3/f *

N A S A TECHNICAL

NASA m X-2539

MEMORANDUM
CO

i
X

IS*

<c

3E

COPY

DESIGN OF TOROIDAL TRANSFORMERS

FOR MAXIMUM EFFICIENCY
by James A, Dayton, Jr,
Lewis Research Center
Cleveland, Ohio 44135
NATIONAL AERONAUTICS AND SPACE ADMINISTRATION WASHINGTON, D. C.

APRIL 1972

1. Report No.

2. Government Accession No.

3. Recipient's Catalog No.

NASA TM X-2539
5. Report Date

4. Title and Subtitle

April 1972

DESIGN OF TOROIDAL TRANSFORMERS FOR

MAXIMUM EFFICIENCY

6. Performing Organization Code

8. Performing Organization Report No.

7. Author(s)

E-6784

James A. Dayton, Jr.

10. Work Unit No.

9. Performing Organization Name and Address

112-27

Lewis Research Center

National Aeronautics and Space Administration
Cleveland, Ohio 44135

11. Contract or Grant No.

13. Type of Report and Period Covered

12. Sponsoring Agency Name and Address

Technical Memorandum

National Aeronautics and Space Administration

Washington, D.C. 20546

14. Sponsoring Agency Code

15. Supplementary Notes

16. Abstract

The design of the most efficient toroidal transformer that can be built given the frequency,
volt-ampere rating, magnetic flux density, window fill factor, and materials is described
here. With the above all held constant and only the dimensions of the magnetic core varied,
the most efficient design occurs when the copper losses equal 60 percent of the iron losses.
When this criterion is followed, efficiency is only slightly dependent on design frequency and
fill factor. The ratios of inside diameter to outside diameter and height to build of the magnetic core that result in transformers of maximum efficiency are computed.

17. Key Words (Suggested by Author(s))

Transformer
Toroidal transformers
Transformer efficiency
Transformer mass
19. Security Classif. (of'.this'report)

Unclassified

18. Distribution Statement

Unclassified - unlimited

20. Security Classif. (of this page)

Unclassified

21. No. of Pages

20

' For sale by the National Technical Information Service, Springfield, Virginia 22151

22. Price*

$3.00

DESIGN OF TOROIDAL TRANSFORMERS FOR MAXIMUM EFFICIENCY

by James A. Dayton, Jr.

Lewis Research Center
SUMMARY
In many electronic applications such factors as the efficiency of electrical power,
sources and the ratings of key components may determine the transformer current.
However, no theorem exists for determining the design of the most efficient transformer
when current is not a variable. This report provides a criterion for maximum efficiency for a toroidal transformer with a given current, frequency, volt-ampere rating, magnetic flux density, window fill factor, and materials.
The expressions derived are specifically for the toroidal transformer, both because
this type has wide applicability and because its description may be reduced to dependence on a single dimension, height H, and two geometric factors, the ratio of inside to
outside diameter Y and the ratio of height to build Z. However, appropriately restated, the basic results presented here probably hold for any type of transformer.
A continuum of wire and core sizes and shapes are permitted in the analysis. Two
winding schemes are considered, the basic two winding transformer with equal current
densities in both windings and the inverter transformer with a center tapped primary
designed for twice the current density of the secondary.
As the dimensions of the transformer are varied, which in this case of .fixed current
correspondingly varies the current density in the windings, it is found that the most efficient desigif for a given Y and Z is reached when the conduction or copper losses
equal 60 percent of the iron losses. It is shown that the efficiency can be maximized byvarying Y and Z to the values Yj>, and Z-,, which do not depend on any design parameter other than the fill factor. The values of Y., and ZM are determined for the
two winding arrangements considered here, the basic two winding transformer and the
inverter transformer with center-tapped primary.
Using this procedure it is shown that for given materials the efficiency of the transformer designed depends only slightly on the frequency.
A numerical example demonstrates that, at its maximum, transformer efficiency is
only slightly dependent on fill factor.

INTRODUCTION
Transformers are used in many applications where circuit efficiency is a primary
concern. Previously, if a designer wished to maximize transformer efficiency there
were no guidelines to follow in the case where current was not a design variable. Such
cases arise, for example, when current must be limited by the ratings of other circuit
components or when the power supply must operate at a particular current. The purpose
of this report is to present a criterion for the maximization of toroidal transformer efficiency when current is not an appropriate variable.
The design equations of the toroidal transformer have been reduced to dependence
on a single dimension, the height H, and two geometric parameters, the inside to outside diameter ratio Y and the height to build ((DQ - D,)/2) ratio Z. In this way, an
analysis can be made of the dependence of efficiency of the toroidal transformer on core
size and shape.
In a previous study (ref. 1) in which the design equations of the toroidal transformer
were computerized, certain trends in efficiency were observed. However, these were
somewhat obscured by the restriction of the program to core sizes that are available
commercially as catalog items and to standard sizes of round wire. The present study
allows a continuum of core sizes and shapes and permits an explicit expression for the
core dimensions needed to produce the most efficient toroidal transformer possible at
a fixed current, voltage, fill factor, magnetic flux density, core material, winding
material, and operating temperature. A continuum of wire sizes is also assumed here,
permitting the designer to choose whatever conducting material and shape the application
dictates.
Two cases are considered: the simple transformer having an untapped primary and
an untapped secondary and the basic parallel inverter transformer with a center-tapped
primary and untapped secondary. The maximum efficiency criteria are derived, and
some numerical examples are presented.

SYMBOLS
effective cross sectional area of magnetic core, m
- ,,H 1
O

\V!

area of conducting portion of primary winding material, m

y JL

- wi,
,Tri &9
c,

area of conducting portion of secondary winding material, m

i , wi,
.,i
^i , wi,
. 0^

area of primary winding material, m

area of secondary winding material, m 2

window area of core box, m 2

temperature coefficient of resistivity,

BM

maximum magnetic flux density, T

Cj

function defined by eq. (35)

C2

function defined by eq. (40)

C3

function defined by eq. (36)

C4

function defined by eq. (37)

C5

function defined by eq. (38)

Dj

inside diameter of core iron, m

Djrp

inside diameter of core box, m

DQ

outside diameter of core iron, m

DQT

outside diameter of core box, m

Dj

function defined by eq. (43)

D2

function defined by eq. (49)

Dg

function defined by eq. (44)

D^

function defined by eq. (46)

D5

function defined by eq. (47)

Dg

function defined by eq. (45)

6^

instantaneous voltage applied to primary, V

fill factor, dimensionless

frequency, Hz

height of core iron, m

H^j

height of core iron for maximum efficiency, m

Hrp

height of core box, m

Ij

primary current, A

\2

secondary current, A

current density, A/m

JM

current density for maximum efficiency, A/m

Jj

primary current density of basic transformer, A/m

kj

function defined by eq. (24)

function defined by eq. (25)

2
2

L1

length of primary conductor, m

L2

length of secondary conductor, m

mass of transformer, kg

mass of conducting material in transformer, kg

M.

mass of transformer iron, kg

primary turns

N_S

secondary turns

P
P.

transformer conduction losses, W

< transformer iron losses, W

equivalent winding resistance reflected to primary, J2

Rj

primary resistance, fi

Rg

secondary resistance, fi

operating temperature, C

TO

reference temperature in eq. (12), C

time, sec

V-.

primary voltage, V

Vp

secondary voltage, V

Wf

waveform factor, dimensionless

Wj

specific core loss, W/(kg)(Hz)(T)

W.

insulation increment, dimensionless

diameter ratio, dimensionless

YM

diameter ratio for maximum efficiency, dimensionless

Z'

height to build ratio, dimensionless

ZM

height to build ratio for maximum efficiency, dimensionless

j]

efficiency, percent

electrical resistivity of conduction material,

density of magnetic material, kg/m

density of conducting material, kg/m

PQ

resistivity at TQ,

cp

magnetic flux, We

DERIVATION OF CONDITION OF MAXIMUM EFFICIENCY

The first case to be considered is the simple toroidal transformer having a single,
untapped primary coil of Np turns and a single, untapped secondary coil of Ng turns.
Beginning with Faraday's Law,

dt

An expression for the effective voltage

as

(1)

induced in the primary coil may be derived-

= 4WffNpBmAc

(2)

where Wf is the waveform factor (1.0 for a square wave, 1.11 for a sin wave), f is the
frequency, B is the peak magnetic flux density, and A^-, is the effective cross sectional area of the core.
A second relation is available that links the number of turns to the dimensions of
the toroid,
N

PAT,wi, 1 ~ A W T

(3)

where A
. + is the total area of the primary wire, AVV is the effective area of the
-L Wi j J.
core window, F is the fraction of the core window to be filled with windings, and the
secondary and primary occupy equal fractions of the window area.
The derivation that follows reduces the equations describing the toroidal transformer
design to dependence on a single variable, the height H of the iron core. The first step'
in this development will be to describe the length of the transformer windings in order
to compute their resistance.
In a previous publication (ref. 1) expressions were presented for the average length
of conductor in a toroidal transformer. For the primary the conductor length L, is
L1=Np

(4)

and for the secondary

L

2=NS

3-2

(5)

where HT is the height of the core box, DQ^ is the outside diameter of the core box,
and Djrp is the inside diameter of the core box.
Equations (4) and (5) are derived assuming that the secondary is wound on top of the
primary as shown in figure 1. Furthermore, each winding completely fills its fraction
of the window, leaving no voids, and is uniformly distributed around the toroid.

Figure 1. - Cutaway drawing of toroidal transformer with tape-wound core.

The resistance R, of the primary coil and Rg of the secondary are then
pL,
(6)

(7)

c,wi,2

where p is the effective resistivity of the conductor and AC wi is the area of the conducting portion of the winding material.
The area A,,L*. \VJL- I-, is related to ATA j WAj
. -,*- by the relation
(8)

where the increment Win represents the increase in area of the coil due to the insulation of the wire and the introduction of insulating tape, if necessary.
When the corrections due to excitation current, leakage reactance, and winding resistance are neglected, the relation between voltage, current, and the number of turns
in the transformer is simply
(9)

The equivalent transformer winding resistance R

reflected to the primary side is

(10)

eq

For the usual case in this simple transformer where the primary and secondary
windings operate at the same current density, the equivalent resistance can be found
after substitution of equations (4) to (7) and (9) into equation (10); it is

R eq

PN,

2DQT +

4-4

(ID

c, wi, 1

When skin effect is eliminated through appropriate choice of conductor material,

size, and shape, p is not a function of frequency and may be written as

(12)
For the case of copper conductors (ref. 2) PQ = 1. 73xlO~ ohm-meter, a = 0. 00393
C~ , and TQ = 20 C.
The incremental dimension required for the core box is generally approximately
0.1 times the height H of the iron core. With the symbols Y used to denote the ratio
of inside Dj to outside DQ diameters of the iron core and Z to signify the ratio of
iron height H to iron thickness or build (DQ - Dj)/2, the dimensions of the core and the
core box may all be expressed in terms of H, Z, and Y.
1

HT = 1.1H

(13)

DQ=
0

2H

(14)

Z(l - Y)

n =
1

2YH

(15)

Z(l - Y)

(16)
- Y)

DITT T = H [

2Y

-0.1]

I.ZU - Y)

. These equations may now be used to describe other core dimensions introduced in
equations (2) and (3). Allowing 15 percent for insulation between layers of magnetic
material, the effective core cross- sectional area is written
H2
An = 0. 85^-

(18)

The window area is simply

A

oI

= lH2
4

9V

(19)

0.1

Iz(l-Y)

The expression for equivalent resistance can then be restated by substitution of equations (2), (3), (8), (13), and (16) to (19) into (11). This results in a relation for R
terms of the basic design parameters for the toroidal transformer.

o ,[

eq

/;

I;

\1

p(l + W.
)vfz^
2.1 + 0.24/1- *.+ 0.1Vl- F +
i
(2+ 4Y- 4 Y t / l - - - 2YVl - F
m
X
L
T
2
Z(l - Y) \
>
2
^
_
12
2.2698B^f2w;FH5 ^
C
"A

I f7 /1
I ii^l -

The total copper losses in the transformer P

V\
I^

are then simply

(20)

The iron losses P. are approximated by taking the specific iron loss to vary linearly
with frequency and flux density.
<22>

where pM is the density of the magnetic material and WA is the specific core loss in
watts per kilogram per hertz per tesla.
The efficiency rj of the transformer, expressed in percent, is
lOOCV, L - P-,, - PJ
LI
2H L

c
o
i o O - k 1 H - 5 - k 02H 3

(23)

where

2
44.06p(l + W.
)V,I,Z
2 . 1 + 0.2*/1 --+
m
ll
I
T
2

0. iVl - F+

i
[2 + 4Y - 4Y4/1 - ^ - 2Y>/1 - F),
Z(l - Y) \
T
2
/J

- Y)

. .
(24)

o- 1J

and

2
Since Y and F must be less than 1, the functions kj and ^ are always real and
positive.
When equation (23) is differentiated with respect to H the following relation is obtained
^2. ^ S kX^ H " 6 - 3k0H2
9H

(26)

Since a second differentiation would yield a negative value, it can be stated that the
efficiency of the transformer is maximized when
5k n H" 6 - 3kH2 = 0

(27)

or, multiplying through by H and referring to equation (23), when

(28)
Thus as the core dimension H is varied, which is equivalent to varying the current
density when I-, V,, BM, F, f, Y, and Z are all constant, the most efficient design
is reached when copper losses are 60 percent of iron losses.
This result must not be confused with the conventional theorem of the most efficient
operating point for a given transformer. That theorem is derived on the basis that only
current is varied and results in the assertion that the most efficient operating point is
reached when P CxU = P..
i
The expression derived here is of practical importance in any design where current
is not a permitted variable and efficiency is critical.
Solving for H in equation (27) and designating this particular value of H as HM
results in

(29)

1/8

M='

4
0. 734p(l + W.m )V?I?Z
2.1 + 0.24/1- +0.
1 i
L
T 2

- F+

(30)

2 + 4Y - 4YW 1 - i - 2YV1 - F
. - Y) V_
?
2
/

2Y

,B3 f3W2w.Fp
M
f i "

y)

The variation of the losses in the transformer designed to its most efficient point
can be found by substituting HM into equations (21) and (22):
cu

.-1/8

(31)

Therefore, when this criterion is applied in transformer design, the efficiency increases slowly as frequency is increased for constant BM> However, this effect may
be more than offset in practice by the nonlinearity of W. at high frequencies.
The expression for efficiency can be rewritten by substituting equation (29) into
equation (23) to obtain

10

TJ = 100-

Inspecting equation (32) shows that the choice of transformer materials can be demonstrated to effect the most efficient designs with losses increasing as the 5/8 power of
W. and the 3/8 power of p.
The maximum efficiency that can be reached for a toroidal transformer designed
with a fixed Y, Z, and F is represented by equation (32). But the efficiency can be
further improved by varying Y and Z for a particular value of F. The fill factor F
will generally be controlled by the mechanics of transformer construction, so that it will
be considered to be fixed. However, when F is available as a variable, it too can contribute to an ultimate enhancement of the efficiency.
The following expression is obtained by differentiating equation (32) with respect to
Z, holding F and Y constant, and setting the result equal to zero:
8k1
ak9
3k 9 i+Sk!=0
L

* sz

dz

(33)

To implement this expression it is necessary to rewrite equation (24) in the form

(34)

where
44. 06p(l + W-

(35)

B?, f W?F

(36)

C4 = 2.1 + 0.2-|/1--+ 0.1 Vl - F

(37)

and
11

2 + 4Y - 4YV 1 - ^ - 2YV1 - F
C,D =
1
2
1- Y

(38)

Equation (25) is rewritten

(39)

k 9 =_2
Z2

where
C0 = 100?7pMWif ivi/l/^^jc
+ Y\ ,
1 Y

(4Q)

V - /

After the indicated differentiations and substitutions are performed equation (33)
becomes
0. 4C4Z2 + Z(2C3C4 + 0. 7C5) - C3C5 = 0

(41)

Similarly, the variation of Y with Z and F fixed, leads to a maximization of efficiency. In this case equation (24) is rewritten as
/
D6 + D4Y\
DJD + _ - _]
IV o
iv/
1-Y
k = - i/

(42)

\1 - Y 10/

where

44.06p(l + W-m
- )V 1 I 1 Z 2
= - _9 9 9
B
fW F

12

(43)

D, = 2.1 + 0.2 \l 1 - - + 0.1>/1 - F = CA

(44)

(45)

4 - 4 t / l - ^ - 2V1- F

'4* =

<7

(46)

DD5 = ^ Z

(47)

v -TV U + y)

(48)

The expression for ko is rewritten:

1- Y

where

1007rpMMW.fBMM
,
i

=-

49)

Again, differentiating equation (32) with respect to Y and setting the result equal to
zero result in
3k.,
9k9
3kz9 + 5k!x - = 0
3Y
3Y

(50)

The following is obtained by making the necessary substitutions and differentiations:

Y2fp5(7D4 - 4D3 + 3Dg) + 1.3D4 - DS + 0.3Dg1
+ 10D3 - 3D4) - D4 + Dg + 2DJ - 6D5(Dg + Dg)
- l.SDg- D3 - 0.3D4 = 0

(51)
13

The procedure and equations for maximizing efficiency described are applicable to
any toroidal transformer. The particular expressions for kj and kg will depend on
the winding configuration, however. For example, if the basic parallel inverter transformer is considered, that is, a transformer having a center tapped primary designed
for twice the current density of the untapped secondary winding, the following adjustment
must be made in k, to account for the difference in primary resistance.

kj

44.06p(l + W i )V 1 I 1 Z 2 3.2 + 0.34/1-1+ 0.1 V T - F +

1
(3 + 5Y-6Y,.,.
2
=

L
*
Z(l-Y)l
Y
2

,_
2

^|

(52)

2Y

f B F[
.c
M
I Z(l - Y)

The maximization expressions (41) and (51) are unchanged, but new expressions for
C4 and C5 must be used with equation (41) and for D3, D4, and Dg in equation (51).
These expressions are
I

C, = Do = 3 . 2 + 0.34/1 --

.
+ 0.1\1 - F

(53)

3 + 5Y - 6Y4/1 - - - 2YV1- F

(54)
1 -Y

5-6
D4 =
!

(55)

(56)

When equations (41) and (51) are solved simultaneously, the values of Z and Y that
maximize efficiency, ZM and YM, are the result. That efficiency is maximized will be
demonstrated by a numerical example in the next section. The values of YM and ZM
do not depend on the transformer materials, frequency, flux density, insulation thickness,
signal waveform, temperature, or volt-ampere rating; they are functions solely of the
type of winding and the fill factor. This is because Cp C2, Dp and D2 drop out of
14

equations (41) and (51). Therefore, this geometrical relation need be solved only once
for a particular type of toroidal transformer; the result will apply to all future designs.
The mass M of the copper and iron in the transformer can be computed by relying
on the foregoing development. The mass of the copper M is simply
M

PMC(L1AC, wi, 1 + L2Ac, wi,

When equations (3) to (5) and (8), (9), (13), (16), and (17) are substituted into equation (57), the mass of the copper in both the simple and inverter transformer can be
written

TT ft
TTW r
nMC
2Y
MC
2Y
8(1 + W
W.in)) LZ(1
[2(1 - Y)
Y)

Q1

n9
\
J

I
\A
4.2 + 0.' I
I[
T

F
2

/
"

4
/
Z(l-Y)

^S8\

The mass of the iron MA was already expressed in equation (22) as

(59)

It follows that M be the sum of equations (58) and (59):

M = MC + M.

(60)

The current density in the basic transformer is the same in both primary and secondary windings. For the primary of the basic transformer the current density J.. is
written

(61)
A

c,wi,l

When equations (2), (3), and (8) are substituted into equation (61) the current density
becomes
15

-rr-

g
1

2W fB

f M

"m-

(62)

FA A

c W

Further substitutions from equations (18) and (19) into equation (62) yield the expression
for current density in the primary or secondary of the basic transformer or in the secondary of the inverter transformer. This general value of current density will be called
simply J and expressed as
J =

2.353V 1 I 1 Z(1 + W. )
11

(63)

,W f fB M FH 4 [_fL_-0.l] 2

The current density in the primary of the inverter transformer is twice J.

When the expression for HM (given in eq. (30)), is substituted into equation (63) to
obtain JM, the current density at the most efficient point, it found that
J

but is not a function of V

or

APPLICATIONS
That the solutions of equations (41) and (51) do indeed produce values of Y and Z
which maximize efficiency is demonstrated by figure 2 where losses are plotted as a
function of Z and Y for the inverter transformer. The magnitude described in figure 2
depends on the particular choice of materials, power rating, frequency, insulation, and
magnetic flux density; the minimum, roughly in the neighborhood of Z = 1. 5, Y = 0. 85
for a fill factor of 0.5 (fig. 2(a)), would be the same for any transformer of this type.
The minimum at a fill factor of 0.2 for the inverter transformer as seen in figure 2(b)
is near Z = 1.0, Y = 0.9. The data on Supermendur (ref. 3)(49 percent cobalt, 49 percent iron, 2 percent vanadium) was taken from Frost et al. (ref. 4).
A comparison of figures 2 (a) and (b) indicates that the most efficient point tends to
come at higher values of Y and lower values of Z as the fill factor is reduced. This
trend is verified in figure 3 where Y, and ZM, found by iteratively solving equations (41)
and (51) on a digital computer, are plotted as functions of fill factor in figure 3(a) for the
basic transformer and in figure 3(b) for the inverter transformer. Also plotted on fig16

-O

"I
CSJ

CO

CM

I
\O

I
OO

UJUI 'W^

I
OO

lA

~l

'w 'sseui |

-O
I

Cvj"

vo

.3
- o

CM"

ui/v '^r 'A)|suapiuajjno

i-

'It 'A

5T

O
C\J

t^>

o^*

'8J03 UOJ| JO)l)6]8H

lU3DJ3d 'S8SSOT

I
evj

fe"fe

.-O

I
oo

OO

-^-T

oo

i^y^i

CXJ
t

e>j

^j.

1-

CD Ifc-

is!

~ ra 3 '

=>

*:::-

a> -o a,
l/ CI *

3. f gB

G | :

= i !
E

LJ.

" o
^
S C
^
ZJ
W

c &
o c

is S
o> **~

H-

ert<p

.s; .a

s
J5.&0

17

ure 3 for purposes of illustration are the efficiency, transformer mass, current density,
and height of the iron core. These four quantities depend on the specifications stated
there, but Y and ZM depend only on fill factor and transformer type.
The actual values of YM and ZM approximated from figure 2 are found in figure 3 (b). At a fill factor F of 0.2, YM = 0.91 and ZM = 1.07; and at F = 0. 5, YM =
0.85 and Z = 1.35.

CONCLUSIONS
A new rule of thumb emerges from the results of this report. When current and
voltage cannot be varied, but virtually all other design parameters are flexible, the most
efficient design of a toroidal transformer will be that which produces copper losses equal
to 60 percent of the iron losses.
A second result can be stated. For a given fill factor, the shape of the toroidal iron
core can be specified to produce a transformer of the maximum efficiency possible for
the materials and design used.
The first result may be generally true for any type of transformer whose design
equations can be reduced to dependence on a single variable, such as the iron height
used here, since the dimensional relation would probably be the same.
It should be noted that the most efficient design may not be the best design in every
case. For example, by increasing current density somewhat from the value of J-*,
(current density for maximum efficiency), a smaller value of H (height of iron core)
would be specified, and the transformer mass would be correspondingly lowered. The
resulting reduction in efficiency might in many cases be more than compensated by the
decrease in mass.
It should be particularly noted in figure 3 that by the correct choice of Y and Z
efficiency is virtually unaffected by variations in fill factor. For the basic transformer,
efficiency increases by only 0. 6 percent for a variation in fill factor from 0.2 to 0.9.
For the inverter transformer the change in efficiency is 0.7 percent over this range.
Lewis Research Center,
National Aeronautics and Space Administration,
Cleveland, Ohio, February 22, 1972,
112-27.

18

REFERENCES
1. Dayton, James A., Jr.: Toroidal Transformer Design Program with Application to
Inverter Circuitry. NASA TM X-2540, 1972.
2. Rosebury, Fred: Handbook of Electron Tube and Vacuum Techniques. AddisonWesley Publ. Co., 1965, p. 220.
3. Gould, H. L. B.; and Wenny, D. H.: Supermendur, A New Rectangular-loop Magnetic Material. Electrical Eng., vol. 76, no. 3, Mar. 1957, pp. 208-211.
4. Frost, R. MO; McVay, R. E.; and Pavlovic, D. M. : Evaluation of Magnetic Materials for Static Inverters and Converters. NASA CR-1226, 1969, pp. 63 and 65.

NASA-Langley, 1972

10

E-6784