"I
, 1l"!lit

\1

If'1,j"l
f ;':l '4; ")~l
.' 1, oj, :~,

<W!t

' ].ii, ), '~,'i

': .

I"i. " ,~
I '"

'.,

':~I

,jell

j " ' i~

JJ"

'J

: ; " ",I

"'fill
:t ~'~I~
, I'

','' '~f:, ' ~ , ,11 ~il,'i

i'I' ';':'1'"

. \r'H,

, 1'" ;;1-Jt!

+ j2STW

, John Wiley,
c. S. Brown nnd D. P. Campbell, Prillf;p!e s uJ Servu"w('!lallisms

New York. \')40, p. 244.

jWT
Ii;! ""j'

/;

'i

(;

"I

+j

jWTUun;+ 1)

.~ I '

'. )1,.....

q.

'h

,\;j

t-.., .
"~

.~~;

(jWT)2 + j2sWT +

~~1
It w

;,~y

jW'1'

I
I

,j~ "

(jwT)a +

jWT(jW'1'

..

'-.-{/w

'f

I(. --'),

a> 1

jw-r + I

I)

~~' - - -

.',.'
'. "~

",

----

'1~~

"~

(jW'T)2

.}lO

/J"t

=t=?v,

+ j2{W7 + 1

-,~!~ ..

~~,

..~
t"

~'t

n.
't Figure 57 is taken from Brown and Campbell, op. cir., 1'. 157, by permissio

l'

fig. 5-7.

Nyquist plots of transfer functions.

',: "1"1" .::tr:,

, ,.
'1'

; ,j

I.

rc

jW'T((jw-rj2 +-j2'WT + 1]

,I

!I

.,1~

+ 1]1

represen t
These terms appeari ng in the system transfer functio n C(jw)
(2) a

origin,
the
at
zero
or
respecti vely in the comple x ,\. plane (I) a pole
of
pair
a
(3)
and
axis,
real
e
pole or zero at some point on the negativ
plane.
x
comple
the
of
half
left
the
in
conjup,atc comple x poles or zeros
quantiti es
The ,tbility to readily visualiz e the polar locus of these three
t plot 0:'
Nyquis
the
izing
synthes
for
e
valuabl
arly
proves to be particul
terms.
such
of
groups
of
ed
compos
s
function
transfer
more complic ated
s

function
transfer
several
of
plane
x
comple
the
in
loci
the
of
Typical plots
usual1\'
is
it
are given in Fig. 5-7.t In sketchi ng the Nyquis t plots
as W -+ 0
sufficient to conside r the asympt otic behavio r of the functio n
the cente:
in
plot
t
and w --, O. Once these arc determi ned, the Nyquis
r and
behavio
otic
range of frequen cies can be drawn from the asympt

. D.

",I

l)l

.1

(1) (jw)l
(2) (jWT

-~-+--..~

c~r

.!

i~'

369

Locus in Complex Plane

d:
.,''k.,. tit'

Ite~"

depend s strongly on the dampin g ratio

function
The transfer function s of Eqs. 5-7 and 5-15 with the additio nal

~ ~~ ~
';;'1 ~

ang curve
be quite large for small' -+ 0) and for large' (, -+ co). The
degrees
90
and
for Eq. 5-15 is 0 for w -+ 0, -1 ~O degrees for w -+ co,
= Wn
w
near
w
to
at W = W The rate 'of change of angle with respect
n

t'~ <
,1"

FUNDAMENTALS OF SYSTEM DYNAMICS

ELECTROMECHANICAL ENERGY CONVERSION

368

if.

~i:
./
.....""'-",
~~---- ,-~=.,

,,'

"

c...

"

ELECTROMECHANICAL ENERGY CONVERSION

FUNDAMENTALS OF SYSTEM. DYNAMICS

several points in the intermediate range of frequencies, recognizing that.

the transfer function is a well-behaved function over most of the center
range of frequencies. In a Nyquist diagram, especially in considerations
of stability, the plot includes negative frequencies. The part of the plot
of Fig. S-7 that is for negative frequencies is shown by the dashed lines.
When the Nyquist plots are available for two transfer functions that
are to be cascaded, the total cascaded transfer function is obtained by
multiplying the magnitudes and adding the angles at each frequency.
Because of this difficulty the Nyquist plot is normally used only for stability
studies where transfer functions do not have to be multiplied.

expense of gain, then, negative feedback can be used to shorten the time
constant (specd up the response) of a system. Note, however, that for
both the open-loop and closed-loop transfcr functions the ratio of gain
to time constant is the same. This is the familiar concept from feedback
amplifier thcory that the gain-bandwidth product of an amplifier cannot
be changed by the use of feedback.

370

5.3

,rr

Fig. 5-9. First-order system with positive

feed-back. ...

Now consider the system of Fig. 5-9 in which the open-loop function
is given by Eq. 5-16 but in which the feedback is positive. In this case.
if K < 1, the closed-loop transfer function is

In Sec. 5-1 the effect of feedback on static gain was briefly illustrated.
In this section the effect of feedback on the dynamic properties of a
transfer function will be illustrated.
Consider the simple closed-loop system with negative feedback shown
in Fig. 5-8. The open-loop transfer function is

I'
\.

~rf,,~

:1

Effect of Feedback on Dynamic Properties

.f
E =

-"".

G(s) = 1

,;

,.

~~IC.
_~
l+rs
~

'T.

When the system is operated closed loop, the

R=

G(s}
1 + G(s)

~,

C
E
=

(S.17a)

G(s)

K/(l + K)

R=

+ [T/(1 +

K)]s

(5-18)

1-

K;CK - I)
HCK - I)]s

(5-19)

,.ott'

--:,7".- - - ; - :

(5-20)
'j,

'if.
r:j':;

(S17b)

! .f-' f~"-t

The closed-loop system is characterized by a lower gain KI(l + K) and

a shorter time constant T/(l + K) than the open-loop system. At the

,II

;,,~R~

j'.,:"
I:~

l
"

[,./(1 - K)]s

<,1

Substitution of Eq. 516 into S-l7a leads to

K)

. This is an unstable system because the response to any disturbance will

contain a term exp [( K - 1)/,.]1 which increases exponentially with time.
.This exal11 pie shows how too much positive feedback can cause instability.
. It is also possible for too much negative feedback to cause instability.
For example. consider the system of Fig. 5-10 which has the third-order
open-loop transfer function [with H(s) = 1]

K/(1 -

R= -

Fig. 5-8. First-order system with negative

feedback.

a single time constant

transfer function is

Positive feedback has increased both the gain and the time constant by
the factor! /(1 - K) but the ratio of gain to time constant has not been
changed.
If the gain in fig. 5-9 is greater than unity (K > I). the closed-loop
transfer function can be written as

where s is the complex frequency (T + jw of Laplace transforms. When

the system of Fig. 5-8 is used open loop, it is characterized by a gain K and
.,
~

.,;;

>.",

R= 1+

(5-16)

+ TS

37\

,.,,",

:,ti,i
:t"~-

Fig. 5-10.

Simple closed-loop system.

'iler

I:U::.L I ROMECHANiCAL ENERGY CONVERSION

j/J.

J:'

FUNDAMENTALS OF SYSTEM DYNAMICS

The closed-loop transfer function is

R=

G(s)
1 + G(s)

(5-21)

= s(l + Tls)(1 + T2S) + K

This transfer function can be rewritten as

R=

~,

K
TIT2SJ

(TI

',""'"", ,
",J"

(5-22)

+ T2)s2 + .I' + K

",

It is entirely possible for the cubic in the denominator to have roots with
positive real parts when K is large enough, in which case the system will
be unstable.
Some simple illustrations have been given in which the effect of feedback
on dynamic properties has been illustrated. It has also been pointed
out that instability may occur. In the following section the question of
stability will be treated.

,
,,>\

5.4 Stability
Absolute stability is a binary quantity: either the system is stable or
it is not. However, even though a system has satisfied the necessary and,
sufficient conditions for stability, it may display damped oscillations that
are too large in amplitude and take too long to be damped out when a
step input is applied. It is this latter condition that is of considerable
interest and is spoken of as degree of stability or relative stability. It
goes without saying that all control systems must be stable in the mathe
matical sense (absolute stability); it is only the relative stability that is
subject to judgment.
In the closed-loop amplitude-frequency response plot, the absolutely
unstable condition appears as a discontinuity in the curve, i.e., the curve
goes to infinity at the frequency of instability. In the practical case,
however, system amplitudes never go to infinity in the true sense of the
word. Either saturation in various parts of the system limits the amplitude
of oscillations to some steady value, or the oscillations build up to a
point where a part of the system fails in some fashion (e.g., mechanically,
electrically). Saturation can occur electrically in amplifiers, magnetically
in motors, generators, and transformers, hydraulically in pressure-limited
systems; in fact, any physical component is subject to some type of
saturation or nonlinearity as the amplitude of input increases. Even in
the stable case, saturation in some form is desirable because it prevents
mechanical damage. In the unstable case, saturation holds the sustained
oscillations to a constant-amplitude value. Besides the fact that satura
tion occurs in all systems, an appreciation of the linear situation is basic
I '
(~:

~-"
~,

:r,
.l'

to the ulHkrstanding of nonlinear behavior. Although in many cases

saturation has a stabilizing inl1uencc upon a system in which linear
analysis would predict an unstable condition, the converse is also true.
In systems where saturation can be expected in normal operation, it is ~:
always wise to examine what changes, if any, the effective gain reduction 'b'

caused by this saturation will produce in terms of system stability.

'':~

Before turning to mathcmatical critcria for stability, the basic physical if

reasons for instability need to be considered. If in Fig. 5-10 a certain
sinusoidal input is placed at the reference input, there will be a return
sigmd B which wi]] be developed at the summing point by virtue of the
amplification of the forward and feedback loops G and H. The return
signal B is compared with the reference input. If, when the signal has t
arrived at the summing point, the phase shift is 180 degrees, the reference 4)
input signal will be reinforced, producing greater output and still greater
return signal. This continues until the system saturates when the
amplitude of oscillation becomes constant. Note that even if the reference
input is removed, the system continues to oscillate.
It is not necessary to impress a sinusoid upon the unstable system to
cause it to break into oscillation. Sometimes a small change in amplifica
tion such as that which may be caused by tube changes may bring the
system into the oscillation mode. Most people have experienced this
situation '.,ith public address systems whcre the system is quiescent until
a calculated or random increase in gain establishes the necessary conditions
for oscillation.
To produce an oscillating system there must be two energy storage
media in the loop; in electric circuits inductance and capacitance serve in
this capacity, and in mechanical systcms inertias and springs will serve as
the energy storages. Oftcn in clectromechanics one energy storage
medium is electrical in nature and the other is mechanical. When the
oscillations are sustained, the energy is transferred back and forth
between the storage media with the power supply or external energy source
providing for the losses in the system. When the external energy source
does not provide for the losses attendant upon the continual intertransfer
of energy, the system comes to rest.

5."1.1

Absolute Stability
<~

The mathematical definition for absolute stability of a system may be

stated as follows:

':1~:

'h

A system is stable if, to an arbitrary input, the response function or

the maxima of the peaks of the response function are not monotonically
increasing with time.

';';

",\~~
'=!'

373

:r,;v

~\
374

J'4'

"""

FUNDAMENTALS OF SYSTEM DYNAMICS

In effect, this defmition requires that all system responses come to a steady
state value as time approaches infinity. Note that a function whose steady
state is a sinusoid is considered by the mathematicians to be stable. This
is somewhat at variance with the ideas of control system engineers; they
abhor any system whose final value contains a steady-staL\" sinusoid
(Le., an oscillator). They consider such a system as unstable. This
inconsistency need not be a deterrent to the use of the mathematician's
definition; in this treatment, the mathematician's definition will prevail.
In working with linear constant-coellicient systems the Laplace trans
form equation for the closed-loop response of the system will be of the
form*

to predict stability from a knowledge of the G(s) R(s) product. The

Nyquist stability criterion provides such a prediction of stability.
Since the denominator of the closed-loop transfer function is (I + GR),
tne Nyquist criterion is concerned with the values of s that make
(1 + GH) = O. More specifically, the Nyquist criterion has to do with
the vector G(jw) H(jw) encircling the point -1 + jO as the frequency is

F(s)

A(s)

(5-23)

B(s)

r'''~M~

55",-I+jl

-2

59

!53=jl

= -1

58

.1-1+jl

=0

57

where A(s) and B(s) are polynomials in .\' of' the form (or reducible to the
form)

+ a,,-lsP--l +
+ bq_lsq I +

a{J'fP

s<i

+ als + 00

+ hls + bo

..,..,. ..... ,

~~

1':....._
~-

(5-24)

where the a's and h's are real constants, p and q are positive integers, and
Such an equation in s will yield a system time response containing
only the terms c'\', eJ"'l', and rme'.', where Sk is a root of 13(.1'), jWI is an
imaginary root, and m + I is the number of times Sk appears. It can be
seen that thc real portions of Sk must be negative to result in a stable
system given by the definition above. Thus, the poles of F(s) determine
the stability of the closed-loop system. This is tantamount to saying that
all the roots of 13(.1') must lie in the left half of the .I' plane or along the
imaginary axis. Thus in Fig. 5-11, .1'1> .1'2 = I jl and .1'7 = 1 contain a
positive real portion and will yield an unstable system. Roots $3' $4
would yield a stable oscillatory system (an oscillator must satisfy this
condition). Roots .1'5' .1'6 = -1 j I and .1'9 = - 1 will yield stable
systems; also, root .1'8 = 0 will yield a stable system. A pair of roots at
zero would yield an unstable system.
If the order of B(s) is above two, it is difficult to ascertain the factors .
In such cases a conformal mapping technique can be employed to
determine whether or not there are any roots in the right half of the
$ plane.
Most often, though, the denominator polynomial 13(.1') of the
closed-loop function is not available. Instead the G(s) and H(s) functions
of Fig. 5-10 are present in factored form. Consequently, it is convenient

.....~

.,.;.,-'-I .... ~- ~ ......'.

X
56" -1

X
54 --jl.52- I - j1

-ji

..
'\

varied from - joo to

criterion states:

\.{1I

,,

.,
Fig. 5-11.

.j

1~~Ul

'

p < q.

. \ t\

=+ 1

Real axis
,tl"

375

ELECTROMECHANICAL ENERGY CONVERSION

+ joo

B(s) root locations.

in a Nyquist or polar plot.

The Nyquist

The number of zeros of (1 + GH) is equal to the number of net

encirclements of the -l.O,jO point by the vector G(jw) H(jw) plus the
' ~
number of poles of G(s) H(s) in the right-half plane.

":'i
.~

.''

* M. F. Gardner and J. L. Barnes, Transients in Linear Systems, Vol. I, John Wiley.

New York, 1942, p. 163.

,~il-

For a stable system the number of zeros of [1 + G(s) R(s)] must be

zero. Formalizing the procedure to determine if this condition is met:
(1) Determine the number of poles of G(s) H(s) in the right-half plane.
This is denoted as P. (2) Plot the function G(jw) H(jw) for all values of
jw from jro to - jro (3) Determine the net number of times a vector
drawn from -1, jO to the locus of G(jw) H(jw) makes an encirclement of
-1.0, jO. This is denoted as N. (4) Add the number of net encircle
ments (with due regard to the sign, considering counterclockwise rotation
as positive and clockwise as negative) to the number of poles (2 = P + N)~
If Z is zero, the system will be stable; if Z is positive, the system will be
unstable; if Z is negative, the analysis is meaningless-the system is
physically unrealizable or a mistake has been made.

376

ELECTROMECHANICAL ENERGY CONVERSION

The locus of G(jw) H(jw) in the region near zero (0+ > w > 0-)
somctimcs prcscnts confusion. If the locus goes to infinity at w = 0 it
usually does so asymptotically to one of the axes of the G(s) H(s) plane
as shown in Fig. 5-12. The infinite semicircular part of the locus (Fig.
5-12) must be drawn on the proper side of the diagram. The decision
on which way to go when one reaches the point w = 0+ will always be

1
.;~

'~J!li,~

jw= -jO=jO

--.....

w < 0-,.."/

.p

I
I

..........

.\

fr

..

-..

{,.It.

.'~

.ll~ ,~">J

rtP'<"1 {;; m

..:-

<

li;\~ <t~,

jw.= +jO

= jO+

t1

-""-"........ \ ~e

"j......

"

,tV!(
.~

).~1r'~,!t

.~

-I

l.0

,
t ..

. -;~ ....;.; ,,~:,

.;~;'

"~,'

,[i"..

,"'"

:\

;:

H:'i'

.f.

..f:I'~~~t:~

-.: ''lk:,\ i!r~ti

'>t'm

-uU

correct if it is kept in mind that G(jw) H(jw) is a conformal map of the

imaginary axis of the s plane.. At w = 0 the angle of s (in the s plane)
changes abruptly from +rr/2 to -rr/2 and it does so in a negative direction
(clockwise). The angle change of G(jw) H(jw) in the region of s = 0 is
then easily computed. For examplc, the locus shown in Fig. 5-12 has a
transfer function of the form G(jw) H(jw) = K/[jw(l + jWT1)(1 + jW'l"2)]'
As w nears zero, G(jw) H(jw) will approach KUw.'Since the angle of

!l'()

'~j~
;]1;):.

f""'rjl bio " '.

: ; '.'

w =0 R
',It'
~e

"{';.

Fig. 5-12. Determining the number of encirclements of the Ny.; f

Quist point (- I. jO). In this case the number of encirclements is .~: .. ,
two. Transfer function G(jw) H(jw) = K/[jw(l + jw'''l) (l + }=z)]. I

,;;.~~! '

J.

,Il

.'

w=+-13
---;:

I"
W=+oo
I~,
-

I
/

i'(;;

~'I

,(I;

if'

;~"m,'r4HJ

, / . '

:'1

/ , .

//

i"~

w<O

.,,----------~

...

LA'

Sense
convention
'~"

w=oo

\.!.)

, t~, '

'"" \

')\f)T

'\,

!,

jw changes by -1T at w = 0, the angle of G(jw) H(jw) will change by +11'.

This is shown as the infinite semicircle in Fig. 5-12 which gives the +rr
angle change of G(jUJ) H(jw). In general. it may be said that for inte
gration tcrms (i.e., l/jw terms), the angle change from w = 0+ to
w = 0- is +rr radians for each term. Thus. jf G(jw) H(jw) approaches
1/(jw)2 whcn w is small, the connection bctwecn 0 + and 0 - is 2rr radians;
if G(jw) /l(jw) approaches I /(jw P it is 3rr radians. ctc. On the other
hand, differentiating terms (i.e., jw terms) produce a - rr radian change
11m

"

,.\

\
\

;'i~'1:'

"

377

1'.11;

Rotating vector

,~I::-

FUNDAMENTALS OF SYSTEM DYNAMICS

~.

:'.

.'

'''',~

~:iW '
.1,

'.,
J

,'~>

........

e . . . .' . .

.!

'

; "

= I/O

+ jWT)3.

for each term. Thus if G(jw) H(jw) approaches jw when w is small, the

conncction between 0+ and 0- is made by tracing out an area described

when a - rr rotation is made; if G(j<o) H(jw) approaches (j/O)2, the

surface is found by making a -2rr rotation. etc.
Consider for an example of absolute stability the case where

:r

''t'lf.

Fig. 5-13. Nyquist plot of G(jw) H(jw)/K

G(jw) H(jw) = (.
JWT

K I)'l

(5-25)

The Nyquist plot of G(jw) H(ju))/K is shown in Fig. 5-13. If the plot of
Fig. 5-13 is mUltiplied by K to determine the plot of G(jw) H(jw), it is

evident that as long as K < 8, the G(jw) H(jw) plot will not encircle

-1 + jO. Thus thc system whose .. open-loop" function is defined by

Eq. 5-25 is stable for K < 8.

"

'!tA.'"

'I,

,t,
'f~

378

ELECTROMECHANICAL ENERGY CONVERSION

5.4.2

Relative Stability

In any particular control problem, absolute stability is a necessity for

proper operation of the system. Since a wide range of paramder settings
may produce an absolutely stable system, the best parameter setting must
be found in view of the design specifications. A better picture of the
problem of system design can be had by considering a specific example.
In Fig. 5-14, a d-c motor is driven by a separately excited d-c generator.
The position of the motor shaft is measured by a potentiometer, and the

liOi

Constant current

"
.~H_i

/)/

"

Constant velocity

W~'T

~C"
""

""

.~}

1" ~1 .~
..

"

t:',.1
t-f>
c';
...:.-.:

;I ' ~.
'i.

Va

where

!~.

;I~

if

Ka

"

,'. . ~

= generator field current (amperes)

. The generator is assumed to operate at constant speed so that the generated

voltage is proportional to the field current (see Eq. 4-57a) or
V,

;;: ..........t'.'.l

where Kg

vg

~iZ'

.}.;"'"'1<1
J.':.'
:,~
;'

',t

Positional servomechanism with Ward-Leonard drive.

output voltage of the potentiometer is compared with the reference

voltage provided by a reference potentiometer. The error voltage is
amplified and is used to drive the field of the doc generator. Such a
control system is called a positional servomechanism. The combination
of separately excited generator and motor is called a Ward-Leonard
drive. To achi..:ve proportional action, the field of the doc motor is
supplied from a separate doc source.
Examination of Fig. 5-14 shows that the elements of a feedback control
system do indeed appear, namely, error measurement, amplification, and
power modulation. To analyze the behavior of the system of Fig. 5-14,
each component must be examined to determine its differential equation.
A block diagram is then constructed so as to focus attention on the
system dynamics.
The potentiometer error-measuring system produces a direct voltage at
the input of the amplifier which is proportional to the difference between

constant of proportionality (volts per ampere)

open-circuit generator voltage (volts)

Vr

!: ;~
f.,'
:g

'

(5-28)*

= K,if

If the armature inductance of both machines is neglected, the

differential equations of the generator and motor armature circuit and the
doc motor shaft are (see Eqs. 4-57a and b)

.1
..

t, ".

(5-26)

= constant of proportionality (amperes per volt)

,~,~ ;~"

"f" "

= K.(R - C)

Va = amplifier input voltage (volts)

K. = constant of proportionality (volts per radian)
The assumption is made that the amplifler has a high output impedance
so that the inductance of the generator field is negligibh:. In this case,
the amplifier produces a field current which is proportional to the amplifler
input voltage, or:
(5-27)*
if = KaVa

l.

'~".

The equation of the

where

~~
.~W;~~

Fig. 5-14.

a..

the posilions of the sliders of the potentiometers.

error-measuring system is therefore:

~~;..

~,

i,

379

FUNDAMENTALS OF SYSTEM DYNAMICS

....,..4...-.
where

Vb =

iaRa

(5-29)

Tm = Ie + TL = Kmi"
Vb = KmC
,
..'!

(5-30)
(5-31)

"

= motor back electromotive force, or generated voltage (volts)

ia = armature current (amperes)

Vb

Tm =
TL =
C =
Ra =
J =
Km =
,;,

motor electromagnetic torque (newton-meters)

load torque (newton-meters)
\'
. .~
motor shaft position (radians)
lola! armature circuit resistance (ohms)
Iota! moment of inertia (kilogram-meters 2)
constant of proportionality (newton-meters per ampere
volts per radian per second)

f,

Of.;

Note that negligible mechanical damping and linear operation of the'

entire system are assumed.
;''',.

* Note that the dimensions of the constants K K., Kg, and K m are aU different.
This is unfortunate but foUows the adopted convention used by control engineers for
treating quantities which are kept constant for the conditions of the problem under
consideration.

380

ELECTROMECHANICAL ENERGY CONVERSION

FUNDAMENTALS OF SYSTEM DYNAMICS

Equations 5-26-5-31 can now be represented in block diagram form

as shown in Fig. 5-15. The block diagram can be reduced to simpler
form using the techniques of Sec. 2.1.5, giving the result of Fig. 5-16.
In Fig. 5-16

K1

KeK~KaKm (newton-meters

381

Examination of Eq. 5-32 shows that the control system exhibits a

second-order response and is absolutely stable. In addition, if a step is
applied to the input, after the transient has settled down there will be a
zero steady-state error (i.e., C = R in steady state).
( 'Il.
Rewriting Eq. 5-32 yields

per radian)

C(s)
R(s) = (s(w,y

K2 = Ra(K;, (radians per second per newton-meter)

"

JR,,(K;, (seconds)
:!,

The block diagram of Fig. 5- I6 is a frequently recurring form. In

particular, assuming that the motor-generator combination and error
measuring system have been specified along with the load characteristics

where

Fig. 5-15.

't~

L I - - - - - IKmsl'
i
-IW

"

l~

Il"

Examination of Eq. 5-33 shows that the natural frequency of the system
can be increased by increasing K" (proportional to the amplifier gain)
but at the same time this will reduce the damping ratio of the system so
that the transient response will be highly oscillatory (but still stable !).
The degree of oscillation in the transient response of a closed-loop system
is a measure of its relalive stability.
Three situations will be examined.
(a) A step input is applied with the system initially at rest and
, < I (TL = 0).
R(t) = ROu- 1(t) ........... .....
(5-34)
..
where U-I(t) = unit step.

I
4S.

'K,

= VKJ.,. = undamped natural frequency

_ -_.,.._

'Y.'t

Block diagram of positional control system of Fig. 5-14.

'j;:~".

(5-33)

t.>

, = 1(2vKv = damping ratio

Vb
"'}~/

2~(s(wn)

. ~~;

<i

,,;

Wn

.,. =

n;'t'"

R(s) = o

(5-35)

K~

. Irs +

1)

r';,'.;

)l'.,

C(/) = ROU-l(/) -

,p

Ro
VI~-'2 ('

t/2T SIO

,F-

(v I - '2 '0,,1

+ 1 (5-36)

tan -I ,IT -=-12

... t\

(5-37)

OIl

'1" j

(b) A sudden constant velocity is applied at the input with the system
initially at rest and' < 1 (TL = 0).

Fig. 5-16. Simplified block diagram.

R(t) = QOU-2(t)

(moment of inertia in this case), the design problem is the proper choice
of the amplifier gain since all the other constants are fixed. The choice
amounts to finding a proper value of K\. For the moment, the load
torque TL is ignored and the closed-loop transfer function C(R is
considered.
C(s)

R(s) = (-r(Kv)S2

I" .. >'l."._ . i . : ,
(l(Kv)s + I tU1f '... !In'

where K. = KIK z (radians per second per radian).

Q
R(s) = 2"o

(5-39)

')

E(t)

(5-32)
-:"1I~

(5-38)

where u-z(t) = unit ramp.

t/J

'll:i:

;'

e.,

"
~

Q
Q e- t / 2T
"
---..2 0
" sin(~wlJt
K. wnVI _ ~2

tan-I

2,vr-=-~2
""'......
..

,11

+ t/J)

(5-40)
'it

"r'
'~

"

(5-41 )

11

~.

382

ELECTROMECHANICAL ENERGY CONVERSION

FUNDAMENTALS OF SYSTEM DYNAMICS

In Fig. 5-17 it is seen that the modulus response exhibits a peak at al;
frequency W = WR. the resonant frequency of the system. The magnitude:
of the peak is M p (peak magnitude) with log modulus Lm (Mp ). By
differentiating Eq. 5-46 with respect to wand setting the derivative equal
to zero, the following relations result:

(c) A step of load torque is applied with the system initially at rest and
, < I (R = 0).

TL(t)

Tou-I(t)
T(S)=To
L

(5-42)

(5-43)

To
e- t / 2, .
.-
(t) = K U-l(t) - -y.FI _'2 Sin (VI - '2 Wilt
1
4> = tan- I

.'
.,

.'

+ 4

vi - '2

_ ')

1
I( '(0)2 + 2S)
(.

:)
I

UJ"

OJ )

I]

WR

nvr _ '2

(5-47) ;
(5-48) .

= wnvl - 2,2

Examination of Eqs. 5-47 and 5-48 shows that the peak magnitude
increases with decreasing ~ while the resonant frequency is approximately.
"1')1

5i

i i

Lm (MpJ
or-------;~-~-~

,
i

.l''.

fltJ
!l{fflJi,.

-5
-10

ttl

,'; d

tJ','

')U1

'"
]l
.~

t~:n

r", 0.5

)i lI')'"

<,,:.

'< 11.,. (,,~

-15

"C

=[""'1

~:~ -20

,UI":.

'l.i:"i.
'~"F;

--'

[\' lil
~! '~'

-25
-30
-35
-401

0, 1

J:~

(5-46)

'.

"i),

I
A

I
-

_.

!
A"

I
_.

!
-

I
-

'\I

J:!!.

. ~';. ~

Wn

Fig. 5-17.

The log-modulus response for one value of' is shown in Fig. 5-17.

11

Examination of the transient responses described by Eqs. 5-36, 5-40,

and 5-44 establishes several important ideas. Note that the rate of
decay of the envelope of the transient portion of each response depends
only on the time constant T and is independent of amplifier gain. The
response to a ramp input exhibits a steady-state error which is inversely
proportionul to the constant K v The response to a step load disturbance
exhibits a steady-state error which is inversely proportional to the Constant
K 1 Thus. increasing the amplifier gain would increase both K v and Kl>
reducing errors due to constant velocity inputs and steady loads.
Unfortunately, the frequency of the transient oscillation is also increased
while the settling time of the system (measured by the rate of decay of the
transient) is unaffected by an increase in amplifier gain. This last fact
means that increasing the gain would merely increase the number of
oscillations exhibited by the system during the settling time. Such per
formance is highly undesirable since rapid oscillations would eventually
destroy the system components. In this case the system is said to be too
lightly damped.
In any real design, therefore, the choice of amplifier gain is a com
promise between high static accuracy and relative stability or degree of
damping. In practice a damping ratio lying between 0.4 and 0.7 has
been found to be reasonable for a system of the sort being discussed.
Once the damping ratio is specified, the amplifier gain is fixed so as to
meet this specification and complete the design.
Another point of view that can be taken in determining relative stability
is to examine the frequency response of the system for a sinusoidal input.
In particular. the log-modulus response introduced in Sec. 5.2 conveys
useful information. For the system under study

rcuw )]

"------

(5-44)

(5-45)

----;;-,

Lm LR(jw) - _Olog\o

383

Log-modulus response of second-order system with' = 0.5.

proportional to the natural frequency W II for small values of ~. Recalling!

the results of the transient study, the peak magnitude M p is directly
related to the damping ratio for this system, and a specification of M p is
as effective a design criterion as the specification of t. In practice, values
of M p ranging from about 1.0 to 1.7, giving Lm (Mp ) of about 0 to 3
decibels have been found reasonable.

385

ELECTROMECHANICAL ENERGY CONVERSION

FUNDAMENTALS OF SYSTEM DYNAMICS

Since the damping ratio only applies to a second-order system, this

performance criterion has limited application. However, the concept of
peak magnitude at resonance as a measure of relative stability has been
extended to higher order systems. Graphical techniques have been
developed whereby a plot of the open-loop frequency response
[C(jw)/E(jw)] can be used to determine M p and WR for a system regardless
of the order of the differential equation of the system. In effect, M p is
a measure of the nearness of the open-loop frequency response to the
point -1 +}O in the polar plane. Thus M p is closely tied to the absolute
stability of the system because the point -1 +)0 is the critical point in
a study of absolute stability.
The graphical techniques mentioned above include the use of constant
M contours on a Nyquist plot which yields the closed-loop response from
a plot of the open-loop response. A similar technique is employed with
a Nichols chart in which a log magnitude is plotted versus phase angle
with frequency as the parameter. t

With the system fixed as above, the steady-state error E due to a steady
load torque TL is
(5-50)

384

"~~.,:.

:1, = ~I

~-,);

"'-;i."

Error voltage from

reference (R)
and output (C)
potentiometers
of fig, 5-14

'!

i.

-1

i%h

Ej

Fig. 5-18.

R2

Eo

To amplifier
input In
Fig, 5-14

Ill'f
".II
,-)t,

Lead compensation network.

',.;n
TJ.
I

i'.,

.'
"

'1
c

K~

l+r 1" )

o:Ki ( 1+""Z~;

.(" + I)

.... 'f

f(

i.t.'.''i

'n

ik

'il

~;
;~

Fig. 5-19.

Block diagram of system with lead compensation.

negligible loading on its input circuit and that it is not loaded by the

input impedance of the amplifier, it has a voltage transfer function

Eo(s)
E;(s)

= ex

( I
1

+ TIS)
+ <XTIS ',,;n

'~'

where
ex = -::------"----;:R1 + R2

with K v = K I K 2 . Thus, selection of the motor and the relative stability

' ,
," .
fixes the whole system.

tfn

(5-51)
,/

R 2

~J

",

,,'f.

v'

(5-49)

" i,e,

R1

5.4.3 CompensationI'

,= 2VK;r

1I

.-"":!":-:-~""

Compensation of one form or another is used in most closed-loop

systems to improve performance. The reason for this is that the dynamic
behavior of certain fixed components (such as a motor) is such that
straight amplification will not yield a satisfactory system. The un
satisfactory performance usually results when the gain required for static
accuracy is so high that the relative (or sometimes absolute) stability is
unsatisfactory. In these cases compensation can be used to improve the
relative stability while maintaining the over-all gain high.
As an example of two simple types of compensation consider the
example of the preceding section for which the block diagram was given
in Fig. 5-16. Assume that the motor has been selected to provide the,
proper torque and speed (and therefore power) to the load. Once the
motor is chosen the gain K 2 and time constant T are fixed. The value of
, for proper relative stability sets the gain K1 according to the expression
below Eq. 5-33:

j
~

Now suppose this amount of error is not tolerable in the system and must
be reduced while maintaining the relative stability the same. There are
two simple ways of achieving this.
The first method to be used is called lead compensation. Lead
compensation in a d-c system such as this can be accomplished with the
simple R-C circuit of Fig. 5-18. Assuming that this circuit causes

"':,

and

TI

= RIC

It

With this network cascaded between the error voltage and the amplifier
the new block diagram for the system is shown in Fig. 5- I9, where the
amplifier gain is changed to yield the new gain K[.

Brown and Campbell, op. cit., Chap. 7.

t ibid., Chap. 10.

Brown and Campbell, op. cit., p. 220.

I"

J,Z:~

,\j:~

;if;

,I

;1:

'~~:

-'~~'

386

ELECTROMECHANICAL ENERGY CONVERSION

FUNDAMENTALS OF SYSTEM DYNAMICS .

The usual process in lead compensation is to make a have a value of

0.1 to 0.2, and, although it is not necessary, the time constant 'TI is often
made equal to 'T. The values assumed here are
a

= 0.1

and

Tl = 'T

(5-52)

where K; = aK;K2
constants

s(a'TS

K'
+ 1) + K;

:\

(5-53)

The denominator polynomial yields the system

w~ =

Jri

and

" =

2V;K~'T

(5-54)

The damping ratio of the system without compensation is given by

Eq. 5-49:
~ ~?
I
(5-49)
, = 2vf(:/r

..

Thus the improved performance is expensive but not

by a factor of
impossible to achieve.
Although the object of the compensation was decreased sensitivity to
load torque, other changes resulted which can be either good or bad
depending on the application. It was shown in Eq. 5-40 that the error
due to a constant ramp of input R is proportional to 11K. in the un
compensated system. In the compensated system this error is proportional
to 11K: and since K~ = 10K. (Eq. 5-55) the error due to a ramp input is
decreased. This is usually a desirable result. The bandwidth of the
system (w;, in Eqs. 5-54) is increased by a factor of 10 by the compensation.
This means that the system will respond more rapidly to disturbances,
which is usually a good effect; however, in many systems there is noise in
the range of frequencies included in the increased bandwidth which can
saturate the amplifier or otherwise cause undesirable effects. Conse
quently, it is often necessary to limit the bandwidth of the system to
exclude undesirable noise frequencies.
When it is desired to decrease the error due to load torque and not
change the bandwidth (response time), a second method of compensation
called lag or integral compensation can be used. Lag compensation

:I

With the lead compensation introduced in the system the gain K; must
be determined. In introducing the lead compensation it was decided
that the system should have the same relative stability as with no
compensation. The closed-loop transfer function including compensation
and with TL = 0 is

C
R

.;).

't"i!

Error voltage from

"t,,,",, tll)
and output Ie)

To make' = t' the gain must be

potentiometers
of Fig. 5-14

-K~ = -. = I 0

K.

(5-55)

t'

KI = lOOKI

(5-56)

This shows that for the same relative stability the amplifier gain must be
increased by 100; consequently, with unchanged input and output
impedances this means a required increase of 10 4 in power gain for the
f;
amplifier.
The object of the compensation was reduced error for constant load it
torque. From Fig. 5-19 this transfer function at steady state is
~:

EI
T'

Liss

1
I
= 7(i = 10K
a

I",
,

which is an improvement of a factor of 10 over the system without

compensation (see Eq. 5-50). This was accomplished by using an amplifier
with a voltage gain illcn:ased by a factor of 100 and a power gain incn:ased

Ej

0) T"mp"fi~

J: C

Eo

I~
n

Eo(s)
Ej(s)

:"'}

t' ,

where

, ;

'T2

= (R l

R 2 )C

1 + f3'TzS
I + 'T2S
and

(5-58)

fl=~
R + R

Cascading this network between the error voltage and the amplifier leads
to the block diagram of Fig. 5-21 where the gain K'( indicates that the
amplifier gain will be different from that used in the uncompensated system.

t(

~'

t'

Brown and Campbell, op. cit., p. 265.

f ~b

'f

Input In
Fig 5-14

can be obtained by using the network shown in Fig. 5-20 interposed

between the error voltage and the amplifier. Assuming that the network
causes no loading of the error voltage potentiometers and that the amplifier
causes no loading of the network, the voltage transfer function is

'

l13.

(5-57)

It 1

Fig. 5-20. Lag compensation network.

./>:'

f .."".

= -.;
a

1
[
~

from which

K~

387

104

_--:.....'--~---:.---'---"-

c--:'

ELECTROMECHANICAL ENERGY CONVERSION

3BB

''',~,~II'I_~

and the damping ratio is

In lag compensation networks the constant {3 is usually set in the range

0.1 to 0.2. For this system
(5-59)
f3 = 0.1

,_

(5-64)

= 2vK';,{3T

It is easy to show that the resonant peak given by Eq. 5-62 is the largest..n

The time constant {3T2 is adjusted to be

{3T2 = lOT

389

FUNDAMENTALS OF SYSTEM DYNAMICS

in this system; thus when Eq. 5-64 is constrained to be the same as for
the uncompensated ,system (Eq. 5-49), the gain K~ has the value

(5-60)

The placement of {3T2 indicated by Eq. 5-60 can be varied somewhat but it
must be an order of magnitude larger than T in order that the lagging
phase angle due to the factor (I + (3T2s)/(1 + T2S) will not occur near
the natural frequency of the uncompensated system.

Kn _ Kv

(5-65)

K"1
- -K-1

(5-66)n1'.

-p-

and the gain

K7 is
f1

TL

K'{ (

K2

1 + (3T2')
1 + T2 s

'(TO

The increase by a factor of 10 in the gain is accomplished in the amplifier.

If the input and output impedances are unchanged, the amplifier power
gain must be increased by a factor of 100.
",":lLI~w
The steady-state error due to load torque is given by
,

+H

1
T

Ls$

According to the stated specifications the relative damping must be

the same as for the uncompensated system. To set the gain K; for this
relative damping the closed loop response with TL = 0 is used

C
S(TS

K;(1 + {3T2 S)
1)(T2S + 1) + K;(l

--:;;'Ii

".f,j
'r.1

(5-61)

+ {3T2S)

whcre K: = K;K 2 Since the denominator polynomial of this transfer

function is a cubic, the relative damping cannot be set exactly as it was for
the uncompensated and lead compensated cases. However, the relative
damping can be set approximately by realizing that with S = jw and for
w

liT,

i::

T2w
I

>1

{1T2w

>I

"

1 'f'
I

Rw'Z lIT ~

"
S(TS

Ki

I)

+ K;{3

."

;~

r,'~t,'~)':

Jl

,I ~

-#,<\1
~;'t~

; ~ :, ..';

which is one-tenth as much error as in the uncompensated system.

Another system characteristic that is improved by the above lag
compensation is the steady-state error due to a constant velocity input at
R. The lag compensation reduces the velocity error by a factor of 10.
The addition of the Jag compensation has not changed the system
bandwidth and as a result the transient response will be essentially the,
same as in the uncompensated system.
Two simple techniqucs for improving the dynamic and steady-stato
performance characteristics of a closed-loop systcm havc been illustrated.
Therc are many other techniques about as varied as the imagination of
engineers.

5.5

Summary

'I'"

':t' ~

'

,'#

~.~~,

'1;,4 .~f

'

liYl
,,

.~,j;,:

~-

"""J:l:'1

. ":~.,t .

II,

...

';~

';0-,,' -:

~.,._

.. J';

(5-63) ~

tl'

(5-67)

This chapter has presented some of the fundamental tools and ideas
that are lised for the analysis and design of linear closed-loop feedbaclc~!,
control systems. Emphasis was always placed on dynamic behavior,
and design was found to be a compromise between static accuracy ando,!
relative stability.
An example of a positional servomechanism was used to illustrate the
techniques. The effect of load disturbances was included because this is
often of primary importance. Two simple schemes were employed tGlw

(5-62)

For this transfer function the natural frequency is

w: == J~{3,

S'~'

'"
iii"

,,;

Consequently, near the resonant frequency o~ the uncompensated system,

Eq. 5-61 becomes approximately

C/

.,:,
\

: "::-~

Fig. 5-21. Block diagram of system with lag compensation.

R=

K7

1
= lOKI

1
(~

390

ELECTROMECHANICAL ENER.GY CONVER.SION

illustrate how the dynamic behavior of the system can be changed to allow
a higher gain for better static characteristics. This process could just as
well have been directed at improving the dynamic response of the system.
In either case the improvements are always obtained .tt the cost of added
equipment, e.g., compensation network and incr.:ascd amplifier gain.

I
1

-i

."j:t>f

(1) Generator: Driven at constant speed w.

r resistance R,
Series winding ~ inductance L,
.
Lspeed-voitage coefficient K,
A
t
{reSistance R.
rma ure
inductance L"

I"

PROBLEMS

,ltl_'

5-1. The separately excited d-c motor shown schematically in Fig. 5P-I has
an armature resistance of 2 ohms and negligible armature inductance. The
motor supply voltage is V. T L represents an external torque applied to the
shaft in the negative <p direction.
(0) Write the equilibrium equations for the system assuming that V and TL
are unspecified functiQns of time.
(b) Convert the equations of (0) to a form suitable for block diagram
representation and draw the block diagram.

.
..

.'

391

Obtain the transfer function lis)! VI(S) relating armature current to injected
armature voltage.
Parameters of the system:

FUNDAMENTALS OF SYSTEM DYNAMICS

;l:'

(2) Motor:

Constant field excitation I r..

N,; ;'.,

Pure inertia load J

Armature resistance R.
Armature inductance L,.
Armature speed W m
Torque coefficient Km = L::;llm

'i

.\

"

= L:'w.

i<,;

r,
\

ia

...
',:
'1'<

=0.5 kg - meter

rl

J:m,

Ug

d
=0,3661 f"I,;

'.t1

l'L

2.0 ohms

"

dt

= rad/sec

Control
Windi ng """"

~'.'
~.

t1f=5am p

"

'

'
w\A.,",',"
Polarity of speed voltage
due to serie, field with
positive i a and positive Wg

Direction of flux
due to positive i a

,l),

lftl O

~.

<

1""':"'

series)
winding s

"1

Wg

=constant

fJ.A

,.

5-2. A frequency-response tcst is performed on a Ward-Leonard control

system. This test is carried out by driving the generator at constant speed, by
open-circuiting its control field, and by injecting a variable frequency alternating
voltage 10 the.: armature circuit as shown in Fig. 5P-2.

If1ft = constant

Fig. SP-I.

(c) Using the block diagram and assuming T L = 0, find the steady-state
speed of the machine when V = 230 volts. Also find the steady-state armature
current.
(d) Now assuming V = 0, find the steady-state speed of the machine and
the armature current when TL = 50 newton-meters.
(e) Stilt assuming V = 0, find the steady-state speed of the machine and the
armature current when T L = 50 cos wI. Be sure to make use of your knowledge
of the sinusoidal steadv-state solution of electric circuits.
(j) Write an exrre~ssion for 1(1) and i.,(t) when V = 230 volts and TL
= 50 -j 50 cos WI.

Di rection of
flux due to
1f",

Fig.SP-2.

Direction of speed voltage due

to a positive If", in shunt
field and positive W 1ft
;'

5-3. We wish to determine the response of the motor-generator set of

Fig. 5P-3 to a torque disturbance. For this test the control field of the generator
is open-circuited.
Parameters of the system:

(1) Generator: Driven at constant speed w.

rresistance R,
Series winding ~ inductance L,
Lspeed-voltage coefficient K,
resistance R.
A rmat ure
.
{ mductance
L.

L:'w.

-l

FUNDAMENTALS OF SYSTEM DYNAMICS

ELECTROMECHANICAL ENERGY CONVERSION

392

Find the transfer function wm(s)!TL(s) between the motor shaft speed CUm and
the applied torque disturbance T L
ia

Direction of flux due

to a positive i a

DirectIOn of speed voltage due

to a positive Ifm in shunt

r - - - - - - - - - -...--~I

.~

5-5. A d-c motor is to be used as a spring-loaded position control as shown

in Fig. 5P-5. The field of the motor is suppiied with a constant current I,.
The equilibrium position of the torsion spring is 1> = o.

Ii
~:

>"~

lleld and positive

Polarity of speed voltage for

Wm

I \

= 0

Motorl
t

~I/ = constant

~~~

,r ~ '.

1.

~,

1"..

E~ A't.

Va

(OOOl.,

I'm = constant

!i

,u
If =

D"eclion of speed voltage J

du~ to , positive I a In
selles field and POSItive wl/

TorSion

e,.

:',

"',

Flg.5P-J.\

Fig. SP-S.

;!

-:

5-4. The Ward-Leonard drive shown in Fig. 5P-4 consists of the following
system components:
(I) A triode amplifier having a plate resistance of rD'
(2) A doc generator running at a constant speed. The generator has two
field windings: the one connected to the amplifier has a resistance R, and
inductance L r and a speed-voltage coefficient Gar. The series winding has a
speed coefficient Ga.. The mutual inductance between the two windings is

.U.t~,,

spring

constant K

"

'.""1.
. . . . .'.

,, .'r

d\('-

constant
,H~

1, ; ,
'~,~
.

t. "c.;.
.'
:f"'

..

'

;.,'

I;:'

."

-/;

The machine is described by the usual parameters except that mechanical

damping c< is assumed to be zero.

Find the transfer function r/)(s)! Vn(s) betwe~n armature voltage and rotor
posi~ion.

S-l>. A closed-loop voltage control system is constructed using a l-kw.

lOO-volt d-c generator to supply a load of II. amperes at a load voltage of VL

volts, The control system is shown 111 Fig. 5P-6.

,:!'i'i'

L".

"l,...";::

if

Constant
current

f~'
','"

.~~j

-~~

....

0'

:,
'OIl'

Vs

UL

\f~' ~~t,;,,;,:-

Fig. SP-4.

I,: ~

-'J'

'I~

tu

-=
.,j......,. '.

Fig. SP-6.

~i
'./~I;

"

(reference)

;+
~

l~

-=

~'''t 11",

Amplifier

;j-'

,76'onstant
speed

"';';4; ,
,<.,".'~'

toro';!~

"li

t .>,,:~

"i<'~

positive If and positive tf,

Direction 01 flux due .... )

Control

1f,

(3) The motor. which has a constant field excitation, drives a pure inertia
load J. The motor torque constant is Km ,
The total armatur-:: resistance and inductance of both motor and generatotli
including the series winding are R, and L, r c s p e c t i v e l y . ,
An altcrnating voltagc v, is inJccted in thc armature circuit to provide ~
dither action to overcome friction. Determine the transfer function yieldinl,
the current ripple in the plate circuit of the input amplifier,
.

Constant field excitation ['m = constant

Pure inertia load J (no mechanical damping and motor shaft
pcrfcctiy rigid)
Armature resistance Rm
Armature inductance L m = 0
Torque coefficient Km = U;J'm

(2) Motor:

393

_~_

r !I

.,

,:...~

~j

394

ELECTROMECHANICAL ENERGY CONVERSION

The generator parameters are:

Field inductance L, = 300 henrys
Field resistance R, = 1500 ohms
Armature inductance La = negligible
Armature resistance R a = 10 ohms
Rotor speed w, = constant
Open-circuit generated voltage per unit field current
Kr = ISOO volts/amp
The input impedance of the amplifier is infinite and its output impedance is
zero.
(0) Draw the block diagram of the system of Fig. SP-6 showing UK as input,
VL as output, i L <1S disturbance, and VI! as error.
(b) The amplifier gain is K o volts/volt. Find K o if the time constant of the
clused-Iuop system is to be 0.1 sec. This is the time constant of the response
VL(s)/ VR(s) with i L = O.
Shown in Fig. SP-7 is the block diagram for a positional control
system. The blocks represent the following:
G I = Transfer function of an amplifier set for a gain of 2.0 newton-meters/my
G 2 = Transfer function of a pure inertia load where the moment of inerti.
is 1.0 newton-meter/sec 2
HI = Transfer function of a potentiometer whose constant is SO mv/rad
H 2 = Transfer function of a tachometer (velocity generator) whose constant'
is 10 mv/rad/sec

.wE

. ...

'

.to

j~'.

"~.
,~

;
t1~(,.

Dynamics
of
Transducers

.:'
11

"q.

j:

d~;;'

,HI, .

.n

t;

,if
~.~

;"fjm~tj.i

't

/'-.

"'.~ : J

~~~W:l'"

it \

5-7.

<

6.0

ft
"

.
~

;$1"'<:'
..
,

"i!

!l'
c
';"

..; il'

/.

Fig. SP-7.

"

At what frequency will the system oscillate when switch S is open?

If switch S is closed, will the system be stable? Justify your answer.
Find the output-position-to-torque disturbance transfer function C'(s)/TL(s)
with S closed.
(0)
(b)
(c)

.," .. ,"~ _ili\

":I~

,~.

Introduction

:.em
:~

,j

.'i:I'l

~1,!r;:"

is

In the broadest sense a transducer

a device for converting energy
from one form to another. Electromechanical transducers include aU
conversion devices that convert energy between electrical and mechanicai
forms. This dcfinition includes rotating machines. For the purposes 'of
the present chapter the term electromechanical transducer is used to
denote incremental motion translational and rotational devices. The
terr~ incremental motion as used here does not imply linear operation.
Since electromechanical transducers are energy conversion devices and
therefore are inherently nonlinear, complete and general analyses require
techniques such as computers for solving the nonlinear differential.
equations. As pointed out in Sec. 2.1.5, much useful information can'
often be obtained by linearizing the equations of motion. When this:
is done the general techniques presented in Chaps. 2 and 5 for linear
systems can be used for analysis and synthesis. In the present chapter
two examples of electromechanical transducers will be treated. The'
nonlinear equations of motion will be obtained and then linearized to allow
the use of general techniques in treating the dynamic behavior of the'
transducers.
The two examples of transducers, the torque motor and the microphone,

treated in this chapter are used merely to illustrate techniques. The forms

of the dcvices chosen for analysis are unimportant and they may be

replaced by other forms of transducers in the future. As long as the

new transducers are electromechanical the techniques to be used in this
chapter will apply.
395

DYNAMICS OF TRANSDUCERS

t:Lt:<.:r ROMECHANICAl ENERGY CONVERSION

J'Ib

6.1

397

Torque Motor

In many high-performance, high-power-level feedback control systems

a hydraulic motor is superior to an electric motor as a source of mechanical
power. Figure 6-1 shows a typical electrohydraulic position servo
mechanism. The reference input R and feedback signal B are electrical;
the electronic amplifier amplifies the signal and feeds it to a torque or
stroke motor which controls the position of a pilot piston in the hydraulic
preamplifier. The hydraulic preamplifier and hydraulic motor constitute
a two-stage hydraulic amplifier in which the velocity of the output shaft
position C is proportional to the displacement of the pilot piston. The
hydraulic system obtains power from a hydraulic pump that is driven by
some (possibly electric) prime mover. The position of the output shaft
is sensed electrically and fed back to the input.
The analysis of a typical torque motor for use in a system such as
shown in Fig. 6.1 is the subject of this section. The configuration to be
analyzed is shown in Fig. 6-2* in end view. The stator and rotor are
made of suitable magnetic materials; the stator is composed of thin
laminations stacked perpendicular to the paper in Fig. 6-2; the rotor is
not laminated but is made of a high-resistivity magnetic material. The
rotor is constrained to move only with a rotation in the. {J direction
shown in Fig. 6-2. This movement occurs against the' torque of a
restraining spring which is usually the rotor shaft. Four coils, each

,-~

~~
,.,0

Rotor aXIs/

Pole'

f-.-'

'--_.- a

n/

0)

..j

7/~Rotor
1

,l'li '
Coil
'if

Fig. 6-2.

Construction of torque motor.

'~ "'):

The torque motor can be connected and operated in two alternative

ways. One connection for proper operation is shown in Fig. 6-3a in
which coils I and 3 are connected series-aiding in one side of a push-pull
amplifier and coils 2 and 4 are connected series-aiding in the other side
of the amplifier output. When thc signal current i is zero, the currents
are balanced and in the absence of an external load the rotor is in the
neutral (0 = 0) position. With reference to Fig. 6-2, when the signal
current i goes positive, the magnetic field intensity decreases under poles

'I
r." - ' - ' - - - . "

,/.0~

:/;2'

'.'
'-1-1"1

~ ~.~~.J

.,ji;"".

'\~j~ I '

~::~';~

" I ....
,.<-,.,;
"".

'r'

III
,.'

-w

'iii. '

...

I'
, I

'I'
, h'

.':'

":'(:~;,'

I ,~j

~' "j

-r-

\'
',1

,':,

(a)

_'W,
,Fig. 6-1. Electrohydraulic servomechanism. hI},' >

'T' I

, ::~~.

it,

.:"

having N turns, are wound on the stator as shown in Fig. 6-2 with positive (,
current flowing into the paper on the unprimed side and out of the paper \~
on the primed side. The dimensions of the device are shown in Fig. 6-2.,.1
The construction shown in Fig. 6-2 and the analysis to be given follow closely the 1 ,
treatment in .. An Electric Valve Actuator for Hydraulic Servomechanisms," R. H.
Frazier and R. D. Atchley. Meteor Report No. 42, Dynamic Analysis and Control
Laboratory, MIT, Sept., 1949.
. i;
Itt
,~

/<'~

IJ)
~ , t'

"

(H'

"
t~
j,
{~

~},
i;~

'j
"jf

,
J,j

-----

-.

..........--~

,-

....

~.,,_.--

".,-

_.

(b)

Fig. 6-3. Two coil connections for proper operation.

: (.

J,'

ELECTROMECHANICAL ENERGY CONVERSION

398
(I) lint! (3) lUlU im:rclIscs unucr IWlcs t2) lind (4), tcnding to t~lrn the rotor

in the

+ e direction. With the alternative connection shown in Fig.

The electrit:al frequencil:s and mechanical velocities at which the

torque motor operates are low enough that the lumped-parameter approach
of Chap. 1 can be used to find the equations of motion.
In order to simplify the analysis to one that yields valuable information
about the dynamic operation while still retaining the key characteristics
of the device, several assumptions are made. First, the magnetic
materials of both the stator and the rotor arc assumed to have iniinite
permeability; thus the only regions of nonzero magnetic field are the air

i'\

6.1.1

however, it has already been specified that the coils will be interconnected
as shown in Fig. 6-3a. The amount of algebra required to determine
parameters is considerably lessened by considering the coils in pairs as
.:It.T

,;::,;;[1["
.1

~--;

.,

-#:~

:1\

+0 ._-_._......

=:::t-'

,"'!f

length

..

go =

under pole (2), go - a8 ,

under pole (3), go + atJ
under pole (4), go - aO

_..J

4 La

I,

II

Va
,

..

d (see Fig. 6-2)

~~~ ~.

;.,;'

FIg. 6-5.

Equivalent cirCuit.

,/.L:,'14

::~,

gaps. Next, the fringing fields around the edges of the air gaps are

neglected, meaning that in the air gaps the magnetic field lines arc: '",,<.

perpendicubr to the magnetic surfaces. When the rotor is displaced an

angle 0 from the neutral position the air gap becomes wedge-shaped;

however, the angular displacement 8 is assumeu to be small enough that

the wedge-shaped air gaps can be approximated by a uniform air gap

whose length is eq ual to the average length of the wedge-shaped gap.

This approximation is illustrated by the exaggerated drawing of Fig. 6-4.

" ,
....
'.

Thus with a displacement () the gap lengths areny' ""~ '11"

under pole (1), go + a(J ' "~.'
"'"
"
"

with go = D

i'"

Fig. 6-4. Illustrating air-gap approximation.

III

0+

D-d
'tl-~

,..j)

~-

2'

+ Oa =average gap

... i

'''Oh'

,:;-=.o==~.:::j--=----\,-='- -----f
'
go

i1

T
go - lJa = average gap

"

Equations of Motion

In the determination of parameters and the equations of motion the

four coils can be treated in general as having four independent excitations;

"::::::::::''''::::'::::::':''':1

399

The next step in the analysis is to determine the parameters of the

device. Since this is a magnetic field device the electrical properties can
be characterized by inductances and resistances and the inductances can
be determined from the stored magnetic energy or flux linkages obtained
from the solution of the static magnetic field problem.

6-3b, the opt:ration is esst:ntially tht: same.

."

DYNAMICS OF TRANSDUCERS

(6-1)

.,

'~.

given by the circuit of Fig. 6-30. Thus the current i 1 is as'sumed to flow
through coils 1 and 3 in series and the current i 2 is assumed to flow
through coils 2 and 4 in series. As yet, the currents i 1 and i 2 are not
restricted to have the values given in Fig. 6-3a. Thus the equivalent
electric circuit to be treated is shown in Fig. 6-5 in which the self-,
inductances and the mutual inductances are functions of the mechanical.
angle fJ.
If each coil has a resistance R e , then
ih,
R}

R 2 = 2R c

(6-2)

The inductances can be determined either from stored magnetic energy

or from flux linkages. The latter technique will be used here because of
its simplicity. To use this technique the fields in the air gaps must be
known. With the assumptions of no fringing or leakage fields and of
uniform air gaps even with a displacement 8, the magnetic field will be,
uniform undl:r each pole face. Tn addition, because coils 1 and 3 have

400

ELECTROMECHANICAL ENERGY CONVERSION

DYNAMICS OF TRANSDUCERS

the same current and the gaps under poles (I) and (3) are always the
same, the fields in these two regions are the same and will be designated
H t . Similarly, coils 2 and 4 have the same current and the gaps under
poles (2) and (4) are the same so the fields in these two regions will be
designated H 2 The fields are given by their components

= H Il +
Hz = H Z1 +

H1

'401

1n a similar manner for excitation i 2 with i 1 '= 0, we have

l/22

Ni 2
go - aO

(6-9)

and
Ji IZ = 0

H l2
H zz

(6-3)

(6-10)

',l

The mutual fields H 12 and H Z1 are zero, indicating that the mutual
inductance between circuits 1 and 2 is zero:

where the second subscript on the components indicates the excitation,

i l or i z. The positive direction of all components is indicated in Fig. 6-6.

L 12 = L 21 = 0

(6-11)

-II};

!!

The flux (inking circuit lowing to current II is found by first finding the
flux density

~---------------~----------~

,I

I
I

Contour 2./

directIon
of H field

(j}
Positive

H
tHll :t
, '2

tH~~

: tHIl

_--___ rr::-:

AH 21

-.~!

I
I

~..

I
I

,,

iI",

0
tH---:J

B l1 = f1-o H ll
The cross-sectional area of each pole is

I.

(6.12)

A p = wi

tH'2

,,

t>

"11

Cont~

~"

= 2Nli-!jJ. oH ll

Contours for integration to find H field.

(6- 15)

Since inductance is defined as the flux linkage per ampere, the self
inductance of circuit I is

Assuming that current i 1 is applied and current i z -';7 0 and using the
Maxwell equation
"
.

,~H

c:f~ ; t'

around contour 1 in Fig. 6-6, we have

2(go

.ff!/'

,.,

",:~. )~t..: ,<t<

;.,~' ;,~.

go

+ aO)

t),

"H': ,j,

I,

,,~

'.

H21 = 0

'-~

,.
g

r~-

.t

'd/ti'

'z

_~,'i,

~{ft

~,.,

= 2IV!/LON2

.tt";:

go - aO

; u)

Designating

L II
t"::., L zz

''''

(6-16)
, .'

Ll
,I

(6-17)
'f;

,':~W

(6-18)

go

Lo

= I + (aB/g o)

=1-

II'

~.VI

(6-8)

),:1

a8 ,,;j

Eqs. 6-16 and 6-17 can be written as

Substitution of Eq. 6-6 into 6-7 and solution for H 21 yields

i

",22

(6-7),

- H 21 (go ...:. aO)- Nil

L = 2l1'!jJ. oN2

(6-6)

The use of Eq. 6-4 around contour 2 yields "1"i- ~,

HII(go

L 22 =

(6-5)

a8

go

~,..

= 2wljJ. oNZ

Similarly, the self-inductance of circuit 2 is

t!.

Nil

".11
/1

(6-4)

"!:';;';ld,/JIl':' dl.t!1 }l~;, .ili:

_
II -

L l1 =

~H,)

+ aO)H ll = 2Ni1 : .i!t,lrl~ '\ '(~'-

from which

_ 2wlf1-oNZ/l
go + aO

II -

H' ds = Ni.

-1

"

.'\1

(6-14)
"",",,",,'''''

Substitution of Eq. 6-6 into 6-14 yields

Fig.~.

(6- 13)

hence the flux linkages with circuit J are

La
(a8/g o)

:IJ

(6-19)

.1,)

1"'~'f1

(6-20)

';lift,.

402

ELECTROMECHANICAL ENERGY CONVERSION

The volt-ampere equations for the torque motor expressed in terms of

the parameters given by Eqs. 6-2, 6-11, 6-19, and 6-20 may be written
from Fig. 6-5 by inspection and in matrix form are simply

~ = 'L~Rc + pL II

O'

2R c

+ pLEJ ~

+ apO + K8 - T,

i , = 10
iz
10

V, =

'tt

mechanical torque of electrical origin

J = moment of inertia of the rotor
a = damping constant
K = constant of restraining spring

"

(
2R c 10

.)

Lo
di
(I') a
Lo
d8 (6 271)
[1 + (a8/g o)] Ji - 0 - 1 g;; [I + (aB/g )2 dt

Vz

"

= 2Rc(lo +
dZO

.
I)

Lo
di
[1 _ (a8/g o)] dJ

dB

T = J -dt-Z '+a dt

K8

w;"

-!Llli~

+ -tLzzii

(6-23)

VI

T,

OW;"
---:-;--/1
L

I .2 aLII

= "1_ 1 1 --,,.-U
+
U

1,2 aL n

1/
_ 2 ----C;-U
L

(6-24)

ii'

t
.'!'"
'~;.

t'

Substitution from Eqs. 6-19 and 6-20 into Eq. 6-24 and evaluation of
the derivatives yield:

nv

Lo

,I.!

Lo
2 I go [1 -t (aO/go)]2

!i 2 !!.

+ ~i~!:.

Lo
2 - go (1 - (aO/go)]2

(6-25)

(6-29)
i'.

,I

.kl~

'... ,./ ' ' 'f"''f

T = J d- 8
~2

(6-30)

a dB

di

+ I) + L o -df + loLo go
--d
f

dO + K8 + a -~

(6-31)

)2

2 -a Loloi - 2L o/ o2( -a 0
~
~

(6-32)

The electrical equations can be simplified further by realizing that if

each of the vacuum tubes in the push-pull arrangement of Fig. 6-3 has a
plate resistance r" and if the tube feeding circuit I has an incremental
internal voltage -(' and the tube ft:eding circuit 2 has an incremental'
internal voltage +e, Eqs. 6-30 and 6-31 can be subtracted to obtain

di

2e = (4R c + 2rp )/ + 2L o dii

.
di
a dO
= 2R (10 - I) - L o - - 10L o - C
,dt
godt

,I 1
~H,

""j~,t

Vz = 2R c (lo

I,

".J

1)2 go -=.-....,......::,o

With the assumption of small electric signals i <{ 10 and small meche,
anical displacements aO/go <{ I, Eqs. 6-27-6-29 can be linearized to

from which the electromagnetic torque T, is

:f'r

. a
Lo
d8., )
I) go [1 _ (a8/R'n)12 dt (6-_8

'

'\

+ (10 +

]2
- 2 (10 :'1: 1' )
go2
[I -)(a8/g
o

'it

The torque motor is completely characterized by Eqs. 6-21 and -6-22,

once the torq ue i~. is found in terms of system parameters and variables.
This torque can be found from the method of arbitrary displacement
given in Sec. 1.2.2. The electrical variables used here are currents so the
coenergy will be used (see Table I-I). The magnetic coenergy of the
system is given by

ii:

111'

ni

+ 2(10 -

!<,

= externally applied torque

=

(6-26)

"h,

T,

.\ .f_, ,)-u,...t

i
i

"

403

wliere 10 is a constant, quiescent current. Using Eqs. 6-26 in Eqs, 6-21

and 6-22, the equations of motion can be written out as

.:~

,i

where

;'

and in the mechanical equation (T,)., With the electrical connections

shown in Fig. 6-3a the currents are

(6-21)

(6-22)

.. ~lf;!'

DYNAMICS OF TRANSDUCERS

= +

where p indicates the time derivative d/dt. This completes the deter
mination of the electrical parameters and the vOlt-ampere equations,
The mechanical part of the system is also assumed to be linear; thus
the mechanical equation of motion is

T = Jp 28

'ill'~.;

a dO
+ 21oLo --d

go t

(6-33)

Making the definitions

"),,,;

2R c

+ rp

...

a8
10Lo - ... G

The equations of motion for the torque motor contain nonlinearities

in both the elel:trical equations (because L II and L 22 are dependent on 8)

go

:o? .,

(6-34)

404

and dividing Eq. 6-33 by 2 yield

The Laplace transformation of Eqs. 6-35 and 6-37, assuming zero

initial conditions and using Eq. 6-38, yields:

'"t:,'"

.
di
e = Rl + L o dt

dO

(6-35)

+ G dl

'

';
-~

(s) = (R

(;J

Los) I(s)

(6-39)

Gs O(s)

T(s) = (Js 2 + as + Kl) 8(s) - 2G I(s)

If for Eq. 6-32 the definition is made

2Lo!Z

405

DYNAMiCS OF TRANSDUCERS

ELECTROMECHANICAL ENERGY CONVERSION

These equations can be put in the more meaningful forms

(6-36)

= Ke

(6-40)

(s) = R(1

then the torque equation can be rewritten as

T(s)

7"es) I(s) + Gs 8(s)

= J(s2 + 2'mw".s +

(6-41)

(6-42)

.0;') B(s) - 2G I(,f)

,~...;Ul

d 28
T = J dt 2

dO
<X dt

+ (K - Ke)8 - 2Gi

where

(6-37)

Equations 6-35 and 6-37 are the linear differential equations that describe
the torque motor of Fig. 6-2 for small currents i and small displacements O.
From Eq. 6-37 it is evident that there is an effective electrical spring
constant K e which gives rise to a force that opposes the mechanical
spring force KfJ. Whenever K - K, < 0, i.e., the elcctrical spring constant
is greater than the mechanical spring constant, the device is unstable.
For this condition a small current will displace the rotor until it hits a
mechanical stop. Thus for stable operation the condition must hold
that K > K., and in most practical cases K is made several times larger
than K.. To simplify the mechanical equation further the definition is
made
, ,
\ . i,\
,:l,
(6-38)
Ko = K - Ke

~14:i

Wm

J~o is the mechanical natural frequency

= 2.I

<X
. r--'-

vJKo

IS

~l~;

.", 1:

' ld amplOg
. ratto
.
t he mech
aOlca

'.1

'~..

'

One form of block diagram representing Eqs. 6-41 and 6-42 is shown
in Fig. 6-7. It is evident from Fig, 6-7 that there is interaction between

I"'~I:",'.

~- r., ~"

2G

'R(1
- -+-T,.S)
--

'

,'<.-,'\1

'i

9(8)

;'1.

;';"k,~'2

$'

~~

"

Gs

I"

,;-1t

}tJ~ H;~~":~J: ,'-',ll'

Fig.

~7.

Block diagram of torque motor.

... ! , , ; , " l "

the electrical and mechanical systems because the input to the electric
circuit is composed of the input voltage and the voltagc of energy
conversion Gs8. Similarly, the input to the mechanical system is com
posed of the applied torque T and the electromagnetic torque T e This
interaction will affect the dynamic properties of the torque motor.
Before treating the dynamic behavior it is a good idea to write down
the steady-state transfer functions in order to observe the way in which

The dynamic operation of the torque motor of Fig. 6-2 can be studied
using a variety of techniques as indicated in Chap. 2. Following the
techniques of Chaps. 2 and 5. block diagrams will be used to study the
torque motor because they present a graphic illustration of the physical
,r,'
,.J i ; '/. t "
processes in the system.
See also Frazier and Atchley. loc. cit.

't

".

,i:

""~~;
":'-!

It

-!

~'.."

r~'~

":t
.:J(s ~ +2,'",w",s+w 2
""
+
~
m)
r

~~,

,""'''',

'1"'1

:~(

~ ..

~;,

T(~)

' i,:.",::_,
" "1.\;

't,'

1 ",

., .\.~,
.,

~i:~_'r.,;;11!~

~--

I ,~',i

flit
,::

with the restriction that for stable operation Ko > 0."" ,

The above consideration of stability was based on the linearized equa
tions of motion. A more comprehensive treatment of stability can be,
made with the more general nonlinear equations in a manner similar to
that used in Example 2E5 of Chap. 2....

6.1.2 Dynamic Operation

I~ is the electrical time constant

I'

".11"

'"

\,m

"

'T.

.~!

~!.;

oj,

,""

406

ELECTROMECHANICAL ENERGY CONVERSION

DYNAMICS OF TRANSDUCERS

407

~, :,

the various parameters affect the static sensitivity.

transfer function from E to 8 with T"" 0 is
-81

2G
= - = - 2G

E ss
RJw~
RKo
and the steady-state transfer function from T to 0 with E = 0 is
- 01 ',==

- 1 == - 1
Jw~
Ko

T ss

(6-43)

I1

(6-44)

The steady-state

For the greatest change in angle for a given change in input voltage, the
factor 2Gj RKo should be large. To obtain the smallest change in angle
() for a given change in applied torque, IIKo should be small or K o large.
The dynamic properties of any linear system can be expressed in the
frequency domain or the time domain with the same amount of informa
tion contained in each domain. Both types of expressions will be used
here.
The usual mode of operation of the torque motor is such that the
external mechanical load consists of inertia and damping so that the
applied torque T(s) in Fig. 6-7 can be assumed zero and the load inertia and
damping can be lumped with the inertia and damping of the torque
motor. Consequently, in what follows the applied torque T is assumed
zero.
The transfer function from E to 8 in Fig. 6-7 is

ill,

8(s) _
E(s) - (TeS

",.

,e

'~

2GjRJ

1)(s2

nmwmS

2
Wm)

2G2

+ RJ

2GjRJ
,J

T ..I

(I

2YSmWmTr)S 2

2
TrW m

') "

Y
2SmWm

(r) ,+
RJ

..

(w~/w;)(2G/ IU)
(n'

I )(S2 -1- 2'wlI"l'

+ w~)

.'.',;.-.

2
Wm

(6-46)

-1._

r::::;=-,

;siN

t~

31~
;;:.Ic;r

- 20 db/dec

log

= -2-

7".

but the natural frequency w" is difficult to determine for the general
literal case.
The frequency response of the system, which is obtained from Eq.
6-46 by setting s = jw, consists of a flfSt-order lag and a quadratic lag
in cascade. It is easy to sketch the asymptotic log-modulus curve (see

"I:

logw

',.,i,:

~,

'j,

.5

.,
.~~

1;
~

'k

Fig. 6-8. Asymptotic sketch of frequency response." .

The step response of the torque motor is easily calculated from the
transfer function of Eq. 6-46. A step of Eo volts is assumed to Occur
at t = 0 with the torque motor at rest. The Laplace transform of the
position is
8(s)

[(w;,/w;,,)(2GjRJ)]Eo
SeTS + l)(s2 + 2'w"s + w~)
1

(6-48)

Expansion of this expression in partial fractions and inverse transformation,

yields the time response
_."
:

r_

_ 2GEo
OCt) - RJW211
m

e-t!T

[( ,

)2 + (l

;,

'2)]

WnT

Wd

(6-47)

Wm

","Ie,:,

+
2
W"

'>

"

w"

slope

From Eg. 6-46 it is apparent that

"

'3~~I
0
log r

4"

r' ...

(6-45)

Since u cubic polynomial can always be factored into a l1rst-ordcr factor

and a quadratic factor, Eq, 6-45 can be put in the form
0(.1)
Us) =

1')
i

Sec. 5.2) for one particular set of parameters. Such a sketch is shown
in Fig. 6-8. The actual log-modulus and angle curves for a typical
torque motor are shown in Fig. 69.'" From Fig. 6-9 it is apparent that the
frequency IjT is lower than Wn and that the damping ratio' is less than
unity.

e-Cwnt
T

[(, -

sin

.p =
and

<LId

tan-I

vlr-=-r + tan- t V
- ,

+ .jJ)

~r + (I

where

.:1:'; ",... L

(Wd t

(6-49)

- {2)f

'2

1(ljWnT) - ,

= Wn~

This figure appears as Fig. 5.3 in Frazier and Atchley, op. cit.

>i

ELECTROMECHANICAL ENERGY CONVERSION

408

+ 90

+ 121-

+ 60I
log

modulus .........

+30I

+61-

..

]'"

~.

o~

o)

'"~

"-

l.

~13

. 0>'3
;;'Jj'"
~ ~,,\'"

L::.i-J

2lI "I'"
N

,310- 1

30-6 0

~ '"

.,..
~. - 90

..j

-,18
I

,~..,'-

!~!.1 t
~;"

t~jft'

,)

")"1'"

,
' I!

- 301-

.!,

..

,-,"

~ '/

- 42

- 21

1
,'ti

f'j,i':'k'

~-:

,'1

10

lL,.'

'~-'

20

100

200

"',

-'

:,~;}

400 600

Frequency, cycles per second

'1'."'."
.~

.. "

'..

" % ' - " "

Fig. 6-10.

,)~>~,

'._H"""" ., .....

&to' " " ' . .

{,

"~,,

"~'"

'i

"'l1...,

II"'!,
~

,;

W II

~~

'm

."

....

in
This conditi on of operati on has the effect of causing the feedback path
tely
Fig. 6-7 '0 be ineffective. Thus the first-order lag factor is comple
cy

determined by Lo and R of the electric circuit. The natural frequen

ters
parame
al and mechanical
W n is determ ined by a combin ation of electric
because according to Eqs. 6-42, 6-38, and. 6-36

..

'

.~~.,

..",-
:M:~ill'

~i ,""

Ir'"",~ ~

~ ...
1l-,J -j

This figure appears as Fig. 5-2 in Frazier and Atchley, op. cit.

"h'" .'

, II

'01.
"

Step response of typical torque motor.

'T ;:::; 'T

~#:.

as
Case 1. When the coils are driven by high impedance sources such
2G2/RJ
term
the
making
large.
is
R
pentode vacuum tubes, the resistance
in Eq. 6-45 negligible and

Frequency response for a typical torque motor.

term
occurrence is that l/T ~ ~wn, making the time constan t of the second
term.
in Eq. 6-49 much shorter than the time constan t of the third
are
When a torque motor like the one for which experimental data
in
shown
that
as
given above is supplied in a closed-loop system such
the
than
less
rably
Fig. 6-1, the closed-loop system bandwi dth is conside
t peak
natural frequency of the torque motor. Consequently, th~ resonan

+or,"

~:.i

: r,;.,., I

~:'"~l

Fig. 6-9.

f1

id441

~';..

.J

\'

.'

,~", ;. ~

- 180I

""
,

.:...i

' X>''i

~~d: ,~

- 361-

"':"!~"'4;;:;M.M'#!i%1

-, .

"''<

.",' t~-4t~

-IS' )

"~'c'1.p""'''i!''

,NijilA:g;;;iP". .....
(

t,

- 12 )

- 24t-

not
in Fig. 6-9 and the oscillatory behavior illustrated in Fi~. 6-10 do
ent
compon
some
that
means
show up in the output of the system. This
the
than
band
pass
smaller
a
has
in the loop, usually the hydraulic motor,
torque motor.
of
The constan ts T, " and. W n in the denomi nator of Eq. 6-46 are
the
in
peak
t
resonan
and
dth
bandwi
interest because they determine the
in the
frequency domain or thc speed of response and relative stability
cases.
special
two
for
imated
time domain. These constan ts can be approx

,l

: ~

409

"~

Angle

t.
~$
~

-3' )
-6
.
I~I~I
;i~I~1

<>

1"'11
.C/

Ie.

~'-.,.

~1\

DYNAM ICS OF TRANSDUCERS

~.

state
In Eq. 6-49 the first term in the brackets represents the steadyely
ultimat
that
displacement and the other two terms represent transients
cy
frequen
the
decay. A step response for the torque motor for which
this
for
that
Note
response in Fig. 6-9 was taken is shown in Fig. 6-10. *
masks
particu lar motor the oscillatory third term in Eq. 6-49 essentially
the
for
reason
likely
the simple decay of the second term. The most

+ 18 r

".~":'
I

/go)2
JK - 2Lo/~(a
J

.....

">'1~,-'"

(6-50)

:~

:~

."
i
I'

/,."

'v

1
-i
j

410

6.1.3

Thus the natural frequency can be controlled by both mechanical and

electrical parameters. The damping ratio, according to Eqs. 6-42, 6-38,
and 6-36 is
1
a
(6-51)
, =

2 VJ[K - 2Lo/~(a/go)2]

r~,:;

"
:!

I'

....

S$

2I()Lo(ah~o)

2G

(6-52)

RKo = R[K - 2LoI~(a/go)F

permeability is valid.
Case 2. When there is negligible mechanical d~mping ~m ~ 0, the
constants in the denominator of Eq. 6-46 become

'".

ii'
'[~~

19'ft:.fJ

'~.' .' "

'~.

~~

#Vii.J.l

RVJlK

---2L~J~(algo)]2

~j

'"IK

2G

RJw~

-.1 ,;'p':t

~ _;~;

l~it.

,,,'

'J'li
;\

I;L,fi

..

:~

(6-54)

.. \ tl

'''If!

.n,',

-~.

,~j,H

t lH.d; ;jJ:tD:\}\~ .~:"

"'''''''~' ,<~...~ ~."",... ~...-..-.-..,

',l

.,".j>.

"

~.;.~j,

. {

,4 1 ,

..:q),,~ ,

i~~

~~;

~ie;'
.. ,?

I
i

'--:--."

.'-

"'l
'r~ '"
"",

.."J"

,. ...,:

r,:'l-:( _.+~
j

~......

...-J

~:

Hydraulic preamplifier
and motor

Torque motor

",~.~

(TS

K m
+ 1 )(5+ 2\~ + 1)
.

wI'

"

Kh

8(~~+21h~+lJ.
h

Fig. 6-11. Block diagram of an electrohydraulic servomechanism.

hydraulic preamplifier. The transfer function of the hydraulic motor

includes the gain of the hydraulic preamplifier. The preamplifier is !
assumed to have no other dynamic effects than those already noted. The!

(6-53)

~.

>Ic,

The transfer functIOn of the hydraulic motor has not been derived here. The
derivation IS straightforward and is given in several references. for instance, G. S.
Brown and D. P. Campbell, Principles 0/ Servomechanisms. John Wiley, New York,
1948. pp. 136-140.

'(~1;'L"
y

RKo

I' If;i:H ~1;:

r~

~',

. ~ ~l1Jo..lr:W'

= 2G'OSIJ.1U

'''-f"_~

'1-1,

Thus the damping ratio can be controlled with quiescent current 10 which
also affects W I1 or with the resistance R which atlects the time constant
T.
Note that a change in R affects the static sensitivity (Eq. 6-52) in the
same way that it affects both ~ and r.

".

I(

./'i'(.,
~f,

to yield displacement in meters at radius '0 where the output is attached,

includes the inertia and damping reflected to the torque motor from the

i,

G2 - - - -G2
-=
""' RJw n - RVJKo

L~I~(al.~o)2

,( ,'1

The time constant T and natural frequency W n are the same as in Case 1
but now all thc damping is supplied electrically by the voltage of energy
conversion GsO in Fig. 6-7. In terms of the definitions of Eqs. 6-34,
6-36, and 6-38 the damping ratio is

,.

"

,,...,

Wn ~ Wm

Application in Electrohydraulic Servomechanism

.J'

~'tt

Thus, increasing the quiescent current also increases the static sensitivity
by a factor that is greater than the factor by which 10 is increased. This
shOWS that when lc::ss bandwidth can be tolerated, the other operating
characteristics of this torque motor can be improved by increasing the
quiescent current. How large the quiescent current can be depends both
on the iZR losses in the coils and the magnetic saturation, the latter effect
setting a limit to the range over which the original assumption of infinite

411

The torque motor of Fig. 6-2 which is described by the transfer function
of Eg. 6-46 will now be integrated into the electrohydraulic servo
mechanism of Fig. 6-1, and some system characteristics will be investigated.
The block diagram of the system is shown in Fig. 6-11. The amplitier
is a pure gain Ka except that the plate resistance of the output tubes has
been included in the transfer function of the torque motor (see Eqs. 6-34).
The transfer function of the torque motor, in which

and it can also be controlled by both electrical and mechanical parameters.

It is interesting to note that increasing the quiescent current 10 decreases
the natural fn:l\ucncy and increases the damping ratio. This improves
the relative stability but makes the bandwidth less wide. Reference to
Eqs. 6-43 and 6-34 shows that
() 1

..

DYNAMICS OF TRANSDUCERS

ELECTROMECHANICAL ENERGY CONVERSION

~~rr.

412

ELECTROMECHANICAL ENERGY CONVERSION

DYNAMiCS OF TRANSDUCERS

the Nyquist stability criterion (Sec. 5.4.1) will be applied.

by jw in the open-loop transfer function of Fig. 6-12 yields

gain K E represents the sensitivity of the sensing device that changes

position error to volts. The dimensions of the various constants are:
Ka
Km

volts per volt

meters per volt
radians per meter-second
volts per radian
reference position in radians
position error in radians
output position of hydraulic motor in radians

Kh
KE
R
E

i'i" t

- <

'T

C(jw) _
E(jw) - jw(-rjw

w"

1
W/o <'T

and

11

~~

l)[(~:r + 2~:: + 1] [(~:r + 2~h~ +

When this transfer function is sketched on a polar plot for values of w

from - 00 to + 00, Fig. 6-13 results. The key features of this figure are

,.
'i

usually have the

w=o

-n- _ ......
1m

I
.r

(6-55)

;~

, ,

~.

fV.-i>

",,

1l.

l)(S~
W"

~+
h

, jl,y;:

,""'.
,,~

'"

sr!'

,;1

;,

"

....

~.:

\ll

Fig. 6-12.

Simplified form of Fig. 6-1 I.

_1\,~~."

,/

'I
.

w .. 0+

Fig. 6-13.

The block diagram of Fig. 6-11 can be simplified to that shown in'
Fig, 6-12 in which the gain K v is

Kv = KKaKmK/o

(6-56)

In the design of a system such as that shown in Fig. (i-12 the character
istic frequencies of the torque motor and hydraulic motor are usually
fixed and the problem remaining is to set the gain K" by adjusting the
amplifier gain K a until the system is suitably stable. If the system thus
arrived at does not meet the required specifications, then compensation
can be added (see Sec. 5.4.3).
The first question that arises in the setting of the amplifier gain is:
Can the system of Fig. 6-12 ever be unstable? To answer this question
..
,,'
f"

{,

t
'"",

if
}""

- --- --

."'I ?,u'rf'f
R.
~:

I
.,

3-~',

'

:'left

(I.,

V"f

./
,/

tl!

\..,n"
:

1)

,,,-.-~.

. "",.:.,!t..

+2110

- :~

;.~i;

,:14I-,I;r

,i

Ky

bt

" \

/,'

/~. fW'
/

l'

~.

S(TS+ 1)(5 +2l"~ n +

(6-57)

1]

This is natural because the hydraulic motor operates at a much higher

power level than the torque motor, is bigger, and therefore should have
slower response than the torque motor.

Replacing.r

Kv

'\

liT, W no and Wh of Fig. 611

The characteristic frequencies

relations

413

'f
. (fr~{~' ;.1

t) l1.i

{.J'

Polar plot of open-loop transfer function.

that the function of Eq. (i-57 behaves as Ijjw near w = 0 and as l/Uw)6
near w = <:1), It is evident from Fig. 6-13 that if the gai n K" is too high
the locus can enclose the point -1 + jO twice; and, since there are no
right-h~llr-plane poles in the open-loop transfer function, the system will
be ullstabk:. Thus there is an upper limit on the gain for absolute stability.
There is no lower limit, however. The actual setting of the gam will
depend on the relativc stability desi red.
The rClative stability of this system can 'be set by the techniques given
in Chap. 5. With the relative stability set by the M p criterion, the
bandwidth will be approximately Wh when the characteristic frequencies
satisfy the relations of Eqs. 6-55.

'Q

..~
I

)j

414

6.2

ELECTROMECHANICAL ENERGY CONVERSION

6.2.1

Electrostatic Microphone

Insulator
L':"

,l~~'

Equations of Motion

dq+ 1.
= R df
C

if

Eo

1'1

i = dq

Microphone

71~

dt

"

~.

~$

~._

J!.

1- ',.

:..=..

Eo

Perforated metal
back plate

'.

"

Fig. 6-14. Simple capacitor microphone.

',.

."

,.

e(t)

iii

Fig. 6-1 S. Electric bias circuit for capacitor

microphone.,
"

'1:,,0
.~-.

If the area of the diaphragm and back plate in Fig. 6-14 is A and~!'
fringing fields are neglected, and if the diaphragm is assumed to move as'
a rigid piston, the capacitance of the microphone is

which in turn alters the terminal voltage which is the output variable.
As the diaphragm deflects. air is forced from the space adjacent to the
diaphragm through the perforations in the b'lCk plate into tht,; other air
space. The perl'orations resist air flow through them, providing damping
for the spring.-mass system. The air inside the air spaces. by virtlle of its

compressibility, provides a spring constant against which the diaphragm

acts.
The operation just described includes only those parameters which

affect the bandwidth of the microphone. A practical capacitor micro

phone has additional second-order effects which will not be included

here.
A microphone is used to record or reproduce sound; thus the bandwidth
or the variation of the characteristics with frequency is of interest.
Conscquently. til,' g.oal of tlK' present analysis is to study thc I'requcncy
response of thc capacitor microphone of Fig. 6-14.
."-t

,+' !

I~?

,.

0+

Cit)

Air space

i.

(6-58)

where q is the instantaneous charge on the capacitor and C is the

instantaneous capacitance.

-t))

415

The circuit conventionally used for electrically biasing a capacitor

microphone is shown in Fig. 6-15. Since the capacitance can vary with
time, the electrical equation of motion of the circuit of Fig. 6-15 is

A second example of an electromechanical transducer that will be

analyzed is the electrostatic or capacitor microphone. The configuration
assumed is the extremely simple one of Fig. 6-14. In the construction
shown the movable diaphragm is deflected by the pressure of a sound
wave. The diaphragm and perforated back plate, which is assumed

rigid in this treatment, form a capacitor which is biased with essentially

constant charge. The deflection of the diaphragm changes the capacitance

!~

DYNAMICS OF TRANSDUCERS

"'/, ~,;:,;:r~

oA
= -x

....

'!~~;",J!",t

..

"

.; i

: 1 .~

'J

(6-59) ;

When the distance x is expressed as

"

X(f) = Xo

+ oCt)

l!..ff.",

(6-60) ,

where X o is the rest displacement with Eo applied and 8 is the deflection

caused by sound pressure, Eq. 6-59 can be written as
"0 \C\WMKil'j(j;"t '1.""

Co

C= I

(6-61 )

+ (8/xo)

where
" ', ..,J

~-.'

"'''W~~'<~''''!'f C' _ ~oA

,

L. L. Beranek, Acoustics, McGraw-Hill, New York, 1954, pp. 157-164.

:.j

.~

0-

Xo

,;o(tilttf,\'-,

416

Substitution of Eq. 6-68 into Eq. 6-64 yields the complete mechanical
equation of motion

The electrical equation of motion, Eq. 6-58, becomes

dq
E o '- Rdt

417

DYNAMICS OF TRANSDUCERS

ELECTROMECHANICAL ENERGY CONVERSION

0)

+ -Coq ( 1 +Xo

(6-62)

.
d 2x
j = M dt 2

+ ex

dx
dt

,
K(x - xo)

q2

(6-69)

+ 2 COXo

The output voltage e shown in Fig. 6-15 is

= !L
(1 + ~)
Co
Xo

The elc;ctrical equation of motion, Eq. 6-62, and the output voltage,
Eg. 663, are nonlinear because of the qx product. and the mechanical
equation of motion, Eq. 6-69, is nonlinear because of the q2 term. These
equations can be linearized by assuming small-signal operation. Thus
the charge is defined by
(6-70)
q(t) = Qo + ql(t)

(6-63)

Next the mechanical equation of motion must be derived. The sound

pressure is assumed to be uniform over the whole diaphragm and the
diaphragm is assumed to move uniformly as a piston. The sound pressure
exerts a force f on the diaphragm. The diaphragm has a mass M D and
a spring constant K D The air on the left of the diaphragm acts as a
mass M A' The air spaces to the right of the diaphragm react on the
diaphragm with a spring constant K A and the perforations in the back
plate introduce the mechanical damping constant IX into the system. *
With the system thus defined the mechanical equation of motion is

d 2x

M -d
2
t

dx

a: -d
t

+ K(x - x o) -I.,

where Qo is the rest charge and is given by

Qo

(6-64)

The mechanical displacement is small when S/xo ~ 1. Using these

assumptions in Eqs. 6-62, 6-63, and 6-69 and neglecting products of
incremental quantities (qj and S) yield (realizing that Qo/Co = Eo)

Eo

(6~65)"

ox

J q2
1 q2x
W =--= - 2 C 2 O A

,114

(6-66)'-1

Substitution of Eq. 6-66 into 6-65 yields

1 q2 ~:

! .... - 2 6' oA

I'

Co

.b

J. = -

Eo 0
Xo

Co

(6-73)

,:~

(6-74)

Xo

r' .

f=

d 20

Md2

Eo

do

+ adt + KS + -qt
Xo

(6-75)

.~

:.;?,'~

o=

R dql

dt

+ !JJ.. + Eo (,
Co

Xo

,;.$

.
;

;,

/t"d

(6-76)

e - Eo

= '1

ql
Eo ~
-C + 0
o Xo

(6-77)

Equations 6-75, 6-76, and 6-77 describe the properties of the capacitor
microphone that are of interest in the reproduction of sound.

".

The terms -Kxo and -1[Q~/(Coxo)] combine to yield Kxo in the mech
anical equation. When the quiescent terms are subtracted from Eqs.
6-73 and 6-74 the incremental electrical equations result:

~~

For the derivation of the approltimations to the acoustic system used here see
Beranek, loc. cit.

I ~:
1 (nt'
"

~I"

(6-68)

Itt

~;~-

Putting this expression in terms of Co (Eq. 6-61) yields

1 q2
2 CoXo

'Lt',

ii
:;1'

[,

(6-67)

,I',; 1J'1.

+ !JJ.. +

'j

;,.
~~'

Qo
Co

e = Eo + !lJ.. + Eo S

~
~~

",:1

The electrical energy W. stored in the capacitance is

= R dql

dt

The method of arbitrary displacement and conservation of energy

derived in Sec. 1.2.3 can be used to determine the force!. by Eq. 1-58. . i

oW/q, x)

(6-71)

CoEo

and qj(t) satisfies the inequality ql ~ Qo. The output voltage is defined
as
(6-72)
e = Eo + ej(t)

where M = M D + M A
"._j
K = KD + KA
f. = mechanical force of electrical origin
_-.1\
X o = rest position of diaphragm with Eo = 0 ,!

1'. _

418

ELECTROMECHANICAL ENERGY CONVERSION

6.2.2

DYNAMICS. OF TRANSDUCERS

low frequencies the w terms in the denominator are negligible, giving the
frequency response
, TEo
jWEll
Kxo
(6-84)
F...--. o =
E1,Co
1--
K~

Frequency Response

Since a microphone is used to record or reproduce sound, it is desirable

to have a microphone with a flat frequency response over the range of
frequencies contained in the sound. Thus the dynamic charactcristics
of microphones are usually given in terms of the frequency response, and
often only the magnitude of the frequency response is given. The
frequency response of the capacitor microphone will be found.
The force J due to the sound wave is assumed to be a steady-state
sinusoid
J(t)

= F cos wI =

Re (Fe iw1 )

(6-78)

Ij
oj

]'"

.~

'0

In line with the techniques of Sec. 2.1.4, Eqs. 6-75, 6-76, and 6-77 can
be rewritten in terms of complex amplitudes as
F

[(jw)2M

o=

(jwR

E 1 = QI

Co

+ jwa +

K]il

~
Xo

QI

;Jo;';. ';'"

(6-79)

.'

t.;,! '

~'

E1 _
F -

+ ~JQ1 + ~: tJ.

(6-80)

r.

(6-81)

Xo

-0

jwR

. ex
+ jW+
K

i!~.

!c.1I~1

.5

'.,

loglO

'1

~~
..

,=

"

; \"'"

1t'<J11?;l-;.!

_ !('fc~

ex
T

(6-82)

RCo
ij'

,Qf'

[( j::!..)2 + 2,jw + I](jwr + I) _

~

loglO CAl"

Iq" '.

f'

~ ~T

loglO CAl

f::;-.....

"2~

'i,"

'?Q.~

TEo

~~d ~

<,.;;j

E)
F

jw Kxo

~
jwr

E~Co
K~

(6-85)

The break frequency of the denominator tenn is

E~C()

TEo
jw Kxo
~

-~

E5~o

(6-83)

above which the asymptotic plot of Fig. 6-16 goes to zero slope, yielding
the mid-frequency asymptote

The asymptotic log-modulus plot of Eq. 6-83 is shown in Fig. 6-16 for
the type of response normally obtained with capacitor microphones. For

il

This gives the initial slope of 20 decibels per decade in Fig. 6-16. As the
frequl:ncy is incretlsed from zero the first frequency-sensitive term to
become appreciable in the denominator is thejwT tenn, giving the response

Xo

... J

E1 _

1 (jwRC o + 1) _....J!

2YKM

Eq. 6-82 can be rewritten as

~~,

.,~\

= J~

Fig.6-I6. Asymptotic frequency response of capacitor

microphone.

Definingt.
wn

I - 40 (db/dec) slope-l
I
I

I:

;,it

20 (db/dec) slope

j'

+ Eo tJ.

Kxo
M
E C l(jw)2 K

;f

where tJ., QI> and E, are the complex amplitudes of 3, ql> and e\ respectively.
Simultaneous solution of Eqs. 6-79-6-81 for the ratio EilE yields

419

Ell

'Flmid

~.

Eo
= Kxo

(6-86)

420

ELECTROMECHANICAL ENERGY CONVERSION

As the frequency is increased further the frequency terms in the quadratic

factor in the denominator of Eq. 6-83 become appreciable, leading to the
approximate transfer function

-:--:....::K:.-.xo~ _ _

w-+CO

- ()2
j::':'" + 2~j _

Ji.

(6-87)

.)J

which gives the -40-decibel-per-decade slope starting at

6-16.
The break in the asymptotic curve of Fig. 6-16 at

Wn

in Fig.

E~Co
':"..-r) .

is due to a first-order lag and therefore the actual magnitude curve will
be below the asymptote in this region (3 decibels at the break frequency).
On the other hand, at the break frequency W n a resonant rise can occur
whose peak depends on the damping ratio~. For an underdamped
microphone (, < 1) the actual response curve will be above the
asymptote in the region of W = W n
In the design of a capacitor microphone the region of flat response
shown in Fig. 6-16 must be adjusted to include all sound frequencies that
are to be reproduced. The lower limit of the pass band is determined
primarily by the time constant T = RCo, and can therefore be controlled
somewhat by the resistance R in the bias circuit. The upper limit on the
pass band, W m is determined by the mechanical parameters of the dia
phragm and the air spaces. The damping ratio ~. which determines the
height of the resonant rise near W = W m is also determined by the para
meters of the diaphragm, the air spaces, and the perforated back plate.
The sensitivity in the pass band is determined by the bias voltage. the
spacing of the diaphragm and back plate, and the net spring constant of
the system according to Eq. 6-86.

'.

,':,

.10

,11
~ ~-il

(~l

,0
oiL

/ .''1

;::~Ll;':-J'C\~O

"1;;'7:

v,lfl'll

~\

-~

w=

Q'1

Dynamics
of
Commutator Machines

Wit

(J)n

(Jt'~

Eo

Ell

-F

';'..H\~,'

7.0

:'"

~'iI:
,If
,~

r'

~ii)

''

'>:;:.; ,

'1"

"

,~)t1V r!!ji d

')'. /

':4!fi~

)r!'

.,,' ;-.

'rlri
tl~

Introduction

~}

:,

,.' \'

"wi

..,. '\.

Commutated windings are used in a large class of rotating machines.

The characteristics of these commutator machines depend upon the
manner in which connections arc made to the stator and rotor windings
through slip rings or through a commutator, and upon the frequencies,
magnitudes, and phase displacements of the terminal voltages.
Of these commutator machines, the d-c machine is probably the most
flexible insofar as control and inherent characteristics are concerned.
D-c machines are used (1) as motors for providing bulk mechanical
energy with either inherent or external control of their torque-speed
characteristics; (2) as generators for providing bulk electrical power with
either inherent or external control of the output; and (3) as power
amplifiers for providing electrical power in accordance with variable
command signals in control systems.
Although this chapter will be concerned mainly with d-c machines
because they are perhaps the most important commutato'r machines, it
should be recognized that there are a-c machines-many of them rather
uncommon-which also utilize a commutator, and to which the present
analysis is still applicable. Examples of a-c commutator machines are
the a-c series and repulsion motors and their variants. the several types of
multispeed polyphase commutator motors (such as the Schrage motor).
and synchronous converters.
This chapter will deal with some of the more commonly known
C01l1l~,~:tator machines and will illustrate how their behavior and character
istics can be derived from the generalized rotating electromechanical
421
.":,.1:p!"

';'1

.j

f.J ",:

~1

"l.

....

,;;.

. \ .. ,

'''>

.~... ~ ;

;:r'i'~ ,t,

0422

ELECTR.OMECHANICAL ENERGY CONVERSiON

DYNAMICS OF COMMUTATOR MACHINES

' .p

energy converter introduced in Chaps. 3 and 4 through the imposition of

specific constraints which define the particular configuration. The
emphasis will bt: upon the dynamics of these machines because of their
importance in control and power systems.

423

These rotor windings have the property that the fields set up by currents
in them are stationary in space, but have voltages generated in them due
to rotation.
'~r 1

7.1

Equations of Motion of the

Four-Brush Commutator Machine

~'"

~
'..

+9

I~

In Chaps. 3 and 4 the equations of motion for the generalized electro

mechanical energy converter are written in several equivalent forms,
depending upon the choice of coordinate system. In particular, when
both stator and rotor coordinate systems (reference frames) are attached
to the stator, i.e., stator afJ and rotor dq variables, the equations of motion
as given in Eqs. 4-50a and band 4-55b are of the general form

.,

sr = (PJls.r
'v ~~.dq
u#,dq

and

s. r
J <!J S"~,dq
)iafJ.dq
"
'f'

'ii,

,,'

q+

'

(7-la)

.;.,

t'

+ p.:f""uP.dq +

U'

= Jep:I + a'llJ

'1 ,

eaiJ,dq/

, .r
"~.dqi uP,dq

f'"

u r

(7-1b)

The dq system of rotor currents is one producing fields in space that are
stationary with respect to the stator (Sec. 4.3). It can be obtained by
means of fixt:d hrushes sliding over the rotor surface. Ina physical
machine, the fixed brushes actually are sliding over an extension of the
rotor conductor structure called the commutator. * Essentially, the
commutator provides a set of available rotating terminals for each coil
placed on the rotor. From these terminals, which are formeo II1to a
cylindrical structure, currents are either injected into or collected from the
rotor through brushes which bear upon the commutator to act as sliding
contacts.
The space position of the stationary field is a function of the position
of the brusht:s, To establish the dq system dcflnt:d previously, tht: brushes
on the rotor must ht: so locatt:d that in the rotor field the d axis lines up
with the a axis of the stator and. similarly. the q axis lines up with the
fJ axis of the stator. When viewed from this system of coordinates. the
generalized electromechanical energy converter may bt: repn:senteu as an
equivalent two-pole, four-brush commlltator machine illllstratt:d in Fig.
7-1. Notice that in this figllre the rotor is represented by two Sl<ltionary
windings whid\ an: in lint: with the dircct- and quadnttlll'l:-axis brllshes,
For a discussion of commutators. see M. Liwschitz-Garik and C. C. Whipple.
Ell'ctri<' Machines. Vol. I. Fundamelllals 0/ D-C Alachilles. Van Nostrand. New York,
1946, E. Fitzgerald and C. Kingsley. Electric Machinery. McGraw-Hill, New York,
1952.

---<>

/3' axis

'.

,r

Iq

6.

'IJ

.6
+

Uq

u{j

Fig. 7-1. Schematic representation of the four-brush commutator machine.

i

The equations of motion governing the dynamic performance of the

four-brush commutator machine are those of Eqs. 7-10 and b which have
been derived in Chap. 4. These equations may be written explicitly using
the notation established in Eqs. 4-560 and b as
J;

"

v'fJ

ji

I~
f

R~

v'

vd r
vqr

+ pL~

pL~d

Rl + pLl

pL'i.

pL'fu
.pGrs

J,e"
) JfJ

q"

rs
pLqfJ

R'" + pL'd
~Grr
.1<1
rr
_ .pCqd R'q + pL'q

'I

I"

lfJ

/'1

(7-20)

.,"

Iq

and

~l

= J~ + a~

s
[Crsi'i
J#dfJ - Crsj'j'
qaqa

rr
+ (Cdq

_ Crr);,;']
qd dq

(7-2b)

424

DYNAMICS OF COMMUTATOR MACHINES

ELECTROMECHANICAL ENERGY CONVERSION

In deriving Eqs. 7-2a and b in Chap. 4 (Eqs. 4-56a and b), it was stated
that in conventional commutator machines the speed coefficients G differ
in magnitude from the inductance coefficients L. For example, with
sinusoidal current sheets on stator and rotor, nL~: = G;~; but, for practical
commutator machines which have nonsinusoidal current sheets,
nLd: *- G~~. The speed coefficients and inductance coetJicients of a
machine with nonsinusoidal current distributions are derived in Chap. 11
by the technique of space harmonic analysis. It is shown in Chap. II
that the inductance coefficient, e.g., L jk , is determined from the flux
linkage of the jth winding per ampere in the kth winding, whereas the
speed coefficient, e.g., GJk , is determined by the space integral between the
jth pair of brushes of the product of flux density produced by the kth
current and the turn density of the jth winding. For a commutator
machine with uniform turn density on the rotor, which is the construction
normally used, the speed coefficient becomes merely the product of the
total flux between the jth brushes produced by the kth current and the
number of conductors in the jth winding. Speed coefficients can be
found in practice from a knowledge of the flux per pole and the number
of rotor conductors, or they can be found directly from test data on an
unsaturated machine. Since the speed coefficients are determined by the
pole flux, it becomes quite easy to account for saturation effects in machines.
Strictly speaking, saturation invalidates the use of linear transformations
such as the dq transformation; however, since the speed coefficients are
determined by the total flux and turns, the effect of saturation on the total
pole flux determines the effect of saturation on the speed coefficients.
In commutator machines saturation is important and must be considered.
It will be treated after the linear (unsaturated) operation has been
illustrated.
The commutator machine shown schematiqlly in Fig, 7-1 can be used
in a variety of ways. For instance, if the windings are suitably eliminated
from one stator axis and one rotor axis, the result is a representation
of conventional doc motors and generators. When additional stator
windings are added in the axis in which the stator winding was retained
above, the additional windings can be used to control the characteristics.
of the doc motor or generator. One example is series-compounding of
motors. Retention of all the windings in Fig. 7-1 yields a machine which
can be operated as a two-stage control-type generator (metadyne).
When additional stator windings are added to this configuration, the
possibility of several types of feedback between the two stages results.

This feedback can be used to obtain the desired characteristics for a

partiC1J:.lf control application. For example. amplidyne characteristics
can be obtained.
In the sections to follow, single-axis machines will be studied first.
Simple types of operation will he considered. and these will he generalized
to describe an M-field, single-axis d-c machine including magnetic
saturation. Then the two-axis machine wiIi be considered. building up
from simple examples to the most general type of operation including the
effects of magnetic saturation.

7.2

Single-AXiS D-C t1achines

;~.

The term "single-axis doc machine" means the use of a single axis
on the stator and a single axis on the rotor. Inspection of the electro
magnetic torque in Eq. 7-2b shows that in order for a single-axis machine
to convert energy the a stator and q rotor or f3 stator and d rotor axes
must be retained. Either choice is suitable; however, to obtain motor
action for T. > 0 and ~ > 0 the f3 stator and d rotor axes are retained
(see Sec. 4.3.3).
The retention of the f3 stator and d rotor axis windings yields the
schematic representation of Fig. 7-2 in which the convention of defining
the stator winding as the field winding (subscriptf) and the rotor winding

----0

'I

,l..;

'1."

,.I-,j

,I

v..

'.
I

rHl!t

i --

;0
:ii)

Vf-+

!If

:'d'f

'~

,r.:-.

'11

Ii.
".";,,,, ,:t
.~.,.

Ad,:

~".

.#

'1~'

"ti ..... t

'''',.1
~.

Fig. 7-2. The single-axis d-c machine.

".lft,;

~~. \,~

"

"""';l

if

For additional infonnation on metadynes, see A. Tustin, Direct Current Machines

for Control Systems, Macmillan, New York, 1952. J. M. Pestarini, Metadyne Statics,
Technology Press, Cambridge, and John Wiley, New York, 1952.
.
, ) '1

-425

I,

).

iJj

426

ELECTROMECHANICAL ENERGY CONVERSION

DYNAMICS OF COMMUTATOR MACHINES

as the armature winding (subscript 0) has been used. Eliminating the IX

stator and q rotor variables from Eqs. 7-20 and b and changing the
nomenclature to agree with Fig. 7-2 yield the equations of motion for the
single-axis doc machine shown in Fig. 7-2.

an easily applicable method for solving dynamic problems. Using the

techniques of Sec. 2.1.5, Eqs. 7-30 and b can be represented in a block
diagram form which is suitable for use with analog computers. Such a
block diagram is shown in Fig. 7-4 in which the operator p appears only
as an integration lip. Note that in Fig. 7-4 there are three inputs vf'
Va' and T and three outputs if' i a , and~.
Thus, the volt-ampere and
torque-speed characteristics of external devices can be added to the
representation of Fig. 7-4 to obtain a complete system representation for
a system containing a single-axis doc machine.
Although the representation of Fig. 7-4 is general and can be used with
an analog computer to solve general dynamic problems, many problems
can be solved analytically by linearizing the representation of Fig. 7-4.
Such solutions are less accurate than the general solutions; however, the
accuracy is often sufficient for engineering applications and the small
loss in accuracy is more than offset by the added ease of solution. In
addition, linearization of Fig. 7-4 allows the use of the general theorems
of linear systems to more easily define behavior.
For. the study of such problems as small-signal stability of feedback
control systems, it is often helpful to linearize nonlinearities around
steady-state operating points. The techniques for accomplishing this
linearizatio ll are given in Sec. 2.1.5. Defining the variables in Fig. 7-4 as

_ fRf

+ Lfp

.:'d - LGaA
T = J~

IX~

l i;l

Ra + L~ ~

T~ = J~

IX~

(7-30)

Gafifia

(7-3b)

The machine (Fig. 7-2) represented by Eqs. 7-30 and b is essentially

the three-terminal-pair device shown in Fig. 7-3. This figun: in con
junction with Eqs. 7-30 and b can be used to illustrate the energy conversion
and control properties of the device. Inspection of Eg. 7-30 shows that
the field circuit is a simple R-L circuit independent of armature and
mechanical quantities. On the other hand, the field current if appears in
Field
-----.
If

_..'0_ .-

~t

Armatur .. i Iio

n-----L
Fig. 7-3.

,I

tif

D-c

,I
J

'J

j.

i
~

~q;,

machine

,,:'

TI

Shaft

i/ = If 0
Va =

The single-axis do<: machine as a three-terminal-pair

T =

t'

VaO

+ Val

+ i al
To + T I
Q O + Wt

i a = laO

network.

the armature and mechanical equations. The mechanical speed ~

appears in the armature equation and the armature current ;" appears in
the mechanical equation, indicating bilateral coupling between the
armature and the mechanical system. Thus Fig. 7-3 can be interpreted
with the aid of Eqs. 7-30 and h to represent a system in which power can
flow between armature and mechanical terminal pairs with the field
circuit acting as the controlling element. Since the control power
required in the field is usually less than the controlled power flowing
between the armature circuit and the shaft, Fig. 7-3 represents an elect!"o
mechanical power amplifier.
In general, the equations of motion, Eqs. 7-30 and b, for a single-axis
doc machine are nonlinear due to the rf,if product in the armature equation
and the ilia product in the torque equation. These nonlinearities make
the solution of a general dynamic problem by analytic techniques difficult.
However, when such generality is necessary, an analog computer atTords

+ V/I.
+ i/I ;

v/ = VfO

I-----()

. ,~~.- ~

~ =

427

(74)

where the subscript 0 represents constant values at a quiescent point and

the subscript I represents small amplitude time functions. The use of
the linearization of Fig. 2-2 for the two product nonlinearities of Fig. 7-4
yields the linearized incremental representation of Fig. 7-5 in which
VfO = IfoRf
VaO = laoRao

To = aDo -

GafQolfO

(7-5)

GaflfOlao

The linearized block diagram of Fig. 7-5 is useful in studying the small
signal properties of the single-axis doc machine either as a separate
component or as one component in a larger system. For instance, the
study of the small-signal regulating properties of a doc generating system
in which the dynamic properties of the mechanical drive must be included.

I.~ -~~---------""':~~~~--l'

428

ThcII; are scveral additional ways in which Eqs. 7-30 and h or Fig. 7-4
can be linearized, depending on the application. These techniques of
linearization will be illustrated in the following examples.
.".

EXAMPLE 7E/
.~

.~
.c:

.....

a'?

~e

'2

.~

btl

.S
<II

~
';;j

g.

,d

.>oc

.J:J

"3

.g

.;4

].

"

~f' ..

~, .!'

'"

~;!,.i";.

ii:
(~

'J

)~

'1~

'l~i

"~', ,:~'" ~'

'"

.~

,,'

'.~

'!i;;,

u.

,.il{
1,,:
'1:
,I".

.'"). .r'
~ \'

;i" ';"'I;'~

;~'"

;~t

.1r,.

~~

t.

t~,

'Pt"ft

.,:

:61 ' "

,

" !

~) !'~

1'1
.;.:

"oJ

,',,",

nq

,r~

1J ,

(Jp

(7EI-2a)
(7EI-2b)

(GafIa)if

and it reduces the representation of Fig. 7-4 to that shown in Fig. 7EI-1.
The torque-speed (T versus J,) characteristic of the mechanical load on
the motor can be added to Fig. 7EI-l to complete the representation.
I'

(/J

~i-

"'k,,

+ Lfp)if
+ a)</> -

'~

;,(~,,;
It;
;,~., .

.!.O

; j"J'

(7El-I)

I a = constant

J!", ','

Vf = (Rf

"C

~,;

i"

This constraint reduces Eqs. 7-3a and b to

~
.rl

i:

,]

,'11;, .

'0

'0

,-"

)'

I""='

j"

'?

"C

The D-C Shunt Motor

When the machine of Fig. 7-2 is operated as a motor, the linearized

representation of Fig. 7-5 can be used for small signal studies; or, for
special types of constraints, Eqs. 7-3a and b or Fig. 7-4 become linear.
These constraints are:
(a) Field control with constant armature current.
(b) Armature control with constant field current (often called Ward
Leonard type of c o n t r o l ) . l l
(a) Field control
.;;,
, .;;.\t4
For field control the armature current i a is constant:
-tf

.~

429

DYNAMICS OF COMMUTATOR MACHINES

ELECTROMECHANICAL ENERGY CONVERSION

/)f

t4

I_I '4>i~

Tf

t:.

!$ .

{.,

I{

,~

',tfF~~ I '.:1:,;:'1
:'1~!t

(!t'j

1.l1V<'

.:

if ."

!'

Flg.7EI-I. Block diagram for field control (ia

fa

= constant).

.', '

11:
',,'lE .
~:t

tt:~

'

! , if:

. ,,,

..

''?t.''j':{
",' .,."
if,: . "

4'

The block diagram of Fig. 7El-l can be simplified to that shown in

Fig. 7EI-2 in which it is seen that the field voltage vf controls the speed
through two cascaded time constants: the field time constant 'r! = Lf/Rr
and the mechanical time constant T m = Jla. The applied shaft torque T
affects ~~e speed </> through a single time constant: the mechanical time
constant T m'

..<I
. <I

~~.<~.~.

..;;

..

'~

430

ELECTROMECHANICAL ENERGY CONVERSION

DYNAMICS OF COMMUTATOR MACHINES

(b) Armature (Ward-Leonard) control

For armature control the field current if is constant:
if

If

= constant

,pI
V;; 7'000

(7EI-3)

This constraint reduces Eqs. 7-3a and b to

Va = (Gaflf)~
T = (Jp

(R a

LaP)ia

a)~ - (Gaflf)ia

(7EI-4b)

..,

l'

IJ{

= L(/R{

Tm

..'f'
,;

L~\

1/(Ga/I/)

- 1

(7EI-6)

+ TjP + Tj'TaP 2

where'Ta = LalRa and 'Tj = JR a/(Gaf l f )2.

It is interesting to note that Eq. 7EI-5 is of the same form as an R-L-C
series circuit for which v = i(R + Lp + I/Cp). The quantity J/(Gafl )2
f
may therefore be considered as an equivalent capacitance C (or 7) = RaC).
Thus a motor with armature control, driving an inertia load, can be
treated as capacitive loading. Often the armature inductance of the
motor is small enough to have negligible effect upon the system dynamics;
in this case, the time constant 'Ta is assumed zero in Eqs. 7EI-5 and 7EI-6.

(7EI-4a)

and it reduces Fig. 7-4 to the simplified linear representation shown in

Fig. 7EI-3.

T{

..31

and

!/>

:;t

J,

=J/Cli

Fig.7EI-2. Simplified block diagram for field contro!'

:'1::&

1',," L,,/R a

\~f{j>1

Tm
1

=J/Cli

'

, "f

('i>
~

!/>

r,
"

.1

liRa

+ TaP +

(lhp)

E.XAMPLE 7E2

Separately EXCited D-C Generators!

surplying torque T at speed ,p as depicted in Fig. 7E2-1. Since the

armature current i a was defined in terms of power flowing into the terminals
as shown in Fig. 7E2-1, the load current is given by

The block diagram of Fig. 7El-3 can be simplified to that shown in

Fig. 7EI-4 in which it is evident that the armature voltage Va affects the
speed ,p through two cascaded tilne constants with negative feedback
around them. The shaft torque T affects the speed ,p through one time
constant T", with the other time constant To in a negative feedback loop."
In many probkms using armature control the motor is orten assumed ;;i
to be driving ,I pure inertia load (a = 0). In this case the following
transfer functions can be obtained from Fig. 71-3:
'!I.' '\
ia\

Flg.7EI-4. Simplified block diagram for armature control.

ZL(P) connected across the armature terminals and a mechanical driv~

Block diagram for armature control (i, = I, = constant).

V;; T-O

,.: . ,lffi~j;

Consider now'a'separately excited doc generator with a general load

J
Fig.7EI-J.

,.

.~ \,

iL = - ia

. (7E2-1)

and the load voltage is given by

VL

ZL(p)iL

-ZL(p)iu

Va

(7E2-2)

The general problem of the doc generator can be solved by using the
general representation of Fig. 7-4 or the small-signal problem can be
solved by using Fig. 7-5. In both cases the constraint of Eq. 7E2-2 must
be added to the representation. It often happens that the mechanical
drive is a sufficiently good speed source for changes in the generator

(7EI-5)
"

ELECTROMECHANICAL ENERGY CONVERSION

DYNAMICS OF COMMUTATOR MACHINES

load to cause a negligible change in speed. In this case the speed can be
assumed constant
(7E2-3)
~ = WQ = constant "

In this case the generator and load can be represented by the block
diagram of Fig. 7E2-3. In this figure the generator is considered to
contain unly the field time constant 7f

432

ia

k----- Generator ----;'MI....- -

if

, ". uf

Uf

[]
+

7.2.1

Zr}p)

I,

'~?:'l:l,
.,;,j'

,"l'

(t~~tt

. :-~t-' ,

lI:,I

and Eqs. 7-3a and b become, with the addition of the constraint of
Eq.7E2-2,
(7E2-4a)
VI = (RI + LIP)il

+ ZL(p)]iL

Fig. 7E2-3. Block diagram representation of the d-c

generator with load.

Fig. 7E2-1. Separately excited d-c generator.

(Gafwo)if = [R a + LaP

) ;j{r~'- i,~;:{

'&

-------0

433

'"

Magnetic Saturation Effects

The general block diagram of Fig. 7-4 and the subsequent simplified
block diagrams are representations of single-axis doc machines with no
magnetic saturation. In many applications it is necessary to include the
effects of saturation. Strictly speaking, the equations of motion, Eqs.

7-3a and b, do not apply if saturation is present; however, when saturation

is included in these equations and block diagrams in an approximate
way, sufficiently accurate results can be obtained for many applications.
Flux

(7E2-4b)

linkages
(Vn

)0('

q,

These equations can be represented by the linear block diagram of

Fig.7E2-2.

'r--'~

,~,~

--.;.

....._;
,

l"Air-gap line

/
/
I

r
1
R a + LaP

i&

!,I f

Field current if

+ ZL(P)
"
I

""'1

!~ :~

;.;"1"

Fig. 7-6. Saturation curve of single-axis

d-c generator neglecting hysteresis.

Fig. 7E2-2. Block diagram of separately excited d-c generator driven by

constant speed source.

.' i. ~

"~

IT

"f
t

It is often convenient to lump the armature parameters La and R a with

the load impedance

RQ

+ LaP + ZL(P)

= Z~(p)

(7E2-5)

,;VIi\,,;'
. ,:; ~

and to define the internally generated voltage ,;

'-,ii,
:",'

i"
"

.\i

! ,~~.

!J&'

JH\

vg = (GaJWo)il

"d,

(7E2-6)

,j{1f

1:;_,'f
I",'

1~'

In a single-axis doC machine saturation occurs principally in the flux in

the field axis. This saturation has two effects: (1) it causes a reduction
in the field circuit inductance; and (2) it decreases the voltage generated
in the armalure circuit. The effects of saturation in the armature axis are
neglected in most single-axis machines.
To a first approximation there is no interaction between the field and
armature fluxes; consequently, a measure of the flux saturation in the

>'"

'"f;

.1

.,....."l ..""

.,,,,.>
.. "~' f.

;.-.1,.';

I
i

,;.t

!
J,(

"-

-l

>1' '

ELECTROMECHANICAL ENERGY CONVERSiON

DYNAMICS OF COMMUTATOR MACHINES

field axis is obtained by plotting a curve of the open-circuit armature

voltage per radian per second of speed as a function of field current.
Such a curve is shown in Fig. 7-6 in which

When the field leakage inductance L/1 of Eq. 7-7 is included in any
external impedances or neglected. Egs. 7-3a and b can be written in terms
of the saturation factor SUI) defined by Fig. 7-7 as

<43<4

.pm/

A (Va)oc

where A is a proportionality constant, and .pm/ is the part of the field flux
linkages that link the armature circuit. The self-inductance of the field
is defined as
(7-7)
L/ = L/I + Lm/

Lml
Lmlo

= ----'!!..-

i
f

if '" S

Galo

+ SLm/op)i,

(7-9a)

Va = SGQ/oi,

+ (RII + LaP)iQ

(7-9b)

T = (Jp

+ rx)~

- SGa/oiai/

(7-9c)

In these expressions, since S is a function of if' it should beoperated on

by p in Eq. 7-9a. The resulting term is assumed to be negligible to
simplify the analysis.
8- 1

v/ - pL/h = v = (R,

(7-6)

435

""-Unity slope

/~Unlty slope
/

/
I

);

:C'

)'J

L.".

if

L."o If

G., .

= G.,o If

I
I

,,!\.-,.;

Fig. 7-7. Saturation curve of speed

coefficient.

'(

~);\

Fig. 7-8. The inverse saturation factor

'" 8-1 = liS.
._':

.;

where L/I is leakage inductance representing flux that does not link the
armature and L nll is the self-inductance due to IPml which does link the,
armature. From the definition of inductance as flux linkages per ampere'
L",/ =

.p,:"

(7-8)

1/

and in terms of the unsaturated mutual inductance L m / o , the inductance

L ntj is given as a function of if in Fig. 7-7.
In addition to the effect on the self-indllctance of the fteld. saturati9n
also reduces the voltage generated in the armature circuit. This saturation
can be accounted for by making the speed coefflcient Gal a function of
the field current. If the constant Gala is defined as the speed coefficient
in the unsaturated machine, the speed coefflcient for any fteld current can
be found from the saturation curve defined in Fig. 7-7. Note that at the
origin the curve of Fig. 7-7 has unity slope.

1'...:.'

,J..
J
I:
jif
I '~<

,}! "
,.~

..

i_
t<

Equations 7-9a-c can be represented easily in block diagram form when

the inverse saturation factor S-1 is defined by interchanging the axes of
Fig. 7-7 to yield the curve of Fig. 7-8. In terms of this inverse saturation
curve S-l the block diagram for the single-axis doc machine is shown in
Fig. 7-9. Figure 7-9 can be compared with Fig. 7-4 to find how saturation
affects the machine. The block diagram of Fig. 7-9 is suitable for use
with analog computers because the operator p appears only as an
integration lip and there is one nonlinearity S-l which can be represented
with a nonlinear function generator.
.
The nonlinear block diagram can be linearized for small signals around
a quiescent point in much the same way that Fig. 7-4 was linearized in
Fig. 7-5. In this case the block S-l is represented by a constant equal
to the slope of the curve at the quiescent point.
The saturation function S-l can easily be included in the simplified block
diagrams of Examples 7EI and 7E2.

-436

ELECTROMECHANICAL ENERGY CONVERSION

DYNAMICS OF COMMUTATOR MACHINES

7.2.2

ol

~
<>

c:

.g
~

.~

~}

'~

<> 0
.c
<J

:e;

"0

.2

General Source of Field Voltage'

When a single-axis doc machine is used in a feedback control system,

the field winding is often fed from a frequency-sensitive source. This
occurs mostly when the field is fed from another machine or through a
compensating network.
Suppose that the field winding is fed by the general source depicted in
Fig. 7-IOa. The coupling network is linear and passive but may have
energy storage elements in it. The source can be represented by its

~.~

.~

437

':J

<J .

c:"'O
._
"'0

o .ia'"
I-

" -5

~ ~c:

l1(t)

;g

C ~
o .-
.~ t; "';:.':'"
... 0 <>
3 E ~

Linear
passive
coupling
network

:
, ..;

..

lJ

(a),)
'

'"
'"
c: ::l
0

u '''::
v

b1

'"
<>
.c

~.::

IZ'(p)!

'" >. 0

.... ... , E .0
' ...

t>Il

,5

'l;l

II,

"

0+

c:

.g

-g ; I::.
v -

.5 ....-
'V

(b)

..

"" ';;

.5 +

~ ~-~

'E1;']
" .

'~l

(J)

:1'

'f"

Fig. 7-10.

.~ 13'~

Thcycnin equivalent shown in Fig. 7-lOb in which the open-circuit voltage

is

~~"O

;,.qtf:,:.

-Itl--
104

I~ ,.

; }J:I,' "

<l;

i";,

'----To-:;H
f

:'

:e
L~g

.I"

~ ~:r,

11
;f ;{'

:"

.~~

,:

U.

){rI.~

.,

:G

.5 c

>

'~.l~n

....,

t ..

"

~{
1j-::~.i\

,
~ ,

'"

-~.~

~'t

I,e:'

I '\.

.\

(7-10b)

The circuit of Fig. 7-10b is described by the equation

J. '

)J.'(

,r

&

"v-;

w.

..".

c:

'0

''',1

(7-lOa)

I se

;~>l4f.;

:'}1'

,.,'

= O(p)rJ

Z'(p) = ~oe

" . '<llf.

8 :.:!

,p

. I

(I)

.,; c::

voe(p)
and the short-circuit impedance is

'

>,i11

c:

_ '-l

..!. ;g
. g

ll~,t

g ].5
~
ps .5 r:-'

t ; ~.rr'11f ~: t, , ;

1, '_P':\"\

'itJ1"J'

::l

'0 ~.8

..\ l

:!!

fJ., (1'

",,,,

"1;l.o"

J'

1,

"ilJ .. E

.5 E

bi)
.~

l~'

(a) General network supplying field voltage.

Y",,-
'U-.

vel::.

\."1,".'

General field supply circuit.

(b) TMvenin equivalent circuit.

-\:

O(p).v = IJZ'(p)

v,

,."
(7-11)

When Eg. 7-11 is combined with the field volt-ampere equation, Eq~
7-90, which includes magnetic saturation, there results
>~

O(p)v = [R[

+ pL[1 +

SLm[op

+ Z'(p)Ji[

(7-12)

t:'!'

'

,,;.,

t:~

~:;,.

438

ELECTROMECHANICAL ENERGY CONYERSION

DYNAMICS OF COMMUTATOR MACHINES

439

When a new impedance Z(p) is dctlned as

/'

Z(p)

i ,
.~

..,

O(p)v = [Z(p)

,~

'~

cc:
g

...

=- R/ + pL/1 + Z'(p)

(7- J)

Eq. 7-12 becomes

~.

-5

'1t

t)

,"

(7.14)

Equation 7-14 can be combined with Eqs. 7-9b and c to yield the nonlinear
block diagram shown in Fig. 7-11, which represents the single-axis doc
machine with frequency-sensitive networks in the field circuit and with
magnetic saturation. It must be remembered that the impedance Z(p)
contains the resistance R f and the leakage inductance L/lp as additive
factors.
7.2.3

J. .t.;'.

+ SLmfop]if

Single-Axis D-C Machine with M Field Windings

,fi;

Single-axis doc machines are often constructed with more than one
field winding. For instance, compounding windings in doc motors and
generators are additional field windings because they are wound on the
stator and have the same magnetic axis as the primary field winding.
Control-type doc generators may have several field windings for control,

'(
"0

'~

---<>

--<>

-0

.~

'0

~)

,,,h

c:

.~
~

-I?~
--<I

-r-

("

.i..,
.;(

g .

~
. ~- l:,.:

....

Fig. 7-12.

e/l

'.

<\)

OJ)

1lO':::
ii: ~.

c:,'i!":)

~J

M-field, single-axis d-c machine.

:50

o:J

~,vri

compensation, feedback, and other purposes. Thus, to complete the

treatment of single-axis doc machines, the eq uations and block diagrams
previously presented will be generalized to represent a single-axis doc
machine with M field windings.
A single-axis doc machine with M field windings is represented
schematically in Fig. 7-12. Since all of the control windings have a

-,

.
440

DYNAMICS OF COMMUTATOR MACHINES

ELECTROMECHANICAL ENERGY CONVERSION

common magnetic axis, each control circuit will have a volt-ampere

equation similar to the field volt-ampere equation of Eqo 7-3a except
that each equation in the M-field case will contain mutual inductance
coupling terms with all of the other (M - I) field windings. In addition,
the armature volt-ampere equation will contain a generated voltage from
each field current and the electromagnetic torque T, (see Eqo 7-3b) will
have a term for each field current. The changes caused by having M
field windings are illustrated in the equations

effective in generating armature voltage, the relation among speed

coefficients is

";j

Gak
N k

,;-

1
"

where

Te = Jf>

+ a~

Ga1itia - Ga2 i2i a - .. ',- GaMiMia

(7-15b)

c~f

Vk

= LJ/,

i,j

= 1,2,3, ... , M

L k)

Nk

NkNJ

= open-circuit voltage-transfer function of coupling network

Zk(P)

in kth winding
short-circuit impedance of coupling network in kth wind
ing including resistance R k and leakage inductance pLkl

','

IV;
L k ) = N k L kk

,j.

;'

j!;

!:!~ ;;1' , '

2:

Gaji)

j-I

+ (R a + LaP)ia

;<i
(7-21)

(7-22)

The quantity I GajiJ which appears in Eqs. 7-21 and 7-22 can be obtained
j=1

4.1

from Eq. 7-20 by solving for ik> multiplying by Gak , and summing over
all k(l, ... , M).

~ Gak1k. --

L.

k=1

~ VkOk(P)Gak

L.

k-1

Zk(P)

ak ~ L .]
~
[G
L.
- - L. P kJ'j
k=1 Zk(P) j_1

Lkj

= --.!..Lkk
Nk

r
'~

(7-23)

f"""-'

Nj

(7-18)

'~,

,t'e

(7-24)

~ G . L.
ak1k -

,,1

(7-25)

Thus Eq. 7-23 can be rewritten as

k-l

~: 1 '

Gal = N Gak

--

0'

';1'

!:

From Eq. 7-18 the self-inductances are related as

~~

' i,

,I

where N k and N) are the numbers of turns in the kth and jth' control
windings respectively, Similarly, since the mutual flux. of the field is
l'

At

and from Eq. 7-19 the relation between the speed coefficients is
~

i'

j-I

(7-17)

or:

i'

kth winding

T = Jf + a~ - ia L GaiJ

-2=-

.",

(7-20)

j~

and the torque equation is given by Eq. 7-15b as

where L kl is the leaka~ inductance and L kk is the mutual inductance.

To simplify the analysis and yield a representation that is simple but still
useful, the assumption is made that the same mutual flux links all windings.
This assumption is reasonable because all of the field windings are
physically wound on the same magnetic structure. The condition of
equal mutual flux leads to the relation among self- and mutual inductances
L kk

+ L1 pLkJj

o k(P)

Va

(7-16)

+ L kk

Zk(p)i k

The armature volt-ampere equation is, from Eq. 7-15a,

After a block diagram representation has been obtained for no saturation,

saturation will be included in the approximate way described in Sec. 7.2.1.
Each self-inductance in Eq. 7-15a can be written as a sum of a leakage
inductance and a mutual inductance; thus, for the kth control winding
L k = L kl

= signal voltage for

In, writing Eqs. 715a and b no magnetic saturation was assumed; con
sequently, the field mutual inductances obey the relation

LIj

(7-19)

N)

VkOk(P)

~.

N + a~

= Gaj

Equations 7-15a and b, 7-18, and 7-19 can now be used to obtain a
block diagram representation. For generality, assume that each control
winding is fed by a source through a linear, passive network. Then,
according to the treatment of Sec. 7.2.2, the kth control winding circuit
can be described by the expression

~'

/ ..1

t:;:

T=

441

~ VkOk(P)Gak _ ~ :;! pL kk G .
( )
L. L. -~aJ')
Zk P
k-1 j-I Zk(P)

L.
k-I

(7-26)

\lrlJ

+42

ELECTROMECHANICAL ENERGY CONVERSION

~ G'
.:..

u"/I<'"

'1- v"O"(p)Ga,, _ [~ pL u ~ G .]
Z I< ( P)
"-' z ( ).:.. uJ'!
1<-1 -I< P 1~1

k-I

(7-27)

.!

is simply

Ii

(7-28)

The physical meaning of the term

L Gu"i"

can be established from Eq.

"-1

7-21, where it is seen that the open-circuit armature voltage

Vuo

Vuo = </>

2:

1-1

Gail

J.-;

j.

1
~.
,

2: Ga"i"

from which it follows that

is the open-circuit armature voltage

I<~I

per radian per second.

Vao

Thus Eq. 7-29 can be written as

</>

~ v"O"(p)Ga,, _ ~ pL u

Vao

I<~" Z,,(p) ~

ZIp)

1<::'1

(7-29)

"

oj

or
\ 1!"OIp)Ga"
Vao

Zk(P)
L

1 k::'1
'I'

1 + ). P
":;'1

. "'.*.
''tt
~

(7-30)

kIt

2: to

give for either il< or

k=1

.
/" =

"

1"

1
Vk = o-()
k

i
1

~'I

t.

v"

1{.

Vk0IP)
pLkk Vao
Z,,(p) - Zip)Gul<

or

..
1-.'

Z,,(p)

A simple relationship between the open-circuit armature voltage Vao and

the inductances and impedanccs in the M field circuits has been obtained.
Another relationship can be derived using Eqs. 7-27 and 7-28 which is
of particular valllc in analyzing multifield machines in which the field
windings are connected to the output. For these cases, which will be
studied in Example 7E4, it is adYantageous to be able to evaluate either
Vk or i k in terms of the open-circuit voltage. This can be obtained from

Eq. 7-27 by eliminating the

[z ().
kP

I"

pLu I Vao ]
+ -,-

Cal< '!'

(7-31)

(7 32)

The usefulness of Eqs. 7-31 and 7-32 will become apparent in the later
examples.
Now that the general relationships needed to characterize the M field
machine have been derived, Eq. 7-27 or 7-30 can be combined with Eqs.
7-21 and 7-22 to yield the block diagram shown in Fig. 7-13; this represents

r;

l~

! .

i
I,
.'

~--

Ii.,

iii

DYNAMICS OF COMMUTATOR MACHINES

The last term in Eq. 7-26 can be rearranged and the indices of summation
changed to yield
1<-1

443

a single-axis doC machine with M field windings, each fed through a linear
coupling network. The representation does not take magnetic saturation
into account.
.
The representation of Fig. 7-1 j is general and nonlinear in that i it
contains two product terms. This representation can be used with
analog computers or it can be linearized for special cases in the same
manner as were the block diagrams in Sec. 7.2.
The effects of magnetic saturation can be included in an approxim~te
way in Fig. 7-13 in the same way as they were in Sec. 7.2.1. When the
assumption is made that all inductance terms L kk and speed coefficients
Gak saturate with the factor S as defined in Sec. 7.2.1 and when the time
derivative of S is neglected, saturation effects are included in Fig. 7-13
by the addition of the S-l nonlinear transfer function shown in the block
diagram of Fig. 7-14. This representation can be used with an analog
computer or it can be linearized by the techniques discussed previously. i
The generality of the block diagram of Fig. 7-14, coupled with the fact
that the doc machine can operate as an electromechanical power amplifier,
shows that the characteristics obtainable are quite flexible. The multi
plicity of characteristics that may be achieved with d-c machine amplif1ers
arises from the various ways in which the field windings may be arranged
to introduce feedback in machine systems.
Several basic kinds of feedback may be distinguished in a machine
system: external feedback and internal feedback (with the provision Of
extra windings inside the machine); scalar (direct) feedback which com
prises series (current) and shunt (voltage) types of feedback, and feedback
through frequency-sensitive elements; positive (regenerative or cumulative)
feedback, and negative (degenerative or suppressive) feedback. These
various kinds of feedback may, in turn, be classified in two essential
groups: operational feedback and modifying feedback. Operational
feedback improves the performance of a machine but does not substantially
modify its characteristics. Examples of operational feedback are the
passing of armature current through interpole windings to improve
commutation, and through pole-face (compensating) windings to neutralize
the cross-magnetizing effects of armature reaction, particularly in d-c
machines subjected to violent fluctuations in load. The reduction in
armature inductance resulting from the addition of a compensating
winding is an incidental effect that can be readily taken into account in
the formulation of the dynamic analysis. Operational feedback will not
be considered in a general form, but some types of modifying feedback
will be examined which will show how steady-state and transient
characteristics of d-c machine systems can be controlled and adapted to
meet given specifications. In all cases except a few operational feedback

--..:--- -,---- -

...

.;

:!~

'3

:.")'

.a

-_.--

.~

'\

r
m

de
.+':'\

-'

_.

;,,:

'.

.---

'J!

....'"
?

"

~...

;,J

3:

()

I ~

fi
.~

-:

Lmk

.(p)

R
)

;p

r
m

m
;,J

G)

-<

<-,~

()
~

o
z

1IJ11

<

::.

-.

~ ~~ ;.' ~ 'F.

a
.:..'"

- .... *
'"

;:

..

i: .'.

az

Fig. 7-1), Block diagram representation of single-axis d-e machine, with M field windings and no magnetic saturation.

...

~. ~ .~
~

-"'i;i

::. j:.' i

:T

~~

;.'

i- _.....

-i.

4;1 :.

,.,

;;10

V\

.~

'~_"il!:J.f1-'lf4ri),_ Wr"
'7t~~}~

"

jf - . -

<

ti(

,-- <-t1i

r{ , Kfti'

....

,}

.:~

k\><

~,

,.

-<

111

3:

-{

V\

t
I
Y - -
\ ' p L mJw

II.

.~J

"()

112

.~.

11",,8

3:

3:

Z/w(p}

lS-'

~
0

;;10

3:

()
~'

Z
m
VI

fia,7-1-4. Block diagram representatloQ of single-axis doC macbiDe

,
'....:

;!>.

....1>:_

with M field windings and magnetic saturation.

....

.. ,- d

'~,;;

ELECTROMECHANICAL ENERGY CONVERSION

cases, the block diagram of Fig. 7-14 can be specialized to describe the'
operation. Several examples of feedback in doc machine ~y~tems will
now be treated.

;',

DYNAMICS OF COMMUTATOR MACHINES

+47

.:'

EXAMPLE 7-4:. Scalar Feedback

Two simple types of scalar feedback are commonly used in doc machines:
I

EXAMPLE 7E3 Multiple Excitation

The simplest case of the M-field doc generator is the one in which all
field windings are directly excited by voltage sources (some of these
sources may be zero). It is assumed that the leakage inductance of each
field winding is negligible; thus

(a) Shunt (voltage) feedback.

(b) Series (current) feedback.

j'

These types of feedback are often used to achieve compounding as well

as other special characteristics to be cited later.

(a) Shunt feedback

Zk(P) = Rk

and

A d-c machine arrangement with positive shunt (voltage) feedback 'is

illustrated in Fig. 7E4-1. This machine is equipped with two field

(73-1)
Ok(P) = 1

The rotor is assumed to be driven by a constant speed source.

~
I

, I

w =

constant

ia

Vaol
vk Iro

Gak/Rk
= Gak/Rk
I + ~ P LmJ
I + ~ fJ7'J
)-1

'iI'

R)

(7E3-4)

)~l

where T) = L m)/ Rj is the time constant of the jth field winding. Equation
7E3-4 indicates that with respect to voltage generated in the armature
circuit of an M-field doc generator, the system behaves dynamically as a
single-time-lag system in which the equiv<lknt time constant is equal to
the sum of the individual time constants for each field winding. Thus the
addition of more field windings will increase the over-all time constant
and reduce the speed of response.
If saturation is present, the block diagram of Fig. 7-14 shows that
saturation produces an equal reduction of both gain Gak/R k and time
constant Lkkl Rh At any instant of time corresponding to a certain
operating point on the saturation curve, the gain and the time constant
are reduced by a factor equal to the slope of the saturation function at
the instantaneous operating point.
It is worth mentioning at this point that the effects of eddy currents in
the field structure of the machine may be conveniently included in the
analysis if represented by an equivalent short-circuited field winding
having appropriate resistance and inductance values.

f'

/'

(73-3)

The transfer function from the kth signal voltage Vk to the open-circuit
armature voltage is found easily from Fig. 7-13 or 7-14 or Eq. 7-30 to be

(73-2)'

This assumption, plus the assumption of no magnetic saturation, reduces

the problem to a linear one. When the armature is open-circuited

: I

~.:Jj.." "

~~

{,

j'
5,
l'

II

i~;

--0

R"",

'Ve

I,

. .i

:t:

"F ~ r ' I '

.;'..
'i4"i'

Flg.7E....'. Positive shunt feedback in a d-c machine.

{,

."'1

I~i

v.1t

. iii"
(.

I,

'i

lIa

R.I.. L.1t I

j,

! 1

---:>

:>1
'\

'j

\I

\ j

'r' '

".(;, rl
l '."i

\
.I
,I
I

,I

windings: one. denoted by c, is separately excited and is called the control

winding; the other, labeled sh, is the shunt winding, connected across the
machine armature terminals. The machine can operate as a motor or 4~
a generator with either field acting alone, or with both excitations present.
The block diagram of Fig. 74-2 is derived from the general representation
of Fig. 7-14 by noting that
;

M= 2,
ZI(P) = R e
Zz(p) = R,1t

'1,.

;ih

R,

k == (1. 2),
OI(P) = 1,
Oz(p) = 1,

Gal == Gc

(74-1)

GaZ = G,lt

The successful operation of the machine as a generator with shunt

excitation only is inherently dependent upon the nonlinearity S-l. This
condition can be deduced directly from a consideration of the block
.~

-~~-_._--,

.,

449

ELECTROMECHANICAL ENERGY CONVERSION

DYNAMICS OF COMMUTATOR MACHINES

diagram of Fig. 7E4-2. (Winding c in this case may be short-circuited

and assumed to simulate the eddy-current effects.) A required condition
is that the gain Gsho/(R + R sh ) must be larger than unity. The steady
state characteristics are obtained from the block diagram representation
by setting P = O. Dynamically, the presence of some residual or initial
voltage of suitable polarity is necessary to build up the excitation of the
machine. The output voltage is controlled by adjusting the resistance
R. in the shunt circuit, i.e., by varying G.ho/(R + R..}.

assumptions allow simplification of Fig. 7E4-2 to the linear block diagram

of Fig. 7E4-3. From Fig. 7E4-3 the transfer function from control
voltage V c to armature voltage Va is:

448

.~

i
!

~c =

Gc~J/Rc
I - R

Gsh~)
+

R.h

(
TC

(7E4-3)

+ T,'h

1 __Q.~p
R + R 8Ii

This relation shows that an increase in gain results from positive shunt

I .'

Va

lid
" 'q~f

,J;

"
,I

L mco

T"" '"'

---rr;-'

.I,.~

feedback but an identical increase occurs in the time constant.

Eq. 7E4-3 is written as

q,

L"..ho
+R

Toho R,.h

I
l,;

\l' '

Fig. 7E4-J. Simplified block diagram of doe machine with positive feedback.,

~'

Fig. 7E.....2. Block diagram of doe machine with positive shunt feedback.

II

~ = cu

= constant

II
I
I

"
t,

.: ~

1). ,These

>/l ~ :

,?,'

.~.

Tp

Vc

\ ',

\.

K =

Gcw/R c

and

G"h(O

0'

Ji,
J

(7E4-4)

where

'J.

9'li'
~ '-''

,;~,,'

*~t

,ji/
';i

(7E4-2)

anq is operating in the magnetically linear region (8- 1

,'j'

-=-

'\

The doc shunt generator as a machine amplifier is one in which the

command or control is effected by the adjustment of a parameter of
performance: namely, a resistance. This means of control is often
known as rheostatic control; and it may assume various forms, depending
upon the regulator design-such as the carbon-pile, the vibrating, or'
finger-contact types of voltage regulators.
Another process of control of a shunt doc machine, involving what is,
sometimes called dynamic voltage regulation, is achieved by injecting a
signal voltage into the control field c in Fig. 7E4-1. To study this type of
operation, assume that the machine of Figs. 7E4-l and 7E4-2 is operating
as ,a generator, has negligible armature impedance (Ra + LaP) ~ 0, is
driven by a constant speed source

'k'"

Va

+ R.II

the ratio of gain K to time constant

"'-.

{Sf!)

<1.'1

G.hw
+ R.h
'Ii) flU

is

Gc~)/Rc
-K = ----"--'
-....::.
'T
T c + T.h

,;"

't "''''

,_l

+ 'T.h
R

When

tn
"Jf!

(7E4-5)

which is independent of the feedback except for the small effect of R on the
time constant T.h' Equation 7E4-5 expresses a relation similar to the
gain-bandwidth product of vacuum-tube amplifier theory.
As long as the feedback loop gain Gshw/(R + R.h) of Fig. 7E4-3 is less
than unity, the machine is stable and can operate at any voltage. If this
gain becomes greater than unity, self-excitation results and the voltage
Va is determined by the saturation curve.
When this gain is exactly equal

'V

~."'"''

i.'
;",.,,'

0450

DYNAMICS OF COMMUTATOR MACHINES

ELECTROMECHANICAL ENERGY CONVERSION

45'
I

The general block diagram of the series-connected doc machine can bb

derived directly from Fig. 7-14, using the definitions given in Eqs. 7E4-7;.
To take into account the interconnection between field winding s, thf.
armature circuit a, and the load L, it is necessary to evaluate both 1's an(1
Va in Fig. 7E4-4 in terms of known variables in the block diagram of Fig.
7-14. The desired relation for Va is, from Fig. 7E4-4

to unity (this is sometimes called an astatic condition), the transfer

function of Eq. 7E4-3 becomes

~ = (Gcw/R c
7"c

Vc

)!

(7E4-6)

+ 7"sh P

which indicates that the generator acts as a perfect integrator. Mathe

matically, this means an infinite steady-state gain. In practice,
achievement of the integration is very sensitive to slight changes in speed
and resistance which affect the tuning or astatic condition Gshw = R sh + R.
Machines operating on this principle have been used extensively, and are
often known by trade names such as Regulex (Allis-Chalmers), V.S.A.
(Reliance Electric), and Magnavolt (English Electric).

V"

iL

v.

Vc

u.

.: 6

+
Fig.7E4-4. Series feedback in a d-c generator.

~*

through a feedback winding and the complete armature circuit must be

included in the dynamic analysis of the doc machine. The general
configuration of series positive f~edback in a doc generator is illustrated
in Fig. 7E4-4. In order to introduce a minimum of complexity, the
leakage inductances of the two field windings are neglected and the
definitions are made
M = 2,
VI

V2

= V eo
= V.,

ZI(P)

= Reo

01(P) =

Z2(P)

= R"

02(P)

= I,

I'",

:i"i'i<",',>

(I, 2},,'I'

I,

':

Gala

G eo

Ga20 = Gs ,

Vs

and this can be evaluated

+ pLs Vao
G,

(7E4-1O)

Zo(p) - R a

+ R, + pL" + ZL(P)

~'

(7E4-l1j

"0"

til

e
t

(7E4-9)

-i"

In the representation of Fig. 7E4-6 it is seen that the load impedance

ZL(P) which has been absorbed in Zo(p) modifies the field circuit time
constant as LsIZo(p) and for large values of Zip) may be neglected.
The inductive coupling between field circuit and load circuit appears as
the block I - (pLslG"w). The compounding effect of the series windin~
appears as a positive feedback aiding the excitation V c and producing an;
increase in gain but also a slower response.
;
The dynamic characteristics with any type of load impedance can be:
calculated from Fig. 7E4-6. Since this machine has positive feedback,:
. it is possible to adjust the parameters (usually the load resistance) to make;
,this machine act as an integrator with characteristics similar to those:
described in part (a) of this example for shunt feedback. A machine:
operated in this way is sometimes called a Rototrol (Westinghouse).

,~

11

"

ZL(P)

R" L,

(7E4-8)

Using Eqs. 7E4-7, 7E4-8, 7E4-l0 and the general block diagram of Figl
7-14 gives the block diagram shown in Fig. 7E4-5. Figure 7E4-5 can b~
greatly simplified if the speed is assumed to be a constant. For eP = ""
the block diagram of Fig. 7E4-6 results, after considerably simplifying
and rearranging and defining,~.a... ,
.. i
i

'1

"

= is =

v, _ Roi,

---<>

+ iLZL(P)

The remaining unknown is clearly seen to be

from Eq. 7-32 to yield

iL
----..,...

---<>

V,

where the current i L is

(b) Series feedback

A second type of feedback in common use in doc machines is series,
feedback. When series feedback is employed, armature current is passed

"

0=

= Lc
L 220 = L s

Lila

(7E4-7)

:"
"

.~

,:

i~
,~

Both shunt and series scalar feedback discussed in the preceding example
have been of the positive or regenerative type. If, instead, negative
feedback is resorted to in a doc machine, the effects are just opposite
and may be taken care of in the block diagrams of Figs. 7E4-3 and 7E4-6
by making appropriate changes in signs. Negative feedback increases the

';l'

L
..:
.,'. .>~.
.~

~:

..'
-~-'-

452

ELECTROMECHANICAL ENERGY CONVERSION

453

DYNAMICS OF COMMUTATOR MACHINES

.-'

'"'

."9

speed of response of the machine, and often has a stabilizing tendency

in a control system, since it effectively reduces the time lags. This
improvement, however, is obtained at the expense of decreased
amplification.

';'1"<'"
~! .:

~,

:'

,r,~

~f

."~i'
.J;,

,,J:",:

.(-:tn;

iiI-

Eo-.

(] S-1

]
SM

't':
e'

.......
',j'

.s'i

~~+

Zo(p) = R a + R. + pLa + Zdp'

Fig. 7E4-6. Block diagram of d-c generator with series feedback when driven at
constant speed';' = w,

~ t:
~'

-6

'0

, l,l.t

'"I

",)(,j 1

El
~ ' .,

1'41' ,

J,

1.

.--

,bb

T\ '"

,t ",\.

g
as o<'!IH "~'V ~

lC1)I

..ll~

,
.. 'I-,

!',f,

... 1

t ..
i "\

"')),I!Xl, :'

<

~t:~

, ..

,: i

~~,i'

_1~}

7.204

;:

,~,

I' !

;,l!:l;;"

'il:,H'i4L:

. ~t

';lJ

l~.vr~.}~"
(. ,~

"'!',.

.1

If

,t'.ftt<.,

1 IJ'

'.1 I

;I "

'~"I

.,,.

~t,

t i~'

, ,n,t~i'

:1/:;

t, {I I ' \

'

~r'1

"
t:

:~~

,,~n!\ 1~1l:t.M;~

kl'l.t,~:)'l!' r~it,i"t!\'

1'1

i
fl

,nii
I hi '

~',-

",.
1'-'
, ';.

...

'

.,:J"

;i.;

.-,,#

II

Multistage Machine Systems

t,

,~t '
'f

t"

(u .

"/

'.

Ii

;Yl,_~:

t't;

r0

~,

Nt
\'JI (

~III

~
---,.-

:.e

\'~"

,<)

'~

The combination of various feedback arrangements in d-c machines

introduces some interesting possibilities for altering their characteristics.
A general combination of scalar feedback would include separate, series.
and shunt excitations. The usefulness of the formulation of the dynamics
of machine systems presented in this chapter appears more evident when
more complex feedback combinations are investigated. The techniques
established here can be used directly to handle more complicated machines
with many feedback windings.

.f1'

~,~.~,7'tttf'i:

:.,~i

,~-.t<

..

'

:;iff~~~."

II'

;jJ ,

~'

The amplification of a single d-c machine can be increased by designing

a large field system and/or by introducing some form of positive feedback.
Aside from purely constructional difficulties, there is a limit to this increase.
since it is normally accompanied by an increase in the time constant.
which has a tcndency to make the over-all systcm less stable or, at best.
more difficult to stabilize sufficiently. The importance or power amplifica
tion in machine systems and the relationship between design and the
dynamic characteristics is a subject of vital importance. but it will not be
treated here and attcntion will be confined to only the signal character
istics. Where large amplification is required. multistage machine
arrangemcnts provide a solution in which amplification is achieved with

~\
_l~, ." ~

~..."t

,AI_,
.;'f_~'~
,~

r~

.II

,...

,~
454

ELECTROMECHANICAL ENERGY CONVERSION

only the time Jags introduced by each stage, and these lags may be small
indeed. Two types of multistage machine systems are encounten:d:

DYNAMICS OF COMMUTATOR MACHINES

t
11

d axis cause a magnetic field centered on the stator IX axis and currents in
the q axis cause a magnetic field centered on the stator f3 axis.
I

la'

(0) The cascaded type in which a sequence of single doc machine

amplifiers are placed in succession, each machine constituting one single

stage.
.
(b) The cross-field type in which two stages are combined in a single
machine.

4SS

.,

~.~

axis

it'

I",.

+~.
u
a.

The feedback combinations of even the simplest types are indeed

innumerable when applied to multistage machines. The techniques and
methods developed in this chapter are very well suited to the study of
these feedbacks. Within the limits of the basic assumptions, the general
ized block diagram of Fig. 7-14 can be extended to cover any multistage
arrangement of cascaded machines. In the next section the cross-Held
type of machines will be treated.

\
I
I

"..

i
I

.,

la,
+~

U"'2

+~JaN.I .
'.

$c,

rl-"

uaN

1tiN',

7.3

"

;".Ji

Multifield Two-Axis D-C Machines

The basic doc machine treated in previous sections of this chapter has
been restricted to the single-axis machine which has one distinguishing
characteristic that makes it particularly useful for control purposes-the
absence of transformer coupling between the armature and field. There
are other configurations of commutator machines which also have
particular advantages. Of the possible configurations the two-axis d-o
machine is probably the most important. Each stage of a two-axis
machine is merely a single-axis machine, but by locating the two axes on a
common magnetic structure. mutual couplings between each stage exist
and must be taken into account. Certain inherent characteristics of tho
two-axis machine are particularly useful, and by employing various
combinations of lield windings on each axis it is possible to obtain a wide
range of characteristics with two-axis machines. The utilization of the
two-axis machine as an electromechanical power amplifier is of particular
importance. "fhe analysis of the two-axis machine developed in this
section will stres~ the signal characteristics of the machine, taking into
account the el1'ects of several field windings on each axis in a manner that
is particularly lIseful for studying the dynamics of the machine as an
electromechanical amplifier.
The configuration of a two-axis (cross-field) doc machine is shown in
Fig. 7-15. For generality. the machine is assumed to have N windings on
the stator IX axis and M windings on the stator f3 axis. The rotor terminals
consist of two sets of brushes, the d and q axis brushes. Currents in the

..

.~J

,,~

;p

u'
d

,
1.

~ .
t;'
.

.:,
...

.r

lq

M~"
1 1

'1\
rii;~
1 11 1

u!
ON

'.

'.....

...

8' axis

II'
~

;.: f!

.,;

,,,..

u!

u!

p z . B,

,;':

"~

.:t

Fig. 7-15.

'. I

~ . ~.'l.'

MultifieJd two-axis commutator machine.

...

, ,

f'

t;
~,~!'
rt
(lilt

i;,

7.3.1

Equations of Motion of Multlfield Two-Axis DC Machines

Equations 7-20 and b describe a two-axis commutator machine with

one winding on each stator axis. These equations can be generalized by
recognizing that all windings in one axis will be described by similar'
equations and that transformer coupling will exist among all windings'
on the same axis. In a manner analogous to that used for the general
ization of a single-axis machine in Sec. 7.2.3, Eqs. 7-20 and b can be
generalized to describe the multifield machine of Fig. 7-15. The resulting
equations are given in Eqs. 7-33a and b. Since machines of this type are
normally used as amplifiers (generators), a general load voltage is included
in Eq. 7-330.

456

ELECTROMECHANICAL ENERGY CONVERSION

I~_.. ,,_e ... ~J

~iD. ,,~

r.::l

~J ~ ~_.

tip

'="'

"<Xi ~
"&I. ... ";:;:
~
~'"

... 0

0::..

I~
"" ~ t+

The general dynamic representation of the multificld two-axis machine

can be derived in a manner exactly analogous to that used in Sec. 7.2.3
for the single-axis machine. The essential dilTerence between the two-axis
machine and the single-axis machine is the mutual coupling which exis~s
betwet'~ the armature circuits and the field circuits. To derive the general
dynamic representation from the equations of motion, Eqs. 7-33a and b.
it is advantageous to assume equal mutual flux in any given axis. This
leads to the following relationships between inductances and speed
coefficients in the two axes:

I ...

0::..

l:l,

I 0

!"'_
""

"_
~

t;
~O ~o ... ~
0::..

l:l,

""
+

-&-

""

I ~ .....- Q~,:i

"... "
"
~
.. ~

.. -

... 0

l:l,

::.:

I 0

\:.l

'-'

:)

.
t:,-

;:;:

:;CQ. ~Q).
"'-l "'-l

. 0

"ii
+ I ~~
.......- ~ct
l...

0::..

l:l,

;:;:

I,)
-&-

I 0

"'-l

l:l,

...

... 0

+ '"

~~

.1:1,

'0

(Q.

"'-l

1:1,

~CQ.

"" 2.
~

~ '" +~
~.

~O
0::..

~.
~
~
N

~.

~~

I
I'

f'

~ ::l.~
+.
~

il:;0
L

"'-l

0::..

U" (N;)
NS
ql3k

i;

!'lOF

I.

I,;)

""S- ~i
.::4 ~
1:1,
I

.. 0

I~.g~ t:;;~i

..:. 0

"'-l -&-

1:1,

"

l:l,

"..1.

:'

+
'.~

'!t.~

,f-

"j'

O'

.1"
h

~""'
:i
\:.l

'Q-

.~ r;~

,....
'" t.,: I
I

""

(N~/)

N'13k -_ G"d131

(7-34e)

Gr.,

= Lrr
qq

(7-34f)

Gd13k N~k

~kl3/

(7-34d)

qd

d13k

rs

( Nr)

r,.

= Gdq

(7-34g)
(7-34h)

p.. ; I

I
'

r.

("

Direct Axis
0:n(p)1/" n

0'd (p)v dr

;:l

, I

I
!

Z: (p)i~_ +
n

-n

L pL:' aki:k + pL':di;

n

k-l

(7-35a)

tl

'j

-&-

II

where N represents the number of turns in a particular winding.

The other general condition which is useful in characterizing the
multifieJd two-axis machine is to assume that a linear coupling network
is connected across the terminals of each field winding and each armature
winding. Following the procedure established in Sec. 7.2. J, the resistance
and leakage reactance of a given winding will be lumped in the coupling
network. The volt-ampere equations which apply to Fig. 7-16 can be
derived from the general relationships of Eqs. 7-33a and b modified to
include the effects of the coupling networks and they are:

:::::
I

..

+
~""'
r:,i

I~::..,
. ,,11"" 'h'::'::',..
z ~iD. ~cll:'"
I .~ nl::.
::I

':'-~

"

rr

= G

ak

LIS

13k

G
+

:.~

Nr)
(--.:!..
N"

::! ....

"

(7-34c)

~! ;.~-.A

11m

,-.,

1:1,

~ ~ ~_g
1:1,

~~....

"'-l

"'-l

~/

13k13m

"'-
:i
I,)
'(r

"

I 0

"'-l-e-

l:l,

... ::l

,{

(N'
"j:jS ) --

L"

~e

fb..

",,(Q,

.,..

",

'G:'"

......

...

"-8

... 0

lj

~o

1:1,

G"
qak

.,nHt:u;]'!f ,.('

Quadrature Axis

~~

+ ...

-&-

""'1tJiit'l;--*'v"~._' (' 1~.~'

N

G'

"li

G~

'~

(7-34b)

tid

ok

~~

z'

1:1,
,.!..;

=~
"'-l
0::..

"'-l

"'-l"'-l
1:1,

_
01

~~

~~

::l

...

:;:;~

... 0

~
~of:

~CQ.

'-i~

N') = L"

Lr
., (~d
dak NS

c:..

G"
(a/)
qak N:
= G"'1"/
k

~~
N

N'

(7-34a)

a.

~tl

~""

N'

L'"akan ( N$a l ) -_ L"

aka}

+ ...

~'" "'-l

~'"

~
~<Q,

Direct Axis

~-~

~""

Dynamic Representation of the Multifteld Two-Axis D-C

Machine

7.3.2

6 6

~~--

457

DYNAMICS OF COMMUTATOR MACHINES

...,.....~ :Q'
.....
...,

= zr(p)i'
d
d

+ pL"dddi r + ,(,.".

k-l

pL"
is
dOkOk

+ L.'\'

k-l

.lG"
i"
'I'
dl3kl3k

+ 'I'lGrri'
dqq (7-35b)

."~

.\,t'
;~ \Y~

I-..

"'~"'4>:.

iF

>4~.;
458

where

where

S + pLs + ZS' (p)

Zsan (p) = R an
Un,
n

R; + pLJ1 + Z'd'(p)

L'd = L~,

Zs/lm (p)

(7-35c)

l.1

Z~(p) =

459

DYNAMICS OF COMMUTATOR MACHINES

ELECTROMECHANICAL ENERGY CONVERSION

(7-35d)

+ LXJ

S + pLs + zsr (p)

= R tlm
11
13 m

(7-36c)

R; + pL;, + Z;'(p)

(7-36d)

10 ,

Z;(p) =

(7-35e)

L;

= L;, + L~;

(7-36e)

The first step in the analysis is to define the open-circuit generated

voltages of each axis which result from all excitation, field and armature.'
These voltages defined as vdo and v;o are:

.'., l;P)

t'

Direct Axis

'j

\)

r
vdo

....

-,',':.,

-,

Z:2(pJ

v''10 = -

.'..

~~,'

>,

" ~

lJ

'0

'. 'fL.'"
'~"~'.'

fl

allS
-----0

il. .'
~ ~ ~'.
..

.a

/.

d~;

.~

O;(pJ

f';frl,:

'l~

t,J
&;~

Z:(p)

----v;_+

UdAl +

(7-38)

k-I

IId r

liJ,

The open-circuit voltages vdo and v;o are a direct measure of the mutual
air-gap flux and as such will prove valuable in simplifying the equations
of motion. Using Eqs. 7-37 and 7-38, the derivation of the desired
voltage transfer relationship for each axis follows the same general method'
used in Sec. 7.2.3 for the single-axis machine, except that it is necessary'
to include the effects of armature currents as well as field currents on
the generated voltage in each axis. To carry out the desired rearrange
ment of Eqs, 7-35 and 7-36 it is advantageous to make use of the fact that
v'doN and v;oN are the mutual flux in each axis; thus the following
relationships can be derived with the aid of Eqs. 7-34a-h:

I,"

-+
;. ,

,'t-';

,;

"!sr.r
L
pLana/ak + pLandld =
.IS

..", '\'

.t.<~ '{~

Direct AXis

"L

~ t i ~:'~,,~

'"
iGrs is _ 'fiGrr
i,>AU
L 'f qak at
qdd

,"

,'XI'1'

.):o,.}.

(7_37)1

Quadrature Axis

-~

~'

"lGrs's
iGrr 'r
= k-t
L 'f df1t'f1t + ep dqlq

fl';

~,li

"I

k-I

')

Fig. 7-16. Circuit for general operation of two-axis doc machine,

',t-~'i

51

Grs (pL anan )

'l"a .,

P~~tn

(NL G"""I,.. + Gqdld

rS.$

k- I

rr . , )

(- v~o)

11

(7-39) .

~ "-'

l'o:l

Quadrature Axis

0;

..

~.:l:. I

~.

(P)v~ = Z~ (p)i~
"'m

+ k-l
L pL~srn(3);.
+ pLrqiq'
'" ""

I :I;:

"'\'

rs's

~ pLduklak

(7-36a)

In

M
N
O'(p)V r = Z'(p)i r + pLrri r + ') pL rs is _ )' iG rs is _ iGrr;r (7-36b)
q
q
""
qq" k::l q/lktlk k::t'f 'l"k.... 'f qdd

iF

(i"'~~'1

x-I

rr 'r _
+ pLddl
d-

I
rs
G's (pL da.)

__

(N

qa,.

pL'Jd (_ V~)

G"
qd

1.
ep

k-l

rs s
Gqat'at

rr or
+ Gqdld

(7-40)

-l

DYNAMICS OF COMMUTATOR MACHINES

ELECTROMECHANICAL ENERGY CONVERSION

460

Summing Eq. 7-45 over all n field windings and adding to Eq. 7-46 give
an equation for the effective inputs to all direct-axis windings. After
using the identity defined by Eq. 7-38, there results

Quadrature Axis
M

SI"

pLflmflbk

x-I

+ pLflmqlq

".,

,13

= G rs

(pL flmflm )

dll,.

k-I

"'s

Gdt3k ' flk

rr .r)

Gdqlq

Direct Axis

= pLPmllm (V~o)
rs
Gdll
m

(7-41)

rs
OsOn(p)v"l'ln G~

1'1

""

L., - - - -

n-I

~ Lrs's
+ PLrr"
1 (Lr.,)
qllklllk
qqlq = Grs
P qllm

L. p
k-I

dt3 m
rr

(% Gr"
L.
k-I

.,

dllklflk

L O:"'(p)v~n +
n-I
m-I

L _"m_'_ "m

m-I

dll

O~(p)v~

O;(p)v;

pL"
V')
dd ]} (
qo
Zd(P)
- -:{

(7-47)

f'~

;;

,;.,

Jt'-

dq

",.

).'A.'l~'

{,';.

,J,,~

'

Z;(p)
{

i~f

pI.fS
'''mllm
[
m-1Z~"m(p)

pL l}(V~ )
+~
rr

(7-48)

Z'(p)j
q

Equations 7-47 and 7-48 give simple and useful relationships between
the inputs to the fields and the open-circuit generated voltages.
To complete the characterization of the two-axis machine, the currents
id and i; are needed plus the torque equation. The current id can be
det.ermined from Eqs. 7-35b, 7-37, and 7-40 as

-.J:

1 [
id = Zr( ) O~(p)v~ P

r.::M v'~o ]
v~o + ~rr

(7-49)

'I'

qd

Similarly, using Eqs. 7-36b, 7-38, and 7-42, the quadrature current

.
, '!

'r

Zr( )
q p

Iq =

}'
'-"

The simplest relationship between input and output is not obtained using
either Eq. 7-43 or 7-44, but results by employing a modification of these
inputs. The desired form of the equation for the direct axis can be
obtained from Eqs. 7-35a and b after using Eqs. 7-39 and 7-40 and
rearranging to give

qo

Z#m(p)

(7-43)
(7-44)

O;"'(p)v~m

pUs
ana n

Quadrature Axis
r
r
M 0" (n)p" GrJ
m [or(p)v - v ]G"

= L

An analogous development for the quadrature axis utilizing Eqs. 7-36a

and b, 7-41, and 7-42 gives

(7-42)'

The four relationships, Eqs. 7-39-7-42, have one essential character,

which is that they rclate the transformer voltages due to all currents in
one axis to the derivatives of the speed voltage in the other axis. These
relationships result directly from the assumption of a common mutual
flux in a given axis and lead directly to considerable simplification in the
equations of motion.
The relationship to be derived from the equations of motion is one
that relates the inputs (field voltages) to the outputs (armature voltages),
and that is in a form which allows saturation to be taken into account.
The inputs are all the field excitations on a given axis, plus the armature
winding on that axis. The inputs to the direct and. quadrature axes
re~pectively are:

input to quadrature axis

Zd(P)

- { + l LN Z~n(p) +

n= 1

Grr
1<P
dq

input to direct axis =

Z~n(p)

\ ..,

r
rr
[O'(p)v'
- vdo
]Gqd
d
d

Grr
or)
dqlq

pL (V do )
= '::..::.'!!l

""C'

461

[r
, ,
Oip)vq - vqo

pL~~ V do ]
Grr 1
dq

i; is
(7-50)

<P

The remaining relationship needed is the torque equation which can be

written from Eq. 7-33b as
:.i't'

= J</> + a~

'f

Id

rr'" )
M Gdllmlilm + Gdql
(N
q + Iq n~1
(2L
~

rs"

'f

--..

fS

'1

Gqanlan

rr 'r

Gqdl d

(7-51)
1'0

LSI ( V')

rs

osan ( P)v'an Gqa n

Z~n(p)

GrS"1
I

[O~(p)V~ - vdo]G;d
Zd(P)

qan an

P ana"
qo
+--
-
Z~n{p)
eP

= Grrj' +
qd d

pL',M ( _

Zd(P)

V;o)

and using Eqs. 7-37 and 7-38 for vdo and v;o' this becomes,.
(7-45)

(7-46)

J.r

=~,

+.1

a'I' -

'r

'"

do

-:j; -

'r

v'flO

I q -:{

l~

(7-52)

The complete set of dynamic equations of motion of the multifield

two-axis machine is given by Eqs. 7-47-7-50 and 7-52 in a particularly
'. '4

~k
?I,.,

m
r

()

-i

o
~

Fig. 7-17. General block diagram of the multifield two-axis commutator machine.

()

::I:

u'q

()

C)

-<
()

o
z

<

Vl

(5
Z

,~ . _;,

,%t;. ,.....

,~trc:*v #....) " .

i" awr~Y:

~_.~ "~*s.~,,; ..,, ~

,~-~

";'~f

~,----~

..
~-

:.

J:.

':..;...

..:.

.J:

::"
~

-<

~.

()
VI

Vd

I. ? L"!m6,~ + pL;~ I ~ =4!9~. ~ I (

()

"

m~1 Zs..(p'

Z~p'

~
~

-i

~
)
()

::I:
r

Vd

:.

Z
m

VI

.
-_ ..

_ ._._.__.
~---'--

.r-0'

_.;;,;-",'O:'::':"&_,,,:-... ~- -,--_.-

.:..

______.;:.J-----

~ --

conven ient form. The block diagram representation of this set of equations
is givcn in Fig. 7-17. Saturation is included in this diagram in a manner
exactly analogous to that developed in Sec. 7.2.1 for the single-axis
machinc. For the two-axis machine it is necessary in general to have
diflerent saturation characteristics in the two axes. These inverse
saturation curves designated as Si 1 and S;; 1 are obtained from the open
circuit voltage characteristics of the machine just as in Fig. 7-6 and 7-7.
The curve Si 1 is determined from V~D versus the direct-axis field currents
i~n' and S;; 1 is determined from vdD versus the quadrature-axis field currents

if

~~f

IT

s
IlJm'

v'

i"

7.3.3

Signal Transfer Properties of Two-Axis D-C Machines

Two-axis doc machines have their widest use as electromechanical

amplifiers to provide power amplification in control systems. In this
type of operation the machine acts as a generator driven by a mechanical
source. A complete treatment of the dynamics of the machine must
include the drive characteristics. Although such a complete treatment is
mathematically complex, it can be done using Figs. 7-16 and 7-17. To"
show the utility of this general representation an example will be worked
in detail. A general treatment including drive dynamics will not be
given; instead, the machine is assumed to be driven by a constant speed
source and the electric signal properties only will be derived. This type
of treatment proves useful in the study of the dynamics of feedback control
s~stems containing two-axis doc machines as amplifiers.

l'~
Vd

vq

VdL

vqL

pLfc
pLfd
R/ + pL/
0
pL cl
pLed .
.0
R" + pL"
pLdl
pLdc
R d + pLd
wGdq
-wGqC
-wGq/
-wGqd R q + pLq
0
0
0 ,1
oi
!
0
0
0
0

in which

0
ZdL(P)

idz.

ZqL(P)

iqL

= w = constant

Vd

;~.

iq

.~I

Vq

;,;';
.f,

)0

+ 6

---J-+ 9
QL

0
I

iqL

"

::f

0+
V~

'~'

5.

ZdL (p)

i.,:'
i ,L.
I

Z'IL (p)

'1l!

.~

'-:;f/

0
i/

0
i"

0
ld

0_,_ i q

0
0

r"'+
,

:XAMPL 75 The Two-Axis Rotating Amplifier

Consider the two-axis doc machine shown schematically in Fig. 7E5-1.
Note that this schematic is similar to the one in Fig. 7-16. The volt
ampere equations that describe this machine are easily obtained from
Eq. 7-33(1 and are
~.
Vf

465

DYNAMICS OF COMMUTATOR MACHINES

ELECTROMECHANICAL ENERGY CONVERSION

'f

,(7E5-l)

'--~

Fig- 7ES-I_

Two-axis doc machine.

The usual method of operation of the machine shown in Fig. 7E5-1

as an amplifier is in the connection shown in Fig. 7E5-2. The input
signal is vf and the output signal is either load voltage VL or load current
iL . The winding c which is connected in series with the load circuit is
called the compensating winding.
The dynamics of the machine in Fig. 7E5-2 will be obtained by using
the general block diagram of Fig. 7-17 to find the transfer function for
the machine. To use Fig. 7-17 the following quantities are defined using
Figs. 7E5-1 and 7E5-2.
The subscripts
al

(7E5-2)

is assumed.

a2
"",;~~"""",--""~,-"""",,,,>

,.

=f
=C

(7E5-3)

'II

"
.~.

,'$'t.i

-466

ELECTROMECHANICAL ENERGY CONVERSION

The open-circuit transfer functions

O~(p)

OJ(p) = I

O;(p)

O;(p)

=
=

(7E5-4)

The current constraints

I'

id

-iL

467

DYNAMICS OF COMMUTATOR MACHINES

,I

(755)

ic = iL = - id
The voltage relations_""",,

vq

Vd = Vc
11 f:

Vc

.."1

=0

+ iLZL(P)

-Zc(p)id

= Vc - idZL(P)

(75-6)

cc ( - Vqo)
+ pL
G
-</>

(7E5.7)

L~

U3
l

i ..

qC

:i
~ 1

.5

Equations 7E5-3 through 7E5-6 all follow directly from the connection
diagram in Fig. 7E5-2. Equation 7E5-7 for the voltage V c is obtained
from Eqs. 7-350 and 7-39 and expresses V c in terms of quantities which
are explicitly expressed in the general block diagram of Fig. 7-17. Using

.~
, .'

~\ ~~.I

l'

.,'

i.

I....

...ra

*j.f'

~.

6 ,;

:~1
+6

)0'<

Uc

<:-.

J'\.~

.... ~ ~

' - ~l~-;"o '

e.

'~

,.\ll

, T" .

"i

. I'

lq

l
.~.

i"
1.

'.

iXl

12 , (pJ

I
l'

f
.:l'l,

Connection of two-axis d-c machine as rotating amplifer.

I
h

tt'

....w

., .

j-',

.it
II.

:_;11

,";\~- ;',> !

+~

"J,.:.,

.m.

r;

:.a .
-I

; lq

<'

Fig. 7E5-2.

""I""
.:J/e;'"
Q,3

.l

OJ

df

,~,.,

468

ELECTROMECHANICAL ENERGY CONVERSION

DYNAMICS OF COMMUTATOR MACHINES

the relationships of Eqs. 7E5-3 through 7E5-7 and the general block
diagram of Fig. 7-17, it is possible to draw the block diagram of Fig.
7E5-3. The block diagram of Fig. 7E5-3 is unnecessarily complex for a
machine operating at constant speed. By a process of reduction using
the block diagram algebra introduced in Chap. 2 it is possible to obtain
the much simpler diagram shown in Fig. 7E5-4. The block diagram of
Fig. 7E5-4 presents in a very clear fashion the essentials of the two-axis

Assuming that saturation is negligible and with the perfect compensation

of Eqs. 7E5-8 and 7E5-9, the block diagram of Fig. 7E5-4 reduces to the
one shown in Fig. 7E5-5. This figure shows that internally the amplifier
has two cascaded time constants, 7'1 = Lf/Rf of the field winding and
"/I ... L q / R q of the quadrature-axis armature circuit.
The resistance and
inductance of the armature direct-axis circuit are included in the
impedance ZAp). Magnetic saturation effects can easily be included in
the block diagram of Fig. 7E5-5, if desired.

:',

Uf

--1Ri+~H

Gqfw

H~H

GdqW

..

...

Zd(P)

,.
~

Gqd - GqC

=0

Fig. 7ES-S.

Lc/

=0
-"

= R d + LdlP + R c + Lc/p + Z L(P)

,.
I

'I;":

t.'~,~". .'~.\"
~.

I.
I I

I'

iL

(7E5-8)

Zd(P)

Block diagram of rotary amplifier with perfect compensation.

This type of compensation makes the rotating amplifier an amplidyne

(General Electric). When the condition of Eq. 7E5-8 is fulfilled, the
inductance also satisfies the condition

Ldl

The representation of the two-axis machine as a rotary amplifier given

in Fig. 7E5-4 is based on the assumption of equal mutual flux in any given
axis. This assumption leads to a very simple characterization of the
two-axis machine, placing in evidence the essential character of the inter
actions between axes. The techniques of analysis of the two-axis machine
presented in developing the general block diagram of Fig. 7-17 which
focus attention on the mutual air-gap flux result in the simplest representa
tion of the machine taking into account the interactions between input
and output quantities. For example, it is possible to derive a general

Reduced block diagram of two-axis machine in Fig. 7ES-2.

amplifier of Fig. 7E5-2. The pertinent characteristics are the time constants
of field circuit and quadrature-axis armature circuit in the forward path.
the transformer voltage feed forward path which causes a change in output
whenever there is a time rate of change in direct-axis flux (vqo/w), and the
negative feedback between the output current and the input voltage VI'
When this negative feedback is appreciable, which is the case whenever
the compensation is small (GqC ~ Gqd), the machine acts very much as a.
current source independent of the load for a constant applied field voltage.
The two-axis machine as a rotating amplifier is particularly interesting
when it is compensated in such a way that the negative feedback term
proportional to w in Fig. 7E5-2 is made zero, i.e.,

H dPT t-

'{:i1.

Fig. 7E5-4.

469

(7E5-9)

Fig. 7E5-6.

-,
'~J'

/~~:

;\~.

= R d + L"p + R c + Lcp + ZL(P) -

2L cd P

Block diagram of two-axis machine as rotary amplifier.

~70

ELECTROMECHANICAL ENERGY CONVERSION

block diagram of the two-axis rotary amplifier of Fig. 7E5-2 directly

from the equations of motion of Eq. 7E5-1 subject to the constraints of
Eqs. 7E5-5 and 7E5-6. Such a block diagram is shown in Fig. 7E5-6.
This block diagram indicates derivative feedback between i L and vI' but
in actuality this effect does not exist, for by suitable reduction after
assuming equal mutual flux, Fig. 7E5-6 can be converted to Fig. 7E5-4.
Both Figs. 7E5-4 and 7E5-6 reduce to Fig. 7E5-5 for the case ..;f perfect
compensation. The general techniques presented in Fig. 7-17 become
even more useful when employed with configurations having many feed.
back windings in both axes.

.'-./

2.5 amp. When the machine is run at 1200 rpm the open-circuit terminal voltage
of the armature varies with the field current according to the relationship

V.

= 50/,

where V. is the open-circuit terminal voltage and I, is the field current. The
rotor of the machine has a moment of inertia of 30.0 kg-meters 2 and is not
acted upon by any external spring or frictional forces.

:1

PROBLEMS

471

DYNAMICS OF COMMUTATOR MACHINES

rltlf.2.5.mp ;,

7-1. Steady-state tests at a constant speed of 1200 rpm on the doc machine
shown in Fig. 7P-1 show that the open-circuit voltage between the brushes and

Flg.7P-2.

('

ll'I

If at I = 0 a torque impulse of 3.8 x 10 3 newton-meter-sec is applied to the

rotor, find the current iu(I). In your calculation neglect armature resistance
and armature inductance.

iu

n'",~bt.1

Q+
'. ~ (.:.'

,,'j

..,

~~
UII

7-3. D-c generators often have a series field which is connected in series with
the armature as shown in Fig. 7P-3. The purpose of the series field is to
prevent the terminal voltage VI from ,decreasing as the electrical load increases...

If>'"'' ..... '

"""~"~

-~~

:co::;

Ra

Fig.7P-I.
It' "",

~!'

the flux linkages with the field vary approximately according to the following
relationships:
-5 < I, < 5
I, > 5
I, < - 5
V, = 50/, + 5
V. = 255
V. = -245
A, = 60.0
A, = 12.0/,
\ = -60.0
where I, = field current in amperes
V. = open-circuit brush voltage in volts
A, = field flux linkage in webers
The field is connected to the armature at t == 0 by closing the switch S.
Answer the following questions. In your calculations you may neglect armature
resistance and self-inductance.
(0) Find the current i,(t) and the voltage v(t) if the field resistance is 40 ohms.
(b) If the tleld resistance is increased to 60 ohms, what will happen when
the switch is closed?
7-2. An inductance is attached to the armature of a doc machine as shown
in Fig. 7p2. The tleld of the machine is excited with a constant current of

;1 \ '

La

.,eontro1r:l
fi~eld,

:\ti

.:}:I~
,

I '

l'<1

ii

,>\,

,,:.

.:5.,

ii:~' :

""O.
"
!J

,.,
~

Rf,L f

...
RL

.
I

'j

't

'J"

Fig. 7P-3.

ii'

f~l.;'I;:;'

iI.,

.?
;<!

The total resistance of the armature and series field is R a and the total seir
inductance of the armature and series field is La. The speed voltage V, is related
to the control Held current and armature current by the relation
v.

Lf

"i;;:,,:/';-' .;

:~i

l~

..J~ .

)J

,'.J.t:

.,~, i

= Ga,wi, +

Ga,wi.

The generator is driven at constant speed w.

(0) Write the two equilibrium equations for the machine.

~. ((b) Laplace-transform the equilibrium equations and draw a block diagram

. ''ii
representation of the equations with the control-field voltage V,(s) as the input
~:~ . and the terminal voltage V.(s) as the output.

,I'.;

.'

t;

4n

(e) Write the transfer function for V,(s)/ VI(S),

(d) How should Ga,w compare with R. for the terminal voltage to be

independent of the armature resistance?

(c) What is the physical explanation of the result of part (d)?
(f) If VI has a constant value, what is the expression for the steady-state
terminal voltage when the condition of part (d) is satisfied?

Polarity of voltage for

posit,ve i l i 2 and'w"

"~

+"-U2-

,
I.

r1t:,

,;;

"','

.~

::~'

.~

(O.366i r + O.0183)!ft volts;~= rad/sec

';l

Fig.7P-4.
(a) An external torque TL of 128 newton-meters is applied to the shaft of the
machine in the positive direction of,p. The field current if is 4.95 amp and
constant. Find the steady-state speed, armature current, and power flowing
into the 230-volt source.
(b) Find the steady-state speed of the machine for the conditions shown in
the figure if the applied torque TL = O.
(e) If the steady-state conditions of (h) prevail and the field current suddenly
becomes zero, find the speed drp/dt, and the armature current as a function of
time.
(d) At what final speed will the machine run under the conditions of part (e)?

7-5. A doC exciter used in a control system is shown in Fig. 7P-5. It has a
separately excited control field I and a shunt field 2 that is degeneratively
connected across the exciter output through a series R-C network. Neglecting
armature internal impedance and assuming perfect coupling between field
windings, determine the transfer function va(p)/Vj(p) with and without the
presence of the feedback winding 2. Explain briefly the effect of this type' of
feedback.
/.'
,J
~ kl"!:r
The data for the figure are:
R I = 1600 ohms
R 2 = 1000 ohms
L, = 16 henrys
L 2 = 30 henrys
G.tw = 2190 volts/amp
G. 2 w = 3000 volts/amp " ;-\ '1 ft"

C = 100 10'1

~' I

"

'I", '

: f",

t\

", ).In,,

7-6. Figure 7P-6 shows a Ward-Leonard speed-control system employing

feedback by means of a differentially connected series field. Ra and La are the
total resistance and self-inductance of the armature circuit and are shown in
the diagram as lumped clements. V, is the speed voltage of the generator
which is produced by the generator field current and the armature current
through the series field. Vm is the speed voltage of the motor and varies directly
since the field of the motor is separately excited.
with the motor speed
(a) Write the equilibrium equations for the system.
(b) DU1W a block diagram representing the equilibrium equations determined
in (a).
(e) By suitable manipulation of the block diagram, find the transfer function
between the output speed and the field voltage vr when the load torque is zero.

+,

~lJ~

',"

"

'" 2,

":f,
i
J{

UI/

;w'

",

= (G",wl/i, - Gas wi)

II (.

Fig. 7P-6"

:'

A
). ,..,'
,., ....1

'>'",,~fw:

I,

(d) Find the transfer function between T L and ~ when VI is zero.

(e) Using the results of (c) and (d), find the steady-state relationship between
TL, and VI' That is T L , VI have had constant values for a long time.
(f) Show that your answer determined in (e) is correct by solving the original
equilibrium equations for under the condition that s = O.

+,

: orr

L 12 = VL IL 2

~i

Fig.7P-5.

':, .. 0.50 kg- meter 2

U6 '"

__ UI~~~tlllk

'Py' ~.

1"

oJ

Ua

The separately excited doc motor shown in Fig. 7p-4 has an effective
armature circuit resistance of 0.2 ohm and negligible armature inductance.
The relationship between the speed voltage of the machine V" the field current,
and speed is given in Fig. 7P-4.
! 7-4.

_ ur

-473

DYNAMICS OF COMMUTATOR MACHINES

ELECTROMECHANICAL ENERGY CONVERSION

(g) One of the principal reasons for using differential series fields as in this

~i}
~i

f
:",,1,,"
.J'_ ;

.'1'1.

example is to prevent large transient currents in the armature circuit. Show that
if L I , = 0 the effect of the series field is equivalent to placing an additional
"Ji

',.

-474

ELECTROMECHANICAL ENERGY CONVERSION

resistance of G w. ohms in series with the armature. Why does this prevent
large transient armature currents? Why does one resort to feedback rather
than actu<tlly increasing the armature resistance?

'~-'~

DYNAMICS OF COMMUTATOR MACHINES

Control
:,winding

7-7. The Ward-Leonard system shown in Fig. 7p7 provides one method of
controlling the speed of a d-c motor. Its operation is to be analyzed in this
problem.
The motor and generator are identical units. The generator is maintained
at constant speed wI( at all times by an oversized drive motor. The 'idd
current of the motor is maintained at all times at 10 by a current source.
Ra

Generator

Drive motor

maintaining
constant speed

u"

Q
- u, irt

R/'Lr,~qr

i:'"
'f "

~~;

~.

_A""~J.n .:}~

.. ~:,

Rc,Lc,G"c+
-Load

1-

u.

rJ'..~Bearmg with

<.:!J

fnctlon

"r~

'=

(a

= turns-ratio)

Flg.7P-8.

k ';"d ' 'Id '"'''0'

VISCOUS

Dw

i!;

. t.

R. represents the total armature resistance of the motor and the generator
plus that of brushes and connecting wires. The inductance of the armature
circuit may be neglected. as may the etTect of saturation.
(a) Obtain a differential equation relating output speed W m in radians per
second to v"
(b) Sketch a graph of W m in the steady state versus different d-c values of v"
(e) Assume that the system was in the steady state with a constant field
voltage V,o and that at , = 0 the value of v( was instantly raised to 2 V(o ..
Obtain an expression for W m as a function of time after t = O.
Sketch this solution, assuming that the mechanical time constant is twice as
long as the electrical time constant.
(d) Let capacitance be a:l.alogous to moment of inertia and conductance be
analogous to damping.
Devise an ekctric circuit which is equivalent in behavior to the load on the
generator. Express the electrical parameters in terms of the parameters of
system.
Obtain the bll)ck diagram representation of the al1ll"idync system
shown il) hg. 71'-H; the feedback winding is energil.cd by the output voltage
thn)ugh a" damping transformcr." The al1\l'lidyn~' rotnr is driven at a constant
spccd (II. ror thc low-frequency range of interest, assume perfect coupling in
the transfllrmer and in the lield structure. Also assume linear operation and
nqlll,,'1 '1l1\l'lidylll' nrmat\ll'c ill1l'cL!:lI1l:c ill the olltl'lIt ("inTt-axis) ..:irelli!.
7-8.

B)

Rr.Lr.G"r

FI,.7P-7.

~,

\mpensating winding ~

Stabilizing
winding

~/

Motor

III

"tFlIwg

f)

475
j

I
'.

-,:

!'

~!

;- ,;.'

J.'.!.

,,~

.,l

7-9. The closed-loop voltage regulating system shown in Fig. 7P-9 comprises
the following components:
(I) A vacuum-tube amplifier for which the input and output impedances
may be considered infinite (this situation may be approximated by using
pentodes). The amplifier gain is K a = 0.001 amp/volt.
(2) An amplidyne exciter running at constant speed W a The quadrature
brushes are short-circuited while the direct-axis brushes supplying the output
are connected in series to two stator windings, denoted by c (for compensator)
and s (for series) respectively. Control excitation furnished by the amplifier
is supplied to the control winding, labeled f, located in the direct axis. The
series winding s is connected in such a manner as to produce excitation in the
quadrature axis in the same direction as that created by the short-circuited rotor
circuit. Th6 series winding c in the direct axis acts to produce exactly 100%

II~: .:
f:
;~

I
I

I:
"'j
~

,i

id

i r

Amplifier

."
~

:~"

,.~'.t,

,t.~1
,:",

I.,. .

C[~' ...

Ex~j~:~

..JI

ct
'.'....
'~l
<

Flg.7P-9.

Main
generator

.-

+.j.J>

''')

"~;;l. I
~d,

'i i~

"

':

,i

'<

r-------------- ..,I

,~.~!t

-;7;

476

ELECTROMECHANICAL ENERGY CONVERSION

compensation. (The polarities for the various windings of the machine are
indicated by dots in the figure.) The parameters of this machine are:

Control field I

Rr
1000 ohms
L, ... 200 henrys

Kr =- G.rw. = 400 volts/amp

R, = 10 ohms
Rotor (both direct and
L,= 1 henry
quadrature axes)
{
K, = GaaW. = Gadw. = 50 volts/amp
Rc'" 10 ohms
Compensating winding c
L c '" 1 henry
{
K ... Gqcw. = 50 volts/amp
R. = 15 ohms
Series winding s
L ... 2 henrys
{
K. = Ga,w. = 85 volts/amp
The gain K indicates the value of a speed voltage appearing along one rotor
axis for a unit current excitation in a winding located in the other axis. The
mutual inductive couplings between windings located on the same axis arc to be
neglected in this problem.
(3) A main d-c generator running at constant speed
The constants of
this machine are:
field resistance R, = 50 ohms
field inductance L. = 100 henrys
open-circuit voltage per unit field current K. = G.rw. = 1000
volts/amp
armature resistance R. = 4 ohms
armature inductance negligible

\
'J

"
-!'

-r,"
~

>

VL(s)
Ve(s)
VL(s)
VR(s)

Vs(s)

VR(s)

'th I
WI

.
WIth

L.

h ... 0

l'i"
~ '.

.,.,i

~ "

'th I
WI

L ""

Ve(s)
h(s)

th V,
R

(c) What is the undamped natural frequency of the closed-loop system and
its damping ratio with h = O?
(d) For steady-state d-c conditions (s = 0), and with h = 0, what is the
value of VEIVR for the two following cases: '
(1) The ampJidyne exciter is equipped with the series winding s. \
(2) The series winding s is completely removed.

;f)

Dynamics
of
Induction Machines

"~

liM;
-,:

-'~-',=-,,~

\"

,Uj'-'"

:~-(

1.\\
...:;

f
<1

,,1

4'i

1
d

.~

WI

.f;

,.

.~

,,""

f"

(a) Derive a block diagram for the whole regulating system indicating the
reference signal VR , the error Ve , the output VL., and the disturbance fL.'
(b) Obtain the following transfer functions:
:

w..

,~

..",
.J
I

I
:1

:J

,.,
J;,~i. .

8.0

Introduction

A~ pointed out in Sec. 4.6.2, an induction machine has a-c power

supplied to the stator from external sources. The rotor power is supplied
by induction from the stator. Consequently, the rotor is a passive
circuit with no conductive coupling to sources. The speed of an induction
motor affects the frequencies of the voltages induced in the rotor; hence
the condition for torque production discussed in Sec. 3.5.1 is satisfied at
all speeds.
Three principal types of induction motors are in use at the present time.
Large three-phase induction motors are used for supplying large amounts
of mechanical power at essentially constant speed in such applications as
rolling mills, punch presses, pumps, etc. Small single-phase induction
motors are used for supplying small amounts of mechanical power to
such devices as home refrigerators, air conditioners, blowers, etc. A
single-phase induction motor has no starting torque so an auxiliary
winding is used for starting and it may be left in the circuit during
normal operation as in capacitor start, capacitor. run motors, or it may
.be switched out of the circuit automatically when the motor comes up to
speed. With the starting winding in operation the machine is essentially
operating as an unbalanced, two-phase machine. A third type of
induction machine widely used is the two-phase servomotor. This type
of motor is designed for stable operation around zero speed and is
principally used in position servomechanisms. The advantage of an
induction servomotor is the low static friction because it has no brushes.
Conventional operation of induction motors occurs with the rotor

4n

I..
478

479

DYNAMICS OF INDUCTION MACHINES

ELECTROMECHANICAL ENERGY CONVERSION

';L.,.

circuits short-circuited. Most induction motors have a squirrel-cage

rotor in which all the rotor conductors are short-circuited together at
each end of the rotor. The rotor conductors and end rings form a sort
of squirrel cage, hence the name.... In <ome instances it is desirable to be
able to control the rotor resistance for better torque characteristics over
the whole speed range. In this case the rotor windings are brought out
to slip rings which can be short-circuited through external, variable
resistances. In some cases the speed of a wound-rotor induction motor
is controlled by controlling the frequency of the rotor currents. t
Operation of this type has been classified in Sec. 3.6.3 as synchronous
induction motor operation and will not be treated in the present chapter.
Only the conventional operation of induction machines will be discussed
here.
In the sections to follow the dynamic operation of induction machines
will be studied. First, an analog computer diagram suitable for analyzing
all types of induction machines including all nonlinearities except satura
tion will be developed. Then simplified linear representations will be
obtained that give approximate solutions rapidly in many special cases
of operation.

Electrical
?"f"
nll,I'ft = ("],f""
. Ilfl,"" -1- pY'"uJJ.rtJI)/.,'
.tt/l,lfJJ

l~L ~J

dll

.,

<

:~

"

[r;:/J] + f'[I~fJ1

p[/;'~:lI,s]

]l

pr/;'i~lIl)]

[r;',sJ

.I.'. .;

,'~

..

t.;,l(Y~

+ p[l;',sJ

(~i.la)

".tt

'

'I

"

(8:lb)

~
~I'
0
cosnol-sinnf ~
'l ~
0
R' + pL' pU', sin no cos no_ Ii I
~ PL,tosnOl
sinn~ ~I
0 j ~
- sin nr/> cos nr/>
0
R' + pL'
i;

..i
4

I
:

!
(8~lc)

v~

.~,

L-

Mechanical

..,.;;., I
-~

T = T", - Te

(8-2a)

.,)j

8.1

.
'

.l

(8-2b)

;}',

Equations of Motion

f~'~ l,~~~rl

Induction motors usually have smooth rotors;t consequently, the

equations of motion developed for the smooth-air-gap machine in Chap. 3
are directly applicable. In the case of a squirrel-cage rotor, the rotor
equations arc represented with the same number of phases as the stator,
which means that analytically the squirrel-cage rotor has been replaced
by an equivalent winding having the same number of phases as the stator.
In addition, the equivalent rotor for analytical purposes is assumed to
have sinusoidally distributed conductors whereas II squirrd-cage rotor
has a uniform conductor distribution. For the consequences of this,
which are not too detrimental. see Chap. lIon harmonic analysis.
The equations of motion for a two-phase, smooth-air-gap machine in
terms of actual coil variables are given as Eqs. 4-1<.)0-c and 4-20a-d.
These equations are:
.

T =

r,

f'

(J/> + at/ - ~ IW!fJ

'~

#.'.~:

~~.

1;""

:1

,Ii<

.jf

~,,"

'~

nLsr

~~I,'h:~

:oat"

:n;~;

Il!ail

B-Hl
I

Sf

nL

-sin nrj>

cos nej>

- cos nr/>

- sin nr/>

= (14) + at/ -

- sin nr/>

I - cos nr/>

cos nrj>

I -sin nrj>-

r-

.:::::;

"

s
IIJ

'
j
i;
~Hl

I.

nLsr[U~ip - ipi:) cos nr/> - U=i:

".

(8-2c)

ipip) sin nr/> J (8-2d)

-~

These equations are repeated in their various forms to illustrate once again
the definitions of the symbols used in the matrix notation.
Eq uations 8-1 and 8-2 describe directly a two-phase induction motor.
They also describe a single-phase induction motor if one of the stator
equations is eliminated. In addition, as discussed in Sec. 4.0, Eqs. 8-1

* A. E. Fit/gerald and C. Kingsley, Jr., Electric Machinery, McGraw-Hill, New York,

1952, pp. 125-127.
,',
jJ
"
t Ibid., pp. 42R-429.
t Small singlephase motors designed to start as induction motors but to run at
synchronow, speed have salient rotors .md utilize the reluctance torque due to saliency
for sylll;hronous opt,r;ltion.

,.

),:-

and 8-2 describe the energy conversion properties of a machine with any
number of phases on rotor and stator when the variables and parameters
are defined properly.
In a later section, three-phase induction machines will be treated;
consequently, the equations of motion of a three-phase, smooth-rotor
machine will be given and reduced to the af3 components of Eqs. 8-1 and
8-2. From Sees. 10.2-10.2.4, the equations of motion of a three-phase,
smooth-rotor machine in terms of the coil variables are:

(I~c] = L~ Icos

21T

41T

COST

COST

2;

21T

211

41T
cos- cos
3
3

-.J

vabc I

p(I:hc]

[r'abc]

+ P[/;bcl

i:mc

T =

N + (X~

~WJ

~[/rs ]

v~
'~

,:

~bc I = lib ' (8-Sa?;:,

C

"

,.

'1,

vr~ ~h

~
'abc = ib

(8-5e) ~

'~,'

[I~beJ

'u,1

'e

'

"~.-t>

'abe

'

0
R aS

(8-5e)

,..

00

~ R: '~
,'d

2;)

'

'~l'!

4;)
cos (n</> + 2;)

2;) cos (nq,

cos n</>

tw
,

(851)

cos (n</> _

,
;j

~7T)

'0'

' (8-5/)

-.

0'

<1

',:

,'-j

I
(il~

i;

t:.,

'~

,...

Vi

v'2

~'1

R" 0 "0
[r: bc ] = 0 a
0
,
0 0 Rr
'.
a

~~'

xa -

'1

0 .

; ,,' ",~..:

:1:

jL

2;)

..

cos nq,

n()i~;M kJ.

(8-5d)

are the three phase variables, Le., voltages Va, Vb, and V c are the line-to
neutral voltages of the abe three-phase system, and i ib , and ic are the
line furrents. The parameter matrices are:

. [<0R~ OJ

(n</> _'

\'

\ 'e ;

'

cos (n</>

[/;~c] = [l:~c]t

~~ '.;
~ = ~.

IJ

"'1S.c ,.'

(8-5)

The two-phase, real-component transformation from Fig. 10-11 is

:;"1

21T

cos (n</> _

'!"

(8-Sb)

(;

'S
~,.

rr:bC] =

=- L:;' cos

'a

'$

(8-5h)

COST

cos n</>

(8-4)

f.

211

<rI, ,"

I .

i;bc

(:&";.l '. \ '

..
ia

211

Icos 3'

i:be

v: bC ) = vb I
v'

1\

,"

~[I;bC]

abc

;1.

where

8</>

[/~c) = L:

41T

o~(l;bC]

t Ili;bClllli;bctll

COST

cos- cos3
3

id.

---,

(8-3)

II

{I j,

41T

COST

~~

p[l:;'J

(8-5g)

cos3

21T

V~bc I = ~bC] + P[/:bc]

~I

DYNAMICS OF INDUCTION MACHINe:;'

l:Nl:RGY CONVERSION

1;1..C\..II\VI It:LHANIl.AL

~uv

"

I'

1:1

Xa

-t
v'3

-t
v'3

v'2

U!I

"~f,31_

~-

[r~J

+ p[l~)

p[[~)

p[/~J
VQatl]

.1"

,r~,'"
I

-2

(8-6)

Xb

,i:t,,!UP~

Mi

Xc

where x indicates a voltage or a current. Application of the trans

formation of Eq. 8-6 to Eqs. 8-3 and 8-4 to change the variables from abc
to Oaf' yields:
<':j" "~

'1"

If}"

fJJ

P[/~\
~
-.J3

i..j A

(8-7)

".,!,\', .It ._

-.

'it

""'4f'<1

f
'4 " ~

..82

ELECTROMECHANICAL ENERGY CONVERSION

T = Jf

+ a,p - !

I i~l lltOaP' II

~
8~[/oap] ~

~[Ioap]

- ' '--
I

-W']
84, 0atI
8

~[I'.'
o,p OaP]

It is nppnrcnt from Eq,s. R-7 lind 8-8 thut the zCro-SCqllCIlt'~' part or I ill'
impedance matrix contains no indllctnnce terms, citht'f st'lf or mUllin\'''
Thus the torque does not depend on 10 and I~. Furthermore, when
balanced three-phase voltages are applied:

(8-8)

v: = V: cos wi

----l

where

v~

'oatil =
(r'OatI]

= R~

[/~]f1;" tL~

"

'0

~o o~

[/~]

1 0

[I"
Oap ] = ~l"
2 ~a
~

cos nt/J

-s:nJ

sin

n,p

cosnJ
0

cos nt/J

sin n;1

-sin nt/J

001

.~

;rt:.

(8-10)

= 0

." J

;.?

8.2

i'

.~)

==
==

tL:
tL:

'.

.~.

. (8-11)

..u~I!t(~;
\",\.j

I'
"

s,
1"".. 1.L
2

;\ I

",~

"

lj

,.w ...,..

'

~f1

Induction Machine Constraints

.,~;

The excitations for conventional operation of induction machines

consist of short circuits on the rotor and polyphase alternating currents
on the stator.
Looking first at the rotor circuit constraints, for the two-phase 'case
v~

','"

the unit matrix

L
L'
S

i~~

cosn~

100

[II] =

is zero:

q.~~

R'=R'..

tL:;

v~

In any case, the zero-sequence currents i oand i~ do not affect the torque;
thus the zero-sequence parts of Eqs. 8-7 and 8-8 can be omitted and the
problem is reduced to a{3 components. This means that Eqs. 8-1 and
8-2 can be used to describe the dynamic energy conversion properties of
a three-phase, smooth-rotor machine when the parameters in Eqs. 8-1
and 8-2 are defined as
,00
:q
.
'J ~

~o~
..

tL; 0 1

~0

~
~ ~

[lo~p] = [I~p]1 =

\,

, (H-I))

i~~(

v~ = ~(v~ + t1, + v~)

, lI'

= R~(u]

"
,:'\

cos (wt -

Equation 8-6 shows that the zero-sequence voltage

'p

["0atI]

.~

or

'OatIl =

'.:
Ip

271) '...
-3

.,
(wi -"3
471)
'v: ::=V:cos

~:

or

'.'

'0
s

.~"- V

v'P

[u]

.;-,

V~tlr = v~

~= Va

as

;- f1"~

'

v~

7l

-un

DYNAMICS OF INDUCTION MACHINES

= v~ =

(8-12)

The application of the transformation of Eq. 8-6 to these voltages yields

the result that Eq. 8-12 holds also for the Oa{3 representation of the
three-phase case.
The zero sequence impedance will contain inductance terms if the leakage reactance
is included in the self-inductance. For balanced impedances, however, there will
never be any mutual coupling between the zero sequence and the other currents.

486

~r
R

viJ

-=

"

+ pL

IR' + pL'

L
r-sr21
P R'

"

Ip ~

i;
(8-23)

[pJ n~l E+p;1 n~~ ~ ~:'i

lc/>.. + a~

sr2

[-n~Kll+p~= ~

......./

's'r,

n 7i!(lti'd

's or,

la1q)

'.'"

--~la."

3 '-- FGl
T =

:f

.I .,\.

"-0,

~.

l;

<':'~,:

i'f

~.,1.
I".

"
:;;l

.~

1, -8

It is apparent from the foregoing section that the transformation from

afJ to dq coordinates replaced the angular dependence in the impedance
matrix by speed dependence. In addition, it removed the angular
dependence from the electromagnetic torque. Nevertheless, Eqs. 1:>-23
and 8-24 are nonlinear because of the existence of products of currents
..lnd of speed and currents. Consequently, the solution of a dynamic
problem which includes all of the effects given in these equations is
difficult to obtain unless machine computation is used. The anarog
computer techniques described in Sec. 2.3 will be used to derive the analog

4,
\'.:

..

>

-.,

(8-24)

The equations of motion as expressed by Eqs. 8-23 and 8-24 arc

especialIy useful when a machine already in existence is to be analyzed
because it is possible to determine only the four parameters in these
equations from electrical tests on the assembled machine. This is so
because a no-load test yields two paramder values and a blocked rotor
test yields two parameters. Only four parameters are necessary, though,
because rotor variables are only auxiliary variables and the actual currents
do not have to be known. In fact, a squirrel-cage rotor is replaced by a
mathematically equivalent, sinusoidally distributed winding.

8.3.2 Analog Computer Representation

DYNAMICS OF INDUCTION MACHINES

,
-t."'
;

When the rotor equations are divided by R' and the transformations of
Eqs. 8-22a and b are used, Eqs. 8-20 and 8-21 assume the form

,
v'"

';"

ELECTROMECHANICAL ENERGY CONVERSION

'.
:r

,{
'J;:.

0481

computer representation of the dynamic behavior of the induction

machine as given hy Eqs. 8-23 and 8-24.
The analog computer representations of each of the four volt-ampere
equations or Eq. S-23 and of the torque relation of Eq. 8-24 arc shown
in Fig. 8-lu-e. The H.ve parts are interconnected to yield the complete
representation shown in Fig. 8-2. Note that the representation has three
inputs, v~, v~, and T, and one output, either t/> or~. In addition, any
other variables could be considered outputs.
With respect to the a{3-dq coordinates, the analog computer representa
tion of Fig. 8-2 is completely general. Consequently, it can be used to
study the conventional operation of induction machines and also to study
unconventional operation. For instance, by exciting one stator winding,
say v~ with direct current, the voltage induced in the other winding is
_ proportional to the angular acceleration. Such a problem can easily
be studied using the analog computer representation of Fig. 8-2. Such
devices as a-c tachometers can also be studied with the aid of Fig. 8-2.
Whenever the operation of three-phase induction machines with

unbalanced excitation is of interest the representation of Fig. 8-2 com

pletely describes the energy conversion properties. However, to have a
complete representation of the electrical system from which three-phase
variables can be obtained by transformation, the zero-sequence volt
ampere equations which were omitted from Eq. 8-7 must be included
in the representation. Since these equations are independent of the a{3
equations and are independent of each other, these equations can be
represented by two independent circuits. Inspection of Eq. 8-7 shows
that the zero-sequence impedances are pure resistances; however, if
leakage reactance is included in each phase the zero-sequence impedances
will be R-L circuits.
When the representation of Fig. 8-2 is to be used to describe the
operation of a machine with a number of phases greater than three, all
of the nontorque-producing electrical variables must be included if the
line currents are to be found. All the equations describing the nontorque
producing components are independent (see Chap. 10) so their repre
sentation is quite simple.
Although the analog computer affords an excellent opportunity for the
study of the dynamic behavior of induction machines, the method is
complicated, and involves money for the computer and time for setting
up the computer. In some problems approximations can be made that
allow rapid solution for dynamic properties. The resulting solutions,
though approximate, are entirely adequate for many purposes.

489

DYNAMICS OF INDUCTION MACHINES

II

'.' 1.

v:"o----l

..

pi do-------(

..

'1

PlfS

\\'

.,

Ola

la

;p~

,';
.

'.

.r'
Olq

(0)

r'
ido

."

Up

Plq

')

(d)

-'j'

~\, """'t!

..

.r'

')

;~i\

OlfS

..

,~:

IS
1 1

J
-l

.r,

Id

."

; (b)

.,

~\,.

,
r:i'

!.

.,

... t~\I.'~

,_i~

.,.

p'

',1

la

!:U~~

.r.

i in

Iq

,.

".

J~'"t

'1'"

....

,.
i'

(e)

1J.\
Pld

+
....,..,..

.,

.r'

Old

i'~;

8.4 Two-Phase Servomotor

j..,:.l
I'

.if\l

"lA

.,~.E'

~Jl'~

: .' ,of

~'t

,It

(c)

Fig. 8-1 (continued).

.r'

t'"

<><1IIl'",

:1

~-t

I.~

Fig. 8-1. Analog computer representation of five equations or motion. (a) Stator CII'
equation. (h) Stator fJ equation. (c) Rotor d' equation. (d) Rotor q' equation.'
(e) Torque equation.

,fie

;~,

.~

'i

'"
f. ..
\ ,(

',,:,'

:.

A two-phase servomotor is a two-phase squirrel-cage induction motor

especially designed for use as the output member of an automatic position
control system. A constant-amplitude alternating voltage is applied to
one winding of the stator, designated the 'reference phase. An adjustable
voltage 90 degrees out of phase with the reference voltage is applied to the
second stator winding, designated the control phase. By varying the
magnitude of the control voltage, the speed or torque of the motor can.
01 h/

'.

..'

> . . . .'

"

.~,

~t1
'1', '\V:t~

490

ELECTROMECHANICAL ENERGY CONVERSION

,4tl

DYNAMICS OF INDUCTION MACHINES

be made to change as shown in Figs. 8-3 and 8-4, If the phase sense of
the control voltage is reversed. the motor will reverse and exhibit the
same characteristics in the opposite direction, Thus, the two-phase
servomotor provides a compact and ruggcd mcans for convcning .In a-c
,
Torque

; 4!,:

,_.<::-./.".--.\,.
/ ,
.
,/

'

."I. .....(/'
'

VII constant

.;

.5

-~

c:
o

';::

"0

...o

.S

W
,:l~

CSIO:J~

II=V
~~-,

liN" VNCOS wI

,~

c:

')

.2

\ '\:

;.;J

0.

e
~

",

,<t

..........,

'J1""iII:

""'.........

',,=

Speed

\<

'~.'

i
8

"

'I: '

0.
E

"~l-:

f)

'0

....
I~

......

.:,j

:t
I

Fir. 8-3.

u:

."1/

"n".fl'flI~

.'

-,.;'

'oN'

Torque-speed characteristic of a two-phase servomotor.

signal of varying amplitude and phase into a shaft velocity or torque.

The drooping torque-speed characteristic of the two-phase servomotor
makes possible stable operation anywhere in the speed range, as contrasted
with a conventional induction motor. This characteristic is obtained by
using considerably higher rotor-circuit resistance than in standard motors,
Two-phase servomotors with ratings above 10 watts are usually equipped
with motor-driven fans to dissipate the heat developed internally by the
combination of high rotor resistance and operation in the vicinity or
standstill.

..

t:Lt:I...1

~~

KUMt:CHANICAL ENERGY CONVERSION

DYNAMICS OF INDUCTION MACHINES

~;

When a two-phase servomotor is used in a feedback control system,

the dynamic characteristics of the motor influence the operation of the
complete system. The analog computer representation of Fig. 8-2 could
be used with an analog computer representation of the rest of the system
UN

torque-speed curves describe the reaction of the electrical system on the

mechanical system even under dynamic operation. In addition, in a
limited region around zero speed the torque-speed curves can be closely
approximated by straight lines. In a position-control system operation
occurs mostly near zero speed; thus for such considerations as relative
stability the linear approximations to the torque-speed curves are adequate.
In Sec. 4.6.4 the linear approximation to the torque-speed curves of a
two-phase servomotor was derived. Following that derivation, when the
reference voltage is applied to the stator ex winding

.oj

:f.

VH
Reference
voltage

~'-.,

;~;

V~

"1\

]it

rw

viJ

~
'.'.
':, Jr,

.,
.!'I'Il-,J

~J'r,.

!,~

'

(T~)avi

d'JtAI"'~

"

x=

voltage

amplitude
..#'",~~~<'~'
~~

t'
/

....

"
I~'.~J.

./

'"

iJ>

..

I
Il.,

Reversal in direction
of rotation

Shaft

(8-25)

VR cos wt

speed

~,.~~.~..-.",..".~_ ...",.~ -'~

=ig. 3-4. Waveforms showing the variation in two-phase servomotor speed in
lccordance with variations in arnplitud~ of control-phas~voltage.... ,J.. ' QJ:'!1,

l .. _.

1'-

~t

(8-26)

= Vc sin wt

Vc

the resulting linearized torque-speed characteristic is given by Eq. 4-256.

When this expression is rewritten in terms of the parameters defined in
Eqs. 8-1 and 8-2, the result is:
"'

...,-,......

Control

VR

",{,'

J!'"

Yo

and when the control voltage is applied to the stator f3 winding

'.

Control

voltage

493

(n/2)L
Dm Ds\?

sr2

VR

/.r1\71

(2wVc - VRn~)

(8-27)

The average torque is used because the torque that pulsates at the second
harmonic of the electrical frequency w will cause no appreciable response
in the mechanical system due to its low-pass nature. The linearized
torque-speed curves are shown in Fig. 8-5.
In the application of the two-phase servomotor the amplitude VR of the
reference voltage is constant but the amplitude Vc of the control voltage
varies with time. The variation of Vc is so slow, however, that the control
voltage is still essentially sinusoidal and Vit) is used to describe the
envelope of the control voltage.
When Eq. 8-27 is rewritten as .
(T~)avB = K vV c - K~~

".A- .. ,

iV.

.w'..

"1"

,,' \

~,.~U;

,,~,';'~

''*-'~:'' fl
'l>f'~ .!

i . ,

, I

i. I
I'

(8-28)

/'

where

K"

"I

nLsr2wVR

I'

i"

= K[(RS)2 + ( .. _-

~
~.

or design or analysis purposes; however, a simpler, though less accurate,

nethod does exist.
It usually happens that in a two-phase servomotor the electrical time
onstants are much smaller than the mechanical time constants. Thus

he electric transients occur at essentially constant speed; and, from' the,

it"wpoitH of the mechnnicnl system. all electric transients appear dis- "
OIHinllo\ls. T<l n good degree of approxim,ution the' steaJy-state

:..ni

'I,.

(n 2 /2)U,.zV;
K. = R'(Rs )2 +' ... _

f,~

it can be combined with Eqs. 8-2 to yield

'T( )
I

""'

J d </>
dt 2

+ a d</>
dt

.. V ( )
c t

fi"

This expression can be rearranged to the form

,- t,:~~

'J':-

,,'J
;{,

+ fi~ d</>
dt
'

(8-29)

-49-4

ELECTROMECHANICAL ENERGY CONVERSION

T(t).

K uV,(t)

tl 2</>

= J -d
t 2 + (a +

d,!,

The trllllsl'er rUllction of Eq; 1\::12 is the rOml nOI'/lllllly used in tlte study

(~-30)

K.,.) -,tf

This is u linear dilrl:rl:ntial l:~IUlilioll wilh Cl)Il~llllll \~l)dlil.it:llh

11111.1,

Torque

(~lll1trolsystems (~ontaining

of fecdoack
run~~li(l1l

as

servomotors.

This same Irnl1sl'cr

is olkn (,hlllilwd hy findill", lilt' l'onslanls K" find K.,. ~ral'lti\':t11y

,from a measured sct or torquc-spccd cUl'ves.'" This graphical kchniquc

is also ortelt used to determine the incremental transfer function at some
spccd ollwl'thun !'cro, for which application the usc of the amdyli(,;
torque-speed expression entails rather involved algehraic manipulations.

such, it can be handled using any of the linear techniqucs or Chap. 2.

l'.h,h

(1~)avg

VR = constant

"I

.J,

'8.5' Three-Phase Induction Machines'

Yc

'r

increasing

.,); 1 ~

\1

i '

/ ,ji, ~'1'

Speed
,f

.\;

;p

T~rque

r;

k;..

!.~

. " t1$

'.';; ~'.

J:w'

,.,C/
".,'j,

'n,
'~. I> ~~~'~;':).l..

~Normal

.,

~r

~' j ,.

1';,,1

'j

I,

Linearized torque-speed curves of a two-phase servomotor.

'1

I
I
I

,,)

;;.I

= [Js2 +

(a

K</s] ep(s)

!,

Km

SeT mS +

where

Km

1)

(8-31)

,~~,

(8-32)

Ku

= a

K</>

,~l41,

:'

0.J_
1

+----- Speed
w

If'

ilL,

w'"
Synchronous
speed
Slip

Fig. 8-6. Typical torque-speed curve of conventional induction machine.

and then solving for ep(s)j V,(.\') to obtain

ep(s)
Ve(s)

'tt

I'

:l~,?Y't

;'.

<'."

The transfer function from control voltage Vc to position</> with the

applied torque T equal to zero can be found by first Laplace-transforming
Eq. 8-30 with zero initial conditions to obtain

KuVe(s)

.:~d" ,11:

operating
range

<

Fig. 8-5.

-'1r

Large induction motors (l horsepower and up) are normally used to

supply continuous or intermittent mechanical power at a speed ncar the
,synchronous speed. Such motors have steady-state torque-speed curves

f7>a

~?S

DYNAMICS OF INDUCTION MACHINES

'.;

of the general shape shown in Fig. 8-6. The region of normal operation
is shown in Fig. 8-6 as the region of negative slope just below synchronous
speed.
For high-power-Ievel applications (10 horsepower and up) induction
motors are usually of the three-phase variety and are operated from
balanced three-phase voltages. The rotor is usually a squirrel cage but
it may have windings if speed control or starting-torque control is to be
achieved by control of the rotor resistance.
See. e.g., G, S. Brown and D. P, Campbell, Principles of Servomechanisms, John
Wiley. New York, 1948. pp. 131-133.

Tm=a+K.

............ " V , ' I L. ... nl"'ll ........ I"'IL.

CI .... C"U I

\..VI'I v el\.:>IVN

DYNAMICS OF INDUCTION MACHINES

Although the steady-state energy conversion properties in the normal

operating range are always of interest, the dynamic behavior in this region
must often be taken into account. In addition, the transients that result
when the motor is started are usually of interest.
The equations of motion for a three-phase induction machine have
already been given in Sec. 8.1 where it was shown that the (x{3 components
of Eqs. 8-1 and 8-2 can be used to describe a three-phase machine. In
addition, it was shown that for balanced three-phase excitation the
equivalent two-phase voltages are balanced. Only balanced three-phase
operation will be treated in this section.
For a complete solution for the electric and mechanical transients
that results from any disturbance, an analog computer can be used with
the representation of Fig. 8-2. However, reasonable approximations can
be made that allow rapid analytical solution, without using an analog
computer, that are completely satisfactory for most analysis and system
design purposes.
The same type of approximation is made here that was made for the
two-phase servomotor in Sec. 8.4; Le., the mechanical time constants are
much greater than the electrical time constants in most large machines
operated from constant-voltage busses. Consequently, during any electric
transient the speed of the motor changes very little with the result that
electric transients occur at essentially constant speed. This approxima
tion allows the electrical system to be analyzed using the equations with
constant mechanical speed-i.e., steady-state equations-and from the
viewpoint of the electrical part of the system a transient in the mechanical
system proceeds slowly through a succession of steady-state electrical
modes of operation. Using this approximation, the reaction of the
electrical system on the mechanical system is given simply by the steady
state torque-speed curve.
The electric transients are so fast relative to the mechanical system that
from the viewpoint of the mechanical system all electric transients appear'
as discontinuous time functions (steps). For instance, from the view
point of the mechanical system a step in line voltage causes a step change
in motor currents (both stator and rotor) from the initial steady-state
value to the final steady-state value, both current values being calculated
at the mechanical speed that exists at the instant of application of the
line voltage step. After the initial step the currents proceed through a
succession of steady-state values as the speed changes in the mechanical
transient.
The approximations discussed above will be used in the ensuing analyses
of induction machines under normal running conditions and during

,;

11;f;""

..

The steady-state characteristics, which will be used in the transient

analyses, havc already been describe in tcrms of + -, fh components in
Chap. 4. The transformation relating a{3 and + - components is, from
Fig. 4-28,

Ii"

~.

,.; ~-

V2

and the transformation relating

~(

ITIJl~

IT

+-

andfb componcnts is

OJ

:-1\\"
"

',Ii

'.t'

"",

(8-33)

1-2) ;]

~ = pn~ I
~
~ La I e-j~ ~

e~

(8-34)

where x represents a voltage or a current. With the balanced three-phase

cxcitation of Eqs. 8-9 and the definitions of Eq. 8-17, the phasor
symmetrical-component voltages are (see Sec. 4.6.2)
,
1
V: = Y~eJ6+ = v2 (V~ + jVi;) = V3 V:
(8-35)

ys = y s e)' =

"~

~
v2 (ysa -J'V~fJ} = 0

With these voltages applied, the electromagnetic torque developed in

an induction machine has been calculated in Sec. 4.6.2 and is
~f

't

l!. wLsr2 R' vs2

T

1',
<:~,

:'?J.';'
'~"' '
1,

'

r = [R- SrR'-

2
s
sr2
]
2
s
w2 (L'L - L )

(8-36)

+w

2(

L r RS

+L

Rr) 2
S

j,;
'i

I' ,

In Eq.

in which the parameters are defined by Eqs. 8-1, 8-7, and 8-11.
8-36 the slip s is given by
I~~.''~

.<,",

,-./,~U/.,

s =

w-nw

nW m

t .

J .

-: !

,~

"'-<'!

~~

"

(8~37)

~v'

!"

Equation 8-36 combined with the torque equations, Eqs. 8-2, yields
-:',

dwm
T= J dt

aW m -

Te

"~

r"

-'J
t~',

"

(8-38)

. <I

..;-,~.l

,.~."
. ~-- ... ~j,.,

where W m is mechanical speed, and it will be used to describe mechanical

transients. Note that Eq. 8-36 is nonlinear because the mechanical
,'A
speed W m is contained in the slip S"

sturting tmnsicnlS.
~

497

.io:

,,,'"

~."

';:4f'

t.l'~ "~";'i
1."~# \t_~.~

..

'1" '
~

,"

"'

498

EL ECTROMECHANICAL ENERGY CONVFRSION

DYNAMICS

The only ekelric transients will be eurn.:nl transients beeau'" cxcitation

voltage is assumed to he eonslant. With tlie assumption of n.:lalivc!y
short e!<':etrical time constants. purely ek'<:trie Ir;ll1sicnls (i.l' .. currcnt
transients) arc disregarded and the current is considered 10 he in the
steady state for whatever instantaneous speed exists. Tile slcady-statc
symmetrical-component stator currents have been found ill SCI.:. 4.(1.2.
Using Eqs. 4-231 a, 4-232a, and 8-35,
[(Wls) + jwL']V~
+ jwL'](W + jwL') + (wl./')'!

/$

I:

[(Rrls)

The only function of time on the right-hand side of Eq. 8-39 is the slip s;
thus during a mechanical transient both the magnitude and phase of the
current change continuously with time due to changes in the mechanica)
speed. The actual three-phase line currents during the transient can be
found by using the transformation from + - to a{3 components in Eq.
8-33 and the transformation from Oaf3 to three-phase variables in Eq. 8-6.
This transformation can be made directly by using the proper trans
formation from Fig. 10-11.

"J

"I' I "
<'

. .I.dw".
. .. I ("

til

1/

I
l'
t

wmO =

Normal Operating Range

(1112)/.1),,2 V~.f
W(R,l + w 2L$z)

T
'1'

rf

(8-40)

Substitution of Eq. 8-37 into Eq. tI-40 and the use of that result in
Eq. tI-38 yield the linearized mechanical equation of motion: io!; ,;~,

J dw,.,
til

+ aw _:! LsrzV~(w - nw m )
m
2 R r (R,2 + w 2L,2)

If we define the electrical damping constant as

?

nU

lt

= 2"

:fl6t#.

,d.'

(8-41)
f"i ~:~'.1 '-" 1"tl
:~J!

L"l V,l
+

R'(R

SI

+ w2L$')

(8-42)

u ..) ...,,",

(HI.l)

(a :e aJ ~

(8-44)

When the mechanical damping constant a is zero the no-load speed is

synchronous because no average power is necessary to keep the rotor
turning.
For operation around synchronous speed (.I' "" 0) the current of Eq.
8-39 can be simplified by multiplying numerator and denominator by s
and neglecting terms multiplied by s in the denominator. The result is
/+

The dynamic behavior of the induction motor with balanced voltages

in the region of normal operation near synchronous speed (see Fig. 8-6)
is simple. This region is one of small slip .I' (s = 0 at synchronous speed);
consequently, the electromagnetic torque of Eq. 8-36 can be linearized
with respect to .I' by multiplying numerator and denominator by .1'2 and
neglecting terms multiplied by .I' in the denominator. The linearized
result is

~9'1

This is a first-order, linear dif1'crential equation with constant cocflicients.

Any or thc lilH.:ar !c<.:hnilIUCS of Chap. 2 I.:an be uset.l in thc solution or
transien t pro blcms deseri lll:d by this cljuation.
The term ",.('''/11) in Eq. X-43 is a constant and determines the speed :It
which the motor will run when the applied torque T is zero. Define this
no-load speed as wmo, and it is found from Eq. 8-43 as

8.5.1

INDUCTION MACIIINC$

cl\uution ~-41 C:III b~ writtcn in thc Corm

':i

(8-39)

= 0

or

(W + jwsL') V~
K(R$ + jwL$)

(8-45)

Substitution of Eq. 8-37 into this result yields

l. [R

r
H"

+ jew
R' (R S

- nwm)L'] V~
+ jwLS )

(8-46)

It is evident from this expression how the magnitude and phase of the
stator current change with mechanical speed for operation around
synchronous speed.
.~.,

EXAMPLE 8EI

- 'fl~'- (7

.':~,~r.~ . <I

A good example of a high-power-Ievel application of induction machines

in which the dynamic characteristics in the normal operating range arc
important is a punch-press drive. In this application the motor is
subjected to impact loading; i.e., the load torque occurs in the form of
pulses of short duration. An impact load will at least momentarily slow
the motor down. If consistent punching is to be obtained, the motor
must be allowed to return to its normal speed before the next impact
This is essentially the same result as obtained in W. Y. Lyon, Transient A naly.l i,
of Alternating-Current Machinery, Technology Press, Cambridge, and John Wiley,
New Yark, 1954, pp. 259-260.

. -

\0.

,,""'

,-VI~

u r 1"11"\1-11'-;)

V CI'\.,)&VI'I

IX e

~n = J ddWm
+
t

T(t) dt] uo(t)

Wmi

(8EI-I)

',on

WmO

Llwme- tfT

(8El-6)
.. ~

fl.

IX

IX

(~) ~
IX

u~

rtf.,
J

(8El-7)

'This response is shown in Fig. 8EI-2.

,(.1

r
I

...

.i

tH I'f"

'All!

Wm

.+

!ftl:>1!

'n~

wiitlilrr

iT
:.'e It)'J

Time t

,~

Impulse [uow .

"I,

+ "'7

~"

ni

~~;r--

" ~i

(II)

W,

Fig.8EI-I.

.101

:AJ

':(,1.

;~'

.
in}

~hape.

'ii"

~ I

'~li1 i o~

't'i~l :.,)!

-""

'~'R

(8EI-5)

:j"
"I'

U)l

,1JI,-

:/.<.J

(a)

~wm

Substitution of Eqs. 8-44 and 8EI-3 into 8El-6 yields

:;'~~,

t)l

T(t)

Time t

Wmo

r;.

,t -:,'.
Torque

(8El-4)

IXe)W m

where

where uo(t) is the unit impulse occurring at t = O. To find the response

of the induction motor to an impact load applied at t = 0, the impulse of
Eq. 8EI-l is assumed.

T(t)

(0:+

The solution for Eq. 8El-4 with the initial condition of Eq. 8EI-5 is

Torque

;)VI

I-II"\,-nll'll:.)

with an initial speed

Wm

U:

=-

lI"1UU'-IIVI"I

For I > 0, the torque T is zero and the differential equation becomes

load is applied. Consequently, the time that the motor requires in

returning to the unloaded steady-state speed defermines the allowable
duty cycle of the motor under impact loads.
An impact load on the induction motor occurs as a pulse of negative
torque that lasts for a relatively short duration, possibly in the form
shown in Fig. 8El-l a. If the duration 8 of the torq ue pulse is small
compared to the mechanical time constant, the torque pulse can be
approximated by the impulse function (Fig. 8E1-lb)

luo(t)

vr

Approximation to impact load torque.

(a) Impact load torque pulse .

:j~,

Fig.8EI-2. Impact load response of induction motor.

I,'
I'

(b) Approximation to be used.

The classical method of solution given in Sec. 2.1.1 will be used for
solving Eq. 8-43 with the initial speed at 0- given by Eq. 8-44 and the
driving torque of Eq. 8EI-l. The application of the impulse of torque
causes an instantaneous drop in angular momentum of
Ll(Jw m )

= J Llw m =

or a step change in speed of

:,:.

-1

t..,

'~,f.iit:l~.% 0,

(8E1-2)

!;,

I
.dIu", -]

'lb. T',.i'<

1m!:,

(8El-3)
\ S~

The key characteristics of the response of Eq. 8El-? are the speed
drop L1w m under impact load and the time constant T which determines
the time required for the motor to return to the steady-state speed. The
impact speed drop is often reduced by the addition of extra inertia J; in
fact, the addition of inertia is often necessary to ensure that the motor
doe~ not leave the normal operating range and stall under impact load.
In many cases, however, the addition of inertia is not desirable because it
increases the time constant T which reduces the allowable duty cycle of
the machine. The time constant is smallest when the electrical constant
U r given by Eq. 1\-42 is as large as possible.
From Eq. 8-42 it is apparent
what parameters govern this constant. Incidentally, the constant IX. is

'.,

." ~f,-~'

r,

J_:'~

502

ELECTROMECHANICAL ENERGY CONVERSION

DYNAMICS OF INDUCTION MACHINES

the slope of the torque-speed curve in the normal opemting range (sec
Fig. 8-6); hence, for best opemtion under impact loads this slope should
be as large as possible. This is usually accomplished by making the rotor
resistance R' low.
The stator current that flows during the transient can be evaluated by
substituting Eg. 8EI-7 into Eq. 8-46 to obtain

to thc stClu.ly-stutC,llO-ltHlll valuc us the Il1cchanicul transk'nl dies. The

inslllnlllneous change ill currcnt ducs lIul occur ill prm:llcc bccause for
rapid changes ill specd the dl:ctrie trullsicnl rI:S!WIISC musl aIst) bl: (;011
sidercd. 1n rcpl'att'd opcrati()f1 of tl1\: mllchinl' in II punch-!'rcss dri\'l'
care must be exercised to ensure that the repeated current lx'aks do nOI
overheat the mw.:hine.

{ R'

J'

+j

[(_a_)w
+ nl e-'/T] c} V~
+ ",.
J
IX

R'(R'

+ =

(!ml-8)

8.5.2

+ jwL')

.'

I'

IS

= ~

R'

R,2

[(_a_)w
+ '!..! r'fT lJ L,z V~
+
J
2

ex

a,
vi R s2

The speed transient that results when an induction motor is slarted is

often of interest. Inspection of the torque-speed curve of Fig. 8-6 shows
that the torque does not vary linearly with speed; consequently, the
starting transient involves the solution of a nonlinear differential equation.

(8El-9)

+ wZlf

The resulting stator current transient is sketched in Fig. 8E J -3 where

the envelope is given by Eq. 8EI-9. Note that in this transient the
current increases instantaneously as the speed drops and then falls back

'f

Torque

...

Stator
cutrent

. HJf~~

Starting Transient

;1

This yields an amplitude of

~OJ

;"iY'~~lI

,_.)):i:J.)

Load
........... torque

"'; fM
.t

\1

~; ~ ~

'+

Wm

moo

Speed

(a)

l'"

rll:

'::.(

l:f'

~.~Il-

"r--

,~

"

,'.::it'1 ~. j:.J i
.: ,';1,.;

:~~ih"'"

'

I, J

"f;

;.-;
"~

...

__ !

\.",;

"

,"

LI//A

Wm

(b)

Speed

moo

Time
Col m

f'

___l _____

__

~I-

Vl

Wm_

r----------=-
...- - - - -

!-

VV

tnb

;,00

,,~'~"

I [I,

""

(1:)

Fig. 8-7. Calculation of starting transient for induction motor.

curves. (b) Graphical integration. (c) Speed transient.

Fig.8EI-3. Stator current transient for impact load.

S"

(a) Torque-speed

......... 1 ","vlll:,-n/,\I'lI,-AL. eN el'\.I", l'

' f:Z

I...UNVcl\~ION

Under simple conditions this problem can be solved graphicaJly using the
approximation discussed earlier that the steady-state torque-speed curve
represents the instantaneous electromagnetic torque.
The method is now described. * The steady-state torque-speed curve
and the load torque versus speed are plotted to the same scale, as shown
in Fig. 8-7a. At any mechanical speed W m the torque difference JT is
available for accelerating the inertia J of the system according to the
differential equation

AT

= Jdw m

PROBLEMS

8-1. A two-phase servomotor has its control winding excited by a suppressed

carrier current
ie = llt) cos wet
and its reference winding excited by a sinusoidal current
iF.

where

(8-47)

dt

(w m J dwm
)0 JT

We

= carrier frequency

lR = constant
l,(r) = modulating signal amplitude

T~

(8-48)

Thus if the quantity JjiJT is plotted as a function of W m by measuring

.tJT in Fig. 8-7a, the result shown in Fig. 8-7b represents the integrand on
the right-hand side of Eq. 8-48. By performing the graphical integration
of Fig. 8-7b the speed-versus-time curve shown in Fig. 8-7c results.
The speed-versus-time curve of Fig. 8-7 c can be used with Eq. 8-39 to
obtain an envelope function for the stator current. This will give a good
approximation to the electric transient that results when the motor is
started.

;~~

'~,

,~

r b.

In this chapter the dynamic operation of induction machines has been

covered. When the details of both electric and mechanical transients
must be known with considerable accuracy,
analog computer with the
representation of Sec. 8.3.2 affords probably the simplest method 0;'
obtaining the result. This representation includes all nonlinearities
except saturation.
For some applicutions in which the mechanical transients predominate,
fhe simrler, approxima(c (cchhiqucs of Sees. 8.4 and 8.5 will often suffice.
Thesc tcchniqucs can he applied without the aid or (:oll1pulcrs to give
useful answers in a minimum of time.

= .Qo

+ J(dr/

T= To

+ JT(t)

;~,;

)#,

."

f>t ,:,

;!"., "
I :

(Q159 0
0
fJf~ = fJf;:m = I 0
0.159 0
~
0
0.159
:P~:,

Stator

= 'yJ;;m

O.OJ38
-0.0135
-0.0135

-0.0135
0.0338
-0.0135

i-:'

'f":t

,~,:;.

r-1.'~' ~

.'

..

ohms

-O.OI~5

-0.0135
0.0338

;1/

V.
henrys

rotor 1l1ll1\l:11 induClance cons(:1111 I." ~. 0.0304 henry,

(0) Express the sla(or to rolor mu(ual inductancc matrix .:1':,:" and the rotor
1051(\(01' Illalrix .:1';,:.. ill terms or the relative displacement angle ,/,.
10

.if,

,';( , i;

dl

Obtain the linearized incremental equations and compare the coefficients with
Xv and K defined in Eq. 8-28.
8-3. A 10-hp, 60-cps, six-pole induction motor wound with three phase
windings on both its stator and rotor has the following paramcter matrices:

t\'

Thi.. 1l.111l'hkul meel""" "I' WIUli\ll1 follows 1\. E. Fillgcrald and C. Kingsley, Jr.,
n.'('lri<' Machil/ery. McGraw-Hili. New York. 19S2. pp. .~77-S79.

,(~'*:,l~.H;' l't ':,:T'

"

an

,..

Kle(t)

Ve = V,o + J VC<t)
dl

Summary

Determine the value of K.

(This type of servomotor operation is frequently used in feedback control
systems in which signal information obtained from synchro components is in
the form of suppressed carrier. The motor inherently acts as a demodulator.)
8-2. Equations 8-29 and 8-30 were obtained by linearizing the equations of
motion for the two-phase servomotor for small signals about a steady state of
zero.
Perform a similar linearization for small operation about the quiescent point
defined by VeO , .Qo. and To by writing

drj>
8.6

= lR sin wet

The frcquencics of the modulating signal are assumed to be small compared

with the carricr frequency so that 1,(1) can be treated as a constant.
With the motor at standstill (slip s = I), show that the main component of
average torque is of the form

If the transient is initiated at t = 0, the time required for the machine

to reach the speed W m is given by integration of Eq. 8-47 as
t =

50S

DYNAMICS OF INDUCTION MACHINES

.1

....,

',-5"~

507

ELECTROMECHANICAL ENERGY CONVERSION

DYNAMICS OF INDUCTION MACHINES

(b) Obtain thc instantaneous equilibrium equations (volt-ampcl'c am! torLluc:

relations) in terms of symmetrical components,
(c) With the rotor short-circuited, a balanced sct of voltagcs is applied to thc
stator:

8.5. Thc systelll depictcl! in Fig, SI'S is lin im!ucti,m III0 I,"' ,hiving II 1,lllU
throul,th 1\ tlt'\ihk sh:i\'l, Tht' inc."rliil Ill' rill' I11l1tllr is .1,.111111111(' incrti" Ill' th(
load is ),., Thc torsi()flal spring constant or thc shan is K. NC/llect mechanical
lblllping lIlld usc tlte lincari:lcd equations or ~ec, !l.S.1 to rind lite equlllwns
lhat dl'sl:rilx' this system with bil!:II\Ct'(\, C(lnslnnt-i1l11l'liludc 1I1'1'licd vlllti\~('~.
Obtllin 1I ditt"crcntilll cqu.ltion relating load torque Tr. and 10'ld spccd W,._
Whllt is the: rcsponse in wI. to a stcp of load tor4ue '!

506

v~

+ 27f)

y' cos

(WI

v: =

V' cos

(WI 237f)

y' = 127v'2 volts

with

.'.~-

w = 120rr rad/sec

v; = Y' coswt

Balanced
voltages

lf.

( (T

Induction
motor J",

"h

Determine the steady-state behavior of this motor at constant speed

m = (I - s)(w/n) mechanical rad/sec.
(d) Express the vOlt-ampere relations for steady-state operation using
complex phasor notation. Show that stator and rotor currents are of the form:

Flexible
shaft K

i!.. = I'e}wl

and

if =

Flg.8P-5.
, :~

l'e 1wl

Also show that the volt-ampere relations reduce to:

,:

, ,...

~
o

'

c' :,j~'+ _,j!.Ift_~

1ft

~ + JX'

I s

l'

\j

"t"

T,

= p

nr; L" (jJ'''/' nm

jJ'l'*)

.,-'\
J

"1
';1, ,

;-~.

"

,.v

,.

.
L" Re(jl'''l')

~\

Also show that this torque relation can be written as

')I:,
,l,!

Tt =

P R'

II'

12

n--~- ;r~,

w S

(f) Compute the speed, electromagnetic torque, stator current, and power
factor for the two cases:
s= 1
,;it
s = 0.02

8....

i~'"
',"

;;t

What are the values of the parameters in the above impedance matrix?
Derive an equivalent circuit that represents this impedance matrix.
(e) Show that. for steady-state and balanced conditions, the instantaneous
torque is also equal to the average torque by proving that
,~,,~

~.

Use the linearized equations of motion developed for the normal

operating range of an induction motor in Sec. 8.5.1 to find the response in the
speed to a step change in load torque with constant applied voltagc amplitude.

it'

~'
'~r~

\
.)t

.j~

in this chapter the rotor windings are three-phase instead of two-phase.

Thus, the general methods established in Chap. 3 for calculating the
parameter matrices and the general inductance parameters of Eqs. 3-31
through 3-36 become immediately applicable. Rather than repeat the
detailed calculations, the parameter matrices are written below by
inspection.

''""
c

-a

on

<.l

n;

.!l!

;(

'"

..
1\
!!!

)'1'

."

n;
.j,!
c

'"

'""

II

,tl~

on

;; I'

'''""" "''"'" "

."

"0

I
I

I
...

~
.::.

c
.2

.,

.l;!.:i

...

..

"9I

~.

'"cc:

,t

.8
>>c:

'"

:-e~ a r I
~I "I',

_It

+Q-="

~!:
~I

..."'c::

'a

"

--e.
::t

-IJ
~
I

tl

(1;0] =

'"-&

.~

....,

_,,~

::'"

,tj<o>

bO

.Ej

-a

"

+ L~2

S0

...

""!""!
':'?

-t

.~-:~

"

~
~

flitljr~j
"'I II
~

i ~

II
I

:IQ
, ...

.... '"

_abc,ao

UJi.J('

(9-20)

1\

(/~bc]

J!'

L~ ~ TI + L~2 ~ - TI

(9-2b)
,1'0,11'

iJ

\'.\

'II
~WC

'-"'i'

rJ,;,ft!-

,.... )

cos n(24) + 27T(3n) cos n(24) + 47T/3~n

os n(24))
cos n(24) + 27T(3n) cos n(24) + 47T(3n) cos n(24))
cos n(24) + 27T(3n)
cos n(24) + 47T/3n) cos n(24))

\".

_/it,

'h'

(9-2c)

ll;b,abc]

-0

.~

kW

[(1~b] I [l~~,ab~]

I[I' '

_~

0:

(9-1)

[l.l .. L~ -1
I-t
-! -!" 1

oS

.~

.c::
(>

"

,,,,

'"
~
.g,

t
-I"::

"'t .- ~'-'

0
R' 0
R'
0

10

where

:!OJ
+

"'u

filS.'
ab,aoc

0
0

~l

c::

" ,..,

R' 0

...

"~

01

,~

-L

om

,,~~

....

): I

~""

1-"';

"0

~~ 1~~O+

~I~

ab,abc
~..

-5

~+

I)! _IJ j ~.~

lot

<.l

":_'"

.91

IRS 0 I
I 0 Rb

kit

.!!

511

DYNAMICS OF SYNCHRONOUS MACHINES

".~

I:LI:C\ ROMECHANICAL ENERGY CONVERSION

)IU

~>:

~r

h"

iI:

..

leos n(e/

Icos n(4) + 27T(3n)

~J

+ U'

L'" cos n(<(,)

cos nCr/> + 27T/3n)
cos n(<p + 47T(3n)
... cos nCr/> - 7T/2n) cos n(<() - 7T(2n + h(3n) cos n(<p - 1T(2n + 47T(3n)

1'2

ES 1//> + 1T(2/1)

cos "(4)

+ 7T/2n + h(3,,)

Icos n(4) + 41T(3n) .

cos n(ef>

'.

..,

+ 1T/2n + 47r/3~
(9-2d)

lI~~c.ab]

..

lI;:~."bc]"

(9-2e)

,(~,I.;;~,~:"fi1Jl#

#,,~

'!It
f",~
.'
~t

, .~;
'II't

,r'-"

.....L~

"
~.

~.'

.,,,".,,",,

, ,.1-'1 ~".

,,,,~,-~.'

hit, ",L",.

~~'pil-,.
:';"ilo.1

~';" ;,1 .",'"

'f' I ~',

"

Dynamics

'.

"<'~:

(J'

r"

j,'{.

'------"

of

Synchronous Machines
"H

'l

~ ,j> I l:'

l;'i~ =\

c,.'" ' ,.II

\~

:,'

,.rtl'
'''1 1

~'

I'

9.0 Introduction
~~~ ;

'

:~k

'~j'(C

4,"'-

/.,

,~~"

'

,.!l"',
.,fI:.~

;'. ~

Jp

dr

_".f'"

~I',..":

,I
\

'~,'

~.

':ili
:~

'; ~:

,,'.

\'?i4.,

t':~

I,

:1 '\

"1 ~I

ll 1

"'!

::\ JI

.. t,f.

';lilt

'j'.1

,
','I',V

''-'I

l' ~;:

;.~

lI
I

Iii

'ill.,
,n"
~!,

.,~;

;';,;-4')

11".

f, "
t~,<'.t.'fl9(:J

t-..J
"" >h.",:.'\:'
,)1
f

The synchronous machine is the single most important electric machine

for the generation of a-c power. As a source of power it may be used as
an isolated unit or it may be used as one of a whole network of synchrOnOl!S
machines which are interconnected to supply power to a common load.
The intcrconnecting network may be a short bus for two machines in c!o,e
proximity, or it may consist of transmission lines and auxiliary equipment
of many miles in length which convert generating systems located in
different geographical regions into a single connected network. The
problems of system dynamics for the complicated interconnected synchronous machine system become very diftlcult, and even the dynamics
of a single synchronous machine taking into account its mechanical drive,
voltage regulating system, and load is an involved dynamic problem. It
is not feasible, nor within the scope of this book, to attempt to analyze
an involved interconnected system of synchronous machines. The general
analytic techniques of machine analysis introduced in Chaps. 2, 3, and 4
:tnd ~encralized in Chap. 10 make it desirable to show how the problem of
syncltJ'l)110US machine dynamics can be approached using the techniques
already established. For this purpose a three-phase generator will be
chosen as the device to study since three-phase synchronous generators
arc ~tandard for power generation in the United States and most other
countries of the world. Two problems will be investigated: (1) the
dynamics of voltage regulation in a single machine and (2) the dynamic:
of two parallel-connected generators including the voltage regulatin~
system and the drive system. The drive dynamics will not be develope,

~l

DYNAMICS OF SYNCHRONOUS MACHINES

509

" explicitly. but the, synchronous machine representation will be developed

'"' so as to allow the introduction of the drive dynamics.
The approach which will be followed is to develop computer representations for the synchronous machine and system. A direct analytic
,solution in closed form of the equations of motion for a general dynamic
problem is difficult, if not impossible, without restricting the case considered to one of extreme simplicity. With the availability of modern
computers, both analog and digital, it is not necessary to limit analytic
work to the simple cases and solve the complicated problems of machine
dynamics by experimental testing of actual equipment. Dynamic
representations of synchronous machines and systems using computing
:, "machines are feasible, and give a whole new degree of freedom to the
power system analyst.
,
The examples of computer representation of the synchronous machine
will be developed for an analog computer. It is just as possible to use the
fundamental equations of motion of the machine and system and solve
, them using digital equipment. The choice depends on the complexity of
the problem and the equipment available. The general methods of
system reduction presented herein are applicable in either case.
~

9.1

The Equations of Motion for Idealized Synchronous i,

Machines
':
I

il

The analysis of the three-phase synchronous machine will be dev610ped

utilizing the techniques established in Chap. 3 for the generalized two-phase
machine. In summary, those techniques are to establish parameter
matrices which define inductances and resistances as measured at each
set of terminals and then to derive equations of motion in terms of actual ,
coil variables utilizing Lagrangian or other techniques. These equations
of motion are transformed to new sets of variables which convert the.
equations of motion into the most useful form for the problem under
consideration. The two-phase transformations are given in Chap. 4
and are generalized for n phases in Chap. 10.
The physical model which will be used in the analysis is obtained from
Fig. 9-1 by assuming that the actual windings are representable by
sinusoidally distributed current sheets. The magnetic surfaces are
,assumed to be smooth (slot effects are neglected) except that the stator
has saliency which introduces a second-harmonic space variation in the
inductances. These are the same idealizations that were used with the
generalized two-phase machine in Chap. 3, the only difference being that

508

,i!

'Ill '"

"q .'
i~1 ,:

,!\i.r

:~.~

\,

i~.

'ii

:Ii

~;~'-~

in this chapter the rotor windings are three-phase instead of two-phase.

Thus, the general methods established in Chap. 3 for calculating the
parameter matrices and the general inductance parameters of Eqs. 3-31
through 3-36 become immediately applicable. Rather than repeat the
detailed calculations, the parameter matrices are written below by
inspection.

.!!
~

'6

.."

.!!!

'"

. ..
.. ...
II

";Q

t:i

'0

'0

-!3

"i6

5:. ""-I~'~

M"

~, .

n'

-9-1-

IJ

) I

:i~!i

"l-e,

11
t4-f~

'b.

I'
II
,I

0::

..

....
II

;: ....,,);

+" '{

-."

"I~

I,

'fh.'

::,'

.J"

.er-i ";ll,l'.

N
0

~<ln

Ii'

:t{;;

'L.e~/l"s.~

fV~ ":".

.ol~, I'
I ~~
.

'"....o

~(

" .,. ; , I

~~ ~ t"1I... .;:dKij

!1.t,'

where

0.

r{~bc]

'I)

~
~

r.

f"'-,

-: L~2 cos n(2'"

cos n(2</>

8
:i'!

-..;
..,

-'"

of

[lobe.ab]

(labe l':lUbC;'l1I1J

(9-2a)

-'

(9-2b)

ff

+ 41T'/3~n

.. _.

(\

""

'

';;

~f

::1 ;::!!j-,.

{I

os n(2r/J)

eo

1 --! -~
= L~ -t
l-t
-t -1
I

'

'5~

~[l~b] I [l;~'Ub~

[l~b] = L:~ TI ~~:, ~ -r"""'~,

..c:

""'.! t,"

ob,obc -

.a
u

::!.

~llil'

If.,r

"

'i'"

.S

",I j~I'

I, ,0,

....o
~

,"1

~"-~r
I

','

II

, '"

>,t

!!.,

", .

'1'

~
~

"

....
+Q~
'-

'"

--9+

"'~

""...c

~.

~1Jl
.~

oS

--0+

~I"

. l' tv,

J;l~

iI'

h~ ..

><:

! ,

(9-1)

R' 0 r 0 I
R 0
o 0 Rr

0.

'"

Io

.a

'i

lI
1,.,

_
ab,abc -

~.r

:E :E

~b!

g'a

";

..

II

':}

\1

,.........."
0

511

DYNAMICS OF SYNCHRONOUS MACHINES

ELECTROMECHANICAL ENERGY CONVERSlo'N

510

+
+

cos n(2.p
217/311) cos n(2</>
417/3n) cos n(2</

+
+

217(3n) I cos 11(2</)

417/3n) cos n(2r/
cos n(2>

+ 217/3n)
(9-2c)

:ttt-

<.' " [l~~.ak]

L8r fZos n(</ "t,>

1 cos n(r/> + 217/311)
I cos n(4) + 417/3n)
~s n(rp - .,,/2n) cos n(c{> - .,,/2n + 217/3n) cos n(4) - 17/2n + 417/3'Qj

:J:

= "

M
it

",~,

+ L"
"2

I
,'-.

"

cos n(</
cos n(</>

+ 17/2n)

cos n(>
n(c{>

r cos

+ 217/3n)
+ .,,/2n + 217/3n)

cos nCr/>
cos n(r/J

+ 4r./3n)
+ .,,/2n + 4.,,/311)

0, ",

(9-2d)
[l~te,ab]

= [/~~.abc]t

If)

(9-2e)

ELECTROMECHANICAL ENERGY CONVERSION

512

The equations of motion ill terms of the parameter matrices, the voltage
V~;.ube' and current i~~ul)(" ;lrc:

"

DYNAMICS OF SYNCHRONOUS MACHINES

513

Mcc!ranil'u/

.Jf
'/' -I-

'1'
f
,,"

.1
f('I'

~. fI [/"""/'
'.lIh. J ,..

I""'"
.".. ',,1'1

+ (I"
~dq

I")""]
'qd ',1 1'1

t9-41

where
Electrical

1J~:i~."J1('

(:~:',i:.u'l('

(1)-:'\1/)

+ P.Y/~\,i:,I~tl(')/~:i~."'h'

Mechanical

..
T = J</>

+ rt,j, _

(().!f/"I,'

nV'"fJb./lI,c ) t ~lIIJC
()t!J

(9-1h)

V"r
)
,~IJ.llbt

It
~

',.~'I.
.~.

:1

v~

~~

Ib

!".'.:

'.,"

Vb
1/),r

-'

Va

I.lh.ahc -

's r

(9-3c)

(ab.abe

I"

r
Vh

ij;

v~

Ie

I
~fi

.,

'"

,p :..

loads.
The general transformation from phase variables to dq variables is
given in Chap. iO for any number or phases. Since the three-phase
transformations arc particularly important, they arc taken from Chap. 10
and are listed specifically in graphical form in Fig. 9-2. Using the
transformations indicated by Fig. 9-2 and the general equations of motion
given by Eqs. 9-1 through 9-3 there results, after a transformation to dq
rotor variables, the following set of equations of motion:

,,1

.'

va

v'b
v'd =

R~

;I
~

:1

i
"!J.

'A

Electrical

;1
i

v~;.dqO = f?l'~~.dqOt'~~.(lqO

+ pL~
0
pL',fa

0
Ri,

+ pL~

n~L'Jr,

V'q

_n~Lrs
qll

pL~b

z>{)

-aIII.,

pL':d

pL';q

Rd + pL d

n~L"dq

i~1

lq'r

-n~L~d R~

+ pL~
0

R~

+ pLo

Ib

'r

10

(9-4a)

I,:' =

(L~

"M

+ 1':,1 ""
I.:rrlo + I.tl .,."
,c

/."
.. 1 ,.,
~bq ,~qb

= L'"db

L~,

- /'~1) ... L~I

L'"qlJ = \/~(L"
..
IA

I:'
"" L"
ltd
.Ju

V J(L"
2
I"

+ LIT)
""1
- L"r)
1"2

+ L~I = ~(L~ + L~2) + L~/

L'",q + L~I = 1(L: - L~2) + L~I
~'(L~ - L~1) + L~I = L'",q + L~I
~(L: + L:1) + L'dl ... L'",d + L~,
L:(1 -,2(';;) + L~I = L~l

L~ ... L'",d

L'o

'1-.'.'
", "

(I,;. + 1':'1) -t

L; =
L4
L;;d

(9-3d)

The equations of motion in terms of the phase (abc) variables are usually
not the most convenient form for the solution of dynamic problems. In
most instances the af3-dq variables are a more convenient set for handling
synchronous machine dynamics, particularly for the condition of balanced

~,~

.. r

..

where

I,~ ... I.:....

,R'd

i;,

= R;

Ro= R'

and vd, v~, v~, i~,

and ioare the direct, quadrature, and zero-sequence
variables as defined in Chaps. 10 and 4. Note that in defining the four
self-inductances L~, L b, L'd, and L~ the terms L~/' Lbl , and L~I have been
introduced. These terms were not in the idealized model of Eqs. 9-1
through 9-3. It is well known that in actual physical machines there are
slots, end turns, etc., which introduce in the self-inductance of each phase
additional inductances called leakage inductances which are not included
in the idealized model. When extending the equations of motion of
Eqs. 9-4a and b to describe practical machines, excellent correlation is
obtained if the leakage inductances are included. Since leakage inductances affect each phase independently, they can be added to the equations
of motion as if they were additional inductances in series with each pair
of terminals. This does not modify the torque equation and only affects
the volt-ampere equations through the self-inductances as shown. In
introducing the leakage inductances, the subscript I has been used to
denote leakage and a subscript rn has been introduced for the mutual
part of the self-inductance. This notation will be used throughout the
chapter. The equations of motion given by Eqs. 9-4a and b are for a
machine with a two-phase salient stator and a three-phase smooth rotor.
The rotor phases were chosen to be symmetrical, i.e., of equal resistance
and inductance per phase, which conforms with usual design practice in
synchronous machines.
The model to establish the equations of motion from Fig. 9-1 was
chosen to correspond to the generalized two-phase machine treated in

1
I

;r,

1
,4.

'rr
eJT eJT

eJT eJT

,4.

2.

I
I

--

-;;
(

~
I
0,

I
I

+,-

ein.p

Symmetrical
components

e-jnq,

is. b. c

r
r
r

/'1

.,

Rotating - field
components

~,

J
~

O,f,b

I
I

('

;:

V3

2,..

,4.

eJl eJT \--.

'rr
eJT eJT

",~

..

ii)
)<

Q>

=>

0-

3
Q>

<:2.

'.

=>

, ..

::..::

'"

.'

--l

C
2
<

"

;;t

C
z

I
:r

"2
"C;x

c;;

II>

eM

'.

r--

e 1n

I
I
I

i.}

I
I

.<"

..

~
I

..

4:.-

.'

""

(")

I
I
I

o
I

Y2

"0

:r

-.fi.

-.fi.

-.fi.

-j

72

---,

'c'-'

L ...... " " . "

J..

-.j2

-j

Yi

-j

,~

_.'~

,:.,f.;
4

__

2'
'-3

'\2
2

..[3

"'2

'\2

~
..i

.~

~,,j

..[3

''2 -2' -2

~'_'~.-

'.

r-LI

nq,

nq,

-SIO

cos

cos

sin

nep nq,

I
I

--

'~;

....

,~,it

:;~

"2
1'3

-.fi
2'

Yi

Two-phase real
components
O,a,{:J

-2 --;1'

~
~

OJ

0-

"tf

Cross - field
components
O,d,q

1
cos

-sin

sin

cos

nep
nq,

%-(8}-%'

[a]

xl =x']

nep

nep

Q>
g.

::>

II>

-<

}>

:1

--

"T

-<

r
:r

;;t

...;

,"'VI;

v
:3

):

C
:I

fiC' 9-2. Block diagram repr*ntation of senecal three-phase transformations.

'>,,,-'

::>

I
I
I

II>

I
I

-J-

.....-

-.f3
Y3
"'2 -2
Key

-~ -~

;u

'\2
2"

-j
----'.

<II

c;.b,c

1
".

I
I

~;-,

....

j
~

.<- -- - ' . - _. - - .- - -

~_._

_.
--12
-2-

~-'-.-""

"

;..~-

-"
"';$;

-~

0'

r
'~

'"
'"

.~

~.
',.;

VI

0-

.;;;'

1
~

1
1

a,b,c r - -

4..- . 2..-

e JT e JT ~
'2..-

..

.4..-

e-jnq,

()

-I

;;0

eJT eJT

m
,...

-+

0,

~
...

:I

e Jno

Symmetrical
components

+,-

Rotating- 'Ield

I
I

()

compon~nts

)-

O,f,b

I
I

n
)-

r
m

Z
m

I
~~

1
1

4..-

e J'1[- eJT ~
. 4".

jE.

eJT e 3

I
I

'.

\
~

J
I
j

-<

()

<

m
;;::

::!

""
<5

o'

"

or.

~I

"0

::r

~i

<

I'

2"

-/--- I - -.'= ,

C'l

'""

!
/

=l

iO
)<

...,..-"\..'

I
.'

;::;

-0

.......>'

0
0

!,

::r

re- jnQ

~
:?

,~

e jno

,~

"._.-

.1

-- t----J . i

f
.. -{.

."""'i"-'
I

~~
-v3

1
-2 -2

~
I
I
I

l~
..

a,b,cU

--:::,

~tl

-2"

-"2

-v3

_ii

..

I
I
I

.,

Two-phase real
components
0, cr, (J

:'~

!!.

:>
~

~.~

cos

-s~1

n<;>

I
I

sin

cos

rnt>

ncb

ncfJ

-<

!!.

)-

Cross .. field
compcnem3

'"

:>

o'

:I

V>

O,d,q

L4I
I

...

I
I
I
I

<l>

:::l

~~nJ ~ .

I
I

v.

x~x')[aJxJ=x'].

Key

1nm
'1

I nIP
sin
~

",;f

,~

lAlS

.f

li-l

_1

I
I
I
I
I
I
I

~_.J

~:

Fig. 9-3. Conventional transfonnations from three-phase system abc to Clarke's a/3 variables, Park's cU; variables, t Lyon's - - vap.:.bles.:
and K u's fb variables. These transfonnations were not used in this or previous chapters. The transformations of Fig. 9-2 have been u'!Ied
throughout.
.

-<

:;x:

o
z
o
c

v.
:I

>
r.
I

rr.
v.

E. Clarke, Circuit Aoolysis of A-C Power Systems, John Wiley, New York, 1943.

t R. H. Park, ibid.
t W. V. Lyon, Tra.nsient Analysis of Alternating Currellt A/achiller.'", Technology Press, Cambridge, and John \Viley, New )"Clrk:_ 150')4.
Y. H. Ku, "Transient Analysis of Rotating Machines and Stationary Networks by Means of Rotating Reference Frar-..e>.. Tr::JJU. AlEE, "V<
Vol. 70, PI. I, 1951, pp. 943-957.
.
":::
.--u

';1"

CL.c ....

DYNAMICS OF SYNCHRONOUS MACHINES

II\Ul"ltLHANICAL ENERGY CONVERSION

519

.j

'~

Chaps. 3 and 4 so that the results of those chapters could be used with a
minimum of confusion. This model does not conform, however, to the
vast amount of literature available in the field of synchronous machine
theory. The most significant differences are as follows:

..(.
~,'

1. The field is usually taken as being on the rotor.

2. The phase currents abc are usually defined as positive out of the
machine for generator action, i.e., the negative of the current direction
assumed in Fig. 9-1 for the three-phase currents on the rotor.
3. The conventional transformation to dq components from three-phase
variables as defined originally by Park* introduces a factor of t in
transforming from abc to dq variables, whereas the transformation used
in this chapter introduces a factor of v1. Thus the dq variables used in
this treatment differ by a factor of Vf from Park's dq variables. The
main advantage of this transformation which has the vlf for both the
direct and inverse transformations rather than the conventional method
which uses t for transforming from abc to dq and 1 for transforming from
dq to abc has to do with the power invariance of the transformations and
the form of the impedance matrix obtained. With the transformation
used in this chapter the mutual inductance L:;d = L'1a. giving symmetrical
coupling terms between the a phase (field circuit) on the stator and the
direct d axis circuit on the rotor. The transformation following Park
would establish that L~d = t(L'1a), giving unsymmetrical inductances
between the a phase (field circuit) and the direct-axis circuit of the rotor.
This lack of symmetry can be eliminated by the use of per-unit
impedances. The general discussion of the per-unit impedances in the
literature, taking into account unsymmetrical mutual inductances between
field and direct axis, is unnecessarily involved, and since it can be
eliminated entirely by a simple scale change in the dq variables this is,
in the authors' opinion, the preferred approach and has been the one
followed. The conventional transformations from abc to dq variables as'
used by Park are given in Fig. 9-3 for the convenience of those who want
to correlate the results of this chapter with the published literature on
synchronous machines.

9.2

The Synchronous Generator

The Rcnernl cqllofions or motion for u 211-1'011:, two-phase salient stator,

I hrl't'-phllsq SI1Hl\1th rotor, ~kctrio ml\chin~ Wtll'C lIcl'ivtlu In S~c.9. L
These cqlll\ti(~lls will now he speciali1.cd fM u synchronolls gcncrator and
R. II. Pur~, "Tw(l-RI'"~linn Th~or)l of l'ynQhronotl~ Mi1\lhlnl!ll." A.IEE '!'fflllJ,j
VIII. 411, Iny. Il. 116.

placed in a form which agrees with conventional synchronous machine

theory, except that the dq variables of Fig. 9-2 will he retained. The
physical configuration which will be considered is shown schematically
in Fig. 9-4 and consists of a three-phase statnr and a salient rotor which
has three windings-a field winding in 1he direct axis. a damper winding

. totlb----

-tJ,,_
+

i
L
..--

,J

i'

Fig. 9-4. The lhree-phase synchronous alternator with direct- and quadrature-axis
dampers. Sign conventions and rotation as conventionally adopted for synchronous
alternators.

in the direct axis, and a damper winding in the quadrature axis. The
directions of current now and shaft rotation which arc lIsually adopted in
. the study or synchronous alternators are indicated in Fig. 9-4.
The equations or motion for the machine in Fig. 9-4 can be derived
dircclly frolll Eqs. 1)-4(/ and h. The factors which must he considered are
Lhut saliency exists 011 the rotor and that the direction of rotation assumed
corrc~pond~ to rlltation in the -,I, direction in Fig. I)-I.
The current
direlJtiol\N ill phllNe~ {//I(' lire the negatives or those in Fig. 1).1. The two
windings in the dirccl axis, the field Hnd direct-axis damper winding,

<//=0

"!'>J.~ ~_p.

!;,,_.- ..t,''''r'

'-~';'e

.,.....

~N~~

'\:0'..

'\" i<t

\.

_.,..
,.{

. I<""'~"
.

....t.Ar

....

~.

'"

.. <.;...jt.\r.O,
'li~*

,.

~,

,. ,''-. t'- .. ~,

:").1<1"

ht~~,

"j

11
S10

Iwo windings in Jlhasl~ 1/ 01\ I hI' stallll' in h).'., ') I, The

interchanging 01' stalor :ll1d rotor pillS th\: olhn ch:lngvs illdil'all'd lIhow
are accomplished hy till' rollowing ",'ddinilions or ljllanliti\'s. (S,'\' I'ahh'
9-1.) Jn terms of the l:liangcs dcpil:lcd hy Tah!c \)-1 t hl~ l'qlla IiOllS
motion for thl' sym:hronolls f,'.\:ncrator in hI-'.. \},t aI',' wrilll'1I dirl:\:'il y
from E4S, \)-4<1 anu IJ in I 'IS, \).) II allli IJ,
'
corrcsf)ond

or

~..

Ii

10

J,

.p

"?~

.;;

-c/>

'/~

-'I'

,,'

"

I.

'.
'.

v',

'.

,,'0

Vb

"

"

,,'

'd
"

,I

"

Vo

"

'0

U,.d twk. to{

give two wind- i~
ings in the i~
direct axis

'S

Used twice

{ L~~
L~~
L'a

L~~

L'o

rs
Lrs
- L Qd
da-

= L~d + L~,
= L~a +

L~,

R~=R~=Ro=R'

Used twice
f>

R~
{ R'
a

R'b

-(,
- ill

1'.

- io

&/0

v:

= L~~ = L~:
L~~ = L~: = L~~
L~

v,

i ka

,.\',

-i,

i1c4

L"
ad

-ill
- ib

/ 11

v'd

{L:
LS
L'b

-4,
- '/~

i,

Va

Ib

Used twice

L:411ivulcnt QlI4111titics for hg, \)4

'f>

/.

l.j:-

j':'

', :u
I'-"-

.~.

: '. . . :

"

;"j .

-)iJ

!:

,d)r~\

...3'
~...3'

;,)~~

.s

"

cr:::

Vka
1

+
..}

~,

)i"

'"'" ...
~ ~

f'

'.f

I,:

.J

,.1.1

+-S-O

:s::

'"

l:::

L al

Lo

,.~>

Ra

'" 'of

~
~

R,

~
~

R kd
R ka

.....

The equations of motion, Eqs, 9-5a and "', describe the synchronous
machine in Fig, 9-4, Certain simplifications can be made by converting
to a per-unit system which establishes all the mutual reactances in any
axis (d or q) as equaL The proper choice of a per-unit system will make
L Jd , L kdd , L mJ , L",kd' L md all equal; similarly, L mkq , Lk'N' and L",q will be
equal. This corresponds in essence to a change in scaling or a change in
impedance level.

,t1,;,.,

g:

"'-l

+
._'
,,,:-."

II:::

~ ~

..J"'-lO

-S-

II:::

~-S-

cr:::'"

tl

:-e"")

...

...:l..... 0

II:::

;;;,...

;.,'" ;:,Q

,cr:::

"~'O;',

II

I . . '"
;.:.""

,I

II

...:l"-=i 0

;:,.... :::>'"

Il:P

"#
IN~ i'l

....

...

..:s
~

...:l" +

Eo-..

I
.-, ... ,.

~:

'),-,,'-'i,

...

.);

\."

..:s
~

"'-l '" 0
OO+-S-...:l

~I

.
r"

.......

.~

~ =ll
"'-l
0

"I

.1'" ,,~

r:c,"

~.

(tM)

.,

'" "'-l'" ...:l

~-S- ~
"'-l :;:: + 0
~I

t)

.:s

...

~4!J

VIc4

,-'"
,,,:,

~:r::-t-~--'~-~

1\/

v,

"1l':"
I

it.-.

r:<'O

,', ;,. - 'I ;

~.'-'ii

1\f!1

N,

It

00000+
'. !::{: ,;l.--Jf~~.

Vb

u"~ --~'I

'- '[ '[ ,'j

r-........--..-- ....;., .'

'"

-if,

L, = L m, + L n
L kd = L mkd + L kdI
L ka = L mka + L kal
L[d
L kdd
L kaa
L d = L md + L al
Lad = L md + L ar
La = L ma + L al
L da = L ma

--'

,. I ' ~;,-

I'J" ,

~
~

~'
';J\,"

.,;

in hg, 9-1

l i... '

J t',:\/

TABLE 9

QlIantities

511

DYNAMICS OF SYNCHRONOUS MACHINES

ELECTROMECHANICAL ENERGY CONVERSION

522'

9.2.1

~~

ELECTROMECHANICAL ENERGY CONVERSION

A Per-Unit System

ductances on the turns per phase and a permeance factor taking into
account the geometry of the machine can be determined.
The symbol for conductor density used in Eqs. 3-31 through 3-36 was
ZS for stator windings and Z' for rotor windings. To apply Eqs. 3-31
through 3;36 to the present case in terms of the configuration of Fig. 9-4,
define 7.[ = Z' and Za = Z' where the subscripts/and a refer to the field
and armature respectively. The dependence of the various inductances
on conductor densities using Eqs. 9-4 and 3-31 through 3-36 is

The per-unit method of representing parameters of a machine is

especially advantageous in the study of a large number of machines
because it provides a procedure for nondimensionaiizing the various
quantities, thereby introducing a simpler way of comparing the signal
and power properties of the machines. The per-unit method is particularly
useful in the development of equivalenl circuits and block diagram
representations since it brings into focus the true similarities and differences
among components by referring all quantities to a common set of base
quantities. There are many per-unit systems, but for machine systems
the per-unit system derived in this section proves to be particularly
convenient.
To illustrate the method, consider the case of a three-phase synchronous
machine and to avoid unnecessary algebraic complications ignore the
damper windings. The volt-ampere equations, in terms of two-axis
(d, q) quantities, are from Eq. 9-5a: .

VI = (R I
Vd

+ pLI)il + pLldid

= pLldil + (R"

vq = n<f,Lldii + n<f,Ldid + (R" + pLq)iq

L m , = (Z,)2Pd

(9-8a)

L md = 1'(Za)2 Pd

(9-8b)

L mq

= 1'(Za)2Pq

'r

(9-8c)

Lid = V~(ZfZa)Pd

(9-8d)

where Pd and Pq are permeance factors (i.e., geometric factors) for the d
and q axes respectively. With the mutual inductances defined in terms of
the field turns, armature turns, and the factors of l' or vf introduced by
the transformation from phase quantities to dq variables. it is possible to
establish a system of equations in which all the direct-axis mutual
inductances are equal and all the quadrature-ax.is mutual inductances are
equal. This is accomplished by the appropriate scaling of voltages,
currents, and inductances.
Using the relationships of Eqs. 9-8a-d as a guide, rewrite the VOlt-ampere
relationships of Eqs. 9-6a-c as follows:

(9-6a)

+ pLd)id - n<f,Liq

523

DYNAMiCS OF SYNCHRONOUS MACHINES

(9-6b)
(9-6c)

In this system of equations the mutual inductance between the field

and direct axis is Lid and the mutual components of the self-inductances
L f , L d, and L q have been defined as 9 I
,~

..

.,~,

= LI

...

L ml

- L ll ,"

(9-7a)

V
+

...

L md = Ld - L,,~ _

(9-7b)

if:!'

L mq .~ Lq - L al .~

(9-7c)

"

L)] (J23 ZaZf).If + (J3-:2 ZfZ" L[d) pi

Zf vf
(J:23Zu)
[2:3(Za)2(R
ZI
I + P f
.Vd -- (JJZ
(J23 ZoZI If. ) + (Ra + pLd)rd. - n<p(Lmq + Lar)rq
:2 Z/ L)
fd P
=

(9-9a)

(9-9b)

It:r

,"

The inductances Lll and L al are the leakage components of flux due to
slot~, end turns, and similar flux paths which exist in physical machines.
None of the mutual inductances L ml , L md , L mq , or Lid are necessarily equal
in the system of equations that has heen derived for the machine, Eqs.
,}-6. This ean be seen by referring to Eg. '}-4 and Table Y-I where the
above inductances are defined, and to Eqs. 3-31 through 3-36 where the
terms L;:, L;';, L;', L;:2' L;'2' L:~ LIre culculated fmlll fkld llUUn(itks. These
hasic definitions apply to this chapter since the same physical model was
chosen. Using these relationships. the dependence of the mutual in-

f~f'

(J~ ~; L fd )(J~ ~~ if)n<f, + n<f,(Lmd + L"I)id + (~~ + pLq)iq

(9-9c)

Clearly Eqs. 9-90-(' are identical to Eqs. 9-60-('. The ratios of conductor
densities and (he scale factors which have been introduced, when multiplied by the inductanccs as indicated in Eqs. 9-9a-c, arc such as to refer
all induClalll:es 10 lhe armatufe condudor dellsity Za. The seale factors
introduced also fefer the ficld voltages and currents to the level of the
voltages and currents in the armature circuit. The scale factors introduced

.'ii

:<"'\)

~~:
1f;
ELECTROMECHANICAL ENERGY CONVERSION

52..

into
are

Eqs. 9 I)u l'

and

which rder

all quantities to the armattlre circuit

(J 23 ZrZ,,)
i zJ)'
(if).rm. = (J "3 z"

(vr)arm,

(9-100)

1'/

Lr arm.

(9-IOc)

(9-10d)

'fd

't~i
'~~

~(Z,,)2L
2 Zr

f'"

ntI

~(Zu)2 Z2

, J

r',

JZ2
2

(9-11)

"P,1

X aO

~ao (ohms)
00

\1:\

in

(9-13(')

'

XII

(9- Dh)

= laO

= wo(Ld L q , L md , L nrq )]

...

(J2Z
3 Za

\)

-;=- ) -

r,,: XII

per unit [Rf ; X f = ,w-'L,/

0\
, L mf)] .rm., = rI'. X
f

>=

if

(9-13d)

/00
:

X 'h~

a
X
-a

C9~ 13e)

(J32 Z/
2 )2 Rr; XI
X
- -

-aO

(9-13f)

. ~,

per unit (X/d = WoLld)ar'm. = x/d =

(J~ ~;) ,~:

(9-13 g)

"I"

The per-unit electrical quantities expressed in terms of the armature base

quantities are given by Eqs. 9-13a-g. Note that in these equations all
per-unit resistances and reactances are written as lower case letters and
voltages, and currents are still written as lower case letters. The fact that
the method of writing the variables is not changed in the per-unit system
should cause no difficulty because a system of equations with lower case
r and x for resistance and reactance means that the equations are in terms
of per-unit quantities. The definitions of Eqs. 9-l3a-g establish a perunit system in which all direct-axis mutual inductances are equal and all
quadrature-axis mutual inductances are equal.
One problem which occurs in practice in trying to use Eqs. 9-13a-g for
finding per-unit quantities is that the conductor density ratio Zn/ZI is
usua!1y not known. It can, however. be determined by the simple
techniques of observing that, in the q axis, voltages n~Lf,tir and n~LJiti
are generated. For example, when balanced base armature current /uo

rated current (amperes peak)

woLao

(1)-13a)

-- (J~~II)+
2 Zi v

. C')
per umt
If arm. =

per unit [R,,:

<,~

a()

VaO = rated voltage (volts peak)

laO =

<........

',: I Vi

Thus the scalc factor ~(Z.JZr)l converts an inductance as measured in the

field circuit into an equivalent inductance referred to the armature
circuit. In addition, Eq. 9-11 is identical to Eq. 9-8b; thus scaling of
Eq. 9- IOe or 9- IOd is easi Iy shown to produce only two mutual inductances
L md or L mq , Eq. 9-8b or 9-8e, as measured in the armature circuit. The
scaling makes all direct-axis mutual inductances equal and all quadratureaxis mutual inductances equal. The aim of a single set of mutual
inductances for both field and armature is accomplished by the scaling
factors in Eqs. 9-lOa-d.
The next step is to establish a set of base quantities to use for defining
a per-unit system. It has been widely accepted that the base quantities
are determined by the rating of the machine with electrical variables
defined by armature ratings. The base quantities are.

l'

"a
= V'lli

IIO

pcI' unil i u

-;J.

2 Zr (~fPtl) -

per unit (vr)arrn.

d~

l~l lh~'

Eqs. 9-10 and 9-12 uS

per unit Vu

The effect of the scaling can be seen by considering anyone of the above
relationships. For example, consider Eq. 9-10e. Using Eq. 9-8a for
L mf and the scale factor t(Za/Zr)l given in Eq. 9-10e, there results
'l~

The pel'-ul\il llclJ quanlillcs Icl'clTcd

Ihe per-unit :lrmllt lire qlllllllitics urc uclerllli ned fro II I

lInd

\;,'

(J Z")l
32z/

bust' ekCfrir:t1 quanlities.

Hrn1alllr~~

(9-IOh)

)
3(Zu)2J
= [2
Z/ (Rr Lr

(L/d).rm, =

nrrllllturc qUlllltities lllH.! the ~llIlIll(ltICS I'cJ'cI'l'CU to (hc 1ll'tlIlllllre hy (h~'

If

(R

,J1

51',

DYNAMICS OF SYNCHRONOUS MACHINES

(9-12)

To = rated torque (newton-meters)

Wo = rated frequency (radians per second)

The per-unit electrical quantities are obtained by dividing the actual

,j':

526

ELECTROMECHANICAL ENERGY CONVERSION

flows in all three armature phases and the machine is not loaded (torque
angle S = 0), the d and q currents that result are:

v! laO

Ido =
I qo = 0

n~Ldldo = n~LdV! laO

(9-15)

n~L/dl/o = VqO = n~Ldvt 1.0

from which

(9-16)

LjdI,o = Liv1 l ao

The inductances L d and L,d are expressed in terms of the conductor

density ratio by Eqs. 9-8b and 9-8d. Substitution of these expressions
into Eq. 9-16 gives

"'\

v1 (Z/Za)PdI,O = t(ZD)2PiVtlao

~='Hoo
Zo

, I

"

(9-17)

I/o

where I/o has been defined as the field current which generates VqO
according to Eq. 9-16. Since the phase voltage is
Vqo ), it follows that
'IO i~ found fl'Om the open-circuit voltugc characterislic as that value of
field current which generates on the open-circuit air-gap line (see Fig. 9-7)
a phase voltage equal to the base mechanical speed times the armature
oircc\-l\xis illuuclUllce times the base vulue of armature' current.
i.e. II \'olll\ge of (wll/n)L dJa\!, Tlms the ratio ofcol1uuctor dcnsitles
nclwcrn field nnd nrmntllrc ill cosily del('I'mined' \Ising Eq. 917
Substituting Eq, 9-17 into Eqs. 9lJa-g estahlishes a per-llnit syslcmin
whid~ all direct-axis mlltllal inductances lire cqlllli and nil qUlIdrature-axis,
IllUlll~1 il\\IUClIlll~cS nrc lll}uaJ.

vt(

rh~ ht\~~ illlpcdt\l\~c,

\.lIlTC'nl.

hlIM:'

Thus from Eq. 9-13b

vi VaolaO

basc field voltage V/0 =

(9-18)

/0

The technique of deriving a per-unit system for the alternator has been
dcveloped. A per-unit system for the dampers is derived in the same
manner; but. since in practice these windings are short-circuited and their
effect can only be determined experimentally by tests on the armature
circuit, it fo))ows that experimentally (1~tcrmined parameters are already
expressed in terms of the armature circuit base quantities. Consequently,
they wi)) satisfy the desired condition of having the same mutual inductances
as for quantities referred to the armature. The per-unit system developed
here will be used in the rest of the chapter.

rrl'lIt'lhly,

bnsc
IIlld

1\1'111t\ltll'c

11lIso

voltage

11ll'ljUd

lInticlIl'I'Cl1t. bn~c

hllve belln established.

field
The

.,,.$

,r

:1"
$!I

9.2.2 The Per-Unit Equations of Motion

'ii~

The per-unit system derived in Sec. 9.2.1 simplifies the equations of

motion of Eqs. 9-5a and b. A second simplification results by restricting
the alternator to be connected ungrounded wye, which is the connection
commonly used in practice and which forces the zero-sequence component!
io to zero (io = 0, ungrounded wye). A further convenience is to define ~
the actual frequency corresponding to the speed in electrical radians per
second as w, where w = n~, and define the base frequency or speed as ji
wo0 In terms of these changes and in terms of flux linkages, Eqs. 9-501,
and b become
,11.

,EleCtrical

:uti

L~~'~1

r"

k:

Vf=

t;:t. '

:~'f;i

rIll

I'hd

= 0 =

Vkq

+ pt/J,

l~' (_:,::'it;t ..~

rlcdilc,l

rkikq

+ Pt/Jkq :,P.

fJ'Pkd

I'd "-'

- ",,;d

-I-

I"/Id

,~}rlJ

(O'/Iq

+ p'/'q +. wY/II

ill

(9-19a) II
(9-19b) ..

J~

(9-19c)
~";;

V q "-' - r(,iq

id,j l

:.;",.~

("

Jp -

"

lJ)

a-

"

(9-19d)

,r.

(9-1ge)

fur,

Al('(""(11/ i( 'Ill

I'l'i..

c,

,.1 1

tJ;

H.X;

.~

t'HS\, (i1'11\ \'ollll~~ 11111~t nlso he known. This. h()WCV~I', rol1uw~ Jlrc.:tly
('I'om F'\. 1).),1i1 I1\lW I hilt /.,1/) is c.'slnhlishcd. The hnsc vullle of field
t:

~t>

f;,~};"

'\'.'

/ .

527

-"

If next the armature is open-circuited and the field is excited, a voltage

will be induced in the quadrature axis. Define I/o as the field current
which generates the VqO of Eq. 9-15; i.e.,

or

~i

current is onc per-unit field current by definition.

and using Eq. 9-17 for 2,/20 we have

(9-14)

With the field open-circuited (i/ = 0):

VqO

DYNAMICS OF SYNCHRONOUS MACHINES

... ' ...

I .)
+ n{!jI"I'I-"ql"

(9-19f)

~,

'#,Ii'

!.,,-.
!':'t;..",

, -'1,"'"

l'~

11

52.

DYNAMICS OF SYNCHRONOUS MACHINES

""

ELECTROME<...HANICAL ENERGY CONVERSION

9.3

In the above expressions the several nux linkages arc given as

.-=

-.![(X",d + x//)i/
Wo

+ "'d/

-.![xmd i,
Wo

+ iflq/

~ [Wo

"'md

+ ifll /

ifld = iflmd

ifl,

iflq

iflkd

iflmq

(x mq

= iflmd + iflkd/

-.![xmd i/
Wo

+ .pkq/

-..!..- [ -

.pkq = "'mq

.I.

't'md = -

Wo

.pmq

Wo

(Xmd

x,,/)iq

Xmdid

xmqiq

(. .
. )
Id + Ikd

iq

xmdikdJ

(<J-19 g)

xu/)id

+ xmdikd]

(9-19h)

+ xmqikq ]

(9-19i)

+ (xmd + Xkd/hd]

(9-19)

+ (x l7lq +

(9:19/c)

xkq/)ikq ]

x m/

Xmkd

xld

....

;,

',Vt'1

.~, ~l4f'
.',to

x mq =

Xmkq

Xkqq =

X/kd

f
i..

l:
..

I
'}

= Xkdd
(9-19n)

per-unit quadrature-axis mutual reactance

(9-190)

X/I
Xul

.....J

f1;

Xkdl
Xkq/

=
=
=
=

x/ -

Xmd

Xd -

X l7l d = x q -

(9-19p)
(9-I9q)

x mq
~i"

"t'

,';j

Xkd -

Xmd

(9-19r)

Xkq -

x mq

(9-19s)

The armature leakage reactance Xul (Eq. 9-19q) is assumed to be the same
in the direct and quadrature axes because the armature leakage reactance
depends mostly on slot etfects and end turn reactances which are not.
strongly affected by the saliency. This assumption agrees well with
experimental results.
The per-unit equations of motion, Eqs. 9-19a-f, can now be used to
obtain computer representations for the three-phase salient-pole
alternator.

Computer Representation of the Synchronous GCIlCI,;}tor

for Volt;)gc Rcgllbclon Problems

The voltage regulation of synchronous generators 'is an important

problem in power gcncration studies. Synchronous generator voltage
regulatiol' has been extensively studied, but most ot' the puhlished work
the maximum voltage drop in
has hccn confined 10 the determination
altcrnator terminal voltage which follows the suddcn applil.:ation tlf a
large load. Hanlert and .lamest have used analog computation to solve
certain aspects of the voltage regulation problem, but their idealized
representations of the alternator do not portray with sufficient completeness the transient characteristics of the machine as encountered in actual
system operation. The MIT Energy Conversion Group, and in particular
Riaz have made extensive studies of the voltage regulation problem of
aircraft alternators, and their results will be the ones used and presented.
It is the object of this section to develop a reasonably complete, yet
simple, block diagram representation of the synchronous generator that
is applicable to voltage and frequency regulation system problems
including parallel operation of several generators. This representation,
which will be suitable for use as an analog computer representation, will
account for variable-speed operation, saturation effects, variable load power
factor, and the effects of damper windings (amortisseurs) and saliency..
The general equations of motion, neglecting saturation, are given by
Eqs. 9-19a-[ Using the techniques of Sec. 2.3, these equations can be
represented in the block diagram of Fig. 9-5 in which the operator p occurs
only in the form of an integration lip. This is essential for use with analog
computers (see Sec. 2.3). The dotted path and nonlinear factor shown
in Fig. 9-5 is included to account for saturation and will be discussed in
I
Sec. 9.3.2.

or

The leakage reactances, which are indicated by the subscript I, are the
differences between self- and mutual reactances and are defined as follows:
'-'

0'

.~

rt

per-unit direct-axis mutual reactance

(9-19m)

i kq )

The above equations are in per-unit form with all direct-axis mutual
reactances equal and all quadrature-axis mutual reactances equal. The
mutual reactances, denoted by the subscript m, are related as follows:
Xmd

i1.,

(9-19/)

~~

Xmd 1/ -

= ...!..x"",( Wo

Xmdid

519

:l
~l.

I
"

..

See, e.g., E. L. Harder and R. C. Cheek, "Regulation of A-C Generators with

Suddenly Applied Loads," I-AlEE Trans., Vol. 63, 1';';4, pp. 310-318; II-AlEE
Trans., Vol. 69. Part I, 1950, pp. 395-406. H. C. Anderson, "Voltage Variation of
Suddenly Loaded Generators," Gen. Elec. Rev., Vol. 48, 1945. pp. 25-33. C. Concordia, Synchronous Machines: Theory and Performance, John Wiley, New York. 1951.
t E. L. Harder. "Solution of the General Voltage Regulator PrOblem by Electrical
Analogy," AlEE Trans., Vol. 66, 1947, pp. 815-825.
t H. B. James. "The Use of an Analogue Computer to Optimize the Transient
Response of an Aircraft-Type Generator-Regulator System." AlEE Trans., Vol. 72,
Part II, 1953, pp. 363-368.
M. Riaz, "Dynamics of Direct Current Machines' and Synchronous Machines in
Regulating Systems." WADe Tech. Note 56-419, Servomechanisms Laboratory, MIT,
Cambridge, Apr. 1956. M. Riaz, .. Analogue Computer Representations of Synchronous
Generators in Volt'lge Regulation Studies.... AIEE Trans., Vol. 75, Part III, Dec. 1956,
pp. 1178-1184.

530

DYNAMICS OF SYNCHRONOUS MACHINES

ELECTROMECHANICAL ENERGY CONVERSION

9.3.1
;,

)]

]
;"1
.::8

l.

.;'+]
~

]
,,"~

.~

15

""0

~~

.::

",-

.'

.~

i.

..

531

The Phase Variables

When a synchronous generator is applied in a voltage regulation system

the quantity normally regulated is the phase voltage. The block diagram
of Fig. 9-5 involves only the d and q voltages; thus, if Fig. 9-5 is to be
suitable for voltage regulation studies the phase voltage must be obtained
as a function of Ud and uq In this treatment only balanced operation is
, considered; therefore. only one phase voltage needs to be obtained.
Reference to the transformations of Fig. 9-2. taking into account the
reversal of rotation given in Table 9-1. shows that the a phase voltage is
given in terms of Vd and llq as:

,I.
'...
.: .!
,(.

Va

'~ \

vj- (Vd cos n</> - vq sin n</

,,,,,,'~

The angle n is given by

,;
0

(,.,,It . -~J

-;;;

...c

n =

(9-20)

+ ",..~.'JO'

f wet) dt + oCt) .. \i}w,I

n

'

(9-21)

...
d

When the time variations in wand 0 occur slowly compared to the base
period 2-rr/wo. the frequency wand angle 0 can be assumed constant during
one period 2-rr/w and Eq. 9-20 can be rewritten in the approximate form

11 '

l-l

J-, ,'Y, j,

c:
'co..."

":'i'

:.0

:'l'
,-::.'

~.~.

OJ

.loI:

.2i=Q

313' .~

'i
t>O

ii:

"
:I
, ,

........

:
~~

..

..

Va =

vi [Vd cos (wt + ~) -

vq sin

(wt + ~)]

(9-22)

This approximation is quite accurate for most voltage regulation systems

containing rotating machines. From the approximate expression of
Eq. 9-22 an envelope voltage V, can be defined as the per-unit amplitude
of the phase voltage and is given by
1....,(1\ !

.~

V/
~-.q

= IVai = vS(v~ + v;>

(9-23)

In a similar manner, the per-unit amplitude of a phase current can be

obtained as
.:, ':1

. = 1'1
'n = ...h.(,2
J Id +

1/

.2)

I"

'f

(9-24)

'0

'g

~Ij

~1

.H.t,,"!
'I

Although Eqs. ()-2J and 9-24 are written in terms of a phase variables,
they apply equally well to the other two phases in a machine operating
with a balanccd load.
Equations 9-23 and 9-24 can he included in the generator representation

".
~"

1';'.-

\;1

'iJ~'~.

...

"..;,., "

''.i.4i~

S32

ELECTROMECHANICAL ENERGY CONVERSION

DYNAMICS OF SYNCHRONOUS MACHINES

reactancc'" 10 dcll'l"lnine' a satllrntinn clIrve'. This nH,tlll.d .slill lIlTllllllh

for saturation ill lwth axes. In tlw present Ircatll\t'1I1 t\l'llr Ill' tl\\',r
IlH:lh(lds will he fllllowed: inslt'ad a 1110re apprOxil1l11tc hut ..till II"C I'll I
techniquc will he WiCtl.
In thc unllature. lhe resultant nux dictates the saturatioll level alld
altt:rs the relationship bclween this tlux and the (;orresponuing IlIagnc(ll'
molive rOr(~cs. Oil tilt' olhn hund, in th~ poll's. the saturation is Plilll.\lily
due to the d-axis component of mutual flux and the field leakage flu:,.
This affects the relationship between the d axis mutual nux and the
d axis magneto motive force. The representation of all of these c1rects
in both armature and poles would require the use of two nonlinear function
generators, one or which would depend upon two variables. Information
in addition to the presently available open-circuit characteristic would be
needed to specify these two nonlinearities. A rather crude yet satisfactory
approximation involving only one nonlinearity is based on the assumption
that all of the leakage reactanc.es and the mutual reactance x mq are imkpendent of saturation, since they correspond to paths mostly in air. Only the
direct-axis mutual reactance Xmd is considered to vary with flux level.
Thus only one nonlinearity is required in the block diagram representation
as shown by the dotted path of nonlinearity in Fig. 9-5.
The saturation nonlinearity to account for the variation of Xmd with
flux level is derived in the following manner. When there is no load on
the alternator, the terminal voltage VI is given by

by adding the hlock diagram of Fi~. 9-6 to thaI of Fil!.. IJ-:'i. When
this is done the generator block diagram provides the terminal variables
that arc usually mcasured am! controllcd in voltage regulation studies,.. ,

"d
__

"
Iq

Fig. 9-6.

9.3.2

()

{l~,1+ v.f)

----:1 yltiJ+ ':J

I- -.v,
I

;33

d.

Block diagrams to obtain terminal (phase) variables from dq variables.

Representation ofthe Saturation Nonlinearity of the Alternator

The representation of the alternator is made much more accurate by

including in some manner the effects of saturation in the iron. Strictly
speaking, the presence of saturation nonlinearity in a machine invalidates
the use of two-axis quantities, which is essentially based upon the
application of the superposition principle. Saturation in one axis affects
saturation in the other, and vice versa, so that, in effect, coupling terms
between the two axes are introduced by saturation. The coupling due
to saturation is seldom considered in machine studies and will be neglected
here.
It is customary, when dealing with saturated machines, to simplify the
problem by adjusting the values of the reactances along the two axes by
means of saturation factors. Various more or less accurate nu:thods
have been proposed to introduce these correction factors. A more
accurate method is to consider the saturation to be determ' '-'~d by the
voltage behind armature leakage reactance, i.e., in the voltage generated
by the total mutual air-gap flux. For application of this method, saturation curves taken for the machine while operating under various degrees
of loading arc necessary. Such load saturation curves are not usually
available. An approximate method which retains the advantage of
treating the total air-gap flux but does not require load saturation curves
is to use the open-circuit saturation curve and the voltage behind Potier

(V,)no

load

vivq = Vf w o.p1lfil

(9-25)

In this case the only current that flows to set up a flux is the field current
if. Consequently, to account for saturation Xmd can be written as a
function of if to yield the result
(Vt)no
;1~~

load =

V"fwO.pmd == vtXmd(if) i,

(9-26)

When the generator is loaded, the mutual flux linkage is given by Eq.

9-19/;

= xmAIi)[i, - i d + i kd )

(9-27)

assumed that under load the relation

i{ - i d + i kd is the same as that between Xmd
circuit condition of Eq. 9-26, as shown graphically
In Fig. 9-7a the open-circuit voltage is plotted as

between x md and
and i{ in the openin Fig. 9-7a and b.
a function of field

WO.pmd
It

is

"Li

For a discussion of Potier reactance see any standard text on electric machines,
e.g., R. R. Lawrence. Principals of Allemming-eurnm, Machinery, 4th cu., McGrawHill. New York, 1953.

S. B. Crary, Power Syslem Slabilily. Vols. I and II. John Wiley, New York, 1945
and 1947. D. I L\llIJi-Sepcn, .. Saturation Effects in Synchronous Machines," AlEE
Trans. Vol. 73, Part Ill-B, 1954, pp. 1349-1353.

.'

.
j

_ - - - - -

----,
j

DYNAMICS OF SYNCHRONOUS MACHINl:::l

ELECTROMECHANICAL ENERGY CONVERSION

~H

53~

which allows Eq. 9-28 to be rewritten in the form

current. In Fig. 9-7b the open-circuit characteristic is replotted under

the assumption for loaded operation discussed above. In Fig. 9-7 b the
functional relationship can be written in terms of the departure J(WOtfmd)
from a linear relationship by expressing the sum of currents as

(9-30) .

x mdo[2:i - isJ = WOtfrttd

The saturation effect described by Eqs. 9-29 and 9-30 is included in the
block diagram of Fig. 9-5 by the portion shown in dotted lines.
This method for includi ng saturation seems rather crude; however, it
does give good results in the study of voltage regulation problems.

(9-28)

.
9.3.3
i 8 = !(WOtfmd)

t'

~ (,(Air-gap line

+~/

r~

11/
$Ill>

<;j (

VI

(at no load
v,

e;

,oj'

~(

Per-unit
terminal
voltage

Representation of the Alternator Load

The equations and block diagrams developed to this point represent

the internal behavior of the alternator as seen from its electrical and
mechanical terminals. Consideration must now be given to the representation of the alternator load. For use with Fig. 9-5 this means the
calculation of Ud and uq given id and iq
Consider, for example, a balanced inductive load of per-unit resistance
rL ohms per phase and per-unit reactance XL ohms per phase. The
per-unit load v'olt-ampere relations can then be written as:

J~

= --If Wo Vtmd)

'I

iil

r._,
.. ' J

Va = rL1a

{u~

'jt\

.,'\

!,:,,',O'

';i

Vb

t\:;
f; :'>'

;otS'

f(hJnt/.I",,,)
1;1

, .

Iii)

"
L;
(lIn

,... r~I"" v....,'

U}

H
,

','

"'"h

.. ~ .-: l;

. ~ 'i<' 'It

~ 10,,1 -' ..

n:;"0.ffl.f,.~~

"

r"

,f1rifWI
\11}~~~

.-

;';'

'.~

~''''-

rL'e

,.

.,

(9-31)

Wo

1 De

XI- -

p .

Wo

Ie

;li,lJ

:,;'j

ho'

lAI

II" ,= ',.I". +

fJ '
XL 1"0
'" -

(u

X, fIJI)
'.... ~., J

(9-32)

I'j-";i'"

fJ .

''1
~ ~'

r l /,/ 1- x,. --- '" 1- x/

'''0

til

- I"
(,iO

V-,nl(

lillll,

11I",IIl,lIuII lIr slIlllnllloll lIpproxllllllll,1I1 1I,,:d III hille'''' dlllHrll'l" rcprescnlll(II) Ol'lln-~ir\l\lit\lhi\rl\QI~rilitiQ~ (h) HllpIOI11'I\ Op(ll)-rln'lIll dUHi'\'I~ri~lk. (I') I,

v\'nUI!'I

"'uV""'"

riC 9-7.

~'
I

' ... ,

"Ii"

f e'

For inclusion in the block diagram of Fig. 9-5 the load equations must be
expressed in the same reference frame (dq coordinates) as the armature
equal ions. (ising the transformations of Fig. 9-2 and taking into account
the rt~versals of signs given in Tahle' 9-1. the rer-unit load volt-ampere
equal ions for tic( variahles are

'.1;,,~i!

hJ O

:i

rL'b

<v1

. + XL -p.'b

/'

,(,

~j1f!p.~,

ti

,~

"

(a)

if
Per-unit field current

.. ,

,f

Wo

~~ ~

:\

. + XL -p.la

;(l.'H)"

;~

j.

;rhc zero-scq IIcnee term has heen omitted hecause ill = 0 has been specified
already.
For most voltage regulation problems the transients due to pia and pi.

f,.

ELECTROMECIIANICAL ENmr.Y CONVFI~SION

536

DYNAMICS OF SYNCHRONOUS MACHINES

Sl7

hllve Ill'}.digihlt' dft'cfJIlll fhl; syskm dlllflll'tl'ri,til1S. . Thus lIlt'S(' 11'1"11'. lin>

~Gl-_t{~) ----.

;iI
fron, .111

C'lilllillilfl:d IIlld Fq'l.l)_,l) l\f~ wrifll.'n in Ih(' IIPl'nnilllilll" I'MI\I

...

"tJ

III all
:d

>~,

\ j.

,~i.';

(;iij

XI.

/I., "-

..,

I'Ll., - ,\' l. ru
- i.,

I
IJ" "'" ""1

,i>~

(1)-.1.1)

'" .

+ XI.-I"
. I"ll

The expressions or Eqs. 9-:n can be rut in block diagram form us shown.,
in Fig. 9-8 for addition to the block diagram of Fig. 9-5.
Equations describing several types of loads are summarized in Table 9-2.

from drive

,t

~J

.__..1,,,,]

9.3.4

:;~~

iq

from all.

Fig. 9-8.

Vq

Approximate: representation of balanced inductive load.

'+,.,......

"""'-....

TABLE 9-2.

Equations (or the Various Loads'

Type of Load

Equations

--------------------~
Series resistance and inductance
.
XL.
Xl.
(a)

rL

XL

per-unit n:sistance
.. d
.

= per-unIt

In

uctlVC reactance

V4

IdrL

+-

Pld -

Wo

= -

Wo

Shunt resistance and capacitance

(b)

1',

X,

= per un~t resist~~ce

= per-unit capacItive reactance

Vd

V.

Wo

= -

V -

PXL

WOrt.
- - Id

pxl.

wOrL.

-- I

(d)

Open circuit

(e) Short circuit

,j

Id

WaX, .
=- Id + WOV,
-- - WaX,
- - Vd

WOX,

pi,

Vd

rLid

V.

rLi,

let I' ~
let

1~t1

'~ -

W .
I,

+-

PXL ' .

Wo
-Vd -

,J

Resistance
(c) rL = per-unit resistance

'1.1

Wo

Wo

Vd -

1. Armature resistance r a is neglected. This assumption, which is

well justified for most practical machines, makes the electromechanical
power conversion zero so the. problem becomes purely electrical.
2. The effects of the damper windings are neglected. This is justified
because the dampers have the greatest effect during transients in which
the torque angle changes, which requires transients in real power outpuL

'jtf r~: h

XL_
+ -Xl..
pl. + - Wid

PXL

...if'

----" WI,

Wo

V, = l.r L

14

1,'

When the voltage regulation properties of an alternator are of interest,

a particularly illuminating study can be made with a zero power-factor
load. Such a load leads to <l;n extremely simple block diagram for the
combination of alternator and load when some well-justified further
'
assumptions are made:

Iii

~-1!'

.tGl----~
to alt.

Zero Power-Factor Load

00

r~O

in (c)
in (c)

pl'.

WoX,
--

PI','

.~

I
:

'\:".

'1:.. "

:;

J!L from drive

Wo

Fig. 9-9. Approximate block diagram for zero-power-factor load on alternator.

538

ELECTROMECHANICAL ENERGY CONVERSION

DYNAMICS OF SYNCHRONOUS MACHINES

3. Transient terms due to the Pif;d and pif;q terms in Eqs. 9-19d and e
are neglected. These terms give unidirectional armature transients which
tcnd to cancel with rcspcct to the average of the three-phase voltages.
which is usually the variable of interest in voltage regulation studies.
The three assumptions given above force Vd and i q to be zero. When
these assumptions are applied to Figs. 9-5, 9-6, and 9-8 the simple block
diagram of Fig. 9-9 results.
Consideration of this special case for the study of voltage regulating
systems offers a simple visualization of the transient performance of the
alternator. In fact, the design of the regulating system on the basis of
zero-power-factor loading is greatly simplified, though leading to somewhat conservative results when compared with the conventional transient
specifications on such systems.

9.4

"n1'l1'~

~<",,,,

. ~,

~~

~ ~I.

~o

,~

t1

>~

~]I

I~~j
1;i~1l

'3-':

>~

2!

0::

..;
';}~:Il

~
>.

'"00

Jj

.* ~

.5

DynamiC Representation of Parallel-Connected

Alternators

i~~
fg 8

,!Il

The evaluation of the dynamic performance of two alternators consisting of separate drives, separate voltage regulators, and a common load is
a classic problem of power system analysis. Considerable literature on
the parallel operation problem is available. In most parallel operation
studies the alternators are greatly simplified and point by point steady-state
methods of calculation arc lIsed to study the dynamic performance. This
approach makes it difficult to consider the effect of a fast governor or a
fast voltage ret;nllator on the system performance. With the ready
availability of modern computing equipment it is more feasible to represent
alternators, drives, regulators, and loads so that accurate dynamic studies
C;l1l be performed.
III this section the block diagram representation of' an alternator that
was devd()lX~d previollsly and which is given in Fig. 9-5 will be lIsed ill the
dl:Vellll11llClll or II block. diagram representation ("or a cOll\pkle system of'
I'Dr:i1kl'((I\\I\cctcd allcrnallll's. The system IU;I\.I will he restrictcd (0 a

539

-;;j

...c:....
...

'.

00

{~

'?
o:l

,--,.....~.-

];

-;;;

!tJ)j.

....~0
c

.Sl

l.

-;;j

'Ur

, I . (~

'E

C'"O ;z

~ ~~~

~. :~

-,..,.;,,,, .

tl

8
<:>

-il'

0fil

---(0)

r~

lIO

-,

i.i:

...~

":II:111('('d load allhollp.h ;III condilions or U. I., alld (' '''ads illdlldil1jJ. opell-

alld shMh'irclIit conditions nrc nllowed,

<'

Tlte basic system which will be considercd is shown schematically in

hg. 9-10 lind a block diagl'llm or the systcm sltowing Ihe pertinent
vnll:llJ!cs nnd C\IIl\I'III\Cllts ill the inlerc()nnccled .syslcm as lIsed ror the
hl'ld. dillglllll\ I'cprcscntlliioll is shown ill l-'ig. \)-11.
+ ~;",'. f\11 il\~tl1"I:C, {'1IIl'Y.

J;.j

l.\

~.

('\1l1conlill. /"". ,'/1.: Ie 1{lId('lIhC'rl>t, '/hlll.rlrlll

/""("/1"."",1' "f "(1'"".". ("'11"1""\:'',\''1'''''\'' I\k(!l<ll\' Ilili. Nnv \' .. ,~, 1')\0 . 11111 Y. II. 1-;11,

~-,~'

.. H,)1,,1 ing-l'I"ldTIll'llry lind ( 1,'\\('1'111 !\ 1\1I1ysis nr SYlI,'llI'llnllllS 111111 1nil 11<" 1ion MlldliI\Q_,"
.ttllmIrlIU.. ,1:. (1.0111101\). Vol. 1)<), 1'111'1 IV. 1'I~2, pp. "Ill 42l!,

'!.:(;t
j-'

i:rt~

.l.

,",__ :h
t.,.

;1.
,Jf

,
.riM,_

>

".

...-- .

tl

C :' .3

o"~~

:'.

"., / I Ij

:i'

1",' ..II.,

n'~

- - - . : I1il1

'i'

'111r,

"\" tt\]
"'{ ,..". .,'J,

. .. '-' ...
,nr>Jo' 1,
1.~l""",

1;

hvn'
~.

j'
,

(~)

540

ELECTROMECHANICAL ENERGY CONVERSION

~'

r
I

I
I
I

'"

-r------'OE

...
3

..;P'

0"''''

.- >''''0

U"'.,

tI

_~l -1=----.:~~=

o
.-

,~T~}

'g

U~1V
~

O>c

.,.,
"

.:,'
',!
..

VI

II)

I\I.U
-

U
0>

o d;o

Ecoc

.5
...,""

1.'1

... J4 .... 'IJ

III lU
..
C_
~

I~'

0'

\~

"0
.~

-j- - ~
- - - - -~,

-0

--'

~ ~I
~2~

v""~

:;
~

I--

lJ1j . ~
o

hill'
"o~

~O~V1

~ v~

"

E
0>

~~+
I

..

t-.

;'0 L'"-t~1 .;'

~
~

. ',-

v~

~'-

V CI

VCI

:];

Vdl =
~ ~

vq2

..

(9-35)

sin

+ vql sin 012

12 + vql cos 012

(9-36)
(9-37)

+
+

i~ =

iLl

iLl

i CI

+ iLl
+ iJ.,.

(9-38)

Of course, balanced operation is assumed. Application of the dq transformations of Fig. 9-2 with the same assumption about the constancy of
W2, and 8 12 used above leads to the results in terms of dq currents,

Q1

w"

I
~,

0-

::;
~

'.
u

'" .,
~

'::0

'> ~

cos 012

-V,/I

i CI

I ;,1,

<1>0
~: .=:: ::J.
OU)

Vdl

i b1

"n:

The current constraints for parallel operation with two loads as shown
in Fig. 9-11 are
i al + i az = iLl + i L2

u::

',~ }

:5'

With the parallel-

These transformations are based on the assumption that WI> W2, and 3
12
do not change appreciably during the period of one cycle of the base
frequency 27T/W{}.

'"

l.!l

(I> - 34")

til)

'q

and using the dq quantities of machine I as reference, the dq transformations of Fig, 9-2 applied to both machines yield

.,;ti

c -

"'I
-

Ol)

.~
"0

q:

'0

'0

~
0>

==

~4:

.S

-j:
fF ~./

(wWo

"n,

IJ,.,

:;-

fJ

Vb l

::l

>

0'

'"
.....
<112
-- -Wo

Wo
whcrc Wo is the busc frcyucllcy for the system.
connection constraints
.

.~

,jf
-If

q:

l
c

_"f.::!!
:-J ." nl

~u.;:.~
c :;:) Q., ~
-

tlI)

"

_- - - - - - - ,

o
n;

Pn, .,1I1l1 0pcroLlolI

IIIC

_..

S"I

WllclI lwo or 1lI0rl' alternator.s lire cnlllleded ill parallel the cnllslraillts
tliut tile (CllllillUI voIIU!.'.lS IIc Ilie SUIIIl' alld llial tlil' Slllll 01' tlie I<"rlllillUl
rllI'lCIII:1 IIlllsl l'qllnl I Ill' load ":lIllI:'1I1.
l'Iu:'sc l'llllslmillls apply III I'haSl'
V()ltagl~s IIl1d clIrn:llts; thllS, hel'ore they CIIII be ilpplied to the pHlullel
conlH:ction 01" two 1I1l1'nwlors liS I'cprcsrl\ll'l! ill I,'ijl.. ,),,5 lhe l"'11Slnl1l1(S
111 \1st he expr,'ss,'d in tnl1lS 01" tlfl vllrinhJ(,Il,
This Inlllsl"nl'lll"liOIl or
conslraints is done hy the hlock tillcd "Tmnsl"ormalion of 1I1krnlllor
rolor rcl"erences" in Fig. 9-11.
('(lmidcr Ihc CUSe (II' twu, pllrallcJ-C\lIlIICClell llllel'lllllol's lIl' SI Il.lIulel!
ultcl'llll(or I lIlIU allernator[ 2, NUllIerical subscripls witt retcr 10 the
1I1ll'rnator nllmher. In gelleral, the two ultcfIllIl"rs will be nllll'lill~ lit
/dilrerent speeds '''I and ,oJ!. The instanlaneous dilrl."I'('nn' 1)('1\1'1',11 tl\c
,tllfqUC angles \II" the Iwo machines expressed in electrical radians is

...

I"

-tIG

9.-'1.1

'1

~
c

+~ ...

~ ~.Q

VI

~l .!.--- ... --.-L

~--_......-_--

c~

."

I
,.J

..
co>

.;.:} ..:Y

. .+

g'~

~~

;:'

~I

DYNAMICS Of SYNCHRONOUS MAOIINES

.. ,

~.!!

,~

""~~

f, ~~-~n
,"J

of

~ ~~
S!

'j,

i'h - idLI = - idz = -(jel2 cos 812 - iq2 sin 8d

(9-39a)

== - i;2 = - (iJz sin 8\2 + iq2 cos 012)

(9-39b)

iq,

iqLI

DYNAMICS OF SYNCHRONOUS MACHINES

ELECTROMECHANICAL ENERGY CONYERSION

542

9.4.2
~ -6

,
: #

'8'3

U, UMOuS SI

Z# JOleuJa1lV

....

..;

i -- - - -_. - - t -- - - ----------1-

~j_ ..

I
I!

,j

3"10
3

L:

...

i-t----------1---l
I~

:--1----;-----;-

~~I

.8

E1J"

~ 1
... I

+_....,

~"""!-.~".

<s'
-,
~I

'0

t!'

>-1

>-

~.

___ .J

'--

,f

I,

I
t

Ii'? '\'"

r''''

...

~"'t

:-...~

,1'1'

,-

<

++

.,l

'.

~.

~ >..

i':,'Jl\:'.

J:

:'i

0>0

-i~1
. ~.i,:,

c'

'

<!~

i:.'~"

:.. .' ..,

.

....
' ..!':

(, '\'
,\,-1)

.~l'i.l

tl

it

';,>.
L.:

... '

,il':'

"~I'

.' .

"

!i, 'j

I
,.

I'
f.- ~t 1111 IINIlI"_ ~'" I l, t"llfll1t1UV

1
ii:

".r

"

h'?,l ~f,~:~l~-,

. 'l. '~

:1'

1~:

;' I

,I~';

"',

,-I

'.

:a

J::c

,..a .,

. ~~,)

..

f!'

$$ -~;

,ll

-~

t:,. ..~k~_

,,/t.~

:,

.}t

'.,:

Summary

The block diagrams for alternators, loads, and parallel connections have
been derived and are presented in Figs. 9-5, 9-8. and 9- I2. To complete
the study of a voltage regulation system. either a single machine or
parallel-connected machines, the block diagrams for the exciter, voltage
regulator, drive, speed control, and load division circuits must be obtained.
The detailed block diagrams of these components will not be derived here
because they depend on the particular type of component selected for
each application. The general techniques for obtaining these additional
block diagrams are the same as those used in the study in this chapter.
A eOlllputer representation of a complete two-alternator system for an
, nircrart power system has been developed and checked by the staff of the
, MIT Fnergy ('on version Laboratory using the teehniq ues of this chapter. *
The results showed excellent correlation between equipment tests and
.computer lests.
As I';,s( vollage regulators and speed govcrnors become increasingly
IIsed, co III plel c a Iterna tor represen tal ions of the type developed in this
chapIn will he used more eXll'nsively hy pown systcm analysts.
Thi, \\,,'rk was SIlpp"rled ill pari hy Ihe II,S, Ai, hlrce, HCI'''!'ls covering this
WOI k are: I.. V, 1I0lli alld V, n, II aas. .. A nalol~ ('"mputer H crrcscnt;llion~ of
A11,'11 '" 1"", I,,, 1',1I"lkl (ll',',al'''".'' "''''.'1': fillfls . VI'I -'(1. 1''''1 I, May. )957. pp.
I ~ \ 1"<' 1 V 11"11,, M, Hill/,lIlIt! I', I':, SlIltl", .il .... A,,;,Jog ('01111'111<:1' Representation
01

1\ itt"!:\1 I

~Pl.

,~r

Block Diagram for Parallel Operation

The constraints of parallel connection expressed in dq coordinates by

Eqs. 9-3(\, 9-37, and 9-39 can be applied with any of the loads of Table
9-2 and the rcpresentation of Fig, 9-5 to obtain a block diagram
represC"'ation of two parallel-connected alternator-load systems. An
example of parallel connections with inductive loads is shown in Fig. 9-12.
Similar block diagrams for other types of 10ad~ can be developed by the
same techniques using the load equations of Table 9-2,
When more than two alternators are to be parallel-connected these
same tcchniques can he applied. The alternators arc numbered
1,2,3, , . , . n and the transformations of Eqs, 9-37 and 9-39 arc used to
refer all constraints to the dq coordinates of one machine, If machine I is
taken as reference, then angles "12, "13, "14, ... , "In are defined as in Eq.
9-34 and all transformations are with respect to machine I as reference.

9.5

'!

---- ....,

I
I
I

...

E <l>

~,~

co,

~ t ~
E
I...;:::;.'0.
~
~
_..L

\.\

7d

1.-i5

Ii

'I

1\114'111;11 iliH ('lInt'lll

I'/sh. AI' \ \((.lh),12'12,

A'""all i\11\'lIlallll~:
U"}'I'11 -)(,

; ~,.',

543

-III, :';(Opl

{'WTCIlI
It)'Hl , AI'

(;('Iwrlll inp, SV"fCI)!....

JVA /)(. 1",.(

II.

Noll" .56.JH4.

V II, Ilaas, 1', L. SlIlilh, and I), {' While. "Study of

I kc'lr;.: I'owcr {;ellcralill/t Systellls." WADe Tech.
\ \(l,fh).\}ll..

't,

5+1

ELEC I KOMEUfANICAL ENERGY CONVERSIUN

PROBLEMS

9-1. Assume that an altl'l'nator has a zero power factor 10:1(1 as rcprt'st'nle<!
in block diagnllll form in hg. 9-9. Assume thai lhe spet:o is consl;\Il1 at the
base value woo Neglect saturation.
(0) Determine lhe transfer function from field voltage 10 load vol!;Ie',t:,
(b) For a step of tield vollage
applied at I = O. Wh.lt is the load voltage
response'?

v,

iL

,..

/;

1',

:"I,

Generalized Analysis.

t'

of

the n-m Winding Machine

I,',

.;

'I

1
t

'~

"

,I

. :'f1:

,..

T.

:1;

10.0

where 8 is the torque angle:

(D) Obtain a linearized incremental block diagram showing .::lv, and .::lTL as
inputs and .::l8 as output.
(b) Find the response of the torque angle Ll8 to a step change in L1TL with
.::lv, = O.
(c) Find the response of the torque angle Ll8 to a step change iii Llv! with
jjh=O.
(d) Find the response of the torque angle ,,18 to simultaneous steps in .::lvl
and jjTL
(e) Sketch the frequency response [Ll8(jw)/jjT1.(jw))) with .::lv, = 0 assuming
the system to lx underdampcd. Which parameters affect the magnitude of the
resonant peak?

lil

TL = TLO + jjTL (1)

v1 = V,o + jjv1(/)
S = So + ,,18(1)

,.)

:', i

9-2. A two-pole, three-phase synchronous motor is excited by a balanced

set of three-phase voltages. The frequency and amplitude of these voltages is
kept fixed. The inertia of the motor plus load is J. the load torljue is Tl.. and
mechanical damping can be neglected. The field is excited by a voltage source
v,. The machine has damper windings in both axes. Magnetic satur<\tion is
neglected.
For small excursions from a steady-state operating point, specified by

.-tof

"

-~

....:

Introd uction

The rotating electric machines studied in previous chapters have been

restricted to be either two-phase or three-phase. In all cases the idealized
two-phase machine has been used as the basic device for studying the
energy conversion properties of electric machines. This limitation to twophase machines was imposed on previous investigations because in the
majority of practical cases an n-phase-stator, m-phase-rotor machine
can be reduced for purposes of analysis of the energy conversion characteristics to an equivalent two-phase machine. The analytical techniques
necessary for reducing the nom phase machine to a two-phase machine
and the limitations of such an analysis will be introduced in this
chapter.
In developing a generalized mathematical theory applicable to a large
variety of rotating electric machines, it must be recognized that an allencompassing analysis valid for all types of rotating electromechanical
devices under the varied conditions of operation encountered in practice
is difficult and will not be attempted. It is necessary at the outset to
.introduce certain assumptions and idealizations and, further, to consider
only a limited class of devices and problems.
The type of machines to be dealt with here is characterized by the
presence of geometrical symmetry on at least one of the two sides of the
air gap. The effects of slots are neglected, but not the effects of salientpole structures when these are confined either to the stator or to the
rotor (but not to both). However, the application of the methods of
analysis to be developed in this chapter to salient-pole machines is only
545

I,

546

ELECTROMECHANICAL ENERGY CONYERSION

possible when saliency in these machines can be considered to introduce

at most a second space harmonic in the air-gap permcance. Examples
of machines for which the proposed analytic techniques do not apply are
the inductor-alternator which exhibits saliency on both stator and rotor,
hysteresis machines, and permanent-magnet machines. On the other
hand, problems involving polyphase induction machines, synchronous
machines with smooth and salient l1cId structures, and commutator
machines can be solved, subject to certain constraints, with the use of the
proposed mathematical methods. Certain specific problems such as
harmonic effects of unbalanced loads will not be considered. Neither will
consideration be given to magnetic saturation and other parameter
nonlinearities. Problems related to the detailed calculation and measurement of the" constants" or .. parameters" of the machine and the actual
distribution of the windings are considered as design matters which,
though very important, must be deferred to the particular treatment of a
given situation.
The approach to be adopted is based on the fundamental consideration
that machines can be viewed in terms of sets of linear lumped circuits in
relative motion. The presence of this motion introduces the question of
which of several coordinate systems (reference systems) should be used
to express the equations of motion. The specific choice is usually
governed by the particular problem and the desired form of the equations
, of motion.
Tahlc 10'-1 summarizcs the various descriptions of the classes of
rotating machines to which the analysis is applicable.
'I',

GENERALIZED ANALYSIS OF THE nom WINDING MACHINE 547

'.~K

....o

0.

....
'"
o

>:
IJ,J

t1
'0..
P

...

e ..

~
l::
0

III

Os

'0 0

;020tn

'" E 0 E

~ 0
0 c::

r.%

...

...
'3

:.::l..c
o.O'

..c

.~

'g

Cll -

....

;:l

'r

.....

o~

.so

l:l::
"0

13
.-

eoo

13 eoo
._

t\~:.o

t\~:O

>.

.~~

..c::

<

";

8,

'1J

"'- Of)

'V
.D

c:

CIl

v, "'.0

~0...0
~

;:l

.:' .:~ l..J

8.S

E~
o

L"

.,~

:l!

~ -~

.... i
0
y

'"

y
Cl

.:

-t:;;j 0E]
1\ E c::;
;.-,';;
::: E'~ - i~ ..
en'
~

l::

8-

E:;gc:

E '';

-8

';:;'
0

>l)

.. c:: ...

:1.,,J

.s
~

-f~

Vi

.,

0 ~
~

.-

'-'

I::

0
U

C;

~~~.

ftI~

~:2~

., c:
</)
o:"j
C"j_

~ ..c

'" l::7-

15

t::
~

"0

-g
o

-;:;eoo

"cu

e]

.-'" .g'"' 'E

:E'~ ~,

<U

:;.S

E'-

o ~
U

l'3

c:

l~~
~8~

-; ;

t:.- .&

&0

Y
....
0
y

"0

on'"
""C'1)!!

~'3 ,~

'" 0'

Zl

r5~~

..(

CIl

t b8

.5 E~
....... _

en

11\

I\\E]II\i::

l::

::: E ';

....

VJ' ....

"I

.~

;..

l C

o~
~ () t:l

';:;'
';:j
0
-L..oc
.~ 0/) _ _
~

")II

t\

'"

tI'>_

0..0

_c

Il)

O/l..c
c:: P.- >-.

..c",

.- 19.s:

'5

-~ ~

Vl
l
ro_

-a

~
e""
',: c: '" v~
E'-

''.1

.~~

o
.~

-og

.:

~ ''>

"..f

..

c::

..;'"
l:i-.':l
.S2 ~ l

C1\,> ...

Q::j

<:l
::1"0

1\

"'0
0

~~
:--. ., c: '"

v!::

/, .!.:

I"

:=

>:

'-c
o

"""
0

~:~
'" u
;:l

-.-

,s

oU ro'"'

(;

~eoo

"t'
~

E6~

"0

>.

CIl

rf.j

(j

~.~ ~ $1
o l:I..,. 0

.~

:.:::

U~"'~

"0

Ec:
Ec::
E: E'~ E: E'~

.s
S

::IU

c:E~o..

... ::I

:;

1
~

~
<U
<C

...

J5~

:':::0

y
...
Cl
0

{iY

..

E.!::
Z

III

<
I-

c:;

c8

CIl,

=_)...jC~L...~

$'B

Y~

.g~ c:: ~
5 E~c

f~
c: ()

"0 0

.0

...

....

'1

CIl

CIl

CIl

~8

Il)

~i~~6

<Ji':<Ji'

'13

;gouEc::]
.,j' ;q ~,~ ~

.... "':.:::o..yc:-c

r-F=

V
;:: _0

'" 0

U
.!

CIl

;'i! ~

....

'"

"...,

'"
09'"'

Vi

,J

~f

....

.09
Il)

~
~ '0

09o

8.

c::

~ '" Il)
~
13

e ..c~

>(
~IJ,J

~'o

o'"'

or

Il)

'3

'"

til

~E..c;-o
..cp->...c
u .... "' ..
O'I ......
O"VC'::l..cd..c::

u.J
;:J'o

Th~ annlysis \11' the 1/-/1/ winding, nlll~hine .inYolws two esscntial ~ters:
(I) 1~tHhlishill~ til\" l'llllaliolls or mol ion ill 'l'rms or a scI 01" variahles,
lind (1) l'liminaling as many 01" the variables as possible without losing
any l'ss\'ntial ill/\)l'Jllalion. The estahlishment or (\11': equal ions of motion
(ollows dosely the method established III Chap. .I Ill'tlie stuuy
the
l"'Il-pl\llsc ;i.cncrali/cd nWl'hinc. oml Ihe e1iminatioll or systcm YlIl'illhlc'"
IIldi/<'~. 1!Ie' IIlnllis HlId Il'nll"rOI'lIUllioll Ie'l'hlli'lllcs illll'l"lt"'l'd i" Chap, 4.
'III\' tl'ill\sl'.lrl\lilti,\IlS "hi,'h pl"\\'\' \lS,'I'"I in n'd\lcil\f!, Ill\' variahks in Ihe
11-/11 phase 1l1al'11i,w arc P.l'Il\~r:i1i/.ld versions or 1IIl' complex symmelrical
cOlll\'l)ncnt lransl'ormatinn and Ihc rcal 1r:lllsrl1l'malion, These transI"III'1lHltiol\s will he inlrllduced hefme all IInalysls or Ihe /I-III wlnuing
machine is undertaken.

E ::I
~-g
....._

()

....

"0
<:l

6~:= ~O~(Jbu

III

AnalytiC Techniques

..

<l)
'"

c
~.9
>.r-5
-'0

.:(

c::
>.

III

lIS

'~ ~

Il) '"
' " (";I
(";I.e

i)'~"

'"

.t::

tI'>

'./>

10.1

oE

"0

'l)

05,g

CIl

..>;

"0

rl

""'r"
..

ll::

J:l

~
.....,

CIl

1\

'[

~ ~->
E u ...
.... 1J';
B -5 ~
0 .... '"'
'"' 0

:n

-g~c:

"'1J:a

o""!i::
., Us ,,_

.~-

.
v

o c:
~{;
;;. 1"11

I- ,-.

/,

c:
r

,n

~)

:J

L. -:J

fJ

C':l

:::3

v "0 '"

;,., c: E
5 6
0' ...... 1;'0
5'
Po..

CIl

('oj

I,

:c

,2.~
,::: 2o..u.c

v.'~
~u~

..r:

~ c

on
;:l

I)

'"

C:._

o'"'+''/>
-17J

'V
c::

0..c
.... u

:5:2u

~
E
;,.,

I)

.c:

c:: '"
E
E

on
::l
f)

VI
>l)

c::
0:.2
.... 0
c:

.c: '"

gE

.... +--

.no

8"5
E 03
o Ei

til

I/l

.....

-i

vi

"..

VI

'\)

c:
::l'-

';;jell
-

~ :.a -8.
o ~

-V

'-_
~ c:
~ V).!t

"''''~

~ ~++

+-

.~
l

5<48

10.1.1

Syrlllllclrlc.:al LUIllPOllclll

-I

CONVEI{~ION

ELECTROMECHANICAL ENEKGY

C;rNF.HAI11FD ANAl YSIS or TII[

,.,\II~lonll.llIoll.

I'cl'rC:,CJlIIIIIIIIl .. I Iht,

.c!':rs

or

[/ =

Vn

ee 2

a- 4

-2

(II

'l

2)

-2(n-2)

(~

c:

(II

2("

,j

"J

ft' ~

.~.

I -

n:" r

(10-.1)

':'1

\:,.,<

I ',J;

:)'

PO-oJ)

:~

where

'~..'..:'~'

III

of the origillal varia hies is gi\'l:1l by

..

a-(n-l) (l-2(n-l)

a-(n-I)(II-2) a-(II-I)(tr-l)

1/-1 =

which a = el2 ,,/n = eJ3 is the nth root of unity and S is called the
characteristic angle 27T/n. A typical clement in the ith row and kth
column is given by a-(l-I)(k-l).

vii

til

u n 2

a2

a"

a 2 (n-2)

a,,-I

a 2 (n-l)

In

axis
a.0I- 4
.h

ill

Real axis
1=

014

'c')

'"

a(n-1)(1I-2)

un

a 2 (n-l)

01 5

a(n-l)(n-l)

The symmetrical component transformation matrix satisfies the

following condition:
[ / = [/(-1*
(10-6)
In the field of electrical engineering, systems are encountered whose
matrix descriptions f have symmetries which allow diagonalization of the
mat.rix by the symmetrical component transformation. The resulting
description, Le., the diagonal matrix fs, in terms of the new symmetrical
component variables affords a great simplification in the ensuing analysis,
since the inverse of fs involved in the solution is then readily obtained.
A type of symmetry often found in machine configurations is exhibited
by the matrix

)\1

a-I

all

012

au

014

...

01 (n-I)

01"

Din

all

012

013

...

al(n-2)

l(n-I)

01(11-1)

al"

all

a12

...

1(n-3)

1(11-2)

013

a1'+

.'j'

!?E=

Fig. 10-1. Phasor representation of the n roots of unity (drawn for n = ".

.............. . ......... . ...............

The n roots of unity expressed by the various powers of IX are complex
numbers, which can be conveniently represented graphically in terms of
phasors in the complex plane. Figure 10-1 illustrates this phasor
Books on the subject: W_ V. Lyon. Applications of the Method of Symmetrical
Components. McGraw-Hili, New York, 1937; W. V. Lyon. Trallsi('flt Analysis of
Alternating-Currefll .\Iachillery, John Wiley, New York, 1954; E. Clarke, Circuit Analysis
of A-C Power Systems, Vol. I, John Wiley. New York, 1943.

(10-5)

~~

Imaginary

91-

'I"

TIle illvcrse lrallSllll'lllalioll c.\pressillg the symmctrical 1.:OttlPOllCllts

.rs

(10-2)

1/

",.

1 ... 1- cr" J.. -I-

4. 4 ,.

I)

I)

ul IIlI11y rul' lilc special elise III wlllch

where i -/- ,., tt alld I,. r. and i are integers.

tcrlll~

IX

"

( 10- J)

where the Xs are l~e symmetrical wmponents

Ille origillal variables ./;.
The symmetrical component transformation malrix .'/' is

a-- I

""II~

WINDING MACIIINL HI

The rllllllwillg rdal inlls arc lIseful:

The symmetrical component transformation, originally dcwlopcd by

Fortcscuc, is a Illteal' cOlllplcx trtlllst'UrJllallllJl oj vill'lahlcs lkllill:d lIy
:1; =

1/

It'"

al2

..
J

a1S

...

a1/1

(10-1)

all

This symmetry may be termed" circulant": the second row of the matrix
is written by placing the nth element in row I at the start of row 2 and
shirting all other elements to the right; similarly, row 3 can be generated
from row 2, etc" until the nth row is reached. Observe that all the
elements along a line parallel to the main diagonal are equal.

'.-c ...

550

GENERALIZED ANALYSIS OF THE n-m WINDING MACHINE 551

ELECTROMECHANICAL ENERGY CONVERSION

G'

Assume that the matrix !Z of Eq. 10-7 is an impedance matrix relating

the voltages v and currents i as
(10-8)
v = ~e'

..I

o
......,

[i~.1

Transforming v and i using Eq. 10-4 gives

s = 9"-1. v
is = .9"-1 , t'

(10-9)

(10-10)

and

Vs '\ 9"-I!!r.9"e's

(10-11)

!Zsis

i. __ "

= 9"-I!!r.9"

(10-12)

.\

!Zs =

o
o

o
where
t

...
...
2 ...

0
0
0

. ,.

0,,-2

= k-I
L 0lk

01

L"

l,.l'!l

~,~~.tl"" ~

~~if ~

02

1,1

"~

n-I

.1'11

.1'1'

(/~

I, 2, ' , , ,

II, .... ",,-,

,~atisrying
I'

the relation:
1/(;1.

10-14)

slll:h thaI
,. .... I, 2, ...

.--..

~IE::

~I~ ~

0
u

"

i.'"

.,~

I, ,.

11/2

iii
'1:>.

" ,,":l.,,4;1

E::

--- --- --(.)

"" ",,,~\,, ... ~, rb~dw

~ IE:

Vl

~~".;

'"

(.)

II: I

~t
,,

-DVl

J
'I

-- --- ~~-j
-e

,,,

."0

(.)

,;1;. ..,.

-IE::

~Il::: ~I~
I
I
-{l

'/

,.:,)';,;

:,f iVl

-~:; ~~;.

(.)

0
tJ

'!"'1

-/

...-.....-..

~
. <

'''l

t'

!f'S'-

,...(}~~

(10-15).

~.\'!..

'.

"'r

J 'l

'Ill.

:1,t

:;i,;J1;

1/ -

'"

'-"

IE:

+
~
,,,

I I::

I:

('I

~~

only if /I i... ('VI'Il,

II', in addilion II) bl'ing rcal and possl'ssing a l'ircnlalll syn'lnH'lry,
l11atl'lX :.J is also Syllll1lclrical, I1lcaning thai :Y " '.'f,. so Ihal In c1l't;ct
It I~ ~YIIIIII~lrl~lIl lIhUlil IWIII dlllgllllal" tile diagonal 11Iatll!\ :.7 s " ahll
ar~

.t

'\"

I,

1/

E::

n}\\\
...-..

(.)

TWl\ d~ll\~nls. ht)w~v~r. ure always r.elll: lin and ""I.~' Ihe laller cxisting

re:ll :ll\d its ckll\cl\ts

ctiagnnal mntrix 10-13 are complex ql1antities

'

~I~

Vl

"i

H the clements of matrix ::r in Eq. 107 are real quantities, the above
rdati'"ls indicall: IIl;lt all the clemcnts cxccpt two ill lhc tl'Hnsronncd
tl r

r-I

...-..N

<1

1)lIlk

Il(A

..-

fl

11

I,,, .
("

,~1l

~(

= Ie-I
2: a- 2 (k-I)0Jk ' ;;,
~ It
k-I

(10-13)

1 ~

i"

,,,
8

<tit." ~,

n
II"

cc(k-I)alk
I e - I " ) \"

'-"

Vl

"I"If

If

L:!:..-J
.:::

'~\

',-,

.--..

.~

00

,dT

~ I t:

o(.)

0
0
0

0
0

.'< "'){!

5
~,

Using the transformation relationship of Eg, 10-12 with Eqs, 10-5, 10-7,
and 10-2 for .9'-1, fL ,and .9' yields the following result:

0
0

'-"

The new impedance matrix is

!Zs

~I
7

,~.

I~I':J::,.
'
,.'~~
),1

-,

>

552

<

GENERALIZED ANAL. YS'/$ OF THE

Anothe r example of a particu lar kind of symmetry for which

the
symmetrical component method achieves the diagonalization process
is
the rectangular (n x m) matrix 10-16. Matrix 10-16 is best rewritte
n
in exponential form by using the identity

,,'

,~,.

<~
ELECT ROMEC HANIC AL ENERGY CONVE RSION

.,

fr:!.... \
:~

cos

"

Ej~.
+

k~

= t,e j </>f3l a -k + '!e-j</>fJ- iak

The same diagonalizing procedu re can, be used on the transpose

o(
:1
'
I
"
~'nm, - [/_1
2'nm, 9'Ifn
mm'
" with the result

,2'Nfl' i.e.,

, ,.
~

vnm

.~.

,,'"
!,J,

'"

..

.p

nm-

~l'

{3a-(n-l)

l,
'\

{rl

~
I

j
e- </> \

" . : "....

'I'

+2-

.;'

;:,t".

.,

1;: ' )
,'~.

1'101 -/
'!i -\",,;

l~; ,.:'.~::~
I,. ~ .. Ii,""

.t'l'it .'..I(r~-""""'"
i$! .; , ..
I ~~
....II.'
\ 0$"
~

/.

'lI',~"t

,J,
~
~

..,,".,
(\

."
,

I""""'!

."

...

f3m-1 a- (n- I)

. ..

f3-(m-l)

...
.. .

f3-1 a

f3-(m-I) a

...
I
...... ...... ...... ...... .....
a2

fJ-I a 2

n-l

{3-1 a n-1

L:..:-

~~f'~'

~)'

.'

,41' ....

t"

(10-12,)

{3-(m-1l a n-l

When dealing with a rectangular matrix ;'l'n", in Eq. 108, the matricts
representing the changes of variables expressed by Eqs. 10-9 and
10-10
are necessarily of different order. The voltage transfo rmation of
Eq.
10-9 is of order n, whereas the current transfo rmation of Eq. 10-10
is
of order m:
Vi = :;r;n l V
(10-19)
i' = !r;;,~l

(10-20)

Performing this operation on matrix 2'''111 of Eq. IO-U~. the result

is
surprisingly simple:

.'j'

'p'

- nm

vnm
2

0
0

I~

J
e

]]

.j

(10.22)\

li

0 . eJ-

Xa

':';

Xb
",,~

,.',

x=

Xci'
'(~jMl~"

..,

(lO-23a)

..

l '

Xk

The symmetrical compon ent transformation 10-1 introduces a new

set
of n quantities whicll are, in gCl1cr~d, complex. They are denoted by
the
column matrix Xs in the sequence shown:

~;l: ~,

"

(l0-21)

o
o
o

where no~ th~ "poly nonzero elements are located in the 2, 2 and
m, It i
positions.
:'
The importa nce of matrices such as Eqs. 107 and 10-16 in the field
or.
rotating electric machinery will be evident later. The next step in develop
-:
ing the symmetrical compon ent variables is to interpret in some detail
the,
meaning of the symmetrical compon ent variables in relation to the origina
l
'set of variables. To this end, consider the change of variable introdu
ced:
by the symmetrical compon ent transformation as expressed in Eq. 10-1.
The original n quantities are assumed to be of the same kind, having
values X a , Xb' Xc, , Xk where Xk is the last or the nth of the quantiti
es. "
These values may be real, imaginary, complex, constant, or variabl
e.:\
,They may represent either fixed or rotating phasors. Withou
t yet ",
inquiring into what the quantities actually represent, they will be written
',I
as the column matrix x
'

The transformations of Eqs. 10-19 yield the new impedance matrix

;'l'~m = 9'/~nt . :':zn", . ,'/'",m

0
0
0

,i

{3m-I a -l

it"l,1

.Hi ,."'. ~; ..."

1'1'1

f3a- Z

a-(n-I)

j'{

"~l;

a- 2

'

I,t.rj"

"1,

fJa- 1

"'2

\.,.

lit",

a-I
eN

I0 o

fJm-l

f3

,,'.,

.P~ml =-2-

are respectively the nth and mth roots

eJZ,,/m

~,

I,. 'r'>'

o o
o e-j </>

:"i"

(l0-17)

' ii;'

where a = ejZ,,/n and fJ =

of unity. Thus

n-m WINDI NG MACHINE 553

,

':\

,}

Xz

xs =

Here all elements are zero except the two located in positions 2,
2 and
Xn-I

n,m.

._:f,~

(1O-23b)

554

ELECTROMECHANICAL ENERGY CONVERSION

It is important to emphasize that the symmetrical components as given

in Eq, lO-23b are not only functions of the n original quantities, but
depend also upon the order or sequence in which these quantities are
given in Eq, 1O-23a.
It is customary to call X o the zero-sequence component, Xl the positivesequence component. and X n - j the negative-sequence component.
Because of the importance of these last two components in the study of
machine behavior (this can be inferred at this state by observing that the
nonzero elements in matrix. 10-21 are precisely those two sequence
quantities), they will specifically be labeled as

GENERALIZED ANALYSIS OF THE nom WINDING MACHINE 555

are the instantaneous values of the voltages and currents (assumed to be

actual voltages and currents and therefore real) in three electric circuits,
the symmetrical components, written as

~
o

Vs = V+

'O

are, in accordance with the transformation of Eq. 104,

1
v)
(va + Vb + ve)

for the positive-sequence quantities

X+ = Xl

(10-24)
X_

The meaning of these symmetrical components will be investigated

considering the specific case of electrical three-phase systems, which are
probably the most commonly encountered electrical powel; systems.

.'/ =

v'3

1
1

The inverse transformation is

.'/-1 =

V3

1Ij

1 a
1 (X2

Il" I
./,

~"

of:

ex = e jZ ,,/3 =
a 2 = eJ4-tr13 =

If
f,J

~
Vb

Ve

.,'i

(1O-25a)

...

~b
]

'e

I (.

.
Ib

'.e)

I (.'a +'
2')
v3
alb + a 'e

. =

1 (.'a + a 2'lb + ale')

vi

(1O-26b)

I,
1

= ~3 (vo + v+ + v~)
1

Vb =

v3 (vo + a 2v+ + av_)

"e

v3 (vo + aV+ + a 2v_)

(10-290)

I
'a = v3
(io + i+ + i-)

*'~,

'I

(J0-28b)

and

:~

.'.>

e ...

(10-280)

The values of the instantaneous quantities in terms of the symmetrical

components are:

(10-25b)

f l'

-0.5 + jO.866
-0.5 - jO.866

and

.~;,.,

+ aVe)

. =

Po

'a

(10-26a)

a2Vb

a 2 ve)

1+

1_

I,

where

v3 'a +

a~

a2

,,~

l'

':;,,,

,
';

,,2

v3 (Va +

10

.'\'

Li

a-~

0:-2

'.

C(~~ = J3 ~

v_ =

aVb

and

, 'fl.

The matrix. of the symmetrical component transformation for a threephase system is obtained from Eq. 10-2 by letting fl == 3:

_1_

V3 (Va +

~t

'1,

10.1.2 Three-Phase Symmetrical Components

1
,el

l;
t'

V+

for the negative-sequence quantities

X.-l

(l0-27b)

1_

V-

Vo =

and

is = ~+

and

(lO-27a)

lb

Ie

[
v3
(io + a.2i+ + ai-)

v3 (io + ai+ + a 2i-)

(10-29b)

'

Note that, since the voltages '!-) and currents i are real quantities and
since a and 2 are conjugate complex quantities, the positive-sequence

'.

~ ~:!;:r '\i1.

.~;.

_..-'
>'

~ .. ~h:'~

556

"t~,,-

ELECTROMECHANICAL ENERGY CONVERSION

GENERALIZED ANAlY'SIS"OF THE nom WINDING MACHINE 557

/U'<
~

i~'

'\.~

components (of instantaneous real variables) and the negative-sequence

components are always conjugates, i.e.,

V+ =

V~

and

".,"'"

,j

',:"

a=a- 2

\
\

-----~;.,

;;
';
,_.

1
If'

1't..:
t

"'1"

''1l

,..,

1=

LIe

0'3

JIt'= -

{'b

):

"

= i7' . v

= il,.5I'j.5l'v s

(10-33a)
/

and then, by Eq. 10-6, this becomes,

Va

p(t) ... ~':, V s = [i~

a 2 .. a-I

la'

= i~vo

-Y311_

(b'

d'

i:

i~ /. v+

(I0-33b)

+ itv+ + i!v_

,"'",

Fla. 10-2. Phasor representation of three-phase instantaneous symmetrical components.

k~

,I'h.. . .

operating on a given quantity performs a rotation of that quantity through

the characteristic angle 1) = 27T/3 in the complex plane (positive rotation
being in the conventional counterclockwise direction) and a 2 ... a-I
produces a rotation of -(2rr/3), i.e., in the negative direction .
Figure IO-2h illustrates the graphical representation of the symmetrical
component transformation of three real quantities Va' 1'", ve (in that
sequence). Note that, since these quantities are real, they must be
shown as three phasors on the real axis. The positive- and negativesequence components are clearly conjugate phasors.
The instantaneous power in thl: tlm:e phases is given by

t:; ""'.""'"''
t'! . ",",1
-,

'4<''''-

~'~i. .
-.<,~""",.

,~ ~.
(

...

,.j

~ , ::.'~~'

. ,.t

1 ~

~
I
~

.. '

t :'~
J'
:r

~!

'a

p(t)

= (', . 'v = [ill

i"

A,~," '

r,.
\:-.

i c I . Vb

where Va' Vb, Ve indicate peak values and 'Ya' 'Yb, 'Ye are phase angles.
These relations can be written in the exponential form as
Va =

., ~ ,f."

(10-31)

. ~ ;'t

Ve

'.
I

li'l""
. ;iP"

I'

(10-32)

v_

, ';\

Vbl

,~

j,

'I:t . '

..

_-----

c-"

'i\

+ it v" + i:vc

r . . .,.
','
"

p(t)

"i~' I'
,I' (~.~ I.

i:1

which is exactly the same as Eq. 10-31 since all quantities are real.
Equation 10-31 becomes then a special case ofEq. 10-32 for real variables.
The power relationship of Eq. 10-32 can be expressed in terms of
transformed variables using the transformation relationship of Eq. 1()..1
which gives,
;

('lol-,'

,j

= i:Va

-J311+

~;'I;,

'B-,

it

;',H.!

aXIs

~~

ii v = Ii:

p(t) =

Imaginary

".
"1"
, .,..
, ,''''''.':'C

(10-30)

i+ = i!

The zero-sequence components are always real.

In order to better visualize the above relations between the variables
and their symmetrical 'components, the method of n:prescntation of
complex quantities by means of phasors is helpful. The three roots of
unity are n:presented by the three unit phasors oriented as shown in
Fig. 1O-2a. They can thus be interpreted as rotating operators: a

'~', ._~~""' 'J' .

l~

When dealing with complex transformations of variables, it is desirable to

rewrite the power relation 10-31 in a more general form as
'

;.

iava

ibVb

iev e

<,~,.

Vb
Vc

1-( Vae jw1 + V:r jw,)

= t(Vbej"" +
= -H Vee jw1 +

V:e-}<ut) :~
V:e- jw,)

'.'

"!

fi

(10-35)

where V = Vejy is a complex number which has a magnitude of V and

an angle of 'Y, and V = Ve- jy is the conjugate number (see Fig. 10.3).

558

GENERALIZED ANALYSIS OF THE nom WINDING MACHINE 559

ELECTROMECHANICAL ENERGY CONVERSION

The instantaneous quantity Va is thus interpreted as the resultant of two

oppositely rotating phasors which are always conjugate. At time I = 0,
they are in positions indicated by Va and V,; in Fig. 10-3. In conventional steady-state sinusoidal analysis, only the phasor Va is usually
retained (though expressed in rms units) to represent the quantity Va in
a "phasor diagram." At any subsequent time t, the" phasors" Va and
will be rotated through opposite angles wI as illustrated in the figure.

(boldface italic type indicates a phasor or complex magnitude) gives;

1

,,

~,' V~e-jwl

0\,

It'''''tr;:

Fig. 10-3. Phasor representation of a steady-state sinusoidal voltage.

V,

Substituting Eqs. 10-35 in 1O-28a, the following instantaneous symmetrical component voltages result:
" Ii ill :l'.!"

....\1
I

jw, + - -.. (V*

v.o '" - (Va + Vb + V)e
2v3
C
2V3.

V ...

+ aVb + a 2Vc)eJwt +

I '
+ a 2 Yb + aYc)ejw,

. reV.

2v 3

+ V*b

I
2\73(Y:

~.:;

jwl '
+ V*)e<

+ aV: +

l ' .,
a 2 V b*

+ 2v. f'ii'(V*
+
3 a

Yx s

= y-lS

.'\.

i'"') Y3
,

I I

II

_----::..-:J

v.

a V.

'iH

~;;:"f

n:J~.f;a;l-

v,.

(a)

.,,:'1.
(c)

~17';~

'!'l

Sequence components of three-phase phasor quantities. (0) Zero-sequence.

(c) Negative-sequence.

Using the phasor symmetrical components of Eqs. 10-38, Eqs. 10-36

reduce to:
Vo = -t( Voe jwt + Vti e- jwt )
(lO-39a)
v + = t( V + eJwt + V~ e-Jwt)
~
"J ..
, J"
jw
iwt
v_ = t(V_e
+ Vte- ,)
'~

Similarly, the instantaneous symmetrical components of currents may be

expressed in terms of the phasor or steady-state symmetrical components:
io

(.1O-37a)

= t(loejwt + I;e- jwt )

i+ = t(/+eJwt +
C = t(I_e Jw , +

'';'\

r ..J,,,

",

~11

~,,~f

+ aV*)e"J-I.
c

\''11'

Xs

",

(b) Positive-sequence.

a2V~)e-j...t (10-36)

'!

Fig. 10--4.

'''Wj:f(\'1 ltO:'.

Now defining the symmetrical component transformation of phasor

variables as
I' ,.
X

"lpfl ,'(15,;1' ..

1;_

,v+
i

aVe)

Vb

va

I
= 2v'3(Va

+ a2 Vb +

aV,

.",

I
Y- = v] (Va

(10-38)

".

Real axi

Vc )

Note that, in the case of phasor symmetrical components, the positiveand negative-sequence components are no longer conjugate.
Figure 10-4 illustrates the resolution of the sequence components of
three-phase phasor quantities.

Yo

Vb

+ aYb + a 2 Ve)

v3

V:

Vo = -/' (Va
" 3
I
V+ =
(Va

(1Q-37b)

1
,
.illf

I~e-Jw,)
I~e-iwl)

(IQ-39b)

'"

560

...... '"

",",<~A.c

.,

,.

,....

GENER.ALlZED ANALYSIS OF THE nom WINDING MACHINE 561

Substitution of Eqs. 10-39 "Illd 10-38 in Eqs. 1O-33h yields:

,~~,

,,~-

ELECTROMECHANICAL ENERGY CONVERSION

p(t)

"fii ",.-:~~.~' ,.'

(iovo

These balanced systems may be expressed by two sets of three equ$l.

magnitude and equally displaced (at 120 degrees) phasors as shown in
Fig. 10.5, i.e.,
Va'" Vel,.

+ i_v-t + i+v_)

= iloVS

+ tItVo + tIoVoe2lwI + iIi Vir 2jw1

+ tI- v!. + !I~ V+ + tI_ V+e 2lw1 + tI~ V~e-2lwl (10-40)

j, :
t

'..l

iIi.n

and'

+ il+ V~ +'i/!. V - + tI+ V_e 2}w1 + tl!. V~r2lwl

Vb

= a 2 Va

Vc

= aVa

.\

The sum of the terms that are independent of wI represents the average
value of the power, which may also be written as

"JI(I~'

.fh

,~'.'~1~~

~-)->.

p(t)aV8

~'l.'

~.

I,

",
.,I
I"

= i(I; Vo + I: V + + r. V_)
+ t(I;Vo + l~V+ + I~V_)*
=

.' 1-'1

t Re(I:Vo + l:V+ + I~V_)

l;;

: I

NI \."

I 1

:1, I... L.;

c, '

I;,
.1,'
., ."'.\. . . \"
',
~~

Ii

"

a 2I,.

"

III =

~,

)'

Ia = left!

"I

It: ,. ala

.j

!. '. i l

~)';'

I,

,.u..,.,
:';;'~,!.J

ji,

<\oo,*"411~4

,"I- ,.

i!~~,r., .

P ".....

. Ir!/, ,.,

'1- ,,~ " ..

I"'

t",

,"~'. ~

/;

'.li

':

. l#l'';

',fJ

(10-41)
f1"~

R+tl~I!1 . ~

..

Real axis

v.,
II rtffl\l'

where Re denotes the" real part of."

The sum of the six terms in Eqs. 10-40 which are functions o'f time
gives the instantaneous value of the double-frequency pulsation in power.
An important special case arises with balanced three-phase systems,
i.e., when the system of voltages and currents are of the form

\.,

t.'

(10-43b)

~'t!

l.{.

".J \',Hjim
';~j':~-i!'1<-j

ij. '

- t

II')

J,.,t

'H'.i.t.( ~

; dh )

.'"
, ".,' <,,'
.'':'
{1 ,'..

'"

-J-O

(1O-43a)

l,'.,

L',

,-,

~-

,.,'J' lol "

1<':; . ,~1;f~'Ji: .. t

.'1 ,',

Vb

:i
'.~

:!

~l

Fig. 10-5. Three-phase system of balanced currents and voltages

:'".,

t" .. .....- .

~I

1;1;

'.'
'

"

_',

.',

r ,.
1.

The phasor symmetrical components of the above phasors

= Vcos (wI

Vb

= V COS

(WI + y

Vc

= V cos

(WI + y + 2;)

y)

,.., . ~,..".

.j\~

2;)

.\.

ia = I cos

ib

,~:",

=;;

I cos

i c = I cos

~t
,

a2

V+

= V3 V,,(1 +

a3

+ ( 3)

a4

+ ( 2) = 0

V_ =

(WI + 'I)

2;)

(WI + 'Y)

(wt +

+ ~17T)

'Y)

Vo = v')Va(l

(10-420)

and
1;"

f,

a~e

,'1'

1.'I.",rj,.......~:;

11

Va

"

V3V,,(1 +

....

v'3V..

;~

Similarly,
(10-42h)

a) = 0

(10-44a)

!'l.

10 = 0
U,

P"" ','

1+
1-

=
=

(lO-44b)

v3Ia

1:

_J

562

ELECTROMECHANICAL ENERGY CONVERSION

GENERALIZED ANALYSIS OF THE nom WINDING MACHINE 563

Hence, in a balanced system, the zero-sequence and negative-~equence

phasor components are zero. The instantaneous symmetrical components

on the other hand, are, from Eqs. 1O-39a and 10-39b:
I

v_

Xl =

1(xa

+ Xb + Xc +
+ jXb - Xc -

X2 =

!(Xa

Xb

X3 =

!(Xa

jXb -

Xo

Vo = 0
V+

When the original sequence of real variables is denoted by

and Xd, the symmetrical components are:

io = 0

[.

= V3Vaejwt

(l0-45a)

and

i+ =

= V3V*e-jwl
a

;_ =

V3Ia e}w1
V3I:e- iwl

(I0-45b)

'The zero-sequence variables are still zero, but the positive and negativesequence components are conjugates and consist of only one term,
The instantaneous power in the balanced case is:

pet) = i:v+

i+v:-

= (3)(:t)(I:Va
=
=

la V:)

tRe{l:Va }
iRe {Ie-iT) Ve iY}

= t/V cos (y "~"

= !(xa

(10-46)

TJ)

Four-Phase and SemifourPhase (or Two-Phase) Symmetrical

Components
~ ''1t

If \,.

!\

",,"0'

:~",,;""i'

[I ],
[I ]

1 1 -j
7=211-1

1,
",

and the inverse matrix is

J.

1
.... 1
1
-1

-I

Xa

9'-1

=.!-

1
2 1
1

j
-1
-j

-H

1
-1

-I

= -Xc

(10-50)

and they reduce the original four variables to two independent variables.
Equations 10-49 then become:
Xo

= 0

Xl

= Xa

X3

'fV:

+ jXb

X2 =

(l0-51)

= X" - jXb

There can be no zero-sequence component in a two-phase system.

To maintain invariancy of the bilinear form, the positive- and negativesequence components in the two-phase system are defined as:

-j

1
= 'I/2(x
a + jXb)

t;!I.;!'\uT

("

(l0-52a)
"< (

(10-49)

+ jXd)

Xb = -Xd

( ....IIt,~...

(10-47)

jXd)

Both Xo (the zero-sequence component) and X2 are real quantities, and

and X3 are the conjugate terms which constitute the positive- and
negative-sequence components of the four-phase system.
The so-called two-phase symmetrical component transformations cannot
be obtained directly from the general n-phase case. The characteristic
angle or the two-phase system is not. as could be expected, 271/2 = 71, but
71/2 as in the four-phase system. In fact, the two-phase system is a special
case of the four-phase system in which particular constraints are imposed
On the variables. These constraints are

The four-phase symmetrical component matrix is derived from the

general n-phase matrix Eq. 10-2 by letting n = 4. In this case the
characteristic angle :5 equals 271/4 = 71/2 and the operator CI. becomes
equal to j. Hence, the matrix of transformation is:

Ii !

Xc

Xd)
Xd)

xc.

Xl

This is also equal to the average power. The pulsating component of

instantaneous power is zero in a balanced multiphase system.

10.1.3

+ Xc

X a , Xb,

x_ = V2(xa

...:..:

jxb)

(10-4&)
i,e., they are equal to 1/\/'2 times the corresponding positive- and negativesequence components as defined by Eqs. 1o-? 1 in the constrained four-phase

:'"
t.

564

system. The variables x a and Xb are usually denoted by x a and xp.

The. two-phase matrices of transformation are then:

'!
"
~

ELECTROMECHANICAL ENERGY CONVERSION

;.~ '=::~~\
.. f

"""",-'\.,

""'~""' '\

'1">-~"'."''''''

,.....

;.

;".

(lO-S2b)

~ rt
V2l2.

(10-52c)

and

.. ,;>t.i1 ...... ,.

y-l

j,j

_.~.

Y'=~ll 11
V2 L=-j ~
=

'.~

1:: ':: '~-.

"

.. . ",
.... ',

>t.~'~

-~ ~' -

I;

~"':; ~,

.;l~.

'Ii

"

.1

Ii

_t''i'

;~;,J

,:1

.~

..... 10("

).'

\1

"~"l"

Real axis

/'/-v2x",

- ",

.~ I

///

.....,.,

.ioI'-,

x_

.,,, ...

1',1

,1

Fig. 10-6. Geometrical relations between" f3 and

II .............. .
1

1'1

;I., ""."'.' .
I .
~I

l-'.'_ilWlJi1>':.'"

J,

Itt

11

'. """"1~

,I

I.

'14tH:

. -...

il:'~'t-'
t'-j.'t.,~",~

' . . "",

'"''~'-:'''''

IIJ
I

li "

!I

i"

components.

,\

The transformations of the instantaneous voltages and currents in a

two-phase system assume the form:
1
V+ = ~(Va + jVfj)
Va = V2(v+ + v_)
v2
(l0-53a)

..

: : ".

~ I';j

vfj = V2( - jv+

:\

+-

' -."

liilc:.t\
,t,

x = Y'. Xs
(10-55)

.x

= [/-1 , X

. are complex transformations.

If the original n variables in the sequence X of Eq. 10-23a are all real,
the corresponding symmetrical components in Eqs. 10-24 have the
following property:

x, = x:_,

...

- ,....,c~,' ,

It
lit

The symmetrical component transformations which were defined in

r '""' I, 2, ' ..

11 -

I, r ':f O. n/2

(10-56)

:,

:J;:i:

1'1

~!f'

i+v_

Sec, 10.1.1 as

-x_

\1.

---------

.....!

(10-54)

10.104 Generalized Two-Phase Real Component Transformations,

Xs = Y'~

'I}

i'

:1

tj

+ ipvp
i:va + il Vfj
i~v+ + i~v_
iava

and

I~

., ,,

= i_v+ +

!l
-~

"j

!l
Li~ 1j..

p(t)

'--

-..t",l>

;1,

The expressions for the instantaneous power are:

The two-phase system should preferably be referred to as a semifourphase system for which the symmetrical component transformations are
expressed by Eqs. 10-52. Figure 10-6 is a geometrical phasor representation of these transformations.

I:: ~""

GENERALIZED ANALYSIS OF THE nom WINDING MACHINE 565

+ jv_)

V- =

1
v,-/v

a -

jVtJ)

,1j(,t'J'l"

and
.

la =

1
V2(i+

L)

1+

V2(ia + jitJ)
(IO-53b)

.
Ifj

V2( -ji+ + jL)

- -- V2(0I"

The two components Xo and X n /2 [the latter component exists only if 11

is even and is located in the (n/2 + l)th row of column matrix 10-23b]
are always real quantities. All other components appear in conjugate'
pairs. It should therefore be possible to combine these conjugate pairs'
of symmetrical component variables and obtain a new set of n independent
variables which are real. This set of real variables, which will be called
the generalized two-phase components, serves the same purpose as the
symmetrical components: it reduces the mathematical complexity in the
description of certain electrical systems.
The generalized two-phase components can be derived from the
symmetrical components by performing, on each pair of conjugate
symmetrical components, a semifour phase (or two-phase) transformation
. as defined by Eqs. 10-52. The generalized semifour-phase (or two-phase)
symmetrical component transformations are then expressed as

jip)

Xs

gj 'XT

(1O-57a)

XT

= gj-l Xs

(10-57b)

and

566

ELECTROMECHANICAL ENERGY CONVERSION

where X s is tht: symmetrical component matrix 10-24 and

responding two-pha't: real ~omponent matrix defined as
1O'1fl"

..

Xr

(l0-58)

Xt2

X e(n-I)

The matrices of transformation are:

v2 0 0
o 1 0

:1

,.~

'<):J

t'!

1
PA = v2

'

567'1

,l

Xo
xr

GENERALIZED ANALYSIS OF THE nom WINDING MACHINE

Note that the element Vi located in the center of these matrices eXists!!
only if the order of the matrices is even.
i,
I
Analogous to XI being called the positive-sequence component X+, and'i
x n - l the negative-sequence component X_, the generalized real two-phaSe, i
system of Eq. 10-58 will have xn designated as X a , the a component,,:
and xt(n-'h as xfj, the P component. It must also be noted that the;;
generalized two-phase components in Eq. 10-58 appear in pairs symmetric;
about the (n12 + I)th row. Only the zero-sequence and (n/2 + l)th;:
components are single components which are not functions of thei
characteristic angle S; as before, x n /2 exists only if n is even.
il
Combining Eqs. 10-55 with 10-1, the direct transformation between'!
original variables x and their two-phase components XT results:
I

is the cor-

Xo

..

". ~(;

'It.,t,,

:1

or . xr

(10-61)

and
XT

''il
n

v2

= !I' . PA XT =

:=

!!J- I . !I'-l . X

= or-I. X

(10-62) :

The matrix of the two-phase rea! transformation is thus:

,i

:~'

vi

.} - .

-j

1'1:

c",
'~j

f.~. _
,.~

T~'

-j
0

0 1
1 0

"r"
';Jl'"

---.

and

ij"i

vi

0 0
0 1
1 0

010

..
"".

"

~'1

or.

0
-j
-l
(10-59)

: .~,

d""".,._

,,.; i

", ,

:~;''''

iI

,;
I

3d':

;..,~

'"

"

:.,,:

.U;,

vi

cos S

cos 28

vi ...
.. . -2

cos 215

cos 415

...

cos 38

cos 615

.,

v2
v2
-2

~l'

. :).;

"

l:

O-j

sin 20

sin 0

. ..

sin 615

sin 315

,
I

This column exists

only if n is even
~
Ii

vi

o -j

(10-63),

v2

. .. sin 48 sin 28

".\;' ~)Pa$. l(.!i)$',-j'; l{l 1

"

As should be clear by now, all these transformations keep quadratic

forms invariant and hence
:,

~h

~lr

:,

;1

{fK~H

vi
2
1 -vi
2

;"

"., It.',+
.~.

.} ~J)2_

5', -

i'" ,

l\

fJI- =

'-'w'- ,t"

':

\~'

1".',.
~ "'j'

:!'.

1,/. ;~"

vi

.. .

-j
''<.:

"

,:';?I.

.;,,.,

.""
.....:

,,~;"

", .
<I;

or-I = or,
as can be checked by performing the multiplication
81- 1 .9'-1

(10-60)
"."

-,

,'1.'".'
..'.'.'.'
.,;

(10-64)'

(10-65)

ELECTROMECHANICAL ENERGY CONVERSION

568

produced by a given coil will be postulated on the basis of the results of

Chap. J and this will be used to calculate inductances by the well-known
flux linkage approach. This will prove su{l]cient to estahlish the general
fonn of the inductance matrices which will characterize the machine.
The techniques of analyzing the n-m winding machine will be developed
by considering a machine Wilh a rotor and stator structure that is physically
symmetrical around the air gap. Around the periphery of the stator of
the machine there will be located n identical cosinusoiJally distributed
coils per pair of poles spaccd 27T/n electrical radians apart; and the rotor
will have located around its periphery 111 identical coils per pair of poles
space 2rrill1 electrical radians apart. A typical layout for such a machine
is shol-ll1 in Figs. 10-7 and IO-l.'. I n this schematic type or drawing, each
coil is labeled a. h, l' . . . n on the stator and a, h, C, .. m on the rotor. '
It is assumed that each coil-although shown lumped in one slot-is
actually cosinusoidally distributed in space on a smooth magnetic structure
and that the mutual flux density between rotor and stator produced by
this distributed coil is a sinusoidal function of space. In practical
machines, coils (usually called phase belts in machine theory) may be
distributed in several slots and may have a fractional pitch (i.e. a coil
th row a r less than". electrical radians between coil sides). For distributed
and fractional pitch windings the equivalent turns per coil may be derived
in a straightforward manner. For each phase belt in the n identical
stator phascs and In identical rotor phases the turn density and terminal
current are defined as

As a specific example. consider a sCl or II ~. llnce rcal variables X,t>

x h and XC' In this case the two-phase real componcnl transformation
matrices are:

i~.

(10-66)

;1
t

(10-67)

The real two-phase variables form the sequence xo. X a x s' Furthermore,
in this three-phase case. the transformations between symmetrical
components and two-phase components are:

f'~

x~

x+
x_

I
v2

IV' .~!
0 0I
0 1

-)

L-

and

~
x ..
xlJ

v2
1

L'

0
0

t;r

o
1

-j

~
I-

x..
xlJ '

x+

(10-68)

Stator
Z'

r--,

maximum turn density per phase belt in a cosinusoidally

distributed stator coil
if = instantaneous current in stator coil j, the terminal current of j

Jm

GENERALIZED ANALYSIS OF THE nom WINDING MACHINE 569

(10-69)

zr =

x_

i; =

10.2 The n-m Winding Machine

10.2.1

The previous sections have developed the general transformations which

can be used to analyze an nom phase machine. The first step in the
analysis of the nom phase machine is to develop a physical description of
the device and find its equations of motion. The techniques for doing this
have been established in Chap. 3 in the study of the generalized two-phase
machine. In this chapter a complete field analysis of the physical model
will not be made. Rather, the essential form of the magnetic field

,
1'

Rotor

rr""

maximum turn density per phase belt in a cosinusoidally

distributed rotor coil
instantaneous current in rotor coi 1j, the terminal current of j

Resistance Matrices of Stator and Rotor

The resistance matrices for the n-m winding symmetrical machine are
derived assuming that all stator coils are physically identical and similarly
that all rotor coils are physically identical. Thus the coil resistances can
be defined as
'.
R~~ = R~~ = R;; = ... R~~ .... R S
(10-70)
and
R~~ = R'M, = R;; == ... R';;,m == R'
(10-71)

I'"

-"I

;,.'

~.

./

""

...

0.-''' _ _~---.:_.:.::=~

J.

_.~

-:r,''';'!'~

! -U__ ~-..:-.~.~._rll! '- .,. ~.J<.

'U"~

:x!'C.!!:. ~,.~\ .. _~~.~ __ .:.:.:::::'::~_:..:.'.-~ _ ,._~~.;:!'.::,,~ ~ -~ r.'=':-;,:.~~_,_~_~

*~r;=;;~r

~ ~

--,,-=-:-

-.:e:e."":::""C"

.~~

.f'~-?!'llF.......

';,.

~ ~ -:~._

-:

-I

'

.~

~, ~~:

; (:

2 ~ In 4,,/P mechanical radians

tl' ~ ;' ~ t ? ~ ;~...~ t

===1

~
m
m

Air- gap flux density produced by eQui,,'aJent stator coil a

~_.-

B;

=Z~i~J10 i sin 0'

=Z:i~J.loi-sin (8' + 6)

;;0

.,.

v
ThIS flu). distribution is for a

t'wo~poIe

machine, for a P pole machine there

()

are P/2 complete sine waves around

the penphery of the air &ap

:>

Z
()

:>

Rotation

r
m

;;0

G)

-<

()

<
m
;;0
V'

<5

,r=2'K'/m

0;= (R-JI4I

:(

Physical model of n-m phase machine.

Fig. 10-7.

NOles:
1. B' and S' are plotted at time' = constant.
2. I.. ~ I
. Jtcd by the ',;
,q. ,)~

:$

a ;

,.

.:'11 unsle.

:1!ff

',;;.;~~:;

. ~.--

-~'

..

,.:-

';f,.:
,

'..,.

'11

t j:,;.

~.

"

o'

3!':
",,,,

.,

c:

,.....c:

';'

.:c

F>
a

;r:

'"

. .
"
p

...

-<

-i.

0<>

,""

....

:;;~

'"
.,0'

:>

3 ..

c.

::>

....
"'"
"'"'...
~~

0
::>

.,0

:>

:T~

()

.,:,
0
.,<>

N
m
o

:'.0
,,0

;:S'
~

?0

;;0

.0:

::::

'-

'<

"

l1

0..

'0

::l

:>

...

"

-;
I

-l

()

C)

',~

~.

~~

$.

"r-t ,"
f},'1

b/f('

.~';J '\;j?
~1~
3

~
'7

;;:

a.

d /:

-~ if
,.

V1

iii

'T1

-I
I

:>

3
~

-~
3

:>

()

Z
m
1.11

......

572

GENERALIZED ANALYSIS OF THE nom WINDING MACHINE 573

ELECTROMECHANICAL ENERGY CONVERSION

s:;

The radial flux density

produced by this current sheet can be derived
as in Eq. 3-28, to obtain the result that

and in matrix form the stator and rotor resistances become

~;-

1.I'\_llU

, 0

o
;:jf'fS

'nn

o
R~%

0o

~
0

R~~

Bj'

:"'1

'~.

LI>
If

and

rr
Raa

o R"hb

o
.9t';;;m

where Aij

=0

ij

(10-73)

R'o/I'"

flux linkage of ith stator coil produced by the jth stator

current
= jth stator current

R;:m

(/>"). =

III
iI

I r'aa

'I

L'"ba

I'
I~
Ii

::e."nn = I L'"ea

if
:,

SS
L ab

ss
Luc

..

L~~

L'"bb
L....

ss
L be

..

L'"bll

cb

L'"cc

"

'.l

L~:

:,>'

02

- 2/ cos (8" - Sj)

(10-79)

'f'; f

The flux

<Pj produces

.~

a Rux linkage with the ith stalor coil of

~.

~.
p.

(10-74)

"

L~~

{o' + 'If
a2
Aij' = Jo"
ijZ' J.L~ 2/ cos (8' - Sj)Z' cos (8" - S~)a de-

'i"
j ~...

."

(10-75)

ij to be the terminal current, the current distribution is

Iii)

(l0-76)

I'
~

~~.1i"

~""'\I}

...

'".,

""-1"
. ~ ,.."

~"r

(l O-~O) , '~ .. "

""- " .... """"~

,...

J.L a3
= t\Z')2_o_- 17/ cos (S~ - S'j)

,1:1

Substituting )..;; (Egs. 10-80) into Eq. 10-78 gives for the inductance:
L~f =

ZS cos (fJ' - Sj)

iiZ' cos (fJ' -

t"

:-t.t'

The flux produced by the jth stator coil has been assumed to be a
sinusoidal function of space produced by a current sheet which is a
cosinusoidal function of space. Taking the jth current sheet to be
centered around the angle fJ' = 8j, the conductor distribution is

and taking

. L en
S

= i;Z'/~o

,;

Inductance Matrices of Stator and Rotor

The self- and mutual-inductances of the stator coils and the rotor coils
will not be calculated in detail: however, the general form of these
inductances will be determined. The stator inductance matrix is defined as

,.

" ...

o('l .. ~

(I {tr' + 1T iJ~Z'ILO ~ sin ({J' - Ina dfJ' dz

J Jos

-'1-

.,

:j

11

(10-78)

With the assumption of a sinusoidal flux density the flux produced by

the jth current is:

where 'PIli is the n x n unit matrix.

10.2.2

Xij
= i'~
-

~"~

o"

(10-77)

o
Rcerr

.')
OJ

where a is the radius of the rotor and g is the air-gap length (see Sec.
3.1.2). Each term in the inductance matrix can be determined from
the definition of inductance in tcrms of flux linkage as

(10-72)

R'''1"

a. (DS
= I...j Z'
~J.Lo-SIn v -

a3
g

(Zj2 fLo - 171 cos

(S~

- oj)

Equation 10-81 establishes that the self-inductance (i

is identical, i.e.,
L~: =

Lbi,

= ...

(10-81)

= j)

of each coil

a3
L::" "'" (Z')2 J.Lo g TTl

Similarly, thc even character of the cosine function. cos 0

establishes that

Lij

Li:

(10-82)

= cos (-0)
(10-83)

"j.,.

j' "il

\1

576

GENERALIZED ANALYSIS OF THE nm WINDING MACHINE

ELECTROMECHANICAL ENERGY CONVERSION

and substituting Eq. 10-92 into Eq. 10-89 gives for the inductance

L:;

sn

G'
0\

;2

L:; == (Z')(T)1TJ-Lo -ag I cos (8 + 8; - 8;)

(10-93)
For ditTerent values ofj and i the angles S; and 8: are as given in Table 10-2,

'-'

In :~I

where 8' == 21T/m and 8S == 21T/n (see Fig. 10-7).

TABLE 10-2

:-s

<A:>

r--<

8'I

0
18'
28'
:

(m -

b
c

18'
28'

(n - 1)8'

<A:>
..-.,

3-

a
b
c

1)8'

8'j

I
E::

ct>

'"o

......,

LJ

' 1.

8; - 8n

'.

(10-94)

"!!l'1

l'l

Ljf

L'" cos (8 + 8j - oj)

L" == L

+ t:o

-s

+ +

ct> ct>
'-'
'-'
trJ

r/)

o(.)

(.)

t'J)
(! ;'

'.7'1.(

!),"_,U'.

LJ
'"o
(.)

G:l
:::;-

~:

c;"'5;

ct>

II

-S

CS
(.)

'-' ' - '

<A:>('l

..(H\

(10-97)

== + constant
== .if~'",l

o'"

ct>

ct>

88

Lfl

Equation 10-97 is sufficient to show that

2~n

'j

-<

~
(.)

and that
sr

~J:l:)f1!

(10-95)

and the mutual inductance matrix .2':,'", of Eq. 10-88 can be written as
shown in Eq. 10-96. This inductance matrix is seen to be of the general
form of Eq. 10-16. which was shown to diagonalize under a symmetrical
component transformation. The inductance matrix 2';:n is now needed
to complete the description of the nom winding symmetrical machine.
A procedure similar to that used in deriving Eqs. 10-89 through 10-93
establishes that

'-".

..-.,

Thus the inductance becomes

Lf; == Lsr cos (8 +

. ."..,.,

o
c;"'c;"'
c.o ('l
c.o

II ".\

'v

(10-98)

Thus the complete set of parameter matrices of the nom symmetrical

machine is established and is given by Eqs. 10-72, 10-73, 10-85. 10-87,
10-96, and 10-98.

.~

.~

-;~
+,,~

;l;;,4If

'"o(.)

3
L" == (ZS)(Z')1T,uO a I == (positive) constant

" - /;;..1

Further. it has been assumed that all stator coils are identical and all
rotor coils are identical. Following the sign convention for the assumed
current directions in Fig. 10-7 establishes that

-1fo',

~ l:
~"

~
~.

....

,. '.' . .
.;q_.~'-"~~*~',4
t.
.,.' ..

~~ ~;

J~

578

ELECTROMECHANICAL ENERGY CONVERSION

F ~

'.

10.2.4

I,

ld!"': ..
~

:t-,--

~.".'",

.. ,
,

,.,~".

.-;"-'>:"
."lolI!.~

,.

,
...

j'

......

..,:.IJ.!'

r"'"
'\'

::~'

.,

; Electrical Equations of Motion

Stator
k = a, b, ... ,

Equations of Motion for the nm Phase Machine

The nom phase machine has n + m electrical terminal pairs and one
mechanical terminal pair, where terminal pairs are defined as in Sec. 1.2.1.
The independent electrical coordinates are the coil currents and the
only significant stored electrical energy is assumed to be the magnetic
energy associated with' these currents (this assumes that interwinding
capa-:itance is negligible and hence the electric field stored energy is
negligible). The independent mechanical coordinate is the relative angular
displacement between stator and rotor and the only significant stored
mechanical energy will be the kinetic energy of the rotating structure.
The (n + m + I) generalized coordinates for this machine are shown in
Table 10-3.

'-,jo."'.

Vk =.
',1,

Rotor

".

"~(;

"1

Rotor-m

T = d-T.<tI~) + a~
,

- 2-

Rotor
;0'

q.

q.

i.

r"

P.

I.
Q.

.. ,l
....
~ 4.
'''I~\-''~

.
-

y .... ,

tf ...

.<t... l"

1:.. '1'.....

.-", .. '

I
I

"o""'It"-

! >1'.,

_.lh

.~(."

Viscous
friction or
loss factor

'.
'S

,V

...
...

'.,\,

'S

"

'n

...

J~

~.""'"

,;

~~

.~ .'1 ..

'li !. H
! '
I

f~

:l,

.,.,.
,

~
;

j=a

)-a

v:

...

Vn

v'

...

v'III

R"a.

. ..

R"nn

R"

...

R"mm

IX

""

1 ~.. ~
-2.c.
.c.

",tt

"'f

{Oil

I)

I-a

L.
I-.a

a(L'.....
'r .,
IIJI/il

:l-l.
V'f",

(10 99 )
- . C

,,

h"e1~ MP;"

i:..n';~

.' . : . 1 1

~~~
'z.':m

~S

= _n_

t/'m

t "

Vn

(IOIOOa)

'd'1

Va

"(

. E\\ ~

Vb
;.

v'm
and
l{~~~

" '~,

'S

'"

'S

~~ "
i

'~:

;0

~.

'/1

a
t s

n
is' = __
nm
0'
~m

~,

"

Va

,h:l~. ,;,' i~

,hi

,
L\

*
i

1ft

1 ~
2-.c.

e(LSI) 1/1)
's s
-

00-1.

I-a U'f"

s
In
'r

'a
Ii,

I"

, >l

+ R"kk i'k

vb
4'1

The equations of motion of the nom winding machine can be derived

using any of the techniques developed in Chap. 1 and employed for the
generalized two-phase machine in Chap. 3. Because of the similarity of
the previous treatment a detailed derivation of the equations of motion
will not be given. The equations of motion will be established assuming
that each electrical terminal pair is excited by a voltage sllurce and that
all the terminal currents i;;. ii~, .... i;;, i~, ' . , , i:', and the mechanical
position are independent variables. Any interrelation among these
variables will be introduced as a constraint in the solution of a specific
prohlem. Thus, using the parameters detined in Table 10-3 the equations
of motion become:

L"i')
kJ J

L 2 Oo)Lij,)i:iJ - -2 I-a
L )_.
L U'f"
OoiL';FJI[ij
I-a )-a u,/,

b--1\

-.

.....

:!( L'SjS
+
d'
kJ)

(1O-99b)
\

",',

.:.

"'
If

I-a

The (n + m + I) equations of motion (Eqs. 1O-99a, b, ,c) can also be

written in matrix form.
Define the voltage and current matrices", and e' as
."

ep

J:.

(I0-9~a)

:"

L':Jij + ~ L':Ji;) + Rkkifc

t I-a

1-.

;ii"" ... '

!'I:. ,1.;

k=b ... k=m

...

+ R"kklk's

Mechanical Equation of Motion

d
1"'''' a

''!'';.

k=a k=b ... k-n k=a

dd (

11

+ R"kk'".,
= dtdC")
I\k

'.n ,

Mechanical

Stator-n

v,k

~)

=
!g'

Electrical

d(,,)
I\k

.-dt

;1

k = a, b, ... , m

TABLE 10-3

'"

I: .

GENERALIZED ANALYSIS OF THE n-m WINDING MACHINE 579

'r

1m

(1O-IOOb)

\,

ELECTROMECHANICAL ENERGY CONVERSION

580

GENERALIZED ANALYSIS OF THE nom WINDING MACHINE 581

Using the parameter matrices defined previously and the symbol p

the equations of motion in matrix form become:

== d/dt,

current. Using the transformation equation for voltage and current

defined by Eq. 10-1 gives
i~ = sdNeN
(10-102)

Electrical

sn

~ ""

EJPH + f):'.rs
nn

v;-"

Mechanical

T == p2J4>

+ pexr/>

nn

p2':.n

i I e'~ Ie':" 1

I 8'
.r, P!sr
,,..
II dim'"
+ p!t' m~ em
nm

. f..

(lO-lOla)

.zS'~ ]S

N(} .z.rs

Tij.
(} n",

o CLH,

8 .:/H'

nn

8r/> -L mn

0</1' '""'-.J

v~ = .# N'1..';"

(10-103)

i;" = !JJMl M
v;;' == !JJMVM

(10-105)

where the lower case subscripts nand m apply to the actual currents.
voltages, and parameters of the physical coils on stator and rotor and
the capital subscripts Nand M refer to the transformed symmetrical
component currents, voltages, and parameters. See Table 10-4.

en

(lO-lOlb)

.,
em

TABLE 10--4.

Equation 10-IOla represents a set of n + In differential equations and

these equations plus Eq. IO-IOlb are the (n + In + 1) equations of
motion for the nm phase machine.
A significant reduction in the
amount of labor involved in solving Eq. JO-JOla can often be effected by
a change of variables. In essence, the problem reduces to one of
diagonalizing the coefficient matrix of the set of simultaneous differential
equations. The techniques of diagonalizing matrices and the importance
of various types of symmetry were discussed previously and the types of
symmetry exhibited by the parameter ma trices .0f. and ! are of the type
that the symmetrical component transformation or the generalized,
two-phase transformation will partially diagonalize.
10.2.5

(10-104)

Symmetrical Component Transformation Applied to the nom

Machine

The equations of motion of the nom machine have been established in

terms of the terminal voltages, currents, torques, displacement, and
resistance and inductance parameters. The resulting difTerential equations
are a set of (n + m + I) simultaneous nonlinear differential equations.
1n their present form these equations are very difficult to interpret and
nearly impossible to solve. These equations can be simplified considerably by a change of variables.
The equations of motion (Eqs. 10-101a and b) will be transformed
to a new set in terms of symmetrical component currents and voltages
using the symmetrical component transformation [/ defined in Eq. 10-2.
To avoid confusion with symbols, define sd N as the stator symmetrical
component transformation for both voltage and eu rrent and .04M, as the
rotor symmetrical component transformation for both voltage and

,~
r
1

IN
l:'
e'M

e.'S
eo'S

e~

i: ...

e'~

It

e'i ...

"N-l

(;.~

e';;

e'; ... e':;'

if

e'rl

'S

.,

eo

'r

. . . (;M-l

The symmetrical component transformations .0/N and ,<IN I are shown

in Eqs. 10-2 and 10-5 to be of the form

.\

"

:.A

., '

_~

"','n

;4
:i,

"'YN--

VIi

'~~~l

,J
..

e"~

Actual stator
Symmetrical component stator
Actual rotor
Symmetrical component rotor

1-

Symbolism

1
J

(X-I

a- 2

. ,

a2

(X

(X-2

cc 4

..

a4

ex 2

(1O-106a)

I ;., 'a~' ~~ . ....... ~~~ ~_~

~ ex a 2 , a-2 a-I

!
I

, .~'

.1

'~

.#;,1

= _1_ \ I

VIi

_H'

a2

(X2

...

a- 2

...

a-

a-I
a-

.......................
a- 2
0:- 1

ex- 4

...

a4

a2

a- 2

...

a2

(lo-l06b)

where a = e j (2,,/n) and j = v=T, .#Nand 'd'N 1 are n x n matrices.

The transformations !JJ M and .q;;/ are of exactly the same form as d N
and .sYNI except that each clement a: is replaced by an clement f3 where
f3 = ej(2,,/m), and .q;M and Bdi./ arc m x m matrices.

'I'

582

,;4

4f-.J,.

~>{,

'1

ELECTROMECHANICAL ENERGY CONVERSION

GENERALIZED ANALYSIS OF THE nom WINDING MACHINE 581

Similarly, the rotor inductance matrix becomes

1, ~e:-

....'_-~.l'.

.,

10.2.6

~,;

Symmetrical Component Transformation of Volt-Ampere

Equations

J"!"l

..

~
sN

-;

...$:-,'
~ _:~:.,

..

"~',' , ......

v'M =

-.

c..'"

.,{,.... '/Irt"- ~
-- ~"'"

ml

;. 1

...

"

....;~l<o* ...

""

gg-l(plf'S
)~/ N
lvf
mn

I .d

- 1('pYS' ):-:$

nm

,~,'...,!'I;>~",,~',.

:jf'/.tM

' ....."".,-.

.,'
......,'.

""~I'

= 88-;/:!J',;,m88M = :JtJ"At' R'!JItmHM = R'l!M

(10-109)

~c'

;~'~I

!"f' ;, ~

.... ..
: "[:4 .-,

I 0
0

L'j~

'-~:}~.

L:fo

= L~~ + L~~ + L~~ + L~~ + ' .. L~~ = k-I

2:" L~k

L 11

ss
= L au

L"S

= L SSua +

SS

fro

~.

22

a-ILss

ab

a- 2L ss

"b

a- 2L"

tJc

+ ' , 'aL""n = ~L

+ -4L'" + .,.
(X

tiC

k-l

(.(

2lss
~(Jn

C(

ft

LI}v_I)(N_I)

= L~~ + aL~~ + a2L~sc

and a = e it(hlu) .
valuable later.

+ .. ,a-IL~~

SS
Ob servet h at L SSii = L (N-i)(N-I)'

i
,->

k-I

+ ... (3L:'", = L

P-(k-I)L~~

, I
1

I,

== r:..
~ I-'R-Cm-I)(k-I)L"
ak
k=\

->;(~.,

!l"NM

= ~;/:f~'",H M

~rtfN =

(10-112)

f!!"At I y',:,ndN

(10-113) :

To evaluate these matrix products it is advantageous to express the

matrices ~::m and ff'':,n, which are term by term cosine functions, in their
exponential forms by using the identity

.In

cos X

==

ejx + e- jx
2

Then Eq. 10-96 for ~::m can be written as

,)~:~.

-(k-I)L"
ok

ejf

i>.

2';;"

ej(U-I')

= 2""

ej(9-Z~)

I.".. .-

a-(n-')(k-I)L~~

-'~i~'

.1

L sr

+2

,~;

~."

' f!j(H~')

. ,

e j (Hll'-26).

e-jf

e-j(HI')

e-j(9-I')

e-j(9+1 f -IS)

e-j(9-21')

e-j(HIl'-ZIl')

ej[u+(m-l)I~J

I,

:t

I
I

;j;9~;~~I;'~~(:~I')i2J
e-j(HCm-IWl

.............................................
e-j[U-(n-1 )1']

Th'IsequHltywl
I'
'II prove

'-"'",

e j (9+1'-I')

;j;9~;n~; ):,~ . ';j;O:~f~('n~ ;)~,; . ,',' .'

t!

~ .\

'l~;

sr

"~

.i.
.~.". . :

'--

L~~

= L"aa + /3L"ab +
j3 2L"
'a
c + ... p-'L"
am

~,.

k~1

k-I

and (3 = eit(Z"lm). Observe that L~; == L(M-i)(M-i)'

Defining ~s:m and !l'~~fN as the symmetrical component' mutua'!!
inductance matrices between rotor and stator and between stator and,
rotor respectively and using Eq. 10-96 for !l'~'", and Eg, 10-98 for y~",

= L,
,. a -2(k- IJ L ale
n

',~I:.

".~

.:,!.~;

+ p-zL~:

L"
CM-l)(M- J)

,t.."

SI

t,

+ L';" + L':e + ... L:'",

= L:~ + f3- IL':"

o 'J
L(;'~'KM-"

(10-111)

~:;lt

."

L~~

,c

.................................................

L[fv_I)(N_1J

Where

,,~.

0
0

o
o LV.
o o
o o

k=1

(10-110)

o
o

0
0
0

M -

L"
II

~
LOO ==

o
o

0
0
LSI
22

0
0

rnm

'" 'l'

, .,

~1*h~

~~

0
NN = d-lyss
N . nnd N --[ 0

ff7U

.} ;.v

.,!'

0
Lss
II

where

Defining !f~~N and .:f~~M as the stator and rotor symmetrical component inductance matricl:s and using Eq. lO-gS for Y;,:, and Eq. 10-87
for If:;m yield

.""

(10-107)

(10-108)

SS
L 00
0

...

'f

Similarly,

~.:t.

_ 11- I~"!!I

MM -

(.M

= RS!JIt N

:#:~N = d"N1d:;:,d N = d"NIRsu7lndN

"

~.

IN

~-I(;jfrr
+ pyrrmm )~M
M
mm

:1
:~

p!lm)d

'

~"

.~:

Taking the parameter matrix term by term and carrying out the indicated
matrix products result in the following new parameter matrices for the
symmetrical component currents. These new symmetrical component
parameter matrices will be denoted by subscripts of capital Nand M
(i.e., :r~N' :!l':.IM' etc.).
Defining ;;f;'N and :3f'/.tM as the stator and rotor symmetrical component
resistance matrices and using Eq, 10-72 for :#~~ and Eq. 10-73 for ;jf':.m give

", ........,
-)..,.,

t'

1:"1:11'1- 1(:#" +.

oo

!.

The volt-ampere relations of Eq. 10-10Ia, when transformed from the,

phase variabks to the symmetrical component variables using Eqs.
10-102 through 1O-106b, become

e-j[9+ll f -(n-I)I']

e-j[U+(m-l)4'-(n-I)IlSJ

(10-114)

584

ELECTROMECHANICAL ENERGY CONVERSION

'

1"

IX

eJ~S

f3

eJ~'

and

GENERALIZED ANALYSIS OF THE nom WINDING MACHINE 585

se::

Equation 10-114 can be considerably simplified by referring to Fig.

10-7 where the angles SS and S' are defined as 3s = 21T/n and S' = 21T/m.
Referring now to the transformation matrices d Nand !JtJ M' recall that
elements a and f3 of these matrices were defined as a = ej(l,,/nl and
f3 = e j (211;m). This immediately allows the definition of the elements et
and f3 as

In Eq. 10-98 it was shown that

Thus, taking the transpose
n = se~'"",
of Eg. 10-117 and performing the matrix multiplication of Eq. 10-113
yield the symmetrical component mutual inductance matrix between
stator and rotor
.!1m

, MN -

&J M- I.!f'rs
r;,1 -:JtJ - l.!f'sr .9/
mn' N M
nm,
N

0000001

(10-115)

_ vmn

This equality of the physical displacement between the n stator coils

and the angle 21T(n used in the symmetrical component transformation is
imperative in reducing a matrix of the form of Eg. lO-l14. The same is
true for the rotor. The symmetry of the location of the coils around the
periphery of the stator or rotor is thus essential if the symmetrical
component transformation is to prove valuable.
Proceeding now to capitalize on the definitions foc a and f3 in Eg.
10-115, we can rewrite the matrix of Eg. 10-114 as

f3
lim _ -L" e}6
nm

Sf

+ L2

r}8

f3cc 1

a- 2

{3a- 2

a-(n-l)

(Jet-(n-I)

f3- J
f3-l a

p- 1a2

a(n-I)

,8-l a (n-l)

..

.. .

..
,

'.

>

'i,"~'
"

,B-(m-lla(n-ll

>

j-

-2

I0

0 0 '..
0000
eJ6

~ .~

0
0

. ' ~ .. ~ .' ,. , ~~}e

'''S',
;;' "4

"

In terms of these variables the volt-ampere relations ace

I= ~~N + p2'~N i

lop"

[ ( op..z.
rJ
MN

... ..
",

. t'+f!'

,to......."
..,

"

P!f':.M

<7I.MM

p..z.(orr
MM

......""

'
~N

~.~.

(10-119)

tiM

~~i

. l.1t

,'

Substituting in the volt-ampere relation the symmetrical component

parameler matrices derived in Eqs. 10-lO8, 10-lO9, 10-110, 10-111, 10-117.
and 10-118 yields the set of volt-ampere equations 10-120,
1
The volt-ampere relations in terms of the symmetrical component
voltages and currents are, except for four equations

:~
~1'j

"

= I, j = (N

(10-117)

v~ -= (R

1
j
.j

,if"

- 1) on the rotor

Carrying out the matrix products

'"
I')"

;:.\ (10-121a)

+ pL::O)i~

r)~

= (R S +

v~

= (R S

v~

(R'

+ pL;])i~
+ pL~~)i~

V(N_I)

[DS
~

~p (N-I)(N-nl(N-I)+P

i ... '

.)
~

= (M

Stator Volt-Ampere Equations

'1

ii

- I) on the stator, and k = 1, k

in diagonal form and are independent.

indicated in Eq. 10-120 gives:

"j

00 .. , 0

,1

',t

(10-116)

.. i'Qlr~

roo

. ., ,.

.9/NI :t'~;"!JiJ M

-~Ls,

I v-;" I tiM I

'liM

{3-(",-I)

(10-118),

,.:

",,'

'.

'.'

r}' 0 0 0 0 I

?.I"r.,.

f3-(m-l)

;II

All the narameter matrices have now been determined for the transformed symmetrical component currents
~':V e"~
and voltage$

, ~

With the definition of 2';:" given by Eq. 10-116, the matrix product
shown in Eq. 10-112 which defines the symmetrical component mutual
inductance matrix between rotor and stator has become surprisingly
simple, namely,
!'NM

{3(..-l)a- 1 .

,8(m-J)a-(n-l)

- -z-L : : - - . : : :..

{3(m-\}

a-I

s,

pL';~)i: + p v;mLsrej6i~

(lO-121c)

,t..

(10-121d)

].S

LJ.f

8'"A,.""

';~J. (10-121b)

""'<;

(vnm
-2- Ls,e -J6)"

I(M_t>

(l0-l21e)

J
S86

ELECTROMECHANICAL ENERGY CONVERSION

'l..0

'''}~.'

1/"1

' '"'

,,_~

,
,._~
~

"_'"'

._~

._0

.!'

.....

00001~~

h:::E

0000

...

rt, ''f'I"

!.

~,

0000

000

,'~-.

0000

00

0000

......

,.

O\!

il'

->I

.,

0000

10

f
~

'00

C>.
+000

0000

0000

~
0

0000

;.J-f~ l

,<>

~"

-.l
::1
~N

0000

::6

,"-".,"",

,.i.....

10

-,"
~"'.1<f ..

0000

'0

0000

.00

10

0000

C>.
+000

10

0000

000

00

"'/"';j

I . _"
"'1"""

"ll'"

'~

~~ ..q""l"

.....

~t

":.,

.~

JI!'

~,l"

, t,e ,..)

I ..,....

:s

~.';/'.

"~.
./

..

f.;,.

:-

0000

"t

OOOC:>

10

+ pLl)i~ + P

= (~ + pL')i'.. + P

, =
v+

v':.
0000

7
v;m

L"e)ei:

(l0-124a)

i,-

(l0-124b)

e -j8"1+

(I 0-1 24c)

LSTe- j8

+ PL')"1+ + P

hJ

v;m

L ST ei8 i'.-

L = L~~ = Ltfv-l)(N_1)
L' = L~~ = L(~_l)(M_I)
S

' . f .

II

I~ ~ ~'"' ... J Ih~ h~

= (R' + pL')i'- + P

vnm
2 LST

where by Eqs. 10-110 and 10-111

I'l::

h';

(R'

';'4

C>.

0000

(R'

?it!

,(Ii

::1:..

">1

Zl::
::8

t;

if_

"

I::::

1,(.,

'\

:r

(10-123)

= v'+
= v'_

with analogous definitions for the currents and impedances. Writing

volt-ampere equations for the + - components on stator and rotor using
Eqs. lO-12Ib, e, g, i and Eqs. 10-123 yields
if+ =

Zl::

ill.

V(M_l)

Zl::
10

(10-122)

v(N _I) = V"_

C>.

= (RIcIc + pLlck)ik

v~

"

0000 1

"i';1l""''i,

~;,

I(N_ J}

These equations. of which there are (n + m - 4), are linear, independent, and directly solvable. The remaining four are a set of four
simultaneous nonlinear differential equations. Following the general
procedure established in Sec.. lO.l.1, Eqs. 10-24 define the positive- and
negative-sequence voltages as
v~ = if+ -~
:i

,i

(vnm
-2- LSTeie).,

.
]I(M_
., I) + P
+ PL"(M-I)(M-l)

Vk
0000

I'
" ;I "".
..

[R'

The volt-ampere equations are written in terms of the symmetrical

component variables in Eqs. 1O-12Ia-i. These equations show clearly
that all except four of the voltage equations (v:, Eq. 1O-12Ib, vtN_I~
Eq. 10.12Ie. v~, Eq. 10-12Ig, and V(M-O' Eq. 10-12Ii) are of the form of

"

; ..,,'t

(lO-12lh)

(l0-12l i)

Q:::

~",

1 '1-

V(M_O

~:::

." "1 ."

C>.

t
~

t"
."
it

l.

..,

,I'

(l0-12Ig)

-.l

t'

".

,.,'

Q:::

~..

~!::

ll'

(l()..12If)

It;.

:'

= (R' + pL~~)i~

vi' = (R' + pL~~)ir+ p(7 L:'e-J6 )i:

v'=
(R' + pL"\i'
2
221 2 ..

.......................

~~

C>.
+

v~

'-"

C>.

Rotor Volt-Ampere Equations

:::E

h'"

GENERALIZED ANALYSIS OF THE nm WINDING MACHINE 587

I
.j ~
~

I-

(l0-124d)

ELECTROMECHANICAL ENERGY CONVERSION

588

-j

The four volt-ampere relationships of Eqs, 10-124a-d arc the only voltampere equations expressed in terms of symmetrical component variables
which arc functions of the mechanical angle J "" O/(P/2), and hence
should be the only terms related to the torque or energy conversion
properties of the machine, This will be confirmed in the next section
when the torque is evaluated in terms of symmetrical component variables.
Comparing the volt-ampere relationships of Eqs. 10-124a-d to the
symmetrical component volt-ampere equations derived in Eq, 4-194a for
the special case of a two-phase machine shows that they arc identical in
form when Eq, 4-194a is restricted to the case of a smooth-air-gap machine
(L;;; = 0, L~2 = 0), Since all the other (n + m - 4) symmetrical component volt-ampere equations except the + - components for stator and
rotor are linear, and nontorque-producing, it follows that the symmetrical
component transformation reduces the nom machine to an equivalent
two-phase energy converter plus a set of independent networks (n + m - 4)
in number. Once the n-m machine is reduced to the equivalent two-phase
machine all the analytical techniques of Chap, 4 become immediately
applicable.

".

10.2.7

GENERALIZED ANALYSIS OF THE nom WINDING MACHINE 589

Substituting from Eqs, 10-102 through 10-105, the electromagnetic

torque T, in terms of symmetrical component variables becomes

T =

(",.

":

~=rd'NI

5J L0

OJ ~

(10-125)

I.9lJ ~

'f,l'

;'

.,

T= p2J.p + pac/> - t

Ii~~ I (..~~ I 81>_o !i'''S II ~2""

01>
\ ~"
Ie
:.p!i'~ I o.p8
lin

11m

~x

...

rorr

.z, mm

- p2Jrf>

+ pa<{> -

T.

'r

WI

Te =

l"2"

I ' '. I 'r*,

(N,

(M,

G> Q '!::>. ~"

N,

!=0"'' "

M,

8q,
i1 !/!"
11m
.'

.~M ~:"
N
-

() (mr ill'
81> -L mm"7~ M

(10-127)
'r

tM

and

0)0*
_ r1b-1
V<1M, -;;;OM

(10-128)

rGu

rs

&f, .;z NN

I o</> 2'MN

8 CDsr ,~.
II Oct>..L
NM ~

(10-129)

i3
rr
"
8.p.!i'MM I (. M

Substituting for .!i'NN' !E'MM which are constant matrices and !I";'N' !ENM
which arc functions of IJ = (P;'2) and performing the indicated differentiation and matrix multiplication yield

T = P
4'

[,,,* -2vnm L'fr( ')e

1+

jO'r
1+

's*
1_

i' ,

vnm

'r
-2- LSI( -;') e - jO1_

vmn
+ 1_
"* vmn
+ 1'r*
+
2 L ,,(_ J')e -j9's
1+
2 LIr (.)
) ej8'S]
1_

:!
~ .'

'~'..

!:;

jl;'
(10-126)

_,I
..z. mn'w N

:-*"

a</>

(/Jr,'

1.9/*

established for all symmetrical component transformations in Eq. lO-6

and the fact that d Nand ,'!IJhi have constant clements so that the derivative
8/0</> can he interchangcd with thc matrix multiplication, the transformed
inductance matrices defined by Eqs. 10-110, 10-111, 10-117, and 10-118
reduce Eq, 10-127 to

()

,"''/;, = ,'JI N1

d'

where Nand EflM are symmetrical component transformations of order

Nand M respectively (Eqs. 10-106 and 10-107). Since the symmetrical
component transformation is complex, the torque expression must be
written as i,*(o2'/o(.'just as power for complex variables is defined as
iiv. This convention was initially introduced in Eqs. 4-13 and 4-18.
In terms of phase variables the torque equation 10-IOlb becomes.

d*N, 3-!.e"".f:J!
(l,p /III N
,

Using the conditions that

The torque expressed in terms of phase variables is given by Eq.

10-10lb, The transformation equation for the current is, from Eqs.
10-102 and 10-104

.~~:;

Ils*.
I l~!1'I
_ N, ~

,'M>*
V<1 M,

~\

Symmetrical Component Transformation of Torque

1.
"2"

1:
0/

<;-~

_ !:. vn;1
'r*'s)' )9 _ ('"",.r + 'r.
2 L'''( 1 + 1 + + 1_
1_ ;e
1_ 1_
1 + 1+

- 4

's. 'r

's ) ' _

je

10]

(10-130)

Using the identities established in Eq. 10-30 or Eq. 4-181 gives

V nm L'sr}'[('.'.")
)9
T. = !:.
2 2
1+ 1+ e

('S 'r",) -Jl/]

1+1+ e

(10-131)

Equation 10-131 shows that the torque is only dependent upon the
positive- and negative [I and (N - l)]-sequence currerHs of the stator
and the positive- and negative [1 and (M - 1)]-scquenec currents of the
rotor, Furthermore, the electromagnetic torque T. for the n-m phase
~ /~,

i'-

/'~'% . :'

".,t.-'\;,

~,

;"-\'::~~,

..

~!'

590

~ 4~

,'-i"''',.,.

...

~-"'.

..

;1>";;,#.h

'.-

"

.. _~.1

.......

-,

machine is identical in form to that of the two-phase smooth-air-gap

machine given in Eq. 4-194b. Therefore, the only quantities which must
be considered to determine electromechanical energy conversion properties are the positive- and negative-sequence currents of an equivalent

Equation 10-134 has some interesting aspects. There are three classes
of terms obtained by subdividing p(t) as follows:

two-phase machine.

Physically these three terms are easily seen to be

.. '
... ,\>;-.

.,""",,,,~.,"

3]

pet) =

= \

iN~

i:J, 1 sI'N1*sI'N
,0

; :~:'I\I'

= I (;~~ 1 i~~ I
~

M,

..

'''\' "

Substituting for v~ and

,~

'~I'

-.,p

3.

+ <J,v'M
.= [io*v~ + i:*v~ + i2*vi +
+ im_l)v(N-l)J
+ (i(;*v(; + i~*v~ + i;*v2 +
+ i(,~_I)V~M-l))

pet) = (;,~v;'

0,(1,

\ 14

(10-132) ,

Performing the indicated matrix multiplication, the power becomes

".

l)M their values from

pet) = io*(RS

+ pLoo)io +

i:*(RS + pLmi:

ST

L e

(10-133)

J8

e";J,R ri M

+ i'J,P:?':tMe'M

+ [i(~ _ I)pi[M_ 1) +

'.:

(10-136);
!

i~* pine - JO}

(10-138)

Pr(t) is the instantaneous power delivered to the mutual coupling field

between stator and rotor. This term represents the electric coupling
between stator and rotor, and as such it must supply all the power going
into electromechanical conversion as well as any change in mutual stored
energy.

)i~

+ i2*(R" + pL'{z)i!], + i~*(RS + pL;S3)i~ + ...

'~

. ,-' 'I'

~\

.,'* (R
s*
/ V nm L" e - J8) I(M_l)
'r
+ I(N_I)
+ PLss(N-l)(N-I) ).sI(N-1) + I(N-l)P\-2-

10.3

",!ri';:

,.,
,.,.,<;.1,,,,

.~'

1if""~'

tl..l

"j,4)-o'

+ ir*(Rr
+ pI")i'
+ ir*(R
+ pl")i
\l
-0\) \l
1
~II 1 +
r

.j~;i'
./-

(R r
I(M-l)
'r

).r
PL"
(M-I)(M-I) I(M_ I)

~~t',,"

sr J8
i'*p(vnm
I
2 L e- )i'(i",,:,
I ~

.,
+ ,]. ,(R
+ pL"22)1'r2 + .. ,

(V

(10.137),

:if

Eqs. 10-121 gives

+ Wp( V~m

":,

s
] jB
V nm L,r,[ s*.r r.
Pr(t ) == -2- 1 I, PIJ + '(M_I)PI(N_l) e

, ,
:t'"

~.

,,

(10-135)

FL{t) is the time derivative of the instantaneous stored magnetic energ/

in the self-inductances of the stator and the rotor. It does not contain'
the stator to rotor coupling terms and hence is only the instantaneous
power delivered to the stator magnetic field and the rotor magnetic field'
independently. It does not include the power delivered to the mutual
field between stator and rotor. For the special case of balanced polyphase
voltages and steady-state operation, the term FL(t) can easily be shown'
to be zero.

I:16- I*~-IM M
v
'

~1/S
--.!!.'V

PL(t) = iN~p2'~Ne'N

2.

I I"
~' i

+ Pr(t)

PR(t) is the power dissipated in stator and rotor winding resistances.

~l-'N

PR(t) + FL(t)

PR(t) = lN~Rse'N

l.

Symmetrical Component Transformation of Power

__.t.:

.,,...~.

The power in terms of the symmetrical component variables can now

be investigated. The definition for power p(t) given in Eq. 4-13 establishes
the total electrical power as

,-~

pet)

'

10.2.8

GENERALIZED ANALYSIS OF THE nm WINDING MACHINE 591

ELECTROMECHANICAL ENERGY CONVERSION

;""

'r*
nm L sreJ8)"I(N-1)
I(M_ op -2(10-134)
i

if

~'-.

Summary

The nom phase machine has been analyzed using the general symmetrical
component transformation and it has been shown that for purposes of
studying torque production and energy conversion the nom phase machine
reduces to an equivalent two-phase machine expressed in terms of the
positive- and negative-sequence compont.:nts for st<l~or and rotor. With
this fact established the analysis of the nom phase machine is complete
since any further investigation of the eq uations of motion has been treated

592

- .
~

S=J,
-!...J ,

aln'

, ., ! ,
,
I

S' A -,

.:1

ti2

l-a.:3-J a- 4
.-3 I <i6
~ _':..._..)-I

~
~b

x,

2. 1
Xk

-'-~-'f ..:_- ~-- ---: .

I

"t.

2) ",2(0-1)
i<t(n-I/ i2(n-l)

a:

I
I

-I

-I

,
-,

a: 1
",4

a1

ia~6

-\

I
I

_a:!i":!!l~:2In-1/
(i.3:n'.~~l:~1
. I
-I

.!.~

'ilonln:tI'll\,.1)

-I",(n-')I'I) ~~
I

ai' Ixn-/ a(;;:ifrii

'"a

.<:

a.
,

ml

c:

~/Z

r--J=.R.
~.E.
~

x,

J-'~.J "'2.
~
~o;20 i (,(lS40
lZ!1
GDS

3S \ coHo

'jJ1ll
; 1fl2
0

I-I2/?

''"In-I)' ""1(n-1)6
a!,

fl/z

-I

"'16
'106 S 'I'H

o I

x.

cos2'

is''l{n.,)5,,~

(J

-I

to'6'

1m
(j)~~
'o'l\n1)1

I-J2!2

12/1 :-/2/2
o

'0,)6

':"'_I_~

S'1l2b~T.sjIl48

f---~-~

,In,S

,mo

3&

Sin

r!l12

tOS 5

m f l l IIJ~ l-I'l~1--:
I
1-i...J_,-"SB::f'4S

......:.-1-.

I
;J1",R,

I J1/1

~---i--~
.
o ,In I

I

,in 4t

$I"~~ ( 511t

71/2

}';i'~

~
~

"nlln-I)!
I,in(n-I)I

'I

;':1~'1,

-T

12

I
I

I
I

-!-l-.

Ij

-;

!-.
J
j

-J

X<r~ X ,
I~

Gener.',zed

two-phase

campanenl5
~-"\h

''i

'.

' !.

. J2 .
.

1-'-

J
,

-J

,
"41,.

'

(fl. Jz

'.').-.!

I I

l-:-P -:-+-+- C-J

j
I

'I

.:t"" .

<B"~J

II

I'+,
~C- '--'--

-J

r#;:'~t ~'tr

~T

-j

'~"}'-' '.

,.

;'\

'"v

'"1';'

'6

1-'~

~~

~I

,j

,
"

:t.,"

~.l,

:"t ... \/'\

.'4-

.i}>
' 1:

Flg.10-9.

tll"'lof"lC!

'L,

II
.!'\;'
j "i-.

;;,;,41;

e--:..- Xs
1--_ Xn-l

t.:!

Symmetrical components

-I

?;"

XX '
-'...!......L

'a 3In-1 "lin-I)

..,

:~

"

-;-+:('-~tr~?j

;!!n:!1 ~(n-I)' ;

-I

~
.-!J_

'I

~f,n- 0-1

''-In-2}j..'''('~

, a(n-ll a2In-/
Taln-I) -a21n-l)

.'

--;-f'-. )i . I '.

-:---:----:(n-Ix'
''"'~~

-,

OF THE nom WIN DING MACHINE 593

GENERALIZED ANALYSIS

ElECTROMECHAN1CAL ENERGY CONVERSiON

Block diagram representation of n-phase time-in variant linear transfonna-

NQles:

" .. el<J and S = 2.,,/n,

See Fig_ 10.12 for conventions on block diagrams.
In each of the above n x n matrices, the center, or (n/2 + l)th, row and column exist
only if n is even_
Excepting Xo. the Xs and xr variables always appear in pairs, ~ymmelrical about tho
(n/2 + l)/h row.

-"'-,
.:vA
.'it;

!.

~.

'>

_d.:--::;"':- ' -

.7 :.

~i-.-Y ~

.. }~,~.. t~:~ =
.

;~ . ~ -:.~ .. ::

-U~i,*t~~~'f~:

., -

.....

,.

..

r.

" ;

_ ;,i-j

;-

...

.-:::-:.'~

::;,

;.~

;,~ ~

--~

\.,.

~r;

-.,

_~

it

~-

~~-

ff~ f ; ~: ~

~t

~'~,:~~.'

H~ r:~-

-. - - - - - - -

>~

t ..~~

..

'"
<oD

4"

2"

. 2"

4"

m
r
m

V3

1
~-

e lT elT

~
I

f--

1 e lT e lT

e- j8

I
I
I

Symmetrical
components

f--

b
:.I

e J8

,.
()

J:

O,f,b

0,+,-

a,b,c

Rotating - field
components

Z
()
)-

Vi

lfE

e )

e J'[.j

e j!'!.3

1"

oo

l+

r--

"
-6

f--

eJ~

n
)C

m
;1:J

Cl

-<
()
~ 0
g. Z
3
<

e- jO

eJT

0-

;1:J

::>

V)

'" o
z

I
--i

:T

"

~I

:k

1
!

'1-,:-1-"--

Y2 _. --_. --- 1
I

~11
Yi

_.1
2

11--

-:;rz

"

._---

f---- .

, -;

r~f~r~"-~'~~'~f '-~(;r"~'-;
-.(2

Y2

"0
:T

--.
I

----

.. _-

.i

I
I

cas

2"
I

-2
V3

-2

I
I
I

::0
co

-sin

cas

~
~

co

::J

",

0'

--J3

2" --2

I
I

_v'

"':

,~,

I
I
I
I
I

I;TTI
cas

Cross-field
components

-sin/--

sin

cas

O,d,q

m
;1:J

)-

N
m

'3

)-

I
I

_1

:l!-.

~:;

sin~

-T

I
1

Cl

I
I
I

Yi

I
I
I

IV3

V2

-~.-

a,b,c

Yi

_._-

-2 - 2

Two - phase real

components
0, a, {J

-.{2

...

:oJ

)-

r
-<
en

Vi

0
-n
-l
J:
m
::J

Fig. 10-10. Block diagram representation of general three-phase transformations.

~
Z
0
Z

Cl

,.

3:

()

J:

m
V1

<oD

V1

596

GENERALIZED ANALYSIS OF THE nom WINDING MACHINE 597

ELECTROMECHANICAL ENERGY CONVERSION

AClual Powtr Comronenl!l

:~~;

t,'

1~lIo

Fig.
To,quo

Vollllmpcre

.;"", L
II' + pL'

'd

R' + pL'

p,in

\Inn; L-

.:
-

vi

2
P

vn,i'LIII'

v~

P=;'"
_ vim, (I'

r'"

p,in

. 'P
P Sin
2 '"

f 1-

,/;"' Lit'

/I'

+ pL'

stator

I;

l';

R' + pL'

-2- L" P

vnm

II

L'i

,;--

~~

L"

.;n;;; L n p

-2-

-Ct:..
2

/I'

+ pL'

Real component.

~, .. "I

II' + pL'

;~

.....

;~-""

t ~

",-Ir"

:>""'''

v',,;,; L"
2

()

R'

ROlor referred

Rotor referred

to stalor

to SI'Hor

+ pL'

;~

i;

Pvnm
T,. '2 -2- L '"

(I:.2 ';nm
r.")
2

(iji~

- i;i;)

0',

R' + pL'

t"_

Il' + pL'

';nn'L" ( _ i.!.~)
2
P 2 1\

()

j(i~.'l- i',i;)-'''

'i!

-/; j I!,...
"'h ' 1. .:. .,
,"lI,'"~1,','"

v,

'~

;"],,' '.:

.;}

-2- L p

I~

iF

R' + L'(p

iVnniL"(p+~~)
i 2

;~

v/iTm "

-~~)
R'

+ t:

(p + ~ ~) I

,'!", ',:,'-

.1

..'

'.,"

i,,~ ~,~
.~;

'i",

i;

i~

':omplcx componen..

....

',' i,

,.' if:
1 '

_ . '

",

viim Litp

-2-

'f'>

~-r'

, .''''\Or .r~~

i;i~) sin ~f' "

;'

'J'f

vn", L"

v',

....--.
T _

II' + pC

---r

Iro]

Ii

v;

-(i','/:)j.

if>

Rolor rererr~
to slalor

~LP

~.

-I-

'/m" L "

p<'r"

rcrerrecS

v'"", L"
2

i~

"(0

I.

11.' + pL'

~L-P

O:i~

i;

l'~

U;I.

/I' + pL'

()

0'

v'_

P
pco, 1. '"

10

[(1:11 - I,,:) co>

,.

W~i~)j.IP

vn;;; Lt"

ROlor

7;-(I~?L')

co,

II' + pL'

1,1,

/WI!-

r'" P r'" I'

11.' + pL'

~In

v;m L"

i:

vtim L"
2

Volt-ampere

_ vim'L_

"

pco~ ~tb

VOlt-ampere and torque equations.

i'~ .~ ....

L:

-~

.. ~ft..'k'

"dI
~~

0/<'.'
'y.,.,

598

r,.)'>
,Ii\;'"

effect the change to real variable s.

phase
There is one word of caution about the simplifi cation of the nom
analytic
for
chosen
model
machin e which must be observe d. The
similarl y in
purpose has identica l impeda nces in each stator phase and
If instead
d.
restricte
not
and
general
is
each rotor phase. The excitati on
they are
or
etrical
unsymm
are
nces
impeda
the stator or rotor phase
ns
equatio
m
+
n
the
of
n
reductio
the
load,
etrical
unsymm
connec ted to an
the
by
effected
be
longer
no
will
ns
equatio
ng
produci
to four torquetwo-pha se
symme trical compon ent transfo rmation nor by the general ized
no
general
transfo rmation . For unbalan ced impeda nces there is in
neous
simulta
1)
guarant ee of reducti on in the comple xity of the (n + m +
possible
nonline ar equatio ns and the problem must be solved in its worst
cases
special
no
but
form. For special cases simplifi cations are possible

",,'';''If ....
e"

t'~

'

"'".

~,..,'

""', 4,c,..'

.~

\' '

~ I'

;'j1
tI. I ,1

ii!

'

~.",

..........,

, ';"'j.I
"

,It,

"~I'~ i~,,

">'

~.

z- -0 -,

.....

~I '

,,;

A" = JI

d~,~,.. ,~

matrix
10-1. A three-phase transfor mation of currents is given by the
equatio n

,; - :Je.,;'
in which

J %U

,; = ~&
I,

.."

:+":r,f'; .

-I

~
. t:_I- -- Y
-

r -

'j

'.v=

,.. .. V&

....

"

~
~
.~
.~

in terms of :!t.
s of matrix.)l(.
(b) Carry out the indicated operatio ns in (0) to get the element
ase
10-2. The volt-ampere matrix for a four-pole machine with a three-ph
ents
compon
rical
symmet
neous
instanta
of
terms
stator and a five-phase rotor in
is given below:
'S
v0
10

l I .\.

v1
v~

Vo
VI

;~i

v~

')':i

;~\

vi

V4

,
F

where !r

"~

Fig. 10-12.
1010, I t)... I I.

RInck

di~gram

'J=.

representations of malrix operatio ns used in Figs. 10-9.

RS + pL'o

;:/J '!I = J
~'36 .f'I . A ' " J

-'-1
." .'ill

VI

v = .)1(. v'

i:

o
.'.,

_
~n~n~
J

Vo

= V,

v'

find the matrix" ," of the voltage transfor mation

",t,

. .v .... = y

H'f

II

V,

ftj,

';' = ~.

1
-1

in which

'i"

:j J

(0) If the power form is to be invariant, i,e.,

e'.v = i;* v'

'~M:~. '.v ....,

,,~

PROBL EMS

" .... ~.l\t'\'"U."

..

599
GENERALIZED ANALYSIS OF THE nom WINDI NG MACHINE

will be investig ated here.

those
To put in readily availab le form the results of this chapter and
the
e,
machin
phase
nom
of Chap. 4 for the case of a smooth -air-gap
in
given
are
motion
of
general transfo rmation and resultin g equatio ns
Figs. 10-9, 10-10, 10-11, and 10-12.

" . r'\''''d'~

...., !,

~~

of the nom
in conside rable detail in Chap. 4. The equatio ns of motion
n to +additio
in
fb
or
dq,
af',
of
terms
in
ed
phase machin e express
among
ips
ationsh
interrel
the
from
directly
d,
obtaine
be
can
compon ents
- 4)
In
+
variable s establis hed in hg, 4-2X. The non torque- produci ng (n
of
terms
in
terms, since they are il1llcpendent, need not be express ed
In
result.
can
other variable s since no further simplifi cation of form
rather than
some instance s, howeve r, it is desired to use real variable S
real and
between
ips
comple x variable s and in this case the interrel ationsh
used to
be
may
comple x variable s establis hed in Eqs. 10-59 and 10-60

I ....' ..

ELECTROMECHANICAL ENERGY CONVERSION

0
0
0
0
0

0
o
o
R' + pL s
RS + pL'
0

p:i[fL"e- l
0
0
0

'S

II

-'S

12

-.,

10

=;?l'-.,
11

-.,
12
-.,
13

---

i4
-

0
0

pY[h"e l
0

0
0
0

0
0
0
R' + pL'O
0
o R' + pL' 0
0
R' + pL~
0
o
R'
0
0
0
o
pij}U'e l1 0 0 0

0
0
0

0
0

p v'215L" e- I

0
0

o
o

+ pL~
R'

+ pL'

_.
600

j
~
~

ELECTROMECHANICAL ENERGY CONVERSION

(b) Write the transformation matrices by which the voltage, current, and

1.
~

j ...:h6l

"

impedance matrices are multiplied to change the above volt-ampere equations

to fb components referred to the stator.
(c) Using part (b), find the volt-ampere equation in terms of fb components.

10-3. (a) In the unbalanced three-phase circuit shown in Fig. IOP-3, write
the equilibrium equations in terms of the variables I", I b, and I,.
(b) We wish to solve the problem by phasor symmetrical components.
Therefore, make the following change in variable:

-~

.. =

I . ;~. :,.
..

,. '-,).

I>

~: ~:. ~.'
~;,.. ,..~~ t ',.

:l "........ ,-t'.
I

... ,.., \

I'

~ I I:
~,,.,

I..

. '~"

,i?

!"I

,.,'I
i

.. ~

I.
Ib =
I,

~m ~

1 1 1
1 a2 a
1 ex a 2

10
II;

E.
Eb =

E,.

I 1
a2
1 1

~I
1
-

:r,

where IX = e I2,,'3. Show that the equilibrium equations in terms of the phasor
symmetrical components reduce the circuit to three independent, single-loop
circuits. Draw the independent circuits and solve for the phasor symmetrical
component currents.
'."

"'i"

'I

rl

).

,)a

1,"I;i..',

II

-."'
5
i.'.j(

..'

" .'~'f

.'

.. ~.
,<.;.:'t

51!r....

E b .. 10/-120
E c .. 10/+120

9l'.:..... ~.l
~

2~ - [-~.4

.,. ".

~.J

0
0.1
0

-0.4

-on '~ [~.4

-0.4 ;

-0.4

v~1

.....

L=-0.4

---l

v~

-S

-0.4

-o.~

-0.4

-0.4

, t,) '"

,.

v' ;,

1)~

.,
.,

I.

e'~

v',

~--v'

.f~:';~

Fig.IOP-3.

is'

,1

0.10
0.1

(b) Find the transformed impedance matrix for the two-phase real com.
ponents. Also find the components of the transformed voltages and currents in
terms of the components of the actual voltages and currents, the actual voltages
and currents being specified as

.,

Note: All voltages and currents

are III rms values

tl;.

'~'I'

The stator to rotor mutual inductance constant is L" = 0.8 henry.

(a) Obtain the rotor to stator mutual inductance matrix '?:I;. and the stator
to rotor mutual inductance matrix 2';:n' Leave IJ as the only literal factor in
these expressions.

In

~1

0
0

r:

'1;~

Ea =

~l oo OJ

-0.4

I .

+ 'E c

II

rtf

!.!'-

! .l.

1:!.
i

;~

E2

'r

t'

~-

Eo
EI

10-4. Given a three-phase machine in which the parameter matrices are

(with resistances in ohms and inductances in henrys):

~ nn

r .~.~.~.:

HI I !

GENERALIZED ANALYSIS OF THE nm WINDING MACHINE 601

(u) Find the torque in terms uf instantaneous symmetriclli components.

U.t

rl<l: LJI) Y

Ib

.,

I,

.,

I.

'r
~.

"

'b
'r

""~'"
'

1""4

.,'~ .I~ ~
!.

1.1,.

til': .,; . ~,:,

Ii

'~

(c) Determine the actual phasor currents I I b, and I, from the computed
phasor symmt:trical component currents.
(d) Show that the power (P + jQ) in terms of the phasor symmetrical
components is

+ jQ =

3(16Eo

ItE I

liE 2)

Calculate the power from the above equation and compare it with that
calculated by use of the actual phase currents.

'.' '.. '

~:

',.,~
.
t.

';
I
.~ 'p

\"

~ )

(c) Find the transformed impedance matrix for the symmetrical components.
Also find the components of the transformed voltages and currents in terms of
the components of the actual voltages and currents.
(d) Verify the results of (b) and (c) by performing the direct transformation
from two-phase real components to symmetrical con.ponents or vice versa.
(e) Further transform the results of (b) to obtain the two-axis (dq) components,
referring rotor components to the stator. The desired results are the transformed
impedance matrix and the components of voltage and current in terms of the
two-phase components obtained in (b).

602

ELECTROMECHANICAL ENERGY CONVERSION

(f) Further transform the results of (e) to obtain the rotating (fb) components, referring rotor quantities to the stator. The desired results are the
transformed impedance matrix and the rotating comronents of voltage and
current in terms of the symmetrical components obtained in (e).
(g) Verify the results obtained by transforming directly from the two-axis
components to the rotating components or vice versa.

E R

E V E N

,c.

'It;:

10-5. A smooth-air-gap machine equipred with three stator and three rotor
phases operates in steady state as a synchronous motor under the dictates of the
following impressed voltages:
Stator-three-phase balanced voltages

Space Harmonic Analysis

(wt + S + 1T) -.v: I = V cos (wt + a + 2;)

V.. cos (wt + a + 1T + 2;)

Machines

.In

Vm cos

1T _

.\

11.0 Introduction

where 8 is an angle function of the load on the motor.

Rotor-a direct voltage is applied to phase a; phases band c are shortcircuited

-Dr

v~
~,~

= 0

v~

,i

The motor revolves at synchronous speed, i.e.,

= w. "~;.

. '.

(a) Find the torque-producing voltage components as expressed by

:'~;;t4

it
~,,;.~

Vd
S

Vq

to'

= 11~

v'fJ

i.e., in terms of two-axis real components with stator quantities referred to the
rotor.
(h) Obtain the volt-ampere relations governing the steady-state behavior of
the synchronous motor in terms of the two-axis components given above. The
.parameters of this symmetrical rotating machine are
R S , R', LS , L', L sr , and 8 . q .
(c) Assume negligible stator resistance eITects (R~ 0) and solve the equations
for the steady-state currents:
"

'S

14

e,= i;

.r

fa

i~

-:':,.~

.,.1

! ~

':

,,,.~,

The I'rcvious chapters dealing with rotating electric machines have all
assumcd that each phase or coil on the stator or rotor of a machine was
sinusoidally distributed in space. This assumption greatly simplified the
problem of evaluating thc electromagnetic fields in the air gap of the
machine. The re~ults obtained with this restricted model are quite general
and the analytic techniques introduced can be used for machines which
cannot be characterized by single sinusoidal current sheets on stator or
rotor. One important class of machines in which the sinusoidal approximation is not valid is the commutator machine. To ohtain the energy
conversion characteristics of commutator machines the actual current
distribution on stator and rotor must be known. Even for synchronous
or induction machines it may be desirable to take into account the effects
produced by nonsinusoidal current distributions. The techniques for
doing this following the general rrocedure develored previously is one of
harmonic analysis. Given any arbitrary space distrihution of current on
stator or rotor. this current distribution can he expanded in a Fourier
series of space harmonics. Each harmonic current can then be analY7.ed
by the techniques already established and the final equations of motion
obtained by applying current or voltage constraints hctween harmonic
current sheets to yield the exact current distrihution obtained in the
actual machine. This 111l.:thod will be developed for a two-phase smoothair-gap machine and then applied to the specific case of a two-axis
commutator machine .
603

.fi.t.

;,
1\.1

~.

., ;fJ'f'

; ...}i,"1'r

-~~.

'1""'('0'10

;.

Two-Phase Machine with Nonsinusoidal Current Sheets

on Stator and Rotor

The physical model which will be analyzed is shown in Fig, 11-1. Both
the stator and rotor are assumed to have two-phase current sheets.
The two nonsinusoidal current sheets on stator or rotor which are
designated as a and b are located in space quadrature; i.e., they are
displaced 1Tj2 electrical radians apart. The physical model will be
restricted to be a two-pole configuration with a smooth air 6;lp The
choice of a smooth air gap is not necessary but it results in considerable

........p.'

SPACE HARMONIC ANALYSIS IN MACHINES

ELECTROMECHANICAL ENERGY CONVERSION

604

:~

.~
;~

,l.

"I
{

, I

I
!

!
ft,

'tit

. ,";..~:

!"''1''oI';'' ,

', .. '

'i 'i~t.

\.~".

,ill~'"

.t: .

'1- ,":'

,,,tL
,1

"
........

,~

r
ll~h

'~

II.~

600........
"f~.)

,!hl>

k,

. ,,/,' '/.//,// ,/./.,'X ..... ' ... ~.....

,--

.,. . ~i',.

-;

.. \

'11

"J\ ";

'"

':'./".-.m

.... 1

;:.

,,,

...

I,.

,\1-.

'~I"
.:

;\

~!I

'I.

t'"":",,,,I""
t,r.;., .__. I ' '

".>;>, ,... , ...

K~W) = K~l

~ar
a8

Ii,

111,

.i,

'f'!

<,..
I

Jl

Fig. II_I. TW'\'!'"k structure with <:yJlndli<:al 'iymmctry on both members .lI1d arbitrary
two-phase t:urrclll ,hcels 011 statuI' ami rotor. U = ekt:trical angle, '" = mechanical
angle.

K(~

+ 17) =

. ,.

-K(</J~

the surface current density can be expanded in a Fourier series.

example, the a current sheet on the stator can be expanded as

;~

,/az

simplification in the analysis because there is then no interaction between'

harmonic windings of different order. If the machine had salient poles
such interaction would take place. Later it will become apparent where:
the additional terms in the equations would appear in the salient case, ;
an.d further comment is reserved for that time.
Assum~ that each of the four current sheets (two on the stator, two on '
the rotor) sets up an axial-flowing current density K when current flows,
in the respective winding. Without specifying the form of the current,
sheet other than that it be an odd harmonic function *

,,~

;~

.,
'~'~

605

cos </J'

+ K~) cos 3</J' + ' .. + K~T) cos 7J</J' + . ..

For'

(II-I)

Before commenting on Eq. ll-I the notation will be clarified. The

superscript s indicates a stator quantity, and later the superscript r will
be used to designate rotor quantities. The subscript a refers to the :a'
phase winding of stator or rotor, and the subscript b will refer to the b
phase winding. The numerical subscripts I, 3, ... , 7J indicate the
order of the space harmonic component.
Equation 11-1 expresses K~(4)S), which is the equivalent surface current
distribution set up by current flowing in the a winding on the stator, as
a function of cosinusoidal space harmonics. The cosine terms result
because of the choice of the origin of the stator angle 4>s.
A physical interpretation of Eq. I I-I is not only useful but is a prerequisite to any comprehension of the derivations which follow. It must
be understood that the one physical a winding of the stator can be thought
of as an infinite number of a windings, each with a current flowing
through it which contributes one sinusoidal component to the total
nonsinusoidal current density. In addition, it is convenient to consider'
that the 1)th harmonic winding is composed of 7J c"i!s distributed around
the air gap in 21T electrical radians which are physically identical and have:
the same current flowing through them and the same voltage across them.
Each of these 7J coils has two tl:rminals, and it is at these terminals that
the 7Jth harmonic voltage and current will be dd'tned. The primary
reason for this concept is that thl:se 1) coils of the 7Jth harmonic winding
will be connected in series or parallel to produce the proper magnitude
of the 7Jth harmonic current distribution for a given case. This will
Specifying a winding as an odd harmonic function of angle means thnt each winding
forming a pole is the same as all others.

606

~.l$ '.

>

ELECTROMECHANICAL ENERGY CONVERSION

= i~~Z~1J

':1..1.".
...

11.1.1

(11-2)

where i~~ is the current flowing in 7]th winding and Z~1J is the amplitude
of the conductor density in a fictitious 7]th harmonic winding. Both of
these quantities are defined in Eq. 11-2 for the a winding on the stator.
Simiiar cxpressions are defincd for thc other windings. Equation 11-1
can be rewrittcn in terms of the harmonic current and the conductor
distribution as

x~( ,{/) = i~lZ~l cos cps

+ i~JZ~J cos 3cp' + ... + i:~Z;;1J cos 7]cps '+ . ..

(11-3)

The expression for Kt(r/l ), the surface current density due to the b
;winding on the stator, will be similar to Eq. 111 or 11-3 except that </I'
is replaced by (cp' - 11/2) because the a and b windings are in space
quadrature. T h u s '}:.'
i

.~~.

:~\,

; t~

Kj,(,p')

Xb1 cos

(cpl -

i) + it cos 3 (.pI -~) + : ..

-"'I

. "'.' n i

.\
;"

iI'

'"

cos "'(4l + X'b~"

:!.)
2 . + .. '.
'.' "
k

, ~ . ~;{""

,a.d:.

(11-1)

;t"

The Air-Gap Magnetic Fields

The expansion of the current sheets into series of space harmonic

components leads to the assumption that each 1)th space harmonic of
surface current density is produced by a fictitious, sinusoidally distributed
winding having 1) cycles in every cycle ofthc fundamcntai space component.
In Sec. 3.2.1, the air-gap magnetic field due to a sinusoidally distributed
winding having n cycles in 21T mechanical radians was found. This result
can be applied directly to find the air-gap magnetic ficlJ due to the 7]th
space harmonic current sheet. The magnetic field can then be used to
define the self- and mutual-inductances at the terminals of each fictitious
harmonic winding using the techniques developed in Chap. 3. Since a
smooth air gap has been assumed. there will be no coupling between
harmonics of unequal order and the problem can be handled in general
by finding the fields and inductances for the 1) order. If there were
saliency (nonuniform air gap), mutual coupling between harmonics of
different order would occur. This wouid not invalidate the analytic
method used here, but it would add considerable algebraic detail to the
solution.
Restricting the treatment to a two-pole machine with a smooth air
gap and using the surface current densities defined hy Eqs. 11-1, 11-4,
1I 5, am] I 1-6. the fields produced by the ')th harmonic are obtained
directly from Eq. 3-22 by recognizing that the 71th harmonic current has
the same distribution in 21T mechanical radians as does the n-pole-pair
winding defined in Chap, 3. Thus the air-gap field due to the "1th stator
current K~" of phase a is

This same technique of series expansion when applied to the rotor current
sheets yields
..' : \
fl'
~j. ~?4
..JIlt 'hIt'" ~
K;(cp') = K;I cos </>' + K;J cos 3c/>' + ... + X:~ cos 7]</>' + ... (11-5)"

HS
Q1J

0'

=_

[(r/o)1J-1 + (r/a)-'I-1]
a, (b/a)H _ (b/a)-H K;~ sin

t', i{. (

K;(</>')
'-J

j;'

':;;,'i'''

K bl cos (<</>' -

i) + Kb3 cos 3 (e,b' -i) + ...

,;;

i) + ...

1~;

r' t

do

,.~

/i,/' d; H' ,.

,+ K~ cos 71 (1/ -

,.
I.

:,.

c:i{

'I

It-'

,....

',i"k"\

(11-7)

f .... l,~'

[(r/a)H' - (r/a)-H]

a~ (b/a)1J-1 _ (b/a)-1J-l K;1J cos 7]</>'

f""

K:

Substituting for the current density

1J its value in terms of the terminal
current i;;1J and the conductor density Z:1J from Eq. 11-2 gives

(11-6)

:The four equations for K:(</>'), Kt(</>'), K;(cp'), and K;W), Eqs. 11-3, 11-4
i 11.5, and 11-6, complete the expansion of the actual nonsinusoidal current
sheets into series of sinusoidal space harmonics. The circuit representation of each harmonic winding can now be determined by the methods
;. .developed in Chap. 3.

7]4l

"~~

601
',"t,<;t

become clear in the analysis of the commutator machine latcr in this

chapter.
To complete the description of the current sheets, define

x:."

SPACE HARMONIC ANALYSIS IN MACHINES

H.t _
[(r/a)'1-1
Q1J - -a r (h/a),! 1

+ (r/a)-H]
_

(r/o)1J-1 - (r/a)~H]
-a", [(b/a)~~1 _ (b/a)-'I-1 i;1JZ:'l cos "1/>'

H:1J.

Equation 11-8isthe desired expression for .

I
I

I
I,

"~"

I'

(6/a)-1J-I i;1J Z :1J sin "1</J'

ri;)

(11-8)
~q. ~

;.'4i'

608

',t'

-l~'
~

ELECTROMECHANICAL ENERGY CONVERSION

...... "',....
,

Fig. 112 in which 'I identical coils have identical instantaneous current~
flowing in them. The voltage across one coil expressed as
'

The field set up by the stator b-phase current it", can be inferred from
Eq. 11-8. The effects of Kt", are identical in form to the effects of K:""
except for a phase displacement of -1T/2 radians. Thus

-~,

,V'f .,

,../.1,

i,\;~)~.

,;"

.'.

'di

v = L dl

"li~

.......

609

SPACE HARMONIC ANALYSIS IN MACHINES

,,,.,f,,,

',.

(1I-f3)

,.,,'...... 11,.

.,

H bS

",""."

'"

= _ 8,

[(r/a)"'-l + (r/a)-TJ-1] . S ,
(b/O)'I-.I _ (b/a)-"'-I 'b",Zb",

SIn

, I

contains the effects of all the other coils because di/dt is the same in all.
coils. Now if the 'I coils are connected in series, the total voltage VI is I

'YJ (</>' - 1T/2)

(11-9)
_

~,~;"".

,. #,..-_1.,;,' -~-

[(r/a)""-I - (r/a)-TJ-1]
8. (b/a)"'-I _ (b/a)-"'-I it."Zb'l/ COS 'YJ (</>' - 1!'/2)

VI

tt"'{_,.'h'

.,,,",.,..
,.~

..

,'

The fields set up by the two rotor current sheets, K~", and K;Tj' can
also be obtained directly from Eq. 3-22. Using the condition that
cpr = +' _ cp, the rotor 7]th harmonic fields are:

~,'

f:

-~

H at)
r

= _ a,

>t. 'r

[-(r/b)TJ-1 - (r/b)-TJ-1] 'r r '

$
(a/b)TJ-1 _ (o/b)-r,-I lu."Zu."Sln'YJ(c/>

-</

:Ij
_

"h

".""..
~~4''f.
:'!'IJ"

I)

'

"ttt ..

'~

~"

,\'"t"'~

..~

fi,; ;.

'\+'~I"

,.

'i+',f:~'.,

,,'_,","t

-Sr

[- (r/W-I - (r/b)-TJ-1]
(a/b).,,-1 _ (u/b)-Tj-l

i~Z~ sin .,,(4)' - 4> - ."./2)

(11-15)

dt

!i !i li

Ii

(11-11)

-(r/W- 1 + (r/b)-.,-1]
[ (a/W- 1 _ (a/b)-.,,-1 i;''Z~cos.,,(4)' - </> - 1T/2)

II I,;;,.

II

II

II

,\

...~,.
~~

==

-s.

"",'.
-t~."

H'

bTJ

"

"<"f'-/

.~,

[-(r/b).,,-I + (r/b)-TJ-1]
(a/W-I _ (a/b)-H i:"'Z~Tj cos 'YJW - cp)

C'

.. ..... ..,..,.

... '

8.

(11-14)

From Eq. 11-14, then, it is apparent that the inductance L can be con
sidered as 1/'YJ of the total inductance of the 'YJ coils connected in series.

(11-10)

-~~~.:

!" :

1f

it - 'YJ i =

""

."V

where L is defined by Eq. 11-13. On the other hand, if the 'I coils are j
connected in parallel, the total current it is
.

, r-'tol",-..-.

"

di
= 'YJL (it

c.

,~,- .

Adding Eqs. 11-8, 11-9, 11-10, and 11-11 gives the total 'YJth harmonic H
field in the air gap:
(11-12)
H." = H~., + H~ + H~ + H~
In the derivation of parameters it is necessary to define the parameters
of each cycle of each harmonic current sheet as a separate coil because
in the analysis of any arbitrarily distributed winding the several coils of
a given harmonic will be connected in series, in parallel, or in seriesparallel combinations. In all cases, the current and voltage of every
cycle in a given harmonic current sheet will be the same. The self- and
mutual-inductance parameters could be defined for each cycle separately;
however, this would add unnecessary complications to the analysis. The
restriction that all cycles of a given harmonic current sheet have exactly
the same voltage and current allows a simpler inductance parameter
description. To understand this point consider the situation (kpicted in

~~f,

Coil no. 1

.~

Fig. 11-2. Illustrating definition of self-inductance parameter for one cycle of '1th '
harmonic current sheet.
"

Alternatively, the inductance can be defined from Eq. 11-15 as 1/'Y} times'
the inductance of all the coils connected in parallel.
A treatment similar to that given in conjunction with Eqs. 11-14 and
11-15 can be given for the mutual inductance between one coil of a set of
." coils in which identical currents flow and another set of 7] coils in which
another current flows.as depicted in Fig. 11-3. The voltage across one
. coil of set 1 is

di t

VI

',I""

11

,"I"

'~:!

"

,,:;,

'---

di2

= L ll dt + L 12 dt

(11-16)

,
I
tt.;

610

~:

ELECTROMECHANICAL ENERGY CONVERSION

At

4?~.
'I,

where L ll is defined as in Eq. 11-13 above. The mutual inductance L l2

includes the effect of all "7 coils of set 2, When all the coils of set I
are connected in series, the total voltage Vlt becomes

= -"]Vi

VII

= "7L lI

di l
dr

di 2
"'11 dt

SPACE HARMONIC ANALYSIS IN MACHINES

;f

two coils in a set are connected in parallel and this combination is connected in series with one additional coil in the set as shown in Fig. 11-4,
the currents in all three coils are no longer identical and the inductance
parameters defined in Eq. 11-16 no longer hold.
As illustrated by the development connected with Eqs, 11-16 and 11-17,
the inductance parameters, both self- and mutual, can be determined
by finding 1/"7 of the total inductances that would result if all "7 cycles
of the ryth space harmonic current sheet for each phase on both rotor and
stator were connected in series. These individual inductances are easily
determined by evaluating 11"7 of the total magnetic energy stored in the
air gap by the 7)th space harmonic currents, This can be accomplished
by evaluating the stored magnetic energy in the region 0 < z < I,
a < r < b, and 0 < / < 21T/T]. Thus, defining (Wm)~ as the stored energy
that defines the apparent inductances for one cycle of the '7th harmonic
current sheet, the stored energy becomes

....Ii.. .

"~

.~t..

;;'.\1,
:"'::m

(11-17)

Thus L I2 can be considered as 1/"7 times the mutual inductance between

coil sets 1 and 2 when all coils of one or both sets are connected in series.

li

lil li

li

ij
1.

U
1

"t

l"t

VI

Coil set 1

(Wm)~

!J,-t -',. _'\1,

'P,

fh~

t!

1~',

'
/7
<;

~"

Ij.

.,

f ..
;
'.

x;$

X
$ ;c.' ,.
0,/

,~

S~t<

'"
'?

""'''''

.,

Using Eqs. 11-8-11-13, the stored energy is

.{

~ t~

bf21T{
-.;
2. 0

'~

;;

.\
/"l~

\'"

I'

The above technique of stating that the self- and mutual-inductances

are 1!"7 of the corresponding inductances when all coils of a set are
connected in series merely provides a simple method for calculating the
inductances but docs not impose any restriction on the generality of the
inductances thus defined, provided interconnections within a coil set are
made so as to keep the currents identical in all coils, For instance, if

,'.\.,

(r/a)-H
_ (b/a)-H]

i:~Z:q sin 7J</>s

(r!a)-~-l]

_ (b/a)-~-l i~Zb" sin TJW

- ?T/2)

- (r/b)~-1 - (r/b)-~-l]
[ (a/b? 1 _ (a/b)-.-I i;"Z;~ sin 7](l - </

(r/a)~-l - (r/a)-"-I]'
.
'S
J
( (b/a?-I - (b/a)-~-l 1."Z"" cos .,,</>'

(r/a)H - (r/a)-"-l] 'J S

[(b/a)H _ (b/a)-H It>,ZJx, cos "1(1/ - ?T/2)

[-(r(b)H

-(r/b)~-l
1 _

- (r/b)-,,-t] 'r r ' ,

2
(a/b)-H It>,ZJx, sm 7](</> - </> - ?T/2)

(r/b)-H]

(a(b)H _ (a/h)-H

i;~Z;" cos "7W

- ,p)

(r/a)H

i~

([(b/a)~-I

r (a/b)"

'.'!

'~

11

Fig. 11-3. Illustrating definition of mutual inductance,

= 1-401

(r/a)~-I

.'

I,:

+ [ (b/a)~-l

t~'

(11-18)

1'1

"~'

1 b 21T

(W)

"0

~">~'

2 f-Lo(H~ H~)d7' ~ = Jr0 "Jr0-'; ~o (H~ H~)r dr d</>" dz

= 1

,.;. J . ; ",~.

Coil no, 1

611

-<r/b)H + (r(b)-H]
.
[ (a/W-I - (a/b)-o-I i;"Z~ cos ''7(4)' -

,p - ?Tf2)Y}r dr d</>'
(11-19)

~"."."."...,..

Jc
""'~-' ,j:., ..

612

.... ~-;-i',.""

'.r

f'~.".>

Performing the integration indicated by Eq. 11-19 and collecting terms,

I]

11.1.2 The Parameters

I]

W) = (is Z')2 1-'07Tlb2[(b/a)2~ +

+ (i" ZS)2 1-'07Tlh2[(h/a)2~ +
m ~
"'I -u~
27]2 (b/aF~ _ 1
b~ b~
21]2
(b/a)2~ - 1

+
,

(i' Z')2 J1-oTtla 2

Ql)

,''t~!."

Q~

27]2

It was shown in Chap. 3 that the energy conversion characteristics

depend only on resistance and inductance parameters, which will be
defined as those for one of the "I coils which comprise an TJth harmonic
winding subject to the restriction imposed by the derivation of Eq. 11-16
that the current must be identical in all 1) coils of the 7]th harmonic winding.
The parameters will be defined in matrix form as

[I1.;-+ (a/b)2l)
(a/b)2~] ('r zr)2 l-'oTtl0 [I + (alb)2~]
+ Ib~ b~ 27]2 1 - (a/b)2~
2

.....+
:'l"

Z' ) 1-'0Ttlab [
2
] cos 7]4>
lu~ a~IQ~ "" ~ (b/a)~ - (b/a) "

( 's Zs "

Stator

~s

Resistance
,.

l('1'''

" zs 'r Z' ) ""oTt/ab [

2
]
(.J.
(la',
~Q'/~ b~ ~ (b/ap _ (bla)"" cos 7] 'I'

'S

tin,

zs 'r Z')
b,/al) a~

Tt

lah[
2
]
---:;;z(b/ar' _ (b/a)-~
/1.(1

cos 7]

P4s,

I'J)

Tt ~

'1'-

Tt/2

(.J.

;,:~

'II

"
I, ~", I-to-

~ 1;-"

~i.,

1 .

'-; , '

2
., zs ., Z' \ J1-oTtlab[
( I~
bl)1bl) bl)I ~ (b/a)~ _

.J.
(b/a)-~]cos
"1'1'

11-20)

~t'; 'ii-"'h'

Equation 11-20 gives the stored energy necessary to define the inductances
for one cycle of the 1]th harmonic current sheet subject to the restrictions
imposed in connection with Eq. 11-16.

1,,0,,1 ..,;
I

,~\

li_~' Ji~.

.'

, ...

.'
' ... '
,,

ltil-

,1' ....:/""'11

....

~, .-l... -t-i... ~"'

,..........Jo,ItA

Resistance
P4" _

Inductance

2" =

Rotor to Stator

1"_~'''-''

MutuQ/lnductQn~';

~~~

l' .

Fig. 11-4. Type of connection for which inductances arC' not valid.
i'
'i'lo<,-""
>~,.+:

, t>;
;, I '"

,'1'/

....!-:",

,,'i

Q~

It should be noted that no terms involving (i;~ib~) or U;,,/;,) appear

in the expression for stored energy. This indicates that the mutual
inductance hetween the two stator axes is zero, and that the mutual
inductance between the two rotor axes is zero; i,e., there is no coupling
between windings on the same member. This is not unexpected, for the
axes are in space quadrature.

;~
1t

(11-21a)

b-qb"

L"~
L:iS
a~l)

,~.

(11-2Ib)

b~l)

L"~

(11-2Ic)

R~'J

(II-21d)

a~"

L~""'"

b~l)

~' L'~

(1l-21e)

~' L"~

(1l-21f)

Q~

L"
b ~l)

:1''' =
Q~l)
"
L,$

b~'1

,,'

\#

l)

~'~},

.IIi:<$if"

~,

"-LO

'~"

(,

::,:':1.:;:

;J

11'

Rotor

, .".;.,,~

v'

a~

b"a"

'. .l,J..;.~

~,

"-ls'

2"-

~_.

...~,'i

Stator to Rotor
Mutual Inductance

C~~

2 ss ,= al)Ql)
~
L'S
.
b"a"

;I)

~:I

<{.I

''';;':'Ti. J

Inductance

1 ,\ '
{il

t",

613

SPACE HARMONIC ANALYSIS IN MACHINES

ELECTROMECHANICAL ENERGY CONVERSION

a~"

L"
b~l)
a~~

L"
b,p~

Consider first the resistances. Since the two stator windings are not
necessarily identical, Eq. 11-21a cannot be simplified except in notation,
to give;
S
S
R',S
= R Ql)
R"s
= Rbl)
(11-22)
"'I"l)
b~'1
The a and b windings on the rotor are not necessarily identical either;
" thus (from Eq. 11-21d),
(11-23)
R"
- R'1>"
R"
. R'til)
a~l)
f>.r,b~ -

614

ELECTROMECHANICAL ENERGY CONVERSION

G" ~

'
G

(11-25)

-'

In Eq. 11-20 the magnetic stored energy was calculated from the magnetic
fields. The magnetic energy can also be expressed as

W'I

!f'~r

IL:
~G'J

'l

!f''' =

(W",)'1 =

2! 2: 2: LU'Ii''1iJ'I

(11-26)

Expanding Eq. J1.26, remembering that Li/~ must equal LJi'l' and com, paring the result with Eq. 11-20, we have the following definitions for
the inductances:

L 3

QTJ"'l

L~S

""

)2 I L o."fh2 f(hj ll)27)_-i:"_}]

"'l
"1 2 L(bla)27) - 1

= (2"

Q'I

= L = (Z' )2 /ioTr1b 2 [W~FTJ +

b'l

L" = L'
Q'l"7)
Q'"

b.,

1]

(11-28)

[l.
(aLb.)~7)]
1 - (alb

(11-29)

7)2

= (Z' )2 /io-rr!a 2
a"

,?2

~I

)2TJ

-.[ L::h"

L"
b-qb-q

= L'

= (2' )2 /io-rr!a

b.,

LSI
Q"lb-q = L"
Q'lbTJ

b'l

l b)2TJ ]
[J1 +- (a(a/b)2."

.J.

SF

21 2' IL o7Tlab [

L::TJ~ = L::"" cos TJ(

a'!

--:;j2 (blat

Q'I

Lf."." = Lbo'l cos .,,(9 Q

=
"

LbTJbTJ

.,

7T12) ,
7T
Z zr fJ-0 !ab
b.,

.J.

= L~ cos '7'/'

Q'I

7)2

cos '7<?
cos .,.,(</> +

b.,

""

"'/2)

L;;;' ~ cos '7(</> L';" cos 1]

2
]
.J.
=--(bla'i"
cos '?'/'

(11-32)

_2 (bla)-'1 cos '7 (.J.

't'

,~

,~

/2) (11-33)

[---~_._]
(b(a)'1 - (b(a)-'1
2

cos riC</> - '/T(2)

r;~"

;J=

L?

__ ,

.~

.J.

G" 2'~"

'm~t
Ii

"'/1)

2"'"
- I ~
'1_
a.p".a"b'l - 1.!'~$.2:; -

"n"

".

2'''

'I,

Ri,.,

R'

,,"

(11.39)

000

n
:J

-:-

L"a'l cos ry'/>

L',,"

L'b'l

(1l-40)

L~~~ cos." (.p

i) ,

LJ:a'l cos 'rJ

(11-34)
(11-35)

Equations 11-27 through 11-35 give numerical values (and angular

dependence, if any) of the various inductances.

L~:, cos 'rJr/>

L;:,,'1 cos

'f}

(c,6

";J
1;6
~~

I
I
~~~

"

L~~'l cos." (eP + ~)

rr
L b'l
cos

-~)

.,,9

J~,t

r~

~ (b(a)'1 ::::<:b(a)-'1 cos '?'/'

'~

;"1 AtJ Ii.

t1

71'

(11-38)

"

ZI Z' ILo",!ah [

Lt:." cos.,.,4>

The parameters of one of the '7 coils in an .,.,th harmonic winding have
been completely defIned.
In summary. the parameter matrices :Jf.~'b
a'" <ll1d .!/.";;',)
n" for one
1') l'j. ll'''7J
1) 'I' rr~fJ

coil of each of the ryth windings are:

j:1

7T1ab

rr/2)

(11-30)

+ '/T12)

- s zr fLo
[
= 2a'l
b." ---:;j2" (b(a)'1

L:.b'I cos .,.,(</> + ?r/J)

L~'

(11-31)

(It-37)

rM -

:!f~"h".a.,b~

=0

L~."QTJ = L a7) cos TJ'/'

"

")2

L::'I cos '?eP

. 'r

(11-36)

L~

L';a~ cos

!f"S _

(11-27)

(blaF'" - 1

b-qbTJ

ru

I'--Q'I

Rb'l

'I

a>u

Q'I

615

.Referring to Eqs. 11-2Ib, 112lc, 11-2Ie. and 11-21/, and making the
substitutions indicated by Eqs. 11-27 through 11-35, we have:

(11-24)

"'I

"OR;"
9f"

SPACE HARMONIC ANALYSIS IN MACHINES

Therefore Eqs. 11-21a and It-2Jdbccome:

!!iSS =

'',~.'
. ...

....

(4) -;)

L ."r

"

Lf'I cos'rJ<P
0

If"

(11-41)

616

ELECTROMECHANICAL ENERGY CONVERSION

11.1.3

SPACE HARMONIC ANALYSIS IN MACHINES

617

Equations of Motion
~

'n

"f

The previous analysis which has <.letermined the paramelers has been
carried out by considering only one coil of the 1)th harmonic
the actual.
winding. There arc r/ coils located in 2Tr radians for each lJth harmonic.
In writing the equations of motion this fact must be recognized and
taken into account. To maintain a high degree of Ilexibility in the
equations of motion, the electrical variables (voltages and currents) will
be chosen as the current through and the voltage across one coil of a Y]th
harmonic winding. The torque, however, must be defined as the total
torque acting on the mechanical shaft to produce a mechanical displacement <p. The equations of motion can be developed llsing the
Lagrangian technique of Chap. I, since all system variables are defined in
the reference system of the actual windings. Rather than go through this
procedure formally, since the method has been clearly established in
previous chapters, the equations of motion will be written by inspection.
The volt-ampere equations for one coil of the Y]th winding will be of the
general form

or

v s.r
where

:J?~.rb a
TJ 71'

'117

and

+ p!,s.r
. 1 l,.r
a~b~.a~o.,J a.,b~.a,l>~

_ [&i s.r

a.,bT/'a.,b~ -

a~b~.a.,b~

!,~.r" a

h
"1 ry. TjTJ

' -'

.... I ,.":.~.... II

a.,b~.a~

f1

s
Vbf1

vr

11

af1

..

:c:F

:1

.. '1
-

's

'"

oTjb TJ ,GTf "1

--'iii

la~

__ h

'0')

it.;

=.r

la."

,:"?'.i'

(11-44)

~-::

...
f

<A"

'.\

":.~

1~

ls.r

.
~
+""
f

.,,"-a

'~;

Y".

~t

,>t

+""

1-

and

il
~

;.-~

+""

.;

...

.f

I~~

;:'i

...

"
<'. J
:',---.

'r

~,
\

" ..
,

i;"

The complete volt-ampere equations, written in terms of the parameters

derived previously and taking into account all harmonics. are given in
Eq. 11-45 in which only one coil of each harmonic current sheet is included.
Since there are actually 1) coils in the Y]th harmonic windings, (i - 1)

Ii

.,f'

Vb~

.,

(1l-42)

(11-43)

:;j-"
~

'-:0

v~~
b

~.

~~

the voltage and current matrices are

"--

"t-.
~.

are defined by Eqs. 11-40 and 11-41 and

S,r

..
'-~{
.....

.....

'"

.. or
~"
~

.,-

~"

...

~l

t'"

iT

618

ELECTROMECHANICAL ENERGY CONVERSION

SPACE HARMONIC ANALYSIS IN MACHINES

equations for the ith harmonic are not included because they are identical.
This fact will be taken into account when the interconnections between
the YJ harmonic windings are derived.
The torque equation, assuming an inertia J and a viscous friction ct, is
simply

form. However, the magnitudes of the coefficients of speed terms and

of time-derivative terms could differ.
To show this effect and also to give an example of the application of
the method of harmonic analysis, the model of a commutator machine
shown in Figs. 11-5 and 11-6 will be used to derive the complete equations

..

T = Je/>

1 [ (is,,
) 82'S,'
is''
+ at/> - L _
YJ a."b~.a.,b~ t . a."b~.a,,b,, a."b~.a."b~]
~ odd

oe/>

(11-46)

The factor of YJ in the expression for the electromagnetic torque occurs

because 2'~"h a"..."
h is only for one coil of the YJth harmonic current sheet
and the total air-gap energy of the YJth harmonic is '7 times larger than this.

619

r--~'
F.<"r

.,..~.

II./ A

Summary of Harmonic Analysis

A two-pole, two-phase, smooth-air-gap machine with nonsinusoidal

current distributions has been analyzed by expanding the nonsinusoidal
current distribution into a Fourier series of harmonic current sheets.
This has enabled analysis of the device by the sinusoidal techniques
already introduced in Chaps. 3 and 4. The equations of motion in
terms of the fictitious harmonic currents, voltages, resistances, and inductances have been developed and are given in Eqs. 11-45 and 11-46.
To solve any particular pI:oblem it is only necessary to relate the harmonic
currents and voltages to the actual coil currents and voltages of the physical
device. The technique for accomplishing this will be developed by using
the method of harmonic analysis to analyze a commutator machine.*

11.2

,q'
I

eLI
~

(,)

::J

.:J+:

_.D

o ''::

VI",

.~.:o

-"s

.,,-

;~

q axis of current
distribution

---d

,$-

The Two-Axis Commutator Machine

'tto, : ,~

,
'.i.

The analysis of synchronous and induction machines using only the

fundamental space component of air-gap flux density is in many cases
perfectly satisfactory. The analysis of commutator machines, however,
always requires a complete representation of the air-gap flux density if
correct results are to be obtained. In Chap. 4 the commutator machine
was developed from the generalized two-phase machine with sinusoidal' \,
current sheets using the dq transformation. It was pointed out at that
time that in practical commutator machines a sinusoidal current distribution never occurs, but that the results obtained using only the fundamental
component of current gave equations of motion which were of the correct
* Other examples and some general conclusions are given in C. O. Riddleberger.
.. Parameter Determination of Commutator Type Machines," S. M. Thesis. Dept. of
EE. MIT. 1957.

Fig. 11-5. The idealized two-axis commutator machine.

'I.,.
'~'

..,

',,,,,,,,,.

of motion of a two-axis commutator machine. The physical model is

assumed to have two concentrated coils on the stator located in space
quadrature and a uniform current sheet located on the rotor. Two sets
of brushes are in contact with the rotor conductors by means of a
commutator and thus the rotor current density remains fixed in space in
a position determined by the axis of the brushes. The stator current
density will be approximated by impulses of current density and the
rotor current density will be approximated by square waves of current
density. The stator has two windings in space quadrature designated as

"

SPACE HARMONIC ANALYSIS IN MACHINES

ELECTROMECHANICAL ENERGY CONVERSION

620

~--l

pole

tr

radlans~

a'

I
I

(l

_~u
41'!O

K~(d>')

K~(</>')

,~:;

b'

."

I-Z_J

1+1

It

(j)

1r12

tr

37r/2

2IT

~'i

,..

~
l~~
~'.j

= i;Z~(cos d>' + cos 3</>' + cos 5</>' + ... )

~~~

windings a and h. The terminals of the rotor windings an.: the hru,hcs.
The two sets of hrushes arc in space quadrature and the terminal curren Is
arc designated as i~ and i;. Before the equations of motion can he written
in terms of the currents iJ and i;, it is' necessary to find the relationship
between the terminal currents i J i~ and the actual coil currents. This
result can be accomplishl.:d hy a harmonic analysis of the actual nonsinusoidal current distribution on the rotor and hy the use of the dq
transformation gcneralized to apply to the 1]th harmonic of current
density. Before starting the detailed analysis of the commutator machine
it is advisable to derive the 1)th harmonic dq transformation since it is
necded to relate the coil currents of the equations of motion in Eqs.
11-45 and 11-46 and the lWO terminal currents of the rotor of the commutator machine.

11.2.1
......
K;(41')

K ;(41') = i~Z~[cos (d>' - 1rlz) + cos 3(41' - trlz) + cos 5(1/>' - 1r'IZ)

Stot"

+ ... 1

:>

7ll/lIZZI?ZIZ?Z???????????????????????????????????????;

I K;(I/>')=i;Z'(Cos</>'-1/3cos3I/>'+1/5cos54>'-)

K,W'

Rotor

I ., l_-----db'"'""~.,
K~(4)')

K;'.',
',-

~ ~",'

-,."

'~.r~

Ii;

q 't'

d'

tid

q'

K;(1)')

"~

Fig. 11-6.
machine.

1r12

11"

31r12

'f +

K~} cos 31>'

Kf,l cos (1)' - 1T/2)

+ K;'" cos 7J(,p'

+ ... + K~T/ cos 7]1>, + . ..

+ K b} cos 3W -

- Tr/2)

11/2)

'

K~" cos

i ;. ~;( '..~\

t i;

K~ cos TJ(,p' -

"11>'

Tr/2}

= K~" cos
=

(11-47)

+ ...

+ ...

(11-48)

Taking the 7]th harmonic term and substituting 1>'

---cl-----=-ED-=-+----:=--=,-----'::BLs=-.:)--:e=--=e-----=-e-De e e e
ED 0-:. f '
ED@@EDEBoeeeeeeeeeo
4>'

To analyze the commutator machine by harmonic analysis it is necessary

to express the brush currents as equivalent currents in a coordinate
system attached to the rotor. From the work of Chap. 4 this implies the
need of an ab-dq transformation similar to the one known for the simple
sinusoidal case. The derivation of this transformation is not a complicated one. and the result can be expressed as a matrix operator no
more complex than that found for the simple dq transformation introduced
in Chap. 4.
A~suming two-phase windings on the rotor in space quadrature but of
any arbitrary current distribution, it is possible to express the current
densIties defined as K~(rf/) and K~(rf/) in Fourier series as
K~(rf{) = K~l cos

1/3c053(41' -1r12) + 115cos5(4)'-1r!2) - ... )

t I ,.. 1 ,

lid
d t

~~.

I::rL':"""

= i{,Z'[cos(4)'-1rIZ) -

.,

i~.l

* '/

,,":j

The 7]th Harmonic ab-dq Transformation

-..."

"ill

I'

621

1>' -

,p give

''1(1)' - 1

(11-49)

K b" cos ''1(1)' - 1> - Tr/2)

(11-50)

The use of trigonometric identities to expand the right-hand sides of

Eqs. 11-49 and 11-50 yields, from Eq. 11-49
,. . ; t
t
K;" cos 7](1)' - 1 = K~T/ cos 7]q/co~ 7]1> + K~T/ sin "71>' sin "71>
= (K;T/ cos 1]1 cos "71>" + (K;T/ sin '71> sin 1)1T12) cos 1](1)' - 1T12) (II-51)

211"

Current distribution in the two-pole. two-axis. four-brush commutator

.,,~

622

ELECTROMECHANICAL ENERGY CONVERSION

and from Eq. II-50

K;" cos 7)[(</>' - 7T/2) - </>]

= K;" cos 7)(</>' - 7T/2) cos 7)</> + K;" sin 7)(</>'
= (K;" cos 7)</

The inverse of this transformation is' also useful.

':T] [c~s

- 7T/2) sin TJ</>

'r

cos 7)(</>' - 7T/2) - (K;" sin TJ</> sin TJ7T/2) cos "I</>'

IIJ.r,

(11-52)

K;".. thus obtaining:

=

sin "I</>
r
KaT] cos 7)</> - K;" sin 7)7T/2

K r - K r sin "I</>
'iT] 0" sin 7)1r/2

+ Kr

~
r

KdT]

K;'

r:

~n":i

SIn 7)r/>
-.
/2
SIn 7)7T

c os "1</>
.

COS
" ~(\

'1'1
,,, '.

d
'r

lq"

si n 7)1>
sin 7)7T/2

'Hi,

Kr
~

Infinitely distributed conducting grid into

which a sinusoidally snaped brush Injects
a sinusoidal current density that remains
fixed in space relative to the brush

:v(

(11-55)
:~"

K;'

:,

Fig. 11-7.

"

,~.

;1\

.~

1:

!~~.'::,f.-

01)

-.J

::J

'i'

'
Aif

OTJ

sin~
~.r
7T/~ I

cos "11>

(11-57)

qr

:',

sin

7)'('

'r

'~i.

zr,

~OS7)'"

cos

T] I

dr

Whenever the harmonic windings of phases a and b on the rotor can

be taken to be equal, i.e., z~ = z~ =
then Eq. 1I-55 can be used to
transform the terminal currents. For example, since each K is found by
multiplying the appropriate Z by the appropriate i, if the Z's of the
rotor harmonics are identical Eq. 11-55 may be expressed as:

'd"

7)7T/2

A..

Id

t,

Comparison of Eqs. 11-53 and 11-54 with the dq transformation for

the sinusoidal case in Eq. 4-42 indicates a similarity in form. Equations
11-53 and 1I-54 define the 7)th harmonic ab-dq transformation. and they
may be written in matrix form as:
f'

~n

"11>

7)~ ~r

sin7)7T/2
sin

(11-54)*

b"

.,

-SIn

"1t/J

-"

(11-53)

cos 7)cP

It is:

Equations 11-56 and 11-57 are the generalized 7)th harmonic ab-dq
transformations. Thus the equations of motion in terms of ab components on the rotor can be transformed to rotor quantities in the dq

From the right-hand sides of Eqs. II-51 and II-52, define the
coefficients of cos 7)</>' as K~" and the coefficients of cos YJ(</>' - 7T/2) as

r
KdT]

623

SPACE HARMONIC ANALYSIS IN MACHINES

',:'1

l ....

Idealized rotor structure in dq coordinate system.

coordinate system, which corresponds to the brushes and commutator

on the actual machine. The interpretation of dq currents as brush currents
follows directly from the discussion of Sec, 4.4. For the 7)th harmonic
dq transformation Fig. 11-7 illustrates the physical meaning of the transformation. It is drawn for the current sheets of order 7) = I and 7) = 3.
The interconnection of the various current sheets to represent the actual
commutator will be apparent when the actual current sheets are specified.

(11-56)

11.2.2 Transformation of the Rotor Currents to dq Variables"......

i~"

The equations of motion expressed in terms of a~b" variables on stator

and rotor derived for a general nonsinusoidally wound two-phase machine
in Sec. 11.1 can be transformed to dq variables on the rotor, and by

Since 'I is an odd integer, sin '1",/2 assumes only the values I. Therefore. it may
be placed in the numerator or denominator of a term, whichever is the more convenient.

:'7

,j,

.hl

624

SPACE HARMONiC ANALYSIS IN MACHINES

ELECTROMECHANICAL ENERGY CONVERSION

matching coel1icicnts of the various harmonic terms for stator currents

and for rotor brush currents the eljuations of motion of the commutator
machine will hc ohtained, Tllt:re is one simplification which is ill order
before using the general relationships of Eqs, 11-45 and 11-46, Since the
rotor winding of the commutator machine is a continuous winding and is
symmetrical, the rotor conductor densities of phases a and h are identical
and it will be assumed henceforth that

Z;"

= Z(,,,

Z;,

R'a" = R'b." = R "r

(11-60a)

S
'
L Sab"
= L a."
'

(l1-60b)

r
L b"

( 11-60c)

sr - Lsr
L ba."
b"

(11-60d)

L'."

Lb'l
"

La."

ib.,,1

s
La..
s
'b."

cos 7],p

If.

sin 7],p
sin rrrr/2

i;."

(11-61)

I SIn 7]17/2

d~[

~-"\

d,

d-'

tJ

, d,'

cos 7],p

"

i'q."

.91" =

IE ~ 11-----

cos 7],p
sin 7],p
sin 7]77/2

::::=-=-:7]</>;:-1
sin

sin 7]17/2

is,,

cos 7],p

(11-64)

(11-65)

.91

v""

"a~"I'd."q"

- .91 es.,

." a~".drflTJ

(11-66)
(11-67)

It should be emphasized that these voltages and currents are for only one
coil out of 7] in the 7]th harmonic winding.
To transform the equations of motion first consider the electrical
equations. In terms of a."b." variables, the general 1]th set of equations is
given by Eq. 11-42 which can be written as

v s .r
(11-62)

From Eq, 11-45 it is evident that the volt-ampere equations are

described by a diagonal impedance matrix in terms of 4 x 4 submatrices.
This is a consequence of the lack of coupling between harmnnics of
dilferent order. The transformation matrices of Eqs. 11-64 and 11-65
arc also diagonal in terms of the 4 x 4 submatrices of Eqs. 11-62 and
11-63. Consequently the ab-dq transformation introduces no coupling
between harmonics of different order. Thus, the transformations for
the general 7]th voltage and current matrices are:
a~'1,a~'1 -

This transformation alters the rotor variables but leaves the stator
variables unchanged, which is the desired result. Now define:

(11-63)

cos rl</>

l
... iJ

a~.".a."b." -

("

...

-1

vs,r

sin 7],p
Sin 7]17/2

si 11 'fit.

s;-n -yrr/2

cos "I</>
sin 7]</>

All other symbols defined in Sec. 11.1 and used in Eqs. 11-45 and 11-46
are unchanged,
The 7]th harmonic currents on the rotor can now be transformed from
a."b." variables to d'lq'l variables by the transformation:
La."
~
's

(II-59)

L'a." =~
=

(11-58)

This will affect all the inductances which arc functions of Z,; or Z(, and
also the resistances. The affected terms are six in number and are
redefined as follows:

..to/"

,I'

615

, _ f/'sl r
ls,r
a"b.".a."h" a~",a.,b" a"h.",a'lh"

(11-68)

in whieh the resistance and inductance matrices have been combined to

form one impedance matrix.

0N
0-

r
m

Q
;xi

3:

,,'
-"\

R~I

+ pL~l

()

pL~;

", I

}>

"I

r:t l
V41

v'

"

R~

+ pL:,
~L"
bl

pL'J,.
-~L"
"I

Ri

pL~

~Li

+ pL~

-~Li

pL;~

;~

Ri

}>

Z
i'11 I m
='"
Cl
I~ f -<

+ pL~

..

()

R~3

+ pL~3

pL~~

i;' 1

<

v'~3

d3

U,3

Rt3

+ pL:3

pL~~

-3~L~

3~L"
"3

PI-:;

0
R~

+ pL~

3~L~

,~

~\'

-'3

Z
()

-3~L~
R~

='Vl"

i'tJ ,

pL;;

(5
Z

i'3

+ pL~

i'I)

,,"IUI;1 J'T7n".IT:r:~Cil"I' 7.~JJI.n;~"ffl!.Jn.'(~'~';";~'.~i#',"

...

...

- .-

..

..

.~
:)

..

}>

R:'I + pL~

V:'I

Vl

-0

.,.'"

sr
pL"1)

i:

T1

()

v'"
v'4'1

!i

pL~
.

+ PL:"

i:"

p4~

~L:~(Sin ~)

R:.. + pL~

~L~(S~ 'i)

'4"

pL"
b.,

-TJ~L~(sin i)

~ +pL~

i:"

~:

'r

'r

V~1)

R'"

~.

TJ+L~~( sin i)

}>

='"

3:

o
Z
n
}>
z}>
r

-<

Vl
Vl

(11-72)

Z
3:

i!)
I

Z
m

V\

~
......,

..,.

41,

628

ELECTROMECHANICAL ENERGY CONVERSION

u~b".d"q~

I !!l'.f"
.9/ (-'5"
~
urybry.a~b~
~ aryb~.dryqry

= d -

"-1

7)2
drS,'
J
a~",d~q"

1!fLs"
<:1.91ry
'I
aryb~,a~~

u"b".d~qry -

/.f.r

~ =

a~ry,dryQ~ a~~.dryQ~

1" ('s,
)
-2 "odd
L.
(- a;P~.a~b" t '7

a,/b",a~"

Ol./..

sin 1)Tr L ,r
2 u"

7)2

'S'

to a;P~.a~b~

a1J

sin 1)rr L.rr

b~

_ _I

T~ --

" :L
2 II r"U1J isb~ i'd~ i'q~ x
~ odd
L.

t.'
i,
;"1'

'I'"

'.) {J

,'.., ....

.J

( -sin

~.,

rr

. '7
( 5m 2

;\,

:t

;rr)L:~

)Lsr

Ib"

(11-77)
i;~

h"

.,

Tr)Lsr

-sin 7I
2

'j

ln~
'5

b1J

( sin 1)rr)L'"
(

(,I,/,

7/," LV

(11-76)'

82'S"

sin

(11-71)

The matrix equation for one coil in any general harmonic winding can
be used to write the volt-ampere equations for all orders of harmonics.
The use of Eqs. 11-45, II-59, 1I-60, 11-62, 11-63. 11-64, and 11-65 in
Eqs. 11-70 and 11-71 leads to the volt-ampere equations given in Eq.
11-72. It is once again emphasized that Eq. 11-72 contains a description
of only one cycle of each harmonic current sheet.
The torque relationship. Eq. 11-46, must also be transformed to dq
rotor currents from the ah rotor currents. In the torque expression
Eq. 11-46 only the electromagnetic torque is affected by the change of
variables. Define the electromagnctic torque as T, given by Eq. 11-46 as:
T

1/2

.~ '72 si n 1)'" L"

2 b'l

(11-70)

which allows the writing of Eq, 11-69 in the form

_ :.!l",'i,r

11-75a can

(11-69)

a~bry.dryqry - .

v~.r

or tq
_

From Eq. 11-69 a new impedance matrix can be defined as

!fL'"

629

of energy conversion. '1he torque coellicient matrix

be evaluated using Eqs. 11-46 and IJ-62 to obtain

Substituting Eqs. 11-66 and 11-67 into Eq. 11-68 and prcmultiplying
the result by .,?/~-I yield the vOlt-ampere equation for one coil of the
1)th harmonic winding in terms of d"q" variables on the rotor
1/5"

SPACE HARMONIC ANALYSIS IN MACHINES

1t('J

n~

(11-73)

T~

Making the substitution for dq rotor currents defined by Eq. 11-67 yields:

=".L.

'~ \'~.~,

odd

TJ

sr
sin TJ7T
i" i' )
2 (L b~ isb~i'd1J - L"
a~ U1J t('J

(11-78)

".~ ...

T - 1 "(is.,
) .91
~ - -2 "odd
Lu"b~.d~q~ t ~l 1)

82'S"

a~".a~ry d (is,,
,;,./..

(,I,/,

"a"b".d"q"

(11-74)

;..- .

The torque coefficient matrix is defined as

. ,,\.'?P"'.~). If,

...

"
ff Sa~ry,dryQ"
-.91~l 82'~~",a~~
-- SIf~
- 1)
8</>

(l1-75a)
';~i

which makes the electromagnetic torque of Eq. 11-74 assume the form
1.;".;

1 L.
"

T~ = -2

odd

('s,'
) ,-rrS,'
( .5,'
)
(- a"b~.drfl" t J a~~,d~q~ (-a~~.dryQry

'il

~rtj ~

(l1-75b)

The torque coefficient matrix ff for the 1')th harmonic current sheet is
similar in form to the torque coefficient matrix defined in Eq. 4-50e. In
Chap. 4 it was shown that the torque coefficient matrixff and the matrix
of speed coefficients ' were related since they both contain all the terms

,.~~'

Equation 11-76 is the '7th harmonicff matrix. Substituting Eq. 11-76

into the torque equation 11-75b and carrying out the indicated multiplication yield for the electromagnetic torque the results shown in Eqs.
11-77 and 11-78. Substituting the electromagnetic torque T~ into Eq.
11-46 gives

T =

Jq, + a~ -

T. =

Jq, + a~ - 1Jodd
2: TJ2 sin 1)2'" (Li: it id ~,,~

L:; i;

i; )

"1J1J

(11-79)
Equation 11-79 is the torque equation for the commutator machine in
terms of all the harmonic currents expressed in terms of ab variables on
the stator and dq variables on the rotor. The transformation of the
equations of motion to the rotor dq coordinate system is thus completed,
and Eqs. 11-72 and 11-79 are the electrical and mechanical equations of
motion respectively.

630

ELECTROMECHANICAL ENERGY CONVERSION

11.2.3

SPACE HARMONIC ANALYSIS IN MACHINES

'1
J
,

Interconnection of the Harmonic Windings

The equations of motion for a two-pole,' two-phase, smooth-rotor

machine with nonsinusoidal current on each phase have been derived and
expressed in terms of ab stator coordinates and dq rotor coordinates. The
magnitude of the harmonic currents and the harmonic turn densities are
as yet unspecified except in general terms. If these variables can be
related to the true terminal currents and actual turn densities of a
particular commutator machine, the machine equations of motion will be
determined. The a~b~ currents of the stator are hypothetical coil currents
expressed in the stator coordinate system; thus these terms can be related
directly to the actual stator currents shown in the physical model of a
two-axis commutator machine in Fig. 11-6 to evaluate the magnitude of
the stator harmonic currents and turn densities. Similarly, it was shown
in Section 11.2.1 that the d~q~ components of the rotor are the equivalent
of rotor brush currents expressed in the stator coordinate system. Thus
these currents are in the same coordinate system as the actual rotor brush
currents of the physical machine in Fig. 11-6 and hence can be related to'
the true rotor brush currents to evaluate the harmonic currents and turn
densities of the rotor d"q~ variables.
The current sheets of the two-pole, two-axis commutator machine were
specified as impulses of current density on the stator and as square waves
of current density on the rotor that are fixed in space by the axes of the
brushes. Since the brushes are stationary in the stator coordinate system
the rotor current density is expressed relative to the stator coordinate
system. The current densities for stator phases a and b and rotor coils
d and q are plotted in Fig. 11-6. These current densities can be expanded
in Fourier series in terms of the current densities composed of the terminal
currents i:, it, id, i; and the coil conductor densities Z~, Zt, Zd =
= Z'.
.A Fourier analysis of the stator impulse of current density and rotor
square wave of current density gives:

,~

These four equations are the Fourier expansions of the stator and
rotor current densities of the two-pole, two-axis commutator machine of
Fig. 11-6. These equations show clearly that the harmonic currents and
harmonic turn densities can be defined as follows:

,~li

X'll>/
oS

lU'l
;~;

+ cos 7J(</>' - 71'/2)

\,

+ ... ]

".' +

s
Kb~

and
I

I
\

i;Z'[cos (4)s - 71'/2) -

+ ! sin 'fJ7T cos 71(4)'

71

cos 3(</>s - 71'/2)

- 71'/2)

+ ... ]

.~ .

,j;y"4
.' /~'l

;j "'"

s
10

(1l-84b)
(11-84c)

s Zs

-Z-

(11-85a)

= Ib'l "." = '" b

'.,b., = 'sb

(11-85b)

Z~=z;,

(1l-85c)

7)71') Id,Z'
x dft," ":'.'. ld.,~zrd~ -_ 7)-1(.SIll -2

(11-860)
j

'.1

or

10) ~r~

lJ

I '

."

rd'l _ !(sin
Tl7I')i'
'fJ
2 d

(l1-86b)

and
i
}; ,~

t A. ),:"
.~

lh

~.

,j;

Zd., =

zr'

(l1-86c)

r = i r Z'
Xq'l
l/TIq'l

= !(sin
'Y/7T)i'Z'
'fJ
2 q

(11-87a)

or
~;)

. 't.

'~,:l

, ~tJlj,,,,

/l..

i' = !(sin 7171') i'

71

(1l-87b)

(11-80)

and
Z;~ =

(11-81)

cos 'fJ4>' + ... )

+ ...
(11-83)

I:

Z'

(l1-87c)

The four relationships, Eqs. 11-84b, 11-85b, Il-86b, and 11-87b, relate
harmonic currents to the terminal currents. Equations 11-84b andll-85b
show that the stator terminal currents and all orders of stator harmonic
currents are equal. Thus by connecting all harmonic windings in series
for a given phase, all terminal pairs of the harmonic winding are reduced
to a single terminal pair and the actual stator current flowing in this
terminal pair will produce the correct current density distribution on

(11-82)

X;W)

(11-84a)

or

\"<>

1- cos 3</>' + ... + ~ sin ;

szs0

10

Z~'l = Z~

K;M') = iJZ'(cos </>s -

I1'l

Rotor

'f

and

.'1';;

K~(1/) = i~Z~(cos c/i' + c;s 34>' + ... + cos 'fJc/i' + ...)

K~(c/iS) = ibZf,[cos (4)' - 7T'/2) + cos 3(4)s - 7T'/2) + ...

= lo~ZSIl7I =

or

Z;

Slalor

631

~.:

- -

I'

632

~I

ELECTROMECHANICAL ENERGY CONVERSION

the stator. Fl]lIations II-XM, and 11-R7h for the rolor show Ihat the
1)th harmonic rotor current is I h times as large as the rotor brush current.
Since there are 1] coils in the 1)th harmonic winding, all coils of a given
harmonic arc connected in parallel, and each harmonic group totaling 'TJ
parallel-connected coils is connected in series with every other harmonic
group. The polarity of adjacent windings (harmonic groups) is different
because of the sin (1)7T/2) factor which has the value + I or - 1; thus it is'
necessary to reverse the 3rd, 7th, 11 th, etc., groups. This type of interconnection of the harmonic coils will produce a single terminal pair and
the actual brush current flowing in this terminal pair will produce the
correct current density distribution on the rotor.
These various interconnections of the harmonic windings can be
represented in matrix form by defining a connection matrix for the
harmonic current and a connection matrix for the harmonic voltages.
Define the current transformation due to interconnection as
fJiJ ?,s"

(-'s,r
_
a~b~."~q~ -

;1

:~,

; II

~i.

establishes

.'!Ill'

as:

.~.~ ~ ."

3
p=

3'

[1

~)

sin 1)'11'/2

-I
-I ,',.

Ji
:~

<.

f!l'S"

';1"'

11-88 and

where

..,

.,

\..
~.,

Bilv

~.

,~,

s"

!!4i

j:'~

(11-92b)

a.,bT),dT)q~

(11-89)
Performing the matrix multiplication indicated by Eq. 11-92b and using
Eqs. 11-72b, 11-89, 11-91, and 11-92a yield Eq, II-nc.
The interconnection matrix :JtJ, can also be u~ed to exprcss the torque
in terms of the actual tcrminal variables i~, i,:, (~. i;. Using hi. 11-89 for
Bill' the torque cquation of Eq. 11-75a can be transformed to the terminal
variables i;;'dq' The same result can be obtained by using the torque
exp-~ssion of Eq. 11-79 and the relationships of Eqs. 11-84b, 11-85b,
11-86b, and 11-87b to obtain

:'J ')'

may wonder why .'lJ 1 defined by Eq, 11-88 and 61" defined by Eq. 11-90 are
.. reciprocal" in a sense. The reason is thal ,YJ j and ,ii" are defined in such a way that it
will not be necessary to find ,111 1- 1 or til,,-1 which is impossible since neither "il j nor $1.
is a square matrix.

T =

~I)'1

j 'f'
J.

1. a'f'

S
[""(L
'i}" bl'

+ 3LbsrJ +
sr
- ;si'(L
q
til
Q

+ 1)L"b + .. "\'-~
~

3Lsr

U)

+ ... + nLsr + ...)]

"f

OTJ

(11-93)

>,

":;f
;~.

..:1J.h,.

'~\"

.ij;
~1

"

f!l""
'ab,dq --

In an entirely analogous way, the transfonnation for voltages due to

interconnection is
S,r
-!!l vs,r
(11-90)*
Vab,dq v a~b~,"T/tlTI

* One

11-92 i~ reduced
(l1-92a)

sin 1)'11'/2
sin 1)'11'/2
1)

11-90. Eg.

v'

-t

TJ

(11-92)

'V~~,dq = f!l'~'b,dqi~;,dq

}1.\!,"i\

-t

lS"

a".,bTJ.dT/qrJ "anbTJ'cI,,Q'rI

and substituting as indicated by Eqs.

to

Bil j =

'in~d

('

$; and /Ji v arc the connection matriccs, and indicate the intcrconnection
of the various coils and windings in operational form. These connection
matrices can be used to find the rcsultant equations of motion in terms
of actual terminal currents of the commutator machine which are i~, it,
i~,
Considering first the electrical volt-ampere equations which are
comprised of 7J equations of the form
1)""
",.,h f / I (!T/QT1 -

633

Using thl' ilk:ls 01 series and p:1I'allcl connections as dlscussl'd pll'ViOlI<,ly

Equations 11-84b, 11-85b, II-86b, and 11-87b establish fJiJ j is

~pt

MACHINE~

(11-91 )

(11-88)

i 'ab.dq

SPACE HARMONIC ANALYSIS IN

1tr
"'

The complete equations of motion of the two-pole, two-axis, commutator

machine are thus givcn by Eqs. 11-92c and 11-93.

634

ELECTROMECHANICAL ENERGY CONVERSION

o...l

:f

~IN

'Vi

:s

_..
~~

..

~. ~IN

.......
:;,

II.~
o...l_
_

'"
+

~..,'

~
.,
-:;

-.
.

..

_ .....
l:l,

:s

:;, ~ .....
+
+ - I l='" ~~
~""""'4
+"'

Li,

~.

I
~~

-----~

..

(l1-94f)

1'"

+
~
:;...
o...l

o...l

10
1=:- ,

.S

1-'

'"

+
+ ...
_ .. o...l
o...l l='"
'-"
~I +

~;,"

L:;;' -

, 1]

L~

+ ... + sin ('Y)1Tj2)L~ + ...

(11-94g)

L"b3

+ ... +

+ . ..

(11-49h)

+ 30"03 + ... +- 'ilLS'

+ ...
./ ""

(l1-94i)

G&, = L b'

+ 3L:;3 + ... + 'Y)L:;" + ...

(11-94j)

G';d

01

sin ('Y)7Tj2)L';:'1

G"
= L"
qa
411

;~

G',/q

= L~

- L;

+ ... +

sin

(1]7Tj2)L~

+ ...

(11-94k)

Substituting Eqs. 11-94a through 11-94k into Eqs. 11-92c and 11-93 gives
for the equations of motion:
''it

't~_"~i!

ii
,q, ',:.:.'.'
-I

\..~:'J'
'(l,

'~.r.'1:;1-. :.
,.{:

L"bq = L"b
= L bs,1
q

,.

+'"

+
~..,
~...

."

L:d = L'la

~IN

DI'

o...l

o...l

.(11-94d)

L~ + !L) + ... + !L:; + ...

+ ,-..

= Lb1 + 3L:' + .. " + "f}Lb" + ...

L' =

:;~ .

._
'"

~..
"'-l

(11-94c)

" ;1
:'"l

t::

+ 3L'03 + ... + 'ilLS

+ ...
'/ 0"

(l1-94e)

-IS::-

c'iK.,

~.., "'"
~.. ~IN

(l1-94b)

Rbi + 3Ri,3

1,.., + ...
+;1'~R'3 + ... ,'.+ -A"J

::.r

tlJ'

+ ... + TJRi,,, + ...

fit

---~
~

(l1-94a)

S
LS ... La1

R'o = R'01 + 3R'U3 + ... + TlR'

./ 0" + ...
Rt

. . ~
+ ~
_.f
+
~ + ""
~
... :: ~u
+ ~
s::'r'
"" +
+

.I~ :~

.; "'~

~ ..~ ~t

R' =

~-

Several interesting conclusions can be developed from these equations.

First, the following symbols are defined:

j}'fl

... -=-

635

~ +

- '-'

+
-~
+ .
.....)
... Z'
~
""
M
+
~
+
M'fI.:'+
M~,,"
~...
+ ....l='" ~~...""
+ o...l
_,;: + ...o...l..c. ~l = ' " : ; ' ; : ~
~
1:(
o...l+'''''
+
-4).
+

+---

~)

:;m

I
~~
5
o...l'

-II:"'

SPACE HARMONIC ANALYSIS IN MACHINES

"

'7'

-..

'0000'

'"

... .:' +

+ - I l='"
+
~
_ ....

.5

+-

~J'

--...

.......

'"

..-4

..:=....-

~IN

1.5

..:=....-

+
~..,

I~

--

~~

-~?J

;.

,,:

Electrical

v~v~

vd

l~ +0 pL~
=

v'q

pL'la
- 'fJG'"qa

o
R~

+ pL~

~Gd~
p L's
qb

pL:d

R' + pL'

JG dq
'f

- q,C;d

pL';:q
rr

R'

J
l

+ pL'

i;
it

i'd

(l1-95a)

i;

Mechanical
T = l'fI

C%'.1f

., "G"
+ 'a1q
q'"

' f 'Gr>'
'ih
db

(11-95b)

~.u
',11'(

Equation 11-95 is the complete set of equations of motion of the idealized

two-pole, two-axis, commutator machine which has impulses of current
density on the stator and a rectangular wave of current density on the
rotor.
It is apparent from the results obtained that for commutator machines
it is indeed necessary to consider the fact that the actual current sheets
are nonsinusoidal. A simple sinusoidal analysis would have given

f'

\:<,

636

ELECTROMECHANICAL ENERGY CONVERSION

Il'c';~:

parameters of the wrong amplitudes, and furthermore would have shown

that the G''' and L" parameters arc cq ual and that the L' and (;;;" parameters
arc equal. Such a result is true only if the fundamelllal space component
alone is considered, but for a commutator machine such a simrlc arproach
is not allowable. This fact was discussed in Chap. 4 and the present
development justifies the K, N, M factors introduced arbitrarily in Sec.
4,3.2. The present development also shows clearly how the method of
harmonic analysis can be used to fmd the correct equations of motion for
a given rotating machine when nonsinusoidal current sheets exist and
must be considered.

SPACE HARMONIC ANALYSIS IN MACHINES

l!J

.wr:...

Substitutj ng

;~

The equations of motion have been developed by starting with the fields,
defining lumped parameters for each space harmonic current density,
and proceeding in a formal way to obtain the equations of motion. In
the resulting equations of motion voltages appear which arc proportional
to the speed <p. These voltages are given by the terms G~:+, Gd~~' and
G~~~ multiplying currents and are seen to have the character of a v X ,",oH
voltage induced by a moving conductor in a magnetic field.
To show that this is indeed the case, consider the problem of finding
the voltage induced in the q axis of the rotor by the motion of the rotor
conductors in the stator magnetic field. The net voltage due to v X ,",oH,
where v = a<p"~' found by integrating over the total conductor between
the q axis brushes, would contain terms of the form r~(H),. In terms)
of the harmonic fields defined by Eq. 11-7 and the conductor distribution
defined in Fig. 11-6, the net v X ,",oR voltage in the q brush axis due to
the a stator field would be obtained by integrating v X ,",oH~ at the rotor
surface (r = a). Defining this voltage as v;~ gives

"

v;~ =

J- (v X ,",oH~)

r-a

d (conductors) \' ,

(11-96)

...

'n

..

."1~,.Uh'if;~;l

'<~

"I

vi/a

1
J.:
1I

it',

;t

l.,
'1
o!;

"~

rs _

Vqa

(1 (2" ",.

Jo Jo

'1tdd Q<p

sin <{>')a dt/J' dz = _ ~

( _ Z'
TJ
TJ

--

.. ~

.1 [

2f L o(bla)

] '.' Z"
(bla)'1 _ (bla)-'1 10 '1 0

SJn

1'J'I"

2ab
_] hLOZ~Z'7r i' (11-97)
(bla)'1 - (bla) '1
~
a'1

2: [
'10UU

.1.6

J "c,r$
= -.f.IJ',",

lqa

1 ,',
.la,

L';

(tl

from Eq. 11-33 givc~

(ll-n

~
~,'

~,
-~.,

1\
,

,rl~."';,
j'

'~

(11-99

This proves that the G~;l term is obtained from the line integral 01
v X ILoH,: over the conductors in the q axis. A similar de\!elopmenl
would show that G;i;. is obtained from the line integral of v x l'oH~ Ovel
the conductors hetween the d brushes. that (;~:., is ohtained froll) the lim'
integral of v x ilOH~ between the d brushes, and that G;':/ is obtained from
the line integral of v x fLoB;/ between the q brushes. This establishes (l
simplified approach which can often be used to advantage to evaluate these
speed coefficients. It is also possible to show that the inductance terms
such as L;:~/ could be obtained by calculating the flux linkage with the (J
stator phase per unit ampere in the d rotor coil.
These physical intefl?retations
each of the various terms often prove
invaluable in llnderst)1~ the operation of devices and should be
developed whenever pq~ible. The general harmonic approach can,
however, be applied'\'Vf.{enever intuition fails and it illustrates the
generality of the space' harmonic analysis method which has been the
underlying analytic technique used throughout the analysis of rotating
electric machinery in this and previous chapters,

ot

,,~,l

1;;'
,~,I:,'

'1/

L..

'1 odd

rs

'~

,!'If't'f.t'L'I.'.

or in explicit form

1 ',\. ,.

tp1u

'I

which from Eli. ll-Y4i is simply

'~

all conductors
between q brush.

c/

"I'

= i~ from Eg. 11-84b and L~~

Vi/a

,I

A Physical Interpretation of the Equations of Motion of the

Commutator Machine

'I

>-

~.

11.2.4

i;;

63i

~.

:"t'i~j'

,j

,;~I

~/'

J..
tt

't

f:1

~'1:~

~~
r~/

11
, I

,ll '

!l;~I

'll

~l
.':') I

"

,.,:; 'I
.~.~

~. ;'JI:

Alternat or. See Synchro nous machine s

Analog compute r represen tation, induction machine , 486-489
servomo tor, two-pha se, 490
Angle, space, electrica l, 172
mechani cal, 172
Arbitrar y displace ment, 4, 17, 18

-;'

:.
\fl
.~

.:":1,;

.I
'iI
,
i

yl

I
i

'1
.;;

(~4

\ :

.!I
1

r
i

Ic

I:

t
I

;;"',,1

-1\

L _

.~

.010;.'

~~

!~

Block diagram s. See Systems , block diagrams for

Bode diagram . See Frequen cy response
Calculu s of variation s, 32
Capacita nces, 7
Change of variable s, 254
Christof fel symbol, 73
Coenerg y, electric, 26, 28, 29
kinetic, 49, 58, 59
magneti c, 14, 19, 20, 23, 49
Commu tator, idealized , 266-270
Commu tators. See Machine s with non
sinusoid al current sheets
Comput ers, analog, 158
represen tation, transduc er, magneti c
field, singly excited, 161, 162
solution of linear equation , 161
symbols , 159
Conserv ation of energy, 11
Conserv ative, Lagrang ian, 62
system, 39

Constra ints, external , 12

holonom ic, 40, 42
internal , 9, 17
internal electric, 27
nonholo nomic, 40, 42
Continu ity of charge, 4
Continu ity of space, 3
Convert ers. See Machine s, Transdu cers
Coordin ates, generali zed, 42, 43
ignorabl e, 69
indepen dent, 40
quasi, 69-72, 74, 75
true, 70-72, 74, 75
Current sheets, 174

.~

II

d'Alemb ert's principle , 3

D-c commut ator machine s, 277
four brush (two-ax is), 285
doc transfor mer, 286-288
equation s of motion, 422, 423
power transfer , 286, 287
ideal paramet ers, 278
physical paramet ers, 278
power flow in, 284
D-c commut ator machine s, single-axis,
278, 425
armatur e control of, 430, 431
deflnitio n of variable s in, 280
equation s of motion for, 426
field control of, 430
general block dillgra~ for, 428, 43~)
639

I'

,40

INDEX

!)c COmnll!l;'(Ur

In:h,,:hIl1l"'. 'lrlrk-~Ixis.

inl,.: I L'I It\.' III al

t... 1I.

bl,t\,:'-.

111.1g.1';111l

~~2K

scrics-conncc.:tcd. 282. 2HJ

;ndlll'I,IIlCCS in, 457

'I'r.:cd cor.:tlkir.:llls in, j57
rotating amplifier, as, 41>1470
blor.:k diagr;,ms for, 1(,7..4(\9
compens;,tioll in. 46X. 4(,1)
r.:quations of motion lor, 41)4
,,:hematic of, 466
-c induction m;,chines, )09 -) 12
,'qllarions "f motion for, ) J()
torque spccJ CUIV,. fIll', ) I I
,C tf/uhf"rm.:r. 2x62XX
cgr.:es of (reedom, 9, 10,40
l)iH~f,-=nll.tI C(I":ltioll.". !ill..:ar, ch.lfat..:tcristic cquati'"h of, <)()
c!a"ic;l! \,'Iuti,'n of, X<)
compleOlcnt<lry solution for. 90
c"nSlant coetlicients in, lI7, lIlI
hOfllo ..cnous, '.10, ',IX
inl ... )!r.llillg facf<\r in, 149

nth ,)ruer, K'.I

J'<lrt,cular solution f,'r, ')0, <JK
I 'Ul('

(If.

((",:lIil-'ll'llb

in.

Frequency response, nHlltnr.:tic field

transducer. 127
Nyquist diagram, 129, 130
polar plot. J 29-13 J
Function, co Hamiltonian, 53, 54
co-Lagrangian, 53, 54
Hamiltonian, 51
Hamiltonian state, 47
Lagrangian, 52
Lagrangian stllte, 411
state. 52
variation of a, 34

X7.

serlcs "'Iution. 155157

n,'nlinear, 81. 15K
DissipatIon, 39
Rayleigh fllnction. 61
Electric .... uchinc ... 170
See illso

Ma~hines

Electromechanical coupling network, 6

Energy. electric, lI, '.I, 24, 25, 2Y, 56-5Y
cnlh,t1py. 54, 55
Gibbs free, 54, 55
Helmholtz free. 54, 55
internal. 54, 55
kinetic, %, 57
magnetic, ll. 9, 13. 15. 17,56-59
calculation of, 15. 16
potential, 4H, 49, 56-59
total. 54. 55
total stored, 5
Energy convasion. frcquency conditions for, salient air gl'p, 211
uniform ilir gap, 20K
voltage of. 20 I. 2U2, 204
Entn,I'Y, 54
I::qllilihrilllll, equlltion. St'" EUlcrL~lgf"lllgC

stablt.:. 95, %, lOS, 106, 118

ste;IJy-st;,te, 125
unslllhlt.:. n, 109
EulcrLaJ:range cqualion. 35 -37, 61, 70
Boltzmann-Hamel f.,rm, 73. 7f.
Extremum, variational. 33, 35
Faraday's law, 10
FeeJback, rrlncirlc of, 361. 362
Se,' ",1.'0 Systems, Ct'lS.:tl 1001'
Flux link .. gcs, 7
Force, .:!r:ctrOfllechanical coupling, II,
12, I f.--Ill

from electric tie Ids, 27-29

from magnetic l1cltls. 21
gcncndiz~d. 49
nonconservative, 60
of constraint, 60
Frequcncy rcsponse, !lotle diagram, 128,
13 J

()t)

SifUtll!"n"ous, '.10.

valYlfll..~

t inh.~

lillear.

IO~

definition of, 121. 1::!.2

~r

.~ .l
~

~l
,\.:'

~I'
;:'

series feetlback in, 450-453

shunl-conne<.:tcd. 2K I, 2K2, 429
shunt (cedback in, 447-.150
Thevenin equivalent ticlJ circuit
for, 437, 431<
commutator machines, two-axis,
multitielJ, 454 jf.
equation, of motion for, 455, 456
general bloc!" diagr;lm for, 462,
46)

\.
~,

u.\ftl:lt jult~.

147-1\1)

lincafll.cd cquation, "", ,127

Mfields in, 4~<J

equal ions of moilon fIll, 440

gcnCI;l! hlock Ji'I).:'alll\ 1'01, 444~

445
inducrances in, 440
speeJ cOetlicients in, 441
mulliplc excitation of, 446
m~lpk "eld.s in. 4.19
mU;t;sh,gc, 453, 454
saturation in. 4)3-436
.c.dar fccdb.".:k in, 447
separately excited. 431-433

IC

INDEX

f)ilrl~f(':llt i~a1

.~
~,

';:i1
.,:
~

Hamiltonian, 44
Humilton's principle, 30, 37, 60
:t-hlrmonic analysis, space. See Machines
with nonsinusoidal current sheets
Hysteresis, 39

Impulse, response, 112

of timevarying system, 154
Generalized machines. commutator,
unit, 112
analysis of, 266-270
Incremental, equations of motion, 95converted power in, 202
97, 107, 119. 126
current sheets in, ,181
variations, 93, 95, 96, 107, 117, 125
def\nition of, 171-175
Inductanc.:s, 7
dynamic variables in, 175-179
Induction machines, 477 ff.
equations of motion in. 197, 199
analog computer representation of,
af3 variables, 259-262. 348
486-489
dq variables, 271 ff.. 348
definition of. 209
jb variables, 324-327, 349
equations of motion for, afJ variables,
'Y~ variables. 302-304, 348
479
general transformation, 263-265
Jq variables. 485
summary of afJ. Jq, "",
Ih,
normalizeu dq variables in, 48534X. 349
4H9
symmetrical components
SU also Servomotors, two-phase
317-32[.349
Induction machines. three-phase, coninductances. self and mutual, 187straints on equations of motion
188
for, 483, 484
slot and end turn, 180
dynamic operation of. 496
Lagnmgian for, 195
impact load response of, 499-503
magnetic energy in, 186, 187
normal operating range of, 495
magnetic fields in, 180-186
linearized equ01tions for, 498, 499
parameter matrices for, 178, 179, 190
starting transient of, 503, 504
induetuncc, ::!.63
torque-speed curve for, 495
resistance, 263
Induction machines. two-phase, 238
sp.:etl coefficient, 264
as servomotors, 246
torque coemcient, 263, 264
bulanced operation, torque speed
power balance in, 20 I, 202
curve for. 340
.
schematic representation for, twoconstraints in. 32ll. 330
pole, 126
equations of motion for, 331
""At'\,
torque coetlicient in" 203
equivalent circuits of. 241
torque equation for, 1'>2-194, 199,
m:,ximum torque in, 245. 339
200
phasor diagram~ f<)I'. ::'4 I
volrage of energy conversion in, 20 I,
rotating flclds in, 336. 337
202
slir in, 233. 3:\2

<+-),

'~I

'
l
'~I

:f

"I
'1,
;\1

~l

'I

!'
~

J~.

:~'~
~

,~

J
i!

'il
Ii!.,
,j
,
"

,.............

Generalizcd machines, voltampere

r.:'1uatiom for, 191, 192, 199
Generators. See Synchronous generators, Dc commututor machines

+-,

:1\
:~ I

., ..... _

641

,__

_.__,,

";'

titl '..tHN"""

,.

~~/'IfI"

642

+I

rI

Machines with nonsinusoidal current

sheets. 604 If.
currcnt sheets. dC'finition of. 604-606
equations of motil)O [or. (,16-618
inductanccs in. 60'.1, 610. 613-615
interel1nn~l'lion of. 612
parameter mat rice' for. 613-615
two-oxis commutator. 61X. 619
complete d" equations of motion
for, 633-(,35
connection matrices for currents
in. 632
connection matrices for voltage in,
633
current di,trihutions in. 620
dcfinition of parameters for, 635
Kirchhoff's Jaws, continuity of current, 4
interconnection of harmonic windloop equations, 4
ings in, 630-633
torque equation for, 628, 629
Lagrange equation, 64
volt-ampere equations, 'lth harSee also Euler-Lagrange equation,,~

monic, 625, 627

Lagrangian. nonconservolive, 60, 61
quasi coordin;lte, in. 70-72
Matrices. circulant. 549. 5.51
symmetrical component transforLaplace. Set' Transform
mation of, 552, 553
Linearization, functional nonlinearity,
complex, 258
137-139, 146
Hermitian orthogonal, 258'
product nonlinearity, 107, 117, 125,
impeJancc, 257
135-137
inductance, 263
Linearized differential equations. See
inverse. 256
Incremental
rcsistance, 263
Loads, general representation of, abc
speeJ cocffil'ient, 264
variables, 535
torque coefficient, 263, 264
dq variables, 535-537
transformation, 256, 257
zero power-factor, 537, 538
unit, 257
Loop equations, 46
unitary. 258
See also Transformations
Machines, classification of, 547
Microphone. electrostatic, 414-420
equations for constant speed, 213
equations of motion for, 415-417
two-phase, 171
linearized, 417
Su also D-c commutator machines;
frequency response of, 418-420
D-c induction maehincs; Electric
Momentum, generalized, 43, 49
machines; Energy conversion;
Generalized machines; Induction
fl-m winding machines. dq variables,
machines; Machines with nons iequntions of motion in, 596, 597
nllsoidal current sheets; nom
jh variahlc', equations of motion in,
winding mnchines; Rcpulsion
5\16. 597
machines: Servomotors, twonom winding machines. 545 If.
phase; Synchronou, mochines;
pha,e variables. definition of, 578
Threc-phase machines
equations of motion in, 579, 580

Induction m:1chincs, two-phase, squirrcl-ca!,:e, 327

slendy-st:ltc operntion of, 331 If.
torque, starling, 246
torque in, 242
nver:lge. 335
jn~tnr.tancous. 335
torque-speed ~.urve for, 245
unbalanced operation of, 333, 338
torque-speed curve for. 340
variables, dr.r'njtion of. 32X, 330
Initial eondition~, 101, 133, 134
Integral. line. 58
Internal constnlints, 8

643

INDEX

INDEX

n"'"

~.
,,1

;;

1~'"

f,:f:
.~.

j~

ti

"\"

t,

I,

~"'i'
,:.'"
. .t {l
:~li

10\.1

,.

:t;~!
,
~W

,/~

.~t'

...

t',

'~I

"J

.-~'.,i~ .',l',~
.

winding machines, phtlsC v,lriaoic\,

parameter m;,triccs for, 5(,')-577

real vari .. hles. j(cncrolized. equations
o[ mot,on in. 59(,. 597
symmetrical component variables,
dcfinition of. 58 I
equations o[ motion in. 59(,. 597
:,,,rameter matrices for, 582-585
power in. 590. 591
torque equation in. 588. 589
lorqlJe-producing components of,
5ti7. StiR
transformation m;ltriees for, 581
VOlt-ampere eqllations in, 585-.587
Node equations. 46
Nonconservative, forces. 60, 63
kinetic coenergy, 61
La~rangian. 60
potenti;]ls. 60
Nonlinearities. functional, 137
product type. 93
Nyquist dia~ram. 129, 130
~"r "I.w Frcqucncy response
Nyquist stability criterion, 375, 376
See also Systems, closed loop

."i

"

~
~

":
,~

.~

.'

I":
.~
"

'.\

':tl

-lr,

"

.it!:
"J'

~~ "
~;
;.;\

~
f.~

'~

"\

"

Partial fraction expansion, 103, 109,

110
Permeability, space-dcpendent, 174
Per-unit systems. 522-527
bnse quantifies for. 524, 526, 527
equations of motion in, 527, 528
paramcters in. 523-525
vari;lhlcs in. 524, 525
Phasc helts. 174
Phase space. 43
Pha'or,. steady-state, 315
Sn! aI.w Symmetrical components
Poles. definition of, 174
salient. 174
Potent!nl, nonconservative, 60
Power, complex vari;,hles, in, 258
rcal variables. in. 257
speed (fependent. 276
symmctric;>! component variahles. in.
instantaneous. 5~7, 562. 565,
590.591
ph,,,,,,,. 5(.0..V,2
Sl',' 01.\0

Machines

.'

Power convcrlt.:d. ;lvcrage.

2()f)

instantaneous. 205
Power invarianee, 257
Rayleigh dissipatIOn function. (,I
Real componenh. "eneralj~eJ. 5(,.~
three-phase. transiormatian matrices
for ..~6li
variahles in. 56!!
transf0rmation matrices (or, 566, 567
vari;lhks in. 5(.5. 566
Repulsion machines. 304-309
equ;ltian, of motion for, )06
torque speed curves for, 309
Saturation in. d-e machines. 433-436
inverse satlJration factar, 435
saturation factor, 435
synchronous mochine.~, 532-535
Servomechanism, electrohydraulic. 396.
411
block diagram of. 411, 412
stahility of. 413
transfer function for, 413
'Jr
positional, 37i1 ff,
Servomotors. two-phase, 246, 341-345
analog computer representation of.
490
linearized equations for, 343-345,
493. 494
torque speed curves for, 342, 343,
491, 494
transfer function for, 494
Slip, 233. 332
Space harmonic analysis. See Machines ,1
with n()n'"lllsoidal current'!lheets '"
Stability. SI'(' Systcm
State function, 7. II
definition of, 38
Superposition, integral, 112
"b
principle of, xx
Symmet rieal components. instantaneous,
313 fT,
steady-state. 3 J 5
Symmetrical components. four-phase.
transformation matrices for, 562
vnriahll..~s

in,

.~f).~

Symmetrical components. n-phase, imI,,'don('(' l11a',,~ in, ~.~n

phllsor vlll.ahics for, 55\!

"

Symmetrical components, II-phase,

tramfornj.,tions for, 54X, 549,
.~A6

tl
'Ii,,~
...

,,".:

INDEX

INDEX

644

v;lri"hks, lklillilion of, 553, 554

negative sequcnce, .~54
positive sequence, 554
Symmetrical components, three-phase,
phasor representation of, 556,
501,562
rhasor variahles in, 559
powa in, 557, 560, 562
transform;ltions for, 554, 594, 595
variahks in. 555
Symmet rical ctlmponent~, two-phase
(semifour pila~e), ~6:1
power in, 505
transform,Ition matrices for, 564
variabks ill, 564
Synchronous generators, oefinition of
par,lmetcrs in, 520
oetinition of variables in, 519, 520
equatillns of motion for, 521
loaos on, 535-53X
p;lralle]-connccted, 53X-543
block diagrams of, 540, 542
transformation of rotor references
in, 54l
See alsu Synchronous mach ines
Synchronous-induction machine, definition of; 209
two-phase smooth rotor, 221, 229,
232, 234
equivalent circuits for, 221, 229,
232,234
phasor oiagrams for, 229, 232
power balance in, 236
slip in, 233
torque in, 235,236
Synchronous machines, 508 If,
definition of, 209
single-phase, 2 I3
two-phase, smooth-air-gap, 215
equivalent circuit for, 221
pha~or oiagram for, 221, 222
torque in, 223, 224, 225
Synchronous machines, three-phase,
hlock diagram of, 530
[520
deflnition of parameters in, 513,
detinition of variables in, 519, 520

Synchronous machines, threc-phase,

equations of motion for, 521
ab, ahc variable", 50lJ-5 12
ll/), Ot"l variables, 512, 513
Park's variables for, 51li
per-unit systems for, 522-527
base quantities for, 524, 526,527
equations of motion in, 527,528
pGrameters in, .~23, 524, 525
variables in, 524, 525
saturation characteristic in, 534
saturation nonlinearity in, 532-535
Syncbronous machines, two-phase, balanced operation of, 2\}4 IT.
oelinit ion of variables (dq), 291.
292
equivalent circuit (dq), 292
schematic of, 291
steady-state characteristics of, 295.
296
sum and difference components (5
and D), 2\}3ff.
torque-angle characteristics, nonsal ient, 296
salient, 297
unb;danccd operation of. 293 If.
Synchronous speed, 209
System, conservative, 39
disturbances in, :164
feedback in, negative, 370
positive, 371
final state of, 58
graph of, 45
input to, 89
nonconservative, 60
open-loop, 363
response of, 89
stability of, 89, 372
absolute, 372, 377
Nyquist, 375-377
relative, 372, 378
static, 40
steady-state, 42, 43
transfer function of, 111-114, 120
angle vs. log-frequency for, 365.
366
asymptotes for, 366, 367
hreak frequency of, 367

\'."
I.

i~!.

IfI'

r:

~
\(

7'
:f

j.
,"It~
',.

".k1...\',
~t i

:;~

\.'",:J'
.' l'

11
~.

:'1

'~~

. ,j

..'..

..

':/i .

,.'!f'

~;.f

.....'.

, .;~ ,<

. .::

~,

'; "1

",'1

('~

'"I

~I
~fl

:1

System, transfer function of, cascaded,

366
damping ratio for, 367, 368
first-order system, 366, 367
froJll convolutlllll integral, 1/2114
from differential equ'ltion, 115,
119, 120
log modulus of, 365, 366
natural frequency. 367, 368
Nyquist diagram of, 368, 369, 375
oscillatory, 374
polar plot of, 36~, 369
polynomial expansion of, 374
roots of, 374, 375
second-order, 367
stable, 374
third-order, 371, 372
time domain, 120
unstable, 374
Systems, block diagrams for, 131
algehra for, 131,139-141
frequency domain, in, 134
functional nonlinearity in, 137, 138
product nonlinearity in, 135, 136
reduction of, 139-141
R-L circuit, of, 133
symbols in, 132
time domain, in, 133
Systems, closed loop, compensation of,
384 tf.
controlled variables in, 362
disturbances in, 363, 364
elementary, 362
error variables in, 362
gain in, 370
lag compensation in, 387-389
lead compensation in, 385, 386
,M lJ criterion for, 383, 384
Nyquist stability criterion for, 375377
reference variables for, 362
relative stability in, 378-384
stability, absolute, in, 375-377
time constants in, 370, 371
Terminal pairs, 13 t
electrical, 9-10
mechanical, 9-10

645

Three-phase machines. gener:t1 cqllations of motion, III". vari:')lcs.

4X(J,481
Ont! variahlcs, 48 I-Ali)
Torquc, eoetlicienl. 20.1
coefficient no,Mix, 204
complex vari"hlcs, in, 259
tI'l variables, in, 27f>
real vari;,bles, in. 25H
running, 20ti, 207
starting, 200, 207
See aim Machines
Torquc motLlr. 396 fr.
block diagram of, 405
construction of, 397
dynamic operation of, 404 fT.
equations of motion for, 399-402
lineariz'ltion 01',403.404
equivalent circuit of, 399
frequency response of, 407, 408
step response of, 409
Transducers, 395
"Iectric tklo, 104
magnetic field, 91, 92
doubly exciteo, 123, 142
frequcncy response of, 127
linearized block diagram for, 146
nonlinear block diagram for, 145
singly excited, 116, 150
analog computer representation
of, 161, 162
Transfer function. See System
Transform, laplace, 100
basic theorems, 101
pairs, 102
Transformation matrix, linear, 256
Transformations, Cauchy-Heavyside,
88, 89
complex, 258, 3 12 If,
dq (real). 265 ff.
Ib (complex), 321-324
four-phase symmetrical component,
562
Fourier, 8H, 89
')'<5 (real), 300 fr.
laplace, 88, 89
legendre, 52, 53
n-phase generalized real, 565, 592,
593

"

JI

~il

lk, .

I.

l(l
"

tMrlto..

INDEX

16

Transformations, three-phase. powerinvariant. 514. 515

two-ph'l.se general real, 263-265

lnsformations, n-phase symmetrical

component, 548. 549, 592. 593
'cal. 256. 257
summary of, 345-347
symmetrical component, 548, 549
I
circulant matrix m, 549
fOllr-ph,,~c. 562
n-phase. 548. 549. 592, 593
rectangular matrix. 552. 553
semifour-phase. 562
two-phase (semifour-phase). 562564
symmetrical component (+ - ), instantaneous, 313 ff.
I
~teady-state (phasor), 315-317
synchronous machine, in rotor refer'f
ences, of, 541
! three-phase. conventional, 516, 517
generalized real. 594, 595

Variables, associated. 44
electrical, 54
extensive. 44, 54
independent. 45, 46
intensive. 44, 54
generalized, for electromechanical
system, 54-57
mechanical, 54
primitive, 45, 46
state, 44
Variational method. 30

't .'

(,1

"l!
~

.~

:
i

Ward-Leonard drive, 378 If.

!liP

., I

Zero power-factor loads, 537. 538

,i I
!t~!il .!I
oil

t.l .

,~l,

.~.

~:I!

"'f'll;

on.., ,,,1!i{)

I'll!:

j:-I

~f !

r .If:

'.' I
'. I
:

I:.
J1>

,,10;

tlf(;
'. ('~ 1~

ri, .; ;t,-,(

"

"I

;1

'1",_
~,

~Ij

<. '.,1

:; I'

,~

{
'::;J\

!".H4:,b

,~!;

~.;

-;&:;"

.,.~

:l~
IH

" ,1'~'1

"y .

~~t

4~

t'l

,~~-I
;'\

~:

' ' '~:*:':1

f.i.,('J

,~.

i..~l::l

~~\/1
;.'':1

~il

:' I

:q

::'1

l!:t

;.V!"
'i ...:,f,~~.tl

....

,~;,

.-(;r

'",~,;.)

\-