00 SRM D Design For Automotive - 2008 PDF

Design of a
Switched Reluctance Machine Drive

for Automotive Applications
By
Anthony Murphy, B.Eng.
A thesis presented for the award of
MASTER OF ENGINEERING
Research Supervisor
Mr. Jim Dowling
School of Electronic Engineering

DUBLIN CITY UNIVERSITY
July 2008
Declaration
I hereby certify that this material, which I now submit for assessment on the programme
of study leading to the award of Master of Engineering, is entirely my own work, that I
have exercised reasonable care to ensure that the work is original, and does not to the
best of my knowledge breach any law or copyright, and has not been taken from the
work of others save and to the extent that such work has been cited and acknowledged
within the text of my work.
Signed: ____________________
ID No.: ____________________
Date: ___________________
ii
Abstract
Design of a Switched Reluctance Machine

Drive for Automotive Applications
Anthony Murphy
Automotive electrical systems are currently 12Vdc supplied from an engine driven
alternator. In an effort to meet electrical system demands of the future this voltage will
be increased to a 42Vdc standard.
Operating at this increased voltage level will facilitate numerous electronic systems to
be incorporated into the vehicle in order to improve safety, increase fuel efficiency and
reduce emissions. As part of the initiative the starter and alternator will be integrated
into a single unit. The 42-volt bus will allow for a more powerful integrated starter
alternator (ISA).
The switched reluctance motor (SRM) could be considered as a viable solution for
implementing integrated starter alternator. The SRM is a low maintenance, low cost,
rugged, brushless motor without permanent magnets. It has a rugged construction that is
ideally suited for automotive applications with its harsh under-bonnet environment. It
provides high torque and power relative to motor volume. The principal disadvantages
of this type of motor include the motor design and control difficulties Modern motor
design software and lower cost digital signal processing chips for motor control have
overcome these difficulties.
This thesis discusses the automotive electrical systems, switched reluctance motors
(SRM), converter configurations. The operation and the design of a 3.5kW full bridge
converter and its associated control circuitry to facilitate the SRM to operate in starter
and alternator modes are described in detail.
iii
Acknowledgements
I wish to express my sincere gratitude to my supervisor, Mr. Jim Dowling, for all his
guidance, advice and support throughout the duration of this project.
I would like to thank PEI Technologies and DCU for financial support and assistance
during the course of this research.
I am very grateful to the staff in PEI technologies, Liam Sweeney, Ciaran Waters and
Claus Agersbaek, for their guidance and support during the research. They provided
ideas, comments and valuable assistance throughout.
Thanks also to the other PEI postgrads and exchange students, Amra Pasic, Francis
Leonard (RIP), Markus Florian and Ralf Schlosser, for their help, support and
suggestions.
A special thanks to Eoin Kennedy for assistance in the areas of software development
and technical writing and for the opportunity to contribute to a publication.
I also wish to thank all the technical staff of the school of Electronic Engineering, who
provided the necessary equipment and help to complete this research.
Thanks to Breda McManus for her kind words of support and encouragement.
To my fiancée, Shona, my most profound thanks for your ceaseless support and
encouragement and your constant understanding. It has made all the difference.
Finally, I wish to thank my family and friends for all their encouragement and support.
iv
Contents
Declaration ii
Abstract iii
Acknowledgements iv
Contents v
1 Introduction 1
1.1 Motivation overview……..….…….………………………………………8
1.2 Outline of thesis………………………………………………………….. 5
2 Operating principles and features of the SRM 8

2.1 Historical review…………………………….…………………………….8
2.2 Fundamentals…………………………………………………………….13
2.3 Mathematical modelling of the SRM…………………………………….21
2.3.1 Analysis of magnetically linear operation ………………….. 21
2.3.2 Analysis of magnetically nonlinear region…………………..24
2.3.3 Analysis of generator operation……........................................27
2.4 Dynamic operation of the SRM………..………………………………. 29
2.4.1 Low speed motoring…………….………..………………… 30
2.4.2 High speed motoring……………………………..………… 33
2.4.3 Low speed generating………………………………………..34
2.4.4 High speed generating……………………………………….38
2.5 Summary …………………………………………….....……………….39
3 Converter topologies 41
3.1 Selection criteria and topology groups…...….…………………………. 41
3.2 Bridge converters….………………………..………………………….. 43
3.2.1 Asymmetrical half-bridge converter………………………. 43
3.2.2 Shared phase winding converter …….………………...….. 45
3.2.3 Shared switch converter…………………………………….47
3.2.4 Full-bridge synchronous rectifier converter………………...48
v
3.3 Capacitive energy recovery converters……………………………...…. 50
3.3.1 C-dump converter……………………………………………50
3.3.2 Buck-boost converter………………………………………...53
3.3.3 Split DC converter…………………………………………...55
3.3.4 Sood converter……………………………………………….57
3.4 Magnetic energy recovery converters ……………………. …………...58
3.4.1 Bifilar converter……………………………………………...59
3.4.2 Auxiliary commutation winding converter…………………..60
3.5 Dissipative converters…………………………..……..………………. 62
3.5.1 R dump converter………………………………………….…62
3.6 Self-commutating converters……………………………………….…. 64
3.6.1 Series resonant converter………………………………….…64
3.6.1 H-bridge converter…………………………………………...66
3.7 Converter selection………………………………..…………………....67
4 Preliminary analysis 69
4.1 Operating conditions………………………………….…………...……69
4.1.1 SRM specifications……………………………………....….. 70
4.1.2 Performance conditions………………………………………70
4.1.3 Operating environment………….…….……..……………….73
4.1.4 Conduction periods and duty cycles………………………….74
4.1.4.1 Starting and low speed motoring………….………..75
4.1.4.2 Low speed generating………………………………76
4.1.4.3 High speed generating…………………………...…77
4.1.5 Automotive electrical parameter limits………………………79
4.2 Switching component analysis………………………..……….……….81
4.2.1 Asymmetric half-bridge converter……………………...……82
4.2.1.1 Efficiency during starting mode………………..…..87
4.2.2 Full-bridge synchronous rectifier configuration….……...…..89
4.2.3 Parallel configuration……………………….………………..90
4.3 Reverse recovery analysis………………………………….……..……93
4.3.1 Simulation analysis………………………………..…………93
4.3.2 Experimental analysis……………………………..…………95
4.4 Summary………………………………………………………….……97
vi
5 Converter design 99
5.1 MOSFETs in parallel……………………………………………...…. 99
5.2 Heatsink design………………………………………………………102
5.2.1 Fundamentals………………………..……………………...102
5.2.2 System description…………….……………………………105
5.2.3 Determination of Rθs-a ……………………………….……..108
5.2.3.1 Determination of Rθ for heat spreader…………….109
5.2.3.2 Determination of junction temperature, Tj………..111
5.2.4 Step-by-step determination of Rθs-a – starting mode…..……111
5.3 Gate drive design……………………………………………...……..115
5.3.1 Gate drive configuration……………………………………116
5.3.2 Determination of peak gate drive current…………………..119
5.3.3 High current buffer……………………………………...….121
5.3.4 Determination of peak current for buffer gate drive circuit..122
5.3.5 Bootstrap components…………………………………...…123
5.3.5.1 Calculating the bootstrap capacitor value………....123
5.3.5.2 Selecting the bootstrap diode…………………...…125
5.3.5.3 Gate resistor selection………………………….….125
5.3.6 Blanking time delay……………………….………………..126
5.3.7 Electrical isolation………………………………………….126
5.4 DC source capacitor analysis……………………………………...…128
5.4.1 Capacitor parameters…………………………………….....128
5.4.2 Specific application requirements…………………………..128
5.4.3 Determining capacitance value………………..………….129
5.5 Bus bar analysis………………………………………………...……131
5.5.1 Bus bar implementation………………………………….....131
5.5.2 General bus bar structure…………...............................……132
5.6 Current sensors……………………………………………………….132
5.6.1 Shunt resistor……………………………………………….132
5.6.2 Current transformer……………………………………...…133
5.6.3 Hall effect transducer……………………………………....134
5.6.4 Sensor selection and implementation………………...…….136
5.7 Circuit layout considerations………...………...……………..………137
5.7.1 Full-bridge circuit……………………………………….…137
5.7.2 MOSFET gate driver………………………………………139
vii
5.7.3 Gate drive buffer circuit…………………………..……….139
5.7.4 Optocoupler electrical isolation…………………………...139
5.7.5 Bus bars……………………………………..…………….140
5.7.6 Current sensor…………………………….……………….141
5.8 Final converter assembly………………………………………..……141
5.9 Summary……………………………………………………………...143
6 System implementation 144

6.1 Hardware….…………………………………………………………. 144
6.1.1 The SRM………..………………………………….……145
6.1.2 The digital torque measurement system………..…….….146
6.1.3 Experimental RL load setup and 1kW power supply…....147
6.1.4 The LV power supply and interface unit…………..…….147
6.1.5 DSP controller………………………………………........148
6.1.6 Converter prototyping……………………………...….....151
6.1.6.1 Circuit layout……………….……………..……….151
6.1.6.2 Mechanical layout…………….………….......……154
6.1.6.3 Converter assembly…………….……………….....156
6.1.6.4 Preliminary converter analysis…….………………158
6.2 Software……………………………………………………………....158
6.2.1 Control structure……………………………...………….159
6.2.2 Control implementation……………………………….....162
6.2.3 Design verification…………………………………….…167
6.2.3.1 Preliminary converter and source code evaluation...167
6.2.3.2 Current controller evaluation……………...……….172
6.3 Summary……………………………………………………...………176
7 Conclusion and recommendations 177

7.1 Conclusions……………….………………...…………………...……177
7.2 Recommendations……..……………………………………………...178
Bibliography 180
Appendix A: Power analysis derivations A1

A.1 Analysis equations………..………....….….………………….…….. A1
A.2 Converter efficiency and power transfer…………………………..…A2
viii
A.2.1 Starting and low speed motoring………………………...A6
A.2.2 Low speed generating………………………............……A7
A.2.3 High speed generating……………………………....….A10
A.3 Component Analysis………………………………...……………....A13
A.3.1 Asymmetrical half-bridge…………………............……A15
A.3.1.1 Starting and low speed motoring – Q1…………...A16
A.3.1.3 Starting and low speed motoring – D1…….……..A18
A.3.1.4 Starting and low speed motoring – D2….………..A19
A.3.1.5 Low speed generating – Q1……………….……...A20
A.3.1.7 Low speed generating – D1………………..……..A22
A.3.1.8 Low speed generating – D2….…………….……..A23
A.3.1.9 High speed generating – Q1…………...….……...A24
A.3.1.10 High speed generating – Q2……….......…...…...A25
A.3.1.11 High speed generating – D1………………...…..A26
A.3.1.12 High speed generating – D2….…………............A27
A.3.2 Full-bridge synchronous rectification…….............……A28
A.3.2.3 Starting and low speed motoring – Q3…….……..A29
A.3.2.4 Starting and low speed motoring – Q4….………..A31
A.3.2.7 Low speed generating – Q3………………..……..A32
A.3.2.8 Low speed generating – Q4….…………….……..A34
A.3.2.9 High speed generating – Q1……………………...A35
A.3.2.10 High speed generating – Q2……………..……...A35
A.3.2.11 High speed generating – Q3………………...…..A35
A.3.2.12 High speed generating – Q4….……………..…..A37
ix
Appendix B: Power analysis computations B1
B.1 Operating conditions......……..……..…………………………………B2
B.2 Asymmetrical half-bridge – single devices……………………...……B3
B.2.1 MOSFET – ATP20M11FLL, Diode – SSTH20002TV…....B3
B.3 Full-bridge with synchronous rectification………….…………...……B4
B.3.1 MOSFET – ATP20M11FLL – single devices………......…B4
B.3.2 Parallel MOSFETs – FDP047AN08A0 – 2 devices…….…B5
B.3.4 Parallel MOSFETs – FDH038AN08A1 – 2 devices…….…B7
B.3.5 Parallel MOSFETs – FDH038AN08A1 – 4 devices…….…B8
B.3.7 Parallel MOSFETs – FDP060AN08A0 – 4 devices…...…B10
B.3.8 Parallel MOSFETs – SPP80N08S2L-07 – 2 devices…......B11
B.3.9 Parallel MOSFETs – SPP80N08S2L-07 – 4 devices…..…B12
B.3.10 Parallel MOSFETs –BUK75/7606-75B – 2 devices….…B13
B.3.11 Parallel MOSFETs – BUK75/7606-75B – 4 devices……B14
B.3.12 Parallel MOSFETs –PSMN009-100W – 2 devices…...…B15
B.3.13 Parallel MOSFETs – PSMN009-100W – 4 devices…..…B16
Appendix C: Thermal analysis computations C1

o
C.1 FDP047AN08A0 – 2 devices – Ta = 45 C......……..…………C4
C.2 FDP047AN08A0 – 4 devices – Ta = 45oC......……..…………C5
C.7 FDP060AN08A0 – 2 devices – Ta = 90oC......………………C10
C.8 FDP060AN08A0 – 4 devices – Ta = 90oC......………………C11
C.9 BUK75/7606-75B – 2 devices – Ta = 45oC......………...……C12
C.10 BUK75/7606-75B – 4 devices – Ta = 45oC......………….…C13
C.11 BUK75/7606-75B – 2 devices – Ta = 90oC......……….……C14
C.12 BUK75/7606-75B – 4 devices – Ta = 90oC......……….……C15
x
Appendix D: Experimental system D1
D.1 Schematics……………………………….......……..…………D1
D.2 Photographs of the experimental system.........……..…………D2
Appendix E: ADSP-21992 EZ-KIT lite evaluation system board E1

E.1 Overview of the analogue to digital conversion unit..…………E2
E.2 Overview of the PWM generation unit..………………….……E4
E.3 Overview of the auxiliary PWM generation unit..………..……E7
E.4 Overview of the peripheral interrupt unit………….……..……E9
E.5 Overview of the Flag I/O peripheral unit..……………....……E14
Appendix F: VisualDSP++ 3.0 and implementing the ISR F1

F.1 Overview of VisualDSP++ 3.0 features……………..…………F1
F.2 Implementation of the intrinsic functions……........……………F2
F.3 Implementation of the ISR in C………………..………………F4
Appendix G: DSP implemented source code G1

G.1 srm_pipwm_current_reg.c…………………………..…………G1
xi
Chapter 1 – Introduction
Energy efficiency has become increasingly important in the 21st century due to the
increased cost of fossil fuel resources combined with the need to reduce the levels of
greenhouse gases. Combining the capabilities of an electric machine to act as a motor
and a generator in a system significantly reduces the carbon footprint of that system by
eliminating the need for an extra machine. In order to maximise the advantage of this
system integration, the electrical machine must be very efficient and require drive
circuitry of limited complexity while providing equal or better levels of efficiency.
In the automotive environment, the increase in fuel efficiency and the reduction in
greenhouse gas emissions have become critically important issues. The reduction in the
number of motors used, together with providing new opportunities to harness their
regenerative capabilities, is an area under continuous investigation. The integrated
starter-generator (ISG) is an immediate application for the electric machine to operate as
a motor and a generator in the automotive environment. This has expanded into the area
of mild hybrid through to a full hybrid electric vehicle (HEV) that provides varying
degrees of propulsion assistance and regenerative braking.
Vehicles with an ISG and a 42V electrical system are defined as a mild hybrid (Jones
2003). The ISG allows for start-stop mode of operation, where the engine can be shut
down when the car is braking or stopped but when required it can be restarted quickly
and cleanly. At start up the large motor allows the engine to spin up to operating speed
before commencing the engine combustion process. It can also provide some levels of
propulsion assistance and, when the car is slowing down, energy can be recovered
through regenerative braking. All these capabilities contribute to improving the vehicle
fuel efficiency and reducing emissions.
The HEV uses an internal combustion engine (ICE) together with batteries and electric
motors as propulsion sources (Sreedhar 2006). The batteries and machine in this
application can provide some or all vehicle propulsion capabilities and the motor also
provides regenerative braking capabilities when the vehicle is slowing. The ICE that is
currently fuelled by petroleum products (petrol, diesel, etc) is transitioning so that it can
also operate with various forms of biofuels (bioethanol, biodiesel, etc) in order to reduce
1
the dependency on crude oil. The increasing popularity of hybrid vehicles, for example
the Toyota Prius, together with the sharp increase in cost of petroleum in recent years
has led to intense interest in the application of electric machines for the purpose of
vehicle propulsion assistance and regenerative braking capabilities.
1.1 Motivation
The switched reluctance machine (SRM) is particularly suitable for use in automobiles
(where there is a harsh environment of high temperature and vibration) as it has a
rugged construction and is a brushless motor without permanent magnets (DiRenzo
2000). The SRM is one of the oldest electric motor designs having been used in
industrial applications as far back as the 1830s. Despite its advantages, it has been
largely unused for many years because of design and control difficulties.
The SRM has a very simple construction as shown in Figure 1.1(a). Only the stator has
windings, while the rotor has no conductors or permanent magnets. The windings on
two opposing stator poles are connected to create a single phase winding. Both the rotor
and stator poles are salient in order for the machine to produce reluctance torque. A
magnetic field is created when the stator phase-winding is energized and reluctance
torque is produced by the movement of the rotor to its minimum reluctance position.
Rotor Stator
Poles Poles
Rotor
Phase
Winding
Stator
Figure 1.1: The SRM Layout
2
The saliency of both the stator and rotor poles creates non-linear magnetic
characteristics which complicate the analysis and control of the SRM. The control
problem arises because phase to phase switching of the drive current must be precisely
synchronised with rotor position for optimal operation under motoring and generating
conditions. Magnetic circuit analysis software has reduced the problems associated with
SRM design, while the arrival of powerful, yet moderately priced, digital signal
processors (DSP) has enabled the control to be addressed in a cost effective manner.
These advances have fuelled a new interest in the SRM for a wide range of consumer
and industrial applications.
The work presented in this thesis was undertaken as part of a larger research project
entitled “Switched Reluctance Design for Automotive Applications”. The project was
carried out by three PEI Technologies (PEI) centres: the University of Limerick
Analogue Centre (ULA), University College Dublin (UCD) centre and the Dublin City
University (DCU) centre. The Dublin City University (DCU) centre was the principal
investigator and had overall management responsibility for the project.
The DCU centre was responsible for selection and design of a prototype machine
controller. This involved the selection and design of generic models for the SRM as a
motor and as a generator, the simulation and implementation of closed-loop control of
the SRM under motoring and generating conditions, and the design and construction of
a suitable power converter for the SRM and the construction of a laboratory test
apparatus. The key sections are illustrated in Figure 1.2.
Figure 1.2: Key project sections for the Switched Reluctance Design for Automotive
Applications
3
The specific part discussed in this thesis is the design and construction of the power
converter, the closed-loop current control strategy and the laboratory test apparatus. The
main purpose of the power converter design and the control scheme for the SRM was to
maximise efficiency when operating as a motor or a generator in low voltage, battery
powered applications. The development of the converter is divided into five stages:
converter selection, preliminary analysis, converter design, experimental setup and
converter control implementation.
The converter is targeted for operation under the 42V automotive specification in order
to maximise the capabilities of automotive power semiconductors. In order to meet the
additional requirements of future electrical systems the automotive industry has
established a new 42V voltage standard (42V PowerNet) to replace the current 14V
system (Kassakian 1996) and (Kassakian 2000). Improvements in fuel economy and
reduction in emissions are the main arguments for the introduction of the 42V standard.
A number of converter topologies were reviewed, the details of which are discussed.
The asymmetrical half bridge and full bridge converters were selected for preliminary
analysis. The analysis includes the evaluation of a number of semiconductor devices to
determine component efficiency, based on a range of operating conditions that are
discussed in detail and to verify reverse recovery times. The full-bridge converter using
synchronous rectifier operation proved to be the most efficient solution for both
motoring and generation over a wide range of operation.
Configuring of semiconductors in parallel was central to the implementation of the full-

bridge converter design and this required particular attention to power and thermal
analysis, layout of the printed circuit board and the implementation of specific control
strategies in order to achieve an efficient design operation. Discrete MOSFET
semiconductors mounted in parallel significantly increase the current handling
capability of the circuit but numerous design issues had to be addressed, including stray
inductances, circuit layout and thermal coupling.
Control strategies were designed to coordinate the full-bridge circuit switching signals
for both motoring and generating modes with synchronous rectifier capability, to
4
employ blanking time capability and to implement accurate proportional-integral pulse
width modulated (PI-PWM) current control.
Two pieces of laboratory test apparatus were constructed as part of this project. The first
apparatus consisted of the inductive load, the DSP evaluation board (to implement the
switching and current control strategies), low voltage and high voltage power supplies
to power the converter. It was constructed in order to evaluate the converter operation,
to develop the switching strategies for motoring and generating and, from there, the
development of the current control strategy. The second laboratory apparatus consisted
of an SRM coupled with a load/drive machine and torque measurement system. This
apparatus was constructed for the integration of the converter with the SRM control
algorithms for motoring and generating.
The contribution of this work is to illustrate that this converter configuration together
with the control strategy would improve the efficiency of the switched reluctance drive
system operating in both motoring and generating conditions. This topology allows
independent operation of each SRM phase to maximise control flexibility and to enable
the converter to be applied to an SRM with any number of phases. The laboratory
apparatus facilitates evaluation of SRM control strategies under motoring and
generating conditions to further enhance this machine’s application potential.
1.2 Overview of thesis

The fundamental principles of operation of the switched reluctance machine under
motoring and generating conditions are discussed in Chapter 2. The essential concepts,
modes of operation and benefits and limitations are presented for the machine. The
mathematical model for torque production is investigated for both motoring and
generating modes and a review of the dynamic operation of the machine under motoring
and generating conditions is presented for the machine. The chapter also includes a
historical review of the SRM.
Chapter 3 reviews a variety of converter topologies in order to determine the most

suitable selection for this application. The converter topologies are classified according
to a predominant feature and a description of operation and main advantages and
5
restrictions are then presented for each converter. The selection of the most suitable
converter topologies for preliminary analysis is then presented.
The preliminary analysis of semiconductor components to determine the most efficient

converter configuration is discussed in Chapter 4. The operating conditions are defined
in order to set limits of converter operation and these include: SRM specifications,
operating environment and electrical requirements. A detailed component power
analysis to determine the most efficient semiconductor solution is then presented. The
simulation analysis and electrical tests on the most efficient components to determine
the final semiconductor selection together with the most suitable converter configuration
are then examined.
Chapter 5 discusses the final converter design, including a review of the issues that
needed to be addressed when implementing MOSFETs in parallel: the design of
heatsinks, gate drive circuitry, the determination of the dc-link capacitor, bus bar
configuration, the current sensing circuit and the specific printed circuit board layout
considerations.
The experimental setup is described in Chapter 6. A description is given of the

laboratory apparatus consisting of an SRM coupled with a load/drive machine and
torque measurement system for the purpose of integrating the converter and the SRM
control algorithms for motoring and generating operation. The laboratory apparatus to
evaluate the converter under various load conditions and to implement the switching and
current control strategies is then discussed. The key components in this apparatus
include the inductive load, the DSP evaluation board and low voltage and high voltage
power supplies. A detailed description is provided of the converter prototyping
including the circuit board layout and the mechanical layout of the converter through to
final assembly of the complete converter.
Chapter 7 describes the development and implementation of the current control strategy
implemented with the converter. A detailed description is given of the converter
switching sequence including experimental results. The current control strategy,
including the implementation of the proportional integral controller, is outlined and
experimental results presented to illustrate converter operation at low and high current.
6
Finally, a summary of the thesis, conclusions and suggestions for possible future
research are given in chapter 8.
7
Chapter 2 – Operating principles and features of
the SRM
In order to design a power converter for the SRM to operate as both a motor and a
generator a detailed understanding of the machine is required. Knowledge of the
dynamic modes of operation of the SRM as a motor and a generator is central in order to
define the current and voltage waveforms applied to each machine phase. These
waveforms are used to implement a detailed power analysis for each of the switching
components in the converter. To understand the dynamic operation of the SRM, the
fundamental principles of the machine are first discussed and this is followed by a
detailed mathematical analysis. A historical review provides a chronological history of
the SRM up to the present day.
2.1 Historical review

Developments in the switched reluctance machine can be divided into different periods
and under specific categories. Early evolution in this machine would be defined as the
period up to the 1960s and modern developments would be defined thereafter. The
specific categories would include: the developments in the machine design from its
early beginnings through to its present day configuration implemented as both a motor
and a generator; the progress in converter design, the growth in control strategies for
motoring, generating and sensorless control operation; and, finally, applications of the
SRM both past and present and possibly in the future.
Early switched reluctance machines were known as “electromagnetic engines”, due to

the strong influence of steam engine design, and were some of the first machines
developed in the 1830’s and 1850’s (Miller 2001). It began with the horseshoe
electromagnet of William Sturgeon in 1824 and was improved upon by Joseph Henry in
1831. A model for the electromagnetic behaviour was initially developed by Rev.
William Richie in 1833. It was not until John Hopkinson’s work on magnetic circuits
that the electric machine could be designed systematically. These early machine designs
were largely by trial and error and some examples include machines by Rev. Nicholas
Callan in the late 1830’s, a number of his machines can still be seen in Saint Patrick’s
College, Maynooth [(Miller 2001) and (IEEE 2006)], the machines built for Charles
8
Wheatstone by William Healey in 1842, examples built by Taylor and others built by
Davidson (Miller 2001). They all had very poor magnetic circuits due to the use of solid
instead of laminated iron for the rotor and stator. They also had no means to recover the
stored inductive energy and the pulsating forces yielded structural problems. These
machines were superseded by the ring wound armature dc motor by Pacinotti (1864)
and Gramme (1869) and the polyphase induction motor of Tesla (Keville 1994) that
provided vastly superior performance.
The transition to the modern era of what is now termed the ‘switched reluctance
machine’ occurred in the mid 1960s and it was brought about by four parallel
advancements in industry: developments in power switching devices (thyristors and
power transistors); the evolutions of microprocessors and digital circuitry to implement
control algorithms; the development of high speed computers to analyse and design the
machine using advanced programming languages; and, finally, the general expansion in
the application of the machine and the control systems. According to (Miller 2001),
some of the key features of modern reluctance machines and their drives were
developed at Queen’s College, Dundee, in the early 1960s, specifically in the area of
motor design, electric switching circuits using thyristors and open-loop and closed-loop
control.
In the late 1960s through the early 1970s, Prof. Byrne and his colleagues at University
College Dublin published significant material on the influence of saturation to improve
energy conversion efficiency of the machine [(Byrne 1973) and (Byrne 1976)] that was
verified by [(Miller 1985) and (Stephenson 1989)]. They also did valuable original work
on machines with low phase numbers (Byrne 1973), electronic drives with fewer power
switches (Byrne 1985) and addressed the issue of current imbalance on machine
performance (Devitt et al 1981). Burnice Bedford of the General Electric Company,
USA, published two patents in 1972 describing all the key features of the modern SRM
and the controlled switching circuits [(Bedford 1972a) and (Bedford 1972b)].
Around this period, Lawrenson and Stephenson at the University of Leeds, began work
on the switched reluctance machine and their research efforts culminated in the
landmark paper published in 1980 (Lawrenson 1980). It addressed the many design
issues regarding number of poles, pole arcs and choice of phase numbers and described
9
the favourable static and dynamic performance of the machine in four quadrants when
compared to an induction motor of equivalent size. This paper marked the beginning of
renewed interest in the switched reluctance machine that continues to this day. Their
research work lead to the formation of SR Drives Ltd. (bought out by Emerson in 1994)
and produced about half the commercial applications for the machines in the early 1980s
(Miller 2001). Since 1987 Professor Miller and his colleagues of the Speed Consortium
at the University of Glasgow have made significant developments in the design and
analysis of the SRM and developed a widely used CAD program called PC-SRD
[(Miller 1990) and (Fuengwarodsakul 2005)].
Up until the 1930s, electronic switching between phases in electrical machines was
controlled using a commutator (Miller 2001). Early electronic switching circuits
consisted of mercury arc rectifiers or thyratrons but offered poor reliability at a high cost
and this severely limited the application of these devices. The advent of thyristors and
power transistors in the early 1960s allowed for the development of reliable (but still at
a relatively high cost) electronics switching circuits to be developed for electrical
machines. Dr. Cruickshank and his colleagues at Queens College, Dundee, developed
some of the first switching circuits using thyristors for the SRM and also developed
some of the fundamental switching techniques for these machines (Miller 2001).
Burnice Bedford of General Electric (GE) Company, USA, developed all the key
features of the modern controlled electronic circuits for the SRM by exploiting new
power semiconductors being developed by GE and the company’s capability to
manufacture a wide range of speciality motors. He initially developed the asymmetric
half-bridge converter (Bedford 1972b) which was further refined and developed by Ray
(Ray 1979) and Lawrence (Lawrence 1980). He also originally developed the split-dc
converter (Bedford 1972a) that was further refined by Graseby Controls Ltd. UK
(Miller 1993). Early work on the C-dump topology was implemented by Prof. J.V.
Byrne (Byrne 1985) and further developed in detail by J.T. Bass (Bass 1987).
During the 1980s through the 1990s, significant developments took place in electronic
switching circuits for the SRM due mainly to the advances in semiconductor
technologies from the development and advances in microprocessor design. This led to
greater power handling capabilities for the power transistor beginning with the bipolar
10
junction transistor (BJT) but more significantly the MOSFET and the IGBT. These
devices allowed for forced commutation and current regulation using pulse-width
modulation working at switching frequencies higher than was possible with the
thyristors. Numerous topologies facilitating different circuit configurations and
component count were realized during this period and up to the present day. Examples
of these converters include: the R-dump converter (Krishnan 2001), the Sood converter
(Sood 1992), the series resonant converter (Park 1992) and the full-bridge converter
(Krishnamurthy 2005). A useful comparison of many different converter topologies are
included in both (Vukosavic 1990) and (Barnes 1998). A detailed review of a number of
converter topologies is presented in chapter three.
The development of microprocessors and microcontrollers and related digital circuits

during the late 1970s and the early 1980s allowed for the implementation of complex
control algorithms and ushered in a new era of control capability for the SRM (Chappell
1984). Reduction in the torque ripple that is inherent in SR motor operation was a key
area that was targeted using these complex algorithms. A comprehensive review of the
different control approaches to torque ripple reduction is provided by (Husain 2002).
Since the 1990s, the operation of the SRM as a generator has received considerable
attention. Radun emphasised the need for closed loop control of the switched reluctance
generator (SRG) when he discussed the instability of the open loop operation of the
SRG with fixed turn-on and turn-off angles in an important paper from 1994 (Radun
1994). The SRG has been under investigation for application as an automotive
starter/generator [(Kokernak 1999), (Mese 2000) and (Fahimi 2001)], wind power
generation [(Torrey 1993) and (Cardenas 1995)] and various aero space applications
[(MacMinn 1989), (Radun 1994), (Radun 1998) and (Cossar 2004). A thorough
overview of the development of SRG control is provided by (Miller 2001) and (Torrey
2002).
In recent years, significant research has being directed towards the operation of the
SRM without a position sensor. A number of different methods of indirectly estimating
the rotor position have been proposed but they all make use of the inductance variation
in one way or another. The aim of these ‘sensorless’ position estimation schemes is to
11
reduce system cost while also increasing system reliability. A detailed review of the
various approaches is provided by both (Husain 1996) and (Ehsani 2002).
Products Company
Air-handler A.O Smith

EV drives Aisin Seiki
High-speed motors and controllers AMC NEC/Densei
Commercial vacumm products Ametek Lamb Electric
High speed centrifuge systems Beckman Instruments
Electric doors Besam A/b
Mining drives British Jeffrey Diamond
Energy saving screw air compressor CompAir Ltd.
Automotive cruise control DANA Corp.
High air flow-rate hand-dryer Dyson UK
250kW low-speed drive Elektro Magnetix Ltd
Washer drive Emerson/SRDL
Pumps,HVAC motion control Emotron A/b
Plotter drive Hewlett-Packard
Several Mavrik Motors
Train air conditioning Normalair Garrett
Megatorque direct-drive NSK Ltd
General purpose industrial drives Oulton, Tasc Drives
Weaving machine servos Picanol
Fork lift/pallet truck drive Radio Energie
Starter/Generator for articulated vehicle SR drives Ltd.
Mild hybrid-electric powertrain SR drives Ltd.
3kW 12V belt-drive starter-generator SR drives Ltd.
Industrial drives Sicmemotori
VSR high pressure pumps Weir Group Plc.
Table 2.1: Switched reluctance applications
While a significant number of developments have occurred with the SRM over the past
number of decades, it has yet to receive significant acceptance in industry. Some of the
most popular applications for the SRM include the Maytag Neptune washing machine
and vacuum cleaners by Ametek, LG and Dyson (Fiedler 2005). In addition Dyson
introduced a new hand dryer in October 2006 that used a SRM operating at 100,000
rpm to generate a 180ms-1 air flow-rate that is claimed to dry hands in 10s (Nathan
2006). A number of current and previous applications of the SRM are provided in Table
2.1 [(Krishnan 2001), (Miller 1993), (Miller 2002), (SR drives Ltd 2007a), (SR drives
Ltd 2007b) and (Dyson UK 2007)].
12
2.2 Fundamentals
Electrical machines can be broadly defined into two categories by how they produce
torque: either electromagnetically or by variable reluctance (DiRenzo 2000). With the
first type, motion occurs due to the interaction of two mutually coupled magnetic fields,
one generated by the stator and the other by the rotor. This produces electromagnetic
torque, which tends to bring the magnetic fields into alignment. DC and induction
motors operate on this principle. In the second type the motion is produced by variable
reluctance in the air gap between the rotor and the stator. Switched reluctance machines
fall into this second group.
The switched reluctance machine is an extremely simple electrical machine. The

machine is doubly salient in that it has both salient rotor and stator poles made of
magnetically soft, low loss steel laminations. The rotor pole does not contain magnets or
windings, while the stator pole has windings on each pole, with one phase consisting of
two diametrically opposite poles wound in series to form a single phase winding.
Numerous combinations of phase, rotor pole and stator pole combinations are available
to yield different SRM geometries. The phase winding and the cross section of a four
phase, eight stator pole and six rotor pole arrangement of an 8/6 four-phase SRM is
illustrated in Figure 2.1 (a). A typical converter arrangement is shown in Figure 2.1 (b).
Figure 2.1: (a) Cross-section of an 8/6 four-phase SRM. (b) Classic converter.
13
Magnetic flux is generated when current flows in a phase winding and results in torque
being produced due to the tendency of the rotor pole to align with the stator pole in
order to yield a minimum reluctance path. In Figure 2.1 (a), the rotor and stator pole
pairs, R1 and S1 illustrate an unaligned position and R2 and S2 illustrate the aligned
position. The flux-linkage and therefore the phase inductance varies with rotor position
where it is at the minimum value (Lu) at the unaligned positions (θ1and θ4) and rises to a
maximum value (La) at the aligned positions (θ2 and θ3) as illustrated in Figure 2.2.
Physically, position θ1 corresponds to the start of overlap where the leading edge of the
rotor pole is aligned with the leading edge of the stator pole and θ4 corresponds to the
end of rotor and stator pole overlap.
Figure 2.2: The variation of idealised phase inductance with rotor position
The rotor angle at which the voltage is applied across the phase winding is called the
turn-on angle and it is removed at the turn-on angle, while the difference in these angles
is defined as the dwell angle or firing angle. The complete cycle from the start of one
variation in inductance cycle until it repeats again is called the pole pitch. The region
where the phase inductance remains constant is defined as a dead zone. The rotor speed
at which the back-emf and the applied voltage across the phase winding are equal in
magnitude is defined as the base speed.
14
Figure 2.3: The idealised phase inductance, current and torque for motoring operation
The SRM can operate either as a motor or as a generator depending on the range of
rotor angles at which the voltage is applied to the phase winding. The machine’s
operation is independent of the direction of current in the phase windings but is a
function of rotor position with respect to the energised phase. Accurate rotor position
information is essential to ensure smooth operation of the SRM when motoring and
generating. The term “switched reluctance” refers to the switching of sequential phase
excitations that achieve the rotor rotation (Miller 2001).
Figure 2.3 shows the variation in idealised phase inductance and torque with rotor angle
for a constant phase current. For motoring operation, the firing angles are selected so
that current flows in the phase winding when the rotor and stator poles approach
alignment which is when the phase inductance in increasing and this produces a positive
motoring torque in the direction of rotation. The electrical energy consumed in the
alignment of the rotor pole and stator pole pair is converted to mechanical energy in
driving a load attached to the rotor shaft. For generating operation, the firing angles are
selected so that the current flows in the phase winding immediately after the rotor and
stator poles have passed alignment when the phase inductance is decreasing. The rotor
produces negative or braking torque opposing the direction of rotation. Generating
occurs when the mechanical energy expended by a prime mover in pulling the rotor pole
away from the excited stator pole is converted to electrical energy.
15
To illustrate the fundamental operation of the SRM as a motor and as a generator, only
motoring operation at low speed, and generating at the base speed, are discussed here.
The SRM operation as a motor and a generator at other speeds is discussed in Section
2.3.
Under motoring operation it is clear that negative torque is undesirable. For this reason,
the phase current must be reduced to zero before the θ3 – θ4 interval when the poles are
separating (The operation in the intervals θ1 – θ2 and θ2 – θ3 are discussed in the
following paragraphs).Therefore, the ideal motoring current waveform should be a
series of pulses where the position of the pulses coincides with the rising inductance
interval. Thus, the ideal motoring torque waveform for a single-phase SRM has the
shape shown in Figure 2.4.
Figure 2.4: The idealised phase inductance, current and torque for motoring operation
The inductance at the unaligned position (θ1) tends to be a very small value resulting in
the value of current in the phase winding increasing very quickly. The rectangular
current waveform (current pulse) can then be approximated by chopping the current
along θ1 – θ2. To avoid production of negative torque the current must be reduced to
zero before alignment at θ3 and this can be achieved by reversing the voltage across the
phase before alignment.
16
As can be seen in Figure 2.4, the single-phase SRM produces regions of discontinuous
torque. Continuous rotation of a single-phase machine relies on either the momentum of
the machine when motoring or on the prime mover when generating. Similarly, the
machine cannot start from every position, as at least two phases are required to
guarantee starting and at least three phases are required to ensure starting direction
(DiRenzo 2000). It is shown in Section 2.3 that the torque can only be produced in
regions of increasing and decreasing inductance. Hence, no torque is produced in the
‘dead zones’ between θ2 and θ3 at the aligned position and in the unaligned position
between θ4 and θ1.
To produce continuous positive motoring torque and to ensure self-starting capability

from all positions, the full 360 degrees of rotation must have areas of increasing
inductance so that appropriate placed current pulses produce continuous torque output.
Figure 2.5 shows the idealised inductance, current and torque waveforms for the four-
phase 8/4 SRM of Figure 2.1. The energy conversion (torque production) that takes
place in discrete cycles through the interaction of one rotor and stator pole pair is called
a stroke. The number of strokes per revolution, S, is related to the rotor poles, NR, and
the number of phases, P, by S = NRP. Hence, for the 8/6 four-phase machine in Figure
2.1(a), S = 6(4) = 24.
Figure 2.5: The idealised phase inductance, current and torque waveforms for a four-
phase 8/6 SRM under motoring conditions
Under generating operation, the ideal generator current should be a series of pulses
where the position of the pulses coincides with the falling inductance interval as shown
17
in Figure 2.6. The rectangular current waveform can be approximated when the back-
emf and the applied voltage balance and the current stays constant until pole overlap
ends. Current pulses must be applied at areas of decreasing inductance over the full 360
degrees of rotor rotation to produce continuous generating torque as illustrated in Figure
2.7.
Figure 2.6: The idealised phase inductance, current and torque for generating operation
Phase Phase Phase Phase Phase Phase Phase Phase Phase Phase
D A B C D A B C D A
Inductance
0o 60o 120o Rotor Angle
Current Phase Phase Phase Phase Phase Phase Phase Phase Phase Phase
C D A B C D A B C D
0o 60o 120o Rotor Angle
0o 60o 120o
Rotor Angle
Phase Phase Phase Phase Phase Phase Phase Phase Phase Phase
Torque
C D A B C D A B C D
Figure 2.7: The idealised phase inductance, current and torque waveforms for a four-
phase 8/6 SRM under generating conditions
18
Like other motors, the machine torque is restricted by maximum allowed current, and
the speed by the available bus voltage. The current must be maintained within a
maximum allowable limit until the rotor reaches the base speed. At this speed the back-
emf of the machine is equal to the applied voltage to the phase winding and the current
will not increase further. Above the base speed the back-emf continues to increase so
the conduction angles can be enlarged to maintain constant power output. At very high
speeds, when the conduction angle cannot be increased any further, the torque decays
rapidly. The SRM can still operate at very high speed but only under light loads. The
torque/speed characteristic for motoring in a particular direction is illustrated in Figure
2.8. The characteristic for generating in the same direction of rotation would be
achieved by reflecting the waveform in the speed axis, while that for motoring and
generating in the opposite direction is achieved by reflection in the torque axis.
Figure 2.8: Torque/speed characteristic of the SRM
The main benefits of the switched reluctance machine include:

• Simple construction in that only the stator poles have windings and the rotor has
no windings or magnets but is constructed of steel laminates stacked on a shaft.
• Simplicity of the rotor construction which means that it is suitable for extremely
high-speed operation as it has a low moment of inertia.
• Ease in cooling as the stators are located nearest the machines outer casing and
only the stators have windings which are the main source of heat generation.
• Extremely robust machine primarily due to is simple construction and
additionally due to the electrical independence of the phases, which enables the
SRM to remain operational even with a failure of one of the phases.
• Lower maintenance requirements than a dc machine as the SRM is a brushless
machine
19
• Good starting torque like a dc series motor as the torque is proportional to the
square of the current (explained in Section 2.3).
• Power density equivalent to, or better than, induction machines but less than
permanent magnet synchronous, and brushless dc, machines for speeds up to
20,000rpm, while above this speed, power density is comparable to, or greater
than, these other machines (Krishnan 2001).
• A number of sensorless position estimation schemes are possible using the
principle that the phase inductance variation of the SRM is uniquely dependent
on rotor position and the phase current. This allows the elimination of the
position sensor to further improve reliability and reduce cost.
• Independence of torque with respect to current direction which means that the
converter only needs to facilitate unidirectional current. Converter topologies
can use less than two switches per phase, leading to a reduction in cost.
The main limitations of the switched reluctance machine include

• Complex control algorithms are required to eliminate the torque ripple produced
due to the strong non-linear magnetic characteristics of the SRM.
• Radial magnetic forces that act on the stator structure during rotor shaft rotation
produce acoustic noise levels higher than other machines. The noise can be
reduced by using complex torque control strategies and through careful design of
the SRM structure.
• Need for electronic commutation for its operation and fact that it cannot work
directly from a dc or ac line. Additionally, a separate path must be provided for
current in the phase winding to flow at the end of the commutation cycle. This
requires additional components that contribute to the overall cost of the
converter.
• Precise rotor position information is required in order to synchronise the
excitation of the phase windings with rotor position in order to implement
effective control strategies. The method of obtaining the position feedback
information has a significant bearing on the complexity and the reliability of the
SRM drive. If a position sensor is used it can have a significant effect on the
reliability and robustness of the drive while sensorless position schemes
contribute significantly to the complexity of the drive.
20
2.3 Mathematical Modeling of the SRM
Because each phase is identical and energised in sequence, the analysis is confined
initially to a single phase. Firstly it is assumed that no saturation occurs and a linear
approach is taken. Later, saturation is taken into account using non-linear analysis. The
following discussion on linear and non-linear operations is from the analysis by (Miller
2001). The linear and nonlinear analysis discussed here for motoring can be applied to
generating.
2.3.1 Analysis of magnetically linear operation

Linear analysis is centred on the voltage and torque equations. The terminal voltage
generated across one phase is related to the change in magnetic flux in the winding
according to Faraday’s law. The following simplifying assumptions are made:
• hysteresis and eddy current losses are neglected in the analysis of the magnetic
circuit
• there is negligible mutual coupling between phases
• the winding resistance is constant and independent of the current waveform and
frequency
The terminal voltage generated across one phase is:
v ph = i ph R ph + v Lph (2.1)
Where:
vph = the applied phase voltage
iph = the phase current
Rph = the phase winding resistance
vLph = the induced voltage on the phase inductance
Subscript ‘ph’ is left out for clarity, from here on.
The induced voltage vL is defined by the magnetic flux linkage, ψ, that varies as a
function of the phase current, i, and the rotor position, θ.
21
dψ (i,θ ) ∂ψ (i,θ ) di ∂ψ (i,θ ) dθ
vL = = + (2.2)
dt ∂i dt ∂θ dt
Therefore
∂ψ (i,θ ) di ∂ψ (i,θ ) dθ
v = iR + + (2.3)
∂i dt ∂θ dt
Where, dθ/dt is the angular velocity, ω, in rad/s, and ∂ψ(i,θ)/∂i represents the
instantaneous inductance L(i,θ). For simplicity, it is assumed that there is no magnetic
saturation, which means that the phase inductance is unaffected by variations in phase
current, then L(θ) = ψ(θ)/i. Equation (2.3) can be further expanded using linear analysis
to yield:
di dL(θ )
v = iR + L(θ ) + iω (2.4)
dt dθ
Thus, the applied phase voltage is equal to the sum of the voltage drops on the right
hand side of Equation (2.4), the resistive voltage drop, the inductive voltage, and the
induced emf or ‘back-emf’. The back-emf, e, is equivalent to:
dL(θ )
e = iω (2.5)
dθ
The instantaneous electrical power, vi, is
di dL(θ )
vi = i 2 R + L(θ )i + i 2ω (2.6)
dt dθ
The time derivative of magnetic stored energy at any instant is given by
d 1 2 1 2 dL(θ ) di 1 dL(θ ) di
 L(θ )i  = i + L(θ )i = i2ω + L(θ )i (2.7)
dt  2  2 dt dt 2 dθ dt
22
Assuming no losses other than i2R and in order to agree with the law of conservation of
energy, the mechanical power must equal to the electrical input power less the resistive
power losses (i2R) and the rate of change in the magnetic stored energy. Therefore, the
instantaneous mechanical power, which is equivalent to ωT, where T is the
instantaneous electromagnetic torque, can be written as:
1 2 dL(θ )
ωT = iω (2.8)
2 dθ
Therefore the expression for electromagnetic torque (T) is:
1 2 dL(θ )
T = i (2.9)
2 dθ
Equation (2.9) indicates that the torque is independent of the sign of the phase current
and is instead determined by the sign of the dL/dθ, where the absolute value of dL/dθ
contributes to the amount of torque produced. For this reason, SRMs are generally
designed to have a large ratio between the aligned and unaligned inductance in order to
attain high torque levels. Large air gaps between the stator and rotor pole at the
unaligned position increases the inductance ratio (Miller 1993). Typically, an SRM with
a low phase number and therefore a low rotor and stator pole count has larger air gaps.
Figure 2.9: Equivalent circuit for a single phase of the SRM
From the perspective of the phase winding terminals of the SRM, equation (2.4) appears
to indicate the equivalent circuit shown in Figure 2.9 comprising of a resistance (R), an
inductance (L) and a back-emf (e) that is proportional to speed. However, with L and e
varying with both rotor position and current, the torque cannot be calculated from a
23
simple equivalent circuit but requires direct solution of equations (2.1) and (2.9). A
method to directly solve these equations requires non-linear analysis of the SRM.
2.3.2 Analysis of magnetically nonlinear region

The practical SRM operates most effectively and efficiently in the magnetically
saturated region. Operating the machine in the magnetically linear region leads to
inefficient operation of the drive. Nonlinear analysis takes into account the saturation of
the magnetic circuit in the SRM and this allows a full understanding of the
electromechanical energy conversion for both motoring and generating operation. One
type of analysis is based on the magnetization curve shown in Figure 2.10; this is a
sample of the flux linkage (ψ) versus phase current (i) characteristic at a particular rotor
position.
Figure 2.10: Magnetization curve indicating magnetic field stored energy (Wf) and the
magnetic field co-energy (Wc)
The energy stored in a magnetic field is
ψ
W f = ∫ i (ψ ,θ ) dψ (2.10)
0
It is more desirable to express torque in terms of current instead of magnetic flux and
this is achieved using the term of co-energy (Wc) instead of energy. Magnetic field co-
energy is defined as
i
Wc = ∫ψ (i,θ )di (2.11)
0
24
The relationship between field energy and co-energy is illustrated in Figure 2.10 and
defined as
W f + Wc = i ψ (2.12)
Figure 2.11: Magnetization curve showing the effect of the rotor moving from position
A to position B
A rotor movement from its original position at A to a new position at B through an

angular displacement ∆θ yields the magnetization curves for both positions shown in
Figure 2.11. Assuming that the current remains constant during this rotor rotation, the
electrical energy (We) exchanged with the supply is
dψ
∆We = ∫ ivdt = ∫ i dt = ∫ idψ = ABCD (2.13)
dt
The change in stored field energy is
∆Wf = OBC – OAD (2.14)
25
Therefore, the mechanical work is the change in electrical energy minus the change in
magnetic field energy is expressed as:
∆Wm = ∆We – ∆Wf (2.15)

∆Wm = ABCD – (OBC-OAD)
∆Wm = OABCD – OBC
∆Wm = OAB (shaded in Figure 2.11)
(2.16)
This value corresponds to the increase in the magnetic field co-energy, ∆Wm = ∆Wc. The
mechanical work done during the displacement ∆θ, from A to B, can be expressed as:
∆Wm = T∆θ = ∆Wc (2.17)
Therefore, in the limit, when ∆θ → 0, the instantaneous torque (T) is the partial
derivative in co-energy with respect to angular displacement.
 ∂W 
T = C (2.18)
 ∂θ  i = cons tan t
In a motor with no saturation all the magnetization curves would be straight lines and
the stored magnetic field energy would equal the magnetic field co-energy at all times to
yield:
1
W f = Wc = L(θ )i 2 (2.19)
2
This expression for Wc implies that equation (2.18) for electromagnetic torque reduces
to:
1 2 dL(θ )
T = i (2.20)
2 dθ
This is identical to equation (2.9) from linear analysis.
26
For a multi-phase SRM, the instantaneous torque equation becomes the summation of
the individual torque values for each phase and is of the form:
k
T = ∑T
n =1
n (2.21)
where Tn is the torque produced by the nth phase and k is the total number of phases.
2.3.3 Analysis of generator operation

In the description of the generating operation provided earlier, the firing angles are
selected so that the current flows in the phase winding immediately after the rotor and
stator poles have passed alignment when the phase inductance is decreasing. The rotor
produces negative or braking torque opposing the direction of rotation. A negative
voltage is then applied to the phase winding and the mechanical energy expended by a
prime mover in pulling the rotor pole away from the excited stator pole is converted to
electrical energy to yield a generating current.
The circuit diagram for the SRM to operate as a generator is shown in Figure 2.12. An
example of the generator current and the idealised inductance is shown in Figure 2.13
io iL
VS
iin
Q1 D2
C Vs Load
Phase A iout
iph
D1 Q2
Figure 2.12: Switched reluctance generator for single phase
27
Figure 2.13: Switched reluctance generator phase currents and idealised inductance
The instantaneous current flowing in the phase winding is defined as iph. As illustrated
in Figure 2.12 and 2.13 at the angle θon, the voltage (VS) is applied to the phase winding
and current (iph = iin) begins to flow. Negative voltage (-VS) is applied at θoff and current
(iph = iout) continues to flow due to the generating action until the end of the rotor and
stator pole overlap at which point the current in the phase winding decays to zero at θext.
The current flow in the phase winding between θon and θoff is called the excitation
current (iin), while the current flowing between θoff and θext is defined as the generated
current (iout).
The average of the excitation current, Iin, over a single stroke is determined as follows:
θ off
1
I in = ∫θ i dθ (2.22)
θ off − θ on
ph
on
The average of the generated current, Iout, over a single stroke is determined as follows:
θ ext
1
I out = ∫θ i dθ (2.23)
θ ext − θ off
ph
off
The net generated average current, Io, over a single stroke is expressed as:
Io = Iout – Iin (2.24)
28
The power conversion in the SRM when operating as a generator can be characterized
by the excitation penalty, ξ, (Miller 2001) and is expressed as:
Pext I I in
ξ = = in = (2.25)
Pout I out I o + I in
where Pext is the mean electrical excitation power and Pout is the mean electrical output
power. An excitation penalty that is as small as possible is desired in order to minimise
losses.
2.4 Dynamic operation of SRM

The continuous operation of the SRM requires the switching of current pulses from
phase to phase during shaft rotation. For motoring operation, this can be defined for low
speed and high speed ranges. The generating mode of operation is analysed at low
speed, base speed and high speed. The classic or asymmetric half-bridge converter
shown in Figure 2.14 is the converter topology used to assist in the description of all the
dynamic modes of operation of the SRM.
Figure 2.14: (a) Static view of the converter. (b) Magnetization mode. (c) Freewheeling
mode. (d) Forced demagnetization mode
29
2.4.1 Low speed motoring
The current pulses must be regulated at low speed as back-emf is insufficient to limit the
current. There are two main methods of current regulation at low speed: voltage-PWM
and current hysteresis. The regulation of the phase current is implemented by chopping,
which is the switching on and off the power switches in Figure 2.14 at a frequency that
is much higher than the fundamental frequency of the current waveform (Miller 1993).
There are two types of chopping schemes: soft chopping and hard chopping.
The soft-switching strategy has three modes of operation as illustrated in Figure 2.14b, c
and d. The first mode is called the energization mode or magnetization mode or
positive-voltage loop (PVL) as indicated in Figure 2.14b. During this time both switches
(Q1 and Q2) are on and the current rises rapidly in the phase winding as indicated in
Figure 2.15a (I). During the second mode, the freewheeling state or zero-voltage loop,
only one switch and one diode (Q2 and D1) are on as illustrated in Figure 2.14c. There is
zero volts applied across the phase winding and the current continues to flow through
the switch and the diode, although it is gradually decaying as indicated in Figure 2.15a
(II). No energy is transferred to or from the supply during this period. The third mode,
the demagnetization state or the negative-voltage loop, occurs when both switches are
off and the energy returned to the supply via the freewheeling diodes (D1 and D2) as
shown in Figure 2.14c. Both diodes are forward biased and a negative voltage is applied
across the phase winding, forcing the current to rapidly decay to zero as indicated in
Figure 2.15a (III).
The hard-switching strategy has only two modes of operation, the magnetization and
demagnetization states. The magnetization and demagnetizations waveform for the
hard-switching scheme is illustrated in Figure 2.15b (IV) and Figure 2.15b (V)
respectively. During demagnetization, the decay in the current waveform is greater
leading to greater current ripple but reduced dynamic response time.
During the zero-volt period of the soft chopping operation the change in the phase
current is very small, which allows for a reduced switching frequency and current ripple
rating for the converter filter capacitor (Miller 2001). Additionally the soft chopping
scheme produces lower acoustic noise and lower EMI levels (Miller 1993).
30
L L
Lmax Lmax
Lmin Lmin
θON θ1 θ2 θ3 θ4 Rotor θON θ1 θ2 θ3 θ4 Rotor

Angle Angle
Vg Vg
Q1 Q1
Vg Vg
Q2 Q2
+VS +VS
Vph Vph
-VS -VS
iph iph
I IV
II III V
Rotor Rotor
Angle Angle
Torque Torque
θON θ1 θ2 θ3 θ4 Rotor θON θ1 θ2 θ3 θ4 Rotor

Angle Angle
(a) (b)
Figure 2.15: Low speed motor of operation for both (a) soft switching – (I)
energization, (II) freewheeling and (III) demagnetization and (b) hard
switching - (IV) Energization and (V) Demagnetization
The energy conversion process starts when the power switches are turned on and current
begins to flow in the phase winding at θon, the current increases rapidly to the required
value before rotor and stator pole overlap at θ1. Torque production begins as the rotor
and stator poles approach overlap and it then rises rapidly to the required value. By
maintaining a constant current value between the beginning (θ1) and end of overlap (θ2),
31
a constant torque value can be maintained over this period. The power switches are
turned off at θ2 and the phase voltage has a negative polarity, which leads to the current
freewheeling through the diodes. The current falls rapidly to zero and must be at zero
before θ3, in order to prevent the production of a negative torque. The process is
illustrated in what is called the energy conversion loop illustrated in Figure 2.16 for low
speed operation. The maximum energy conversion area, W, is bounded by the aligned
and unaligned magnetization curves and is traversed in a counter clockwise direction
from the unaligned to the aligned position. A single cycle of the loop is achieved during
each stroke.
Figure 2.16: Low speed motoring energy conversion loop
As noted previously, there are two main methods of current regulation at low speed:
voltage-PWM and current hysteresis. Voltage-PWM implementation requires that one or
both of the power switches are turned on and off at a fixed frequency, while the duty
cycle varies depending on the current requirements. This method can be implemented
with either hard or soft chopping switching schemes. Under soft chopping operation,
one of the power switches are turned on and off and the other power switch remains on
during the conduction period, while with hard chopping both of the power switches are
turned on and off together.
Current hysteresis necessitates that one of the power switches be switched off when the
current exceeds a particular set point value and switches back on again the current fall
below a second level. The range between the current limits is referred to as the
hysteresis-band. The switching frequency varies, which can lead to unwanted audible
noise and current harmonics. Both soft and hard chopping switching are possible with
this type of current regulation. Delta modulation is a variation of current hysteresis in
32
which the current is sampled at a fixed frequency. When the phase current exceeds a
reference current value one of the power switches is turned off and when the current
drops below the reference value, the power switch is turned on again. The switching
frequency can vary up to the fixed sampling frequency value. Soft and hard chopping
schemes are also possible with this type of current regulation.
2.4.2 High speed motoring

At high speed, no current regulation is required as the back-emf is large enough to limit
the current. The torque is controlled only by varying the firing angles of a single pulse
of current to yield what is called the single pulse mode of operation. The single pulse
mode waveforms for the idealised inductance profile, the phase voltage, the single pulse
current and the resultant torque are provided in Figure 2.17
.
Figure 2.17: Single pulse control for single phase motoring at high speed
33
The energy conversion process starts when both power switches are turned on after
passing the unaligned position of the previous stroke at θ4 and current rises rapidly in
the phase winding due to the low value of inductance and no back-emf. At θ1 the
inductance begins to increase and the back-emf quickly exceeds the supply voltage and
forces the di/dt to become negative. The greater the speed, the faster the drop off in
current in this region and only an increase in the supply voltage can counteract this
effect. The dwell angle must be increased linearly with speed above the base speed in
order to maintain a reasonable torque output although it will be decaying in value. The
energy conversion loop for high speed operation is illustrated in Figure 2.18 and is
traversed in the counter clockwise direction.
Figure 2.18: High speed motoring energy conversion loop
2.4.3 Low speed generating

Generating at low speed is similar to motoring at low speed except the firing angles are
retarded in order for the current pulse to coincide with the region of decreasing
inductance. Current regulation is required during the low speed generating cycle, as the
magnitude of the back-emf is smaller than the combined magnitude of the applied
voltage and the resistive voltage drop. Hard chopping is the preferred switching strategy
as energy is returned every switching cycle whereas no energy is returned during the
zero-voltage cycle of the soft switching strategy. Voltage-PWM control can be used to
regulate the current when generating at low speeds by varying the average applied
voltage in the SRM (Miller 2001). The excitation and generation waveform at low speed
34
is illustrated in Figure 2.19. It is assumed that the inductance is linear and that the
current reaches its set point before θ3 and the conduction cycle ends at θ4.
Figure 2.19: Voltage-PWM current regulation at low speed
In one PWM period the excitation current corresponds to the area A1 and the generated
current corresponds to the area A2 in Figure 2.19. Assuming that the duty cycle d is
constant over a conduction period and the current ripple due to PWM is negligible
compared to its average peak value ip, then
A1 = D1tpwmip
A2 = D2tpwmip (2.26)
Where D1 = d and D2 = 1 – d and tpwm is the PWM period (tpwm = 1/fpwm). The net
generated current in one PWM period (Io,pwm) is:
Io,pwm = A2 – A1
= (D2 – D1)tpwmip (2.27)
In the analysis, the initial excitation current in the area indicated (1) in Figure 2.19 and
the generated current after the turn-off angle at θ4 in the area indicated (2) are neglected.
35
The ratio between excitation power and the generated power yields the excitation
penalty, ξ, (Miller 2001) for the current regulated with PWM is defined as:
D1 d
ξ= = (2.28)
D2 − D1 1 − 2d
During the excitation period, the terminal voltage of the phase winding is expressed as
follows:
di
v ph = L(θ ) −e (2.29)
dt
where the voltage drop (iR) due to the resistance of the coil is neglected and the back-
emf, e, is equivalent to:
dL(θ )
e = iω (2.30)
dθ
where ω is the rotor speed, dL/dθ is always negative under generating operation and the
negative sign is placed alongside e for clarity. For effective PWM control at low speeds
the condition |e| < |Vs| must always be true, otherwise the generating current is
uncontrollable and increases during the demagnetization period. During the excitation
period the equation is:
diexc
L(θ ) = VS + e > 0 (2.31)
dt
During the generating period, the equation is:
di gen
L(θ ) = − VS + e < 0 (2.32)
dt
Assuming that the change in L over the PWM period is small and can be neglected, in
order to keep the average current constant over a PWM period, then
36
diexc di gen
D1t pwm = D2 t pwm (2.33)
dt dt
VS + e D1 = V S − e D2
Rearranging the above equation and substituting D1 = d and D2 = 1 – d yields the

expression:
1 e 
d =  1 -  (2.34)
2  VS 
From equation (2.28), in order to generate net energy which is greater than the
excitation power, the condition of d < 0.333 must be satisfied. Substituting this
condition into equation (2.34) gives the condition required in order to generate net
power greater than the power consumed during the excitation of the phase winding with
voltage-PWM is:
0.333VDC < e <VDC (2.35)

and
0.333ωb < e <ωb (2.36)
where ωb is the base speed. It therefore can be concluded for net power generation at
low speed using voltage-PWM is possible between one third of the base speed and the
base speed. The energy conversion loop for low speed generating operation is illustrated
in Figure 2.20 and is traversed in the clockwise direction.
Figure 2.20: Low speed generating energy conversion loop
37
2.4.4 High speed generating
Generating at high speed can be defined for two regions: (a) at the base speed and (b)
above the base speed. The power switches turn on at θon and excitation power is applied
to the phase winding and magnetic energy is built up in the air gap and at θoff the power
switches are turned off and regenerated current flows through the freewheeling path
until the current goes to zero at θext.
Figure 2.21: Single pulse control for single phase generating at high speed
At the base speed, the magnitude of the back-emf and the combined magnitude of the
applied voltage plus the resistive voltage drop in the phase winding are in balance and
this yields current that remains constant without the need for current regulation (Torrey
2002) until the overlap of the poles at θ4 and then decays to zero by θext as indicated in
Figure 2.21 (a). Above the base speed, the back-emf in the phase winding has a larger
magnitude than the combined magnitude of the applied voltage plus the resistive voltage
38
drop of the phase winding, and the current increases after the power switches turn off at
θoff and the current increases until pole overlap at θ4 where it decays to zero at θext as
indicated in Figure 2.21(b).
On method of implementing generator operation at high speed is to regulate the applied

voltage with speed in order to use the condition |e| > |Vs - iR|, as indicated in Figure
2.21(b), to allow the generating current to reach a set point value, and then implement
and regulate the condition |e| = |Vs - iR|, as indicated in Figure 2.21(a), to maintain
constant generator current (Miller 2001).
The energy conversion loop for high speed generating operation is illustrated in Figure
2.22 and is traversed in the clockwise direction.
ψ
Aligned
Unaligned
Current (A)
Figure 2.22: High speed generating energy conversion loop
2.5 Summary
The SRM has gone through significant developments since its beginnings in the early
1800s to the present day. The evolution in high speed computers and magnetic analysis
software has allowed significant developments in the design of the SRM. Since the
1980s, significant developments in semiconductor technologies have led to greater
power handling capabilities for power switching devices, particularly the MOSFET and
the IGBT. These devices have allowed for converter operation with forced commutating
39
schemes and for current regulation schemes to operate at much higher frequencies. The
development of powerful microprocessors and in particular low cost digital signal
processors (DSP) in recent years has allowed for the implementation of complex control
algorithms for the SRM. This has increased the application potential of the SRM in
many commercial arenas from starter/generator for articulated vehicle and mild hybrid-
electric powertrain to washing machines and hand dryers.
For the implementation of the SRM in an automotive application as both a motor and a
generator, it is desirable to have high torque capability to drive loads (starting an engine
or actual propulsion assistance) or to operate as a generator, particularly under
regenerative conditions where high braking torque may be beneficial. A two-phase
SRM would be the most suitable for these applications if operating in only one
direction. This SRM configuration would minimise the complexity of the power
converter. The larger areas around the stator poles of a two-phase SRM allow for phase
windings with copper of larger cross section. This enables lower copper losses at higher
current operation. The larger iron sections of the rotor and stator poles keep core losses
low and maintain good mechanical stiffness which is important in minimising acoustic
noise (Miller 1993). The core losses are also reduced by the lower commutation
frequency of the two-phase SRM. The larger air gap, higher current handling capability
and reduced core losses all contribute to the increased torque capability of the two-phase
SRM as discussed in Section 2.3.
From the review and the analysis of the SRM, the converter in this application should
operate under voltage-PWM for both low speed motoring and generating conditions and
single-pulse operation under high speed conditions. Variation in the source voltage
would be desirable under high speed generating to maintain the condition |e| = |Vs - iR|,
in order to maintain constant generator current.
40
Chapter 3 Converter topologies
Significant research effort has been invested in the design of electronic power
converters for the SRM. A review of a cross section of the numerous converter
topologies that are available is presented in this chapter before finalizing the
configuration that is most suitable for this application. The general requirements are
defined for the converter in order to determine the most suitable choice. The converters
are grouped into various classifications, which are based on the predominant feature of
the converter. Each converter from a topology group is described under the heading of
its predominant features, the modes of operation and the main benefits and limitations.
On review of all the groups, the most suitable converter configuration is selected for
preliminary analysis.
3.1 Selection criteria and topology groups

The fundamental selection criteria to determine the most suitable converter for this
application include the following:
• Independent phase operation with respect to the operating modes for the
energization or magnetization, freewheeling and demagnetisation for both
motoring and generating modes of operation.
• Soft chopping switching capability or the ability for the application to implement
freewheeling operation.
• Unidirectional current and bidirectional voltage for all phases.
• Maximum efficiency capability in both motoring and generation modes of
operation.
• Minimum complexity in order to minimise converter costs
The converter topologies are divided into two primary classifications: hard- switching
and soft-switching. The fundamental converter configurations are illustrated in Figure
3.1. The hard-switching classification contains five subgroups: bridge, capacitive
energy recovery, magnetic, dissipative and single phase. The soft-switching
classification contains a single subgroup: self-commutating There are four bridge
converters topologies: the asymmetrical half-bridge, the shared winding converter, the
shared switch converter and the full-bridge converter. The capacitive energy recovery
group has five fundamental configurations but there are variations within each
41
configuration. The configurations are: C-dump converter, voltage boosting converter,
the split dc converter and the Sood converter. The magnetic energy recovery group have
two basic converter schemes: the bifilar converter and the auxiliary commutation
winding converter. The dissipative group has a single elementary configuration called
the R-dump but there are variations defined as the decay converters. The self-
commutating group has four essential configurations but there are some derivatives.
They are: series resonant, zero-voltage transition (ZVT) PWM converter, auxiliary
quasi-resonant dc-link (AQRDCL) converter and the H-bridge converter. The single
phase converter topology is included here for completeness but is not discussed here as
the application requires a minimum of a two phase SRM to implement unidirectional
starting capability.
Converter Configurations
For SRM
Hard-Switched Soft-Switched
Converters Converters
Bridge Capacitive Magnetic Disipative Single-Phase Self-Commutating

Converters Converters Converters Converters Converters Converters
Single
Assymetric C Bifilar R Series
Controllable
Half-Bridge Dump Dump Resonant
Switch
Auxiliary
Shared phase Voltage ZVT PWM
Commutation
winding Boosting Pollock
Winding
Miller Split-dc AQRDCL

Shared-Switch
Full Sood H
Bridge Bridge
Figure 3.1: Classification of power converters for switched reluctance motor
42
3.2 Bridge Converters
Four fundamental bridge converter topologies are addressed in this section and these are
the asymmetrical half-bridge converter, the shared phase winding converter, the shared
switch converter and the full-bridge converter. A brief description is provided for each
converter topology that includes a discussion on the possible modes of operation
together with an illustration. The main benefits and limitations are outlined for each
topology.
3.2.1 Asymmetrical Half-Bridge Converter

This converter is also known as the classic converter and is a widely documented
(Lawrenson 1980), (Ray 1979), (Vukosavic 1990) converter topology using a single rail
power supply as shown in Figure 3.2a. It is defined as a 2N active switch configuration,
where N is the number of phases.
VS
Q1
D2
D1 Q2
(a)
VS VS VS
Q1 Q1 V(D2)
Q1 V(Q1) D2 D2
D2
Vph = VS Vph = 0 Vph = -VS
iph iph iph
D1 V(D1) D1 V(D1)
D1 V(Q2) Q2 V(Q2) Q2 Q2
(b) (c) (d)
Figure 3.2: The asymmetrical half-bridge or classic converter. (a) Static view of
converter. (b) Energization mode. (c) Freewheeling mode. (d) Forced demagnetization
mode
43
This converter has two possible switching strategies, soft-switching and hard-switching.
The soft-switching strategy has three modes of operation as illustrated in Figure 3.2 b, c
and d. The first mode is called the energization mode or magnetization mode as
indicated in Figure 3.2b. During this time both switches (Q1 and Q2) are on and the
current rises rapidly in the phase winding as indicated in Figure 3.3a (I). During the
second mode, the freewheeling state, only one switch and one diode are on as illustrated
in Figure 3.2c. There is zero volts applied across the phase winding and the current
continues to flow through one switch and one diode, although it is gradually decaying
as indicated in Figure 3.3a (II). No energy is transferred to or from the supply. The third
mode, demagnetization, occurs when both switches are off and the energy in the phase
winding is returned to the supply via the freewheeling diodes as shown in Figure 3.2c.
The voltage is reversed across the phase winding which forces the current to rapidly
decay to zero as indicated in Figure 3.3a (III).
Figure 3.3: Half-bridge converter modes of operation for both (a) soft switching – (I)
energization, (II) freewheeling and (III) demagnetization and (b) hard switching - (IV)
energization and (V) demagnetization
44
The hard-switching strategy has only two modes of operation, the magnetization state
and the demagnetization states. The magnetization and demagnetizations waveform for
the hard-switching scheme is illustrated in Figure 3.3b (IV) and Figure 3.3b (V)
respectively. During demagnetization, the decay in the current waveform is greater
leading to greater current ripple but reduced dynamic response time.
The most significant advantages of this circuit are (i) the possibility to independently
control the upper and lower switched for maximum flexibility, (ii) all possible firing
angles can be implemented with soft-switching allowing for maximum regenerative
braking capability and (iii) equal performance in forward and reverse directions. The
main limitation is the high power semiconductor count per phase, which can become
expensive with switched reluctance drives with a large number of phase windings. Other
disadvantages are the relatively low demagnetization voltage at high speeds due to the
fixed voltage supply and the requirement for a large capacitor on the supply voltage in
order to filter the voltage ripple associated with the magnetization and demagnetization
of the phase winding. Demagnetization during generator mode of operation could yield
significant power losses through the freewheeling diodes due to the large forward
voltage drop across the diode coupled with the large operating current.
3.2.2 Shared phase winding converter

This scheme allows for two phase winding to be shared in a modified asymmetrical
half-bridge converter (Krishnan, 2001) as shown in Figure 3.4a. This increases the
utilization of the power devices by increasing the component duty cycles. An SRM with
an even number of phases is required to take advantage of this scheme. A silicon-
controlled rectifier (SCR) is used to steer the current through the desired phase winding.
Only one power switch and one diode are required per phase winding compared to two
power switches and two diodes in the asymmetrical converter. Alternate phase windings
are grouped together to allow for independent control with overlapping currents that do
not exceed one phase cycle duration.
This converter can also be operated with soft-switching and hard-switching strategies.
Soft-switching again has three modes of operation: magnetization, freewheeling and
demagnetization, while hard-switching has only magnetization and demagnetization as
discussed previously with the asymmetrical converter. The main difference is that the
45
desired phase winding becomes active when the SCR is turned on (Phase 1 by SCR1 and
Phase 2 by SCR2 as illustrated in Figure 3.4). The operation of Phase 1 is illustrated
graphically in Figure 3.4 and Phase 2 operates in the same manner when SCR2 becomes
active.
Figure 3.4: The shared-phase winding converter. (a) Static view of converter. (b)
Energization mode. (c) Freewheeling mode. (d) Forced demagnetization mode
The main advantages of this configuration include:

• Only one power switch, one SCR and one diode are required per phase winding.
• Greater utilization of power switches and diodes in the modified asymmetrical
converter as the power devices are now used for two phase windings instead of
one thereby reducing cost in the SRM drive.
• The converter is capable of positive, negative and zero voltage output, which
allows for greater current control flexibility.
• Independent control of phase current is possible with this configuration
The main disadvantages of this configuration are:
• The SCR is always in the conduction path, which can lead to significant losses
during high power applications. This would require a larger heat sink for cooling
thereby reducing the efficiency of the system and increasing system costs.
46
• SCR gate drive circuits would increase the component count and therefore drive
costs, while also increasing complexity with the requirement for gate drive
circuit isolation.
• Current overlap is not possible between phase winding within the same half-
bridge.
3.2.3 Shared switch converter

This converter is also known as the Miller converter when implemented with a three
phase SRM (Pollock 1990) and (Vukosavic 1990). A two phase implementation is
illustrated in Figure 3.5a. The switches and diodes are shared between more than one
phase-winding that yields an N+1 active switch configuration. This is particularly useful
when the SRM has two or more phase windings as it reduces the number of power
device.
Figure 3.5: The shared switch converter. (a) Static view of converter. (b) Energization
mode. (c) Freewheeling mode. (d) Forced demagnetization mode
This converter can also implement soft-switching and hard-switching strategies in the
same fashion as with the asymmetrical half-bridge converter. The operation of phase 1
is illustrated in Figure 3.5, while phase 2 operated in the same way except the Q3 and D3
are now active devices instead of Q1 and D1.
The benefits of this converter are greater utilization of the power switches due to the
shared switch operation and the topology is applicable to two or more phases and has
fewer constraints on phase current overlap when the phase number is greater than four.
47
The main limitations of this topology are that there are restrictions on the current control
options during current overlap conditions. For phase count greater than or equal to four,
two or more phases can be operated independently but not all the phases can be operated
independently.
3.2.4 Full-bridge synchronous rectifier converter

This configuration offers the potential to reduce conduction and switching losses during
freewheeling and demagnetization mode of operation by replacing the diodes in the
half-bridge circuit with MOSFETs power switches as indicated in Figure 3.6a.
Figure 3.6: The full-bridge converter. (a) Static view of converter. (b) Energization
It also provides utmost flexibility in motoring and generating modes of operation

(Krishnamurthy 2005). This converter also has the capability to operate in either soft or
hard switching strategy. The soft-switching incorporates the two modes of operation of
hard-switching, so only soft-switching is discussed here. The magnetization mode
begins when both Q1 and Q2 turn on and both Q3 and Q4 are off as illustrated in
Figure3.6b. Throughout freewheeling, Q3 and Q2 turn on and both Q1 and Q4 are off as
48
indicated in Figure 3.6c. During demagnetization, Q3 and Q4 turn on and both Q1 and Q2
are off as shown in Figure 3.6d. During magnetization and freewheeling modes, Q3
operates as a synchronous rectifier, (Mohan 1995c) while both Q3 and Q4 operate as
synchronous rectifiers during demagnetization cycle of operation. The idealized
waveforms of control signals for all the switches in magnetization, freewheeling and
demagnetization modes of operation and phase winding voltage and current waveforms
are illustrated in Figure 3.7a for soft-switching and Figure 3.7b for hard-switching
strategies.
Figure 3.7: Full-bridge converter modes of operation for both (a) soft switching – (I)
energization, (II) freewheeling and (III) demagnetization and (b) hard switching - (IV)
energization and (V) demagnetization
A delay period, called blanking time, is required between the switching on and off of Q1
and Q2 pair and Q3 and Q4 pair in order to prevent a short circuit of the power supply.
During the blanking period the diode in parallel with the switch conducts until the
power switch turns on.
49
The main advantages are the same as with the half-bridge converter but with improved
operational efficiency during freewheeling and demagnetization modes of operation
using synchronous rectification scheme, which is especially beneficial during generator
mode of operation. The mail limitation is again the same as with the half-bridge with
under utilization of the power switches being further exacerbated in the full-bridge with
the additional power switches. The additional cost of the extra gate drive circuits for the
additional power switches must be taken into account.
3.3 Capacitive energy recovery converters

Four capacitive energy recovery converters are addressed in this section and they are the
C-dump converter, the split dc converter, the Sood converter and the voltage boosting
(buck-boost) converter. A brief description is provided for each converter topology that
includes a discussion on the possible modes of operation together with illustration and
the main benefits and limitations of the circuits. Voltage boosting can be achieved by a
number of different schemes, some of which are highlighted briefly. A detailed
discussion is provided on the buck-boost converter.
3.3.1 C-dump converter

The defining characteristic of the C-dump converter is the sharing of the freewheeling
circuit with the SRM phase (Bass 1985) and (Miller 1993) as indicated in Figure 3.8a.
During freewheeling the charge builds up in the dump capacitor, Cd, and is typically
controlled by a step down circuit, consisting of Qr, Dr and Lr, to maintain an average
value of 2Vs across the capacitor. This defines the converter as a dual-rail circuit. The
step down circuit operates like a large snubber with energy recovery to the supply.
50
Figure 3.8: The parallel C-dump converter. (a) Static view of converter. (b)
Magnetization mode (c) and (d) Freewheeling with energy recovery mode. (e) and (f)
Demagnetization with energy recovery mode
There are three distinct mode of operation: magnetization, freewheeling with energy
recovery and demagnetization with energy recovery. During magnetization, the supply
voltage is applied to the phase winding when Q1 is turned on as indicated in Figure
3.8b. When the phase winding current exceeds a predetermined threshold, switch Q1
turns off and forward biasing diode D1 to allow the current to freewheel and charge the
capacitor, Cd as indicated in Figure 3.8c. This has the effect of reducing the phase
winding current below the reference value. During the next magnetization cycle of the
switch, Q1, a feedback control circuit varies the duty cycle of the energy recovery
switch, Qr, to regulate the voltage across the dump capacitor by transferring energy from
the capacitor to the dc source and the inductor, Lr, as indicated Figure 3.8d. This cycle
of energization, freewheeling and energy recovery is repeated during the conduction
period of the phase winding. At commutation, the demagnetization begins when the
switch, Q1, turns off and the energy in the machine is partially transferred to the dump
capacitor as indicated in Figure 3.8e and the voltage across the capacitor increases. The
voltage difference between Vo and Vs is impressed across the phase winding and the
current decays rapidly to zero. The voltage across the dump capacitor is regulated by
controlling the duty cycle of the energy recovery switch, Qr, as indicated in Figure 3.8f.
The waveforms for the C-dump converter during the complete cycle of operation are
shown in Figure 3.9.
51
Figure 3.9: Waveforms for the C-dump converter with energy recovery circuit
The main advantage of this converter is the ability to have independent phase control
with the minimum number of switches. The main disadvantage is that demagnetization
voltage is limited due to it being the difference in the voltage across the dump capacitor
and the supply voltage. Additional losses are incurred in the energy recovery circuit
resulting in a decrease in the efficiency of the motor drive. Failure of the energy
recovery circuit can lead to uncontrollable charging of the dump capacitor yielding
catastrophic results. The circuit shown in Figure 3.8 is a parallel C-dump converter,
while it is also possible to have a series C-dump converter as illustrated in Figure 3.10a
and this circuit allows for the dump capacitor to be rated at the demagnetization voltage
and not the demagnetization voltage plus the supply voltage as is the case with the
parallel C-dump. A further modification of the C-dump circuit is provided in Figure
3.10b (Mir 1997). This revised circuit reduces the VA rating of the power switches and
the voltage rating of the dump capacitor. A freewheeling mode of operation is possible
with this revised configuration.
52
Figure 3.10: Variations of the C-dump converter circuit, (a) the series C-dump
converter, (b) the modified C-dump converter
3.3.2 Buck-boost converter

For a buck-boost converter, Qa, Da and La and the output capacitor, Cb form a buck-
boost front end stage (Barnes 1998), (Krishnan, 2001) and (Vukosavic 1990) as
illustrated in Figure 3.11a. This dual-rail voltage boosting circuit offers added flexibility
in the SRM control capabilities by separating the magnetization and demagnetizing
voltage and providing a variable supply voltage. The converters operation can be
divided into four possible states.
• Magnetization commences when the switch, Q1, turns on and impresses the
voltage, VCb, across the phase winding as indicated in Figure 3.11b.
• The second state is forced demagnetization: it occurs when Q1 is off and the
diode, D1, conducts transferring the energy from the phase winding to the
capacitor, CS as indicated in Figure 3.11c.
• The third stage is the charging of capacitor, Cb, with Qa turned on. When the
switch, Qa, turns on, and assuming the voltage drop across Qa is negligible, then
the voltage across the inductor, La, is equal to the source voltage as indicated in
Figure 3.11d.
• The fourth and final stage is the charging of capacitor, Cb, with Qa turned off.
During this stage, Qa turns off and the diode, Da, conducts transferring the
energy stored in La to Cb as indicated in Figure 3.11e.
53
VS VS
Qa Qa
Q1 Q1
+ Da D1 + Da D1
VCs CS VCs CS
_ - _ -
La La -
VCb Cb Phase VCb Cb
Vphase Phase
+ +
+
iphase
(a) (b)
VS VS VS
Qa Qa Qa
Q1 Q1 Q1
+ Da D1 + Da D1 + Da D1
VCs CS VCs CS VCs CS
_ - _ - _
+ - -
La VCb + La VCb
Cb
VLa Cb VLa La VCb Cb
+ Vphase Phase +
Phase Phase
- + +
-
iphase
(c) (d) (e)
Figure 3.11: The buck-boost converter. (a) Static view of converter. (b) Magnetization
(c) Demagnetization (d) Charging of Cb with Qa turned on (e) Charging of Cb with Qa
turned off
The most significant advantage of the buck front end allows for the input voltage to be
greater than the dc source voltage to accelerate current buildup in the phase winding,
which is especially useful in generator mode as rapid current build up is required in the
phase winding near rotor and stator pole alignment. Another advantage is the each phase
can be controlled independently of the other even though there is only one switch per
machine phase. The main disadvantages are extra circuit complexity and the higher
voltage rating requirement for the power switches, the voltage across Q1 is VCb, while
the voltage across Qa is VCs plus VCb.
There are a number of other voltage boosting circuits including: (a) parallel dc-link
voltage-boosting converter (top) and series dc-link voltage-boosting converter (bottom),
are shown in Figure 3.12a (Barnes 1998). The boost capacitor in the dc-link voltage-
boosting converters increases the turn on and turn-off voltage applied for part of the
motoring stroke. The auxiliary rail boost converter is shown in Figure 3.12b (Hava
1992), parallel (top) and series (bottom). In these circuit the auxiliary voltage boost
capacitor, Cb, is switched in and out of the energization cycle using the switch, Qb, and
diode, Db. The sequential boost converter is shown in Figure 3.12c (Le-Huy 1990).
This circuit is similar to the series auxiliary rail boost converter but has a boost
capacitor and blocking diode per phase winding. The variable dc-link converter is
illustrated in Figure 3.12d (Krishnan, 2001). Here the switch, Qc, the diode, Dc, the
inductor, Lc, and the capacitor, Cc, form a step-down switching power stage. This stage
54
varies the input dc source voltage to obtain the desirable input voltage to the machine
windings.
Figure 3.12: (a) Parallel dc-link voltage-boosting converter(top) and series dc-link
voltage-boosting converter (bottom), (b) Auxiliary rail boost converter, parallel (top)
and series (bottom) (c) Sequential boost converter (d) Variable dc-link converter
3.3.3 Split DC Converter

In this circuit, the single dc link capacitor is replaced with two capacitors connected in
series as illustrated in Figure 3.13a. This converter is also known as the Oulton
converter circuit (Miller 1993). This circuit has two possible modes of operation:
magnetization and demagnetization, which fundamentally is a hard-switching format.
The operation for a single phase is as follows:
• Magnetization commences when Q1 is closed. Current circulates through Q1,
phase 1, and the capacitor C1. The voltage across the capacitor is applied to the
winding as shown in Figure 3.13b.
• When the switch, Q1, is open at demagnetization, the current freewheels through
the diode, D1, and the voltage across the capacitor, C2, is applied to the phase
winding. The charging of the capacitor, C2, quickly depletes the phase winding
current as illustrated in Figure 3.13c.
55
Figure 3.13: The Split dc converter. (a) Static view of converter. (b) Magnetization
mode. (c) Demagnetization mode
The waveforms illustrating the magnetization and demagnetization modes of operation

are illustrated in Figure 3.14.
Figure 3.14: Waveforms for the split dc converter circuit
56
The main advantage is this circuit achieves the capability of having one switch per
phase without adding extraneous passive components and still achieves significant
control flexibility and efficiency. The main limitation is the requirement for balancing
the charge across the dc link capacitors, C1 and C2, which requires an even number of
machine phases. Additionally, the balancing of energy flow in the capacitors results in
some loss in the independent control of the phase windings. Another disadvantage to
this configuration is that only half of the supply voltage is applied to the phase winding
during magnetization and demagnetization, decreasing dynamic response of the current
control loop and reducing the maximum motoring speed capability.
3.3.4 Sood converter

This converter does without a dc-link capacitor and instead uses the boost capacitor as a
dump capacitor as indicated in Figure 3.15a. It is also considered to be a dual rail
topology and its configuration is similar to the C-dump and buck-boost circuits but the
Sood converter has primarily different modes of operation (Vukosavic 1990). This
converter has four possible operating states:
• The first state is magnetization from dc supply, Vs. The switch, Q1, turns on and
the phase winding is connected to the supply voltage as indicated in Figure
3.15b
• The magnetization from the capacitor, Cb, is the second state. Here both Q1 and
Qa are on, the capacitor voltage, Vc, is larger than the supply voltage, Vs and the
energy flows from the capacitor to the phase winding as shown in Figure 3.15c.
If the capacitor voltage is smaller than the supply voltage the magnetization
reverts to the first state.
• Demagnetization through freewheeling is defined as the third state. In this state,
Qa is on and D1 and Da are on providing a freewheeling path for the phase
winding current as indicated in Figure 3.15d. This allows the phase winding to
demagnetize slowly. S1 must withstand the boost voltage, Vc, during this period.
• Demagnetization by charging the boost capacitor, Cb, is defined as the fourth
and final state. During this period, both Q1 and Qa are off and the diode, D1,
turns on and the phase winding is demagnetized through D1, the boost capacitor,
Cb, and the supply voltage as shown in Figure 3.15e. The boost capacitor is
charged by the demagnetization energy for the phase winding and the supply
voltage.
57
The boost capacitor voltage must be kept within a narrow band. The minimum voltage
is determined by the demagnetization requirements and this value also determines all
switch and boost capacitor voltage ratings. Low speed operation requires PWM current
control and alternates between state one and state four. State three is only implemented
at very low speed when time is not critical. During state four, the switch, Qa, must
remain off until the phase winding current has decayed to zero. The main advantage of
this configuration is the low component count and that the four possible states allow
additional motor control freedom. The main limitation is the complex control required to
implement the four states. There are also additional power losses through the energy
recovery circuit.
.
Figure 3.15: The Sood converter. (a) Static view of converter. (b) Magnetization from
Vs. (c) Magnetization from boost capacitor (d) Demagnetization through freewheeling
(e) Demagnetization by charging the boost capacitor
3.4 Magnetic Energy Recovery Converters

These circuits exploit the fact that the energy stored in the magnetic field of the phase
winding can be transferred to a closely coupled second winding that can be either
returned to the voltage supply or used to energize another phase winding (Barnes 1998)
and (Miller 1993). Two converter configurations are examined in this section: the bifilar
converter and the auxiliary winding converter. A brief description is provided for each
converter topology that includes a discussion on the possible modes of operation
together with illustration and the main benefits and limitations of the circuits.
58
3.4.1 Bifilar Converter
This configuration uses one switch and one diode per phase and using a bifilar-wound
SRM with closely coupled primary and secondary windings to regenerate the stored
magnetic energy to the source [(Krishnan 2001) and (Ray 1979)]. The switch is in series
with the primary winding, while the diode is in series with the secondary winding as
illustrated in Figure 3.16a.
VS VS VS
+
Q1 VQ1 D1 Q1 D1 Q1 D1
-
(a) (b) (c)
Figure 3.16: The bifilar converter. (a) Static view of converter. (b) Magnetization state
(c) Demagnetization state
There are two operating states: magnetization and demagnetization.

• Magnetization commences with the switch, Q1, turns on allowing current, ipri, to
build up in the primary phase winding; during this period the diode is reversed
biased as illustrated in Figure 3.16b.
• At commutation the switch turns off and the polarity of the induced emf in the
secondary winding is such to forward bias the diode, D1. This allows the current,
isec, to circulate through the diode, the secondary winding and the source, which
leads to the demagnetization of the primary as illustrated in Figure 3.16c.
The waveforms illustrating the magnetization and demagnetization modes of operation

are illustrated in Figure 3.17.
59
Figure 3.17: Waveforms for the bifilar converter circuit
The main advantage is the component count is low due the use of the magnetic recovery
winding. The main drawback is that the bifilar winding is not economical for large
motors. The voltage across Q1 at turn off is VS + aVS, where ‘a’ is the turns ratio
between the primary and secondary winding, requiring the switch to have a higher
voltage rating that consequently results in greater power losses. Another limitation is
that the coupling is never perfect and snubber circuits are required to limit the transient
voltages. This adds cost and complexity to what is otherwise a simple configuration.
The bifilar winding also decreases the efficiency of the system due to the poor copper
utilization from the reduced winding area available for the main winding. Soft-switching
is not possible as there is no freewheeling condition.
3.4.2 Auxiliary Commutation Winding Converter

This converter also requires that the motor must have an auxiliary winding (Liang 1994)
as illustrated in Figure 3.18a. This circuit has four possible states: energization and
freewheeling of the main winding are the first two states. The next state is the
demagnetization of the main winding to the auxiliary winding and the final state is
60
demagnetization of the excess energy in the main winding to the source. The
magnetization and freewheeling of the main winding is implemented using a
conventional converter like an asymmetrical half bridge, which facilitates switching
between positive and zero voltage loops as illustrated in Figure 3.18b and Figure 3.18c
respectively. The demagnetization of the main winding to the auxiliary winding is
implemented through magnetic coupling as illustrated in Figure 3.18d. During the
negative voltage loop the diode, D2, is forward biased, which allows current to circulate
in the auxiliary winding. The demagnetization of the excess energy the main winding to
the source is illustrated in Figure 3.18d, while during this period the current freewheels
in the auxiliary winding. During the magnetization of the next phase winding, the
positive voltage loop allows the magnetic energy of the auxiliary winding to couple to
the main winding to allow the current to build up rapidly.
Figure 3.18: The auxiliary winding converter. (a) Static view of converter. (b)
Magnetization state (c) Freewheeling state (d) Demagnetization to auxiliary winding
and to dc-link
The main advantages are that the magnetic energy is recovered in the auxiliary winding
and therefore increases the performance of the system. Another benefit is that current is
allowed to rapidly build up in the phase winding using the stored magnetic energy of the
auxiliary winding. This circuit also allows for soft-switching capability. The main
disadvantage is the extra cost for an SRM with the auxiliary winding.
61
3.5 Dissipative Converters
As the name suggests, these converters do not recover the magnetic stored energy in the
phase winding, but instead dispel all the energy. The most fundamental type of
dissipative converter is the R-dump converter (Barnes 1998) (Krishnan, 1990). It has
two variants known as the decay type converter and a modified version of the decay
converter.
3.5.1 R dump converter

This converter configuration consists of a single switch and single diode per phase
winding in the SRM as indicated in Figure 3.19a (Krishnan 1990). It can only operate in
hard-switching mode with the capability for only two states: magnetization and
demagnetization.
Figure 3.19: The R-dump converter. (a) Static view of converter. (b) Magnetization
state (c) Demagnetization to the dump resistor, Rd
During magnetization, the switch, Q1, turns on and current flows in the phase winding
as illustrated in Figure 3.19b. The demagnetization commences when Q1 turns off and
the phase current freewheels through the diode, D1, charging the capacitor, C1, and later
flows through the dump resistor, Rd as shown in Figure 3.19c. The detailed waveforms
for the circuit operation are provided in Figure 3.20.
62
Figure 3.20: Waveforms for the R-dump converter circuit
At low power levels, the simplicity, low cost and low semiconductor component count
of the converter may outweigh the loss in drive efficiency. Another advantage is that
only grounded gate drive circuits are required, which further reducing converter costs.
The main limitation is that this type of converter is impractical for high power
applications as the dissipation of stored energy would substantially reduce the drive
efficiency. Other limitations are that the negative voltage drop tends to decrease as
current decreases, and the inability to apply zero volts across the voltage winding during
current conduction.
There are two variations on the R-dump circuit; the first provides a switch, Qa, in
parallel the dump resistor as illustrated in Figure 3.21a (Barnes 1998). This circuit is
called the decay converter and it provides for a zero-voltage loop and therefore the
possibility of soft-switching operation. Another version of the decay circuit adds
another switch, Qb, and diode, Db, and this allows for one voltage loop to operate in
63
freewheeling mode, while another phase is in demagnetization mode as shown in Figure
3.21b (Barnes 1998).
Figure 3.21: (a) Static view of the decay converter. (b) The modified version of a dual
decay converter
3.6 Self-Commutating Converters

These converters use various techniques to commutate between the phase-windings of
the SRM. There are a number of different converter schemes to implement self
commutation that include: the series resonant converter, the zero-voltage transition
(ZVT) PWM converter, the auxiliary quasi-resonant dc-link (AQRDCL) converter, the
current source converter and the H-bridge converter. Even though these designs are old
they are included for completeness of this review.
3.6.1 Series resonant converter

The semiconductor devices switch at zero in this converter due to the partially
resonating nature of the circuit (Park 1992) as illustrated in Figure 3.22a. This circuit
has two fundamental modes of operation: magnetization and demagnetization. During
energization of the phase winding, SCR1, SCR2 and S1 are turned on and this causes the
capacitor, C1, to overcharge, which in turn forces the current in the inductor, La, to zero
as illustrated in Figure 3.22b. Then SCR1, SCR2 and S1 turn off and the phase current
continues to flow, until C1 is negatively charged. Then S1 turns on and the
demagnetization current flows via the diodes, D1 and D1, to return energy to the dc link.
The main advantage is that switching at zero volts reduces the losses and the stresses in
the switching components. The main limitation is that the component count is high,
increasing overall system cost.
64
Figure 3.22: The series resonant converter. (a) Static view of converter. (b)
Magnetization state (c) Demagnetization to the dump resistor, Rd
Additional resonant converters include the zero-voltage transition (ZVT) PWM

converter (Park 1992) illustrated in Figure 3.23a and the auxiliary quasi-resonant dc-
link (AQRDCL) (De Doncker 1991) illustrated in Figure 3.23b. The dc link voltage in
the ZVT PWM converter is regulated using a chopping switch, Sa, that is commutated at
zero voltage through the action of a sub circuit comprising of a resonant tank, Lr and Cr,
and an auxiliary switch, Sb, that is hard switched. There are conduction losses due to the
chopping switch and some limitations on independent current control of the phase
windings.
Figure 3.23: (a) The ZVT PWM converter, (b) and the AQRDCL converter
65
The AQRDCL converter uses a resonant circuit to allow for the transition of the main
switches to occur at zero volts to further reduce switching losses. The main limitations
are the additional conduction losses in the resonant circuit switches and limitations in
independent control of the phases due to the zero voltage switching.
3.6.2 H-Bridge Converter

In the H-bridge circuit, two phases must always carry current. The circuit is configured
as illustrated in Figure 3.24a (Corda 1991). There are three stages of operation. In the
first state, the top left switch, Q1, and the bottom-left switch, Q3, turn on and current
conducts through the top-left (phase 1) and bottom-left (phase 3) phase windings as seen
in Figure 3.24b. In the second stage, when Q1 turns off, the bottom-left diode, D3, and
the top-right switch, Q2, turns on so that the sum of the current flowing in the top two
phase winding remains at a constant value as indicated in Figure 3.24c. The third stage
consists of the Q3 turning off and Q4 turning on to maintain a constant current through
phase 2 and phase 4, while D1 and D3 becomes forward bias to demagnetize phase 1
and phase 3 as illustrated in Figure 3.24d. The fourth stage consists of Q2 turning off,
while the diode, D4, and the switch, Q1, turns on so that the sum of the currents flowing
in the top two phase winding remains at a constant value as indicated in Figure 3.24e.
The final stage consists of Q4 turning off and Q3 turning on again to maintain a constant
current through phase 1 and phase 3, while D2 and D4 becomes forward bias to
demagnetize phase 2 and phase 4 as illustrated in Figure 3.24f. The main advantage is
this configuration allows for one switch and one diode per phase and thereby reducing
converter component costs. The main limitation is that current cannot flow in only one
phase and therefore prevents independent phase control. This converter configuration
requires that the machine design must consist of multiples of four phases.
66
VS VS VS
Q1 Q2 Q1 Q2 Q1 Q2
D1 D2 D1 D2 D1 D2
ph1 ph2 ph1 ph2 ph1 ph2
D3 D4 D3 D4 D3 D4
Q3 Q4 Q3 Q4 Q3 Q4
(a) (b) (c)

VS VS VS
Q1 Q2 Q1 Q2 Q1 Q2
D1 D2 D1 D2 D1 D2
D3 D4 D3 D4 D3 D4
Q3 Q4 Q3 Q4 Q3 Q4
(d) (e) (f)
Figure 3.24: The H-bridge converter. (a) Static view of converter. (b) Stage1
(c) Stage 2 (d) Stage3 (e) Stage 4 (f) Stage5
3.7 Converter selection

In this application where the SRM is implementing automotive starter and alternator
functions, there is a requirement for maximum flexibility of converter topology design
to achieve those key objectives, converter efficiency is of critical importance, as is
independence of converter phase operation in order to maximise control strategy
options. Circuit complexity and consequently cost is another key consideration.
The magnetic energy recovery converters that require an SRM with a bifilar or auxiliary
winding can be eliminated as the SRM that is used in this application is not of this type.
As efficiency is of critical importance, the dissipation converter topologies are not
suitable as the power requirements for the SRM are quite large. Of the bridge converter
topologies, the shared switch converter is also unsuitable as the conduction losses in the
switching device in series with the phase winding would have a significant effect on
efficiency. Independent control of phase windings during motoring and generating
67
modes of operation prohibits the use of the shared winding, the shared switch, the split-
dc, and the H-bridge converters.
Minimising circuit complexity and therefore the converter cost is also an important
consideration, so the resonant converters, the auxiliary and the sequential converters are
eliminated for this reason, as the component count is high in these converters. The series
and parallel dc-link converters are also eliminated due to the circuit complexity, as these
circuits have essentially a voltage boost front end on a bridge converter. The AQRDCL
converter is also be ignored for the same reasons as the dc-link converters above but
also due to the reduced efficiency incurred by the passive components in the resonant
circuit outweighing the theoretical zero switching losses under resonant operation The
C-dump is discarded mainly on the grounds of circuit complexity contributing to extra
converter costs and there are additional losses due to energy recovery circuit. The Sood
converter is discarded due to the potential costs of implementing the complex control
may outweigh the reduced component count.
The buck-boost converter potentially offers significant advantages under generating

operation with the capability of provide magnetization voltages greater the supply
voltage. This and the reduced number of switching components per phase must be
weighed against the potential inefficiency of using switching devices with a higher
voltage rating and the additional losses incurred by the passive components in the
chopper circuit. This leaves the asymmetrical half-bridge and full-bridge converters as
the circuits that provide maximum flexibility, potentially high efficiency levels, and
provides full independence over all ranges of operation. These converters are analysed
in the next chapter to determine the most suitable circuit to carry to the final converter
design stage. If variable input voltage is necessary for improved range of operation in
the SRM drive, a buck-boost front end can be added to the circuit at a later stage.
68
Chapter 4 - Preliminary Analyses
In order to design an efficient converter for SRM motoring and generating modes of
operation, power switching components were selected and evaluated to maximise
efficiency and reliability and minimise cost. To identify suitable components to
evaluate, the operating conditions were first defined in which these devices needed to
function. The efficiency of the components and the converter configuration options
were then determined through power analysis. The components were first evaluated on
the half-bridge converter and then on the full-bridge converter. The MOSFET reverse
recovery times were evaluated through simulation and experimental analysis in order to
evaluate worse-case operation for implementation as synchronous rectifiers in the full-
bridge converter. The most efficient converter configuration was determined during this
assessment and the final selection was matched with the optimum components for this
application.
4.1 Operating Conditions

The operating conditions for the SRM converter are determined by the specifications of
the SRM and the application (i.e. it must function as an automotive ISG). The SRM
specification defined the peak power and peak current requirements. The application
specifications include: the performance conditions for the ISG, the operating
environment of the ISG in which the converter must function, the worst case conduction
periods and duty cycles for motoring and generating, and automotive electrical
parameter limits.
The performance conditions in this situation are the modes and speeds at which the
SRM operates as an automotive ISG; these modes are: starting, low speed motoring,
low speed generating and high speed generating. The speed range for each mode
determines the conducting periods and the duty cycles of operation for the switching
devices, and the voltage and current requirements.
The application also determines the operating environment that is encountered by the
converter and therefore by the switching devices, which includes temperature extremes
and cycles and varying levels of vibration.
69
The automotive electrical parameter limits defined the voltage and current limits used to
specify and completely evaluate the switching components. Safety margins were
defined for the voltage and current ratings. Limits were provided for other parameters
including MOSFET on-resistance, diode forward voltage, reverse recovery time, trr, of
diodes and synchronous rectifier MOSFETs, and finally, for the case and junction
temperature limits. This allowed for the evaluation of different switching devices and
converter configurations in order to achieve the most efficient solution, and
consequently defined the requirements for the heat sinks, gate drive circuitry, filter
capacitors and the choice and configuration of the high current conductors.
4.1.1 SRM Specifications

Initially a two phase machine was proposed for the ISG but unfortunately it proved
impossible to obtain and no alternative vendors could be found to supply an equivalent
off-the-shelf 2-phase unit. The SRM that is used in this application is a four phase unit
with eight stator poles and six rotor poles as this satisfies the required voltage and
power rating A summary of the specifications for the machine is provided in Table 4.1
[(Motion System Tech 2002) and (Chapmore Controls Ltd. 2007)].
Phases 4 Aligned inductance (La) (µH) 334

Stator poles 8 Phase winding resistance (mΩ) 8
Rotor poles 6 Maximum speed (rpm) 15000
Power rating (W) 3500 Base speed (rpm) 6000
Rated voltage (V) 48 Rated torque (Nm) 5.48
Rated Current (Avg) (A) 84 Peak torque (Nm) 10.7
Unaligned inductance (Lu) (µH) 47 Weight (kg) 15
Table 4.1: SRM specifications
4.1.2 Performance Conditions

The ISG operates in four possible modes: starting, low speed motoring, low speed
generating and high speed generating. Typical starter motor and alternator specifications
in standard automobiles are illustrated in Table 4.2 for starter motors [(Visteon 2005a),
(Visteon 2005b) and (Visteon 2005c)] and Table 4.3 for alternators [(Bosch 2008)].
70
Type Power range Stall torque Current Efficiency Mass
1 0.8-1.0kW 11Nm 175A @ 5Nm 61% @ 5Nm 3.5kg
2 1.0-1.7kW 15-28Nm 160-220A @5Nm 71% @ 5Nm 2.6-3.2kg
3 1.7-2.5kW 36 - 42Nm 400A @ 15Nm 70% @15Nm 3.9-4.1kg
Note
Type 1 starter for 4-cylinder petrol (1.0L to 1.6L) and diesel (up to 1.4L) engines
Type 2 starter for 4 to 8 cylinder petrol (1.4L to 4.0L) and diesel (up to 1.8L) engines
Type 3 starter for 4-cylinder diesel (up to 2.5L) engines
Table 4.2: Typical starter motor specifications for standard automobile
Power range Min Current Max. Current Mass

1.1-2.1kW 45-82A 80-155A 4.7-6.4kg
Note
Minimum current available @ 1800 RPM
Maximum current available @ 6000 RPM
Table 4.3: Typical alternator specifications for standard automobile
Starting is the primary motoring mode of operation for the ISG, typically in the engine
speed range of 100 – 200rpm for current vehicle engines with equivalent power
capabilities in the range of 0.8-2.5kW [(Bosch 2000), (Visteon 2005a), (Visteon 2005b)
and (Visteon 2005c)]. Low speed motoring applies to start-stop or idle-stop mode of
operation that would be in the range of 500 – 1000 rpm [(Service Tech Magazine 2001)
and (Albertoni, L et al 2003)]. Generating at low speed provides a reduced power
output, equivalent to the maximum power of 1.1kW – 2.1kW for current generators, and
operates at engine speeds in the region of 1000rpm – 2000rpm (Bosch 2008). High
speed generating is defined for an engine speed range of 2000 – 3000rpm and is capable
of providing up to 3.5kW as a standard generator or through regenerative braking. At
very high speeds, 3000-6000rpm, the ISG would be disabled to provide maximum
vehicle acceleration. It was assumed that the SRM rotor would rotate twice for each
engine revolution as this would utilize almost the full SRM speed range and provide
increased torque handling capability for the given ISG package size. The operating
speeds of the SRM for each mode of operation are illustrated in Table 4.4. An
illustration of a potential implementation of the ISG in an automotive environment is
shown in Figure 4.1 and a description of each mode of operation is provided in Figure
4.2.
71
Specification Starting Low speed Low speed High speed
motoring generating generating
Min. speed (rpm) 200 1000 2000 4500
Max. speed (rpm) 400 2000 4500 6000
Table 4.4: Worst case ISG operating speeds
Figure 4.1: ISG automobile layout
Mode Operational Status Description of operation
36V
ISG Converter Battery Power is supplied from the 36V
battery on starting and the ISG
Starting Solenoid Engine Transmission restarts the engine. A level of
Clutch
propulsion assistance may also
be provides during this time
Aux. A/C, Power
steering, etc.
36V While vehicle is stationary, the

ISG Converter Battery
Vehicle engine is switched off and the
Stationary solenoid clutch is switched off
Solenoid Engine Transmission and the ISG operates in low
Clutch
(Low speed speed motoring mode driving the
motoring) auxiliary devices (i.e. the power
Aux. A/C, Power
steering, air conditioning)
steering, etc.
Normal 36V
ISG Converter Battery
driving The engine drive the vehicle. The
battery charge status is
(Low and high Solenoid Engine Transmission continuously monitored and the
Clutch
speed amount of generation is
generating) controlled as required
Aux. A/C, Power
steering, etc.
36V
Deceleration ISG Converter Battery The wheels of the vehicle drive
the ISG during deceleration
(Low and high Solenoid Engine Transmission activating regenerative braking.
speed Clutch
Some of the vehicle driving
generating)
energy is recovered to the battery
Aux. A/C, Power
steering, etc.
Figure 4.2: ISG operation
72
The ISG implementation would use a belt drive coupling system as illustrated in Figure
4.1 and the configuration is similar to systems what has been implemented by a number
of manufacturers [(Delphi 2008), (Visteon 2003) and (Teratani et al 2003)]. The ISG
would be part of a 42V power supply system that uses a 36V battery (Teratani et al
2003). This system would easily integrate into existing powertrain and would still
achieve the benefits of stop/start mode, regenerative braking and potentially launch
assistance. A solenoid clutch would be connected between the engine crankshaft and
crank pulley and would engage and disengage on command when the various operating
modes were implemented as illustrated in Figure 4.2.
4.1.3 Operating Environment

The automotive engine compartment represents a particularly harsh environment for
automotive electronics (Myers 2003). This includes extreme temperature variations
leading to thermal cycle or thermal shock (>1000 cycles), vibration (up to 10Grms),
exposure to dirt, moisture, chemicals and gases, external noises and electrostatic
discharge. The extremes in ambient temperature (-40 oC to 150 oC) are perhaps the most
difficult challenge to address. Figure 4.3 shows a typical automotive operating
environment (Myers 2003).
Figure 4.3: Automotive Operating Environment
Applications that have high power dissipation represent the greatest challenge to
electronic cooling systems design. Motor control electronic modules must be able to
operate in ambient temperatures up to 150oC. This provides a significant challenge to
the reliability of the design, especially considering the requirement for faultless
operation in all weather conditions for 10 years or more (Myers 2003).
73
The ambient temperature requirements for the converter in this application are defined
as –40oC to 100 oC by locating the converter low down and at the front of the engine to
maximise cool airflow or use water cooling to maximise heat transfer. The ambient
temperature experienced by the ISG during cold starting is defined to be -40oC to 45oC,
while during other modes of operation would experience the full temperature range. For
power analysis, the upper temperature is of greatest importance. The proposed worst-
case temperatures to be experienced by the converter under the different modes of
operation of the ISG are shown in Table 4.5 (at the end of section 4.1)
4.1.4 Conduction Periods and Duty Cycles

A detailed analysis used to determine the conduction intervals during starting, low
speed motoring, and low and high speed generating operation is presented in the
following sections, and the details for each speed is provided in Table 4.5 (at the end of
section 4.1). It is assumed that the worst case speed of operation for the ISG would be
200rpm for starting, 1000rpm for low speed motoring, 2000rpm for low speed
generating and 4500rpm for high speed generating. These speeds yield the maximum
conduction intervals for each mode of operation of the ISG.
A typical circuit schematic for a single phase SRM converter is shown in Figure 4.4.
The duration over which a phase winding becomes active and repeats the cycle again is
defined in this application to be the conduction period. The active interval is then
defined to be the conduction duty cycle, DC. These definitions are illustrated for starting
and low speed motoring, low speed generating and finally high speed generating in
Figures 4.5a, 4.5b and 4.6, respectively.
+VS
Q1
D2
Phase Winding
D1 Q2
0V
Figure 4.4: Single Phase SRM Converter
74
4.1.4.1 Starting and low speed motoring
During starting and low speed motoring, it is assumed that current is flowing just before
and during the entire motoring cycle (region of positive dL/dθ) and for a short period
after alignment for demagnetisation as indicated by the interval θon, θ1, θ2 and θext in
Figure 4.5a. An ISG starting speed of 200rpm implies that one revolution takes
approximately 300ms. Since the SRM has 6 rotor poles, each phase is energised six
times for one complete revolution, which yields a conduction period of 50ms for each
SRM phase.
Figure 4.5: PWM current control for single phase operating at low speed.
(a) Motoring - Soft chopping (b) Generating - Hard chopping
75
It is assumed there is a maximum conduction duty cycle for motoring mode of 50%.
Therefore, the MOSFET, Q2, is on for 25ms with the switching frequency of the device
determined by the SRM operating speed, which under starting conditions is 20Hz
(50ms). The MOSFET, Q1, is pulse width modulated (PWM) at 25kHz - 50kHz,
depending on audible noise and allowable levels of component switching losses during
the commutation duty cycle in order to maintain the phase winding current at its set
point value. The PWM duty cycle varies in order to maintain the maximum phase
current and here it is assumed an average duty cycle value of 50%, which yields an
overall duty cycle for Q1 of 25%. The diode, D1, is on when Q1 is off, and vice versa as
part of the freewheeling cycle, yielding an overall duty cycle of 25%, but this device
also conducts during the demagnetisation cycle, which is assumed to be 10% duty cycle
at 20Hz. The diode D2 is only on during the demagnetisation of the phase winding and
therefore a 10% duty cycle at 20Hz can be assumed. These would be deemed as the
worst-case conduction duty cycles for the converter during starting operation. Low
speed motoring assumes the same duty cycles as starting but for a speed of 1000rpm.
The specific duty cycles and switching frequencies for each device are provided in
Table 4.5 (at the end of section 4.1) and a detailed graphical representation of the
current and voltage waveforms for each device is provided in Appendix A.
4.1.4.2 Low speed generating

During low speed generating, the worst case scenario would be that conduction would
commence in the end of the dead zone (region of zero dL/dθ) prior to the generating
region (region of negative dL/dθ) to allow the current reach a set point and then the
generating current regulation would take place during the entire generating region and
allowed to extinguish in the following dead zone area as indicated by the interval θon, θ3,
θ4 and θext in Figure 4.5b. This would result in the worse case duty cycle of
approximately 10% at the end of the previous motoring mode and dead zone,
approximately 40% for generating mode of operation and 10% during final
demagnetisation.
Low speed generating requires current regulation as the magnitude of the back-emf is
smaller than the combined magnitude of the applied voltage and resistive voltage drop
below the base speed. The SRM low speed generating range is defined where back emf
is greater than 0.333 of the DC link voltage and less than the value of the DC link
76
voltage and therefore the rotor speed is greater than 0.333 of the base speed but less
than the base speed (Miller 2001).
Low speed generating uses hard-chopping PWM current regulation and that implies that
all devices are operating at the same PWM frequency with the average duty cycle
assumed to be 50%. This points to an overall duty cycle of 25% for all the devices over
the conduction period. D1 and D2 are on for an additional 10% for demagnetization.
The specific duty cycles and switching frequencies for each device are provided in
Table 4.5 (at the end of section 4.1) and a detailed graphical representation of the
current and voltage waveforms for each device is provided in Appendix A.
4.1.4.3 High speed generating

High speed generating is implemented through single pulse operation and conduction
commences in the motoring mode of operation to allow the current in the phase winding
to reach the desired value. It remains constant through the generating region until
overlap occurs at minimum inductance at which point it tails off to zero in the dead zone
as illustrated by the interval θon, θ2, θ3, θoff ,θ4 and θext in Figure 4.6.
At the base-speed, the generating current is approximately constant due to the source
voltage being approximately equal to the back-emf. The source voltage is 42Vdc during
generating at an ISG operating speed range of 4500rpm – 6000rpm and consequently
provides maximum generating power capability over normal automotive engine cruising
speed. The generator output is assumed to be disabled above 6000rpm.
The worst case high speed generating occurs for 42Vdc at 4500rpm and the worst case
duty cycles are 40% in motoring mode and 30% in generating mode and a further 10%
for demagnetisation. The switching frequencies and associated duty cycles for the
switching devices at 4500rpm are provided in Table 4.5 (at the end of section 4.1) and a
detailed graphical representation of the current and voltage waveforms for each device
is provided in Appendix A.
77
Figure 4.6: Single pulse control for single phase generating at high speed
78
4.1.5 Automotive Electrical Parameter Limits
As noted previously, a 36Vdc battery is used in this 42Vdc power supply system
[(Service Tech Magazine 2001) and (Teratani et al 2003)]. According to the SRM
specifications [(Motion System Tech 2002) and (Chapmore Controls Ltd. 2007)], the
peak power capability is 5900W at 6000rpm at the rated voltage of 42Vdc, which
implies a peak current, Im, of 140A.
Based on the above criteria, it is possible to define the electrical requirements at the
various engine speeds. Under cold starting conditions the available battery voltage is
36Vdc, which assuming a peak current capability of 140A available to the SRM phase
winding during initial engine start-up. During SRM low speed motoring for start-stop
operation, it is assumed that up to 42Vdc is available and the current values are the
same as start-up. During low and high speed generating, the peak voltage and peak
current are assumed to be 42Vdc and 140A, respectively.
A safety margin was assumed for all specifications. With an operating voltage of
42Vdc, and assuming a safety margin of 50%, a minimum voltage rating of 63Vdc is
required for the power semiconductors. The peak operating current rating is 140A, and
assuming a safety margin of 25%, this yields a peak current rating of 175A. The on-
resistance of all MOSFETs and the forward voltage of the diodes must be as low as
possible in order to maximise efficiency. The reverse recovery time, trr, specification of
120ns is based on the average value from the ten most suitable diodes available that
meet the voltage and current specifications [(Microsemi 2004), (International Rectifier
1997), (IXYS 2003), (STMicroelectronics 2004), (Fairchild 2002b)]. The case
temperature of a power semiconductor can range from 90oC to 110oC in practice
(International Rectifier 2005a). An upper value of case temperature was used in this
application. A summary of the electrical requirements is given in Table 4.5.
79
ISA Function Starting Low speed motoring Low speed generating High speed generating
Voltage (VS) 36 42 42 42
PeakCurrent (A) 140 140 140 140
PeakCurrent (A) per Switch 140 140 140 140
Avg Current (A) 77 77 77 77
Avg Current (A) per Switch 77 77 77 77
Peak Power (W) 5040 5880 5880 5880
Avg Power (W) 2772 3234 3234 3234
ISA Speed (rpm) 200 1000 2000 4500
Rev/Sec 3.33 16.67 33.33 75
Period/Rev (ms) 300 60 30 13.3
Rotor Pole 6 6 6 6
Pole Period (ms) 50 10 5 2.2
80
Frequency (Hz) 25000 20 25000 100 25000 200 450 450
Dudy Cycle - Q1 25% 25% 25% 40%
Dudy Cycle - Q2 50% 50% 25% 40%
Dudy Cycle - D1 25% 10% 25% 10% 25% 10% 30% 10%
Dudy Cycle - D2 10% 10% 25% 10% 30% 10%
Notes
o
1. Maximum ambient temperature in all cases (TA ): 90 C 8. Heat Spreader length: 0.05 m
o
2. Maximum semiconductor case temperature (TC): 110 C 9. Heat Spreader width: 0.025 m
o
3. Maximum semiconductor junction temperature (TJ): 110 C 10. Heat Spreader depth: 0.016 m
o
Table 4.5: SRM Worst Case Operating Conditions

4. Minimum semiconductor voltage rating: 42 Vdc 11. Thermal Coupler RθTC-S 0.2 C/W
5. Minimum semiconductor peak current rating: 140 A
6. Minimum semiconductor continuous current rating: 84 A Device Number 1
7. Minimum semiconductor switching frequency: 25 kHz Rotor Pole No. 6
4.2 Switching Component Analysis
Evaluation of the most efficient component options was central to determining the final
converter configuration for motoring and generating operation. A number of suitable
component options were selected from the vendor documentation. A power analysis
was implemented with each converter configuration using what is deemed to be the
most efficient devices from the specifications. The final converter configuration was
selected from this analysis and then devices from a number of vendors were evaluated
in order to determine the most efficient device and converter combination.
Since the SRM is capable of efficiency levels greater than 90% (Chapmore Controls
Ltd. 2007), it is imperative that the converter circuit be equally efficient in order to
maximise the advantages of this motor configuration. The analysis was conducted on
three converter configurations: asymmetric half-bridge converter, the full-bridge
synchronous rectifier converter with single components and the same full-bridge
configuration with a number of switching components mounted in parallel, as illustrated
in Figure 4.7(a), (b) and (c) respectively. The first case for analysis consists of two
MOSFETs and two diodes per phase winding, while in the second case the diodes are
replaced with MOSFETs to implement synchronous rectification. In the third and final
case each MOSFET is replaced with a number of lower current rated MOSFETs
connected in parallel and is defined as the parallel configuration.
Figure 4.7: (a) Asymmetric half-bridge converter, (b) full-bridge synchronous rectifier
converter with single components and (c) full-bridge synchronous rectifier with
switching components mounted in parallel
The power analysis was implemented for motoring operation at two speeds, 200rpm and
1000rpm, and for generating operation at two speeds, 2000rpm and 6000rpm. Detailed
81
power calculations are presented for the starting speed (200rpm) of the asymmetrical
half-bridge converter, while the power calculations for the other speeds were
implemented using spreadsheet program. Appendix A provides a complete analysis of
the switching waveforms for each component of the asymmetrical half-bridge converter
and the full-bridge synchronous rectification converter for each converter mode
(magnetisation, freewheeling or demagnetisation) and at each ISG mode of operation
(starting, low speed motoring, low and high speed generating) and was used to select
the relevant equations for the power calculations, while the parameter values for each
component is provided in Table 4.5 at the end of section 4.1. A summary of the
efficiency calculations for all three converter configurations and suitable component
options for all ISG modes of operation is provided in Table 4.9 (at the end of section
4.2) and the full power analysis results are provided in Appendix B. The most efficient
components were then evaluated for reverse recover time specification under simulated
conditions. Short-listed components from the simulation study were electrically tested
in order to determine final selection.
4.2.1 Asymmetric Half-bridge Converter

The most suitable MOSFET (Q1 and Q2) and diode (D1 and D2) for this application
was the Advanced Power Technology APT20M11JFLL (Advanced Power Technology
2004) and the ST Microelectronics STTH20002TV (ST Microelectronics 2004)
respectively. The power calculations were implemented according to the analysis for
conduction and switching losses that is provided in Appendix A. The procedure for the
power analysis was divided into three stages, (i) the power calculations for each power
semiconductor, (ii) the total power dissipation was implemented for a single phase
converter and (iii) the efficiency was evaluated to determine if it was ≥ 90%. A detailed
set of calculations is presented for the SRM starting speed, to illustrate the process and a
summary is provided for the other operating speeds. The power analysis calculations
were generated using a spreadsheet program to evaluate the various operating
conditions and the efficiency results are provided in Table 4.9 at the end of section 4.2.
The power dissipation of the MOSFET, Q1, consists of conduction losses and switching
losses. The conduction losses depends on the value of its on-resistance (RDS(ON)) and
this value must be derated by the on-resistance temperature coefficient. The on-
resistance temperature coefficient for maximum junction temperature was assumed to
be the worst-case condition. A summary of the relevant data from the component
82
specifications (Advanced Power Technology 2004) and operating conditions to
implement the power analysis that follows is provided in Table 4.6. It was assumed that
the worst case rise (tr) and fall (tf) times of the MOSFET to be 10 times greater than the
value in the specifications as a rule of thumb (Sweeney 2002). The power analysis for
Q1 and Q2 under ISG starting mode of operation is implemented according to the
analysis described in section A3 of Appendix A.
RDS(ON) 11mW
thermal coefficient for RDS(ON)at TJ = 150oC 2.25
RDS(ON) at TJ = 150oC, RDS(ON)HOT 25mΩ
tr 50ns
tf 100ns
Worst case tr, tr(wc) 0.5µs
Worst case tf, tf(wc) 1µS
trr 500ns
Peak current, Im 140A
Minimum switching frequency, fs(Q1) 25kHz
Switching duty cycle (DSW) 25%
Commutation duty cycle (DCOM) 50%
Commutation frequency (Starting), fs(Q2) 20Hz
Demagnetisation duty cycle (DDM) 10%
VDS(Q1) = VDS(Q2) = VDC 36V
V(D1) = V(D2) = VF 1.8V
Table 4.6: Component specifications and operating conditions for power analysis
The conduction power losses are determined as follows:

PC ( Q1) = ( I RMS ( Q1) ) 2 * Rds ( ON ) HOT (4.3)
where
I RMS (Q1) = I m DSW (4.4)
I RMS ( Q1) = 70 A
Therefore:
PC ( Q1) = 121W
83
The switching transition power losses are:
VDS ( Q1) * I m
PS ( Q1) = * (t r ( wc ) + t f ( wc ) ) * f s ( Q1) (4.5)
2
PS ( Q1) = 95W
Total power dissipation for this device, PT(Q1), is the sum of the conduction losses,
PC(Q1), and the switching losses, PS(Q1).
PT(Q1) = PC(Q1) + PS(Q1) (4.6)
PT(Q1) = 216W
The power dissipation of the MOSFET, Q2, consists of conduction losses and switching
losses. In this case, the commutation duty cycle, DC, is 50% and the switching losses
also occur at the commutation frequency, fs(Q2), of 20Hz, otherwise the component data
is the as in Table 4.4. The conduction power losses are determined as follows:
I RMS ( Q 2 ) = 99 A
Therefore:
PC (Q 2 ) = 243W
The switching transition power losses are:
PS (Q 2 ) = 76mW (negligible)
Total power dissipation for this device, PT(Q2), is the sum of the conduction losses,
PC(Q2), and the switching losses, PS(Q2).
PT(Q2) = 243W
84
The losses of the diode, D1, occur in two regions: (I) freewheeling and (II)
demagnetisation. A summary of the relevant data from the specifications (ST
Microelectronics 2004) and the operating conditions to implement the power
calculations is provided in Table 4.7. The power analysis for D1 and D2 under starting
speed of operation was implemented according to the analysis from Appendix A.
Worst case forward voltage drop (VF) 1.8V

Peak current, Im 140A
VR = VS 36V
Qrr @ TJ=125oC and @diF/dt = 100A/uS 275nC
trr 85nS
Minimum switching frequency, fsw 25kHz
Freewheeling duty cycle (DF) 25%
Commutation frequency (Starting), fc 20Hz
Demagnetisation duty cycle (DDM) 10%
Table 4.7: Component specifications and operating conditions for power analysis
(I) Freewheeling
The conduction power losses during freewheeling, PCF(D1), are determined as follows:
PCF ( D1) = I m ( D1) *V F * DSW (4.7)
PCF ( D1) = 63W
The switching transition power losses during freewheeling, PSWF(D1), are determined as
follows:
PSF ( D1) = Qrr ( D1) *V R * f sw (4.8)
PSF ( D1) = 250mW (negligible)
(I) Demagnetisation
The conduction power losses during demagnetisation, PCDM(D1), are determined as
follows:
I m ( D1) = 0.5 * I m (4.9)
85
The current waveform during demagnetisation is of a triangular shape; therefore the
output current is approximately half during this interval.
I m ( D1) = 70 A
Therefore:
PCDM ( D1) = 13W
The switching transition power losses during demagnetisation, PSWDM(D1), are

determined as follows:
PSDM ( D1) = 200uW (negligible )
Total power dissipation for this device, PT(D1), is the sum of the conduction losses,
PCF(D1), and switching losses, PSWF(D1), during freewheeling, and the conduction losses
PCDM(D1), and switching losses, PSWDM(D1), during demagnetisation.
PT(D1) = PCF(D1) + PSF(D1) + PCDM(D1) + PSDM(D1) (4.10)
PT(D1) = 76W
The power dissipation of the diode, D2, only occurs during demagnetisation, which is
be the same as for the D1 during demagnetisation. Therefore the total power loss for
this device is:
PT(D2) = 13W
The power loss for the single phase, PT(SP), during starting is the sum of the total power
losses incurred by each device.
PT(SP) = PT(Q1) + PT(Q2) + PT(D1) + PT(D2) (4.11)
PT(SP) = 550W
86
4.2.1.1 Efficiency during starting mode
The efficiency during starting mode of operation is implemented as described in A2 of
Appendix A. The power transfer occurs in two regions (I) magnetisation and (II)
demagnetisation as indicated in Figure 4.8.
L Stroke or Conduction Period
Lmax
Lmin
θon θ1 θ2 θext θ3 θ4 θs θ
+VS
Vphase
t
I II
-VS
Im
Iphase
t1 t2
t0=0 D1TS t3 = TS t
D2TS
TS
Figure 4.8: Phase winding voltage and current waveforms during starting and low
speed motoring
(I) Magnetisation
During the interval t0 – t1, the current is constant with a peak value, Im, of 140A and the
voltage is a square wave and has a peak value, VS, of 36Vdc and a duty cycle, D1, of
50%. The equation for the average power during demagnetisation (PM) that was
determined in Appendix A is defined as:
PM = 0.5I mV S D1 (4.12)
The output power during magnetisation is
PM = 1260W
87
(II) Demagnetisation
In the interval t1 – t2, the voltage is constant with a peak value, -VS, of -36Vdc the
current is approximated by a triangular wave and has a peak value, Im, of 140A and a
duty cycle, D2, of 10%. The equation for the average power during demagnetisation
(PDM) is:
PDM = − 0.5VS I m D2 (4.13)
The power transfer is from the phase winding to the battery so the average power during
demagnetisation is:
PDM = − 1( − 0.5VS I m D2 ) = 0.5VS I m D2 (4.14)
The output power during magnetisation is
PDM = 252W
The total power transfer during staring mode of operation is
Pphase = PM + PDM (4.15)
Pphase = 1512W
Therefore the efficiency for the converter phase during starting mode of operation is
Pphase
η= (4.16)
Pphase + PT ( SP )
η = 73%
The power analysis calculations were repeated for low speed motoring, low speed
generating and high speed generating as described in A.2 of Appendix A and the results
generated using a spreadsheet program and the efficiency results for each speed of
operation are provided in Table 4.9 at the end of section 4.2, while the detailed power
88
analysis calculations are provided in Appendix B. From the results, it can be seen that
the efficiency for this configuration was < 90% and was therefore not suitable for the
application.
4.2.2 Full-bridge Synchronous Rectifier Configuration

The full-bridge synchronous rectification allows for the use of MOSFETs instead of the
freewheeling and demagnetisation diodes, which was a large source of power loss
[(Mohan 1995a), (STMicroelectronics 2001) and (Excelsys Technologies Ltd 2004)].
The schematic for the converter circuit is shown in Figure 4.9. This scheme reduces the
resistive losses during freewheeling and generating modes of operation. Another
important characteristic when implementing this configuration is that the reverse
recovery time of the MOSFET must be approximately equivalent to that of the diode.
The components that are used as synchronous rectifiers are the same devices used in the
asymmetric half-bridge converter configuration (Advanced Power Technology 2004). A
summary of the specifications for a number of diodes that is tabulated in Table 4.10 in
section 4.3 indicates that the reverse recovery time needed to be ≤ 130ns. It can also be
seen from Table 4.6 that the MOSFET had a reverse recover time of 500ns and
therefore this device would not be suitable for this application.
Figure 4.9: Full-Bridge Synchronous Rectifier
For the sack of completeness, a power analysis study was implemented on this
configuration. The power calculations follow the procedure described in A.3 of
Appendix A. The efficiency calculations are provided for each speed of operation in
Table 4.9, while the detailed power analysis is provided in Appendix B. As can be seen
from the results in Table 4.9, this implementation is less efficient than the asymmetric
half-bridge configuration. This is due to Q3 having significant conduction power losses
89
that do not occur with the diodes in the asymmetric half-bridge configuration. Therefore
this scheme is not suitable on two counts, poor efficiency and reverse recovery time.
4.2.3 Parallel Configuration

This circuit is a derivative of the full-bridge synchronous rectifier scheme but the
individual MOSFET are replaced with a number of lower power devices connected in
parallel as illustrated in Figure 4.10,. The lower power devices have much lower reverse
recovery time and on-resistance values than the single high power device as can be seen
in Table 4.8. The derating for all MOSFETs was taken at the worst-case junction
temperature (TJ) of 175oC.
Figure 4.10: Full-Bridge Synchronous Rectifier parallel configuration
The analysis follows the same procedure as the full-bridge synchronous rectification
configuration for the Q1, Q2, Q3 and Q4 as described in A.3 of Appendix A. The current
through each device was evaluated for different numbers of devices in parallel. The
power loss for the single-phase converter is the sum of the component power losses
multiplied by the number of devices in parallel. The number of devices evaluated in
parallel was two and four in order to determine the minimum quantity required to meet
the efficiency specification of ≥ 90%. The specifications for the most suitable devices
[(Fairchild 2002a), (Fairchild 2003), (Fairchild 2004), (Infineon 2003), (Philips 2002)
and (Philips 1999)] are provided in Table 4.8 and a detailed summary of the efficiency
results for each component is provided in Table 4.9. There were no suitable components
from International rectifier, On Semiconductor, ST Microelectronics, IXYS, APT,
Toshiba and Vishay. The complete analysis for each component is presented in
Appendix B.
90
Make & Model Vds Ids Rds(on) Rds(on) RthetaJC trise tfall trr Package
o
Fairchild (Vdc) (A) (mOhm) Derating ( C/W) (nS) (nS) (nS)
FDP047AN08A0 75 80 4.7 2.34 0.48 88 45 53 TO-220
FDH038AN08A1 75 80 3.8 2.67 0.33 141 126 50 TO-247
FDP060AN08A0 75 80 6 2.17 0.58 79 38 37 TO-220AB
Infineon
SSP80N08S2L-07 75 80 7.1 1.83 0.5 81 78 100 TO-220
Philips
BUK7506-75B 75 75 5.6 2.1 0.5 56 48 86 TO-220AB
PSMN009-100W 100 100 9 2.78 0.5 100 100 80 TO-247
Table 4.8: MOSFET Specifications
It can be seen from the results that only the Philips BUK7506-75B can satisfy the
specification (≥ 90% efficiency) for all ISG modes when 4 devices are used in parallel,
while the Fairchild FDP047AN08A0 and FDP060AN08A0 are marginal. The Fairchild
FDH038AN08A1, the Infineon SPP80N08S2L-07 and the Philips PSMN009-100W
failed for all device numbers and therefore this device was eliminated from further
evaluation. The actual number of switching devices that were implemented in parallel
was a trade off between efficiency, cost, heatsink capabilities, gate drive complexity and
cost. Two devices in parallel were deemed as sufficient to achieve the efficiency
specification and still retain a reasonably low level of circuit complexity and cost. The
selected components were then evaluated for reverse recovery time capabilities in order
to finalize the most suitable candidates.
91
Power Analysis Starting Low speed motoring Low speed generating High speed generating
Speed (RPM) 200 1000 2000 4500
Classic Configuration
Efficiency (% ) 73 76 85 93
SynchronousRectifier
Configuration
Efficiency (% ) 71 74 79 87
Parallel Configuration
Fairchild FDP047AN08A0
Efficiency (%)- 2 devices 86 87 87 97
Fairchild FDH038AN08A1
92
Infineon SPP80N08S2L-7
Phillips BUK7506-76B
Table 4.9: Component Efficiency Comparison (%)

Phillips PSMN009-100W
4.3 Reverse Recovery Analysis
A simulation study was implemented on the most suitable devices to evaluate the
reverse recovery time, trr, of each component. This time needed to be as low as possible
in order to minimise shoot through currents during freewheeling operation of the
converter. An electrical test was preformed on each of the approved components to
verify simulation accuracy and finalize the component selection.
4.3.1 Simulation Analysis

The reverse recovery time, trr, is the period of time that the current conducts in the
reverse direction at turn off of a diode (Mohan 1995b). It is measured from the zero
crossing where the diode current goes from forward bias to reverse bias, to the point at
which a straight line through the maximum reverse recovery current (IRM or IRRM) and
0.25 * IRM passes through zero as illustrated in Figure 4.11 [(Microsemi 2004) and
(Fairchild 2002b)]. The reverse recover time of the intrinsic diode in the MOSFET that
is used this application must limit the amount of shoot through current experienced by
the circuit. A time of ≤ 130ns was chosen in order to compare favourably with the
average value of the diodes that could be used in the asymmetric half-bridge
configuration [(Microsemi 2004), (International Rectifier 1997), (IXYS 2003),
(STMicroelectronics 2004), (Fairchild 2002b)] as indicated in Table 4.10. The reverse
recovery time of the candidate MOSFETs was initially evaluated under simulated
conditions and successful devices were then evaluated in an experimental test circuit to
verify actual device operation.
Manufacturer Microsemi IRF IXYS ST Fairchild

Component APT2X101_100DQ60J 80EPF02 DSEP 2x 101-04A STTH20002TV RURG8060
trr (nS) 160 195 120 85 85
Table 4.10: Diode reverse recovery times
Figure 4.11: Diode Reverse Recovery Waveform
93
The test circuit for reverse recovery simulation is illustrated in Figure 4.12. The pulse
period was adjusted in order to obtain a forward current (IF) through the diode of 20A
and once the desired forward current is achieved, a reading was obtained for the reverse
recovery time (trr) and the maximum reverse recovery current (IRM). The resultant
waveforms trr and dif/dt are illustrated in Figure 4.13, respectively. The simulation
results for all the candidate devices are tabulated in Table 4.11.
Figure 4.12: Reverse Recovery Simulation Test Circuit
Figure 4.13: Reverse Recovery Electrical Test Waveform – trr, IRM
94
4.3.2 Experimental Analysis
The test circuit for the reverse recovery electrical test is illustrated in Figure 4.14. The
pulse period was adjusted for a forward current (IF) of up to 20A and the readings taken
for forward current (IF), diF/dt, reverse recovery time (trr) and current (IRM) are tabulated
in Table 4.11. Examples of the reverse recovery time waveforms are included in Figure
4.15 for forward current (IF) and diF/dt and in Figure 4.16 for reverse recovery time (trr)
and current (IRM). From the results, it can be seen that all of the devices from Fairchild
Semiconductor are suitable for implementation in the converter. The reverse recovery
time of Infinion device is too high, while there was no simulation file available to test
the device from Philips Semiconductor. The Fairchild Semiconductor FDP047AN08A0
was selected for the circuit implementation due to component availability.
+VS
Q1 R Load
L Load
Current Probe Amplifier

Q2 trr
Pulse
Generator Oscilliscope
Figure 4.14: Reverse Recovery Electrical Test Circuit
95
Figure 4.15: Reverse Recovery Electrical Test Waveform – IF
Figure 4.16: Reverse Recovery Electrical Test Waveform – trr, IRM
Reverse Recovery Data Sheet Simulation Study Electrical Test

Analysis trr trr IF IRM trr IF IRM
Parallel Configuration (nS) (nS) (A) (A) (nS) (A) (A)
Fairchild
FDP047AN08A0 53 95 43.4 0.56 67 10 2.5
FDP060AN08A0 37 87 43.4 0.46 58 10 3
Phillips
BUK7506-76B 86 No Pspice model file No Samples Available
Table 4.11: Reverse Recovery Analysis
96
4.4 Summary
Component cost and efficiency are essential aspects of the converter design. The
switched reluctance motor has an efficiency value of over ninety percent, so the
converter must also operate at an equivalent level. The key element in the converter
design is keeping the switching component losses to a minimum. Significant
developments in MOSFET technology have enabled conduction losses to be reduced
significantly. The choice of switching frequency has a significant bearing on the
component losses to the point where these are the dominant losses. Therefore,
component power analysis is essential in determining if single components or multiple
components in parallel provide the optimum solution. From the implementation of the
power analysis, simulation study and electrical tests, the full bridge synchronous
rectifier converter with parallel switching devices offers the most efficient solution for
both motoring and generating operation. The implementation of a number of switching
devices in parallel requires careful consideration in order to obtain efficient and reliable
operation of the converter.
To evaluate the converters configurations and the associated switching components, the
operating conditions were defined in order to set limits for the converter operation. The
operating conditions were determined by the SRM specification and the application of
the machine as an ISG. The application specific conditions included the performance
conditions and operating environment for the ISG, the conduction periods and duty
cycles experienced by all the switching components under all converter modes
(magnetisation, freewheeling and demagnetisation) and ISG modes (starting, low speed
motoring, low and high speed generating), and the specific automotive electrical
parameter limits.
The full-bridge synchronous rectifier with two MOSFETs mounted in parallel provided
the best compromise of efficiency, cost and complexity for implementation with the
ISG under motoring and generating operation. The most efficient MOSFETs were
evaluated for reverse recovery time capability in order to minimise the possibility of
current shoot through during freewheeling mode of operation. The analysis was
implemented under simulated and experimental conditions and a device from Philips
(BUK7506-75B) was deemed the most acceptable component for this application and
97
two devices from Fairchild (FDP047AN08A0 and FDP060AN08A0) were deemed as
marginal.
As can be seen from the power analysis computations in Appendix B, the power
dissipation of the switching devices during low speed generating is excessive. As low
speed generating occurs in the region from engine idle speed (1000rpm) up to engine
cruising speed (2000rpm), it would represent only a small contribution to the total
generating requirements. Generating at around engine idle speed would be rare when
the vehicle is operating in start/stop mode. Eliminating generating operation between
engine idle speed and engine cruising speed would be more efficient as this would
represent the region where vehicle acceleration would occur. Low speed generating is
therefore not considered beyond this preliminary analysis even though it is a
theoretically possible mode of operation.
98
Chapter 5 Converter Design
The previous chapter addressed the power losses in the switching components and
determined that a parallel component configuration is the most effective solution. This
chapter addresses the detailed design process for the converter. A detailed discussion is
presented on the implementation of the parallel component configuration and its
associated requirements. The heat sink design to determine the most appropriate heat
sink configuration for this application is described. Gate drive design is then addressed
incorporating the issues of blanking time and circuit isolation. The selection of the
appropriate value of filter capacitor is explained and a discussion on the high current
conductor configuration is included. Current sensor requirements are addressed and
finally the specifications for the prototype circuit layout are outlined.
5.1 MOSFETs in Parallel

When discrete MOSFETs are configured in parallel, the current handling capability of
the circuit can be increased significantly. However, the actual implementation is more
complex than using single high power modules. Numerous issues need to be addressed,
including thermal coupling, stray inductances, circuit layout, gate drive circuitry and
parasitic oscillations, in order to maximise current balancing between the parallel
devices.
Careful attention needs to be given to the thermal coupling when MOSFETs are
connected in parallel. Under steady state conditions the positive temperature coefficient
of the MOSFET on-resistance tends to equalize the current in parallel devices
(International Rectifier 2004a), (Gauen 1984) and (Mohan 1995). Although, this has
less of an effect under dynamic operation, it is still essential to have close thermal
coupling between devices to maintain junction temperature equilibrium (Gauen 1984)
and (Pelly 2004). Parallel devices should be mounted on the same heat sink and a
current derating of 20% is also required to account for variation in device
characteristics. The requirement for tight thermal coupling weighs against electrical
isolation and when a thermal barrier is placed between the individual devices it tends to
decouple the individual junction temperatures. Therefore, thermally conductive
insulator pads, placed directly at the cooling surface of a TO-220 or TO-247 device or
electrically isolated TO-220 and TO247 packages, are not ideal for parallel device
configuration (Pelly 2004). From this observation, the Fairchild Semiconductor device,
99
FDH038AN08A1, with its TO247 electrically isolated package was deemed unsuitable
for this application. The parallel devices should be mounted on a common heatsink. If
electrical isolation is required, the parallel devices can be mounted on a common
heatspreader to thermally couple the junctions. The isolation barrier is placed between
the heatspreader and the main heatsink as illustrated in Figure 5.1.
Figure 5.1: Using a heatspreader to provide thermal coupling between parallel devices
and electrical isolation with main heatsink.
The effect of stray collector, source and gate inductances can be significant when the
devices are connected in parallel. Stray collector inductance can lead to voltage
imbalance and overshoot conditions but it can be tolerated so long as it does not exceed
device ratings (International Rectifier 2004a). It can also affect the turn-on times by
dropping voltage at the collector. Stray source inductance must be balanced, especially
at high frequencies of operation, as it affects device turn-on times due to the
counteractive voltage drop slowing the rise of the gate voltage (International Rectifier
2004a). It can also lead to different turn-off times in devices causing excessive current
in some devices. Any significant levels of stray gate inductance can lead to dv/dt
induced device turn-on (International Rectifier 2004b).
Circuit layout is of critical importance when implementing a parallel MOSFETs

configuration. Minimising conducting path lengths can significantly reduce stray
inductance, while symmetrical layout of components and routing of connections
equalizes the stray inductance (International Rectifier 2004a), (Forsythe 1981), (Gauen
1984) and (Mohan 1995). The most significant effect of non-symmetrical layout is
current imbalance during switching intervals and consequently unbalanced switching
losses. The stray source inductance is the most critical element to balance. An ideal
layout for balancing the source inductance and the switching losses is the circular layout
100
illustrated in Figure 5.2a (Pelly 2004). Where switching losses are low due to switching
frequency and/or switching voltage is relatively low, the simple in-line layout illustrated
in Figure 5.2b can be used even though it is not electrically symmetrical.
Figure 5.2: (a) Circular layout, (b) in-line layout
Special attention also needs to be given to the gate drive circuitry implementation and
layout and to the prevention of parasitic oscillations. The gate drive circuit must
represent a stiff voltage source for rapid device turn-on and it must provide very low
impedance for fast device turn off in order to help negate some circuit imbalance
(International Rectifier 2004a), (Forsythe 1981), (Fairchild 1993) and (Gauen 1984). It
is also extremely important that the gate drive impedance is matched through careful
layout (Gauen 1984). Close symmetrical layout assists in matching the gate drive
impedances and minimising stray inductances (Gauen 1984).With the low impedance
paths created when using MOSFETs in parallel, parasitic oscillations can become a
significant problem. Individual gate decoupling resistors (10Ω – 20Ω) or ferrite beads
provide damping to prevent oscillations (International Rectifier 2004a), (Forsythe
1981), (Gauen 1984) and (Mohan 1995). Zenor diodes to protect the MOSFET gate
should be placed on the drive side of the gate resistor to prevent oscillations, while any
capacitors in the gate drive circuit can also lead to oscillations (International Rectifier
2004a).
The following are the key areas that need to be addressed when implementing
MOSFETs in parallel:
• Close thermal coupling of parallel MOSFETs to prevent junction temperature
imbalance.
101
• Symmetrical component placement and layout with minimum length connection
paths in order to balance and minimise any stray circuit inductance.
• Gate drive circuit that represents a stiff voltage and has very low impedance.
• Individual decoupling resistors or ferrite beads on the MOSFET gates to prevent
oscillations.
5.2 Heatsink Design

Thermal design must be considered at an early stage when designing power electronic
equipment to operate at high ambient temperature conditions (Mohan 1995). It is critical
to maintain the lowest possible junction temperature as the failure rate doubles for every
10oC rise in junction temperature (Mohan 1995) and (Hill 2004). It is difficult to keep
power semiconductor case temperature below 90oC without using a water-cooled
heatsink (International Rectifier 2004c). This discussion reviews the fundamentals of
heat sink design and the specific requirements for this application. The step-by-step
procedure for the heat sink design is then discussed for the starting mode of operation
and the process concludes with the selection of the most suitable heat sink configuration
for this application.
5.2.1 Fundamentals
Three factors affect junction temperature: the sum of all thermal resistance from the
junction to the ambient environment (air, oil or water), the amount of heat to be
dissipated and finally the ambient temperature (Hill 2004). A schematic (Hill 2004) and
(Polyfet 2005) of the “equivalent circuit” for the thermal resistances is given in Figure
5.3.
Figure 5.3: Schematic for thermal resistances
Thermal resistance, Rθ, is derived from the equation for energy flow from the higher
temperature end to the lower temperature end of a material as illustrated in Figure 5.4
and is determined from the formula (Mohan 1995)
102
λA∆T
Pcond = (5.1)
d
∆T d
Rθ = = (5.2)
Pcond λA
Where Pcond is the energy flow per unit time (W), λ is thermal conductivity (Wm-1oC-1),
A is the cross sectional area (m2), ∆T is the temperature difference, T2 – T1 (oC), d is the
length of material (m) and Rθ is defined as the thermal resistance. The units of thermal
resistance (Rθ) is degrees centigrade per watt (oCW-1).Aluminium is typically used in
heatsinks and has a thermal conductivity of approximate 220 Wm-1-oC-1 when 90% pure
(Mohan 1995), (Tillmann Steinbrecher 2005) and (Electrocomp 2003).
(Heat flow direction)

Pcond
T2 A T1
Figure 5.4: Conduction of heat energy per unit time, Pcond, in a material
The type and size of heatsink can be determined from the following equation (Mohan
1995), (Hill 2004), (Wakefield 2005) and (Polyfet 2005):
Tj - Ta
R θs − a = − (R θj −c + R θc − s ) (5.3)
Pcond
The above expression can also be used to determine the temperature at the different
locations in the heat flow path.
The predominant types of cooling strategies for power electronics are air (forced or
natural) or liquid. Air is favoured over liquid due to the complexities of the hardware
required to control and circulate liquid and the extra design effort and cost (Soule 2005).
The use of a fan with the heatsink provides additional cooling but valuable space is used
for the fan, fan mount, and air flow entry and exit. As fan speed increases the airflow at
103
higher speed generates two sources of noise: the moving fan and the friction of the air
moving between the cooling fins.
Heat dissipation is proportional to surface area of the device and the volume of fluid (air
or liquid) moving along the heat transfer surface (Bertrand 2003). A liquid cooled heat
sink can dissipate more heat with considerable less flow volume, maintain better
temperature consistency and generate less local acoustic noise (Soule 2005). Liquid
cooled heatsinks are also unaffected by elevation, which would be an important
consideration in the automotive environment. The surface area required for liquid
cooled heatsinks is much smaller than that required for air-cooled heatsinks.
There are a number of different schemes for implementing liquid cooled heatsinks
(Bertrand 2003) and these includes: embedded or pressed tube in aluminium plate,
expanded copper tube, gun-drilled plate, one-piece castings, direct cooling, bolted cold
plate and brazed cold plate. The embedded copper tube in aluminium cold plate
illustrated in Figure 5.5a (Aavid Thermalloy 2007a) provides the most cost effective
solution in a light package but with a possible lack of uniformity across the surface of
the plate and differences in thermal characteristics in double sided applications. The
brazed cold plate illustrated in Figure 5.5b (Aavid Thermalloy 2007b) provides the
highest performance and the most uniform cooling across the entire plate and can have a
very compact footprint that can facilitate double sided cooling.
Figure 5.5: (a) The embedded copper tube in aluminium cold plate, (b) brazed cold
plate
104
5.2.2 System Description
The implementation of the converter in this application using MOSFETs in parallel
necessitates that the junction temperature of the devices also be equalised so good
thermal coupling is vital. In order to achieve this, the devices must be mounted on a
common heat sink and, if electrical isolation is required, a heat spreader or bus bar can
be used to thermally couple the junctions. An isolation barrier, called an electrically
isolated thermal conductor, can then be placed between the heat spreader and the main
heatsink (Pelly 2004). Illustrations of the side view and the top view for a possible
configuration are provided in Figure 5.6 and Figure 5.7, respectively.
As mentioned previously, it is critical for reliability to maintain the lowest possible

junction temperature. The MOSFET case temperature, Tc, is defined to be equal to
100oC, as typically the temperature would range from 90 oC to 110 oC (International
Rectifier 2004c). The automotive engine compartment ambient temperature can range
from -40 oC to 150 oC (Myers 2003) and (DuPont 2005). Operating in this environment
necessitates locating the converter assembly low down on the engine and using water-
cooling to limit the ambient temperature of the heatsink to approximately 90oC. For
implementation in an automotive environment it is assumed that a double-sided heat
sink is used for the converter, while for implementation in the laboratory a single force
air cooled heatsink is used for each converter phase. This allows for a simplified heat
sink configuration for the prototype converter as the ambient temperature was assumed
to have a maximum value of 45oC to compare with the maximum ambient temperature
experienced by the ISG during cold starting conditions as defined in section 4.1.3.
The final selection on the number of devices to be implemented in parallel was

determined as part of the heat sink analysis. The fewer devices that can be used, the
more compact the single-phase converter can be, thereby minimising stray inductance.
The number of devices in parallel also dictates the size of the heat spreader and
therefore the thermal conductor. The heat sink analysis describes the step-by-step
process used to determine the thermal resistance requirement for the heat sink during
ISG start mode. A spreadsheet program was then used to evaluate the different
combinations of devices number in parallel, heat spreader and thermal conductor size,
case and ambient temperatures, in order to determine the most suitable configuration.
105
Figure 5.6: Side view of air cooled converter mechanical assembly
Q3a Q3b Q1a Q1b
Heat
Thermal conductor
spreader
Q4a Q4b Q2a Q2b
Figure 5.7: Top view of air cooled converter mechanical assembly
106
Figure 5.8: Heat Flow Path for a Single Phase Converter
107
5.2.3 Determination of Rθs-a
In order to determine the thermal resistance requirement for heat sink-to-ambient path,
one needs to calculate the sum of thermal resistances from the junction-to-sink. The heat
flow path from junction-to-ambient for a single-phase converter, Rθj-a, (assuming device
number equals two) is shown in Figure 5.8. As there are parallel paths for heat flow, the
thermal resistance combine like electrical resistors in parallel (Mohan 1995). First, the
thermal resistance from the junction-to-heat spreader is determined for each component
from the sum of the thermal resistance junction-to-case, Rθj-c, and case-to-heat spreader,
Rθc-hs. Then the parallel sum of these thermal resistances is calculated for the selected
number of devices. The total thermal resistance in parallel is the thermal resistance for
the individual component divided by the number of devices in parallel (N) since the
thermal resistances are the same value (Mohan 1995) and (Fowler 1989). Next, this
value of thermal resistance is added to the value of thermal resistance for the heat
spreader-to-thermal conductor, Rθhs-tc, and the thermal conductor to the heat sink, Rθtc-s,
and this yields the thermal resistance junction-to-sink, Rθj-s, for each component group
(e.g. Rθj-s(Q1)). Finally, this step needs to be repeated for the other three component
groups (Q2, Q3 and Q4) and the parallel sum of the thermal resistances for the four
groups of components yields the converter phase thermal resistance junction-to-sink,
Rθj-s(Phase), when using a single heat sink for each phase. The thermal resistance sink-to-
ambient, Rθs-a, is then determined using equation 5.3. The revised equation to determine
the thermal resistance sink-to-ambient, Rθs-a, is:
 Rθ j − c ( Device ) + Rθ c − hs 
T j( max ) − Ta  ( ) + ( Rθ hs − tc + Rθtc − s ) 
Rθ s − a = ( )− N  (5.4)
PT ( SP )  4 
 
where PT(SP) is the total power dissipation of the MOSFETs from the single converter
phase, Ta, is the maximum ambient temperature of 45oC during ISG cold starting
conditions and Tj(max) is the maximum junction temperature experienced by the
MOSFETs (Q1-Q4) during that particular ISG mode of operation (starting, low speed
motoring and high speed generating).
108
The junction temperature, Tj, determined from the following equation.
T j = Tc + (Rθ j − c PT (device) ) (5.5)
where PT(device) is the total power dissipation for each of the MOSFETs devices (Q1-Q4)
from the single converter phase, Rθj-c is the thermal resistance for the TO-220 package
and Tc is the case temperature of 100oC. (Polyfet 2005)
In order to determine the thermal resistance sink-to-ambient, Rθs-a, the values of thermal
resistance had to be determined for the heat spreader and the electrically conductive
thermal conductor. The step-by-step process of calculating the thermal resistance values
under starting mode of operation is illustrated in section 5.2.4 and the results for this
and the other speeds during motoring and generating modes of operation are included in
Table 5.3. The table also includes results for different numbers of devices in parallel for
each component group and the effects of different ambient temperatures on the heat sink
selection process. The calculations were implemented using a spreadsheet program to
automate the heat sink analysis in order to evaluate the effects of different device
numbers in parallel (2 and 4), variations in ambient temperatures (45oC and 90oC),
using alternative types of thermal conductor materials and different sizes of heat
spreader. These calculations were implemented for each ISG mode of operation
(starting, low speed motoring and high speed generating). The specific objective of the
analysis was to determine the optimum heat sink selection (Rθs-a value). The detailed
analysis for each component is provided in Appendix C, while a summary of the results
is provided in Table 5.4.
5.2.3.1 Determination of Rθ for Heat Spreader
The heat spreader is constructed from aluminium (λ = approximately 220 Wm-1-oC-1).

The worse case situation assumes that only two devices are used in parallel and
therefore this defines the smallest size of heat spreader. The MOSFET dimensions were
determined from the device data sheet for a TO-220 package (Fairchild 1999)
(approximately 17mm long by 11mm wide) and the two parallel devices are mounted
side by side on the heat spreader that is 50mm long by 25mm wide. There are four heat
spreaders, one for each component group (Q1, Q2, Q3 and Q4) as illustrated in Figure
5.4. The maximum thickness (d) of the heat spreader is assumed to be 16mm in order to
allow ample material depth for the mounting screws to attach the MOSFETs and to
109
allow sufficient area to attach electrical connections to the heatspreader/bus bar in order
to complete the connections for the full-bridge converter. These dimensions provide the
worse case sizing of the heat spreader to be used in this application. The thermal
resistance of the heat spreader, Rθhs-tc, is:
d
Rθ hs −tc = = 0.058 o CW −1
λA (5.6)
5.2.3.2 Determination of Rθ for the electrically isolated thermal conductor

Elastomers are used for the thermal conductor in this application. They are made from
an elastomer-binding agent, usually silicone, in conjunction with thermally conductive
filler, typically aluminium-oxide or boron nitrate (Electrocomp 2003). The material is
easy to cut to size at room temperature, has high tensile strength and does not require
thermal grease. A list of products and relevant specifications is given in Table 5.1. The
size of the thermal conductor is dependent on the size of heat spreader determined in the
previous section. The worst case sizing assumes that the insulator has the same area as
the heat spreader but the material was extended beyond the edges of the heat spreader
by 5mm in order to provide sufficient insulation. The worst-case thermal resistance,
Rθtc-s, for the Warth K200 thermal conductor product is:
d
Rθ tc − s = = 0.135 o CW −1 (5.7)
λA
The thermal resistance values for each product (Bergquist 2005), (Bergquist 2004) and
(Warth 2004) are given in Table 5.1. The Bergquist Sil-Pad 2000 provides the best
solution (Rθtc-s = 0.058oCW-1) for this application.
-1 o -1 o -1
Vendor Product Construction λ (Wm - C ) d(max) (mm) Rθtc-s ( CW )
Bergquist Sil-Pad 2000 Silicon/fiberglass 3.5 0.254 0.058
Bergquist Sil-Pad K10 Silicon/film 1.3 0.152 0.094
Warth K200 Silicon/fiberglass 1.3 0.22 0.135
Table 5.1: Sources of Thermally Conductive Insulators
110
5.2.3.3 Determination of junction temperature, Tj
The value of the junction temperature for Q1 during starting mode when using the
Fairchild Semiconductor FDP047AN08A0 is
(
T j = Tc + Rθ j − c PT (Q1 ) = ) 127 o C (5.8)
where PT(Q1) = 55W during ISG starting mode, Tc = 100oC and Rθj-c =0.48oCW-1. The
junction temperature determined for each device (Q1 - Q4) is provided in Table 5.2.
Device Q1 Q2 Q3 Q4
o
Tj ( C) 127 113 127 101
PT(device) (W) 55 27 57 2
Table 5.2: Calculated junction temperatures for each MOSFET (Q- Q4)
during ISG starting mode
5.2.4 Step-by-Step Determination of Rθs-a – Starting Mode

The calculation of the thermal resistance sink-to-ambient, Rθs-a, discussed here
illustrates the step-by-step process that is involved to determine the thermal resistance
value under ISG starting mode of operation of the SRM. The value of thermal resistance
sink-to-ambient for the single phase converter is calculated using equation 5.4. The
following are the operating conditions when using the Fairchild Semiconductor
FDP047AN08A0 (Fairchild 2004):
Rθj-c (oCW-1) Rθc-hs (oCW-1) Rθhs-tc (oCW-1) Rθtc-s (oCW-1) Tj(max) (oC) Tc (oC) Ta (oC) N PT(SP) (W)
0.48 0.1 0.058 0.058 127 100 45 2 283
Table 5.3: Operating conditions
Rθj-c is the thermal resistance of the TO-220 package provided in the device data sheet
(Fairchild 1999), Rθc-hs is the thermal resistance of the heat sink compound (Wakefield
2008), Rθhs-tc is the thermal resistance of the heat spreader determined in section 5.2.3.1,
Rθtc-s is the thermal resistance of the thermal conductor determined in section 5.2.3.2, Tc
is the maximum case temperature of the MOSFET, Ta is the maximum ambient
temperature, and Tj(max) is the maximum junction temperature experienced by the
MOSFETs (Q1-Q4) during ISG starting mode, N is the number of devices connected in
111
parallel and PT(SP) is the total power dissipation of the single phase converter in ISG
starting mode. The thermal resistance sink-to-ambient, Rθs-a, is:
 Rθ j −c ( Device ) + Rθ c− hs 
T j (max) − Ta ( ) + ( Rθhs −tc + Rθtc − s ) 
Rθ s − a = ( )− N  (5.9)
PT ( SP )  4 
 
= 0.19 o CW −1
The thermal resistance sink-to-ambient, Rθs-a, for the different initial conditions were
calculated in the automated spreadsheet program and a summary of the results is shown
in Table 5.4, while the detailed results for the thermal analysis calculations are provided
in Appendix C.
Thermal Analysis Starting Low speed High speed

motoring generating
Tc = 100oC 200RPM 1000RPM 4500RPM
2 Devices @Ta=45oC 0.19 0.17 0.48
4 Devices @Ta=45oC 0.23 0.20 0.88
o
2 Devices @Ta=90 C 0.031 0.030 0.074
4 Devices @Ta=90oC 0.032 0.028 0.14
2 Devices @Ta=45oC 0.18 0.17 0.41
4 Devices @Ta=45oC 0.24 0.21 0.77
o
2 Devices @Ta=90 C 0.026 0.026 0.061
4 Devices @Ta=90oC 0.032 0.028 0.12
Phillips BUK7506-76B
2 Devices @Ta=45oC 0.21 0.19 0.46
o
4 Devices @Ta=45 C 0.27 0.24 0.86
2 Devices @Ta=90oC 0.031 0.029 0.070
4 Devices @Ta=90oC 0.038 0.033 0.14
Table 5.4: Thermal analysis to determine Rθs-a (oCW-1)
From Table 5.4, it can be seen that the Philips BUK7506-76B provides the most
suitable solution for this application but the Fairchild Semiconductor FDP047AN08A0
and FDP060AN08A0 are also acceptable. For the FDP047AN08A0, the worst-case
thermal resistance sink-to-ambient is 0.03 oCW-1 at Ta = 90oC and 0.17 oCW-1 at Ta =
45oC. It is possible to use a forced air-cooled heatsink for implementation in a
laboratory environment as Ta < 45oC. A suitable heatsink would be the HS Marston
112
890SP-01000-A-100 (HS Marston 2001), which provides a thermal resistance of
0.12oCW-1 when using an airflow rate of 22.2ls-1.
A water-cooled configuration would be used in the automotive environment due to the

high ambient air temperature of 100oC - 150oC in the engine compartment, while it is
assumed that the cooling fluid has a maximum Ta < 90oC.. A possible configuration is
an aluminium plate with embedded copper tubing where each converter phase would be
attached to a plate. A possible candidate for the heatsink is the Aavid Thermolly six
pass model 416101U00000G liquid cooled cold plate (Aavid Thermalloy 2007a). The
plate has a thermal resistance of 0.012oCW-1, with a coolant flow rate of 0.06 ls-1. The
use of a brazed cold plate would provide a more significant improvement in thermal
resistance and more uniform cooling. It would also be more applicable to a mass
production environment, as custom plates can be designed for this application. An
illustration of a possible water cooled heatsink layout is given in Figure 5.9.
It can be seen in the detailed analysis provided in Appendix C that when using 2 devices
in parallel that the Tj ≈ 130oC during starting and low speed motoring while the Tj <
115oC when using 4 devices in parallel. The heat sink requirement is almost the same
for 2 or 4 devices in parallel under starting and lower speed motoring conditions but is
significantly lower for 4 devices under high speed generating conditions. Using 2
devices in parallel reduces cost and complexity but leads to a compromise of reduced
reliability for the components due to the higher junction temperatures experienced by
the devices.
113
Figure 5.9: Side view of water cooled heat sink
114
5.3 Gate Drive Design
The full-bridge synchronous rectifier with two MOSFETs mounted in parallel was used
in this application and is illustrated in Figure 5.10. It utilizes a high side switch
configuration (with the MOSFET drain connected to the voltage supply and the
MOSFETs source connected to the load) and this places specific requirements on the
gate drive circuit. To fully turn on the high side switch, the gate voltage must be higher
than the voltage supply, which is frequently the highest voltage in the system. The gate
voltage must be controllable from logic circuits which are normally referenced to
ground. The control signals need to be level shifted to the source of the high side
device, which in most cases swings between positive and negative voltage supplies. The
power absorbed by the gate drive circuitry should not significantly affect the overall
efficiency of the converter.
Figure 5.10: Full-Bridge Synchronous Rectifier parallel configuration
When implementing MOSFETs in parallel, large source currents must be provided by

the gate drive circuit to enable rapid MOSFET turn-on and also provide very low
impedance for fast device turn-off in order to negate circuit imbalance in the circuit.
There are numerous techniques for implementing high side gate drive and some of these
include floating gate drive supply, pulse transformer, charge pump, boot strip, carrier
drive and gate drive integrated circuit (or gate driver IC). A brief review of these
techniques to implement these gate drive configurations is summarized in the following
section. The gate driver IC, or more simply called the ‘gate driver’, was selected for this
application as it integrated most of the capabilities (level shifting of the logic signals,
gate voltage higher than the supply voltage and operation in a wide range of duty cycles
and switching frequencies) for high side gate drive operation in a compact package).
115
The gate driver can provide large peak output current that is acceptable for most
applications. However, when driving MOSFETs in parallel a current buffer may be
required to facilitate the necessary peak current levels. The implementation of a current
buffer is determined by the peak gate drive current requirement of the parallel
MOSFETs configuration.
A dead time period or blanking time needs to be provided between switching states of a
high and low side switch (e.g. Q1 and Q2 in Figure 5.10) in order to prevent current
cross conduction or shoot through condition. The gate driver provides almost no dead
time period (Clemente 1990), so a resistor diode network was placed on the gate driver
output to implement the additional dead time requirement.
Isolation of the logic control signals is required to protect the low voltage control logic
circuitry from the power switching circuits in the event of a fault condition. Isolation in
this application is achieved using optical isolators. The output of the optical isolators
and the gate drivers for all the switching groups (Q1 –Q4) for all converter phases would
be powered by a single isolated power supply.
5.3.1 Gate Drive Configuration

As noted previously, there are a number of techniques for implementing high side gate
drive and some of these include floating gate drive supply, pulse transformer, charge
pump, boot strip, carrier drive and gate drive integrated circuit or ‘gate driver’. An
outline of each technique is provided in Table 5.5 (Clemente 1990). The main limitation
of the floating gate supply is the requirement for an isolated supply for each high side
MOSFET which would be prohibitive in this application requiring multiple high side
MOSFETs. The pulse transformer is a simple and inexpensive solution for high side
drive but has limited switching frequency range. The transformer is designed for
operation in a specific frequency range whereas the MOSFETs in this application
operate at frequencies from 20Hz up to 25 kHz. Slow turn-on time’s factor against
selecting charge pump for this application. While simple and inexpensive, limitations in
the range of duty cycles provided using bootstrap capacitor circuit negate this
configurations selection. The gate driver was selected for this application as it provides
level shifting, gate voltage higher than the supply voltage and operation in a wide range
of duty cycles and switching frequencies in a single compact package.
116
Method Basic Circuit Main Features
Vs
Full gate control for infinite periods of time.

Floating Gate
One isolated supply required for each high side
Floating Supply Drive MOSFET => significant cost.
Level shifting required for ground referenced
Gate Drive signal => must sustain full voltage and switch fast
Supply Level Shifter Load
or with low power consumption
Or
Opto isolator Low side Opto isolators expensive with limited bandwidth
device
and noise sensitive
Vs
Simple and cost effective but has limitations.

Operation over a wide duty cycles requires
additional circuitry.
Pulse
Transformer size increases significantly as
Transformer frequency decreases.
Load
or Significant parasitics create less than ideal
Low side
device
operation with fast switching waveforms.
Vs
Can be used to generate a gate voltage higher than
the supply voltage (Vs) controlled by a level
shifter or to “pump” the gate when device is on.
Charge Requires level shifting circuitry and turn on times
Load
Pump or
too long for fast switching applications.
Low side Gate can be turned on for an infinite time period.
Oscillator device
Inefficiencies in voltage pumping circuit may
require more that two stages of pumping
Vs
Simple and inexpensive to implement but has

Gate same limitations as pulse transformer, in that duty
Drive
cycle and on-time are constrained by bootstrap
Bootstrap capacitor refresh rate.
Load Power dissipation can be significant when
or
Level
Shifter Low side
capacitor charged from high voltage supply.
device Requires level shifting circuitry
Vs
Gives full gate control for infinite period of time.

Carrier
Improved switching performance requires
Drive Load
or additional circuitry.
Low side
Stop
device
Oscillator
Vs Provides level shifting referenced to ground

signal in a compact package that requires only a
Gate few addition components
Floating
drive integrated circuit Supply Gate Is able to sustain full voltage applied to MOSFET
driver
and provide very fast switching speeds with low
(IC) Control
with
level power consumption.
Logic
or with
shifter Load
or Operate in bootstrap mode or with floating
Optical
Gate driver isolation
Low side
device
supply.
It can operate over a wide range of frequencies
and duty cycles.
Table 5.5: Gate drive configurations
The specific device selected for use in this application is the International Rectifier
IR2110 gate driver (International Rectifier 2005b) is illustrated in Figure 5.11
(functional block diagram) and Figure 5.12 (typical connection diagram). This device is
a high voltage and high speed gate driver with independent high and low side referenced
output channels. The floating high side channel is designed for bootstrap operation and
117
can drive a MOSFET that operates up to 600V. The input logic is TTL/CMOS
compatible down to 3.3V logic. The signals from the input logic are coupled to the
individual channels through high noise immunity level translators. The high side
channel is built into what is defined as an “isolation tub” that is capable of floating from
600V to -5V with respect to ground. This ‘tub’ floats at a potential of VS, which is
established by the voltage applied to VB. The gate charge for the high side MOSFET is
provided by the bootstrap capacitor, CB, which is charged by the VCC supply through the
bootstrap diode during the time when the device is off (assuming that VS swings to
ground during that time, which generally is always the case). As the capacitor is charged
from a low voltage source the power consumed to drive the gate is small.
Figure 5.11: Functional block diagram with lead definitions for the IR2110 gate driver
Figure 5.12: Typical connection diagram when using IR2110 gate driver
118
A number of issues need to be addressed when using the IR2110 gate driver including,
the selection of the bootstrap components (capacitor, CB, and diode, DB) and minimising
negative spikes at the high side floating supply pin, VS. Negative spikes are minimised
by reducing parasitic inductances through careful layout, discussed in section 5.7.2,
improved local decoupling by using low ESR capacitors in parallel with CB and
capacitor at low side supply (VCC to COM), selecting the gate resistor value to minimise
the amplitude to the negative spike.
5.3.2 Determination of Peak Gate Drive Current

Power MOSFETs have a large gate capacitance that must be charged to achieve device
turn on. The gate drive circuit must provide sufficient current to charge the equivalent
gate capacitance in the required time interval. The total gate charge is the key
specification in determining gate drive requirements as it brings together all device
charge values and accounts for the “Miller effect” [(Dunn 2003), (Dunn 2004),
(Supertex 2001), (Barkondarian 2000) and (Andreycak 1999)].
The process of charging the gate is defined in three stages as illustrated in Figure 5.13
[(Dunn 2003) and (Barkondarian 2000)]. The first stage is the charging of the gate-
source capacitance, CGS, the gate-drain capacitance, CGD, or the Miller capacitance is
also charging but it is very low. Once CGS is charged up to the gate threshold voltage,
the device begins to turn on and the current ramps up to its full value. On reaching the
full current value, the drain-to-source voltage begins to collapse. It is at this point that
the gate voltage levels out due to the Miller capacitance being charged as the drain
voltage falls. When the drain voltage has fallen to its final level, the gate capacitance
(both CGS and CGD) is charged the rest of the way to the gate drive voltage.
Figure 5.13: (a) Gate capacitance model for MOSFET and (b) gate charge waveform
119
The gate charge value simplifies the calculation of the current required from the gate
drive circuit to turn on and turn off the MOSFET, as illustrated in equation 5.10 [(Dunn
2003), (Dunn 2004), (Barkondarian 2000), (Hussain 2002), (Dorf 1999), (Winters 2002)
and (Andreycak 1999)].
QG(tot)
IG = (5.10)
∆t
where IG is the gate drive current, QG(tot) is the total gate charge for the MOSFET and ∆t
is either the turn on or rise time, tr, or the turn off or fall time, tf, for the MOSFET,
which are specified in the device data sheet. The rise and fall times and the total gate
charge specifications for the output devices determine the amount of gate drive current
to source or sink. The relevant specifications for all suitable MOSFETs for the
converter are provided in Table 5.5.
The power dissipated by the IR2110 gate driver is typically < 500mW (Clemente 1990)
and this would have a negligible effect on the overall converter efficiency. The power
dissipated to charge the device gate capacitance Qg(tot) whenever a capacitor is charged
or discharged through a resistor is defined by equation 5.11, [(Dunn 2003), (Dunn
2004), (Supertex 2001) and (Andreycak 1999)].
PG = Qg(tot) VG f (5.11)
where Qg(tot) is the total gate charge, VG is the gate voltage and f is the switching
frequency of 25kHz. The step by step calculation of the peak source and sink currents
for the Fairchild FDP047AN08A0 are provided in detail in the following section and the
results for all the devices are included in the Table 5.5. The gate voltage of 12V is
sufficient, as a greater value has little effect on reducing the on-resistance of the
MOSFET (Andreycak 1999).The peak source current and peak sink current are
determined as follows:
IG(turn on) = 3.14A
IG(turn off) = 6.13A
120
The power dissipated to charge the device gate and consumed by the gate drive circuitry
are determined as follows:
PG = 41.4mW
where Qg(tot) is the total gate charge provided in Table 5.5, VG is the gate voltage of 12V
and f is the switching frequency of 25kHz.
Gate Drive Data Results

V G = 12V tr tf Q G(tot) I G (turn on) I G (turn off) PG
Device No. = 2 (ns) (ns) (nC) (A) (A) (mW)
Fairchild
FDP047AN08A0 88 45 138 3.14 6.13 41.4
FDP060AN08A0 79 38 95 2.41 5 28.5
Philips
BUK7506-76B 56 48 91 3.25 3.79 27.3
Table 5.5: MOSFET data and gate drive calculations
As can be seen from the results in Table 5.5, the current requirement for the MOSFETs
in parallel is significantly more than the IR2110 gate driver can provide so a high
current buffer was a requirement in this application.
5.3.3 High Current Buffer

The high current gate drive buffer circuit used in this application is described in
[(Clemente 1990) and (Kiraly 2004)] and is illustrated in Figure 5.14.
Figure 5.14: High current gate drive buffer circuit
This circuit can deliver 8A peak output current, has high input impedance and low
output impedance. When the input changes states the resistor, R1, limits the current
121
through Q1 and Q2 when both transistors are on for a few ns. The output stage formed
by Q3 and Q4 can be sized to suit the peak current demand and also corrects the logic
flow from the gate drive input to output. The truth table for the input, output and
transistor states is provided in Table 5.6.
Input Q1 Q2 Q3 Q4 Output
H Off On On Off H
L On Off Off On L
Table 5.6: Truth table for gate drive buffer
There is a delay in the turn on, due to the RC time constant formed by R1 and the input
capacitance of the output transistor, which is beneficial in providing a dead time in the
overall converter operation. The configuration as it is implemented in this application is
shown in Figure 5.15.
Figure 5.15: High and low side gate drive using the IR2110 gate driver and high
current buffer circuits
5.3.4 Determination of peak current for the buffer gate drive circuit
The calculation of the peak input gate drive current for the buffer circuit determines the
gate driver selection. When 12V (VG) is applied to the buffer input, the p-channel device
(Q1) turns off, the n-channel device (Q2) turns on and Q3 turns on and current flows into
122
the gate and thereby charges the gate capacitance. Therefore peak current requirement is
determined using the combined gate charge values for Q1 and Q2, and using either the
fall time of Q1 or the rise time of Q2, which ever is greater. The power dissipation is
determined from the following device data as illustrated in Table 5.7 [(Fairchild 2002a,
2004) and (Philips 2002)].
MOSFET QG(tot) (nC) tr (ns) tf (ns)

Philips BUK 7506-76B 91 56 48
Fairchild FDP047AN08A0 138 88 45
Fairchild FDP060AN08A0 95 79 38
IRF IRFD9110 8.7 27 17
IRF IRFD110 8.3 16 9.4
Table 5.7: MOSFET specification data
The peak source and sink current are determined using equation 5.10, while the power
dissipated to charge the gate drive capacitance is determined by equation 5.11. The
results are given in Table 5.8.
IG(turn on) (A) IG(turn off) (A) PG (mW)

1 0.63 5.1
Table 5.8: Buffer circuit calculations
The gate driver circuit must source up to 1A and it must be capable of sinking 0.63A.
The IR2110 (International Rectifier 2005b) can sink and source 2A, so is suitable for
use in this application. The total power dissipation for the gate driver is ≈500mW
(IR2110 <500mW and buffer <5mW) and therefore has negligible contribution to the
overall converter power losses.
5.3.5 Bootstrap Components

As noted in section 5.3.1, the voltage between the Vb and Vs pins (Vbs) of the IR2110
gate driver supplies the high side gate driver circuitry and should remain between 10-
15V in order to fully turn on the MOSFET. This requires the selection of the
appropriate bootstrap capacitor and diode values.
5.3.5.1 Calculating the bootstrap capacitor value

The minimum charge, Qbs, that needs to be supplied by the bootstrap capacitor, Cbs, is
determined by the equation (Adams 2004):
123
I qbs (max) I Cbs ( leak )
Qbs = 2Q g + + Q LS + (5.12)
f f
where Qg is the gate charge of high side MOSFETs, f is the frequency of operation,
ICbs(leak) is the bootstrap capacitor leakage current and QLS is the level shift charge
required per cycle which equals 5nC for the IR2110 (International Rectifier 2005b). In
order to minimise the amount of ripple on the high side gate voltage (VBS), the
minimum charge in the bootstrap was multiplied by a factor of 15 (rule of thumb)
(Adams 2004). The capacitor value is determined by the equation (Adams 2004):
 2Qbs 
C bs ≥ 15  (5.13)
VCC − V f − VLS − VMin 
where Vf is the forward voltage drop across the bootstrap diode, VLS is the voltage drop
across the low side MOSFET (or load for the high side driver) and VMin is the minimum
voltage between VB and VS of the IR2110 gate driver.
The calculation for the bootstrap components is implemented step-by-step for the
Fairchild FDP047AN08A0 and the results are tabulated in Table 5.9. The gate charge
for the high side MOSFET was determined from the combination of the high side buffer
circuit and the two high side MOSFETs of the converter. The total gate charge values
for the devices are given in Table 5.7. On high side circuit activation, the charge (Qg)
must be supplied to the gates of Q2 and Q3 of the buffer circuit and the two parallel high
side MOSFETs of the converter and on deactivation, charge is only supplied to Q1 and
Q 4.
Q g (nC) I Cbs (leak) (µA) Q LS (nC)1 f (kHz) V f (V) V Min (V) V LS (V)2 V CC (V) I qbs(max) (nA)
293 3 5 25 0.875 13 0.5 15 230
1
IR2110
2
assumes a worst case of 50% duty cycle
Table 5.9: Bootstrap capacitor operating conditions
The minimum charge that needs to be supplied by the bootstrap capacitor is determined
from equation 5.12.
Qbs = 590nC
124
The bootstrap capacitor value is determined from equation 5.13. A 47µF 100V capacitor
was used in order to provide sufficient margin.
C bs ≥ 28µF
5.3.5.2 Selecting the Bootstrap Diode

The diode, Dbs, must be able to block the full supply voltage of 42V when the high side
device is switched on. It must have a fast recovery time to minimise the amount of
charge that is fed back to the voltage supply, VCC, of the IR2110 gate driver. The high
temperature reverse leakage current is important if the capacitor has to store charge for
long periods of time. The current rating for the diode is the product of the charge, Qbs,
and the switching frequency, f, (Adams 2004). The relevant component data and current
calculation data is provided in Table 5.10. The diode that meets these requirements is
the MUR120 (ON Semiconductor 2005). It has a worst-case forward voltage, Vf, drop of
0.875V.
If = Qbs f (5.14)
If = 15mA
V rrm (V) t rr(max) (ns) I f (A) V f (V) f (kHz)

42 100 15 0.875 25
Table 5.10: Bootstrap diode data
5.3.5.3 Gate resistor selection

Increasing the value of the series gate resistance rapidly reduces the amplitude of the
negative spikes due to stray inductance (e.g. track length, component lead length above
the circuit board) at the high side return (Vs). The value of the series gate resistor
reduces the switching speed and therefore the value of dv/dt and is based on the graph
of turn-off time versus series gate resistance shown in Figure 5.16 (Clemente 1990).
Selecting a resistor value just on the “knee” provides a good trade-off between the spike
amplitude and the turn-off speed. The gate resistor dampens any oscillations in the gate
circuit caused by a resonant tank formed by parasitic inductances and capacitances
inherent in the MOSFET and contributed by circuit connections (Fairchild
Semiconductor 1998). A 22Ω gate resistance was selected for the input to the gate drive
125
buffer circuit, while a 4.7Ω gate resistor was placed on each MOSFET in a parallel
configuration and was located as close as physically possible to the gate of the device.
Figure 5.16: Gate resistance vs. negative voltage spike amplitude and turn-off time
5.3.6 Blanking Time Delay

A time delay needed to be provided between the switching states of the high and low
side switches in order to prevent a current shoot through condition. This time delay is
defined as the blanking time. There are a number of elements contributing to the
blanking time. There is a 25ns difference between the turn-on and turn-off propagation
delay times of the IR2110 gate driver (International Rectifier 2005b), to ensure no
conduction overlap of the converter power devices. A resistor diode network was
placed at the input to the buffer circuit to provide additional margin of safety as
illustrated in Figure 5.17. The diode would be reversed biased at turn-on thereby
generating a delay corresponding to the RC time constant of the gate resistor and the
input capacitance of the buffer circuit. The diode would be forward bias at turn-off to
provide a low impedance path for fast turn-off. A 1N4148 (Fairchild 2007) diode was
used for this application.
5.3.7 Electrical isolation

Electrical isolation for the control logic signal for the gate drive circuit was provided
using HPCL-263A optocouplers (Agilent Technologies 1999a). This device provides
high dv/dt rating up to 1000V/µs and can operate at speeds up to 10 Mbit s-1. The
implementation of this device required an isolated power supply for the optocoupler
output and the gate drive circuit. The optocouplers were mounted on an interface board
or the microcontroller board to further reduce susceptibility to common-mode noise
generated by the full-bridge converter circuit (Agilent Technologies 1999b).
126
Figure 5.17: Full Bridge Converter
127
5.4 DC Source Capacitor Analysis
Essentially during magnetization and demagnetization, the full-bridge circuit mimics
the operation of a boost converter, while during freewheeling mode, it operates like a
buck converter during freewheeling operation. The dc source needs to have low internal
impedance for the instantaneous current and a large capacitor can provide this low
impedance path (Mohan 1995). It provides the regulated dc connection between the dc
source and the inverter (Krein 1998) and (Lander 1981). Selecting the dc source
capacitor requires knowledge of the general capacitor parameters, the specific
application requirements in order to determine the correct value for the application.
5.4.1 Capacitor Parameters

Capacitors are selected based on the required capacitance, the rated operating voltage,
the RMS current, and frequency, (Mohan 1995), (Vishay 2004), (Nichicon 2004) and
(BHC 2004). Aluminium electrolytic capacitors provide large capacitance per unit
volume and high ripple current capability (Mohan 1995) and (BHC 2004). The rated
voltage is the peak value of pulsed voltage that is applied continuously to the capacitor.
Any pulsating voltage across a capacitor yields an alternating current flowing through
the capacitor and the RMS value of this alternating current is called the ripple current.
The heat generated by the ripple current is an important factor in determining the useful
life of the capacitor and higher frequency operation yields greater reactive power
dissipation and temperature rise in the capacitor (Vishay 2004).
5.4.2 Specific Application Requirements

DC source capacitors supply high currents when there are peak requirements, so they
tend to be abruptly charged and discharged and they need to have a high thermal rating
to handle the relatively high reactive power dissipation (Siemens Matsushita 2004).
Under the proposed 42-Volt electrical system, electrical components, including
capacitors, must withstand higher operating temperatures of 150oC and above (Epcos
2003). The requirement for large capacitance values can necessitate that the capacitors
be connected in parallel to achieve the correct value. Individual protection may be
required against sudden discharge of the entire capacitor bank into a defective capacitor
(Vishay 2004) and (BHC 2004). It is also important that there is a balance of the current
distribution in all capacitor branches (Nichicon 2004) and (Power Designers 2004).
128
5.4.3 Determining Capacitance Value
The percentage voltage ripple on the dc source capacitor, CS, determines the actual
capacitor value. High speed generating defines the worst case cycle of operation for the
voltage ripple as the switching period is determined by the speed of the machine. It has
a minimum value of 2.2ms at 4500rpm. It is assumed that the capacitor voltage is
maintained at 42V maximum and that excess current is directed to the battery voltage
supply by QG as illustrated in Figure 5.18. All the ripple current through Q4 flows
through the capacitor. The shaded area in Figure 5.19 represents the charge ∆Q.
Therefore the peak-to-peak voltage ripple is represented by equation 5.15, which when
rearranged allows for the calculation of the capacitor as illustrated in equation 5.16
(Hart 1997).
∆Q I S DTs VS DTs
∆VS = = = (5.15)
CS CS RC
VS DTs
CS = (5.16)
∆VS R
The peak-to-peak voltage ripple in this application is assumed to be the dip in the dc
source voltage, VS. The percentage ripple, ∆VS /VS, on the dc source capacitor is defined
to be 10%, the duty cycle, D, for the magnetization period during high speed generating
is 40%, and the period, Ts, is 2.2ms. The resistance value, R, is determine from the
series resistance, ESR, of the electrolytic capacitor (typically > 25mΩ) (Epcos 2006),
the phase winding resistance of the SRM (typically 0.8mΩ) (Motion System Tech
2002), and the on-resistance of the MOSFETs (typically 4mΩ) (Fairchild 2004). Using
the above data, the value of dc source capacitor is determined to be 0.3F. This is quite a
large capacitor value and would add excessive cost to the overall system. As the
capacitor requires a voltage rating of 100Vdc (double the maximum power supply
value), the most cost effective device would be the 10,000uF capacitor. In order to
implement the converter with this device, the magnetisation duty cycles needed to be
limited at each speed of operation. Table 5.11 illustrates the magnetisation duty cycles
and percentage voltage ripple values for the various generating speeds of operation
when using this capacitor value. Ultimately, the capacitor selection is a compromise
between cost and acceptable performance limits.
129
Figure 5.18: Full-Bridge Converter with charging circuit
Lmax
Lmin
Rotor
θ1 θon θ2 θ3 θoff θ4 θext Angle
IQ3
& ∆Q
IQ4
IQ1
&
∆Q
IQ2
IQG
VS
Figure 5.19: Voltage ripple on the dc source capacitor, CS
130
∆V S /V S (%) D (%) T s (ms) Speed (rpm)
5 25 0.04 2000
10 1 2.22 4500
20 3 2.22 4500
50 7 2.22 4500
10 2 1.67 6000
20 4 1.67 6000
50 10 1.67 6000
Table 5.11: Magnetization duty cycles and percentage voltage ripple values at various
generating speeds of operation when using a 10,000uF capacitor
5.5 Bus Bar Analysis

Two types of power distribution technologies are used in modern equipment: wiring
harness and bus bars. Wiring harnesses are suitable for low current low volume
applications while bus bars are more suitable to high current that require low inductance
(Hill Technical Sales 2004). Large di/dt and leakage inductances can be a significant
problem in the power supply leads of circuits. When stray inductance are reduced in the
leads so to are the transient voltage spikes and noise that lead to significant EMI and
parasitic oscillations (Mohan 1995), (Hanley 2003) and (Hill Technical Sales 2004).
Bus bars have many advantages and with the correct layout out can provide an ideal
solution for high current applications.
5.5.1 Bus bar implementation

The use of bus bars allows for the reduction in conductor losses, reduce stray inductance
and simplify power bus assembly and routing. Bus bars provide a greater surface area so
can handle more current with reduced losses (Dossert 2004) and (Hanley 2003). They
yield an increase in circuit reliability due to fewer connections, as would be the case
with cabling (Hanley 2003). Rectangular bus-bars reduce the self-inductance by ⅓ to ½
when compared to round conductors of equal length and cross sectional area (Hanley
2003) and (Hill Technical Sales 2004). A laminated structure helps to control both self
and mutual inductance (Hill Technical Sales 2004). It consists of interleaving layers of
ground planes and voltage layers of different polarities. Bus-bars can simplify routing of
power conductors and can reduce probability of errors in assembly due to their rigid
structure (Hanley 2003). Conductors can be made from a variety of metals, including
copper, nickel, gold or aluminium, but cost and weight are the deciding factors in an
application (Mohan 1995), (Hanley 2003) and (Eldre 2004). While more aluminium
may be needed in a conductor than copper, 1.66 times more cross sectional area is
required to conduct the same amount of current, it has 55% less weight when compared
131
with copper and is less expensive (Hanley 2003) and (States Department of the Interior
2000).
5.5.2 General bus bar structure

The structure of the bus bar is critical for the reduction of stray inductance, elimination
of noise and arcing, improving current distribution in parallel capacitors and reducing
the need for snubber circuits for the power devices. A laminated structure with large
plates in a sandwich with thin insulators is the most space efficient bus bar structure
(Mohan 1995), (Power Designers 2004), (Hanley 2003) and (Eldre 2004). The large
plates create the interleaving layers which minimise mutual inductances (Hill Technical
Sales 2004). Traditionally bare bus bars do not suppress noise, are bulky, have potential
to arc between the plates and are costly to assemble (Eldre 2004). The laminated
structure reduces noise due to the interleaving layers (Hanley 2003). Creepage and
arcing is eliminated by increasing the distance the insulator protrudes past the plates
(Mohan 1995). Laminated bus bars allow for equal current distribution to parallel
capacitors (Mohan 1995) and (Power Designers 2004). The structure must be as
compact as possible as long interconnection distances between switching devices
increases bus inductance, which requires complex snubber circuits for protection,
adding to system cost (Hill Technical Sales 2004).
5.6 Current Sensors

In order to implement precise current regulation in a power converter design, the choice
of current sensor is critical. When implementing current regulation, consideration needs
to be given to the type of current being measured, the operating range of the current, the
required output signal, measurement accuracy, available power supply and the working
voltage. Additionally, dynamic operating parameters need to be considered, which
includes frequency range, di/dt and dv/dt. Finally, consideration must be given to the
environmental variables – temperature, vibration and shock, and the presence of
external fields. Three current sensing schemes – shunt resistor, current transformer and
Hall effect – were evaluated for suitability in this particular application.
5.6.1 Shunt Resistor

With this scheme, the voltage drop across the shunt resistor is proportional to the
current flowing in the circuit (Dickinson 2002) and additional circuitry is required to
improve signal quality (Lepkowski 2003). This measurement can be implemented in
132
either high side or low side configuration as illustrated in Figure 5.20. The high side
measurement is the preferred method, as it is less intrusive as long as the sense resistor
is small. The low side measurement is easier to implement but disrupts the ground path
of the load and this can cause noise problems in the system. In either case, the load
current, IL, is determined using the equation (Lepkowski 2003).
VM
IL = (5.17)
RM
The shunt resistor measurement scheme is inexpensive to implement but it requires
additional circuitry to improve signal quality and it has no electrical isolation (Mohan
1995) and (Lepkowski 2003). This scheme is impractical for use in circuits where the
current measurements are above 20A and has poor accuracy over a wide temperature
range (Dickinson 2002) and (Lepkowski 2003).
Figure 5.20: Shunt resistor measurement schemes, (a) High-side and (b) low-side
5.6.2 Current Transformer

These sensors use the principle of transformers where the ratio of the primary current,
IP, to the secondary current, IS, is a function of the turns ratio, N, as illustrated in Figure
5.21 and described using equation 5.18 (Lepkowski 2003). They are simple to
implement and have no drive circuits. The scheme provides galvanic isolation but an
AC signal is required to prevent the transformer from saturating (Mohan 1995),
(Dickinson 2002) and (Lepkowski 2003).
IP
IS = (5.18)
N
133
Figure 5.21: Current sensing transformer
5.6.3 Hall Effect Transducer

Hall effect is based on the principle that a voltage, VH, is developed when current flows,
IC, in a direction perpendicular to a magnetic field, B, as illustrated in Figure 5.22
(Melexis 1997) and (Lepkowski 2003).
Figure 5.22: The Hall Effect principle
There are two main techniques: open loop and closed loop (Dickinson 2002) and
(Lepkowski 2003). The open loop scheme amplifies the Hall generator voltage to
provide an output signal as illustrated in Figure 5.23a. Closed loop transducers uses the
output voltage of the Hall element to develop a secondary or compensation current in a
secondary coil as illustrated in Figure 5.23b. This secondary current in conjunction with
the secondary ratio generates a magnetic flux that is equal in magnitude but opposite in
polarity to the flux created in the primary current to yield a total flux of zero
(Lepkowski 2003).
Both techniques are suitable for measurement of dc, ac and complex current waveforms
in power applications that need galvanic isolation (Melexis 1997) and (LEM
Components 2004). The two schemes provide a robust solution to environmental
disturbances, including vibration, moisture and dirt or oil film (Gilbert 2002). The
closed loop design has the advantage of very good accuracy and linearity, low gain drift,
134
wide bandwidth (0Hz - 200 kHz), and fast response time, but the current measurement
is limited due to the finite compensation current (Mohan 1995), (Dickinson 2002),
(Melexis 1997), (Lepkowski 2003) and (LEM Components 2004). Closed loop
transducers have very good reaction times (less than 1µS) and are able to measure di/dt
of 50 – 500AµS-1 and greater, which makes them suitable for short circuit protection
(LEM Components 2004).
Figure 5.23: (a) Open loop Hall Effect transducer, (b) closed loop Hall Effect
transducer
Recently, a hybrid scheme (LEM Components 2004) has being developed that works
like an open loop Hall Effect transducer up to 10kHz and like a current transformer at
higher frequencies. Both Hall effect and current transformer signals are electronically
added to form a common output signal as illustrated in Figure 5.24.
Figure 5.24: Combined Open loop and closed loop Hall effect transducer
This scheme is similar to closed loop Hall effect transducers but only requires a small
secondary power supply (LEM Components 2004). At higher frequencies, the current
transformer yields high bandwidth and fast response. Open loop at low frequencies
135
implies that there are gain and offset drifts with temperature and a moderate level of
accuracy. Low frequency open loop operation requires more expensive construction
when compared with the closed loop scheme (LEM Components 2004). Bandwidth,
response time and di/dt behaviour are similar to closed loop transducers although it is
slightly less efficient at high frequencies (LEM Components 2004).
5.6.4 Sensor Selection and Implementation

The closed loop Hall effect sensor provides the most suitable option for this application
for the reason that the sensor provides electrical isolation, has large bandwidth, fast
response time and is therefore able to provide the response times necessary for current
overload protection. The specific device selected for this application is the LA 100-
P/SP13 (LEM Components 2005). This device allows the bus bar connection to the
motor phase terminal to pass through the sensor aperture as illustrated in Figure 5.25(a).
In order to implement the current sensor in this application, the following items had to
be addressed: the layout of the power conductors and the sensor output signals, the
power supply requirements, the determination of the measurement resistor and finally
any signal conditioning requirements for the transducer output signals. The specific
layout issues are discussed in section 5.7.6. The closed loop sensor requires a bipolar
power supply (LEM Components 2005) ±15V power supply provides the power
requirement.
The appropriate measurement resistor, RM is determined based on the range provided in

the data sheet, and depends on the maximum ambient temperature. The maximum
measured voltage, VM, is determined using equations 5.19 and 5.20
IS = IP(NP/NS) (5.19)
VM = RMIS (5.20)
where IS is the secondary current, IP is the primary peak current, NP is the number of
primary turns and NS is the number of secondary turns (LEM Components 2004).The
connections to the current sensor are implemented as shown in Figure 5.25b (LEM
Components 2005).
136
Current
direction
LA 100 – P/SP13
M - +
Sensor
IS
aperture
RM
OV - +
(a) (b)
Figure 5.25: (a) Current sensor physical layout, (b) current sensor connections
5.7 Circuit Layout Considerations
5.7.1 Full Bridge Circuit

In a typical bridge circuit, the critical stray inductances that need to be reduced are in
the high current path of the circuit. Stray inductances in the “dc path” are due to the
wiring inductances between the MOSFETs and the decoupling capacitors (dc source
capacitor), while stray inductances in the “ac path” are due to the wiring inductances
between the MOSFETs (Clemente 1990). Stray inductance was reduced by using a
laminated bus structure to minimise connection lengths (input connections from the
source capacitor, MOSFETs interconnections and output connections to the SRM).
Parallel MOSFETs were mounted in a symmetrical pattern in order to further minimise
stray circuit inductance and they were mounted to a common heat spreader as this
provided close thermal coupling between devices. The detailed layout of the full bridge
circuit can be seen in Figure 5.26a (top view), 5.26b (side view) and Figure 5.26c
(circuit schematic). The MOSFET drains were connected directly to the heat spreader,
while the sources were connected to the printed circuit board mounted above the
MOSFETs (Murray 2001). The source connections were connected symmetrically using
copper bus conductor. The laminated bus bar was mounted centrally to maximise
symmetry and is described in detail in section 5.7.5.
137
Figure 5.26: Complete converter illustrations (a) top view (b) side view and (c) circuit
schematic
138
5.7.2 MOSFET gate driver
Negative voltage spikes are generated at the Vs pin of the IR2110 gate driver due to
di/dt in the stray inductance in the ac path (Clemente 1990). Once stray inductance was
reduced in the ac path, decreasing switching speed with a gate resistor further reduced
the di/dt-induced spikes. The circuit layout minimised the stray inductance on the gate
drive charge/discharge loop by minimising track lengths. This reduces oscillations and
improves switching speed and noise immunity to prevent dv/dt turn-on (Clemente
1990). The connections went directly from the gate driver pins to the MOSFET pins
using parallel tracks in a PCB layout. The IR2110 gate driver was placed as close as
possible to the power switches and local decoupling was improved by using one low-
ESR capacitor for the bootstrap capacitance and in the VCC-COM and VDD-VSS pin
locations of the IR2110. The low side capacitance, VCC-COM pins, was ten times
greater than the bootstrap capacitor (Clemente 1990). A ceramic capacitor was placed
as close as possible to the VB and VS pins of the gate driver when using an aluminium
electrolytic capacitor for the bootstrap capacitor, as the low-ESR capacitor provided
good de-coupling.
5.7.3 Gate drive buffer circuit

A 10µF electrolytic and a 0.1µF ceramic decoupling capacitors were mounted
physically close to the buffer output to nullify the effects of stray inductance (Kiraly
2004). Short parallel tracks in a PCB layout were used to connect the buffer output to
the parallel MOSFETs. A single point ground (ground plane of the centrally located
laminated bus bar) was used to connect the sources of the low side power devices to a
common point, which defined the common ground. A low-ESR decoupling capacitor
was selected for use in the power stage, typically 10µF to 1000µF depending on the
layout and switching current.
5.7.4 Optocoupler electrical isolation

A 0.1µF ceramic disc bypass capacitor was placed on the power supply line and it
needed to be placed within 10mm of the IC (Agilent Technologies 1999a). The
recommended layout for the optocoupler provided by the manufacturer is illustrated in
Figure 5.27 (Agilent Technologies 1999a).
139
Figure 5.27: Manufacturers recommended layout for optocoupler
5.7.5 Bus bars

The bus bars were mounted centrally on the converter as shown in Figure 5.17a. The
bus bars are separated into four layers: positive motor phase layer, A+, negative motor
phase layer, A-, positive power layer, +VS, and negative power layer, -VS, with an
insulating material separating each layer The bus bars are approximately a half-inch
wide. The specific interconnections to the MOSFETs are as follows:
• The negative motor phase plane, A-, is connected to the drain of Q2 via the heat
spreader and the source of Q4 via the stud connection located between the two
parallel devices.
• The positive motor phase plane, A+, is coupled to the drain of Q3 and to the
source of Q1 using the same scheme as the negative motor phase layer.
• The negative power plane, -VS, is connected to the source terminals of Q3 and Q2
the stud connection located between the two parallel devices.
• The positive power plane, +VS, is attached to the drains of Q1 and Q4 via the
heat spreader.
This format equalizes the dimensions and provides symmetry to the layout, which
further reduces stray inductance; the bus bar layers are stacked in the following
configuration: the bottom layer of the bus bar is the positive power plane, +VS, next is
the negative power plane, -VS, then the positive motor phase plane, A+, and finally the
negative motor phase layer, A-. A thin layer of insulation is placed between each motor
and power plane to implement a laminated bus bar construction. This configuration
minimises mutual inductance due to the close association of opposite polarity currents.
140
The symmetrical connections between MOSFET and the source stud were implemented
using copper tined conductor as illustrated in Figure 5.26a.
5.7.6 Current sensor

The layout of the current carrying conductor through the sensor aperture provides
maximum separation from the return conductor or other high current conductors and the
conductor completely fills the sensor aperture (LEM Components 2004), (Besnier 2004)
and (LEM Components 2005) as illustrated in Figure 5.26a and 5.26b, respectively. The
transducer output wiring must have minimum loop area and also must have minimum
length traces separated from power leads (LEM Components 2004). The current sensor
is mounted in the top corner of the PCB as illustrated in Figure 5.26a. Copper trace
material was cleaned away from around the circuit to provide isolation from the high
power circuitry and all signal and power connections were brought to the circuit through
a dedicated connector. Power and ground planes for the current sensor power supply
were implemented to minimise noise interference.
5.8 Final Converter Assembly

The complete single phase full-bridge converter assembly incorporating all the
requirements previously discussed and illustrated in Figure 5.26 can be seen in its
complete form in the images shown in Figure 5.28 – 5.31. The overall final assembly is
shown in Figure 5.28, the top view is provided in Figure 5.29, the side view is provided
in Figure 5.30 and the end view can be seen in Figure 5.31.
Figure 5.28: Single phase full-bridge converter final assembly
141
Figure 5.29: Top view of the single phase full-bridge converter
Figure 5.30: Side view of the single phase full-bridge converter
Figure 5.31: End view of the single phase full-bridge converter
142
5.9 Summary
The full-bridge converter with synchronous rectification using two MOSFETs in
parallel provides the best compromise of efficiency, cost and complexity for
implementation with the ISG under motoring and generating operation. Using
MOSFETs connected in parallel improved the current handling capability of the circuit
but the implementation was more complex than using individual modules. To finalize
the converter design, a thermal analysis was conducted in order to determine the most
suitable heat sink configuration. When using MOSFETs in parallel, the gate drive
circuit needs particular consideration in order to provide sufficient current to charge the
gates of the MOSFETs and effective operate the high side devices of the full-bridge
configuration. The provision for blanking time was addressed in order to provide
prevent current shoot-through conditions and electrical isolation was implemented using
optoisolators as part of the gate drive configuration, in order to protect the control logic
circuitry. A source capacitor value was selected in order to provide instantaneous
current for motor drive and act as a filter capacitor to reduce ripple during generating
operation. The main converter interconnections and power conductors in this
application were provided by a laminated bus bar structure in order to minimise stray
inductance in the circuit. A closed loop Hall effect current sensor was selected for this
application in order to implement current regulation during converter operations and
allow for the potential inclusion of short circuit protection. Specific layout conditions
needed to be addressed for the full bridge converter, the gate drive circuit with electrical
isolation, the bus bar configuration and the current sensor in order to achieve the most
efficient operation of the circuit in this application.
143
Chapter 6 – System Implementation
Practical implementation in this application involves both hardware and software. The
issues discussed range from the experimental setup for converter evaluation through to
the development of control strategies to provide precise operation of the converter under
motoring and generating modes. Two experimental setups were developed, one
consisting of the SRM and a load/drive motor with a torque measurement system and a
second setup comprising of a bench load to model the resistive and inductive
characteristics of a single phase the SRM. Experiments were only carried out on the
bench load.
6.1 Hardware
The experimental setup comprising of the SRM and load/drive motor with a torque
measurement system was developed to provide a load or facilitate drive under motoring
and generating operating modes of the ISG, respectively. The experimental set-up for
the converter evaluation and the development of the current control strategy under
bench conditions consists of a resistive-inductive load to model the SRM phase
winding, a 1kW power supply, a low voltage power supply and interface circuit, the
full-bridge converter and the digital signal processing development board (DSP
controller) as illustrated in Figure 6.1. The development of the converter prototype
included the converter layout, mechanical component considerations, converter
assembly and preliminary evaluation of the converter to verify that it satisfies functional
tests.
1 k W P o w er S u p p ly
C o n verter L V P o w er S u p p ly
& PC
In terfa ce U n it
RL Load
S et-u p
D S P C o n tro ller
Figure 6.1: The complete experimental setup for converter evaluation

144
6.1.1 The SRM
The project initially involved developing an electronic drive for a 2-phase SRM. The
use of a 2-phase SRM is suitable for applications with high initial starting torques and
where the cost of the converter needs to be minimised. The large peak torque capability
also yields a significant regenerative braking capability, which can be highly
advantageous. Unfortunately, the original project plan for the 2-phase SRM to be
designed and constructed by a project partner was not fulfilled so an alternative SRM
was sourced for the project application. An SRM of equivalent power rating was
sourced commercially but since switched reluctance motors have a limited availability
in off-the-shelf configurations, compromises had to be made in order to meet the desired
power requirements. The SRM that most closely matched the specifications was a
3.5kW, 48V, 4-phase, 8/6 SRM (Motion System Tech 2002). While an SRM could be
custom designed to the project specifications (Motorsoft 2007), this direction was
deemed to be prohibitively expensive and time consuming for this single application, so
the commercially available SRM was selected instead, in order to prove the validity of
the hardware application. The nominal parameters for the 4-phase SRM are provided in
Table 6.1.
Manufacturer Motion System Tech

Model RA165L
Parameters Value
Rphase 8mΩ
Laligned 334µH
Lunaligned 47µH
Moment of Inertia 127 kg cm2
Rated current 100A
Rated voltage 48V
Stator/rotor poles 8/6
Phases 4
Rated torque 5.68 N-m
Rated speed 6000 rpm
Table 6.1: Nominal parameters of the SRM
145
6.1.2 The Digital torque measurement system
The digital torque measurement system consists of a dc-machine (MV 1042) attached
via ball bearing at each end of the machine to the machine housing, a machine bed
(MV1004) and a control unit (MV 1041) (Terco 2003). The machine bed aligns and
secures the dc machine to the test machines. Torque measurement is achieved using a
torque arm attached beneath the machine on which is mounted a strain gauge transducer
and this constitutes the signal source for the torque measurement. The machine also
contains a speed indicator, which generates a pulse train proportional to speed. The
control unit consists of digital instruments for torque, speed, and field and armature
currents, while the front panel of the control unit also contains a potentiometer for
control of the field rectifier for load tests. The field can also be supplied by an external
source by removing the jumpers and there are terminals on the front panel for
connecting the armature winding to an external power source (motor drive) or to a load
resistor (generator drive). There are analogue outputs for speed and torque and an
analogue input for control of load torque.
An interface chassis was developed to align and connect the SRM with the dc machine
and machine bed. A suitable coupler was acquired to interface the SRM with the dc-
machine. It is a general-purpose coupler with a flexible rubber insert for flexibility and
shock resistance. It is capable of accommodating accidental misalignments and has a
torque range of up to 90Nm (Fenner 2006). The digital torque measurement system is
illustrated in Figure 6.2.
Control Unit DC-Machine Coupling SRM
Torque
Manchine bed
transducer
Figure 6.2: The digital torque measurement system
146
6.1.3 Experimental RL load setup and 1kW Power Supply
The RL load setup consisted of an inductor that modelled the worse case phase
inductance and winding resistance when the SRM was in the unaligned position and the
inductance was at the minimum value. It was used as part of the bench test setup to test
the converter operation and develop the current regulation strategy. The experimental
RL load setup with converter is illustrated in Figure 6.3.
RL Load Setup
Q1 Q4
R L
Q3 Q2
Converter
Figure 6.3: Schematic of converter with RL load
A 1kW power supply was used for the dc link power supply for the converter. It
provides dc power up to 1kW over a wide range of voltage and current combinations.
The power supply has a voltage range of 0-60V and a current range of 0-50A.
6.1.4 The LV power supply and interface unit

This unit provides all the low voltage power supplies and signal interfacing for the
converter and the DSP controller. This unit was developed as part of a prototype
assembly that was designed and built for a 3phase SRM (described in the report
(Murphy 2002)). A detailed schematic of the power supply circuit is included in
Appendix D. The power supply provides the following voltages:
• isolated +5V supply for the optical isolators output stage on the interface circuit
• ± 5V supplies for the DSP controller
• +3.3V supply for the CMOS logic circuits and the optical isolator inputs on the
interface circuit to provide logic level compatibility with signals to and from the
DSP controller
• ± 12V supplies for the current sensing circuit on the converter hardware
• + 15V for the converter gate drive circuit.
• +12V, 2A power supply for cooling fans on the converter
147
The interface circuit’s primary purpose is to provide optical isolation for the SRM gate
drive signals from the DSP controller circuit. This reduces the risk of damage to the
DSP controller circuit in the event of an electrical failure on the converter. The circuit
also provides a distribution point for the power supply voltages for the converter and
DSP controller circuits. Finally, this circuit provides an interface for signals being
routed to and form the DSP controller circuit, i.e. current sensing signals and the output
enable switch. A detailed schematic of the interface circuit is provided in Appendix D.
A bock diagram of the interface circuit and the low voltage power supply is given in
Figure 6.4; while a detailed interconnection schematic for this system is included in
Appendix D.
Figure 6.4: Block diagram of interface and low voltage power supply
6.1.5 DSP controller

An Analog Devices digital signal processor (ADSP21992) was employed in the
experimental implementation of the current control strategy. It was suitable for this
application, as it had a three-phase PWM generator unit and dual auxiliary PWM
outputs, which allowed the implementation of current regulation using PWM for a four
phase SRM. Additionally, using this hardware allowed for integration of the converter
current controller with the SRM motor and generator control strategies developed on the
same hardware.
The ADSP21992 is a mixed signal DSP controller that integrates the fixed point ADSP-
219x family base architecture (Analog Devices 2003a). A functional block diagram of
the DSP controller is provided in Figure 6.5 and the key features are listed in Table 6.2.
(The ADSP21992 costs approximately €19 when ordered in quantities of between 1000
148
and 5000). The ADSP-21992 EZ-KIT Lite development board was employed in the
experimental set-up for evaluation of the converter and development and test of the
current regulation scheme (Analog Devices 2003b). The integrated special purpose and
motor control peripherals (the analog-to-digital conversion (ADC), the pulse-width
modulation (PWM), the programmable interrupts and the programmable Flag I/O pins)
were employed in the converter evaluation and in the current controller implementation.
An overview description of the development board and how these specific peripherals
are setup are provided in Appendix E.
• ADSP-219x DSP core, 160MHz

• 14-bit, 8-channel, 20 MSPS ADC with on-chip voltage reference
• 16K data memory, 32K program RAM
• Three-phase PWM generation unit, with additional dual auxiliary
PWM outputs
• 16-bit general purpose flag I/O port
• Incremental encoder interface unit
• Three 32-bit, general purpose timers
• External memory interface
• Peripheral interrupt controller
• Synchronous serial (SPORT) and SPI communications port
• Integrated controlled area network (CAN) interface
Table 6.2: Key features of ADSP-21992
In order to implement the numerous capabilities of the ADSP-21992 DSP, a full

understanding of the software development environment called VisualDSP++ was
required (Analog Devices 2002).. It provided a single, integrated project management
and debugging environment for advanced code development and debug. VisualDSP++
3.0 has a number of built in or intrinsic functions designed to make programming a DSP
in a C environment even more user-friendly. The specific software requirements to setup
and use the required peripherals in this application are provided in Appendix F.
149
Figure 6.5: Functional block diagram of ADSP-21992
150
6.1.6 Converter prototyping
The prototyping of the converter is discussed under four specific headings: electronic
circuit layout, mechanical layout, converter assembly and preliminary converter
evaluation. As image of the complete converter assemble is shown in Figure 6.6.
Figure 6.6: Single phase full-bridge converter final assembly
6.1.6.1 Circuit layout

The converter circuit schematic was initially drawn in a CAD software schematic
capture package and a netlist was generated in order to create a prototype printed circuit
board (PCB) layout for test and evaluation. The central aspect of the layout is the
MOSFET positioning that defines source and drain layout to minimise stray inductance
and maximise symmetry and maximise circuit balance (through equal track lengths)
between the parallel devices. After much iteration a final layout compromise was
achieved for the parallel switching component layout. Significant attention was focused
on addressing the gate drive circuit layout for the gate driver and the buffer circuit
where the layout was optimised for minimal stray inductance and maximum circuit
balance. All gate drive signal groups used ground planes in order to minimise noise and
stray inductance and maximise gate drive signal quality. The layout for the current
sensor required that the device was located as far as possible from the high current
switching signals and their corresponding switching emissions. The laminated bus bar
connection to the SRM (A+ and A-) separated a sufficient distance from the current
sensor to minimise sensor signal interference and maximise signal quality as illustrated
151
in Figure 6.10a. A dedicated ground plane was placed in the sensor circuitry area to
minimise noise contamination of the sensor signal. The printed circuit board layout
illustrating the various ground planes is illustrated in Figure 6.7. The complete
converter circuit schematic is illustrated in Figure 6.8.
Current Sensor
ground plane
Gate drive output

ground planes
Gate drive input

ground planes
Figure 6.7: Printed circuit board layout illustrating the various ground planes
152
Figure 6.8: Converter circuit schematic
153
6.1.6.2 Mechanical layout
The mechanical outlines of all the major components were defined during the PCB
layout. This allowed for detailed sizing of all mechanical components (heat spreaders
heatsink, cooling fan and bus bars) to be finalized including determination of the exact
locations of all the components. There were interactions between the various mechanical
components and the PCB, which included the bus bar drain connections (heat spreader),
locating the sources studs on the PCB to attach the bus bar source connections,
attaching the PCB to the heat sink while electrically isolating the heat spreader, routing
the bus bar through the current sensor and locating of the cooling fan was also finalized
during the PCB layout. The specific hole sizes and locations for all the mechanical
components (heat spreader, heatsink and bus bar) and the PCB were also defined during
the layout. An illustration of the converter assembly with the heat spreader, thermally
conductive insulators and the heatsink is shown in Figure 6.9, while a detailed drawing
of the complete converter assembly is provided in Figure 6.10.
A number of templates were created as part of the PCB layout process that would allow
marking of the exact hole locations on the heat spreader (both top and sides), the exact
size and mounting hole locations for the thermally conductive insulator and the exact
mounting hole locations on the heat sink to attach the heat spreaders and the cooling
fan. A number of template outline drawings were also created for the bus bars to
specify the exact size and mounting hole locations for each bus bar layer and insulator
layer.
The holes drilled in the top of the heat spreader were through-holes for the mounting
bolts to attach the PCB heat spreader assembly to the heat sink. The hole diameter
needed to be sufficient to accommodate the mounting bolt and an insulating spacer.
Holes were drilled on the side of the heat spreader and tapped for M4 bolts to attach the
relevant laminated bus bar connections. The thermally conductive insulators were cut to
a sufficient dimension so as to allow four millimetres of the electrically insulating
material to protrude all around the heat spreader as illustrated in Figure 6.9. Through-
holes were punched to just allow the PCB heatspreader to heat sink mounting bolts to
pass through in order to maximise electrical isolation.
154
Figure 6.9: Converter assembly with heat spreaders and heatsink
155
6.1.6.3 Converter assembly
The assembly of the converter was divided into various stages: the PCB assembly, the
heat spreader with MOSFET assembly, the installation of the bus bar assembly, the heat
sink been put in place and finally the fitting of the heatsink cooling fan. The PCB
assembly involved the installation of all the circuit components except for the power
MOSFETs. Two MOSFETs were installed to each of the four heat spreaders and the
each assembly was then attached to the printed circuit board at the four locations for Q1,
Q2, Q3 and Q4. Insulating spacers were placed between the heat spreader and the PCB
and additional insulating spacers were placed in the through-holes for the bolts attaching
the PCB heat spreader assembly to the heat sink as illustrated in Figure 6.10b. This
provided electrical isolation between each head spreader MOSFET drain connection that
would otherwise be compromised when the PCB heat spreader assembly was bolted to
the heat sink. The bus bar lamination was routed through the current sensor and then
affixed to the appropriate drain and source locations as described in section 5.7.5 of
chapter 5 and filler strips were inserted to the bus bar at the current sensor to improve
signal strength as illustrated in Figures 6.10a and 6.10b. The remaining bus bar layers
were attached to the appropriate drain and source connections together with appropriate
thin insulator separating each layer as illustrated in Figures 6.10a and 6.10b. The
thermally conductive insulators were placed at the appropriate locations on the heat sink
and the PCB - heat spreaders assembly was aligned with the heat sink and bolted in
place. The PCB - heat spreader assembly to heat sink mounting bolts were tightened to a
torque value of 15Nm following a diagonal pattern to provide uniform pressure on all
the heatsink, thermal conductive insulator and heat spreader junctions in order to
achieve the best possible thermal equilibrium between the power switching devices. The
heatsink cooling fan was then attached to the heatsink. The complete converter
assembly drawings are shown in Figure 6.10.
156
Drain Drain
Source Source
Figure 6.10: Complete converter assembly (a) top view (b) side view and (c)
approximate 3D view
157
6.1.6.4 Preliminary converter evaluation
The initial evaluation of the converter involved visual inspection of the converter and
the electrical test of the converter in incremental stages through to full converter
switching operation on a load resistance. The first area to be addressed was the visual
inspection of all components to determine if the values and device orientations were
correct. The gate driver IC was removed and power is applied to the circuit to verify no
short or open circuits and that all power supply levels were available. The output pins of
the gate driver were shorted to their respective common pins to prevent a floating input
to the buffer circuit. A square wave test signal was then applied to verify that the input
signal lines were functional. The gate drive IC was then inserted but with the output
pins floating and the test signal was again applied to verify the output gate drive signal
was available. A differential probe was used to verify the all-high side gate drive
signals. The gate drive IC was then completely inserted and the test signal applied once
more to verify the outputs of the gate drive buffer circuit, in order to verify that the
correct signals were present and at the correct levels. Finally a low voltage and low
current power supply was applied once more to the dc link without the presence of the
dc source capacitor and the test signal was applied to verify the output switching signal
of the converter across a large value resistor in order to evaluate that the converter was
functional. The converter was then ready for full operation and for the development and
test of current control strategies.
6.2 Software
Precise control is required for the SRM and the full-bridge converter to operate
efficiently. The level and timing of the current that enters each phase winding of the
SRM is extremely important to ensure smooth torque generation. In order to evaluate a
converter design, the first area of control to be addressed tends to be current regulation.
Initially, this can be achieved using an open loop scheme at very small duty cycles, but
is really only suitable for functional tests. The operation of the converter precisely over
a wide range requires closed loop current control as shown in Figure 6.11. In this
project, it was implemented using voltage-PWM with a Proportional-Integral (PI)
controller to determine the duty-cycle values
158
Figure 6.11: Converter current control scheme
The converter was only implemented as a single-phase unit and a resistive-inductive

(RL) load was used to replicate the phase winding resistance and inductance in the
unaligned position. This provided the worst case conditions to evaluate both the
converter and the current control strategy. The control source code was developed such
that it could be ported for implementation with a 4-phase SRM.
The overall control structure was first defined in order to achieve precise step-by-step
operation of the converter and following this the precise implementation of current
regulation was then addressed. The control strategy was then implemented on the
ADSP-21992 DSP evaluation board using the C programming language through the
VisualDSP++ application. The central function in the project source code was the
interrupt service routine (ISR) for the PWM duty-cycle update. A number of
experiments were conducted to evaluate the fundamental operation and the worst-case
performance of the converter and the current control strategy.
6.2.1 Control Structure

Current control was implemented using a voltage-PWM as described in Chapter 2,
section 2.4.1, and the duty-cycles were determined using a PI controller. PWM provides
fixed switching frequency operation, which allows for a constant level of switching
losses and therefore improves converter efficiency. Additionally the use of a PWM
current controller reduces the current harmonics, which yields a reduction in acoustic
noise during SRM operation (Miller 2001). The current controller design was based on
the scheme of designing the controller for the unaligned phase inductance value that
results in a stable design for all rotor positions (DiRenzo 2000).
159
The implementation of the switching strategies for precise converter operation was
addressed prior to the design of the current controller. These switching strategies control
the operation of the individual converter MOSFETs in order to execute the four modes
of operation (magnetisation, freewheeling, demagnetisation and shut-down) of the
converter as shown in Figure 6.12 and described in Chapter 2. The PWM output signals
were connected to the MOSFETs, Q1 and Q3, and I/O logic level signals were
MOSFETs, Q2 and Q4.
Figure 6.12: The full-bridge converter. (a) Static view of converter in shut down mode,
(b) magnetization mode, (c) freewheeling mode, and (d) demagnetization mode
SRM control requires precise rotor position information in order to determine the
correct turn-on and turn-off angles, which in turn defines the conduction interval over
which the current flows into and out of the phase winding of the SRM. For operation as
a single-phase unit with an RL load, signal pulses were generated to simulate the turn-
on and turn-off angles instead of reading and decoding the actual rotor position
information. The position information allowed for the establishment of control signals to
precisely implement switching strategies as shown in Table 6.3.
There are five distinct cycles that happened during the conduction interval for the
converter: start-up PWM, PI voltage-PWM, freewheel PWM, demagnetisation PWM
and shut-down. Start-up PWM and PI voltage-PWM implements magnetisation and
freewheeling operation as shown in Table 6.3. At the turn-on angle, there is a start-up
160
routine using PWM operation but at fixed duty-cycles in order to allow the current in
the SRM phase winding to rise quickly and controllably to a threshold current (Ith)
value. The current is then controlled by PI voltage-PWM operation and regulated at the
reference current (Iref) value and this requires PWM operation with variable duty-cycle
values determined by a PI controller. If the current exceeds an upper limit current (Ihigh),
assigned to be 10% above Iref, then freewheel PWM operation is required to allow the
current to decrease sufficiently. This mode applies 0% duty cycle to Q1 and 100% duty
cycle to Q3.
At the turn-off angle, demagnetisation of the phase winding commences and when the
value of the current is above the threshold level the MOSFETs, Q3 and Q4, are turned
on. Q3 requires PWM operation and is implemented at 100 kHz with 100% duty cycle.
This higher frequency operation allows more frequent current measurement to determine
if the minimum current threshold has being reached. Finally, demagnetisation at low
current and shut-down requires that all MOSFETs are turned off and the remaining
current conducted through the intrinsic diodes of the MOSFETs. Shut down also signals
the end of close loop control for that particular phase.
Position Modes Full bridge converter

logic of Q1 Q3 Q2 Q4
A B Operation Imeas<Ith Imeas<Ith<Ihigh Imeas>Ihigh Imeas<Ith Imeas<Ith<Ihigh Imeas>Ihigh
1 1 Mag/free Start PWM PI PWM Free PWM Start PWM PI PWM Free PWM On Off
1 0 Demag 0% Shutdown Demag PWM (100%) Off On
X 0 Shutdown Off
Table 6.3: Rotor position logic and converter switching cycles
The angle data and current measurement (Imeas) data determine the transition points
between the different cycles. The current measurement defines the point to toggle
between start-up PWM and PI voltage-PWM (PI PWM) and the changeover between
demagnetisation at high current (demag PWM) and shut-down if the shut-down angle
has not yet being reached. The angle data defines when PWM (start-up and PI)
commences and when demagnetisation commenced.
The PI controller determines the new duty-cycle value required to regulate the current at
a predetermined reference value. The control law used for the PI controller is shown in
equation 6.1 (Dorf 1998).
161
T
u (t ) = K p e(t) + K i ∫ e(t) dt (6.1)
0
In this application, u(t) is the new duty-cycle value, Kp the value of gain applied to the
error signal, e(t). The error signal, e(t), is the difference between the reference current,
Iref, and the actual measured current, Imeas. Ki is the value of gain applied to the integral
of the error signal. The time increment, dt, is the PWM period (40µs for 25kHz
switching frequency), as the integral of the error is determined at a fixed time interval
during the ISR for the PWM update.
The PI controller was tuned manually. The resultant values of Kp and Ki, were Kp =
0.015 and Ki = 0.1.
6.2.2 Control Implementation

The source code was structured for the implantation with a four-phase SRM. Since the
converter was only evaluated as a single-phase on an RL load, only the source code for
a single-phase (phase A) was fully developed and tested. The ISR for the PWM was
only developed for a single-phase (phase A) but could easily be expanded for phases B
and C. Phase D would require the implementation of an ISR for auxiliary PWM but was
not developed further than the initialization stage in order to verify it was capable of
matching the operation of the other three phases. The source code contains six
functions: read_rotor_angle(), commutation_four_phase(), motor_control_a(),
pwm_isr(), Initialization(), and the main function.
A tutorial on how to implement the configuration requirements for the DSP modules
(ADC, PWM, Auxiliary PWM, Program sequencer and Peripheral interrupt controller,
and Flag I/O) used in this project are documented in Appendix E and tutorial on using
the VisualDSP++ development suite to create the source code is provided in Appendix
F. The complete source code is documented in Appendix G
The main function calls the functions in the appropriate sequence for successful
program operation for the PWM current control update. A number of DSP modules
(Clock speed, Interrupts, ADC, PWM, Aux PWM and FIO) registers needs to be
configured before program execution. These setup requirements are grouped together in
162
a single initialization function, initialization().The flowchart for the overall structure of
the program source code is shown in Figure 6.13.
Figure 6.13: Overall source code structure
The central function in the project source code is the ISR for the PWM update. A
number of other functions are required for the ISR implementation and to provide
structure in the execution of the source code. In order to implement the PWM
successfully, the exact rotor position is required so a function, read_rotor_angle(), was
developed to provide rotor angle information. As the converter was only evaluated as a
single phase with a RL load, the function only provided the angle output data needed to
test the converter operation. The angle information is used by the commutation function,
commutation_four_phase(), to generate the appropriate control signals to implement the
various modes of operation for each phase and disable the other phases at the same time
as illustrated in the flowchart shown in Figure 6.14.
163
Figure 6.14: The commutation function to determine the control signals
The control signal information is then used by the motor control function,
motor_control_a(), to configure and enable the PWM outputs and the FIO outputs for
phase A, as illustrated in the flowchart shown in Figure 6.15. This implements each of
the five distinct cycles (startup PWM, PI PWM, freewheel PWM, demag PWM and
shutdown) that occur during the conduction interval of the converter. A duty cycle
update control variable is then set to allow the ISR to be called.
The steps involved to implement the ISR function for the PWM, PWM_ISR(), are
illustrated in the flowchart shown in Figure 6.16. The first step in the ISR is to clear the
interrupt and then check if in the 2nd half-cycle of the PWM signal as an update is only
implemented at this time. This provides a half PWM period between the ADC reading
and the duty-cycle update. The ISR then reads the ADC value and calculates the current
value, and then using this information calculates the duty-cycle value from the PI
difference equation. Then depending on the control signal for the converter mode of
164
operation (magnetisation, freewheeling and demagnetisation) and the level of current,
the final duty-cycle update codes are determined and assigned to the appropriate
registers. The duty cycle update control variable is then cleared before exiting the ISR.
Start
Motor_Control
Selection
Start-up PWM
Demagnetisation Shut down
and
PI Voltage-PWM
Imeas
Imeas > Ith Imeas < Ith
Set PWM Frequency- 25kHz Set PWM Frequency - 100kHz

Invert low side PWM outputs Invert Low side PWM outputs No invertion of low side
SR Alternate Mode SR Alternate Mode PWM outputs
Enable PWM outputs with Enable PWM outputs with Disable SR Alternate Mode
PWM Sync signal & double PWM Sync signal & double Disable PWM outputs and
update mode update mode PWM Sync signal
Enable Q2 Flag I/O Disable Q2 Flag I/O Disable Q2 Flag I/O
Disable Q4 Flag I/O Enable Q4 Flag I/O Disable Q4 Flag I/O
PWM ISR
End
Figure 6.15: The motor control function to initial the appropriate converter cycle
165
Start
PWM ISR
Clear Interrupts
Test for 2nd half cycle
Selection
Yes No
Current Reading
Read ADC
Scale reading to true
current value
PI Controller
Calculate Error
Calculate Integral
Calculate duty cycle (D)
Selection
Startup PWM
and Demagnetisation
PI PWM
Imeas Imeas
Imeas > Ihigh Imeas > Ith Imeas < Ith

Imeas < Ith Ith < Imeas > Ihigh
Set D = 2.5% for Q1 Set D = D for Q1 Set D = 0% for Q1 Set D = 0% for Q1 Set D = 0% for Q1
Set D = 2.5% for Q3 Set D = 1-D for Q3 Set D = 100% for Q3 Set D = 100% for Q3 Set D = 0% for Q3
Start-up PWM PI PWM Freewheel PWM Demag PWM Shut-down
End
Figure 6.16: The ISR for the PWM duty-cycle update
166
6.2.3 Design Verification
As noted previously, the control strategy was implemented on an ADSP-21992 DSP
development board. Only a single phase converter was evaluated and an RL load was
used to model the phase winding resistance and inductance at the unaligned position.
This provided the worst-case condition to evaluate both the converter operation and the
current control strategy. Initially the elementary converter and source code operation
was evaluated under three separate conditions: the gate drive operation, the gate drive
switching operation under the different cycles of operation and finally the current
measurement capability. The current control algorithm was then evaluated at different
current levels in order to verify the converter operation and current control accuracy.
6.2.3.1 Preliminary Converter and source code evaluation

The initial assessment of the converter necessitated that the gate drive switching signals
be evaluated from the DSP evaluation board output through to the power MOSFETs
(Q1-Q4) inputs in order to verify that signal levels were correct at each stage The
switching signal chain from the DSP output through the gate drive buffer circuit output
of the converter is shown in Figure 6.17.
Figure 6.17: Gate drive signal path for Q1 at DSP 3.3V output (Channel 1), at the
IR2110 gate driver 5V input (Channel 2 1V = 50mV) and at buffer circuit 15V output
(Channel 3 1V = 20mV)
167
The current measurement by the DSP was verified in conjunction with the basic
MOSFET switching operation using the setup illustrated in Figure 6.18. The MOSFET
switching signals for start-up PWM, PI PWM, demagnetisation or demag PWM and
shutdown were evaluated at this time. The signals for the transition between start-up
PWM and PI PWM, and between PI PWM and demag PWM are shown in Figures 6.19
– 6.20, respectively.
Figure 6.18: Initial current measurement test setup
Figure 6.19: Transition between shutdown and start-up PWM operation, Q1 (Channel
1), Q2 (Channel 2), Q3 (Channel 3) and Q4 (Channel 4). Q1 has fixed duty cycle of 2.5%
and consequently Q3 has a duty-cycle of 97.5% during start-up, while Q2 is on and Q4
is off
168
Figure 6.20: Transition between PI PWM and demag PWM operation, Q1 (Channel 1),
Q2 (Channel 2), Q3 (Channel 3) and Q4 (Channel 4). The Q3 PWM gate signal operated
at 100 KHz and 100% duty cycle during demagnetisation.
As can be seen from Figure 6.20, unstable PWM operation occurred during the
transition between PI PWM and demag PWM. During this interval of approximately 10
– 20µs, Q1 turned on while Q3 was on leading to a shoot through current condition. A
number of attempts were made to prevent this glitch from occurring including adding a
delay in the angle reading source code to shut down the converter between transitions,
but a solution could not be found. As this occurred at the reference current, which could
be up to 100A and as the result of one board failure at much lower current, it was
decided to disable this feature and go directly to shutdown from PI PWM as illustrated
in Figure 6.21. While demag PWM is possible, this glitch would need to be resolved in
order to implement both demagnetisation mode at high motor speed and generator mode
at high efficiency and achieve the maximum advantage of this converter design. As the
converter was designed for operation with a duty cycle of 10% during demagnetisation
mode and the value experienced during testing as <1% at ISG starting speed (200rpm),
the demag PWM mode was disabled to achieve stable software operation. Therefore,
during shutdown the demagnetisation current would flow through the intrinsic diodes of
Q3 and Q4.
169
Figure 6.21: Transition between PI PWM and shutdown operation, Q1 (Channel 1), Q2
(Channel 2), Q3 (Channel 3) and Q4 (Channel 4).
Once the current measurement and MOSFET switching operation was verified, the PI
current controller was developed for operation at 5A, initially. The controller was tuned
heuristically with the test results shown in Table 6.4 for proportional gain, Kp, integral
gain, Ki, 5A reference current, Iref, peak current, Ipeak, steady state current, Iss, percentage
overshoot, OS%, rise time, tr, settling time, ts, and steady state error, ess, to illustrate the
transient and steady state response of the system. The resultant values of Kp and Ki, that
yielded the best performance characteristics were Kp = 0.015 and Ki = 0.1.
Kp Ki Ipeak (A) Iss (A) OS% tr (µs) ts (µs) ess

0.001 0 2.8 2.5 -44 270 960 -2.5
0.01 0 4.6 4.3 -8 430 810 -0.7
0.1 0 16.3 4.9 226 640 5680 -0.1
0.05 0 8.5 4.7 70 700 3180 -0.3
0.02 0 5.3 4.5 6 680 1500 -0.5
0.015 0 4.9 4.4 -2 700 1100 -0.6
0.015 0.001 5 4.5 0 700 1100 -0.5
0.015 0.01 5.3 4.8 6 700 1160 -0.2
0.015 0.1 5.3 4.9 6 7200 1100 -0.1
0.015 1 5.4 4.9 8 700 1100 -0.1
0.015 10 6 4.9 20 700 1840 -0.1
0.015 0.5 5.3 4.9 6 700 1260 -0.1
Table 6.4: Results from heuristic tuning of the PI current controller
170
A comparison between the measured current of 5A in the RL load and the value
calculated in the source code and plotted by the VisualDSP++ application is shown in
Figures 6.22 and 6.23, respectively. This verified the converter current sensor operation
and measurement accuracy using the source code calculation.
Figure 6.22: Measured current in the RL load at 5A (Channel 3 1A = 10mV/div)
Figure 6.23: Current of 5A calculated during source code execution and plotted by the
VisualDSP++ application. Current measured by current sensor on the converter
171
6.2.3.2 Current controller evaluation
The performance was then evaluated under the different operating cycles (start-up
PWM, PI PWM, demagnetisation PWM and finally shut-down). Figure 6.24 illustrates
the full-bridge converter modes of operation for soft switching to provide a reference for
comparison with the actual results. Operation under start-up PWM conditions is shown
in Figure 6.25, while the transition between start-up PWM and PI PWM conditions is
shown in Figure 6.26. The current ramped up with a fixed duty cycle value until the
threshold current, Ith, was reached and then continued under PI PWM operation until the
reference value, Iref, of 5A was reached. Once at the reference current, the PI PWM
control strategy regulated the current at the reference value. At demagnetisation, all the
MOSFETs (Q1 - Q4) were turned off and the transition between PI PWM and shut-down
is shown in Figure 6.27. The current handling capability was then evaluated at three
different current levels, 20A, 40A and 100A, and are shown in Figures 6.28 – 6.30,
respectively
Figure 6.24: Full-bridge converter modes of operation for soft switching (I)
magnetisation, (II) freewheeling and (III) demagnetisation
172
Figure 6.25: Start-up operation at transition period between shutdown, start-up PWM
and PI PWM, Q1 (Channel 1), Q3 (Channel 2), RL voltage (Channel 3) and RL current
(Channel 4 – 5A/div).
Figure 6.26: Current waveform and control signals at transition period between start-up
PWM and PI PWM, Q1 (Channel 1), Q3 (Channel 2), RL voltage (Channel 3) and RL
current (Channel 4 – 5A/div ).
173
Figure 6.27: Current waveform and control signals at transition period between PI
PWM and shut-down, Q1 switching signals (Channel 1), Q2 switching signal (Channel
2) and RL load voltage (Channel 3), Current (Channel 4 – 5A/div)
Figure 6.28: Current and voltage waveform and control signals for 20A RL load
current. Q1 switching signals (Channel 1), Q2 switching signal (Channel 2) and RL load
voltage (Channel 3), Current (Channel 4 – 10A/div)
174
current. Q1 switching signals (Channel 1), Q2 switching signal (Channel 2) and RL load
current Q1 switching signals (Channel 1), Q2 switching signal (Channel 2) and RL load
175
6.3 Summary
The system implementation consisted of both hardware and software. The hardware
included two experimental setups and the full-bridge converter prototyping. One
experimental setup consisted of the SRM and a load/drive motor with a torque
measurement system and a second setup comprising of a bench load to model a single
phase winding of the SRM. Experiments were only carried out on the bench load. The
full-bridge converter prototyping included the electronic circuit layout, the mechanical
layout, the converter assembly and the preliminary converter evaluation.
The software operation centred on the switching strategies to control the operation of
the individual converter MOSFETs and the implementation of the PWM current
regulation strategy. The execution of precise current control required an understanding
of the DSP evaluation board used in this application. The specific areas included the
ADC, the PWM, and the auxiliary PWM, the program sequencer, the peripheral
interrupt controller and the flag I/O. An understanding of the software development
environment, used in conjunction with the DSP evaluation board, was required in order
to implement all the DSP functions required in this application.
The design was evaluated under a range of operation from initial functional testing
through to full power operation and the limitations and the results are presented to
illustrate each level of operation.
176
Chapter 7 - Conclusion and Recommendations
7.1 Conclusions
In this thesis a number of converter topologies for implementation with the SRM were
presented. The full-bridge converter with synchronous rectification using two
MOSFETs mounted in parallel was chosen for the design that would be used as part of
integrated starter generator (ISG). Control strategies were designed to provide precise
control of the full-bridge converter switching devices and implement current regulation
using voltage PWM with a proportional-integral controller.
The opening chapters presented a review of the SRM and a number of converter
topologies. This provided the criteria to define the operating conditions when operating
the SRM as an integrated starter generator and narrowed the converter selection to the
half-bridge asymmetrical converter and the full-bridge converter. Preliminary analysis
on the switching devices and the converter topologies determined the most efficient
configuration for operation under motoring and generating conditions and also defined
the limits for the converter operation under these conditions. The switching frequency
was defined to be fixed at 25kHz and the current was regulated using PWM and the
duty cycle value determined using a proportional-integral controller.
Based on this analysis, the full-bridge with synchronous rectification using two
MOSFETs mounted in parallel was deemed to provide the most efficient solution for
SRM operation as a motor and especially when operating as a generator. MOSFETs
connected in parallel improved the current handling capability of the circuit but the
implementation was more complex than using individual modules. The issues that had
to be addressed centred about maximising the current balance between parallel devices.
This was achieved through minimising stray inductance by using symmetrical layout
and using laminated bus bars for all high current connections, designing a gate drive
circuit to provide the significant charge required by the parallel devices at turn on and a
low impedance path for fast device turn off, and through close thermal coupling of
parallel devices by mounting them on a heat spreader. Thermal analysis determined the
most suitable heat sink configuration while a closed loop Hall effect transducer was
selected for the current measurement.
177
The converter was evaluated using an experimental setup comprising of a bench load to
model the characteristic of a single SRM phase winding. C code was developed to
implement the switching strategy for the converter switching devices to implement
magnetisation, freewheeling and demagnetisation modes of operation based on angle
information data. PWM with the duty cycle updated through a proportional-integral
controller as the method used to implement current regulation.
The software was implemented on a DSP evaluation board and required a

comprehensive understanding of a number of key features of the DSP including ADC,
PWM and interrupt implementation. The hardware and software were validated by
experimental results presented in Chapter 6. The experimental results verify the
switching strategies, but highlighted unstable PWM operation in the transition between
PI PWM and Demag PWM. Attempts were made to resolve the problem but none were
found. Demag PWM mode was disabled (the converter went straight to shutdown mode)
in order to prevent a current shoot trough condition occurring that would result in a
catastrophic converter failure. The experimental results verify the current measurement
accuracy and the converter current handling capability up to 125A. While the current
handling capability was verified for the converter, this was due in part to an error that
was occurring in the current measurement. While the current measurement was accurate
up to 40A, the error in the measurement when it had reached 125A was >25%.
A second experimental setup was constructed consisting of a 3.5kW 4 phase SRM and a
load/drive motor with torque measurement system but only the bench load setup was
used for converter testing and software development.
7.2 Recommendations
While software was developed and tested to implement the PWM strategies to operate
the converter in start-up PWM, PI PWM and in demag PWM modes, the unstable
transition between PI PWM and demag PWM need to be resolved in order to achieve
stable system operation.
With full motor operation and precise position information it could be possible to apply
0% duty cycle to Q1 and 100% duty cycle before turning off Q2 and turning on Q4.
178
Developing a test circuit to implement precise rotor position information would be the
next logical stage for prototype software development.
The 1kW power supply was completely saturated in attaining the 125A measurement,
therefore, to operate a 3.5kW SRM at high power levels a 3.5kW – 5kW power supply
would need to be obtained or designed to fully evaluate the converter with the SRM.
The current measurement accuracy would need to be analysed to determine if the non
linearity in the reading could be corrected in software or if an alternative current sensor
would be required.
There is a possibility that the start-up PWM mode could be eliminated as this would
allow faster rise time of the current waveform. Experiments would need to be conducted
to verify that the current overshoot was not excessive, especially at high current levels.
Three more single phase converters would need to be constructed in order to evaluate
the 4 phase converter operation with the load/drive motor. To evaluate a 4 phase
converter, the low voltage power supply and interface circuit would have to be
redesigned to facilitate of a four phase converter.
Further source code development would be required to implement the PWM current
control of the fourth phase using the auxiliary PWM module of the DSP evaluation
board. Software development would be required to implement and adapt control
strategies for operating the 4 phase SRM as a motor and a generator.
From the preliminary analysis it is better to operate the SRM at a higher speed in order
to reduce the losses during starting conditions and to have the majority of the generating
and regenerative braking occurring at a higher speed in order to maximise converter
efficiency.
179
Bibliography
(Aavid Thermalloy 2007a) Aavid Thermalloy, L.L.C., Hi-Contact Liquid-Cooled Cold
Plates – Hi-Contact - 6 Pass Models, P/N 41610U00000G, [Online], Available from:
http://www.aavidthermalloy.com/, [Accessed 20 June 2007].
(Aavid Thermalloy 2007b) Aavid Thermalloy, L.L.C., Turbo Fin Liquid Cold Plates,
[Online], Available from: http://www.aavidthermalloy.com/, [Accessed 14 April 2008].
(Adams 2004) Adams, J., Bootstrap Component Selection for Control IC’s,
International Rectifier Design Tip DT 98-2a, International Rectifier [Online]. Available
from: http://www.irf.com, [Accessed 6 Aug. 2004].
(Advanced Power Technology 2004) Power MOS 7 FREDFET, APT20M11JFLL Data

Sheet, Advanced Power Technology, www.advancedpower.com, 2004
(AEC 2006) AEC Council, AEC History. [Online], Available from:

www.aecouncil.com/AECHistory.html, [Accessed 12 Dec. 2006]..
(Agilent Technologies 1999a) Agilent Technologies, HCMOS Compatible, “High

CMR, 10 MBd Optocouplers”, HPCL-263A Data Sheet, Document No. 5968-1087E,
[Online], Available from: www.semiconductor.agilent.com, Nov. 1999, [Accessed 12
July 2005].
(Agilent Technologies 1999b) Common-Mode Noise: Sources and Solutions, App. Note
1043, Doc. No. 5965-5979E, [Online], Available from:
www.semiconductor.agilent.com, Nov. 1999, [Accessed 12 July 2005].
(Albertoni, L et al 2003) Albertoni, L., Balluchi A., Casavola A., Gambelli C., MoscaE.,
and Sangiovanni-Vincentelli A.L., Idle Speed Control for GDI Engines using Robust
Multirate Hybrid Command Governors. Proceedings of 2003 IEEE Conference on
Control Applications, CCA200, Vol. 1, Turkey, pages 140--145, June 2003.
180
(Analog Devices 2002) VisualDSP++ 3.0 C/C++ Compiler and Library Manual for
ADSP-219x DSPs, Revision 4.0, Analog Devices Inc., Norwood Mass. USA, July 2002
(Analog Devices 2003a) ADSP-21992: Mixed Signal DSP Controller with CAN, Data
Sheet, Rev. 0, Analog Devices Inc., Norwood Mass. USA, 2003
(Analog Devices 2003b) ADSP21992 EZ KIT Lite Manual, Revision 1.1, Analog
Devices Inc., Norwood Mass. USA, 2003
(Analog Devices 2003c) ADSP-2199x Mixed Signal DSP Controller Hardware

Reference, Preliminary Revision 0, 2003, Analog Devices Inc., Norwood Mass. USA,
2003
(Analog Devices 2003d) EE-188: Using C to implement Interrupt-Driven Systems On

ADP-219x DSPs, Analog Devices Inc., Norwood Mass. USA, March 2003
(Andreycak 1999) Andreycak, W., Practical Considerations in High Performance

Mosfet, IGBT and MCT Gate Drive Circuits, Unitrode Application Note U-137, Texas
Instruments Inc., Dallas, TX., 1999 [Online]. Available from:
http://focus.ti.com/lit/an/slua105/slua105.pdf, [Accessed 28 April 2005].
(Barkondarian 2000) Barkondarian, V., Power MOSFET Basics, International Rectifier

Technical Papers, [Online]. Available from: http://www.irf.com, [Accessed 6 Aug.
2004].
(Barnes 1998) Barnes M. and Pollock C., Power electronic converters for switched
reluctance drives, IEEE Transactions on Power Electronics, Vol. 13, No. 6, pp. 1100-
1111, 1998.
(Bass 1987) Bass J.T., Miller T.J.E., Ehsani M. and Steigerwald R.L., Development of a
unipolar converter for variable reluctance motor drives, IEEE Transactions on
Industrial Applications, Vol. 1A-23, pp. 545 - 553, 1987.
181
(Bedford 1972a) Bedford, B.A., Compatible permanent magnet or reluctance brushless
motors and controlled switch circuits, US Patent No. 3,678,352, 18th July 1972a.
(Bedford 1972b) Bedford, B.A., Compatible brushless reluctance motors and controlled
switch circuits, US Patent No. 3,679,953, 25th July 1972b.
(Bergquist 2005) The Bergquist Company, Sil-Pad@ 2000: Higher Performance, High
reliability Insulator, Data Sheet No. PDS_SP2000_0305E [Online]. Available from:
http://www.bergquistcompany.com/objects/data_sheets/PDS_SP2000_0904.pdf, [Accessed
25 April 2005].
(Bergquist 2004) Sil-Pad@ K10: Higher Performance, Kapton-Based Insulator, Data

Sheet No. PDS_10043_SP_K10_1104 [Online]. Available from:
http://www.bergquistcompany.com/objects/data_sheets/PDS_10043_SP_K10_1104.pdf
[Accessed 25 April 2005].
(Bertrand 2003) Liquid-Cooled Heatsinks Enable Smaller, Cooler Devices [Online].

Available from: http://powerelectronics.com/mag/power_liquidcooled_heatsinks_enable/,
June 2003, [Accessed 15 April 2005].
(Besnier 2004) Besnier, R., Current Sensors Enquiry, CRC-European Sales and
Technical Support-Sensing and Control [Online]. Available from:
http://www.honeywell.com/sensing/products, [Accessed 15 Dec. 2004].
(BHC 2004) BHC Components Ltd, Technical Data, [Online] Available from:
http://www.bhc.co.uk/electric.htm, [Accessed 20 September 2004].
(Bosch 2000) Bosch Product Overview and Catalogues - Alternators and Starter Motors
[Online], Available from: http://www.bosch.com.au, [Accessed 23 June 2003]
(Bosch 2008) Bosch generators product brochure, Robert Bosch GmbH [Online],
Available form: www.bosch.de/k, [Accessed 13 Feb 2008]
182
(Byrne 1973) John Byrne and James Lacy, Electrodynamic system comprising a
variable reluctance machine, Patent No. GB1321110, The Patent Office, London, 20
June 1973.
(Byrne 1976) Byrne J.V. and Lacy J.G., Characteristics of saturable stepper and
reluctance motors, Proceedings of the IEE Conference on Small Electrical Machines,
IEE Conference Publication No. 136, London, UK, pp. 93-96, 30 - 31 March, 1976.
(Byrne 1982) Byrne, J.V. and McMullin, M.F., Design of a reluctance motor as a
10kW spindle drive, Proceedings of MOTORCON '82 Europe, PCIM '82 Europe, the
4th International Intelligent Motion Conference, Geneva, Switzerland, pp 10 – 24, 28 -
30 September, 1982.
(Byrne 1985) Byrne J.V., McMullin M.F. and O' Dwyer J.B., A high performance
variable reluctance drive: a new brushless servo, Proceedings of MOTOR-CON '85,
SATECH '85, the 7th International Intelligent Motion Conference, Chicago, IL, USA,
pp. 147-160, 21 - 25 October, 1985.
(Campbell 2004) Campbell, R.J., Rajashekara, K., Evaluation of power devices for
automotive hybrid and 42V based systems, 2004 SAE World Congress, March 2004.
(Cardenas 1995) Cardenas R., Ray W.F. and Asher G.M., Switched reluctance
generators for wind energy applications, PESC '95 Record, 26th Annual IEEE Power
Electronics Specialists Conference, Atlanta, USA, Vol. 1, pp. 559-564, 18 - 22 June,
1995.
(Chapmore Controls Ltd. 2007) TracMotion2000 model RA165L Specifications

[Online]. Available from: http://www.chapmore.co.uk/Products/AMCT/ra165.htm, July
2000, [Accessed 6 Aug. 2004].
(Chappell 1984) Chappell, P.H., Ray W.F., and Blake, R.J., Microprocessor control of
a variable reluctance motor, IEE Proceedings, Part B, Electric Power Applications, Vol.
131, No. 2, pp 51 – 60, March, 1984.
183
(Clemente 1990) Clemente, S., and Dubashi, A., HV Floating MOS-Gate Driver ICs,
International Rectifier Application Note AN-978 [Online]. Available from:
http://www.irf.com, [Accessed 6 Aug. 2004].
(Corda 1991) Corda J. and Skopljak E., Four-phase switched reluctance drive using
pulse-width modulation control with four switches, Fifth International Conference on
Electrical Machines and Drives, 1991, Conf. Publ. No. 341, pp. 77 – 80, 11-13 Sep
1991.
(Cossar 2004) Cossar C. and Sawata T., Microprocessor controlled DC power supply
for the generator control unit of a future aircraft generator with a wide operating speed
range, Proceedings of PEMD '04, the 2nd International Conference on Power
Electronics, Machines and Drives, Edinburgh, UK, Vol. 2, pp. 458 – 463, 31 March - 2
April, 2004.
(De Doncker 1991) De Doncker R.W. and Lyons Jr. J.P., The auxiliary quasi-resonant
DC link inverter, 22nd Annual IEEE Power Electronics Specialists Conference, PESC
'91 Record, pp. 248 – 253, 24-27 June 1991.
(Delphi 2008) Delphi 42-Volt Belt Alternator Starter Stop-Start System Data sheet,
Delphi Corporation [Online], Available from: www.delphi.com, [Accessed 13 Feb
2008]
(Devitt et al 1981) Harris M.R., Bolton H.R., Ward P.A., Byrne J.V., Smith G.B.,
Merrett J., Devitt F., Paul R.J.A., Schwarz K.K., Mangan M.F., Anderson A.F., Bourne
R., Lawrenson P.J., Stephenson J.M., Fulton N.N., Davis R.M., and Ray W.F.,
Discussion on variable-speed switched-reluctance motor systems, IEE Proceedings,
Part B, Electric Power Applications, Imperial College of Science Technology and
Medicine, UK, Vol. 128, No. 5, pp 260 – 268, September, 1981.
(Dickinson 2002) Dickinson, R. and Milano, S. Isolated Open Loop Current Sensing
Using Hall Effect Technology in an optimised Magnetic Circuit, [Online] Available
from: http://www.allegromicro.com/en/Products/Design/current_sensing/bsp_v1_52.pdf, Nov
2002. [Accessed 18 Nov. 2004].
184
(DiRenzo 2000) DiRenzo M., Switched Reluctance Motor Control – Basic Operation
and Example using the TMS320F240, Application Report SPRA420A, February 2000
[Online]. Available from: http://focus.ti.com/lit/an/spra420a/spra420a.pdf, [Accessed 2
July 2007].
(Dorf 1998) Dorf, R., and Bishop, R., Modern Control Systems, 8th Ed., Menlo Park,
CA; Addison Wesley Longman, Inc, 1998.
(Dorf 1999) Dorf, R., and Svoboda, J., Introduction to Electronics Circuits, 4th Ed.,
New York, NY; John Wiley & Sons, Inc, 1999.
(Dossert 2004) Dossert Corporation, Flat Bus [Online] Available from:

http://www.dossert.com/technicalinfo/flat_bus.htm, [Accessed 25 May 2004].
(Dunn 2003) Dunn, J., Determining MOSFET Driver Needs for Motor Drive
Applications, Microchip Application Note AN898 (DS00898A) [Online]. Available
from: http://www.microchip.com, [Accessed 6 Aug. 2004].
(Dunn 2004) Dunn, J., Matching MOSFET Drivers to MOSFETs, Microchip

Application Note AN799 (DS00799B) [Online]. Available from:
http://www.microchip.com, [Accessed 6 Aug. 2004].
(DuPont 2005) Ceramic Solutions for Automotive [Online]. Available from:

http://www.dupont.com/mcm/cer-automotive.html, [Accessed 3 Feb 2005]
(Dyson 2007) Dyson Airblade™. [Online], Available from: www.dyson.co.uk, Dyson

Appliances, UK, 2007 [Accessed 11 Jan. 2007].
(Ehsani 2002) Ehsani M. and Fahimi B.F., Elimination of position sensors in switched
reluctance motor drives: state of the art and future trends, IEEE Transactions on
Industrial Electronics, Vol. 49, No. 1, pp. 40-47, 2002.
(Eldre 2004) Eldre Corporation, IGBT Information and Applications [Online] Available
from: http://www.busbar.com/power/igbt.asp, [Accessed 26 May 2004].
185
(Electrocomp 2003) A look at heat-management for electronics [Online]. Available
from: http://www.dataweek.co.za/news.asp?pklNewsID=9550&pklIssueID, [Accessed
12 April 2005].
(Epcos 2003) Epcos AG, Single-Ended Aluminium Electrolytic Capacitors for

Automotive Applications, Product Profile 2003 [Online] Available from:
http://www.epcos.com/web/publikationen/pdf/se_automotive_2003.pdf, [Accessed 20
September 2004].
(Epcos 2006) Epcos B41560 Series Compact 105oC Long Life Capacitors Data Sheet,
Epcos [Online], Available from: http://www.epcos.com, Dec. 2006, [Accessed 20 June
2007].
(Excelsys Technologies Ltd 2004) Test your power supply knowledge, [Online],
Available from: http://www.excelsys.com/newsletter/EN_Jan04/Jan04_quiz.htm, Feb
2004. [Accessed 9 Nov 2004].
(Fahimi 2001) Fahimi, B., A switched reluctance machine based starter/generator for
more electric cars, Proceedings of IEMDC 2001, IEEE International Electric Machines
and Drives Conference, Cambridge, MA, USA, pp. 73-78, 18 - 20 June, 2001.
(Fairchild 1993) Parallel Operation Of Semiconductor Switches - Application Note

7513 Rev. A1 [Online]. Available from: www.fairchildsemi.com, June 1993, [Accessed
6 Aug. 2004].
(Fairchild Semiconductor 1998) Suppressing MOSFET Gate Ringing in Converters:

Selection of a Gate Resistor, Application Bulletin AB-9, Rev. 1.0.2 [Online], Available
from: http://www.fairchildsemi.com, Jan. 2007, [Accessed 12 July 2005].
(Fairchild 1999) TO-220AB Package Dimensions Data sheet, Fairchild Semiconductor

[Online], Available from: http://www.fairchildsemi.com, Sept. 1999, [Accessed 15
April 2008].
(Fairchild 2002a) FDP060AN08A0 - Discrete Automotive N-Channel PowerTrench
MOSFET Data sheet, Fairchild Semiconductor [Online], Available from:
http://www.fairchildsemi.com, Nov. 2002, [Accessed 12 July 2005].
186
(Fairchild 2002b) RURG8060 Ultrafast Diode Data Sheet, Rev B, Fairchild
Semiconductor [Online], Available from: http://www.fairchildsemi.com, Jan. 2002,
[Accessed 4th April 2005].
(Fairchild 2002c) New Trench MOSFET Devices Provide Industry's Lowest RDS(ON)
and Exceptionally Low Gate Charge for 42V Automotive Applications [Online],
Fairchild Semiconductor Corp., Available from: http://www.fairchildsemi.com, May
2002, [Accessed 11 Jan. 2007].
(Fairchild 2003) FDH038AN08A1 - N-Channel PowerTrench; MOSFET Data sheet

Fairchild Semiconductor [Online], Available from: http://www.fairchildsemi.com, Feb.
2003, [Accessed 12 July 2005].
(Fairchild 2004) FDP047AN08A0 - 75V N-Channel PowerTrench MOSFET Data

Sheet, Fairchild Semiconductor [Online], Available from:
http://www.fairchildsemi.com, June 2004, [Accessed 12 July 2005].
(Fairchild 2007) 1N4148 Small Signal Diode Data Sheet, Fairchild Semiconductor
[Online], Available from: http://www.fairchildsemi.com, Jan. 2007, [Accessed 12 July
2005].
(Fenner 2006) HRC coupling – general purpose coupling with flexible rubber element
[Online]. Available from: http://www.fptgroup.com/products/hrcjaw.asp, [Accessed 23
Sept 2003]
(Fiedler 2005) Fiedler, J.O., Kasper, K.A. and De Doncker, R.W., Acoustic noise in
switched reluctance drives: an aerodynamic problem? 2005 IEEE International
Conference on Electric Machines and Drives, pp1275 – 1280, 15-18 May 2005.
(Forsythe 1981) Forsythe, J.B. Paralleling HEXFET power MOSFETS for higher
power output, Proc. Eighth International Solid-State Power Conversion Conference
(Powercon 8), Apr. 1981.
187
(Fowler 1989) Fowler, R.J. Electricity: Principles and Applications, 3rd Ed., New York,
NY; McGraw-Hill Book, Inc, 1989.
(Fuengwarodsakul 2005) Fuengwarodsakul, N.H., Fiedler, J.O., Bauer S.E. and De

Doncker, R.W., New methodology in sizing and predesign of switched reluctance
machines using normalized flux-linkage diagram, Proceedings of IAS 2005, 40th IEEE
Industry Applications Society Annual Meeting, Hong Kong, China, Vol. 4, pp 2704 –
2711, October, 2005.
(Gauen 1984) Gauen, K. Paralleling power MOSFETs in switching applications,

Motorola App. Note AN-918 [Online], Available from: http://www.onsemi.com, 1984,
[Accessed 6 Aug. 2004].
(Gilbert 2002) Gilbert, J. and Dewey, R., Application Note 27702A, Linear Hall-Effect
Sensors, Allegro MicroSystems, Inc., NH, [Online] Available from:
http://www.allegromicro.com/en/Products/Design/an/an27702.pdf, 2002, [Accessed 18
Nov. 2004].
(Graf 2001) Graf, A., Semiconductors in the 42V PowerNet, Intertech, 42V Automotive
Systems Conference, Sept. 2001.
(Hanley 2003) David Hanley, Why choose laminated bus-bar assemblies? [Online]
Available from: http://www.electronicproducts.com/, [Accessed 26 May 2004].
(Hart 1997) Hart, D., Introduction to Power Electronics, Upper Saddle River, NJ,
Prentice Hall, Inc., 1997.
(Hava 1992) Hava A.M., Blasko V. and Lipo T.A., A modified C-dump converter for
variable-reluctance machines, IEEE Transactions on Industry Applications,
Vol. 28, Issue 5, pp. 1017 – 1022, Sept.-Oct. 1992.
(Hill 2004) An Introduction to Heatsinks and Cooling ( Ver. 1.01) [Online]. Available
from: http://homepages.which.net/~paul.hills/Heatsinks/HeatsinksBody.html, [Accessed 07
September 2004].
188
(Hill Technical Sales 2004) Hill Technical Sales, Laminated Bus Structures [Online].
Available from: http://www.hilltech.com/lambus.html, 2004. [Accessed 26 May 2004].
(HS Marston 2001) Ultra-Fin High Density Cooling, P/N 890SP-020000-A-100

[Online]. Available from: http://www.farnell.com/datasheets/48732.pdf, [Accessed 25
April 2005].
(Husain 1996) Husain I., Indirect rotor-position estimation techniques for switched
reluctance motors - a review, Electromotion, Vol. 3, No. 2, pp. 94-102, 1996.
(Husain 2002) Husain I., Minimization of torque ripple in SRM drives, IEEE
Transactions on Industrial Electronics, Vol. 49, No. 1, pp. 28-39, 2002.
(Hussain 2002) Hussain, A., Driving Power MOSFETs in High Current, Switch Mode
Regulators, Microchip Application Note AN786 (DS00786A) [Online]. Available from:
http://www.microchip.com, [Accessed 6 Aug. 2004].
(IEEE 2006) IEEE, Milestone unveiled for Callan, IEEE Region8 News, Vol. 9 No. 4,
Dec. 2006.
(Infineon 2003) SPP80N08S2L-07, Low Voltage OptiMOS MOSFET Data sheet,

Infineon Technologies [Online], Available from: http://www.infineon.com, May. 2003,
[Accessed 12 July 2005].
(International Rectifier 1997) 80EPF02 Fast Recovery Diode Data Sheet, International
Rectifier [Online], Available from www.irf.com, June 1997, [Accessed 12 July 2005].
(International Rectifier 2004a) Paralleling HEXFET power MOSFETS, International

Rectifier Application Note AN-941 (v.Int) [Online]. Available from www.irf.com,
(International Rectifier 2004b) HV Floating MOS-Gate Driver ICs - Application Note

AN-978 [Online], Available from www.irf.com, [Accessed 6 Aug. 2004].
189
(International Rectifier 2004c) Current Rating of Power Semiconductors, International
Rectifier Application Note AN-949 (v.Int), [Online]. Available from:
http://www.irf.com, [Accessed 6 Aug. 2004].
(International Rectifier 2004d) HEXFET Power MOSFET, IRFD9110 Date Sheet,

International Rectifier [Online]. Available from: http://www.irf.com, [Accessed 6 Aug.
2004].
(International Rectifier 2004e) HEXFET Power MOSFET, IRFD110 Date Sheet,

International Rectifier[Online]. Available from: http://www.irf.com, [Accessed 6 Aug.
2004].
(International Rectifier 2005a) “Current Rating of Power Semiconductors, [Online],

International Rectifier Application Note AN-949 (v.Int), Available from:
http://www.irf.com/technical-info/appnotes/an-949.pdf, [Accessed 4 March 2005].
(International Rectifier 2005b) International Rectifier, High Side Low Side Driver
IR2110 Data Sheet No. PD60147 Rev.U [Online]. Available from: http://www.irf.com,
[Accessed 20 June 2007].
(IXYS 2003) DSEP2x101-04A HiPerFREDTM Epitaxial Diode Data Sheet, IXYS

[Online], Available from www.ixys.com, 2003, [Accessed 12 July 2005].
(Jones 2003) Jones, W.D., Hybrids to the rescue [hybrid electric vehicles],
IEEE Spectrum, Vol. 40, Issue 1, Jan. 2003, pp70 – 71
(Kassakian 1996) Kassakian, J.G.; Wolf, H.-C.; Miller, J.M.; Hurton, C.J., Automotive
electrical systems circa 2005, IEEE Spectrum, Vol. 33, Issue 8, Aug. 1996, pp 22 – 27
(Kassakian 2000) Kassakian, J.G., Automotive electrical systems-the power electronics

market of the future, Fifteenth Annual IEEE Applied Power Electronics Conference and
Exposition, APEC 2000, Vol. 1, 6-10 Feb. 2000, pp3 - 9
190
(Keville 1994) Keville B., Switched reluctance motor for automotive applications,
M.Eng.Sc. Thesis, The National University of Ireland, University College, Dublin,
1994.
(Kiraly 2004) Kiraly, L., High Current Buffer for Control IC’s, International Rectifier
Design Tips DT 92-2A, [Online]. Available from: http://www.irf.com, [Accessed 6 Aug.
2004].
(Kjaer 1994) Kjær P.C., Cossar C., Gribble J.J., Li Y. and Miller T.J.E., Switched
reluctance generator control using inverse machine model, Proceedings of ICEM '94,
International Conference on Electrical Machines, Paris, France, Vol. 2, pp. 380-385,
5 - 8 September, 1994.
(Kokernak 1999) Kokernak J.M., Torrey D.A. and Kaplan M., A switched reluctance
starter/alternator for hybrid electric vehicles, Proceedings of INTELLIGENT MOTION
'99, Powersystems World '99, 40th International Intelligent Motion Conference,
Chicago, IL, USA, pp. 74-80, 7 - 11 November, 1999.
(Krein 1998) Krein, P., Elements of Power Electronics, New York, NY; Oxford
University Press, Inc, 1998.
(Krishnamurthy 2005) Krishnamurthy M., Edrington C.S. and Fahimi B., Comparison
of Various Converter Topologies for Bipolar Switched Reluctance Motor Drives, 2005
IEEE 36th Conference on Power Electronics Specialists, pp. 1858 – 1864, 11-14 Sept.
2005.
(Krishnan 1990) Krishnan R. and Materu P., Design of a single-switch-per-phase

converter for switched reluctance motor drives, IEEE Transactions on Industry
Applications, Vol. 37, Issue 6, pp. 469 – 476, Dec. 1990.
(Krishnan 2001) Krishnan R., Switched reluctance motor drives: modeling, simulation,
analysis, design and applications, Industrial Electronics series, CRC Press LLC, ISBN:
0-8493-0838-0, 2001.
191
(Lander 1981) Lander, C. W., Power Electronics, Berkshire, UK, McGraw-Hill, 1981.
(Lawrenson 1980) Lawrenson P.J., Stephenson J.M., Blenkinsop P.T., Corda J. and
Fulton N.N., Variable-speed switched reluctance motors, IEE Proceedings, Part B,
Electric Power Applications, Vol. 127, No. 4, pp. 253-265, July, 1980.
(Le-Huy 1989) Le-Huy H., Viarouge P. and Francoeur B., Unipolar converters for
switched reluctance motors, Conference Record of the 1989 IEEE Industry Applications
Society Annual Meeting, Vol. 1, pp. 551 – 560, 1-5 Oct. 1989.
(Le-Huy 1990) Le-Huy H., Viarouge P. and Francoeur B., A novel unipolar converter
for switched reluctance motor, IEEE Transactions on Power Electronics, Vol. 5, Issue
4, pp. 469 – 475, Oct. 1990.
(LEM Components 2003) LEM Components, Closed Loop Transducers with Small
Footprint up to 100 A Nominal, Technical Document No. CH99102 aE/US [Online].
Available from: www.lem.com, Feb 2003, [Accessed 26 May 2005].
(LEM Components 2004) Isolated current and voltage transducers: Characteristics –

Applications – Calculations (3rd Ed.) [Online]. Available from:
http://www.lem.com/images/stories/files/Products/P1_5_1_industry/CH24101E.pdf,
2004, [Accessed 18 Nov 2004].
(LEM Components 2005) LEM Components, Current Transducer LA 100-P/SP13 Data

Sheet [Online]. Available from: www.lem.com, [Accessed 31 May 2005].
(Lepkowski 2003) Lepkowski, J., Motor Control Sensor Feedback Circuits, Microchip
Application Note AN894 (DS00894A) [Online]. Available from:
http://ww1.microchip.com/downloads/en/AppNotes/00894a.pdf, 2003, [Accessed 18
Nov 2004].
(Liang 1994) Liang F., Liao Y. and Lipo, T.A., A new variable reluctance motor
utilizing an auxiliary commutation winding, IEEE Transactions on Industry
Applications, Vol. 30, Issue 2, pp. 423 – 432, March-April 1994.
192
(MacMinn 1989) MacMinn S.R and Sember J.W., Control of a switched reluctance
aircraft starter-generator over a very wide speed range, Proceedings of IECEC '89,
IEEE 24th Intersociety Energy Conversion Engineering Conference, Washington DC,
USA, Vol. 1, pp. 631-638, 6 - 11 August, 1989.
(Melexis 1997) Melexis Microelectronics. Hall Applications Guide, [Online] Available

from: http://www.melexis.com, 1997, [Accessed 18 Nov 2004].
(Mese 2000) Mese E., Sozer Y., Kokernak J.M. and Torrey D.A., Optimal excitation of
high speed switched reluctance generators, Proceedings of APEC 2000, IEEE 15th
Annual Applied Power Electronics Conference and Exposition, New Orleans, LA, USA,
Vol. 1, pp. 362-368, 6 - 10 February, 2000.
(Microsemi 2004) APT2X101DQ60J Ultrafast Soft Recovery Rectifier Diode Data

Sheet, 053-4208 Rev A, Microsemi Power Products [Online], Available from:
http://www.microsemi.com, Dec. 2004, [Accessed 4th April 2005].
(Miller 1985) Miller T.J.E., Converter volt-ampere requirements of the switched

reluctance drive, IEEE Transactions on Industry Applications, Vol. IA-21, No. 5, pp.
1136-1144, 1985.
(Miller 1990) Miller T.J.E. and McGilp M.I., Nonlinear theory of the switched
reluctance motor for rapid computer-aided design, IEE Proceedings, Part B, Electric
Power Applications, Vol. 137, No. 6, pp. 337-347, November, 1990.
(Miller 1993) Miller T.J.E., Switched Reluctance Machines and their Control, Magna
Physics Publishing and Clarendon Press, Oxford, U.K, 1993.
(Miller 2001) Miller T.J.E., Electronic control of switched reluctance machines,

Newnes Power Engineering series, Reed Educational and Professional Publishing Ltd,
ISBN: 750650737, June, 2001.
(Miller 2002) Miller T.J.E., Optimal design of switched reluctance motors, IEEE
Transactions on Industrial Electronics, Vol. 49, No. 1, pp. 15-27, 2002.
193
(Mir 1997) Mir S., Husain I. and Elbuluk M.E., Energy-efficient C-dump converters for
switched reluctance motors, IEEE Transactions on Power Electronics, Vol. 12, Issue
5, pp. 912 – 921, Sept. 1997.
(Mohan 1995) Mohan N., Undeland T.M. and Robbins W.P., Power Electronics -
Converters, Applications and Design, Second Edition, John Wiley & Sons, Inc., 1995,
ISBN 0-471-30576-6.
(Motion System Tech 2002) Switched Reluctance Motor Specification, RA165217 Data
Sheet No. SS0002-RA165217a, Japan, June 2002.
(Motorsoft 2007) Turn key prototyping services for all types of electric machines
[Online]. Available from: http://www.motorsoft.com/about.htm, [Accessed 17 Aug.
2006]
(Murphy 2002) Murphy A.J., WP5n – Hardware description, PEI Technologies, Dublin
City University, Ireland, 2002.
(Murray 2002) Murray, C. J., Integrated starter-alternator next on automakers’ wish list,
EE Times [Online], Available from: http://www.eetimes.com/story/OEG20021218S0027,
2002, [Accessed 17 June 2003].
(Murray 2001) Murray A.F.J., Wood, P., Keskar, N., Chen J.,& Guerra A., A 42V
Inverter/Rectifier for ISA using Discrete Semiconductor Components, Future
Transportation Technology Conference, August 2001.
(Myers 2003) Myers, B.A, Cooling Issues for Automotive Electronics, [Online],
Available from: http://www.electronics-cooling.com, ElectronicsCooling Online, Aug 2003,
[Accessed 3 Feb 2005].
(Nathan 2006) Nathan, S., Dyson's quick on the dry, The Engineer Online [Online],
Available from: www.e4engineering.com, Nov. 2006, [Accessed 4 Jan. 2007].
(Ness 2006) RMS (Root Mean Square) and Average values of typical current
waveforms [Online], Available from: http://home.sun.rr.com, March 2006, [Acessed 15
Aug. 2006].
194
(Nichicon 2004) Application Guidelines for Aluminium Electrolytic Capacitors,
[Online] Available from: http://www.nichicon-us.com/english/seihin/alm_larg/param_f.htm,
[Accessed 20 September 2004].
(ON Semiconductor 2005) SWITCHMODE Power Rectifiers, MUR120 Data Sheet No.
MUR120/D, Rev. 9 [Online]. Available from: http://onsemi.com, [Accessed April
2005]
(Park 1992) Park S.S. and Lipo T.A., New series resonant converter for variable
reluctance motor drive, 23rd Annual IEEE Power Electronics Specialists Conference,
Toledo, Spain, PESC '92 Record. Vol. 2, pp 833 - 838 29 June - 3 July, 1992.
(Pelly 2004) Pelly, B.R. Choosing between multiple discretes and high current modules
- Design Tip DT 94-2 [Online], Available from: http://www.irf.com, [Accessed 6 Aug.
2004]
(Philips 1999) PSMN009-100W; N-channel TrenchMOS (tm) transistor Data sheet,

Philips Semiconductors [Online], Available from: http://www.nxp.com, Oct. 1999,
[Accessed 12 July 2005]
(Philips 2002) BUK75/7606-75B - TrenchMOS (tm) standard level FET Data sheet,
Philips Semiconductors [Online], Available from: http://www.nxp.com, Sept. 2002,
[Accessed 12 July 2005]
(Pollock 1990) Pollock C. and Williams B.W., A unipolar converter for a switched
reluctance motor, IEEE Transactions on Industry Applications, Vol. 26, Issue 2, pp.
222 – 228, March-April 1990.
(Polyfet 2005) Power RF MOSFET Transistors [Online]. Available from:

www.polyfet.com/-rftopics.pdf, [Accessed 07 September 2004]
195
(Power Designers 2004) Power Designers, Electrolytic Capacitors [Online] Available
from: http://www.powerdesigners.com/InfoWeb/design_center/Design_Tips/Electrolytics/Caps.shtm,
[Accessed 26 May 2004].
(Radun 1994) Radun A.V., Generating with the switched reluctance motor, Proceedings
of APEC '94, IEEE 9th Applied Power Electronics Conference and Exposition, Vol. 1,
pp. 41-47, 13 - 17 February, 1994.
(Radun 1998) Radun A.V., Ferreira C.A. and Richter E., Two-channel switched
reluctance starter/generator results, , IEEE Transactions on Industry Applications, Vol.
34, Issue 5, pp. 1026 – 1034, Sept. – Oct. 1998.
(Ray 1979) Ray W.F. and Davis R.M., Inverter drive for doubly salient reluctance
motor: its fundamental behaviour, linear analysis and cost implications, IEE Journal on
Electrical Power Applications, Vol. 2, No. 6, pp. 185-193, December, 1979.
(Rundle 2002) The Coming 42-Volt Electrical System, Electrical Rebuilder’s Exchange
[Online], Available from: http://lees.mit.edu/public/In_the_News/Electrical+Rebuilder's+Exchange.pdf,
March 2002, [Accessed 17 June 2003].
(Service Tech Magazine 2001), 42-V Update, [Online], Available at:

http://www.sts.sae.org/servicetech/sep-2001/42-v.pdf, Service Tech Magazine 2001,
[Accessed 24 June 2003].
(Siemens Matsushita 2004) Applications – DC Capacitors [Online] Available from:

www.epcos.com/inf/20/50/db/power_99/00170019.pdf, 2004. [Accessed 16 September
2004].
(Sood 1992) Sood P., Power converter for a switched reluctance motor, US Patent No.
5,115,181, 19th May 1992.
(Soule 2005) The Benefits of Liquid Cooling over Air Cooling for Power Electronics
[Online]. Available from: http://www.icepak.com/prod/icepak/solutions/articles/-iceart19.htm,
[Accessed 15 April 2005].
196
(States Department of the Interior 2000) United States Department of the Interior,
Electrical Connections for Power Circuits [Online] Available from:
http://www.usbr.gov/power/data/fist/fist3_3/vol3-3.pdf, August 2000. [Accessed 23
September 2004].
(SR Drives Ltd 2007a) SR Drive motor/generator for mild hybrid-electric powertrain,
SR Drives Ltd,.[Online], Available from: http://www.srdrives.com/hybrid-
powertrain.shtml, [Accessed 9 Jan. 2007].
(SR Drives Ltd 2007a) SR Drive starter-generator system reduces emissions and
improves fuel economy, SR Drives Ltd, [Online], Available from:
http://www.srdrives.com/starter-generator.shtml, [Accessed 9 Jan. 2007].
(Stephenson 1989) Stephenson M. and El-Khazendar M.A., Saturation in doubly salient

reluctance motors, IEE Proceedings, Part B, Electric Power Applications, Vol. 136, No.
1, pp. 50-58, January, 1989.
(STMicroelectronics 2001) TD340 - H-Bridge Quad Power MOSFET Driver for DC

Motor Control [Online], Available from http://www.st.com/stonline/books/pdf/docs/7048.pdf,
Dec 2001, pp 8 [Accessed 4 Nov 2004].
(STMicroelectronics 2004) STTH20002TV Turbo 2 Ultra Fast High Voltage Rectifier

Data Sheet, STMicroelectronics [Online], Available from www.st.com, July 2004,
[Accessed 12 July 2005].
(Supertex 2001) Calculating Power Dissipation and Supply Current in HV91 Series
Parts, HV91 Series Application Note AN-H21, Supertex Inc., Sunnyvale, CA, Dec.
2001[Online]. Available from: http://www.supertex.com/pdf/app_notes/AN-H21.pdf ,
(Sweeney 2002) Sweeney, L. Technical Guidance, PEI Technologies Ireland, DCU,

2002
197
(Teratani et al 2003) Teratani, T., Kuramochi, K., Nakao, H., Tachibana, T., Yagi,
K., Abou, S., Toyota Motor Corp., Aichi, Japan;, Development of Toyota Mild Hybrid
System (THS-M) with 42V PowerNet, IEMDC'03., IEEE International Electric
Machines and Drives Conference, Wisconsin, USA, Vol. 1, pp 3-10, 1-4 June 2003.
(Terco 2003) Digital Torque Meter System MV 1040 Technical Specification and
Operation Manual, TERCO AB, Stockholm, Sweden, 2003
(Tillmann Steinbrecher 2005) The Heatsink Guide - All about PC cooling [Online].
Available from: http://www.heatsink-guide.com/heatsinkinfo.shtml, [Accessed 12 April
2005].
(Torrey 1993) Torrey D.A., Variable-reluctance generators in wind energy systems,

PESC '93 Record, 24th Annual IEEE Power Electronics Specialists Conference, Seattle,
WA, USA, pp. 561-567, 20 - 24 June, 1993.
(Torrey 2002) Torrey D.A., Switched reluctance generators and their control, IEEE
Transactions on Industrial Electronics, Vol. 49, No. 1, pp. 3-14, Feb. 2002.
(Vishay 2004) Aluminium Capacitors, Document Number: 28356 [Online] Available

from: http://www.vishay.com/docs/28356/intro.pdf, March 2004. [Accessed 20
September 2004].
(Visteon 2003) Press release archive, Visteon SpeedStart12™ Improves Fuel Economy,
Reduces Emissions, Visteon Corporation [Online], Available from: www.visteon.com,
July 2003, [Accessed 13 Feb 2008]
(Visteon 2005a) Series A Starter Motor Data sheet, Visteon Corporation [Online],
Available from: www.visteon.com, July 2005, [Accessed 13 Feb 2008]
(Visteon 2005b) Series LXB Starter Motor Data sheet, Visteon Corporation [Online],
(Visteon 2005c) Series LXD Starter Motor Data sheet, Visteon Corporation [Online],
198
(Vukosavic 1990) Vukosavic S.N. and Stefanovic V.R., SRM inverter topologies: a
comparative evaluation, The 1990 IEEE Industry Applications Society 25th Annual
Meeting, Seattle, Washington, USA, Part II, Conference Record of IAS '90, pp. 946-
958, 7 - 12 October, 1990.
(Wakefield 2005) Introduction to Thermal Management, Wakefield Thermal Solutions

[Online]. Available from: http://www.wakefield.com, [Accessed 12 April 2005]
(Wakefield 2008) Silicone oil-based thermal joint compound, Wakefield Thermal

Solutions catalogue [Online]. Available from www.wakefield.com, [Accessed 27 April
2008]
(Warth 2004) K200: High Performance Thermally Conductive Interface Material

[Online]. Available from: http://www.warth.co.uk/site2003/pdf/thermal/K200.pdf,
2004, [Accessed 25 April 2005]
(Williams 1987) Williams, B.W., Power Electronics: devices, drivers and applications,
London, UK, Macmillan Education, Ltd., 1987
(Williams 2004) Williams, A., Power on the move, IEE Review, Vol. 50, Issue 5, May
2004 pp38 – 41
(Winters 2002) Winters, T., Design basics for Power Circuits [Online]. Available from:
www.eetasia.com/ARTICLES/2002APR/2002APR08_ICD_RFD_CT_POW_TAC.PDF, [Accessed 28
April 2005]
199
Appendix A – Power Analysis Derivations
The detailed analysis provided in the following sections presents the relevant equations
and parameter values that are used to implement the power analysis for each component
in both the asymmetrical half-bridge converter and full-bridge synchronous rectification
converter and to determine the overall converter efficiency. The analysis is divided into
three specific ISG modes of operation: (a) starting and low speed motoring, (b) low
speed generating and (c) high -speed generating. A number of equations are defined for
the current, voltage and power calculations and these equations are used throughout this
appendix. A method is presented to determine the efficiency for a single phase converter
phase in each ISG mode of operation. The current and voltage waveforms are presented
for each component (Q1, Q2, D1 and D1) of the asymmetrical half-bridge converter and
the relevant equations and parameter values are then selected for at each ISG mode of
operation. The same procedure is implemented for each component (Q1, Q2, Q3 and
Q4) of the full-bridge synchronous rectification converter.
A.1 Analysis Equations

The voltage and current waveforms are assumed to be periodic during the converter
operation. The specific current and voltage waveforms for each switching device in both
the half-bridge and full-bridge converters configurations are illustrated in Sections A.3.
i(t)
Im
DT t
T
Figure A.1: Periodic square waveform
The average value of current for the periodic square wave shown in Figure A.1 is
determined from the equation (Hart 1997):
T
1
I AVG = ∫ i (t ) dt = I m D (A.1)
T 0
A1
The RMS value of current for the periodic square wave is determined from the equation
(Hart 1997):
T
1 2
T ∫0
I RMS = i (t )dt = I m D (A.2)
i(t)
Im
DT t
T
Figure A.2: Periodic triangular waveform
The average value of current for the periodic triangular wave shown in Figure A.2 is
determined from the equation (Ness 2006):
T
1
I AVG = ∫ i (t ) dt = 0.5I m D (A.3)
T 0
The RMS value of current for the periodic triangular wave is determined from the
equation (Ness 2006):
T
1 2 D
T ∫0
I RMS = i (t )dt = I M (A.4)
3
The conduction power losses for the MOSFET are determined by the equation
(Williams 1987):
Pc = ( I RMS ) 2 Rds ( ON ) HOT (A.5)
The hot on-resistance (RDS(ON)HOT) is the MOSFET drain to source resistance at the
maximum allowable junction temperature.
A2
i(t)
v(t)
Im Vm
tf t
Figure A.3: Typical voltage and current at turn-off switching transition for an inductive
load. Current and voltage are interchanged at turn-on
The MOSFET switching transition power loss is determined using the equation for an
inductive load as shown in Figure A.3 (Williams 1987):
Vm I m
Ps = (t r + t f ) f s (A.6)
2
Vm = VDS = VS (the applied voltage) for Q1 and Q2 of both the half-bridge and full-bridge
converter, as shown in Figure A.9 (page A15). Vm = VD = VSD (the forward voltage of
the intrinsic diode) for Q3 and Q4 of the full-bridge converter, as shown in Figure A.22
(page A28). Im equals the maximum phase current, while tr and tf are the worst case rise
and fall times, respectively.
The conduction losses for the diode are determined using the equation (Williams 1987):
Pc = I m VD D (A.7)
Where VD = VF (the forward voltage of the diode) and Im is the peak current and D the
duty cycle of the current, an example of this is shown in Figure A.9 (page A15).
The switching losses for the diode are determined using the equation (Williams 1987):
Ps = Qrr VR f s (A.8)
Where, Qrr is the reverse recovery charge, VR = -VD = VS, the reverse voltage applied to
the diode just after turn-off and fs is the switching frequency.
A3
The total power loss for a single phase is the sum of all the component losses, both
conducting and switching losses, and is determined by the equation:
PT ( SP ) = PT (Q1) + PT (Q 2 ) + PT ( D1) + PT ( D 2) (A.9)
A.2 Converter Efficiency and Power Transfer

The efficiency of a power converter is defined as the ratio of the output power to the
output power plus losses (Hart 1997). In this application the converter transfers power
in both directions (motoring and generating) so the output power is defined as the phase
power.
Pphase
η= (A.10)
Pphase + PT ( SP )
During magnetisation the power is transferred from the battery voltage source to the
phase winding while during demagnetisation and generating the power transfer is from
the phase winding to the battery. Power losses occur in the converter, PT(SP), during all
stages of operation, so the total output power of the converter is the sum of all power
transfers, Pphase.
The current and voltage waveforms for the phase winding of the SRM are nonsinusoidal
periodic waveforms as shown in Figures A.4 (page A6), A5 (page A8) and A.6 (page
A10). The current and voltage waveforms can be represented by the Fourier series as
shown in Equations A.11 and A.12 (Hart 1997).
∞
v (t ) = V 0 + ∑V
n =1
n cos(nω 0 t + θ n ) (A.11)
∞
i (t ) = I 0 + ∑I
n =1
n cos(nω 0 t + φ n ) (A.12)
The average power is determined using equation A.13
T
1
T ∫0
P = v(t )i (t )dt (A.13)
A4
The average power for nonsinusoidal periodic voltage and current waveforms is
expressed as
∞
 Vn ,max I n ,max 
P = V0 I 0 + ∑ 
n =1  2
 cos(θ n − φ n ) (A.14)

The total average power is the sum of the powers at the frequencies in the Fourier series.
For practical analysis in order to determine the worst case power losses, the idealized
periodic voltage waveforms are constructed from a sequence of square waves, while the
idealized periodic current waveforms are constructed from a sequence of square waves
and triangular waves. The average power losses were analysed for three specific ISG
modes of operation: (a) starting and low speed motoring, (b) low speed generating and
(c) high-speed generating.
The power transfer for the converter is assumed to be positive when it is from the
battery to the phase winding, while during demagnetisation and generating the power
transfer is assumed to be negative. The power transfers in either direction contribute to
power losses by the converter components so the magnitude of all the power transfers is
assumed to contribute to the output power of the converter. In order to determine the
total power transfer for the converter, the power transfer during generating and
demagnetisation is multiplied by -1.
A5
A.2.1 Starting and Low Speed Motoring
The calculation of the average power during starting and low speed motoring is divided
into two regions, (I) magnetisation and (II) demagnetisation as indicated in Figure A.4.
The phase winding voltage is assumed to be square waves during the magnetisation
region and constant during the demagnetisation region, while the current is assumed to
be constant during magnetisation and a triangular wave shape in the demagnetisation
region.
L Stroke or Conduction Period
Lmax
Lmin
θon θ1 θ2 θ3 θ4 θs θ
+VS
Vphase
t
I II
-VS
Im
Iphase
t1 t2
t0=0 D1TS t3 = TS t
D2TS
TS
Figure A.4: Phase winding voltage and current waveforms during starting and low
speed motoring
The time corresponding to the turn-on angle (θon) is t0 = 0, the turn-off angle at the
alignment angle (θ2) occurs at t1, the end of demagnetisation occurs at t2 and the end of
the stroke (θs) is at t3 = TS. The value of Ts is determined from the speed of the SRM.
The duty cycle for the magnetisation region (I) is D1 and the duty cycle for the
demagnetisation region (II) is D2. The PWM square wave voltage waveform has a
switching frequency of 25 kHz and has an average duty cycle of 50% as described in
Chapter 4, Section 4.1.4.1. This yields an overall duty cycle for the voltage square wave
of 0.5D1.
A6
The equation for the average power (P) over the stroke during starting and low speed
motoring is defined as:
TS
1
P =
TS ∫ v(t )i(t )dt
0
(A.15)
1  t1 t2 t3

 v(t )i (t )dt + v(t )i (t ) dt + v(t )i (t )dt 
 t∫ ∫ ∫
P= (A.16)
TS 
0 t1 t2 
During magnetisation (I) the interval t0 – t1, the current is constant (Im) and the voltage
square wave has a peak value of VS as shown in Figure A.4 The equation for the average
power during magnetisation (PM) is defined as:
t1 D1TS
1 0.5I m
PM =
TS ∫t v(t )i(t )dt = TS ∫ v(t )dt = 0.5I
0
m VS D1 (A.17)
0
During demagnetisation (II) in the interval t1 – t2, the voltage is constant (-VS) and the
current is approximated by a triangular wave and has a peak value of Im as shown in
Figure A.4. The equation for the average power during demagnetisation (PDM) is:
D2TS + D1TS
− VS
t2
1
PDM =
TS ∫t v(t )i(t )dt = TS ∫ i(t )dt = − 0.5V
D1TS
S I m D2 (A.18)
1
demagnetisation is:
PDM = − 1( − 0.5VS I m D2 ) = 0.5VS I m D2 (A.19)
The average power loss during the interval t2 – t3 is zero as the voltage and current is
zero during the interval. Therefore, the average power (P) over the stroke during
starting and low speed motoring is defined as:
P = 0.5 I mVS D1 + 0.5I mV S D2 (A.20)
A.2.2 Low Speed Generating

The calculation of the average power during low speed generation is divided into two
regions, (I) magnetisation and generating and (II) demagnetisation as indicated in Figure
A7
A.5. The phase winding voltage waveform is assumed to be square waves during the
magnetisation and generating regions and constant during the magnetisation
demagnetisation region, while the current is assumed to be constant during
magnetisation and generating and a triangular wave during demagnetisation.
Stroke or Conduction Period

L
Lmax
Lmin
θ1 θ2 θon θ3 θ4 θs θ
+VS
Vphase
t
-VS
I II
iexc igen
Im
Iphase
t1 t2
t0=0 D1TS t3 = TS t
D2TS
TS
Figure A.5: Phase winding voltage and current waveforms during low generating
The time corresponding to the turn-on angle (θon) is t0 = 0, the turn-off angle at the
unaligned angle (θ4) occurs is t1, the end of demagnetisation occurs at t2 and the end of
the stroke (θs) is at t3 = TS. The value of Ts is determined from the speed of the SRM.
The duty cycle for the magnetisation and generating region during the interval t0 – t1 is
assumed to be the same at D1 and the duty cycle for demagnetisation during the interval
t1 – t2 is D2. The PWM square wave voltage waveform has a switching frequency of 25
kHz and has an average duty cycle of 50% as described in Chapter 4, Section 4.1.4.2.
This yields an overall duty cycle for the voltage square wave of 0.5D1 for magnetisation
and 0.5D1 for generating.
The equation for the average power (P) over the stroke during low speed generating is
defined as:
A8
1  t1 t2 t3

 v(t )i (t )dt + v(t )i (t ) dt + v(t )i (t )dt 
 t∫ ∫ ∫
P= (A.21)
TS 
0 t1 t2 
During magnetisation in the interval t0 – t1, the current is assumed to be approximately

constant (Im) and the voltage is a PWM square wave with a peak value of Vs and a duty
cycle of D1 as indicated in Figure A.5. The average power during demagnetisation (PM)
is:
t1 D1TS
1 0.5I m
PM =
TS ∫t v(t )i(t )dt = TS ∫ v(t )dt = 0.5I
0
m VS D1 (A.22)
0
During generating in the interval t0 – t1, the current is assumed to be approximately

constant (Im) and the voltage square wave with a peak value of -Vs and a duty cycle of
D1 as indicated in Figure A.5. The average power during generating (PG) is:
t D1TS
1 1 0. 5 I m
PG = ∫
TS t0
v(t )i (t )dt =
TS ∫ v ( t ) d t = − 0. 5 I
0
mVS D1 (A.23)
generating (PG) is:
PG = − 1( − 0.5 I mVS D1 ) = 0.5 I mVS D1 (A.24)
During demagnetisation in the interval t1 – t2, the voltage is assumed to be constant (-Vs)
during the interval and the current is approximated by a triangular wave with a peak
value of Im as indicated in Figure A.5. The average power during demagnetisation is:
D2TS + D1TS
− VS
t
1 2
PDM = ∫
TS t1
v(t )i (t )dt =
TS ∫ i(t )dt = − 0.5V
D1TS
S I m D2 (A.25)
demagnetisation is:
PDM = − 1( − 0.5VS I m D2 ) = 0.5VS I m D2 (A.26)
A9
zero during the interval. Therefore, the average power (P) over the stroke during low
speed generating is:
P = I mVS D1 + 0.5I mVS D2 (A.27)
A.2.3 High Speed Generating

The calculation of average power during high speed generation is divided into three
regions, (I) magnetisation, (II) generating and (III) demagnetisation as indicated in
Figure A.6. The phase voltage waveform is assumed to be constant during all regions,
while the current waveform is a triangular wave for both magnetisation and
demagnetisation and is a constant for the generating region.
Figure A.6: Phase winding voltage and current waveforms during high speed
generating
The time corresponding to the turn-on angle (θon) is t0 = 0, the magnetisation turn-off
angle (θoff) occurs is t1, the end of generating occurs at the rotor and stator pole overlap
(θ4) occurs at t2, the end of demagnetisation occurs at t3 and the end of the stroke (θs) is
A10
at t4 = TS. The value of TS is determined from the speed of the SRM during high speed
generating. The duty cycle for the magnetisation during the interval t0 – t1 is D1, the
duty cycle for generating during the interval t1 – t2 is D2 and the duty cycle for
demagnetisation during the interval t2 – t3 is D3.
The equation for the average power (P) over the stroke during low speed generating is
defined as:
1  t1 t2 t3 t4

 v(t )i (t ) dt + v(t )i(t ) dt + v(t )i(t ) dt + v(t )i(t ) dt 
P=
TS t∫ ∫ ∫ ∫ 
(A.28)
0 t1 t2 t3 
During magnetisation in the interval t0 – t1, the voltage is assumed to be approximately

constant (VS) and the current is a triangular wave with a peak value of Im as indicated in
Figure A.6. The equation for the average power during demagnetisation (PM) is:
t D1TS
1 1 V
PM = ∫
TS t0
v(t )i (t )dt = S
TS ∫ i(t )dt = 0.5I
0
m VS D1 (A.29)
During generating in the interval t1 – t2, the current and voltage are assumed to be
approximately constant at Im and -Vs, respectively, as indicated in Figure A.6. The
average power during generating (PG) is:
D2TS + D1TS
− VS I m
t
1 2
PG = ∫
TS t1
v(t )i(t )dt =
TS ∫
D1TS
d t = − V S I m D2 (A.30)
generating (PG) is
PG = − 1( − VS I m D2 ) = VS I m D2 (A.31)
During demagnetisation in the interval t2 – t3, the voltage is assumed to be constant (-Vs)
during the interval and the current is approximated by a triangular wave with a peak
value of Im as indicated in Figure A.6. The equation for the output power during
demagnetisation is:
A11
D3TS + D2TS + D1TS
− VS
t
1 3
PDM = ∫
TS t2
v(t )i(t )dt =
TS ∫ i(t )dt = − 0.5V
D2TS + D1TS
S I m D3 (A.32)
demagnetisation is:
PDM = − 1( − 0.5VS I m D3 ) = 0.5VS I m D3 (A.33)
zero during the interval. Therefore, the average power (P) over the stroke during
starting and high speed generating is:
P = 0.5I mVS D1 + I mVS D2 + 0.5I mVS D3 (A.34)
A12
A.3 Component Analysis
The analysis for each component of both the asymmetrical half-bridge and full-bridge
synchronous rectification converter is divided into three specific ISG modes of
operation: (a) starting and low speed motoring, (b) low speed generating and (c) high
speed generating. The mechanical rotor position, inductance profile, the switching
waveforms for all components (Q1, Q2, D1 and D2), and the associated phase voltage
and current are illustrated in Figure A.7 for (a) starting and low speed motoring and (b)
low speed generating.
Figure A.7: Waveforms at speeds of operation (a) starting and low speed motoring (b)
low speed generating
A13
The mechanical rotor position, inductance profile, the switching waveforms for all
components (Q1, Q2, D1 and D2), and the associated phase voltage and current are
illustrated in Figure A.8 for high speed generating
Stator
s s s s
Poles
Rotor
R R R R
Poles Direction of
rotor rotation
Conduction Period
L
DC
Lmax
Lmin
Rotor
θ1 θon θ2 θ3 θoff θ4 θext Angle
+VS
Q1
+VS
Q2
+VS
D1
+VS
D2
+VS
Vphase
-VS
Im
|e| = |VDC|
Iphase
I II III Rotor
Angle
Figure A.8: Waveforms for high speed generating operation
A14
A.3.1 Asymmetrical half-bridge
The determination of the relevant equations and parameter values for evaluating each
component (Q1, Q2, D1 and D1) in each ISG mode of operation (starting, low speed
motoring, and low and high speed generating) for the asymmetrical half-bridge
converter is presented in the following sections. The schematic of the asymmetrical
half-bridge converter including the converter modes of operation (magnetisation,
freewheeling and demagnetisation) are illustrated in Figure A.9
Figure A.9: The asymmetrical half-bridge or classic converter. (a) Static view of
converter. (b) Energization mode. (c) Freewheeling mode. (d) Forced demagnetization
mode
A15
A.3.1.1 Starting and low speed motoring – Q1
Lmax
VS
I(Q1) Lmin
Q1 V(Q1)
D2 θon θ1 θ2 θext θ4 Rotor
Angle
PWM
+VS
D1 Q2
V(Q1)
Im
I(Q1)
(a) (b)
Figure A.10: (a) Converter schematic with voltage and current labelled for Q1.
(b)Voltage and current waveforms for Q1 during magnetisation
The conduction losses are determined using equation A.5. The RMS current is
determined using equation A.2 as the current waveform is assumed to be of the form of
a square wave with a peak value of Im as shown in Figure A.10b. D is the duty cycle of
continuous operation, which is determined in chapter 4 to be 25%.
The switching losses are determined using equation A.6. Vm = VS (the applied voltage)
for Q1 as shown in Figure A.10a and Im is shown in Figure A.10b. The switching
frequency, fs, is 25 kHz. The tr and tf time are obtained from the specific MOSFET data
sheet.
A16
(b)Voltage and current waveforms for Q2 during magnetisation and freewheeling
determined using equation A.2 as the current waveform is assumed to be a square wave
with a peak value of Im as shown in Figure A.11b. D is the duty cycle of continuous
operation, which is determined in chapter 4 to be 50%.
frequency, fs, is determined from the speed of the SRM. From the analysis in chapter 4,
starting at 200RPM yields 20Hz and low speed motoring at 1000RPM yields 100Hz.
The tr and tf time are obtained from the specific MOSFET data sheet.
A17
A.3.1.3 Starting and low speed motoring – D1
Lmax
VS
Lmin
Q1
D2
Rotor
θon θ1 θ2 θext θ4 Angle
V(D1)
D1 V(D1) Q2
-VS
I(D1)
Im
I(D1)
I II
(a) (b)
Figure A.12: (a) Converter schematic with voltage and current labelled for D1.
(b)Voltage and current waveforms for D1during freewheeling and demagnetisation
The conduction and switching losses are divided into two regions of operation, (I)
freewheeling and (II) demagnetization.
(I) Freewheeling
The conduction losses are determined using equation A.7. The current waveform is
assumed to be of the form of a square wave with a peak value of Im as shown in Figure
A.12b. VD = VF (the forward bias voltage) for D1 as shown in Figure A.12a and D is the
duty cycle of continuous operation, which is determined in chapter 4 to be 25%. The
switching losses are determined using equation A.8 where VD = VS (the reverse voltage)
across D1 and fs is 25 kHz. Qrr for the specific diode is obtained for the diode data sheet.
A18
(II) Demagnetization
assumed to be of the form of a triangular wave with the peak current is assumed to be
0.5Im. D is the duty cycle of continuous operation, which is determined in chapter 4 to
be 10%. The switching losses are determined using equation A.8 where VD = VS (the
reverse voltage) across D1 and fs is 20 Hz for starting and 100Hz for low speed motoring.
Qrr for the specific diode is obtained for the diode data sheet.
A.3.1.4 Starting and low speed motoring – D2
Lmax
VS Lmin
Q1 V(D2) D2 θon θ1 θ2 θext θ4 Rotor

Angle
I(D2)
V(D2)
D1 Q2 -VS
Im
I(D2)
(a) (b)
(b)Voltage and current waveforms for D2 during demagnetisation
The conduction and switching losses are the same as for D1 in the second region of
operation, (II) demagnetization.
A19
A.3.1.5 Low speed generating - Q1
L
Lmax
VS
I(Q1) Lmin
Q1 V(Q1)
D2 θon θ3 θ4 θext Rotor
Angle
PWM
+VS
D1 Q2
V(Q1)
iexc
Im
I(Q1)
(a) (b)
a square wave shape with a peak value of Im as shown in Figure A.14b. D is the duty
cycle of continuous operation, which is determined in chapter 4 to be 25%.
frequency, fs, is 25 kHz. The tr and tf time are obtained from the specific MOSFET data
sheet.
A20
A.3.1.6 Low speed generating – Q2
The conduction and switching losses for both regions are the same as for Q1.
A21
A.3.1.7 Low speed generating – D1
L
Lmax
VS
Lmin
Q1
D2
θon θ3 θ4 θext Rotor
Angle
D1 V(D1) Q2 V(D1)
I(D1)
-VS
igen
Im
I(D1)
I II
(a) (b)
(b)Voltage and current waveforms for D1during generating and demagnetisation
generating and (II) demagnetization.
(I) Generating
assumed to be of a square wave with a peak value of Im as shown in Figure A.16b. VD =
VF (the forward bias voltage) for D1 as shown in Figure A.16a and D is the duty cycle of
continuous operation, which is determined in chapter 4 to be 25%. The switching losses
are determined using equation A.8 where VD = VS (the reverse voltage) across D1 and fs
is 25 kHz. Qrr for the specific diode is obtained for the diode data sheet.
assumed to be of the form of a triangular wave with the peak current is assumed to be
A22
0.5Im. D is the duty cycle of continuous operation, which is determined in chapter 4 to
be 10%. The switching losses are determined using equation A.8 where VD = VS (the
reverse voltage) across D1 and fs is 200 Hz. Qrr for the specific diode is obtained for the
diode data sheet.
A.3.1.8 Low speed generating – D2
L
Lmax
VS Lmin
Q1 V(D2) D2 θon θ3 θ4 θext Rotor

Angle
I(D2)
D1 Q2 V(D2)
+VS
igen
Im
I(D2)
I II
(a) (b)
(b)Voltage and current waveforms for D2 during generating and demagnetisation
generating and (II) demagnetization. The conduction and switching losses for both
regions are the same as for D1.
A23
A.3.1.9 High speed generating – Q1
Lmax
VS
I(Q1) Lmin
Q1 V(Q1) θ1 θon θ2 θ3 θoff θ4 θext Rotor

D2
Angle
+VS
D1 Q2 V(Q1)
Im
I(Q1)
(a) (b)
Figure A.18: (a) Converter schematic with voltage and current labelled for Q1. (b)Voltage
and current waveforms for Q1 during magnetisation
a triangular pulse shape with a peak value of Im as shown in Figure A.18b. D is the duty
frequency, fs, is determined from the speed of the SRM. From the analysis in chapter 4,
high speed generating at 4500RPM yields 450Hz and 6000RPM yields 600Hz. The tr
and tf time are obtained from the specific MOSFET data sheet.
A24
Lmax
Lmin
VS
Rotor
Q1 θ1 θon θ2 θ3 θoff θ4 θext Angle
D2
I(Q2)
+VS
D1 V(Q2) Q2 V(Q2)
Im
I(Q2)
(a) (b)
and current waveforms for Q2 during magnetisation
The conduction and switching losses for both regions are the same as for Q1.
A25
A.3.1.11 High speed generating – D1
Lmax
VS Lmin
Q1 θ1 θon θ2 θ3 θoff θ4 θext

Rotor
D2 Angle
+VS
D1 V(D1) Q2 V(D1)
I(D1)
Im
I(D1)
I II
(a) (b)
Figure A.20: (a) Converter schematic with voltage and current labelled for D1. (b)Voltage
and current waveforms for D1 during generating and demagnetisation
generating and (II) demagnetization.
(I) Generating
assumed to be of the form of a square pulse shape. D is the duty cycle of continuous
operation, which is determined in chapter 4 to be 30%. Switching losses are determined
using equation A.8 where fs for high speed generating at 4500RPM yields 450Hz and
6000RPM yields 600Hz.
assumed to be of the form of a triangular pulse shape with a peak current is assumed to
be 0.5*IM. D is the duty cycle of continuous operation, which is determined in chapter 4
to be 10%. Switching losses are determined using equation A.8 where fs for high speed
generating at 4500RPM yields 450Hz and 6000RPM yields 600Hz.
.
A26
A.3.1.12 High speed generating – D2
Lmax
VS
Lmin
Q1 V(D2) D2
Rotor
I(D2) θ1 θon θ2 θ3 θoff θ4 θext Angle
D1 Q2 +VS
V(D2)
Im
I(D2)
I II
(a) (b)
Figure A.21: (a) Converter schematic with voltage and current labelled for D2. (b)Voltage
and current waveforms for D2 during generating and demagnetisation
regions are the same as for D1.
A27
A.3.2 Full-bridge with synchronous rectification
The determination of the relevant equations and parameter values for evaluating each
component (Q1, Q2, Q3 and Q4) in each mode of operation for the full-bridge with
synchronous rectification is presented in the following sections. The schematic of the
full-bridge synchronous rectification converter including its modes of operation are
illustrated in Figure A.22.
VS
Q1 Q4
Q3 Q2
(a)
VS VS VS
Q1 Q4 Q4 Q1 V(Q4) Q4
Q1
V(Q1)
Vph = VS Vph = 0 Vph = -VS
iph iph iph
Q3 Q2 Q3 V(Q3) V(Q2)
Q2 Q3 V(Q3) Q2
V(Q2)
(b) (c) (d)
Figure A.22: The full-bridge converter. (a) Static view of converter. (b) Energization
A28
A.3.2.1 Starting and low speed motoring – Q1 and Q2
The conduction and switching losses are the same as for Q1 and Q2 from the asymmetric
half bridge.
Lmax
VS
Lmin
Q1 Q4
Rotor
θon θ1 θ2 θext Angle
iph PWM
+VSD
Q3 V(Q3) Q2 V(Q3)
-VS
I(Q3)
Im
I(Q3)
I II
(a) (b)
and current waveforms for Q3 during freewheeling and demagnetisation
A29
Freewheeling and (II) demagnetization.
(I) Freewheeling
a square pulse shape with a peak value of Im as shown in Figure A.23b. D is the duty
The switching losses are determined using equation A.6 with the switching frequency, fs,
of 25 kHz and Vm is equal to the peak voltage, VS, across Q3 as shown in Figure A.23a
and I(Q3) = Im as shown in Figure A.23b. The tr and tf time are obtained from the specific
MOSFET data sheet.
The switching losses are determined using equation A.6. The switching frequency, fs, is
determined from the speed of the SRM. From the analysis in chapter 4, starting at
200RPM yields 20Hz and low speed motoring at 1000RPM yields 100Hz. Vm is equal
to the peak voltage, VS, across Q3 as shown in Figure A.23a and I(Q3) = Im as shown in
Figure A.23b. The tr and tf time are obtained from the specific MOSFET data sheet.
A30
Lmax
VS
I(Q4)
Lmin
Q1 V(Q4) Q4
Rotor
θon θ1 θ2 θext Angle
iph
V(Q4)
Q3 Q2 +VSD
-VS
Im
I(Q4)
(a) (b)
and current waveforms for Q4 during demagnetisation
The conduction and switching losses are the same as for Q3 in the second region of
operation, (II) demagnetization.
A31
A.3.2.4 Low speed generating – Q1 and Q2
The conduction and switching losses are the same as for Q1 and Q2 from the
asymmetric half bridge.
Pole
L Pitch
Lmax
VS
Lmin
Q1 Q4
Angle
iph
Q3 V(Q3) Q2 PWM
+VSD
V(Q3)
I(Q3)
-VS
igen
Im
I(Q3)
I II
(a) (b)
Figure A.25: (a) Converter schematic with voltage and current labelled for Q3. (b)
Voltage and current waveforms for Q3 during generating and demagnetisation
Generating and (II) demagnetization.
(I) Generating
The conduction losses are determined using equation A.5 The RMS current is
A32
The switching losses are determined using equation A.6 with the switching frequency, fs,
of 25 kHz and Vm is equal to the peak voltage, VS, across Q3 as shown in Figure A.25a
and I(Q3) = Im as shown in Figure A.25b. The tr and tf time are obtained from the specific
MOSFET data sheet.
determined from the speed of the SRM. From the analysis in chapter 4, low speed
generating at 2000RPM yields 200Hz. Vm is equal to the peak voltage, VS, across Q3 as
shown in Figure A.25a and I(Q3) = Im as shown in Figure A.25b. The tr and tf time are
obtained from the specific MOSFET data sheet.
A33
Pole
L Pitch
Lmax
Lmin
VS I(Q4)

Q1 V(Q4) Q4 Angle
V(Q4)
iph +VSD PWM
Q3 Q2
-VS
igen
Im
I(Q4)
I II
(a) (b)
and current waveforms for Q4 during generating and demagnetisation
regions are the same as for Q3.
A34
A.3.2.7 High speed generating – Q1 and Q2
The conduction and switching losses are the same as for Q1 and Q2 from the
asymmetric half bridge.
Figure A.27: (a) Converter schematic with voltage and current labelled for Q3. (b)
Voltage and current waveforms for Q3 during generating and demagnetisation
Generating and (II) demagnetization.
(I) Generating
A35
determined from the speed of the SRM. From the analysis in chapter 4, high speed
generating at 4500RPM yields 450Hz and 6000RPM yields 600Hz. Vm is equal to the
peak voltage, VS, across Q3 as shown in Figure A.27a and I(Q3) = Im as shown in Figure
A.27b. The tr and tf time are obtained from the specific MOSFET data sheet.
determined from the speed of the SRM. From the analysis in chapter 4, high speed
generating at 4500RPM yields 450Hz and 6000RPM yields 600Hz. Vm is equal to the
peak voltage, VS, across Q3 as shown in Figure A.27a and I(Q3) = Im as shown in Figure
A.27b. The tr and tf time are obtained from the specific MOSFET data sheet.
A36
and current waveforms for Q4 during generating and demagnetisation
regions are the same as for Q3.
A37
Appendix B – Power Analysis Computations
In the following appendix, the complete power analysis calculation results generated by
a spreadsheet program are presented for the asymmetrical half-bridge converter and
full-bridge synchronous rectification converter. The specific components that were
evaluated are tabulated in Table B.1, while a summary of the operating conditions are
provided in section B.1. The power calculation results for the asymmetrical half-bridge
converter are presented in section B.2 and the results for the components that were
evaluated for the full-bridge synchronous rectification converter are presented in
sections B.3 through B.9, respectively. A number of component configurations were
evaluated for the full-bridge synchronous rectification converter and these included
single device and 2 or 4 devices mounted in parallel at each switching location (Q1, Q2,
Q3 and Q4).
Device Manufacturer Model Number

Mosfet APT APT20M11JFLL
Diode ST Microelectronics STTH20002TV
Mosfet Fairchild FDP047AN08A0
Mosfet Fairchild FDH038AN08A1
Mosfet Fairchild FDP060AN08A0
Mosfet Infineon SPP80N08S2L-7
Mosfet Phillips BUK7506-76B
Mosfet Phillips PSMN009-100W
Table B.1: Components evaluated
B1
ISA Function Starting Low speed motoring Low speed generating High speed generating
Voltage (VS) 36 42 42 42
PeakCurrent (A) 140 140 140 140
PeakCurrent (A) per Switch 140 140 140 140
Avg Current (A) 77 77 77 77
Avg Current (A) per Switch 77 77 77 77
Peak Power (W) 5040 5880 5880 5880
Avg Power (W) 2772 3234 3234 3234
ISA Speed (rpm) 200 1000 2000 4500
Rev/Sec 3.33 16.67 33.33 75
Period/Rev (ms) 300 60 30 13.3
Rotor Pole 6 6 6 6
Pole Period (ms) 50 10 5 2.2
Frequency (Hz) 25000 20 25000 100 25000 200 450 450
B2
Dudy Cycle - Q1 25% 25% 25% 40%
Dudy Cycle - Q2 50% 50% 25% 40%
Dudy Cycle - D1 25% 10% 25% 10% 25% 10% 30% 10%
Dudy Cycle - D2 10% 10% 25% 10% 30% 10%
Notes
B.1 Operating Conditions
o
1. Maximum ambient temperature in all cases (TA ): 90 C 8. Heat Spreader length: 0.05 m
o
2. Maximum semiconductor case temperature (TC): 110 C 9. Heat Spreader width: 0.025 m
o
3. Maximum semiconductor junction temperature (TJ): 110 C 10. Heat Spreader depth: 0.016 m
o
4. Minimum semiconductor voltage rating: 42 Vdc 11. Thermal Coupler RθTC-S 0.2 C/W
5. Minimum semiconductor peak current rating: 140 A
6. Minimum semiconductor continuous current rating: 84 A Device Number 1
7. Minimum semiconductor switching frequency: 25 kHz Rotor Pole No. 6
B.2 Asymmetric Half-Bridge – Single Devices
B.2.1 MOSFET- APT20M11FLL, Diode –STTH20002TV
Power Analysis
Component Manufacturer Model Manufacturer Model
MOSFET: Q1, Q2, D1 and D2 APT APT20M11JFLL ST STTH20002TV
Data Device No.: 1 VF 1.8 V
trr 500 ns Max Peak Current Qrr@di/dt=100A/µs 275 nC
o
RDS(ON)@Tj=175oC 0.02 Ω 184 A (TC = 115 C) RDS(ON) Coefficient @Tj=175oC 2.25
tr 500 ns Max Cont. Current Rise and Fall time assumed worse
tf 1 µs 80 A (TC = 144oC) Multiply by a factor of 10
Power Analysis Starting Low speed Low speed High speed
Speed (RPM) 200 1000 2000 4500
Q1
Conduction Losses
IRMS(Q1) (A) 70 70 70 51
PC(Q1) (W) 121 121 121 65
Switching Losses
PS(Q1) (W) 95 110 110 2
Total Losses
PT (Q1) (W) 216 232 232 67
Q2
Conduction Losses
IRMS(Q2) (A) 99 99 70 51
PC(Q2) (W) 243 243 121 65
Switching Losses
PS(Q2) (W) 0.076 0.4 110 2
Total Losses
PT(Q2) (W) 243 243 232 67
D1
Freewheeling or Generating
Conduction Losses
PCF(D1) (W) 63 63 63 76
Switching Losses
PSF(D1) (W) 0.25 0.3 0.3 0.01
Demagnetization Losses
Conduction Losses
PCDM(D1) (W) 13 13 13 13
Switching Losses
PSDM(D1) (W) 0.0002 0.001 0.002 0.01
Total Losses
PT (D1) (W) 76 76 76 88
D2
Total Losses
PT (D2) (W) 13 13 13 13
Power Loss Single Phase
PT(SP) (W) 547 563 552 234
Power Transfer of
Single Phase
Im (A) 140 140 140 28
VS (V) 9 11 11 42
PM (W) 1260 1470 1470 1176
Im (A) 0 0 140 140
VS (V) 0 0 11 42
PG (W) 0 0 1470 1764
Im (A) 7 7 7 7
VS (V) 36 42 42 42
PDM (W) 252 294 294 294
Efficiency of Single Phase
Pphase (W) 1512 1764 3234 3234
Pphase + PT(SP) (W) 2059 2327 3786 3468
Efficiency (%) 73 76 85 93
B3
B.3 Full-bridge with Synchronous Rectification
B.3.1 MOSFETs- APT20M11FLL – Single Devices
Power Analysis
Component Munufacturer Model
MOSFET: Q1, Q2, Q3 and Q4 APT APT20M11JFLL
Data Device No.: 1 VSD 1.25 V
trr 1000 ns Max Peak Current
o o o
RDS(ON)@Tj=175 C (Coeff: 2.25) 0.02475 Ω 184 A (TC = 115 C) RDS(ON) Coefficient @Tj=175 C 2.25
o
tf 1000 ns 80 A (TC = 144 C) Multiply by a factor of 10
Speed (RPM) 200 1000 2000 4500
Q1
Conduction Losses
IRMS(Q1) (A) 70 70 70 51
PC(Q1) (W) 121 121 121 65
Switching Losses
PS(Q1) (W) 95 110 110 2
Total Losses
PT (Q1) (W) 216 232 232 67
Q2
Conduction Losses
IRMS(Q2) (A) 99 99 70 51
PC(Q2) (W) 243 243 121 65
Switching Losses
PS(Q2) (W) 0 0 110 2
Total Losses
PT(Q2) (W) 243 243 232 67
Q3
Conduction Losses
IRMS(Q3) (A) 70 70 70 77
PCF(Q3) (W) 121 121 121 146
Switching Losses
PSF(Q3) (W) 95 110 110 2
Demagnetisation Losses
Conduction Losses
IRMS(Q3) (A) 26 26 26 26
PCDM(Q3) (W) 16 16 16 16
Switching Losses
PSDM(Q3) (W) 0 0 1 2
Total Losses
PT (Q3) (W) 232 248 249 166
Q4
Total Losses
PT (Q4) (W) 16 17 249 166
PT(SP) (W) 707 739 960 465
Power Transfer
of Single Phase
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PM (W) 1260 1470 1470 1176
Im (A) 0 0 140 140
Vm (A) 0 0 42 42
PG (W) 0 0 1470 1764
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PDM (W) 504 588 588 294
Pphase (W) 1764 2058 3528 3234
Pphase + PT(SP ) (W) 2471 2797 4488 3699
Efficiency (% ) 71 74 79 87
B4
B.3.2 Parallel MOSFETs- FDP047AN08A0 – Two Devices
Power Analysis
MOSFET: Q1, Q2, Q3 and Q4 Fairchild Semi FDP047AN08A0
RDS(ON)@Tj=175oC (Coeff: 2.34) 0.011 Ω 184 A (TC = 115oC) RDS(ON) Coefficient @Tj=175oC 2.34
tf 450 ns 80 A (TC = 144oC) Multiply by a factor of 10
Speed (RPM) 200 1000 2000 4500
Q1
Conduction Losses
IRMS(Q1) (A) 35 35 35 26
PC(Q1) (W) 13 13 13 7
Switching Losses
PS(Q1) (W) 42 49 49 1
Total Losses
PT (Q1) (W) 55 62 62 8
Q2
Conduction Losses
IRMS(Q2) (A) 49 49 35 26
PC(Q2) (W) 27 27 13 7
Switching Losses
PS(Q2) (W) 0 0 49 1
Total Losses
PT(Q2) (W) 27 27 62 8
Q3
Conduction Losses
IRMS(Q3) (A) 35 35 35 38
PCF(Q3) (W) 13 13 13 16
Switching Losses
PSF(Q3) (W) 42 49 49 1
Conduction Losses
IRMS(Q3) (A) 13 13 13 13
PCDM(Q3) (W) 2 2 2 2
Switching Losses
PSDM(Q3) (W) 0 0 0 1
Total Losses
PT (Q3) (W) 57 64 65 20
Q4
Total Losses
PT (Q4) (W) 2 2 65 20
PT(SP) (W) 283 312 508 111
Power Transfer
of Single Phase
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PM (W) 1260 1470 1470 1176
Im (A) 0 0 140 140
Vm (A) 0 0 42 42
PG (W) 0 0 1470 1764
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PDM (W) 504 588 588 294
Pphase (W) 1764 2058 3528 3234
Pphase + PT(SP) (W) 2047 2370 4036 3345
Efficiency (%) 86 87 87 97
B5
B.3.3 Parallel MOSFETs- FDP047AN08A0 – Four Devices
Power Analysis
o o o
o
Speed (RPM) 200 1000 2000 4500
Q1
Conduction Losses
IRMS(Q1) (A) 18 18 18 13
PC(Q1) (W) 3 3 3 2
Switching Losses
PS(Q1) (W) 21 24 24 0
Total Losses
PT (Q1) (W) 24 28 28 2
Q2
Conduction Losses
IRMS(Q2) (A) 25 25 18 13
PC(Q2) (W) 7 7 3 2
Switching Losses
PS(Q2) (W) 0 0 24 0
Total Losses
PT(Q2) (W) 7 7 28 2
Q3
Conduction Losses
IRMS(Q3) (A) 18 18 18 19
PCF(Q3) (W) 3 3 3 4
Switching Losses
PSF(Q3) (W) 21 24 24 0
Conduction Losses
IRMS(Q3) (A) 6 6 6 6
PCDM(Q3) (W) 0 0 0 0
Switching Losses
PSDM(Q3) (W) 0 0 0 0
Total Losses
PT (Q3) (W) 25 28 28 5
Q4
Total Losses
PT (Q4) (W) 0 1 28 5
PT(SP) (W) 225 254 450 61
Power Transfer
of Single Phase
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PM (W) 1260 1470 1470 1176
Im (A) 0 0 140 140
Vm (A) 0 0 42 42
PG (W) 0 0 1470 1764
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PDM (W) 504 588 588 294
Pphase (W) 1764 2058 3528 3234
Pphase + PT(SP) (W) 1989 2312 3978 3295
Efficiency (%) 89 89 89 98
B6
B.3.4 Parallel MOSFETs- FDH038AN08A1 – Two Devices
Power Analysis
MOSFET: Q1, Q2, Q3 and Q4 Fairchild Semi FDH038AN08A1
o o o
o
Speed (RPM) 200 1000 2000 4500
Q1
Conduction Losses
IRMS(Q1) (A) 35 35 35 26
PC(Q1) (W) 11 11 11 6
Switching Losses
PS(Q1) (W) 84 98 98 2
Total Losses
PT (Q1) (W) 95 109 109 8
Q2
Conduction Losses
IRMS(Q2) (A) 49 49 35 26
PC(Q2) (W) 22 22 11 6
Switching Losses
PS(Q2) (W) 0 0 98 2
Total Losses
PT(Q2) (W) 22 22 109 8
Q3
Conduction Losses
IRMS(Q3) (A) 35 35 35 38
PCF(Q3) (W) 11 11 11 13
Switching Losses
PSF(Q3) (W) 84 98 98 2
Conduction Losses
IRMS(Q3) (A) 13 13 13 13
PCDM(Q3) (W) 1 1 1 1
Switching Losses
PSDM(Q3) (W) 0 0 1 2
Total Losses
PT (Q3) (W) 97 111 111 18
Q4
Total Losses
PT (Q4) (W) 2 2 111 18
PT(SP) (W) 430 488 881 103
Power Transfer
of Single Phase
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PM (W) 1260 1470 1470 1176
Im (A) 0 0 140 140
Vm (A) 0 0 42 42
PG (W) 0 0 1470 1764
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PDM (W) 504 588 588 294
Pphase (W) 1764 2058 3528 3234
Pphase + PT(SP) (W) 2194 2546 4409 3337
Efficiency (%) 80 81 80 97
B7
B.3.5 Parallel MOSFETs- FDH038AN08A1 – Four Devices
Power Analysis
MOSFET: Q1, Q2, Q3 and Q4 Fairchild Semi FDH038AN08A1
o o o
o
Speed (RPM) 200 1000 2000 4500
Q1
Conduction Losses
IRMS(Q1) (A) 18 18 18 13
PC(Q1) (W) 3 3 3 1
Switching Losses
PS(Q1) (W) 42 49 49 1
Total Losses
PT (Q1) (W) 45 52 52 2
Q2
Conduction Losses
IRMS(Q2) (A) 25 25 18 13
PC(Q2) (W) 5 5 3 1
Switching Losses
PS(Q2) (W) 0 0 49 1
Total Losses
PT(Q2) (W) 5 6 52 2
Q3
Conduction Losses
IRMS(Q3) (A) 18 18 18 19
PCF(Q3) (W) 3 3 3 3
Switching Losses
PSF(Q3) (W) 42 49 49 1
Conduction Losses
IRMS(Q3) (A) 6 6 6 6
PCDM(Q3) (W) 0 0 0 0
Switching Losses
PSDM(Q3) (W) 0 0 0 1
Total Losses
PT (Q3) (W) 45 52 53 5
Q4
Total Losses
PT (Q4) (W) 0 1 53 5
PT(SP) (W) 383 441 835 62
Power Transfer
of Single Phase
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PM (W) 1260 1470 1470 1176
Im (A) 0 0 140 140
Vm (A) 0 0 42 42
PG (W) 0 0 1470 1764
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PDM (W) 504 588 588 294
Pphase (W) 1764 2058 3528 3234
Pphase + PT(SP) (W) 2147 2499 4363 3296
Efficiency (%) 82 82 81 98
B8
B.3.6 Parallel MOSFETs- FDP060AN08A0 – Two Devices
Power Analysis
o o o
Speed (RPM) 200 1000 2000 4500
Q1
Conduction Losses
IRMS(Q1) (A) 35 35 35 26
PC(Q1) (W) 16 16 16 9
Switching Losses
PS(Q1) (W) 37 43 43 1
Total Losses
PT (Q1) (W) 53 59 59 9
Q2
Conduction Losses
IRMS(Q2) (A) 49 49 35 26
PC(Q2) (W) 32 32 16 9
Switching Losses
PS(Q2) (W) 0 0 43 1
Total Losses
PT(Q2) (W) 32 33 59 9
Q3
Conduction Losses
IRMS(Q3) (A) 35 35 35 38
PCF(Q3) (W) 16 16 16 19
Switching Losses
PSF(Q3) (W) 37 43 43 1
Conduction Losses
IRMS(Q3) (A) 13 13 13 13
PCDM(Q3) (W) 2 2 2 2
Switching Losses
PSDM(Q3) (W) 0 0 0 1
Total Losses
PT (Q3) (W) 55 61 62 23
Q4
Total Losses
PT (Q4) (W) 2 2 62 23
PT(SP) (W) 286 311 483 130
Power Transfer of Single Phase
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PM (W) 1260 1470 1470 1176
Im (A) 0 0 140 140
Vm (A) 0 0 42 42
PG (W) 0 0 1470 1764
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PDM (W) 504 588 588 294
Pphase (W) 1764 2058 3528 3234
Pphase + PT(SP) (W) 2050 2369 4011 3364
Efficiency (%) 86 87 88 96
B9
B.3.7 Parallel MOSFETs- FDP060AN08A0 – Four Devices
Power Analysis
o o o
Speed (RPM) 200 1000 2000 4500
Q1
Conduction Losses
IRMS(Q1) (A) 18 18 18 13
PC(Q1) (W) 4 4 4 2
Switching Losses
PS(Q1) (W) 18 21 21 0
Total Losses
PT (Q1) (W) 22 26 26 3
Q2
Conduction Losses
IRMS(Q2) (A) 25 25 18 13
PC(Q2) (W) 8 8 4 2
Switching Losses
PS(Q2) (W) 0 0 21 0
Total Losses
PT(Q2) (W) 8 8 26 3
Q3
Conduction Losses
IRMS(Q3) (A) 18 18 18 19
PCF(Q3) (W) 4 4 4 5
Switching Losses
PSF(Q3) (W) 18 21 21 0
Conduction Losses
IRMS(Q3) (A) 6 6 6 6
PCDM(Q3) (W) 1 1 1 1
Switching Losses
PSDM(Q3) (W) 0 0 0 0
Total Losses
PT (Q3) (W) 23 26 26 6
Q4
Total Losses
PT (Q4) (W) 1 1 26 6
PT(SP) (W) 217 242 414 70
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PM (W) 1260 1470 1470 1176
Im (A) 0 0 140 140
Vm (A) 0 0 42 42
PG (W) 0 0 1470 1764
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PDM (W) 504 588 588 294
Pphase (W) 1764 2058 3528 3234
Pphase + PT(SP) (W) 1981 2300 3942 3304
Efficiency (%) 89 89 89 98
B10
B.3.8 Parallel MOSFETs- SPP80N08S2L-07 – Two Devices
Power Analysis
MOSFET: Q1, Q2, Q3 and Q4 Infineon SPP80N08S2L-07
RDS(ON)@Tj=175oC (Coeff: 1) 0.016 Ω 184 A (TC = 115oC) RDS(ON) Coefficient @Tj=175oC 1
o
Speed (RPM) 200 1000 2000 4500
Q1
Conduction Losses
IRMS(Q1) (A) 35 35 35 26
PC(Q1) (W) 20 20 20 10
Switching Losses
PS(Q1) (W) 75 88 88 2
Total Losses
PT (Q1) (W) 95 107 107 12
Q2
Conduction Losses
IRMS(Q2) (A) 49 49 35 26
PC(Q2) (W) 39 39 20 10
Switching Losses
PS(Q2) (W) 0 0 88 2
Total Losses
PT(Q2) (W) 39 40 107 12
Q3
Conduction Losses
IRMS(Q3) (A) 35 35 35 38
PCF(Q3) (W) 20 20 20 24
Switching Losses
PSF(Q3) (W) 75 88 88 2
Conduction Losses
IRMS(Q3) (A) 13 13 13 13
PCDM(Q3) (W) 3 3 3 3
Switching Losses
PSDM(Q3) (W) 0 0 1 2
Total Losses
PT (Q3) (W) 98 110 111 29
Q4
Total Losses
PT (Q4) (W) 3 3 111 29
PT(SP) (W) 469 521 873 165
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PM (W) 1260 1470 1470 1176
Im (A) 0 0 140 140
Vm (A) 0 0 42 42
PG (W) 0 0 1470 1764
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PDM (W) 504 588 588 294
Pphase (W) 1764 2058 3528 3234
Pphase + PT(SP) (W) 2233 2579 4401 3399
Efficiency (%) 79 80 80 95
B11
B.3.9 Parallel MOSFETs- SPP80N08S2L-07 – Four Devices
Power Analysis
MOSFET: Q1, Q2, Q3 and Q4 Infineon SPP80N08S2L-07
RDS(ON)@Tj=175oC (Coeff: 1) 0.016 Ω 184 A (TC = 115oC) RDS(ON) Coefficient @Tj=175oC 1
Speed (RPM) 200 1000 2000 4500
Q1
Conduction Losses
IRMS(Q1) (A) 18 18 18 13
PC(Q1) (W) 5 5 5 3
Switching Losses
PS(Q1) (W) 38 44 44 1
Total Losses
PT (Q1) (W) 43 49 49 3
Q2
Conduction Losses
IRMS(Q2) (A) 25 25 18 13
PC(Q2) (W) 10 10 5 3
Switching Losses
PS(Q2) (W) 0 0 44 1
Total Losses
PT(Q2) (W) 10 10 49 3
Q3
Conduction Losses
IRMS(Q3) (A) 18 18 18 19
PCF(Q3) (W) 5 5 5 6
Switching Losses
PSF(Q3) (W) 38 44 44 1
Conduction Losses
IRMS(Q3) (A) 6 6 6 6
PCDM(Q3) (W) 1 1 1 1
Switching Losses
PSDM(Q3) (W) 0 0 0 1
Total Losses
PT (Q3) (W) 43 50 50 8
Q4
Total Losses
PT (Q4) (W) 1 1 50 8
PT(SP) (W) 385 437 789 92
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PM (W) 1260 1470 1470 1176
Im (A) 0 0 140 140
Vm (A) 0 0 42 42
PG (W) 0 0 1470 1764
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PDM (W) 504 588 588 294
Pphase (W) 1764 2058 3528 3234
Pphase + PT(SP) (W) 2149 2495 4317 3326
Efficiency (%) 82 82 82 97
B12
B.3.10 Parallel MOSFETs- BUK75/7606-75B – Two Devices
Power Analysis
MOSFET: Q1, Q2, Q3 and Q4 Phillips Semi BUK75/7606-75B
o o o
o
Speed (RPM) 200 1000 2000 4500
Q1
Conduction Losses
IRMS(Q1) (A) 35 35 35 26
PC(Q1) (W) 14 14 14 8
Switching Losses
PS(Q1) (W) 33 38 38 1
Total Losses
PT (Q1) (W) 47 53 53 8
Q2
Conduction Losses
IRMS(Q2) (A) 49 49 35 26
PC(Q2) (W) 29 29 14 8
Switching Losses
PS(Q2) (W) 0 0 38 1
Total Losses
PT(Q2) (W) 29 29 53 8
Q3
Conduction Losses
IRMS(Q3) (A) 35 35 35 38
PCF(Q3) (W) 14 14 14 17
Switching Losses
PSF(Q3) (W) 33 38 38 1
Conduction Losses
IRMS(Q3) (A) 13 13 13 13
PCDM(Q3) (W) 2 2 2 2
Switching Losses
PSDM(Q3) (W) 0 0 0 1
Total Losses
PT (Q3) (W) 49 55 55 21
Q4
Total Losses
PT (Q4) (W) 2 2 55 21
PT(SP) (W) 254 277 430 116
Power Transfer
of Single Phase
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PM (W) 1260 1470 1470 1176
Im (A) 0 0 140 140
Vm (A) 0 0 42 42
PG (W) 0 0 1470 1764
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PDM (W) 504 588 588 294
Pphase (W) 1764 2058 3528 3234
Pphase + PT(SP) (W) 2018 2335 3958 3350
Efficiency (%) 87 88 89 97
B13
B.3.11 Parallel MOSFETs- BUK75/7606-75B – Four Devices
Power Analysis
MOSFET: Q1, Q2, Q3 and Q4 Phillips Semi BUK75/7606-75B
o
RDS(ON)@Tj=175 C (Coeff: 2.7) 0.012 Ω 184 A (TC = 115 RDS(ON) Coefficient @Tj=175oC 2.7
tf 480 ns 80 A (T C = 144oC)
Multiply by a factor of 10
Speed (RPM) 200 1000 2000 4500
Q1
Conduction Losses
IRMS(Q1) (A) 18 18 18 13
PC(Q1) (W) 4 4 4 2
Switching Losses
PS(Q1) (W) 16 19 19 0
Total Losses
PT (Q1) (W) 20 23 23 2
Q2
Conduction Losses
IRMS(Q2) (A) 25 25 18 13
PC(Q2) (W) 7 7 4 2
Switching Losses
PS(Q2) (W) 0 0 19 0
Total Losses
PT(Q2) (W) 7 7 23 2
Q3
Conduction Losses
IRMS(Q3) (A) 18 18 18 19
PCF(Q3) (W) 4 4 4 4
Switching Losses
PSF(Q3) (W) 16 19 19 0
Conduction Losses
IRMS(Q3) (A) 6 6 6 6
PCDM(Q3) (W) 0 0 0 0
Switching Losses
PSDM(Q3) (W) 0 0 0 0
Total Losses
PT (Q3) (W) 20 23 23 5
Q4
Total Losses
PT (Q4) (W) 0 1 23 5
PT(SP) (W) 193 215 368 62
Power Transfer
of Single Phase
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PM (W) 1260 1470 1470 1176
Im (A) 0 0 140 140
Vm (A) 0 0 42 42
PG (W) 0 0 1470 1764
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PDM (W) 504 588 588 294
Pphase (W) 1764 2058 3528 3234
Pphase + PT(SP) (W) 1957 2273 3896 3296
Efficiency (%) 90 91 91 98
B14
B.3.12 Parallel MOSFETs- PSMN009-100W – Two Devices
Power Analysis
MOSFET: Q1, Q2, Q3 and Q4 Phillips Semi PSMN009-100W
RDS(ON)@Tj=175oC (Coeff: 2.7) 0.024 Ω 184 A (TC = 115oC) RDS(ON) Coefficient @Tj=175oC 2.7
Speed (RPM) 200 1000 2000 4500
Q1
Conduction Losses
IRMS(Q1) (A) 35 35 35 26
PC(Q1) (W) 30 30 30 16
Switching Losses
PS(Q1) (W) 63 74 74 1
Total Losses
PT (Q1) (W) 93 103 103 17
Q2
Conduction Losses
IRMS(Q2) (A) 49 49 35 26
PC(Q2) (W) 60 60 30 16
Switching Losses
PS(Q2) (W) 0 0 74 1
Total Losses
PT(Q2) (W) 60 60 103 17
Q3
Conduction Losses
IRMS(Q3) (A) 35 35 35 38
PCF(Q3) (W) 30 30 30 36
Switching Losses
PSF(Q3) (W) 63 74 74 1
Conduction Losses
IRMS(Q3) (A) 13 13 13 13
PCDM(Q3) (W) 4 4 4 4
Switching Losses
PSDM(Q3) (W) 0 0 1 1
Total Losses
PT (Q3) (W) 97 108 108 42
Q4
Total Losses
PT (Q4) (W) 4 4 108 42
PT(SP) (W) 506 550 844 238
Power Transfer
of Single Phase
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PM (W) 1260 1470 1470 1176
Im (A) 0 0 140 140
Vm (A) 0 0 42 42
PG (W) 0 0 1470 1764
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PDM (W) 504 588 588 294
Pphase (W) 1764 2058 3528 3234
Pphase + PT(SP) (W) 2270 2608 4372 3472
Efficiency (%) 78 79 81 93
B15
B.3.13 Parallel MOSFETs- PSMN009-100W – Four Devices
Power Analysis
MOSFET: Q1, Q2, Q3 and Q4 Phillips Semi PSMN009-100W
o o o
o
Speed (RPM) 200 1000 2000 4500
Q1
Conduction Losses
IRMS(Q1) (A) 18 18 18 13
PC(Q1) (W) 7 7 7 4
Switching Losses
PS(Q1) (W) 32 37 37 1
Total Losses
PT (Q1) (W) 39 44 44 5
Q2
Conduction Losses
IRMS(Q2) (A) 25 25 18 13
PC(Q2) (W) 15 15 7 4
Switching Losses
PS(Q2) (W) 0 0 37 1
Total Losses
PT(Q2) (W) 15 15 44 5
Q3
Conduction Losses
IRMS(Q3) (A) 18 18 18 19
PCF(Q3) (W) 7 7 7 9
Switching Losses
PSF(Q3) (W) 32 37 37 1
Conduction Losses
IRMS(Q3) (A) 6 6 6 6
PCDM(Q3) (W) 1 1 1 1
Switching Losses
PSDM(Q3) (W) 0 0 0 1
Total Losses
PT (Q3) (W) 40 45 45 11
Q4
Total Losses
PT (Q4) (W) 1 1 45 11
PT(SP) (W) 379 423 717 127
Power Transfer
of Single Phase
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PM (W) 1260 1470 1470 1176
Im (A) 0 0 140 140
Vm (A) 0 0 42 42
PG (W) 0 0 1470 1764
Im (A) 140 140 140 140
Vm (A) 36 42 42 42
PDM (W) 504 588 588 294
Pphase (W) 1764 2058 3528 3234
Pphase + PT(SP) (W) 2143 2481 4245 3361
Efficiency (%) 82 83 83 96
B16
Appendix C – Thermal Analysis Computations
The following sections present the calculations to select the most suitable heat sink for
the full-bridge synchronous rectification converter. A spreadsheet program was used to
implement the thermal calculations for a number of MOSFETs (Phillips BUK7506-75B,
Fairchild FDP047AN08A0 and FDP060AN08A0). Calculations were implemented in
order to evaluate the effect of different number of devices in parallel (2 and 4), variation
in ambient temperatures (45oC and 90oC), using alternative thermal conductor materials
and heat spreader sizes. These calculations were implemented for each ISG mode of
operation (starting, low speed motoring and high speed generating). The specific
objective of the analysis was to determine the optimum heat sink selection (Rθs-a value).
The detailed spreadsheet calculations for the above conditions are presented in sections
C.1 through C.12.
The value of sink-to-ambient thermal resistance (Rθs-a) is determined using Equation 5.4,
where PT(SP) is the total power dissipation of the MOSFETs from the single converter
phase, Tj(max) is the maximum junction temperature experienced by the MOSFETs (Q1,
Q2, Q3 or Q4) in the single phase converter, Ta, is the maximum ambient temperature,
Rθj-c is the thermal resistance of the TO-220 package, Rθc-hs is the thermal resistance of
the heat sink compound, Rθhs-tc is the thermal resistance of the heat spreader and Rθtc-s is
the thermal resistance of the thermal conductor.
The projected junction temperature experienced by the devices is determined using

Equation 5.5, where Tc, is the maximum case temperature and PT(device) is the total power
dissipation of the each MOSFET devices (Q1, Q2, Q3 or Q4).
The thermal resistance of the heat spreader (Rθhs-tc) and the thermal resistance of the
thermal conductor (Rθtc-s) is determined using Equation 5.2, where d is the thickness of
the material (m), λ is thermal conductivity (Wm-1oC-1) of the material and A is the cross
sectional area (m2)
The specific converter layout is illustrated in Figure C.1 for implementing two devices
in parallel and a side view of the structure illustrating the different thermal resistance
C1
layers (case, heat spreader, thermal conductor and heat sink) is illustrated in Figure C.2.
The heat flow path from junction to ambient is illustrated in Figure C.3.
Q3a Q3b Q1a Q1b
Heat
Thermal conductor
spreader
Q4a Q4b Q2a Q2b
Figure C.1: Top view of single phase converter layout
Figure C.2: Side view of single phase converter layout
C2
Figure C.3: The heat flow path from junction-to-ambient for the single phase converter
C3
Heatsink
Component Munufacturer Model Thermal conductor (tc) Heat spreader (hs)
MOSFET: Q1, Q2, Q3 and Q4 Fairchild FDP047AN08A0 Rθtc-s (o CW -1 ) 0.058 Rθhs-tc(o CW -1 ) 0.058
2
Data A (mm ) 1250 L (mm) 50
o -1
Rθj-c ( CW ) 0.48 d (mm) 0.25 W (mm) 25
o -1 -1o -1
Rθc-hs ( CW ) 0.1 λ (Wm C ) 3.5 d (mm) 16
Thermal Calculations Starting Low speed High speed A=LW(mm2) 1250
motoring generating λ (Wm-1oC-1) 220
Q1 200 rpm 1000 rpm 4500 rpm
o -1
Rθj-hs(Q1) ( CW ) 0.29 0.29 0.29 Operating conditions
o
TJ ( C) 127 130 104 Device # 2
o
Q2 TC ( C) 100
C4
o -1 o
Rθj-hs(Q2) ( CW ) 0.29 0.29 0.29 TA ( C) 45
o
TJ ( C) 113 113 104
Q3 Notes:
o -1
Rθj-hs(D1) ( CW ) 0.29 0.29 0.29 Rθj-c (oCW-1) = Thermal resistance - junction to case
o
TJ ( C) 127 131 109 Rθc-hs (oCW-1) = Thermal resistance - heatsink compound
Q4 Rθhs-tc (oCW-1) = Thermal resistance - heat spreader
o -1
Rθj-hs(D2) ( CW ) 0.29 0.29 0.29 Rθtc-s (oCW-1) = Thermal resistance - thermal conductor
o
TJ ( C) 101 101 109 Rθs-a (oCW-1) = Thermal resistance - sink to ambient
Single phase converter
o -1
Rθs-a(Single P hase) ( CW ) 0.19 0.17 0.48
C.1 FDP047AN08A0 – Two Devices – Ta = 45oC
Heatsink
2
Data A (mm ) 1250 L (mm) 50
o -1
Rθj-c ( CW ) 0.48 d (mm) 0.25 W (mm) 25
o -1 -1o -1
Rθc-hs ( CW ) 0.1 λ (Wm C ) 3.5 d (mm) 16
Q1 200 rpm 1000 rpm 4500 rpm
o -1
o
TJ ( C) 112 113 101 Device # 4
o
Q2 TC ( C) 100
C5
o -1 o
Rθj-hs(Q2) ( CW ) 0.15 0.15 0.15 TA ( C) 45
o
TJ ( C) 103 103 101
Q3 Notes:
o -1
o
o -1
o
o -1
C.2 FDP047AN08A0 – Four Devices – Ta = 45oC
Heatsink
2
Data A (mm ) 1250 L (mm) 50
o -1
Rθj-c ( CW ) 0.48 d (mm) 0.25 W (mm) 25
o -1 -1o -1
Rθc-hs ( CW ) 0.1 λ (Wm C ) 3.5 d (mm) 16
Q1 200 rpm 1000 rpm 4500 rpm
o -1
o
TJ ( C) 127 130 104 Device # 2
o
Q2 TC ( C) 100
C6
o -1 o
Rθj-hs(Q2) ( CW ) 0.29 0.29 0.29 TA ( C) 90
o
TJ ( C) 113 113 104
Q3 Notes:
o -1
o
o -1
o
o -1
C.3 FDP047AN08A0 – Two Devices – Ta = 90oC
Heatsink
2
Data A (mm ) 1250 L (mm) 50
o -1
Rθj-c ( CW ) 0.48 d (mm) 0.25 W (mm) 25
o -1 -1o -1
Rθc-hs ( CW ) 0.1 λ (Wm C ) 3.5 d (mm) 16
Q1 200 rpm 1000 rpm 4500 rpm
o -1
o
TJ ( C) 112 113 101 Device # 4
o
Q2 TC ( C) 100
C7
o -1 o
Rθj-hs(Q2) ( CW ) 0.15 0.15 0.15 TA ( C) 90
o
TJ ( C) 103 103 101
Q3 Notes:
o -1
o
o -1
o
o -1
C.4 FDP047AN08A0 – Four Devices – Ta = 90oC
Heatsink
2
Data A (mm ) 1250 L (mm) 50
o -1
Rθj-c ( CW ) 0.48 d (mm) 0.25 W (mm) 25
o -1 -1o -1
Rθc-hs ( CW ) 0.1 λ (Wm C ) 3.5 d (mm) 16
Q1 200 rpm 1000 rpm 4500 rpm
o -1
o
TJ ( C) 125 128 105 Device # 2
o
Q2 TC ( C) 100
C8
o -1 o
Rθj-hs(Q2) ( CW ) 0.29 0.29 0.29 TA ( C) 45
o
TJ ( C) 116 116 105
Q3 Notes:
o -1
o
o -1
o
o -1
C.5 FDP060AN08A0– Two Devices – Ta = 45oC
Heatsink
2
Data A (mm ) 1250 L (mm) 50
o -1
Rθj-c ( CW ) 0.48 d (mm) 0.25 W (mm) 25
o -1 -1o -1
Rθc-hs ( CW ) 0.1 λ (Wm C ) 3.5 d (mm) 16
Q1 200 rpm 1000 rpm 4500 rpm
o -1
o
TJ ( C) 111 112 101 Device # 4
o
Q2 TC ( C) 100
C9
o -1 o
Rθj-hs(Q2) ( CW ) 0.15 0.15 0.15 TA ( C) 45
o
TJ ( C) 104 104 101
Q3 Notes:
o -1
o
o -1
o
o -1
C.6 FDP060AN08A0– Four Devices – Ta = 45oC
Heatsink
2
Data A (mm ) 1250 L (mm) 50
o -1
Rθj-c ( CW ) 0.48 d (mm) 0.25 W (mm) 25
o -1 -1o -1
Rθc-hs ( CW ) 0.1 λ (Wm C ) 3.5 d (mm) 16
Q1 200 rpm 1000 rpm 4500 rpm
o -1
o
TJ ( C) 125 128 105 Device # 2
o
Q2 TC ( C) 100
C10
o -1 o
Rθj-hs(Q2) ( CW ) 0.29 0.29 0.29 TA ( C) 90
o
TJ ( C) 116 116 105
Q3 Notes:
o -1
o
o -1
o
o -1
C.7 FDP060AN08A0– Two Devices – Ta = 90oC
Heatsink
2
Data A (mm ) 1250 L (mm) 50
o -1
Rθj-c ( CW ) 0.48 d (mm) 0.25 W (mm) 25
o -1 -1o -1
Rθc-hs ( CW ) 0.1 λ (Wm C ) 3.5 d (mm) 16
Q1 200 rpm 1000 rpm 4500 rpm
o -1
o
TJ ( C) 111 112 101 Device # 4
o
Q2 TC ( C) 100
C11
o -1 o
Rθj-hs(Q2) ( CW ) 0.15 0.15 0.15 TA ( C) 90
o
TJ ( C) 104 104 101
Q3 Notes:
o -1
o
o -1
o
o -1
C.8 FDP060AN08A0– Four Devices – Ta = 90oC
Heatsink
MOSFET: Q1, Q2, Q3 and Q4 Phillips Semi BUK75/7606-75B Rθtc-s (o CW -1 ) 0.058 Rθhs-tc(o CW -1 ) 0.058
2
Data A (mm ) 1250 L (mm) 50
o -1
Rθj-c ( CW ) 0.48 d (mm) 0.25 W (mm) 25
o -1 -1o -1
Rθc-hs ( CW ) 0.1 λ (Wm C ) 3.5 d (mm) 16
Q1 200 rpm 1000 rpm 4500 rpm
o -1
o
TJ ( C) 123 125 104 Device # 2
o
Q2 TC ( C) 100
C12
o -1 o
Rθj-hs(Q2) ( CW ) 0.29 0.29 0.29 TA ( C) 45
o
TJ ( C) 114 114 104
Q3 Notes:
o -1
o
o -1
o
o -1
C.9 BUK75/7606-75B – Two Devices – Ta = 45 oC
Heatsink
2
Data A (mm ) 1250 L (mm) 50
o -1
Rθj-c ( CW ) 0.48 d (mm) 0.25 W (mm) 25
o -1 -1o -1
Rθc-hs ( CW ) 0.1 λ (Wm C ) 3.5 d (mm) 16
Q1 200 rpm 1000 rpm 4500 rpm
o -1
o
TJ ( C) 110 111 101 Device # 4
o
Q2 TC ( C) 100
C13
o -1 o
Rθj-hs(Q2) ( CW ) 0.15 0.15 0.15 TA ( C) 45
o
TJ ( C) 103 103 101
Q3 Notes:
o -1
o
o -1
o
o -1
C.10 BUK75/7606-75B – Four Devices – Ta = 45 oC
Heatsink
2
Data A (mm ) 1250 L (mm) 50
o -1
Rθj-c ( CW ) 0.48 d (mm) 0.25 W (mm) 25
o -1 -1o -1
Rθc-hs ( CW ) 0.1 λ (Wm C ) 3.5 d (mm) 16
Q1 200 rpm 1000 rpm 4500 rpm
o -1
o
TJ ( C) 123 125 104 Device # 2
o
Q2 TC ( C) 100
C14
o -1 o
Rθj-hs(Q2) ( CW ) 0.29 0.29 0.29 TA ( C) 90
o
TJ ( C) 114 114 104
Q3 Notes:
o -1
o
o -1
o
o -1
C.11 BUK75/7606-75B – Two Devices – Ta = 90 oC
Heatsink
2
Data A (mm ) 1250 L (mm) 50
o -1
Rθj-c ( CW ) 0.48 d (mm) 0.25 W (mm) 25
o -1 -1o -1
Rθc-hs ( CW ) 0.1 λ (Wm C ) 3.5 d (mm) 16
Q1 200 rpm 1000 rpm 4500 rpm
o -1
o
TJ ( C) 110 111 101 Device # 4
o
Q2 TC ( C) 100
C15
o -1 o
Rθj-hs(Q2) ( CW ) 0.15 0.15 0.15 TA ( C) 90
o
TJ ( C) 103 103 101
Q3 Notes:
o -1
o
o -1
o
o -1
C.12 BUK75/7606-75B – Four Devices – Ta = 90 oC
Appendix D – Experimental System
D.1 Schematics
Figure D.1.1: Interconnect block diagram
D1
Figure D.1.2: Interconnect block schematic
D2
Figure D.1.3: Schematic of converter circuit
D3
Figure D.1.4: Schematic of interface circuit (Murphy 2002)
D4
Figure D.1.5: Schematic of power supply circuit (sheet 1 of 2) (Murphy 2002)
D5
Figure D.1.6: Schematic of power supply circuit (sheet 2 of 2) (Murphy 2002)
D6
D.2 Photographs of the experimental system
Figure D.2.1: Bench test experimental system
Figure D.2.2: Converter and RL load
D7
Figure D.2.3: ADSP-21992 EZ-KIT Lite board
Figure D.2.4: Power supply circuitry
D8
Figure D.2.5: Interface electronics board
Figure D.2.6: Motor rig – SRM and Terco test apparatus
D9
Appendix E – ADSP-21992 EZ-KIT Lite
Evaluation System Board
The ADSP-21992 EZ-KIT Lite was developed by Analog Devices in order to allow for
fast and efficient system development. This board was employed in the experimental
set-up for the development and test of the SRM converter operation and current control
strategy. The EZ-KIT Lite development board facilitates access to all the capabilities of
the ADSP-21992 digital signal processor (DSP). The board’s features include (Analog
Devices 2003b):
• Analog Devices ADSP-21992 160 MHz, mixed-signal DSP

• USB debugging interface
• Analog input circuitry
• 8-channel 12-bit DAC (AD5328BRU) on SPI interface
• PWM outputs interface
• External memory interface
• Encoder interface circuitry
• General-purpose I/O interface
• UART interface (RS-232)
• CAN interface circuitry
• Flash memory (512K x 8)
• External SRAM (64K x 16)
• Interface connectors
• 14-pin emulator connector for JTAG interface
• Analog inputs connector
• DAC outputs connector
• PWM outputs connector
• Encoder interface connector
• SPORT connector
• RS-232 connector
• External memory interface connector
This development board was designed for use in conjunction with the VisualDSP++
development environment. VisualDSP++ runs on a Personal Computer (PC) and
provides a single, integrated project management and debugging environment for
advanced code development and debugging tasks to be performed. Access to the ADSP-
21992 processor on the EZ-KIT Lite board from the PC is achieved through a USB port
or an optional JTAG emulator. An assembly drawing for the ADSP-21992 EZ-KIT Lite
development board is shown in Figure E.1 (Analog Devices 2003b).
E1
Figure E.1: Assembly drawing for the ADSP-21992 EZ-KIT Lite development board
The ADSP-21992 is ideal for SRM control applications because of the number of
integrated special purpose and motor control peripherals. In the work described in this
thesis, a number of these special purpose units were employed in the operation and
control of the converter, which included the Analog-to-Digital Converter (ADC), the
PWM Generation unit, the Auxiliary PWM Generation unit, the Program Sequencer and
the Peripheral Interrupt Controller and the Flag I/O peripheral unit. A brief description
of each of these units is included below.
E1. Overview of the Analog-to-Digital Conversion unit

The ADC unit on the ADSP-21992 accurately converts up to eight independent analog
signals, such as the signals required for high performance current regulation, into digital
signals through a 14-bit pipeline flash ADC. The ADC clock rate is programmable with
a maximum ADC clock rate of 20MHz. The full conversion of a single channel takes
7.5 ADC clock cycles.
All eight analog inputs applied to the analog input connector on the EZ-KIT Lite board
(Connector P4 in Figure E.1) must be in the range from –1V to +1V. The analog
interface circuitry then converts the ±1V signals on the input connector to signals
centred on the ADSP-21992 reference voltage level (either the internally derived 1V
E2
level or the externally provided 1.024V level). Effectively, the analog interface circuits
offset the analog connector inputs by the reference voltage level.
A functional block diagram of the ADC unit of the ADSP-21992 is shown in Figure E.2
(Analog Devices 2003c). As can be seen, the eight input signals are divided into two
banks of four signals. VIN0-VIN3 makes up one bank while the other is comprised of
the signals on VIN4-VIN7. The internal multiplexers are used to connect the various
analog inputs to the ADC.
Figure E.2: Functional block diagram of the ADC unit
There are a number of different conversion modes that can be selected using bits 4-6 in
the ADCCTRL register. However, during the course of this project only one mode of
operation was employed: simultaneous sampling mode (selected by clearing bits 4-6).
In this mode, two analog inputs (one from each four-signal bank) are sampled
simultaneously. VIN0 and VIN4 are sampled first followed by the pairs VIN1/VIN5,
VIN2/VIN6 and VIN3/VIN7 with two cycles of the ADC clock between the sampling
of one pair of analog signals and the next. After each pair of inputs are converted, the
14-bit digital numbers are written in 2’s complement, left-aligned format to a dedicated
16-bit, register i.e. the ADC register ADC0 stores the converted result for the signal on
VIN0 etc. In addition, a dedicated bit is set in the ADCSTAT register. After the ADC
E3
has finished with all of the channels, an interrupt may be generated. Alternatively, the
ADCSTAT can be polled to detect successful conversion of a given pair of inputs.
There are a number of ways in which the conversion process can be started, determined
by bits 0-2 in the ADCCTRL register. For the work described in this thesis, the
conversion process was started by setting bit 1 of the SOFTCONVST register (a mode
selected by setting bits 0-2 of the ADCCTRL register).
E2. Overview of the PWM Generation Unit

The PWM generation unit is a flexible, programmable, three-phase PWM waveform
generator. The unit contains special functions that simplify the generation of the
required PWM switching patterns for control of the electronically commutated motor
(ECM). A special mode can also be implemented for the SRM. A block diagram of the
main functions of the PWM Generation Unit is given in Figure E.3 (Analog Devices
2003c).
Figure E.3: Block diagram of the main functional units of the PWM Generator unit.
The PWM generator produces six PWM output signals that consists of three high-side
drive signals (AH, BH and CH) and three low-side drive signals (AL, BL and CL). The
polarity of the PWM signals is determined by the PWMPOL input pin of the
E4
PWMSTAT register, so that either active HI or active LO PWM patterns can be
produced by tying the PWMPOL input pin high or low. The switching frequency and
the dead time of the generated PWM patterns are programmable using the PWMTM and
PWMDT registers. The PWMDT is internally forced to 0 by hardware when
PWMSR is low to signify that the switched reluctance mode is active. The duty-cycles
of the three pairs of PWM signals are directly controlled by the three duty-cycle control
registers (PWMCHA, PWMCHB and PWMCHC). On the EZ-KIT Lite board shown in
Figure E.1, access to the 6 PWM output signal pins on the ADSP-21992 chip is via
connector P10 (Analog Devices 2003b).
The PWMSTAT register provides status information about the PWM system. In this
project, the PWM polarity is always active HI PWM outputs (this requires placing
jumper JP5, on the EASY-KIT Lite board, in position 2-3 to connect the PWMPOL pin
to +3.3V), there is no external trip (this requires placing jumper JP5, on the EASY-KIT
Lite board, in position 2-3 to connect the PWMTRIP pin to +3.3V) and SRM mode is
activated (this requires placing jumper JP6, on the EASY-KIT Lite board, in position 1-
2 to connect the PWMSR pin to GND). The status of which half cycle is active is
provided by PWMPHASE bit (0 = 1st half and 1 = 2nd half). This register also contains
the interrupt bits to indicate an interrupt has occurred (TRIPIRQ bit =1 when interrupt
has occurred from and external trip and SYNCIRQ bit = 1 when an internal interrupt
has occurred due to the PWM synchronization signal). Only the synchronization
interrupt was used in this project. The source code must write a 1 to clear the interrupt
bit and this is usually done during an Interrupt Service Routine (ISR). An ISR is called
to update the duty-cycle values when the PWM generator unit channel is active.
The PWMCTRL register is central to activating and defining operating mode

characteristics of the PWM generator unit. It is configured to enable or disable PWM
generation (if PWM_EN bit is 1 or 0), implement double or single update mode (if
PWMDBL bit is 1 or 0), set the operation after trip (SYNC_EN PWMSYNC bit is set to
1 to continue interrupts and this requires that the external shutdown be disabled which
requires that the PWMTRIP pin is connected to a logic HI), select internal or external
synchronization signal (external if EXTSYNC bit is 1). As the external synchronization
signal was used in this project, the SYNCSEL bit was ignored.
E5
A number of additional registers must be configured or status read in order to
implement the PWM generator unit, including, the PWMSEG register, the PWMGATE
register and the PWMLSI register. Each of the six PWM output signals can be enabled
or disabled by separate output enable bits of the PWMSEG register. An additional 3
control bits on the PWMSEG register permits independent crossover of the two signals
of a PWM pair but this feature was not used in this project. The PWMGATE register
controls a high-frequency chopping signal that is mixed with the PWM signals to drive
pulse gate-drive transformers. This feature was not used in this project. When the SRM
mode is activate, the PWMLSI register is used to create the four SRM chopping modes:
hard chop, alternate chop, soft chop-bottom on and soft chop-top on. The low side
invert that is configured by the PWMLSI bits (1 = invert) is the only difference between
hard chop mode and alternate chop mode. Soft chop-bottom on uses a 100% duty-cycle
on the low side of the channel, while with the soft chop-top on it is the high side of the
channel that utilizes a 100% duty-cycle. Alternate chop mode was used in this project.
The PWM switching frequency is controlled by the 16-bit PWM period register,
PWMTM. The required code for the PWMTM register is determined using equation E.1
(Analog Devices 2003c).
f CK
PWMTM = (E.1)
2 × f PWM
The value of fCK is the same as that for the peripheral clock, HCLK, which was 64MHz
in this project and fPWM is the desired PWM switching frequency. Therefore, in this
project the PWM switching frequency was 25 kHz and this yielded a PWMTM register
code of 1280. The hexadecimal value (0x0500) was written to the PWMTM register.
Switched reluctance mode was selected in this project and this requires that both the
high-side and low-side duty-cycle values had to be set. The high-side duty-cycle values
are controlled by the six 16-bit duty-cycle registers. PWMCHA, PWMCHB and
PWMCHC registers control the high-side duty-cycles, while PWMCHAL, PWMCHBL
and PWMCHCL registers control the low-side duty-cycles. The high-side duty-cycles
are determined using formula E.2 (Analog Devices 2003c).
PWMCHx = ((d AH − 0.5) × PWMTM ) + PWMDT (E.2)
E6
The generic duty-cycle code PWMCHx is replaced by one of the three high-side duty-
cycle codes, PWMCHA, PWMCHB or PWMCHC. The PWM switching frequency
code was 1280 and the desired duty-cycle value (dAH) ranged from 0 – 1, while the dead
time code (PWMDT) was 0 in this application as the switched reluctance mode was
active. Therefore, a duty-cycle of 10% yielded a PWMCHx code of -512, and the
corresponding hexadecimal value (0xFE00) was written to the register. The low-side
duty-cycles are determined using formula E.3 (Analog Devices 2003c).
PWMCHxL = ((0.5 − d AH ) × PWMTM ) − PWMDT (E.3)
The generic duty-cycle code PWMCHxL is replaced by one of the three low-side duty-
cycle codes, PWMCHAL, PWMCHBL or PWMCHCL; otherwise the equation
components are the same as equation E.2. Therefore, a duty-cycle of 90% yields a
PWMCHxL code of -512, which corresponding hexadecimal value (0xFE00) is written
to the register. To save time during code execution the code for high-side channel was
assigned to the low-side channel as both values were the same as can be seen in the
examples. In this project, the alternate chop mode was selected and when in this
configuration all the low-side channels were inverted. So when the high-side channel
was on for a particular duty cycle, the corresponding low-side channel was off for the
same duty-cycle.
E3. Overview of the Auxiliary PWM Generation unit

The ADSP-21992 contains a two-channel, 16-bit, auxiliary PWM output unit that can
be programmed with variable duty-cycle, variable frequency and may operate in either
an independent or offset operating mode. On the EZ-KIT Lite board shown in Figure
E.1, access to the 2 auxiliary PWM output signal pins on the ADSP-21992 chip is via
connector P10 (Analog Devices 2003b). The switching signals can be externally
shutdown using the input pin, AUXTRIP . In this project, there was no external trip
(this requires placing jumper JP8, on the EASY-KIT Lite board, in position 2-3 to
connect the AUXTRIP pin to +3.3V) (Analog Devices 2003b).
The auxiliary PWM generator unit can operate in either independent or offset modes.
Only offset mode was used in this project, so it is the only mode that is discussed here.
In this mode, there is an offset time between the rising edge of channel one (AUX0) and
channel two (AUX1) and this implements alternate chop mode for switched reluctance
E7
mode operation. The switching frequency for the two channels is controlled by the
AUXTM0 register, the duty-cycles for channel one and two are controlled by the
AUXCH0 and AUXCH1 registers, respectively, and the offset time is controlled by the
AUXTM1 register.
The auxiliary PWM control register (AUXCTRL) enables the auxiliary PWM output
(AUX-EN bit, 1 = enable), defines operating mode (AUX_PH bit, 1 = independent, 0 =
offset) and enables the synchronization signal (AS_EN bit, 1 = enable) to allow for
interrupt operation. The auxiliary PWM status register (AUXSTAT) provides the
interrupt or raw status. The raw status bit (AUXTRIP, 1 = HI, 0 = LO, active trip)
defines if an external trip has occurred. The synchronisation interrupt bit (AS_IRQ, 1 =
occurred) provides notification that an interrupt has occurred on the synchronisation
signal while the trip interrupt bit (AT_IRQ, 1 = occurred) signals that an interrupt has
occurred due to and external trip. A 1 must be written to these interrupt bits to clear
them. A synchronisation interrupt was only used in this project. An ISR is called to
update the duty-cycle values when the auxiliary PWM generator unit is active. The
offset time also has to be updated during his routine.
The auxiliary PWM switching frequency for both channels is controlled by the 16-bit
auxiliary PWM period register, AUXTM0, in offset mode. The required AUXTM0
register code is determined using equation E.4 (Analog Devices 2003c).
TAUX0
AUXTM0 = (E.4)
t CK
The value of tCK is the fundamental timing unit of the auxiliary PWM unit is the
switching period of the peripheral clock, HCLK, and had a frequency 64MHz in this
project. This yields a fundamental time increment of tCK of 15.625nS. The auxiliary
PWM switching period, TAUXO, for a switching frequency of 25 kHz is 40µs and this
yields an AUXTM0 register code of 2559. The hexadecimal value (0x09FF) must be
written to the AUXTM0 register.
In offset mode, the AUXTM1 register defines the offset time from the rising edge of the
signal for the AUX0 output to that of AUX1 output. The required AUXTM1 register
code is determined using equation E.5 (Analog Devices 2003c).
E8
TOFFSET
AUXTM1 = −1 (E.5)
t CK
The offset time, TOFFSET, is equivalent to the on-time for the AUX0 channel, which in
turn depends on the duty-cycle of the channel. Therefore, the offset is equal to the duty-
cycle multiplied by the auxiliary PWM switching period (40µs). For a 10% duty-cycle
for the AUXO channel, this yields an AUXTM1 register code of 255. The hexadecimal
value (0x00FF) must be written to the AUXTM1 register. The code is updated at the
beginning of the next AUX0 switching period.
The duty-cycle or on-time for the AUX0 channel is controlled by the AUXCH0 register
and is updated at the beginning of the next AUX0 switching period. The required
AUXCH0 register code is determined using equation E.6 (Analog Devices 2003c).
d AUX0 × TSW
AUXCH0 = (E.6)
t CK
For a 10% duty-cycle, dAUX0, for the AUXO channel and an auxiliary PWM switching
period, TSW, of 40µs, yields an AUXCH0 register code of 256. The hexadecimal value
(0x0100) must be written to the register. The duty-cycle or on-time for the AUX1
channel is controlled by the AUXCH1 register and is updated at the beginning of the
next AUX1 switching period. The formula to determine the register code is the same as
equation E6. The duty-cycle for AUX1 is determined using equation E7 (Analog
Devices 2003c).
dAUX1 = 1 – dAUX0 (E.7)
E4. Overview of the Program Sequencer and the Peripheral Interrupt

Controller
In order to understand how to implement interrupt service routines with the ADSP-
21992 DSP, a fundamental understanding of the program sequencer and the peripheral
interrupt controller is required. The program sequencer of the ADSP-21992 controls
program flow where it is constantly providing the address of the next instruction to be
executed by other parts of the DSP. Program flow in the DSP is mostly linear yielding
sequential executing of program instructions. There are non-sequential program
E9
structures that direct the DSP to execute an instruction that is not at the next sequential
address and these structures include: loops, subroutines, jumps, interrupts and idle. The
program flow variations for the DSP are illustrated in Figure E.4 (Analog Devices
2003c).
Figure E.4: Program Flow Variations for the DSP
The sequencer manages the execution of these program structures by selecting the
address of the next instruction to execute. As part of this process, the sequencer handles
the following tasks: Increment the fetch address, maintains stacks, evaluates conditions,
decrements the loop counter, calculates new addresses, maintains an instruction cache
and handles interrupts. To accomplish these tasks, the sequencer uses the blocks shown
in Figure E.5. The sequencer’s address multiplexer selects the value of the next fetch
address from several possible sources. This address enters the instruction pipeline and
ends with the program counter (PC). The pipeline contains the 24-bit addresses of the
instructions currently being fetched, decoded, and executed. The PC couples with the
PC stack, which stores the return addresses.
E10
Figure E.5: Program Sequencer Block Diagram
To manage events, the sequencer’s interrupt controller handles interrupt processing,

determines whether an interrupt is masked, and generates the appropriate interrupt
vector address. A set of system control registers configures or provides input to the
sequencer. These registers include ASTST, MSTAT, CCODE, IMASK, IRPTL and
ICNTL.
Interrupts can stem from a variety of conditions, both external and external to the
processor. In response to an interrupt, the sequencer processes a subroutine call to a
predefined address, the interrupt vector. The DSP assigns a unique vector to each
interrupt. The DSP core supports five fixed interrupt sources (Emulator, Reset,
Powerdown, Loop and PC Stack and Emulation kernel Interrupts) and up to 12 user
assignable interrupts. The user assignable interrupts are generated by the peripheral
units of the ADSP-21992 and their connection and prioritization is managed by the
Peripheral Interrupt Control Unit.
E11
The masking of the various interrupts is controlled by the IMASK processor register
and the latching of the pending interrupts is controlled by the IRPTL processor register.
There are dedicated bits of these registers that are associated with the various fixed and
user assignable interrupt sources. Each interrupt has a dedicated 32-bit address in the
interrupt vector table. The dedicated bits and associated vector addresses for each of the
ADSP-21992 five core and 12 user assignable interrupts are shown in Table E.1
(Analog Devices 2003c).
Table E.1: ADSP-21992 Interrupt Mask
E12
Table E.2: Peripheral Interrupt Sources
The ADSP21992 has 32 individual peripheral interrupt sources that are tabulated and
identified in Table E.2 (Analog Devices 2003c). There is a unique 4-bit code that allows
each peripheral interrupt source to be assigned to one of the 12 user-assignable
interrupts of the DSP core. Four of the 4-bit codes are contained in each of the 8, 16-bit
Interrupt Priority Registers (Interrupt Priority Register 0 (IPR0) to Interrupt Priority
Register 7 (IRP7)).
E13
The user may write a value between 0x0 and 0xB to each 4-bit location in order to
connect the particular interrupt source to the corresponding user assignable interrupt of
the ADSP-21992 DSP. Writing a value of 0x0 connects the peripheral interrupt to the
USR0 user assignable interrupt of the DSP core, while writing a value of 0xB connects
the peripheral interrupt to the USR11 user assignable interrupt. The core interrupt USR0
is the highest priority user interrupt, while USR11 is the lowest priority. Writing a value
between 0xC and 0xF effectively disables the peripheral interrupt by not connecting it
to any ADSP-21992 DSP interrupt input. The masking and interrupt flagging are
controlled by the core registers IMASK and IRPTL. There are additional features and
registers that are available if the user wishes to assign more than one peripheral interrupt
to any given DSP interrupt but this scheme was not implemented in this project. Only
two peripheral interrupts were assigned in this project, PWMSYNC_IRQ was assigned
to USR0 and AUXSYNC_IRQ was assigned to USR1.
E5. Overview of the Flag I/O peripheral unit

The Flag I/O (FIO) unit is a parallel I/O interface that supports 16 bi-directional general
purpose I/O signals (PF0-PF15). Each flag bit can be individually configured as an
input or output depending on the contents of the direction (DIR) register. They can also
be used as a source for an interrupt (Analog Devices 2003c).
When a flag is configured as an input, the FIO can be programmed to invert the input
value, latch a level or detect a signal edge (rising, falling or both) depending on the
contents of the POLAR, EDGE and BOTH registers. When a flag is configured as an
output, the output value is driven from the FLAG register. The 16-bit FLAG register
exhibits ‘sticky’ behaviour; only writing a ‘1’ to a bit can modify that bit. Writing a ‘1’
to a bit of the FLAG register at the even address 0x0002 (FLAGC) clears the FLAG bit
while writing a ‘1’ to a bit of the FLAG register at the odd address 0x0003 (FLAGS)
sets the FLAG bit. Writing a ‘0’ to any bit in either FLAGC or FLAGS has no effect.
On the EZ-KIT Lite board shown in Figure E.1, access to the 16 FIO programmable
flag pins on the ADSP-21992 chip is via connector P8.
E14
Appendix F – VisualDSP++ 3.0 and implementing
the ISR
VisualDSP++ 3.0 has a number of built in or intrinsic functions designed to make
programming a DSP in a C environment even more user-friendly. These intrinsic
functions require the inclusion of a number of header files in order to access these
functions and to define symbolic names for the registers of the ADSP-21992 DSP. The
specific registers and functions that were accessed in this project include Flag I/O, ADC,
PWM, auxiliary PWM and the various functions required to implement an interrupt
service routine (ISR) to update the PWM and the auxiliary PWM duty cycles.
F.1 Overview of VisualDSP++ 3.0 Features

The built-in functions eliminates the need for referencing-dereferencing scheme using
cast pointers (*(int*)) to access memory-mapped registers. Access to all non-memory-
mapped registers in C required embedded assembly source code. The intrinsic
functions to access non-memory-mapped system register are sysreg_read and
sysreg_write and the functions to access all the memory-mapped I/O space registers are
io_space_read and io_space_write (Analog Devices 2003d). These four functions and
other bit manipulation instructions are defined in the header file sysreg.h. There is also
an enumerated type in this header file that provides listing of the system registers that
can be accessed using sysreg_read and sysreg_write.
The io_space_read and io_space_write intrinsic functions require the inclusion of the
architecture definition header files, adsp-21992.h, adsp-2199x.h, def2191.h, def219x.h
when using the ADSP-21992 DSP. The definition header file, adsp-21992.h, allows
access the symbolic names that are unique to the ADSP-21992, specifically the CAN
system registers. The common header file, adsp-2199x.h, is required to access the
registers that are common to all the ADSP-2199x DSP family, which includes I/O
registers, interrupt controller registers, PWM registers, encoder interface unit register,
auxiliary PWM registers, watchdog timer registers, ADC module registers. The
definition file, def2191.h, is also required when using the ADSP-21992 DSP as it
contains additional symbolic names including: the DMA Bus Bridge, External Access
Bridge, JTAG debut, clock and system control, interrupt controller, SPORT, SPI and
UART communications controllers and the timer. The definition file, def219x.h, defines
F１
the symbolic names for all the system register bit and addresses that are common to all
ASDP-219x DSP cores. The Flag I/O block defined in the header file, adsp-2199x.h,
supersedes the definitions already called out in the header file, def2191.h. The
definitions for the interrupt controller provided in the header file, adsp-2199x.h, results
in those in header definition file, def2191.h, been undefined.
F.2 Implementation of the intrinsic functions

An example of using the built-in functions is configuring the Flag I/O (FIO) Peripheral
unit to set the general purpose I/O signal pins 0-3 (PF0-3) to be outputs (Analog
Devices 2003d). This information is contained in the DIR register, where a 1 means that
the particular pin is an output. The first step is to access the correct IO page for the DIR
register. The sysreg_write intrinsic function is used to access the IOPG system register.
The first argument is the register to be written to, sysreg_IOPG (as defined in the
enumeration in sysreg.h). The second argument is the value to be written to the IOPG
register, FIO_Page, which is #defined in adsp-2199x.h to be 0x06 (the page offset
required to access the FIO register set).
Once the IOPG register is set appropriately to access the FIO registers, the DIR register
is accessible. The DIR register is an IO space register and is written to using the
io_space_write intrinsic function. Here, the arguments are the address to be written to,
FIO_DIR, which is #defined in adsp-2199x.h to be 0x000 (the physical I/O address on
IO page 6 of the DIR register) and the value to be written to that address, 0x000F,
where bits 0-3 are set to 1 to enable the corresponding PF pins 0-3 to be outputs. The
source code to implement these instructions is shown in Figure F.1. Access to the ADC,
PWM, and auxiliary PWM is done in the same manner, as illustrated in the code
examples shown in Figure F.2 – F.4, respectively.
#include <sysreg.h>
#include <adsp-2199x.h>
main( )
{
sysreg_write (sysreg_IOPG, FIO_Page); //Go to FlO page
io_space_write (FIO_DIR, 0x000F); //Set DIR register for PF0-3 as outputs
}
Figure F.1: Configuring the Flag I/O pins 0 3 (PF0-3) of the FIO Peripheral Unit as
outputs.
F２
#include <sysreg.h>
main( )
{
sysreg_write(sysreg_IOPG,ADC_Page); //Go to ADC page
io_space_write(ADC_CTRL, 0x0F0F); //max clock, sim.samp.mode
//and pwmsync convert trigger (000),
//SOFTCONVST(111)
}
Figure F.2: Go to ADC page, set ADC clock to maximum, implement simultaneous
sampling mode, and trigger when SOFTONVST bit is active
#include <sysreg.h>
main( )
{
sysreg_write(sysreg_IOPG, PWM0_Page); //Go to PWM page
io_space_write(PWM0_SEG, 0x03F); //Disable all pwm outputs 0x03F
io_space_write(PWM0_CTRL, 0x000); //Disable PWM generation
io_space_write(PWM0_TM, 0x0500); //25kHz PWM Switching Frequency
io_space_write(PWM0_CHA, 512); //Set PWM AH Duty Cycle to 90%
io_space_write(PWM0_CHAL, 512); //Set PWM AL Duty Cycle to 10%
io_space_write(PWM_SI, 0x0); //No invertion of low side outputs,
//SR Alternate mode
io_space_write(PWM0_SYNCWT, 63); //Set pwm sync signal width for 500ns
}
Figure F.3: Go to PWM page, disable the PWM output and generation, set the
switching frequency to 25kHz, set the duty cycle for the high-side channel to 90% and
the duty-cycle of the low-side channel to 10%, invert the low-side output due to
switched reluctance mode and enable the PWM synchronization signal with the pulse
width to 500ns
#include <sysreg.h>
main( )
{
sysreg_write(sysreg_IOPG, Aux_PWM_Page); //Go to AUX PWM page
io_space_write(AUX_TMA0, 2559); //25kHz PWM Switching Frequency
io_space_write(AUX_CHA0, 256); //10% initial high-side duty-cycle
io_space_write(AUX_CHA1, 2304); //90% initial low-side duty-cycle
io_space_write(AUX_TMA1, 255); //Offset for low-side duty cycle
io_space_write(AUX_CTRL, 0x000); //Disable outputs and interrupt
}
Figure F.4: Go to auxiliary PWM page, set the switching frequency to 25kHz, set the
duty-cycle for the high-side channel to 90% and the duty-cycle of the low-side channel
to 10%, set the off-set for the low-side output, disable outputs and disable interrupts
F３
F.3 Implementation of the Interrupt Service routine in C
The interrupt handling in C utilizes the header-defined interrupt(signal, subroutine)
module to take care of everything [(Analog Devices 2002) and (Analog Devices
2003d)]. This function associates a specific ISR module to be run for a given signal that
could be received during run-time. The list of possible signals (SIG_INT value) is
detailed in the signal.h header file. The interrupt handling function also sets the correct
bit in IMASK to enable servicing for that interrupt. In this project, the option to
prioritize the interrupts was utilized.
When using the ADSP-21992, the peripheral interrupt sources, PWMSYNC_IRQ and
AUXSYNC_IRQ, must be assigned to the user assignable core interrupts. The PWM
interrupt was assigned to the highest user assignable core interrupt, USR0, and the
auxiliary PWM interrupt was assigned to the second highest user assignable core
interrupt, USR1 (Analog Devices 2003c). The highest user assignable core interrupt,
USR0, has the signal number SIG_INT4 and the second highest user assignable interrupt,
USR1, has the signal number SIG_INT5 (Analog Devices 2002).
PWM_ISR() and AUXPWM_ISR() are the ISR functions that is called to service the user
assignable interrupts, USR0 and USR1, respectively. The ISR function is called to
service the interrupt once the correct bit is latched in the interrupt latch (IRPTL)
register. This bit must be cleared when the ISR is called. In addition to this interrupt-
registering scheme, all interrupts must be cleared or reset at the initial source code
execution, the user must globally enable interrupts by using the intrinsic function
enable_interrupt() that is defined in the sysreg.h header file. The source code
instructions to clear or reset the interrupts at the initial source code execution, assign
the user interrupts to the DSP core, assign the appropriate ISR to in corresponding
signal number and finally to globally enable all interrupts are implemented in an
initialization function in the project source code. The source code to implement these
initialization steps is shown in Figure F.5.
F４
#include <sysreg.h>
main( )
{
//---------------------------Reset Interrupts--------------------------------//
//Clear/Reset All Interrupts
sysreg_write(sysreg_IRPTL, 0x000);
sysreg_write(sysreg_ICNTL, 0x000);
sysreg_write(sysreg_IMASK, 0x000);
// Assign Peripheral interrupt to the core interrupt
sysreg_write(sysreg_IOPG, IntCtrl_Page); //Go to Peripheral Interrrupt
//Control page
io_space_write(IPR2, 0xF0); //Assign PWMSYNC_IRQ to highest user
//assignable core interrupt priority
//(USR0)
io_space_write(IPR6, 0xF1FF); //Assign AUXSYNC_IRQ to second highest
//user assignable core interrupt priority
//(USR1)
interrupt(SIG_INT4, PWM_ISR); //INT4 is for interrupt USR0
interrupt(SIG_INT5, AUXPWM_ISR); //INT5 is for interrupt USR1
//---------------------------Enable Interrupts---------------------------------//
enable_interrupts(); //Globally Enable Interrupt
}
Figure F.5: The source code to initialize the ISR functions.
The ISR functions, PWM_ISR() and AUXPWM_ISR(), have a common requirement to

clear the interrupt latch bit associated to that ISR. In the case of the PWM, this latch bit
is the SYNCIRQ bit of the PWM status (PWMSTAT) register and for the auxiliary
PWM the latch bit is the AS_IRQ bit of the auxiliary PWM control (AUXCTRL)
register. Once the interrupt latch bit is cleared, the specific source code instruction for
that particular ISR can be implemented. An example of the source code to clear this
interrupt bit is shown in Figure F.6
//----------------------------------PWM ISR--------------------------------------//
#pragma interrupt
void PWM_ISR()
{
sysreg_write(sysreg_IOPG, PWM0_Page); // Go to PWM i/o page
PWM_CycleA = io_space_read(PWM0_STAT); // Read the PWM Status Register
io_space_write(PWM0_STAT, 0x20A); // Clear PWM Interupt Latch bit
Figure F.6: The source code for the ISR function to clear the interrupt latch bit.
F５
Appendix G – DSP Implemented Source Code
G.1 srm_pipwm_current_reg.c
//--------------------------------------------------------------------------------------
// File: srm_pipwm_current_reg.c
// Date: 29/06/2008
// Author: Anthony Murphy
// Purpose:This is the source file to implement the PWM and AUXPWM Generator Unit and
// read the ADC system on the ADSP-2199x
// Ref: ADSP-2199x Mixed Signal DSP Controller Hardware Reference,
// P/N:82-000640-01 Rev. 0, 2003
// Detail: This file used to initialise the PWM Generator Unit, then in the ISR
// implements adc reading and calculate and update the pwm duty cycles and
// implement reading a switch to determine if mag/free mode or demag mode. The
// pwm controller was finally implemented in its own function,
// motor_control_a() for phase A.
//--------------------------------------------------------------------------------------
//--------------------------------------------------------------------------------------
//Include files
//--------------------------------------------------------------------------------------
#include <signal.h>
#include <math.h>
#include<sysreg.h>
#include<adsp-21992.h>
//--------------------------------------------------------------------------------------
//Symbolic constants
//--------------------------------------------------------------------------------------
#define P 0.015 // Value of proportional gain

#define I 0.1 // Value of integral gain
//--------------------------------------------------------------------------------------
//Function prototypes
//--------------------------------------------------------------------------------------
void initialization (void);
void PWM_ISR(); // PWM ISR Prototype
void AUXPWM_ISR(); // AUXPWM ISR Prototype
void motor_control_a(); // SRM Phase A current control prototype
void commutation_four_phase(); // SRM commutation of all four phases prototype
void read_rotor_angle(); // Read rotor angle
//--------------------------------------------------------------------------------------
//Global data variables
//--------------------------------------------------------------------------------------
int i=0,j=0,count,loop=10;
int DutyCyleUpdateA=0, DutyCyleUpdateB=0, DutyCyleUpdateC=0, DutyCyleUpdateD=0;
int EnablePhaseA=0, EnablePhaseB=0, EnablePhaseC=0, EnablePhaseD=0;
int EnableSRM=0, MagFreeDemagA=0, MagFreeDemagB=0, MagFreeDemagC=0, MagFreeDemagD=0;
int DemagPWMA=0, DemagPWMB=0, DemagPWMC=0, DemagPWMD=0, PWM_CycleA=0;
int StartupA=1, StartupB=1, StartupC=1, StartupD=1;
int ADCStatusA, ADCValueA, DutyCycleTempA, DutyCycleAL, DutyCycleAH;
int ADCStatusB, ADCValueB, DutyCycleTempB, DutyCycleBL, DutyCycleBH;
int ADCStatusC, ADCValueC, DutyCycleTempC, DutyCycleCL, DutyCycleCH;
int ADCStatusD, ADCValueD, DutyCycleTempD, DutyCycleAUX0, DutyCycleAUX1, AUXPWMOffset;
int AngleMinA=10, AngleMaxA=80, AngleMinB=200, AngleMaxB=220, AngleMinC=230,
AngleMaxC=240;
int AngleMinD=250, AngleMaxD=260, AngleDemagA=0, AngleDemagB=0, AngleDemagC=0,
AngleDemagD=0;
int AngleDemag=20, AngleOffA=0, AngleOnA=0, RotorAngle=0;
int MagFreePeriodPWM = 1280; //25kHz = 1280, 50kHz = 640, 100kHz = 320;
int DemagPeriodPWM = 320; //200kHz = 160, 500kHz = 64, 1Mhz = 32;
float Ith=0, Ia=0, Ib=0, Ic=0, Id=0, Iref = 0;
float error=0, errorlow = 0, errorhigh =0, Da=0, integral=0, deltaT=0.00004;
float I_a[4096], Ilow = 0, Ihigh = 0;
float tempA=0, tempB=0, tempC=0, tempD=0;
G1
//--------------------------------------------------------------------------------------
//Main Program
//--------------------------------------------------------------------------------------
main()
{
initialization ();
Iref = 90.00;
Ith = 5.00;
Ihigh = Iref*1.1;
Ilow = Iref*0.9;
errorhigh = Iref-Ilow;
errorlow = Iref-Ihigh;
while(1) // wait for interrupts

{
commutation_four_phase(); //Activate SRM commutation
}
}
G2
//--------------------------------------------------------------------------------------
//Function: initialization()
//Description: To initial the DSP clock speed, interrupts, PWM, auxiliary PWM, FIO and
ADC
//Inputs: -
//Returns: -
//--------------------------------------------------------------------------------------
void initialization (void)

{
//Increase the dsp clock speed from 32MHz to 128Mhz
//Bypass mode enable and PLL shutoff
asm("iopg = 0x00;");
asm("ax1 = 0x0120;");
asm("io(0x200) = ax1;");
asm("nop;");
//Bypass mode enable and write the correct multiplier value

asm("ax1 = 0x0950;");
asm("io(0x200) = ax1;");
asm("nop;");
//Bypass mode disable and maintain the same mulitplier value as before
asm("ax1 = 0x0850;");
asm("io(0x200) = ax1;");
asm("nop;");
//---------------------------Reset Interrupts--------------------------------
//Clear/Reset All Interrupts
sysreg_write(sysreg_IRPTL, 0x000);
sysreg_write(sysreg_ICNTL, 0x000);
sysreg_write(sysreg_IMASK, 0x000);
//----------------------Initialize Interrupts--------------------------------
// Assign Peripheral interrupt to the core interrupt
sysreg_write(sysreg_IOPG, IntCtrl_Page); //Go to Peripheral Interrrupt
//Control page
io_space_write(IPR2, 0xF0); //Assign PWMSYNC_IRQ to highest user
//assignable core interrupt priority
//(USR0)
io_space_write(IPR6, 0xF1FF); //Assign AUXSYNC_IRQ to second highest
//user assignable core interrupt priority
//(USR1)
interrupt(SIG_INT4, PWM_ISR); //INT4 is for interrupt USR0
//interrupt(SIG_INT5, AUXPWM_ISR); //INT5 is for interrupt USR1
//-------------------------Enable Interrupts---------------------------------
//Globally Enable Interrupt
enable_interrupts();
//-------------------------Intitialize PWM-----------------------------------
io_space_write(PWM0_CTRL, 0x000); //Disable PWM generation
io_space_write(PWM_SI, 0x0); //No invertion of low side outputs,
//SR Alternate mode
io_space_write(PWM0_SYNCWT, 31); //Set pwm sync signal width for 500ns
//-------------------------Intitialize AUX PWM-------------------------------

// Go to AUX PWM i/o page
sysreg_write(sysreg_IOPG, Aux_PWM_Page);
io_space_write(AUX_CTRL, 0x000); //Disable outputs and interrupt
//-------------------------Intitialize FIO-------------------------------------
sysreg_write(sysreg_IOPG,FIO_Page);
io_space_write(FIO_DIR, 0xFFFC); // set outputs except PFO and PF1
io_space_write(FIO_EDGE, 0x0000); // PF0 and PF1 are level sensitive
io_space_write(FIO_POLAR, 0x0000); // PF0 and PF1 are active high input
//-------------------------Intitialize ADC-------------------------------------
sysreg_write(sysreg_IOPG,ADC_Page); //Intitialize ADC
io_space_write(ADC_CTRL, 0x0F07); //max clock, sim.samp.mode
//and pwmsync convert trigger (000),
//SOFTCONVST(111)
}
G3
//--------------------------------------------------------------------------------------
//Function: PWM_ISR()
//Description: The pwm interrupt service routine
//Inputs: -
//Returns: -
//--------------------------------------------------------------------------------------
#pragma interrupt
void PWM_ISR()
{
PWM_CycleA = io_space_read(PWM0_STAT); // Read the PWM Status Register
io_space_write(PWM0_STAT, 0x20A); // Clear PWM Interupt Latch bit
PWM_CycleA = PWM_CycleA & 0x0001;
if(PWM_CycleA==1) // If in the 2nd half cycle, update pwm

{
sysreg_write(sysreg_IOPG, FIO_Page); // Go to FIO page
io_space_write(FIO_FLAGS, 0x0020); // Activate (PF5)
io_space_write(FIO_FLAGC, 0x0010); // Disable (PF4) =>2 for 2nd half cycle
}
else // If in the 1st half cycle, no update of pwm
{
sysreg_write(sysreg_IOPG, FIO_Page); // Go to FIO page
io_space_write(FIO_FLAGS, 0x0010); //Activate (PF4)
io_space_write(FIO_FLAGC, 0x0020); //Disable (PF5) =>1 for 1st half cycle
}
if((DutyCyleUpdateA==1)&&(PWM_CycleA==1))
{
//-------------------------------ADC Start --------------------------------
sysreg_write(sysreg_IOPG,ADC_Page);
io_space_write(ADC_SOFTCONVST, 0x1); // Set the soft convert start bit
ADCStatusA = 1;
while(!(ADCStatusA & 0x0001)) //Wait till conversion of ADC4
ADCStatusA = io_space_read(ADC_STAT); //is complete
ADCValueA = io_space_read(ADC_DATA4); //Read value from ADC4

ADCValueA=ADCValueA>>2; //ADC value is leftmost 14 bits =>
//shift right 2 bits
tempA=(float)ADCValueA; //Cast convert to a float for
//calculations
Ia=(tempA/0x1FFF)*90.09; //Determine current measurement
//value, divide by 8191 or 0x1FFF
//and scale 90 if 11.1ohms
tempB=ceilf(Ia*10); //Current value rounded off for
Ia=(tempB/10)+1.6; //to one decimal place
if(j<4097)
{
I_a[j]=Ia; //Capture initial current data for plotting
if(j==4096)
j=-1;
j++;
}
//-------------------------------Duty Cycle Update Algorithm --------------
tempA = ceilf(tempA); //Duty cycle value rounded off for

//duty cyle calculation
error = Iref-Ia;
integral = integral+(error*deltaT); //Calculate new integral value
Da = (P*error)+(I*integral); //If error in range, cal duty cycle
//-------------------------------Duty Cycle Update Start-------------------//

if(MagFreeDemagA==1) //Mag/Free mode of operation
{
if((Ia>Ith)&&(Ia<=Ihigh)) //Implement full PWM duty cycle update
{
DutyCycleAH = (int)((Da-.5)*1280); //Calculate high side duty cycle
DutyCycleAL = DutyCycleAH; //Due to SR mode and alt mode low and
//high side dutycycle are the same.

io_space_write(PWM0_CHA, DutyCycleAH); //Update duty cycle for high side
io_space_write(PWM0_CHAL, DutyCycleAL); //Update duty cycle for low side
StartupA=0;
}
G4
else
{
if(Ia>Ihigh) //Current too high=>Freewheel
{
DutyCycleAH = (int)((0-.5)*1280); //Calculate high side duty cycle
//hide side dutycycle are the same.
sysreg_write(sysreg_IOPG, PWM0_Page); //Go to PWM i/o page

io_space_write(PWM0_CHA, DutyCycleAH); //0% duty cycle for high side
io_space_write(PWM0_CHAL, DutyCycleAL); //100% duty cycle for low side
StartupA=1;
}
else //Implement Startup PWM duty cycle update
{
DutyCycleAH = (int)((0.025-.5)*1280);//Calculate high side duty cycle
//hide side dutycycle are the same.
io_space_write(PWM0_CHAL, DutyCycleAL);//99% duty cycle for low side
}
}
}
else
{
if(MagFreeDemagA==0) //Demag mode of operation
{
if(Ia>Ith)
{
Da = 0;
DutyCycleAL = (int)((.5-Da)*320); //Calculate low side duty cycle
DemagPWMA=1;
}
else
{
Da = 0;
Da = 1;
DutyCycleAL = (int)((.5-Da)*320); //Calculate low side duty cycle
DemagPWMA=0;
}
}
}
DutyCyleUpdateA=0;
}
G5
//--------------------------------------------------------------------------------------
//Function: motor_control_a()
//Description: Implements the converter modes of operation magnetisation/freewheeling,
// demagnetisation depending on control signals
//Inputs: -
//Returns: -
//--------------------------------------------------------------------------------------
void motor_control_a()
{
if ((MagFreeDemagA==1)&&(EnablePhaseA==1))//Mag/Free mode of operation - Commutation
{
if(DutyCyleUpdateA==0) //Implement Mag/Free mode PWM - ISR
{
//------------------------------- PWM Start ---------------------------//
sysreg_write(sysreg_IOPG, FIO_Page);// Go to FIO page
io_space_write(FIO_FLAGC, 0x0100); //Disable MD2 (PF8)
io_space_write(FIO_FLAGS, 0x0200); //Activate MQ2 (PF9)
// Go to PWM i/o page

sysreg_write(sysreg_IOPG, PWM0_Page);
io_space_write(PWM0_SEG, 0x00F); //Activate AH and AL)
//Enable Phase A PWM outputs, 0 = enable,
//1 = disable. Disable all pwm outputs 0x03F,
//enable all pwm outputs 0x0, only enable
//phase A 0x00F, only enable phase B 0x033,
//only enable phase C 0x03C
io_space_write(PWM_SI, 0x1); //Inverts low side outputs, SR Alternate
//mode, disables when phase is off
//Invert all=0x7, invert none = 0x0, invert
//A=0x1, invert B=0x2 and invert C=0x4
io_space_write(PWM0_CTRL, 0x007); //Enable PWM generation with PWMSYNC signal.
//double update (0x007), single update (0x003
DutyCyleUpdateA = 1;
}
else;
//Do nothing, update duty cycle next time
}
else
{
if ((MagFreeDemagA==0)&&(EnablePhaseA==1)) //Demag mode of operation
{
if(DutyCyleUpdateA==0) //Implement Demag mode PWM
{
if(DemagPWMA==1) //Implement Full DemagPWM mode
{
sysreg_write(sysreg_IOPG, FIO_Page);//Go to FIO page
io_space_write(FIO_FLAGC, 0x0200); //Disable MQ2 (PF9) => end of commutation
//(mag and free mode)
io_space_write(FIO_FLAGS, 0x0100); //Activate MD2 (PF8) => demag mode
//1 = disable. only enable phase A
//Invert none = 0x0
//double update (0x007), single update (0x003
}
else
{
if (DemagPWMA==0) //Disable Full DeMag PWM mode
{
sysreg_write(sysreg_IOPG, FIO_Page);//Go to FIO page
io_space_write(FIO_FLAGC, 0x0300); //Disable MQ2 (PF9) and MD2 (PF8)
//1 = disable. Only enable phase A 0x00F
//Invert none = 0x0
//double update (0x007)
G6
}
}
}
else; //Do nothing, update duty cycle next time
}
else
{
if ((MagFreeDemagA==0)&&(EnablePhaseA==0)) //Shut down Phase A operation
{
sysreg_write(sysreg_IOPG, FIO_Page); //Go to FIO page
io_space_write(FIO_FLAGC, 0x0300); //Disable MQ2 (PF9) and MD2 (PF8)
io_space_write(PWM_SI, 0x0); //No invertion low side outputs,
//Invert none = 0x0
io_space_write(PWM0_CTRL, 0x000); //Disable PWM generation with
//PWMSYNC signal.
DutyCyleUpdateA=0; //This allow motor control to start pwm again
}
}
}
}
G7
//--------------------------------------------------------------------------------------
//Function: commutation_four_phase()
//Description: Selects the converter modes of operation magnetisation/freewheeling,
// demagnetisation depending on rotor data
//Inputs: -
//Returns: -
//--------------------------------------------------------------------------------------
void commutation_four_phase()
{
read_rotor_angle(); //Read current rotor angle
if((AngleMinA<RotorAngle)&&(RotorAngle<=AngleMaxA))
{
AngleDemagA=AngleMaxA-AngleDemag; //60
AngleOffA = AngleMaxA-10; //70
AngleOnA = AngleMinA+10; //20
EnablePhaseB=0;
EnablePhaseC=0;
EnablePhaseD=0;
if((AngleOnA<RotorAngle)&&(RotorAngle<=AngleDemagA))
{
MagFreeDemagA=1; //Implement Phase A Mag/Free mode
EnablePhaseA=1;
io_space_write(FIO_FLAGS, 0x0040); //Activate (PF6)
io_space_write(FIO_FLAGC, 0x0080); //Disable (PF7) =>1
}
else
{
if((AngleDemagA<RotorAngle)&&(RotorAngle<AngleOffA))
{
MagFreeDemagA=0; //Implement Phase A Demag mode
EnablePhaseA=1;
io_space_write(FIO_FLAGC, 0x0040); //Disable (PF6)
io_space_write(FIO_FLAGS, 0x0080); //Activate (PF7) =>2
}
else
{
MagFreeDemagA=0;
EnablePhaseA=0; //Shut down Phase A
io_space_write(FIO_FLAGS, 0x00C0); //Activate (PF6 & PF7) =>3
}
}
motor_control_a(); //Activate Phase A PWM current control
}
}
G8
//--------------------------------------------------------------------------------------
//Function: read_rotor_angle()
//Description: Determines the rotor position value depending on input pulse combination.
//Inputs: -
//Returns: -
//--------------------------------------------------------------------------------------
void read_rotor_angle()
{
// Go to FIO page
sysreg_write(sysreg_IOPG, FIO_Page);
//Get value of PF0 Flag
EnableSRM = io_space_read(FIO_DATA_IN);
if ((EnableSRM & 0x0003) ==3) //If Signal pulse is high => implement mag and free mode
{
RotorAngle = 40; //Constant angle in turn on angle range
}
else
{
if ((EnableSRM & 0x0003) ==1) //If Signal pulse is low => implement demag mode
{
RotorAngle = 75; //Constant angle in turn off angle range
j=0;
}
else
{ //Shut down phase operation
RotorAngle = 75; //Constant off angle
j=0;
}
}
}
G9

00 SRM D Design For Automotive - 2008 PDF

Uploaded by

Copyright:

Available Formats

00 SRM D Design For Automotive - 2008 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

00 SRM D Design For Automotive - 2008 PDF

Uploaded by

Copyright:

Available Formats

Design of a

Switched Reluctance Machine Drive

Anthony Murphy, B.Eng.

A thesis presented for the award of

School of Electronic Engineering

Design of a Switched Reluctance Machine

2 Operating principles and features of the SRM 8

6 System implementation 144

7 Conclusion and recommendations 177

Appendix A: Power analysis derivations A1

Appendix C: Thermal analysis computations C1

Appendix E: ADSP-21992 EZ-KIT lite evaluation system board E1

Appendix F: VisualDSP++ 3.0 and implementing the ISR F1

Appendix G: DSP implemented source code G1

Figure 1.1: The SRM Layout

Configuring of semiconductors in parallel was central to the implementation of the full-

1.2 Overview of thesis

Chapter 3 reviews a variety of converter topologies in order to determine the most

The preliminary analysis of semiconductor components to determine the most efficient

The experimental setup is described in Chapter 6. A description is given of the

2.1 Historical review

Early switched reluctance machines were known as “electromagnetic engines”, due to

The development of microprocessors and microcontrollers and related digital circuits

Air-handler A.O Smith

Table 2.1: Switched reluctance applications

The switched reluctance machine is an extremely simple electrical machine. The

To produce continuous positive motoring torque and to ensure self-starting capability

0o 60o 120o Rotor Angle

0o 60o 120o Rotor Angle

Figure 2.8: Torque/speed characteristic of the SRM

The main benefits of the switched reluctance machine include:

The main limitations of the switched reluctance machine include

2.3.1 Analysis of magnetically linear operation

The terminal voltage generated across one phase is:

Subscript ‘ph’ is left out for clarity, from here on.

The instantaneous electrical power, vi, is

The time derivative of magnetic stored energy at any instant is given by

Therefore the expression for electromagnetic torque (T) is:

Figure 2.9: Equivalent circuit for a single phase of the SRM

2.3.2 Analysis of magnetically nonlinear region

The energy stored in a magnetic field is

A rotor movement from its original position at A to a new position at B through an

The change in stored field energy is

∆Wf = OBC – OAD (2.14)

∆Wm = ∆We – ∆Wf (2.15)

∆Wm = T∆θ = ∆Wc (2.17)

This is identical to equation (2.9) from linear analysis.

2.3.3 Analysis of generator operation

Figure 2.12: Switched reluctance generator for single phase

Io = Iout – Iin (2.24)

2.4 Dynamic operation of SRM

θON θ1 θ2 θ3 θ4 Rotor θON θ1 θ2 θ3 θ4 Rotor

θON θ1 θ2 θ3 θ4 Rotor θON θ1 θ2 θ3 θ4 Rotor

Figure 2.16: Low speed motoring energy conversion loop

2.4.2 High speed motoring

Figure 2.18: High speed motoring energy conversion loop

2.4.3 Low speed generating

Figure 2.19: Voltage-PWM current regulation at low speed

During the generating period, the equation is:

PCF ( D1) = I m ( D1) V F DSW (4.7)