Control Strategies and Power Electronics in a Low-cost ...6350/Thesis_CDU_6350_Yan_X.… · Control Strategies and Power Electronics in a Low-cost Electric Vehicle Propulsion System

Control Strategies and Power Electronics in aLow-cost Electric Vehicle Propulsion System

Employing a Brushless DC Machine

By

Xinxiang Yan, M.Eng

School of EngineeringFaculty of Technology

Northern Territory University

A thesis submitted in fulfillment of the requirements forthe degree of Doctor of Philosophy

December 2000

i

Abstract

Battery-powered electric vehicles (BEVs) are a promising part of the solutions to

urban pollution and global warming. Existing problems of BEVs are short drive

range and high cost. Energy and power control strategies in EV propulsion remain a

substantial research topic which require further investigation and power electronics

plays a key role in EV development. This thesis studies the control strategies and

power electronics for a short range, low cost EV employing high efficiency

permanent magnet axial flux brushless dc motor (BDCM). Many of the technologies

presented can be used in more general applications.

Based on the drive cycle data obtained from an ICE vehicle and its analysis, a

novel energy and power management scheme has been proposed. The proposed

energy and power management scheme employs the newly developed non-flowing

zinc-bromine battery (NFZBB) and ultracapacitors (UCs). NFZBBs are optimized for

specific energy, low cost and long life. Ultracapacitors featuring high power density

and long life are used to complement the NFZBB to achieve good acceleration

performance at low speeds. The thesis shows how combining the two different types

of energy storage can optimize the EV propulsion system to achieve high energy for

drive range, high power for good acceleration performance, long lifetime, and low

cost.

A novel multifunctional dc-dc converter topology has been proposed for the

overall energy and power control. This single, simple dc-dc converter is used for both

the control of the energy which is drawn out of and placed into the ultracapacitors

and the speed extension of the BDCM. Several advanced control technologies of

BDCM are studied.

Zero Voltage Transition (ZVT) soft-switching techniques are utilized in the

multifunctional dc-dc converter to achieve high efficiency, reduction of volume and

weight, and lower electromagnetic interference (EMI) level. A prototype of the ZVT

dc-dc converter with 96% efficiency has been developed.

ABSTRACT ii

At high power levels, due to large dimensions of ferrite cores and air gaps, the

influence of leakage flux and fringing flux on the inductor design within the dc-dc

converter becomes significant, and is difficult to predict accurately. With

conventional design of high power ferrite cored inductors, difficulty exists in

achieving a satisfactory overall performance. A novel design for high power ferrite

cored inductors employing a dual-coil or distributed winding scheme is proposed by

the author to combat the problems. Results from both 3-Dimensional (3D) finite

element analysis (FEA) and measurement verify the viability of the novel design for

high power inductors. The proposed inductor design allows better optimization by

reducing the localization of saturation, leakage flux, and EMI levels, achieving good

utilization of the magnetic cores.

A new intelligent gate drive module for high power IGBTs has been developed

and presented. This module features high current capacity, powerful functions, high

performance, wide application, and convenience of use, and is suitable for high

power applications such as in EV drives. The detailed functions, operation,

performance, and practical application issues are addressed.

A soft-switched on-board EV battery charger with unity power factor has been

developed. It employs a zero-current-switching (ZCS) boost converter with power

factor correction (PFC) and an asymmetrically controlled zero-voltage-switching

(ZVS) half-bridge (HB) dc-dc converter in cascade. Several state-of-the-art power

electronics techniques are used. They include power factor correction, a ZVS

technique requiring few additional components, control of steady-state duty cycle of

the charger's dc-dc converter for high efficiency, a novel active pre-load with

minimized power rate, and an optimized charging profile.

iii

List of the Author's PublicationsDuring His Ph.D. Candidature

1. Xinxiang Yan and Dean Patterson, "Improvement of drive range, acceleration and

deceleration performance in an electric vehicle propulsion system”, The 1999

IEEE Power Electronics Specialist Conference (PESC'99), 27th June-1st July 1999,

Charleston, South Carolina, USA

2. Xinxiang Yan, Dean Patterson and Byron Kennedy, "A multifunctional dc-dc

converter for an EV drive system", 16th International Electric Vehicle Symposium

and Exhibition (EVS-16), 12-16 October 1999, Beijing, China

3. Xinxiang Yan and Dean Patterson, "Novel power management for high

performance and cost reduction in an electric vehicle", 1999 World Renewable

Energy Congress, 10-13 February 1999, Perth, Australia. Also published on

Journal of Renewable Energy (UK), Vol. 22 (1-3), 2001

4. Xinxiang Yan, Dean Patterson, Dan Sable and Guichao Hua, "Design of an

intelligent gate driver for high-power IGBT modules", The 1998 IEEE

International Conference on Power Electronics Drives and Energy Systems for

Industrial Growth (PEDES'98), 1-3 December 1998, Perth, Australia (IEEE

PEDES'98 prize paper)

A LIST OF THE AUTHOR’S PUBLISHCATIONS DURING HIS CANDIDATURE iv

5. Xinxiang Yan and Dean Patterson, "A high-efficiency on-board battery charger

with unity input power factor", International Journal of Renewable Energy

Engineering, April, 2000

6. Xinxiang Yan, Dean Patterson, Dan Sable, Xinfu Zhuang and Guichao Hua,

"Design of power electronics for a satellite flywheel energy storage system", IEEE

PEDES'98, 1-3 December 1998, Perth, Australia

7. Dean Patterson, Xinxiang Yan and S. Camilleri, "A very high efficiency controller

for an axial flux permanent magnet wheel drive in a solar powered vehicle", IEEE

PEDES'98, 1-3 December 1998, Perth, Australia

8. Byron Kennedy, Dean Patterson, Xinxiang Yan, and John Swenson,

"Development of a light short range electric commuter vehicle", International

Journal of Renewable Energy Engineering, April, 2000

9. Xinxiang Yan and Dean Patterson, "Design of an intelligent gate drive module for

high-power IGBTs", 1998 Australian University Power Engineering Conference

(AUPEC'98), 27-30 September 1998, Hobart, Australia

10. Paul Chandler, Xinxiang Yan, and Dean Patterson, “Novel design of high-power

ferrite cored inductors with ease of design and improved overall performance”,

2001 IEEE Power Electronics Specialist Conference (PESC'01)

v

Declaration

I hereby declare that the work herein, now submitted as a thesis for the degree of

Doctor of Philosophy of the Northern Territory University, is the result of my own

investigations, and all references to ideas and work of other researchers have been

specifically acknowledged. I hereby certify that the work embodied in this thesis has

not already been accepted in substance for any degree, and is not being currently

submitted in candidature for any other degree.

Xinxiang Yan

December 16, 2000

vi

To my family

vii

Acknowledgements

I would like to thank several organizations and people that made this work possible.

The work was undertaken at the Northern Territory Centre for Energy Research

(NTCER) and the School of Engineering of Northern Territory University. Funding

for the project was provided by the Northern Territory Power and Water Authority,

and the Australian Research Council. The author was financially supported during

the project by the Northern Territory University Postgraduate Research Scholarship

and the International Postgraduate Research Scholarship from the Department of

Employment, Education, Training and Youth Affairs of Australia.

A special gratitude is for my academic supervisor, Professor Dean Patterson, for

his guidance, encouragement, challenges and methodical engineering view. Working

with him was a great pleasure.

My fellow Ph.D. students Paul Chandler and Andrew Tuckey who were also the

computer system administrators at NTCER have been very helpful during different

phases of this thesis. Paul also did 3D FEA for the verification of my proposed novel

high power ferrite cored inductor design. Thanks also to Research Associate Byron

Kennedy who gave me much technical support during the project development.

I would like to thank the two experienced technical officers Wayne Flavel and

Howard Pullen for their technical support. Wayne contributed many hours to the

fabrication of the power block of the converter prototype.

I want to recognize the work of senior mechanical research fellow John Swenson

in helping to develop, install and test the vehicle data logger system. Thanks are also

extended to the meter-reading technician Sergio Sepulveda at the Northern Territory

Power and Water Authority for helping to obtain the vehicle drive data.

And finally, the deepest appreciation from my heart is for my parents for

bringing me up and for my wife and my daughter for their love, companionship, care,

and patience during the long academic journey.

viii

Contents

List of Figures…………………………………………………………….…...xiv

List of Tables…………………………………….…………….…..………..…xxi

Table of Abbreviations………………………...….………………………..xxii

Chapter 1 Introduction …………………………………………………….…1

1.1 Electric Vehicle Background…………………………………………….......1

1.2 Benefits of Electric Vehicles…………………………...…………………....2

1.2.1 Reduction of Air Pollution and Global Warming……………….…...2

1.2.2 Reduction of Dependence on Depleting Fossil Energy Sources…….3

1.2.3 Other Benefits…………………………………………………..…....4

1.3 BEVs, HEVs and FCEVs……………………………………………..……...5

1.3.1 Battery-Powered Electric Vehicles (BEVs)……………………….…5

1.3.2 Hybrid Electric Vehicles (HEVs)…………………………….……...6

1.3.3 Fuel-Cell Electric Vehicles (FCEVs)…………………………...…...7

1.3.4 A Few Important Notes……………………………………..….…....9

1.4 EV Research Project at the Northern Territory Centre

for Energy Research………………………………………………..……......9

1.5 Contributions and Outline of the Thesis……………………………………10

Chapter 2 Data Logging from an ICE Vehicle and the Data

Analysis………………………………………………………….14

2.1 Design of a Data Logger for Collection of Vehicle Drive Cycle Data…….14

CONTENTS ix

2.1.1 Data Acquisition…………………………………………………….14

2.1.2 The Physical Parameters Measured………………………………....16

2.2 Typical Drive Cycles of a Meter-Reading Vehicle…………………………17

2.3 Analysis of the Measured Drive Cycles…………………………………….17

2.3.1 Comparison between the Obtained Drive Cycle and the FUDS…....17

2.3.2 Acceleration Rates……………………………………………….....24

2.3.2.1 0-60 km/h Acceleration Data…………………………….....25

2.3.2.2 0-80 km/h Acceleration Data……………………………….26

2.3.3 Frequencies of Acceleration to Different Speeds……………….….26

2.3.4 The Average Speed and the Maximum Speed……………………...28

2.4 Specifications of the NTCER EV Propulsion System……………………...28

2.5 Review……………………………………………………………………...29

Chapter 3 Several Advanced Control Technologies in Permanent

Magnet Brushless DC Motor Drives……………………....30

3.1 Permanent Magnet Brushless DC Motors (BDCM) for EV Propulsion……30

3.2 An Overview of a BDCM Drive System and its Control………………...…33

3.3 Impacts of Different PWM Strategies and Commutation Schemes on

Performance of BDCMs………………………………………………….…37

3.3.1 PWM with Bipolar Voltage Switching vs. PWM with Unipolar

Voltage Switching ……………………………………………..…...37

3.3.2 Performance Improvement of BDCM in Regeneration Mode…...…41

3.3.2.1 Method I – Combining Unipolar Voltage Switching

PWM and Bipolar Voltage Switching PWM…………….....43

3.3.2.2 Method II – Twelve-Step Commutation…………………....45

3.4 Speed Extension and Constant Power Operation of the BDCM……….…...47

3.4.1 Flux Weakening Method…………………………………………....48

3.4.2 Mechanically Varying the Air Gap………………………………....49

3.4.3 Advancing the Phase Angle of the Excitation Transition Points.…..50

3.4.4 Voltage Boosting Method Employing a Bidirectional Direct

DC-DC Converter……………………………………………….......50

3.5 Current Control of BDCM at Speeds above the Base Speed in

CONTENTS x

Motoring Mode when using the Voltage Boosting Method………….…..…51

3.5.1 Method I: The Converter – Current controlled;

The Inverter – Commutation Only……………………………….....51

3.5.2 Method II: The Converter – Voltage Source Output;

The Inverter – Current Mode Control………………………………56

3.6 Control of Regenerative Braking of BDCM above the Base Speed………..60

3.6.1 When Motor Speeds are much Higher than the Base Speed…….....61

3.6.2 When Motor Speeds are only Marginally Higher than the

Base Speed………………………………………………………….62

3.7 Review……………………………………………………………………...64

Chapter 4 Novel Energy and Power Management in the EV

Propulsion System…………………………………………..…..66

4.1 Power Requirements for the Electric Vehicle Propulsion System………....66

4.1.1 Road Load Characteristics……………………………………….....67

4.1.2 Components of Vehicle Power vs. Speed for the FUDS……….…..69

4.1.3 Observations……………………………………………………......73

4.2 Overview of Commercial EV Batteries…………………………………….74

4.3 Non-Flowing Zinc-Bromine Batteries………………………………….......76

4.4 The Load-Leveling Concept and High-Power Energy Storage

Devices in EVs……………………………………………..…………..…...78

4.4.1 The Load-leveling Concept in EVs…………………………….…...78

4.4.2 High-Power Batteries………………………………………….…....79

4.4.3 Flywheels……………………………………………………….…..79

4.4.4 Ultracapacitors……………………………………………………...80

4.4.5 Summary…………………………………………………………....81

4.5 A Novel EV Power Management Scheme Employing Zinc-Bromine

Batteries and Ultracapacitors…………………………………………….....81

4.6 Simulation and Optimization of the EV Power System…………………....83

4.6.1 The Simulation Model Using Matlab / imulink…………………….83

4.6.2 The Simulation Results and Determination of the Power Rating

of the Ultracapacitor Bank and the DC-DC Converter…….…….…84

CONTENTS xi

4.6.3 Control Rules for the DC-DC Converter for Load Leveling

the Batteries…………………………………..…………...………...89

4.6.4 Simulation Results on the Voltage and Current Transients

of the Short Term Power Unit………………………….………...…90

4.6.4.1 Voltage and Current Transients of the STPU

in Boost Mode…………………………………………...….91

4.6.4.2 Voltage and Current Transients of the STPU

in Buck Mode…………………….……………………...….94

4.7 Overall Power Control in the NTCER EV…………………………….…....96

4.8 A Novel Multifunctional DC-DC Converter Topology for the

Overall Power Management in the NTCER EV………………….......97

4.8.1 Topology Derivation……………………………………..……...….97

4.8.2 Operation Principles and Topology nalysis……………………......100

4.8.2.1 At Motor Speeds below the Base Speed…………….……..100

4.8.2.2 At Motor Speeds above the Base Speed………………..….102

4.8.2.3 Zero Current Switching (ZCS) Conditions of the

Contactor…………………………………………...……...104

4.9 Review…………………………………………………………….….....…104

Chapter 5 Development of the Soft-Switched Bidirectional

DC-DC Converter………………………………..………....…....106

5.1 Introduction…………………………………………………………….......106

5.2 Evaluation of Soft Switching echniques……………………...…..…….….107

5.3 ZVT Operations of the Bidirectional DC-DC Converter………………..…110

5.3.1 ZVT Operation in Boost Mode…………………………………….110

5.3.2 ZVT Operation in Buck Mode……………………………………..119

5.4 Implementation Issues and Experimental Results…………………………121

5.4.1 Implementation Issues ……………………………..…….……......121

5.4.2 Experimental Results……………………………………...…….....124

5.5 Novel Design of High-Power Ferrite Cored Inductors with Ease

of Design and Improved Overall Performance ……………………………132

5.5.1 Problems with Conventional Design of High-Power

CONTENTS xii

Ferrite Cored Inductors……………………………….……..….....133

5.5.1.1 Conventional Core and Winding Configurations of

High Power Ferrite Cored Inductors………………….…...133

5.5.1.2 Observations of Inductor Configuration #1 and

Inductor Configuration #2 Aided by 3D EFA……………..135

5.5.2 Novel Design for High Power Ferrite Cored Inductors………...…139

5.5.2.1 A Novel Winding Scheme for High Power

Ferrite Cored Inductors……………...………………….....139

5.5.2.2 3D FEA Results for Inductor Configuration #3…………...140

5.5.2.3 Measurement Results, Comparison and Conclusions….….141

5.6 Review……………………………………………………………………..143

Chapter 6 Development of an Intelligent Drive Module

for High Power IGBTs……………………………………....…145

6.1 Functions and Features of the New Intelligent IGBT Drive Module….…..147

6.2 Functional Diagram and its Operation…………………………………….148

6.2.1 Intelligent Functions……………………………………………….148

6.2.2 Bipolar Drive Outputs……………………………………………..149

6.2.3 The Built-in, Well Regulated and Isolated Floating Bipolar

Power Supply……………………………………………………...150

6.3 Measurements……………………………………………………………..152

6.4 Practical Application Issues…………………………………………….....155

6.4.1 Fault Status Feedback……………………………………………...155

6.4.2 Selection of Gate Drive Resistor……………………………….….155

6.5 Review……………………………………………………………….….....156

Chapter 7 Soft-Switched On-Board EV Battery Chargers

with Unity Power Factor……………………………….....…157

7.1 Power Stage Configurations for EV On-Board Battery

Chargers………..…………………………………………………………..159

7.2 An Example — A 780 W High-efficiency, Low-cost,

Unity-Power-factor Battery Charger………………………………………161

CONTENTS xiii

7.2.1 Design Specifications…………………………………….………..161

7.2.2 Power Stage and its Soft Switching Operation……………………161

7.3 Optimization of the Charge Profile………………………………………..164

7.4 A Novel Active Pre-Load with Minimized Power Rate for the

On-Board Battery Charger……………………………………….…….….166

7.5 Constant Steady-State Duty Cycle Control of the DC-DC Converter…….168

7.6 Measurements……………………………………………………….….…170

7.7 Review………………………………………………………………….....172

Chapter 8 Conclusions and Recommendations for Further

Work ………………………………………………………..…...173

8.1 Conclusions………………………………………………………………..173

8.2 Recommendations for Further Work……………………………………...177

Appendix A The Non-Flowing Zinc/Bromine Battery for

Electric Vehicle Applications …………………….……...179

Appendix B Schematic Diagrams ……………………………………….190

Bibliography ……………………………………………………………..…...200

xiv

List of Figures

1.1 Block diagram of a BEV propulsion system…………………………....…...5

1.2 Block diagram of a series HEV propulsion system………………………....6

1.3 Block diagram of a parallel HEV propulsion system…………………….....6

1.4 Block diagram of a FCEV propulsion system…………………………….…8

2.1 Functional diagram of hardware setup of the data logger system……...…..15

2.2 A photograph of the data logging equipment………………………………16

2.3 Drive cycle of the domestic route, Jan-20 1999……………………………18

2.4 Drive cycle of the domestic route, Jan-21 1999…………………………....19

2.5 Drive cycle of the domestic route, Feb-02 1999…………………..…...…..20

2.6 Typical daily drive cycle of the commercial route, Jan-25 1999………......21

2.7 The Federal Urban Driving Schedule (FUDS)……………………….….....23

2.8 Average speed vs. time to 60 km/h - all routes………………………..…...25

2.9 Average acceleration vs. achieved speed up to 60 km/h - all routes…….....26

2.10 Average acceleration vs. achieved speed up to 80 km/h - all routes…….…27

2.11 Average acceleration vs. time to 80 km/h - all routes…………………...…27

2.12 Histogram of terminal velocity after acceleration from standstill - all

routes………………………………………………………………….….…28

3.1 A photograph of the NTCER motor disassembled………………………....33

3.2 Ideal back EMF waveforms and current waveforms of a BDCM……...…..34

3.3 A BDCM drive system with closed-loop torque control…………………...35

3.4 A BDCM drive system with double-closed-loop control………………......35

LIST OF FIGURES xv

3.5 Using a full-bridge dc motor drive system to model the BDCM drive

system in 60��

3.6 PWM with bipolar voltage switching………………………………………38

3.7 PWM with unipolar voltage switching ………………………………….....39

3.8 "Lubricous torque" offered by bipolar voltage switching PWM scheme…..40

3.9 Unipolar voltage switching PWM with six-step commutation…………….41

3.10 Simulated phase current waveform in regeneration mode using unipolar

voltage switching PWM with six-step commutation scheme………………42

3.11 Illustration of the cause of the distorted motor current in

regeneration mode……………………………………………………….....42

3.12 Bipolar voltage switching PWM with six-step commutation……………...43

3.13 Illustration of phase current transition in regeneration mode using

bipolar voltage switching PWM………………………………………...….44

3.14 Simulated phase current waveform in regeneration mode using unipolar

voltage switching PWM with six-step commutation scheme……………....44

3.15 Twelve-step commutation scheme………………………………………....45

3.16 Illustration of phase current transition in regeneration mode using

twelve-step commutation scheme…………………………………………..46

3.17 Simulated phase current waveform in regeneration mode with

twelve-step commutation scheme…………………………………………..46

3.18 Natural torque-speed characteristic of a BDCM and desired torque-speed

characteristic for traction application………………………………………48

3.19 Speed extension of BDCM by employing a dc-dc converter……………….51

3.20 Motor current control Method I…………………………………………….52

3.21 Motor current control loop with unspecified control plant (Method I)….....53

3.22 Root locus of Kci(s)………………………………………………………....55

3.23 Influences of system parameters on the location of the poles

and zeros of Kci(s)…………………………………………………………..56

3.24 Equivalent system diagram using Method II of the motor current control

at the motor speeds above the base speed………………………………….57

3.25 Outer control loop with unspecified transfer function of plant

LIST OF FIGURES xvi

using Method II……………………………………………………………..57

3.26 Root locus of Kcv(s)………………………………………………………...58

3.27 Mathematical model of the closed voltage loop using Method II………….59

3.28 Step response of the closed voltage loop…………………………………...59

3.29 Buck-regeneration operation when the motor speeds are much higher

than the base speed………………………………………………………….61

3.30 Hybrid boost-regeneration mode when the motor speed are

marginally higher than the base speed……………………………….……..63

3.31 Simplification of Figure 3.30………………………………………..……..63

3.32 Two states of the circuits when San is executing PWM switching…….…...63

3.33 The equivalent boost circuit…………………………………………….….64

4.1 Components of vehicle power vs. speeds………………………………….69

4.2 Wheel power on FUDS……………………………………………………..70

4.3 Components of the vehicle power on FUDS…………………………...…..71

4.4 Required battery power and energy consumed on FUDS…………………..72

4.5 Power for climbing vs. speeds………………………………………….......72

4.6 Comparison among lead-acid, MiCd and MiMH batteries………….……...75

4.7 Connection configuration of the ultracapacitor bank

in the propulsion system for load leveling purpose………………………...82

4.8 Matlab / Simulink model of the dc-dc converter in boost model…………..84

4.9 Output power capacity and efficiency of the STPU

employing 20 ultracapacitors…………………………………………….....86

4.10 Output power capacity and efficiency of the STPU employing 30

ultracapacitors……………………………………………………….…...…87

4.11 Output power capacity and efficiency of the STPU employing 40

ultracapacitors………………………………………………………….…...88

4.12 Control diagram of the bi-directional dc-dc converter for the purpose

of load leveling the batteries ……………………………………………….90

4.13 Simulink block diagram and simulated step response of the

current loop of the STPU in boost mode……………………………...…....92

4.14 Battery current, bus voltage and currents of the STPU

LIST OF FIGURES xvii

in boost mode by simulation………………………………………...……...93

4.15 Simulink block diagram and simulated step response of the

current loop of the STPU in buck mode………………..…………………..94

4.16 Battery current, bus voltage and currents of the STPU

in buck mode by simulation………………………………………………...95

4.17 The concept of the overall power management…………………………….97

4.18 Two-stage scheme for the overall power management using two

separate dc-dc converters in cascade………………………………...…..….98

4.19 The proposed multifunctional dc-dc converter topology

for the overall power management …………………………………………99

4.20 Operation mode at motor speeds below the base speed………...…………101

4.21 Regeneration mode at motor speeds below the base speed……...……..….102

4.22 Battery charging the ultracapacitors……………………………………….102

4.23 Constant power operation (in motoring mode) when motor

speeds are above the base speed…………………………….……...……...103

4.24 Buck-regeneration operation when motor speeds are much higher

than the base speed………………………………………………………...103

4.25 Hybrid boost-regeneration mode when motor speeds are

marginally higher than the base speed…………………………….…….…103

5.1 Diagram of the ZVT bidirectional dc-dc converter………………………..110

5.2 Equivalent circuit in boost mode………………………………………..…111

5.3 Key circuit waveforms in boost mode…………………………………..…112

5.4 Circuit operation diagrams in boost Mode………………………………...113

5.5 Simulated typical circuit waveforms in boost mode……………………....118

5.6 Hua’s ZVT dc-dc boost converter……………………………………....…119

5.7 Equivalent circuit in buck mode………………………………………......119

5.8 Simulated typical circuit waveforms in buck mode……………………....120

5.9 A photo of the prototype of the ZVT bidirectional dc-dc converter………121

5.10 Circuit diagram of the practical ZVT bidirectional dc-dc converter

including two saturable inductors and two RD snubbers………………….122

5.11 Control diagram of the bidirectional dc-dc converter……………………..123

LIST OF FIGURES xviii

5.12 Gate drive waveforms Vgs1 and Vgsx in boost mode……………………….125

5.13 Waveforms of Vgs1 and Vds1 showing soft switching operation

of the main boost switch…………………………………………………...125

5.14 Waveforms of Vgsx and Vds showing soft switching operation

of the auxiliary boost switch……………………………………………….126

5.15 Waveforms of Vgs1 (lower trace) and VD1 (upper trace) in boost mode…...126

5.16 Waveforms of Vgsx (lower trace) and ilrx (upper trace) in boost mode…….127

5.17 Waveforms of Vgs1 (lower trace) and Vc (upper trace) in boost mode.…....127

5.18 Measured efficiency vs. output power in boost mode

at Vin = 100 V and Vo = 170 V……………………………………….…....128

5.19 Logical gate signal waveforms Vgs2 and Vgsy in buck mode……………....129

5.20 Logical level gate signal waveform and the real gate drive

signal waveform of the main buck switch, showing the

time delay between them…………………………………………………..129

5.21 Waveforms of Vgs2 (logical level) and Vds2 showing soft switching

operation of the main buck switch……………………………………...…130

5.22 Waveforms of Vgsy (logical level) and Vdsy …………………………...….130

5.22 Waveforms of Vds2 and VD1 in buck mode………………………………..131

5.23 Waveforms of Vgsy and ilry in buck mode………………………………....131

5.24 Measured efficiency vs. output power in buck mode at Vin =100 V

and RL = 1.5 ohm……………………………………………………….…132

5.25 Inductor Configuration #1……………………………………………...…134

5.26 Inductor Configuration #2………………………………………………...134

5.27 View highlighting the ¼ section used in the 3D FEA model of

Inductor Configuration #1………………………………………………...136

5.28 3D FEA of a ¼ model of Inductor Configuration #1 showing excessive

flux leakage and some flux localization around the central leg……..…….137

5.29 3D FEA results of Inductor Configuration #2……………………………..138

5.30 The proposed dual-coil winding configuration, Inductor

Configuration #3…………………………………………………………...140

5.31 3D FEA results of Inductor Configuration #3………………………….….141

LIST OF FIGURES xix

5.32 A photo of the inductor as built using the proposed dual-coil

winding scheme………………………………………………………..….142

6.1 The functional diagram of the new intelligent IGBT drive module….…..148

6.2 The closed-loop voltage regulation scheme………………………….…...151

6.3 Two different flyback transformer configurations………………………..152

6.4 Turn-on gate signal……………………………………………………......153

6.5 Turn-off gate signal…………………………………………………...…..153

6.6 Typical turn-on waveforms for 300A, 600V IGBTs…………………....…153

6.7 Typical turn-off waveforms for 300A, 600V IGBTs…………………..….154

6.8 The termination of the pulse following the determination of a fault

condition…………………………………………………………………..154

6.9 A photo of the intelligent high power IGBT drive module

for general applications………………………………….………………...154

6.10 Optimized gate connection configuration…………………………..……..156

7.1 Block diagram of an EV on-board battery charger……………….....…….159

7.2 Diagram and key waveforms of the ZCS boost PFC converter…….……..162

7.3 Diagram and key waveforms of the ZVS HB converter…………….….....163

7.4 Optimized charge profile…………………………………………….…….166

7.5 Circuit of the active pre-load…………………………………………..….167

7.6 Diagram of principle of constant steady-state duty cycle control…………169

7.7 Overall efficiency measurement…………………………………….……..170

7.8 Oscilloscope photograph of typical input line voltage and input

line current waveforms………………………………………………….…171

7.9 Loop gain measurement in constant current charging mode……………...171

7.10 Loop gain measurement in constant voltage charging mode……………...172

B1 Power block schematic…………………………………………………….191

B2 Gate driver schematic……………………………………..……………….192

B3 Control Schematic #1……………………………………………….…..…193

B4 Control Schematic #2……………………………………………………...194

B5 Control Schematic #3………………………………………………..…….195

B6 Control Schematic #4……………………………………………….……..196

LIST OF FIGURES xx

B7 Control Schematic #5………………………………………………….…..187

B8 Control Schematic #6……………………………………………….…..…198

B9 Control Schematic #7……………………………………………………...199

xxi

List of Tables

2.1 Analysis result of the measured typical daily drive cycles……………..…..22

2.2 Analysis results of the FUDS…………………………………………...…..24

2.3 Specifications of the NTCER EV propulsion system…………………....…29

3.1 A survey of the world's major electric vehicles (1996- )……………….......31

3.2 Main electrical specifications of the NTCER motor…………………….....32

3.3 Trigger signals for four-quadrant operations…………………………….....37

3.4 Comparison between bipolar voltage switching PWM and unipolar

voltage switching PWM………………………………………………..…...40

4.1 NTCER EV data………………………………………………….………...67

4.2 NFZBB projected performance………………………………………..……77

4.3 Specifications of Maxwell ultracapacitor PC7223………………….……...80

5.1 Comparison results of the three different inductors studied………………142

6.1 Definition of input pins of the drive module………………………………148

xxii

Table of Abbreviations

Abbreviation Meaning

3D 3-Dimensional

ACRDCLC Actively Clamped Resonant DC Link Converter

BDCM Brushless DC Motor (Machine)

BEV Battery-Powered EV

BJT Bipolar Junction Transistor

CARB California Air Resources Board

CC Constant Current

CP Constant Power

CV Constant Voltage

CWT Coaxial Winding Transformer

DCM Discontinuous Current Mode

EMC Electromagnetic Compatibility

EMF Electromagnetic Force

EMI Electromagnetic Interference

EOCC End of Charge Current

EPA Environment Protection Agency (US)

ESR Equivalent Series Resistance

EV Electric Vehicle

EVS Electric Vehicle Symposium & Exhibition (International)

FB Full-Bridge

LIST OF ABBREVIATIONS xxiii

FCEV Fuel Cell Electric Vehicle

FEA Finite Element Analysis

FOC Field-Oriented Control

FUDS Federal Urban Driving Schedule (US)

GFCI Ground Fault Current Interrupter

HB Half Bridge

HEV Hybrid Electric Vehicle

IC Integrated Circuit

ICE Internal Combustion Engine

IEC International Electrotechnical Commission

IGBT Insulated Gate Bipolar Transistor

IM Induction Motor (Machine)

IPM Intelligent Power Module

LHP Left Half Plane

MCT MOS Controlled Thyristor

MOSFET Metal Oxide Semiconductor Field Effect Transistor

NFZBB Non-Flowing Zinc-Bromine Batteries

NiCd Nickel Cadmium

NiMH Nickel-Metal Hydride

NTCER Northern Territory Center for Energy Research (Australia)

NT PAWA Northern Territory Power and Water Authority

PCB Printed Circuit Board

PI Proportional-Integral

PID Proportional-Integral-Differential

PM Permanent Magnet

PMSM Permanent Magnet Synchronous Motor (Machine)

PFC Power Factor Correction

PWM Pulse Width Modulation

RC Resistor-Capacitor

RD Resistor-Diode

RHP Right Half Plane

LIST OF ABBREVIATIONS xxiv

RMS Root Mean Square

SCR Silicon Controlled Rectifier

SOC State of Charge

SRM Switched Reluctance Motor (Machine)

STPU Short Term Power Unit

UC Ultracapacitor

ULEV Ultra Low Emission Vehicle

UVLO Under Voltage Lockout Protection

ZCS Zero Current Switching

ZCT Zero Current Transition

ZEV Zero Emission Vehicle

ZVS Zero Voltage Switching

ZVT Zero Voltage Transition

ZVZCS Zero-Voltage and Zero-Current Switching

1

Chapter 1

Introduction

1.1 Electric Vehicle Background

The concept of the electric vehicle (EV) was conceived in the middle 1800’s while

gasoline-powered internal combustion engine (ICE) vehicles were introduced toward

the end of the same century [1]. A competition between EVs and ICE vehicles began

following the introduction of the latter. The competition has turned out to be a

dramatic "Marathon" that is still ongoing, and more importantly, a new exciting

"turning point" is expected to come. EVs, despite being self-starting, energy-efficient,

and nonpolluting, were and are plagued by poor-range and short-lived batteries. Early

gasoline-power vehicles were capable of covering long distance but were started by

hand cranks. As a comfortable and easy-to-operate method of transportation, battery-

powered EVs were preferred by the wealthy elite and became a dominant means of

auto transportation in the early 1900's. Then EVs remained in existence side by side

with the ICE vehicles for several years. However, since an electric starter motor for

ICE vehicles was invented by Charles F. Kettering in 1911 [1], battery-powered EVs

lost the battle to ICE vehicles and had to stand by and watch the world go by while

patiently awaiting new technology. Since 1911, ICE vehicles have evolved, been

improved in design, and have become the dominant automobile. An automobile is

now an indispensable part of modern life and has contributed to the prosperity and

the economic growth of mankind.

CHAPTER 1 INTRODUCTION 2

Despite the fact that an ICE vehicle has been a very popular means of

transportation for many decades, owing to its advantages of range, refueling time,

power and the low price of the fuel, EV interest has never disappeared totally.

Whenever there has been any crisis regarding the operation of the ICE vehicle, we

have seen a renewed interest in the EV. The energy crisis in the 1970’s brought EVs

back to the streets. The most recent environmental awareness and energy concerns

have imposed, for the first time since its introduction, a serious threat to the use of

the ICE vehicles. One significant event is the 1990 California Mandate. In October

1990, the California Air Resources Board (CARB) established rules (with some

modified thereafter) that mandated 10% of all vehicles sold in California by 2004

must be zero emission vehicles (ZEVs). The California Mandate has reverberated

worldwide. Since now not only academia, but also governments and industries are

pushing EVs onto the roads. The availability of several commercial EVs is important

evidence.

1.2 Benefits of Electric Vehicles

There is a range of benefits arising from the use of EVs, which are briefly addressed

below.

1.2.1 Reduction of Urban Air Pollution and Global Warming

The most immediately apparent benefit of the use of EVs is their contribution to

reduction in pollution, both atmospheric and acoustic, especially at their points of

use. ICE vehicles have been responsible for most of the air pollution in the world’s

major cities as emissions from millions of vehicles on the road add up. ICE vehicles

also contribute to the world’s global warming. According to figures released by the

US Environment Protection Agency (EPA), conventional ICE vehicles currently

contribute 40-50% of ozone, 80-90% of carbon monoxide, and 50-60% of air toxins

found in US urban areas [2]. Two thirds of all oil consumed in US is used in the

transportation section, which results in 40% of the USA's contribution to global

warming [3].


Australia is producing the second highest greenhouse gas per capita in the world

[3]. In Australia, 21% CO2 comes from ICE vehicles. It should be noted that in

Australia, the concentration of vehicles is particularly non-uniform. Although

Australia has a very large land area, over 80% of Australians live in only a few big

cities, such as Sydney and Melbourne, and thus vehicle density in those places is very

high. This situation makes urban air pollution from ICE vehicles in Australia an

especially serious issue. Where vehicle density is high, the problems associated with

motor vehicles are severe.

In developing countries, such as China, India and Indonesia, the numbers of

automobiles have been increasing rapidly due to very large population growth and

generally fast growth of their economies. Air pollution in those areas, largely

contributed to by tremendous numbers of ICE vehicles, is already serious and is

getting worse [4].

EVs produce no pollution at the points of use and therefore are one of the

promising solutions to urban air pollution. Although generation of the required

additional electricity will increase power station emissions, it is much easier and

more efficient to control unwanted products emitted from a small number of power

stations than from millions of individual exhausts. This is especially so considering

the fact that for ICE vehicles, emission levels increase with the vehicle age. The

increased use of electricity generated from natural gas, nuclear, hydraulic power and

other renewable energy should largely reduce the demand of the electricity derived

from "polluting" coal and significantly reduce the air pollution.

1.2.2 Reduction of Dependence on Depleting Fossil Energy Sources

Another major benefit arising from the replacement of ICE vehicles by EVs is

reduction of our dependence on the depleting fossil energy sources. Road transport

depends almost completely on naturally occurring petroleum. Oil provides 40% of

the world's energy of which transport consumes 60% [5]. Throughout the world,

there is growing concern about the depletion and rising cost of the resource.

According to [4], oil is expected to last hardly for more than 100 years. The future

availability of affordable fossil fuel is definitely in question. Clearly, there is a


limited amount left to find. We should be saving our liquid fossil fuel for airplanes.

Obviously, diverse primary energy sources should be used and those that can replace

oil should be in their early development stages now.

It is very important to note that renewable energy is a commodity, just like any

other form of energy. It has a major role to play in meeting the needs of global

demand and combating the danger of global warming. Although presently renewable

energy represents only 5% of all primary energy needs, it has been predicted that this

will reach 70% by the year of 2060 [3].

1.2.3 Other Benefits

As well as the benefits of lower environmental degradation and diversified fuel

sourcing, there are other benefits of the use of EVs as listed below.

� There are advantages to the electricity supply industry arising through the load

leveling capabilities of EVs. Utilities are generally under pressure to improve

load factor to lift the utilization of efficient base load plant. EVs provide a battery

load that, to a large extent, could be charged under the control the utilities by

offering attractive off-peak rates, for example, an overnight charging rate. In

many areas the capacity for power that can be used to charge EVs in off-peak

hours already exists. No additional pollution will result from the use of that

power. The increase in total energy supplied by the utilities will provide

opportunities for developing alternative energy sources as required.

� It is a general scientific knowledge that EVs work more efficiently than ICE

vehicles do. For a conventional ICE, total energy efficiency from fuel to vehicle

traction is less than a mere 15%. Some tightly controlled best heat engines can

achieve total energy efficiency of 35-40% but only at peak power and optimum

speed [6-8].

� Reducing dependence on depleting liquid fossil energy sources through the use of

EVs should benefit economies of most of the oil-importing countries overall. It is

a well-known fact that 8 countries have 81% of all of the world's crude oil

reserves [3]. It is reasonable to assume that oil prices will increase considerably

as the limited oil sources are depleting, not to mention that an oil crisis can be


triggered and worsen by other factors, such as politics and regional conflicts.

Widespread use of EVs would allow nations to shift away from their over

reliance on foreign oil and tie the nations’ energy future to domestic, clean, and

renewable resources, benefiting the nations’ overall economy.

1.3 BEVs, HEVs and FCEVs

An electric propulsion system is described as the heart of an electric vehicle. Its

function is to interface the power source with the vehicle wheels, transferring energy

in either direction under the control of the driver. In terms of the types of EV

propulsion systems, there are three types of EVs. They are battery-powered EVs

(BEV), hybrid EVs (HEVs) and fuel cell EVs (FCEVs).

1.3.1 Battery-Powered Electric Vehicles (BEVs)

Battery-powered EVs are referred to as zero emission vehicles (ZEV) as they don't

produce pollution at the points of use, sometimes known as Pure EVs. A block

diagram of a basic BEV propulsion system is shown in Figure 1.1. It consists of

batteries, motors and motor controllers, transmission devices and wheels. It may be

desirable to directly drive wheels without a transmission.

EV propulsion systems with brake energy regeneration and local zero emission

demonstrate their best performance in typical stop-and-go city traffic with low-speed

cruising. As can be seen in the following sections, an ideal BEV propulsion system is

simpler than those of an HEV and an FCEV. Currently the main barrier to success for

BEVs is their short drive range per charge, resulting from very large, expensive and

short-lived battery packs.

B atte ry M o to r T ran sm issio n W h ee ls

Figure 1.1 Block diagram of a BEV propulsion system


1.3.2 Hybrid Electric Vehicles (HEVs)

An HEV has two or more energy conversion paths, one of which is bidirectional.

Usually it consists of a heat engine and electric converter. The heat engine may be a

piston engine or a turbine, and the electric converter consists of an electric motor and

inverter, which can be run bidirectionally either as a motor or as a generator.

Generally, an HEV has an electric energy storage device -- a battery, flywheel or

ultracapacitor-- that captures some of the energy converted from kinetic energy in

braking and stores it for use the next time the vehicle is accelerated.

In terms of the configuration of these energy sources, a series HEV and a parallel

HEV have been widely discussed [9-12]. A series hybrid is an HEV in which only

one energy converter can provide propulsion power. An ideal series HEV is shown in

Figure 1.2. The battery-powered electric motor drives the propulsion wheels, while a

downsized ICE drives a generator, the output of which is connected to the battery.

The ICE can be downsized because the battery buffers the highly variable power

demand, and the ICE needs only deliver the average power. The battery output is

controlled to determine the voltage waveform applied to the motor, and hence how

fast the motor spins, which in turn determines the vehicle speed. All propulsion

power comes from the electric motor.

B atte ry M o to r T ransm issio n W hee lsG enerato rE ng ine

Figure 1.2 Block diagram of a series HEV propulsion system

B attery M o to r

T ran sm issio n W h eelsD riveS h aft

E n g in e C lu tch

C lu tch

Figure 1.3 Block diagram of a parallel HEV propulsion system


A parallel HEV, shown in Figure 1.3, is one in which more than one energy

converter can provide propulsion power directly to the wheels. Usually a parallel

HEV runs on the electric drive in slow city traffic, on the engine when cruising at

high speed on highways, and on both when vehicle acceleration is required. Series

HEVs offer significant advantages over parallel HEVs, such as mechanical

simplicity, design flexibility and modularity allowing easier incorporation of

technological advances, except that final output power can not be shared.

HEVs can largely reduce emissions and thus is referred to as ultra low emission

vehicle (ULEV).

HEVs date far back to the earliest days of the automobile industry. Diesel

electric locomotives are series HEVs. The objective of the development of an HEV is

to overcome the problem of short travelling range associated with BEVs. So far

HEVs have demonstrated the feasibility of achieving the desired fuel economy, but

have been too complicated, too expensive, and too heavy. Now, with better

understanding of BEV technology, efforts are been made to explore whether BEV

technology can be synergistically combined with the technology of the internal

combustion engine in a cost-effective product. This product would emit much less

than the ordinary vehicle in the way of pollutants and deliver more in the way of fuel

efficiency.

1.3.3 Fuel-Cell Electric Vehicles (FCEVs)

A fuel cell reverses the process of electrolysis, in which an electric current breaks

down water into oxygen and hydrogen gas. Since the beginning of 1990s, several

automobile manufacturers have begun launching intensive R&D activities based

around the fuel cell. Fuel cells are regarded as attractive by the automobile

manufacturers because those automobile manufacturers believe that [13]:

� Fuel cells are more efficient than internal combustion engines.

� They can greatly reduce greenhouse gas emissions from road vehicles. With

hydrogen-powered fuel cells or direct methanol fuel cells one can produce

near zero emission vehicles.


� They don't suffer from drawbacks of battery powered EVs, such as short

range and short-lived batteries.

� They are virtually maintenance free.

� They are suitable for mass-production, and no expensive metals except for

platinum are used.

Figure 1.4 shows one configuration of FCEV drive system. In this concept,

methanol liquid which is a relatively low-cost by-product of waste materials and does

not need special safety measures, is converted into hydrogen via an on-board water-

vapor reformation process. In turn, the hydrogen gas is then fed into the fuel cells

where it reacts with atmospheric oxygen to produce the electrical energy that powers

the vehicle. The process gives near-zero emission levels and eliminates the need for

the bulky hydrogen tanks or heavy batteries previously required. In terms of driving

dynamics and travelling range, this hydrogen-powered car is on a par with

conventional petrol and diesel-powered vehicles [14].

However, many technical and economical obstacles remain on the path to

produce a fuel cell vehicle [14]:

� It is not entirely sure that FCEVs really "work" in the fuel economic,

technological and social sense.

� Can FCEVs be produced at commercially acceptable costs?

� There is no general consensus about which fuel should be used.

� There is as of yet no distribution infrastructure for methanol or compressed

hydrogen.

F ue lC e lls

M o to r T ransm issio n W hee lsW ate r-vapo rR e fo rm atio n

M eth ano lL iq u id

Figure 1.4 Block diagram of a FCEV propulsion system


1.3.4 A Few Important Notes

� Many different alternatives to ICE vehicles, BEVs, HEVs and FCEVs, seem

possible at this stage.

� Development of each EV technology, BEV, HEV and FCEV, is beneficial to

the other two. For example, an ultracapacitor bank can be used for BEVs,

HEVs and FCEVs to achieve good performance of acceleration and

deceleration, and a small fuel cell can be used as a range extender for a BEV.

Progress of development of different EV technologies, BEV, HEV and

FCEV, should eventually lead to the best solution to problems associated with

ICE vehicles.

� The most important fact is that no matter which of EVs, HEVs, or FCEVs

ultimately dominate, all of them employ electric motors to drive the vehicles.

Electric drives and their power electronics for vehicle propulsions deserve

more investment in research and development.

1.4 EV Research Project at the Northern Territory Centre

for Energy Research

The Northern Territory Center for Energy Research (NTCER) at Northern Territory

University is currently undertaking a wide range of R&D projects in the area of

renewable energy and power electronics, including an electric vehicle research

project, NTCER EV, supported by the Northern Territory Power and Water Authority

(NT PAWA) in Darwin, and an Australian Research Council grant. The goal of this

EV project is to modify an existing ICE vehicle to produce a short range, commuter

EV. This type of vehicle has many applications in the Darwin region where the

terrain is flat and the distances to cover are short. In particular the NT PAWA is

interested in using NTCER EV in their fleet of vehicles used by personnel who

record readings on power meters. Other interests already shown for the NTCER EV

are for parking inspectors, for general commuter use, and for demonstration.


This research concentrates on, but is not limited to, the development of the

NTCER EV. The technologies developed in this research are also valuable for

general EV applications, and even wider applications [15-23].

The NTCER EV project has been divided into two phases. Most of the technical

outcomes of the Phase 1 project are presented in this thesis. Phase 2 will see the

refinement and adjustment of the chosen technique, an increase in power, and in-

vehicle experimental trials. These are further work, and therefore are not included in

this thesis.

1.5 Contributions and Outline of the Thesis

Energy and power control strategies in EV propulsion remain a substantial research

topic which require more investigation, and power electronics plays a key role in EV

development. This thesis studies the control strategies and power electronics for the

NTCER EV. Many of the technologies discussed can be used in more general

applications. The following describes the contributions and outlines of this thesis.

Chapter 2 presents the analysis of measured data on daily drive cycles obtained

from an ICE meter-reading vehicle. To understand the expected travel profile of the

NTCER EV and its expected operating performance, a data logging system was

developed and installed on an ICE meter-reading vehicle to measure the daily drive

cycles of the vehicle. The obtained drive cycle data were closely examined and

analyzed, and much meaningful data, in terms of drive range, average speed, top

speed, acceleration rate, braking frequency and ratio, etc., were obtained. The

measured drive cycle data and its analysis provide a basis for the EV propulsion

system design. The analysis results were also compared with the (US) Federal Urban

Driving Schedule (FUDS). It has been concluded that the NTCER EV is similar to a

typical urban EV, making the NTCER EV research more generally applicable.

Chapter 3 discusses several advanced control techniques for BDCMs, with

emphasis on EV applications.

Firstly, the impact of PWM strategies and commutation schemes on the

performance of BDCMs is explored. A comparison between bipolar voltage

switching PWM and unipolar voltage switching PWM is made and selection criteria


are given. Two solutions are proposed to the phase current distortion problem of

BDCMs in regeneration mode. One is a combination of unipolar voltage switching

PWM, bipolar voltage switching PWM and a six-step commutation scheme, and the

other is a twelve-step commutation scheme. It is concluded that the former is a better

control scheme for the EV propulsion system in terms of efficiency, rectangularity of

phase current waveforms and circuitry simplicity.

Secondly, several techniques for speed extension and constant power operation

of BDCMs are discussed. Those include flux weakening, advancing the phase angle

of the excitation transition points, mechanically varying the air gap of the motor, and

the voltage boosting method. Of those, the proposed voltage boosting method has

distinct advantages.

Thirdly, two different current control methods of BDCMs at speeds above the

motor base speed, when using a boost converter for speed extension of the BDCMs,

are studied. Control system stability analysis in the s-plane when using the two

different methods is presented. A clear conclusion is arrived at.

Finally, this chapter studies the control of regeneration current of BDCMs when

the motor speeds are above the base speed. This is a well-known difficult issue in

BDCM control. An effective control method is proposed.

In Chapter 4, the power and energy requirements for the NTCER EV propulsion

system are analyzed. After an overview of EV batteries, it is concluded that none of

the currently available EV batteries can ideally meet EV requirements in terms of

power density, energy density, lifecycle and cost. Accordingly, a novel energy and

power management scheme has been proposed. The proposed energy and power

management scheme employs the newly developed non-flowing zinc-bromine battery

(NFZBB) and ultracapacitors (UCs). NFZBBs are optimized for specific energy, low

cost and long life. Ultracapacitors featuring high power density and long life are used

to complement the NFZBB to achieve good acceleration performance at low speeds.

Combining the two different types of energy storage can optimize the EV propulsion

system to achieve high energy for drive range, high power for good acceleration

performance, long lifetime, and low cost.


The overall power control characteristic of the EV propulsion system, resulting

from the combination of the constant power control scheme above the motor base

speed, seen in Chapter 3, and the short-term power assist scheme, discussed in

Chapter 4, is presented. This maximum torque – speed envelope is similar to the

conventional torque-speed characteristic of an internal combustion engine plus

transmission. That is, the propulsion system provides high torque at low speeds and

lower torque at high speeds, with which human beings are familiar.

Both the short-term power assist scheme at low speeds and the constant power

control scheme at speeds above the base speed require a bidirectional dc-dc

converter. Instead of using two separate dc-dc converters, which will add

considerable complexity and cost to the propulsion system, a novel multifunctional

dc-dc converter has been proposed. It adds only three components to a simple

conventional bidirectional dc-dc converter and can implement all the functions

required by the overall energy/power management scheme. The operational principle

and the control rules of the multifunctional dc-dc converter are presented in detail.

Chapter 5 discusses two main topics. Firstly it covers the development of the

soft-switched multifunctional dc-dc converter. After an evaluation of various soft-

switching techniques, zero voltage transition (ZVT) soft switching technique is

chosen for the multifunctional dc-dc converter. The ZVT dc-dc converter has

minimum switching loss resulting in high efficiency, reduced electromagnetic

interference (EMI), and improved reliability for the power devices in the dc-dc

converter. Another important feature of the ZVT dc-dc converter is that no additional

switching voltage and current stress are imposed on the power devices, and the

capability of bidirectional operation of the basic dc-dc converter remains. Simulation

results are presented and verified by the measurement results of the built ZVT dc-dc

converter. The measured efficiency of the ZVT dc-dc converter is above 96%.

Secondly, it presents a novel design of a high power, ferrite cored inductor. At

high power levels, due to the large dimensions of ferrite cores and air gaps, the

influence of leakage flux and fringing flux on the inductor design becomes

significant, which is difficult to accurately predict. With conventional design of high

power ferrite cored inductors, difficulty exists in achieving a satisfactory overall


performance. A novel design for high power ferrite cored inductors employing a

dual-coil or distributed winding scheme is proposed by the author to combat the

problems. Results from both 3-Dimension (3D) finite element analysis (FEA) and

measurement verify the viability of the novel design for high power inductors. The

proposed inductor design allows better optimization by reducing the localization of

saturation and leakage flux, or EMI levels, achieving good utilization of the magnetic

cores.

In Chapter 6, a new intelligent gate drive module for high power IGBTs is

presented. The intelligent functions of this module include under voltage lockout

protection and overcurrent protection with fault status annunciation. The module

contains dual drivers which make it especially suitable for phase leg drive, and

outputs bipolar signals so that the “off” status of the IGBT is ensured when the IGBT

is required to be “off”. It needs only a unipolar dc supply due to the built-in, well-

regulated, and isolated floating power supply. This module features high current

capacity, powerful functions, high performance, wide application, and convenience

of use, and is suitable for high power applications such as in EV drives. The detailed

functions, operation, performance, and practical application issues are addressed.

Chapter 7 presents the development of a soft-switched on-board EV battery

charger with unity power factor. Guidelines for the selection of a suitable power

stage topology and soft-switching technique in EV on-board battery chargers is

given. An optimized charge profile is presented which prolongs the battery life while

maintaining the feature of fast charge. The steady-state duty cycle of the charger's dc-

dc converter is kept close to 45% to help achieve high efficiency in the charger. A

novel active pre-load with minimized power rate ensures that the battery will be fully

charged and also protects the charger on open load. The value of these techniques has

been verified by measurement results from a constructed on-board battery charger for

EV applications, which employs a ZCS boost PFC converter and an asymmetrically

controlled ZVS-HB dc-dc converter in cascade.

Chapter 8 summarizes the outcomes of this research and recommends further

work to be done.

14

Chapter 2

Data Logging from an ICE Vehicle andthe Data Analysis

2.1 Design of a Data Logger for Collection of Vehicle

Drive Cycle Data

To determine the existing vehicle acceleration performance, the drive range required,

and the energy used, a data-logging system was developed and installed on a meter-

reading vehicle, currently a conventional 5-speed ICE vehicle.

Figure 2.1 shows the block diagram of the data logger system. It consists of an

installed vehicle speed sensor, a relay, a Datataker DT-500 – a data acquisition

instrument manufactured by Data Electronics (Aust.) Pty Ltd, a host computer for

programming the Datataker and data loading, and uses the vehicle's brake switch,

ignition switch and battery.

2.1.1 Data Acquisition

The Datataker DT-500 is a microprocessor based intelligent data acquisition

instrument able to monitor, record and raise alarms for a wide variety of physical

parameters including temperature, pressure, flow rates, counts, events, etc [24]. It is

operated in conjunction with a host computer, which is used to enter commands and

programs, to read the data returned after the input channels are scanned. It has 10

CHAPTER 2 DATA LOGGING FROM AN ICE VEHICLE AND THE DATA ANALYSIS 15

H ighS peed

C ou n ters

C 1

D ig ita lI/O

D 1

D 2

G N D

R em ovab leR A M C ard

M ic ro-P rocessor

64 180

R S 232C O M M S(Iso la ted )

H ost C om pu ter

D T -500

In sta lledS peedS ensor

A C /D C

V eh ic leB rakeS w itch

R elay

V eh ic leIgn it ionS w itch

V eh ic leB attery

12 V

Figure 2.1 Functional diagram of hardware setup of the data logger system

differential or 30 single ended analog input channels, 4 digital input/output channels,

and 3 high-speed counters.

The Datataker can be supervised by a host computer. Communicating between

the Datataker and the host computer is done through an RS232 serial interface. An

application software package, DeCipher, is written specifically for supervising and

programming the Datataker. The Datataker is also capable of operating off line,

acquiring data into the memory for later recovery once it is programmed by the host.

This function is very useful here since the data logging is conducted with the

instrument placed in a moving vehicle and it is desirable that there is no involvement

of a host computer and human intervention while the vehicle is on duty.

Analog and digital input signals are converted to data in ASCII format, suitable

for reading directly into the host via the RS232 serial interface. The scanning of the

analog and digital input channels is at intervals of 1 second. All the scan data is

returned from the Datataker in standard ASCII character strings, and then imported

into spreadsheets for analysis.

The internal battery backed data storage memory of the Datataker is 64 Kbytes,

which is not big enough for this application. Thus, the data storage capacity is

extended by a plug-in memory card.


2.1.2 The Physical Parameters Measured

Figure 2.2 shows a photograph of the data logger. The following data was measured

and logged:

� Vehicle velocity vs. time;

� Brake application and release vs. time;

� Time of start and finish of each engine run.

(1) Vehicle velocity

To measure the vehicle velocity, tail shaft revolutions per second were measured by

using a Hall Effect sensor with a magnet attached to the front universal joint on the

tail shaft. The differential ratio and tire revolutions per kilometer were applied to this

data to obtain the vehicle's velocity according to Equation 2.1.

Velocity = (Pulses/second)*(Tire circumference)*3.6/(Differential ratio) (2.1)

(2) Application and release of the brake

The application and release of the brakes were obtained by sensing the voltage

applied to the brake lamp circuit of the vehicle.

Figure 2.2 A photograph of the data logging equipment


(3) The start and finish of each run

The start and finish of each engine run was obtained by using the vehicle's ignition

system to switch the data logger on and off.

2.2 Typical Drive Cycles of a Meter-Reading Vehicle

The ICE meter-reading vehicle alternates between two different types of route. Firstly

there is a domestic route which encompasses driving to the appropriate suburbs,

usually along a highway at high speeds, followed by short, slow driving between

domestic streets. In residential areas, the driver stops the vehicle and walks to read

meters from house to house. Figure 2.3 to Figure 2.5 show the daily drive cycles for

the domestic route for three days, including charts of both speed vs. time, and

applications of brake vs. time.

The second type is a commercial route, in contrast, which requires driving to the

location, parking the vehicle and walking to measure the meters from business to

business. In this type of route, the driver usually leaves the engine on while reading

meters, to maintain air-conditioning. Figure 2.6 shows the charts of both speed vs.

time and applications of brake vs. time for a typical daily commercial route.

It should be noted that in all the graphs of braking vs. time, data for braking

while the vehicle was not moving were excluded.

2.3 Analysis of the Measured Drive Cycles

2.3.1 Comparison between the Obtained Drive Cycle and the FUDS

Spreadsheet calculations on the obtained drive cycles of the meter-reading vehicle

have resulted in much meaningful data, as shown in Table 2.1. They are illustrated

below.

It is shown that the average speeds are quite low and the engine idling time

accounts for a considerable portion of the engine-on time, 18.4% on average, and up

to 41% in the case of the commercial route. That is, the ICE vehicle spends much of

the time idling, its most inefficient and most polluting mode. In contrast, even if an


0

20

40

60

80

100

120

0 2000 4000 6000 8000 10000

Time (s)

Spe

ed (

km/h)

(a)

1 - application of brake; 0 - release of brake

0

1

0 2000 4000 6000 8000 10000

Time (s)

Bra

king

(b)

Figure 2.3 Drive cycle of the domestic route, Jan-20, 1999(a) Speed vs. time

(b) Braking vs. time


0

20

40

60

80

100

120

0 1000 2000 3000 4000 5000 6000

Time (s)

Spe

ed (

km/h

)

(a)

1 - application of brake; 0 - release of brake

0

1

0 1000 2000 3000 4000 5000 6000

Time (s)

Bra

king

(b)

Figure 2.4 Drive cycle of the domestic route, Jan-21, 1999(a) Speed vs. time



020406080

100120

0 2000 4000 6000 8000

Time (s)

Spe

ed (

km/h

)

(a)

1 - application of brake; 0 - re lease of brake

0

1

0 2000 4000 6000 8000

Time (s )

Bra

king

(b)

Figure 2.5 Drive cycle of the domestic route, Feb-02, 1999(a) Speed vs. time



020

4060

80100

120

0 5000 10000 15000

Time (s)

Spe

ed (

km/h)

(a)

1 - applica tion of brake ; 0 - re lease of brake

0

1

0 5000 10000 15000

Time (s )

Bra

king

(b)

Figure 2.6 Typical daily drive cycle of the commercial route, Jan-25, 1999(a) Speed vs. time



Table 2.1 Analysis result of the measured typical daily drive cycles

DomesticRouteJan-20

DomesticRouteJan-21

DomesticRouteFeb-02

CommercialRouteJan-25

Average

Number ofbraking

189 162 140 454

Total brakingtime (s)

786 351 565 1616

Total vehiclemoving time (s)

8066 3665 5461 8242

Total engine-ontime (s)

9279 4351 6126 14059

Idling time (s) 1213 1189 665 5817Braking timeover vehiclemoving time

9.7% 9.6% 10.3% 19.6% 12.3%

Idling timeover engine-ontime

13.1% 8.1% 10.9% 41.4% 18.4%

Drivingdistance (km)

108.7 54.7 73 73.3 77.4

Average speed(km/h)

48.5 53.7 48.1 31.8 45.5

electric vehicle, when stopped, needs to provide power to air conditioning, it does not

generate pollution and its efficiency is much higher than an ICE vehicle.

It is also shown that braking was performed quite frequently. The ratio of the

braking time over the vehicle moving time was very high, 12.3% on average, and up

to 19.6% in the case of the commercial route. This implies that a large amount of

regenerative braking energy is available when an EV is used.

In conventional vehicles, the mechanical energy produced while braking a

moving car is dissipated by friction between the rubbing surface of the brakes, with

the creation of heat energy as the result. The idea behind regenerative braking is to

recover the energy lost in braking or deceleration and then reuse it to accelerate.

During deceleration or when travelling downhill, the EV propulsion motor can be

operated as a generator to charge batteries or other energy storage devices through

regenerative braking.

Recovery of braking energy can significantly contribute to the energy saving and

consequently the extension of the drive range, which will be explored in the later


chapters. The data obtained have confirmed that the meter-reading vehicle is a typical

electric vehicle application, in this respect.

From the drive cycles shown previously it can be seen that the meter-reading

vehicle needs to start and stop very often. This suggests high peak power demands

during quick accelerations, while the average power usage when cruising is much

lower. Based on this characteristic, a novel energy and power management scheme

which employs a power assist unit to load level the EV batteries for vehicle

performance optimization and cost reduction has been proposed, as will be presented

in Chapter 4.

It is interesting to compare the obtained drive cycles of the meter-reading vehicle

which we obtained with the (United States) Federal Urban Driving Schedule (FUDS).

Figure 2.7 shows the profile of speed vs. time of the FUDS. It can be found that a

great similarity exists between the FUDS and the drive cycles of the meter-reading

vehicle. They both have "stop-and-go" characteristics, that is, the vehicles run mostly

in town at low speeds with many starts and stops, and run at high speeds on the

highway for some time.

Spreadsheet analysis of the FUDS was further carried out to provide quantitative

comparison between the driving cycles of the meter-reading vehicle and the FUDS.

Table 2.2 shows some numbers resulting from analysis of the FUDS. The ratio of

0

20

40

60

80

100

0 500 1000 1500

Time (s)

Spe

ed (

km/h

)

Figure 2.7 The Federal Urban Driving Schedule (FUDS)


Table 2.2 Analysis results of the FUDS

Total vehicle moving time (s) 1112

Total engine-on time (s) 1374

Idling time (s) 262

Idling time over engine-on time 19.1%

Average speed (km/h) 38.6

Maximum speed (km/h) 90.1

Maximum acceleration rate (m/s²) 1.60

Maximum deceleration rate (m/s²) -1.47

vehicle idling time over engine-on time is as high as 19.1%, which emphasizes the

"stop-and-go" characteristics. While the maximum speed is 90.1 km/h, the average

speed on the FUDS is only 38.6 km/h. These numbers are very similar to those of the

drive cycles of the meter-reading vehicle. It was not possible to get real data on

braking from the FUDS. However, a maximum acceleration rate of 1.6m/s² and a

maximum deceleration rate of –1.47 m/s² on the FUDS were extracted. These rates

will be part of the design basis in setting up the specifications of the NTCER EV

propulsion system.

In short, the comparison between the drive cycles of the meter-reading vehicle

and the FUDS has indicated that the NTCER EV will be similar to a typical urban

EV. Therefore, the research on this NTCER EV will have significance for general

application EVs.

2.3.2 Acceleration Rates

Acceleration performance is one of the most important performance criteria of a

vehicle and has significant impact on propulsion system design, such as in

determining the maximum power of the vehicle. From the drive cycles obtained from

the meter-reading vehicle, many accelerations from rest to the peak speed and back to

zero, or to a low speed were collected. They were grouped according to terminal


velocity. Data from every individual acceleration were closely examined and all of

them were averaged to provide more practical data.

2.3.2.1 0-60 km/h Acceleration Data

Particular attention was paid to acceleration rates up to 60km/h, as they are of most

concern in urban driving. The average acceleration vs. time to 60 km/h (16.7 m/s) is

shown in Figure 2.8. It can be seen that, it took, on average, about 14 seconds to

reach 60 km/h from rest.

The average acceleration vs. achieved speed, up to 60 km/h is shown in Figure

2.9. Variation of acceleration rates due to the change of gears can be clearly observed

in the chart.

An electric vehicle potentially can have better acceleration performance than an

ICE vehicle. This is because a properly designed electric motor employed in an EV

for propulsion has linear torque-current characteristics. Theoretically, as long as

motor currents are large enough, satisfactory acceleration performance can be

achieved. The targeted acceleration performance is 0-60 km/h (0-16.7 m/s) in 9

seconds, which is better than the measured ICE vehicle acceleration performance.

0

2

4

6

8

10

12

14

16

18

0 2 4 6 8 10 12 14

Time (s)

Spe

ed (

m/s

)

Avg speed Jan 20

Ave speed Jan21

Ave speed Jan 25

Avg speed Feb 02

Average speed

1st Gear 2nd Gear

Target - 60km/h (16.7m/s) in 9 seconds Extended part

Figure 2.8 Average speed vs. time to 60 km/h - all routes


0.0

0.5

1.0

1.5

2.0

2.5

3.0

0 2 4 6 8 10 12 14 16

Speed (m/s)

Acc

eler

atio

n (m

/s²)

AverageAcceleration

1st Gear

Gea

r C

hang

e

2nd Gear

Required acceleration to obtain 60km/h in 9 seconds (1.85m/s²)

3rd Gear

Gea

r C

hang

e

Figure 2.9 Average acceleration vs. achieved speed up to 60 km/h - all routes

Practically, the maximum motor current will be limited by several factors, such as the

maximum current capability of the batteries

and the EV power electronics. These will be discussed in the following chapters.

2.3.2.2 0-80 km/h Acceleration Data

0-80 km/h acceleration data were also obtained, as shown in Figure 2.10 and Figure

2.11. It should be stated that accelerations from rest directly to 80 km/h or higher

speed do not occur very often. Accelerations to 80 km/h often start from a lower

cruising speed.

2.3.3 Frequencies of Acceleration to Different Speeds

Numbers of accelerations to different achieved speeds were counted and the results

are shown in Figure 2.12. It is very interesting to notice that most of accelerations

performed were at low speed range. 76% of all accelerations from rest are up to

speeds of 35 km/h or less. The speed profile on the FUDS in Figure 2.8 reflects a

similar feature, although some differences exist in terms of speeds to which the

vehicle accelerates from rest. The design of an acceleration assist unit will be based

on this fact.


0.0

0.5

1.0

1.5

2.0

2.5

3.0

0 5 10 15 20 25

Speed (m/s)

Acc

eler

atio

n (m

/s²)

Gea

r C

hang

e

Gea

r ch

ange

Gea

r ch

ange

1st Gear 2nd Gear 3rd Gear

Figure 2.10 Average acceleration vs. achieved speed up to 80 km/h - all routes

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

0 2 4 6 8 10 12 14 16 18 20

Time (s)

Acc

eler

atio

n (m

/s²) Ave Acc Jan 20

Ave Acc Jan 25

Ave Acc Feb 02

Average Acceleration

Figure 2.11 Average acceleration vs. time to 80 km/h - all routes


92 91

46

24

12 10

58

0

140

<20 20-25 26-30 31-35 36-40 41-45 >45

Speed Range (km/h)

Fre

quen

cies

27.6% 27.3%

13.8%

7.2%

3.6% 3.0%

17.4%

68.8%

76.0%

55%27.6%

Figure 2.12 Histogram of terminal velocity after acceleration from standstill - all routes

2.3.4 The Average Speed and the Maximum Speed

The average speed of the meter-reading vehicle was 45 km/h across the day,

excluding time when the vehicle was stopped. Although the maximum speed of the

meter-reading vehicle reached 100 km/h occasionally, speeds on most of the roads in

Darwin region are actually limited to 80km per hour. Because the need to provide

some margin for traffic condition, it is reasonable to design the maximum speed of

the meter-reading EV at 90 km/h, which is about the maximum speed on FUDS.

2.4 Specifications of the NTCER EV Propulsion System

Through data logging from the ICE meter-reading vehicle and data analysis, the

mission that the NTCER EV is intended to achieve has been well understood. Based

on this, appropriate specifications for the NTCER EV have been set up, as shown in

Table 2.3. The NTCER EV should provide better acceleration and much better

efficiency than an ICE meter-reading vehicle, as well as the benefits stated in Chapter

1. As indicated in the above analysis, the obtained meter-reading vehicle drive cycles

are similar to typical urban drive profiles, such as the FUDS. Therefore, the NTCER

EV will also be suitable for use as a city commuter car.


Table 2.3 Specifications of the NTCER EV propulsion system

Symbol Description Value

� Vehicle average speed 45 km/h (12.5 m/s)�m Vehicle top speed 90 km/h (25 m/s)�_3% Speed at 3% grade 70 km/h (19.4 m/s)�_6% Speed at 6% grade 50 km/h (13.9 m/s)

a Vehicle acceleration 1.85 m/s²(0-60 km/h in 9 seconds)

2.5 Review

To understand the expected travel profile of the NTCER EV and its expected

operating performance, a data logging system was developed and installed on an ICE

meter-reading vehicle to measure the daily drive cycles of the vehicle. The obtained

drive cycle data were closely examined and analyzed, and much meaningful data, in

terms of drive range, average speed, top speed, acceleration rate, braking frequency

and ratio, etc were obtained. The measured drive cycle data and its analysis provide a

basis for the EV propulsion system design. The analysis results were also compared

with the (US) Federal Urban Driving Schedule (FUDS). It has been concluded that

the NTCER EV will be similar to a typical urban EV, making the NTCER EV more

generally applicable.

30

Chapter 3

Several Advanced Control Technologiesin Permanent Magnet Brushless DCMotor Drives

3.1 Permanent Magnet Brushless DC Motors (BDCM) for

EV Propulsion

As with general-purpose variable speed systems, EV electric propulsion systems

described as the heart of EVs have recently experienced evolution from brushed DC

motor drives to AC motor drives. All the EVs exhibited at EVS-14 (1997), EVS-15

(1998) and EVS-16 (1999) employed AC motor drives except the SAXO from

CITRÖEN, France, which is still using a brushed DC motor drive. This evolution is

because of the problems with brush structures and recent rapid development of AC

motors and power electronics technologies. R&D efforts have been applied to three

main types of AC motors, induction motors (IM), permanent magnet brushless dc

motors (BDCM) and switched reluctance motors (SRM). A recent survey of the

world major electric vehicles (Table 3.1) shows that IMs and BDCMs have been used

in EVs by major EV makers and no use of SRMs [25], though some researchers have

claimed that SRMs are suitable for EV propulsion [26]. More research has been

conducted to find out which of the three types of the motor is best suited to EVs but

no clear conclusions have been reached [2, 25, 27-32]. Probably it is still too early

CHAPTER 3 SEVERAL ADVANCED CONTROL TECHNOLOGIES IN PERMANENTMAGNET BRUSHLESS DC MOTOR DRIVES

31

Table 3.1 A survey of the world's major electric vehicles (1996- )

Motor BatteryCompany Model Type Power

kWType Voltage

V

TopSpeedkm/h

Accel

km/h/s

Range

kmSeats Note

GM EV-1car

IM 100 LeadAcid

128 128 0-96/9

112~144

2

Ford RangerEV

IM 67 LeadAcid

312 120 0-80/12.5

92 2

Chrysler EPICminivan

IM 73.5 LeadAcid

128 0-96/9

96 5

Honda EVPLUS

car

PM 49 Ni-MH

288 130 0-96/17.7

150 4

Nissan Altra EVminivan

PM 62 Li-lon 120 0-80/12

200 4

Toyota RAV4-EV

sportsunility

PM 49 Ni-MH

110 0-96/17

200 5

Toyota Priuscar

PM 30 Ni-MH

5 HEV

* Luciolemini car

PM 36x2 LeadAcid

224 150 0-40/4.3

140 2

Fiat SeicentoElecttra

car

IM 15-30 LeadAcid

216 100 0-50/8

90 4

VolksWagen

GOLFIV

conceptcar

IM 52.5-150

LeadAcid

336 0-96/7.8

112 4

Volvo FL6truck

IM 65-185x2

Ni-Cd

216 90 15-20 HEV

ZYTEK ELISEsports car

PM 75x2 Ni-MH

300 144 0-144/11.2

192

Solectria Forcecar

PM 42 Ni-MH

180 112 0-80/18

168 4

CITRÖEN

SAXOcar

DC 11-20 Ni-MH

120 90 0-50/8.3

80 4

Renault Nextcar

PM 7x2 Ni-MH

148-167

HEV

Daimler-Benz

ELECTRO-VITO

108Evan

IM 40 Na-NiCI

2

280 120 120-170

to make such a conclusion, and more likely, different types of motor may best suit

EVs with different specifications and applications.


32

The NTCER EV propulsion system employs an axial flux permanent magnet

brushless dc motor developed by the NTCER. This is believed to be a good choice

for a number of reasons. Firstly, the availability and decreasing price of rare earth

permanent magnet materials with high energy product have been bringing the cost of

BDCMs down, and the availability of high-speed, low-cost MOSFETs and IGBTs

has lowered converter costs significantly. Therefore the cost of the whole BDCM

drive system compares favorably with brushed dc drive systems while offering

significant benefits in the area of reliability, ruggedness, and efficiency. Secondly,

compared to an induction motor, the BDCM offers the advantages of high power

density and high efficiency, reducing weight and giving longer driving range for a

given battery size.

In addition, the axial flux geometry of the motor has substantial advantages over

the more common radial flux machine for two reasons [33]. First, the aspect ratio can

fit comfortably into a wheel if so desired, and there are significant volume saving

over the more usual radial flux geometry, for which much of the internal volumetric

of the rotor does not contribute to power output. This advantage of volume efficiency

is very important for the implementation of the advanced in-wheel drives (or direct

drives) which are superior to the traditional geared drive trains. Second, a technique

for flux weakening that relies on mechanical adjustment of the air gap, and which

does not impinge significantly on the efficiency, becomes possible. Table 3.2 shows

the main electrical specifications of the motor, and Figure 3.1 shows a photograph

Table 3.2 Main electrical specifications of the NTCER motor

Type 3-Phase permanent magnet brushlessNumber of Poles 12Rated Voltage 180 VRated Power (continuous) 10 kWMaximum Speed 3000 rpmRated Torque 30 NmFlux Type TrapezoidalPhase Resistance 13.1 mohmPhase Inductance 91 mHEfficiency 96%


33

Figure 3.1 A photograph of the NTCER motor disassembled

of the motor, disassembled. The first version of this motor, used in NTCER solar

racing car, "Desert Rose", in 1993 World Solar Challenge, received the General

Motors Technical Innovation Award [33].

3.2 An Overview of a BDCM Drive System and its Control

A BDCM is similar in construction to a permanent magnet synchronous machine

(PMSM). The difference between them is that the former has trapezoidal flux and

current waveforms, while those of the latter are sinusoidal. Since the trapezoidal flux

waveform results in a higher average magnetization of the magnetic material of the

machine with a factor of round 1.3, there is an increase in the utilization of this

material resulting in increase of power density [34-36]. In addition, BDCMs use

concentrated windings as opposed to the distributed windings of PMSMs. The use of

concentrated winding patterns in conjunction with trapezoidal flux patterns result in a

trapezoidal induced back EMF waveform across each phase winding.

Figure 3.2 shows ideal back EMF waveforms and current waveforms of a

BDCM. At any time, only two motor phases carry currents and the current of the

third phase is forced to zero. Each phase conducts 120 electrical degrees and

commutation takes place every 60�. This is known as Six-Step Commutation or Six-

Step Control.


34

ea

ec

eb

180 °120 °60°0°-1 80° -6 0°

i a

i b

i c

-1 20°

(a)

ea

ec

eb

1 80 °1 20 °6 0°0 °-1 8 0° -6 0 °

i a

i b

i c

-1 2 0°

(b)

Figure 3.2 Ideal back EMF waveforms and current waveforms of a BDCM:(a) motoring mode, (b) braking mode

Figure 3.3 shows a functional BDCM drive system with closed-loop torque

control. The motor is star-connected so that it requires only six power devices. The

active power switches can be any type of power semiconductor switches that can be

turned on and off from control terminals, such as MOSFETs, IGBTs and MCTs. The

primary function of an EV controller is torque control and therefore, a typical EV

propulsion is a closed-loop torque control system. The driver gives current

commands through accelerator and brake pedals. The motor speed is open loop

controlled according to the driver's commands. When a “Cruise Control” function,

which is optional in vehicle designs, is required, a closed loop speed control system

with an inner current loop can be employed, as shown in Figure 3.4, where the speed

error is used to control the torque.


35

AB

C

S1 S3

S4S2

S5

S6

M

C om m u tationLog ic

G a te D rivers

P ositionS en sor

A /B /C

A /B /C

S1 ~ S6

A n a logu eS w itch es &O p -A m p s

H ysteresisC on tro ller

Im o to r

Ire f(T orq u e C om m an d )

V bat t C bu s

Figure 3.3 A BDCM drive system with closed-loop torque control

AB

C

S1 S3

S4S2

S5

S6

MV bat t

C om m utationLog ic

G ate D rivers

P ositionS en sor

A /B /C

A /B /C

S1 ~ S6

A na logueS w itch es &O p-A m p s

H ysteresisC on tro ller

Im o to rF /V

C onverter

+

V ref

(Sp eed C om m an d)

Iref

P IC on tro ller -

M oto r S peed

Cbus

Figure 3.4 A BDCM drive system with double-closed-loop control

Directing current into and out of the three phase stator windings is determined

by rotor position. Rotor position information is obtained from a simple Hall-effect


36

rotor position sensor or an indirect position sensor [22, 37-40]. The amplitudes of the

currents are proportional to the desired torque.

Modern electric drives and power supplies always operate in switched mode.

Popular methods of switching modulation include Pulse Width Modulation (PWM)

using a ramp comparison technique, Hysteresis Band Modulation, Delta Modulation,

etc. [41-45]. The regulation of the motor currents can be achieved through the control

of the duty cycle. For the Hysteresis Band Modulation and Delta Modulation,

switching frequency is not constant.

The ramp comparison technique where the current error acts as the modulating

wave and a triangle waveform acts as the carrier wave, has the advantage of limiting

the maximum inverter switching frequency to the triangle waveform frequency and

producing well-defined harmonics. However, this method has its shortcomings.

Multiple crossing of the ramp may become a problem when the time rate of change

of the current error exceeds that of the ramp.

Hysteresis current control is the simplest and most extensively used method. The

hysteresis comparator is used to impose a dead band or hysteresis around the

reference current. This scheme provides excellent dynamic performance since it acts

very quickly. An inherent peak current limiting capability is also provided. The

operation of the current loop is absolutely stable, as the actual current always follows

the current command. The main disadvantage of hysteresis current control with a

fixed hysteresis band is that the switching frequency varies with both motor speeds

and bus voltage. For applications where varying switching frequency is undesirable,

an adaptive hysteresis band current control PWM technique [44] or a switching

frequency feedback technique [46] can be adopted to effectively overcome this

disadvantage.

Four-quadrant operation of the BDCM, that is forward motoring, forward

braking, reverse motoring and reverse braking is achievable, based on the current

command, the motor revolution direction command, and the rotor position. Table 3.3

shows the required trigger signals for all switches, for four-quadrant operations of

BDCMs.


37

Table 3.3 Trigger signals for four-quadrant operations

Activated DevicesRotor PositionSignals Forward Direction Reverse Direction

A B C Motoring Braking Motoring Braking0 0 1 S2 S5 S6 S1 S6 S2

0 1 1 S5 S4 S6 S6 S3 S4

0 1 0 S4 S1 S2 S3 S2 S4

1 1 0 S1 S6 S2 S2 S5 S6

1 0 0 S6 S3 S4 S5 S4 S6

1 0 1 S3 S2 S4 S4 S1 S2

A dedicated inverter was successfully developed for NTCER solar car drive with

a maximum efficiency of 98.8% [22]. With some modification, mainly power, this

inverter can be used for this EV propulsion.

3.3 Impacts of Different PWM Strategies and

Commutation Schemes on Performance of BDCMs

Both PWM strategies and commutation schemes can have impacts on the

performance of BDCMs. They will be addressed in this section.

3.3.1 PWM with Bipolar Voltage Switching vs. PWM with Unipolar

Voltage Switching

For simplicity, a full-bridge dc motor drive system is used to model the BDCM drive

system in each 60� interval, shown in Figure 3.5. This modeling is valid, as only two

A

Sap

San

V d Cbus

Ea

B

Sap

Sbn

L aRa

Figure 3.5 Using a full-bridge dc motor drive system to model the BDCM drive system in 60� ��


38

phases of a BDCM carry current during its operation. In Figure 3.5, Ea equals the line

to line back EMF of the BDCM, and Ra and La are the line to line resistance and

leakage inductance of the motor windings, respectively.

PWM with bipolar voltage switching and PWM with unipolar voltage switching

are two typical PWM schemes. The operation of PWM with bipolar voltage

switching is illustrated in Figure 3.6. The diagonally opposite switches (Sap, Sbn) and

(San, Sbp) are treated as two switch pairs. Two switches in a pair are simultaneously

turned on and off. The two switch pairs are switched complementarily with a small

dead time in between to avoid “shoot through” in a phase leg. The output voltage

switches between –Vd and +Vd voltage levels.

In PWM with unipolar voltage switching, the switches in the two legs are not

switched simultaneously as in the bipolar voltage switching PWM scheme. Instead,

only one switch of one switch pair is executing switching while the other one is kept

on constantly, and the other switch pair is kept off, as illustrated in Figure 3.7. In this

type of PWM scheme, when switching occurs, the output voltage changes between

zero and +Vd or between zero and –Vd.

A

Sa p

Sa n

V d C bu s

Ea

B

Sa p

Sbn

L aR a1

2Ia

Sa p

Sa n

Sbp

Sbn

1

2

1 1

2 2Ia

V A B

V d

-V d

Figure 3.6 PWM with bipolar voltage switching


39

To determine which scheme should be selected in the application, the two PWM

schemes are compared. The comparison results are shown in Table 3.4. These results

are obtained based on the following assumptions:

� The motor operates in forward motoring mode (Quadrant I);

� The armature currents increases and decreases linearly in a switching period;

� The armature current is in continuous current mode unless otherwise specified;

� The on-voltage or on-resistance of the switches, and the forward voltage of the

freewheeling diodes are zero;

� The dead time between the two complementary PWM signals is negligible.

It should be noted that the bipolar voltage switching PWM scheme offers very

good dynamic performance due to the bipolar voltage. While the motor current with

unipolar voltage switching PWM scheme may be discontinuous at light load, where

the duty cycle is close to zero or very small, that with the bipolar voltage switching

PWM scheme is always continuous even when the average motor current is zero.

This is illustrated in Figure 3.8. With the bipolar voltage switching PWM scheme,

A

Sap

San

V d C bus

Ea

B

Sap

Sbn

L aR a1

2Ia

Sap

San

Sbp

Sbn

1

2

1 1

2 2Ia

V dV A B

Figure 3.7 PWM with unipolar voltage switching (motoring mode, continuous current)


40

Table 3.4 Comparison between bipolar voltage switching PWMand unipolar voltage switching PWM

Bipolar Voltage Switching PWM Unipolar Voltage SwitchingPWM

OutputVoltageEquation

aaa

aad Ridt

diLEV �� (1)

aaa

aad Ridt

diLEV �� (2)

aaa

aad Ridt

diLEV �� (1)

aaa

aad Ridt

diLEV �� (2)

AverageOutput

Voltage V0

dDV

1~0�DdVD )12( �1~5.0�D

RippleCurrent

sa

TDL

V)1(

2 0 � sa

TDL

V)1(0 �

Number ofExecutingSwitches

2 1

“Lubricous”Torque

Yes No

Note Duty cycle s

on

T

TD � ; Ton -- on time of San, Ts -- switching period

Sap,Sbn

Ia

San,Sbp

Figure 3.8 "Lubricous torque" offered by bipolar voltage switching PWM scheme

when the duty cycle is 50%, although the average current is zero, there exist regular

ripple currents in the motor winding which results in a "lubricous torque". This

lubricous torque is especially helpful in a servo motor system and in a wide speed

range motor drive system for excellent dynamic performance at very low speeds.

However, this characteristic is not so important to an EV drive system.

In sum, PWM with unipolar voltage switching offers the advantage of lower

switching loss and lower current ripple, while PWM with bipolar voltage switching


41

offers better dynamic performance. As efficiency is a very important performance

criterion, while the lubricous torque is less important for EV propulsion, the BDCM

drive system the NTCER EV employs a unipolar voltage switching PWM scheme.

3.3.2 Performance Improvement of BDCM in Regeneration Mode

Six-step control method is commonly used for the control of BDCMs. While six-step

control in cooperation with a unipolar voltage switching PWM scheme, shown in

Figure 3.9, works well in motoring mode, the phase current during commutations is

not completely controlled in regeneration mode [21]. Figure 3.10 shows a single

phase current waveforms in regeneration mode, simulated in SPICE. As can be seen,

in motoring mode, the commutation of phase current is very fast and the motor

nonzero phase current is exactly 120� wide. In regeneration mode, however, the

commutation of the phase current in a half cycle becomes much slower and the

nonzero current waveform becomes wider than 120�. The cause of this problem is

illustrated in Figure 3.11. At the transition point, �t = 0�, the current in phase B

should be transferred to phase C. However, since there are no switches in the phases

involved in the phase current transitions, the phase current transition occurs in a

relatively uncontrolled fashion.

S1

S6

S5

S4

S3

S2

1 8 0 °1 2 0 °6 0 °0 °-1 8 0 ° -1 2 0 ° -6 0 °

P W M C on stan tly O n

Figure 3.9 Unipolar voltage switching PWM with six-step commutation


42

Figure 3.10 Simulated phase current waveform in regeneration mode using unipolar voltageswitching PWM with six-step commutation scheme

S2: P W M S w itch

S2: P W M S w itch

AB

C

S1 S3

S4S2

S5

S6

MV batt C bus

AB

C

S1 S3

S4S2

S5

S6

MV batt C bus

Figure 3.11 Illustration of the cause of the distorted motor current in regeneration mode

This current distortion results in the decrease of the RMS value of the phase

current, and further, the decrease of the average regeneration power, unless the peak

regeneration current limit is increased. However, an increase of the current limit is

undesirable for the whole system. The issue of this distorted current in regeneration

mode has gained little attention since for most BDCM applications operation in

regeneration mode is not required often, and is of less importance. In the EV


43

application, however, regenerative braking is important, and is performed very often.

Therefore, it is valuable to improve the regeneration performance of the BDCM. Two

methods have been proposed for this improvement as following.

3.3.2.1 Method I – Combining Unipolar Voltage Switching PWM and Bipolar

Voltage Switching PWM

The use of bipolar voltage switching PWM can eliminate the current distortion

problem of BDCMs in regeneration mode. Bipolar voltage switching PWM with six-

step commutation is shown in Figure 3.12. Figure 3.13 illustrates the phase current

transition in regeneration mode using bipolar voltage switching PWM. At the

transition point �t = 0�, current in phase B needs to be transferred to phase C. Since

PWM switches are always in the phases involved in phase current transitions, such

transitions always take place in a controlled fashion. Figure 3.14 shows the simulated

phase current waveform in regeneration mode using bipolar voltage switching PWM.

As can be seen, phase current is exactly 120� wide and the commutation is very fast.

However, bipolar voltage switching PWM is undesirable in motoring mode

owing to high switching loss. A combined PWM scheme has been proposed in this

project, that is, using bipolar voltage switching PWM in regeneration mode and

unipolar voltage switching PWM in motoring mode. By this method, both high

S1

S6

S5

S4

S3

S2

1 8 0 °1 2 0 °6 0 °0 °-1 8 0 ° -1 2 0 ° -6 0 °

P W M

Figure 3.12 Bipolar voltage switching PWM with six-step commutation


44

S3: P W M S w itch

S5: P W M S w itch

AB

C

S1 S3

S4S2

S5

S6

MV bat t C bu s

AB

C

S1 S3

S4S2

S5

S6

MV bat t C bu s

Figure 3.13 Illustration of phase current transition in regeneration modeusing bipolar voltage switching PWM

Figure 3.14 Simulated phase current waveform in regeneration mode using unipolar voltageswitching PWM with six-step commutation scheme

efficiency and good motor current waveforms in all operation modes can be

achieved. The changes of the PWM schemes can be easily determined by the braking

commands. This method improves the performance of regeneration braking at no

cost.


45

3.3.2.2 Method II – Twelve-Step Commutation

Another solution to the distorted current waveforms in regeneration mode is to

modify the six-step commutation into twelve-step commutation. This was

successfully implemented by the author in a flywheel energy storage system

employing an extremely high-speed BDCM for a satellite [21]. Twelve-step

commutation divides the 360� into 30� segments, as illustrated in Figure 3.15. During

each 30�, the three-phase inverter is still equivalent to a full-bridge inverter. Figure

3.16 illustrates the phase current transition in regeneration mode using twelve-step

commutation. At the transition point �t = 0�, current in phase B needs to be

transferred to phase C. Since PWM switches are always in the phases involved in

phase current transitions, such transitions always take place in a controlled fashion.

Figure 3.17 shows the simulated phase current in regeneration mode using twelve-

step commutation.

It is necessary to note that the twelve-step commutation in BDCM drive and the

twelve-step operation in a multilevel converter [45] are different issues, although

their names are similar.

If a discrete rotor position sensor is to be used, twelve-step control requires

modification of the conventional rotor position sensor that usually takes the form of

three inexpensive hall-effect sensors or optocouplers. In general, the discrete rotor

S1

S6

S5

S4

S3

S2

1 80 °1 20 °6 0°0°-1 80 ° -1 20 ° -6 0°

P W M C onstan tly O n

Figure 3.15 Twelve-step commutation scheme


46

S3: PW M S w itch

S5: PW M S w itch

AB

C

S1 S3

S4S2

S5

S6

MV batt Cbu s

AB

C

S1 S3

S4S2

S5

S6

MV batt Cbu s

Figure 3.16 Illustration of phase current transition in regeneration modeusing twelve-step commutation scheme

Figure 3.17 Simulated phase current waveform in regeneration modewith twelve-step commutation scheme

position sensor and / or the position signal procession circuit for twelve-step

commutation will be more complicated than that for six-step commutation. This is

because six-step commutation divides the 360� into six 60� segments while twelve-

step commutation divides the 360� into twelve 30� segments and then requires more

detailed information of the rotor position.


47

It is very interesting to point out that there are strong links or similarities

between Method I and Method II. They can be easily observed from Figures 3.11 –

3.17. With twelve-step commutation the operation and performance of the BDCM in

motoring mode is the same as that using unipolar voltage switching PWM plus six-

step commutation, while its operation and performance in regeneration mode is the

same as that using bipolar voltage switching PWM plus six-step commutation.

Nevertheless, Method I and Method II have their own most suitable applications.

In EV propulsion, Method I is preferred, since Method II needs a more complicated

rotor position sensor or a rotor position signal processing circuit. In the space

application, on the other hand, Method II is preferred, as the control signal for the

change of the PWM modes, like the braking command in the EV operation, is not

available in the space application.

3.4 Speed Extension and Constant Power Operation of the

BDCM

The desired torque-speed characteristic for traction applications is constant torque

below the base speed of the motor and constant power above the base speed, which

implies a reduction in the achievable torque as speed increases. This torque-speed

characteristic minimizes the power rating of the motor and inverter, and

consequently, the cost, weight, and volume of the whole propulsion system. This

torque-speed characteristic has been very much imprinted on human experience since

ICE vehicles slow down for gradients but can negotiate very steep inclines at low

speeds.

However, the natural torque-speed characteristic of a BDCM is constant torque

that is proportional to the armature current, and the maximum speed of a BDCM is

determined by battery voltage. Figure 3.18 shows this difference of the torque-speed

characteristic.

In general, a large ratio of the maximum motor speed to the base speed in

constant power suggests an effective reduction of the motor power rating [2]. The


48

S p eed

To

rqu

e

B D C M n a tu ra l ch arac teris tic"C on stan t T orq u e"

D esirab le ch arac teris tic fo rtrac tio n ap p lication"C on stan t P ow er"

Figure 3.18 Natural torque-speed characteristic of a BDCM and desired torque-speedcharacteristic for traction application

desired ratio varies with specific motor design and the drive system design. In EV

applications, battery voltage can vary as much as ±35% from the nominal value. This

factor should be taken into account in designing the EV propulsion system.

It should be noted that power electronics allows the motor to operate at any point

in the torque-speed plane, below the envelope defined by the maximum torque, the

maximum speed and the maximum power. Using power electronics to achieve speed

extension will be addressed shortly.

While an EV can be designed with direct wheel drive, since an electric motor has

a broad constant torque range, a single-gear-ratio transmission can free up the motor

speed from the vehicle speed and allow motor size reduction, the optimization of the

motor design and propulsion system design. The gear ratio and size will depend on

the maximum motor speed, maximum vehicle speed, and the wheel radius.

The single significant problem with permanent magnet motors in EV

applications is their speed limitation, determined by battery voltage. Several methods

to extend the speed range of BDCM will be investigated below.

3.4.1 Flux Weakening Method

For an induction motor and a non-permanent-magnet dc motor, extension of speed

range and constant power operation region can be satisfactorily achieved by flux-

weakening control [47-49]. However, extension of speed range into constant power

region by flux weakening has not been satisfactorily achieved in most PM machines.


49

Flux weakening control in some PMSMs is possible when the machines are designed

with large armature reaction effects [50-53]. In these machines, a component of the

stator current can be employed to weaken the air gap field produced by the permanent

magnet. Unfortunately, the large circulating currents involved significantly degrade

the efficiency and this method requires complicated rotor design and fabrication.

Moreover, this conventional flux-weakening control cannot be readily applied to

the PM brushless dc motor drive that is fed by squarewave current [28]. This is

because that the conventional d-q coordinate transformation, which is always used in

the flux-weakening control algorithm, can not be directly applied to the squarewave

motor drive. Although Fourier analysis can be employed to solve this problem by

resulting the squarewave into the corresponding fundamental and harmonic

sinewaves, the method is too cumbersome to be implemented. Shifting the “phase” of

the trapezoidal current will be addressed later.

3.4.2 Mechanically Varying the Air Gap

The method of speed extension of BDCM by mechanically varying the air gap has

been explored at NTCER [33, 54]. The axial flux permanent magnet brushless dc

motor developed by NTCER for its solar racing car can be readily dismantled and

reassembled with a range of spacers on the shaft, providing different air gaps. During

the race, the motor can be adjusted every evening to suit the predicted conditions for

the next day.

Versions allowing such adjustment in real time during operation are now

commercially available. This simple flux-weakening technique does not impinge

significantly on the efficiency. Within a reasonable band, increasing the air gap

increases the copper loss as the torque constant decreases, but decrease the iron loss

as the flux density reduces.

It should be noted that such mechanical variation of the air gap is of course not

possible in a BDCM with radial flux geometry.


50

3.4.3 Advancing the Phase Angle of the Excitation Transition

Points

Another way to extend the speed range of BDCM is to advance the phase angle of the

excitation transition points relative to the back EMF [55]. With this method, current

in the on-coming phase is given a controlled time interval (advanced angle between

0° to 60°) to build up before the back EMF increases and the speed range can be

extended by 2:1 or more.

It should be recognized that speed extension with this approach comes at a price

of increased peak phase current amplitude and significant increases in the torque

ripple and copper loss. Also, a more sophisticated rotor position scheme than a basic

Hall sensor scheme is needed. A high-resolution encoder or resolver that is more

expensive needs to be used.

3.4.4 Voltage Boosting Method Employing a Bidirectional Direct

DC-DC Converter

All the above methods of speed extension of BDCMs, special characteristic of the

motor were explored. The research in this thesis addresses an alternative method to

achieve speed extension of BDCM by boosting the battery voltage to a higher voltage

through a direct dc-dc converter [17, 18]. Figure 3.19 shows this scheme.

This method has a number of significant features, as described below.

1. It suits all different types of BDCMs and PMSMs;

2. It does not add constraints on the motor design, so that the motor design can be

optimized with larger freedom;

3. With this method, a very large speed extension range is achievable;

4. It allows the regenerative braking current of a BDCM or PMSM above the base

speed to be well under control, by operating the dc-dc converter in current-

regulated buck mode. It is well known that control of regeneration current of

BDCM or PMSM above the base speed in a conventional motor/inverter is very

difficult since the regenerative current goes through freewheeling diodes, as in

an uncontrolled rectifier.


51

AB

C

S1 S3

S4S2

S5

S6

M

Inverter/m otor

L 1

B att Sbo o st

Sbu ck

D C -D C C onverter

Cbu s

Figure 3.19 Speed extension of BDCM by employing a dc-dc converter

It is very important to note that in the NTCER EV, this bidirectional direct dc-dc

converter serves not only for the purpose of speed extension, constant power

operation and controlled regenerative braking above the base speed, but also for

power assist purposes in a novel power management scheme, which will be

addressed in Chapter 4. The cost of the converter is easy to justify, especially

considering the multifunction it provides.

3.5 Current Control of BDCM at Speeds above the Base

Speed in Motoring Mode when using the Voltage

Boosting Method

This section studies two different current control methods of BDCMs at speeds

above the base speed when using a boost converter for speed extension of the

BDCMs. The study focuses on control system stability analysis in s-plane when using

the two different methods.

3.5.1 Method I: The Converter – Current controlled; The Inverter

– Commutation Only

Since a boost converter is employed, the question arises, can we use the boost

converter to regulate the motor current while we let the inverter perform


52

commutation only in order to minimize the switching loss? Allured by the

minimization of switching loss, the author in [56] tried to implement this control

scheme but did not succeed. This section 3.5.1 looks at this control issue.

As the inverter only does the job of commutation, it can be neglected in

modeling the control system. In a 60° interval, the three-phase BDCM is equivalent

to a dc motor, as shown in Figure 3.20. This particular circuit configuration suggests

that the characteristic of the boost converter output is neither a current source nor a

voltage source. The large bus capacitor makes the circuit lose the characteristics of

current source output, while the intended closed-loop motor current control will

make the circuit lose the characteristics of voltage source output.

For a better understanding, the dynamic behavior of the system should be

approximated by a mathematical model. The system is a double current loop control

system. The current reference of the inner loop is obtained from the output of the PI

controller of the outer loop. In the inner current loop, since there is only one energy

storage component L1 that implies a first order pole in the left half plane (LHP), very

good control performance can be readily achieved. Then the transfer function of this

closed current loop, Gi1(s), can be optimized as a small inertia element. For further

L 1

V in

Sbo o st

D

Cbu s

Ea

L a

Ra

R in

Ia

V bu s

I in

Ia_ ref

P I_ 2-

++-

P I_ 1

P W MM od u la tion

& D river

B D C ME q u iva len t

C ircu it

Figure 3.20 Motor current control Method I


53

simplicity, we assume that the inductor current follows its reference perfectly. Thus

we have

1)(1 �sGi . (3.1)

The motor current control loop is then simplified as shown in Figure 3.21 in which

Kci(s) is the transfer function of the converter. Now we go back to Figure 3.20 to

derive Kci(s).

Assume that the switch and diode are ideal (no loss and no energy storage). The

instantaneous power equation relating input and output quantities of the system is

obtained by differentiating the energy quantities

)2

1()

2

1( 222

1 busbusininn vCdt

diL

dt

dRiiV inini ��

aaaaaa EiRiiLdt

d�� 22)

2

1( . (3.2)

Linearizing the power equation around the operation point (Vin0, Iin0, Ea0, and Ia0)

yields

busbusbusininininininin vVsCiIsLiIRiV �� 00102

aaaaaaaa iEiIRiIsL �� 000 2 . (3.3)

Considering

aaaaaa

abus iRsLRidt

diLdv �� )()(0 (3.4)

and simplifying Equation (3.3) give

2001 [ sVCLiiIsL busbusaainin ��

]2)( 000 aaaaabusbusa EIRsILVCR �� . (3.5)

Therefore, we have

I aP I_2 G i1(s)= 1 K ci(s)

Ind u cto r C urren tTem p la te

+-

I a_ ref

Figure 3.21 Motor current control loop with unspecified control plant (Method I)


54

00002

0

0001

2)(

2)(

aaaaabusbusabusbusa

inininin

EIRsILVCRsVCL

VIRsIL

iniaisciK

��

��

�

��

edscs

bas��

��2

(3.6)

where,

01 inILa �� (3.7)

002 ininin VIRb �� (3.8)

0busbusa VCLc � (3.9)

00 aabusbusa ILVCRd �� (3.10)

002 aaa EIRe �� . (3.11)

At the rated operation point,

aaaainininin RIEIRIIV 2000

2000 �� . (3.12)

From Equation (3.12), we can find Iin0. With the false solution rejected, we have

in

aaaaininin

R

RIEIRVVinI

2

)(40

2000

200 ��

� (3.13)

In addition,

aaabus RIEV 000 �� (3.14)

Using the values from Table 3.4, Vin0 = 120V, Ea0 = 170V, Ia0 = 80A, Rin = 0.06ohm,

Ra = 0.026ohm, Cbus = 4700uF, La = 182uH, L1 = 40uH, we have Iin0 = 122.18A,

Vbus0 = 172.6V, and

16.17403559.0000147.034.105002224.0)(

2 ��

ssssciK . (3.15)

Kci(s) contains a zero and two poles:

41 107365.4 �z

31 10)0825.11143.0( �� ip

32 10)0825.11143.0( �� ip

Figure 3.22 shows the root locus of Kci(s). As can be seen, the two poles are a

complex conjugate pair that are very close to the imaginary axis and far from the real


55

0 2000 4000 6000 8000 10000-6000

-4000

-2000

0

2000

4000

6000

Real Axis

Imag

Axi

s

Figure 3.22 Root locus of Kci(s)

axis. The root locus strongly suggests that little margin is left for designing a stable

and high-performance closed-loop system for the purpose of motor current control.

No matter whether a PI or a PID controller is used, it is very difficult to effect

satisfactory compensation.

The control difficulty mainly originates from the existence of the big capacitor

Cbus. If Cbus is decreased, the poles would move away from the imaginary axis and

towards the real axis. Unfortunately, a big Cbus is preferable to absorb ripple current

and make the bus voltage stable.

Increasing La will have a similar effect as decreasing Cbus, in terms of movement

of the poles. In addition, the system specifications such as maximum motor current,

battery voltage, motor speeds, etc, can marginally effect the location of the poles and

zero. Figure 3.23 shows their influences.

In some other applications where the motor inductance is large and a small bus

capacitor can be used, and dynamic performance is not required to be very fast, this

motor current control scheme above the motor base speed may be workable with

careful design of the controller.


56

Imag

Axi

s

L a increa ses

R eal A x is0

Cb us decreases.S ign ifican t e ffec t.

I a increa ses,Ea decreases.

M arg ian l effect.

I a increa ses,Ea increases.

Figure 3.23 Influences of system parameters on the location of the poles and zeros of Kci(s)

3.5.2 Method II: The Converter – Voltage Source Output; The

Inverter – Current Mode Control

An alternative method is to control the inverter and the converter separately, that is,

the boost converter is controlled as a voltage source output and the inverter is current

regulated with a voltage source input, as illustrated in Figure 3.24. Voltage source

inverters are used far more broadly than the current source topologies in BDCM drive

applications for two reasons. Firstly, the cost, size, and weight of electrolytic

capacitors tend to be significantly lower than DC link inductors in comparably rated

voltage and current source configurations. Secondly, voltage source inverters allow

the use of lower voltage rating MOSFETs or IGBTs, so that the best device

utilization and performance can be obtained.

The big capacitor in between decouples the two stages and makes the control

much easier. Excellent performance of the current control of the BDCM can be easily

achieved with a current-regulated voltage source inverter [21, 22]. Based on this fact,

for the purpose of the analysis and design of the boost converter, the current-

regulated inverter/motor can be modeled as a dc current source I0. In the same way as

in Section 3.5.1, we assume the transfer function of the inner current loop Gi(s)� 1.

Therefore the outer control loop will have the structure shown in Figure 3.25.


57

L 1

V in

Sbo o st

D

Cbus

I0

R in

V bus

I in

V bus_ref

P I_v-

++-

P I_ i

P W MM odu la tion

& D river

E q u iva len tC ircu it o f

M otor/in verter

Figure 3.24 Equivalent system diagram using Method II of the motor current control at themotor speeds above the base speed

Vbu sP I_v G i(s)= 1 K cv(s)

In d uc to r C u rren tT em p la te

+-

Vbu s_ re f

Figure 3.25 Outer control loop with unspecified transfer function of plant using Method II

Following the analysis based on energy balance of the power system in Section

3.5.1, we obtain the transfer function of the converter Kcv(s) in Equation 3.16.

00

0001

22

)(IsVC

VIRsIL

inibusv

scvKbusbus

inininin

��

��

��

dcsbas

�� (3.16)

where,

01 inILa �� (3.17)

002 ininin VIRb �� (3.18)

0busbusVCc � (3.19)


58

02Id � (3.20)

At the rated operation point,

002

000 IVRIIV businininin �� . (3.21)

With the false solution rejected, we have

in

busininin

R

IVRVVinI

2

40

00200 ��

� (3.22)

Again using Vin0 = 120V, Vbus0 = 170V, I0 = 80A, Rin = 0.06ohm, Cbus = 4700uF, L1

= 40uH, we have Iin0 = 120.61A, and

160799.053.105004824.0)(

��

ssscvK . (3.23)

Kcv(s) contains a RHP zero and a LHP pole:

4101874.2 �z

3.200��p

Figure 3.26 shows the root locus of Kcv(s). Let the transfer function of the PI

controller be:

ssscvG 160799.0)( �� (3.24)

0 0.5 1 1.5 2 2.5

x 104

-0.1

-0.08

-0.06

-0.04

-0.02

0

0.02

0.04

0.06

0.08

0.1

Real Axis

Imag

Axi

s

Figure 3.26 Root locus of Kcv(s)


59

Vbu s1

+-

Vbu s_ref

160799.053.105004824.0

�

��

ss

ss 160799.0 �

In d uc to r C u rren tTem p la te

G cv(s) Gi(s) K cv(s)

Figure 3.27 Mathematical model of the closed voltage loop using Method II

Hence, the mathematical model of the outer control loop is shown in Figure 3.27.

With the pole of Kcv(s) cancelled, the open-loop forward transfer function is:

)()()( scvKscvGsoG �

ss 53.105004824.0 �� (3.25)

The closed-loop voltage transfer function is:

)(1)(

)(sG

sGscG

o

o

��

53.1059952.0

53.105004824.0��

ss (3.26)

which only has a first-order pole in the LHP, suggesting a stable system. The step

response of the closed voltage loop is shown in Figure 3.28.

0 0.02 0.04 0.06 0.08 0.1-0.2

0

0.2

0.4

0.6

0.8

1

1.2

Time (seconds)

Am

plitu

de

Figure 3.28 Step response of the closed voltage loop


60

In the analysis above, we assumed that the transfer function of the inner current

loop Gi(s)� 1 for simplicity in addressing the motor current control method. More

practically, Gi(s) should be modeled as a small inertia element. With a PI controller,

the closed voltage loop can be compensated into an optimum second-order system

that can still have a very good dynamic performance. This has been adopted in the

NTCER EV since it offers advantages of high stability and easy control. The

switching loss is minimized through a soft-switching technique, which will addressed

in Chapter 5.

It is worth noting that a two-stage quasi-current source inverter scheme for a

BDCM, consisting of a current-regulated buck converter followed by a three-phase

inverter is workable [45, 57, 58]. It behaves generally as a type of current source

inverter except during phase current commutation intervals when it reverts to

operation as a voltage source inverter. Therefore it captures some of the best features

of both a current source inverter and a voltage source inverter. It is more natural than

the scheme employing a current-regulated boost converter followed by a three-phase

inverter that is only for commutation. This is because that the output of the buck

converter can behave as a current source due to the location of its energy storage

inductor.

3.6 Control of Regenerative Braking of BDCM above the

Base Speed

In a conventional BDCM drive system, it is very difficult to control regeneration

current when the motor speed is above the base speed, that is, when the motor line-

line back EMF is higher than the battery voltage. This is because the regeneration

current will go through the freewheeling diodes of the inverter back to the dc voltage

source and the inverter switches lose their control of the regeneration currents. In this

case, the motor / inverter acts like an uncontrolled rectifier. The natural regeneration

current Iregen-n is determined by


61

eqv

battllnregen R

VVI

��

�

(3.27)

where, Reqv is the total equivalent resistance of the current path, mainly including the

resistance of the battery, resistance of two phase armature windings of the motor, Vl-l

is line to line motor back EMF voltage. As Reqv is usually very small, this

uncontrolled regeneration current can be unacceptably large.

This section studies the control of regeneration current of BDCM when the

motor speeds are above the base speed.

3.6.1 When Motor Speeds are Much Higher than the Base Speed

In this case, the objective of the regeneration current control is to limit the

regeneration current to some maximum value. By operating the dc-dc converter in

current controlled buck mode, regulation of the regeneration current when the motor

line-line back EMF is higher than the battery voltage is possible, as shown in Figure

3.29. The reference for the inductor current is set according to the required

regeneration current of the motor. Therefore the regeneration current can be

effectively regulated not to exceed a maximum current limit. This operation mode

can be called Buck-Regeneration Mode.

L 1

B att

Sbuck

Sbo o st M

In verterC om m uta tion

on ly

C bus

R ectif ie r

Figure 3.29 Buck-regeneration operation when the motor speedsare much higher than the base speed


62

3.6.2 When Motor Speeds are only Marginally Higher than the

Base Speed

The overall objective of regeneration current control is to limit the regeneration

current to a maximum value and maintain it at this value for as long as possible to

achieve fast regenerative braking. When the motor speeds are only marginally higher

than the base speed, it is not possible to get high enough regeneration current even

though the buck switch is controlled constantly on. This is because the difference of

the motor line-line back EMF and the battery voltage is too small.

In this case, PWM switching operation of the inverter is activated while Sbuck in

the dc-dc converter is kept on and Sboost off constantly. As before, we use a full-

bridge dc motor drive system to equalize the BDCM drive in each 60��

shown in Figure 3.30. In this mode, operation of the inverter is similar to its usual

regeneration operation when motor speed is below the base speed, that is, boost-

regeneration mode [59]. For better understanding of this, we derive the equivalent

circuit below.

Due to the existence of the big electrolytic capacitor Cbus, the inductor L1 in the

dc-dc converter only affects very high frequency ripple and is neglected. For further

simplicity, we remove Sboost and short out Sbuck. Thus we have the equivalent circuit

as shown in Figure 3.31. To effectively perform regenerative braking, let San execute

PWM switching. Figure 3.32 shows the two states of the circuit when San is

executing PWM switching. It can be seen that the operation of the circuit is basically

equivalent to that of a boost circuit, as shown in Figure 3.33.

More accurately, this operation mode is a hybrid boost-regeneration mode

because here regeneration current can be controlled in the range from a minimum

value to a maximum limit, while in usual boost-regeneration mode regeneration

current can be controlled in the range from zero to a maximum limit. The minimum

regeneration current is determined by Equation 3.28. It should be stated that the


63

L 1

B att

Sbuc kC o nsta nt ly O n

Sbo o s t

C bus A

Sa p

Sa n

Ea

B

Sa p

Sbn

L aR a

Figure 3.30 Hybrid boost-regeneration mode when the motor speedsare marginally higher than the base speed

B att C bu s A

Sap

San

Ea

B

Sap

Sbn

L aR a

Figure 3.31 Simplification of Figure 3.30

B att C bus A

Sap

San

Ea

B

Sap

Sbn

L aR a

2

1

Ia

Figure 3.32 Two states of the circuits when San is executing PWM switching


64

B att Cbus

Sap

SanEa

Sbn

L aRa

Ia

Figure 3.33 The equivalent boost circuit

existence of the minimum regeneration current is not a problem at all since the

objective of regeneration current control is to limit the regeneration current to a

maximum value and maintain it at this value for as long as possible to achieve fast

regenerative braking.

The time to start the hybrid boost-regeneration mode will be determined by

measuring the difference of the bus voltage that approximates the motor line-line

back EMF, and the battery voltage. The threshold needs to be determined

experimentally for a specific drive system.

It should be noted that hard braking at high vehicle speeds when the motor

speeds are above the base speed would not be performed very frequently. In the case

of gentle braking, the regenerative braking energy is put back into the battery. In the

case of hard braking, a mechanical friction brake needs to be used. Mechanical

brakes will provide extra braking once the current limits are achieved.

3.7 Review

After a brief overview of general BDCM drive systems, this chapter focuses on

several advanced control techniques for BDCMs, with emphasis on EV applications.

Firstly, the impacts of PWM strategies and commutation schemes on the

performance of BDCMs are explored. A comparison between bipolar voltage

switching PWM and unipolar voltage switching PWM is made and the selection

criteria are given. Two solutions are proposed to the phase current distortion problem


65

of BDCMs in regeneration mode. One is a combination of unipolar voltage switching



control scheme in terms of efficiency, rectangularity of phase current waveforms and

circuitry simplicity.



of the excitation transition points, mechanically varying the air gap of the motor and




base speed, when using a boost converter for speed extension of the BDCMs, are

studied. Control system stability analysis in the s-plane when using the two different

methods is presented. A clear conclusion is arrived at.

Finally, the control of regeneration current of BDCMs when the motor speeds

are above the base speed is discussed. This is a well-known difficult issue in BDCM

control. An effective control method is proposed.

66

Chapter 4

Novel Energy and Power Managementin the EV Propulsion System

4.1 Power Requirements for the Electric Vehicle

Propulsion System

The main component of an EV is its electromechanical power train. The principal

objective of an EV system design is to determine the minimum drive weight, volume,

and cost that will meet the system specifications. As seen in Chapter 2, the driving

cycles of the NTCER EV is similar to FUDS. To simplify the analysis and design and

to extend the NTCER EV research to more general urban EVs, it is assumed in this

chapter that FUDS will be the driving pattern of the NTCER EV, unless otherwise

specified. As the FUDS is a standard testing cycle in urban driving, this assumption

also allows for the proposed system to be compared with other studies.

To optimize the design of the drive system, first of all the power requirements

for EV propulsion need to be looked at. Table 4.1 shows the NTCER EV data,

assuming Genesis sealed lead-acid batteries are used (Phase 1 Project). The variables

which define the continuous power and the peak power include:

1. rated vehicle velocity (speed);

CHAPTER 4 NOVEL ENERGY AND POWER MANAGEMENT IN THEEV PROPULSION SYSTEM

67

Table 4.1 NTCER EV data

Symbol Description Value

� Vehicle average speed 45km/h (12.5m/s)�m Vehicle top speed 90 km/h (25m/s)�_3% Speed at 3% grade 70 km/h (19.4m/s)�_6% Speed at 6% grade 50 km/h (13.9m/s)

a Vehicle acceleration 1.85 m/s²(0-60km/h in 9 sec.)

d Range between charges 60 kmm Vehicle mass 800 kgCrr Rolling resistance coefficient 0.008� Air density 1.2Cd Coefficient of aerodynamic drag 0.3A Front area 1.9 m²

�w Head wind velocity * �w = 0 used for calculation

� Road grade ** 3% and 6% grade used forcalculation

� Drive train efficiency 83% (assumed)

2. maximum vehicle velocity;

3. initial acceleration rate;

4. road load characteristic determined by a range of vehicle data;

5. electric motor power rating;

6. extension of constant power speed range beyond the rated speed;

7. gear ratio between the motor shaft and the wheel shaft (a variable transmission is

not essential with an EV).

This section will analyze the power determined by variables 1-4 while Chapter 3

discussed the power determined by the remainder.

4.1.1 Road Load Characteristics

The road load is the total force in Newton, as a function of speed, needed to be

provided by vehicle system to maintain that constant speed. The road load Fw

consists of rolling resistance Frr, aerodynamic drag Faero, and climbing force Fclm.

clmaerorrw FFFF �� (4.1)


68

The rolling resistance Frr is caused by the tire deformations on the road:

)cos(�� gmCF rrrr (4.2)

where Crr is the tire rolling resistance coefficient; m, g and � represent the vehicle

mass, the gravitational acceleration constant and the road grade angle, respectively.

Here, we assume the rolling resistance is independent of the vehicle velocity, which

is usually acceptable.

Aerodynamic drag Faero is the viscous resistance of air acting upon the vehicle:

2)(2

1wdaero vvACF �� (4.3)

�� Cd, the aerodynamic drag coefficient, A, the vehicle frontal

�� w, the head wind velocity.

The climbing force Fclm is determined by the vehicle mass, the gravitational

acceleration constant, and the road grade angle:

)sin(�� gmFclm . (4.4)

��Fclm have negative signs.

The wheel power Pw to overcome Fw is

clmaerorrclmaerorrww PPPvFFFvFP �� )( (4.5)

The power components required to overcome rolling resistance Frr, aerodynamic drag

Faero, and climbing resistance Fclm, respectively, are

vgmCP rrrr � )cos(� (4.6)

3)(2

1wdaero vvACP �� (4.7)

vgmPclm � )sin(� (4.8)

For simplicity, zero head-wind velocity and a ground level are assumed in

analyzing the vehicle power requirements, unless otherwise specified. Given the

NTCER EV data in Table 4.1, we can plot the required wheel power, power

components and power out of the batteries vs. the vehicle cruising speeds, as shown

in Figure 4.1. At low speeds, wheel power is used mainly to overcome the rolling

resistance while at high speeds, >70km/h, for example, wheel power is used mainly

to overcome the aerodynamic drag.


69

0

2

4

6

8

10

0 20 40 60 80 100

Speed (km/h)

Pow

er (

kW)

P_rrP_aeroP_wheelP_batt

Figure 4.1 Components of vehicle power vs. speeds

The vehicle not only needs to provide the power to overcome all the resistance

for cruising, but also needs to provide power for acceleration or overtaking:

vamPaccel � (4.9)

where, a is acceleration rate. The total power for cruising and acceleration is

accelw PPP ��max . (4.10)

Energy for rolling resistance and aerodynamic drag are not recoverable. The

reminder of the kinetic energy of the vehicle excluding that to overcome rolling

resistance and aerodynamic drag can be recovered by regeneration.

4.1.2 Components of Vehicle Power vs. Speed for the FUDS

In Chapter 2, it has been seen that in real driving, the acceleration rate is not constant

during acceleration and accelerations may not start from zero speed. Calculation of

the power requirements when travelling according to the FUDS was carried out based

on the EV data in Table 4.1 and the power formula above. Figure 4.2 shows the total

required power for the FUDS and Figure 4.3 shows the power components due to

rolling resistance, aerodynamic drag and accelerations and decelerations,

respectively. The required battery power and energy consumed on FUDS is shown in

Figure 4.4.


70

0

20

40

60

80

100

0 500 1000 1500

Time (s)

Spe

ed (

km/h

)

(a)

-30

-20

-10

0

10

20

30

0 500 1000 1500

Time (s)

Whe

el P

ower

(kW

)

(b)

Figure 4.2 Wheel power on FUDS(a) FUDS (b) Wheel power

Peak power 22kW, average power 3.22kW (motoring)

Figure 4.5 shows power required for climbing at different speeds on 3% grade

and 6% grade, respectively.


71

0.0

0.5

1.0

1.5

2.0

0 500 1000 1500

Time (s)

P_r

r (k

W)

(a)

0

1

2

3

4

5

0 500 1000 1500

Time (s)

P_a

ero

(kW

)

(b)

-30

-20

-10

0

10

20

30

0 500 1000 1500

Time (s)

P_a

ccel

(kW

)

(c)

Figure 4.3 Components of the vehicle power on FUDS(a) Power due to rolling resistance(b) Power due to aerodynamic drag

(c) Power due to acceleration and deceleration


72

-20

-10

0

10

20

30

0 500 1000 1500

Time (s)

Bat

tery

Pow

er (

kW)

(a)

0.0

0.5

1.0

1.5

0 500 1000 1500

Time (s)

Con

sum

ed E

nerg

y (k

Wh)

(b)

Figure 4.4 Required battery power and energy consumed on FUDS(a) Required battery power (b) Energy consumed

0

10

20

30

40

0 20 40 60 80 100

Speed (km/h)

Pow

er fo

r C

limbi

ng (

kW)

3% grade

6% grade

Figure 4.5 Power for climbing vs. speeds


73

4.1.3 Observations

Several very important observations emerge from the calculation results above:

� The maximum power of the NTCER EV is determined by the acceleration

requirement, not the maximum vehicle cruising speed. The maximum power for

acceleration is 22 kW while the power for cruising at 90km/h is 10.25 kW, of

which 3.07 kW is due to rolling resistance and 7.18 kW due to aerodynamic drag.

This tends to be true for almost all high-performance EVs, as quick acceleration

is a critical performance criteria and demanded by customers.

� Although the peak wheel power is as high as 22 kW, the average wheel power

for the whole cycle is only 3.22 kW, with a ratio of 6.83:1. If 83% system

efficiency is assumed, the required battery power and the energy consumed are

plotted in Figure 4.4. The high ratio of peak power to average power necessitates

a battery load-leveling scheme which will be addressed in Section 4.3.

� The regenerative braking energy available is large. The energy over the whole

cycle consumed to drive the wheels is 1227.3 Wh and the regenerative braking

energy available is 404.3 Wh representing 32.9% of the total energy consumed in

the wheels. Assuming 83% system efficiency, the energy taken out of the battery

is 1478.7 Wh. Therefore, regenerative braking energy available accounts for

27.3% of the energy taken out of the battery. This has confirmed that regenerative

braking is a big contributor to energy saving, extending driving range. However,

how much regenerative braking energy can be recovered will depend on the

regeneration efficiency. Thus improving regeneration efficiency is important.

� A peak power of 22 kW allows the vehicle to climb a 3% grade at 70 km/h and a

6% grade at 50 km/h. Human driving experience is that a vehicle slows down for

gradients, but can negotiate very steep inclines at low speeds. Fast acceleration is

usually not expected during vehicle’s climbing. Therefore, the maximum power

of the vehicle is determined by acceleration performance, not high speed cruising.

4.2 Overview of Commercial EV Batteries


74

To make EVs widely accepted by the public, they must be able to demonstrate a

satisfactory drive range, acceleration and deceleration performance, and low cost.

Currently the significant problems with EVs compared to ICE vehicles are their

relatively short drive range per charge and their high cost. The cost is determined by

very large, expensive, heavy and short-lived battery packs. Of all the sub-systems

used in EVs, the battery remains the main barrier to EV success. While some other

sub-systems in EVs need to be further improved, they are not the limiting factors to

either vehicle performance or large-scale production of EVs.

In order to optimize the design of the energy and power system of the NTCER

EV, this section looks at current EV batteries. There are many battery technologies

currently being developed for EV applications, with Lead-Acid, Nickel-Metal

Hydride (NiMH) and Nickel Cadmium (NiCd) being given the most attention.

Lithium batteries that are expected to be able to meet long-term goals for EV

batteries, featuring high energy density and numerically high cycle life, are still being

developed.

Figure 4.6 compares the current lead-acid, NiMH and NiCd batteries in terms of

specific energy, specific power, specific cost and the cycle life [6, 60]. In Figure 4.6,

the specific energy is defined at C/3, the specific power for 30 seconds at 80% DOD

and the cycle life when discharged to 80% DOD.

Lead-Acid Batteries

Lead-acid batteries are the most common technology, and are currently used in

commercially available electric vehicles since they are cheaper than alternative

technologies. Flooded lead-acid batteries were invented before the beginning of the

20th century and are still the best economically feasible technology for storing electric

power at this scale. Flooded batteries have the disadvantage that they require

electrolyte maintenance. Single-point watering systems that are available greatly

simplify that procedure.


75

0

200

400

600

800

1000

1200

1400

1600

Lead-acid 35 200 440 120

NiCd 40 175 1250 600

NiMH 80 150 1500 540

Energy (Wh/kg)

Power (W/kg)

Life (Cycles)Cost

(US$/Wh)

Figure 4.6 Comparison among lead-acid, MiCd and MiMH batteries

Recent advances in lead-acid chemistry management and construction have

produced batteries that are generally sealed and contain little if any free liquid

electrolyte. Their price is about double that of the flooded lead-acid battery, but

should drop as production volume increases. So far they have all demonstrated lower

number of cycles than flooded lead-acid batteries, but new battery management

techniques may improve that situation [61]. The exact life also depends on driving

and charging patterns [62].

Current EVs using lead-acid batteries as their main energy storage include GM

EV1, GM S-10 Electric Pickup, Ford Ranger Pickup, Chrysler EPIC Minivan, and

Fiat Ceicento Electtra [25].

NiCd Batteries

NiCd batteries have demonstrated greater specific energy, and larger cycle life than

lead-acid batteries and reasonably high specific power. The Volvo FL6 hybrid

electric truck uses NiCd batteries as auxiliary energy storage. However NiCd

batteries are much more expensive than lead-acid batteries. Cadmium is highly toxic,

so recycling efforts have to be managed very carefully.


76

NiMH Batteries

NiMH batteries are highly regarded as they currently have the most energy per unit

mass, highest cycle life of the three different batteries and possess high power

density. Current EVs using NiMH batteries as main energy storage include the Honda

EV PLUS, Toyota RAV4-EV, ZYTEK ELIES, Solectria Force and Citroen SAXO

[25]. Some HEVs use NiMH as auxiliary energy storage, such as the Toyota Prius

and Renault Next.

Unfortunately NiMH batteries are 4 to 5 times more expensive than lead-acid

batteries. Although the cost of NiMH batteries can be reduced as the production

volume increases, it will not be as low as lead-acid batteries due to the raw material

cost.

4.3 Non-Flowing Zinc-Bromine Batteries

As none of the present EV batteries are ideal, it is worth giving more attention to the

development of emerging battery technologies. Among numerous emerging battery

technologies, the non-flowing Zinc-Bromine battery (NFZBB) is a very promising

EV battery technology [19, 63].

The non-flowing Zinc-Bromine battery is a promising spin-off from work on the

flowing electrolyte zinc-bromine battery, by ZBB Energy Corporation in Western

Australia. This emerging battery technology was first reported in [63]. The NTCER

has been closely working with ZBB aiming to use the NFZBB for EV applications.

The new NFZBB will suit EV applications very well because it offers the following

advantages:

� High energy density for a long EV driving range;

� Capacity to withstand several thousand full discharge cycles without damage so

that the batteries will last for the life of the vehicles;

� Low cost production.

A prototype of the NFZBB which will suit EV applications is currently being

developed. Table 4.2 outlines the projected performance built in 24, 48 and 96 volt

configuration under continuous deep discharge cycling conditions. The analysis and


77

Table 4.2 NFZBB projected performance

Estimated Weight

Voltage Energy Power(continuous)

Power(30 sec.) Total Electr.

EstimatedEnergyDensity

(V) (kWh) (W) (W) (kg) (kg) (Wh/kg)24 1.0 800 2,000 20 12 5048 2.0 1,600 4,000 34 24 6096 4.0 3,200 8,000 62 48 64

design of the power and energy system of the NTCER EV in this chapter will be

based on these NFZBB data.

The NFZBB battery stack consists nearly entirely of low-cost polyethylene

plastic materials, with only a thin metal screen on the back side of the terminal

electrodes which is necessary to direct the electrical current in the x-y plane of the

electrode. Polyethylene can be formed and joined using common industrial

techniques such as injection moulding, extrusion and heat welding techniques. All

these techniques lend themselves to highly automated and low cost mass production.

The electrolyte is based on zinc bromide salt solution, which is commonly available

at low costs. The low cost of both materials and manufacturing ensures that a low

cost battery can be produced.

The technology is considered safe, since none of the materials used in cell builds

or electrolytes are classified as toxic. Only when a battery is charged the corrosive

element bromine is produced in the positive cell halves, however, only a very low

concentration of bromine is present in a discharged battery. The health concern with

bromine is vapour inhalation and skin and eye contact. Several levels of protection

have been implemented to ensure that human contact with bromine will not occur

[63]. The level of risk of NFZBB compares favorably with lead acid batteries and

gasoline powered internal combustion engines.

The relatively low power density is the main drawback of the NFZBB. The

projected power density for the NFZBB is 100W/kg and the 24 V NFZBB pack, for

example, can deliver 2 kW power. Correspondingly, the internal resistance of the

NFZBB is relatively high. However, this drawback can be compensated by a novel


78

power management scheme employing a bank of high-power energy storage device,

which will be addressed in Section 4.4 and Section 4.5.

Since the NFZBB is a new battery technology and there is so far only one

publication introducing this technology, the details of the NFZBB are given in

Appendix A (courtesy of the author of [63]).

4.4 The Load-Leveling Concept and High-Power Energy

Storage Devices in EVs

4.4.1 The Load-leveling Concept in EVs

As stated earlier, of all the sub-systems used in EVs, the battery remains the main

barrier to success. The challenge is that design choices for energy density, power

density, lifetime, weight, volume, and cost of the battery are all compromises. The

three leading performance criteria of a battery, energy, power and life are

inextricably linked within an “eternal triangle” [64]. Improvement in any one comes

at the expense of one or both of the others. For example, the common way to

improve the power performance of a battery is to use thinner electrodes, but this

usually lowers both the energy density and the life expectancy. Cost is another major

concern. The lead acid battery is the most commonly used battery in EVs since it is

cheaper than alternative technologies. However its energy density is low. Using

advanced batteries such as NiMH and NiCd can improve EV range performance.

Unfortunately they are very expensive.

For a high-performance EV, both the energy requirement for drive range and the

peak power requirement for acceleration, overtaking and deceleration must be met.

The ratio of the required peak power to average power can be many times. Generally

speaking, in a battery, peaks of power imply increase of losses and temperature and

so decrease of lifetime [65].

Employing high-power energy storage devices in electric vehicle drivelines to

load level the battery can significantly reduce the peak power requirement for the

battery [19, 66-70]. The high-power energy storage devices supply the peak power

during acceleration while the batteries supply the continuous power for normal


79

driving. This offers opportunities to design and use EV batteries that are optimized

for energy density, life and low cost with less attention being given to the peak

power.

At this time, advanced high-power energy storage technologies with high

potential include high-power batteries, ultracapacitors, and flywheels [71], which are

briefly described below.

4.4.2 High-Power Batteries

Among Lead-acid, NiCd and NiMH technologies, lead-acid batteries have the highest

power density (or specific power) as shown in Figure 4.7 previously. In the future, it

is anticipated that bipolar lead-acid batteries will have twice the specific power of

both NiCd and NiMH batteries [6].

For power assist purposes, batteries have to be frequently deep-cycled.

Unfortunately, lead-acid batteries have historically had a relatively short life when

deep-cycled. Bipolar lead-acid batteries have potential of 4 to 5 times longer life than

today’s lead-acid batteries [6] and then might become suitable for power assist

purpose in EVs and HEVs, in this respect. However, recharging time of any batteries

may be too long for the power assist purpose in EVs.

4.4.3 Flywheels

Flywheels are being considered as a way to mechanically store kinetic energy in EVs.

Once set in motion, a flywheel will continue to rotate unless energy is added to speed

it up or removed to slow it down. Modern flywheels employ a high-strength

composite rotor that rotates in a vacuum chamber to minimize aerodynamic losses. A

motor/generator is mounted on the rotor’s shaft both to spin the rotor up to speed

(charging) and to convert the rotor’s kinetic energy to electrical energy (discharging).

A high-strength containment structure houses the rotating elements and low-energy

loss bearings stabilize the shaft. Interface electronics are needed to convert the

alternating current to direct current, condition the power, and monitor and control the

flywheel. Flywheels potentially have excellent high power density and long service

life compared with batteries. They are being used in standby power supplies [72],


80

spacecraft energy store systems [21], and electric vehicle propulsion [73]. A flywheel

is sometimes referred to as an “electromechanical battery”.

While optimistic estimates for specific power of flywheels range between

2000W/kg (short-term) and 8000W/kg (long-term), current prototype flywheels have

much lower specific energy and power density [6].

Although the principle behind employing a flywheel for EV propulsion is fairly

straightforward, building a flywheel system for EV application has been challenging.

Flywheels in EV applications are still in development stage. Efforts must be made to

increase their specific energy and specific power, reduce material costs, design

lightweight containment, and develop simplified system integration before they can

be commercially viable for EVs.

4.4.4 Ultracapacitors

Ultracapacitors are higher specific energy versions of electrolytic capacitors, devices

that store energy as an electrostatic charge. Table 4.3 shows the specifications of

Maxwell’s ultracapacitor PC7223.

The most significant feature of the ultracapacitor is its high power density. This

feature makes ultracapacitors very much suitable for applications needing repeated

bursts of power for short time such as power assist during accelerations and hill

climbing, as well as high-efficiency recovery of kinetic energy during braking.

Table 4.3 Specifications of Maxwell ultracapacitor PC7223

Capacitance 2700 FStored Energy 7142 JMaximum Voltage 2.3 VRated current 400 ADC Series resistance 0.85 mohmPeak power @ 2.3V 3055W/kgEnergy Efficiency 98% @ 60 seconds,

83% @ 6 secondsCycle life Hundreds of thousands timesWeight 800 kg


81

It has been well documented that high regenerative pulse current shorten battery

lives for most types of batteries [74]. With the help of the ultracapacitors to absorb

the regenerative current, the battery life can be extended.

Another key aspect of ultracapacitors is that they have excellent cycle life. It is

possible to cycle ultracapacitors very quickly and deeply without seeing large

decreases in cycle life that most chemical batteries experience. They are expected to

last as long as the vehicle. They also have high cycle efficiency compared to

chemical batteries.

4.4.5 Summary

At this stage, technically, ultracapacitors are more practical than flywheels and high-

power batteries to be used as auxiliary energy / power storage for load-leveling

purpose in EVs. Price of ultracapacitors is high for the time being but should drop as

the production volume increases. The specifications of ultracapacitors such as

specific energy and ESR are also being further improved [67, 71, 75]. High-power

batteries are able to provide high pulse power at reasonable cost, however the short

cycle life and long charging time are serious limitations. Ideally flywheels are

suitable for load-leveling purposes in EVs, but they are still some distance away from

commercial viability for EVs.

4.5 A Novel EV Power Management Scheme Employing

Zinc-Bromine Batteries and Ultracapacitors

The NTCER EV will employ the NFZBB, which features high energy density, long

lifetime, low cost and recyclability, as prime mover to provide continuous power for

cruising, and a secondary power storage--a high-power ultracapacitor bank to load

level the NFZBB, processing peak power during acceleration, climbing and

regenerative braking. This novel power management scheme fully takes advantages

of the NFZBB and the ultracapacitor, providing a good example of the

implementation of the load leveling in EV applications. In other words, by employing

the high-power energy storage devices to load level the NFZBB battery in the

NTCER EV, the battery energy density and peak power density requirements are


82

decoupled. Employing the NFZBB with peak power assist using ultracapacitors in an

EV has not been reported by other authors.

Ultracapacitors have the feature that their voltage is directly proportional to their

state-of-charge. The main problem of connecting ultracapacitors directly to the

batteries is that only a small part of the ultracapacitor capacity is utilized, limited to a

narrow, high state-of-charge region. Besides, a high voltage ultracapacitor bank (at

the battery voltage) is usually expensive and /or oversized.

A power electronics interface is needed to fully utilize the energy storage and provide

adequate control to make the operation of the ultracapacitor bank effective. Several

different ways to insert an ultracapacitor bank into drive trains [65, 70, 76-78] have

been reported. A typical scheme is shown in Figure 4.7, which employs a

bidirectional dc-dc converter between the ultracapacitor bank and the batteries. In

this scheme, the energy flow in the ultracapacitor bank can be controlled

independently of battery voltage and power. With the battery directly connected to

the bus, the dc-dc converter only processes peak power for a short time, minimizing

the energy efficiency degradation. With the ultracapacitor connected to the lower-

voltage side, this scheme does not require a high voltage ultracapacitor bank that is

usually expensive. Using a dc-dc converter gives flexibility in choosing the power,

energy and voltage of the ultracapacitor with affordable cost, and allows full control

of the output power of the ultracapacitor.

N F Z B B P W MInverter

B id irec tiona lD C -D C

C onverter

U CM

Figure 4.7 Connection configuration of the ultracapacitor bankin the propulsion system for load leveling purpose


83

Using six 24V NFZBB packs that can store a total energy of 6 kWh should allow

60 km drive range, but can provide only 12 kW peak power (corresponding to 90 A

peak current). That is not high enough for good acceleration performance of the

vehicle. The extra power required, about 10 kW, will be provided by the

ultracapacitor bank.

4.6 Simulation and Optimization of the EV Power System

4.6.1 The Simulation Model Using Matlab / Simulink

One of the keys to successfully employing ultracapacitors to load level EV batteries

is to correctly determine the power ratings of the ultracapacitor bank and the dc-dc

converter. The efficiency of the dc-dc converter will vary substantially with the

operating points in the calculations, making calculation challenging. Due to the

switch-mode operation of the dc-dc converter and variations in circuit parameters, it

is more accurate and convenient to use Matlab/Simulink simulation to look at the

operation of the power system. By simulation, the bus voltage, maximum output

current of the dc-dc converter at different ultracapacitor voltages, discharging

currents and the inverter currents can be obtained. Output power and efficiency of the

dc-dc converter and the whole power system can then be calculated. Based on these

data, the sizing of the ultracapacitor bank and the dc-dc converter power rating can be

determined.

Matlab / Simulink models were developed for these tasks. Figure 4.8 shows the

model of the dc-dc converter in boost mode. According to [79], the ultracapacitor

(UC) can be modeled as an ideal capacitor in series with a resistor that has the value

of the cell ESR. For discharges longer than 2 seconds this model is accurate enough

[80]. For the MOSFET, the on-resistance, lead inductance and a resistor-capacitor

(RC) snubber are included. For the diode, the forward voltage drop, the lead

inductance and an RC snubber are considered. The boost inductor is modeled as an

ideal inductor with a resistor in series. The value of the series resistor is determined

by both the calculation and measurement. The NFZBB is modeled as an ideal voltage

source in series with an internal resistor.


84

G a te B o o st

b u s

+-

v

V b u s

B u sL in k

T o Wo rksp a ce 2

In d Cu rre n t

T o Wo rksp a ce 1

t

T o Wo rksp a ce

S te p Re sp o n se

Sawwav es

S a wwa veRu c

>

Re la ti o n a l

O p e ra to r

Rb a tt1 5 .4 (s+5 9 4 )

s

P I Co m p

M u x

M u x4M u x

M u x2

M u x

M u x1

L i m i t1

1 5 0 AL 1

i+

-

In ve rte r

1 5 0

I_ u c1 _ cm d

1 2 0

I_ lo a d3 .2 e 4

s+3 .2 e 4

Fi l te r2

1 .3 4 e 4

s+1 .3 4 e 4

Fi l te r1

Clo ck

Cb u s

B u sl i n k2

B u sL in k

1 g

2 m

B o o st S wi tch

ak

m

B o o st D io d e

1 4 4 V

B a tt UC

+i-

I_ u c2

+i-

I_ u c1

+i-

I_ b a tt

+i-

2

+i-

1

+i -

Figure 4.8 Matlab / Simulink model of the dc-dc converter in boost mode

It is believed that the whole simulation model is accurate enough for the

purposes of sizing the ultracapacitor and the dc-dc converter, and providing design

guidelines for the propulsion power system.

4.6.2 The Simulation Results and Determination of the Power

Rating of the Ultracapacitor Bank and the DC-DC Converter

Simulations were done at various operating points:

1. With different numbers of ultracapacitors;

2. At different ultracapacitor voltages or state-of-charge (SOC), i.e., the full voltage,

90%, 80%, 70% and 60% of the full voltage;

3. At different load currents (drawn by the inverter);

4. At different ultracapacitor charging currents.

A numbers of voltages and currents were recorded, including:

Vuc – the ultracapacitor voltage;

Vbus – the bus voltage (the battery terminal voltage);


85

Iload – the current drawn by the inverter;

Iuc1 – the discharging current of the ultracapacitors;

Iboost – the output current of the boost converter.

Based on the measurements above, the following were calculated:

INFZBB – the battery current;

Puc – the power out of the ultracapacitors assuming 100% efficiency;

Pboost1 – the power out of the ultracapacitors or the input power of the boost

converter;

Pboost2 – the output power of the boost converter;

Effi_boost – the efficiency of the boost converter;

Effi_boost_uc – the efficiency of the short-term power unit (STPU) including the dc-dc

(boost) converter and ultracapacitors.

Firstly, the simulations were carried out with 20 ultracapacitors. Figure 4.9

shows the output power and efficiencies of STPU with 20 ultracapacitors in series

(46 V when fully charged), at different discharging currents. Figure 4.10 shows

similar results with 30 ultracapacitors in series (69 V when fully charged).

Several observations may be made from the results of the simulations.

1. At the discharging current rate of 150 A, the STPU with 30 or less number of

ultracapacitors is not able to provide the required 10 kW power, even when the

ultracapacitors are fully charged. While increasing the discharging current rate

can result in higher power output, the efficiency of the STPU drops significantly

as the discharging current rate increases, which is undesirable. It is desirable to

use a high voltage ultracapacitor bank to achieve high efficiency of the STPU.

2. The output power capability and efficiency of the STPU both decrease with

decreasing the ultracapacitors’ state of charge. For high output power capability

and efficiency, the voltage variation of the ultracapacitor bank should be limited

to a manageable range.


86

0

2

4

6

8

20 25 30 35 40 45 50

UC Voltage (V)

PP

AU

Out

put P

ower

(kW

)I_uc1=150 A

I_uc1=200 A

(a)

60

65

70

75

80

85

90

20 25 30 35 40 45 50

UC Voltage (V)

PP

AU

Effi

cien

cy (

%)

I_uc1=150 A

I_uc1=200 A

(b)

Figure 4.9 Output power capacity and efficiency of the STPU employing 20 ultracapacitors

Based on the above, the STPU with 40 ultracapacitors in series (92 V when fully

charged) appear to be able to function effectively, in terms of output power capability

and efficiency. Figure 4.11 shows the simulation and calculation results. It can be

seen that, with 150 A current discharging the ultracapacitors, the STPU can provide

output power ranging from 12.2 kW to 8 kW when the ultracapacitor voltage varies

from its full voltage to 70% of the full voltage, while the efficiency of the STPU

stays between 88.5% and 82%. These numbers are satisfactory, implying that this

STPU has adequate efficiency and high power output capability in a wide range of

variation of the ultracapacitor voltage.


87

0

2

4

6

8

10

12

30 40 50 60 70 80

UC Voltage (V)

PP

AU

Out

put P

ower

(kW

)I_uc1=150 A

I_uc1=200 A

(a)

60

65

70

75

80

85

90

30 40 50 60 70 80

UC Voltage (V)

PP

AU

Effi

cien

cy (

%)

I_uc1=150 A

I_uc1=200 A

(b)

Figure 4.10 Output power capacity and efficiency of the STPU employing 30ultracapacitors

It should be noted that the simulation results also indicate that the magnitude of

inverter current has little effect on the output power capability and efficiency of the

STPU when the ultracapacitors are discharged at a constant current rate. On the other

hand, the magnitude of the inverter current does affect the amount of the battery

current. The STPU should be controlled, such that the STPU would never charge the

battery, to avoid any unnecessary energy flow and loss.

It is necessary to examine the energy available and energy utilization of the

ultracapacitor bank with 40 ultracapacitors when the ultracapacitor voltage varies

from its full voltage to 70%. Based on the data for Maxwell’s PC7223 ultracapacitors

in Table 4.3, half of the stored energy of 40 PC7223 ultracapacitors is


88

02468

101214

50 60 70 80 90 100

UC Voltage (V)

PP

AU

Out

put P

ower

(kW

)

(a)

70

75

80

85

90

50 60 70 80 90 100

UC Voltage (V)

PP

AU

Effi

cien

cy (

%)

(b)

Figure 4.11 Output power capacity and efficiency of the STPUemploying 40 ultracapacitors

147.84 kW-second, allowing discharging for about 15 seconds at an average rate of

10 kW. This is adequate to meet the acceleration performance requirement.

Discharging the ultracapacitor from its full voltage to 70% corresponds to 51%

ultracapacitor energy released, indicating that a large amount of ultracapacitor energy

is controllable.

40 Maxwell’s ultracapacitors PC7223 weigh 32 kg and a 10 kW dc-dc converter

will weigh about 13 kg, giving a total weight of 45 kg. However, if only the NFZBBs

were used to achieve the required maximum power, then double the number of

NFZBB packs (12 NFZBB packs) would be needed, requiring an additional 6

NFZBB packs which weigh 120 kg. Thus the use of the STPU offers 75 kg weight

saving. It is well known that vehicle energy consumption is mass dependent. Energy


89

consumption of the NTCER EV with and without the additional 75 kg mass on the

FUDS were calculated as 1.048 kWh and 0.9795 kWh, respectively. These numbers

indicate that the 75 kg weight saving resulting from employing the STPU represents

about 7 % energy saving in the NTCER EV.

From all the above analysis, it can be concluded that the STPU employing 40

Maxwell’s ultracapacitor PC7223 with discharging current of 150 A is a good

solution. It can provide enough power and energy for good acceleration performance

with high efficiency, offer vehicle weight saving representing energy saving, and

allow good utilization of the ultracapacitors.

4.6.3 Control Rules for the DC-DC Converter for Load Leveling the

Batteries

1. To maintain high efficiency in the STPU, the ultracapacitor voltage is allowed to

vary from full voltage to 70% of full voltage, corresponding to about half of the

total stored energy to be controlled.

2. The batteries charge the ultracapacitors when the power demand of the electric

vehicle is low.

3. The dc-dc converter is activated only when the motor current command exceeds a

threshold (positive or negative) and the power out of or into the battery exceeds a

certain value. Too frequent charging and discharging between the batteries and

the ultracapacitors should be avoided, to reduce energy loss.

4. While the dc-dc converter is working, it is desirable that the ultracapacitor current

(the inductor current) is kept constant and the remainder of the current required is

provided by the batteries. This control strategy simplifies the control circuit.

5. The STPU should be controlled such that the STPU would never charge the

batteries, to avoid any unnecessary energy flow and loss. To ensure this, the

ultracapacitor current (the inductor current) command is given according to the

motor current command. Considering Control Rule 4, the inductor current

command is given in a form of step constant according to the motor current

command. For example, the inductor current is set to 110 A when the motor


90

current command is between 60 to 90 A, and 150 A when the motor current

command is between 90 to 120 A.

4.6.4 Simulation Results on the Voltage and Current Transients of

the Short Term Power Unit

In establishing the effectiveness of the STPU, simulations on Matlab/Simulink were

carried out to observe the voltage and current transients of the STPU. According to

the control rules for the dc-dc converter for load leveling the batteries in Section

4.6.3, while the dc-dc converter is working, the ultracapacitor current (the inductor

current) is kept constant (Rule 4 & 5 in Section 4.6.3). Therefore, a negative

feedback loop is closed on the inductor current, as shown in Figure 4.12. The current

source I_load represents the inverter motor combination which is current-controlled.

The inductance value is 40 �H and the switching frequency of the MOSFETs is fixed

at 20 kHz. A PI compensation network has been designed for the current control.

The bus capacitor of the inverter, Cbus, is considered as part of the STPU. I_uc2 in

Figure 4.12 is defined as the output current of the STPU in boost mode when it is

positive, and as the input current of the STPU in buck mode when it is negative.

L

Sboos t

D boos t

U C

D buck

Sbuck

V bat tV uc

Dr iver Dr iver

SbuckSboos t

Swi tch ingSteer ingCircu i t

PI_i

C bus Bat tery

I _uc1

I _ u c 1 _ c m d

I _ load

I _uc2

I _uc1

Figure 4.12 Control diagram of the bi-directional dc-dc converter for the purpose of loadleveling the batteries


91

Two types of simulation results during transient conditions in both the boost

mode and buck mode respectively are presented in the following sections. The first

type is the response of the ultracapacitor current (the inductor current) to a step input

command. The second type is the current and voltage waveforms at the bus link,

including I_uc2, I_batt, I_load and V_bus, under the same stimulus.

4.6.4.1 Voltage and Current Transients of the STPU in Boost Mode

Figure 4.13 shows simulated step response of the current loop of the STPU in boost

mode. It can be seen that the inductor current (the ultracapacitor current) follows its

command very well. It takes about only 150 �s for the inductor current to reach 150

amperes from zero, without overshoot.

The ripple current in the ultracapacitors is determined by the inductance value

and the switching frequency and is limited to 20 A, as can be seen in Figure 4.13

(b). It should be noted that Maxwell’s ultracapacitor PC7223 is rated at 400 A.

Therefore, with the ripple current well controlled, the ultracapacitors can handle 150

A average current with ease, ensuring their long lifetime.

Figure 4.14 shows the battery current, bus voltage, and output current of the

STPU in boost mode by simulation. When the inverter current is +40 A, only the

batteries provide the power and the STPU is disabled. At an inverter current demand

of +120 A, the STPU is activated, and both the STPU and the batteries provide

power to the motor / inverter. When the ultracapacitor voltage is 82.8 V, 90% of full

voltage, and the motor / inverter demand is set to 150 A, the STPU provides 89.5 A

and the batteries 39.5 A, as shown in Figure 4.14 (a). With the same motor / inverter

current setting but at 64.4 V ultracapacitor voltage (70% of full voltage), the STPU

provides less current, being 61 A, and the batteries provide more, being 59 A, as

shown in Figure 4.14 (b). In both cases, transients take about 2 ms and the variation

and ripple of the bus voltage are very small.


92

G a te B o o st

b u s

+-

v

V _ b u s

B u sL i n k

T o W o rksp a ce 2

In d Cu rre n t

T o W o rksp a ce 1

t

T o W o rksp a ce

S te p _ l o a d

S te p _ I_ u c1

S te p R e sp o n se

S awwav es

S a w wa ve

>

Re l a t i o n a l

O p e ra to r

R_ u c

R_ b a tt1 5 .4 (s+ 5 9 4 )

s

P I C o m p

M u x

M u x4M u x

M u x2

M u x

M u x1

L i m i t1

1 5 0 A L 1

i+

-

In ve rte r3 .2 e 4

s+ 3 .2 e 4

F i l te r2

1 .3 4 e 4

s+ 1 .3 4 e 4

F i l te r1

C l o ck

Cb u s

B u sl i n k2

B u sL i n k

1 g

2 m

B o o st S w i tch

ak

m

B o o st D i o d e

UC

+i-

I_ u c2

+i-

I_ u c1

+i-

I_ b a tt

B a tt

+i-

2

+i-

1

+i -

(a)

(b)

Figure 4.13 Simulink block diagram and simulated step response of the current loop of theSTPU in boost mode


93

(a)

(b)

Figure 4.14 Battery current, bus voltage and currents of the STPU in boost mode bysimulation

Step load : initial value –– 40A, final value –– 120ACurrent command in the ultracapacitors (the inductor): 150A

Trace 1: bus voltage (from top to bottom)Trace 2: current command in the ultracapacitors

Trace 3: output current of the STPUTrace 4: battery current

(a) Ultracapacitor voltage is 82.8V, 90% of full voltage(b) Ultracapacitor voltage is 64.4V, 70% of full voltage

1

2

3

4

2

1

3

4


94

4.6.4.2 Voltage and Current Transients of the STPU in Buck Mode

The voltage and current transients of the STPU in buck mode are similar to those in

boost mode, except that all the currents in buck mode are in the opposite directions

and become negative in value.

G a te B u ck

b u s

+-

v

V b u s

B u sL in k

T o Wo rksp a ce 2

In d Cu rre n t

T o Wo rksp a ce 1

t

T o Wo rksp a ce

S te p _ lo a d

S te p _ I_ u c1

S te p Re sp o n se

Sawwav es

S a wwa veRu c

>

Re la ti o n a l

O p e ra to r

Rb a tt1 5 .4 (s+5 9 4 )

s

P I Co m pNO T

No t

M u x

M u x4M u x

M u x2

M u x

M u x1

L im i t1

1 5 0 A L 1

i+

-

In ve rte r3 .2 e 4

s+3 .2 e 4

Fi l te r2

1 .3 4 e 4

s+1 .3 4 e 4

Fi l te r1

Clo ck

Cb u s

B u sl i n k2

B u sL in k

1

g

2

m

B u ck S wi tch

a

k m

B u ck Dio d e

UC

+i-

I_ u c2

+i-

I_ u c1

+i-

I_ b a tt

B a tt

+i-

2

+i-

1

+i -

(a)

(b)

Figure 4.15 Simulink block diagram and simulated step response of the current loop of theSTPU in buck mode


95

(a)

(b)

Figure 4.16 Battery current, bus voltage and currents of the STPU in buck mode bysimulation

Regenerative current : -120ACurrent command in the ultracapacitors (the inductor): -150A

Trace 1: bus voltage (from top to bottom)Trace 2: battery current

Trace 3: current taken by the STPUTrace 4: current command in the ultracapacitors

(a) Ultracapacitor voltage is 64.4V, 70% of full voltage(b) Ultracapacitor voltage is 82.8V, 90% of full voltage

2

2

1

3

4

1

3

4


96

The excellent step response of the current loop of the STPU in buck mode can be

seen in Figure 4.15. Figure 4.16 shows the battery current, bus voltage, and input

current of the STPU in buck mode. When the ultracapacitor voltage is 64.4 V, 70%

of full voltage, and the motor / inverter current is set to -150 A, the STPU absorbs 73

A of 120 A regenerative current and the batteries take the rest (47 A), as shown in

Figure 4.16 (a). With the same motor / inverter current setting, but at 82.8 V

ultracapacitor voltage (90% of full voltage), the STPU absorbs more regenerative

current, 95 A, and the batteries take 25 A, as shown in Figure 4.16 (b). In both cases,

transients take about 2 ms and the bus voltage is very smooth.

The simulation results on voltage and current transients of the STPU in both

boost mode and buck mode have shown the excellent dynamic performance of the

STPU. These, along with the high efficiency the STPU presented in Section 4.6.2,

have demonstrated the effectiveness of the STPU.

4.7 Overall Power Control in the NTCER EV

In Section 4.5 – 4.6, a short-term power assist scheme, which employs a high-power

ultracapacitor bank to load level the NFZBB, where the NFZBB is designed and

optimized for high energy, high life cycle, and low cost was presented. The STPU

operates when the motor speeds are below the base speed, which is proportional to

battery voltage. At motor speeds above the base speed, a speed extension and

constant power operation control scheme was discussed in Section 3.4. The overall

power control scheme in the NTCER EV for all speeds is illustrated in Figure 4.17.

At very low speeds, all the power comes from the battery. High torque can be

obtained, since the required power, which is proportional to motor speed, is within

the output power capability of the battery. Above a threshold motor speed, n1, the

STPU will be activated to provide additional power when high acceleration is

required. While the shaded triangle shows the extra power required for high


97

M ax im u m T orq u e

S p eednb nm0

C on stan tP ow er

P ow er

P ow er P rov id edb y U C

E xtraP ow er

R eq u ired

n2n1

Figure 4.17 The concept of the overall power management

acceleration, the STPU actually provides power shown in the hatched rectangle and

the battery provides the remainder of the power, in order to simplify the control. The

operation of STPU is disabled at speed n2, due to the upper power limit of the STPU.

At motor speeds above the base speed, nb, the battery voltage is boosted to higher

voltage for speed extension and constant power operation of the motor.

In the whole speed range, from zero to nm, the maximum torque envelope is

shown as a dashed line in Figure 4.17. This maximum torque – speed envelope is

similar to the conventional torque-speed characteristic of an internal combustion

engine with a transmission. That is, the motor provides high torque at low speeds and

lower torque at high speeds, which is familiar to ICE vehicle drivers.

It should be noted that in Figure 4.17, nb could be extended up to nm, determined

by the power ratings of the battery and the STPU, and device ratings.

4.8 A Novel Multifunctional DC-DC Converter Topology

for the Overall Power Management in the NTCER EV

4.8.1 Topology Derivation

As can be seen from Section 4.5, the EV battery load-leveling scheme employing an

ultracapacitor bank requires a bidirectional dc-dc converter with the ultracapacitors


98

connected to its lower-voltage side and the battery to its higher-voltage side. On the

other hand, the constant power operation of the brushless dc motor at speeds above

the base speed, as discussed in Chapter 3, requires another bidirectional dc-dc

converter with the battery connected to the lower-voltage side and the motor/inverter

to the higher-voltage side. If two separate bidirectional dc-dc converters are

employed and connected in cascade, as shown in Figure 4.18, the whole circuit will

be very complicated and expensive. Furthermore, the conduction losses will be

excessive and then an efficiency penalty will be imposed since there are too many

loss-causing components in the current paths. A few other dc-dc converter

configurations for EV propulsion were reported in the literature [69, 70, 81], but they

addressed either a fewer number of functions or some different functions.

It is desirable to use only one bidirectional dc-dc converter for all these

functions. This requires three switches to disconnect the converter and turn the

converter around for boosting the battery voltage. Using three semiconductor

switches for this purpose is one possibility, however this will make the circuit more

complicated since these switches must be bidirectionally controllable. Furthermore,

semiconductor switches will cause considerable conduction loss in this high current

application.

A novel multifunctional dc-dc converter has been proposed for the overall

energy / power management. Figure 4.19 shows its power stage schematic. It consists

U C

L2

N FZB B Sb o o st2

Sb u c k 2L1

Sb o o st1

Sb u c k 1

D C -D C C onverter 1 D C -D C C onverter 2

C PW MInverter M

Figure 4.18 Two-stage scheme for the overall power managementusing two separate dc-dc converters in cascade


99

LC on tac to r

AB

C

S 1 S 3

S 4S 2

S 5

S 6

1

2

M

In verter/m oto r

B a tt

U C

Sbo o st

Sbu ck

D sch o t.

S C RC

Figure 4.19 The proposed multifunctional dc-dc converter topologyfor the overall power management

of a simple bidirectional dc-dc converter, a Schottky diode Dschot, an SCR and a

changeover contactor. To achieve a simple layout, a half-bridge semiconductor

switch module can implement Sbuck and Sboost. The diode, the SCR and the

changeover contactor are used to move the battery from the higher-voltage side to the

lower-voltage side of the converter for constant power operation of the brushless dc

motor. Dschot and the SCR form a very cost-effective and efficient bidirectional

switch in this particular application. The use of a magnetic contactor here is practical

since the contactor switches at very low frequencies, only when the operation mode

needs to be changed. It offers two significant advantages over semiconductor devices.

Firstly, it is a bidirectional device with very simple control. Secondly, it causes little

conduction loss. It is obvious that the proposed multifunctional dc-dc converter is

superior to the two-dc-dc-converter-cascaded scheme in circuitry simplicity, cost, and

efficiency.

It is very important to note that the contactor is always closed or opened under

zero-current-switching (ZCS) condition. This is necessary; otherwise the necessity of

arc extinguishing will make the contactor oversized and very expensive. The ZCS

condition can be ensured by proper control of the dc-dc converter, as will be

discussed in Section 4.8.2.3.

It should also be noted that a large electrolyte capacitor inside the inverter

decouples the operation of the dc-dc converter and the inverter/motor. Therefore,

analysis of the complete drive system can be performed through analysing each


100

power stage separately. The following section illustrates the operation principles of

the proposed multifunctional dc-dc converter. Operation and control issues of the

three-phase inverter / motor (BDCM) have been discussed in Chapter 3. In brief, the

BDCM for the EV propulsion is current-controlled or torque-controlled. While vector

control or field-oriented control (FOC) is widely used in induction motors [47, 117-

119], it is not used in BDCM control. Using field-oriented control, a highly coupled,

nonlinear, multivariable induction motor can be simply controlled through linear

independent decoupled control of torque and flux, similar to separately excited dc

motors. Field-oriented control also sees its application in some permanent magnet

synchronous motors which possess sinusoidal back EMF waveforms and relatively

large armature reaction [47, 120]. However, as armature reaction in BDCMs is little

and the back EMF waveforms and phase current waveforms are trapezoidal / square,

excellent performance of torque control can be easily achieved through simple

current control based on rotor position information. This is one of distinguished

features of BDCMs.

4.8.2 Operation Principles and Topology Analysis

4.8.2.1 At Motor Speeds below the Base Speed

At motor speed below the base speed, the contactor is in Position 2 and the dc-dc

converter works for load leveling. During normal driving, Figure 4.20 (a), only the

battery provides the inverter/motor power through a low forward voltage Schottky

diode, Dschot. and the dc-dc converter is disabled

During acceleration, the dc-dc converter is working in boost mode and the

ultracapacitors provide power within the maximum limit of the dc-dc converter. The

battery provides the rest of the power beyond the capacity of the dc-dc converter.

This mode is illustrated in Figure 4.20 (b).


101

MP W MInverter

L 2

Batt .

Sboos t

Sbuck

1

2

U C

Contactor

C 1

S C RD schot .

(a)

MP W MInver ter

L 2

Bat t .

Sboost

Sbuck

1

2

U C

Contac to r

C 1

S C RD schot.

(b)

Figure 4.20 Operation mode at motor speeds below the base speed.(a) During normal driving, the dc-dc converter is disabled.

(b) During acceleration, the dc-dc converter is working in boost mode. Both thebattery and ultracapacitors provide power to the inverter/motor.

Figure 4.21 shows the sub-topology during regenerative braking. The inverter is

working in boost-regeneration mode and the dc-dc converter is working in buck

mode with a maximum current limit in the inductor. The SCR is triggered by the

regeneration command when the contactor is in Position 2 to allow the battery to take

the rest of the kinetic energy recovered from regenerative braking. It is desirable to

repetitively trigger the thyristor to ensure its conduction any time during

regeneration.

During deceleration at very low speeds, for example, below 15 km/h, the

mechanical braking will dominate.


102

MPW MIn verter

L

B att.

Sbo o st

Sbu ck

1

2

U C

C on tac to r

C1

D scho t. SC R

Figure 4.21 Regeneration mode at motor speeds below the base speed. The dc-dc converteris working in buck mode and the inverter is working in boost-regeneration mode.

MPW MIn verter

L

B a tt.

Sbo o st

Sbu ck

1

2

U C

C on tacto r

C 1

D scho t. SC R

Figure 4.22 Battery charging the ultracapacitors. The dc-dc converter is workingin buck mode

The kinetic energy recovered from regenerative braking may not be enough to

fully charge the ultracapacitors. When the power demand of the electric vehicle is

low, for example, when idling or driving at very low speed, the batteries charge the

ultracapacitors. This operation mode is shown in Figure 4.22.

4.8.2.2 At Motor Speeds above the Base Speed

When the motor speed increases close to the base speed, the controller will switch the

contactor to Position 1 and keep the SCR off. Figure 4.23 shows the sub-topology in

motoring mode. The dc-dc converter is working in boost mode, making the bus

voltage higher than the battery voltage to achieve speed extension and constant

power operation of the motor. As mentioned earlier, a Schottky diode is used for

Dschot to minimize the conduction loss. This is practical since the reverse voltage

across Dschot is only the difference of the maximum bus voltage and the battery


103

L

B att.

Sbu ck

1

2

C ontacto r

C 1Sbo o st

D scho t.MPW M

InverterSC RU C

Figure 4.23 Constant power operation (in motoring mode) when motor speeds are abovethe base speed. The dc-dc converter is working in boost mode.

L

B att.

Sbu ck

1

2

C on tacto r

C1

Sbo o st MPW M

In verterC o m m u tat io n

o nly

D scho t.

SC RU C

Figure 4.24 Buck-regeneration operation when motor speeds are much higher thanthe base speed. The dc-dc converter is working in buck mode.

L

B att.

Sbuck

1

2

C on tac to r

C 1Sbo o st

D scho t.

M

P W MIn verterin H ybrid

B o o st-R egeneratio n

M o de

S C RU C

Figure 4.25 Hybrid boost-regeneration mode when motor speeds are marginally higherthan the base speed. Sbuck in the dc-dc converter is kept on constantly.

voltage, less than one hundred volts in the NTCER EV. When this constant-power-

boost operation is needed, Dschot has negative bias and turns off automatically.

Regulation of the regeneration current when the motor line-line back EMF is

much higher than the battery voltage is shown in Figure 4.24. The inverter is

controlled for commutation only, similar to a three-phase rectifier, and the dc-dc


104

converter in this case works as a buck converter to regulate the regeneration current.

This operation mode is called buck-regeneration mode.

To obtain a high, controlled regeneration current when the motor speed is only

marginally higher than the base speed, PWM switching operation of the inverter is

activated and Sbuck in the dc-dc converter is kept on constantly, as illustrated in

Figure 4.25. In this mode, operation of the inverter is in hybrid boost-regeneration

mode. The details of the operation of a BDCM in buck-regeneration mode and hybrid

boost-regeneration mode have been discussed in Section 3.6.

In the case of very gentle braking, the regenerative braking energy is put back

into the battery. In the case of hard braking, mechanical brakes need to be added.

4.8.2.3. Zero Current Switching (ZCS) Conditions of the Contactor

The changeover contactor must be closed or opened under ZCS condition to avoid

the necessity of arc extinguishing that would make the contactor much larger and

more expensive. Properly controlling the dc-dc converter can achieve the ZCS

conditions. The following control steps can ensure the ZCS conditions of the

contactor.

Transition from Position 2 to Position 1:

(1) Disable Sbuck and Sboost drives;

(2) Switch the contactor from Position 2 to Position 1;

Transition from Position 1 to Position 2:

(1) Activate the trigger signal of the SCR;

(2) Disable Sbuck and Sboost drives;

(3) Switch the contactor from Position 1 to Position 2;

Each transition is expected to take about 15 milliseconds, mainly determined by

the contact time of the changeover contactor. 15 milliseconds later it is safe to enable

Sbuck and Sboost drives for appropriate control.

4.9 Review

In this chapter, the power and energy requirements for the NTCER EV propulsion

system are analyzed. After an overview of EV batteries, it is concluded that none of


105

the currently available EV batteries can ideally meet EV requirements in terms of

power density, energy density, lifecycle and cost. Accordingly, a novel energy and

power management scheme has been proposed. The proposed energy and power

management scheme employs the newly developed non-flowing zinc-Bromine

battery (NFZBB) and ultracapacitors (UCs). NFZBBs are optimized for specific

energy, low cost and long life, and ultracapacitors feature high power density and

long life and are used to complement the NFZBB to achieve good acceleration

performance at low speeds. Combining the two different types of energy storage can

optimize the EV propulsion system to achieve high energy for drive range, high

power for good acceleration performance, long lifetime, and low cost.

The overall power control characteristic of the EV propulsion system, resulting

from the combination of the constant power control scheme above the motor base

speed, seen in Chapter 3, and the short-term power assist scheme is presented. This

maximum torque – speed envelope is similar to the conventional torque-speed

characteristic of an internal combustion engine plus transmission. That is, the

propulsion system provides high torque at low speeds and lower torque at high

speeds with which human beings are familiar.


control scheme at speeds above the base speed each require a bidirectional dc-dc

converter. Instead of using two separate dc-dc converters which will add considerable

complexity and cost to the propulsion system, a novel multifunctional dc-dc

converter has been proposed. It adds only three components to a simple conventional

bidirectional dc-dc converter and can implement all the functions required by the

overall energy/power management scheme. The principle of operation and control

rules of the multifunctional dc-dc converter are presented in detail.

106

Chapter 5

Development of the Soft-SwitchedBidirectional DC-DC Converter

5.1 Introduction

Switch-mode power conversion requires high frequency PWM operation to achieve

small size, light weight, and excellent dynamic performance of converters (including

inverters). PWM converters operating at higher switching frequency can potentially

offer better dynamic performance, as a minimum control delay and bandwidth of a

control loop are determined by switching frequencies.

Conventional hard-switching PWM converters process power by interrupting the

power flow by switching power devices abruptly. Hard switching results in

significant switching losses, electromagnetic interference (EMI) noise and switching

stresses, especially at high frequencies.

At high switching frequencies, switching losses of power devices and EMI

become serious issues. High switching losses reduce power conversion efficiency and

result in the need for larger and heavier heatsinks. Severe EMI may not only cause

stability problems in the converters and even damages of the power devices, but also

interfere with the normal operation of nearby electric/electronic equipment or

devices. To comply with international electromagnetic compatibility (EMC)

standards [82, 83], EMI must be limited to a certain level. Although some of the

CHAPTER 5 DEVELOPMENT OF THE SOFT-SWITCHED BIDIRECTIONALDC-DC CONVERTER

107

problems related to hard-switching techniques can be solved by using dissipative

snubbers and filters, it is often difficult to find a good compromise.

Soft-switching technologies have been good solutions for the problems

associated with high frequency operation of PWM power converters. In a soft-

switched converter, in general, power switches are commutated at zero-voltage

switching (ZVS) or zero-current switching (ZCS) condition or both zero-voltage and

zero-current switching (ZVZCS). Thus switching loss can be eliminated or in reality

minimized. Soft switching allows a reduction in heatsink size or an increase of power

density. EMI noise can be significantly reduced by using soft switching. Reliability of

the converter is also improved as a result of improved efficiency and reduced EMI

noise. It should be noted that for most soft switching techniques, the benefits of soft

switching are achieved at the cost of extra components and control complexity.

In presenting the development of the multifunctional dc-dc converter for the

NTCER EV propulsion system, this chapter highlights two main topics. Firstly, it

details the soft switching technology employed for the bidirectional dc-dc converter,

including the operational principle, simulation results, implementation issues, and

experimental results. This follows the evaluation of various soft switching

techniques. Secondly, it presents a novel design of a high power ferrite inductor with

more accurate design procedures, lower localization of core saturation, better

utilization of the magnetic cores and lower EMI levels, compared to conventional

high power ferrite inductor design.

The multifunctional dc-dc converter is basically a bidirectional dc-dc converter,

operating in boost mode for speed extension and acceleration assist, and in buck

mode for charging the ultracapacitors and the control of regenerative current. In this

chapter, the terms, multifunctional and bidirectional are interchangeable.

5.2 Evaluation of Soft Switching Techniques

Suitable soft-switching techniques and a suitable topology are highly desirable for the

bidirectional dc-dc converter to achieve efficiency improvement, size and weight

reduction, and EMI minimization. Although various soft-switching methods and

topologies have been proposed in the literature recently, many of them have


108

application limitations. It is a very important and practical issue to determine

appropriate soft-switching techniques to meet specific application requirements. In

determining which kind of soft-switching technique is most desirable, the following

system parameters should be taken into account:

� Voltage and current stresses on power devices imposed by the soft-switching

techniques;

� Extra loss caused by the auxiliary unit for soft switching;

� Complexity of the circuit;

� Cost of the soft-switching technique;

� Effective soft-switching range.

Other considerations should include the suitable applications of the soft-

switching technique and its limitations, and the overall benefit vs. the cost introduced

by the soft-switching technique.

Among various soft-switching techniques and topologies for dc-dc converters,

three attractive soft-switching techniques deserve close examination. They are:

1. Lossless passive soft switching methods for PWM converters;

2. Actively Clamped Resonant DC Link Converter (ACRDCLC);

3. Zero-Voltage Transition (ZVT) PWM Technique.

Lossless passive soft switching methods for PWM converters can provide soft

turn-on and turn-off conditions to the power switches [84-87]. A distinguishing

feature of the lossless passive soft switching methods is that they do not employ any

additional active components. They use resonant inductors, capacitors, and diodes to

provide soft turn-on and turn-off of the switches. Therefore, lossless passive soft

switching methods do not need additional control for soft switching and therefore the

reliability of the converters is inherently high. Unfortunately, existing lossless

passive soft switching methods are only suitable for unidirectional operation of

converters. It appears to be very difficult, if not impossible, to employ this kind of

technology for bidirectional converters, such as in this application.


109

Among various resonant-type soft switching techniques, the Actively Clamped

Resonant DC Link converter invented by Divan [88-90] possesses unique features.

An ACRDCLC employs a LC resonant tank to achieve zero voltage condition for all

power switches connected to the dc bus, and an active switch for the purpose of

voltage clamping. The simple topology makes this active clamped resonant technique

especially attractive to a three-phase inverter, which usually employs six power

switches.

One common property of all resonant-type converters, including ACRDCLCs,

and lossless passive soft switching PWM converters, is that they all employ a

resonant inductor in series with the power switch or the rectifier diode to shape

switch voltage or current waveforms. Soft switching is achieved by utilizing

resonance between this resonant inductor and certain resonant capacitors, which are

usually in parallel with the semiconductor devices. However, since resonant elements

are placed in the main power path, the resultant resonant converters are always

subjected to some inherent drawbacks. Firstly, since all the power flows through the

resonant inductor, high circulating energy is always flowing, which results in a

substantial increase in conduction losses. Secondly, the resonance inevitably

generates additional voltage stress on the power devices, although some voltage

clamping measures can be taken to limit the voltage stress. Besides, most resonant-

type converters are unable to maintain soft switching for a wide input voltage and

load range. This is because the energy stored in the resonant inductor depends

strongly on the input voltage and load current and consequently the soft switching

condition is sensitive to the changes of the input voltage and load current.

Instead of using a series resonant tank, an alternative way to achieve soft

switching is to use a shunt resonant network across the power switch so that the

resonant elements are removed from the main power path, alleviating the above-

mentioned limitations. Zero voltage transition (ZVT) and zero current transition

(ZCT) soft switching PWM techniques adopt such a concept [91-96]. ZVT and ZCT

soft switching techniques realize soft switching in a PWM converter without

imposing additional switching voltage and current stress and conduction loss.

Because the auxiliary ZVT circuits are not in the main power path and are actively


110

controlled, the capability of bidirectional operation of the basic dc-dc converter

remains.

Based on the evaluation of various soft-switching techniques as above, it is

concluded that ZVT techniques are most favorable for the bidirectional dc-dc

converter, and thus have been adopted.

5.3 ZVT Operations of the Bidirectional DC-DC

Converter

Figure 5.1 shows the circuit diagram of the ZVT bidirectional dc-dc converter.

Symbols with lower case subscript ‘x’ represent the additional components for ZVT

boost operation, and symbols with lower case subscript ‘y’ represent the additional

components for ZVT buck operation, while symbols with lower case subscript ‘xy’

represent the additional common components for both ZVT boost operation and ZVT

buck operation.

L

S1

V in

D xSx

D 2

D x2 D x3L rx

Crx

CrxyD 1

V o

A

CB

D x1

SyD y2

L ry DD y1

S2

D y

Figure 5.1 Diagram of the ZVT bidirectional dc-dc converter

5.3.1 ZVT Operation in Boost Mode

As fully-controlled active power devices are employed in the auxiliary circuits, the

operation of the converter in boost mode and buck mode are independent. To

simplify the analysis of the circuit, the equivalent circuit in boost mode is drawn in

Figure 5.2. It can be seen that the ZVT boost dc-dc PWM converter differs from a


111

L

S1

D xSx

D 2

D x2 D x3L rx

C rx

Crxy

D 1

A

CB

D x1V lo w V hig h

(V o)

Figure 5.2 Equivalent circuit in boost mode

conventional boost PWM converter by possessing an additional shunt resonant

network, or snubber circuit. The shunt resonant network consists of a resonant

inductor Lrx, an auxiliary switch Sx, two diodes Dx1 and Dx2, a flying capacitor Crx,

and an resonant capacitor Crxy. D2 is actually the body diode of S2. The operation

principle of the ZVT boost converter can be illustrated using the idealized current

and voltage waveforms for one switching period under the following assumptions:

� All snubber diodes have negligible reverse recovery current and forward voltage

drop;

� The MOSFETs (not including their body diodes and output capacitors) are very

fast and have negligible on-resistance;

� The inductance of the main inductor, L, is large enough to keep the inductor

current almost constant during transition intervals;

� The output voltage remains constant.

The assumptions above are very close to the real conditions. However the

notorious reverse recovery characteristics of the slow body diodes of the main

MOSFETs will be considered in the following analysis in order to yield more

practical analytical results.

In one switching period, the auxiliary switch is turned on for a short time before

the turn on of the main switch. There is a small overlap of the turn-on time of the

main switch and that of the auxiliary switch to ensure reliability of operation. One


112

switching period can be divided into eight intervals. Figure 5.3 shows the idealized

current and voltage waveforms for one switching period and Figure 5.4 shows the

equivalent circuits at different intervals of one switching period.

1. Interval t0 - t1

Before time t0, both the MOSFETs are off, and the diode D2 is on, delivering the

power from the input to the output. At time t0, the auxiliary MOSFET, Sx, is turned

V gs1

V gsx

i S x

i L rx

i D 2

i S 1

V ds1

(V A )

V dsx

i D x3

V C

V D 2

t0 t1 t2 t3 t4 t5 t6 t7

i L

i L

t8

i L

V o

V o

i D 2r

V o

V o

i L

i L rxm

i L rxm

Figure 5.3 Key circuit waveforms in boost mode


113

S1

i L

D xSx

D 2

D x2 D x3L rx

C rx

C rxy

D 1

V o

A

CB

D x1

i D 2i L rx

S1

i L

D xSx

D 2

D x2 D x3L rx

C rx

C rxy

D 1

V o

A

CB

D x1

i D 2ri L rx

S1

i L

D xSx

D 2

D x2 D x3L rx

C rx

C rxy

D 1

V o

A

CB

D x1

i L rx

S1

i L

D xSx

D 2

D x2 D x3L rx

C rx

C rxy

D 1

V o

A

CB

D x1

i L rx

S1

i L

D xSx

D 2

D x2 D x3L rx

C rx

C rxy

D 1

V o

A

CB

D x1

i L rx

S1

i L

D xSx

D 2

D x2 D x3L rx

C rx

C rxy

D 1

V o

A

CB

D x1

(a ) (b )

(c ) (d )

(e ) (f)

S1

i L

D xSx

D 2

D x2 D x3L rx

C rx

C rxy

D 1

V o

A

CB

D x1

(g )

S1

i L

D xSx

D 2

D x2 D x3L rx

C rx

C rxy

D 1

V o

A

CB

D x1

(h )

Figure 5.4 Circuit operation diagrams in boost Mode

on, then current iLrx

in Lrx builds up linearly and the current in D2, iD2, decreases. The

equivalent circuit during this interval is shown in Figure 5.4(a). Due to the resonant


114

inductor Lrx, Sx turns on under ZCS condition. At time t1, i

Lrx reaches the level of the

main inductor current, iL, and the current in D2 crosses zero. The main diode D2 then

turns off under ZCS condition. The mathematical descriptions of the currents are

given in the following:

)( 0ttL

Vii

rx

oLrxsx

�� (5.1)

)( 02 ttL

Viiii

rx

oLLrxLD �� . (5.2)

The duration of the interval t1 - t

0 can be found from Equation 5.2 by letting i

D2 be

zero at t = t1:

o

rxL

V

Littt ��

� 0110 . (5.3)

2. Interval t1 - t2

During this interval, the auxiliary switch Sx is on and the main switch S1 and the

diode D2 are off. At time t

1, the diode D2 recovers its blocking characteristics. The

reverse recovery current value is iD2r

. The equivalent circuit during this interval is

shown in Figure 5.4(b). The current iLrx

(= iSx

) continues to increase. The balance of

current iLrx

, iD2r

, and iL starts to discharge capacitor Crxy. As Lrx and Crxy form a

resonant network, so the increase of iLrx

is now sinusoidal. The mathematical

descriptions for the resonant process are given as follows [91]:

)(sin)(cos 111

112 ttZ

Vttiiii o

rDLSxLrx �� (5.4)

)(sin)(cos 112111 ttiZttVV rDoA �� (5.5)

where

rxyrxCL1

1 �� (5.6)

rxy

rx

C

LZ �1 . (5.7)


115

This interval terminates at t2 when the voltage VA (or the voltage across Crxy) changes

its polarity and the diode D1 turns on, clamping VA to zero. Thus, the duration of this

interval can be found from Equation 5.5 by letting VA = 0 at t = t2:

��

��

��

�

rD

o

iZ

Vttt

2111221 arctan

1

�. (5.8)

Before t = t2, the voltage across Crx remains close to zero. Therefore VB follows VA.

Normally the reverse recovery current of the diode D2 has already dropped to zero

before t2. The current iLrx

reaches its maximum value, iLrxm

, at t2:

1Z

Vii oLLrxm �� . (5.9)

3. Interval t2 - t3

Between time t2 and t3, both the auxiliary switch Sx and the diode D1 are on and they

short the inductor Lrx, as shown in the equivalent circuit in Figure 5.4(c). Under ideal

conditions, the current in this loop remains constant at the value of iLrxm

, as given in

Equation 5.9. Although S1 is gated during this interval, there is no current flowing

through S1 since the anti-parallel diode D1 conducts. This interval ends at t = t3 when

Sx turns off under ZVS condition. Consequently, the conduction of D1 ceases and S1

conducts. It is evident that the main switch S1 turns on under ZVS condition and no

power losses occur. In particular, the reverse recovery current of the main diode D2

does not impose a turn-on loss in the main switch S1, while in a hard switched power

converter, the reverse recovery current of the diode is a notorious problem.

4. Interval t3 - t4

The equivalent circuit during this interval is shown in Figure 5.4(d), with two

independent circuit loops. The current iL is shorted by S1 which is on, while current

iLrx

is flowing through Dx2

and Crx, decreasing sinusoidally as expressed below:

)(cos 321 ttIiii LrxmCrxDxrx �� (5.10)

where,


116

rxrxCL

12 �� (5.11)

The voltage across Crx can be easily found as:

)(sin 322 ttZIV LrxmCrx��

where,

rx

rx

C

LZ �2 (5.12)

Usually Crx is chosen such that VCrx

can reach Vo. That is, the energy stored in Lrx

during the intervals t0-t1 and t1-t2 is larger than required to charge Crx to the value of

Vo. This condition can be met by satisfying the following inequality:

oLrxm VZI �2 (5.13)

The interval t3 - t4 ends at t4 when V

Crx reaches Vo and consequently Dx2 conducts,

clamping VB or VC to Vo. To find the duration of t3 - t4, let V

Crx = Vo at t = t4, i.e.

orx

rxLrxm Vtt

C

LI �� )(sin 342� (5.14)

Therefore,

��

��

��

�

rx

rx

Lrxm

o

L

C

I

Vttt arcsin

1

23443

�. (5.15)

5. Interval t4 - t5

The equivalent circuit in this interval is shown in Figure 5.4(e). While part of the

energy stored in Lrx is used to charge Crx to Vo, the rest of the energy stored in Lrx is

transferred to the output through Dx1, Dx2 and S1. iLrx linearly ramps down to zero.

The series diode Dx3 prevents any oscillations between Lrx and Cdsx.

As iLrx

decreases linearly, it can be easily found that:

4324

4554 cos)(

��

��

�� tV

IL

V

ttiLttt

o

Lrxmrx

o

Lrxrx � (5.16)

It should be noted that current in S1 during this interval exhibits a valley since

the current of the diode Dx3 and Dx2 reduces the current in the main switch S1.


117

6. Interval t5 - t6

Figure 5.4(f) shows the equivalent circuit during this interval. S1 is kept on and the

voltage across Crx remains at Vo until t = t6 when S1 turns off. The duration of this

interval entirely depends on the required duty cycle.

7. Interval t6 - t7

Due to Crxy, S1 turns off at t6 under ZVS condition. After S1 turns off, current in the

main inductor charges Crxy and discharges Crx through Dx3, as shown in Figure 5.4(g).

Voltages across Crxy and Crx are expressed respectively as:

)( 6ttCC

iV

rxrxy

LCrxy �

�� (5.17)

)( 6ttCC

iVV

rxrxy

LoCrx �

�� (5.18)

At t = t7, VCrxy = Vo and V

Crx = 0. The duration of this interval is:

L

orxrxy I

VCCttt )(6776 ��

�

(5.19)

8. Interval t7 - t8

At t = t7, voltage VA reaches the value of Vo and thus D2 conducts, as shown in

Figure 5.4(h). At t8, the auxiliary switch Sx turns on again, which marks the end of a

complete switching cycle.

Simulation of the ZVT dc-dc converter on SPICE was carried out and Figure 5.5

shows the simulated key waveforms of the ZVT converter in boost mode. It is noted

that is1

in Figure 5.5 is actually the sum of the drain current of S1 and the current in

the antiparallel diode of S1. It can be clearly seen that the main switch S1 turns on and

turns off under ZVS condition, and both the auxiliary switch Sx and the main rectifier

D2 turn on under ZCS condition and turns off under ZVS condition.

In Figure 5.2, if the flying capacitor Crx is removed, (in this case only one of the

diodes Dx2, Dx3, is needed), the ZVT dc-dc boost circuit will become Hua’s ZVT dc-

dc boost circuit [93], as shown in Figure 5.6. This will reduce the number of the


118

0.16 0.18 0.2 0.22 0.24

Time (ms)

Figure 5.5 Simulated typical circuit waveforms in boost modeFrom top to bottom: Vgs1, Vgsx, iL, irx, is1, Vds1, isx, Vdsx, id2, Vd2, Vc, iDx3

Note: is1 includes the drain current of S1 and the current in the antiparallel diode of S1

shown as the negative part

components of the auxiliary circuit, however the auxiliary boost switch Sx will not

operate under soft switching condition. Therefore the efficiency will be lower

compared to the ZVT boost dc-dc converter with the flying capacitor Crx.


119

L

S1

V lo w

D xSx

D 2

D x2L rx

C rxy

D 1

A

C

D x1 V hig h

Figure 5.6 Hua’s ZVT dc-dc boost converter

It is interesting to note that the auxiliary ZVT circuit can be seen as a ‘baby’ dc-

dc boost circuit considering its topology and the fact that it processes only small

amount of the total power.

5.3.2 ZVT Operation in Buck Mode

The equivalent circuit of the dc-dc converter in buck mode is drawn in Figure 5.7.

Similarly to the ZVT boost dc-dc converter, the auxiliary circuit in the ZVT dc-dc

buck converter can be seen as a ‘baby’ dc-dc buck circuit. Considering the duality of

dc-dc boost converter and dc-dc buck converter, the ZVT dc-dc buck converter

shown in Figure 5.7 is a dual circuit of Hua’s ZVT dc-dc boost converter as shown in

Figure 5.6. Due to the duality of the both circuits, it is not necessary to give a detailed

analysis of the ZVT operation in buck mode here, which is similar to that in Section

5.3.1.

L

V lo w

Sy

S2

D y2

L ry

C rxy

D 1

A

D y1

D

V hig h

Figure 5.7 Equivalent circuit in buck mode


120

Figure 5.8 shows the simulated key waveforms of the ZVT dc-dc converter in

buck mode. It is clearly shown that both the buck switch S2 and the rectifier

(freewheeling) diode D1 switch softly.

However, since the ZVT dc-dc buck circuit in Figure 5.7 is a dual circuit of

Hua’s ZVT dc-dc boost circuit, rather than the one shown in Figure 5.2, the auxiliary

buck switch Sy doesn’t operate under soft switching condition. Of course this will

limit the efficiency improvement in buck operation. Choice of such a design for the

ZVT buck operation is based on the following reasons. Firstly, the bidirectional

0.66 0.68 0.7 0.72 0.74

Time (ms)

Figure 5.8 Simulated typical circuit waveforms in buck modeFrom top to bottom: Vgs2, Vgsy, iL, iry, is2, Vds2, isy, Vdsy, id1, Vd1


121

dc-dc converter works mostly in boost mode. So the efficiency in buck mode is not as

important as in boost mode. Secondly, the auxiliary buck switch only handles a much

lower RMS current than the main buck switch does, and thus a smaller power device

with lower output capacitance can be used as the auxiliary buck switch (this also

applies to the auxiliary boost switch), resulting in smaller turn-off loss. Thirdly, the

rectifier diode D1 in the ZVT dc-dc buck circuit always operates under ZCS

condition. Thus it does not suffer from a reverse recovery problem. For a hard-

switched PWM converter, however, the reverse recovery loss of the rectifier

(freewheeling) diode normally dominates the total switching loss in high voltage

applications, where p-n junction diodes are used. Fourthly, the auxiliary ZVT buck

circuit is simpler than the auxiliary ZVT boost circuit which is shown in Figure 5.2.

Last but not least, the auxiliary ZVT buck circuit does not affect the bidirectional

operation of the ZVT dc-dc converter.

5.4 Implementation Issues and Experimental Results

5.4.1 Implementation Issues

A prototype of the soft-switched bidirectional ZVT dc-dc converter was constructed

and tested. Figure 5.9 shows a photo of the prototype. The actual circuit diagram of

Figure 5.9 A photo of the prototype of the ZVT bidirectional dc-dc converter


122

D xSx

D 2

D x2 D x3

L rxC rx

BD x1

S2

D y

SyD y2

L ry

D y1

D

R yyD yy

L sy

L sx

C

L

S1

V lo w

C rxy1D 1

A

R xx

D xx

C rxy2

V h ig h

Figure 5.10 Circuit diagram of the practical ZVT bidirectional dc-dc converter includingtwo saturable inductors and two RD snubbers

the power stage, shown in Figure 5.10, includes two saturable inductors and two

small diode-resistor snubbers. The two saturable inductors in series with the boost

resonant inductor Lrx and buck resonant inductor Lry, respectively, damp the parasitic

oscillation caused by the reverse recovery of the ZVT diodes. One diode-resistor

snubber is connected between Node C and the power ground for boost operation and

the other between Node D and the positive bus for buck operation. These two small

diode-resistor snubbers have proven very helpful in eliminating noises caused by the

reverse recovery of the ZVT diodes in the practical circuit. Consider the boost

operation as an example. At t = t5 (Refer to Figure 5.3 and 5.4), due to reverse

recovery of Dx1

and Dx2

, the current in Lrx reverses direction. When the ZVT diodes

Dx1

and Dx2

suddenly block after reverse recovery, the voltage at Node C would

quickly drop down below zero without this diode-resistor network. Then significant

negative voltage ringing (noise) at Node C would occur. The diode-resistor network

provides a path to release (dissipate) the small amount of energy stored in Lrx caused

by the reverse recovery current.

In addition, Crxy is split into two capacitors, Crxy1 and Crxy2, while the same total

capacitance is kept, with Crxy1 connected across the drain-source of S1 and Crxy2

connected across the drain-source of S2. In theory, the two capacitors are in parallel

and therefore only one capacitor is needed. However, practically, distributing Crxy at


123

two locations can do a better job than using a single capacitor in suppressing voltage

ringing across S1 and S2 (D2).

The main switches and auxiliary switches were all 200 V MOSETs (IXFN

180N20, IXYS) during Phase I of the NTCER EV project, due to the availability of

the power devices at the laboratory. Obviously MOSFETSs with much lower current

rating can be used for the auxiliary boost switch and buck switch to save cost.

One of the successes of the development of the soft-switched bidirectional dc-dc

converter is a novel design for the high-power main inductor with ease of design and

improved overall performance. It will be presented separately in the next section due

to its significance.

Figure 5.11 shows the control diagram of the bidirectional dc-dc converter. For

L

S1

D xSx

D 2

D x2 D x3L rx

C rx

C rxyD 1

D x1

SyD y2

L ry

D y1

S2

D y V hig hV lo w

D riverD river D riverD river

SyS2S1 Sx

S w itch in gS teerin gC ircu it

P I_ i

P I_ v

M od eS elec tion

B oost o r B u ck ?

C u rren tC om m an dS elec tion

V eh ic le S p eed

M otor C u rren t C om m an d

C u rren tC om m an d s

in B u ck M od eV hig h

C om m an d

Figure 5.11 Control diagram of the bidirectional dc-dc converter


124

the multiple functions of the dc-dc converter detailed in Chapter 4, three additional

components, a changeover dc contactor, a Schottky diode and a thyristor are added to

a conventional bidirectional dc-dc converter. The three additional components

require little, or very simple control. Basically, in boost mode the dc-dc converter is

double-loop controlled, with an inner PI current loop and an outer PI voltage loop.

The current reference comes from the output of the voltage PI regulator. In buck

mode, the dc-dc converter is closed-loop current controlled only, with the outer

voltage loop disabled.

Layout of the prototype, especially the power block, proved to be very critical in

minimizing noise during the development of the dc-dc converter.

5.4.2 Experimental Results

The prototype of the ZVT bidirectional dc-dc converter was tested in both boost

mode and buck mode in the laboratory.

In Boost Mode

Figure 5.12 to Figure 5.17 show a set of oscilloscope waveforms in boost mode. The

current waveforms of the power switches and diode were not measured as it was not

convenient to measure them due to the compact layout of the prototype. However

soft switching operation for both turn on and turn off of the main boost switch S1 and

the auxiliary boost switch Sx can still be clearly observed from Figure 5.13 and

Figure 5.14.

Figure 5.18 show the measured efficiency vs. output power at Vin = 100 V and

Vo = 170 V (rated). The efficiency is well above 96% over a wide output power

range. It should be noted that the rated input voltage of the boost converter (EV

battery voltage, for the function of motor speed extension) is 120 V and the rated out

power 10 kW. Preliminary measurements were limited by both the available lab dc

power supply limits (100 V / 50 A) and the test load bank. However during operation

the EV battery voltage can drop to 100 V, or less. It is expected that at rated Vin =

120 V and full output power, the efficiency of the converter would be even higher.


125

Figure 5.12 Gate drive waveforms Vgs1 (lower trace)and Vgsx (upper trace) in boost mode

Figure 5.13 Waveforms of Vgs1 and Vds1

showing soft switching operation of the main boost switch


126

Figure 5.14 Waveforms of Vgsx and Vdsx

showing soft switching operation of the auxiliary boost switch

Figure 5.15 Waveforms of Vgs1 (lower trace) and VD1 (upper trace) in boost mode


127

Figure 5.16 Waveforms of Vgsx (lower trace) and ilrx (upper trace) in boost mode

Figure 5.17 Waveforms of Vgs1 (lower trace) and Vc (upper trace) in boost mode


128

90

92

94

96

98

100

1000 2000 3000 4000 5000

Output Power (W)

Effi

cien

cy (

%)

Figure 5.18 Measured efficiency vs. output power in boost modeat Vin = 100 V and Vo = 170 V

As the ZVT technology does not impose additional voltage stress on the power

devices, it allow the use of lower voltage power devices, which usually have better

conduction and switching performance than higher voltage power devices. The use of

the lower voltage power devices, as well as the soft switching of the power devices,

also contributes to the high efficiency achieved.

In Buck Mode

Figure 5.19 to Figure 5.23 show a set of oscilloscope waveforms of the converter in

buck mode powered by a 100 V /50 A dc power supply. It should be noted that since

only one channel of differential probe was available in the lab, logical level gate

signal waveforms of the main buck switch and auxiliary switch were recorded instead

of the real gate signal waveforms. There are some delays between the real gate

signals and the corresponding logical gate signals, as can be seen in Figure 5.20. Soft

turn on and turn off of the main buck switch can be roughly observed from Figure 5.

21 considering the time delay between the logical gate signal waveform and the gate

drive waveform shown in figure 5.20.


129

Figure 5.19 Logical gate signal waveforms Vgs2 and Vgsy in buck mode

Figure 5.20 Logical level gate signal waveform (lower trace) and the real gate drive signalwaveform (upper trace) of the main buck switch, showing the time delay between them


130

Figure 5.21 Waveforms of Vgs2 (logical level, lower trace) and Vds2 (upper trace)showing soft switching operation of the main buck switch

Figure 5.22 Waveforms of Vgsy (logical level, lower trace) and Vdsy (upper trace)


131

Figure 5.22 Waveforms of Vds2 (upper trace, 100V/div) and VD1 (lower trace, 50V/div)in buck mode

Figure 5.23 Waveforms of Vgsy (lower trace) and ilry (upper trace) in buck mode


132

90

92

94

96

98

100

500 1500 2500 3500

Output Power (W)

Effi

cien

cy (

%)

Figure 5.24 Measured efficiency vs. output power in buck modeat Vin =100 V and RL = 1.5 ohm

Figure 5.24 shows the measured efficiency vs. output power at Vin =100 V and

RL = 1.5 ohm, with 96% efficiency at Po = 3.4 kW. Even higher efficiency can be

expected in real driving operation, because the input voltage of the converter in buck

mode (the bus voltage) and the output power will be higher.

5.5 Novel Design of High-Power Ferrite Cored Inductors

with Ease of Design and Improved Overall

Performance

The usage of switched mode power supplies or power converters has undergone a

significant growth. Inductors are present in most power electronics circuits, serving

as a filter inductor as in a buck converter, or an energy storage inductor as in a boost

converter, or for other purposes. Generally speaking, inductors in power electronic

circuits with capacitors create sinusoidal variations of voltage or current. This limits

the rate of change of current and sharp transient current transitions.

Inductors, as well as transformers, are frequently the heaviest and bulkiest items

in power converters and are not generally available off the shelf. This is particularly

true for designing high-power and high-frequency converters.

Inductors have a significant effect upon the overall performance and efficiency

of the systems. Skilled inductor design has proven to be a critical to the success of a


133

power electronics product. Because of the interdependence and interaction of

parameters, judicious tradeoffs are necessary to achieve design optimization. At low

power levels, traditional inductor designs with conventional inductor core and

winding configurations can achieve quite good results with ease [97]. However, at

high power levels, due to large dimensions of the cores and winding coils, and large

air gaps, etc., second-order effects such as fringing flux and leakage flux become

significant. This makes the traditional inductor designs with conventional inductor

core and winding configurations inefficient and simple design methods inaccurate.

Issues like high localization of flux with the cores, high-level leakage flux, and hence

high levels of EMI, and low utilization of the cores become of concern, and it is often

difficult to find a good compromise.

In the development of the soft-switched bidirectional dc-dc converter, one of the

contributions is a novel design of the high-power inductor with ease of design and

improved overall performance. In this section, problems with conventional design of

high-power ferrite inductors are addressed and examined with aid of 3-dimensional

(3D) finite element analysis (FEA). A novel winding scheme for high-power ferrite

cored inductors is then proposed to combat the problems.

5.5.1 Problems with Conventional Design of High-Power Ferrite

Cored Inductors

5.5.1.1 Conventional Core and Winding Configurations of High Power Ferrite

Cored Inductors

In high power and high frequency applications, inductors can be air-gapped ferrite

cored, powered iron cored with a distributed and non-adjustable air-gap, or air cored.

At a high power level, large cross-sectional areas and window areas of cores are

necessary. U cores are generally all that is available and choices are limited. Cores

come in discrete sizes, and ferrite cores in very large sizes are often difficult to

obtain. A practical solution is to use a group of ferrite cores which are put together in

certain configurations. Once again, U cores are suitable for such an application.


134

Two common and simple core and winding configurations using a group of U

cores are shown in Figure 5.25 and Figure 5.26. For simplicity, inductors with core

and winding configurations as shown in Figure 5.25 will be referred to herein as

Inductor Configuration #1 and inductors with core and winding configurations as

shown in Figure 5.26 will be referred to herein as Inductor Configuration #2.

An example of an Inductor Configuration #1 is a 5 uH, 175 A inductor designed

and realized by other researchers, as detailed in [98, 99]. It used four large U100/57

ferrite cores (with AE = 645 mm2) of grade F5A. However, upon construction of the

inductor, a number of design criteria were not met accurately. Firstly, it was found

that the measured inductance was 14 uH, with an error in inductance exceeding

180%. The air gap was increased and the number of turns reduced to attain the

Figure 5.25 Inductor Configuration #1

Figure 5.26 Inductor Configuration #2


135

expected 5 uH inductance. Secondly, unexpected levels of EMI were so high that

some ferrous materials used in the supporting structure became hot during testing.

The high-level of EMI was obviously a noise source for the control circuitry, which

could cause circuit reliability problems.

The case study aforementioned is seemingly typical of conventional high power

magnetic circuit design with such core and winding configurations. Conventional

inductor design is base on the assumption that all the flux linked by the winding is

constrained by the core. In fact, there are significant second-order effects. Since

permeabilities of ferrite cores and air differ by only one or two orders of magnitude,

an additional leakage component of unlinked flux is almost always present. The

leakage flux increases with the core dimensions, the winding scheme, the radius of

the winding coil, and the air gap. In addition, the flux across the gap contains an

additional component that bulges outside the core cross section, known as fringing

flux. Fringing flux is a function of gap dimension, the shape of the pole faces, and the

shape, size and location of the winding. Its net effect is to increases the effective

cross section of the gap or shorten the air gap, decreasing the total reluctance of the

magnetic path. Therefore, fringing flux increases the inductance by a factor to a value

greater than that based on the assumption that gap flux is normal to the pole faces.

Fringing flux is a larger percentage of the total for larger air gaps.

While fringing effect can be approximately calculated, it is very hard to quantify

leakage flux unless FEA, preferably 3D EFA, is performed.

5.5.1.2 Observations of Inductor Configuration #1 and Inductor Configuration

#2 Aided by 3D EFA

To gain an almost ideal picture of the behavior of the complex magnetics involved,

3D FEA was employed to examine an inductor utilizing Inductor Configuration #1,

as built with the inductance of 5 uH, after increasing the air gaps to achieve the

design inductance. Figure 5.27 shows the view highlighting the ¼ section used in the

3D FEA model of Inductor Configuration #1.


136

Z

X

C o re 1

C o re 2Y

Figure 5.27 View highlighting the ¼ section used in the 3D FEA model of InductorConfiguration #1

Definition of Core Utilization

In comparing different core configurations and winding topologies, a measure of

utilization of the ferrite cores is needed. A measure that defines how much flux

travels through the core sections with respect to the total flux generated by the

winding structure is chosen. Using 3D finite element analysis, the total flux generated

by the winding and the flux that travels through each core section are calculated via

integrals of flux density on the z = 0 plane of symmetry. Consider Inductor

Configuration #1 as an example, shown in Figure 5.37, the core utilization factor,

)(coreuK , is defined as:

winding

corecoreuK

�

� 2)( � (5.20)

where 2core� is the flux in core section 1 that travels perpendicularly through the z = 0

plane and winding� is the half of the total of magnitude of flux that travels

perpendicularly through the z = 0 plane.

It should be noted that because of the conditional nature of (5.20), this definition

is based upon the core section with the worst ratio of core-carried flux to total


137

induced flux. In some cases, due to symmetry, the flux distributions in the two core

sections are identical, such as Inductor Configuration # 3 described later.

3D FEA Results for Inductor Configuration #1

Figure 5.28 shows a plot of flux density upon a cut-plane through a ¼ 3D FEA

model. The units of the scale shown at right are Tesla and the windings have been

hidden for clarity. The ferrite material used in the inductor has a B-H characteristic

that begins to saturate at 300 mT. The inductance obtained through 3D FEA is 4.63

uH. This agrees closely with the measured value of inductance, 5 uH. Furthermore,

3D FEA also provides other important information about its performance. Firstly,

59% of the magnetic flux generated by the copper windings is not constrained to the

magnetic core, hence the core utilization factor, )(coreuK , is 0.41. Secondly, large

portions of the core show high variation from the designed average flux density,

including areas of under utilized core material.

3D FEA Results for Inductor Configuration #2

An example of an Inductor Configuration #2 is a 55.4 uH (measured), 150 A

inductor, built at NTCER for comparison purposes. It used eight U100/57 ferrite

Figure 5.28 3D FEA of a ¼ model of Inductor Configuration #1 showing excessive fluxleakage and some flux localization around the central leg (represented at left)

( B̂ = 228mT, L= 4.63 �H, )(coreuK = 0.41)


138

(a)

(b)

Figure 5.29 3D FEA results for Inductor Configuration #2(a) a ½ 3D FEA model

(b) a plot of flux density upon a cut-plane showing excessive saturation

but less flux leakage than Inductor #1(B̂ = 356 mT, L= 53.0 �H, )(coreuK = 0.90)

cores, the same core as used in Inductor Configuration #1. This inductor was also

modeled with 3D FEA. Figure 5.29 shows a ½ 3D FEA model and a plot of flux

density upon a cut-plane of Inductor Configuration #2.

The results of the 3D FEA also show significant amounts of flux leaving the

right-hand U-core section prior to crossing the air gap. Not only can this result in a

very high level of EMI, but also causes poor sensitivity of the air gap adjustment on

inductance, and causes bad utilization of the constant cross section core material.


139

Conventional designs of high power ferrite cored inductors try to make a trade-

off among peak flux density, core losses, EMI level and utilization of the cores.

Unfortunately, it is difficult to achieve a satisfactory overall performance of the

inductors. The conventional winding topologies presented in Figures 5.25 and Figure

5.26 rely on the difference in permeability of the magnetic core material and the

surrounding air to constrain magnetic flux within the typical ferrite core structures.

As the physical dimensions and magnitude of excitation of these structures become

large, areas of the core approach saturation cause the effective permeability of the

longer-length, ferrite-bound paths including air gaps to converge towards the short-

length path though the surrounding air. This ultimately results in the large percentage

of flux excursion from the magnetic core. This problem is illustrated in the analysis

of the conventional winding schemes depicted in Figures 5.28 and Figure 5.29 where

significantly higher flux in the core leg internal to the winding is observed than in the

other limbs.

5.5.2 Novel Design of High Power Ferrite Cored Inductors

5.5.2.1 A Novel Winding Scheme for High Power Ferrite Cored Inductors

Both Inductor Configuration #1 and Inductor Configuration #2 have a common

property, that is, they all employ a single coil winding. In formulating the design of

the inductor (40 uH, 150 A) for the bidirectional DC-DC converter, a novel winding

scheme has been proposed by the author to reduce localized saturation and leakage

flux, and improve magnetic core utilization. This inductor uses the same core

configuration of Inductor Configuration #2 as shown in Figure 5.26, but with a dual-

coil or distributed-coil winding scheme. The resulting configuration, Inductor

Configuration #3, is shown in Figure 5.30. Compared to Inductor Configuration #2,

one-half of the number of turns used previously are moved to the opposite side of the

magnetic circuit while the total number of turns is preserved.


140

Figure 5.30 The proposed dual-coil winding configuration, Inductor Configuration #3

5.5.2.2 3D FEA Results for Inductor Configuration #3

Due to the symmetry of the cores and windings, only a ¼ model of Inductor

Configuration #3 is needed for simulation, as shown in Figure 5.31 (a). The 3D FEA

results of Inductor Configuration #3 are shown in Figure 5.31 (b). It can be seen that

the proposed winding configuration significantly reduces the localization of

saturation, with Bm=270 mT, and encourages good utilization of the magnetic cores,

with )(coreuK = 0.91. One direct benefit of good utilization of the magnetic cores or

low leakage flux, is significant decrease of EMI levels.

This novel topology offers a performance that is close to the near-ideal toroidal

inductor topology, which exhibits very high utilization factors by enclosing the

magnetic core within the winding. The advantages of the dual-coil winding topology

over that of the toroidal variety is that it offers the freedom of inserting required air

gaps, simple core size manipulation by stacking and much easier mounting.


141

(a)

(b)

Figure 5.31 3D FEA results for Inductor Configuration #3(a) a ¼ 3D FEA model

(b) a plot of flux density upon a cut-plane showing much less localized saturation

( B̂ = 270 mT, L= 41.2 �H, )(coreuK = 0.91)

5.5.2.3 Measurement Results, Comparison and Conclusions

A photo of an inductor as built utilizing the proposed novel dual-coil winding

scheme is shown in Figure 5.32. It looks somewhat like a single output transformer

but with only two terminals totally. Each end of the winding conductor has been

terminated using 2 lugs for manufacturing reasons.


142

Figure 5.32 A photo of the inductor as built using the proposed dual-coil winding scheme

Table 5.1 summarizes the characteristics of the three different inductors and

provides an overall comparison.

From Table 5.1, it can be easily concluded that Inductor Configuration #3 using

the novel dual-coil winding scheme offers apparent advantages over Inductor

Configuration #1 and Inductor Configuration #2 which use conventional single-coil

winding scheme in terms of core utilization, EMI levels, and reduction of core

saturation level. Overall performance optimization at high power level has been

achieved for Inductor Configuration #3, while for Inductor Configuration #1 and

Inductor Configuration #2, it is difficult to achieve a good design compromise.

Considering Inductor Configuration #1 as an example, increasing the air gaps reduces

Table 5.1 Comparison results of the three different inductors studied

InductanceInductor Measured

(uH)3D

FEA(uH)

Error(%)

PeakFlux

Density(mT)

Saturationlevel

CoreUtilizationKu(core)

EMILevel

#1 (with smallerair gap)

14.0 15.6 11.4 356 Very High 0.676 High

#1 (with largerair gap)

5 4.63 7.4 228 Low 0.412VeryHigh

#2 55.4 53.0 4.3 356 Very High 0.902 Low

#3 39.1 41.2 5.4 270 Low 0.910 Low


143

peak flux density, however it also results in excessive EMI levels and very poor core

utilization. The converse is also true.

It should be noted that while 3D FEA has been the most accurate design tool to

closely examine the inductor’s flux distribution and verify the new design, it is the

proposed novel dual-coil winding scheme that makes it possible to achieve design

optimization. From the practical point of view, the 3D FEA software, which is

relatively expensive at present, is not a necessity in designing high power inductors if

the proposed winding scheme is employed. With the novel dual-coil winding scheme,

leakage flux in the inductor has been reduced to a lower level, so that fringing flux

effect is now the only major effect to be considered. Unlike leakage flux, fringing

flux can be estimated based on analytical formulae. Therefore the proposed winding

scheme for high-power inductors has allowed the use of traditional design theory to

quite accurately predict parameters such as inductance and peak magnetic induction

with confidence.

5.6 Review

In presenting the development of the bidirectional dc-dc converter, this chapter

studies two main topics as summarized below.

(1) Implementation of Soft Switching Operation of the Bidirectional DC-DC

Converter

After an evaluation of various soft-switching techniques, the zero voltage transition

(ZVT) soft switching technique is chosen for the multifunctional dc-dc converter.

The ZVT dc-dc converter has minimum switching loss resulting in high efficiency,

reduced electromagnetic interference (EMI) and improved reliability for the power

devices in the dc-dc converter. Another important feature of the ZVT dc-dc converter

is that no additional switching voltage and current stress are imposed on the power

devices. Simulation results are presented and verified by the measurement results of

the built ZVT dc-dc converter. The measured efficiency of the ZVT dc-dc converter

is above 96%.


144

(2) Novel Design of High Power Ferrite Cored Inductors with Ease of Design

and Improved Overall Performance

At high power levels, due to large dimensions of ferrite cores and air gaps, the


significant. Conventional designs of high power ferrite inductor using traditional core

and winding configurations try to make a trade-off among peak flux density, core

losses, EMI level and utilization of the cores. Unfortunately, it is often difficult to

achieve a good compromise. A novel design of high power ferrite inductors

employing a dual-coil or distributed winding scheme is proposed by the author to

combat this problem. Results from both the 3-Dimension (3D) finite element analysis

(FEA) and measurements verify the viability of the novel design for high power

inductors. The proposed inductor design allows better optimization, reducing the

localization of saturation and leakage flux, or EMI levels, and achieving good

utilization of the magnetic cores.

145

Chapter 6

Development of an Intelligent DriveModule for High Power IGBTs

In recent years, IGBTs have been widely used in various power electronic systems,

particularly in high power and high voltage applications. This is due to their inherent

characteristics of voltage control, low forward voltage drop and high current density,

resulting from their hybrid structure of MOSFETs and BJTs [100]. Both turn-on and

turn-off of IGBTs are achieved by means of a voltage applied to the gate, requiring

only low driving power and a cheap drive stage, similar to that required by

MOSFETs. Moreover, IGBTs have low forward voltage drop and high current

density, like BJT devices, which mean lower conduction loss than MOSFETs in high

power applications. In the power range of tens of kilowatts to hundreds of kilowatts,

IGBTs have dominated the applications, since MOSFETs, the rivals of IGBTs in

middle power range (for example, one to ten kilowatts), are unable to be competitive

with them due to MOSFET’s high conduction loss in high current applications.

As stated in Chapter 1, although the power electronics design for Stage 1 of our

EV project employed primarily MOSFETs as switching devices due to time and

budget constraints, in Stage 2 these devices will be replaced by IGBTs to provide

more power.

Like other power semiconductor devices, IGBTs need gate drive circuits to

control them. Experienced power electronic engineers agree that drive circuits are

CHAPTER 6 DEVELOPMENT OF AN INTELLIGENT DRIVE MODULE FORHIGH POWER IGBTS

146

very critical to the reliability, efficiency, noise sensitivity, and size or volume of

power electronic systems. Modern power electronic equipment such as general

purpose inverters, power supplies, numerically controlled machine tools, industrial

robots, as well as electric vehicle propulsion systems have increased the demands for

high efficiency, high reliability, low noise, intelligent functions and reduced size, and

have exploited the convenience of use of IGBTs. Although drive principles and

techniques for IGBTs have been quite well developed [100-104], some effort is still

needed to develop intelligent, downsized, easy-to-use IGBT gate drivers with high

current drive capacity. Various commercial MOSFET/IGBT drive ICs can be

classified as three types:

� Pure-amplifier drivers such as TC4420/4429

� Intelligent gate drivers such as MC33153 (Motorola)

� High-voltage half-bridge drivers (high-side driver) such as HIP2500 (Harris

Corporation), and IR2110 (International Rectifier).

However, few commercial IGBT drivers can, at present, satisfactorily meet all

the desired requirements. Commercial IGBT drivers are restricted to certain

applications and are subject to one or more of the following limitations:

� Peak current no larger than 2 amperes;

� Single driver;

� No intelligent functions such as overcurrent or desaturation protection;

� No fault status annunciation;

� Unipolar drive outputs;

� Need for external isolated dc supply;

� Inconvenient to use.

This chapter presents the development of a new intelligent gate drive module for

high-power IGBTs. The IGBT drive module features high current drive capacity,

powerful functions, high performance, wide applications, and convenience of use,

and is especially suitable for high power applications such as in EV drives.


147

6.1 Functions and Features of the New Intelligent IGBT

Drive Module

The new intelligent IGBT drive module provides the following major functions and

features:

1. High current drive capacity (> 6A), being able to drive up to 400A/1200V

IGBTs.

2. 3000V galvanic isolation between drivers and power potentials.

3. Bipolar drive outputs, typically +13V/-4.5V, ensuring the “off” status of the

IGBT when it is required to be “off”.

4. Dual gate drivers, especially suitable for phase leg drive.

5. Intelligent functions:

� Overcurrent protection

� Fault status annunciation

� Temperature feedback for intelligent power module (IPM) applications

� Under voltage lockout protection (UVLO) with hysteresis

6. Built-in, well regulated, and isolated floating power supply, requiring only one

incoming unipolar dc supply, and allowing large fluctuation of the dc supply.

7. Suitability to high-power MOSFETs, as well.

8. Convenience of use, since the design has:

� A size of only 100*23 mm (3.94 by 0.905 inch). These “standard”

dimensions allow the drive module to be fixed inside a standard dual

IGBT module package to form an IPM in a phase leg configuration, as

well as more general applications.

� A small number of input connectors (see Table 6.1).


148

Table 6.1 Definition of input pins of the drive module

Pin Connection

1 +10 to 18 volt dc supply

2 High side IGBT command

3 Temperature feedback (only for IPM)

4 Temperature feedback (only for IPM)

5 Fault signal

6 Low side IGBT command

7 DC supply ground

6.2 Functional Diagram and its Operation

Figure 6.1 shows the functional diagram of the intelligent IGBT drive module.

C u rren tA m p li f ie r

+

T ran sm it ter

R ec eiv er

T ran sm it ter

R ec eiv er

S w itch -m o d eP ow er S u p p ly

F au ltL og ic

+ 1 2 V

In 1

F au lt

In 2

T em p 1

G n d

T em p 2

D esatu ra tio nD etec t ion

C u rren tA m p li f ie r

-

D esatu ra tio nD etec t ion

Figure 6.1 The functional diagram of the new intelligent IGBT drive module

6.2.1 Intelligent Functions

Most intelligent functions discussed above are implemented by the HCLP-316J

(Hewlett Packard), a newly released 2 Amp gate drive optocoupler with integrated

desaturation or overcuurent detection and fault status annunciation [105]. The HCLP-


149

316J provides not only functions such as desaturation or overcurrent detection, fault

status annunciation and UVLO as in other intelligent MOS gate drive ICs, but also

two built-in optocouplers for isolated fault status feedback and isolated gate signal

coupling. These two built-in optocouplers make the gate drive PCB board design

simpler.

Overcurrent protection is implemented using the principle of desaturation. Under

a short circuit condition, the IGBT collector-emitter voltage drop increases as it

begins to operate in the linear region. This increase in voltage is monitored as an

indication of excessive current. Upon detection, the gate pulse in progress is

terminated. In addition, the fault signal is “set” to allow detection of the IGBT

condition by the drive control logic.

6.2.2 Bipolar Drive Outputs

When one IGBT in a half-bridge is turned on, a quite high dv/dt will be generated

across the other one in the same bridge leg, due to the recovery of the antiparallel

diode of the IGBT. This dv/dt can cause the second device to conduct momentarily,

leading to additional losses [102]. A negative gate bias can help. In general, whether

a negative gate bias is necessary will depend on the level of the gate threshold

voltage and the characteristic of the collector-gate capacitor of the particular IGBT,

and the impedance of the gate drive circuit that determines the amplitude of the

induced voltage. Thus it is a good practice to apply a negative voltage to the gate of

the IGBT when it is required to be off, to ensure that the gate voltage cannot rise

above the threshold level. Normally a negative bias voltage of 4 to 5 volts is

sufficient for this purpose. A larger negative voltage will cause excessive drive

power dissipation. This new drive module provides bipolar output gate signals with

positive magnitude of 13V and negative magnitude of 4.5V while requiring only one

unipolar incoming dc supply. This is achieved by a built-in, well regulated and

isolated DC supply with two output channels, each of which provides both positive

and negative voltages, as will be discussed in the next subsection.


150

Since the output current capacity of HCLP-316J is only 2 amperes, an additional

current amplifier has been employed, which enables the drive module to provide over

6 ampere peak current and drive up to 400A/1200V IGBTs.

6.2.3 The Built-in, Well Regulated and Isolated Floating Bipolar

Power Supply

In order to provide the necessary floating power supplies, the incoming voltage is

converted to two separate supplies for each IGBT by a switch-mode power supply

connected in the flyback topology. Thus, complete galvanic isolation is maintained

both between the input and each output as well as between the two outputs. To get

the bipolar drive outputs preferred for insuring the “off” status of an IGBT, each of

the output supplies provides both positive and negative voltage. Current-mode

control is employed in the flyback dc supply to achieve a good dynamic

characteristic. For a wide input dc voltage range, the current-mode controller should

be able to work above 50% duty cycle. The current-mode control chip UC3843

(Unitrode Integrated Circuit Corp.) meets this requirement and was selected [106].

Since above 50% duty cycle is required, slope compensation is added for reliability

considerations.

Usually for output voltage regulation, the output voltage is sensed at the

secondary side and fed back to the primary through an optocoupler. Then this

feedback voltage is compared with a reference to get an error voltage. After being

amplified, the error voltage is used to control the output voltage. However, this

control scheme has a problem resulting from the nonlinear characteristic and the

temperature characteristic of an optocoupler that will cause a measurement error.

Since loop control is unable to compensate for a measurement error that is beyond

the loop, this detection error will lead to error in the output voltage regulation. The

effect of temperature is much larger than that of nonlinearity in this application.

To solve this problem, a different closed-loop voltage regulation scheme has

been used, as shown in Figure 6.2. Instead of feeding back the output voltage signal

directly to the voltage regulator in the primary side for voltage regulation, an

amplified error signal, the difference between the detected output voltage and the


151

-+

V ref

+ 1 2 V

S

P W MC on tro l ler

+ 1 2 V

C I

T h ree-term ina l a m p lifie r

Figure 6.2 The closed-loop voltage regulation scheme

output voltage reference, is obtained in the secondary side and fed back to the

primary side through the optocoupler to control the duty cycle, which in turn controls

the output voltage. Since the output voltage error has been amplified with infinite

gain through integration, and this amplified error voltage signal is used to control the

bias current of the optocoupler, the effect of the nonliearity and temperature change

of the optocoupler is eliminated. The three-terminal amplifier with internal voltage

reference, ZR431, together with the integral capacitor, CI, easily implements the

voltage error integration.

It is worth noting that the size of the flyback transformer in this application

should be very small. Therefore it is necessary to make the structure of the

transformer as simple as possible to make the fabrication of the transformer easier

and the isolation among different windings more reliable. For this reason, a simple-

structure transformer together with a simple circuit as shown in Figure 6.3 (a) was

employed to obtain the bipolar output voltage while a transformer with

asymmetrically center-tapped output windings shown in Figure 6.3(b) is avoided. For

the same reason, an auxiliary winding, as seen Figure 6.3(b), which is for isolation,

detection and feedback of the output voltage in closed-loop voltage regulation, is not

recommended. Instead, the scheme of output voltage detection and closed-loop

control shown in Figure 6.2 has been employed. Obviously the transformer used in

this application in Figure 6.3(a) is much simpler than the one in Figure 6.3(b) which

is widely used in conventional flyback converter design.


152

+

-

+

-

A u xila ry vo ltag esensing w ind ing

P r im aryw ind ing

(a ) (b )

+

-

-

+

Figure 6.3 Two different flyback transformer configurations

6.3 Measurements

The proposed intelligent IGBT gate drive module was prototyped and tested. All

components were surface-mounted on a PCB board, which minimizes the size of the

drive module. The drive module worked very well. Figure 6.4 to Figure 6.8 show the

measured typical waveforms. Typical on times and off times are shown in Figure 7.4

and Figure 6.5. Both the low to high propagation delay time and the high to low

propagation delay time are around 550ns. From these two figures we can also read

that the positive level of the gate signal is about 13 V and the negative level is

approximately -4.5 V. Typical turn-on and turn-off waveforms for a 300A, 600V dual

IGBT module are shown in Figure 6.6 and Figure 6.7.

Figure 6.8 shows the termination of the output pulse following the determination

of a fault condition. There is a delay of approximately 2us due to the input

propagation delay and a blanking time necessary to prevent false indications due to

diode reverse recovery current. To avoid misoperation, a blanking time of 500ns is

inserted at the beginning of each gate pulse.

When a fault condition is sensed, the corresponding fault signal will be latched

in the “low” state. The fault signal will remain in the "low” state until reset by the

rising edge of the next gate pulse for the "faulted" channel.


153

Figure 6.4 Turn-on gate signal

Figure 6.5 Turn-off gate signal

Figure 6.6 Typical turn-on waveforms for 300A, 600V IGBTs


154

Figure 6.7 Typical turn-off waveforms for 300A, 600V IGBTs

Figure 6.8 The termination of the pulse followingthe determination of a fault condition

Figure 6.9 A photo of the intelligent high power IGBT drive modulefor general applications


155

Several versions of PCB layout of the intelligent drive module were designed for

various applications. They are very convenient to use. Figure 6.9 shows a photo of

one of the versions.

6.4 Practical Application Issues

6.4.1 Fault Status Feedback

Since the fault signals are provided by open collector drivers with the fault state

defined as logic “low”, two fault signal outputs are combined in a “wired OR”

configuration to develop a composite fault signal. In a three-phase system, more such

fault signals can be combined in a “wired OR” configuration to generate a composite

fault signal for input to the control logic. The control system monitoring the fault

signal is required to latch the fault condition since the internal fault logic is reset at

the end of each input pulse. It is good practice to latch the fault signal with user logic,

as it may be possible in some control systems to command the next gate pulse before

the fault feedback signal is acknowledged.

6.4.2 Selection of Gate Drive Resistor

Each IGBT requires a gate drive resistor inserted in series with the gate connection.

Very low values of series gate resistor will not only produce high diode recovery

voltage transients, but could give rise to unacceptable ringing during recovery, by

switching the device too quickly. This is due to the existence of stray inductance and

parasitic capacitance which form a resonant circuit which can be set into damped

oscillation. This electrical noise could interface with control and protection circuits

and would contribute to the total system EMI. On the other hand, too big a gate drive

resistance will slow the switching transient and lead to high switching losses. A

compromise needs to be made based on specific operating conditions of a circuit.

There are many trade-offs in determining the proper resistor for a specific

application. Factors influencing the choice of gate resistors include the IGBT

manufacturer’s recommendation, switching frequency, proximity of IGBT device to

the gate driver (potential oscillations), IGBT device snubbering, importance of


156

switching losses, acceptable level of EMI, and minimum deadband requirement.

There is an inherent 2-ohm resistor present in each output to limit the maximum

current. An external resistor or resistor network may be installed in accordance with

the application where smaller drive current is required.

To optimize the turn-on and turn-off times, it may be desirable to use separate

gate resistors, as shown in Figure 6.10.

Gate resistors have a high-peak power requirement relative to the average power

dissipation. This is due to the nature of driving a capacitive load. Carbon

composition resistors are superior to metal film types for this duty and are, therefore,

recommended.

R on

Roff

Figure 6.10 Optimized gate connection configuration

6.5 Review

This chapter presents a new intelligent gate drive module for high power IGBTs. The

intelligent functions of this module include under voltage lockout protection and

overcurrent protection with fault status annunciation. The module contains dual

drivers which make it especially suitable for phase leg drive, and outputs bipolar

signals so that the “off” status of the IGBT is ensured when the IGBT is required to

be “off”. It needs only a unipolar dc supply due to the built-in, well-regulated, and

isolated floating power supply. This module features high current capacity, powerful

functions, high performance, wide application, and convenience of use, and is

suitable for high power applications such as in EV drives. The detailed functions,

operation, performance, and practical application issues are addressed.

157

Chapter 7

Soft-Switched On-Board EV BatteryChargers with Unity Power Factor

Battery chargers are important components in the development of electric vehicles.

There are two types of chargers for EV application. One is a stand-alone type which

can be compared to a petrol station aimed at fast charge. The other is an on-board

type which would be appropriate for slow charge from a house utility outlet during

nighttime and opportunity charge. Since demand of electricity during nighttime is

low while electricity generation facilities have been already installed and

considerable amount of electricity has been already generated in nighttime, the

electricity distribution system is running inefficiently. In some areas, electricity is

simply wasted during nighttime to some extent. Under some sort of market

regulations, nighttime charge can effectively shift loads from the "peak" in daytime to

the "off-peak" in night time and greatly benefits an electricity distribution system.

Night-time charge will also benefit customers. They do not have to spend too much

time in charge stations unless they urgently need a fast charge. Introducing "off-peak"

prices of electricity, which encourages customers to charge within discount hours,

should further benefit customers financially [107].

On-board battery chargers can also allow opportunity charging of EVs. Given

that a major problem inhibiting wider use of EVs is the relatively short range

between battery charges, a good deal of effort has been put into seeking ways to

CHAPTER 7 SOFT-SWITCHED ON-BOARD EV BATTERY CHARGERS WITHUNITY POWER FACTOR

158

ameliorate the situation by providing opportunities of frequent charging [108]. Most

vehicles operating in urban environment spend far more time parked than they do

moving. On the other hand, electricity outlets are already very widely available in

urban areas, and in many cases it is possible to connect an EV to power outlet where

it might be parked. Optimum use of opportunity charging can extend the daily

operating range well beyond single charge range.

Considering the popularity of EVs in the near future and the fact that an on-

board battery charger is always carried by an EV, special requirements to such an on-

board charger must be set. They include:

� Single phase AC input;

� High efficiency;

� Low harmonics;

� Nearly unity input power factor;

� Low cost;

� Minimum size and weight;

� Safety of consumers operating the equipment.

The power factor correction (PFC) requirement is necessary to comply with

recently introduced IEC 1000-3-2 line harmonic current regulations. In presenting the

development of a high-efficiency, unity-power-factor EV on-board charger, this

chapter discusses several valuable techniques for designs in an EV on-board battery

charger. They include:

� Selection guidelines for power stage topologies for EV battery charger;

� Optimization of the charge profile;

� Efficiency improvement of the battery charger;

� Employment of a minimum-power-level active pre-load for ensuring that the

battery is being properly charged, and protecting the charger.


159

7.1 Power Stage Configurations for EV On-Board Battery

Chargers

Due to the PFC requirement, the two-stage approach, which has a PFC pre-converter

followed by a DC-DC converter becomes a natural choice. An input filter needs to be

included to prevent conducted harmonics and noise from entering the power supply

system. For safety considerations, the output of an EV battery charger must be

isolated from the mains. A typical block diagram of an EV on-board battery charger

is shown in Figure 7.1.

Galvanic isolation by a transformer provides a measure of safety and the design

of this transformer allows different power stage configurations or converter

selections. To increase the power density and lower material cost, the transformer in

the dc-dc converter is operated at high frequencies, usually in the range of 20 kHz to

100 kHz. In terms of connecting an EV to a battery charger, the conventional means

using a plug and receptacle is most suitable for an on-board battery charger, while an

inductive interface using a coaxial winding transformer (CWT) is more attractive for

a stand-alone type [109].

While the power level of a stand-alone type charger for EV fast charging would

be very high, up to hundreds of kilowatts, the power required for an on-board EV

battery charger is quite low, within a few kilowatts, limited by the maximum branch

circuit current rating in a residential house.

For the PFC stage, a soft-switching boost PFC circuit has been almost a standard

practice due to its simple circuitry, and high efficiency. For the DC-DC stage,

PFCPreconverter

D C /D CC onverter

L in eF ilterA C L ine

D C B u s B a tte ry

Figure 7.1 Block diagram of an EV on-board battery charger


160

however, many topologies can be considered as candidates. Among them, the most

attractive topologies are [110-112]:

(1) A soft-switched full-bridge (FB) DC-DC converter;

(2) An asymmetrically controlled zero-voltage-switched (ZVS) half-bridge (HB)

converter;

(3) An active-clamped soft-switched forward converter.

All these three DC-DC converters can achieve very high efficiency and very

good device utilization, due to employment of soft switching techniques. Further

selection of the DC-DC converter will depend on application specifications,

including power level, input line voltage, battery voltage, initial capital cost, long-

term operation expense, and some economics and business philosophy.

It should be noted that there are also some other configurations of the power

stage for EV battery chargers, such as a one-stage type charger [113] and an integral-

type charger [114, 115]. The one-stage type combines the PFC function and tight DC

output regulation. It reduces the number of the power components while its control is

more complicated and its performance is not necessarily better than the two-stage

type. An integral battery charger employs the hardware of inverters of the EV

propulsion system, reducing the component cost. The main drawback is the need to

provide a Ground Fault Current Interrupter (GFCI) on the charging cable line for

safety against electric shock due to the fact that the main battery and commercial

power source are not isolated from each other. These kinds of battery schemes have

not been as popular as the two-stage scheme since they are related to specific types of

converters or hardware and particular attentions need to be given to their control and

reliability, and safety of the consumers operating the equipment. However, they

present new concepts to implement EV battery chargers and, therefore, deserve

further interest in the development of EV battery chargers in the future.


161

7.2 An Example—A 780 W High-efficiency, Low-cost,

Unity-power-factor Battery Charger [15]

7.2.1 Design Specifications

� Output power: 780 W

� Output voltage: 35-43.5 V

� Input line voltage: 100-132 V/47-63Hz

� Input power factor: >0.95

� Maximum output current: 20 A

� Overall efficiency: >90% at full load and nominal line voltage

� Cooling: natural convection

� Low cost

The high efficiency requirement is critical to meeting the thermal design criteria,

since cooling relies on natural convection. This battery charger was originally

designed for battery charging in recreational vehicles. It is also suitable for battery

charging in electric bikes, electric scooters, and industrial material handling vehicles.

With some modification, mainly extension of power level, it can be used for battery

charging for commuter electric vehicles.

7.2.2 Power Stage and its Soft Switching Operation

The battery charger employs a zero-current-switched (ZCS) boost PFC converter and

a ZVS-HB dc-dc converter in cascade. The diagram and key operation waveforms of

the ZCS PFC converter are shown in Figure 7.2. The boost PFC converter is operated

in the critical conduction current mode and the rectifying diode switches at zero

current. As a result, switching loss and EMI associated with the reverse-recovery of

the diode, which are usually severe problems in hard-switched circuits, are

eliminated. Therefore, high efficiency at high frequency can be achieved and an

inexpensive rectifying diode can be used. This soft switching technique can be

classified into Boundary Switching ZCS, compared to those in Continuous Mode


162

V gs1

V in

i L 1

S 1o n

D 1o n

A CL ine S1

D 1

C 1

L 1

+

-

V bus

i L 1 i D 1

i S 1V inC in

Figure 7.2 Diagram and key waveforms of the ZCS boost PFC converter

converters / inverters, such as being discussed in Chapter 5. It is very attractive in this

application as it does not need any additional components to achieve soft switching

while most soft switching techniques involve considerably complicated auxiliary

circuits. The only cost is the need for some increase in the size of the input filter

inductor in order to keep line harmonic current small.

In this example, the 780 W power level played a key role in determining the

topology of the DC-DC converter. At this power level, the ZVS-FB converter is

apparently complex and costly for the applications due to the need for four power

switches and phase-shift control, although it is capable of providing excellent

efficiency (especially with well-regulated intermediate DC bus voltage). The forward

topology, on the other hand, is simpler and less costly. However, it may not offer an

efficiency advantage over the ZVS HB converter. In addition, the forward converter

requires a big output filter inductor whose dimensions are comparable to those of the

transformer.

A HB converter is much simpler than a FB converter. The conventional HB-

PWM DC-DC converter operates with symmetrical duty cycle control and the

switches are hard switched. Consequently the efficiency is low. To improve the

efficiency of the converter, ZVS operation of a HB-PWM converter can be achieved


163

V o

T 1

T

D 2 on

V b

V 2

V g s3

V g s2

i L f

D 3 on

S3

S2

T r

L 2

C2

V bu s

V a

D 2

D 3

C 3V 2

V o

+

-

i D 2

i D 3

i L 2

Figure 7.3 Diagram and key waveforms of the ZVS HB converter

by using asymmetrical duty cycle control [110]. Figure 7.3 shows the key waveforms

of the ZVS HB DC-DC converter. The two switches are turned on and off

complementarily with a small dead time in between. In this way, ZVS of the switches

is easily achieved by utilizing the energy stored in the leakage and magnetizing

inductance of the transformer.

Before S3 is on, the reflected secondary current and the magnetizing current of

the transformer are flowing into the dot of the primary winding. When S3 is turned

off, the output capacitance of S2 is charged, and the capacitance of S3 is discharged

by the energy stored in the leakage inductance and the magnetizing inductance, until

the anti-parallel diode of S3 is on. Then S3 is turned on under zero voltage condition.

The ZVS mechanism for S2 is similar [115].

As seen above, the soft-switched power stage of the on-board battery charger is

very simple, with no need of additional components which are usually required by

most soft switching schemes. This soft-switched battery charger scheme can be

especially attractive for ultra-small on-board chargers in a battery management

system. According to [116], it has been recently found that longevity of series-

connected lead-acid batteries is influenced by the differences between individual


164

batteries caused by charging. A battery management system was developed to solve

this problem. A thin and compact measurement module that can monitor the voltage,

current and temperature of the batteries was installed. Each module is for up to three

batteries. The information provided by these modules is analyzed by the battery

management system and the optimal charging is done by ultra-small on-board

chargers that were installed in each group of the batteries according to the individual

charge situations such as the voltage and temperature. In such an application, since

multiple on-board chargers are needed, they must be ultra-small while maintain high

performance. Soft switching is obviously desirable for the high efficiency, high

switching frequency, and consequently, small size of the chargers. On the other hand,

it is also desirable that the achievement of the soft-switching operation does not

significantly increase the circuitry complexity. The proposed soft-switched on-board

battery charger can suit this application satisfactorily due to its high performance and

simple circuitry.

7.3 Optimization of the Charge Profile

Battery performance and lifetime are extremely sensitive to charging conditions,

particularly in cyclic applications such as EVs. Therefore, charge schemes must be

optimized. There are some critical charging requirements in an EV environment,

which affect the design of an on-board battery charger. For most types of battery, the

charging requirements include:

(1) Avoiding overcharge;

(2) Avoiding undercharge;

(3) Reducing charging time without affecting the battery lifetime;

(4) Maintaining good quality of the charging current waveform.

Both overcharge and undercharge will make a battery fail prematurely. In EV

applications, undercharging is generally more likely than overcharging. Furthermore,

the results of undercharge appear much faster than overcharge. Therefore, the issue of

undercharge is of critical importance in EV applications.


165

One of effective methods of recharging a battery is using a single value constant

voltage with the highest possible current limit, known as constant voltage (CV)

mode. However, as state of charge (SOC) of the battery gets higher, the charge

current gets smaller. Therefore the time required for fully charging the battery is long.

An alternative charge scheme is to use a constant current (CC) to accomplish the

recharge. While this method usually completes the charging process in a shorter time

period compared to CV charge method, greater care must be taken to control the

charging. Unlike the CV method where the battery itself regulates the charge current

at any point of the charge cycle, the CC method continues to inject a set current

regardless of the battery's state of charge.

In many instances, a combination of CC and CV charge is employed. This can

cut down the charge time significantly without increasing the chances of damaging

the battery permanently. The charger starts working in CC mode and continues until

the battery voltage reaches its peak. This serves as a signal to trigger the charger to

switch to a CV mode.

Charging a battery in CC mode when the battery voltage gets close to its peak

requires the highest power from the charger during the charging process. To reduce

the maximum power requirement of the charger while maintaining the fast charge

feature, a further modified charge algorithm is to insert a constant power (CP) mode

between the CC mode and the CV mode [15], shown in Figure 7.4. This CC/CP/CV

charge algorithm was successfully implemented in the battery charger we have

developed. The CC/CP/CV charge algorithm optimizes the charger in reducing

charge time and minimizing the maximum power requirement of the charger. In other

words, the CC/CP/CV charge algorithm allows utilizing the maximum capabilities of

the converter and circuit size, while providing a high charging speed.

It should be noted that for some types of battery, a full recharge must put in a

few more ampere-hours, for example 2% to 5% additional ampere-hours over what

was taken out on the previous discharge, to obtain maximum lifetime. In addition, for

different applications, charging parameters may vary.


166

4 3 .5 V B atteryV o ltage

3 9 V3 5 V

2 0 A

C h arg in gC u rren t

C h arg in gP ow er

C C M od e C P M od e

C V M od e

M ax im u m P ow er L im it

Figure 7.4 Optimized charge profile

7.4 A Novel Active Pre-Load with Minimized Power Rate

for the On-Board Battery Charger

A novel active pre-load with minimized power rate has been designed for the on-

board battery charger. The employment of the active pre-load ensures that the battery

will be fully charged and provides effective low-load or even open-load protection to

the charger, as described below.

No matter which charge algorithm is being used, CV or CC/CV or CC/CP/CV, a

complete recharge process always ends in CV mode. As stated earlier, ensuring that

the battery will be fully charged is very beneficial. However, simply leaving the

charger on until the end of the charge may cause a stability problem in the charger.

This is because the end of charge current (EOCC) can be extremely small for some

kinds of battery and operating the charger at such a light load will threaten the

charger's safety. When the battery is nearly fully charged the HB DC-DC converter

will operate in discontinuous current mode (DCM) with very small duty cycle. This

increases the dc bus voltage, and the voltage stress on one of the secondary rectifying

diodes due to asymmetrical duty cycle control, and could also cause a circuit stability

problem and damage to the power devices.


167

Some battery designs shut down the chargers before the batteries are fully

charged, which is not good for the lifetime of the battery. Using an active pre-load is

a satisfactory solution. When the battery current drops below a critical level, the

active pre-load automatically starts. One basic requirement for this active pre-load is

that its rated power should be as small as possible, providing the DC-DC converter

can work properly. Conventional design of such an active pre-load usually requires

extra load current sensing. On the one hand, this current sensor should be very

accurate and insensitive to noise at very low-load current. On the other hand, it

should not be saturated and should not dissipate much power at very high-load

current. These design requirements make the conventional active pre-loads

complicated and costly.

A novel active pre-load circuit has been proposed with minimized power rate, as

shown in Figure 7.5 [15]. This active pre-load can automatically start without the

need for a current sensing circuit, which is usually complicated and costly. Its

operation principle is briefly described in the following. When the battery current

becomes extremely low, the pulse voltage of point A increases significantly due to

the asymmetrical drive of the two switches of the HB DC-DC converter. At a critical

load current level, this pulse voltage is high enough to turn on transistor Q. Capacitor

C is used to hold this voltage to keep transistor Q on, providing the converter a

D2

D3

Tr L2

C3

Battery

Q

D R C

R

D

R

R

A

activepre-load

. ..

C

z

b

D

Z

Figure 7.5 Circuit of the active pre-load


168

resistive load Rc. C can also eliminate noise to avoid turning on Q falsely. The

breakdown voltage of the zener diode, DZ, is chosen to set up the minimum battery

current where the active pre-load automatically starts. A Darlington transistor is used

as Q to reduce the base drive current. Since the active pre-load works only in the

interval T1 instead of the whole switching period T (referred to Figure 7.3), the rated

power, and consequently, the size and the cost of the active pre-load are significantly

reduced.

With the employment of this active pre-load, the charger worked very well at the

end of charge, ensuring that the battery would be fully charged. There is no danger of

overcharge even if it is left connected to the mains for long periods without

supervision. Repeated tests have also shown that it provided open load protection for

the charger. Even if a careless user removes the battery before he turns off the

charger, or turns on the charger without connecting batteries to the charger, the safety

of the charger need not be a concern.

7.5 Constant Steady-State Duty Cycle Control of the DC-

DC Converter

Unlike a general DC supply application where a constant output voltage is required,

the output voltage of a battery charger varies with the battery's state of charge. In the

two-stage scheme, if the DC bus voltage is fixed, the minimum steady-state duty

cycle of the HB DC-DC converter can be as small as 20%. Such a small duty cycle

will severely reduce the efficiency and increase current ripple. To optimize the

performance of the converter, it is highly desirable to operate the ZVS-HB-PWM

converter at close to 45% duty cycle. This is explained below.

As the duty cycle of the HB converter falls, the magnetizing current of the

transformer increases to maintain charge balance of the DC blocking capacitor in

series with the transformer primary winding. This, in turn, demands that the

transformer core is properly gapped and this further increases the magnetizing

current. As a result, the conduction loss of the HB converter increases significantly.

In addition, the size of the output filter inductor becomes much larger as the duty

cycle decreases. When the duty cycle of the HB converter is less than 30% (break


169

point), the overall performance of the HB converter becomes worse than that of the

forward converter. Therefore, it is essential to keep the duty cycle of the HB

converter at close to 45% over the entire input and output voltage range to optimize

the performance of the battery charger. This is also true when other types of DC-DC

converter are employed.

To solve the problem, the PFC stage output voltage was regulated proportional

to the battery voltage in this design. As a result, the steady-state duty cycle of the

ZVS-HB DC-DC converter was kept close to 45% in both CC mode and CP mode,

and the performance of the converter were optimized. Figure 8.6 shows the diagram

of principle of the control. When using a commercial PFC control chip, the PFC error

Amp and the reference of the PFC output voltage Vref are inside the chip and the

value of Vref is fixed. The sensed battery voltage VO is then added to the sensed PFC

stage output voltage Vbus at the voltage feedback input of the PFC controller. This is

equivalent to controlling the reference voltage of the error amplifier, Vref.

It is worthwhile to noting that the bus voltage regulation loop should be designed

to be very slow, with the crossover frequency being only a few Hz, since the battery

voltage changes very slowly. Fast response of this loop can cause a stability problem

in the system.

+

-

-1V o

V bus

C co m pR 1

R 2

R 3

V ref

P FC S tageE rro r A m p

Figure 7.6 Diagram of principle of constant steady-state duty cycle control

7.6 Measurements

A 780 W prototype of the on-board battery charger for a 36 volt battery was designed

and built, which employed a ZCS boost PFC pre-converter and an asymmetrically


170

controlled ZVS HB PWM converter. Later, a slight design modification was made

for 48 V and 24 V battery charging. Efficiency of each stage was around 95%. Over

the entire input and output voltage ranges, above 90% overall efficiency, as shown in

Figure 7.7.

Figure 7.8 shows the oscilloscope photograph of typical line voltage and line

current waveforms. It can be seen that both waveforms have good sinusoidal shape

and are well in phase. Greater than 0.995 input power factor have been achieved. The

input current harmonic distortion was limited to 3% typically.

The control parameters have been optimized and thus the operation of the

charger was very stable. The measurements of the loop gain of the dc-dc converter,

shown in Figure 7.9 and Figure 7.10, indicated a phase margin of 40� in the Constant

Current Charging Mode and an even larger phase margin, 89�, in the Constant

Voltage Charging mode, respectively. The difference of the phase margin in the

different charging modes was because of the different control planes in the CC mode

and CV mode. The similarity was seen in the analysis of control characteristics of the

boost converter in CV mode and CC mode in Chapter 4. All the protection functions

worked satisfactorily, including open load protection, due to the employment of the

active pre-load.

80828486

889092

35 36 37 38 39 40 41 42 43 44 45

Battery Voltage in charging (V)

Effi

cien

cy (

%)

Figure 7.7 Overall efficiency measurement


171

Figure 7.8 Oscilloscope photograph of typical input line voltage (upper trace)and input line current (lower trace) waveforms

Figure 7.9 Loop gain measurement in constant current charging mode


172

Figure 7.10 Loop gain measurement in constant voltage charging mode

7.7 Review

Guidelines for the selection of a suitable power stage topology, together with a

suitable soft-switching technique in EV on-board battery chargers is given. An

optimized charge profile is presented which prolongs the battery life while

maintaining the feature of fast charge. The steady-state duty cycle of the charger's dc-

dc converter is kept close to 45% to help achieve high efficiency in the charger. A

novel active pre-load with minimized power rate ensures that the battery will be fully

charged and also protects the charger on open load. The value of these techniques has

been verified by measurement results from a constructed on-board battery charger for

EV applications, which employs a ZCS boost PFC converter and an asymmetrically

controlled ZVS-HB dc-dc converter in cascade.

173

Chapter 8

Conclusions and Recommendations forFurther Work

8.1 Conclusions

The following summarizes the technologies developed for EV propulsion presented

in this thesis.

(1) Energy and Power Management and Control Strategy in the

NTCER EV (Chapter2, 4)

In BEV development, the battery remains the main barrier to success. The challenge

is that design choices for energy density, power, lifetime, weight, volume and cost of

the battery are all compromises. Accordingly, a novel energy and power management

scheme has been proposed for the NTCER EV. The proposed energy and power

management scheme employs the newly developed non-flowing zinc-Bromine

battery (NFZBB) and ultracapacitors (UCs). NFZBBs are optimized for specific

energy, low cost and long life, and ultracapacitors feature high power density and

long life and are used to complement the NFZBB to achieve good acceleration

performance at low speeds. Combining the two different types of energy storage can

optimize the EV propulsion system to achieve high energy for drive range, high

power for good acceleration performance, long lifetime, and low cost.

CHAPTER 8 CONCLUSIONS AND RECOMMENDATIONS FOR FURTHER WORK 174

The energy and power management scheme proposed is based on the drive cycle

data obtained from an ICE meter-reading vehicle, comparison of results of the

obtained data and FUDS, and the analysis of the NTCER EV propulsion system on

the drive cycles. Measurements show that the ratio of peak power to average power is

as high as 6.83:1, and most of the accelerations from rest are up to speeds of 35km/h

or less. Employing ultracapacitors to complement the NFZBB to achieve good

acceleration performance at low vehicle speeds load levels the EV battery.

Combining the short-term power assist scheme with the constant power control

scheme above the motor base speed gives the overall power control characteristic of

the EV propulsion system. The resulting maximum torque – speed envelope from the

overall power control scheme is similar to that of an internal combustion engine plus

transmission. That is, the propulsion system provides high torque at low speeds and

lower torque at high speeds with which human beings are familiar.

Since the NTCER EV profile is similar to that for a typical urban EV, the

NTCER EV research is applicable to general EVs.

(2) Several Technologies in BDCM Control (Chapter 3)

Chapter 3 discusses several advanced control techniques for BDCMs, with emphasis

on EV applications.

Firstly, the impact of PWM strategies and commutation schemes on the

performance of BDCMs is explored. A comparison between bipolar voltage

switching PWM and unipolar voltage switching PWM is made and selection criteria

are given. Two solutions are proposed to the phase current distortion problem of

BDCMs in regeneration mode. One is a combination of unipolar voltage switching



control scheme in terms of efficiency, rectangularity of phase current waveforms and

circuitry simplicity.



of the excitation transition points, mechanically varying the air gap of the motor and





base speed, when using a boost converter for speed extension of the BDCMs, are

studied. Control system stability analysis in the s-plane when using the two different

methods is presented. A clear conclusion is arrived at.

Finally, the control of regeneration current of BDCMs when the motor speed is

above the base speed is discussed. This is a well-known difficult issue in BDCM

control. An effective control method is proposed.

(3) A Multifunctional DC-DC Converter Topology and the

Implementation of Soft Switching in bidirectional Operation

(Chapter 4, 5)


control scheme at speeds above the base speed each require a bidirectional dc-dc

converter. Instead of using two separate dc-dc converters which will add considerable

complexity and cost to the propulsion system, a novel multifunctional dc-dc

converter has been proposed. It adds only three components to a simple conventional

bidirectional dc-dc converter and can implement all the functions required by the

overall energy/power management scheme. The principle of operation and control

rules of the multifunctional dc-dc converter are presented in detail.

After an evaluation of various soft-switching techniques, the zero voltage

transition (ZVT) soft switching technique is chosen for the multifunctional dc-dc

converter. The ZVT dc-dc converter has minimum switching loss resulting in high

efficiency, reduced electromagnetic interference (EMI) and improved reliability for

the power devices in the dc-dc converter. Another important feature of the ZVT dc-

dc converter is that no additional switching voltage and current stress are imposed on

the power devices. Simulation results are presented and verified by the measurement

results of the built ZVT dc-dc converter. The measured efficiency of the ZVT dc-dc

converter is above 96%.


(4) A Novel Design for High Power Ferrite Cored Inductor with

Ease of Design and Improved Overall Performance (Chapter 5)

At high power levels, due to the large dimensions of ferrite cores and air gaps, the


significant and is difficult to accurately predict. With conventional design of high

power ferrite cored inductors, difficulty exists in achieving a satisfactory overall

performance. A novel design for high power ferrite cored inductors employing a

dual-coil or distributed winding scheme is proposed by the author to combat the

problems. Results from both 3-Dimensional (3D) finite element analysis (FEA) and

measurement verify the viability of the novel design for high power inductors. The

proposed inductor design allows better optimization by reducing the localization of

saturation and leakage flux, or EMI levels, achieving good utilization of the magnetic

cores.

(5) A New Intelligent IGBT Drive Module for High Power IGBTs

(Chapter 6)

Existing commercial IGBT drivers are restricted to certain applications and are

subject to one or more of the following limitations:

� Peak current no larger than 2 amperes

� Single driver

� No intelligent functions such as overcurrent or desaturation protection

� No fault status annunciation

� Unipolar drive outputs

� Need for external isolated dc supply

� Inconvenient to use.

A new intelligent gate drive module for high power IGBTs has been developed

and presented in Chapter 6. The intelligent functions of this module include under

voltage lockout protection and overcurrent protection with fault status annunciation.

The module contains dual drivers which make it especially suitable for phase leg

drive, and outputs bipolar signals so that the “off” status of the IGBT is ensured when


the IGBT is required to be “off”. It needs only a unipolar dc supply due to the built-

in, well-regulated, and isolated floating power supply. This module features high

current capacity, powerful functions, high performance, wide application, and

convenience of use, and is suitable for high power applications such as in EV drives.

The detailed functions, operation, performance, and practical application issues are

addressed.

(6) Several Technologies in the Soft-Switched, Unity-Power-Factor

Battery Charger (Chapter 7)

In the development of a soft-switched on-board EV battery charger with unity power

factor, which employs a ZCS boost PFC converter and an asymmetrically controlled

ZVS-HB dc-dc converter in cascade, several state-of-the-art technologies are

presented and verified.

Firstly guidelines for the selection of a suitable power stage topology, together

with a suitable soft-switching technique in EV on-board battery chargers is given.

Secondly, the charge profile is optimized by inserting a constant-power mode

between the constant-current mode and the constant-voltage mode. The optimized

charge profile prolongs the battery life while maintaining the feature of fast charge.

Thirdly, the steady-state duty cycle of the charger's dc-dc converter is kept close to

45% to help achieve high efficiency in the charger. Finally, a novel active pre-load

with minimized power rate ensures that the battery will be fully charged and also

protects the charger on open load.

The value of these techniques has been verified by measurement results from a

constructed EV on-board battery charger.

8.2 Recommendations for Further Work

The following research and development is recommended for further work.

� Modifying the main inductor in the dc-dc converter using Litz wire for

optimization of efficiency and ease of manufacture when finalizing the NTCER

EV propulsion design and construction [98].


� Developing user-defined control algorithms, such as the control of the charging

status and the charging /discharging current of the ultracapacitor bank according

to vehicle speeds, command of the motor current, etc. This can optimize the

propulsion system to achieve further improvement in efficiency, battery lifetime,

and drive range.

� Testing the completed system, including the NFZBB batteries, the ultracapacitor

bank and the high-power inverter.

� Increasing the power by employing a battery bank with higher voltage, and

IGBTs to replace the MOSFETs in the power electronics controllers.

179

Appendix A

The Non-Flowing Zinc/BromineBattery for Electric VehicleApplications

Courtesy of Bjorn Jonshagen [63]

APPENDIX A 180

The Non-Flowing Zinc/Bromine Battery for Electric VehicleApplications

Bjorn JonshagenZBB Energy Corporation, 16 Emerald Tce, West Perth, Western Australia

Phone: +61-08-93109380, Fax: +61-08-93109381, e-mail: [email protected]

Abstract: The Non-Flowing Zinc/Bromine Battery technology is a promising spin-off from work on the flowing electrolyte Zinc/Bromine Battery, by ZBB EnergyCorporation. The new Non-Flowing Zinc/Bromine Battery (NFZBB) is ideally suitedto Electric Vehicle (EV) applications. The advantages offered by the NFZBB are:

� High energy density for a long EV driving range;� High power density for good EV acceleration;� Capacity to withstand several thousand full discharge cycles without damage so

that the batteries will last for the life of the vehicle; and� Low cost production.

The paper, the first publication about this new technology, outlines the constructionof the battery; it’s performance characteristics and the most recent test results. Thetechnology uses multiple cell, bipolar electrode “battery stacks” that can have sixty ormore cells giving individual batteries with voltages of over one hundred volts. Highvoltage vehicle power systems give the advantage of reduced weights and dimensionsfor the power control, wiring and motor systems. The battery has a simpleconstruction with no electrolyte circulation. A single cell prototype, currently on lifecycle test, has passed the 3,300 cycles mark and still achieves very goodperformance. During this test the battery is fully discharged to 1.0 volt per cell ateach cycle.

Keywords: Battery, bike, bi-polar, cycle life, energy density, energy storage, powerdensity, safety, vehicle performance, zinc bromine.

The Chemistry

The Non-Flowing Zinc/Bromine Battery (NFZBB) uses an aqueous electrolytecontaining Zinc Bromide salt. During charge, zinc is electroplated on the anode sideof each bipolar electrode, and bromine is evolved at the cathode side. A microporousseparator divides the two cell halves. A complexing agent in the electrolyte is usedto reduce the mobility, reactivity and vapour pressure of the elemental bromine byforming a polybromide complex. On discharge, zinc is oxidised to zinc ions, and onthe cathodes, the bromine is reduced to bromide ions.

The electrochemical reactions during charge are given as follows:

APPENDIX A 181

Overall: ZnBr2 � Zn + Br2Anode: Zn2+ + 2e - � ZnCathode: 2Br - � Br2 + 2e -

Bromine Complex: QBr - + nBr2 � Q(Br2)nBr -

QBr - = Complexing Agent

Construction and Materials

In a multiple cell bipolar electrode battery stack (Figure 1) only the two endelectrodes have terminal connections. The bipolar electrodes (e) that divide twobattery cells are not accessed from the outside and form the positive electrode(cathode, +) in one cell and the negative electrode (anode, -) in the adjoining cell.Each cell is divided in two cell halves by the separator (s).

Figure 1: Bipolar Stack Build (four cells only drawn)

The bipolar electrode design increases the specific energy and the voltage of thebattery. Multiple cell stacks with sixty or more cells can be built producingindividual batteries with voltages of over one hundred volts. High voltage vehiclepower systems give the advantage of reduced weights and dimensions for the powercontrol, wiring and motor systems.

The NFZBB battery stack consists nearly entirely of low cost polyethylene plasticmaterials. Only a thin metal screen on the back side of the terminal electrodes isnecessary to direct the electrical current in the x-y plane of the electrode. The plasticelectrodes contain carbon for electrical conductivity and glass fibres to minimisewarpage. The separators are made from microporous silica filled polyethylene. Eachelectrode and separator is fitted into a plastic frame. Alternating electrode andseparator frames are then joined together between plastic endblocks to form ahermetically sealed battery stack.

APPENDIX A 182

Polyethylene can be formed and joined using common industrial techniques such asinjection moulding, extrusion and heat welding techniques. All these techniques lendthemselves to highly automated, low cost mass production. The electrolyte is basedon Zinc Bromide salt solution, which is commonly available at low costs. The lowcost of both materials and manufacturing ensures that a low cost battery can beproduced.

Cycle Life

Figure 2 represents the results from continuous life cycle testing of a single cellbattery with a 400 cm2 active electrode area (NF400A-104-01). The test involvesrepeated and continued cycling in an automated test station. Each discharge is deepto near zero state of charge (SOC) when the remaining cell voltage is only one volt.Every 100 cycles a complete “zinc strip” is carried out. This involves discharging allthe way down to zero terminal voltage and leaving the battery short circuited until theopen cell voltage is approximately zero. With this procedure all the plated zinc isremoved from the anode leaving a completely clean electrode substrate for platingzinc during the next charge cycle. After the total removal of zinc and bromine, a fewcycles are required before the full coulombic efficiency is returned.

0

20

40

60

80

100

0 500 1000 1500 2000 2500 3000 3500

EE%VE%CE%

Cycle Number

Eff

eci

en

cy %

NF400-A104-01

Figure 2: Cycle Plot for Cell NF400A-104-01

APPENDIX A 183

Data is logged every 30 seconds and cycle efficiencies are computed for each cycle.The columbic efficiency (CE) represents the capacity discharged from the battery as apercentage of the amp hour charge input. The voltaic efficiency (VE) is the ratiobetween average discharge and charge voltages. The total energy efficiency (EE) foreach cycle is the ratio of DC energy out to DC energy in. The life cycle testing isaccelerated by carrying out several cycles per day. The cycling is carried out at aconstant ambient temperature of 30 �C. Higher return energy efficiency has beenachieved at lower temperatures.

NF400A-104-01 is still performing at 70 % energy efficiency after 3,300 cycles to adischarge voltage of 1.0 volts per cell. This represents nine years of daily cycling ofa vehicle battery. The test is continuing. The cycle life of the Non FlowingZinc/Bromine Battery is not restricted by 100% depth of discharge cycling.

Cycle Performance

Four identical 16 cell prototype batteries labelled NF400A-130-16, NF400A-131-16,NF400A-132-16 and NF400A-133-16 have been built. Figure 3 is a photo of one ofthe batteries. The external dimensions are length: 290 mm, width: 140 mm, height:255 mm.. The batteries are built using the same assembly techniques as was used forthe single cell NF400A-104-01. All cell components and materials, except theseparators, are also identical to that used for NF400A-104-01. The separator materialcomes from an alternative supplier and a separate test series has proved that thismembrane gives an increased coulombic efficiency.

The four 24V nominal (the open cell voltage at fully charged state is around 28 volts)batteries use a secondary containment system consisting of a rotation mouldedpolyethylene box and lid. The space between the box and the cell stack componentsis filled with casting epoxy.

Figure 3: 24 Volt Prototype Battery NF400A-130-16

APPENDIX A 184

The four prototype batteries have been cycled individually on the laboratory benchusing the baseline cycle regime. The cycle performance for the first eight cycles foreach battery is plotted in Figure 4. The cycles are discharged to 16 volts (1 volt percell). As can be seen, the performance is very even between the batteries and thereturn energy efficiency is at a very high level. The first cycle is marginally lowersince no bromine is present prior to the first cycle. On average all the batteriesdeliver a return DC to DC energy efficiency of over 85 %.

80

85

90

95

100

0 1 2 3 4 5 6 7 8 9 10

NF400-A133-16NF400-A132-16NF400-A131-16NF400-A130-16

EE

VE

CE

CYCLE NUMBER

%E

FF

ICIE

NC

IES

Figure 4: Laboratory Cycle Test Performance for the Four Prototype Batteries

APPENDIX A 185

16000

18000

20000

22000

24000

26000

28000

30000

0 1 2 3-4000

-3000

-2000

-1000

0

1000

2000

3000

4000

CURRENTVOLTAGE

TIME/ hr

VO

LTA

GE

/ m

V

CU

RR

EN

T/ m

A

CE (%) VE (%) EE (%)97 90 87

Figure 5: Laboratory Test Cycle #2, Battery NF400A-130-16

Cycle number 2 for battery NF400A-130-16 is plotted in Figure 5 where the constantcurrent charge and discharge characteristics of the system are clearly displayed. Forthis individual cycle the battery returns 87% of the DC charge energy at the pointwhen the discharge voltage reaches 16 Volts.

Three of the prototype batteries have been charged and discharged while connectedtogether in parallel. The batteries were found to share the current evenly indicatingthat scaled up systems can be built by connecting a number of batteries in parallel aswell as by increasing the number of cells in a battery.

Peak Power Density

Voltage measurements on one of the prototype batteries (NF400A-133-16) have beentaken in the first few seconds of discharge at currents up to 80 Amps (Figure 6). Thecurrent was sustained for 60 seconds and the voltage recorded at 30 and 60 seconds.At 80 Amps the voltage was dropping rapidly and the test was terminated after thefirst reading (~3 seconds).

The 50 Amp readings were high, possibly because these readings were taken first,and full recovery was not achieved before the 40 and 60 Ampere tests wereperformed. However, it is still clear that the fully charged battery can deliver around700 Watts for 30 seconds (Figure 7).

APPENDIX A 186

DISCHARGE VOLTAGE NF400A-133-16

0

5

10

15

20

25

30

0 10 20 30 40 50 60 70 80 90

CURRENT [A]

VO

LTA

GE

[V]

After ~3 Sec

After 30 Sec

After 60 Sec

Figure 6: Start of discharge Voltages for Prototype Battery NF400A-133-16

30 second power output

0200400600800

1000

0 20 40 60 80

Current (A)

Pow

er (

W)

Current Voltage Power(Amps) (Volts) (Watts)30 19.7 59140 16.9 67650 15.7 78560 11.3 678

Figure 7: Power output from prototype battery NF400A-133-16

APPENDIX A 187

Energy Density

Prototype batteries have been cycled at a charge loading that return an energy densityin excess of 45 Wh/kg based on the weight of the electrolyte only. Even at thischarge loading, only about 50% of the active chemicals available in the electrolyteare utilised. The most important factor, apart from electrolyte utilisation, is theweight of battery hardware components. The lighter the dry battery can be built, thehigher the discharge energy density.

A new generation of prototypes, with an 800 cm2 active area, is currently beingdeveloped. These batteries are put together in a different way and the hardwarecomponents weigh proportionally less compared with the 400 cm2 generationprototypes. The first eight cell prototype of this new generation is currently on test inthe laboratory delivering in excess of 25 Wh/kg under continuous deep dischargecycling conditions. A battery unit with higher number of cells (higher voltage) candeliver a higher effective energy density since the proportion of “dry weight” to totalweight is less when more cells are put together in one unit. It is estimated that 32 cell,48 volt prototypes using the new hardware will have a practical total dischargeenergy density in excess of 30 Wh/kg.

ZBB Energy Corporation is currently carrying out a research and developmentprogram aimed at increasing the energy density and the power density further. Thisis done by utilising the available active chemicals further and by increasing theconcentration of active chemicals available. Key components of the battery such asthe bromine storage agents, the positive side substrate and the battery separator arealso given particular attention. Table 1 outlines the projected performance from 800cm2 batteries built in 24, 48 and 96 volt configuration under continuous deepdischarge cycling conditions

Table 1: 800 cm2 Non-Flow Zinc/Bromine Battery Projected Performance

EstimatedWeight

# ofcells

Voltage Energy Power(continuous)

Power(30 sec.)

total Electr.

EstimatedEnergyDensity

[V] [kWh] [W] [W] [kg] [kg] [Wh/kg]16 24 1.0 800 2,000 20 12 5032 48 2.0 1,600 4,000 34 24 6064 96 4.0 3,200 8,000 62 48 64

Vehicle Test

The 400 cm2 prototype batteries have been tested on a purpose built bicycle with a48Volt hub motor in the front wheel. The bicycle utilises two series connected24Volt batteries fitted on the rear baggage carrier (Figure 8). The bicycle has good

APPENDIX A 188

performance characteristics with ample power to go up even steep hills withoutpedalling. The top speed, on electric power alone, on level ground is over 30 km/h.

A road test has been carried out on a 6 km test loop that involves some smaller hillsand tight corners requiring acceleration and deceleration. The fully charged electricbicycle was driven, without any pedalling at an average speed of 20 km/h around thetest loop. The bicycle was stopped a few times to change rider. The batteries wereflat after 3 hours and 16 minutes when the tricycle had travelled 66 km.

.

Figure 8: 48 Volt Electric Bicycle

Safety

The technology is considered safe, since neither the materials used in cell builds northe electrolytes are classified as toxic. When a battery is charged the corrosiveelement bromine is produced in the positive cell halves. However, once the batteryhas been discharged, only a very low concentration of bromine remains. The concernwith bromine is vapour inhalation and skin and eye contact. Several levels ofprotection have been implemented to ensure that human contact with bromine willnot occur.

Firstly, the bromine produced during charge is comlexed to form a polybromide“second phase” bromine storage compound. This compound holds virtually all thebromine produced and reduces the vapour pressure of free bromine (150 mmHg @20�C) to a very low level (~3 mmHg @ 20�C) thus limiting the amount of brominereleased to the atmosphere if a cell is ruptured.

Secondly, the polybromide second phase is organic and separates out from theaqueous electrolyte and is soaked into the positive substrate used. This will reduceeven further the amount of bromine released in case of a cell rupture.

APPENDIX A 189

Thirdly, using the bipolar cell build technique only small amounts of bromine areproduced per cell. The NFZBB technology utilises a large number of lower currentcapacity cells rather than a low number of high capacity cells as with the lead acidbattery. This considerably limits the amount of bromine released from a single cellrupture.

The secondary containment used for the prototype batteries provides a spill barriershould one of the cells develop a leak. The mass produced batteries will have asimilar secondary containment and be fully sealed so that no bromine can escape intothe environment. Any bromine containing electrolyte that is accidentally releasedcan be easily neutralised (for example with sodium carbonate) into harmlesscompounds that can be safely washed away with water.

APPENDIX B

190

Appendix B

Schematic Diagrams

This appendix includes schematic diagrams of the bidirectional ZVT dc-dc converter

for Phase I NTCER EV.

APPENDIX B 191

B1: Power block schematic

12

34

56

ABCD

65

43

21

D C B A

Tit

le

Num

ber

Re

visi

onS

ize

B Da

te:

6-D

ec-

2000

S

hee

t

of

File

:C

:\ya

n\..

\App

end

ix c

onve

rter

-pow

er.

sch

Dra

wn

By:

L1

40uH

L2

3uH

E1

Ba

tte

ry

Dx3

DSE

I 2X

101

Dx2

DSE

I 2X

101

Dy-

2D

SE

I 2X

101

Dx1

DS

EI

2X10

1

C3

47nF

/250

V

L3 3u

HD

y1

DSE

I 2X

101

SC

R11

1RK

180

D1

123N

Q10

0

S1

1XFN

180

N20

Sx

1XFN

180

N20

Sy

1XFN

180

N20

S2

1XFN

180

N20

C4

2.2

uF/2

50V

C5

4700

uF

C6

4700

uF

1

2

3

4

CT

1L

A 2

05-S

C7

100n

F/25

0V

R1

330k

R3

4.7k

1234

J1 CO

N4

C9

0.22

uF/2

50V

+N

C -

m

12

J2 CO

N2

12

J3 CO

N2

12J4

CO

N2

12

J5 CO

N2

11.

0

Xin

xia

ng Y

an

Sch

em

ati

c of

Pow

er

Sta

ge

SC

R G

SC

R K

Vbu

s

R2

300k

R4

2.4k

Vba

tt

C8

0.68

uF

40uH

ind

ucto

rou

tsid

e t

he b

oard

Out

put

-

Out

put

+

Rxx

4.7

(2W

)

Dxx

MU

R 8

60 (

3 in

pa

ralle

l)

Ryy

4.7

(2W

)D

yy

MU

R86

0 (3

in

para

llel)

APPENDIX B 192

B2: Gate driver schematic

12

34

56

ABCD

65

43

21

D C B A

Tit

le

Num

ber

Re

visi

onS

ize

B Da

te:

6-D

ec-

2000

S

hee

t

of

File

:C

:\ya

n\T

hesi

s\A

ppe

ndix

FE

T d

rive

rs.s

chD

raw

n B

y:

R1

680

C1

1uF

2

7

8 5

3

U1

HP

221

1

1

8

27 6

5

4

U5

4422

C2

1uF

Ga

te d

rive

r fo

r a

uxi

buck

sw

itch

MO

SFE

T D

rive

rs

21.

0

Xin

xia

ng Y

an

Ga

te d

rive

r fo

r m

ain

Buc

k sw

itch

C8

0.1u

F

C12

0.47

uF+12V

+15V

(boo

st)

C13

0.47

uF

R2

680

C3

1uF

2

7

8 5

3

U2

HP

221

1

1

8

27 6

5

4

U6

4422

C4

1uF

C9

0.1u

F

C14

0.47

uF+12V

+15V

(buc

k-m

ain

)

C15

0.47

uF

Ga

te d

rive

rs f

or m

ain

Boo

st s

wit

ch

R3

680

2

7

8 5

3

U3

HP

221

1

1

8

27 6

5

4

U7

4422

C5

1uF

+15V

(boo

st)

C10

0.1u

FC

160.

47uF

R4

680

2

7

8 5

3U

4

HP

221

1

1

8

27 6

5

4

U8

4422

C6

1uF

C11

0.1u

FC

170.

47uF

123

J1 CO

N3

123

J2 CO

N3

12

J3 CO

N2

1 2

GS

1

CO

N2-

FET

Vin

2

GN

D1

+Vou

t4

0V3

U9

NM

L12

15S

R5

100k

R9

2.7 R

13

6.8

D1

1N58

59

ZD

115

V

1 2

GS

2

CO

N2-

FET

R6

100k

R10

2.7 R

14

6.8

D2

1N58

59

ZD

215

V

Vin

2

GN

D1

+Vou

t4

0V3

U10

NM

L12

15S

1 2

GS

3

CO

N2-

FET

R7

100k

R11

2.7 R

15

6.8

D3

1N58

59

ZD

315

V

C7

1uF

C18

0.47

uF+12V

+15V

(buc

k-a

uxi)

Vin

2

GN

D1

+Vou

t4

0V3

U11

NM

L12

15S

1 2

GS

4

CO

N2-

FET

R8

100k

R12

2.7 R

16

6.8

D4

1N58

59

ZD

415

V

Ga

te d

rive

rs f

or a

uxi

Boo

st s

wit

ch

R17

4.7k

R18

4.7k

R19

4.7k

APPENDIX B 193

B3: Control Schematic #1

12

34

56

ABCD

65

43

21

D C B A

Tit

le

Num

ber

Re

visi

onS

ize

B Da

te:

6-D

ec-

2000

S

hee

t

of

File

:C

:\ya

n\..

\App

end

ix m

ast

er

cont

rol.

sch

Dra

wn

By:

Ma

ste

r C

ontr

ol B

oard

Vbu

sV

batt

Vba

tt-m

Vbu

s-m

volt

age

se

nsor

svo

lta

ge s

ens

ors.

sch

Vbu

sV

batt

Vba

tt-m

Vbu

s-m

Im-c

md

Ve

loci

ty

Vba

tt-m

SC

R d

rive

Dri

ving

, a

bove

Nb

Re

gen,

abo

ve N

b

cont

rol

sign

al

cont

rol

sign

al.

sch

Im-c

md

Ve

loci

ty

Vba

tt-m

SC

R d

rive

Dri

ving

, a

bove

Nb

Re

gen,

abo

ve N

b

Ire

fV

bus-

mD

rivi

ng,

abo

ve N

b

Re

gen,

abo

ve N

b

curr

ent

re

fcu

rre

nt r

ef.s

ch

Ire

fV

bus-

mD

rivi

ng,

abo

ve N

b

Re

gen,

abo

ve N

b

Boo

st p

erm

itte

dB

uck

perm

itte

d

Vba

tt-m

Nor

ma

l V

batt

& c

urre

nt

I-in

d-m

Dri

ving

, a

bove

Nb

Re

gen,

abo

ve N

b

Dis

abl

ed

sign

als

Dis

abl

ed

sign

als

.sch

Boo

st p

erm

itte

dB

uck

perm

itte

d

Vba

tt-m

Nor

ma

l V

batt

& c

urre

nt

I-in

d-m

Dri

ving

, a

bove

Nb

Re

gen,

abo

ve N

b

From

Inve

rte

r

1 2 3 4 5 6

J3 MO

LE

X6

From

Pow

er

Blo

ck

+15V

-15V C

10.

1uF

C2

0.1u

F

Nor

ma

l V

batt

& c

urre

ntB

oost

pe

rmit

ted

Buc

k pe

rmit

ted

Ga

te b

uck

aux

i

Ga

te b

oost

aux

iG

ate

boo

st m

ain

Ire

f

Ga

te b

uck

ma

in

I-in

d-m

I-in

d

PW

M U

C38

23A

PW

M U

C38

23A

.sch

Nor

ma

l V

batt

& c

urre

ntB

oost

pe

rmit

ted

Buc

k pe

rmit

ted

Ga

te b

uck

aux

i

Ga

te b

oost

aux

iG

ate

boo

st m

ain

Ire

f

Ga

te b

uck

ma

in

I-in

d-m

I-in

d

To

Dri

ver

Boa

rd

-15V

+15V

C38

10uF

C39

10uF

C40

10uF

C41

10uF

C3

0.1u

F

C4

0.1u

FC

50.

1uF

C6

0.1u

F

1 2 3 4 5 6J6 MO

LE

X6

SC

R G

SC

R K

SC

R d

rive

conv

ert

er-

driv

ers

2co

nve

rte

r-dr

ive

rs2.

sch

SC

R G

SC

R K

SC

R d

rive

C43

10uF

+12V

(unr

eg)

12345678

J8 MO

LE

X8

1234

J1 MO

LE

X4

+12V

Vin

1

GND2

+5V

3

Re

g1L

M78

05

+5V

+12V

Vin

1

GND2

+12V

3

Re

g2L

M78

12

+12V

+15V

3

Xin

xia

ng Y

an

1.0

APPENDIX B 194


12

34

56

ABCD

65

43

21

D C B A

Tit

le

Num

ber

Re

visi

onS

ize

B Da

te:

6-D

ec-

200

0 S

hee

t

of

File

:C

:\ya

n\.

.\A

ppe

ndi

x vo

lta

ge s

ens

ors.

sch

Dra

wn

By:

Vbu

s

3 21

4 11

U33

A

MC

3317

4P

278

1

54

63

U30

LO

C11

1

R1

10k

+5V

R62

33k

P1

100k

R2 10

k

R67

470

C8

0.1u

F

3 21

4 11

U29

A

MC

3317

4P

C9

0.1u

F

+12V

(se

nsor

)

Vba

tt

3 21

4 11

U13

A

MC

3317

4P

278

1

54

63

U32

LO

C11

1

R4

10k

R64 33

k

P310

0k

R69

470

Vba

tt-m

R5

10k

+5V

10 98

U29

C

MC

331

74P

+12V

(se

nsor

)

Vbu

s-m

VO

LT

AG

E S

EN

SO

RS

4

Xin

xia

ng Y

an

C48

0.47

uFC

490.

47uF

C50

1.0u

F+1

2V

Vin

1

GND2

+12V

3

U28

MC

7812

CO

M5

Vs

1V

out+

6

GN

D2

SY

NC

in14

Vou

t-7

SY

NC

out

8

U27 D

CP

0115

15D

P

+12V

(se

nsor

)

C11

0.47

uF

C7

0.22

uF

C12

0.22

uF

F1 200m

A

C17

0.1u

F

1.0

APPENDIX B 195


12

34

56

ABCD

65

43

21

D C B A

Tit

le

Num

ber

Re

visi

onS

ize

B Da

te:

7-D

ec-

2000

S

hee

t

of

File

:C

:\ya

n\..

\App

end

ix c

ontr

ol s

igna

l.sc

hD

raw

n B

y:

Im-c

md

Vel

ocit

y

12 1314

U13

D

MC

3317

4P

R12

10k

R13

10k

R74 10

0k9 10

8

U15

C

SN

74L

S08

Vba

tt-m

10 98

U13

C

MC

3317

4PP

7 10k

Xin

xia

ng Y

an

3 21

4 11

U12

A

MC

3317

4PR

1610

k

R17

10k

R75

390k

1 23

14 7U15

A

SN

74L

S08

SC

R d

rive

4 56

U15

B

SN

74L

S08

Dri

ving

, a

bove

Nb

Re

gen,

abo

ve N

b

CO

NT

RO

L S

IGN

AL

S

+5V

C13

0.1u

F

+5V

C14

0.1u

F

5

C16

0.1u

F

+5V

1 23

14 7

U14

A

SN

74L

S00

4 56

U14

B

SN

74L

S00

ZD

32.

5V

+5V

5 67

U12

BM

C33

174P

5 67

U13

B

MC

3317

4P

R7

2.2k

+5V

ZD

42.

5V

C15

10nF

C10

0.1u

F

C?

0.22

uF

C?

0.22

uF

1.0

APPENDIX B 196


12

34

56

ABCD

65

43

21

D C B A

Tit

le

Num

ber

Re

visi

onS

ize

B Da

te:

7-D

ec-

2000

S

hee

t

of

File

:C

:\ya

n\T

hesi

s\A

pend

ix c

urre

nt r

ef.

sch

Dra

wn

By:

+5V

R92

22k

R28

22k

R98

22k

C54

100n

F

C55

10nF

P11

10k

5 67

U33

B

MC

3317

4P

+5V

D1

1N41

48

R31

10k

+5V

R32

10k

CU

RR

EN

T R

EFF

ER

EN

CE

1213

1

32

4

5

U10

A

DG

411

C52

0.1u

C53

0.1u+1

5V

Ire

f

10 98

U33

CM

C33

174P

Vbu

s-m

12 1314

U8D

MC

3317

4P

10 98

U8C

MC

3317

4PP

1410

k

P15

10k

+5V

R35

10k

8

67

U10

B

DG

411

Xin

xia

ng Y

an

Dri

ving

, a

bove

Nb

Re

gen,

abo

ve N

b

3 21

4 11

U8A

MC

3317

4P

R36

10k

R77

820k

R10

0

470

LE

D1

Gre

en

+5V

5 67

U8B

MC

3317

4P

R39

10k

R80

820k

R10

3

470

LE

D4

Gre

en

C19

0.1u

F

6

R9

22k

R3

10k

+5V

ZD

5Z

ET

3.3

V 1

%

R8

2.2k

1.0

C?

0.22

uF

C?

0.22

uF

C?

0.47

uF

APPENDIX B 197


12

34

56

ABCD

65

43

21

D C B A

Tit

le

Num

ber

Re

visi

onS

ize

B Da

te:

7-D

ec-

2000

S

hee

t

of

File

:C

:\ya

n\..

\App

end

ix D

isa

ble

d si

gna

ls.s

chD

raw

n B

y:

Boo

st p

erm

itte

d

Buc

k pe

rmit

ted

R44

10k

R81

820k

R10

4

470

LE

D5

Gre

en

R45

10k

R82

820k

R10

5

470

LE

D6

Gre

en

+5V

Vba

tt-m

10 98

U22

C

MC

3317

4PR

4610

k

R47

10k

P17

10k

+5V

R83

820k

Nor

ma

l V

batt

& c

urre

nt

DIS

AB

LE

D S

IGN

AL

S

I-in

d-m

C21

0.1u

F

+5V

R48

10k

R49

10k

P18

10k

+5V

R84

820k

R50

10kR

51

10k

P19

10k

+5V

R85

820k

PR

E4

CL

K3

D2

CL

R1

Q5

Q6

147

U23

AM

C74

F74

+5V

C22

0.1u

F

R89

100k

PB

1 SW

-PB

12 1314

U22

D

MC

3317

4P

R52

10k

R86

820k

R10

6

470

LE

D7

Gre

en

Ove

r cu

rre

ntR

ese

t

Xin

xia

ng Y

an

1 2 1312

14 7

U25

A

SN

74L

S11

3 4 56

U25

B

SN

74L

S11

Dri

ving

, a

bove

Nb

Re

gen,

abo

ve N

b

+5V

C23

0.1u

F

+5V

C24

0.1u

F

+5V

C26

0.1u

F

7

4 56

U17

B

SN

74L

S08

3 21

4 11

U22

A

MC

3317

4P

5 67

U22

B

MC

3317

4P

1 23

14 7

U19

A

SN

74L

S00

4 56

U19

B

SN

74L

S00

R10

22k

R6

10k+5

V

Nor

ma

l V

batt

& c

urre

nt

Buc

k pe

rmit

ted

Boo

st p

erm

itte

d

1 23

14 7

U17

A

SN

74L

S08

132J2 JU

MP

ER

3

132J4 JU

MP

ER

3

+5V

3 21

8 4

U1A

LM

358

5 67

U1B

LM

358

C18

0.1u

F

+5V

1.0

C?

0.22

uF

C?

0.22

uF

C?

0.22

uF

C?

0.22

uF

C?

0.22

uF

C?

0.22

uF

APPENDIX B 198


12

34

56

ABCD

65

43

21

D C B A

Tit

le

Num

ber

Re

visi

onS

ize

B Da

te:

6-D

ec-

2000

S

hee

t

of

File

:C

:\ya

n\T

hesi

s\A

ppen

dix

PW

M.s

chD

raw

n B

y:

+15V

C28

0.1u

F

R11

35.

1k

R10

71k

Q1

2N39

04

C61

1nF

R11

8

56Z

D1

2.5V

1%R

119

2.2k

R10

93.

3k

C29

0.1u

F

R99

10k C56

100n

F

C57

10nF C

66

10nF

R54 5k

R56

10k

R12

14.

7k

P21

10k

C67

330p

R12

02.

2k

R12

633

C30

100n

FR

128

56k

R11

4

10k

R57

10k

+5V

PW

M G

EN

ER

AT

ION

C68

100p

F

9 10 118

U25

C

SN

74L

S11

1 2 1312

14 7U7A

SN

74L

S11

Nor

ma

l V

bat

& c

urre

nt

Boo

st p

erm

itte

d

Buc

k pe

rmit

ted

Ga

te b

uck

aux

i

Ga

te b

oost

aux

i

Ga

te b

oost

ma

in1 2

3

14 7

U3A

SN

74L

S08

R11

01k

12 1314

U12

D

MC

3317

4P

Ire

f10 9

8U

12C

MC

3317

4P

Vcc15

ILIM

9

Vre

f16

Vc13

Out

A11

SS 8

Gnd10

Clo

ck4

NI

2

INV

1

E/A

out

3

Rt

5

Ct

6

Pw

r G

nd12

Out

B14

Ra

mp

7

U2

UC

3823

A

3 4 56

U7B

SN

74L

S11

Ga

te b

uck

ma

in

Xin

xia

ng

Ya

n

I-in

d-m

C33

0.1u

F

+5V

8

+5V

C35

0.1u

F

A1

B2

CL

R3

Cex

t14

Re

xt/C

ext

15

Q13

Q4

168

U5A

SN

74L

S12

3

C65

1nF

R12

3

4.7k

+5V

C36

0.1u

F

9 108

U3C

SN

74L

S08

9 108

U19

C

SN

74L

S00

R61

10K

I-in

d

C44

10uF

C25

1nF

C47

10uF

9 10 118

U7C

SN

74L

S11

+5V

A9

B10

CL

R11

Ce

xt6

Re

xt/C

ext

7

Q5

Q12

U5B

SN

74L

S12

3

C42

1nF

4 56

U3B

SN

74L

S08

1 23

14 7

U9A

SN

74L

S32

+5V

C34

0.1u

F

A1

B2

CL

R3

Cex

t14

Rex

t/C

ext

15

Q13

Q4

168

U6A

SN

74L

S12

3

C51

1nF

R23

4.7k

12 1311

U17

DS

N74

LS

08

C45

1nF

+5V

A9

B10

CL

R11

Cex

t6

Re

xt/C

ext

7

Q5

Q12

U6B

SN

74L

S12

3

C46

1nF

9 108

U17

C

SN

74L

S08

4 56

U9B

SN

74L

S32

C20

0.1u

F

+5V

R18

3.3k

R19

3.3k

R21

6.8k

R14

2.2k

D2

1N41

48

+5V

C31

1nF

+5V

R22

6.8k

+5V

C32

1nF

R15

2.2k

D3

1N41

48+5

V

R20

560k

C27

10nF

1.0

APPENDIX B 199


12

34

56

ABCD

65

43

21

D C B A

Tit

le

Num

ber

Re

visi

onS

ize

B Da

te:

6-D

ec-

2000

S

hee

t

of

File

:C

:\ya

n\T

hesi

s\A

ppe

ndi

x S

CR

dri

ver.

sch

Dra

wn

By:

R12

756

R12

4

4.7

K

+15V

16

34

T1

1:1

Q2

2N22

22

ZD

210

V Z

en

er

D6

1N40

07

R13

0

22R

131

220

D7

1N41

48

SC

R D

RIV

ER

91.

0

Xin

xia

ng Y

an

SC

R G

SC

R K

12 1311

U15

D

SN

74L

S08

SC

R d

rive

TR

IG2

Q3

R4

CV

olt

5T

HR

6

DIS

7

VCC8 GND1

U34

LM

555

C69

0.01

uF

+5V

C37

0.1u

F

R65

33k C72

10nF

R66

10k

R11

56k

9 108

U14

C

SN

74L

S00

+5V

200

Bibliography

[1] S. R. Shacket, The Complete Book of Electric Vehicles, Australia and New

Zealand Book Co Pty Ltd, 1979.

[2] M. Ehsani, K.M. Rahman, H.A. Toliyat, “Propulsion system design of electric

and hybrid vehicles,” IEEE Transactions on Industrial Electronics, vol. 44,

No.1, pp. 19-27, 1997.

[3] A. Sayigh, “Renewable energy - the way forward,” presented at 1999 World

Renewable Energy Congress, Perth, Australia, 1999.

[4] B. K. Bose, “Energy, environment, and advances in power electronics,” IEEE

Trans. on Power Electronics, vol. 15, pp. 688-701, 2000.

[5] B. J. Fleay, “World oil peaks next decade consequence for renewable

energy,” presented at World Renewable Energy Congress, Perth, Australia,

1999.

[6] DOE, “DOE Homepage, http//www.hev.doe.gov,” DOE (US Department of

Energy), 1999.

[7] A. B. Lovins, M. M. Brylawski, D. R. Cramer,T. C. Moore, “HYPERCARS:

materials, manufacturing, and policy implications,” March 1996.

[8] M. K. N. Hayashida, Yusgi Kamiya, “Energy regeneration of city commuter-

car by ultracapacitor and battery,” presented at EVS-16, 1999.

[9] W. Donitz, Karl E. Noreikat, “BEV, FCEV, HEV -- the same electric drive

train?,” presented at EVS-16, Beijing, 1999.

[10] V. Wouk, “Hybrids: then,” IEEE Spectrum, vol. 32, No. 7, 1995.

BIBLIOGRAPHY 201

[11] R. D. King, Kenneth B Haefner, Lembit Salasoo, Rudolph A Koegl, “Hybrid

electric transit bus pollutes less, conserves fuel,” IEEE Spectrum, 1995.

[12] B. Bates, “Getting a Ford HEV on the road,” IEEE Spectrum, 1995.

[13] K. Maruo, “Strategies alliances for the development of fuel cell vehicles,” in

AEVA Electric Vehicle News, vol. 150 March/April 1999. Australia

[14] Merceds-Benz, “Hydrogen-powered FCEV,” in Weekend Australia,

September. Sydney, 1997.

[15] X. Yan, Dean Patterson, “A high-efficiency on-board battery charger with

unity input power factor,” International Journal of Renewable Energy

Engineering, vol. 2, 2000.

[16] B. Kennedy, Dean Patterson, Xinxiang Yan, John Swenson, “Development of

a light short range electric commuter vehicle,” International Journal of

Renewable Energy Engineering, vol. 2, pp. 148-152, 2000.

[17] X. Yan, Dean Patterson, “Improvement of drive range, acceleration and

deceleration performance in an electric vehicle propulsion system,” presented

at The IEEE Power Electronics Specialist Conference (PESC'99), South

Carolina, USA, 1999.

[18] X. Yan, Dean Patterson, Byron Kennedy, “A multifunctional dc-dc converter

for an EV drive system,” presented at 16th International Electric Vehicle

Symposium and Exhibition (EVS-16), Beijing, China, 1999.

[19] X. Yan, Dean Patterson, “Novel power management for performance and cost

reduction in an electric vehicle,” International Journal of Renewable Energy,

vol.22 (1-3) pp. 177-183, 2001.

[20] X. Yan, Dean Patterson, Dan Sable, Guichao Hua, “Design of an intelligent

gate driver for high-power IGBT module,” presented at The IEEE

International Conference on Power Electronics Drives and Energy Systems

for Industrial Growth (PEDES'98), Perth, Australia, 1998.

[21] X. Yan, Dean Patterson, Dan Sable, Xinfu Zhuang, Guichao Hua, “Design of

power electronics for a satellite flywheel energy storage system,” presented at

The IEEE International Conference on Power Electronics Drives and Energy

Systems for Industrial Growth (PEDES'98), Perth, Australia, 1998.

BIBLIOGRAPHY 202

[22] D. Patterson, Xinxiang Yan, S. Camilleri, “A very high efficiency controller

for an axial flux permanent magnet wheel drive in a solar powered vehicle,”

presented at The IEEE International Conference on Power Electronics Drives

and Energy Systems for Industrial Growth (PEDES'98), Perth, Australia,

1998.

[23] P. Chandler, Xinxiang Yan, Dean Patterson, “Novel high power ferrite cored

inductor design with ease of design and improved overall performance,” 2001

IEEE Power Electronics Specialist Conference (PESC'01), 2001.

[24] DataElectronics, “Reference manual for the Datataker data loggers, Version

3.3,” 1993.

[25] J. Yang, “Electric drive system of electric vehicle,” presented at EVS-16,

1999.

[26] N. F. Iiinski, M. G. Bitchkov, “Switched reluctance drive for electric light

EVs,” presented at EVS-16, 1999.

[27] C. C. Chan, K.T. Chau, “An overview of power electronics in electric

vehicles,” IEEE Transactions on Industrial Electronics, vol. 44, No.1, pp. 3-

13, 1997.

[28] C. C. Chan, J.Z. Jiang, W. Xia, K.T. Chan, “Novel wide range speed control

of permanent magnet brushless motor drives,” IEEE Transactions on Power

Electronics, vol. 10, No.5, pp. 539-46, 1995.

[29] Q. An, Jiazhen Yuan, “Complexion and experience of research and

development of Beijing Brand EVs in series,” presented at EVS-16, 1999.

[30] A. M. Nasser, “Comparison of induction, PM, and SR motor technologies in

EV traction systems application,” presented at EVS-16, 1999.

[31] R. L. Hodkinson, “Recent advances in brushless dc motors and drives,”

presented at EVS-16, 1999.

[32] A. Sakai, Y. Sasaki, A. Otomo, F. Kawahata, A. Matui, “Toyota braking

system for hybrid vehicle with regenerative system,” presented at EVS-14,

Orlando, Florida, USA, 1997.

[33] D. Patterson, R. Spee, “The design and development of an axial flux

permanent magnet brushless DC motor for wheel drive in a solar powered

BIBLIOGRAPHY 203

vehicle,” IEEE Transactions on Industry Applications, vol. 31, pp. 1054-61,

1995.

[34] X. Jiang, Jiang Dan, Dongqi Zhu, Yuande Hu, Sixiang Liu, “The optimization

of permanent magnet brushless motor drive system for electric vehicle,”


[35] H. Weh, H Mosebach, H May, “Design concepts and force generation in

inverter-fed synchronous machines with permanent magnet excitation,” IEEE

Trans. on Magnetics, vol. 20, pp. 1756-1761, 1984.

[36] P. J. McCleer, J. M. Bailey, J. S. Lawler, B. Banerjee, “Five phase trapezoidal

back EMF PM synchronous machines and drives,” presented at 1991

European Power Electronics Conference, Florence, Italy, 1991.

[37] S. Ogasawara, H. Akagi, “An approach to position sensorless drive for

brushless DC motors,” IEEE Transactions on Industry Applications, vol. 27,

pp. 928-33, 1991.

[38] T. Furuhashi, S. Sangwongwanich, S. Okuma, “A position-and-velocity

sensorless control for brushless DC motors using an adaptive sliding mode

observer,” IEEE Transactions on Industrial Electronics, vol. 39, pp. 89-95,

1992.

[39] A. Consoli, S. Musumeci, A. Raciti, A. Testa, “Sensorless vector and speed

control of brushless motor drives,” IEEE Transactions on Industrial

Electronics, vol. 41, pp. 91-6, 1994.

[40] C. S. Berendsen, G. Champenois, A. Bolopion, “Commutation strategies for

brushless DC motors: influence on instant torque,” IEEE Transactions on

Power Electronics, vol. 8, pp. 231- 6, 1993.

[41] N. H. Rashidi, “Improved and less load dependent three-phase current-

controlled inverter with hysteresis current controllers,” IEEE Transactions on

Industrial Electronics, vol. 42, pp. 325-30, 1995.

[42] A. Tripathi, P. C. Sen, “Comparative analysis of fixed and sinusoidal band

hysteresis current controllers for voltage source inverters,” IEEE

Transactions on Industrial Electronics, vol. 39, pp. 63-73, 1992.

BIBLIOGRAPHY 204

[43] M. Kheraluwala, D. M. Divan, “Delta modulation strategies for resonant link

inverters,” presented at the 18th Annual IEEE Power Electronics Specialist

Conference (PESC'87), USA, 1987.

[44] B. K. Bose, “An adaptive hysteresis-band current control technique of a

voltage-fed PWM inverter for machine drive system,” IEEE Transactions on

Industrial Electronics, vol. 37, pp. 402-8, 1990.

[45] B. K. Bose, Power Electronics and Variable Frequency Drives, Technology

And Applications. New York: IEEE Press, 1996.

[46] X. Yan, Yanggang Yan, “Techniques of current control for brushless dc

motors,” Journal of Small and Special Motors, 1992.

[47] W. Leonhard, Control of Electrical Drives, 2nd edition. New York: Springer,

1996.

[48] H. Grotstollen, J Wiesing, “Torque capability and control of a saturated

induction motor over a wide range of flux weakening,” IEEE Trans. Ind.

Electron., vol. 42, pp. 374-381, 1995.

[49] S. HIti, Constantin C. Stancu, James M. Nagashima, “Propulsion system

control for next generation of EV1,” presented at EVS-16, Beijing, 1999.

[50] B. Sneyers, D. W. Novotny, T. A. Lipo, “Field weakening in Buried

permanent magnet ac motor drives,” IEEE Trans. on Ind. Applicat., vol. 21,

pp. 398-407, 1985.

[51] T. M. Jahns, “Flux-weakening regime operation of an interior permanent-

magnet synchronous motor drive,” IEEE Trans. on Ind. Applicat., vol. 23, pp.

681-689, 1987.

[52] B. K. Bose, P. M Szczesny, “A microcomputer-based control and simulation

of an advanced IPM synchronous machine drive system for electric vehicle

propulsion,” IEEE Trans. on Industrial Electronics, vol. 35, pp. 547, 1988.

[53] B. K. Bose, P. M. Szczesny, “A high performance inverter-fed drive system

of an interior permanent magnet synchronous machine,” IEEE Trans. on

Industrial Applications, vol. 24, pp. 987, 1988.

BIBLIOGRAPHY 205

[54] D. J. Patterson, “Recent advances in the design and construction of axial flux

permanent magnet machines,” presented at IEAust National Engineering

Conference, Australia, 1996.

[55] T. M. Jahns, “Torque production in permanent magnet synchronous motor

drives with rectangular current excitation,” IEEE Trans. Indus. Appl., vol. 20,

pp. pp. 803-813, 1984.

[56] J. Ofori-Tenkorong, “Permanent-magnet synchronous motors and associated

power electronics for direct-drive vehicle propulsion,” Ph.D. thesis,

Department of Electrical Engineering, Massachusetts Institute of

Technology, 1996

[57] N. Demerdash, T. Nehi, “Dynamic modeling of brushless DC motor -- power

condition unit for electromechanical actuator application,” presented at

PESC'79, 1979.

[58] I. Takahashi, Su G Koganezawa, K Oyyama, “A super high speed PM motor

drive system by a quasi-current source inverter,” presented at IEEE Ind. Appl.

Soc. Annual Meeting, Toronto, 1993.

[59] B. Beihoff, “A survey of torque transduction methodologies for industrial

applications,” presented at 1996 Annual Pulp and Paper Industry Technical

Conference, Birmingham, AL, USA, 1996.

[60] C. C. Chan, E. W. C. Lo, Weixiang Shen, “An overview of battery technology

in electric vehicles,” presented at EVS-16, 1999.

[61] T. Shinpo, H. Suzuki, “Development of battery management system for

electric vehicle,” presented at EVS-14, Orlando, Florida, USA, 1997.

[62] A. F. Burke, J. E. Hardin, E. J. Dowgiallo, “Application of ultracapacitors in

electric vehicle propulsion systems,” presented at the 34th International

Power Sources Symposium, USA, 1990.

[63] B. Jonshagen, “The non-flowing Zinc/Bromine battery for electric vehicle

application,” presented at EVS-16, Beijing, 1999.

[64] G. L. Hunt, “The great battery search,” IEEE Spectrum, November, 1998.

BIBLIOGRAPHY 206

[65] P. Mestre, A. Astier, “Application of ultracapacitors and influence of the

drive control strategies on the performance of an electric vehicle,” presented

at EVS-15, Brussels, 1998.

[66] A. F. Burke, J E Hardin, E J Dowgiallo, “application of ultracapacitors in

electric vehicle propulsion system,” presented at the 34th International Power

Sources Symposium, Cherry Hill, NJ, USA, 1990.

[67] A. F. Burke, S. Somasundaram, “Electric/hybrid super car designs using

ultracapacitors,” presented at the 30th Intersociety Energy Conversion

Engineering Conference, USA, 1995.

[68] P.-A. Gath, J. Falt, “Compact capacitor bank improves electric vehicle drive

performance,” Power Conversion & Intelligent Motion, 1996.

[69] N. Kasuga, Nouhito Ohnuma, Sakar Takagi, Junichi Murakami, Hideki Mori,

Shozo Hashizume, Kiyomi Yamasaki, “Ultracapacitor and battery hybrid EV

with high efficiency battery load leveling system,” presented at EVS-15,

Brussels, 1998.

[70] F. Caricchi, F. Crescimbini, F. Giulii, L. Solero, “Ultracapacitors employment

in supply systes for EV motor drives: theoretical study and experimental

results,” presented at EVS-14, Orlando, Florida, USA, 1997.

[71] R. A. Sutula, Kenneth L. Heitner, Susan A. Rogers, Tien Q. Duong, Pandit G.

Patil, S. R. Venkateswaran, “Driving towards a viable future for electric and

hybrid vehicles: U.S. Department of Energy (DOE) Storage R&D Programs,”

presented at EVS-14, Orlando, Florida, USA, 1997.

[72] R. L. Hochney, C A Driscoll, “Power of standby power supplies using

flywheel energy storage,” presented at IEEE INTENEC'97, Melbourne,

Australia, 1997.

[73] P. P. Acarnley, Barrie C Mecrow, James S Burdess, J. Neville Fawcett, James

G Kelly, Phillip G Dickinson, “Design principles for a flywheel energy store

for road vehicles,” IEEE Tran. on Industry Applications, vol. 32, pp. 105-109,

1996.

BIBLIOGRAPHY 207

[74] R. B. Wright, T. C. Murphy, W. E. Kramer, R. A. Sutula, “Experimental

electrochemical capacitor test results,” presented at EVS-14, Orlando,

Florida, USA, 1997.

[75] T. C. Murphy, P. B. Davis, “Update on ultracapacitor development for

transportation applications,” presented at the Symposium on Electrochemical

Capacitors, USA, 1996.

[76] N. Kasuga, Hiroko Yoshikawa, Nobuhito Ohnuma, “High efficiency battery

load leveling EV,” presented at EVS-16, 1999.

[77] H. Kahlem, “Energy and power from supercapacitors and electrochemical

sources,” presented at EVS-16, 1999.

[78] J. Takehara, Kuniaki Miyaoka, Tetsuo Fukuda, “EV mini-van featuring series

conjunction of ultracapacitors and batteries for load leveling of its batteries,”


[79] F. Brucchi, M. Conte, F Giulii Caapponi, G Lo Bianco, P. Salvati, L. Solero,

“Ultracapacitor test for EV applications: introduction of new equalization

coefficients,” presented at EVS-16, 1999.

[80] J. McClurkin, “Ultracapacitor model and efficiency,” personal

communication, 1998.

[81] A. M. Tuckey, D. J. Patterson, “A new resonant dc link/boost converter

topology applied to extended speed operation of a brushless dc machine,”

presented at 1998 IEEE Applied Power Electronics Conference, Anaheim,

CA, USA, 1998.

[82] IEEE, “Electromagnetic Compatibility Standards Collection,” IEEE, 1996.

[83] IEC, “Electromagnetic Compatibility Standards, IEC 61000 Series,” IEC,

2000.

[84] C.-J. Tseng, Chern-Lin Chen, “A passive lossless snubber cell for non-

isolated PWM dc-dc converter,” IEEE Transactions on Industrial

Electronics, vol. 44, No. 4, pp. 593-601, 1998.

[85] K. M. Smith Jr., K. M.Smedley, “Engineering design of lossless passive soft

switching methods for PWM converters. I. with minimum voltage stress

circuit cells,” presented at APEC'98, Anaheim, California, USA, 1998.

BIBLIOGRAPHY 208

[86] H. Levy, Isaac Zafrany, Gregory Ivensky, Sam Ben-Yaakov, “Analysis and

evaluation of a lossless turn-on snubber,” presented at APEC'97, 1997.

[87] K. M. Smith Jr., K.M. Smedley, “Engineering design of lossless passive soft

switching method for PWM converters. II. Non-minimum voltage stress

circuit cells,” presented at The International Telecommunications Energy

Conference (INTELEC'98), 1998.

[88] D. M. Divan, G. Skibinski, “Zero switching loss inverters for high power

applications,” presented at IEEE Industry Applications Society Annual

Meeting, USA, 1987.

[89] D. M. Divan, “The resonant DC link converter-a new concept in static power

conversion,” presented at IEEE Industry Applications Society Annual

Meeting, USA, 1986.

[90] D. M. Divan, G. Venkataramanan, R. W. A. A. DeDoncker, “Design

methodologies for soft switched inverters,” IEEE Transactions on Industry

Applications, vol. 29, pp. 126-35, 1993.

[91] R. Streit, Daniel Tollik, “High efficiency telecom rectifier using a novel soft-

switched boost-based input current shaper,” presented at The 13th

International Telecommunication Energy Conference (INTELEC'91), 1991.

[92] S. Ben-Yaakov, Gregory Ivensky, Oleg Levitin, Alexander Treiner,

“Optimization of the auxiliary switch components in a flying capacitor ZVS

converter,” presented at APEC'95, Dallas, TX, USA, 1995.

[93] G. Hua, F. C. Lee, “Soft-switching techniques in PWM converters,” IEEE

Transactions on Industrial Electronics, vol. 42, pp. 595-603, 1995.

[94] M. K. Jr. Smith, Keyue M. Smedley, “A comparison of voltage-mode soft-

switching methods for PWM converter,” IEEE Transactions on Power

Electronics, vol. 12, No. 2, pp. 376-86, 1997.

[95] C. Jung-Goo, Ju-Won Baek, Geun-Hie Rim, Iouri Kang, “Novel zero-voltage-

transition PWM multiphase converters,” IEEE Transactions on Power

Electronics, vol. 13, pp. 152-9, 1998.

[96] C. M. d. O. Stein, Helio L. Hey, “A true ZCZVS commutation cell for PWM

converters,” presented at APEC'98, Anaheim, California, USA, 1998.

BIBLIOGRAPHY 209

[97] J. G. Kassakian, Schlecht, F. Martin, Geroge C. Verghese, Principles of

Power Electronics: Addison-Wesley Publishing Company, 1991.

[98] A. M. Tuckey, D.J. Patterson, “A minimum loss inductor design for an active

clamped resonant dc link inverter,” presented at IAS'00, Rome, Italy, 2000.

[99] A. M. Tuckey, “Resonant Link inverter for trapezoidal flux electrically

commutated machines,” Ph.D. thesis, School of Engineering, Faculty of

Technology, Northern Territory University, Darwin, Australia, 2000.

[100] R. S. Chokhawala, Jamie Catt, Laszlo Kiraly, “A discussion on IGBT short-

circuit behavior and fault protection schemes,” IEEE Tran. on Industry

Applic., vol. 31, pp. 256-263, 1995.

[101] C. Licitra, Salvatore Musumeci, Angelo Raciti, Agostino U Galluzzo, Romeo

Letor, Maurizio Melito, “A new driving circuit for IGBT devices,” IEEE

Tran. on Power Electronics, vol. 10, pp. 373-378, 1995.

[102] L. Letor, M Melito, “Safe behavior of IGBTs submitted to a dv/dt,” presented

at Power Conversion Conference, 1990.

[103] R. S. Chokhawala, Jamie Catt Pelly, “Gate drive considerations for IGBT

module,” IEEE Transactions on Industry Applications, vol. 31, No 3, pp.

603-611, 1995.

[104] K. Chen, Thomas A Stuart, “A 1.6-kW, 110-kHz dc-dc converter optimized

for IGBTs,” IEEE Transactions on Power Electronics, vol. 8, pp. 18-25,

1993.

[105] HP, “Preliminary Technology Data of HCPL-316J,” Hewlett Packard

Cooperation, USA, USA November 1996.

[106] Unitrode, Product & Applications Handbook, 1995-1996, USA, Unitrode

Integrated Circuits Corporation (UICC), 1996.

[107] F. Koyanagi, Y. Uriu, “Modeling power consumption by electric vehicles and

its impact on power demand,” Electrical Engineering in Japan, vol. 120, No

4, pp. 40-47, 1997.

[108] D. F. Gosden, Geoffrey W. Crapps, Don Johnson, ERDC 248 Electric Vehicle

Development, Energy Research and Development Corporation, 1994, GPO

Box 629 Canberra ACT 2601 Australia

BIBLIOGRAPHY 210

[109] K. W. Klontz, A. Esser, P.J. Wolfs, D. M. Divan, “Converter selection for

electric vehicle charger systems with a high-frequency high-power link,”

presented at PESC'93, Seattle, Washington USA, 1993.

[110] T. Ninomiya, Norio Matsumoto, Masatoshi Nakahara, Koosuku Harada,

“Static and analysis of zero-voltage-switched half-bridge converter with

PWM control,” presented at ?, 1991.

[111] J. Zhang, C Y Lin, X Zhuang, K Rinne, D Sable, G Hua, F Lee, “Design of a

4kw on-board battery charger for electric vehicle,” presented at Annual VPEC

Seminar, Blacksburg, VA, USA, September 1995.

[112] W. Andreycak, “active clamp and reset technique enhances forward converter

performance,” presented at Unitrode Power Supply Design Seminar, USA,

1994.

[113] Y. M. Jiang, F C Lee, G Hua, W Tang, “a novel single-phase power factor

correction scheme,” presented at APEC'93, San Diego, CA, USA, 1993.

[114] S.-K. Sul, Sang-Joon Lee, “An integral battery charger for four-wheel drive

electric vehicle,” IEEE Tran. on Industry Applications, Vol. 31, No 5, 1995.

[115] T. Ishikawa, Toshiyuki Sekimori, Yutaka Hotta, Akira Suzuki, “Development

of a traction inverter with charging function,” presented at EVS-14, Orlando,

Florida, USA, 1997.

[116] H. Shimizu, Kiyomoto Kawakami, Yoshinori Kondo, “Evaluation and future

development of small EV "Luciole",” presented at EVS-14, Orlando, Florida,

USA, 1997.

[117] B. K. Bose, Power Electronics and AC Drives, Englewood Cliffs, NJ:

Prentice Hall, 1986.

[118] Brain J. Seibel, Timothy M. Rowan, and Russel J. Kerkman, "Field-oriented

control of an induction machine in the field-weakening region with dc-link

and load disturbance rejection", IEEE Tran. on Industry Applications, Vol.

33, No. 6, 1997.

[119] Jung-IK and Seung-Ki Sul, "Sensorless field-orientation control of an

induction machine by high-frequency signal injection", IEEE Tran. on

Industry Applications, Vol. 35, No. 1, 1999.

BIBLIOGRAPHY 211

[120] Alfio Consoli, Salvatore Musumeci, Angelo Raciti, and Antonio Testa,

"Sensorless vector and speed control of brushless motor drive", IEEE Tran.

On Industrial Electronics, Vol. 41, No. 1, 1994.

Documents

Control Strategies and Power Electronics in a Low-cost ...6350/Thesis_CDU_6350_Yan_X.… · Control Strategies and Power Electronics in a Low-cost Electric Vehicle Propulsion System