zabihi sasan flexible high voltage pulsed power...

Flexible High Voltage Pulsed Power Supply

for Plasma Applications

A Thesis by Publication submitted in

Partial Fulfilment of the Requirement for the

Degree of

Doctor of Philosophy

Sasan Zabihi Sheykhrajeh

M.Sc, B.Eng (Electrical Engineering)

Faculty of Built Environment and Engineering

School of Engineering Systems

Queensland University of Technology

Queensland, Australia

2011

III

Acknowledgment

I would like to take this opportunity to convey my appreciation for the help and

support received during the period of this study.

First of all, I would like to express my deepest gratitude to my supervisor,

Professor Firuz Zare, who truly made a difference to my academic perspective.

With his support, encouragement and brilliant advice throughout my PhD

program, I developed a personal interest in the area of power electronics. His

strong capability of conducting projects and creating interesting and novel

scenarios were much appreciated. His honesty, responsibility, and hard work

were also very promising for me and inspired our collaborative relationship.

I also wish to acknowledge the support of my Associate Supervisors Professor

Arindam Ghosh and Professor Gerard Ledwich. It was an honor for me to work

with such a great and prestigious supervisory team.

I would like to convey my sincere thanks to Queensland University of

Technology (QUT) for providing me with a pleasant research area and laboratory

facilities. I gratefully acknowledge the Australian Research council for financial

assistance throughout my research via the ARC Discovery Grant. The assistance

of Laboratory Technicians and the staff of the Research Portfolio were also

greatly appreciated.

I am indebted to many of my colleagues and friends at the Power Engineering

Group for their encouragement, sharing knowledge and for providing a warm

and conductive research atmosphere. I will never forget my time there studying

and researching, I will likewise cherish the pleasant times I had with my friends

during soccer matches and other social events.

I would also like to thank my family for the gracious support they have provided

me throughout my entire life. I must acknowledge my parents who have always

been for me. Without their love, encouragement and assistance; I would not have

achieved this much in my life. My mother, who has dedicated all her life and

scarified all her ambitions for me. My father, who stayed with me throughout my

life and supported me with his precious, thoughtful and heartwarming advice.

They both deserve the best and I wish I was able to compensate a small portion

IV

of their compassion. I also wish to acknowledge my beloved sister Nasim, who

always encouraged me during the hard times and was always on my side. I will

always admire and appreciate her sense of compassion. Thank you to one and all.

V

Abstract

Demands for delivering high instantaneous power in a compressed form (pulse

shape) have widely increased during recent decades. The flexible shapes with

variable pulse specifications offered by pulsed power have made it a practical

and effective supply method for an extensive range of applications. In particular,

the release of basic subatomic particles (i.e. electron, proton and neutron) in an

atom (ionization process) and the synthesizing of molecules to form ions or other

molecules are among those reactions that necessitate large amount of

instantaneous power. In addition to the decomposition process, there have

recently been requests for pulsed power in other areas such as in the combination

of molecules (i.e. fusion, material joining), gessoes radiations (i.e. electron

beams, laser, and radar), explosions (i.e. concrete recycling), wastewater,

exhausted gas, and material surface treatments. These pulses are widely

employed in the silent discharge process in all types of materials (including gas,

fluid and solid); in some cases, to form the plasma and consequently accelerate

the associated process.

Due to this fast growing demand for pulsed power in industrial and

environmental applications, the exigency of having more efficient and flexible

pulse modulators is now receiving greater consideration. Sensitive applications,

such as plasma fusion and laser guns also require more precisely produced

repetitive pulses with a higher quality. Many research studies are being

conducted in different areas that need a flexible pulse modulator to vary pulse

features to investigate the influence of these variations on the application. In

addition, there is the need to prevent the waste of a considerable amount of

energy caused by the arc phenomena that frequently occur after the plasma

process. The control over power flow during the supply process is a critical skill

that enables the pulse supply to halt the supply process at any stage.

VI

Different pulse modulators which utilise different accumulation techniques

including Marx Generators (MG), Magnetic Pulse Compressors (MPC), Pulse

Forming Networks (PFN) and Multistage Blumlein Lines (MBL) are currently

employed to supply a wide range of applications. Gas/Magnetic switching

technologies (such as spark gap and hydrogen thyratron) have conventionally

been used as switching devices in pulse modulator structures because of their

high voltage ratings and considerably low rising times. However, they also suffer

from serious drawbacks such as, their low efficiency, reliability and repetition

rate, and also their short life span. Being bulky, heavy and expensive are the

other disadvantages associated with these devices. Recently developed solid-

state switching technology is an appropriate substitution for these switching

devices due to the benefits they bring to the pulse supplies. Besides being

compact, efficient, reasonable and reliable, and having a long life span, their high

frequency switching skill allows repetitive operation of pulsed power supply.

The main concerns in using solid-state transistors are the voltage rating and the

rising time of available switches that, in some cases, cannot satisfy the

application’s requirements. However, there are several power electronics

configurations and techniques that make solid-state utilisation feasible for high

voltage pulse generation. Therefore, the design and development of novel

methods and topologies with higher efficiency and flexibility for pulsed power

generators have been considered as the main scope of this research work. This

aim is pursued through several innovative proposals that can be classified under

the following two principal objectives.

• To innovate and develop novel solid-state based topologies for pulsed

power generation

VII

• To improve available technologies that have the potential to

accommodate solid-state technology by revising, reconfiguring and

adjusting their structure and control algorithms.

The quest to distinguish novel topologies for a proper pulsed power production

was begun with a deep and through review of conventional pulse generators and

useful power electronics topologies. As a result of this study, it appears that

efficiency and flexibility are the most significant demands of plasma applications

that have not been met by state-of-the-art methods. Many solid-state based

configurations were considered and simulated in order to evaluate their potential

to be utilised in the pulsed power area. Parts of this literature review are

documented in Chapter 1 of this thesis.

Current source topologies demonstrate valuable advantages in supplying the

loads with capacitive characteristics such as plasma applications. To investigate

the influence of switching transients associated with solid-state devices on rise

time of pulses, simulation based studies have been undertaken. A variable

current source is considered to pump different current levels to a capacitive load,

and it was evident that dissimilar dv/dts are produced at the output. Thereby,

transient effects on pulse rising time are denied regarding the evidence acquired

from this examination. A detailed report of this study is given in Chapter 6 of

this thesis.

This study inspired the design of a solid-state based topology that take advantage

of both current and voltage sources. A series of switch-resistor-capacitor units at

the output splits the produced voltage to lower levels, so it can be shared by the

switches. A smart but complicated switching strategy is also designed to

discharge the residual energy after each supply cycle. To prevent reverse power

flow and to reduce the complexity of the control algorithm in this system, the

resistors in common paths of units are substituted with diode rectifiers (switch-

VIII

diode-capacitor). This modification not only gives the feasibility of stopping the

load supply process to the supplier at any stage (and consequently saving

energy), but also enables the converter to operate in a two-stroke mode with

asymmetrical capacitors. The components’ determination and exchanging energy

calculations are accomplished with respect to application specifications and

demands. Both topologies were simply modelled and simulation studies have

been carried out with the simplified models. Experimental assessments were also

executed on implemented hardware and the approaches verified the initial

analysis. Reports on details of both converters are thoroughly discussed in

Chapters 2 and 3 of the thesis.

Conventional MGs have been recently modified to use solid-state transistors (i.e.

Insulated gate bipolar transistors) instead of magnetic/gas switching devices.

Resistive insulators previously used in their structures are substituted by diode

rectifiers to adjust MGs for a proper voltage sharing. However, despite utilizing

solid-state technology in MGs configurations, further design and control

amendments can still be made to achieve an improved performance with fewer

components. Considering a number of charging techniques, resonant

phenomenon is adopted in a proposal to charge the capacitors. In addition to

charging the capacitors at twice the input voltage, triggering switches at the

moment at which the conducted current through switches is zero significantly

reduces the switching losses. Another configuration is also introduced in this

research for Marx topology based on commutation circuits that use a current

source to charge the capacitors. According to this design, diode-capacitor units,

each including two Marx stages, are connected in cascade through solid-state

devices and aggregate the voltages across the capacitors to produce a high

voltage pulse. The polarity of voltage across one capacitor in each unit is

reversed in an intermediate mode by connecting the commutation circuit to the

IX

capacitor. The insulation of input side from load side is provided in this topology

by disconnecting the load from the current source during the supply process.

Furthermore, the number of required fast switching devices in both designs is

reduced to half of the number used in a conventional MG; they are replaced with

slower switches (such as Thyristors) that need simpler driving modules. In

addition, the contributing switches in discharging paths are decreased to half;

this decrease leads to a reduction in conduction losses. Associated models are

simulated, and hardware tests are performed to verify the validity of proposed

topologies. Chapters 4, 5 and 7 of the thesis present all relevant analysis and

approaches according to these topologies.

X

Keywords

Capacitor-diode voltage multiplier (CDVM)

Commutation circuit

Current and voltage sources

Diode rectifiers

Gate turn-off Thyristor (GTO)

H-bridge inverter

High voltage

Insulated gate bipolar transistor (IGBT)

Magnetic pulse compressor (MPC)

Marx generator (MG)

Metal-oxide semiconductor field-effect transistor (MOSEFET)

Micro controller

Multistage Blumlein lines (MBL)

Plasma

Positive buck-boost converter

Pulse forming network (PFN)

Pulse width modulation (PWM)

Pulsed electric field (PEF)

Pulsed power

Reactor

Repetitively operation

Resonant converter

Rising and falling times

Semiconductor technology

Silicon-controlled rectifier (SCR)

Single shot

Solid state technology

Spark gap

Switching transient

Unipolar and Bipolar modulation

Voltage stress

XI

List of Abbreviation

AC: Alternating Current

BJT: Bipolar Junction Transistor

CCM : Continuous-Conduction-Mode

CDVM : Capacitor-Diode Voltage Multiplier

CERN: Conseil Europeen Pour La Recherche Nucleaire (The European

Organization for Nuclear Research)

CSR: Converter Series Resonant

DC: Direct Current

DCM : Discontinuous-Conduction-Mode

DG: Distributed Generation

DNA: Deoxyribo Nucleic Acid

EMC : Electromagnetic Compatibility

EMI : Electromagnetic Interference

EML : Electromagnetic Launcher

GPR: Ground Penetrating Radar

GTO: Gate Turn-off Thyristor

HBRC: Half Bridge Resonant Converter

HFB: Hybrid Full-Bridge

HPM : High-Power Microwave

HTP: Hard-Tube Pulser

HV : High Voltage

IAQ : Indoor Air Quality

IGBT : Insulated Gate Bipolar Transistor

ISL : French-German Research Institute of Saint-Louis

MBL : Multistage Blumlein Lines

MFC : Magnetic Flux Compressor

MG : Marx Generator

MOSEFET: Metal-Oxide Semiconductor Field-Effect Transistor

MPC: Magnetic Pulse Compressor

MVM : Multilevel Voltage Multiplier

PBII&D or PBIID : Plasma-Based Ion Implantation and Deposition

PDM: Pulse-Density Modulation

XII

PDP TV: Plasma Display Panel Television

PEF: Pulsed Electric Field

PEN: Polyethylene Naphthalate

PFC: Power Factor Correctors

PFN: Pulse Forming Network

PID: Proportional–Integral–Derivative

PIII : Plasma Immersion Ion Implantation

POME: Palm Oil Mills Effluent

PPS: Pulsed Power Supply

PT: Piezoelectric Transformer

PWM : Pulse Width Modulation

RF: Radio Frequency

SCR: Silicon-Controlled Rectifier

SDG: Silent Discharge Generator

SDPS: Silent Discharge Plasma Systems

SFPFN: Sequentially-Fired Pulse Forming Network

SMPS: Switched-Mode Power Supply

TRIUMF : Tri-University Meson Facility (University of British Columbia)

(Canada's National Laboratory for Particle and Nuclear Physics)

UPS: Uninterrupted Power Supply

UWB: Ultra-Wide-Band

VRM : Voltage Regulation Module

ZCS: Zero Current Switching

ZVS: Zero Voltage Switching

XIII

Table of Contents

Abstract ................................................................................................................ V

Keywords ............................................................................................................. X

List of Abbreviation .......................................................................................... XI

Contributions ............................................................................................ XXVIII

List of Publications ........................................................................................ XXX

List of chapters according to publications and contributions ................ XXXII

Scholarship and grants ............................................................................ XXXIII

Statement of Original Authorship .......................................................... XXXIV

Chapter 1.............................................................................................................. 1

Introduction ......................................................................................................... 1

1.1. Definition of the Research Problem ............................................................ 2

1.2. Literature Review ........................................................................................ 6

1.2.1. Introduction ..........................................................................................6

1.2.2. Applications .........................................................................................7

1.2.3. Pulsed power supply technologies .....................................................13

1.2.3.1. Magnetic pulse compressor (MPC) .............................................13

1.2.3.2. Pulse Forming Network (PFN) ...................................................16

1.2.3.3. Multistage Blumlein Lines ..........................................................18

1.2.3.4. Marx Generator (MG) .................................................................20

1.2.4. Power electronics in pulsed power generation ...................................23

1.2.4.1. All-solid-state Marx Generator ...................................................24

1.2.4.2. dc-dc Converters .........................................................................26

1.2.4.3. Voltage Multipliers .....................................................................29

1.2.4.4. Pulse Generators Based on Inverters ...........................................32

1.2.4.5. Resonant Converters ...................................................................34

1.3. Account of Research Progress Linking the Research Papers.................... 39

1.3.1. Introduction ........................................................................................39

1.3.2. A new solid-state current-voltage source based pulsed power supply

......................................................................................................................42

XIV

1.3.2.1. Investigating the possibility of producing pulses for plasma

applications through a current source ...................................................... 42

1.3.2.2. Proposition of a novel high-voltage pulsed power supply based

on low-voltage switch-capacitor units ..................................................... 46

1.3.2.3. A new multi-purpose pulsed power supply based on positive

buck-boost converter concept .................................................................. 54

1.3.2.4. A design for producing pulses with higher magnitude ............... 58

1.3.3. New configurations for MG .............................................................. 60

1.3.3.1. A resonant based converter for pulsed power purposes ............. 61

1.3.3.2. A resonant based Marx Generator .............................................. 64

1.3.3.3. A new Configuration for Marx Generator utilizing fast and slow

solid-state switches .................................................................................. 65

1.3.3.4. A new family of Marx Generators based on commutation circuits

................................................................................................................. 73

1.3.4. A high voltage converter based on capacitor –diode voltage multiplier

(CDVM) with a frequency and voltage controller....................................... 78

1.3.4.1. A high voltage converter based on capacitor diode voltage

multiplier (CDVM) with a frequency and voltage controller .................. 79

1.4. References: ............................................................................................... 84

CHAPTER 2 ....................................................................................................... 97

A Novel High-Voltage Pulsed-Power Supply Based on Low-voltage Switch-

Capacitor Units .................................................................................................. 97

2.1. Index Terms .............................................................................................. 98

2.2. Introduction .............................................................................................. 98

2.2. Configuration and analyses ...................................................................... 99

2.2.1. Topology ............................................................................................ 99

2.2.2. Switching modes.............................................................................. 101

2.2.2.1. First mode (SS: on, S1: on, S2: on) ............................................ 101

2.2.2.2. Second mode (SS: off, S1: on, S2: on) ....................................... 101

2.2.2.3. Third mode (SS: off, S1: off, S2: off) ......................................... 102

2.2.3. Discharging Residual Energy .......................................................... 103

2.2.3. 1. Hard methods ........................................................................... 104

XV

2.2.3.2. Soft method, fourth and fifth switching modes (SS: off, S1: on, S2:

off) & (SS: off, S1: off, S2: on)................................................................104

2.2.4. Analyses of load supplying mode ....................................................106

2.3. Control strategy ....................................................................................... 107

2.3.1. Current control .................................................................................107

2.3.2. Voltage control .................................................................................108

2.3.3. Load control .....................................................................................109

2.4. Simulation results .................................................................................... 111

2.5. Components determination and energy discussion ................................. 114

2.6. Experimental results ................................................................................ 116

2.7. Conclusions ............................................................................................. 118

2.8. References ............................................................................................... 119

CHAPTER 3 .................................................................................................... 122

A New Pulsed Power Supply Topology Based On Positive Buck-Boost

Converters Concept ........................................................................................ 122

3.1. Index Terms ............................................................................................ 123

3.2. Introduction ............................................................................................. 123

3.3. Configuration and analyses ..................................................................... 125

3.3.1. Topology ..........................................................................................125

3.3.1.1. General configuration ................................................................125

3.3.1.2. Switching modes .......................................................................126

3.3.1.3. Circuit analyses .........................................................................128

3.3.2. Control strategies .............................................................................130

3.3.2.1. Current source control ...............................................................130

3.3.2.2. Voltage source control ..............................................................131

3.3.2.3. Load modeling control ..............................................................132

3.3.3. Components determination and energy discussion ..........................133

3.4. Simulation results and analyses .............................................................. 136

3.4.1. Simultaneous switching ...................................................................136

3.4.2. Separate switching ...........................................................................137

3.5. Experimental results ................................................................................ 138

3.6. Conclusion .............................................................................................. 141

3.7. References ............................................................................................... 141

XVI

CHAPTER 4 ..................................................................................................... 145

A Solid State Marx Generator with a Novel Configuration ........................ 145

4.1. Index Terms ............................................................................................ 146

4.2. Introduction ............................................................................................ 146

4.3. Topology ................................................................................................. 149

4.3.1. General configuration ...................................................................... 149

4.3.2. Switching states ............................................................................... 149

4.3.2.1. Positive charging mode: (S1:on, S2:off, S3:on, S4:off) .............. 150

4.3.2.2. Negative charging mode: (S1:off, S2:on, S3:on, S4:off) ............ 150

4.3.2.3. Load supplying mode: (S1:off, S2:off, S3:off, S4:on) ................ 151

4.4. Simulation results and analyses .............................................................. 151

4.4.1. Control strategy ............................................................................... 152

4.4.2. Single shot and repetitively operated results ................................... 152

4.4.3. The voltage stresses across the diodes and the current through the

power switches .......................................................................................... 153

4.4.4. The generated voltage adjustability ................................................. 154

4.5. Experimental results ............................................................................... 156

4.6. Structure and performance comparison .................................................. 157

4.7. Conclusion .............................................................................................. 158

4.8. References .............................................................................................. 159

CHAPTER 5 ..................................................................................................... 163

A New Family of Marx Generators Based on Commutation Circuits ........ 163

5.1. Index Terms ............................................................................................ 164

5.2. Introduction ............................................................................................ 164

5.3. Configuration and analyses .................................................................... 165

5.3.1. Topology .......................................................................................... 165

5.3.2. Switching modes.............................................................................. 167

5.3.2.1. First mode: Inductor charging mode ........................................ 167

5.3.2.2. Second mode: capacitors charging mode ................................. 168

5.3.2.3. Third mode: commutation mode .............................................. 168

5.3.2.4. Fourth mode: pulse generation mode ....................................... 169

5.3.3. Control strategy ............................................................................... 170

XVII

5.4. Simulation results .................................................................................... 171

5.5. Experimental verification ........................................................................ 175

5.6. Design features and the component discussion ....................................... 176

5.7. Conclusion .............................................................................................. 178

5.8. References ............................................................................................... 178

CHAPTER 6 .................................................................................................... 182

Using a Current Source to Improve Efficiency of a Plasma System .......... 182

6.1. Introduction ............................................................................................. 183

6.2. Current source topology .......................................................................... 185

6.2.1. Hysteresis current controller ............................................................185

6.2.2. Voltage level and switching stress ...................................................187

6.2.3. Power losses issue ............................................................................189

6.3. Extra capacitor ........................................................................................ 190

6.4. Summary ................................................................................................. 191

6.5. References ............................................................................................... 191

CHAPTER 7 .................................................................................................... 195

A Bidirectional Two-Leg Resonant Converter for High Voltage Pulsed

Power Applications ......................................................................................... 195

7.1. Keywords ................................................................................................ 196

7.2. Introduction ............................................................................................. 196

7.3. Bidirectional resonant converter: topology and operation ...................... 197

7.4. Bipolar control method ........................................................................... 199

7.5. Unipolar control method ......................................................................... 202

7.6. Conclusions ............................................................................................. 204

7.7. References ............................................................................................... 205

CHAPTER 8 .................................................................................................... 208

A High Voltage Power Converter with a Frequency and Voltage Controller

........................................................................................................................... 208

8.1. Introduction ............................................................................................. 209

8.2. Transient .................................................................................................. 211

8.3. Adjustable output voltage level ............................................................... 212

XVIII

8.4. Feeding CDVM through an inverter ....................................................... 213

8.5. Energy discussion for Plasma applications ............................................ 216

8.5. Further analyses ...................................................................................... 216

8.5. Summary ................................................................................................. 218

8.6. References .............................................................................................. 218

CHAPTER 9 ..................................................................................................... 220

Conclusions and Further Research ................................................................ 220

9.1. Conclusions ............................................................................................ 221

9.1.1. Developing and proposing novel solid-state based topologies for

pulsed power generation ............................................................................ 222

9.1.2. Improving conventional technologies that have potential to


adjusting their structure. ............................................................................ 224

9.2.3. A summary of features, advantages and restrictions of proposed

converters ................................................................................................... 227

9.2. Further research ...................................................................................... 230

Developing insulated solid-state topologies for pulsed power .................. 231

Using CDVMs as fundamental voltage boosters for an MG for continuously

high voltage applications ........................................................................... 231

Using PFNs as basic units of an MG ......................................................... 231

Applications ............................................................................................... 231

XIX

List of Figures

Chapter 1

Fig.1. 1. Pulsed power features ..............................................................................7

Fig.1. 2. Time-resolved Schlieren photograph of laser-induced breakdownin a

170 kV switch a few nanoseconds before breakdown [16] ....................................9

Fig.1. 3. Typical still photograph of a discharge induced in water[28] ...............10

Fig.1. 4. Photographs of (a) discharge and (b) concrete scooped out[2] .............11

Fig.1. 5. Two schematics of MPC ........................................................................13

Fig.1. 6. Different types of PFNs .........................................................................16

Fig.1. 7. Samples of typical multistage blumlein pulsers ....................................19

Fig.1. 8. MG with spark gap switching and resistive insulation ..........................20

Fig.1. 9. (a) An all-solid-state MG, (b) Charging mode, (c) Discharging mode ..24

Fig.1. 10. dc-dc converters, (a). Buck (b). Boost (c). Buck-Boost (d). Positive

Buck-Boost ...........................................................................................................27

Fig.1. 11. Capacitor-Diode Voltage Multipliers (CDVM) (a). N-stage Cockcroft-

Walton Voltage Multiplier (b). N-stage Dickson charge pump (c). Another N-

stage CDVM.........................................................................................................29

Fig.1. 12. Voltage source pulsed power supply ...................................................42

Fig.1. 13. Positive buck-boost topology ...............................................................43

Fig.1. 14. (a). A circuit diagram of current source topology (b)&(c). Operation

modes of the current source topology supplying a plasma load ..........................44

Fig.1. 15. Voltages and currents of modeled capacitor with 20, 40, 60, 80 and

100A currents flowing into the capacitive load through a variable current source

..............................................................................................................................45

Fig.1. 16. (a). Inductor energy (b). Capacitor energy (c). Load energy ...............46

Fig.1. 17. Plasma power supply configuration with multi switch-resistor-

capacitor units ......................................................................................................47

Fig.1. 18. A simplified two switch-capacitor unit plasma power supply and the

load model ............................................................................................................47

Fig.1. 19. Switching states of the proposed pulsed power supply circuit (a)

Current source, charging mode (b) Current source, discharging mode (c) Voltage

source charging mode (d) Load supplying mode .................................................48

XX

Fig.1. 20. Possible current loops during short circuit periods. ............................ 49

Fig.1. 21. Two examples of hardware methods for discharging residual energy in

the inductor and the capacitors, (a) Parallel switch-resistor unit located in the

return path (b) A thermistor in the return path .................................................... 49

Fig.1. 22. Circuit’s switching states in association with software method in order

to discharge the remaining energy in the capacitors ............................................ 50

Fig.1. 23. (a). Block diagram of control algorithm (b). Switching signals pattern

............................................................................................................................. 50

Fig.1. 24. Equivalent RLC circuit of power delivery mode of power supply ...... 51

Fig.1. 25. Laboratory prototype of pulsed power supply with double switch-

capacitor units ...................................................................................................... 52

Fig.1. 26. Inductor current, capacitors and output voltages ................................ 53

Fig.1. 27. Pulsed power supply configuration with multi switch-diode-capacitor

units ..................................................................................................................... 55

Fig.1. 28. Switching state of charging capacitors separately ............................... 56

Fig.1. 29. (a). Flowchart of the control algorithm (b). Current and voltage

waveforms accompanied by correspondent switching signals pattern ................ 56

Fig.1. 30. Inductor current, capacitors and output voltages ................................ 57

Fig.1. 31. Developing hardware for higher voltage pulses .................................. 59

Fig.1. 32. The experimental results of developed six-stage buck-boost based

pulse supply ......................................................................................................... 60

Fig.1. 33. (a) Resonant circuit, (b) Half a resonant circuit, (c) Capacitor voltage

and Inductor current of a typical resonant circuit. (d) Capacitor voltage and

Inductor current of a typical half a resonant circuit ............................................. 61

Fig.1. 34. (a).Bidirectional resonant circuit (b).The resonant converter ............. 62

Fig.1. 35. A block diagram of proposed resonant converter ............................... 62

Fig.1. 36. Operation modes of the resonant converter supplied with an inverter

controlled with bipolar method. ........................................................................... 63

Fig.1. 37. Extra states of inverter providing resonant converter with the zero

level of voltage in unipolar control method. ........................................................ 63

Fig.1. 38. Using resonant concept in Marx topology .......................................... 64

Fig.1. 39. The block diagram of proposed converter with a new Marx

configuration ........................................................................................................ 65

XXI

Fig.1. 40. The four-stage simulated model of proposed MG ...............................66

Fig.1. 41. Switching states of proposed MG (a) Positive charging mode (b)

Negative charging mode (c) Load supplying mode, ............................................67

Fig.1. 42. The components voltages and the currents (a) Diodes, (b) Switches ..67

Fig.1. 43. (a). Current and voltage waveforms accompanied by relevant

switching signal patterns, (b). Simulation results of proposed repetitively

operated topology .................................................................................................68

Fig.1. 44. Using switches with anti-parallel body diodes in the inverter .............68

Fig.1. 45. (a). Bidirectional solid-state switching path (b). Proper installation

point of the reserve path (c). Extra switching states associated with the unipolar

control method of the half bridge inverter ...........................................................69

Fig.1. 46. Simulation results for the converters with (a). Anti-parallel body

diodes (b). Reserve path. ......................................................................................70

Fig.1. 47. Hardware set up ...................................................................................71

Fig.1. 48. Experimental results for (a) The capacitors and the output voltages and

the inductor current (b) The voltages across S3 and S4. ........................................71

Fig.1. 49. Block diagram of new Marx topology .................................................73

Fig.1. 50. Circuit diagram of the proposed topology. ..........................................74

Fig.1. 51. Switching states of the proposed Marx generator for single pulse

generation. ............................................................................................................75

Fig.1. 52. Extra switching states of the proposed Marx generator for repetitive

pulse generation. ..................................................................................................75

Fig.1. 53. (a). Control flowchart for a modulator with the repetitive pulse

generation function, (b). The capacitor voltages and the gate drive waveforms of

the converter. ........................................................................................................76

Fig.1. 54. Experimental set up. ............................................................................77

Fig.1. 55. Experimental results. ...........................................................................77



stage CDVM configuration ..................................................................................78

Fig.1. 57. One-stage Cockcroft-Walton voltage multiplier..................................79

Fig.1. 58. Voltage transient of multiplier with 50 Hz input frequency (a).

Identical capacitors (b). Different capacitors (C1=10C2) .....................................80

XXII

Fig.1. 59. Voltage transient of multiplier with 1KHz input frequency (a).

Identical capacitors (b).Different capacitors ....................................................... 80

Fig.1. 60. An ac-dc-ac converter ......................................................................... 80

Fig.1. 61. (a). Schematic of full bridge (two-leg) inverter (b). Bipolar and

unipolar modulations output waveforms ............................................................. 81

Fig.1. 62. Output voltage of inverter and filter for duty cycles of (a). 0.05 (b). 0.5

(c). 0.95 ................................................................................................................ 82

Fig.1. 63. An inverter supplying multiplier with variable frequency and

amplitude ............................................................................................................. 82

Fig.1. 64. (a) Variable input voltage results in variable voltages in the output (b).

Variable output voltage provided by an inverter under unipolar control method.

(c). Inverters’ output waveform with duty cycles of 0.1 & 0.5 & 0.9. (d). Load

connections and voltage rehabilitation capability ............................................... 83

Chapter 2

Fig.2. 1. Marx generator ...................................................................................... 99

Fig.2. 2. Plasma power supply configuration with multi switch-capacitor units

........................................................................................................................... 100

Fig.2. 3. A simplified two switch-capacitor unit plasma power supply and the

load model ......................................................................................................... 100

Fig.2. 4. Switching states of the proposed power supply circuit (a) Current

source, charging mode (b) Current source, discharging mode (c) Voltage source

charging mode (d) Load supplying mode .......................................................... 103

Fig.2. 5. Possible current loops during short circuit periods. ............................ 104

Fig.2. 6. Two examples of hard methods for discharging residual energy in the

inductor and capacitors, (a) Parallel switch-resistor unit located in the return path

(b) A thermistor in the return path ..................................................................... 104

Fig.2. 7. Circuit’s switching states in association with soft method in order to

discharge the remained energy in the capacitors ............................................... 105

Fig.2. 8. Equivalent RLC circuit of power delivery mode of power supply ...... 106

Fig.2. 9. Block diagram of control algorithm .................................................... 110

Fig.2. 10. Definition of voltage and current levels for the control strategy ...... 110

Fig.2. 11. Switching signals pattern .................................................................. 111

XXIII

Fig.2. 12. dv/dt s generated by different inductors with the similar inductive

energy, (a) current of 1mH inductor (b) output voltage of 1mH inductor, (c)

current of 9mH inductor, (d) output voltage of 9mH inductor. .........................112

Fig.2. 13. Output voltages and currents of power supply, (a) Inductor current (A),

(b) C1 & S1 voltage (V), (c) C2 & S2 voltage (V), (d) Output voltage (V), (e)

Load current (A) .................................................................................................113

Fig.2. 14. Inductor current and output voltages of power supply in the case of no

prosperous plasma phenomena, (a) Inductor current (A), (b) C1 & S1 voltage (V),

(c) C2 & S2 voltage (V), (d) Output voltage (V) ................................................114

Fig.2. 15. Times monitoring in a load supplying cycle ......................................115


capacitor units ....................................................................................................117

Fig.2. 17. Inductor current, capacitors and output voltages ...............................118

Chapter 3

Fig.3. 1. A general configuration of proposed concept ......................................124

Fig.3. 2. Pulsed power supply configuration with multi switch-diode-capacitor

units ....................................................................................................................125

Fig.3. 3. A pulsed power supply with two switch-diode-capacitor units and a

non-linear load ...................................................................................................126

Fig.3. 4. Switching states of the proposed power supply circuit (a) Inductor

charging (b) Circulating the inductor current (c) Charging the capacitors (d)

Supplying the load..............................................................................................128

Fig.3. 5. Switching state of charging capacitors separately ...............................128

Fig.3. 6. Flowchart of the control algorithm ......................................................131

Fig.3. 7. Current and voltage waveforms accompanied by relevant switching

signals pattern in separate switching strategy ....................................................133

Fig.3. 8. Times monitoring in a load supplying cycle ........................................135

Fig.3. 9. Output voltages and currents of power supply under simultaneous

switching algorithm, (a) Inductor current (A), (b) C1 & S1 voltage (V), (c) C2 &

S2 voltage (V), (d) Output voltage (V), (e) Load current (A) ............................137

Fig.3. 10. Output voltages and currents of power supply under separate switching

algorithm, (a) Inductor current (A), (b) C1 & S1 voltage (V), (c) C2 & S2 voltage

(V), (d) Output voltage (V), (e) Load current (A) ..............................................137

XXIV


diode-capacitor units .......................................................................................... 140

Fig.3. 12. Inductor current, capacitors and output voltages .............................. 141

Chapter 4

Fig.4. 1. A conventional MG ............................................................................. 148

Fig.4. 2. Block diagram of proposed converter with a new Marx configuration,

........................................................................................................................... 149

Fig.4. 3. Four-stage simulated model of proposed MG, .................................... 150

Fig.4. 4. The switching states of proposed MG (a) Positive charging mode (b)

Negative charging mode (c) Load supplying mode, .......................................... 151

Fig.4. 5. Current and voltage waveforms accompanied by relevant switching

signal patterns, ................................................................................................... 152

Fig.4. 6. Simulation results of proposed topology, (a) Single shot, (b)

Repetitively operation ........................................................................................ 153

Fig.4. 7. The components voltages and the currents (a) Diodes, (b) Switches, . 154

Fig.4. 8. Using switches with anti-parallel body diodes in the inverter, ........... 155

Fig.4. 9. (a). The bidirectional solid state switching path (b). The proper

installation point of the reserve path (c)&(d). The extra switching states

associated with the unipolar control method of the half bridge inverter ........... 155

Fig.4. 10. Simulation results for the converters with (a). anti-parallel body diodes

(b). the reserve path. .......................................................................................... 156

Fig.4. 11. The hardware set up .......................................................................... 157

Fig.4. 12. Experimental results for (a) The capacitors and the output voltages and

the inductor current (b) The voltages across S3 and S4. ..................................... 158

Chapter 5

Fig.5. 1. A conventional Marx generator. .......................................................... 165

Fig.5. 2. Block diagram of the new Marx topology .......................................... 166

Fig.5. 3. Circuit diagram of the proposed topology. .......................................... 166

Fig.5. 4. Switching states of proposed MG for single pulse generation. ........... 167

Fig.5. 5. Extra switching states of proposed MG for repetitive pulse generation.

........................................................................................................................... 169

Fig.5. 6. The capacitor voltages and the gate drive waveforms of the converter.

........................................................................................................................... 171

XXV

Fig.5. 7. Control flowchart for a modulator with the repetitive pulse generation

function. .............................................................................................................172

Fig.5. 8. Simulation results for the proposed converter (single pulse). .............173

Fig.5. 9. Simulation results for the proposed converter (repetitive pulses). ......173

Fig.5. 10. The switches voltages and currents. ..................................................175

Fig.5. 11. The experimental set up. ....................................................................176

Fig.5. 12. The experimental results ....................................................................176

Fig.5. 13. The switch and the diode that connect diode-capacitor units compose a

circuit. .................................................................................................................178

Chapter 6

Fig.6. 1. (a). Block diagram of high voltage source topology (b). N-stage Marx

generator .............................................................................................................184

Fig.6. 2. (a). A circuit diagram of current source topology (b)&(c). Operation

modes of the current source topology supplying plasma load ...........................184

Fig.6. 3. Hysteresis band current control ...........................................................186

Fig.6. 4. Voltage and current of modelled capacitor with 100A inductor current

............................................................................................................................188

Fig.6. 5. Voltage and current of modeled capacitor with 20, 40, 60, 80 and 100A

inductor currents .................................................................................................189

Fig.6. 6. (a). Inductor energy (b). Capacitor energy (c). Load energy ...............191

Chapter 7

Fig.7. 1. (a) Resonant circuit, (b) Half resonant circuit, (c) Capacitor voltage and

Inductor current of a typical resonant circuit. (d) Capacitor voltage and Inductor

current of a typical half resonant circuit ............................................................197

Fig.7. 2. A block diagram of the proposed resonant converter ..........................198

Fig.7. 3. Bidirectional resonant circuit ...............................................................198

Fig.7. 4. Output voltage waveforms with bipolar and unipolar modulations. ...199

Fig.7. 5. Operation modes of the resonant converter supplied with an inverter

controlled with bipolar method. .........................................................................200

Fig.7. 6. Input voltage, inductor current, capacitors and output voltages of a

resonant converter with an inverter controlled with bipolar method in the case of

: (a) fS=f r=15823Hz, (b) fS<f r (fS=10kHz), (c) fS>f r (fS=25kHz), (d) fS>2fr ,

(fS=40kHz) ..........................................................................................................202

XXVI

Fig.7. 7. Extra states of inverter providing resonant converter with the zero level

of voltage in unipolar control method. .............................................................. 203

Fig.7. 8. Input voltage, inductor current, capacitors and output voltages of a

resonant converter with an inverter controlled with unipolar method in the case

of :(a) P.W.=0.4TS, (b) P.W.=0.3TS, (c) P.W.=0.2TS, (d) P.W.=0.1TS ............. 204

Chapter 8



stage CDVM configuration ............................................................................... 210

Fig.8. 2. One-stage Cockcroft-Walton voltage multiplier ................................. 210

Fig.8. 3. Voltage transient of multiplier with 50 Hz input frequency (a). Identical

capacitors (b). Different capacitors (C1=10C2) ................................................. 211

Fig.8. 4. Voltage transient of multiplier with 1KHz input frequency (a). Identical

capacitors (b).Different capacitors .................................................................... 212

Fig.8. 5. An ac-dc-ac converter ......................................................................... 213

Fig.8. 6. (a). Schematic of full bridge (two-leg) inverter (b). Bipolar and unipolar

modulations output waveforms ......................................................................... 214

Fig.8. 7. Output voltage of inverter and filter for duty cycles of (a). 0.05 (b). 0.5

(c). 0.95. ............................................................................................................. 215

Fig.8. 8. (a) Variable input voltage results in variable voltages in the output (b).

Variable output voltage provided by an inverter under unipolar control method.

(c). Inverter’s output waveform with duty cycles of .1 & .5 & .9. (d). Load

connections and voltage rehabilitation capability ............................................. 216

Fig.8. 9. An inverter supplying multiplier with variable frequency and amplitude

........................................................................................................................... 217

Fig.8. 10. (a). Multiplier voltages, and first capacitor current with pulse shape

input waveforms (b). Multiplier voltages, and first capacitor current with

trapezoidal input waveforms ............................................................................. 218

XXVII

List of Tables

Chapter 1

TABLE 1. 1. Specifications of the laboratory prototype circuit ..........................57

TABLE 1. 2. Specifications of the pulser ............................................................59

TABLE 1. 3. Specifications of the Implemented Circuit .....................................70

Chapter 2

TABLE 2. 1. Specifications of the modeled circuit ...........................................116

Chapter 3

TABLE 3. 1. Specifications of the modeled circuit ...........................................135

TABLE 3. 2. Specifications of the laboratory prototype circuit ........................140

Chapter 4

TABLE 4. 1. Specifications of the Modelled Circuit.........................................151

TABLE 4. 2. Specifications of the Implemented Circuit ...................................156

Chapter 5

TABLE 5. 1. The specifications of simulated models .......................................171

TABLE 5. 2. The specifications of implemented hardware ...............................175

Chapter 6

TABLE 6. 1. Variation of (dv/dt)s in the transient of switching ........................189

Chapter 7

Chapter 8

TABLE 8. 1. Circuit specifications ....................................................................215

XXVIII

Contr ibut ions

A new generation of efficient pulsed power supplies based on low-

medium voltage switch-capacitor units

Introducing the current source as the main energy source for this power

supply while the energy transforms to the voltage form at the next stage

Investigating influences of IGBTs switching transients on current

conduction and injection to the capacitors and producing voltage stress

across the output load

Investigating high voltage sharing through solid state switches

Considering hardware based protection methods for the circuit to

discharge the stored energy in case of probable plasma failure.

Developing a smart process of switching modes that allows a safe

discharge of residual energy after each plasma supplying cycle

Evolution of earlier discussed pulsed power supply generation with

respect to positive buck-boost converters concept to supply a wider

range of applications with a proper and more reliable load supplying

process

A novel configuration for all solid state resonant based Marx generator

to reduce required semiconductor rectifiers and switches and

consequently reduce conduction and switching losses

An H-bridge-fed bidirectional two diode-capacitor leg resonant converter

has been studied

The converter is developed for a cascade Marx circuit

A new configuration has been proposed by rearranging the components

and reducing the number of semiconductors in the former resonant Marx

circuit

A new family of Marx generators based on commutation circuit using

inductive energy storage to discharge and recharge the capacitors and

the use of slow thyristors instead of fast IGBTs

Demonstrating diode-capacitor leg based Marx topology

Proposing buck-boost converter as a power supply for the new Marx

topology

XXIX

A high voltage power converter based on capacitor-diode voltage

multiplier(CDVM) with capability of voltage and fre quency control

Comparing different types of dc capacitor-diode networks utilized in

cascade voltage multiplier configurations

Designing an appropriate adaptive control method based on PID feedback

control

Analysing the skills and benefits of Cockcroft–Walton circuit as a high

voltage modulator and studding its results for variable input voltage and

load supplying condition

Hardware implementation of the proposed topologies including the new

buck-boost based pulsed power supply and Commutation based and

resonant based Marx topologies

XXX

List of Publ icat ions

The Queensland University of Technology (QUT) allows the presentation of a

thesis for the Degree of Doctor of Philosophy in the format of published or

submitted papers, where such papers have been published, accepted or submitted

during the period of candidature. This thesis is composed of eleven

published/submitted papers, of which eight have been published and three are

under review. Note that due to overlap of the paper contents, seven papers have

been selected for the thesis as seven chapters.

Published Peer Reviewed Journal Articles:

1. Sasan Zabihi, Firuz Zare, Gerard Ledwich, Arindam Ghosh, Hidenory

Akiyama, “A Novel High-Voltage Pulsed-Power Supply Based on Low-

Voltage Switch–Capacitor Units”, IEEE Transactions on Plasma

Science, Vol.38, No.10, pp.2877-2887, Oct. 2010.


Akiyama, “A New Pulsed Power Supply Topology Based on Positive

Buck-Boost Converters Concept”, IEEE Transactions on Dielectrics

and Electrical Insulation, Vol.17, No.6, pp.1901-1911, Dec. 2010.


Akiyama, “A New Family of Marx Generators Based on Commutation

circuit”, IEEE Transactions on Dielectrics and Electrical Insulation,

Vol. 18, Issue 4, pp.1181-1188, Aug. 2011.

4. Sasan Zabihi, Zeynab Zabihi, Firuz Zare, “A Solid State Marx Generator

with a Novel Configuration”, IEEE Transactions on Plasma Science,

Vol.39, No.8, pp.1721-1728, Aug. 2011.

Published Peer Reviewed International Conference Papers:

XXXI

5. Sasan Zabihi, Firuz Zare, Hidenory Akiyama, “Using a Current Source

to Improve Efficiency of a Plasma System”, Pulsed Power Conference,

2009. PPC '09. IEEE , vol., no., pp.1256-1260, June 28 2009-July 2 2009

6. Sasan Zabihi, Firuz Zare, Hidenory Akiyama, “A High Voltage Power

Converter with a Frequency and Voltage Controller”, Pulsed Power

Conference, 2009. PPC '09. IEEE , vol., no., pp.1250-1255, June 28

2009-July 2 2009

7. Sasan Zabihi, Firuz Zare, Gerard Ledwich, Arindam Ghosh, “A

Bidirectional Two-Leg Resonant Converter for High Voltage Pulsed

Power Applications”, Pulsed Power Conference, 2009 IET European ,

vol., no., pp.1-4, 21-25 Sept. 2009

8. Sasan Zabihi, Firuz Zare, Gerard Ledwich, Arindam Ghosh, “A Novel

High Voltage Pulsed Power Supply Based on Low Voltage Switch-

Capacitor Units”, Pulsed Power Conference, 2009 IET European , vol.,

no., pp.1-4, 21-25 Sept. 2009


Akiyama, “A New Family of Marx Generator Based on Resonant

Converter”, Energy Conversion Congress and Exposition (ECCE), 2010

IEEE , vol., no., pp.3841-3846, 12-16 Sept. 2010


Akiyama, “A New Generation of High Voltage Pulsed Power

Converters”, Australian Universities Power Engineering Conference,

AUPEC2010, Christchurch, New Zealand

11. Sasan Zabihi, Firuz Zare, Gerard Ledwich, Arindam Ghosh, “A

Resonant Based Marx Generator”, Australian Universities Power

Engineering Conference, AUPEC2010, Christchurch, New Zealand

12. Sasan Zabihi, Zeynab Zabihi, Firuz Zare, “A Solid State Marx Generator

with a Novel Configuration”, Presented at 19th Iranian Conference in

Electrical Engineering, ICEE2011, Tehran, Iran

XXXII

List of chapters according to publ ications and contr ibut ions

Scholarship and grants

Fee waiver scholarship from Queensland University of Technology for

PhD degree for 3 years 2008-2011

Supervisor’s living allowances scholarship award through an ARC

Discovery award Funded by the Australian Research Council at

Queensland University of Technology for PhD degree for 3 years 2008-

2011

Travel grant from IEEE Power Electronics Society for attending to the

second IEEE Energy Conversion Congress and Exposition, ECCE2010,

Atlanta, Georgia, United States.

Travel grant from Australian Universities Power Engineering Conference

for attending AUPEC 2010, Christchurch, New Zealand.

QUT grant –in-aid for attendance at ECCE2010 conference in Atlanta,

Georgia, United States, 2010

QUT grant –in-aid for attendance at AUPEC 2010 conference in

Christchurch, New Zealand, 2010

XXXIV

Statement of Original Authorship

“The work contained in this thesis has not been previously submitted to meet

requirements for an award at this or any other higher education institution. To

the best of my knowledge and belief, the thesis contains no material previously

published or written by another person except where due reference is made.”

Signature Date

1

Chapter 1

Introduct ion

2

1.1. Def in i t ion of the Research Problem

Pulsed power, also termed pulsed electrical field (PEF), has been in use in

industry for decades. Recently, it has found more applications due to its

increased flexibility in terms of power, voltage and repetition rates. Pulsed power

has become more appealing these days for an extensively diverse range of

applications due to innovative adaptive methods and their use in pulse producing

processes. The fast growing trend of utilizing pulsed power in industrial,

environmental, medical and military applications is increasing the demand for

even more advanced pulsed power supplies. To date, several pulse generator

topologies using various techniques have been implemented and used for

different applications, including magnetic pulse compressor (MPC), multistage

Blumlein lines (MBL), pulse forming network (PFN), hard-tube pulser (HTP),

and Marx generator (MG). These pulsed power supplies have both common and

individual advantages and disadvantages. These prior-art methods require several

power conversion steps that lead to elaborated circuit structures with low

reliability. Almost all these methods utilize either spark gap or vacuum tube as

high voltage switches in their structure. (However, as an exception, solid-state

technology has been recently utilized in MG architecture.) These magnetic and

gas based switches have basic advantages of providing a high blocking voltage

and a high current carrying capabilities with considerably low transients,

withstanding the action of the current pulse and achieving the desired energy.

However, they also have undeniable disadvantages, including: size, bulk,

expense, inefficiency and limitation in either peak or average power output, short

life span, and low operational frequency.

These pulse modulators also have specific shortcomings. The PFN circuit has

two intrinsic, specific drawbacks; namely, it is very difficult to match the load to

the output impedance of the PFN circuit in plasma applications, and the PFN

pulse length is fixed by the number of LC sections and LC electrical parameters.

The pulse voltage waveforms provided by HTPs are not ideal flat-top types for

applications such as plasma immersion ion implantation (PIII). Moreover, rise

and fall times are about tens of microseconds due to the conduction time of gas

switches. Traditionally used MGs employed resistors as isolators, which made

3

them inefficient. These issues have seriously affected the supply process and

caused numerous concerns. These considerations and concerns lead to two

specific research problems.

Major research objectives in this research work include couple of tasks: First, the

exploration and justification of new topologies that have merits in ensuring the

efficient supply of a wide range of pulse applications, modification of their

structure to create a pulse modulator that can be adjusted according to load

attributes, and introducing them as new generations of pulsed power supplies;

second, improvement of the performance of current pulse generators in

particular, MGs in terms of efficiency, flexibility, reliability and redundancy.

The development of strategies for enhancing the pulsed power supply processes

is the general scope of this thesis. Multifold parts including the power supply

structure, energy transferring path and the control algorithm can be considered,

and several configurations, methods and techniques can be adopted to improve

the various operational parameters of these processes. In particular, increasing

the efficiency of pulse generators can significantly contributes to promoting

pulsed power technology and to making it financially viable for many other

applications. Therefore, substitution of gas/magnetic based switches with solid-

state switches is considered as an appropriate option for cost, loss and volume

reduction considerations. A number of power electronics converters offer

attractive advantages for pulsed power applications; however, their compatibility

with various load characteristics and demands (the generalization) is still a

challenging issue. From these circumstances, Problem # 1 arises:

Problem # 1: Major drawbacks of available pulsed power technologies

including low efficiency, lack of flexibility, short life span and low

operational frequency

Recently developed solid-state technology demonstrates many

favorable switches with high voltage ratings, and switching frequency

that make them suitable candidates for pulsed power generation.

Insulated gate bipolar transistors (IGBT), Metal-oxide semiconductor

field-effect transistors (MOSEFET), and Silicon-controlled rectifiers

(SCR) are the semiconductor switches which can be utilized as

reasonable replacements for existing switches. Nevertheless the

4

voltage ratings of these commercially available switches still cannot

satisfy many application requirements. Thus, several power

electronics topologies with voltage boosting capabilities and their

associated control methods are considered and their performance

properties are fully investigated in order to present an optimum design

for supplying plasma applications with maximum efficiency and

flexibility.

On the other hand, MG has a number of appreciable capabilities to offer in

supplying pulsed power. Its simple structure, control method and principal

switching modes are among attributes which maintain its still practical benefit in

many applications. Generating high voltage based on aggregation of several

lower voltages is the main specification of MG: this makes it an ideal target for

development by utilizing semiconductor switching devices such as IGBTs and

MOSFETs. However, the large number of active devices required in its structure

increases the initial cost and the operation losses. Each stage of conventional MG

consist of one capacitor, one switching device and two power diodes (or resistive

insulators in earlier technologies) that contribute to charging and discharging

paths and increase the losses. Solid-state technology has already been utilized in

MG structure in a number of studies. The outcomes demonstrate solid-state

adjustability and reliability with application demands in voltage stress.

According to reports in previous studies, concerns regarding switching transients

and simultaneous connectivity of switches have been removed. However, the

high level of power losses due to utilizing many active components in its

architecture, and the lack of flexibility in supplying loads with variable voltage

demands, strengthen the necessity for having proper designs for MG topology.

The second research problem is inspired by these concerns.

Problem # 2: Redundancy of solid-state based Marx topology

Reducing the number of semiconductor devices needed for MG units

is a way to cut down the initial cost and the power losses. Replacing

fast switching devices that require elaborate driving stacks with

slower ones is another solution that leads to a circuit with fewer

driving modules. Several issues-including switch specifications and

5

capabilities, and application demands-must be taken into consideration

in the design process. New configurations considering power

electronics circuits and switching techniques can be developed to

reduce the number of needed components and to enhance the

performance of Marx in terms of flexibility and efficiency. Taking the

ladder-shape structure of Marx into account, a few modifications in

the cascade connection of stages can result in a huge reduction in the

number of components.

6

1.2. L i terature Review

1.2.1. Introduction

High voltage, short duration pulses known as pulsed power is very much in

vogue these days. Pulsed power is the term used to describe the science and

technology of accumulating energy over a relatively long period of time and

releasing it very quickly thus increasing the instantaneous power. Steady

accumulation of energy followed by its rapid release can result in the delivery of

a larger amount of instantaneous power over a shorter period of time (although

the total energy is the same). Energy is typically stored within either electrostatic

fields (capacitors), or electromagnetic fields (inductor), as mechanical energy

(using large flywheels connected to special purpose high current alternators), or

as chemical energy (high-current lead-acid batteries, or explosives). By releasing

the stored energy over a very short interval (a process that is called energy

compression), a huge amount of peak power can be delivered to a load. For

example, if one joule of energy is stored within a capacitor and then evenly

released to a load over one second, the peak power delivered to the load would

only be 1 watt. However, if all of the stored energy was released within one

microsecond, the peak power would be one megawatt, a million times greater.

Pulsed power was first developed during World War II for use in radar system. A

massive development program was undertaken to develop radars requiring short

high power pulses. After the war, development continued in other applications

leading to the super pulsed power machines at Sandia National Laboratories,

located in Kirtland Air Force Base, Albuquerque, New Mexico, USA [1].

The pulse attributes vary, based on different applications. The pulse widths and

voltage levels are mostly in the range of 510 1010 −− − s, and 83 1010 − v, respectively.

The required energy is defined based on load demands and varies in the range of

1010 3 −− J. These high voltage pulses with variable pulse widths should have

rising and falling times in the low 69 1010 −− − s regime. Pulse repetition is also

defined by application. Although demand for highly repeated pulses has grown

over last two decades, there are still many applications for single shot pulses. A

pulse pattern including the features indications is depicted in Fig. 1.1.

7

Fig.1. 1. Pulsed power features

1.2.2. Applications

The applications of pulsed power can be classified in four major categories

including industrial, environmental, medical, and military applications. Although

single shot based pulsed power generators with extremely high peak power have

been considered initially for military and nuclear fusion applications, repetitively

operated pulsed power generators with a moderate peak power have been

recently developed mainly for industrial and environmental applications such as

food processing, medical treatment, waste water and exhaust gas treatments,

concrete recycling, ozone generation, material processing, particle accelerating,

engine ignition, and ion implantation [2, 3].

Examples where pulsed power technology is commonly used in the military

include development of radars, electromagnetic launchers, and laser guns. For

medical treatments, many studies are recently being performed on the effect of

pulse electric/electromagnetic fields on a cancer cell and the treatment methods

are progressing. Electroporation by pulsed power has been known for the

medical application of gene manipulation since early times. Recently, many

researchers have interests in using shorter pulse durations for this purpose. For

example, an electrode microchamber with dimensions suitable for live

mammalian cell has been studied in [4]. Extremely short pulse electric fields or

high frequency electric fields that are greater than 10 MHz can be applied to the

nucleoplasm. Then apotosis of the cancer cell can be induced. Nuccitelli et al

applied a pulse electric field to melanomas in a mouse and showed that a

nanosecond pulse electric field caused cell death of the melanomas in the mouse

[5]. The effects to the cell were also investigated by others. Nomura et al.

8

investigated the effect of intense burst ac electric field on the cell [6]. They

showed that a burst electric field of about 25 kHz on the cell caused

electroporation. On the contrary, a burst electric field of about 50 MHz caused

fragmentation or degradation of DNA within the cell.

For biological applications of PEF, electroporation is usually used to sterilize

bacteria. This technique is commonly applied for sterilization in food processing.

El-Hag et al investigated inactivation of microorganisms naturally contaminated

in orange juice [7]. The naturally contaminated microorganisms in orange juice

are more difficult to inactivate by PEF than added unnatural microorganisms.

For crop growth, gas discharges were used for cultivation of mushrooms by

Tsukamoto et al [8]. They used a spark discharge applied to sawdust pots used

for planting fungus. The group with the applied spark discharge had a twofold

gain of Shiitake (Lentinula edodes) mushrooms. Other mushrooms were also

investigated by applying a spark discharge. Those were buna-shimeji

(Hypsizygus marmoreus) mushroom and eringi (Pleurotus eryngii) mushroom.

The crops of those mushrooms increased 15 % by applying the spark discharge.

Pulsed power technologies are also utilized for material processing. Plasma-

based ion implantation and deposition (PBII&D or PBIID) is one of the modern

technologies employing pulsed power technique for surface treatment of

complex shape materials. PBIID can now be considered a mature technology for

surface modification and thin film deposition after pioneering work in the 1990s

[9]. Other applications of the pulsed power technologies in this area can be

counted as material ablation, surface heating (annealing) and new material

synthesizing. Pulsed high-power lasers are employed to ablate solid materials as

ion or neutral particle sources for new material synthesis and/or film depositions

[10]. Pulsed high-power lasers [11] and high-power microwaves are also used to

heat material surfaces (annealing) [12]. Metal foil evaporation with pulsed large

currents are used for synthesizing fine particles (nanocomposite powders),

joining of solid material (ceramics) [13] and plasma generation used as ion

sources [14, 15]. Energetic beams such as lasers are powerful tools to modify the

surface of materials. Various types of laser such as ruby, Nd:YAG, Ti:sapphire,

excimer XeCl, and CO2-laser have been employed for surface modification [11].

High power micro- or millimeter-wave beam can also be employed to heat a

9

material surface. This method is especially effective on dielectric materials [13].

Exploding (fusing) metallic foil or wire can also be used for nanocomposite

particles synthesis, and joining dielectric materials such as ceramics [14, 15].

A 170 kV laser-triggered water switch was developed by Woodworth, et al [16].

They obtained Schlieren photographs of the laser-induced breakdown and

considered the effect of a string of hot dense point plasmas formed with the laser,

as shown in Fig. 1.2.

Fig.1. 2. Time-resolved Schlieren photograph of laser-induced breakdown in a 170 kV switch a

few nanoseconds before breakdown [16]

Decomposition of harmful gases, generation of ozone, treatment of algae bloom

by discharge plasmas in water, and concrete recycling [17-20] are among

industrial and environmental applications in which, repetitive operation and long

lifetime are necessary in the pulsed power generators.

Other industrial applications of pulsed power are as follows:

Gaseous phase pollution control

Non-thermal plasmas produced by a dielectric barrier discharge (a silent

discharge), a surface discharge, a dc corona discharge, and a pulsed corona

discharge, have been well known to have a strong influence on activating

chemical reactions in the gaseous phase. In the last few decades, researchers

have tried to utilize them in many applications such as control of NOX and SOX,

treatment of dioxins, removal of volatile organic compounds, generation of

ozone, and excitation of excimer lasers. On the other hand, the development of

the pulsed power technology has led researchers to use maintenance-free pulsed

power generators with repetitive operation that can continuously produce large-

volume non-thermal plasmas via pulsed streamer discharges. Pulsed streamer

discharges have been utilized to remove numerous hazardous pollutants due to

10

the higher energy efficiency to produce chemically active radicals which react

with pollutants [21, 22]. In the research field of NOX removal by pulsed streamer

discharges, it is well known that a shorter pulse duration of applied voltage to the

discharge reactor has a strong influence on improving the energy efficiency for

reduction of pollutants [23, 17, 24].

Cleaning of lake and dam algea bloom by discharge plasmas in water

Phytoplankton proliferates rapidly in lakes and dams, and it appears as if a

bluish-green powder is scattered on the water surface. When nitrogen and

phosphorous, nutrition for phytoplankton, flow into the lakes and dams, rapid

proliferation of the phytoplankton occurs. The increase of these nutrients is

mainly caused by the increase in human activity around the lakes and dams.

Typical phytoplankton in lakes and dams with the eutrophication state are

Anabaeba and Microcystis. A water surface with the bluish-green powder is

called a water bloom. Water bloom is ugly and smells bad. Furthermore, it

changes the aquatic environment by blocking the sunlight. Toxins have also been

observed in one kind of Microcystis. Therefore, the need to treat water bloom is

a serious environmental problem all over the world. Treatments for water blooms

have been investigated using chemical compounds, ultrasonic [25], microwave,

electrolysis and mollusks. However, these methods are not feasible at the present

time due to economical, effectiveness, and environmental reasons. As a recently

developed method, streamer-like discharges in water produced by pulsed power

generators are considered as a water bloom treatment method [26, 27]. A photo

of discharge in water is given in Fig. 1.3.

Fig.1. 3. Typical still photograph of a discharge induced in water[28]

11

Recycling of concrete by pulsed power

The concrete scrap recycling has been enforced in many countries by justifying

new regulations. Presently, most concrete scrap is recycled as a lower sub-base

coarse material. However, it is foreseen that concrete scrap will increase rapidly

and exceed the demand of road sub-base in the near future. The technology to

produce high quality recycled coarse aggregate must be developed to keep the

percentage of recycled concrete scrap high. Heating and rubbing is one

developed method of producing recycled aggregate. However, the problem is

that too much energy is consumed to heat and rub the concrete. Using pulsed

power is an option to carry out concrete recycling with a higher efficiency and a

proper outcome [29-31]. Fig. 1.4(a) and 1.4(b) show photographs of a discharge

and concrete scooped out by discharges, respectively.

(a) (b)

Fig.1. 4. Photographs of (a) discharge and (b) concrete scooped out[2]

Plasma systems are one of the applications of pulsed power. Exciting desired

materials via these pulses results in separation of components in the form of ions

and electrons and composition of plasma. Generation of plasma has a great

significance in many industrial activities such as fusion energy, ozonising,

recycling and etc. Demands for Silent Discharge Plasma Systems (SDPSs) are

growing as environmental problems are rising. Hence, the enhancement of

SDPSs has significant environmental implications. If SDPSs can be improved,

then there are likely to be substantial benefits in the area of diesel exhaust filters,

waste-water treatment, ozone production, and fuel conversion systems. There

will also be benefits in other areas where plasma plays an important role. SDPS

technology is already being used to address these areas of need, but to a limited

extent because of the inefficient nature of state-of-the-art SDPSs. From whole

12

literature review conducted in SDPSs applications, the use of SDPSs in the areas

of significant need are highlighted that can be classified as follows;

Enhanced SDPSs for diesel engines:

The polluting content from vehicle exhaust arises mainly during the cold starting

of an engine. If, during this period, the hydrogen rich gas generated by a plasma

converter can be directed into the engine to react simultaneously with the petrol,

the combustion rate of the engine can be significantly enhanced. As a result, the

fuel consumption and emission of the pollutants HC, COX, and NOX can be

greatly reduced [32-34]. The removal efficiencies of the pollutants depends

strongly on the type of applied voltage (dc, ac, and pulsed) and on other factors

such as the geometry of the reactors. Research into the most effective applied

voltages provides the means to achieve these voltages, and reactor shape (as

proposed in this project) can substantially improve the effectiveness of exhaust

filtering.

Enhanced plasma technology for indirect conversion of natural gas to

liquid fuels:

Selective methane conversion to valuable chemicals and liquid fuels by indirect

methods is being investigated by a number of researchers. Conventional catalytic

methane conversion has limitations in the area of product yield and selectivity,

required reaction temperature, controllability, and catalyst poisoning due to the

frequent presence of traces of H2S. Research indicates that plasma technology

has substantial potential for overcoming these problems [35, 36]. SDPSs may be

able to provide new, effective, and practical solutions to decomposing harmful

H2S bi-products into useful products (including hydrogen) for full cell and other

energy applications. Enhancing the efficiency of plasma systems is pivotal to

having SDPSs accepted into fuel production technology.

Enhanced use of SDPSs in ozoniser and wastewater treatment:

Bombardment by electrons can break oxygen molecules apart, which in turn

encourages the formation of ozone molecules. The most commonly employed

type of electrical discharge in commercial ozone generators is silent discharge,

which has been employed in the gas phase for a long time. Research clearly

shows that the type of excitation strongly influences the yield and efficiency [37,

38], but it is unknown what type of excitation is optimal. Determination of the

13

optimal excitation waveform, along with practical strategies for creating this

waveform, could deliver more effective and practical ozone production systems.

For all applications discussed above, SDPSs have, as yet, un-tapped potential.

These SDPSs need to be made more efficient, cost-effective, and less prone to

EMI. Practical SDPSs are multi-faceted; they involve advanced electrochemical

(plasma) reactions, advanced power electronic excitation, and advanced

electromagnetic interactions within the reactor. There has been, to date no

significant attempt to jointly optimize the performance of all three components,

and indeed little attempt to get multi-disciplinary experts working together.

Some of the individual components have not even been designed with

established best practice. The power electronic excitation, for example, is

typically created from single level pulse width modulation (PWM), despite the

fact that this kind of approach is known to give high losses, low efficiency, and

high EMI for these applications.

1.2.3. Pulsed power supply technologies

Several studies and reports on the area of pulsed power have been investigated to

identify and classify available pulsed power supply technologies and their

properties, including the advantages and disadvantages. Four topologies have

been arisen from those technologies that seem to have had the most applications.

1.2.3.1. Magnetic pulse compressor (MPC)

MPC is the technology used for decades to produce high-voltage pulses for many

applications. Two schematics of MPC with different configurations are shown in

Fig. 1.5.

Fig.1. 5. Two schematics of MPC

Several applications of MPC are reported in a diverse range of laser radiations.

An all solid‐state MPC with amorphous metals has been used in a research for

pumping a repetition‐rated krf excimer laser [39]. In a similar effort, a Thyristor,

14

silicon controlled rectifier (SCR), switched multistage MPC has been

successfully developed for krf excimer laser excitation in [40]. Being entirely

composed of solid‐state elements, this exciter will provide the extremely long

lifetime at high repetition rate operation that cannot be realized by a conventional

power supply with a discharge switch. To reduce the load on switching elements

in the excitation circuit of excimer lasers, a new circuit has been developed in

[41] where a saturable transformer is utilized in an MPC. The high-

repetition‐rate operation of a transversely excited atmospheric CO2 laser pumped

by an all‐solid‐state exciter consisting of a two‐stage MPC has been tested in

another study [42]. A different topology using IGBT based pulse generator and

an MPC for similar purpose is proposed in [43]. The development of industrial

excimer and CO2 TEA pulse lasers with average optical powers in the kilowatt

range requires pulsing circuits delivering average power levels of tens of

kilowatt. Excitation pulses with voltage rise times in the order of 100 ns and

peak voltages of more than 40 kV are required for efficient operation of the laser.

Pulses of 40 µs duration are supplied to a four-stage series MPC to transform the

pulses down to 150 ns for efficient laser excitation. Development of two

high‐efficiency, all‐solid‐state two-stage MPC as exciters for pumping TEA CO2

lasers is reported in [44], including experimental evidences for their energy

scalability, high‐efficiency and high repetition rate operation. A high‐voltage

pulse output modulator using a step‐up transformer with two stages of MPC

circuits has been developed in [45] as a spiker for the purpose of obtaining the

breakdown of the gas mixture when using the spiker‐sustainer excitation

technique for pumping XeCl discharge lasers. A similar study is reported in [46].

The features of Thyristor-driven pulsers for multikilowatt average power lasers

that are achieved using MPC are addressed in [47]. In an X-ray preionized Ar-Xe

laser using a magnetic-spiker sustainer discharge circuit with an MPC, a

parametric investigation of the laser output versus gas composition and electrical

excitation is performed in [48]. A considerable improvement in the performance

of the discharge excited Ar-Xe laser is demonstrated in this work.

A number of studies report the application of MPCs for waste water treatment.

The application of high repetition and short duration electrical pulses made by

MPC systems for inactivation of spores, bacteria (Escherichia coli) and viruses

15

for the purpose of purification of drinking water has been investigated in [49].

Feasibility studies of EMTP simulation for the design of the pulsed power

generator using an MPC and a Blumlein-type pulse forming network (BPFN) for

water treatments has been presented in [50]. Electron temperature and electron

density of underwater pulsed discharge plasma produced by solid-state pulsed

power generator was studied in [51]. An all-solid state MPC was developed and

used in this research to create the electrical discharge in water.

Gaseous treatment and ozone generation are other areas in that MPCs have been

utilized so far. An MPC was used to control the exhaust gases from a diesel

generator employing a wire-to-plate plasma reactor [52]. Investigation of

operational regimes of a high-power pulsed corona source with an all-solid state

pulser (one-stage MPC) is considered in a report of an ongoing effort on

development of efficient, compact pulsed corona sources for pollution control

applications [53]. However a novel magnetic compressor circuit improving the

coupling to PC discharge was already proposed and evaluated in a similar

manner for pollution control applications through corona source [54].

An MPC in a different application is employed to provide the HV and current

pulses for an experimental investigation of HV short pulsed streamer discharges

in dry air-fed ozonizers under various operating conditions [55]. The production

of ozone is also investigated in another work [56] using a dielectric barrier

discharge in oxygen, and employing short-duration pulsed power generated by

an MPC.

The reusable linear Magnetic Flux Compressor (MFC) is an ideal pulsed power

supply for the electric gun. Effect of mutual inductance on the pulsed current

amplification of MFC is investigated through some numerical experiments in

[57]. Resistance calculation and the design of the reusable linear MFC coil are

mentioned in [58] and [59] respectively. The design, construction, and testing of

miniature, high-power MFCs are presented and discussed toward a

comprehensive task in [60]. Energy conversion and high power pulse production

are made feasible through this research using miniature MFCs. The MFCs are

located inside high-speed, 30-mm projectiles that are launched with a high-

pressure helium gun to velocities of approximately 300 m/s.

16

Some studies have also been undertaken to evaluate the attributes of used

components in MPC’s structure and the effect of utilizing different materials in

MPC’s performance. Simulation of 3-staged MPC using custom characteristics

of magnetic cores is discussed in [61]. A low inductance circuit has been

fabricated in this study in order to obtain shorter time to full saturation of an

unsaturated core and current pulses with shorter widths during core saturation.

An MPC test stand is developed in [62] to evaluate the switching and loss

properties of magnetic core materials that included ferrite and several alloys of

nickel-iron, Metglas, and the nanocrystalline material Vitroperm.

1.2.3.2. Pulse Forming Network (PFN)

PFN has been extensively in use from early 50s due to its attractive method of

compression of electrical field and production of high-voltage pulses. A number

of resonant units including inductive and capacitive components are connected in

ladder shapes compressing the electrical field through magnetizing energy. Six

different configurations of PFNs proposed so far are shown in Fig. 1.6.

+- +- +- +-

Fig.1. 6. Different types of PFNs

Design specifications of PFN are discussed for several applications during these

years. Among them a computer simulation of PFN conducted in CERN (The

European Organization for Nuclear Research) [63] allowed research scientists at

TRIUMF (Canada's National Laboratory for Particle and Nuclear Physics) to

achieve the tightest specification to date for a PFN. While the construction of a

linear PFN for a constant load impedance is relatively easy, the process is more

difficult for a nonlinear or time-varying load. A passive PFN can certainly be

synthesized for nonlinear loads, but is usually large and lacks the flexibility to be

17

truly useful in most practical and research applications. An investigation is tried

to describe the design and construction of a sequentially-fired pulse forming

network (SFPFN) that maintains constant voltage and current for a nonlinear load

[64]. The principal advantages of the SFPFN over its passive analog including its

utility in both linear and nonlinear load applications are discussed in another

report [65]. A PFN with time-varying or nonlinear resistive loads is also looked

into in [66]. A figure of merit that considers frequency and time responses is

defined for PFNs in [67] and a class of filters was found that is best in the sense

of this criterion. Another class of PFN is also proposed in [68] attempting to find

best compromise between the fastest possible rise time and the least possible step

response overshoot for the produced pulses.

The current and voltage wave shapes of a PFN commutation circuit in a newly

proposed modified McMurray inverter have been compared with conventional L-

commutation circuit in [69]. In another study, a recently developed vacuum

switch capable of conducting a current pulse in excess of 300 kA peak for a

duration of approximately 0.5 ms is used in a 1.2 MJ PFN in order to remove the

related concerns to application of vacuum switches in PFNs [70].

Recently, there has been a considerable growth in application of PFNs for

energizing railgunes. A procedure for optimal design of a PFN feeding an

electromagnetic launcher (EML) is presented in [71]. Given the design

parameters (pulse duration, pulse rise time, pulse current amplitude and load

equivalent resistance), the procedure gives the value of the inductances and

capacitances for an optimal design of an L-C ladder feeding network. In another

project, investigation of an alternator charged PFN with flywheel energy storage

to power an 18-shot, salvo fire 30 mm railgun is presented through a conceptual

design in [72]. The PFN program complex for parameter calculation of capacitive

energy store is developed for obtaining a preset shape current in a railgun

launcher. A different scheme analysis of capacitive PFN is given in [73] for

obtaining current pulses with a flat peak in the railgun launcher. A compact and

highly modular PFN, based on semiconducting switches, for electric gun

applications (railguns, electrothermal-chemical guns) has been developed and

built up at the French-German Research Institute of Saint-Louis (ISL) [74]. A

PFN design for blocked-bore plasma armature experiments has been developed at

18

the University of Texas at Arlington (UTA) where the delivery of a single high-

current (with a ramping and a continuously positive di/dt waveform) pulse to a

railgun was required [75]. The development of high density, volume efficient

capacitive PFNs, designed to maximize electrical energy transferred to the gun, is

critical for the future weaponization of both electro/thermo-chemical and

electromagnetic launchers. [76] presents some design considerations for a

fieldable PFN/launcher system.

PFNs are also used in other areas including silent discharge plasma, and

microwaves. The combustion of solid propellants subjected to plasma

augmentation, has been studied with a 300 kJ maximum stored energy PFN in the

range of 1 kJ/g of electrical energy over a 1.2 ms pulse length [77]. An improved

PFN for generation of phase-coherent microwave signals (correspondence) is

reported in [78]. TEM mode networks used in this design to produce phase-

coherent pulse-modulated microwave signals through the S-band portion of the

spectrum. A linear, single-stage, nanosecond PFN for delivering intense electric

fields to biological loads is introduced in [79]. A brief discussion of the

Darlington PFN followed by a theoretical study of the effects of a series

inductance at the input terminals and a shunt capacitance at the output terminals

on the shape of the pulse produced was addressed in an early study [80].

1.2.3.3. Multistage Blumlein Lines

Blumlein pulsers are well-suited devices for high-voltage pulse generation in

nanosecond and microsecond ranges. They are attractive for pulsed power

applications due to their skill in producing flexible pulses with variable durations

and polarities. High power long pulse operation is feasible through these

modulators. These generators have been used with great success in several areas

such as in breakdown tests, X-ray generation, radars, lasers, high-energy plasma

implantation, and biomedical studies. They consist of lengths of transmission

lines charged in parallel and synchronously discharged in series into the load by

using single or multiple switches at the opposite line endings. The configuration

of this pulse supplier is given in Fig. 1.7.

19

Fig.1. 7. Samples of typical multistage blumlein pulsers

The main problem with the device performance is the presence of the shield cable

impedance contributing to the Blumlein power loss especially when using only

one switch. A very well known technique is used in [81] for minimizing these

losses consists of winding the transmission lines to increase the line shielding

inductance if coaxial cables are used.

High-power, repetitive-stacked Blumlein pulsers commutated by a single

switching element are proposed in [82]. The stacked Blumlein pulse generators

developed at the University of Texas at Dallas (UTD) consist of several triaxial

Blumleins stacked in series at one end. The lines are charged in parallel and

synchronously commutated with a single switching element at the other end. In

this way, relatively low charging voltages are multiplied to give a higher desired

voltage across an arbitrary load.

Characterization and analysis of a general purpose pulse power system based on

MG and Blumlein is carried out in another study [83]. A Blumlein configuration

for high-repetition-rate pulse generation of variable duration and polarity using

synchronized switch control is introduced in [84]. A compact high power pulsed

modulator based on spiral water Blumlein line, which consists of primary storage

capacitors, a Tesla transformer, a spiral Blumlein line of water dielectric, and a

field-emission diode, is described in [85]. Design of a 150 kV, 300 A, 100 Hz

Blumlein coaxial pulser for long-pulse operation is investigated in another work

[86]. For large pulsed power generation, one critical issue for such a single-switch

based circuit topology is related to large switching currents. In an article, the

authors propose a novel Blumlein circuit topology based on multiple switches

[87]. The pulsed forming lines are charged in parallel and then are synchronously

commutated via multiple switches. No special synchronization trigger circuit is

needed for the proposed circuit topology; this robust circuit topology is simple

and very reliable.

20

Photoconductive switch enhancements and lifetime studies for use in stacked

Blumlein pulsers are conducted at UTD and the results are presented in [88]. In

another application a reliable 500 kV, 100 kA (each arm) GEMINI dual Blumlein

accelerator is developed and implemented at the Air Force Weapons Laboratory

in New Mexico [89]. Novel isolation, trigger, and trigger timing techniques are

utilized in this device in order to allow two Blumleins to be charged by one MG,

and discharged at different times. A high-voltage pulsed power supply of 100

kV/200 A with output short pulses of the order of 1 µs (based on stacked coaxial

Blumlein technology) was developed for use in surface treatment of materials by

plasma implantation [90].

1.2.3.4. Marx Generator (MG)

MGs have been extensively in use since first introduced by Erwin Otto Marx in

1924. In an ideal performance, the charged capacitors in parallel are connected

through switches in series to produce a high voltage pulse of (nVin) where Vin is

the input voltage and n is the number of capacitors (stages). However due to

several practical constrains the ultimate produced voltage is less than level.

Spark gap technology was traditionally used as switching devices in MGs [91].

The circuit diagram of an MG is shown in Fig. 1.8. The evolution in MG design

and some applications using spark gap based MGs are addressed in this section.

Fig.1. 8. MG with spark gap switching and resistive insulation

The dual channel triggering of a spark gap switch by fiberoptic transported ruby

laser radiation is discussed in [92]. The spark gap is the output switch of a 20-ns

water dielectric Blumlein generator. The Blumlein generator in this set up is pulse

charged in approximately 250 ns by a three-stage Marx bank to 150 kV.

In another work, starting from the project requirements, a 1.2-MJ pulsed power

supply (PPS) module for a high-power laser system has been developed [93]. The

main circuit of this module consists of a high energy density capacitor bank, a

spark gap switch, a magnetic switch, a trigger generator, and a load subsystem.

Since a two-electrode spark gap switch has no separate trigger electrode and must

21

be directly overvolted by the trigger generator, a small Marx generator has been

designed and constructed for the advantage of the proven compactness and

electrical performance.

The development and characterization of a repetitive Marx-Generator-Driven

Reflex Triode System rated at 1 kJ, 300 kV, 12 kA, and 10 Hz and suitable to

drive the load directly is discussed in [94] aiming high-power microwave

generation. Bipolar charging of an MG scheme has been adopted to get both

faster rise time and relatively low charging voltage. The electrical characteristics

and design features of a low inductance, compact, 500 kV, 500 J, 10 Hz repetition

rate MG for driving an high-power microwave (HPM) source are discussed in

another study [95]. Including the spark-gap switches, and benefiting from the

large energy density of mica capacitors, four mica capacitors were utilized in

parallel per stage, keeping the parasitic inductance per stage low.

A repetitive ten-stage wave erection MG is developed in [96] to investigate the

electrical characteristics of such compact devices and potentially provide an

economical approach to realize the miniaturization of intense electron beam

accelerators. Compact design has been made for the generator in this project in

order to achieve a proper stray capacitance of the spark gap electrode with respect

to the ground in each stage because these proper grounded stray capacitances are

critical for obtaining a good wave erection process.

A concept for compact, megavolt MGs has been developed in [97], resulting in

several designs which are approximately half the diameter and half the height of

conventional units. The customized Marx capacitor assemblies utilize multiple

windings incorporated into a single common capacitor case. Spark gap switch

electrodes extend directly from the external capacitor terminals, eliminating the

need for additional buswork.

Laser applications also used spark gap based MGs in a wide range. A coaxial MG

triggered with a UV laser pulse propagating coaxially through multigap switches

is constructed and the design process is reported in [98]. The MG is operated at

maximum voltage of 200 kV with a rise time of less than 10 ns. To trigger

multigap switches in the MG, the laser pulse is passed through fine metal mesh

fitted in the holes formed along the central axis in electrodes of gap switches.

22

A portable hard x-ray source has been developed for nondestructive testing,

medical imaging, and flash x-ray radiography. The source is powered by an MG

that produces a 200 kV, 1.2 kA pulse of 60 ns full width at half maximum [99].

There is another article describing a new simulation model developed with

PSPICE in order to improve the ultra compact MGs designed at the French-

German Research Institute of Saint-Louis (ISL). The proposed model is based on

a Marx elementary unit and is an equivalent electric circuit that matches the

actual configuration of the generator. It consists of a structural description of the

elementary stage of an MG including stray components. It also includes a

behavioral model of the spark gap switches based on the Vlastos formula

determining the arc resistance value [100]. Another paper explains the design and

production of two autonomous ultra-wide-band (UWB) radiation sources. These

sources consist of a high-gain broadband antenna that is driven by one of two

subnanosecond pulsed power sources. Each source is made up of an MG and a

pulse-forming device based on the use of a gaseous spark gap [101]. In a different

study, an optical filter is employed to transport a 15-ns light pulse from a high

power ruby laser for precise triggering of a gas filed high voltage spark gap. Pulse

charging of the Blumlein generator was accomplished by a three-stage MG,

resulting in output voltages up to 250 kV. It was conclusively demonstrated that

an optical fiber will transport a sufficiently intense laser pulse to evince

subnanosecond jitter in the triggering of a pressurized gas switch under the

conditions studied in [102]. A description is given of a nanosecond-rise time,

750-kV spark gap impulse generator that is triggered by an ultraviolet laser pulse

(KrF, λ=249 nm). The SLITS (SF6 laser-induced triggering system) is an add-on

fast switching capability that makes use of the existing equipment (Marx, ac or

dc) and can be triggered at voltages as low as 10% of the spontaneous breakdown

voltage of the gap [103]. An experimental study of the characteristics of hard x-

ray emission in laser-induced vacuum spark discharges has been carried out and

acquired results have been addressed as a paper in [104]. The spark discharge is

performed in a gap (10 mm) of pin electrodes using an MG and a laser pulse to

produce Au plasma on the tip of an anode.

Spark gap over-voltages of fast MGs in the 100-300 kV range were studied in

[105] as a function of stray and inter-stage capacitance for several circuit designs.

23

Overvoltage measurements were made in this work at low voltages by simulating

the spark with a silicon-controlled rectifier (SCR).

A measuring system based on a resistive divider using a copper sulfate (CuSO4

solution) has been developed in a study to measure high-voltage impulses with

rise times of several tens of nanoseconds. The response and temperature

characteristics of the resistive divider have been studied in this work. The divider

has excellent response characteristics (<1-ns theoretical rise time) and good

immunity from interference arisen from an MG and spark-gap switch [106].

After a thorough study of available technologies accompanied by a deep and

particular overview of associated short comes in SDPSs using these topologies, a

few lacunas have been distinguished that do not permit the extensive application

of these systems for industrial purposes. Lack of compactness, and reliability,

being bulky and expensive having intricacy, low repetition rate, and short

effective operational life time are amongst those made available technologies

somehow challenging. Solid-state technology has demonstrated promising

attributes enabling the pulsed power supplies to overcome discussed problems

and performing a proper and more efficient supplying process. Although solid-

state switches have been already utilized in MGs configuration, there are still

other power electronics topologies can be utilized to bring more advantages of

solid-state technology to pulsed power generation

1.2.4. Power electronics in pulsed power generation

Therefore, this thesis has been conducted on designing novel topologies for

pulsed power purposes considering power electronics converters topologies,

applications, control systems, and their benefits, and limitations so that they can

result in the aforementioned objectives in future plasma system and pulsed

power technology. It is essential to have a detailed review of converters such as

developed solid-state MGs, dc-dc converters, capacitor-diode voltage multipliers

(CDVM), power factor correctors (PFC), and high frequency inverters that have

potential to operate as future pulsers.

24

1.2.4.1. All-solid-state Marx Generator

Although spark gap/magnetic switches were conventionally used as switching

devices, solid-state technology being recently utilized at MGs configuration has

improved their performance in terms of efficiency and reliability. The circuit

diagram of a solid-state MG in charging and discharging modes are shown in

Fig. 1.9. A number of recent applications used solid-state MGs are addressed in

this section.

Fig.1. 9. (a) An all-solid-state MG, (b) Charging mode, (c) Discharging mode

An all-solid state MG is employed in [107] to energize microplasma

applications. The proposed circuit in this reference employs two parallel MGs

utilizing bipolar junction transistors (BJTs) as closing switches. The BJTs are

operated in the avalanche mode to yield fast rise times. The design allows for

positive or negative polarity pulses, and can easily be changed to yield higher or

lower output voltage. Due to the advantages of Marx topology, many high

voltage applications are assigned to this generator.

A high-voltage bipolar rectangular pulse generator using a solid-state boosting

front-end and an IGBT based H-bridge output stage is presented in [108] and the

generated pulses are intended to be used in algal cell membrane rupture for oil

extraction. In another study, an all-solid-state pulsed power generator consists of

a Marx modulator based on discrete IGBTs and a magnetic pulse-sharpening

circuit, which is employed to compress the rising edge of the Marx output pulse

25

is proposed in [109] in order to reduce the influence of relatively slow turn-on

speed of the IGBT on the pulse rise time of the Marx modulator. An MG

topology based on commutation circuit is also proposed in [110, 111] that

utilizes IGBTs and SCRs simultaneously. On the other hand, an experimental

MG with MOSFET switches was used in [112] to generate pulsed output

voltages of up to −1.8 kV in order to produce pulsed power microplasma

discharge in N2 gas and N2/NO gas mixture for atmospheric pollution control

purposes. In another application this MG is used for the surface treatment by

microplasma of PEN (polyethylene naphthalate) film using Ar gas and mixtures

of Ar with N2 and O2 [113]. Improving Indoor Air Quality (IAQ) through

decomposition of formaldehyde (HCHO) by a microplasma reactor is another

subject investigated in [114] at a discharge voltage of 1.3 kV using a high

voltage amplifier and an MG with MOSFET switches as pulsed power supplies.

A 200 kV pulsed power supply based on an MG composed of 20 stages and each

stage is made of an IGBT stack, and two diode stacks, and a capacitor has been

presented and implemented in [115]. Authors of [116] have developed MG

technology by substituting solid-state switches like IGBTs and series connected

diodes, instead of insulation components such as spark gaps. They could provide

the pulsed power systems with compactness, reliability, high repetition rate, and

long life time. The rising of pulsed power generators using solid-state devices

eliminates limitations of conventional components, and promises pulsed power

technology to be widely used in commercial applications. A novel bipolar high-

voltage modulator topology, based on MG concept, is proposed by the authors of

[117], for high-voltage repetitive pulsed power applications. The proposed

topology is a generalized version of the negative and positive all-solid-state

Marx modulator concept [118], which takes advantage of the intensive use of

power semiconductor switches to increase the performance of the classical

circuit, strongly reducing losses and increasing the pulse repetition frequency.

Additionally, the proposed topology enables the use of typical half-bridge

semiconductor structures while ensuring that the maximum voltage blocked by

the semiconductors is the voltage of the capacitor in each stage.

Another novel solid-state high power pulse generation technique has been

suggested in [119], suitable for a wide range of pulsed power applications. The

26

technique, termed as multi-level pulsed power converter, can be considered as a

hybrid of the direct discharge type and the MG but with considerably less

complexity in both control and circuitry. It has the ability to generate pulses with

flexible amplitude and duration similar to that of an MG, and unlike the direct

discharge type requires no voltage balancing and snubbing circuitry. As voltage

balancing is inherent to the technique, the timing between switching events is not

critical for balancing the voltage stresses, and as such the driving circuitry of the

converter is relatively simple. In another case, an idea of compact MG for

repetitive applications was discussed in [120] and preliminary tests were

conducted at the condition of low charging voltage using nitrogen gas.

The semiconductor technology is also exploited in low power applications of

MG such as radar transmitter and receiver. High power variable nanosecond

differential pulses generators for ground penetrating radar (GPR) systems based

on avalanche transistor and Marx Bank are investigated theoretically and

experimentally in [121]. Using avalanche transistor as the switch of Marx circuit,

a new type of all-solid-state low-power pulse generator is researched in [122]

that can generate short unipolar pulses.

1.2.4.2. dc-dc Converters

Among all power electronics topologies, dc-dc converters are those that have

found their importance in the early stages. Changing voltage level is a substantial

demand in many applications. Transformers have been traditionally responsible

for voltage conversion. However many axial devices and circuits are required

when using the transformers whereas dc-dc converters are offering simply

practical and effective methods. Although the devices have been utilized in wide

range of applications, they are still gaining popularity in other areas. Switched-

mode power supplies (SMPS) are known as a particularly important class of non-

linear loads which have harmful effects on supply side power quality. As the

equipment is going to be connected and be fed by the grid, the power quality of

distribution system including current harmonics injection and reactive power

flow should be considered. Different kinds of Power Factor Correctors (PFC),

including passive and active ones, could be employed to mitigate these effects.

Dc-dc converters including buck, boost, and buck-boost converters shown in Fig.

1.10 are some types of active PFCs that can be either single-stage or multi-stage.

27

Vdc

SW1

D

L

C

+

-

RSW2

(d)

Fig.1. 10. dc-dc converters, (a). Buck (b). Boost (c). Buck-Boost (d). Positive Buck-Boost

Reviewing the literature demonstrated that typical dc-dc converter topologies are

not considered as pulsed power supplies so far. A number of publications

reported utilisation of these converters in high voltage applications that comes as

follows.

In the case of a switched-mode power supply, a boost converter is inserted

between the bridge rectifier and the main input capacitors. The boost converter

attempts to maintain a constant dc bus voltage on its output while drawing a

current that is always in phase with and at the same frequency as the line voltage.

Another switch mode converter inside the power supply produces the desired

output voltage from the dc bus. This approach requires additional semiconductor

switches and control electronics, but permits cheaper and smaller passive

components. It is frequently used in practice. For example, SMPS with passive

PFC can achieve a power factor of about 0.7–0.75, SMPS with active PFC, up to

0.99 power factor, whereas an SMPS without any power factor correction has a

power factor of only about 0.55–0.65.

A single-stage current-fed full-bridge boost converter with power factor

correction (PFC) and zero current switching (ZCS) has been employed in [123]

for high voltage applications such as medical X-ray imaging, RF generation,

travelling wave tube, and lasers. The single-stage current-fed full-bridge boost

PFC converter can achieve ZCS by utilizing the leakage inductance and parasitic

capacitance as the resonant tank. The variable frequency control scheme with

ZCS is used to regulate the output voltage and to achieve high power factor.

28

Unlike existing single-stage ac-dc converters with uncontrolled intermediate bus

voltage, a new single-stage ac-dc converter achieving power factor correction

(PFC), intermediate bus voltage output regulation, and output voltage regulation

is proposed in [124]. The converter is formed by integrating a boost PFC

converter with a two-switch clamped fly-back converter into a single power

stage circuit. The current stress of the main power switch is reduced due to a

separated conduction period of the two source currents flowing through the

power switch. A dual-loop current mode controller is proposed to achieve PFC,

and ensure independent bus voltage and output voltage regulations.

An efficient power-factor correction (PFC) scheme is proposed in [125] for

plasma display panels (PDPs) to reduce harmonic currents and power

consumption. The proposed high-efficiency interleaved boost converter can

reduce the conduction losses and diode reverse-recovery problems in the

continuous-conduction-mode (CCM) operation. A zero-current switching (ZCS)

condition is obtained in this design to solve the reverse-recovery problems of the

output diodes. In addition, a control strategy is suggested for the use of the

proposed converter in a practical design. A high power factor can be achieved

without sensing the input voltage. Another boost PFC converter is utilized in

[126] to improve the power factor of an MG as a pulsed power supply. The

author of reference [127] has combined an MG with a boost converter to increase

the voltage of photovoltaic cells. In contrast to the present similar circuits, during

the operational process the capacitors are not shortened by the switch and

therefore the large discharge current stresses do not influence the circuit. The

circuit provides the multiplication of the input voltage having a smooth (small

ripples) input current and output voltage. The advantages of the proposed circuit

make it appropriate for use with alternative sources of energy.

Reference [128] has concentrated on a new topology based on a positive buck-

boost converter with multi output. A single output positive buck-boost converter

consists of buck and boost converters in cascade. The proposed topology can be

controlled to achieve advantage of output voltage robustness against input

voltage fluctuation and load changes.

29

1.2.4.3. Voltage Multipliers

The Capacitor-diode based voltage multipliers (CDVM) have been used widely

in space and communication applications. Among them, Cockcroft-Walton

multiplier topology has a remarkable role in voltage promotion in

microelectronics related configurations such as, radio frequency passive

transponders [129], passive wireless microsensors [130] and battery-operated

devices [131]. Three different configurations of these voltage multipliers,

including simple N-stage schematic of both a Cockcroft-Walton voltage

multiplier and a Dickson charge pump are depicted in Fig. 1.11.

Fig.1. 11. Capacitor-Diode Voltage Multipliers (CDVM) (a). N-stage Cockcroft-Walton Voltage

Multiplier (b). N-stage Dickson charge pump (c). Another N-stage CDVM

The advantages of CDVM in those applications are that they are of small size

and weight and have high efficiency and reliability. The main disadvantages of

CDVM in these cases include the delay between input and output and the non-

negligible amount of capacitance needed, but this can be reduced within

acceptable limits by increasing multipliers’ operating frequency via an ac-dc-ac

converter placed in the input of multiplier [132]. In relation to radio frequencies

in particular, Cockcroft-Walton multiplier is widely used to increase alternative

voltage magnitudes to higher dc levels in regard to its stages. The simplicity of

the circuit is the most remarkable benefit of it. Each stage consisting of a couple

of diodes and capacitors escalates voltage one more time. Such stages function as

a complementary extension of a single topology, adding voltage steps to the

output value. Therefore, there is no necessity to use gate turning on switches or

30

transistors and their relative circuits like control boards and driving stacks. It is

obvious that these control blocks make the configuration heavier, more complex,

expensive and less reliable. On the other hand, these circuits have the flexibility

of being fed by any frequent input waveforms except those with a pulse shape.

This means that there is no obligation to give them just sinusoidal waveforms. In

respect to the nature of these circuits which is based on the peak detection, they

are able to increase the voltage magnitude of any alternative waveforms,

including sinusoidal, trapezoidal or even sinusoidal voltage waveforms with

harmonics. However, the voltage stress (dv/dt) across the input should be

controlled in order to control the leakage current through the capacitors. These

specifications support the idea of utilizing these multipliers for pulsed power

applications.

Diode-capacitor multipliers have also been known and widely used as simple

transformerless voltage multipliers [133, 134]. In such multipliers, usually fed

from ac suppliers, the recharge processes of the capacitors occur at the industrial

frequency. As a result, the values of the capacitances must be sufficiently large

[135-139]. CDVM is also employed for low power applications such as passive

UHF RFID (Radio Frequency Identification technology) transponders [140-145]

and passive wireless microsensors [146], but has been seldom considered for

high voltage applications. The basic concept of these topologies is documented

in [147, 148]. CDVMs have been used in several applications as a simple and

reliable means to obtain a high dc voltage from an ac source. Among the

advantages offered by these multipliers, is the fact that the ac source voltage may

be substantially lower than the wanted dc voltage and that capacitor and diode

ratings are lower too, as the sharing of voltage stresses is intrinsically ensured by

the multiplier operation [149-153]. Considering the simplicity of these

multipliers structure and operational features, they have potential to be used in

high voltage applications including pulsed power generation.

The intensive use of semiconductor devices enabled the development of a

repetitive high-voltage pulse-generator topology from the dc voltage-multiplier

(VM) concept. Recently a pulsed power generator is proposed based on CDVM

topology [154]. The proposed circuit is based on an odd VM-type circuit, where

31

a number of dc capacitors share a common connection with different voltage

ratings in each one, and the output voltage comes from a single capacitor.

An expression for the equivalent source resistance of the CDVM circuit is

derived in [153]. The theoretical performance of the CDVM with impressed

current input and constant voltage output has been studied in [156]. A scarce

application of CDVM in high voltage has been reported in [157] with two

separate cascaded Greinacher circuits connected in phase opposition to improve

the dc output quality of called low-ripple compact high-voltage dc power supply.

Reference [158] has also improved and stabilized the generated voltage of

voltage multipliers by placing a voltage regulator in the output of the system.

However, the combination of either MG or CDVM with dc-dc converters could

give more flexible and efficient power devices, which is highlighted in many

reviewed papers [159, 160]. For example, authors in [161] have introduced

application of the voltage multiplier technique (CDVM) for classical non-

isolated dc-dc converters developed in [162]. The major benefits obtained with

the integration of voltage multipliers with classical converters are the operation

with high static gain, reduction of the maximum switch voltage, zero current

switch turn-on, and minimization of the effects of the reverse recovery current of

all diodes with the inclusion of a small inductance. The voltage multiplier also

operates as a regenerative clamping circuit, reducing problems with lay-out and

the EMI generation. These properties allow operation with high static again, high

efficiency, and obtaining of a compact circuit for applications where the isolation

is not required.

Reference [163] has improved a new non-isolated dc-dc converter with high

voltage gain using a three-state switching cell and voltage multiplier stages based

on capacitors. In this topology, the value of the gain can be modified depending

on the requirements of the application by means of the number of multiplier

stages and the duty cycle. The proposed converter can be employed in renewable

energy systems where commonly low input voltages (12Vdc to 48Vdc) are

involved or in uninterrupted power supply (UPS) systems in order to avoid the

necessity of a step-up transformer. An asymmetrically switched class D inverter

with a series-parallel resonant tank and a Walton-Cockroft voltage multiplier for

medium power, high voltage applications is developed in reference [164]

32

operating in a self sustained oscillation mode above the resonant frequency. This

converter is controlled by varying the duty cycle of one switch.

Amongst these options, two switches based dc-dc multilevel voltage multiplier

(MVM) introduced in [165] is a suitable candidate in this application, which has

many benefits. An Nx MVM can be built with two switches, 2N-2 diodes and

2N-2 capacitors, free of magnetic components, to have an output voltage N times

higher than input voltage. It is based on the multilevel converters principle and

designed for unidirectional power transfer applications. Each device blocks one

voltage level, thus high voltage converters can be synthesized with low voltage

devices. The main advantage of this topology is the reduced number of

transistors and gate drives. Possible applications have a wide range from low

power silicon-based voltage multipliers, implemented inside microchips; to

medium power level dc links for multilevel inverters based distributed

generation (DG) systems, where a capacitor's voltage balancing is a challenge for

more than three levels. Other voltage multipliers topologies discussed in [166-

167] are not considered in this section due to either complexity in topology and

control strategy, exerting magnetic elements or design, and component attributes

that made them appropriate for low power applications.

1.2.4.4. Pulse Generators Based on Inverters

In order to have more control and flexibility on output pulses, the modulation of

a repetitive pulse shape waveform voltage over a dc level can be considered, that

can be generated by either conventional two-level or modern multilevel

inverters. The concept was developed among experts however the application of

dc-ac converters was barely considered in this regards. A few researches reported

utilisation of inverters as lateral modules in the pulsed power supply structure

providing bipolar pulse production skill for the plasma systems [168-179].

Another application of inverters in pulsed power system is reported in a research

aiming to drive a silent-discharge-type ozone-generation tube [180]. A high-

frequency linked power conversion circuit for the developed ozoniser is

proposed in this research that mainly consists of three-phase active PFC rectifier,

voltage-source-fed full-bridge load resonant inverter using the IGBT power

modules, and ozone-generation tube load with series-compensating resonant

inductor. The inverter output is connected to the load with the series-

33

compensating resonant inductor via a high-voltage high-frequency transformer

[181–183]. The main characteristic features of the proposed load resonant

inverter-type ozonizer scheme are as follows: operation under working frequency

higher than in previous models, applying zero-voltage soft-switching technique

and trench gate insulated gate bipolar transistor (IGBT) modules with low

saturation voltage, and introducing pulse-density modulation (PDM) to regulate

the ozone production quantity under a stable silent-discharge condition.

Recently, basic inverters were replaced by other topologies such as multilevel

and multi output inverters due to the advantages they brought in different aspects

in comparison with conventional ones. A new contribution to increase the quality

and efficiency of the high or medium voltage system is to use the multilevel

converter instead of the traditional two-level inverter. This permits the

semiconductor devices to operate at lower switching frequencies with higher

efficiency, as well as achieving a high voltage rating for the converter with less

voltage sharing problems and lower voltage stress across switches, which

minimizes EMI. Power converters normally operate at low voltage (600 Volts)

and low frequency (5-500 Hz) in most industrial and power electronic

applications such as motor drive systems. The silent discharge plasma system is

a new application for the multilevel converters where the output frequency and

the output voltage of the multilevel converters have to be 5-100 kHz and 5-15

kV, respectively. There are three candidate topologies [184-187] and their

relative merit for SDPSs has not yet been evaluated. It is important that this

evaluation is conducted.

To feed appropriate power into SDPS applications, state-of-the-art systems

currently use 2-level converters which employ Insulated Gate Bipolar Transistor

(IGBT) power switches. Such switches, however, cannot accommodate the high

voltages needed to make SDPSs work effectively. For this reason, IGBT

switches must be used in conjunction with step-up transformers. The use of these

2-level converters causes major problems [188-192]:

• The voltage stress across the load (dv/dt) is significant and creates much

high frequency electromagnetic noise

• There are heavy losses in the transformer and associated resonant filter

(e.g. core losses)

34

• There are heavy losses due to the quantization noise in the 2-level pulsed

waveform; this problem becomes more serious as the frequency rises

(remember that high frequencies are necessary for the effective operation

of SDPSs)

• Substantial dielectric losses can occur in the SDPSs because of the high

dv/dt values arising when 2-level converters are used

A multi-level inverter could be an appropriate candidate for the above mentioned

problems based on its intricate voltage scattering and flexibility in control

strategies. It is also necessary to analyze this converter to find current loops

during transient times for reducing the stray inductances and capacitances.

1.2.4.5. Resonant Converters

The popularity of resonant converters is based on the zero crossing switching

capability of their topology which is a result from oscillation of inductive and

capacitive components of the circuit. As expected the switching losses are

substantially reduced in these topologies. Particularly in high power applications,

the switching frequency is restricted due to switching losses and the

unavailability of suitable high power transformers [193]. Significant progress has

been made in recent years in soft switching by using resonant techniques. The

applications have been restricted mostly to low power applications, however it

can be extended to high power applications with adequate modifications. There

are several researches on applications of resonant converters for high voltage.

Some of those are discussed below.

A novel hybrid full-bridge (H-FB) three-level (TL) LLC resonant converter is

proposed in [194] for the fuel cell power system. It integrates the advantages of

the H-FB TL converter and the LLC resonant converter. A detailed analysis of

operation and a basic design procedure for a new high-frequency (HF) resonant-

converter technology with phase-shifted regulation is presented in [195]. The

new HF resonant technology has a good potential to be a cost-effective solution

for the voltage regulation modules (VRMs) for the next generations of

microprocessor systems.

The application of resonant converters for SDPS loads are reported in many

different studies for an extensive range of applications. There are a couple of

resonant converter topologies introduced in [196, 197] for plasma torches.

35

Converters for plasma torches generally operate at power levels above 30 kW,

and hard switching topologies are normally used. The zero voltage switching

partial series resonant converter which is considered in [196] has properties that

make the application at high power levels attractive. In this paper the converter is

applied to a plasma torch application where power levels can go up into the

megawatt range. To increase the power rating per unit, a distributed transformer

and parallel operation of a number of converters are proposed in this paper as

well as an improved dynamic controller to improve the operation of system. In

the second paper, the design, implementation, and performance of a half bridge

resonant converter (HBRC) used as an electronic ignition system for arc plasma

torch generation is presented. The significance of the design for this converter

lies in its simplicity, versatility, and low cost. The system operates as a high

voltage supply attached to electrodes before gaseous breakdown and as an open

circuit when an electric arc is established [197].

A simple high frequency resonant power converter is utilized in an experiment

reported in [198] to produce silent discharge ozoniser for colour removal of

treated palm oil mill effluent. Palm oil agricultural and industry activities

generate a great amount of by product, known as palm oil mills effluent

(POME). The treatment conducted using membrane bioreactor has successfully

removed the heavy organic component of POME but the water that remains still

contain colour as it’s by product. The use of a simple silent discharge ozoniser in

colour removal of treated POME is proposed in this paper. The power supply

converted a direct current low voltage input into high frequency and high

sinusoidal voltage output. This high voltage created micro electrical discharges

inside chamber to generate ozone from oxygen molecules.

The analysis and design of a full-bridge (FB) LC parallel resonant plasma driver

at the radio-frequency (RF) operation with variable-inductor based phase control

scheme is presented in [199]. Since the switching frequency of the RF plasma

module is mainly fixed for EMC regulation, the variable-inductor control scheme

can adjust the transconductance amplitude to enable load-current regulation.

Additionally, in order to minimize conduction loss on the switches, the design

criterion of the required dead-time for ZVS condition with the minimal

circulating current of the LC parallel resonant tank is considered in this design.

36

In another contribution, lcscp resonant inverters are employed to drive high-

power HPS lamps. The design criteria for full-bridge series–parallel (lcscp)

resonant inverters suitable for driving high-power high-intensity discharge lamps

are presented in [200]. By using the properties derived from the transfer

functions of the inverter, a soft start up method is proposed in this work. In

steady-state operation, the proposed control minimizes the reactive voltamperes

in the resonant tank. Moreover, the variation of the power delivered to the lamp,

caused by the lamp aging, is limited in order to fulfil the standard. This design

provides cost-effective circuits, simplifying the dc–ac power stage of an

electronic ballast.

A universal resonant converter for equilibrium and nonequilibrium plasma

discharges is introduced in [201]. This new system is proposed to ignite and to

sustain a plasma discharge for different reactor configurations, using a single-

series parallel high-frequency resonant converter. Different operation modes of

proposed converter are analysed in this report, and their performance is verified

in two applications: an equilibrium plasma discharge (electric arc) and a

nonequilibrium plasma discharge (electric barrier at atmospheric pressure)

Research has also been carried out on application of resonant converters in

plasma display panel (PDP) TV. Some of investigations that have been

remarkable in terms of the effectiveness and improvements are those listed and

described below.

Reducing flat transformer temperature in LLC resonant converter for plasma

display is considered in [202]. The trend in PDP TV is towards the thinner

thickness, lighter weight, and fan-less system. To accomplish this goal, it is

desired that advanced power conversion techniques to be used to implement a

low-profile power supply for PDP. As PDP power module, LLC resonant

converter is widely used because it can achieve both high efficiency and

reliability. To achieve a low-profile circuit configuration, it is necessary to

operate the converter at high switching frequency. LLC resonant converter can be

operated at high switching frequency because of soft switching operations.

However, it is a problem that the temperature of flat transformer becomes high at

high switching frequency. In this paper, the method of reduction flat transformer

temperature in LLC resonant converter for PDP is proposed by analysis of core

37

loss and the experiments. A new multi-output LLC resonant converter is

proposed for high efficiency and low cost PDP power module in [203]. In the

proposed converter, zero-voltage (ZV) turn-on of the primary MOSFETs and

zero-voltage (ZC) turn-on and turn-off of the secondary diodes are guaranteed in

the overall input voltage and output load ranges. In addition the primary

MOSFETs and the secondary diodes have the low voltage stresses clamped to

input and the output voltages, respectively. Therefore, the proposed converter

shows the high efficiency due to the minimized switching and conduction losses.

Moreover, by employing the transformer with multiple secondary windings, the

proposed converter can have multiple outputs, which show the great cross-

regulation characteristics. As a result, the proposed converter can be

implemented with low cost and compact size. Experimental analysis of a series

resonant converter for a plasma inertization plant is presented in [204]. This

article showed a dc-ac converter series resonant (CSR) operating high frequency,

using IGBTs and soft-switching technique (zero voltage switching - ZVS). This

full-bridge converter composed of four identical parallel modules operating at

100 kHz to get 400 kHz as a final frequency in the output. In another interesting

research, a new PWM-controlled quasi-resonant converter is presented in [205]

for a high efficiency PDP sustaining power module. The load regulation of the

proposed converter can be achieved by controlling the ripple of the resonant

voltage across the primary resonant capacitor with a bidirectional auxiliary

circuit, while the main switches are operating at a fixed duty ratio and fixed

switching frequency. Hence, the waveforms of the currents can be expected to be

optimized from the view-point of conduction loss. Furthermore, the proposed

converter has good zero-voltage switching (ZVS) capability, simple control

circuits, no high-voltage ringing problem of rectifier diodes, no dc offset of the

magnetizing current and low-voltage stresses of power switches. Thus, the

proposed converter shows higher efficiency than that of a half-bridge LLC

resonant converter under light load condition. Although it shows the lower

efficiency at heavy load, because of the increased power loss in auxiliary circuit,

it still shows the high efficiency around 94%. As already discussed, initiation of

a plasma conduction state requires a relatively large voltage to ionize the gas. A

new version of the series resonant converter is proposed in [206] that uses the

38

magnetizing inductance of the transformer for resonance. This converter is not

suitable for most power supply applications, but the unique load characteristics

associated with plasma loads make this type of converter well suited for arc

striking, while allowing safe operation during the plasma state. A feature of the

resonant converter is that the controller need not be complex, thus making it

suitable for application in competitive industrial systems. Possible transformer

configurations are investigated in this work that includes an air core and a

number of ferrite-cored transformers. The series resonant converter with the

best-suited transformer is verified experimentally in a tungsten inert gas welding

application. A high-voltage power supply based on piezoelectric transformer

(PT) is used in [207] for ozone generation. Even though, nowadays, PTs are only

available with low power rating, there exist several low-power applications of

ozone generation in which the use of this novel technology could be

advantageous. Hence, the aim of this investigation was to evaluate the

possibilities of using PTs in the implementation of high-voltage power supplies

for ozone generation. First, the possible topologies that can be used to drive the

PT were identified. Then, the half-bridge inverter operating under zero-voltage

switching (ZVS) was investigated, and the effect of the silent discharge

generator (SDG) on the converter operation was analysed. A new control circuit

that allows the ZVS operation is proposed in this study. The control circuit

operates in closed loop by measuring the phase between the PT’s resonant

current and the switching pattern and adjusting the switching frequency to the

optimum value to assure ZVS.

According to this survey, several topologies and techniques in the area of power

electronics have been determined that have merit in pulsed power generation and

have never been considered to be utilized for this purposes before that. The other

point revealed through the literature review is that novel methods should be

considered to improve the efficiency of plasma systems. The flexibility and

reliability of pulse supplier are among those characteristics that required to be

promoted. Therefore considering efficient power electronics topologies is a

remedial solution for an enhanced supply of plasma systems.

39

1.3. Account of Research Progress Linking the

Research Papers

1.3.1. Introduction

This project began with a comprehensive literature review of pulsed power

applications, their requirements, and available power supply technologies. Fast

growing industrial, environmental, medical, and military applications have

increased demand for more efficient and flexible pulsed power supplies.

Therefore, the first step in the research process was a review of existing

technologies and techniques used for these purposes. Many published reports on a

number of pulse modulators supplying various applications were reviewed during

this study in order to diagnose and highlight the drawbacks and deficiencies of

these topologies. These investigations demonstrate a dominant trend of using the

following equipment in the pulse supply process:

Marx Generator (MG)

Magnetic Pulse Compressor (MPC)

Pulse Forming Network (PFN)

Multistage Blumlein Lines (MBL)

Each of these pulse suppliers is appropriate for a specific sort of application with

respect to the load specifications. Several aspects of above mentioned modulators

were thoroughly explored in order to distinguish their structure, features and

capability restrictions. This exploration clarified that both mechanical and

electrical devices are facilitated in their structure to produce pulse trains.

Normally, while dealing with such equipment, mechanical losses will often exceed

electrical losses and lead to a lower efficiency and a shorter effective lifespan of

included devices. Electromagnetic or inductive components (such as transformers)

are also used in a number of structures that impose extra magnetic/leakage related

losses to the pulse generation process. Based on these studies, it was also

determined that magnetic/gas based devices are the preferable switching

mechanism utilized in most modulators. Spark gap, hydrogen thyratron and

vacuum tube are already used in MGs, MPCs and PFNs. Although these switches

convey the advantage of prompt switching (in nanoseconde regimes) to the pulse

40

supply, they also deal with a number of significant drawbacks that influence the

cost and the supply process. Some of these shortcomings are as follows:

Being inefficient

Being expensive

Being bulky

Being heavy

Being incapable of operating in a high repetition rate

Having short life span

The insulation of the primary energy source (i.e. the grid) from the load side

during the supply period is a critical parameter in pulsed power applications. In

some cases, undesired resistive collapse (arc phenomena) may occur in the load

side during the supply process; this consumes a considerable amount of energy. To

prevent extra energy being wasted at the arc time, the proper insulation of input

from the load during the supply period would certainly be effective. Furthermore,

having control over power flow to the load and being able to halt the supply

process in a desired stage is another crucial skill in an efficient supply trend.

The power supplies are usually designed based on an application’s inherent

specifications. Some are allocated to inductive loads, some to capacitive ones, and

others are designed for resistive loads. In some cases, applications with a higher

sensitivity degree demand extraordinary pulses with specific properties. Pulses

with absolutely flat tops, variable shapes, and bipolar (alternately positive and

negative) pattern are among them. Sometimes, either load specifications or its

requirements vary in different time stints of a supply process. A flexible pulser

capable of feeding loads with variable requirements is generalized equipment that

can be taken into service for a diverse range of applications.

Based on the equipment gaps and the load demands found as a result of this

survey, specific research aims were then targeted. The most significant of these is

the necessity to introduce cost-effective topologies and techniques to increase

efficiency, flexibility and reliability of the pulsed power supply process. Replacing

existing switching devices with state-of-the-art solid-state switches is a remarkable

step forward to remove the associated concerns. Solid-state switching technology

presents a range of advantages-including higher efficiency, operational frequency,

and long lifespan-in addition to being compact, light and inexpensive. Such

41

qualities make them ideal substitutes for previously used switching technologies.

There were a couple of drawbacks that needed to be resolved in order to remove

all concerns regarding extensive application of solid-state technology in the pulsed

power area. Voltage rating and switching velocity were the major concerns, and

these have been addressed through the latest improvements in the area. For

example, solid-state technology has recently been developed by the introduction of

high power switching devices. A new family of fast solid-state transistors such as

IGBTs and MOSFETs with high voltage and current ratings has been released

commercially during the last decade; these can revolutionise pulsed power supply

technology. Currently, IGBTs with the voltage rating of a few kilo volts and rising

and falling times of a few micro seconds are available in the market. Therefore,

solid-state transistors can be considered as suitable replacements for previously

used switching devices.

Subsequently, proper topologies need to be studied for utilisation of solid-state

technology. In spite of the (discussed) improvements in their structures, the

voltage ratings of modern switches are still far below the range of many pulsed

power applications. Therefore, producing higher levels of voltage through these

devices is still associated with voltage splitting techniques, such as the concept

used in the structure of MGs. Application of solid-state in pulsed power has

already commenced with the implementation of solid-state MGs and their supply

of a number of applications [107-122]. The reported results on their performance

in several cases indicate an acceptable level of satisfaction.

Power electronics offer a wide range of converters and techniques that have the

potential to be utilized in the pulsed power area and to bring many advantages to

the supply process. However, considering the rising application demand, either

new designs or modifications to conventional converter structures are needed in

order to adjust the operational process and the output to this demand. Dc-dc

converters, such as boost and positive buck-boost, CDVMs (including Cockcroft-

Walton VM and Dickson charge pump), and resonant converters are among those

that have apparent potential in high voltage pulse generation. However ac-dc

rectifiers and dc-ac inverters are also utilized in combination with the mentioned

topologies to form a universal pulsed power supply. Commutation techniques and

resonant phenomena are also beneficial and can be exploited in this regard.

42

1.3.2. A new solid-state current-voltage source based pulsed power

supply

1.3.2.1. Investigating the possibility of producing pulses for plasma

applications through a current source

Most technologies used so far to feed plasma applications can be classified as

voltage source topologies since electrostatic energy sources have been utilized in

their structure to store the initial energy needed for producing pulses. As shown

in Fig. 1.12, these configurations can be modeled with a charged capacitor which

is connected in parallel to the electrodes, energizing the load material to form the

plasma.

Fig.1. 12. Voltage source pulsed power supply

A high amount of instantaneous energy delivered to the load in a short time stint

excites and deforms the load. A fast switching device with high voltage rating is

required to rapidly release this accumulated energy. Magnetic/gas based switches

such as spark gap and hydrogen thyratron have been utilized in most of these

topologies to deliver energy to the load. Although these high voltage devices

have a short transient period while switching on or off, they are not able to

operate in a high switching frequency. This means keeping the device in

conduction mode for a short period of time is not possible.

A plasma generation process commenced by imposing high voltage with a fast

rising time (dv/dt) across the load usually follows by an impedance change in the

load side. The load resistivity is substantially dropped by this excitation.

However sometimes this reduction in the load resistivity goes beyond a normal

trend and a sort of short circuit happens in the load side. This follows with an

undesired arc phenomenon in the load that discharges extra energy. To prevent

extra losses, the supplying process should be ceased at a proper instant.

Therefore, a control over power flow is a critical skill for an efficient energizing

process. This considered in present pulse modulators by storing the energy

according to the load demand. Thus, available pulsers are designed for individual

43

applications based on load specifications and consequently a general and multi-

purpose equipment capable of providing flexible pulses for a diverse range of

applications is barely available. New topologies, giving this control capability to

the pulse modulators, can be considered as a solution for this purpose. Using

solid-state technology as high frequency switching devices in currently used

topologies is another remedy for this concern. Supplying plasma applications

through a combination of current and voltage sources is an appropriate candidate

which can brings many advantages to the pulse generators including an efficient,

flexible and reliable supplying process. Basically, voltage and current sources are

more preferred to supply inductive and capacitive loads respectively due to

compatibility issues. As shown in many previous investigations, plasma

applications mostly act as variable resistive-capacitive loads and this additionally

supports the idea of supplying them through a current source. Positive buck-

boost topology, shown in Fig. 1.13, is considered in the earliest stage, as an

appropriate option for producing the pulses.

Fig.1. 13. Positive buck-boost topology

In spite of all merits this topology has in voltage boosting, there are a few

ambiguous points regarding its application as a pulse generator that should be

clarified in advance. The most important concern is to recognize how the

switching transients of solid-state devices affect the power delivery trend. To

generate voltage at the output, the current stored in the inductor has to be

conducted to the capacitor, and to produce a considerable dv/dt, boost switch

SW2 should be switched off promptly to facilitate this function. The switching

transients including active and saturated modes which realistically take a few

micro seconds to be accomplished may influence the current conduction process.

The time taken by the switching transients may negatively interfere to the

voltage stress creation and restrict the produced dv/dt by increasing dt. On the

other hand, having a fast switching transient at high voltage is very challenging.

To inspect and to address the restrictions imposed by the switching transients, a

simulation based study is arranged at the beginning stage. A simplified model of

44

pulse modulator including plasma load is simulated which is given in Fig. 1.14.

A variable resistive-capacitive load model simulates the variations during plasma

formation for the pulse modulator. The resistivity and conductivity of the load

during plasma reaction is simulated in this model by a small capacitor connected

in parallel to a huge resistor and both connected to a small resistor through a

switch S3. Reaction commencement is simulated by connecting the small resistor

to the pulser model at the instant in which the load switch is triggered on.

Fig.1. 14. (a). A circuit diagram of current source topology (b)&(c). Operation modes of the current source topology supplying a plasma load

Although available simulation packages such as Simpower platform in

MATLAB and PSICE are not normally designed to study the switching transient

interactions, IGBT characteristics of this model in MATLAB/Simulink are

determined based on real values to achieve closer responses to the reality. In this

study, the switching transient and the associated effects on velocity of

disconnecting a current path and conduction of current into another circuit are

investigated by assuming a variable current source that injects different amount

of currents into the capacitor. As given in Fig. 1.15, a wide range of voltage

stresses across the load model achieved as a result of flowing different currents

through it, verifies that the switching transient does not have a serious impact on

the voltage rising, whereas the level of flowing current and the capacitive

characteristic of load are determining factors with respects to Eq. (1-1).

dt

dvCtiC =)( &

C

ti

dt

dv C )(= (1-1)

45

0.002995 0.0029975 0.003 0.0030025 0.003005 0.0030075 0.00301-1000

0

1000

2000

Time(s)

Cap

acito

r vo

ltage

(V)

dV/dt

0.002995 0.0029975 0.003 0.0030025 0.003005 0.0030075 0.00301-100

-80

-60

-40

-20

0

20

40

60

80

100

120

Time(s)

Cap

acito

r cu

rren

t(A

)

capacitor voltage(dV), iL=20Acapacitor current(A), iL=20Acapacitor voltage(dV), iL=40Acapacitor current(A), iL=40Acapacitor voltage(dV), iL=60Acapacitor current(A), iL=60Acapacitor voltage(dV), iL=80Acapacitor current(A), iL=80Acapacitor voltage(dV), iL=100Acapacitor current(A), iL=100A

Fig.1. 15. Voltages and currents of modeled capacitor with 20, 40, 60, 80 and 100A currents flowing into the capacitive load through a variable current source

As already utilized in positive buck-boost converters, connecting a small

capacitor at the output can be beneficial in many applications. It acts as extra

energy storage and positively interferes in energy supplying process, so the

inductor current will be kept far from massive tolerances. This capacitor ensures

having necessary dv/dt across the load while the inductor supports the plasma

with delivering the required energy. The capacitor size is determined larger than

load’s capacitive characteristic to avoid the problems raised due to loading

issues. The contribution of each energy storage in supplying a typical load

discussed through Eq. (1-2) is shown in Fig. 1.16. In this collaboration, the

capacitor mostly contributes to achieving desired dv/dt, whereas the inductor is

in charge of supplying the load.

extraCLLoad EEE += )(2

1)(

2

1 2min

2max

2min

2max VVCIIL extra −+−= (1-2)

46

4.9992 4.9993 4.9994 4.9995 4.9996 4.9997 4.9998 4.9999 5 5.0001

x 10-3

4.74

4.75

4.76

4.77

4.78

4.79

4.8

(a)

Indu

ctor

ene

rgy(

j)

4.9992 4.9993 4.9994 4.9995 4.9996 4.9997 4.9998 4.9999 5 5.0001

x 10-3

0

0.005

0.01

0.015

0.02

0.025

(b)

Cap

acito

r en

ergy

(j)

4.9992 4.9993 4.9994 4.9995 4.9996 4.9997 4.9998 4.9999 5 5.0001

x 10-3

0

0.01

0.02

0.03

0.04

(c)Time

Load

ene

rgy(

j)

Fig.1. 16. (a). Inductor energy (b). Capacitor energy (c). Load energy

The above mentioned simulations and analysis have been presented in the form

of a conference paper at 17IEEE-PPC 2009 entitled “Using a current source to

improve efficiency of a plasma system” held at Washington DC, USA [208]. The

6th chapter of this thesis is based on this paper.

1.3.2.2. Proposition of a novel high-voltage pulsed power supply based on

low-voltage switch-capacitor units

According to initial studies that ensured the possibility of supplying plasma

applications through solid-state based current source topologies, a configuration

inspired by positive buck-boost converter topology has been designed and

implemented in the next stage. As shown in Fig. 1.17, this pulse generator is fed

through the grid. A one phase ac-dc converter rectifies the network sinusoidal

input voltage to a dc waveform and a large capacitor, Cin, at the output of

rectifier removes the ripples from the rectified waveform and provides the rest of

the circuit with a continuous and smooth dc voltage, Vin. An inductor, L,

connected to the input voltage through a power switch, SS, acts as a current

source in this topology. The inductor is connected in cascade to a series of

switch-resistor-capacitor units which are in a ladder-shape arrangement. The

capacitors in these units form an integrated voltage source at the output.

Considering variable resistive-capacitive characteristics of plasma applications, a

small capacitor in parallel to a variable resistor simulates the load for the pulsed

47

power supply. The variable resistor composed of a large and a small resistor

connected through a switch, simulates the load resistivity variation in case of

applying a high voltage with a considerable dv/dt.

Switching Pulses

SS

S1

SL

D

R2Load

C1

-

+

+

-

L

S2

C2

+

-

Sn

Cn

+

-

R1LoadControl

Protocol

of

Series

Switches

Plasma Load Model

Load modelling control

CLoad

Current control method

RD1

RD2

AC-DC

ConverterVac

220 V50 Hz

V inCin

Current Source Block Diagram Voltage Source

Current Source Voltage Source

Load

Fig.1. 17. Plasma power supply configuration with multi switch-resistor-capacitor units

1.3.2.2.1. Switching modes

A simplified two-unit model shown in Fig. 1.18 is simulated for further studies

and analyses. Each operation cycle in this topology consists of two separate

parts. The first part, which is the load supplying process includes three principle

switching modes including charging the inductor, circulating the inductor current

and charging the capacitors modes as given in Fig. 1.19(a), (b) and (c)

respectively. Fig. 1.19(d) is the model used for simulation of resistive collapse at

the beginning and during the plasma generation process. The second part,

discharging residual energy, takes place after each pulse delivery in order to

initialize the storing components for the next pulse producing cycle.

Fig.1. 18. A simplified two switch-capacitor unit plasma power supply and the load model

48

Fig.1. 19. Switching states of the proposed pulsed power supply circuit (a) Current source, charging mode (b) Current source, discharging mode (c) Voltage source charging mode (d) Load

supplying mode

According to (1-3) and (1-4), the input voltage located across the inductor

charges it in the first mode where SS, S1 and S2 are on.

)(21 SSSinL VVVVV

S++++++++−−−−==== (1-3)

)()( tiLdtdiLVL ∆∆== (1-4) The inductor current circulates through S1, S2 and the freewheeling diode D when

SS is triggered off in the next mode in order to insulate the input from the load.

This mode is arranged with minimum time duration as the voltages across S1, S2

and D slightly discharge the inductor charge (1-5).

)(21 SSDL VVVV ++++++++−−−−==== (1-5)

The inductor current is led to the capacitors by triggering off S1 and S2 in the

third mode. Pumping a remarkable amount of current into small capacitors

increases the voltage across the capacitors to a higher level promptly, that means

a considerable dv/dt is achieved based on (1-6).

tCIVCItV iCCiCC iiii∆⋅=∆⇒=∆∆ )()()( (1-6)

Assuming identical capacitors (in size) at the output, the same current flows

through all capacitors and therefore charges them in a similar trend. The

generated voltage is also shared by all capacitors equally as given in (1-7). To

ensure an appropriate voltage sharing occurs across the capacitors both sides of

the capacitors are connected to the relevant switches through resistors.

iCout VnV ⋅= (1-7) Where n is the number of switch-resistor-capacitor units which can be extended

to satisfy load demands. Having more units reduces the equivalent capacity and

leads to a higher level of generated voltage with a higher voltage stress (dv/dt).

49

1.3.2.2.2. Discharging residual energy

There is an obligation to discharge the residual energy in the capacitors and the

inductor after either a successful or an unsuccessful pulse supplying cycle in

order to initialize the storing components for the next supplying cycle and to

prevent probable reverse flow of the current. Resistors installed in the common

connections of units are responsible for fully discharging the residual energy in

the capacitors and the inductor after each supplying cycle. They also prevent any

possible short circuit in the units. However there is another loop shown by red

the line in Fig. 1.20 that cannot be protected through these resistors.

Fig.1. 20. Possible current loops during short circuit periods.

Several techniques can be considered as solutions for this problem which are

listed here as hardware and software methods. Installing either a thermistor or a

parallel switch-resistor unit in the return path as shown in Fig. 1.21 are two

hardware methods which can be considered as a remedy in this case. However

these methods impose an increase in the initial cost to the converter.

RD1

S1

SL

D

R2Load

C1

-

+

+

-

L

S2

C2

+

-

R1Load

SS

Vin

-

+

CLoad

(a)R

S

Fig.1. 21. Two examples of hardware methods for discharging residual energy in the inductor and

the capacitors, (a) Parallel switch-resistor unit located in the return path (b) A thermistor in the return path

Another solution is discharging the remaining energy through smart switching.

Protection against unprosperous energy delivery to the load and discharging

residual energy are provided in this method through two extra switching states

given in Fig. 1.22. The alternative and compulsory switching of S1 and S2

forming these states leads to an entire discharge after a few cycles. The

50

triggering signals for the switches are provided by a controller sampling the

inductor current and a capacitor voltage. The control algorithm determining the

switching states is based on the logic provided in the flowchart of Fig. 1.23(a). In

order to protect switching devices from over voltage, the discharging procedure

is conducted in two different schemes. As highlighted by different colours at the

lower part of this flowchart, prosperous and unprosperous delivery processes are

initially distinguished by the control system after load supplying sequences. A

load supplying cycle including pulse generation modes and residual energy

discharging process is shown in Fig. 1.23(b). The inductor current and the

capacitor voltages, accompanied by the switching signals are given in this figure.

Fig.1. 22. Circuit’s switching states in association with software method in order to discharge the

remaining energy in the capacitors

(a) (b)

Fig.1. 23. (a). Block diagram of control algorithm (b). Switching signals pattern

51

1.3.2.2.3. Analyses of load supplying mode and components determination

To study the load supplying process including dynamic and steady-state modes,

the simplified circuit of power supply during plasma reaction shown in Fig. 1.24

is analyzed in the frequency domain.

Fig.1. 24. Equivalent RLC circuit of power delivery mode of power supply

The inductor current of this mode in the Laplace S- domain is

1

000

2 )S+R

L-((LC) S

)) (IR

L)- () )S+(CV ((LCI

(s)=ILCL

L (1-8)

And based on the considered range for the inductor, the equivalent capacitor and

the load resistance, it can be assumed that in any condition:

RCL 4> (1-9) So the time response of inductor current IL(t) can be achieved as:

])()[()](1[)( 21

21211212tt

L ekkekktI αα αααα −− −−−⋅−= (1-10) Whereas:

)0(1 LLCIk = (1-11) )0()()0(2 LC IRLCVk −= (1-12)

)(2])(4)()([ 22,1 LCLC

R

L

R

L −±−=α (1-13)

Since ])([)( dttdILtV LC ⋅−=

])()([)(

1)( )1(

2122)1(

112112

21 ttC ekkekktV −−−− −−−

−= αα αααα

αα (1-14)

Considering equations (1-10), and (1-14), and also the inductor and the capacitor

size ratios, a faster discharge of the electrostatic energy in comparison with the

electromagnetic energy can be expected.

The component sizes, the charges and the switching frequency are determined in

this topology based on load demands. The output equivalent capacitor,

neq CCCC ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅==== 21 , considered at least ten times the load capacitive characteristic

in this design in order to prevent any loading concerns. Additionally, a smaller

equivalent capacitance is preferred at the output in order to produce higher

voltage stress and level. Thus,

Loadeq CC 10==== (1-15) If the capacitors, Ci, are supposed to be identical then nCC ieq = :

52

Loadi CnC ⋅⋅⋅⋅==== 10 (1-16) Where n is the number of switch-diode-capacitor units which is determined by

the switches voltage rating and the demanded output voltage.

Assuming the inductor current is constant during the capacitor charging mode,

the voltage stress can be calculated as follows.

)()( dtdVCCI outLoadeqL ⋅+= (1-17)

LoadL EIL =⋅ 2)21( (1-18) Finally, the recovery time for inductive and capacitive components and the

frequency of pulses generated by the power supply can be determined as follows:

inLLr VILT )(_ ⋅= (1-19) LoutLoadeqCr IVCCT ])[(_ ⋅+= (1-20)

)(])([ 2_ LininoutLoadeqLLCr IVVVCCILT ⋅⋅⋅++⋅= (1-21)

In repetitive pulse generation, a time interval is designated to the load to be

prepared for the next supplying cycle. The frequency of load supply with pulsed

power relies significantly on the load features and requirements, Tr_Load, but

cannot be more than the recovery frequency of the power supply.

)](1[ __max_ dLoadrLCrs TTTf ++< (1-22) 1.3.2.2.4. The experimental results

A double switch-resistor-capacitor unit prototype has been implemented to

investigate the validity of theoretical analyses and simulation results. Due to

restrictions of input power supply in terms of providing high current and voltage,

the laboratory set up is designed to operate in a low voltage scale. As shown in

Fig. 1.25, the set up has been developed to verify the basic concept of this

converter specifically in terms of true voltage sharing through the output units.

The results achieved throughout the experimental examinations shown in Fig.

1.26 indicate that the output voltage is shared and withstood identically by the

switches.

Fig.1. 25. Laboratory prototype of pulsed power supply with double switch-capacitor units

53

(a) (b)

Fig.1. 26. Inductor current, capacitors and output voltages

Based on simulation studies it has been confirmed that no restriction applies

regarding producing higher voltages with considerable dv/dts by increasing the

number of the switch-capacitor units. The outputs can be varied by controlling

the delivered current to the load as well. Utilizing reasonably small capacitors

instead of large ones and having minimum possible diode rectifier in the

architecture of this modulator leads to a huge saving in initial costs. Furthermore,

as another advantage of this design, charging and discharging paths are planned

to have no diodes. That leads to a reduction in the conduction losses and a more

efficient supplying process. Although the smart switching can discharge the

residual energy after halting the supply process and allow the next supply cycle

to be resumed, complexity of the control algorithm can increase the chance of

probable malefactions. Consequently, the control over power flow is still

challenging in this design.

The basic concept of this topology followed by the preliminary calculations and

simulation based studies are presented as a conference paper at IET EPPC2009

Geneva, Switzerland, entitled “A novel high voltage pulsed power supply based

on low voltage switch-capacitor units” [209]. Further concerns including

discharging and protection issues, control algorithm and the experimental tests

are addressed in a more detailed version that is published in the IEEE

Transactions on Plasma Science under the same title [210]. This paper is

presented in chapter 2 of this thesis.

54

1.3.2.3. A new multi-purpose pulsed power supply based on positive buck-

boost converter concept

Lack of control over power flow in most present technologies is a critical

drawback prohibiting the presentation of a general pulsed power supply capable

of producing flexible pulses based on different load demands. As a result of this

limitation, all pulse modulators have to be designed for specific applications

based on their requirements. Otherwise a considerable portion of energy will be

wasted by utilizing the modulator for another application. Therefore, a flexible

pulse supply is a significant contribution to improve the efficiency of supplying

process. Thereby, several techniques are reviewed, analysed and applied to the

current-source topology that has already demonstrated worthwhile advantages in

supplying the capacitive loads.

1.3.2.3.1. Topology features

With respect to this issue, the proposed topology has been revised by substituting

the resistors in switch-resistor-capacitor units by power diodes (switch-diode-

capacitor) as in the positive buck-boost converter. The changes in control

algorithm logic and criteria are performed according to the revisions made in the

architecture of the topology. The complexity of the control method is

substantially reduced due to the elimination of the discharging procedure at the

last stage of the pulse supplying cycle. The reverse power flow direction (from

voltage source to current source) is blocked in this way, so no more action is

required regarding the discharge of residual energy. As a result the residual

energy in the capacitors can be maintained for the next cycle and the converter

can operate in continuous conduction mode (CCM) by selecting a bigger

inductor and keeping the level of stored current in the inductor varying in a

definite band. To guarantee a considerable dv/dt at the output, the boost switches

(S1, S2,…, Sn) are required to be turned off both promptly and simultaneously (in

a precise coordination).

The outcome is introduced as a topology enabling the control over power flow

and pulse supplying process that is flexible enough for use in many applications

by either adjusting the pulse characteristics or ceasing the supplying process in

any stage and consequently saving the residual energy in storing components. A

55

detailed circuit diagram of the topology including the general scheme,

components, load model and controllers is given in Fig. 1.27.

Switching Pulses

SS

S1

SL

D

R2Load

C1

-

+

+

-

L

S2

C2

+

-

Sn

Cn

+

-

R1LoadControl

Protocol

of

Series

Switches

Plasma Load Model


AC-DC

ConverterVac

220 V50 Hz

CLoad


D2

D3

D1

Dn

VinCin



Load

Fig.1. 27. Pulsed power supply configuration with multi switch-diode-capacitor units

1.3.2.3.2. A development in the operation

All switching states and operation analyses of this topology is similar to the

previous converter except there is now the possibility of using it for specific

applications. In some cases, the load demands a high voltage level however the

load resistivity changes by imposing a certain excitation which is not

considerable. A basic high voltage as the fundamental level followed by a

remarkably sharp increase in the voltage level, dv/dt, can lead to the same results

in the load side as whit pulsed power. This converter is able to carry out a

gradual voltage charging process. In addition to the simultaneous charging, the

separate charging of output capacitors is the unique specification of this

proposal, provided by the reverse conductivity preclusion through the diodes.

This has been offered to the topology by an extra switching state shown in Fig.

1.28. The lower capacitors charged in advance during this mode provide a

fundamental voltage level that participates in the load stimulation process.

56

Fig.1. 28. Switching state of charging capacitors separately

1.3.2.3.3. The control flowchart and simulation results

A control algorithm shown in Fig. 1.29(a) has been designed based on the

topology’s properties and the load requirements. Both simultaneous and separate

switching scenarios are considered in this control method so the topology can

operate under both circumstances. The simulation results presented in Fig.

1.29(b) shows the related current and voltage waveforms, as well as relevant gate

drive signal patterns in a supplying cycle.

IL(0)=0

VC1(0)=0,VC2(0)=0

Imax=30A,Vf=500V

Vmax=2/1.5kV

Vmin =100V

IL≥Imax

SS:on, S1:on, S2:on

SS:off, S1:on, S2:on

YES

NO

NO

Load

arranged

NO

SS:off, S1:on, S2:off

YES

VC2≥Vf

YES

SS:off, S1:off, S2:off

Vout≥Vmax

Discharging Process

YES

VC1≤Vmin

NONO

YES

Simultaneous

Switching

Separate

Switching

(a) (b)

Fig.1. 29. (a). Flowchart of the control algorithm (b). Current and voltage waveforms accompanied by correspondent switching signals pattern

1.3.2.3.4. The experimental results

Further analyses have been conducted by developing previously implemented

setup to investigate the validity of this model in producing pulses with

anticipated features. The specifications of this implemented circuit are discussed

57

in Table 1.1. The results obtained from the tests are shown in Fig.1.30. The

current stored in the inductor is conducted to the capacitor bank at the output and

produces a voltage with an acceptable dv/dt. To achieve a higher dv/dt, the

delivered current needs to be more which is not possible through the available

power supply. The voltage sharing across output capacitors is properly

accomplished considering the increasing voltages trend in Fig. 1.30.

TABLE 1. 1. SPECIFICATIONS OF THE LABORATORY PROTOTYPE CIRCUIT

Vin L C1 C2 ILmax 15V 0.4mH 10nF 10nF 7A

VC1

VC2

VOUT

IL

(a) (b)


Concerns associated with simultaneous switching are removed while performing

under separate switching circumstances. A sole switch S1 is triggered off at the

ultimate stage in separate switching technique. This topology can operate in a

higher frequency in comparison with the former design as switching states

associated with the discharge of residual energy are not required in this proposal.

The installation of diodes in the unit circuits prevents the electrostatic energy to

be returned backward and allows the capacitors with remaining charge to be

recharged for the next cycle. The output capacitors in this configuration are

flexible and can be selected as either identical or different with respect to the

load requirements.

Out of these simulations, calculation, tests and analyses, a journal paper is

published at IEEE Transactions on Dielectric and Electrical Insulation entitled

“A New Pulsed Power Supply Topology Based on Positive Buck-Boost

Converters” [110]. This paper was mainly based on the proposed positive buck-

boost based topology for a wide range of applications, and is presented in

Chapter 3. A conference paper entitled “A New Generation of High Voltage

58

Pulsed Power Converters” has been presented at 20th Australian Universities

Power Engineering conference held in December 2010 at Christchurch, New

Zealand, [211] focusing on the energy and component calculations and topology

simulations and analyses of a positive buck-boost topology utilized as a pulse

generator.

1.3.2.4. A design for producing pulses with higher magnitude

An appropriately developed layout is designed according to the positive buck-

boost converter and printed on a copper board (PCB) in order to achieve higher

voltage pulses through the proposed topologies. The pulser includes six stages of

switch-diode-capacitor units. In the design process a number of practical issues

have been taken into consideration to facilitate a proper circuit board and pulser

set up. Some of them are as follows:

High voltage paths (including the connections) and return path are properly

insulated by planning to be located in different sides of the board.

The current paths of the circuit are designed with the target of neutralizing

related magnetic fields to minimize electromagnetic interface (EMI) and

electromagnetic compatibility (EMC) issues.

The current paths and loops are considered with maximum copper width to allow

a large current to be conducted with minimum conduction losses.

The switch-diode-capacitor units are implemented quite symmetrically to let the

contribution of all stages in voltage generation occur similarly and the voltage

sharing to be carried out appropriately.

High voltage diodes with ultrafast reverse recovery are utilized to permit high

quality repetitive operation of the pulser.

A microcontroller with a high operation frequency is used in order to entirely

exploit the capacity of this circuit.

59

Fig.1. 31. Developing hardware for higher voltage pulses

The specifications of used components are evident as in Table 1.2.

TABLE 1. 2. SPECIFICATIONS OF THE PULSER

Vin L Ci (i=1-n) n ILmax 20V 1mH 10nF 6 7A

A SK50 Gar 065 (SEMIKRON) and three SK25 GB 065 (SEMIKRON)

packages are used as solid-state devices in the power board. SK50 Gar 065

package contains an IGBT and a rectifier that makes it ideal for the buck side of

topology whereas SK25 GB 065 package is composed of two similar IGBTs.

Therefore series connection of three of them enables the generation of kV range

pulses. A Texas Instrument microcontroller (TIF28335 DSP) is used to make

60

switching decisions for the circuit. Triggering signals are then produced through

Skyper 32-pro (SEMIKRON) gate drives which provides necessary insulations

between switching signal ground and the power ground.

The pulse modulator board is supplied through a dc power supply that has

restrictions on increasing input voltage and current. Due to the obligatory current

restriction, maximum possible current of 7A is supplied to the pulser and the

output pulse with a voltage level of 1.33 kV is produced. The input current and

produced voltage waveforms are shown in Fig. 1.32(a). As is evident in Fig.

1.32(b) and 1.32(c), the produced voltage is shared relatively equally by the

output stages of the pulse supply. In this case each stage can withstand one sixth

of output voltage that is 220 V. However the rising time shown in Fig. 1.32(c)

which is 700ns can be shortened by increasing the input current.

(a) (b)

(c)

Fig.1. 32. The experimental results of developed six-stage buck-boost based pulse supply

1.3.3. New configurations for MG

The solid-state technology has already been utilized in pulsed power generation.

MG is a pulse supply which has already taken advantage of utilizing compact,

61

efficient and repetitively operated solid-state technology as switching devices.

Although the control and structure simplicity is a remarkable benefit, it brings a

number of disadvantages along including utilizing a huge number of switching

and blocking devices that result in an increase in initial cost and operation losses.

Additionally, it requires fast switching devices such as IGBTs that necessitate

intricate and expensive driving modules. A few design modifications and

utilizing power electronics techniques in switching can end to improved

topologies with far fewer components. This will result in a significant saving in

initial cost, weight, volume and operation losses of MGs.

1.3.3.1. A resonant based converter for pulsed power purposes

Resonance is an energy exchange phenomenon among passive (inductive and

capacitive) components that can be used in different operation modes of MGs to

achieve the abovementioned goals. Considering that, a high voltage converter

which can be developed for pulsed power applications is proposed based on

resonance concept in the beginning. Adding a diode to the resonant circuit results

in an incomplete resonant cycle. However a half a cycle resonance may occur if

the exchange process resumes in diode conduction direction forcing the diode to

be forward biased. The relevant current and voltage waveforms which are given

by Eq. (1-23) and (1-24) can be found in Fig. 1.33.

)cos1()(CL

tVtV dcC

⋅−= (1-23)

)(sin)(CL

tV

L

CtI dcL ⋅

⋅⋅= (1-24)

Cap

acitor

vol

tage

Inductor curren

t

Cap

acitor

vol

tage

Fig.1. 33. (a) Resonant circuit, (b) Half a resonant circuit, (c) Capacitor voltage and Inductor

current of a typical resonant circuit. (d) Capacitor voltage and Inductor current of a typical half a resonant circuit

62

The general block diagram of this converter based on preliminary principle of

using a two-leg diode-capacitor unit fed through an input inductor given in Fig.

1.34(a) is shown in Fig. 1.34(b).

(a) (b)

Fig.1. 34. (a).Bidirectional resonant circuit (b).The resonant converter

Fig.1. 35. A block diagram of proposed resonant converter

As given in Fig. 1.35, a voltage source inverter (VSI) supplies the resonant

converter with an alternative voltage waveform. D1 conducts during the period in

which the input voltage is positive and as a result of the resonant between L and

C1, the capacitor (C1) is charged. In the next switching half a cycle, when the

input voltage is negative, D2 is forward biased and C2 is charged in a similar

trend as C1 but with an inverse polarity. Therefore the summation of voltages

across C1 and C2 appears at the output. If the resonant frequency given by Eq. (1-

25) is either equal to or more than the switching frequency, the capacitors will be

charged twice the input voltage and four times the input voltage will be at the

output. Otherwise the resonant half cycles are not performed properly and the

capacitor voltages could be in any level from zero to twice the input voltage

according to the switching duty cycles. In this case the capacitors are charged

dissimilarly and the ultimate voltage across each capacitor relies on the inverters

switching frequency and the initial voltage across the capacitor at the end of

switching half a cycle. Assuming the switching frequency as (1/2t1), indicating

that the inverter switches are triggered at t1, the voltage and current of the

resonant circuit at t1 given by Eq. (1-26) and (1-27) can be assumed as the initial

condition for the subsequent operation of the resonant circuit.

CLf r

⋅=

π2

1 (1-25)

63

)()cos1()( 111

1 ttVCL

tVtV dcC ==

⋅−=

(1-26)

)(sin)( 111

1 ttICL

t

L

CVtI dcL ==

⋅⋅⋅=

(1-27)

The circuit behaves for the rest of the time until the inductor is fully discharged,

and this leg is disconnected. The ultimate voltage and current across this

capacitor can be found as follows: (Vin=-Vdc during this period).

dcdcC VCL

t

C

LI

CL

tVVtV −

⋅⋅⋅+

⋅⋅+= sin)(cos)()( 11 (1-28)

CL

tI

CL

t

L

CVVtI dcL ⋅

⋅+⋅

⋅⋅+−= cossin)()( 11 (1-29)

The abovementioned operation processes are according to bipolar control

method of the inverter shown in Fig. 1.36, which enables the inverter to supply

the resonant circuit with two levels of +Vdc and -Vdc. As already discussed, the

output voltage adjustment through the increase in switching frequency is not

possible due to asymmetrical charging of capacitors in this method. Producing

zero voltage level intervals between positive and negative voltage polarities

through unipolar control method of the inverter is a technique that enables the

inverter to symmetrically charge the capacitors. The relevant switching states are

given in Fig. 1.37.

Fig.1. 36. Operation modes of the resonant converter supplied with an inverter controlled with

bipolar method.

Fig.1. 37. Extra states of inverter providing resonant converter with the zero level of voltage in

unipolar control method.

The simulation results of the converter supplied by an inverter with both bipolar

and unipolar control methods and several proportions of switching and resonant

frequencies are given in a paper titled “A bidirectional two-leg resonant

64

converter for high voltage pulsed power applications”. The paper has been

presented at IETEPPC2009 in Geneva, Switzerland [212]. The content of this

paper composes 7th chapter of this Thesis.

1.3.3.2. A resonant based Marx Generator

In the next stage, the development of this resonant converter has been considered

in order to produce a higher voltage level. Marx topology has the potential to

exploit the advantages of resonant phenomenon in its operation. Therefore a new

configuration for a resonant based MG is designed to benefit from the energy

transaction process between the inductor and the capacitors. The circuit

schematic diagram of this topology is shown in Fig. 1.38.

D1 D4

C1 C4

LS1

S2

S3

S4

V in

D2

C2

D3

C3

S5

S8

S6 S7

S9+

-

+

- +

-

+

-

D5

D6 D7

Single Phase Inverter Resonant Marx Generator

Vdc

Vinv(t)

Fig.1. 38. Using resonant concept in Marx topology

The preliminary idea is the same as the resonant converter and the connection of

several charged capacitors is the issue considered in this new configuration.

Making this possible, the diode-capacitor legs are arranged in two groups with

opposite directions. An H-bridge inverter in the entrance supplies the resonant

Marx topology with an alternative voltage waveform. Under the bipolar control

method, the inverter has two switching modes converting positive and negative

voltage levels to the output. Therefore each group of Marx capacitors is charged

in an inverter switching mode through a half a cycle resonant with the inductor.

The charged capacitors are then connected in series through S8, and S9 (and any

further). S5, S6 and S7 (and any further) are switched off in this mode. The

operation cycle in this converter includes three principle switching states which

can be extended to five for adjustability purposes. The concept of resonant is

hereby utilized and developed for an MG. The detailed discussions of this

converter accompanied by simulation results are presented as a conference paper

entitled “A Resonant Based Marx Generator” at 20th Australian Universities

Power Engineering conference held in December 2010 at Christchurch, New

Zealand [213].

65

1.3.3.3. A new Configuration for Marx Generator utilizing fast and slow

solid-state switches

Considering the earlier proposed topology for resonant Marx, its pros and cons, a

proper configuration is designed and developed in order to minimize former

short comes and to improve it in terms of having fewer active devices, and

driving modules and also less conduction and switching losses.

1.3.3.3.1. Topology

To fulfil theses desires, diode-capacitor legs arrangement is slightly revised.

Bidirectional diode-capacitor units are formed by connecting diodes with

opposite directions in alternatively arranged legs. The block diagram of proposed

pulsed power supply shown in Fig. 1.39 comprises an ac-dc converter in the

front side, a voltage regulator, a dc-ac converter and an MG topology with a new

configuration. A full bridge rectifier rectifies the grid voltage and supplies the

modulator with a dc voltage. A large capacitor at the output of the rectifier

regulates voltage fluctuations and provides the rest of the topology with a

smooth and continuous voltage level. Subsequently in the next stage, this dc

voltage is inverted to an alternative voltage waveform by a single leg inverter.

The reason behind using a half bridge inverter is utilizing fewer active power

switches however a full bridge inverter could supply MG with more flexibility

enabling the symmetrical adjustment of generated voltage level [212, 213]. This

alternative voltage which has three levels of +Vdc, –Vdc and zero, is applied to an

inductor in the entrance of Marx topology. The configuration presented in this

paper as Marx topology uses a new arrangement of capacitors, power diodes and

solid-state power switches. This topology consists of bidirectional diode-

capacitor units which are connected together through two solid-state switches

with opposite directions. In this configuration each two stages of MG is

composed of two capacitors, two diodes and two power switches that indicates

one diode is reduced in each stage.

AC Grid

Rectifier

AC-DC Converter

Novel Marx topology(Bidirectional diode-capacitor units)

220 V50 Hz Single leg

Inverter, VSI

DC-AC Converter

Voltage Regulator

V inv(t)+

-

Fig.1. 39. The block diagram of proposed converter with a new Marx configuration

66

1.3.3.3.2. Switching Modes

A simplified four-stage MG shown in Fig. 1.40 is simulated and practically

implemented to investigate its operation features and to carry out further analysis

on its performance. The approaches can be extended for a multi-stage MG.

Considering supplied voltage levels through the inverter to the Marx

configuration, +Vdc, –Vdc and zero levels, three principal operation modes are

defined for this topology.

Fig.1. 40. The four-stage simulated model of proposed MG

In positive charging mode shown in Fig. 1.41(a) inverter’s high side switch, S1

conducts whereas all remained switches are off. D1, D3 and DS3 are forward

biased in this mode and as a result a circulating path composed of the inductor,

L, C1, and C3 is completed. The capacitors are charged to double input voltage,

2Vdc through a half a cycle resonance with the inductor. The diodes prevent the

reverse current flowing and consequently next resonant half a cycle is not

occurred. The components behavior during the resonant is thoroughly expressed

through Equations (1-30)-(1-32).

31 CCCeq +=+ (1-30)

)cos1()(+⋅

−=eq

dcCCL

tVtV (1-31)

)(sin)(+

+

⋅⋅=

eq

dceq

LCL

tV

L

CtI (1-32)

This operation mode lasts until S1, S2 and S3 are triggered off, on and on

respectively. In this instant the second switching mode (negative charging mode)

shown in Fig. 1.41(b) is resumed. D2, D4 that are forward biased conduct in this

mode and let another half a cycle of resonance occurs between L, C2 and C4 that

charges the capacitors to double input voltage with an opposite polarity. In the

next switching mode, (Load supplying mode) shown in Fig. 1.41(c) S4 is

switched on whereas the rest are switched off simultaneously to connect the

capacitors in series and to give the aggregation of voltages at the output.

67

Fig.1. 41. Switching states of proposed MG (a) Positive charging mode (b) Negative charging mode (c) Load supplying mode,

1.3.3.3.3. Simulation results

Simulations of this topology are carried out in MATLAB/Simulink platform and

results are captured in order to conduct further analyses on its performance and

to investigate the operation features of this model. Voltages across and currents

through all power diodes and switches are given in Fig. 1.42(a) and 1.42(b)

respectively. It indicates that the voltage across two switches S3 and S4 in the

internal circuit is equally shared by the switches. Fig. 1.43(a) gives switching

pulse patterns in addition to the associated voltage and current waveforms for a

single shot pulse generation. The voltage and current waveforms of this model

working as a repetitively operated pulse generator are presented in Fig. 1.43(b).

The voltage produced by aggregation of charged capacitors in each cycle is

discharged then through the load and allow next supplying cycle to be resumed.

Fig.1. 42. The components voltages and the currents (a) Diodes, (b) Switches

68

0 1 2 3 4 5 6

-200

0

200

Inpu

t vol

tage

(V)

0 1 2 3 4 5 6-4

-20

0

20

Indu

ctor

cur

rent

(A)

0 1 2 3 4 5 6

x 10-4

-500

0

500

1000

1500

Time(s)(b)

C1,

C2,

C3,

C4 a

nd

ou

tpu

t vo

ltage

s(V

)

C1

C2

C3

C4

Output

(a) (b)

Fig.1. 43. (a). Current and voltage waveforms accompanied by relevant switching signal patterns, (b). Simulation results of proposed repetitively operated topology

1.3.3.3.4. Generated Voltage Adjustability

As can be inferred from the circuit analysis, Equations (1-30)-(1-32), and the

simulation results, the inverter’s switching frequency should necessarily be less

than the resonant frequency to have maximum potential voltage generation at the

output of the converter. However, the inverter’s switching frequency cannot be

more than the resonant frequency unless the inverter switches have anti-parallel

diodes as shown in Fig. 1.44.

D1

C1

L

S1

S2

Vdc D2

C2

+

-

-

+

S3

DS3

D3

C3

D4

C4

+

-Vdc

-

+

S4

DS2

DS1

Fig.1. 44. Using switches with anti-parallel body diodes in the inverter

In this case the inductor charge and consequently the capacitor charges will be

different in two half cycles unless the inverter switches duty cycles vary. It

indicates that the capacitors’ symmetrical charging and accordingly the

adjustment of the generated voltage level are relatively impossible in this way.

69

Simulation results, given in Fig. 1.46(a) clarify that capacitor residual charges

after the load supply mode will be different in this case. This is due to the

asymmetrical initial charges and may cause malfunction in normal performance

of the power supply. To give this feasibility to the modulator, two hardware

solutions are available. The first is using a full H-bridge inverter instead of the

half bridge one and controlling it via unipolar modulation method [212, 213].

The second is providing a reserve path for the current which can be

accomplished by installing a bidirectional solid-state switching connection

shown in Fig. 1.45(a) in the junction of the inverter and the inductor as given in

Fig. 1.45(b). In this way a reserve path will be created for the current to be

flowed through it once both the inverter switches become off during a resonant

half a cycle. That is how the unipolar method can be adopted for a single leg

inverter in order to supply the inverters’ load with zero voltage levels in the

middle of positive and negative voltage level intervals. In this way the stored

current in the inductor has sufficient time to be delivered to the capacitors and

the inductor will be free of charge for the next resonant half a cycle. These

devices are just triggered for voltage adjustability purposes. Simulation results

given in Fig. 1.46 confirm that how practical this solution is in symmetrical

charging of the capacitors. Two extra switching states according to this control

method are demonstrated in Fig. 1.45(c)

(a) (b)

(c)

Fig.1. 45. (a). Bidirectional solid-state switching path (b). Proper installation point of the reserve path (c). Extra switching states associated with the unipolar control method of the half bridge

inverter

70

0 1 2 3 4 5 6-500

0

500

Inpu

t vo

ltage

(V)

0 1 2-200

0

200

Inpu

t vol

tage

(V)

0 1 2 3 4 5 6-4

-60-40-20

020

Indu

ctor

cur

rent

(A)

0 1 2-4

-20

0

20

Indu

ctor

cur

rent

(A)

0 1 2 3 4 5 6x 10

-4

-500

0

500

1000

1500

Time(s)(a)

C1,

C2,

C3,

C4,

and

out

put

volta

ges(

V)

C1

C2

C3

C4

Output

0 1 2

x 10-4

-500

0

500

1000

1500

Time(s)(b)

C1,

C2,

C3,

C4,

and

outp

ut v

olta

ges

(V)

C1 voltage

C2 voltage

C3 voltage

C4 voltage

Output voltage

Fig.1. 46. Simulation results for the converters with (a). Anti-parallel body diodes (b). Reserve

path.

1.3.3.3.5. Experimental results

A four-stage laboratory prototype set up has been implemented to investigate the

concept of this circuit practically and to compare the simulation and the

hardware results. TIF28335 DSP is the microcontroller used to run this set up. A

general overview of the prototype including the power board, the control

modules and the gate drives is shown in Fig. 1.47. The components

specifications are addressed in Table 1.3.

TABLE 1. 3. SPECIFICATIONS OF THE IMPLEMENTED CIRCUIT

Vin L C1,2,3,4 finv fr

30 V 445 µH 10 nF NA 53.3 kHz

71

Fig.1. 47. Hardware set up

Experimental tests were conducted in low voltage range due to the voltage

restrictions of the input dc power supply. The input voltage was adjusted to 30 V

and the resonant frequency determined through the capacitor and the inductor

sizes is 53.3 kHz. The resonant time span shown by the inductor current in Fig.

1. 48(a) and the current amplitude, 200 mA, verify the energy exchange process

between the inductive and capacitive components of the circuit according to the

anticipations. As can be seen in Fig. 1.48(a), the capacitors are charged up to

50V each, and the summation of voltages which is 200 V appears at the output at

the last stage of the operation. The summation of voltages across C1 and Cn (n=4

in this case) appeared across the load during initial two modes. The rest of

voltages (VC2+...+VCn-1) are added to this level by triggering on S4 (and its

multiple switches) at the third mode. The voltages across S3 and S4 are shown in

Fig. 1.48(b).

IL

VC1&VC3

VC2&VC4

VOUT

(a) (b)

Fig.1. 48. Experimental results for (a) The capacitors and the output voltages and the inductor current (b) The voltages across S3 and S4.

72

1.3.3.3.6. Structure and Performance Comparison

In comparison with the conventional solid-state Marx topology, it can be seen

that the generated voltage in each stage is twice the input voltage due to the

resonance; therefore the number of needed stages to generate similar voltage

levels is reduced to half of the conventional Marx stages. Furthermore, even the

number of diodes for each stage is decreased to one diode compared to two

diodes in the conventional configuration. Thus, not only the initial cost will drop

but also there will be a noticeable power loss reduction in the capacitors charging

process. Although the number of solid-state switches remained the same as a

conventional MG (one switch in each stage) the type of employed switches can

be varied. In a conventional MG, all switching devices should necessarily be fast

switches like IGBTs, whereas slow switches such as GTOs or SCRs can be

utilized as S3 (and its multiple switches) in this topology. Therefore a fast and a

slow switch can be employed in each of the two stages. That leads to a

significant saving in the driving modules.

On the other hand, the number of solid-state switches in discharging path

becomes one switch associated with two stages. This has been two switches for

two stages in former technology. It means that the load supplying process will be

done with less power losses and accordingly higher efficiency. Another

advantage of this topology is utilizing resonant phenomenon as the operation

method and triggering the switches at the instant at which the flowing current

through them is zero. That leads to keep the switching losses in a minimum

possible level. A single-leg inverter is the only extra device utilized in this

method compared with the previous version. It is quite reasonable by considering

the point that it brought an undeniable number of advantages to this topology. In

this converter, the pulse generation frequency is restricted by the resonant

frequency. The smaller L and Ceq are the higher repetition rate can be achieved.

A paper entitled “A Solid State Marx Generator with a Novel Configuration” and

including the primary concept of this pulse generator and some simulation results

has been presented in 19th Iranian Conference on Electrical Engineering in May

2011 in Tehran, Iran [214]. Extra discussions on the topology features

accompanied by more simulation and experimental test results were submitted as

73

an article with the same title to IEEE Transactions on Plasma Science [215] that

is published in this journal. The paper forms the fifth chapter of this thesis.

1.3.3.4. A new family of Marx Generators based on commutation circuits

Commutation is a term defining techniques used for triggering semiconductor

switching devices in power electronics circuits. Placing a resonant circuit across

a Thyristor and letting it to become off, free of any external force is one of these

techniques called self-commutation. Considering this method in switching

process, another new configuration is introduced for MGs that utilizes self-

commutation connections as auxiliary circuits along with the capacitors in the

MG topology. A new configuration is proposed with a different arrangement of

solid-state switches, power diodes capacitors and inductors. In this way the

number of required fast switching devices such as IGBTs or MOSFETs for the

MG is reduced to half and slow switching devices such as Thyristors that require

fewer driving modules are used instead of them. This configuration also utilizes

fewer power diodes in comparison with conventional Marx configuration.

1.3.3.4.1. Topology

As shown in Fig. 1.49, the power supply is composed of a full bridge rectifier, a

modified buck-boost converter and the new configuration of Marx. A detailed

circuit diagram of proposed configuration is given in Fig. 1.50. The rectifier in

the front side regulates the grid voltage and provides the rest of converter with a

consistent voltage supply. The energy is delivered to the Marx topology through

a modified positive buck-boost converter providing voltage boosting skill and

insulation of load side from input side simultaneously. Marx configuration is

composed of two-leg diode-capacitor units connected together through an IGBT,

S4, and a power diode, D5. The diodes of each unit are connected in similar

directions. One capacitor in each unit is connected to an inductor through a

Thyristor that act as a self-commutation circuit.

AC Grid

Rectifier

AC-DC Converter

Modified positive Buck-Boost Converter

Novel Marx topology(Two leg diode-capacitor units)

220 V50 Hz

Fig.1. 49. Block diagram of new Marx topology

74

D1 D4

C1 C4

L1

Vdc

D2

C2

D3

C3

S3

S4

+

-

+

-

+

-

+

-

D5

S1

S2

SCR1

Df

SCR2

Two leg diode-capacitor unit 1 Two leg diode-capacitor unit 2Modified positive Buck-Boost Converter

L2 L3

+

-

+

-

Fig.1. 50. Circuit diagram of the proposed topology.

1.3.3.4.2. Switching Modes

The switching states of this topology are shown in Fig. 1.51. As given in Fig.

1.51(a), the input voltage charges the positive buck-boosts’ inductor in the first

switching mode. The buck switch S1 controls the inductor charge by

disconnecting the voltage supply in specific times.

dt

diLVV L

Ldc1

1 1 ⋅== (1-33)

dcL V

iLt max

1 ⋅=∆ (1-34)

2max12

11

iLEL ⋅= (1-35)

The converter’s diode Df is in charge of circulating the current when S1

disconnects the input side. Therefore charged inductor circulating current

through S2 and Df acts as a current source prepared to conduct the current into the

capacitors and to charge them. This can be considered as either a separate

switching mode or a combination with next state as defined in this proposal. In

the next switching mode, the energy stored in the buck-boost inductor can be

delivered to the Marx configuration by triggering off and on the boost switch, S2,

and the connecting switch, S4, respectively.

4321 CCCCCeq +++= (1-36)

222

211 2

1))()((

2

111 CeqLL VCtitiL ⋅=−⋅ (1-37)

eqLC C

LiV 1

14,3,2,1⋅= (1-38)

Once the capacitors are charged, S2 and S3 are turned on and off respectively in

order to disconnect the current source from the Marx configuration. The

Thyristors are switched on simultaneously to connect the resonant inductors, L2

and L3, to the associated capacitors, C2 and C4. This circuit creates a path for the

energy stored in the capacitors to be exchanged with the inductors and to be

returned to the capacitors in a half a cycle resonant. This allows the capacitors to

75

be charged with an opposite polarity of voltage at the end of this process. At the

last stage charged capacitors are connected in series through S4. Therefore the

capacitor voltages are collected and a summation of voltages appears at the

output in a significantly short period of time.

Fig.1. 51. Switching states of the proposed Marx generator for single pulse generation.

Furthermore, for a repetitive operation of this pulse supply, a permanent current

source is needed which is provided by a large inductor with a high level of

charge. For this purpose the third and the fourth switching modes are revised by

turning on S1 in these modes. This change permits the inductor to be recharge

during commutation and pulse supplying modes and assists the converter to keep

the inductor continuously charged in a specific level. The two extra switching

states are shown in Fig. 1.52.

Fig.1. 52. Extra switching states of the proposed Marx generator for repetitive pulse generation.

1.3.3.4.3. Control Strategy, Simulations and Experimental Results

The control strategies adopted are based on two functions assigned to this

generator. Two parameters, the inductor current and a capacitor voltage (VC2),

are measured in order to determine the switching times and to make the proper

decisions in these distinguished instants. A controller processes and analyses the

measured samples and makes the decisions. Then the signals are amplified

through the interfaces (Op-Amps) and delivered to driving modules in order to

produce the switching signals. The control algorithm of the converter is given in

Fig. 1.53(a) and the produced signal patterns are shown in Fig. 1.53(b).

76

VC1&3

Vin

S1

S2

S3

S4

SCR1

SCR2

Voltages

Input voltageOutput voltageC

1 voltage

C2 voltage

C3 voltage

C4 voltage

VC2&4

Vout

Fig.1. 53. (a). Control flowchart for a modulator with the repetitive pulse generation function,

(b). The capacitor voltages and the gate drive waveforms of the converter.

Two models have been simulated as single shot and repetitively operated

modulator in order to carry out further analyses on performance features of this

converter. A simple four-stage lab prototype shown in Fig. 1.54 is implemented

then to conduct primary tests and the achievements shown in Fig. 1.55 verify the

validity of proposed modulator. The capacitor and the output voltages and the

input inductor current are shown in Fig. 1.55(a). The operation modes, including

the inductor and capacitors charging modes, followed by the commutation and

the pulse generation modes can be distinguished in this figure. The summation of

voltages across C1 and Cn (n=4 in this case) appears across the load during third

(commutation) mode. The rest of voltages (VC2+...+VCn-1) are added to this level

by triggering on S4 (and its multiple switches) at the fourth (pulse generation)

mode. The energy exchange process in the commutation circuits is illustrated in

Fig. 1.55(b), through depicting the involved capacitor (C2 and C4) and inductor

(L2 and L3) voltage and current waveforms.

77

Fig.1. 54. Experimental set up.

(a) (b)

Fig.1. 55. Experimental results.

Out of these calculations, simulations, tests and analyses, a journal paper has

been published at IEEE Transactions in Dielectrics and Electrical Insulations

entitled “A New Family of Marx Generators Based on Commutation Circuits”

[216]. Topology design process, operation states, control strategies, simulation

and experimental results, and other concerns regarding voltage ratings of

switches, components sizes and Electromagnetic interferes (EMI) issues are

discussed in this paper in detail. The paper is presented in this thesis as chapter 6.

A conference paper entitled “A New Family of Marx Generator Based on

Resonant Converter” has been presented at IEEE Energy Conversion Congress

and Exposition (ECCE2010) held in Atlanta, USA [111]. This paper’s focus was

in the introduction of a new generation of Marx Generators with a new

configuration based on modified positive buck-boost concept and resonant

converters.

78

1.3.4. A high voltage converter based on capacitor –diode voltage

multiplier (CDVM) with a frequency and voltage controller

CDVMs have been widely used in space and communication applications.

Among them, Cockcroft-Walton multiplier topology has a remarkable role in

voltage promotion in microelectronics related configurations such as, radio

frequency passive transponders [129], passive wireless micro sensors [130] and

battery-operated devices [131]. The advantages of CDVM in these applications

include being efficient, reliable and having small size and weight. Simplicity is

another remarkable benefit of CDVM circuits. Each stage consisting of a couple




transistors and their relative circuits like control boards and driving stacks. The

main disadvantages of CDVM in these cases include the delay between input and

output and the non-negligible amount of capacitance needed, but this can be

reduced within acceptable limits by increasing multipliers’ operating frequency

via an ac-ac converter placed in the input of multiplier [132]. Three different

configurations of these voltage multipliers, including simple N-stage schematic

of both a Cockcroft-Walton voltage multiplier and a Dickson charge pump are

depicted in Fig. 1.56.


Multiplier (b). N-stage Dickson charge pump (c). Another N-stage CDVM configuration

79

1.3.4.1. A high voltage converter based on capacitor diode voltage multiplier

(CDVM) with a frequency and voltage controller

CDVMs capabilities and advantages were sufficiently tempting for high voltage

specialists to convince them in order to utilize these topologies in high voltage

applications. The pulsed power generation can also benefit from CDVMs skills.

To study the feasibility of utilizing CDVM in charging the capacitors with initial

voltages for pulse generation purposes, a simple two-stage Cockcroft-Walton

Voltage Multiplier shown in Fig. 1.57 is analyzed theoretically and modeled in

simulations platforms.

Fig.1. 57. One-stage Cockcroft-Walton voltage multiplier

To investigate the dynamic response of this circuit the simulations are conducted

under two different conditions, Similar and different capacitors. The circuit is fed

with a 50Hz, 200V sinusoidal voltage waveform at the input. The simulation

results are given in Fig. 1.56. The voltage transient in case of identical capacitors

takes eight cycles whereas the dynamic mode lasts just three cycles in case of

different capacitors (C1=10C2). The voltage variation across C2 can be followed

through Eq. (1-39).

)1(2

)(22

21

2

21

1 −+

++

= iVCC

CV

CC

CiV CSMC For i>1, SMC V

CC

CV

21

1)1(2 +

= (1-39)

Where VSM and i represents the amplitude of the input voltage and the number of

cycles respectively. In a specific case when (C1=C2), previous equation could be

simplified as:

2

)1()( 2

2

−+=

iVViV C

SMC For i>1 & )( 21 CC = 2

)1(2

SMC

VV = (1-40)

The number of cycles and consequently the time each transient lasts can be

realized through theses equations.

0 0.05 0.1 0.15 0.2 0.25

-200

-100

0

100

200

300

400

Time(s)

Inp

ut a

nd

ca

pa

cito

rs v

olta

ge

s(v)

Input voltage of multiplier(v)Voltage over first capacitor of multiplier(v)Output voltage of multiplier(v)

0 0.05 0.1 0.15 0.2 0.25

-200

-100

0

100

200

300

400

Time(s)

Inp

ut a

nd

ca

pa

cito

rs v

olta

ge

s(v)


(a) (b)

80

Fig.1. 58. Voltage transient of multiplier with 50 Hz input frequency (a). Identical capacitors (b). Different capacitors (C1=10C2)

On the other hand, shortening the input cycles can directly affect the transient

duration. Increasing the frequency of input voltage leads to a faster voltage

boosting at the output. The simulation results in Fig. 1.58 indicate two CDVMs

(with identical and different capacitors) supplied through a 1 kHz sinusoidal

voltage. As shown, the dynamic mode in voltage boosting is significantly

reduced.

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1

-200

-100

0

100

200

300

400

Time(s)

Inpu

t and

ca

pac

itors

vo

ltag

es(v

)

Input voltage of multiplier(v)Voltage over first capacitor of multiplierOutput voltage of multiplier(v)

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1

-200

-100

0

100

200

300

400

Time(s)

Inpu

t and

cap

acito

rs v

olta

ges(

v)


(a) (b)

Fig.1. 59. Voltage transient of multiplier with 1KHz input frequency (a). Identical capacitors (b).Different capacitors

Converting an ac waveform to another ac waveform with a dissimilar frequency

necessitates appropriate converters such as conventional ac-dc-ac converters as

evident in Fig. 1.60 or cycloconverters.

Fig.1. 60. An ac-dc-ac converter

To have a proper selection between these two converters, another parameter is

considered which is having control over the level of produced voltage. The

output voltage of CDVMs is a function of input amplitude. As can be seen in

Fig. 1.64(a), variation of input amplitude results in a change in the level of

produced voltage by a delay. Variable voltage level at the input of CDVM can be

produced by employing an H-bridge inverter controlled with unipolar

modulation method as a supply for CDVM. A brief review of unipolar method

control reveals how variable voltage is available in the output of an inverter. In

the unipolar modulation control method, the output voltage of the inverter has

81

three voltage levels of –Vdc & 0 & +Vdc while in the bipolar modulation, there

are just two voltage levels, –Vdc & +Vdc. Fig. 1.61(b) demonstrates one cycle of

output waveforms for both modulation methods. In both cases, changing TS gives

variation of frequency (fs) in the output. In bipolar mode, changing the average

of the output cycles is possible by changing duty cycles, while in unipolar mode,

the variation of duty cycles not only gives different output averages, but also

leads to a change in rms value of the output voltage. This eventually ends in

having variable voltage magnitudes in the output of the filter. The output voltage

of the inverter cannot be given to the multiplier directly, since high dv/dt s of this

pulsed shape waveform may cause inrush currents in the multiplier’s capacitors.

It is therefore necessary to reduce voltage stress (dv/dt). An LC filter located at

the output of the inverter eliminates high frequency harmonics and delivers high

quality voltage that has variable amplitude with respect to the variation of duty

cycles. Fig. 1.62 shows simulation results for duty cycles of 0.05, 0.5 and 0.95,

while output frequency is 50Hz.

(a) (b)

Fig.1. 61. (a). Schematic of full bridge (two-leg) inverter (b). Bipolar and unipolar modulations output waveforms

82

0.82 0.83 0.84 0.85 0.86 0.87

-200

-100

0

100

200

(a)

Out

put

volta

ges

of in

vert

er a

nd f

ilter

(v)

Inverter

filter

1.82 1.83 1.84 1.85 1.86 1.87

-200

-100

0

100

200

(b)

Out

put

volta

ges

of in

vert

er a

nd f

ilter

(v)

Inverter

filter

2.81 2.82 2.83 2.84 2.85 2.86 2.87-300

-200

-100

0

100

200

300

(c)Time(s)

Out

put

volta

ges

of in

vert

er a

nd f

ilter

(v)

Inverter

filter

Fig.1. 62. Output voltage of inverter and filter for duty cycles of (a). 0.05 (b). 0.5 (c). 0.95

A general block diagram of whole converter consisted of power modulators and

control system is shown in Fig. 1.63. The power side includes the ac-dc-ac

converter, filter, and the CDVM converter whereas a microcontroller

accomplishes all controlling functions including voltage comparing, PID

controlling and bipolar/unipolar PWM techniques.


In simulation results demonstrated in Fig. 1.64(b), the switching duty cycle of

the inverter under unipolar PWM is varied from 10% to 90%. The produced

voltage given in Fig. 1.64(c) provides an overview on how duty cycle variation

differs the produced voltage level. The influence of load connection on system

response is shown in Fig. 1.64(d).

83

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

-200

-100

0

100

200

300

400

Time(s)

Inp

ut a

nd

cap

aci

tors

vo

ltag

es(

v)


0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18-300

-200

-100

0

100

200

300

400

500

600

Time(s)

Inp

ut a

nd

ca

pa

cito

res

volta

ge

s(v)

Input voltage of multiplier(v)Voltage over first capacitore of multiplier(v) Output voltage of multiplier(v)

(a) (b)

0 0.01 0.02 0.03 0.04 0.05 0.06-250

-200

-150

-100

-50

0

50

100

150

200

250

Time(s)

Inve

rte

r ou

tpu

t vol

tag

e(v

)

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1

-300

-200

-100

0

100

200

300

400

500

600

Time(s)

Inp

ut a

nd

ca

pa

cito

rs v

olta

ge

s(v)


(c) (d)

Fig.1. 64. (a) Variable input voltage results in variable voltages in the output (b). Variable output voltage provided by an inverter under unipolar control method. (c). Inverters’ output waveform with duty cycles of 0.1 & 0.5 & 0.9. (d). Load connections and voltage rehabilitation capability

Out of abovementioned analyses and simulations, a conference paper entitled “A

high voltage power converter with a frequency and voltage controller” is

presented at 17th IEEE Pulsed Power Conference (17IEEE_PPC 2009) held in

Washington DC, USA [217] that is contributed in this thesis as chapter 8.

84

1.4. References:

[1] http://en.wikipedia.org/wiki/Pulsed_power, Wikipedia

[2] Akiyama, H.; Sakai, S.; Sakugawa, T.; Namihira, T.; "Invited Paper - Environmental Applications of Repetitive Pulsed Power," Dielectrics and Electrical Insulation, IEEE Transactions on , vol.14, no.4, pp.825-833, Aug. 2007

[3] Akiyama, H.; Sakugawa, T.; Namihira, T.; Takaki, K.; Minamitani, Y.; Shimomura, N.; , "Industrial Applications of Pulsed Power Technology," Dielectrics and Electrical Insulation, IEEE Transactions on , vol.14, no.5, pp.1051-1064, October 2007

[4] Y. Sun, P.T. Vernier, M. Behrend, L. Marcu and M.A. Gundersen, “Electrode microchamber for noninvasive perturbation of mammalian cells with nanosecond pulsed electric fields”, IEEE Trans. Nanobioscience, Vol.4, pp.277-283, 2005.

[5] R. Nuccitelli, U. Pliquett, X. Chen, W. Ford, R. J. Swanson, S. J. Beebe, J. F. Kolb and K. H. Schoenbach, “Nanosecond pulsed electric fields cause melanomas to self-destruct”, Biochem. Biophys. Res. Comm., Vol. 343, pp. 351-360, 2006.

[6] N. Nomura, S. Abe, I. Uchida, K. Abe, H. Koga, S. Katsuki, T. Namihira, H. Akiyama, H. Takano and S. Abe, “Response of biological cells exposed on burst RF fields”, 15th IEEE Pulsed Power Conf., pp.1437-1440, 2005

[7] A.H. El-Hag, S.H. Jayaram and M.W. Griffiths, “Inactivation of naturally grown microorganisms in orange juice using pulsed electric fields”, IEEE Trans. Plasma Sci., Vol. 34, pp. 1413-1415, 2006.

[8] S. Tsukamoto, K. Yamamoto, H. Kudoh, S. Ohga and H. Akiyama, “Development of an automatic electrical stimulator to mushroom sawdust pots”, 15th IEEE Pulsed Power Conf., pp.1437-1440, 2005

[9] J. Pelletier and A. Anders, "Plasma-based ion implantation and deposition: a review of physics, technology, and applications", IEEE Trans. Plasma Sci., Vol. 33, pp. 1944-1959, 2005

[10] B. Sharkov and R. Scrivens, "Laser ion sources", IEEE Trans. Plasma Sci., Vol. 33, pp. 1778- 1785, 2005.

[11] M.S. Trtica , V.F. Tarasenko, B.M. Gakovic, A.V. Fedenev, L.T.Petkovska, B.B. Radak, E.I. Lipatov and M.A. Shulepov, "Surface modifications of TiN coating by pulsed TEA CO2 and XeCl lasers", Appl. Surf. Sci., Vol. 252, pp. 474–482, 2005.

[12] X. P. Zhu, M. K. Lei and T. C. Ma, "Surface morphology of titanium irradiated by high-intensity pulsed ion beam", Nucl. Instrum. Methods Phys. Res. B, Vol. 211, pp. 69-79, 2003.

[13] R.W. Bruce, R.L. Bruce, A.W. Fliflet, M. Kahn, S.H. Gold, A.K.Kinkead, D. Lewis, III and M.A. Imam, "Joining of ceramic tubes using a high-power 83-GHz Millimeter-wave beam", IEEE Trans. Plasma Sci., Vol. 33, pp. 668- 678, 2005.

[14] P.Y. Lee, H. Suematsu, W. Jiang, K. Yatsui and K Niihara, "Synthesis of Al2O3-ZrO2 Nanocomposite Powders by Pulsed Wire Discharge ", IEEE Trans. Plasma Sci., Vol. 34, pp. 1190- 1194, 2006.

[15] K. Takaki, Y. Mikami, M. Itagaki, S. Mukaigawa, T. Fujiwara and S.Nakamura, "Influence of Metal Foil Width on Bonding Strength in Capacitor Discharge Ceramics Joining", IEEE Trans. Plasma Sci., Vol. 34, pp. 1709-1714, 2006.

[16] J.R. Woodworth , D. Chalenski, G.S. Sarkisov and J.R. Blickem,”170-kV Laser-Triggered Water Switch Experiments”, IEEE Trans. Plasma Sci., Vol.33, pp.2051-2059, 2005.

[17] T. Namihira, S. Tsukamoto, D. Wang, S. Katsuki, R. Hackam, H. Akiyama, Y. Uchida and M. Koike, “Improvement of NOx removal efficiency using short width pulsed power”, IEEE Trans. Plasma Sci., Vol. 28, pp.434-442, 2000.

[18] W. J. M. Samaranayake, Y. Miyahara, T. Namihira, S. Katsuki, T. Sakugawa, R. Hackam and H. Akiyama, “Pulsed streamer discharge characteristics of ozone production in dry air”, IEEE Trans. Dielectr. Electr. Insul., Vol. 7, pp.254-260, 2000.

[19] T. Sakugawa, D. Wang, K. Shinozaki, T. Namihira, S. Katsuki and H. Akiyama, "Repetitive short-pulsed generator using MPC and blumlein line”, IEEE 14th Pulsed Power Conf., pp. 657-660, 2003.

85

[20] H. Akiyama, T. Fudamoto, S. Katsuki, T. Namihira and T. Sakugawa, “Industrial Applications of Pulsed Power”, IEEJ Trans. FM, Vol. 125, pp.717-722, 2005

[21] J.S. Chang, P.A. Lawless and T. Yamamoto, “Corona discharge process”, IEEE Trans. Plasma Sci., Vol.19, pp.1152-1166, 1991.

[22] R. Hackam and H. Akiyama, “Air pollution control by electrical discharges”, IEEE Trans. Dielectr. Electr. Insul., Vol.7, pp.654-683, 2000.

[23] T. Namihira, S. Tsukamoto, D. Wang, S. Katsuki, R. Hackam, H. Akiyama, Y. Uchida and M. Koike, “Improvement of NOX removal efficiency using short width pulsed power”, IEEE Trans. Plasma Sci., Vol.28, pp.434-442, 2000.

[24] V. Puchkarev and M. Gundersen, “Energy efficient plasma processing of gaseous emission using a short pulse discharge”, Appl. Phys. Letter, Vol.71, No.23, pp.3364-3366, 1997.

[25] H. Hao, M. Wu, Y. Chen, J. Tang and Q. Wu, “Cyanobacterial bloom control by ultrasonic irradiation at 20 kHz and 1.7 MHz”, J. Environmental Science and Health, Part A: Toxic/Hazardous Substances & Environmental Engineering, Vol.39, No.6, pp.1435-1446, 2004.

[26] H. Akiyama, “Streamer discharge in liquids and their applications”, IEEE Trans. Dielectr. Electr. Insul., Vol.7, pp.646-653, 2000.

[27] Z. Li, S. Sakai, C. Yamada, D. Wang, S. Chung, X. Lin, T. Namihira, S. Katsuki and H. Akiyama, “The Effects of Pulsed Streamerlike Discharge on Cyanobacteria Cells”, IEEE Trans. Plasma Sci., Vol. 34, pp.1719-1724, 2006.

[28] T. Namihira, T. Yamaguchi, K. Yamamoto, J. Choi, T. Kiyan, T. Sakugawa, S. Katsuki and H. Akiyama, “Characteristics of Pulsed Discharge Plasma in Water”, 15th IEEE Pulsed Power Conf., pp.1013-1016, 2005.

[29] I.V. Lisitsyn, T. Muraki and H. Akiyama, “Mechanism of shock waves generation and material destruction in a wire induced surface flashover”, Physica B, Vol.239, pp.6-8, 1997.

[30] H. Inoue, I.V. Lisitsyn, H. Akiyama and I. Nishizawa, “Pulsed Electric Breakdown and Destruction of Granite”, Japanese J. Appl. Phys., Vol.38, pp.6502-6505, 1999.

[31] H. Inoue and I. V. Lisitsyn, H. Akiyama. I. Nishizawa, “Drilling of Hard Rocks by Pulsed Power”, IEEE Electr. Insul. Mag., Vol.16, No.3, pp.19- 25, 2000

[32] D. Wang, T. Namihira, K. Fujiya, S. Katsuki, and H. Akiyama, “The Reactor Design for Diesel Exhaust Control Using, a Magnetic Pulse Compressor”, IEEE Transactions on Plasma Science, Vol. 32, No. 5, October, 2004

[33] M. Okubo, T. Kuroki and Y. Miyairi, “Low-Temperature Soot Incineration of Diesel Particulate Filter Using Remote Non-thermal Plasma Induced by a Pulsed Barrier Discharge”, IEEE Transactions on Industry Applications, vol. 40, No. 6, Nov 2004

[34] R. F. Horng, Y. P. Chang and S. C. Wu, “Investigation on the production of hydrogen rich gas in a plasma converter for motorcycle applications”, Elsevier, Energy Conversion & Management, 2006, 2155-2166

[35] “Application of Plasma Technology in Natural Gas Conversion and Treatment” , Research Institute of Petroleum Industry, http://www.ripi.ir/en/Application%20of%20plasma.asp

[36] T. Nevshemal, “Various Technologies: Plasma Reactors for Gas Decomposition with Advanced Membranes for Hydrogen Recovery”, University of Wyoming Research Products Centre, Dept. 3672, 1000 E. University Ave. Laramie, WY 82071

[37] S. Wang, Y. Konishi and O. Koudriavtsev, “A Novel Silent Discharge Type Ozoniser using Pulse Density Modulated High-Frequency Inverter”, Industry Applications Conference, 1999, Thirty-Fourth IAS Annual Meeting. Volume 2, 3-7 Oct. 1999 pp:764 - 772

[38] I. Suarasana, M. Mudurab, R. Chirab, G. Andreia, I. Munceleanua and R. Mora, “A novel type ozoniser for wastewater treatment”, Elsevier Journal of Electronics, 2005, 831-836

[39] Tanaka, Hiroshi; Obara, Minoru; , "An all solid‐state magnetic pulse compressor with amorphous metals for pumping a repetition‐rated KrF excimer laser," Review of Scientific Instruments , vol.61, no.4, pp.1196-1199, Apr 1990

[40] Shimada, Tsutomu; Obara, Minoru; Noguchi, Akira; , "An all solid‐state magnetic switching exciter for pumping excimer lasers," Review of Scientific Instruments , vol.56, no.11, pp.2018-2020, Nov 1985

[41] Kurihara, K.; Kobayashi, S.; Satoh, I.; Shibata, K.; Shigeta, M.; Masugata, K.; Yatsui,

86

K.; , "Magnetic pulse compressor using saturable transformer to excite excimer lasers," Review of Scientific Instruments , vol.63, no.4, pp.2138-2140, Apr 1992

[42] Hatanaka, Hidekazu; Tanaka, Hiroshi; Obara, Minoru; Midorikawa, Katsumi; Tashiro, Hideo; , "A 1‐kpps transversely excited atmospheric CO2 laser excited by an all‐solid‐state exciter with a magnetic pulse compressor," Journal of Applied Physics , vol.68, no.4, pp.1456-1459, Aug 1990

[43] Hatanaka, Hidekazu; Midorikawa, Katsumi; Obara, Minoru; Tashiro, Hideo; , "5 kW transversely excited atmospheric CO2 laser driven by a solid state exciter employing insulated gate bipolar transistors," Review of Scientific Instruments , vol.64, no.11, pp.3061-3065, Nov 1993

[44] Tanaka, Hiroshi; Hatanaka, Hidekazu; Obara, Minoru; Midorikawa, Katsumi; Tashiro, Hideo; , "High‐efficiency, all‐solid‐state exciters for high‐repetition‐rated, high‐power TEA CO2 lasers," Review of Scientific Instruments , vol.61, no.8, pp.2092-2096, Aug 1990

[45] Giordano, G.; Letardi, T.; Muzzi, F.; Zheng, C. E.; , "Magnetic pulse compressor for prepulse discharge in spiker‐sustainer excitation technique for XeCl lasers," Review of Scientific Instruments , vol.65, no.8, pp.2475-2481, Aug 1994

[46] Smilanski, I.; Byron, S. R.; Burkes, T. R.; , "Electrical excitation of an XeCl laser using magnetic pulse compression," Applied Physics Letters , vol.40, no.7, pp.547

[47] Von Bergmann, H.M.; Swart, P.H.; , "Thyristor-driven pulsers for multikilowatt average power lasers," Electric Power Applications, IEE Proceedings B , vol.139, no.2, pp.123-130, Mar 1992

[48] Qi-Chu Mei; Peters, P.J.M.; Witteman, W.J.; , "Improvement in the performance of an Ar-Xe laser using magnetic spiker sustainer discharge excitation," Quantum Electronics, IEEE Journal of , vol.31, no.12, pp.2190-2194, Dec 1995

[49] Narsetti, R.; Curry, R.D.; McDonald, K.F.; Clevenger, T.E.; Nichols, L.M.; , "Microbial Inactivation in Water Using Pulsed Electric Fields and Magnetic Pulse Compressor Technology," Plasma Science, IEEE Transactions on , vol.34, no.4, pp.1386

[50] Jaegu Choi; Yamaguchi, T.; Yamamoto, K.; Namihira, T.; Sakugawa, T.; Katsuki, S.; Akiyama, H.; , "Feasibility Studies of EMTP Simulation for the Design of the Pulsed-Power Generator Using MPC and BPFN for Water Treatments," Plasma Science, IEEE Transactions on , vol.34, no.5, pp.1744-1750, Oct. 2006

[51] Namihira, T.; Sakai, S.; Yamaguchi, T.; Yamamoto, K.; Yamada, C.; Kiyan, T.; Sakugawa, T.; Katsuki, S.; Akiyama, H.; , "Electron Temperature and Electron Density of Underwater Pulsed Discharge Plasma Produced by Solid-State Pulsed-Power Generator," Plasma Science, IEEE Transactions on , vol.35, no.3, pp.614-618, June 2007

[52] Douyan Wang; Namihira, T.; Fujiya, K.; Katsuki, S.; Akiyama, H.; , "The reactor design for diesel exhaust control using a magnetic pulse compressor," Plasma Science, IEEE Transactions on , vol.32, no.5, pp. 2038- 2044, Oct. 2004

[53] Pokryvailo, A.; Wolf, M.; Yankelevich, Y.; , "Investigation of Operational Regimes of a High-Power Pulsed Corona Source with an All-solid State Pulser," Dielectrics and Electrical Insulation, IEEE Transactions on , vol.14, no.4, pp.846-857, Aug. 2007

[54] Pokryvailo, A.; Yankelevich, Y.; Wolf, M.; Abramzon, E.; Wald, S.; Welleman, A.; , "A high-power pulsed corona source for pollution control applications," Plasma Science, IEEE Transactions on , vol.32, no.5, pp. 2045- 2054, Oct. 2004

[55] Samaranayake, W.J.M.; Miyahara, Y.; Namihira, T.; Katsuki, S.; Sakugawa, T.; Hackam, R.; Akiyama, H.; , "Pulsed streamer discharge characteristics of ozone production in dry air," Dielectrics and Electrical Insulation, IEEE Transactions on , vol.7, no.2, pp.254-260, Apr 2000

[56] Samaranayake, W.J.M.; Miyahara, Y.; Namihira, T.; Katsuki, S.; Hackam, R.; Akiyama, H.; , "Ozone production using pulsed dielectric barrier discharge in oxygen," Dielectrics and Electrical Insulation, IEEE Transactions on , vol.7, no.6, pp.849-854, Dec 2000

[57] Jun Li; Zanji Wang; Ying Wang; Min Gao; Haiqiang Wang; , "Effect of mutual inductance on the pulsed current amplification of magnetic flux compressors," Magnetics, IEEE Transactions on , vol.37, no.1, pp.489-492, Jan 2001

[58] Jun Li; Yinchun Gui; Qibin Deng; Chenda Yu; Ping Yan; Jiansheng Yuan; , "Resistance calculation of the reusable linear magnetic flux compressor coil," Magnetics, IEEE Transactions on , vol.41, no.1, pp. 471- 473, Jan. 2005

[59] Zanji Wang; Jun Li; Xiucheng Liu; Xinjie Yu; Jiansheng Yuan; , "Study on the coil design of the reusable linear magnetic flux compressor," Magnetics, IEEE Transactions

87

on , vol.39, no.1, pp. 442- 445, Jan 2003 [60] Engel, T.G.; Becker, J.E.; Nunnally, W.C.; , "Energy conversion and high power pulse

production using miniature magnetic flux compressors," Plasma Science, IEEE Transactions on , vol.28, no.5, pp.1342-1346, Oct 2000

[61] Jaegu Choi; Namihira, T.; Sakugawa, T.; Katsuki, S.; Akiyama, H.; , "Simulation of 3-Staged MPC Using Custom Characteristics of Magnetic Cores," Dielectrics and Electrical Insulation, IEEE Transactions on , vol.14, no.4, pp.1025-1032, Aug. 2007

[62] Burdt, Russell; Curry, Randy D.; McDonald, Kenneth F.; Melcher, Paul; Ness, Richard; Huang, Chaofeng; , "Evaluation of nanocrystalline materials, amorphous metal alloys, and ferrites for magnetic pulse compression applications," Journal of Applied Physics , vol.99, no.8, pp.08D911-08D911-3, Apr 2006

[63] Fireman, J.; , "Computer simulation of a pulse-forming network," Instrumentation & Measurement Magazine, IEEE , vol.4, no.4, pp.20-22, Dec 2001

[64] Engel, T.G.; Nunnally, W.C.; , "Design and operation of a sequentially-fired pulse forming network for non-linear loads," Plasma Science, IEEE Transactions on , vol.33, no.6, pp. 2060- 2065, Dec. 2005

[65] Nunnally, W.C.; Huenefeldt, S.M.; Engel, T.G.; , "Performance and Scalability of MJ Sequentially Fired Pulse Forming Networks for Linear and Nonlinear Loads," Plasma Science, IEEE Transactions on , vol.35, no.2, pp.484-490, April 2007

[66] Roark, R.M.; Parten, M.E.; Masten, L.B.; Burkes, T.R.; , "Pulse-forming networks with time-varying or nonlinear resistive loads," Electron Devices, IEEE Transactions on , vol.26, no.10, pp. 1541- 1544, Oct 1979

[67] Jess, J.; Schussler, H.; , "On the Design of Pulse-Forming Networks," Circuit Theory, IEEE Transactions on , vol.12, no.3, pp. 393- 400, Sep 1965

[68] Jess, J.; Schussler, H.; , "A Class of Pulse-Forming Networks," Circuit Theory, IEEE Transactions on , vol.12, no.2, pp. 296- 299, Jun 1965

[69] Palaniappan, R.; Abrol, R. K.; Datta, S. K.; Vithayathil, Joseph; , "Modified McMurray Inverter with Pulse Forming Network Commutation Circuits," Industrial Electronics and Control Instrumentation, IEEE Transactions on , vol.IECI-24, no.1, pp.66-73, Feb. 1977

[70] Gully, J.H.; Pratap, S.B.; Headifen, R.N.; Marinos, C.; Dick, W.; Goodel, B.; , "Investigation of an alternator charged pulse forming network with flywheel energy storage," Magnetics, IEEE Transactions on , vol.29, no.1, pp.969-974, Jan 1993

[71] Musolino, A.; Raugi, M.; Tellini, B.; , "Pulse forming network optimal design for the power supply of EML launchers," Magnetics, IEEE Transactions on , vol.33, no.1, pp.480-483, Jan 1997

[72] Scholfield, David W.; Butcher, Michael D.; Hilko, Brian; Dorr, Greg; , "Vacuum switch performance in a 1.2 MJ pulse forming network," Review of Scientific Instruments , vol.79, no.2, pp.024703-024703-5, Feb 2008

[73] Alexeev, A.V.; Baltakhanov, A.M.; Bondaletov, V.N.; , "Pulse forming network for railgun launcher," Magnetics, IEEE Transactions on , vol.28, no.6, pp.3372-3379, Nov 1992

[74] Spahn, E.; Buderer, G.; Wenning, W.; , "A compact pulse forming network, based on semiconducting switches, for electric gun applications," Magnetics, IEEE Transactions on , vol.35, no.1, pp.378-382, Jan 1999

[75] Carey, W.J.; Barrett, B.D.; Nunnally, W.C.; Dillon, W.E.; Eubank, E.L.; , "A pulse forming network design for blocked-bore plasma armature experiments," Magnetics, IEEE Transactions on , vol.29, no.1, pp.915-918, Jan 1993

[76] Dick, W.J.; Goldman, E.B.; Johnson, R.F.; , "Analysis of components in advanced capacitive pulse forming networks for electric guns," Magnetics, IEEE Transactions on , vol.31, no.1, pp.32-37, Jan 1995

[77] Del Guercio, M.; Stobie, I.; Katulka, G.; Oberle, W.; , "A pulse forming network design for electrothermal-chemical combustion characterization of solid propellants," Pulsed Power Conference, 1995. Digest of Technical Papers., Tenth IEEE International , vol.1, no., pp.279-285 vol.1, 3-6 Jul 1995

[78] Ross, G.F.; , "An Improved Pulse-Forming Network for the Generation of Phase-Coherent Microwave Signals (Correspondence)," Microwave Theory and Techniques, IEEE Transactions on , vol.17, no.1, pp. 52- 54, Jan 1969.

[79] Sanders, J.M.; Kuthi, A.; Yu-Hsuan Wu; Vernier, P.T.; Gundersen, M.A.; , "A linear, single-stage, nanosecond pulse generator for delivering intense electric fields to biological loads," Dielectrics and Electrical Insulation, IEEE Transactions on , vol.16,

88

no.4, pp.1048-1054, August 2009 [80] Mathis, H.; Mathis, R.; , "Darlington Pulse-Forming Network," Circuit Theory, IEEE

Transactions on , vol.14, no.3, pp. 332- 335, Sep 1967 [81] Rossi, J.O.; Barroso, J.J.; Ueda, M.; , "Modeling of Wound Coaxial Blumlein Pulsers,"

Plasma Science, IEEE Transactions on , vol.34, no.5, pp.1846-1852, Oct. 2006 [82] Davanloo, F.; Collins, C.B.; Agee, F.J.; , "High-power, repetitive-stacked Blumlein

pulsers commutated by a single switching element," Plasma Science, IEEE Transactions on , vol.26, no.5, pp.1463-1475, Oct 1998

[83] Durga Praveen Kumar, D.; Mitra, S.; Senthil, K.; Sharma, Archana; Nagesh, K. V.; Singh, S. K.; Mondal, J.; Roy, Amitava; Chakravarthy, D. P.; , "Characterization and analysis of a pulse power system based on Marx generator and Blumlein," Review of Scientific Instruments , vol.78, no.11, pp.115107-115107-4, Nov 2007

[84] Rebersek, M.; Kranjc, M.; Pavliha, D.; Batista-Napotnik, T.; Vrtacnik, D.; Amon, S.; Miklavcic, D.; , "Blumlein Configuration for High-Repetition-Rate Pulse Generation of Variable Duration and Polarity Using Synchronized Switch Control," Biomedical Engineering, IEEE Transactions on , vol.56, no.11, pp.2642-2648, Nov. 2009

[85] Liu, Jinliang; Yin, Yi; Ge, Bin; Cheng, Xinbing; Feng, Jiahuai; Zhang, Jiande; Wang, Xinxin; , "A compact high power pulsed modulator based on spiral Blumlein line," Review of Scientific Instruments , vol.78, no.10, pp.103302-103302-5, Oct 2007

[86] Rossi, J.O.; Ueda, M.; Barroso, J.J.; , "Design of a 150 kV 300 A 100 Hz Blumlein coaxial pulser for long-pulse operation," Plasma Science, IEEE Transactions on , vol.30, no.5, pp. 1622- 1626, Oct 2002

[87] Liu, Z.; Yan, K.; Winands, G. J. J.; Van Heesch, E. J. M.; Pemen, A. J. M.; , "Novel multiple-switch Blumlein generator," Review of Scientific Instruments , vol.77, no.3, pp.033502-033502-5, Mar 2006

[88] Davanloo, F.; Dussart, R.; Koivusaari, K.J.; Collins, C.B.; Agee, F.J.; , "Photoconductive switch enhancements and lifetime studies for use in stacked Blumlein pulsers," Plasma Science, IEEE Transactions on , vol.28, no.5, pp.1500-1506, Oct 2000

[89] Noggle, R. C.; Adler, R. J.; Hendricks, K. J.; , "A reliable dual Blumlein device," Review of Scientific Instruments , vol.60, no.11, pp.3540-3548, Nov 1989

[90] Rossi, J.O.; Ueda, M.; , "A 100 kV/200 A Blumlein Pulser for High-Energy Plasma Implantation," Plasma Science, IEEE Transactions on , vol.34, no.5, pp.1766-1770, Oct. 2006

[91] Keller, L. P.; Walschon, E. G.; , "Simple Marx High Voltage Pulse Generator for Wide Gap Spark Chambers," Review of Scientific Instruments , vol.37, no.9, pp.1258-1259, Sep 1966

[92] Harjes, H. C.; Kunhardt, E. E.; Kristiansen, M.; Hatfield, L. L.; Guenther, A. H.; , "Space-Charge Effects in a Laser Fiber-Optics Triggered Multichannel Spark Gap," Plasma Science, IEEE Transactions on , vol.10, no.4, pp.261-265, Dec. 1982

[93] Lee Li; Liu Gang; Zeng Han; Hu Guan; Liu Ning; Cai Li; Lin Fuchang; , "Design, Construction, and Testing of Switches and Trigger Generator for 1.2-MJ Capacitive Pulsed Power Supply Module," Plasma Science, IEEE Transactions on , vol.39, no.1, pp.294-299, Jan. 2011

[94] Sharma, A.; Kumar, S.; Mitra, S.; Sharma, V.; Patel, A.; Roy, A.; Menon, R.; Nagesh, K.V.; Chakravarthy, D.P.; , "Development and Characterization of Repetitive 1-kJ Marx-Generator-Driven Reflex Triode System for High-Power Microwave Generation," Plasma Science, IEEE Transactions on , vol.39, no.5, pp.1262-1267, May 2011

[95] Chen, Y. J.; Neuber, A. A.; Mankowski, J.; Dickens, J. C.; Kristiansen, M.; Gale, R.; , "Design and optimization of a compact, repetitive, high-power microwave system," Review of Scientific Instruments , vol.76, no.10, pp.104703-104703-8, Oct 2005

[96] Jingming Gao; Yonggui Liu; Jinliang Liu; Jianhua Yang; Jiande Zhang; , "Development of a Repetitive Wave Erection Marx Generator," Plasma Science, IEEE Transactions on , vol.37, no.10, pp.1936-1942, Oct. 2009

[97] Ness, R.M.; Smith, B.D.; Chu, E.Y.; Thomas, B.L.; Cooper, J.R.; , "Compact, megavolt, rep-rated Marx generators," Electron Devices, IEEE Transactions on , vol.38, no.4, pp.803-809, Apr 1991

[98] Kusama, H.; Yagi, T.; , "Laser-triggered low timing jitter coaxial Marx generator," Plasma Science, IEEE Transactions on , vol.25, no.6, pp.1431-1434, Dec 1997

[99] Boyer, Craig N.; Holland, Glenn E.; Seely, John F.; , "Portable hard x-ray source for nondestructive testing and medical imaging," Review of Scientific Instruments , vol.69,

89

no.6, pp.2524-2530, Jun 1998 [100] Martin, Benoit; Raymond, Pierre; Wey, Joseph; , "New model for ultracompact coaxial

Marx pulse generator simulations," Review of Scientific Instruments , vol.77, no.4, pp.043505-043505-7, Apr 2006

[101] Cadilhon, B.; Pécastaing, L.; Reess, T.; Silvestre de Ferron, A.; Pignolet, P.; Vauchamp, S.; Andrieu, J.; Lalande, M.; , "High Pulsed Power Sources for Broadband Radiation," Plasma Science, IEEE Transactions on , vol.38, no.10, pp.2593-2603, Oct. 2010

[102] Harjes, H. C.; Schonbach, K. H.; Kristiansen, M.; Guenther, A. H.; Hatfield, L. L.; , "Laser Triggering through Fiber Optics of a Low Jitter Spark Gap," Plasma Science, IEEE Transactions on , vol.8, no.3, pp.170-176, Sept. 1980

[103] Parpal, J.L.; Mercure, H.P.; Mitchel, G.R.; , "A laser-triggered spark gap for 750 kV, 20 ns risetime impulse generator," Electrical Insulation, IEEE Transactions on , vol.24, no.6, pp.1191-1197, Dec 1989

[104] Ohzu, Akira; Ito, Kazunori; , "Characteristics of hard x-ray emission from laser-induced vacuum spark discharges," Journal of Applied Physics , vol.93, no.12, pp.9477-9482, Jun 2003

[105] Morrison, R. W.; Smith, A. M.; , "Overvoltage and Breakdown Patterns of Fast Marx Generators," Nuclear Science, IEEE Transactions on , vol.19, no.4, pp.20-31, Aug. 1972

[106] Yi Liu; Fuchang Lin; Guan Hu; Miao Zhang; , "Design and Performance of a Resistive-Divider System for Measuring Fast HV Impulse," Instrumentation and Measurement, IEEE Transactions on , vol.60, no.3, pp.996-1002, March 2011

[107] T. Heeren, T. Ueno, D. Wang, T. Namihira, S. Katsuki, and H. Akiyama, “Novel Dual Marx Generator for Microplasma Applications”, IEEE Transactions On Plasma Science, Vol. 33, No. 4, August 2005

[108] S. Bae, A. Kwasinski, M. M. Flynn, R. E. Hebner, “High-Power Pulse Generator with Flexible Output Pattern”, IEEE Trans. Power Electron., Vol. 25, pp. 1675-1684, 2010.

[109] D. Wang, J. Qiu, K. Liu, “All-Solid-State Repetitive Pulsed-Power Generator Using IGBT and Magnetic Compression Switches”, IEEE Trans. Plasma Sci., Vol. 38, pp. 2633-2638, 2010.

[110] S. Zabihi, F. Zare, G. Ledwich, A. Ghosh, H. Akiyama, “A new pulsed power supply topology based on positive buck-boost converters concept”, IEEE Trans. Dielectr. Electr. Insul., Vol. 17, pp. 1901-1911, 2010.

[111] S. Zabihi, F. Zare, G. Ledwich, A. Ghosh, H. Akiyama, “A new family of Marx generator based on resonant converter”, 2010 IEEE Energy Conversion Congress and Exposition (ECCE), pp. 3841-3846, 12-16 Sept. 2010.

[112] K. Shimizu, T. Ishii, M. Blajan, “Emission Spectroscopy of Pulsed Power Microplasma for Atmospheric Pollution Control”, IEEE Trans. Ind. Appl., Vol. 46, pp. 1125-1131, 2010.

[113] M. Blajan, A. Umeda, S. Muramatsu, K. Shimizu, “Emission Spectroscopy of Pulsed Powered Microplasma for Surface Treatment of PEN Film”, 2010 IEEE Industry Applications Society Annual Meeting (IAS), pp. 1-8, 3-7 Oct. 2010.

[114] K. Shimizu, M. Blajan, T. Kuwabara, “Removal of Indoor Air Contaminant by Atmospheric Microplasma," 2010 IEEE Industry Applications Society Annual Meeting (IAS), pp. 1-6, 3-7 Oct. 2010.

[115] J. H. Kim, M. H. Ryu, B. D. Min, and G. H. Rim “200KV Pulse Power Supply Implementation”, Power Electronics and Applications, 2007 European Conference on 2-5 Sept. 2007 Page(s):1 - 5

[116] Y. Wu, K. Liu, J. Qiu , X. X. Liu and H. Xiao, “Repetitive and High Voltage Marx Generator Using Solid-state Devices”, IEEE Transactions on Dielectrics and Electrical Insulation Vol. 14, No. 4; August 2007

[117] H. Canacsinha, L. M. Redondoa, and J. F. Silvac, “New Solid-State Marx Topology for Bipolar Repetitive High-Voltage Pulses”, IEEE Power Electronics Specialists Conference, 2008. PESC 2008. 15-19 June 2008 Page(s):791 - 795

[118] M. Sack, C. Schultheiss and H. Bluhm, “Triggered Marx generators for the industrial-scale electroporation of sugar beets”, IEEE Trans. Industry Applications, Vol. 41, pp. 707-714, 2005.

[119] D. J. Thrimawithana, and U. K. Madawala, “A Novel Multi-Level High Voltage Pulsed Power Generator”, 7th International Conference on Power Electronics and Drive Systems, 2007. PEDS '07. 27-30 Nov. 2007 Page(s):445 - 450

[120] H. Heo, S. S. Park, S.C. Kim, J. H. Seo, S. H. Kim, O.R. Choi, S. H. Nam, D.W. Choi, J.

90

Y. Kim, W. S. Lee, J. H. So, and W. Jang, “Compact Marx Generator For Repetitive Applications”, 16th IEEE International Pulsed Power Conference, 2007 Volume 1, 17-22 June 2007 Page(s):874 - 876

[121] W. Ren, H. Wang, R. Liu, “High power variable nanosecond differential pulses generator design for GPR system”, 13th International Conference on Ground Penetrating Radar (GPR), pp. 1-5, 21-25 June 2010.

[122] Y. Xuelin, Z. Hongde. B. Yang, D. Zhenjie, H. Qingsong, Z. Bo, H. Long, “4kV/30kHz short pulse generator based on time-domain power combining”, 2010 IEEE International Conference on Ultra-Wideband (ICUWB), Vol. 2, pp. 1-4, 20-23 Sept. 2010.

[123] J. F. Chen, R. Y. Chen, and T. J. Liang, “Study and implementation of a single-stage current-fed boost PFC converter with ZCS for high voltage applications”, IEEE Transactions On Power Electronics, Vol. 23, No. 1, January 2008

[124] D. D. C. Lu, H. H. C. Iu, and V. Pjevalica, “A single-stage AC/DC converter with high power factor, regulated bus voltage, and output voltage”, IEEE Transactions On Power Electronics, Vol. 23, No. 1, January 2008

[125] Woo-Young Choi; Bong-Hwan Kwon; , "An Efficient Power-Factor Correction Scheme for Plasma Display Panels," Display Technology, Journal of , vol.4, no.1, pp.70-80, March 2008

[126] Sharma, V.K.; Chatterjee, K.; Agarwal, V.; , "Novel Pulse Power Supply Operating at High Input Power Factor," Power Electronics, Drives and Energy Systems, 2006. PEDES '06. International Conference on , vol., no., pp.1-3, 12-15 Dec. 2006

[127] S.V. Araujo , R. P. T. Bascope, G. V. T. Bascope, and L. Menezes, “Step-Up Converter with High Voltage Gain Employing Three-State Switching Cell and Voltage Multiplier”, IEEE Power Electronics Specialists Conference, 2008. PESC 2008. 15-19 June 2008 Page(s):2271 - 2277

[128] A. A. Boora, F. Zare, G. Ledwich, A. Ghosh, “A new DC-DC converter with multi output: Topology and control strategies”, 13th Power Electronics and Motion Control Conference, 2008. EPE-PEMC 2008, 1-3 Sept. 2008 Page(s):468 – 474

[129] C. Chen, M. Baghaeinejad and L. R. Zheng, “Design and Implementation of a High Efficient Power Converter for self-powered UHF RFID Applications,” Proceedings of HDP’06

[130] F. Yuan and N. Soltani, “Design Techniques for Power Harvesting of Passive Wireless Microsensors,” 51st Midwest Symposium on Circuits and Systems, 2008. MWSCAS 2008, 10-13 Aug. 2008 pp. 289 – 293.

[131] D. F. Spencer, R. Aryaeinejad and E. L. Reber, “Using the Cockroft-Walton Voltage Multiplier Design in Handheld Devices,” Nuclear Science Symposium Conference Record, 2001 IEEE vol. 2, 4-10 Nov. 2001 pp. 746 – 749.

[132] L. Malesani and Roberto Piovan, “Theoretical Performance of the Capacitor-Diode Voltage Multiplier Fed by a Current Source,” IEEE Trans. Power Electronics, vol. 8, no. 2, Apr. 1993.

[133] T. Heeren, J. F. Kolb, S. Xiao, K. H. Schoenbach and H. Akiyama, “Pulsed Power Generators and Delivery Devices for Bioelectrical Applications,” Twenty-Seventh International Power Modulator Symposium, 2006. 14-18 May. 2006 pp. 486 – 489.

[134] K. S. Muhammad, A. M. Omar and S. Mekhilef, “Digital control of high DC voltage converter based on Cockcroft Walton voltage multiplier circuit”, IEEE TENCON 2005 Region 10,21-24 Nov. pp:1 - 4

[135] D. F. Spencer, R. Aryaeinejad, E. L. Reber, “Using the Cockroft-Walton voltage multiplier design in handheld devices”, IEEE Nuclear Science Symposium Conference Record, 2001, Volume 2, 4-10 Nov. 2001 Page(s):746 – 749.

[136] G. D. Cataldo, G. Palumbo, “Design of an Nth order Dickson voltage multiplier”, IEEE Transactions on Circuits and Systems, vol. 43, no 5, , pp: 414-418, May 1996

[137] A. Lamantia, P. G. Maranesi, L. Radrizzani, “Small-signal model of the Cockcroft-Walton voltage multiplier”, IEEE Transactions on Power Electronics, vol. 9, no. 1, pp.18-25, Jan. 1994.

[138] D. Zhou, A. Pietkiewicz, and S. Cuk, “A three switch high-voltage converter”, IEEE Transactions on Power Electronics, Volume 14, Issue 1, Jan. 1999 Page(s):177 - 183.

[139] Y. Berkovich, A. Shenkman, B. Axelrod, G. Golan, “Structures of transformerless step-up and step-down controlled rectifiers”, IET Power Electronics, Volume 1, Issue 2, June 2008 Page(s):245 - 254

[140] M. Prudene, L. L. Pfitscher, R. Gules. “A boost converter with voltage multiplier cells”,

91

Proc. of IEEE - PESC, 2005, pp. 2716-2721. [141] B. Emmanuel, J. Gaubert, and P. Pannier, “Conception of UHF voltage multiplier for

RFID circuit”, 2006 IEEE North-East Workshop on Circuits and Systems, June 2006 Page(s):217 - 220

[142] G. De Vita, and G. Iannaccone, “Design Criteria for the RF Section of UHF and Microwave Passive RFID Transponders”, IEEE Transactions On Microwave Theory And Techniques, Vol. 53, No. 9, September 2005

[143] H. Y. Lo, P. Y. Or, and K. N. Leung, “Design Challenges of Voltage Multiplier in a 0.35-am 2-Poly 4-Metal CMOS Technology for RFID Passive Tags”, Electron Devices and Solid-State Circuits, 2007. EDSSC 2007. IEEE Conference on20-22 Dec. 2007 Page(s):433 - 436

[144] G. Gosset, B. Rue, and D. Flandre, “Very High Efficiency 13.56 MHz RFID Input Stage Voltage Multipliers Based On Ultra Low Power MOS Diodes”, 2008 IEEE International Conference on RFID, 16-17 April 2008 Page(s):134 - 140

[145] U. Karthaus, and M. Fischer, “Fully Integrated Passive UHF RFID Transponder IC With 16.7-_W Minimum RF Input Power”, IEEE Journal Of Solid-State Circuits, Vol. 38, No. 10, October 2003

[146] F. Yuan, and N. Soltani, “Design Techniques for Power Harvesting of Passive Wireless Microsensors”, 51st Midwest Symposium on Circuits and Systems, 2008. MWSCAS 2008. 10-13 Aug. 2008 Page(s):289 - 293

[147] K. S. Tam, and E. Bloodworth, “Automated Topological Generation and Analysis of Voltage Multiplier Circuits”, IEEE Transactions on Circuits and Systems, Volume 37, Issue 3, March 1990, Page(s):432 - 436

[148] G. D. Cataldo, and G. Palumbo, “Design of an Nth Order Dickson Voltage Multiplier”, IEEE Transactions On Circuits And Systems-I: Fundamental Theory And Applications, Vol. 43, No. 5, Mar 1996.

[149] M. M. Weiner, “Analysis of Cockcroft-Walton voltage multipliers with an arbitrary number of stages”, Revision of Science Instrument, vol. 40, no. 2, pp:.33C333, 1969.

[150] J. S. Brugler, “Theoretical performance of voltage multiplier circuits’’, IEEE Journal Solid-State Circuits, vol. SC-6, no. 3, pp. 132-135, 1971.

[151] R. C. Cole, “A gated resonant inverter power processor for pulsed loads,” IEEE PESC, 1981, pp. 312-326

[152] F. C. Schwarz, J. B. Klassens, and W. Petiet, “An efficient 600 W high voltage capacitor multiplier”, IEEE PESC, 1980, pp. 316-325.

[153] P. M. Lin, and L. O. Chua, “Topological generation and analysis of voltage multiplier circuits,” IEEE Trans. Circuits Svst.. vol. CAS-24. pp. fi7-530,-0ct. 1977.

[154] Redondo, L.M.; , "A DC Voltage-Multiplier Circuit Working as a High-Voltage Pulse Generator," Plasma Science, IEEE Transactions on , vol.38, no.10, pp.2725-2729, Oct. 2010

[155] J. S. Brugler, “Theoretical Performance of Voltage Multiplier Circuits”, IEEE Journal Of Solid-State Circuit , Volume 6, Issue 3, Jun 1971 Page(s):132 - 135

[156] L. Malesani, and R. Piovan, “Theoretical Performance of the Capacitor-Diode Voltage Multiplier Fed by a Current Source”, IEEE Transactions On Power Electronics, Vol. 8, No. 2, April 1993

[157] F. Hwang, Y. Shen, and S. H. Jayaram, “Low-Ripple Compact High-Voltage DC Power Supply”, IEEE Transactions On Industry Applications, Vol. 42, No. 5, September/October 2006 1139

[158] C. Calligaro, P. Malcovati, and G. Torelli, “Voltage Multiplier With Output Stabilization”, Proceedings of the 38th Midwest Symposium on Circuits and Systems, 1995., Proceedings., Volume 2, 13-16 Aug. 1995 Page(s):905 - 908

[159] L. C. Franco, L. L. Pfitscher, and R. Gules, “A New High Static Gain Non-Isolated DC-DC Converter”, 2003 IEEE 34th Annual Power Electronics Specialist Conference, 2003. PESC '03. Volume 3, 15-19 June 2003 Page(s):1367 - 1372 vol.3

[160] M. Prudente, L. L. Pfitscher, G. Emmendoerfer, E. F. Romaneli, and R. Gules, “Voltage Multiplier Cells Applied to Non-Isolated DC–DC Converters”, IEEE Transactions On Power Electronics, Vol. 23, No. 2, March 2008 871

[161] M. Prudente, L. L. Pfitscher, and R. Gules, “A Boost Converter with Voltage Multiplier Cells”, IEEE 36th Power Electronics Specialists Conference, 2005. PESC '05. Page(s):2716 - 2721

[162] Y. Berkovich, B. Axelrod, and A. Shenkman, “A Novel Diode-Capacitor Voltage

92

Multiplier for Increasing the Voltage of Photovoltaic Cells”, 11th Workshop on Control and Modeling for Power Electronics 2008. COMPEL 2008. 17-20 Aug. 2008 Page(s):1 - 5

[163] M. Rentzsch, H. Guildner, and C. Ditmanson, “Design of an Adjustable High Output Voltage Asymmetrically Switched Class D Converter”, 7th International Conference on Power Electronics and Drive Systems, 2007. PEDS '07. 27-30 Nov. 2007 Page(s):330 - 336

[164] J. C. Rosas-Caro, J. M. Ramirez, and P. M. G. Vite, “A Novel Two Switches Based DC-DC Multilevel Voltage Multiplier”, International Symposium on Power Electronics, Electrical Drives, Automation and Motion, 2008. SPEEDAM 2008. 11-13 June 2008 Page(s):930 - 934

[165] B.P. Divakar, and K.W.E.Cheng, “Voltage Multiplier Integrated HID Ballast and the Application of Triple Insulated Wire in the Design of Igniter Transformer”, 2006 2nd International Conference on Power Electronics Systems and Applications

[166] S. G. Krutchinsky, E. I. Starchnko, A. I. Gavlicky, “Analogous voltage multiplier based on bipolar transistors and MOSFET”, 4th European Conference on Circuits and Systems for Communications, 2008. ECCSC 2008. 10-11 July 2008 Page(s):138 - 141

[167] D. Zhou, A. Pietkiewicz, and S. Cuk, “A Three-Switch High-Voltage Converter”, IEEE Transactions On Power Electronics, Vol. 14, No. 1, January 1999 177

[168] Xiao Bailu; Duan Shanxu; Liu Zhao; Zhu Guorong; Li Xun; , "The sliding mode control for air plasma cutting inverter power source," Electrical Machines and Systems, 2008. ICEMS 2008. International Conference on , vol., no., pp.1806-1809, 17-20 Oct. 2008

[169] Kun-Ming Lee; Chern-Lin Chen; Shin-Tai Lo; , "Resonant pole inverter to drive the data electrodes of AC plasma display panel," Industrial Electronics, IEEE Transactions on , vol.50, no.3, pp. 554- 559, June 2003

[170] Kim, Y.M.; Kim, J.Y.; Jo, M.C.; Lee, S.H.; Mun, S.P.; Lee, H.W.; Kwon, S.K.; Suh, K.Y.; , "Resonance inverter power system for improving plasma sterilization effect," Power Electronics and Motion Control Conference, 2006. IPEMC 2006. CES/IEEE 5th International , vol.3, no., pp.1-5, 14-16 Aug. 2006

[171] Fujita, H.; Akagi, H.; , "A 2-MHz 2-kW voltage-source inverter for low-temperature plasma generators: implementation of fast switching with a third-order resonant circuit," Industry Applications, IEEE Transactions on , vol.35, no.1, pp.21-27, Jan/Feb 1999

[172] Tsai, M.T.; Chu, C.L.; , "Power control strategies evaluation of a series resonant inverter for atmosphere plasma applications," Industrial Electronics, 2009. ISIE 2009. IEEE International Symposium on , vol., no., pp.632-637, 5-8 July 2009

[173] Mohaiminul Islam, S.M.; Kamruzzaman, M.; Sudha, N.-E.-S.; Razzak, M.A.; Takamura, S.; , "RF plasma heating with SIT inverter power source," Plasma Science, 2008. ICOPS 2008. IEEE 35th International Conference on , vol., no., pp.1, 15-19 June 2008

[174] Liangzong He; Shanxu Duan; Baoqi Liu; Xun Li; Guorong Zhu; Xuexiao Luo; , "The IGBT failure mechanism research of DC plasma arc cutting inverter power when arc-starting," Power Electronics and Motion Control Conference, 2009. IPEMC '09. IEEE 6th International , vol., no., pp.1225-1229, 17-20 May 2009

[175] Uesugi, Y.; Imai, T.; Kawada, K.; Takamura, S.; , "Fundamental and third harmonic operation of SIT inverter and its application to RF thermal plasma generation," Power Conversion Conference, 2002. PCC Osaka 2002. Proceedings of the , vol.3, no., pp.1473-1478 vol.3, 2002

[176] Jiang Xiaomei; Rui Yannian; Chen Guoqiang; Liu Kaiqiang; , "Restraining DC Magnetic Bias in PS-PWM Full Bridge Inverter of Plasma Power," Intelligent System Design and Engineering Application (ISDEA), 2010 International Conference on , vol.2, no., pp.306-309, 13-14 Oct. 2010

[177] Kalisiak, S.; Holub, M.; Jakubowski, T.; , "Resonant inverter with output voltage pulse-phase-shift control for DBD plasma reactor supply," Power Electronics and Applications, 2009. EPE '09. 13th European Conference on , vol., no., pp.1-9, 8-10 Sept. 2009

[178] Young-Mun Kim; Ju-Yong Kim; Sang-Pil Mun; Hyun-Woo Lee; Soon-Kurl Kwon; Ki-Young Suh; , "The design of inverter power system for plasma generator," Electrical Machines and Systems, 2005. ICEMS 2005. Proceedings of the Eighth International Conference on , vol.2, no., pp.1309-1312 Vol. 2, 29-29 Sept. 2005

[179] Sun Qiang; Chen GuiTao; Zhao Chen; Yuan Bo; , "A Novel Plasma Cutting Power Supply," Power and Energy Engineering Conference (APPEEC), 2010 Asia-Pacific ,

93

vol., no., pp.1-4, 28-31 March 2010 [180] Koudriavtsev, O.; Shengpei Wang; Konishi, Y.; Nakaoka, M.; , "A novel pulse-density-

modulated high-frequency inverter for silent-discharge-type ozonizer," Industry Applications, IEEE Transactions on , vol.38, no.2, pp.369-378, Mar/Apr 2002

[181] J. Ferreira, A. Ross, and J. Wyk, “A hybrid phase arm power module with nonlinear resonant tank,” in Conf. Rec. IEEE-IAS Annu. Meeting, 1990, pp. 1679–1685.

[182] S. P.Wang, M. Nakaoka, and Y. Uchihori, “New generation electromagnetic induction-heated boiler & evaporator systems using power factor corrected series load resonant PWM IGBT inverter,” in Proc. PCIM Europe, 1996, pp. 719–726.

[183] H. Fujita and H. Akagi, “A 2-MHz 2-kW voltage-source inverter for low-temperature plasma generators: Implementation of fast switching with a third-order resonant circuit,” IEEE Trans. Ind. Applicat., vol. 35, pp. 21–27, Jan./Feb. 1999.

[184] T. A. Meynard and H. Foch, “Multilevel Conversion: High Voltage Choppers and Voltage–Source Inverter”, 23rd Annual IEEE Power Electronic Specialists Conference, Vol 1, pp.397-403, 1992.

[185] J. S. Lai and F. Z. Peng. “Multilevel Converters- A New Breed of Power Converters”, IEEE Transactions on Industry Applications, Vol. 32, No.3, pp.509-517, 1996.

[186] L. Li, D. Czarkowski, Y. Liu and Pillay-P, “Multilevel selective harmonic elimination PWM technique in series-connected voltage inverters”, IEEE Industry Applications Conference, Thirty-Third IAS Annual Meeting, New York, NY, USA; 1998; pp. p.1454-61 vol.2.

[187] Agelidis-VG, “A low-distortion multilevel pulse-width modulated inverter topology”, Proceedings of the IEEE International Symposium on Industrial Electronics, IEEE, New York, NY, USA, 1995, pp.243-7 vol.1.

[188] S. Masuda and T. Ling, “Destruction of Gaseous Pollutants by Surface Induced Plasma Chemical Process”, IEEE Transactions on Industry Applications, vol 29, No. 4, July 1993

[189] R. G. Rice, Ozone in the United States of America ‘state-of-the-art’, in Proc. 13th Ozone World Congress., vol. 2, 1997, pp. 1029–1040.

[190] O. Koudriavtsev, S. Wang and Y. Konishi, “A Novel Pulse-Density-Modulated High-Frequency Inverter for Silent-Discharge-Type Systems”, IEEE Transaction on Industry Application, vol. 38, No. 2, March 2002

[191] H. Ohkawa, T. Akitsu, M. Tsuji, H. Kimura, M. Kogoma and K. Fukushima, “Pulse-modulated, high-frequency plasma sterilization at atmospheric-pressure”, Elsevier, Surface & Coatings Technology 2006, 5829-5835

[192] S. Masuda and K. Akutsu, “A Ceramic-Based Ozoniser Using High-Frequency Discharge”, IEEE Transactions on Industry Applications, vol. 24, No. 2, March 1988

[193] T. Shimizu, H. Kinjyo and K. Wada, “A Novel High-frequency Current Output Inverter based on an Admittance Conversion Element and a Hybrid MOSFET-SIC Diode”, IEEE Transactions on Industrial Electronics, 2003, 2003-2008

[194] Ke Jin; Xinbo Ruan, “Hybrid Full-Bridge Three-Level LLC Resonant Converter; A Novel DC-DC Converter Suitable for Fuel-Cell Power System”, IEEE Transactions on Industrial Electronics, Volume 53, Issue 5, Oct. 2006 Page(s):1492 - 1503

[195] Huber, L.; Hsu, K.; Jovanovic, M.M.; Solley, D.J.; Gurov, G.; Porter, R.M., “1.8-MHz, 48-V Resonant VRM: Analysis, Design, and Performance Evaluation”, IEEE Transactions on Power Electronics, Volume 21, Issue 1, Jan. 2006 Page(s):79 - 88

[196] S.D.Roos, J.A. Ferreira and W.G Odendaal, “A plasma torch converter based on the partial series resonant converter” 28th Annual IEEE Power Electronics Specialists Conference, 1997. PESC '97 Record, Volume 2, 22-27 June 1997 Page(s):1388 - 1394 vol.2

[197] Pacheco-Sotelo, J.; Pena-Eguiluz, R.; Eguiluz, L.P.; de los Rios, A.S.; Sanchez, G.C.;“Plasma torch ignition by a half bridge resonant converter”, IEEE Transactions on Plasma Science, Volume 27, Issue 4, Aug. 1999 Page(s):1124 - 1130

[198] Facta, M.; Salam, Z.; Buntat, Z.; Yuniarto, A.; , "Silent discharge ozonizer for colour removal of treated palm oil mill effluent using a simple high frequency resonant power converter," Power and Energy (pecon), 2010 IEEE International Conference on , vol., no., pp.39-44, Nov. 29 2010-Dec. 1 2010

[199] Ray-Lee Lin; Min-Han Lee; , "Analysis and design of full-bridge LC parallel resonant plasma driver with variable-inductor based phase control," IECON 2010 - 36th Annual Conference on IEEE Industrial Electronics Society , vol., no., pp.77-82, 7-10 Nov. 2010

94

[200] Branas, C.; Azcondo, F.J.; Bracho, S.; , "Contributions to the design and control of lcscp

resonant inverters to drive high-power HPS lamps," Industrial Electronics, IEEE Transactions on , vol.47, no.4, pp.796-808, Aug 2000

[201] Pacheco-Sotelo, J.O.; Barrientos, R.V.; Pacheco-Pacheco, M.; Flores, J.F.R.; Garcia, M.A.D.; Benitez-Read, J.S.; Pena-Eguiluz, R.; Lopez-Callejas, R.; , "A universal resonant converter for equilibrium and nonequilibrium plasma discharges," Plasma Science, IEEE Transactions on , vol.32, no.5, pp. 2105- 2112, Oct. 2004

[202] Sihun Yang; Seiya Abe; Shoyama, M.; , "Design consideration of reduction flat transformer temperature in LLC resonant converter for plasma display," Power Electronics Electrical Drives Automation and Motion (SPEEDAM), 2010 International Symposium on , vol., no., pp.1141-1145, 14-16 June 2010

[203] Chong-Eun Kim; Ki-Bum Park; Gun-Woo Moon; Jun-Young Lee; , "New Multi-Output LLC Resonant Converter for High Efficiency and Low Cost PDP Power Module," Power Electronics Specialists Conference, 2006. PESC '06. 37th IEEE , vol., no., pp.1-7, 18-22 June 2006

[204] Dubut, J.P.; Amado, J.D.; Salazar, A.O.; Junior, L.P.; de Aguilar, M.V.; , "Experimental analysis of a series resonant converter for a plasma inertization plant," IECON 2010 - 36th Annual Conference on IEEE Industrial Electronics Society , vol., no., pp.464-468, 7-10 Nov. 2010

[205] Woo-Jin Lee; Seong-Wook Choi; Chong-Eun Kim; Gun-Woo Moon; , "A New PWM-Controlled Quasi-Resonant Converter for a High Efficiency PDP Sustaining Power Module," Power Electronics, IEEE Transactions on , vol.23, no.4, pp.1782-1790, July 2008

[206] Ferreira, J.A.; Roux, J.A.; , "A series resonant converter for arc-striking applications," Industrial Electronics, IEEE Transactions on , vol.45, no.4, pp.585-592, Aug 1998

[207] Alonso, J.M.; Ordiz, C.; Dalla Costa, M.A.; Ribas, J.; Cardesin, J.; , "High-Voltage Power Supply for Ozone Generation Based on Piezoelectric Transformer," Industry Applications, IEEE Transactions on , vol.45, no.4, pp.1513-1523, July-aug. 2009

[208] S. Zabihi; F. Zare; H. Akiyama; “Using a current source to improve efficiency of a plasma system”, Pulsed Power Conference, 2009. PPC '09. IEEE , pp.1256-1260, June 28 2009-July 2 2009

[209] S. Zabihi; F. Zare; G. Ledwich; A. Ghosh; “A novel high voltage pulsed power supply based on low voltage switch-capacitor units”, Pulsed Power Conference, 2009 IET European , pp.1-4, 21-25 Sept. 2009

[210] S. Zabihi, F. Zare, G. Ledwich, A. Ghosh, H. Akiyama, “A Novel High-Voltage Pulsed-Power Supply Based on Low-Voltage Switch–Capacitor Units”, IEEE Trans. Plasma Sci., Vol. 38, pp. 2877-2887, 2010.

[211] S. Zabihi, F. Zare, G. Ledwich, A. Ghosh, H. Akiyama, “A new generation of high voltage pulsed power converters," Universities Power Engineering Conference (AUPEC), 2010 20th Australasian , pp.1-5, 5-8 Dec. 2010

[212] S. Zabihi, F. Zare, G. Ledwich, A. Ghosh, “A bidirectional two-leg resonant converter for high voltage pulsed power applications”, 2009 IET European Pulsed Power Conference, pp. 1-4, 21-25 Sept. 2009.

[213] S. Zabihi, F. Zare, G. Ledwich, A. Ghosh, “A Resonant Based Marx Generator”, 20th Australasian Universities Power Engineering Conference, (AUPEC) 2010, pp.1-5, 5-8 Dec. 2010.

[214] S. Zabihi, Z. Zabihi, F. Zare, “A Solid-State Marx Generator with a Novel Configuration”, 19th Iranian Conference on Electrical Engineering, 17-19 May 2011, Tehran, Iran.

[215] S. Zabihi, Z. Zabihi, F. Zare, “A Solid-State Marx Generator with a Novel Configuration (Periodical style—Accepted for publication)”, IEEE Trans. Plasma Sci., Vol.39, No.8, pp.1721-1728, Aug. 2011.

[216] S. Zabihi, F. Zare, G. Ledwich, A. Ghosh, H. Akiyama, “A new family of Marx generators based on commutation circuits”, IEEE Trans. Dielectr. Electr. Insul., Vol. 18, Issue 4, pp. 1181-1188, August 2011.

[217] S. Zabihi; F. Zare; H. Akiyama; “A high voltage power converter with a frequency and voltage controller”, Pulsed Power Conference, 2009. PPC '09. IEEE , pp.1250-1255, June 28 2009-July 2 2009

96

I

Statement of Contribution of Co-Authors

The authors listed below have certified that:

1. they meet the criteria for authorship in that they have participated in the conception,

execution, or interpretation, of at least that part of the publication in their field of expertise;

2. they take public responsibility for their part of the publication, except for the responsible

author who accepts overall responsibility for the publication;

3. there are no other authors of the publication according to these criteria;

4. potential conflicts of interest have been disclosed to (a) granting bodies, (b) the editor or

publisher of journals or other publications, and (c) the head of the responsible academic unit,

5. they agree to the use of the publication in the student's thesis and its publication on the

Australasian Digital Thesis database consistent with any limitations set by publisher

requirements.

In the case of this chapter:

A Novel High-Voltage Pulsed-Power Supply Based on Low-voltage Switch-Capacitor Units

Published in: IEEE Transactions on Plasma Science, Vol. 38, No. 10, pp. 2877-2887, Oct.

2010

Contributor Statement of contribution Sasan Zabihi ) Proposed the initial design and conducted simulation studies and data

VJ~yi/ analysis, designed the control strategy, implemented hardware set-up and conducted experimental verifications and wrote the manuscript.

11 Au'g. 2011

Proposed the initial design and supervised the validity studies Firuz Zare including: conducting the simulations and experimental studies and

writing the manuscript. Gerard Ledwich Aided experimental design, and data analysis.

Arindam Ghosh Aided planning the control strategies and writing the paper.

Hidenori Akiyama Provided us with general information about pulsed power supply specifications and its application demands.

Principal Supervisor Confirmation

I have sighted email or other correspondence from all Co-authors confirming their certifying

authorship.

A/Pro f. Firuz Zare 11 Aug. 201 1

Name Date

97

CHAPTER 2

A Novel High-Voltage Pulsed-Power Supply

Based on Low-vol tage Switch-Capaci tor Units

Sasan Zabihi*, Firuz Zare*, Gerard Ledwich*, Arindam Ghosh*, Hidenori

Akiyama†

*School of Electrical Engineering, Queensland University of Technology, GPO

Box 2434, Brisbane, Australia

† Kumamoto University, Japan

Published in: IEEE Transactions on Plasma Science, Vol. 38, No. 10, pp. 2877-

2887, Oct. 2010

98

Abstract— This paper presents a high voltage pulsed power system based on

low voltage switch-capacitor units connected to a current source for several

applications such as plasma systems. A modified positive buck-boost converter

topology is used to utilize the current source concept and a series of low voltage

switch-capacitor units is connected to the current source in order to provide high

voltage with high voltage stress (dv/dt) as demanded by loads. This pulsed power

converter is flexible in terms of energy control, in that the stored energy in the

current source can be adjusted by changing the current magnitude to significantly

improve the efficiency of various systems with different requirements. Output

voltage magnitude and stress (dv/dt) can be controlled by a proper selection of

components and control algorithm to turn on and off switching devices.

2.1. Index Terms

Current source, High voltage stress, Plasma, Pulsed power supply, Switch-

Capacitor units

2.2. Int roduct ion

Steady accumulation of energy followed by its rapid release can result in the

delivery of a larger amount of instantaneous power over a shorter period of time

(although the total energy is the same). The energy is delivered in the form of

high voltage short duration pulses which are called pulsed power. Voltage

magnitude, pulse rising time duration, repetition and energy are significant

specifications of these pulses which are defined based on applications

requirements. Pulsed power converters became widespread industrially with

increasing demands in applications such as ozonising, sterilizing, recycling,

exploding, winery, medical and military applications [1, 2]. Plasma systems are

currently the most substantial application of pulsed power technology [3].

However, there are still specific issues which hinder the wide scale application of

these systems. The main issue is power efficiency which can affect long term

usage of pulsed power supplies in industry.

Conventionally, Marx Generators (MG) [4-6], Magnetic Pulse Compressors

(MPC) [7-9], Pulse Forming Network (PFN) [10, 11], Multistage Blumlein Lines

(MBL) [12, 13] etc, are employed to supply pulsed power systems. They are

mostly classified in voltage source topology category and suffering from major

99

drawbacks such as being inflexible in terms of controlling output voltage levels

and stresses and power delivery to the load. Circuit diagram of a Marx generator

is depicted in Fig. 2.1. Since plasma systems are naturally known as capacitive

loads for power supply equipment, current source topologies are suitable

candidates in terms of flexibility to supply these sorts of applications and

improve efficiency. With respect to this issue, a dc-dc converter based on the

buck-boost converters concept is designed to feed these loads. This topology

aims to generate high voltage with a series of low-medium voltage switches. The

novel idea in this proposal is employing a series of switch-capacitor units in

order to provide high dc voltage with high voltage stress, dv/dt, considering

plasma loads requirements. The modified version of this converter can generate

high dc voltage levels in a few nanoseconds.

Fig.2. 1. Marx generator

2.2. Conf igura t ion and analyses

2.2.1. Topology

The proposed circuit diagram includes an ac-dc rectifier connected to a modified

positive buck-boost converter as shown in Fig. 2.2. An inductor connected to the

dc source through a switch SS acts as a current source. A controller is used to

control the current through the inductor which adjusts the energy required by a

load. A flywheel diode is used to provide a current loop for the inductor when

the switch SS is turned off. A series of switch-capacitor units connected in

cascade to the current source can generate high voltage pulse with significant

dv/dt.

Plasma applications are known as nonlinear resistive-capacitive loads. To

simulate plasma behavior for the pulsed power supply, a simple resistive-

capacitive model with a switch has been chosen to show a high and low

resistivity of load, R1Load & R2Load, in different physical situations. R1Load, in Mega

100

Ohms range represents the load resistivity before a resumption of plasma

reaction in order to model the load’s leakage current. R2Load in the range of a few

ohms represents the load resistivity in the period of plasma reaction. This load

model is highlighted in Fig. 2.2. In a real condition, the resistivity of the load

substantially drops and based on the proposed model, the load current is supplied

by the voltage and current sources.

Switching Pulses

SS

S1

SL

D

R2Load

C1

-

+

+

-

L

S2

C2

+

-

Sn

Cn

+

-

R1LoadControl

Protocol

of

Series

Switches

Plasma Load Model


CLoad


RD1

RD2

AC-DC

ConverterVac

220 V50 Hz

V inCin



Load

Fig.2. 2. Plasma power supply configuration with multi switch-capacitor units

To analyze the pulsed power supply converter, we have considered two switch-

capacitor units as shown in Fig. 2.3 and the analysis and simulations can be

extended for n switch-capacitor units.

Fig.2. 3. A simplified two switch-capacitor unit plasma power supply and the load model

The general concept of this circuit is based on delivering the stored energy in the

inductive and capacitive components to a load. To satisfy this condition, the

inductor current should be pumped into the capacitor bank to charge the

capacitors and create high voltage and high dv/dt across the load.

101

2.2.2. Switching modes

The operation modes of this topology are classified into two separate parts. Load

supplying part and discharging the capacitors which initializes the energy storage

components for the next supplying cycle. The load supplying part which is

consisting three operation modes are shown in Fig. 2.4.

2.2.2.1. First mode (SS: on, S1: on, S2: on)

Fig. 2.4(a) indicates inductor’s charging mode while in other three modes

inductor is being discharged. Based on (3-1) & (3-2), the input voltage locates

across the inductor in this mode and charges it.

)(21 SSSinL VVVVV

S++++++++−−−−==== (2-1)

)()( tiLdtdiLVL ∆∆== (2-2) If the inductor is supposed to be with no initial current charge, ∆i=I , then the

time interval to charge the inductor to a certain level will be ∆t=(L.Imax)/VL.

2.2.2.2. Second mode (SS: off, S1: on, S2: on)

In Fig. 2.4(b) the stored current circulates through a diode D while Ss is turned

off. In this mode the inductor current reduces gently due to the low negative

voltage across the inductor generated by the voltage drop across the diode and

switches. The inductor current can be supposed constant due to the

insignificancy of this voltage. The charged inductor acts as a current source for

the rest of topology.

)(21 SSDL VVVV ++++++++−−−−==== (2-3)

These two switching sates illustrate how the current source receives energy and

keep it ready to deliver to the voltage source and the load subsequently. There

are arbitrary numbers of switch-capacitor units connected in series together and

the whole unit, in parallel with the current source of the system which charges

the capacitors. The number of these units is determined by the required output

voltage. Take it into the account that when the inductor is being charged through

SS all those switches should be closed otherwise there may be an undesired

resonant between the inductor and the capacitors. As soon as the inductor current

reaches a defined current level, indicating the inductor is fully charged; SS is

switched off and the inductor current flows through the diode D as shown in Fig.

2.4(b). The voltage drop across the diode and the switches creates a small

negative voltage across the inductor which slightly discharges the inductor.

102

2.2.2.3. Third mode (SS: off, S1: off, S2: off)

In the next switching mode, that is called capacitors charging mode and is shown

in Fig. 2.4(c)&(d), the switches, S1, S2, are turned off simultaneously, to allow

the inductor current to flow through the capacitors and charge them. This can

create a high voltage across the capacitors particularly while the capacitors are

selected in nF ranges. Assuming the current, IC is constant, then:

tCIVCItV iCCiCC iiii∆⋅=∆⇒=∆∆ )()()( (2-4)

Since the capacitors are identical and the same current flows through all

capacitors, there will be a similar voltage across each capacitor. Therefore, the

summation of these voltages appears at the output of power supply.

iCout VnV ⋅= (2-5)

iCout VnV ∆⋅=∆ (2-6)

Where n is the number of switch-capacitor units which can be extended to satisfy

load demands. Having more units reduces the equivalent capacitance of capacitor

bank in the output of the topology and with a fixed injected current there will be

a higher voltage level and stress in the output of power supply.

nCC ieq = (2-7) Based on switching characteristics of the power switches in terms of internal

resistivity of conducting and non-conducting modes, both sides of each capacitor

are connected to the sides of each switch in order to provide voltage sharing over

the switches.

Plasma reaction is resumed with respect to the stimulation of a high voltage over

related material. But the key point is that, this high voltage should be induced

with an extremely high voltage stress, dv/dt. Pumping stored current into the

series of capacitors which have considerably low capacitances can generate

significant high voltage magnitude and stress to fulfill the plasma creation

requirements.

103

Fig.2. 4. Switching states of the proposed power supply circuit (a) Current source, charging mode (b) Current source, discharging mode (c) Voltage source charging mode (d) Load

supplying mode

2.2.3. Discharging Residual Energy

It should be noted that based on this switching algorithm, all capacitors should

get fully discharged to avoid probable short circuit through the switches in the

switch-capacitor units. In this control algorithm, series switches should be turned

off when a pulse voltage should be generated and turned on when the capacitors

are fully discharged. However there is still a major concern exists. Even when

capacitors are almost discharged the flowing current to the load creates a voltage

over capacitors which may cause a short circuit at the time of closing unit

switches. There are several ways to either prevent or damp this phenomenon

such as putting either reverse blocking components, extra inductive or other

damping elements in the short circuit loop.

Since there is a possibility of not delivering the whole capacitor’s energy to the

load, and unit switches closing for current recovery, there should be an

appropriate number of damper components like resistors or inductors, located in

the switch-capacitor units to prevent the probable inrush current which is due to

making charged capacitors short circuit by closing units switches.

Any inductive elements in the switch-capacitor loop exchanges energy with the

capacitor and leads to oscillations which cannot be prohibited by the diodes. So

they are not suitable options for such a purpose in this topology.

Resistor is another alternative, which can be installed in common paths of units

and damp the remained energy of inductor and capacitors. As shown in Fig. 2.5,

although RD1 can prevent short circuits of two units, there is still another loop

consisted of C1, C2, S1 and S2 which can be shorten circuit. Putting a resistor in

this path causes loss in charging the capacitors and supplying the load periods.

There are couple of ways, hard and soft methods, to discharge the capacitors

with just common path resistors.

104

Fig.2. 5. Possible current loops during short circuit periods.

2.2.3. 1. Hard methods

In the hard method, a parallel switch-resistor unit will be located in the return

path as shown in Fig. 2.6(a). This switch will be on unless the period of

discharging remained energy which will be off. So the resistor will be in the

flowing path in this period and keep the topology away from short circuit.

Thermistors can also be employed in this regard. A thermistor is a type of

resistor whose resistance varies with temperature. They can be installed in series

with switches and protects the circuit in the stint of inrush current. As shown in

Fig. 2.6(b) the resistivity of circuit will be increased in discharging mode.

These two solutions impose an initial cost to the whole topology but, the benefit

they can bring for this power supply is the possibility to stop supplying load in

any stage. Therefore the remained energy in inductor after each supplying mode

can be saved in each supplying cycle.

RD1

S1

SL

D

R2Load

C1

-

+

+

-

L

S2

C2

+

-

R1Load

SS

Vin

-

+

CLoad

(a)R

S

Fig.2. 6. Two examples of hard methods for discharging residual energy in the inductor and

capacitors, (a) Parallel switch-resistor unit located in the return path (b) A thermistor in the return path

2.2.3.2. Soft method, fourth and fifth switching modes (SS: off, S1: on, S2:

off) & (SS: off, S1: off, S2: on)

This should be considered to keep the power loss at the minimum possible level.

In soft method, which is used in this investigation, no extra element is installed

in the return path. Instead, the remained energy is discharged with extra

switching states. As can be inferred from hard methods, the supplying path

105

should be free of any energy consuming elements during load supplying period

otherwise the loss rate will increase substantially. Therefore, (n-1) is the number

of resistors used in this topology. The point is that since there is no resistor in the

supplying path, the switches, S1, S2, cannot be turned on simultaneously,

otherwise short circuit is unavoidable. The unit’s switches are turned on and off

in a manner through that the remained voltage across output capacitors and

plasma reactor will be discharged. Following this concept, a couple of switching

states will be added to the former statuses after plasma creation, and be repeated

alternatively to fully discharge remained voltage and current. These states

composing discharging remained energy part are shown in Fig. 2.7.

Based on designed control algorithm, the inductor and C1 are fully discharged

during first state while S1 is on and S2 is off. In this switching mode the inductor

energy is delivered to C2 and charges it. In the next state, while S1 and S2, turned

off and on respectively, C2 is discharged through RD1. But there is still a

considerable voltage across reactor capacitor, CLoad, which is going to be

discharged through RD1 during this mode. Although C1 is almost ten times CLoad,

this voltage can charge C1 to a potentially hazardous level and it needs to be

discharged again. So, the switching statues changes between these two modes

until capacitors voltage become less than a safe value.

Fig.2. 7. Circuit’s switching states in association with soft method in order to discharge the

remained energy in the capacitors

In case of no prosperous plasma generation, a general step should be considered

to discharge the capacitors before starting a new pulse process. In these

circumstances, the output voltage will increase automatically and may cause a

real trouble. Even if the capacitors can tolerate this high voltage, they need to be

discharged and ready for the next cycle. The mentioned solution can be applied

for such a problem as well. By taking this concern in to the account, the

discharging resistors RD should be considered as much as feasible to handle such

a high voltage and prohibit high current flow in discharging period.

106

2.2.4. Analyses of load supplying mode

In this mode the equivalent circuit of system would be a parallel RLC circuit

shown in Fig. 2.8. To realize how the stored energy in power storage elements of

the circuit will be delivered to the load, the instantaneous inductor current IL(t)

and capacitor voltage VC(t) in this mode are the most prominent indicators of

power delivery which can be achieved as follow:

Fig.2. 8. Equivalent RLC circuit of power delivery mode of power supply

The inductor current of this mode in the Laplace S- domain is

1

000

2 )S+R

L-((LC) S

)) (IR

L)- () )S+(CV ((LCI

(s)=ILCL

L (2-8)

And based on the considered range for inductor, capacitor and plasma resistance,

it can be assumed that in any condition:

RCL 4> (2-9) So the time response of inductor current IL(t) can be achieved as:

])()[()](1[)( 21

21211212tt

L ekkekktI αα αααα −− −−−⋅−= (2-10) While:

)0(1 LLCIk = (2-11) )0()()0(2 LC IRLCVk −= (2-12)

)(2])(4)()([ 22,1 LCLC

R

L

R

L −±−=α (2-13)

Since ])([)( dttdILtV LC ⋅−= (2-14)

])()([)(

1)( )1(

2122)1(

112112

21 ttC ekkekktV −−−− −−−

−= αα αααα

αα (2-15)

Since the capacitors are opted smaller in comparison with the inductor, they

cannot store that much energy and will be discharged considerably faster than the

inductor. Besides above equations, (2-10) & (2-15), simulation results presented

in Fig. 2.12 (a) & (d) can also demonstrate evidences to verify this statement. In

this mode, there will be a high current for a short period of time which

discharges the capacitors. After that if there were any demand for more current,

the inductor would supply energy to the load until fully get discharged otherwise

the supplying process would be stopped by closing series units’ switches

107

alternatively and power supply can get prepared for the next power delivery

cycle. The simulation results exhibited in the Fig. 2.12 confirm this expression.

Several simulations and analyses under different load conditions have been

carried out to verify the validity of proposed topology and control strategy. As

the simulation results will demonstrate in the next parts, this power supply has

the capability of generating pulsed power in an extensive range of amplitude and

dv/dt.

2.3. Contro l s t rategy

Power switches utilized in this configuration have three distinguished functions.

With respect to these functions, appropriate control algorithms are designed to

turn on and off the switches properly and satisfy the desired duties thoroughly.

2.3.1. Current control

To control the inductor current, a current control block determines a duty cycle

for the switch SS located between the dc source and the inductor in order to

charge the inductor at a specific current level. In this strategy, the desired amount

of the inductor current is selected as a limit to turn off SS. The switching state in

Fig. 2.4(a) shows the inductor charging mode. Based on series switches

condition, the energy can be either stored in the inductor or delivered to the load.

On the other hand, while the inductor delivered energy to the load and became

discharged, the control block switches on SS, in order to charge the inductor.

There are also a couple of conditions should be met before turning SS on. The

capacitors voltage should be under a specific level to let the SS turned on. The

control strategy in this block can be found as follows.

If upperinductor II ≥ then SS=0

If upperinductor II ≤ & 2min21 ,, VVVV CLoadCC ≤ then SS=1

Where IInductor, IUpper and ILower are the inductor actual current and desired

inductor’s charging and discharging amounts, respectively. VC1, VC2, & VCLoad are

capacitors voltage and Vmin2 is safe level of voltage for short circuit. The block

can be implemented by a logic device or a simple program in a microcontroller.

108

2.3.2. Voltage control

Series switches, S1, S2, are also controlled with respect to the inductor current,

capacitor’s voltage and the load condition. The switches can be off while the

inductor is fully charged and SS has already become off. Supplying repetition rate

is another parameter which can define switching moments. During off state, the

current flows through the capacitors and charges them. Then at a specific voltage

level plasma occurs and discharges inductor and capacitors but not thoroughly.

The remained energy should be discharged before turning on the unit’s switches

in order to refrain from short circuit and inrush current. On the other hand if

plasma does not take place to start another cycle we need to discharge the

capacitors. Based on alternative switching of S1, S2 but with different logics, two

specific control programs are assigned for each of these cases. The logics are

shown in Fig. 2.9. S1, S2 are turned on and off alternatively to discharge

inductive and capacitive elements. For example the remained energy after each

supplying cycle is discharged according to following process:

In the switching mode shown in Fig. 2.7(a), while S1 is on and S2 is off, C1 and L

are discharged with different time constant. In the next switching state, S1 is

turned off and S2 is turned on simultaneously in order to discharge C2 and CLoad.

This procedure continues and the circuit change between these two modes for a

few times to ensure that the output voltage is under a specific voltage level. The

switches become on together while the inductor and capacitors are fully

discharged. The flowchart shown in Fig. 2.9 demonstrates the control algorithm

designed for this topology.

As can be seen in control flowchart given in Fig. 2.9, some voltage and current

levels are defined as conditions for switching times determination in order to

control the circuit performance with maximum level of safety. With this regard,

Vmax1&2 and Vmin1&2 are selected for 1st and 2nd upper and lower safety level of

voltages, respectively. Fig. 2.10 provides an overview of these values in a

general configuration. Vmax1 indicates a margin above the voltage level in which

plasma phenomenon occurs. Vmax2 presents a summation of voltages can be

tolerated by all the switches. This amount is assigned based on the switches

characteristics and rising voltage above that level can be critical for the switches.

Based on the number of switch-capacitor units in the topology, n, each power

109

switch should handle Vmax2/n volts. Vmin1 is chosen with respect to the level of

voltage in which the load supplying process can be stopped and Vmin2 is a safe

level of voltage for capacitors in situation of short circuit. Imax gives the inductor

charging amount and Imin is a safe level of current which cannot charge the

capacitors more than Vmin2 level. As shown in start block of this flowchart

Vmax1&2, Vmin1&2 and Imax&min values are designated 3500, 4000, 100, 5V and 20 &

0.1 A, respectively for a two unit topology.

2.3.3. Load control

In order to model supplying mode, SL is switched on as soon as the capacitor

charged up to a required voltage level. This voltage is defined by the load and

may change for different applications. Load resistivity drops dramatically to few

ohms by turning SL on. This low resistivity discharges capacitors and inductor

with high and low time constant. Turning SL off, lets the system to be prepared

for next cycle of load supplying. The control strategy adopted to generate gate

signals for this switch is almost similar to the current control, despite of

measuring and comparing output voltage instead of inductor current. The

switching signals pattern of the circuit is shown in Fig. 2.11.

110

Fig.2. 9. Block diagram of control algorithm

Vmax2

Vmax1

Vmin2

Vmin1

0

Voltage

Time Time

Imax

Imin

Current

0

Fig.2. 10. Definition of voltage and current levels for the control strategy

111

Fig.2. 11. Switching signals pattern

2.4. Simulat ion resul ts

According to the (3-4), it can be deduced that a higher current can generate a

higher dv/dt across the capacitors. Fig. 2.12 illustrates the results extracted from

the simulation of the circuit operating in the mode shown in Fig. 2.4(c). In this

figure, the voltage and current stresses of circuit with two different inductors (L1

& L2) and current levels (IL1 & IL2) are depicted. To have a same stored energy in

the both inductors, for 12 LkL ⋅= , the inductor current should be adjusted as:

112

12 )1( LL IkI ⋅= (2-16) Since,

222

211 )21()21( LL ILIL ⋅=⋅ (2-17)

In this example L1 and L2 are 1 & 9 mH, respectively and consequently IL2=3 IL1

in order to have the same energy stored in the inductors, L1 and L2. As it can be

seen in Fig. 2.12(a) & (c), the inductors currents are controlled at 45 A and 15 A

for IL1 and IL2, respectively. Ultimately, the voltage level of the capacitor is based

on the stored energy in the inductor.

0 0.5 1 1.50

20

40

(a)

Ind

uct

or

curr

ent(

A)

0 0.5 1 1.50

20

40

(c)

Ind

uct

or

curr

ent(

A)

0.4460.448 0.45 0.4520.4540.4560.458 0.460

10

20

(b)

Ou

tpu

t vo

ltag

e(kV

)

1.33 1.3321.3341.3361.338 1.34 1.3421.3441.3460

10

20

(d)Time(ms)

Ou

tpu

t vo

ltag

e(kV

)

Fig.2. 12. dv/dt s generated by different inductors with the similar inductive energy, (a) current of

1mH inductor (b) output voltage of 1mH inductor, (c) current of 9mH inductor, (d) output voltage of 9mH inductor.

Leqout ICLV ⋅= )( (2-18)

There are same stored energies in the inductors for both cases. Therefore, the

final values of the voltages are similar and reach the level of 18kV. However,

dv/dts vary regarding the inductors current levels. In the first case, the inductor

current is set to 45 A which creates a 4.5 V/ns voltage stress across the

capacitors while in the second case, the inductor current (15A) causes the output

voltage to rise with 1.5V/ns slope. To show how this power supply circuit works,

and verify the validity and accuracy of foreseen circuit analyses which comes

latter, a 10Ω load is assumed as a resistivity of the load in the plasma reaction

period and results are presented in Fig. 2.13.

0 0.2 0.4 0.6 0.8 1x 10

-4

0

10

20

(a)

Ind

uct

or c

urre

nt(

A)

0 0.2 0.4 0.6 0.8 1x 10-4

0

500

1000

1500

(b)

C1 &

S1 v

olta

ge(V

)

4.46 4.48 4.5 4.52 4.54 4.56 4.58 4.6

x 10-4

0

20

40

1.33 1.332 1.334 1.336 1.338 1.34 1.342 1.344 1.346

x 10-3

0

20

40

113

0 0.2 0.4 0.6 0.8 1

x 10-4

0

1000

2000

3000

(c)

C2 &

S2 v

olta

ge(V

)0 0.2 0.4 0.6 0.8 1

x 10-4

0

1000

2000

3000

(d)

Ou

tput

vo

ltag

e(V

)

0 0.2 0.4 0.6 0.8 1

x 10-4

0

100

200

300

(e)Time(s)

Loa

d cu

rren

t(A

)

Fig.2. 13. Output voltages and currents of power supply, (a) Inductor current (A), (b) C1 & S1 voltage (V), (c) C2 & S2 voltage (V), (d) Output voltage (V), (e) Load current (A)

As can be seen in Fig. 2.13(d), the plasma resumes at 3kV voltage while each

capacitor provides half of this voltage level shown in Fig. 2.13(b) & (c). At the

plasma reaction period, the load discharges capacitors firstly and then the

inductor energy can be either delivered to the load with a bigger time constant or

discharged in assigned resistors. The load voltage created by inductor current

which is remained voltage across capacitors after discharging is balanced

identically. These results remove all concerns in regard of capacitor’s voltage

sharing issues. The remained current and voltage in the inductor and capacitors

are discharged through discharging modes previously described. While L and C1

are being discharged, the current passes through C2 and recharges it to almost

3kV. In the next mode C2 is being discharged while the voltage across CLoad is

recharging C1 again. Hence these switching states continue alternatively in order

to fully discharge the remained energy.

The results shown in Fig. 2.14 demonstrate the case of lack of plasma reaction

and energy delivering. A brief review over these results can be beneficial for the

analyses of power supply operation in this particular case. As anticipated, the

inductor energy is delivered to the capacitor bank and charges it and if this

energy is not depleted the system will face a major trouble.

0 0.5 1 1.5 2 2.5x 10

-5

0

10

20

(a)

Ind

ucto

r cu

rren

t(A

)

0 0.5 1 1.5 2 2.5

x 10-5

0

1000

2000

3000

(b)

C1 &

S1 v

olta

ge(

V)

114

0 0.5 1 1.5 2 2.5x 10

-5

0

1000

2000

3000

(c)

C2 &

S2 v

olta

ge

(V)

0 0.5 1 1.5 2 2.5

x 10-5

0

2000

4000

(d)Time(s)

Ou

tpu

t vo

ltag

e(V

)

Fig.2. 14. Inductor current and output voltages of power supply in the case of no prosperous plasma phenomena, (a) Inductor current (A), (b) C1 & S1 voltage (V), (c) C2 & S2 voltage (V), (d)

Output voltage (V)

There is no plasma at 3kV voltage, thus the capacitors should be discharged

when the output voltage reaches to a higher voltage level (4kV for example). The

power supply switches to the 5th mode but the point is the inductor current is in a

considerable level and may charge C2 to a high level of voltage which is out of

tolerate range for the switches. Therefore it is not possible to entirely discharge

the inductor current in a single discharging state. Hence the switching mode

changes while each capacitors voltage crosses a specific value (3kV for example)

until the inductor current being fully discharged. Then the exchange manner

follows to ensure discharging capacitors. Now, the system is prepared for the

next supplying cycle.

2.5. Components determinat ion and energy

discussion

Efficiency is the main concern when designing a power supply for plasma

applications. In this topology having the least energy losses is considered in

addition to the flexibility of the equipment which needs to be adjusted for a

diversity of pulsed power applications. The inductive and capacitive components

(L & Ci), should be selected appropriately in order to both satisfy load

requirements and avoid energy wasting. As the output voltage level and stress

and delivered energy are defined by load, the elements sizes can be determined

with regard to those parameters.

The output equivalent capacitor, neq CCCC ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅==== 21 , should be at least ten times

of the load capacitance to prevent any loading problem. On the other hand the

equivalent capacitance needs to be as small as possible to generate voltage stress

and level demanded by the load. Thus,

Loadeq CC 10==== (2-19) If the capacitors, Ci, are supposed identical then nCC ieq = :

115

Loadi CnC ⋅⋅⋅⋅==== 10 (2-20) n is the number of switch-diode-capacitor units which is determined by the

switches voltage and the demanded output voltage.



)()( dtdVCCI outLoadeqL ⋅+= (2-21) In the last stage, the demanded energy stored in the inductor defines the

inductance value.

LoadL EIL =⋅ 2)21( (2-22) Finally, the recovery time for inductive and capacitive components and the

frequency of pulsed power generated by the power supply can be determined as

follows:

inLLr VILT )(_ ⋅= (2-23) LoutLoadeqCr IVCCT ])[(_ ⋅+= (2-24)

)(])([ 2_ LininoutLoadeqLLCr IVVVCCILT ⋅⋅⋅++⋅= (2-25)


prepared for next supplying cycle. For instance in plasma generation, produced

plasma needs to be exhausted and the reactor should be filled with fresh material.

This interval is defined by Tr_Load in these equations. The frequency of load

supply with pulsed power relies significantly on the load features and

requirements, Tr_Load, but cannot be more than the recovery frequency of the

power supply.

)](1[ __max_ dLoadrLCrs TTTf ++< (2-26) Load’s capacitive and resistive characteristics in the interval of plasma

phenomena define discharging time for the inductor (Td). The inductor current in

a load supplying cycle is shown in Fig. 2.15 with detailed time intervals.

0 0.2 0.4 0.6 0.8 1 1.2

x 10-4

-5

0

5

10

15

20

Time(s)

Ind

uct

or

curr

en

t(A

)

Td

Tsmax

TrC

TrLoad

TrL

Fig.2. 15. Times monitoring in a load supplying cycle

Regarding above determinations, a model has been designed in

116

Matlab/SIMULINK platform to analyze the performance of the proposed circuit.

The detailed specifications of circuit are given in Table 2.1.

TABLE 2. 1. Specifications of the modeled circuit

Vin L RD Ci R1Load R2Load CLoad fs 200V 0.6mH 10 Ω 10nF 10MΩ 10 Ω 1nF 2kHz

These results indicate how the topology decreases the energy losses and

improves power efficiency. For example presume that the current source and the

units switches, SS & S1,…,Sn get closed when the inductor still delivers 1A to the

load. This means 10 volt still exists across the 10Ω load which is named as Vout-

off. Closing switches results in discharging inductor and output capacitors through

unit loops and loses energy. This energy loss for a delivery cycle can be

estimated as:

)V+CI(L=E out-offeqL-offLoss22)21( ⋅⋅⋅ mj003.3)101061106.0(5.0 2923 =××+××= −−

While the total stored energy in the inductor is:

mjIL=E LTotal 12020106.05.0)()21( 232 =×××=⋅⋅ − Regarding the calculation and analyses, in this strategy for example, if the

intrinsic conduction and switching losses of circuit are neglected, the energy

losses will be almost less than 2.5% of the total stored energy in the inductor.

This loss is negligible in comparison with the delivered energy to the load so, the

efficiency could be considered more than 97.5% in power delivery process.

2.6. Exper imenta l resul ts

To verify the validity of the proposed topology and the control strategy in

satisfying energy exchange among the inductive and capacitive components, a

laboratory prototype of a double switch-capacitor unit converter has been

developed. The instantaneous value of the inductor current is used to control the

level of the stored energy in the inductor. The control signals are fed into the

gates of the main switching devices (SS, S1, S2) through the gate driver circuits

which provide the necessary isolation between switching signal ground and the

power ground. The laboratory prototype including the control and the power

modules is shown in Fig. 2.16.

117

Fig.2. 16. Laboratory prototype of pulsed power supply with double switch-capacitor units

In the power board, a SK50 Gar 065 (SEMIKRON) switch, which is a compact

design of an Insulated Gate Bipolar Transistor, (IGBT) and a diode module

suitable for the buck converter, and a SK25 GB 065 (SEMIKRON) including

series connection of two IGBTs are used as the main switching devices for SS, D,

S1 and S2. The controllers for this set up has been developed utilizing an NEC

32-bit 64MHz V850/IG3 micro-controller. Skyper 32-pro (SEMIKRON) is used

as a gate drive circuit, which can drive two switches independently and is

compatible with the utilised IGBT modules.

This prototype has been designed to conduct a test at low voltage range. Due to

power supply and measurement equipment restrictions, and protection concerns,

the low voltage test performed initially in order to investigate and verify the

general concept of this topology and its control algorithm. The inductor current

range was adjusted around 2.5 Amps and consequently the output voltage was

under 300 Volts. The captured results are demonstrated in Fig. 2.17(a) & (b).

The inductor current, capacitors and output voltages are shown in Fig. 2.17(a) in

a wider time range while a focused shot of these voltages in Fig. 2.17(b)

illustrates the voltage rises in the capacitors. Both the capacitor voltages (VC1,

VC2) shown separately with different colours attain to 110 Volt values while the

output voltage of converter which is the summation of those two voltages and

depicted under them reaches to 220 Volts. The high rising time of voltages and

the low ultimate voltage magnitudes are due to the adjustment of the inductor

current in low ranges.

118

(a) (b)


2.7. Conclus ions

This paper proposed a new topology based on switch-capacitor units in series to

generate high voltage level and stress for pulsed power applications. The general

concept of this pulsed power supply is based on a current source topology

connected to a series of low-medium voltage switch-capacitor units which

considerably improves the efficiency of plasma systems. Simulations have been

carried out to validate the proposed topology and control. The simulation results

show that there is no restrict for the generation of higher voltage levels and

stresses by increasing the number of the switch-capacitor units. A laboratory

prototype is also implemented and the test results verified the whole concept of

the topology in low voltage. Utilizing pretty small capacitors and having no

diode in the configuration of this power supply are some advantages of this

topology in comparison with former technologies, such as Marx generators,

which are utilized as pulsed power supply. In addition, the output voltage level is

flexible and can be adjusted in a high range through the control switching signals

which is not possible in Marx technology. Although the output voltage in Marx

modulator can be adjusted by changing either the number of stages or the input

voltage, the proposed topology has the potential to vary the output voltage in a

wider range by controlling the inductor current which is accomplished through

the duty cycle of SS. Changing a software parameter is clearly easier in

comparison with varying Marx stages or the input voltage (with power supply

restrictions). Having no control over power flow to the load is the main shortage

of this circuit which can be neglected while there is possibility to define the

amount of stored energy with respect to the load demands.

119

2.8. References

[1] H. Akiyama, T. Sakugawa, T. Namihira, K. Takaki, Y. Minamitani, and N. Shimomura, “Industrial Applications of Pulsed Power Technology,” IEEE Transactions on Dielectrics and Electrical Insulation, Volume 14, Issue 5, pp. 1051 – 1064, October 2007.

[2] H. Akiyama, S. Sakai, T. Sakugawa, and T. Namihira, “Invited Paper - Environmental Applications of Repetitive Pulsed Power,” IEEE Transactions on Dielectrics and Electrical Insulation, Volume 14, Issue 4, pp. 825 – 833, August 2007.

[3] T. Namihira, S. Tsukamoto, D. Wang, S. Katsuki, R. Hackam, H. Akiyama, Y. Uchida, and M. Koike, “Improvement of NOX Removal Efficiency Using Short-Width Pulsed Power,” IEEE Transactions on Plasma Science, Volume 28, Issue 2, pp. 434 – 442, April 2000.

[4] Y. Aso, T. Hashimoto, T. Abe, and S. Yamada, “Inductive Pulsed-Power Supply With Marx Generator Methodology,” IEEE Transactions on Magnetics, Volume 45, Issue 1, Part 2, pp. 237 – 240, January 2009.

[5] J. Gao, Y. Liu, J. Liu, J. Yang, and J. Zhang, “Development of a Repetitive Wave Erection Marx Generator,” IEEE Transactions on Plasma Science, Volume 37, Issue 10, Part 1, pp. 1936 – 1942, October 2009.

[6] C. J. T. Steenkamp, and M. P. Bradley, “Active Charge/Discharge IGBT Modulator for Marx Generator and Plasma Applications,” IEEE Transactions on Plasma Science, Volume 35, Issue 2, Part 3, pp. 473 – 478, April 2007.

[7] T. Namihira, S. Sakai, T. Yamaguchi, K. Yamamoto, C. Yamada, T. Kiyan, T. Sakugawa, S. Katsuki, and H. Akiyama, “Electron Temperature and Electron Density of Underwater Pulsed Discharge Plasma Produced by Solid-State Pulsed-Power Generator,” IEEE Transactions on Plasma Science, Volume 35, Issue 3, pp. 614 – 618, June 2007.

[8] J. Choi, T. Namihira, T. Sakugawa, S. Katsuki, and H. Akiyama, “Simulation of 3-Staged MPC Using Custom Characteristics of Magnetic Cores,” IEEE Transactions on Dielectrics and Electrical Insulation, Volume 14, Issue 4, pp. 1025 – 1032, August 2007.

[9] R. Narsetti, R. D. Curry, K. F. McDonald, T. E. Clevenger, and L. M. Nichols, “Microbial Inactivation in Water Using Pulsed Electric Fields and Magnetic Pulse Compressor Technology,” IEEE Transactions on Plasma Science, Volume 34, Issue 4, Part 2, pp. 1386 – 1393, August 2006.

[10] J. Su, X. Zhang, G. Liu, X. Song, Y. Pan, L. Wang, J. Peng, and Z. Ding, “A Long-Pulse Generator Based on Tesla Transformer and Pulse-Forming Network,” IEEE Transactions on Plasma Science, Volume 37, Issue 10, Part 1, pp. 1954 – 1958, October 2009.

[11] W. C. Nunnally, S. M. Huenefeldt, and T. G. Engel, “Performance and Scalability of MJ Sequentially Fired Pulse Forming Networks for Linear and Nonlinear Loads,” IEEE Transactions on Plasma Science, Volume 35, Issue 2, Part 3, pp. 484 – 490, April 2007.

[12] J. S. Tyo, M. C. Skipper, M. D. Abdalla, S. P. Romero, and B. Cockreham, “Frequency and bandwidth agile pulser for use in wideband applications,” IEEE Transactions on Plasma Science, Volume 32, Issue 5, Part 1, pp. 1925 – 1931, October 2004.

[13] J. O. Rossi, M. Ueda, and J. J. Barroso, “Design of a 150 kV 300 A 100 Hz Blumlein coaxial pulser for long-pulse operation,” IEEE Transactions on Plasma Science, Volume 30, Issue 5, Part 1, pp. 1622 – 1626, October 2002.

121

I



I. they meet the criteria for authorship in that they have participated in the conception,









requirements.


A New Pulsed Power Supply Topology Based On Positive Buck-Boost Converters Concept

Published in: IEEE Transactions on Dielectric and Electrical Insulation, Vol. 17, No. 6, pp.

1901-1911, Dec. 2010

Contributor , Statement of contribution Sasan Zabihi \ Proposed principle modifications in the initial design, conducted

I~//'---' simulation studies and data analysis, designed the control strategy, implemented hardware set up, designed and conducted experimental

11 Aug. 2011 verifications, and wrote the manuscript. Proposed the initial design and supervised the validity studies

Firuz Zare including: conducting the simulations and experimental studies, designing the control strategy, and writing the manuscript.

Gerard Ledwich Aided experimental design, and data analysis.

Arindam Ghosh Aided planning the control strategies and writing the paper.

Hidenori Akiyama Provided us with general infonnation about pulsed power supply specifications and its application demands.



authorship.

A/Pro f. Firuz Zare ~ _N_am __ e---------------~--- ~

11 Aug. 2011

Date

122

CHAPTER 3

A New Pulsed Power Supply Topology Based

On Posi t ive Buck-Boost Converters Concept


Akiyama†




Published in: IEEE Transactions on Dielectric and Electrical Insulation, Vol. 17,

No. 6, pp. 1901-1911, Dec. 2010

123

Abstract— Improving efficiency and flexibility in pulsed power supply

technologies is the most substantial concern of pulsed power systems specifically

with regard to plasma generation. Recently, the improvement of pulsed power

supply has become of greater concern due to the extension of pulsed power

applications to environmental and industrial areas. With this respect, a current

source based topology is proposed in this paper as a pulsed power supply which

gives the possibility of power flow control during load supplying mode. The

main contribution in this configuration is utilization of low-medium voltage

semiconductor switches for high voltage generation. A number of switch-diode-

capacitor units are designated at the output of topology to exchange the current

source energy into voltage form and generate a pulsed power with sufficient

voltage magnitude and stress. Simulations carried out in Matlab/SIMULINK

platform as well as experimental tests on a prototype setup have verified the

capability of this topology in performing desired duties. Being efficient and

flexible are the main advantages of this topology.

3.1. Index Terms

Pulsed power supply, High voltage, Current source, Voltage source, Power

converter, dc-dc topology, Plasma.

3.2. Int roduct ion

Pulsed power is the accumulation of energy over a relatively long period of time

and releasing it very quickly which is a process aiming to increase the

instantaneous power. The characteristics of this pulse, including voltage level

and rising time are determined based on the load requirements.

Although single shot based pulsed power generators with extremely high peak

power have been considered initially for military and nuclear fusion applications,

repetitively operated pulsed power generators with a moderate peak power have

been recently developed mainly for industrial applications such as food

processing, medical treatment, water treatment, exhaust gas treatment, concrete

recycling, ozone generation, engine ignition, ion implantation etc [1, 2]. Marx

Generators (MG) [3, 4], Magnetic Pulse Compressors (MPC)[5, 6], Pulse

Forming Network (PFN)[7, 8], Multistage Blumlein Lines (MBL)[9, 10] etc, are

the most popular technologies which have been utilized so far as pulsed power

124

supply. Hiring aged methods and technologies besides lack of agreement

between power supply and load properties cause major issues in pulsed power

area. Not tacking utilization of advanced knowledge and recent approaches in

power electronics and semiconductors into account was due to lack of necessity

to improve power supply technologies during past decades. Efficiency, flexibility

and intricacy are major drawbacks of these power supplies. Controlling power

flow is a critical skill which can improve the efficiency of power supply systems.

On the other hand these pulsed power systems require high voltage, high power

switches in which their voltage blocking and switching time are limited. The

switches technology utilized for pulsed power generation has varied with respect

to the development of power semiconductor devices over, the past few decades.

Thyristor, IGBT, MOSFET, etc are some of those switches mostly classified as

solid state semiconductor switches [11-13]. Since pulsed power applications

demand for high dv/dt, fast switches with short switching transients have critical

role in pulsed power supply topologies [14]. The timescales of these transients

are from nanoseconds to microseconds, including a switching transition of power

semiconductor devices, commutating processes, and drive signal transmissions.

These transient processes directly affect the performance and reliability of power

electronic systems with imposed restrictions over power conduction [15].

Most pulsed power applications have resistive-capacitive characteristics [16];

therefore, a current source topology seems to be a proper candidate to supply

such loads. With respect to this issue a combination of current and voltage

sources is considered in this paper to develop the initial concept of high voltage

pulse generation with low voltage switches. The circuit depicted in Fig. 3.1

reveals a general configuration for the proposed topology. Same sort of fast and

low-medium voltage semiconductor switches and diodes are used between two

energy storages in order to control the energy delivery process. In this

configuration the inductor and the capacitors, which can be supposed as the

current and voltage sources, are in charge of supplying energy and generation of

appropriate voltage level and stress respectively.

Fig.3. 1. A general configuration of proposed concept

125


3.3.1. Topology

3.3.1.1. General configuration

The topology considered in this paper is based on the positive buck-boost

converter concept. The general concept of this topology is presented in Fig. 3.2.

Switching Pulses

SS

S1

SL

D

R2Load

C1

-

+

+

-

L

S2

C2

+

-

Sn

Cn

+

-

R1LoadControl

Protocol

of

Series

Switches

Plasma Load Model


AC-DC

ConverterVac

220 V50 Hz

CLoad


D2

D3

D1

Dn

V inCin



Load

Fig.3. 2. Pulsed power supply configuration with multi switch-diode-capacitor units

An ac-dc converter rectifies grid ac voltage into a dc voltage and supplies rest of

the circuit. The source voltage charges an inductor L through switches, SS, and

S1, S2…, Sn, composing a current source. The level of current, stored in the

inductor during charging mode, can be controlled via an appropriate duty cycle

of SS.

A freewheel diode D which is connected between the switch and the inductor,

conducts the current in order to provide a current loop and keep the current

constant, while SS is switched off. The switches, S1, S2…, Sn, are connected to a

series of capacitors through diodes which compose switch-diode-capacitor units.

These units in a group association act as a combined voltage source and generate

desired high voltage at the output. The inductor current flows through the unit’s

switches while they are on. As soon as the switches are turned off, the inductor

126

current flows to the capacitors through the diodes. The received energy from the

current source is stored in the capacitors in the form of voltage.

Most pulsed power applications have resistive and capacitive properties which

can be modeled as a sample load with a capacitor, CLoad, a switch, SLoad, and two

resistors, R1Load & R2Load as shown in Fig. 3.2. The capacitor represents the

capacitive specification of the loads and switching between large and small

resistors, R1Load & R2Load, simulates the break down phenomena happening while

pulsed power applies to the loads.

As shown in Fig. 3.3, a double unit configuration is investigated in this paper as

a simple model. The results can be extended for a multi unit topology.

Fig.3. 3. A pulsed power supply with two switch-diode-capacitor units and a non-linear load

3.3.1.2. Switching modes

The operation modes of this topology are separated into two major groups.

Switching states depicted in Fig. 3.4(a) and 3.4(b) and Fig. 3.4(c) and 3.4(d) are

classified in current source category and voltage source category, respectively.

3.3.1.2.1. First mode: Charging inductor (SS: on, S1: on, S2: on)

As demonstrated in Fig. 3.4(a), in this switching state, all the switches, including

current source switch SS and units switches, S1 & S2, are turned on to increase the

inductor current. Therefore the input voltage Vin appears across the inductor and

the charging time can be calculated as follows.

)(21 SSSinL VVVVV

S++++++++−−−−==== (3-1)

t

iL

dt

diLVL ∆

∆======== (3-2)

If the inductor is supposed to be with no initial current charge and Ii ====∆ then

LV

ILt

⋅⋅⋅⋅====∆ .

3.3.1.2.2. Second mode: Circulating the inductor current (SS: off, S1: on, S2:

on):

As soon as the inductor current crosses a defined amount, the control system

turns off the current source switch SS and disconnects the input voltage source Vin

127

from the rest of topology. Henceforth, the freewheel diode D conducts and lets

the inductor current to circulate through S1 & S2. In this mode, which is

illustrated in Fig. 3.4(b), the low voltage drop across the diodes and switches

discharges the inductor moderately. As the total voltage across the inductor is not

significant, the discharging effect can be neglected and the circulating current is

considered to be constant. This switching state keeps the current stored in the

inductor and allows the load system to be prepared for the next cycle of

energizing.

)(21 SSDL VVVV ++++++++−−−−==== (3-3)

This is a mandatory switching mode but the converter can stay in this mode for a

short time in order to minimize conduction loss. Then the converter can be

switched to the third switching mode if the load is prepared to be energized.

Neglecting this switching state in order to avoid the conduction losses raises

stability concerns and it is indeed a safety state which is necessary for this

system. During this switching mode, the inductor is isolated from the source by

turning SS off, giving the possibility of disconnecting input voltage from the load

during power delivery period. Therefore, even if there is an arc occurring at the

load side, there will be no chance to waste a large amount of energy through the

input source.

3.3.1.2.3. Third mode: Charging capacitors (SS: off, S1: off, S2: off)

In this switching state, the current source delivers the inductor current to the

capacitors and charges them. As exhibited in Fig. 3.4(c), the unit’s switches, S1

& S2, are turned off in this mode and the inductor current is pumped into the

capacitors and charges them to a certain level defined by the load.

tC

IV

C

I

t

V

i

CC

i

CC i

i

ii ∆∆∆

∆⋅⋅⋅⋅====⇒⇒⇒⇒==== (3-4)

During this state, in reality, the resistivity of load considerably collapses due to

the application of pulsed power to the load and plasma generation reaction. A

plasma phenomenon has been modeled by decreasing the load resistance from R1

to R2 through switch SL as demonstrated in Fig. 3(d). The required energy is

delivered to the load from the voltage and current sources in this mode. The

capacitor bank and the inductor are discharged subsequently according to the

proportion of energy stored in them. Once the load supplying process is finished,

128

the topology can switch from the supplying mode to the charging inductor mode

with no concern.

(a) (b)

(c) (d)

Fig.3. 4. Switching states of the proposed power supply circuit (a) Inductor charging (b) Circulating the inductor current (c) Charging the capacitors (d) Supplying the load

3.3.1.2.4. Fourth mode: Separately charging the capacitors (SS: off, S1: of,

S2: off):

This topology is also flexible in terms of charging capacitors separately. In this

scenario, the unit’s switches will be turned off subsequently and gives the

feasibility of charging the capacitors in an appropriate sequence. As can be seen

in Fig. 3.5, S2 is turned off while S1 is still on in order to charge C2 through D2 in

this mode. The achievement of this strategy is having a voltage storage

continuously charged which provides a basic voltage level for the load. In the

present model, C2 is responsible for this function, so the other capacitors are

dedicated to providing desired voltage stress.

Fig.3. 5. Switching state of charging capacitors separately

3.3.1.3. Circuit analyses

The voltage sharing of capacitors in different operation modes can be calculated

as follows:

While S1 and S2 are turned on and off separately, there will be an initial charge in

C2, shown with fundamental voltage Vf in the following equations.

129

While C2 become charged in the fourth mode (SS:OFF, S1:ON, S2:OFF,

SL:OFF):

0)1(1

====CV (3-5)

fC VV ====)1(2

(3-6)

While S1 is turned off, as well as S2 in the next mode (SS:OFF, S1:OFF, S2:OFF,

SL:OFF), the voltage sharing of capacitors would be:

21

22)2(1 CC

VCVCV fout

C ++++−−−−

==== (3-7)

21

21)2(2 CC

VCVCV fout

C ++++++++

==== (3-8)

Whereas Vout is the required voltage level for the load break down.

When the load discharges the capacitors energy, the capacitor allotted for dv/dt

generation C1 will be fully discharged and then get negative charge from the

capacitor allocated to fundamental voltage C2 since it still has energy and

delivers it to the load which charges C1 negatively. (SS:OFF, S1:OFF, S2:OFF,

SL:ON):

21

2)3(1 CC

VCV f

C ++++−−−−

==== (3-9)

21

2)3(2 CC

VCV f

C ++++==== (3-10)

In a specific case when C1=C2, the above equations change in to:

(SS:OFF, S1:ON, S2:OFF, SL:OFF):

0)1(1

====CV (3-11)

fC VV ====)1(2

(3-12)

(SS:OFF, S1:OFF, S2:OFF, SL:OFF):

2)2(

1

foutC

VVV

−−−−==== (3-13)

2)2(

2

foutC

VVV

++++==== (3-14)

(SS:OFF, S1:OFF, S2:OFF, SL:ON):

2)3(

1

fC

VV

−−−−==== (3-15)

2)3(

2

fC

VV

++++==== (3-16)

The separate switching strategy has a number of advantages in comparison with

simultaneous switching. As already mentioned, a fundamental dc voltage level

which is almost invariable can be generated in this method. This source of

130

energy will be helpful to save a part of the supplying process and increase the

frequency of pulse supply.

3.3.2. Control strategies

Switches used in this power supply have two different functions. A single switch

at the front side of topology SS can charge the inductor at a certain level. A range

of switches, S1 & S2, at the output of the topology either circulates the current or

conducts it to the capacitors. The switch used for modeling the plasma break

down phenomena in the load SL is controlled at a certain voltage level. As

expected, each type of these switches is functionalized under a specific principle

in order to meet assigned duties. A flowchart, shown in Fig. 3.6, describes the

logic of decisions which generate control signals to charge the inductor and

capacitors.

3.3.2.1. Current source control

The first stage is charging the inductor through the front part of the circuit.

Assuming the switches S1, and S2 are on, the inductor can be charged when the

switch SS is turned on. The controller measures the inductor current and turns off

the switch when the inductor current reaches Imax. In this case the energy stored

in the inductor is 2max2

1IL ⋅⋅⋅⋅ .

As can be found in the flowchart shown in Fig. 3.6, turning SS off which means a

transition between inductor charging and current source modes is carried out

with a comparison between actual inductor current and a specific amount Imax set

as charging limit. Whereas, SS is turned on while a load supplying cycle is spent

and the energy is delivered to the load in this cycle. This will be detected for the

system as soon as output voltage becomes less than a specific level; Vmin. Vmin is

relevant to the load energy demand and determined by the programmer. The only

concern which restricts Vmin determination is diode’s breakdown voltage, Vd.

dVV ≤≤≤≤min (3-17)

To increase Vmin level, it is possible to connect a number of low voltage diodes in

series. In this way, the flexibility of stopping load supply in higher voltages will

be brought to this configuration as another advantage.

131

Fig.3. 6. Flowchart of the control algorithm

3.3.2.2. Voltage source control

There are two control strategies, simultaneous and separate switching, applicable

for this range of switches. Each has its own features and benefits.

Initially, the simultaneous switching method is considered to demonstrate the

performance and the capabilities of this topology. In this method, all unit’s

switches will be turned off together after the inductor is fully charged. As a

result, the inductor current will be pumped into all output capacitors and will

charge them at the same time. Each capacitor generates a specific dv/dt and

voltage level with respect to the capacitor amount. Assuming similar capacitors,

the eventual voltage level will be shared among all capacitors equally. Pulsed

power will be generated and applied to the load which subsequently discharges

voltage and current sources. This trend will be repeated for the next pulse

supplying cycles. The simulation results of this strategy are displayed in Fig. 3.9.

132

Separate switching is another scenario which can be considered for controlling

unit’s switches and bringing inevitable advantages for this topology. In this

switching strategy, unit’s switches are turned off subsequently and as a

consequence let the capacitors to be charged separately. This flexibility is

particularly beneficial in the case of asymmetrical capacitors. In this feature, it is

possible to assign different duties to different capacitors and recharge them with

this respect. As an illustration, suppose that C2 is allocated to provide a

fundamental voltage level. On the other hand C1 is allotted for providing

required dv/dt. In this regard, C2 should be appointed more capacitive than C1.

Here in this study, C2 is determined ten times C1. As already discussed in the

switching modes, S2 is switched off while S1 is still on and conducts the current.

As the circuit indicates in Fig. 3.5, the current flows to C2 through D2 and

charges it up to a specific level. Just after charging C2, S1 is switched off as well

and allows the inductor current to be conducted through D1 and charges both C1

& C2 simultaneously. Since C2 is more than C1 a similar current will charge it

less than C1 in a definite time. Therefore, the level of voltage provided by C1 is

significantly higher than that by C2. In this regard, C1 is dedicated to dv/dt, and

C2 is assigned for a basic and rather unvarying voltage level. The current and

voltage waveforms accompanied by switching signal patterns of circuit

controlled with this principal are exhibited in Fig. 3.7.

3.3.2.3. Load modeling control

Load switch, SL, is turned on when the output voltage reaches a specific voltage

level. Therefore the resistivity of load suddenly collapses by turning SL on, in

order to simulate a plasma phenomenon. On the other hand, SL becomes off

while the reaction ends.

Discharging process appeared at the bottom of control algorithm flowchart is a

safety procedure considered in situation of no prosperous plasma reaction. Once

the output voltage increases to more than Vmax level in these conditions, an

external load will be connected to the power supply output and discharges the

stored voltage. Vmax selected with respect to the voltage tolerance of capacitors

and diodes is almost 20% more than the voltage level in which plasma is

expected to take place. In these circumstances, the diodes D, D1, D2 should

tolerate high levels of voltage since the inductor current is supposed to be

133

finished and the output voltage locates across D and D1. As can be seen in Fig.

3.4(c) and Fig. 3.5, these diodes as well as D2 should have breakdown voltages

in an appropriate range in order to handle high levels of voltage

Fig.3. 7. Current and voltage waveforms accompanied by relevant switching signals pattern in separate switching strategy

3.3.3. Components determination and energy discussion

Efficiency is the main concern when designing a power supply for pulsed power

applications. To have the most possible efficient configuration, the topology

structure, control algorithm and components sizes should be in the best

correspondence with the application attributes and demands. In this

configuration, having the least energy losses is considered in the topology in

addition to the flexibility of the equipment, which needs to be adjusted for a

diversity of pulsed power applications. The inductive and capacitive components

(L & Ci), should be selected appropriately in order to both satisfy load

134

requirements and avoid energy wasting. As the output voltage level and stress

and delivered energy are defined by the load, the elements sizes can be

determined with regard to those parameters.

The output equivalent capacitor neq CCCC ⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅==== 21 should be at least ten times

the load capacitance to prevent any loading problem. On the other hand, the

equivalent capacitance needs to be as small as possible to generate voltage level

and stress demanded by the load. Thus,

Loadeq CC 10==== (3-18)

If the capacitors, Ci, are supposed identical then n

CC i

eq ==== :

Loadi CnC ⋅⋅⋅⋅==== 10 (3-19)

n is the number of switch-diode-capacitor units which is determined by the

switches voltage and the demanded output voltage.



dt

dVCCI out

LoadeqL )( ++++==== (3-20)

In the last stage, the demanded energy stored in the inductor defines the

inductance value.

LoadL EIL ====⋅⋅⋅⋅ 2

2

1 (4-21)

Finally, the recovery time for inductive and capacitive components and the

frequency of pulsed power generated by the power supply can be determined as

follows:

in

LLr V

ILT

⋅⋅⋅⋅====_ (3-22)

L

outLoadeqCr I

VCCT

⋅⋅⋅⋅++++====

)(_ (3-23)

Lin

inoutLoadeqLLCr IV

VVCCILT

⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅⋅++++++++⋅⋅⋅⋅

====)(2

_ (3-24)


prepared for the next supplying cycle. For instance, in plasma generation,

produced plasma needs to be exhausted and the reactor should be filled with

fresh material. This interval is defined by Tr_Load in these equations. The

frequency of load supply with pulsed power relies significantly on the load

135

features and requirements, Tr_Load, but cannot be more than the recovery

frequency of the power supply.

dLoadrLCrs TTT

f++++++++

<<<<__

max_

1 (3-25)

Load’s capacitive and resistive characteristics in the interval of plasma

phenomena define discharging time for the inductor (Td). The inductor current in

a load supplying cycle is shown in Fig. 8 with detailed time intervals.

Regarding the above determinations, a model has been designed in

Matlab/SIMULINK platform to analyze the performance of the proposed circuit.

The detailed specifications of the circuit are given in Table 3.1.

TABLE 3. 1. Specifications of the modeled circuit

Vin L C1 C2 R1Load R2Load CLoad fS 200V 0.6mH 10nF 100nF 10MΩ 10Ω 1nF 2kHz

0 0.2 0.4 0.6 0.8 1 1.2

x 10-4

-10

-5

0

5

10

15

20

25

30

Time(s)

Indu

ctor

cur

rent

(A)

Td

Tsmax

TrL

TrLoad

TrC

Fig.3. 8. Times monitoring in a load supplying cycle

These results indicate how the topology decreases the energy losses and

improves power efficiency. For example, presume that the current source and the

units switches, SS & S1,…,Sn, get closed when the inductor still delivers 10A to

the load. This means 100 volt still exists across the 10Ω load which is named as

Vout-off . This voltage and the remaining inductor current can be stored in these

components and be used in the recovery period of the next cycle. Therefore, no

energy will be wasted in this topology in the process of delivering energy. Based

on this issue, the power losses will be restricted to switching and conduction

losses and the efficiency of this power supply in pulsed power supplying systems

can be considered remarkable.

136

3.4. Simulat ion resul ts and analyses

Several simulations at different conditions have been carried out to verify the

validity of the proposed topology performance. The input voltage level, the

components size, the inductor current magnitude and the load breakdown

resistivity are the parameters varied in an extensive range to study the topology

performance in different situations. The results presented in this part are revealed

for two different switching strategies.

3.4.1. Simultaneous switching

In this case, the inductor is charged up to 30A and kept charged in this level until

the load becomes prepared for the supplying cycle.

Then S1 and S2 are switched off simultaneously, allowing the inductor current to

be pumped into the capacitor bank. The inductor energy delivered to the

capacitors is exchanged to the voltage form. The generated dv/dt is in proportion

with the inductor current level and the equivalent capacitors size. In this respect,

the output dc link’s voltage is charged up to 2kV, while each capacitor generates

1kV. This level of voltage, accompanied by an appropriate slope and rise time,

dv/dt, is critical for the modeled load and causes a breakdown phenomenon in

the load. Thus, load resistivity is markedly dropped and consumes the stored

energy. Consequently, the capacitors are discharged in a considerably short time

stint because of a very small time constant.

CR⋅⋅⋅⋅====τ (3-26)

The capacitors are not fully discharged because the inductor still supplies the

load with the current. This current magnitude times load resistivity creates a

voltage across output capacitors during this period. The inductor is discharged

afterwards because of a greater time constant.

R

L====τ (3-27)

The voltage remained at the output is also shared equally between two

capacitors. This supplying process can be stopped at anytime and this moment is

determined by the load demand for energy. The graphs exhibited in Fig. 3.9

demonstrate the inductor current, the capacitors and the output voltage and the

load current for a pulse generation moment respectively.

137

1.66 1.665 1.67 1.675 1.68 1.685x 10-4

28

29

30

31

(a)In

duct

or

curr

ent(

A)

1.66 1.665 1.67 1.675 1.68 1.685

x 10-4

0

1000

2000

(b)

C1 &

S1 v

olta

ge(V

)

1.66 1.665 1.67 1.675 1.68 1.685x 10

-4

0

1000

2000

(d)

Ou

tpu

t vo

ltag

e(V

)

1.66 1.665 1.67 1.675 1.68 1.685

x 10-4

0

100

200

(e)Time(s)

Load

cur

ren

t(A

)

Fig.3. 9. Output voltages and currents of power supply under simultaneous switching algorithm, (a) Inductor current (A), (b) C1 & S1 voltage (V), (c) C2 & S2 voltage (V), (d) Output voltage (V),

(e) Load current (A)

3.4.2. Separate switching

In the next case, the unit’s switches are turned off separately, based on a

particular logic, in order to charge asymmetrical capacitors for a specific

purpose. The relevant results to this strategy are shown in Fig. 3.10 in detail. In

this scenario, different functions are defined for each capacitor. The capacitor

which is opted larger is responsible for storing a definite amount of energy and

serving an almost continuous level of voltage. The smaller one will be charged

afterwards and is in charge for dv/dt. The discharging process is almost the same

as the previous one except for voltage sharing at the end of the process. The

smaller capacitor is discharged with a lower time constant than the bigger one.

7.84 7.85 7.86 7.87 7.88

x 10-4

27

28

29

30

31

(a)

Indu

ctor

cur

rent

(A)

7.84 7.85 7.86 7.87 7.88

x 10-4

-500

0

500

1000

1500

(b)

C1 &

S1 v

olta

ge(V

)

7.84 7.85 7.86 7.87 7.88

x 10-4

-500

0

500

1000

1500

(c)

C2 &

S2 v

olta

ge(V

)

7.84 7.85 7.86 7.87 7.88

x 10-4

-500

0

500

1000

1500

(d)

Out

put

volta

ge(

V)

7.84 7.85 7.86 7.87 7.88x 10-4

0

50

100

150

(e)Time(s)

Load

cur

rent

(A)

Fig.3. 10. Output voltages and currents of power supply under separate switching algorithm, (a)

138

Inductor current (A), (b) C1 & S1 voltage (V), (c) C2 & S2 voltage (V), (d) Output voltage (V), (e) Load current (A)

Although it has been charged to a higher voltage level, in comparison with the

other capacitor, it will be discharged faster due to the lower time constant. C2 is

still charged while C1 becomes fully discharged. Therefore, C2 continues to

deliver energy to the load through C1 and this energy will charge C1 with a

negative polarity. This trend will be continued until the inductor is fully

discharged. During this period, the output voltage, which is the resultant voltage

of capacitors, is inductor current times the load resistor. At the end of this

process, the output voltage will intend to be zero, corresponding with the

inductor current reduction, while C1 and C2 are charged with negative and

positive voltage respectively. This voltage balance creates initial provisions for

the next supplying cycle of the circuit. The next supplying cycle starts with the

inductor recovery modes and the topology will subsequently switch to capacitors

separate charging mode. Although the expectation is the conduction of D2 and

recharging of C2, D1 conducts the inductor current due to a positive voltage

caused by C1 negative voltage across it. Thus C1 and C2 are recharged

simultaneously but with different time constants. This procedure continues as far

as C1 voltage crosses zero level which is the time that the voltage across D1 turns

to be negative. Therefore D1 stops conduction and D2 conducts for the rest of this

switching state. The next switching state commences while C2 is charged up to

the fundamental voltage level, Vf.

Once C2 voltage becomes equal to this specific amount, S1 will become off in

order to charge C1 as well as C2. Although both capacitors are charged again in

this mode the main goal of this switching state is charging C1 to provide a

desired dv/dt for a load. This concern is satisfied due to insignificancy of C1. As

soon as demanded voltage level accompanied by an appropriate dv/dt is

generated the break down phenomena happens in the load and the accumulated

energy will be delivered to the load. The supplying process can be stopped in any

stage by turning S1 and S2 on. These sequences are repeated in all supplying

cycles.


To investigate the validity and accuracy of the proposed topology and the control

139

algorithm, a laboratory prototype has been developed with double switch-diode-

capacitor units at the output of the converter. A photo of mentioned experimental

setup including power and control boards, and the switches’ drivers can be seen

in Fig. 3.11.

SEMIKRON products, SK50 Gar 065 and SK25 GB 065, are utilized as

semiconductor components in the power circuit. SK50 Gar 065 consists of an

Insulated Gate Bipolar Transistor, (IGBT) and a power diode module which are

suitably arranged for buck configuration. This module is used in the power board

as the front side switch and diode, SS and D. SK25 GB 065 is also comprised of

two IGBTs connected in series together. This composition is properly fitted to

the output units design and consequently utilized as the output switches, S1 and

S2, in this converter. The controller for this setup has been developed utilizing an

NEC 32-bit 64MHz V850/IG3 micro-controller. Skyper 32-pro (SEMIKRON) is

used as a gate drive circuit, which can drive two solid state switches

independently and is compatible with the utilized IGBT modules.

The laboratory setup in this study is designed in order to generate an output

voltage level around one kilo volt. The IGBTs assembled in the SK25 GB 065

can withstand 600V each, and based on this fact 1kV is designated as the

ultimate voltage level of power supply in order to have an appropriate safety

margin. There are also other power switches with voltage ratings up to 1.5kV

available commercially which can be utilized in this topology to develop the

voltage escalating skill. This 1kV level is also determined due to power supply

and measurement equipment restrictions, as well as protection concerns.

Hardware with such features complies with the initial purpose of experimental

test and provides sufficient evidences to validate the true performance of

proposed topology. The specifications of the components used in this prototype

besides assigned adjustment level for the control of inductor current are given in

Table 3.2.

140

Fig.3. 11. Laboratory prototype of pulsed power supply with double switch-diode-capacitor units

TABLE 3. 2. Specifications of the laboratory prototype circuit

Vin L C1 C2 ILmax 15V 0.4mH 10nF 10nF 7A

The captured results are demonstrated in Fig. 3.12(a) & 3.12(b). As can be seen

in Fig. 3.12(a), the inductor is charged up to 7 Amps, while the energy exchange

process starts when the current is declined to 6 Amps. This reduction happens

during circulating current mode. By turning S1 and S2 off, the inductor current

flows into the capacitors and charges them up to 1.12 kV. The output voltage is

split pretty equally across both capacitors since each capacitor is almost charged

up to 560V. It indicates that the voltage sharing between the capacitors is rather

done properly. On the other hand, a focused perspective in a more limited time

frame, Fig. 3.12(b), gives a better vision of voltage rising trends. The low dv/dt

in this test is due to a low inductor current level and can be markedly improved

by raising the current level. As anticipated, a portion of delivered energy is

wasted during charging process due to conduction and switching losses. Active

and passive components including power switches and diodes as well as

circulating circuit normally consume a part of energy. In this case, comparing the

stored energy in the inductive and capacitive elements reveals that the energy

loss during exchanging energy procedure is considerable. Utilizing more

efficient components can influence the efficiency of whole converter

substantially.

141

VC1

VC2

VOUT

IL

(a) (b)


3.6. Conclus ion

This paper presents a current source based topology for pulsed power

applications. A range of switch-diode-capacitor units connected in series together

and in cascade to the current source is responsible for generating high voltage

and high dv/dt. In this regard, both current and voltage sources are considered

and utilized in this configuration as power sources despite of having

distinguished duties. The novel contribution in this configuration is utilizing low-

medium voltage switches to tolerate a high voltage at the output. In addition, this

topology has the flexibility of being easily adjusted for a wide range of pulsed

power applications. Having control over power delivery to the load is another

advantage of this power supply, which makes it thoroughly efficient. The

proposal topology’s true performance is investigated through several simulation

models and the acquired results confirm the validity of this model to cover all

desired duties. A laboratory prototype is also tested and the attained results have

verified the initial concept of this configuration in generating high voltage

pulses.

3.7. References

[1] H. Akiyama, T. Sakugawa, T. Namihira, K. Takaki, Y. Minamitani, and N. Shimomura, “Industrial Applications of Pulsed Power Technology,” IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 14, Issue 5, pp. 1051–1064, October 2007.

[2] H. Akiyama, S. Sakai, T. Sakugawa, and T. Namihira, “Invited Paper - Environmental Applications of Repetitive Pulsed Power,” IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 14, Issue 4, pp. 825–833, August 2007.

[3] T. Heeren, T. Ueno, D. Wang, T. Namihira, S. Katsuki, and H. Akiyama, “Novel Dual Marx Generator for Microplasma Applications,” IEEE Transactions on Plasma Science, Vol. 33, Issue 4, Part 1, pp. 1205–1209, August 2005.

[4] H. Li, H. J. Ryoo, J. S. Kim, G. H. Rim, Y. B. Kim, and J. Deng, “Development of

142

Rectangle-Pulse Marx Generator Based on PFN,” IEEE Transactions on Plasma Science, Vol. 37, Issue 1, pp. 190–194, January 2009.

[5] D. Wang, T. Namihira, K. Fujiya, S. Katsuki, and H. Akiyama, “The reactor design for diesel exhaust control using a magnetic pulse compressor,” IEEE Transactions on Plasma Science, Vol. 32, Issue 5, Part 1, pp. 2038–2044, October 2004.

[6] J. Choi, T. Yamaguchi, K. Yamamoto, T. Namihira, T. Sakugawa, S. Katsuki, and H. Akiyama, “Feasibility Studies of EMTP Simulation for the Design of the Pulsed-Power Generator Using MPC and BPFN for Water Treatments,” IEEE Transactions on Plasma Science, Vol. 34, Issue 5, Part 1, pp. 1744–1750, October 2006.

[7] E. Spahn, G. Buderer, and C. Gauthier-Blum, “Novel PFN with current turn-off capability for electric launchers,” IEEE Transactions on Magnetics, Vol. 37, Issue 1, Part 1, pp. 398–402, January 2001.

[8] T. G. Engel, and W. C. Nunnally, “Design and operation of a sequentially-fired pulse forming network for non-linear loads,” IEEE Transactions on Plasma Science, Vol. 33, Issue 6, Part 2, pp. 2060–2065, December 2005.

[9] T. Namihira, S. Tsukamoto, D. Wang, S. Katsuki, R. Hackam, H. Akiyama, Y. Uchida, and M. Koike, “Improvement of NOX removal efficiency using short-width pulsed power,” IEEE Transactions on Plasma Science, Vol. 28, Issue 2, pp. 434–442, April 2000.

[10] D. P. Kumar, S. Mitra, K. Senthil, S. Archana, K. V. Nagesh, S. K. Singh, J. Mondal, R. Amitava, and D. P. Chakravarthy, “Characterization and analysis of a pulse power system based on Marx generator and Blumlein,” Review of Scientific Instruments, Vol. 78, Issue 11, pp. 115107-115107-4, November 2007.

[11] J. Mankowski, and M. Kristiansen, “A review of short pulse generator technology,” IEEE Transactions on Plasma Science, Vol. 28, Issue 1, pp. 102–108, February 2000.

[12] W. Jiang, K. Yatsui, K. Takayama, M. Akemoto, E. Nakamura, N. Shimizu, A. Tokuchi, S. Rukin, V. Tarasenko, and A. Panchenko, “Compact solid-State switched pulsed power and its applications,” Proceedings of the IEEE, Vol. 92, Issue 7, pp. 1180–1196, July 2004.

[13] H. A. Mangalvedekar, K. P. Dixit, D. N. Barve, A. S. Paithankar, and D. P. Chakravarthy, “Development of solid state pulse power modulator using toroidal amorphous core,” IEEE Transactions on Dielectrics and Electrical Insulation, Vol. 16, Issue 4, pp. 1006–1010, August 2009.

[14] S. Castagno, R. D. Curry, E. Loree, “Analysis and Comparison of a Fast Turn-On Series IGBT Stack and High-Voltage-Rated Commercial IGBTS,” IEEE Transactions on Plasma Science, Vol. 34, Issue 5, Part 1, pp. 1692–1696, October 2006.

[15] H. Bai, Z. Zhao, and C. Mi. “Framework and Research Methodology of Short-Timescale Pulsed Power Phenomena in High-Voltage and High-Power Converters”, IEEE Transactions on Industrial Electronics, Vol. 56, Issue 3, pp. 805–816, March 2009.

[16] J. Pelletier, and A. Anders, “Plasma-based ion implantation and deposition: A review of physics, technology and applications,” IEEE Transaction on Plasma Science, Vol. 33, no. 6, pp. 1944–1959, Dec. 2005.

144

I












requirements.


A Solid State Marx Generator with a Novel Configuration

Published in: IEEE Transactions on Plasma Science, Vol.39, No.8, pp.l721-1728, Aug. 2011.

Contributor ~ Statement of contribution Sasan ZJlb..ihi Proposed the initial design and conducted simulation studies and data

/)v-6/ analysis, designed the control strategy, implemented hardware set-up and conducted experimental verifications and wrote the manuscript.

11 Aug. 2011

Zeynab Zabihi Aided simulations studies, and data analysis.

Contributed in initial design process, supervised the validity studies Firuz Zare including: conducting the simulations and experimental studies,

designing the control strategy and writing the manuscript


I have sighted email or other correspondence from all Co-authors confinning their certifying

authorship.

NProf. Firuz Zare 11 Aug. 2011

Name Date

145

CHAPTER 4

A Solid State Marx Generator with a Novel

Configuration

Sasan Zabihi*, Zeynab Zabihi†, Firuz Zare*


Box 2434, Brisbane, Australia.

† Babol Noshirvani University of Technology, Babol, Iran.

Published in: IEEE Transactions on Plasma Science, Vol.39, No.8, pp.1721-

1728, Aug. 2011.

146

Abstract— A new pulsed power generator based on Marx Generator (MG) is

proposed in this paper with reduced number of semiconductor components and

with a more efficient load supplying process. The main idea is to charge two

groups of capacitors in parallel through an inductor and take advantage of

resonant phenomenon in charging each capacitor up to twice as the input voltage

level. In each resonant half a cycle, one of those capacitor groups are charged,

and eventually the charged capacitors will be connected in series and the

summation of the capacitor voltages are appeared at the output of pulsed power

converter. This topology can be considered as a modified MG which works

based on resonant concept. Simulated models of this converter have been

investigated in Matlab/SIMULINK platform and a Lab prototype has been

implemented in a laboratory. The simulation and test results verify the operation

of the proposed topology in different switching modes.

4.1. Index Terms

High voltage stress, Marx Generator, Pulsed power supply, Resonant converter,

Solid state

4.2. Int roduct ion

arx modulator is a popular power supply amongst all pulsed power

technologies. The structure and the control simplicity beside being more efficient

and flexible in supplying various range of applications make it more applicable

in comparison with other topologies like Magnetic Pulse Compressors (MPC)[1],

Pulse Forming Network (PFN)[2], and Multistage Blumlein Lines (MBL)[3].

However a new topology has been recently proposed in [4] based on the positive

buck-boost converter concept and extended in [5] to have more skills in

supplying loads with different demand in pulse shapes. This converter is an

efficient and flexible pulsed power supply having merit to supply wide range of

loads with high repetitive pulses.

A general configuration of the conventional Marx topology is shown in Fig. 4.1.

The initial concept of this topology is charging a number of capacitors in parallel

up to the input voltage level, and then connecting them in series in order to have

the summation of capacitor voltages at the output of the power supply. In this

way, the aggregation of capacitor voltages which is a high level of voltage will

M

147

appear across the load with a fast rising time. As can be seen in Fig. 4.1, each

stage of this generator is composed of a capacitor, a high voltage switch, and two

power diodes.

Recently solid state technology has been utilized in Marx configuration instead

of magnetic switches which were traditionally in use. Insulated Gate Bipolar

Transistor (IGBT), Metal–Oxide–Semiconductor Field-Effect Transistor

(MOSFET), Silicon-Controlled Rectifier (SCR), and Gate Turn-off Thyristors

(GTO) are the semiconductor power switches utilized in recent pulsed power

investigations. A high-voltage bipolar rectangular pulse generator using a solid-

state boosting front-end and an IGBT based H-bridge output stage is presented in

[6] and the generated pulses are intended to be used in algal cell membrane

rupture for oil extraction. In another study, an all-solid-state pulsed-power

generator consists of a Marx modulator based on discrete IGBTs and a magnetic

pulse-sharpening circuit, which is employed to compress the rising edge of the

Marx output pulse is proposed in [7] in order to reduce the influence of relatively

slow turn-on speed of the IGBT on the pulse rise time of the Marx modulator.

An MG topology based on commutation circuit is also proposed in [8, 9] that

utilizes IGBTs and SCRs simultaneously. On the other hand, an experimental

MG with MOSFET switches was used in [10] to generate pulsed output voltages

of up to −1.8 kV in order to produce Pulsed Power Microplasma discharge in N2

gas and N2/NO gas mixture for atmospheric pollution control purposes. In

another application this MG is used for the surface treatment by microplasma of

PEN (polyethylene naphthalate) film using Ar gas and mixtures of Ar with N2

and O2 [11]. Improving Indoor Air Quality (IAQ) through decomposition of

formaldehyde (HCHO) by a microplasma reactor is another subject investigated

in [12] at a discharge voltage of 1.3 kV using a high voltage amplifier and an

MG with MOSFET switches as pulsed power supplies. The semiconductor

technology is also exploited in low power applications of MG such as radar

transmitter and receiver. High power variable nanosecond differential pulses

generators for ground penetrating radar (GPR) systems based on avalanche

transistor and Marx Bank are investigated theoretically and experimentally in

[13]. Using avalanche transistor as the switch of Marx circuit, a new type of all-

solid-state low-power pulse generator is researched in [14] that can generate

148

short unipolar pulses.

By utilizing power semiconductor switches, especially high-voltage IGBTs, as

main switches, Marx modulators have demonstrated many advantages such as

variable pulse length and pulse-repetition frequency, snubberless operation, and

inherent redundancy [7]. This has substantially improved Marx performance in

terms of efficiency and flexibility however there are still other technical issues

which should be considered in order to have a more efficient power supply.

Although configuration and control simplicity has been known as an advantage

for this topology, extra losses caused by using many active and passive

components in charging and discharging passes can be counted as a disadvantage

for this method. Additionally, adjusting the output voltage level with respect to

loads demand is feasible in conventional solid state MG by changing either input

voltage level or the switches duty cycle. In this case, an adjustable dc power

supply is required at the input to vary the input voltage level. It also should be

considered that triggering switches while they are conducting current increases

the switching losses.

A new pulsed power generator based on MG is proposed in this paper that

improves Marx topology in terms of using fewer components and having less

conduction and switching losses. In this proposal, a dc-ac converter is used to

supply a new configuration of switches, diodes and capacitors operating in

resonant modes. Resonant phenomenon is considered in power electronics in

order to minimize switching losses. The concept of the resonant converter has

been developed in such a way that switching transients happen when the current

being conducted passes through the zero level, in order to keep the switching

losses in the power switches to a minimum [15].

Fig.4. 1. A conventional MG

149

4.3. Topology

4.3.1. General configuration

A block diagram of the proposed pulsed power supply shown in Fig. 4.2

comprises an ac-dc converter in the front side, a voltage regulator, a dc-ac

converter and an MG topology with a new configuration. The full bridge rectifier

rectifies the grid voltage and supplies the modulator with a dc voltage. A large

capacitor at the output of the rectifier regulates input voltage fluctuations and

provides the rest of the topology with a smooth and continuous voltage level.

Subsequently in the next stage, this dc voltage is inverted to an alternative

voltage waveform through a single leg inverter. The reason behind using a half

bridge inverter is utilizing fewer active power switches however a full bridge

inverter could supply the MG with more flexibility enabling the symmetrical

adjustment of generated voltage level [16, 17]. This alternative voltage that has

three levels of +Vdc, –Vdc and zero, is applied to an inductor in the entrance of

Marx topology. The configuration presented in this paper as Marx topology uses

a new arrangement of capacitors, power diodes and solid state power switches.

This topology consists of bidirectional diode-capacitor units which are connected

together through two solid state switches with opposite directions. In this

configuration each two stages of MG is composed of two capacitors, two diodes

and two power switches.

AC Grid

Rectifier

AC-DC Converter



Inverter, VSI

DC-AC Converter

Voltage Regulator

V inv(t)+

-

Fig.4. 2. Block diagram of proposed converter with a new Marx configuration,

4.3.2. Switching states

A simplified four-stage MG shown in Fig. 4.3 is simulated in this paper to

investigate its operation features and to carry out further analyzes on its

performance. The approaches can be extended and be considered for a multi-

stage MG. Considering supplied voltage levels through the inverter to the Marx

configuration, +Vdc, –Vdc and zero levels, three principal operation modes are

150

defined for this topology.

Fig.4. 3. Four-stage simulated model of proposed MG,

4.3.2.1. Positive charging mode: (S1:on, S2:off, S3:on, S4:off)

In this switching state given in Fig. 4.4(a), the inverter’s high side switch, S1, is

on while the low side switch, S2, is off. The positive voltage, +Vdc, appears at the

output of the inverter, Vinv(t), (across the inductor and the Marx circuit) due to

conduction of S1. D1, D3 and DS3 (S3’s anti-parallel body diode) are forward

biased in this mode and consequently the capacitors C1 and C3 are in the current

circuit. Therefore the inductor and the capacitors are charged through a resonant

phenomenon. The stored energy in the inductor will then be delivered to the

capacitors however there will not be an opposite energy transmission due to the

presence of the diodes in the resonant circuits. As a result of this half a cycle

resonant between the inductive and the capacitive components of the circuit, the

capacitors are charged up to two times the input voltage while the inductor is

completely discharged. The components behavior during the resonant is

expressed through Equations (4-1)-(4-3).

31 CCCeq +=+ (4-1)

)cos1()(+⋅

−=eq

dcCCL

tVtV (4-2)

)(sin)(+

+

⋅⋅=

eq

dceq

LCL

tV

L

CtI (4-3)

4.3.2.2. Negative charging mode: (S1:off, S2:on, S3:on, S4:off)

In this switching state, the high and the low side inverter switches, S1 and S2 are

turned off and on respectively in order to supply the MG with negative voltage

level, -Vdc. S4 is also switched on simultaneously to complete the circulating

path. The other two diodes, D2 and D4, conduct in this time interval and let the

rest of capacitors, C2 and C4, be charged up to twice the input voltage, however

with reverse voltage polarity. The associated circuit is indicated in Fig. 4.4(b).

151

4.3.2.3. Load supplying mode: (S1:off, S2:off, S3:off, S4:on)

In this stage, the inverter switches, S1 and S2, and also S3 are tuned off. S4 is

switched on in the load supplying mode in order to connect the capacitors in

series and let the aggregation of voltages across all the capacitors appears at the

output of the topology. All diodes are bypassed in this mode, so the energy will

be delivered to the load just through S4. This high voltage pulse which is eight

times the input voltage level, Vdc, is applied to a load connected to the pulsed

power supply. Consequently there will be a break down phenomenon at the load

side due to excitation of the load by this high level of voltage and as a result high

amount of instantaneous power will be delivered to the load. Fig. 4.4(c)

illustrates the relevant circuit to this state.

Fig.4. 4. The switching states of proposed MG (a) Positive charging mode (b) Negative charging mode (c) Load supplying mode,

4.4. Simulat ion resul ts and analyses

The simulation results are provided in this section to verify the validity of

proposed topology. The specifications of the simulated model are presented in

Table 4.1. With respect to the current variation in the resonant circuit, Equation

(4-3), it can be seen that the amplitude of the inductor current relies on the size

of the inductor and the equivalent capacitor. A proportion of the inductor and the

capacitor sizes is selected in the simulations to keep the stored current in an

acceptable range.

TABLE 4. 1. Specifications of the Modelled Circuit


200 V 100 µH 1 µF 10 kHz 11.2 kHz

152

4.4.1. Control strategy

The control simplicity of Marx concept has been relatively maintained in this

configuration. Just an extra switching state has been accommodated that causes

the converter to profit from less power loss while supplying the load. Gates

switching signals, the voltage and the current waveforms of this Marx topology

are shown in Fig. 4.5.

Fig.4. 5. Current and voltage waveforms accompanied by relevant switching signal patterns,

4.4.2. Single shot and repetitively operated results

The simulations are conducted with this model in two situations. A single shot

generator is simulated initially that has been extended then to investigate the

capability of the modulator in generating high repetitive pulses. The attained

simulation results for the single shot and the repetitively operated generator are

presented in Fig. 4.6(a) and 4.6(b) respectively. The input voltage and the

inductor current waveforms are demonstrated in two initial frames shown in Fig.

4.6(a) and 4.6(b), respectively. The capacitors and the output voltages are

depicted in the last frames of Fig. 4.6(a) and 4.6(b). As is apparent in Fig. 4.6(a),

153

C1 and C3 are charged during first half a cycle of the input voltage, Vinv(t), while

C2 and C4 are charged in the next half cycle; both due to the resonant between

the capacitors and the inductor. Ultimately, according to the load supplying

mode of the converter, the aggregation of capacitor voltages which is also

depicted in both last frames of Fig. 4.6(a) and 4.6(b) appears at the modulator

output. The breakdown phenomenon caused by the excitation of the high dv/dt

across the load discharges the capacitor voltages at the last sequence. The

extended simulations for a repetitively operated generator revealed that the

modulator enables to supply the applications that demand high frequent pulses.

0 1 2

-200

0

200

Inpu

t vo

ltage

(V)

0 1 2 3 4 5 6

-200

0

200

Input v

olta

ge(

V)

0 1 2-4

-20

0

20

Ind

ucto

r cu

rren

t(A

)

0 1 2 3 4 5 6-4

-20

0

20In

duct

or c

urr

ent(

A)

0 1 2x 10

-4

-500

0

500

1000

1500

Time(s)(a)

C1,

C2,

C3,

C4 a

nd

ou

tpu

t vo

ltag

es(V

)

C1 voltage

C2 voltage

C3 voltage

C4 voltage

Output voltage

0 1 2 3 4 5 6x 10

-4

-500

0

500

1000

1500

Time(s)(b)

C1,

C2,

C3,

C4 a

nd

outp

ut v

olta

ges(

V)

C1

C2

C3

C4

Output

Fig.4. 6. Simulation results of proposed topology, (a) Single shot, (b) Repetitively operation

4.4.3. The voltage stresses across the diodes and the current through

the power switches

The voltage across and the current through all power diodes and switches are

given in Fig. 4.7(a) and 4.7(b) respectively. The maximum voltage across the

switches and the diodes is four times the input voltage in this case and the

normal currents through the switches and the diodes are 28A and 14A

respectively. There is also a current spark up to 45.5A through the middle

switch, S4, during the load supplying mode which is the delivery current to the

load.

154

Fig.4. 7. The components voltages and the currents (a) Diodes, (b) Switches,

4.4.4. The generated voltage adjustability

As can be inferred from the circuit analyzes, Equations (4-1)-(4-3), and the

simulation results, the inverter’s switching frequency should necessarily be less

than the resonant frequency to have maximum potential voltage generation at the

output of the converter, however the inverter’s switching frequency cannot be

more than the resonant frequency unless the inverter switches have anti-parallel

diodes as shown in Fig. 4.8. In this case the inductor charge and consequently the

capacitor charges will be different in two half cycles unless the inverter switches

duty cycles vary. It indicates that the capacitor’s symmetrical charging and

accordingly the adjustment of the generated voltage level are relatively

impossible in this way. The simulation results, given in Fig. 4.12(a) clarify that

the capacitor residual charges after the load supplying mode will be different in

this case which is due to the asymmetrical initial charges and may cause

malfunction in normal performance of the power supply. To give this feasibility

to the modulator, two hardware solutions are available. The first is using a full

H-bridge inverter instead of the half bridge one and controlling it via unipolar

modulation method [16, 17]. The second is providing a reserve path for the

current that can be accomplished by installing a bidirectional solid state

switching connection shown in Fig. 4.9(a) in the junction of the inverter and the

inductor as given in Fig. 4.9(b). In this way a reserve path will be created for the

current to be flowed through it once both the inverter switches become off during

a resonant half a cycle. That is how the unipolar method can be adopted for a

single leg inverter in order to supply the inverter’s load with zero voltage levels

155

in the middle of positive and negative voltage level intervals. In this way the

stored current in the inductor has sufficient time to be delivered to the capacitors

and the inductor will be free of charge for the next resonant half cycle. These

devices are just triggered for voltage adjustability purposes. The simulation

results given in Fig. 4.10(b) confirm that how practical is this solution in the

symmetrical charging of the capacitors. The two extra switching states according

to this control method are demonstrated in Fig. 4.9(c) & 6.9(d).

D1

C1

L

S1

S2

Vdc D2

C2

+

-

-

+

S3

DS3

D3

C3

D4

C4

+

-Vdc

-

+

S4

DS2

DS1

Fig.4. 8. Using switches with anti-parallel body diodes in the inverter,

(a) (b)

(c) (d)

Fig.4. 9. (a). The bidirectional solid state switching path (b). The proper installation point of the reserve path (c)&(d). The extra switching states associated with the unipolar control method of

the half bridge inverter

0 1 2 3 4 5 6-500

0

500

Inpu

t vo

ltage

(V)

0 1 2-200

0

200

Inpu

t vol

tage

(V)

0 1 2 3 4 5 6-4

-60-40-20

020

Indu

ctor

cur

rent

(A)

0 1 2-4

-20

0

20

Indu

ctor

cur

rent

(A)

156

0 1 2 3 4 5 6x 10

-4

-500

0

500

1000

1500

Time(s)(a)

C1,

C2,

C3,

C4,

and

out

put

volta

ges(

V)

C1

C2

C3

C4

Output

0 1 2

x 10-4

-500

0

500

1000

1500

Time(s)(b)

C1,

C2,

C3,

C4,

and

outp

ut v

olta

ges

(V)

C1 voltage

C2 voltage

C3 voltage

C4 voltage

Output voltage

Fig.4. 10. Simulation results for the converters with (a). anti-parallel body diodes (b). the reserve

path.


A four-stage laboratory prototype set up is implemented to investigate the

concept of this circuit practically and to compare the simulation and the

hardware results. SEMIKRON IGBTs such as SK25 GB 065 and SK50 Gar 065

are used as power switches in the power board. SK25 GB 065, a package of two

IGBTs in series, is utilized as S1 and S2 while two SK50 Gar 065s act as S3 and

S4. TIF28335 DSP is the microcontroller used to run this set up. Skyper 32-pro

(SEMIKRON) gate drives generate the switching pulses to trigger the IGBTs and

provide the necessary isolation between switching signal ground and the power

ground. A general overview of the prototype including the power board, the

control modules and the gate drives is shown in Fig. 4.11. The components

specifications are addressed in Table 4.2.

TABLE 4. 2. Specifications of the Implemented Circuit


30 V 445 µH 10 nF NA 53.3 kHz

157

Fig.4. 11. The hardware set up

Experimental tests are conducted in low voltage range due to the voltage

restrictions of the input dc power supply. The input voltage is adjusted to 30 V

and the resonant frequency determined through the capacitor and the inductor

sizes is 53.3 kHz. The resonant time spam shown by the inductor current in Fig.

4.12(a) and the current amplitude, 200mA, verify the energy exchange process

between the inductive and the capacitive components of the circuit according to

the anticipations. As can be seen in Fig. 4.12(a), the capacitors are charged up to

50V each, and the summation of voltages which is 200 V appears at the output at

the last stage of the operation. The summation of voltages across C1 and Cn (n=4

in this case) is appeared across the load during initial two modes. The rest of


multiple switches) at the third mode. The voltages across S3 and S4 are shown in

Fig. 4.12(b). This simplified model can be extended to have more stages and the

generated voltage level can be increased by supplying the Marx topology

through the rectified grid voltage.

4.6. St ructure and per formance compar ison

In comparison with the conventional solid state Marx topology, the generated

voltage in each stage is twice the input voltage due to the resonant between the

passive components (an inductor and capacitors); therefore the number of needed

stages to generate similar voltage levels is reduced to half of the conventional

Marx stages. Furthermore, even the number of diodes for each stage is decreased

to one diode compared to two diodes in the conventional configuration. Thus,

not only the initial cost will be dropped but also there will be a noticeable power

loss reduction in the capacitors charging process. Although the number of solid

158

state switches remained the same as the conventional MG (one switch in each

stage) the type of power switches can be varied. In conventional MG, all

switching devices should necessarily be fast switches like IGBTs, whereas slow

switches such as GTOs or SCRs can be utilized such as S3 (and its multiple

switches) in this topology. Therefore a fast and a slow switch can be employed in

each two stages.

(a) (b)

Fig.4. 12. Experimental results for (a) The capacitors and the output voltages and the inductor current (b) The voltages across S3 and S4.

On the other hand, the number of solid state switches in discharging path

becomes one switch associated with two stages. This has been two switches for

two stages in former technology. It means that the load supplying process will be

done with less power losses and accordingly higher efficiency. Another

advantage of this topology is utilizing resonant phenomenon as the operation

method and triggering the switches at the instant at which the flowing current

through them is zero. That leads to keep the switching losses in a minimum

possible level. A single-leg inverter is the only extra device utilized in this

method comparing to the previous version. It is quite reasonable by considering

the point that it brought some advantages to this topology. In this converter, the

pulse generation frequency is restricted by the resonant frequency. The smaller L

and Ceq are, the higher repetition rate can be achieved.

4.7. Conclus ion

A new pulsed power converter is proposed in this paper which introduces a novel

configuration as Marx Generator. The whole concept relies on charging two

series of capacitors in parallel over half a cycle resonant and then connecting the

capacitors in series through solid state switches. A half bridge inverter placed in

159

the front side of MG supplies it with alternative voltage levels. Utilizing less

number of semiconductor components, substituting fast solid state switches with

slow switches and consequently having less driving modules in addition to less

switching and conduction losses during capacitors charging and load supplying

processes are the remarkable benefits of this new configuration. Simulations and

tests have been performed and the obtained results verify the proper performance

and operation of the proposed converter in accomplishing desired duties.

4.8. References

[1] J. Choi, T. Yamaguchi, K. Yamamoto, T. Namihira, T. Sakugawa, S. Katsuki, H. Akiyama, “Feasibility Studies of EMTP Simulation for the Design of the Pulsed-Power Generator Using MPC and BPFN for Water Treatments”, IEEE Trans. Plasma Sci., Vol. 34, pp. 1744-1750, 2006.

[2] J. Su, X. Zhang, G. Liu, X. Song, Y. Pan, L. Wang, J. Peng, Z. Ding, “A Long-Pulse Generator Based on Tesla Transformer and Pulse-Forming Network”, IEEE Trans. Plasma Sci., Vol. 37, pp. 1954-1958, 2009.

[3] D. P. Kumar, S. Mitra, K. Senthil, A. Sharma, K. V. Nagesh, S. K. Singh, J. Mondal, A. Roy, D. P. Chakravarthy, “Characterization and analysis of a pulse power system based on Marx generator and Blumlein”, Review Sci. Instr., Vol. 78, pp. 115107-4, 2007.





[8] S. Zabihi, F. Zare, G. Ledwich, A. Ghosh, H. Akiyama, “A new family of Marx generators based on commutation circuits”, IEEE Trans. Dielectr. Electr. Insul., Vol. 18, Issue 4, pp. 1181-1188, August 2011.

[9] S. Zabihi, F. Zare, G. Ledwich, A. Ghosh, H. Akiyama, “A new family of Marx generator based on resonant converter”, 2010 IEEE Energy Conversion Congress and Exposition (ECCE), pp. 3841-3846, 12-16 Sept. 2010.

[10] K. Shimizu, T. Ishii, M. Blajan, “Emission Spectroscopy of Pulsed Power Microplasma for Atmospheric Pollution Control”, IEEE Trans. Ind. Appl., Vol. 46, pp. 1125-1131, 2010.

[11] M. Blajan, A. Umeda, S. Muramatsu, K. Shimizu, “Emission Spectroscopy of Pulsed Powered Microplasma for Surface Treatment of PEN Film”, 2010 IEEE Industry Applications Society Annual Meeting (IAS), pp. 1-8, 3-7 Oct. 2010.

[12] K. Shimizu, M. Blajan, T. Kuwabara, “Removal of Indoor Air Contaminant by Atmospheric Microplasma," 2010 IEEE Industry Applications Society Annual Meeting (IAS), pp. 1-6, 3-7 Oct. 2010.

[13] W. Ren, H. Wang, R. Liu, “High power variable nanosecond differential pulses generator design for GPR system”, 13th International Conference on Ground Penetrating Radar (GPR), pp. 1-5, 21-25 June 2010.

[14] Y. Xuelin, Z. Hongde. B. Yang, D. Zhenjie, H. Qingsong, Z. Bo, H. Long, “4kV/30kHz short pulse generator based on time-domain power combining”, 2010 IEEE International Conference on Ultra-Wideband (ICUWB), Vol. 2, pp. 1-4, 20-23 Sept. 2010.

[15] M. K. Kazimierczuk, A. Abdulkarim, “Current-source parallel-resonant DC/DC

160

converter”, IEEE Trans. Ind. Electron., Vol. 42, pp. 199-208, 1995. [16] S. Zabihi, F. Zare, G. Ledwich, A. Ghosh, “A bidirectional two-leg resonant converter

for high voltage pulsed power applications”, 2009 IET European Pulsed Power Conference, pp. 1-4, 21-25 Sept. 2009.

[17] S. Zabihi, F. Zare, G. Ledwich, A. Ghosh, “A Resonant Based Marx Generator”, 20th Australasian Universities Power Engineering Conference, (AUPEC) 2010, pp.1-5, 5-8 Dec. 2010.

162

I












requirements.


A New Family ofMarx Generators Based on Commutation Circuits

Published in: IEEE Transactions on Dielectric and Electrical Insulation, Vol. 18, Issue 4, pp.

11 81-1188, Aug. 2011.

(

Contributor ~ Statement of contribution Sasan,.Z~i ) Proposed the initial design and conducted simulation studies and data

IH ;/~' analysis, designed the control strategy, implemented hardware set-up and conducted experimental verifications and wrote the manuscript.

11 Aug. 2011 Proposed the initial design and supervised the validity studies

Firuz Zare including: conducting the simulations and experimental studies and writing the manuscript

Gerard Ledwich Aided experimental design, and data analysis

Arindam Ghosh Aided planning the control strategies and writing the paper



I have sighted em ail or other correspondence from all Co-authors confirming their certifying

authorship.

A/Pro f. Firuz Zare 11 Aug. 2011

Name Date

163

CHAPTER 5

A New Family of Marx Generators Based on

Commutat ion Circuits


Akiyama†




Published in: IEEE Transactions on Dielectric and Electrical Insulation, Vol. 18,

Issue 4, pp. 1181-1188, August 2011.

164

Abstract— This paper presents a novel topology for the generation of high

voltage pulses that uses both slow and fast solid-state power switches. This

topology includes diode-capacitor units in parallel with commutation circuits

connected to a positive buck-boost converter. This enables the generation of a

range of high output voltages with a given number of capacitors. The advantages

of this topology are the use of slow switches and a reduced number of diodes in

comparison with conventional Marx generator. Simulations performed for single

and repetitive pulse generation and experimental tests of a prototype hardware

verify the proposed topology.

5.1. Index Terms

Pulsed power, Marx generator, high voltage, resonant converter, positive buck

boost converter, commutation.

5.2. Int roduct ion

HIGH voltage power supplies are required for a wide and increasing range of

applications; and the demand for more flexible and efficient ones is in a fast

growing trend. Pulsed power is an application that frequently demands both a

high voltage stress (dv/dt) and a high voltage magnitude. Currently, pulse

generators are being developed for use in industrial, environmental, medical, and

military applications. However the pulse characteristics such as rise and fall

time, width, repetition rate, voltage and energy levels vary widely in different

applications. Technologies presently used for pulsed power generation include

Marx Generator (MG) [1], Magnetic Pulse Compressor (MPC) [2], Pulse

Forming Network (PFN) [3] and Multistage Blumlein Lines (MBL) [4]. A

recently introduced topology based on the buck-boost converter concept employs

multi switch-capacitor units at the output and has the advantage of being more

flexible and efficient for the generation of high repetitive pulsed power [5, 6].

As shown in Fig. 5.1, the MG uses a simple topology to generate high voltage

pulses. A number of capacitors are charged in parallel from a dc voltage source

up to the input voltage level, and are then reconnected in series to produce a high

voltage across the load. Developments in semiconductor technology saw the

introduction of fast high voltage switches including Insulated Gate Bipolar

Transistors (IGBT) and Metal–Oxide–Semiconductor Field–Effect Transistors

165

(MOSFET). These compact and efficient solid-state devices have replaced bulky,

heavy, costly and inefficient gas and magnetic switching devices (such as park

gap and hydrogen thyratron) used for many years [7-9]. Exploitation of solid-

state technology in the high-voltage generation provides the flexibility of

generating pulses for various load conditions [9, 10]. Although their voltages

ratings do not exceed a few hundred volts, several MGs in the range of hundred

kVs have been designed and implemented using these switches [11, 12]. Several

configurations aiming to ensure the feasibility of generating pulsed power with

adjustable features have been proposed so far. All solid-state MG for generating

bipolar pulses [13], high repetitive pulses [9, 14] and pulses with flexible pattern

[15] have also been designed with different configurations of semiconductor

switches. Although MGs are in use in a wide range of applications, there are still

design and control techniques that can be adopted to improve their performance

in terms of both efficiency and flexibility.

Resonant phenomenon is used in power electronics to minimize switching losses

and the concept of resonant converters has been developed for this purpose.

Switching at the instant at which the conducting current passes through the zero

level, keeps the switching losses in the power switches to a minimum [16]. In

addition, the use of a commutation circuit is another useful technique that utilizes

resonant phenomenon to reverse the polarity of a capacitor voltage. These

techniques are used in the proposed topology to produce pulsed power.

Fig.5. 1. A conventional Marx generator.


5.3.1. Topology

The topology proposed in this paper as an MG, comprises a positive buck-boost

converter that is used as a current source and connected to a number of parallel-

connected diode-capacitor units, as shown in Fig. 5.2. The converter charges the

166

capacitors to a specified voltage and the commutation technique is employed to

connect them in series and to produce the required high voltage at the output of

the system.

As can be seen in Fig. 5.2, a full-bridge diode rectifier connected to the grid

provides a dc voltage for the remainder of the system. The single pulse

applications usually demand high amount of accumulated energy in each shot,

whereas less instantaneous energy is demanded by the repetitive pulse

applications. Although a repetitively operated generator with a moderate peak

power needs less primary stored energy than a single-shot generator [17, 18], the

input source should be able to provide a continuous power supply for the

repetitively operated generator. A three-phase rectifier can be utilized for this

purpose in order to provide primary uninterruptable energy supply from the grid.

AC Grid

Rectifier

AC-DC Converter

Modified positive Buck-Boost Converter

Novel Marx topology(Two leg diode-capacitor units)

220 V50 Hz

Fig.5. 2. Block diagram of the new Marx topology

D1 D4

C1 C4

L1

Vdc

D2

C2

D3

C3

S3

S4

+

-

+

-

+

-

+

-

D5

S1

S2

SCR1

Df

SCR2


L2 L3

+

-

+

-

Fig.5. 3. Circuit diagram of the proposed topology.

A positive buck-boost converter is considered in the next stage to provide the

flexibility of boosting the voltage to any desired level. This converter is

connected to the proposed Marx topology through a power switch that

disconnects the Marx topology from the rest of the circuit after charging the

capacitors.

A detailed circuit diagram of the above topology, including the diode-capacitor

units is shown in Fig. 5.3. As will be seen, the second and its multiple legs

contain a resonant circuit that includes an inductor and a slow semiconductor

switch, Silicon-Controlled Rectifier (SCR). The small inductor is connected to

the capacitor through the SCR to change the polarity of the capacitor voltage.

Such energy exchange process, known as commutation [19], makes series

167

connection of the capacitors feasible. In the proposed method, the polarity of the

capacitor voltages is alternately inverted and subsequently a connection between

the units using a fast switch, S4 is sufficient to provide series connection of the

capacitors. The number of units can readily be increased and a higher voltage can

be produced at the output of the system.

5.3.2. Switching modes

The switching modes of the proposed converter for single pulse generation

consist of the four states which are shown in Fig. 5.4.

5.3.2.1. First mode: Inductor charging mode

In the first state, shown in Fig. 5.4(a), the main inductor, L1, located at the input

of the converter is charged through S1 and S2, while S3 ensures that the rest of the

circuit is disconnected. The charged inductor acts as a current source for the rest

of topology in the subsequent modes. The current level defining the energy

stored in the inductor can be controlled based on the duty cycles of the switches

S1 and S2.

Let us assume that all the semiconductor devices including IGBTs, SCRs and

diodes are ideal components. Then the voltage drop across each of them is zero

when it conducts. According to (7-1), the voltage across the inductor is the input

voltage and the time to charge the inductor to a desired current level, imax, and the

ultimate inductor energy are given in (7-2), and (7-3) respectively.

Fig.5. 4. Switching states of proposed MG for single pulse generation.

dt

diLVV L

Ldc1

1 1 ⋅== (5-1)

dcL V

iLt max

1 ⋅=∆ (5-2)

2max12

11

iLEL ⋅= (5-3)

168

5.3.2.2. Second mode: capacitors charging mode

In the second mode, shown in Fig. 5.4(b), S1 and S2 are turned off and S3 is

turned on simultaneously to deliver the energy stored in the inductor into the

capacitors, and to convert it from electromagnetic to electrostatic form.

Diodes, D1, D2, D3, D4 and D5 conduct the inductor current and charge the

capacitors C1, C2, C3 and C4 to the desired voltage level with positive polarity. In

this state, the buck-boost freewheeling diode, Df, conducts the current to create a

current loop. Assuming the voltage drop across the diodes is negligible, and the

equivalent capacitance of the four capacitors is Ceq, the relation of exchanged

energy between the inductor and the capacitors is as given in (7-5). Instants t1

and t2 respectively are the instants at which the inductor current is fully charged

or partly discharged and can be realized by turning off and on the gate signals for

S3. If the energy stored in the inductor is completely delivered to the

capacitors )0)(( 21=tiL , the final voltage of the capacitors can be expressed as in

(6).

4321 CCCCCeq +++= (5-4)

222

211 2

1))()((

2

111 CeqLL VCtitiL ⋅=−⋅ (5-5)

eqLC C

LiV 1

14,3,2,1⋅= (5-6)

Alternatively, if the inductor current, iL1, is assumed constant (i.e., a large

inductor is used) to provide a permanent current source for a repetitively

operated generator, the voltage across the capacitors can be calculated as

follows.

dt

dVCii eq

eq

C

eqLC ⋅==1

(5-7)

eq

CLC C

tiV

∆⋅=

1max (5-8)

where Ct∆ is the time required to charge the capacitors to maxCV .

5.3.2.3. Third mode: commutation mode

As shown in Fig. 5.4(c), in the third switching state, S2 and S3 are respectively

turned on and off. It is expected that for single shot generator, the inductor will

not have been fully discharged during the second mode (i.e. it is working in a

continuous conduction mode, CCM) and its current needs to be circulated in a

circuit. For a repetitively operated pulse generator, the converter performs in a

169

CCM, and the inductor current never falls to zero. Therefore S2 is turned on to

enable the remaining current to circulate through Df. Simultaneously, switch S3 is

turned off to separate the proposed topology from the buck-boost converter. If

the inductor current needs to be increased to either keep the inductor current

continuous at a specific level in a repetitive application or to charge the inductor

for the next switching cycle in a nonrepetitive application, S1 can be turned on

(Fig. 5.5(a) and 5.5(b)).

The next step is to change the polarity of the second (and any further) capacitor

voltages. In this mode, the SCRs are turned on to change the voltage polarity

across C2 and C4. Resonance occurs between C2 (C4) and L2 (L3), during which

the stored energy in the capacitor is delivered to the small inductor until the

capacitor voltage becomes zero. At this instant, the inductor current reaches its

peak value and the current recharge the capacitor to a reversed polarity. The

energy exchange between the inductor and the capacitor is an inherent

characteristic of the components and is a key factor of the commutation circuits.

Fig.5. 5. Extra switching states of proposed MG for repetitive pulse generation.

Although it appears at first sight that the negative voltage across S4 is almost

twice the capacitor voltage and must be withstood by the switch in this state, the

diode D5 provides the necessary protection by sharing this voltage.

5.3.2.4. Fourth mode: pulse generation mode

Eventually in the final switching state the capacitors are connected in series by

turning on switch S4. This mode begins when the voltage polarities of C2 and C4

are changed and both SCR1 and SCR2 are turned off. By turning on switch S4, the

summation of the capacitor voltages appears across the output of the generator.

In the beginning of this state, the inverse voltage across D5 is almost twice the

capacitor voltage which should be handled by the diode. The relevant power

circuit is shown in Fig. 5.4(d).

170

5.3.3. Control strategy

Two separate control algorithms (switching procedures) are adopted, one for

single pulse generation and the other for repetitive pulse generation. The control

simplicity is an advantage of MGs, and is almost maintained in this

configuration. Instead of the two simple switching states in the conventional

MG, this topology has four switching steps for each pulse generation cycle while

acting as a single pulse generator. These operation modes are necessary due to

the design requirements of the MG, and the gate signals for the power switches

are generated with respect to these modes. In the inductor charging mode, S1 and

S2 are switched on to charge the inductor and the duty cycles of S1 and S2 are

determined through the level of inductor current based on the required storage of

energy in the inductor. A complimentary gate signal is used to trigger S3 on and

off, and therefore S3 is off during this mode as well as S4 and the SCRs. In the

next switching state, the capacitor charging mode, S1 and S2 are switched off

once the inductor is charged up to a certain level. S3 is switched on

simultaneously to conduct the inductor current and so charge the capacitors. In

addition, S3 is switched off when the inductor current fall below a defined level

and the inductor needs to be charged for the next supplying cycle. At this point,

S1 and S2 are turned on to again charge the inductor. As can be seen in Fig. 5.6,

the gate signals for S1, S2 and S3 are determined by the inductor current. In the

commutation mode, the switches in the commutation circuits, SCR1 and SCR2 are

turned on to reverse the voltage polarities across the relevant capacitors (C2 and

C4). The switching signals of the SCRs are determined by the capacitors voltage.

S2 and S3 are switched on and off respectively in this mode, whilst S1 can be

either on or off. Once these capacitors are fully recharged at a negative polarity,

the switching signal is sent to S4 to turn it on and to connect the diode-capacitor

units. S4 will be switched off after the generated pulse is applied to the load and

the capacitors are discharged. The turn off time for S4 can therefore also be

specified by monitoring the discussed capacitor voltage. The above logic

procedure indicates that the control mechanism of the proposed topology can be

designed and implemented by sampling two circuit parameters, the current of the

input inductor and the voltage of the second capacitor. This makes the control

strategy both simple and effective.

171

Input voltageOutput voltageC1 voltage

C2 voltage

C3 voltage

C4 voltage

Fig.5. 6. The capacitor voltages and the gate drive waveforms of the converter.

The gate drive waveforms of all the switches used for the topology of a single

pulse generation in a cycle are shown in Fig. 5.6.

The control algorithm is more complicated for repetitive pulse generation due to

the greater number of safety issues. The flowchart in Fig. 5.7 shows how the

decisions are made for the topology to supply a load with repetitive pulses. To

charge the capacitors alternatively at a high repetition rate, the input inductor

(L1) current should be kept relatively constant at a specific value. A band is

therefore defined to switch S1 and S2 on and off and to keep the inductor charged

steadily. S3 is turned on and off with respect to both the inductor current and the

capacitor voltages. S4, SCR1 and SCR2 are turned on and off as in the former

strategy. The repetitive control strategy contains two switching modes more than

the single shot strategy.

5.4. Simulat ion resul ts

Simulation results for the proposed MG with both single and repetitive pulse

generations are shown in Fig. 5.8 and Fig. 5.9. The circuit parameters used in the

simulations are recorded in Table 5.1.

TABLE 5. 1. The specifications of simulated models

Single Pulse Vin L1 L2 L3 Ci R1Load R2Load

200 V 1 µH 100 nH 100 nH 10 nF 1 MΩ 10 Ω Repetitive Pulses

Vin L1 L2 L3 Ci R1Load R2Load 200 V 433 µH 1 µH 1 µH 10 nF 1 MΩ 10 Ω

172

Fig.5. 7. Control flowchart for a modulator with the repetitive pulse generation function.

The inductor currents and the capacitor voltages in Fig. 5.8 for the single shot

MG can be divided into different time frames according to the switching states.

The first time interval (0 to 0.5 µs) is for the charging state of the inductor up to

100 A. The next time interval (from 0.5 to 0.8 µs) is for the charging state of the

capacitors in the second switching mode. In this mode the inductor current

circulates through all four capacitors and charges them to 500 V. The inductor

current falls to less than 5 A in this switching mode and is maintained in this

level.

As can be seen in Fig. 5.8(b), the voltage polarities of C2 and C4 are reversed

between 0.8 to 0.9 µs, due to the oscillations between the passive components of

the commutation circuits. The inductor currents are also shown in Fig. 5.8(a). All

four capacitors are connected together in series at 0.9 µs by turning on the switch

S4, to generate a voltage at the output of the MG almost four times each capacitor

voltage. To investigate the circuit behavior when supplying a load, a 10Ω

resistor is connected to the output of the MG.

173

0 0.2 0.4 0.6 0.8 1 1.2

x 10-6

0

50

100

150

(a)

L 1, L 2&

L3 c

urr

en

ts (

A)

L1 Current

L2 Current

L3 Current

0 0.2 0.4 0.6 0.8 1 1.2

x 10-6

-500

0

500

1000

1500

2000

(b)Time(s)

C1,

C2,

C3,

C4,

inpu

t &

out

put

volta

ges

(V)

Input voltageOutput voltageC

1 voltage

C2 voltage

C3 voltage

C4 voltage

Fig.5. 8. Simulation results for the proposed converter (single pulse).

6.2 6.4 6.6 6.8 7

x 10-5

19

19.5

20

20.5

(a)

L 1 curr

ent (A

)

6.2 6.4 6.6 6.8 7

x 10-5

0

50

100

150

(b)

L2&

L3 c

urr

ents

(A

)

L2 Current

L3 Current

6.2 6.4 6.6 6.8 7

x 10-5

-500

0

500

1000

1500

2000

(c)Time(s)

C1,

C2,

C3,

C4,

inpu

t &

out

put vo

ltages

(V)

Input voltageOutput voltageC1 voltage

C2 voltage

C3 voltage

C4 voltage

Fig.5. 9. Simulation results for the proposed converter (repetitive pulses).

174

As anticipated, the output voltage falls as all the capacitors lose charge and

therefore voltage, as illustrated in Fig. 5.8(b).

In addition to the change in the component sizes evident in Table 5.1, the current

level in the input inductor, L1, and the switching modes sequence are also

changed in order to adjust the MG for a high repetitive pulse generation. The key

issue in this case is the input inductor size, which should be larger than for single

pulse generation. The higher input voltage can provide the inductor with faster

charging and increase the modulators pulse generation repetition. As shown in

Fig. 5.9(a), the inductor current fluctuations between 20 A and 19.5 A are due to

charging by the input voltage and discharging through the capacitors.

The voltage across and the current through all switches are given in Fig. 5.10. In

this case, medium voltage rate IGBTs can be used as S1, S2, S3 and S4. Simple

SCRs can also be utilized to withstand against the normal range of voltages. All

the fast switches are at a reasonable current level except for S4. In this case, the

current peak is 200 A, although it can be even higher for an MG with more

capacitive units. Although the current level in the supplying mode is critical for

solid-state power switches, semiconductor devices are available which can

handle this level of current, specifically when it flows in the form of pulses.

Normally, when dealing with pulsed currents, an operating level higher than the

rated dc level is possible for solid state components. The analyses of the voltage

and the current of switches in this model reveal that with a proper selection of

components, the proposed topology can accomplish all the anticipated functions.

175

Fig.5. 10. The switches voltages and currents.

5.5. Exper imenta l ver i f ica t ion

A simple four-stage MG is implemented to study the proposed configuration

practically. SEMIKRON IGBTs and SCRs are used to arrange the hardware.

Skyper 32-pro gate drives which are compatible with these switches are used to

provide the gate drive signals needed for triggering the switches. NEC 32-bit

64MHz V850/IG3 micro-controller is used to control the gate drives. The

specifications of the circuit are listed in Table 5.2. The experimental set up is

shown in Fig. 5.11.

TABLE 5. 2. The specifications of implemented hardware

Vin L1 L2 L3 Ci

20 V 433 µH 220 µH 220 µH 10 nF

176

Fig.5. 11. The experimental set up.

The experimental results obtained are shown in Fig. 5.12(a) and 5.12(b). The

capacitor and the output voltages and the input inductor current are shown in Fig.

5.12(a). The operation modes, including the inductor and capacitors charging

modes, followed by the commutation and the pulse generation modes can be

distinguished in this figure. The summation of voltages across C1 and Cn (n=4 in

this case) appears across the load during third (commutation) mode. The rest of


multiple switches) at the fourth (pulse generation) mode. The energy exchange

process in the commutation circuits is illustrated in Fig. 5.12(b), through

depicting the involved capacitor (C2 and C4) and inductor (L2 and L3) voltage and

current waveforms.

(a) (b)

Fig.5. 12. The experimental results

5.6. Design features and the component d iscussion

There are a number of issues which should be considered in the design process.

Firstly, the inductor sizes should be compatible with the capacitor sizes in the

commutation circuits in order to prevent inrush currents. Since the duration of

177

oscillations in the resonant phenomena is defined with respect to the capacitor

and the inductor sizes, according to equation (5-9), small components are

preferred to reduce the energy exchange period and to give the flexibility of

generating pulsed power with a higher repetition rate. Therefore, the selection of

inductors is a trade-off between having the shortest resonance period and having

a reasonable current flowing through the commutation circuits. As a result, the

inductors are selected sufficiently large to control the current peaks.

CLf r

⋅=

π2

1 (5-9)

The number of stages, and consequently, the number of capacitors are

determined by the voltage required by the load, whereas the capacitor sizes are

determined by the required energy. On the other hand, the stored energy in the

input inductor must be sufficient to charge the capacitors to a defined level. A

balance between the inductor size and its current level is necessary to give the

required energy.

Electromagnetic interfere (EMI) is the other issue which should be taken into

consideration when using switching equipment that trigger devices in a high

frequency. The electromagnetic fields which are generated due to this high

repetition rate and switching transients cause interference that may influence

other equipment like optical receivers. To prevent such incidents, all the current

loops in the printed circuit board (PCB) should be laid out with minimum stray

inductance. Using planar busbar configuration is an effective method to

minimize the magnetic fields and the radiated noises in the hardware set up [20].

The collector-emitter voltage, VCE, of power switches should be adequate to

handle the voltage across the switch. Each SCR should withstand the voltage

across the related capacitor. D5 blocks the circuit of C2, S4 and C3, as shown in

Fig. 5.13. Although the voltage sharing across S4 and D5 in the third mode is not

predictable, due to the different characteristics of these components, simulation

results in Fig. 5.10 indicate that they share the voltages across C2 and C3 almost

equally. The summation of C2 and C3 voltages is located across D5 once S4 is

trigged on at the beginning of fourth mode. Therefore D5 is required to withstand

twice the capacitor voltage, in order to block the circuit in the fourth mode,

whereas S4 rating is equivalent to the charge across one capacitor. Diodes, D1,

D2, D3 and D4, also should be able to block a capacitor voltage.

178

The significant achievement of this design is the substitution of fast IGBTs with

slow SCRs. Although, SCRs are slower devices, they require fewer driving

modules rather than IGBTs. In addition, fewer diodes are used in this design.

This leads to reductions in the cost, losses, volume, weight, and system intricacy.

To generate an output voltage that is ten times the input voltage, the proposed

topology requires far fewer modules and components than a conventional MG. A

conventional MG will involve ten fast power switches, and twenty power diodes,

whereas, a ten-stage proposed MG will need only four fast power switches, five

slow switches, and fourteen power diodes. Besides employing fewer active

power elements (such as solid-state switches and diodes), the switching and

conduction power losses will be markedly reduced due to having fewer

components in the discharging path. Furthermore, the whole converter has the

flexibility to increase the generated voltage level through a lower input voltage.

By adjusting the inductor current level, the stored energy in the inductor can be

controlled and the level of voltage in the capacitors can be either boosted or

decreased.

Fig.5. 13. The switch and the diode that connect diode-capacitor units compose a circuit.

5.7. Conclus ion

A new family of Marx generator is proposed in this paper based on the parallel

connection of diode-capacitor units and commutation circuits. This converter

aims to generate high voltage with a topology composed of fast and slow solid-

state switches and it is able to generate a flexible high voltage level at the output

of the converter with a definite number of capacitors. This topology generates

high voltage with fewer components than a conventional MG. The simulation

and the experimental results verify the proposed topology and control in

satisfaction of all expected functions.

5.8. References

[1] T. Heeren, T. Ueno, D. Wang, T. Namihira, S. Katsuki, H. Akiyama, “Novel Dual Marx Generator for Microplasma Applications”, IEEE Trans. Plasma Sci., Vol. 33, pp. 1205-1209, 2005.

179

[2] J. Choi, T. Yamaguchi, K. Yamamoto, T. Namihira, T. Sakugawa, S. Katsuki, H. Akiyama, “Feasibility Studies of EMTP Simulation for the Design of the Pulsed-Power Generator Using MPC and BPFN for Water Treatments”, IEEE Trans. Plasma Sci., Vol. 34, pp. 1744-1750, 2006.

[3] J. Su, X. Zhang, G. Liu, X. Song, Y. Pan, L. Wang, J. Peng, Z. Ding, “A Long-Pulse Generator Based on Tesla Transformer and Pulse-Forming Network”, IEEE Trans. Plasma Sci., Vol. 37, pp. 1954-1958, 2009.

[4] D. P. Kumar, S. Mitra, K. Senthil, A. Sharma, K. V. Nagesh, S. K. Singh, J. Mondal, A. Roy, D. P. Chakravarthy, “Characterization and analysis of a pulse power system based on Marx generator and Blumlein”, Review Sci. Instr., Vol. 78, pp. 115107-4, 2007.



[7] Y. Wu, K. Liu, J. Qiu, X. Liu, H. Xiao, “Repetitive and High Voltage Marx Generator Using Solid-state Devices”, IEEE Trans. Dielectr. Electr. Insul., Vol. 14, pp. 937-940, 2007.

[8] J. H. Kim, B. D. Min, S. Shenderey, G. H. Rim, “High Voltage Marx Generator Implementation using IGBT Stacks”, IEEE Trans. Dielectr. Electr. Insul., Vol. 14, pp. 931-936, 2007.

[9] L. M. Redondo, J. F. Silva, “Repetitive High-Voltage Solid-State Marx Modulator Design for Various Load Conditions”, IEEE Trans. Plasma Sci., Vol. 37, pp. 1632-1637, 2009.

[10] L. M. Redondo, H. Canacsinh, J. F. Silva, “Generalized solid-state marx modulator topology”, IEEE Trans. Dielectr. Electr. Insul., Vol. 16, pp. 1037-1042, 2009.

[11] T. Heeren, J. T. Camp, J. F. Kolb, K. H. Schoenbach, S. Katsuki, H. Akiyama, “250 kV Sub-nanosecond Pulse Generator with Adjustable Pulse-width”, IEEE Trans. Dielectr. Electr. Insul., Vol. 14, pp. 884-888, 2007.

[12] J. H. Kim, M. H. Ryu, B. D. Min, G. H. Rim, “200KV pulse power supply implementation”, 2007 European Conference on Power Electronics and Applications, pp. 1-5, 2-5 Sept. 2007.

[13] H. Canacsinh, L. M. Redondo, J. F. Silva, “New solid-state Marx topology for bipolar repetitive high-voltage pulses”, Power Electronics Specialists Conference (PESC) 2008, pp. 791-795, 15-19 June 2008.



[16] M. K. Kazimierczuk, A. Abdulkarim, “Current-source parallel-resonant DC/DC converter”, IEEE Trans. Ind. Electron., Vol. 42, pp. 199-208, 1995.

[17] H. Akiyama, S. Sakai, T. Sakugawa, T.Namihira, “Invited Paper - Environmental Applications of Repetitive Pulsed Power”, IEEE Trans. Dielectr. Electr. Insul., Vol. 14, pp. 825-833, 2007.

[18] T. Sakugawa, D. Wang, K. Shinozaki, T. Namihira, S. Katsuki, H. Akiyama, “Repetitive short-pulsed generator using MPC and blumlein line”, Digest of Technical Papers, 14th IEEE International Pulsed Power Conference, (PPC) 2003, pp. 657-660, 15-18 June 2003.

[19] S. Zabihi, F. Zare, G. Ledwich, A. Ghosh, “A bidirectional two-leg resonant converter for high voltage pulsed power applications”, 2009 IET European Pulsed Power Conference, pp. 1-4, 21-25 Sept. 2009.

[20] F. Zare, G. F. Ledwich, “Reduced layer planar busbar for voltage source inverters”, IEEE Trans. Power Electron., Vol. 17, pp. 508-516, 2002.

181

I



I . they meet the criteria for authorship in that they have participated in the conception,







5. they agree to the use of the publication in the student' s thesis and its publication on the


requirements.


Using a Current Source to Improve Efficiency of a Plasma System

Published in the proceedings of: I ih IEEE Pulsed Power Conference, PPC 2009, Washington

DC, USA.

Contributor Statement of contribution Sasan Zabihi ~ Proposed the initial design and conducted simulation studies and data

V./ analysis, designed the control strategy, implemented hardware set-up and conducted experimental verifications and wrote the manuscript.






authorship. ~ A/Pro f. Firuz Zare ""'

_N_am __ e__________________ ~ ----D--a-te ____________ __

11 Aug. 2011

182

CHAPTER 6

Using a Current Source to Improve Eff iciency

of a Plasma System

Sasan Zabihi*, Firuz Zare*, Hidenori Akiyama†




Presented and published at: 17th IEEE Pulsed Power Conference, PPC 2009,

Washington DC, USA.

183

Abstract— This paper presents the possibility of utilizing a current source

topology instead of a voltage source as an efficient, flexible and reliable power

supply for plasma applications. A buck-boost converter with a current controller

has been used to transfer energy from an inductor to a plasma system. A control

strategy has also been designed to satisfy all the desired purposes. The main

concept behind this topology is to provide high dv/dt regardless of the switching

speed of a power switch and to control the current level to properly transfer

adequate energy to various plasma applications.

6.1. Int roduct ion

Excessive power losses in plasma generators reveal that a review on power

supply properties is necessary to improve the efficiency of plasma systems. A

high dc voltage level with high dv/dt imposed over electrodes, causes plasma

reaction resumption and during this period of time, an undefined current flow

through the plasma system may lead to heat generation. Thus, a control on

current flow seems to be crucial for power consumption and improving

efficiency. Conventionally, magnetic pulse compressors [1] and multistage

Blumlein lines [2] were utilized to generate high voltage for pulsed power

applications. Even today, there is wide usage of voltage source topologies such

as, either diode-capacitor multipliers, including Dickson charge pump [3] and

Cockroft-Walton multiplier [4, 5] or Marx Generators [6, 7] to feed plasma

applications. As shown in Fig. 6.1(a), these configurations can be modeled with

a charged capacitor which is connected in parallel to the electrodes energizing

the material to form plasma. Fig. 6.1(b) shows a Marx generator as the most

popular circuit which works based on the idea of charging capacitors in parallel

and getting higher voltage during discharging mode while they are connected

together in series. In these configurations, the capacitor’s high voltage over

electrodes resumes plasma formation. In the process of plasma generation, a kind

of low impedance happens across the electrodes which may discharge the

capacitors acting as a voltage source. This high current may cause the capacitor

to get fully discharged and a considerable proportion of the stored energy to get

lost. This phenomenon naturally consumes extra power while there is no control

on the circuit to stop current flow through the low impedance circuit. In fact,

there is no necessity to feed the electrodes at the time of this incident. The

184

system efficiency will be influenced by an effective and smart current source

which controls and limits power flow to a plasma system.

On the other hand, a current source has been known as the most appropriate

power supply for capacitive loads, while voltage source is suitable for inductive

loads. As shown in many previous investigations, plasma acts mainly as a

variable capacitive load and it supports the idea of supplying it via a current

source topology.

High Voltage

GeneratorEnergy Storing

Element

Plasma

Container

(a) (b)

Fig.6. 1. (a). Block diagram of high voltage source topology (b). N-stage Marx generator

A modified positive buck-boost converter operating in a discontinuous

conduction mode with a current controller is a novel suggestion for feeding

plasma systems. This topology can be an appropriate substitution for a voltage

source configuration, in which it can be modeled as a shunt current source. In

fact, the main idea of this circuit is to control the current source magnitude to

control delivered energy to a plasma system. The detailed concept of the

topology including circuit elements, configuration, control blocks and operation

modes are explained in the following section.

Curr

ent

Sourc

e

Load

Curr

ent

Sou

rce

Fig.6. 2. (a). A circuit diagram of current source topology (b)&(c). Operation modes of the

current source topology supplying plasma load

185

6.2. Current source topology

As mentioned, this paper aims to investigate the advantages of a current source

for different plasma applications. Therefore, a non-linear resistive and capacitive

load is used to model plasma reaction behavior approximately with features close

to its real characteristics. As shown in Fig. 6.2(a), a small capacitor connected in

parallel to a big resistor is used to model a plasma system, while the resistor will

be substituted by a small one to simulate the conducting mode. In order to

simulate reactions taking place in a plasma system, the switch S3 turns on

frequently, and connects the small resistance to the system.

As can be seen in Fig. 6.2(a), the first switch, S1, controls the current flow

through the inductor based on a hysteresis band method. A detailed description

of the hysteresis current control method is described in Fig. 6.3. The stored

energy in the inductor can potentially be delivered to the plasma system. Hence,

charged inductor representing the current source for a plasma load. The next

switch, S2, controls delivered energy to the plasma system and the current level

through the inductor. When S2 is turned on, the voltage source, Vin, charges the

inductor and when the switch is turned off, the stored energy will be transferred

in to the load.

One of the most significant advantages of this configuration is having control

over the load current and voltage so that it is possible to stop power supply at

unnecessary times and prevent power loss by a feedback from the load. The other

benefit of this topology is the simplicity of the circuit, which consists of a dc

voltage source, two switches, an inductor and a diode. In this topology, there is

no extra capacitor in parallel with the plasma system as the current source can

generate significant dv/dt and delivers energy to the load. It is also possible to

add a small capacitor in parallel with the plasma system and the capacitance

value can be selected based on a pulsed power level.

6.2.1. Hysteresis current controller

To control the inductor’s current level and determine the switching signal, a

hysteresis current control is used. Supposing that the second switch is turned on,

the first switch can control the current through the inductor. When the first

switch, S1 is turned on, the voltage source is connected to the inductor and it

186

raises the inductor current. When the inductor current exceeds the upper

band, )2

(I

I ref

∆+ a comparator detects the current level and turns off the switch, S1,

in order to keep the current between the bands. In this case, the inductor current

circulates through the diode D1 while the voltages of diode D1 and S1 appear

negatively across the inductor and slightly discharges the inductor as shown in

Fig. 6.3. When the load current crosses the lower band )2

(I

I ref

∆− , S1 is turned on

again and this procedure continues frequently.

)2

(I

I ref

∆+

)2

(I

I ref

∆−refII∆

1T∆ 2T∆

Fig.6. 3. Hysteresis band current control

This control method provides an adjustable and controllable current source to

supply a plasma system. The switching frequency of the current source converter

should be as low as possible to minimize switching power losses. As it is

obvious in Eq. (6-1), switching frequency is defined as a function of the

inductance value, L, the voltage over the inductor, VL, and the hysteresis band

height, ∆I.

T

ILVL ∆

∆= & LV

ILT

∆⋅=∆ (6-1)

When S1 is on ( onTT =∆ 1 ):

2SinL VVV −= & 2Sin

onVV

ILT

−∆⋅= (6-2)

And when S1 is switched off ( offTT =∆ 2 ):

21 SDL VVV −−= & 21 SD

off VV

ILT

+∆⋅= (6-3)

=

+∆⋅+

−∆⋅=

+==

212

111

SDSin

offonswSW

VV

IL

VV

ILTTTf

(6-4)

187

IL

VV

VIL

VVV

VVIL

VVVV SD

in

SDin

Din

SDSin

∆⋅+

≅⋅∆⋅+⋅

≅+⋅∆⋅

+⋅−2121

1

212)(

)(

)()(

To reduce switching losses, the discharging time should be increased and this

means that the negative voltage drop across the inductor should be reduced. It is

also possible to increase the switching time by increasing the hysteresis band

height, but the current ripple should be taken into account. The inductance value

and the input voltage can also determine the switching frequency. However, for a

plasma system, maximum pulse per second which is supported by the current

source is an important factor. This means that if the inductor is discharged, a

minimum time to charge the inductor from zero current to maximum current

determines the charging current. The other factors relate to cost and size issues to

select an inductor. This means that a small inductor is preferred.

6.2.2. Voltage level and switching stress

The most important parameter in a plasma generation is dv/dt and high voltage

level. The high voltage will be generated when the buck-boost converter works

in a discontinuous conduction mode. The voltage over the electrodes will be

dramatically increased since the electrodes have a capacitive characteristic and

can be modeled as a small capacitor. In this mode, a considerable amount of

current would be pumped into the capacitor from the inductor in a short period.

Thus, if the inductor gets fully discharged through the capacitor, the voltage

across the plasma system can be found to be as follows:

22

2

1

2

1CVLI = &

C

LIV .max = (6-5)

The dv/dt across the capacitor depends on the current through the capacitor. In

fact, a power switch, S2, has a minimum switching time, but dv/dt across the

capacitor can be expressed as follows:

dt

dvCtiC =)( &

C

ti

dt

dv C )(= (6-6)

To model a plasma load, a small capacitor is connected in parallel to a big

resistor while both are connected to a small resistance through a switch. Hereby

the resistivity and conductivity of the load during plasma reaction will be

simulated for the power supply.

Fig. 6.4 shows the capacitor voltage and current waveforms. It can be noted that

based on Eq. (6-6), at different current levels, different (dv/dt)s will be obtained.

188

This shows that in this method, the voltage stress over electrodes may change

while the flowing current sharply rises due to the switching transients. This is

accurately illustrated in Fig. 6.5 which shows how different voltage stresses can

be achieved while the current is increasing.

As demonstrated in Fig. 6.5, different current levels that flow through the load

generate different (dv/dt)s over the electrodes. Therefore, the system makes a

wide range of variable voltage stresses available during the current’s rise in

switching transient modes. This can be counted as one of the most remarkable

benefits of the current source topology in plasma systems.

In Fig. 6.5, we have simulated the converter at different inductor current

magnitudes in order to explain the effect of capacitor current on (dv/dt).

Generating a fast switching transient at high voltage is very challenging, but

according to Eq. (6-6), (dv/dt) depends on iC(t) and capacitance value, CLoad.

During a switching transient, the current through the switch S2 decreases while

through the capacitor it is increased. For example, if CLoad is 50pF and the

capacitor current is 1A, according to Eq. (6-6), (dv/dt) is 20 V/ns. When the

capacitor current is increased, (dv/dt) is increased. In fact, with a normal switch

(not very fast) we can create a significant (dv/dt) across the load when a total

capacitance connected to the current source (in this case, the load capacitance) is

not very high. The variation of (dv/dt)s in the transient of switching can be seen

in Table 6.1.

0.01 0.012 0.014 0.016 0.018 0.02 0.022750

800

850

900

950

1000

1050

Cap

aci

tor

volta

ge(

V)

0.01 0.012 0.014 0.016 0.018 0.02 0.022-100

-50

0

50

100

150

Time(s)

Ca

paci

tor

curr

ent

(A)

Fig.6. 4. Voltage and current of modelled capacitor with 100A inductor current

189

0.002995 0.0029975 0.003 0.0030025 0.003005 0.0030075 0.00301-1000

0

1000

2000

Time(s)

Cap

acito

r vo

ltage

(V)

dV/dt

0.002995 0.0029975 0.003 0.0030025 0.003005 0.0030075 0.00301-100

-80

-60

-40

-20

0

20

40

60

80

100

120

Time(s)

Cap

acito

r cu

rren

t(A

)

capacitor voltage(dV), iL=20Acapacitor current(A), iL=20Acapacitor voltage(dV), iL=40Acapacitor current(A), iL=40Acapacitor voltage(dV), iL=60Acapacitor current(A), iL=60Acapacitor voltage(dV), iL=80Acapacitor current(A), iL=80Acapacitor voltage(dV), iL=100Acapacitor current(A), iL=100A

Fig.6. 5. Voltage and current of modeled capacitor with 20, 40, 60, 80 and 100A inductor

currents

TABLE 6. 1. Variation of (dv/dt)s in the transient of switching

Capacitor current (A) dv/dt (V/ns) 20 0.17 40 0.28 60 0.45 80 0.6 100 0.8

6.2.3. Power losses issue

As already discussed, after plasma forming, the material resistance between the

reactor electrodes markedly falls to an insignificant value which causes

impractical power consumption inside the plasma system. To stop this, a voltage

feedback from the electrodes is very likely to work. With regard to this idea, a

control block monitoring the voltage over electrodes closes S2 as soon as it

becomes less than a defined amount. This means that the system intelligently

identifies that the plasma is formed and it’s going to be a short circuit in the

system so that it can distinguish plasma reaction from its consequent incident

and manages to stop plasma energizing in order to prevent excessive heat

generation and power losses in plasma systems. In the load supplying mode in

Fig. 6.2(c), the current source consists of the inductor which delivers the stored

energy to the load. The energies in the current source inductor and in the load

capacitor are calculated as follows:

2

2

1LIEL = & 2

2

1VCE LoadC = (6-7)

190

)(2

1

2

1

2

1 2min

2max

2min

2max IILLILIEL −=−=∆ (6-8)

)(2

1

2

1

2

1 2min

2max

2min

2max VVCVCVCE LoadLoadLoadC −=−=∆ (6-9)

The load capacitance is considerably low; therefore, the stored energy inside it is

negligible. With respect to this fact, the stored energy in the inductor comprises

most of the energy delivered to the load. Even when the capacitor is being

charged, since the current is not sufficient for plasma formation, there may

actually be some microplasmas energized by the energy stored in the capacitor.

The small capacitance cannot store high amounts of energy, so the power loss is

not significant.

The above equations indicate that the output voltage over the electrodes is

variable, based on the circuit inductance and the current flow through it. It is not

possible to elevate inductor current since the inductor may become saturated.

6.3. Extra capaci tor

In some applications, it may be useful to put a small capacitor in parallel with a

load storing energy in charging mode and delivering it to the load. Connecting a

capacitor in parallel with the load can improve the performance of the system for

the expressed situation. It acts as extra energy storage and positively interferes in

energy supply, so the inductor current will be kept far from massive tolerances.

In the process of energy delivery, the stored current in the inductor will be

pumped into the capacitor and will create voltage stress. Because the equivalent

capacitance of the circuit has been increased, the voltage stress will be decreased

according to Eq. (6-1).

The total delivered energy to the load, consisting of stored energy in the inductor

and capacitor, can be defined as:

extraCLLoad EEE += )(2

1)(

2

1 2min

2max

2min

2max VVCIIL extra −+−= (6-10)

Inductor, extra capacitor and load energies shown in Fig. 6.6 can confirm above

statement. In the first mode which starts from 4.9993 ms and continues to 4.9998

ms, 25mj of inductor energy is delivered to the capacitor, while in the load

supply mode in the period of 4.9998 to 5 ms this stored energy in the capacitor

accompanied by the 6.5mj energy delivered directly from the inductor are

transferred to the load. As shown in Fig. 6.6(c) the summation of these energies

which is 31.5mj is received by the load

191

4.9992 4.9993 4.9994 4.9995 4.9996 4.9997 4.9998 4.9999 5 5.0001

x 10-3

4.74

4.75

4.76

4.77

4.78

4.79

4.8

(a)

Indu

ctor

ene

rgy(

j)

4.9992 4.9993 4.9994 4.9995 4.9996 4.9997 4.9998 4.9999 5 5.0001

x 10-3

0

0.005

0.01

0.015

0.02

0.025

(b)

Cap

acito

r en

ergy

(j)

4.9992 4.9993 4.9994 4.9995 4.9996 4.9997 4.9998 4.9999 5 5.0001

x 10-3

0

0.01

0.02

0.03

0.04

(c)Time

Load

ene

rgy(

j)

Fig.6. 6. (a). Inductor energy (b). Capacitor energy (c). Load energy

6.4. Summary

This paper presents the possibility of utilizing a current source topology instead

of a voltage source as an efficient, flexible and reliable power supply for plasma

applications. The benefits of employing this topology instead of a voltage source

are: to decrease power losses with controlling current flow through the load and

the flexibility of generating different voltage levels and (dv/dt) s while having

control on the duty cycle of the switches. Additionally, the topology has the

capability of being set for a range of various applications. Moreover, the

simplicity of the topology and its control strategy is another significant

advantage of this concept.

6.5. References

[1] D. Wang, T. Namihira, K. Fujiya, S. Katsuki and H. Akiyama, “The reactor design for diesel exhaust control using a magnetic pulse compressor,” IEEE Trans. Plasma Science, vol. 32, no. 5, pp. 2038–2044, Oct. 2004.

[2] T. Namihira, S. Tsukamoto, D. Wang, S. Katsuki, R. Hakam, H. Akiyama, Y. Uchida and M. Koike, “Improvement of NO removal efficiency using short-width pulse power,” IEEE Trans. Plasma Science, vol. 28, no. 2, pp. 434–442, Apr. 2000.

[3] M.R. Hoque, T. McNutt, J. Zhang, A. Mantooth and M. Mojarradi, “A high voltage Dickson charge pump in SOI CMOS,” Custom Integrated Circuits Conference, 2003. Proceedings of the IEEE 2003 21-24 Sept. 2003 Page(s):493 – 496.

[4] K. S. Muhammad, A. M. Omar and S. Mekhilef, “Digital control of high DC voltage converter based on Cockcroft Walton voltage multiplier circuit,” IEEE TENCON 2005 Region 10, 21-24 Nov. pp: 1 – 4.

192

[5] D. F. Spencer, R. Aryaeinejad and E. L. Reber, “Using the Cockroft-Walton voltage multiplier design in handheld devices,” IEEE Nuclear Science Symposium Conference Record, 2001, Volume 2, 4-10 Nov. 2001 Page(s):746 – 749.

[6] W.J. Carey and J.R. Mayes, “Marx generator design and performance,” Conference Record of the Twenty-Fifth International Power Modulator Symposium, 2002 and 2002 High-Voltage Workshop., 30 June-3 July 2002 Page(s):625 – 628.

[7] T. Heeren, T. Ueno, D. Wang, T. Namihira, S. Katsuki, and H. Akiyama, “Novel Dual Marx Generator for Microplasma Applications,” IEEE Trans. Plasma Science, Vol. 33, No. 4, August 2005.

194





2. they take public responsibility for their part of ~e publication, except for the responsible







requirements.


A Bidirectional Two-Leg Resonant Converter for High Voltage Pulsed Power Applications

Published in the proceedings of: lET European Pulsed Power Conference, IETEPPC2009,

Geneva, Switzerland.

Contributor Statement of contribution 1---S'-'a-'s_an___,Z~ab.:...i"""h_i _"\--'rl Proposed the initial design and conducted simulation studies and data

'A__£ J/'.-J analysis, designed the control strategy, implemented hardware set-up - / and conducted experimental verifications and wrote the manuscript.

/ l--~l~l~A~u£g .. ~2~0~1 1~-4--------~~----~----~----~--------~--~~ Proposed the initial design and supervised the validity studies

Firuz Zare

Gerard Ledwich

Arindam Ghosh

including: conducting the sirnulations and experimental studies and writing the manuscript Aided experimental design, and data analysis

Aided planning the control strategies and writing the paper



authorship. ~ A/Pro f. Firuz Zare '"

_N_am_ e ______ h_ ---c: s• 11 Aug. 2011

Date

195

CHAPTER 7

A Bidirect ional Two-Leg Resonant Converter

for High Voltage Pulsed Power Appl icat ions

Sasan Zabihi*, Firuz Zare*, Gerard Ledwich*, Arindam Ghosh*



Presented and published at: IET European Pulsed Power Conference,

IETEPPC2009, Geneva, Switzerland.

196

Abstract— A high voltage pulsed power supply is proposed in this paper based

on oscillation between an inductor and a capacitor in an LC circuit. A two-leg

resonant circuit, supplied through an inverter with an alternative voltage

waveform, can generate output voltage up to four times an input voltage

magnitude. Bipolar and unipolar modulations are used in a single phase inverter

to analyse their effects on the proposed resonant converter. Simulations have

been carried out to evaluate the proposed topology and control.

7.1. Keywords

Resonant converter, Pulsed power, Modulation

7.2. Int roduct ion

Resonant converters have recently attracted attentions amongst power electronics

specialists due to the benefits they bring for power circuits in terms of switching

[1] and commutation [2]. The zero crossing of currents which occurs due to the

resonance between inductive and capacitive components of the circuit

significantly decreases the power loss of switching action [3]. The power

exchange between these components is the other advantage of resonant circuits

which can be beneficial for high voltage generation [4, 5].

The initial idea of these converters is based on the resonance and power

exchange of inductive and capacitive elements of the circuit. The inductor

current and capacitor voltage waveforms of a typical resonant converter shown

in Fig. 7.1(a) are displayed in Fig. 7.1(c). In an LC circuit, there is an oscillation

between the inductor and the capacitor which causes the voltage and current of

the components to be changed sinusoidally. As can be seen in Fig. 7.1(c), the

voltage across the capacitor and the current through the inductor oscillate in the

same period in which the resonant frequency of L and C can be calculated as

follows:

CLf r

⋅=

π2

1 (7-1)

In the first quarter of the resonant cycle, the flowing current through the circuit

will charge both the inductor and capacitor while in the second quarter, the

summation of the input and capacitor voltages corresponds with a negative

voltage which appears across the inductor and discharges it. The falling inductor

197

current still charges the capacitor until the inductor current reaches zero. In this

instant, the capacitor voltage would be almost twice the input voltage level. In

the third quarter, the reverse current will discharge the capacitor’s voltage and

this trend will continue during the fourth quarter.

Fig.7. 1. (a) Resonant circuit, (b) Half resonant circuit, (c) Capacitor voltage and Inductor current of a typical resonant circuit. (d) Capacitor voltage and Inductor current of a typical half resonant

circuit

The concept of generating high voltage through resonant converters comes from

this procedure with a few modifications. Diodes with reverse current blocking

capability can stop circuit operation when the current crosses zero at the end of

second quarter, and keep the capacitor voltage fully charged. The circuit and the

results are shown in Fig. 7.1(b) and 7.2(d) respectively. For the next resonant

cycle, the capacitor should be discharged. Therefore, this topology is not suitable

for continuous power supply while it can be utilized for pulsed power

applications. This pulsed power supply is an appropriate candidate for the loads

demanding either low energy with high repetition rate or high energy with low

repetition rate.

7.3. Bid i rect ional resonant conver ter : topology and

operat ion

According to Fig. 7.1(d), instead of discharging the capacitor in the next half

cycle, there is a possibility of using this period for charging another capacitor

and utilizing the summation of voltages across both capacitors. To satisfy this

198

goal, the circuit needs to be fed by an inverter providing alternative voltages for

it. According to this concept, a two-leg resonant converter demonstrated in Fig.

7.3 is presented to benefit the advantage of charging two capacitors, each in a

half cycle of the input voltage. In this configuration, there are two diode-

capacitor legs connected in parallel in which the diodes are in opposite directions

and supplied by an inverter through an inductor. Regarding this, a block diagram

of the desired configuration is depicted in Fig. 7.2 which includes a diode

rectifier, providing dc voltage for the configuration, and a Power Factor

Corrector (PFC) boost converter both increasing the dc voltage and mitigating

the destruction components of the input current, which is injected in to the

network as well as the proposed resonant converter.

Fig.7. 2. A block diagram of the proposed resonant converter

Fig.7. 3. Bidirectional resonant circuit

As we know, an inverter converts a dc voltage to an alternative voltage with

adjustable frequency and magnitude. The inverter includes a number of

transistors with/without anti-parallel diodes as switches. The control signals sent

to the gates of the switches open and close the switches in order to track and

generate a reference waveform in the output of the inverter. In a single phase

inverter, there are two legs including two switches which can be controlled based

on bipolar or unipolar modulations. The bipolar and unipolar methods are two

possible modulation methods which generate different voltage shapes in the

inverter output. As seen in Fig. 7.4, the inverter controlled under the bipolar

method can provide the load with voltage levels of +Vdc & -Vdc while in the

unipolar method, zero voltage level can be achieved in addition to those two

former levels. Ts and fs (Ts=1/fS) are switching cycle and switching frequency,

respectively.

199

Fig.7. 4. Output voltage waveforms with bipolar and unipolar modulations.

7.4. Bipolar cont ro l method

First we consider bipolar modulation in an inverter which results in only +Vdc

and -Vdc. This configuration can operate in two different modes. In the first

mode, which is the first half cycle, S2 and S3 conduct and a positive voltage is

applied to the resonant converter. In this mode, which is demonstrated in Fig.

7.5(a), there is a positive polarity of voltage across D1 which forces it to conduct

and charge C1. The capacitor voltage and the inductor current during the first half

cycle, when Vin=+V dc, can be achieved as follows:

)cos1()(CL

tVtV dcC

⋅−= (7-2)

)(sin)(CL

tV

L

CtI dcL ⋅

⋅⋅= (7-3)

At that moment at which the inductor current crosses zero point, D1 disconnects

this circuit’s loop and C1 is charged twice as the input voltage. In the second half

a cycle when S2 and S3 are switched off and S1 and S4 are switched on, a negative

voltage will be applied to the resonant circuit. In this mode the current flows

through the second leg since there is positive voltage polarity across D2. Again,

while the capacitor C2 and the inductor get fully charged and discharged

respectively - the time that current crosses zero point - D2 stops the circuit

operation. This mode is also indicated in Fig. 7.5(b). The voltage at the output of

this resonant converter would be four times the input voltage level which shows

the benefits of the bidirectional resonance in a two-leg circuit.

200

Fig.7. 5. Operation modes of the resonant converter supplied with an inverter controlled with

bipolar method.

Supplying a double-leg resonant circuit with an inverter controlled with bipolar

method has four scenarios which are discussed below, and the figures are

exhibited in Fig. 7.6.

These scenarios include:

fr>fS , fr=fS , fr<fS , fr<0.5fS conditions,

While fS represents the inverter switching frequency and fr is the resonant

frequency of L & C.

To get the maximum output voltage which is four times the input voltage, the

resonant frequency of L & Ci should be necessarily more than or at least equal to

the inverter’s switching frequency, otherwise the input pulse width would not be

long enough to let the converter components resonant thoroughly. The results

shown in Fig. 7.6(a) and 7.6(b) are achieved in the conditions which both satisfy

sufficient time for an entire resonance. On the other hand, while the inverter’s

switching frequency is more than the resonance frequency of the circuit, the

resonance will not occur thoroughly since there is insufficient voltage to charge

the capacitors to the ultimate level of voltage. This is how we can adjust the

output voltage level up to four times the input level. The results shown in Fig.

7.6(c) and 7.6(d) are obtained under this situation and confirm the above

statements. The proportion of the stored energy in the inductor, which can be

delivered to the capacitor and charge it to an appropriate voltage level, is highly

based on the frequency and pulse width of the input voltage. In a specific

situation, if the input pulse width is not long enough to raise the capacitor

voltage to the input voltage level -because of the voltage difference between the

input and the capacitor voltage levels across the inductor - low current flows

(resonates) in the next cycles until the capacitor voltage reaches the input voltage

level. Fig. 7.6(d) truly presents this situation. As seen in Fig. 7.6(c), if the input

voltage lasts longer, the capacitor will be charged once and its voltage may

201

become more than the input voltage level and less than twice the input voltage

level.

As indicated in Fig. 7.6(c) and 7.6(d), while the inverter changes the polarity of

its output voltage, there are still currents stored in the inductor which take a

while to be discharged. So, the diode used to conduct during this half cycle still

conducts for an adequate period of time after the inverter’s switching in order to

deplete the inductor current, and this time should be dedicated to the system

from the next half cycle. According to this behavior, there are unequal supplying

cycles for each leg and capacitors are variably charged.

If the switching frequency is presumed (1/2t1), which means the inverter

switches at t1, the voltage and current of the resonant circuit at t1 can be assumed

as the initial condition for the following operation of the resonant circuit:

)()cos1()( 111

1 ttVCL

tVtV dcC ==

⋅−= (7-4)

)(sin)( 111

1 ttICL

t

L

CVtI dcL ==

⋅⋅⋅=

(7-5)

The circuit behaves for the rest of the time until the inductor becomes fully

discharged, and this leg is disconnected can be found as follows: (Vin=-Vdc

during this period)

dcdcC VCL

t

C

LI

CL

tVVtV −

⋅⋅⋅+

⋅⋅+= sin)(cos)()( 11

(7-6)

CL

tI

CL

t

L

CVVtI dcL ⋅

⋅+⋅

⋅⋅+−= cossin)()( 11 (7-7)

Since the capacitor charging and the inductor discharging periods are not

identical for different half cycles, the capacitors are not charged similarly. The

asymmetrical response of the resonant converter fed by an inverter controlled

with bipolar method will negatively affect the control process of this

configuration, and make it more complex to control the output voltage.

0 0.05 0.1 0.15-200

0

200

Inp

ut v

olta

ge(

V)

0 0.05 0.1 0.15-200

0

200

Inp

ut vo

ltag

e(V

)

0 0.05 0.1 0.15-200

0

200

Ind

uct

or

curr

ent(

A)

0 0.05 0.1 0.15-200

0

200

Ind

uct

or

curr

ent(

A)

202

0 0.05 0.1 0.15-400

0

400

800

(a)Time(ms)

C1,

C2 &

Ou

tpu

t vo

ltag

es(V

)

C1 voltageC2 voltageOutput voltage

0 0.05 0.1 0.15-400

0

400

800

(b)Time(ms)

C1,

C2 &

Ou

tpu

t vo

ltag

es(V

)


0 0.05 0.1 0.15-200

0

200

Inp

ut v

olta

ge(

V)

0 0.05 0.1 0.15-200

0

200

Inp

ut v

olta

ge(

V)

0 0.05 0.1 0.15-200

0

200

Ind

uct

or

curr

ent(

A)

0 0.05 0.1 0.15-200

0

200In

du

cto

r cu

rren

t(A

)

0 0.05 0.1 0.15-400

0

400

800

(c)Time(ms)

C1,

C2 &

Ou

tpu

t vo

ltaes

(V)


0 0.05 0.1 0.15-200

0

200

400

(d)Time(ms)

C1,

C2 &

Ou

tpu

t vo

ltag

es(V

)


Fig.7. 6. Input voltage, inductor current, capacitors and output voltages of a resonant converter

with an inverter controlled with bipolar method in the case of : (a) fS=f r=15823Hz, (b) fS<f r (fS=10kHz), (c) fS>f r (fS=25kHz), (d) fS>2fr , (fS=40kHz)

7.5. Unipolar contro l method

Unipolar control method of inverter has the capability of solving this difficulty,

since it gives a zero level interval amongst positive and negative polarity of

voltage at the inverter output. In this method, the modulation frequency should

be less than or at least equal to the resonant frequency in order to have a

complete oscillation and obtain a maximum voltage level in the output of the

converter. According to the attributes of this method, the pulse width variation

can define the output voltage level. The advantage of this technique in

comparison with the bipolar method is that the zero voltage intervals let the

inductor current get fully discharged and another leg can be supplied during the

following half cycle. It means the whole period of the next half cycle will be

dedicated to the other leg and supplies it individually.

203

According to this technique, the circuit’s behavior is completely symmetrical for

each cycle and the voltage and current values of the circuit during even intervals

while Vin=0, can be similarly estimated as follows:

CL

t

C

LI

CL

tVtVC

⋅⋅⋅+

⋅⋅= sin)(cos)( 11 (7-8)

CL

tI

CL

t

L

CVtI L ⋅

⋅+⋅

⋅⋅+−= cossin)( 11 (7-9)

While t1 is the pulse width in the unipolar method, which is the period during

which either positive or negative voltage levels are applied to the resonant

converter. It is also possible to acquire the equations of the circuit voltage and

current for odd half cycles and intervals.

The operation modes of the circuit during these intervals are presented in Fig.

7.7. Coinciding the operation of S1 & S3 in on state and S2 & S4 in off state or

vise versa can provide resonant converter with those zero voltage levels. Since a

transistor cannot conduct a current in both directions, there is an anti-parallel

diode across each transistor in which this configuration presents a bidirectional

switch to conduct both positive and negative currents.

Fig.7. 7. Extra states of inverter providing resonant converter with the zero level of voltage in

unipolar control method.

The simulation results for the inverter with unipolar control shown in Fig. 7.8

verify the symmetrical performance of this circuit. In these examples, the

switching and the resonant frequency are identical while the pulse widths are

changed from 40% to 10%. The output voltage of the inverter with the bipolar

control method can be assumed as a unipolar method with 50% pulse width. As

expected, the output voltage levels vary with respect to the pulse width of the

inverter waveform in a linear proportion. Based on the simulation results

presented in Fig. 7.8, for 40, 30, 20, and 10% pulse widths, the output voltages

are 750, 620, 470, and 400V respectively.

204

0 0.05 0.1 0.15-200

0

200In

pu

t vo

ltag

e(V

)

0 0.05 0.1 0.15-200

0

200

Inp

ut v

olta

ge(

V)

0 0.05 0.1 0.15-200

0

200

Ind

uct

or

curr

ent(

A)

0 0.05 0.1 0.15-200

0

200

Ind

uct

or

curr

ent(

A)

0 0.05 0.1 0.15-400

0

400

800

(a)Time(ms)

C1,

C2 &

Ou

tpu

t vo

ltag

es(V

)

C1 voltage

C2 voltage

Output voltage

0 0.05 0.1 0.15-400

0

400

800

(b)Time(ms)

C1,

C2 &

Ou

tpu

t vo

ltag

es(V

)


0 0.05 0.1 0.15-200

0

200

Inp

ut v

olta

ge(

V)

0 0.1 0.2 0.3 0.4 0.5-200

0

200

Inp

ut v

olta

ge(

V)

0 0.05 0.1 0.15-200

0

200

Ind

uct

or

curr

ent(

A)

0 0.1 0.2 0.3 0.4 0.5-200

0

200

Ind

uct

or

curr

ent(

A)

0 0.05 0.1 0.15-400

0

400

800

(c)Time(ms)

C1,

C2 &

Ou

tpu

t vo

ltag

es(V

)


0 0.1 0.2 0.3 0.4 0.5-400

0

400

800

(d)Time(ms)

C1,

C2 &

Ou

tpu

t vo

ltag

es(V

)


Fig.7. 8. Input voltage, inductor current, capacitors and output voltages of a resonant converter with an inverter controlled with unipolar method in the case of :(a) P.W.=0.4TS, (b) P.W.=0.3TS,

(c) P.W.=0.2TS, (d) P.W.=0.1TS

7.6. Conclus ions

A bidirectional two-leg resonant converter is proposed in this paper, which

works in discontinuous mode and has the capability of voltage boosting up to

four times the input voltage level. As the simulation results have shown, the

possibility of adjusting output voltage in this converter is provided based on the

variation of inverter control features, such as switching frequency in bipolar

205

control method and pulse width in unipolar control method. Although the

inverter with the bipolar control cannot charge the capacitors symmetrically, the

unipolar control method of the inverter removes this drawback and gives proper

control over the output voltage. The results confirmed the validity of the

proposed topology in satisfying the anticipated functions.

7.7. References

[1] D. Fu, F.C. Lee, Y. Liu, M. Xu. “Novel multi-element resonant converters for front-end dc/dc converters”, PESC 2008. pp. 250-256, 15-19 Jun. 2008.

[2] M. Pahlevaninezhad, S.A. Khajehoddin, A. Bakhshai, P. Jain. “Voltage ripple reduction in series-parallel resonant converters by a novel robust H∞ control approach”, IECON 2008. pp. 1051-1056, Nov. 2008.

[3] T. Jin, K.Smedley. “Multiphase LLC series resonant converter for microprocessor voltage regulation”, IAC, 2006. volume 5, pp. 2136-2143, 8-12 Oct. 2006

[4] H. V. D. Broeck. “Analysis of a current fed voltage multiplier bridge for high voltage applications”, PESC 2002. volume 4, pp. 1919-1924, 23-27 Jun. 2002.

[5] J. Li, Z. Niu, Z. Zhang, D. Zhou. “Analysis of resonant converter with multiplier”, ICIEA 2007, pp. 326-331, 23-25 May 2007.

207

I



I. they meet the criteria for authorship in that they have participated in the conception,









requirements.


A High Voltage Power Converter with a Frequency and Voltage Controller

Published in the proceedings of: 1 ih IEEE Pulsed Power Conference, PPC 2009, Washington

DC, USA.

Contributor ( Statement of contribution Sasan Z_ab.(hi \ Proposed the initial design and conducted simulation studies and data

/vL / analysis, designed the control strategy, implemented hardware set-up and conducted experimental verifications and wrote the manuscript.



Hidenori Akiyama Provided us with general information about pulsed power supply specifications and its application demands



authorship.

N Prof. Firuz Zare !I Aug. 2011

Name Date

208

CHAPTER 8

A High Voltage Power Converter wi th a

Frequency and Voltage Control ler

Sasan Zabihi*, Firuz Zare*, Hidenori Akiyama†




Presented and published at: 17th IEEE Pulsed Power Conference, PPC 2009,

Washington DC, USA.

209

Abstract— A high voltage power converter is presented in this paper and is

based on a Capacitor-Diode Voltage Multiplier (CDVM) supplied through an

inverter. This power converter has the capabilities of generating variable high dc

voltage with improved transient response. The simulation results which are

presented in this paper verify that due to its fast transient response, this converter

can be used as a high dc voltage source in many applications.

8.1. Int roduct ion

CDVMs have been used widely in space and communication applications.

Among them, the Cockcroft-Walton multiplier topology has a remarkable role in

voltage promotion in microelectronics related configurations such as, radio

frequency passive transponders [1], passive wireless microsensors [2] and

battery-operated devices [3]. Three different configurations of these voltage

multipliers, including simple N-stage schematic of both a Cockcroft-Walton

voltage multiplier and a Dickson charge pump are depicted in Fig. 8.1.

The advantages of CDVM in those applications are that they are of small size

and weight and have high efficiency and reliability. The main disadvantages of

CDVM in these cases include the delay between input and output and the non-

negligible amount of capacitance needed, but this can be reduced within

acceptable limits by increasing multipliers’ operating frequency via an ac-ac

converter placed in the input of multiplier [4]. In relation to radio frequencies in

particular, Cockcroft-Walton multiplier is widely used to increase alternative

voltage magnitudes to higher dc levels in regard to its stages. The simplicity of

the circuit is the most remarkable benefit of it. Each stage consisting of a couple




transistors and their relative circuits like control boards and stacks. It is obvious

that these control blocks make the configuration heavier, more complex,

expensive and less reliable. On the other hand, these circuits have the flexibility

of being fed by any frequent input waveforms except those with a pulse shape.

This means that there is no obligation to give them just sinusoidal waveforms. In

respect to the nature of these circuits which is based on the peak detection, they

are able to increase the voltage magnitude of any alternative waveforms,

210

including sinusoidal, trapezoidal or even sinusoidal voltage waveforms with

harmonics. However, the voltage stress (dv/dt) across the input should be

controlled in order to control the leakage current through the capacitors.


Multiplier (b). N-stage Dickson charge pump (c). Another N-stage CDVM configuration

dt

dvCic = (8-1)

These specifications support the idea of utilizing these multipliers for pulsed

power applications. Of all high voltage applications, pulsed power generators are

the ones which demand novel configurations, including topologies and control

strategies to improve the performance flexibility and power efficiency of these

systems. In pulsed power applications, providing a high level of dc voltage is

challenging.

In this research work, several simulations have been carried out using

MATLAB/Simulink and PSPICE in order to analyze steady state and transient

performance of the converter at different load conditions and validate the control

algorithms. As can be seen in Fig. 8.2, we considered a one-stage Cockcroft-

Walton multiplier in this paper and presented all the simulation results for it. It is

apparent that these simulation results and analyses can be developed for multi

stage multipliers regarding few modifications.

Fig.8. 2. One-stage Cockcroft-Walton voltage multiplier

211

8.2. Transient

As mentioned, due to the load demands in pulsed power applications, it is

absolutely crucial to supply load with a flexible dc voltage with a fast response

time in order to improve the quality of the output voltage. Firstly, a one-stage

voltage multiplier is connected to the grid and supplied by a conventional

sinusoidal voltage waveform, 200V and 50Hz. As can be seen in Fig. 8.3(a), for

the condition of identical capacitors, it takes 8 cycles (0.16s) for the output

voltage to get twice of the input voltage magnitude from zero at the beginning of

the simulation and this period may either increase or decrease according to the

capacitor’s proportions. For example; if C1 is ten times C2, (C1=10 C2=10 mF),

the transient time will markedly decrease to (3 cycles, 0.05s) as shown in Fig.

8.3(b).

The voltage across capacitor C2 is:

)1(2

)(22

21

2

21

1 −+

++

= iVCC

CV

CC

CiV CSMC For i>1, SMC V

CC

CV

21

1)1(2 +

= (8-2)

Where (i) represents the number of each cycle.

For the specific situation when (C1=C2), the former equation could be simplified

as:

2

)1()( 2

2

−+=

iVViV C

SMC For i>1 & )( 21 CC = 2

)1(2

SMC

VV = (8-3)

Hereby, we are able to recognize the number of cycles, taking in each transient

for the output voltage to get the ultimate value. Hence, in regard to input

frequency, the length of time each transient takes can be almost predictable.

0 0.05 0.1 0.15 0.2 0.25

-200

-100

0

100

200

300

400

Time(s)

Inp

ut a

nd

cap

aci

tors

vo

ltag

es(

v)


0 0.05 0.1 0.15 0.2 0.25

-200

-100

0

100

200

300

400

Time(s)

Inpu

t an

d c

apa

cito

rs v

olta

ges(

v)


(a) (b)

Fig.8. 3. Voltage transient of multiplier with 50 Hz input frequency (a). Identical capacitors (b). Different capacitors (C1=10C2)

Another concept which may be considered a solution for decreasing transient

time is supplying the multiplier with a high frequency power supply. As shown

in Fig 8.4, the transient times of the multiplier with identical and different

212

capacitors are reduced to 10 and 3 ms respectively when the multiplier is

supplied by a 1kHz input waveform. Since the grid frequency is always constant

(50 or 60 Hz), a frequency converter in the input of the multiplier is required to

improve the transient response of the system. When feeding the multiplier with a

higher frequency, each cycle lasts for a shorter time and as a result, the whole

transient time will be decreased.

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1

-200

-100

0

100

200

300

400

Time(s)

Inpu

t and

ca

pac

itors

vo

ltag

es(v

)

Input voltage of multiplier(v)Voltage over first capacitor of multiplierOutput voltage of multiplier(v)

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1

-200

-100

0

100

200

300

400

Time(s)

Inpu

t and

cap

acito

rs v

olta

ges(

v)


(a) (b)

Fig.8. 4. Voltage transient of multiplier with 1KHz input frequency (a). Identical capacitors (b).Different capacitors

8.3. Adjustable output vo l tage level

There are many demands for different voltage levels based on the various

applications. A power supply with a capability of providing adjustable voltage

magnitude is highly sought-after equipment. The multiplier’s output voltage is

generated based on the input voltage and the number of multiplier stages.

In regard to Eq. (8-4), for variable voltage magnitude, there are two options:

either changing multiplier stages or the multiplier’s input voltage. Alternative

stages is not reasonable due to the complexities of installation and control

method while it just gives the flexibility of ascending voltage related to the

number of stages times the input voltage magnitude. On the other hand, as

indicated in Fig. 8.9(a), a variable input voltage results in variable voltages in the

output. It is not possible to change the input voltage since a constant voltage is

supplied by grid and we have no control on it unless an ac-ac converter is placed

between the source and multiplier and the multiplier is supplied through it.

inout VnV ⋅= (8-4) As indicated in Fig. 8.5, this converter consists of an ac-dc converter to provide

an adjustable dc voltage, while the input power factor is controlled. The second

converter is a dc-ac inverter which generates ac voltage with variable magnitude

and frequency. In this new configuration, these two converters are connected in

213

cascade, supplying an ac-dc voltage multiplier. The first converter consists of a

diode rectifier with a boost converter which improves the input power factor and

reduces low order harmonics. The controller changes the dc voltage based on the

reference voltage to generate a high voltage at the output of the voltage

multiplier. The second converter is an inverter which generates an ac voltage

with a variable frequency. In a traditional diode capacitor voltage multiplier, the

output voltage depends on the number of capacitors and diodes and input voltage

magnitude. As the grid voltage is constant (220V), it is not possible to change

the output voltage easily. While in this topology, the output dc voltage of the

first converter is controlled. In addition, the output ac voltage of the inverter can

be adjusted in order to have variable voltage magnitude and frequency in the

output.

Fig.8. 5. An ac-dc-ac converter

8.4. Feeding CDVM through an inverter

Inverters are power-switch based pieces of equipment which convert dc voltage

to ac voltage. A control strategy decides the frequent sequence of opening and

closing of switches considering desired output. A schematic configuration of a

single phase inverter utilized in this work is presented in Fig. 8.6(a). There are

also various PWM techniques providing control signals for these switches, such

as bipolar and unipolar modulations. An inverter controlled under unipolar

modulation gives the capability of having variable voltage amplitude in the

multiplier input and subsequently in the multiplier output. It is the most striking

advantage of an inverter switched with unipolar control method.

A brief review of unipolar method control reveals how variable voltage is

available in the output of an inverter. In the unipolar modulation control method

of the inverter, the output voltage has three voltage levels of –Vdc & 0 & +Vdc

while in the bipolar modulation, there are just two voltage levels, –Vdc & +Vdc.

214

Fig. 8.6(b) demonstrates one cycle of output waveforms for both modulation

methods.

(a) (b)

Fig.8. 6. (a). Schematic of full bridge (two-leg) inverter (b). Bipolar and unipolar modulations output waveforms

In both cases, changing Ts gives variation of frequency fs in the output. In bipolar

mode, changing the average of the output cycles is possible by changing duty

cycles, while in unipolar mode, the variation of duty cycles not only gives

different output averages, but also leads to the change of the rms value of the

output voltage. This eventually ends in having variable voltage magnitudes in the

output of the filter.

The output voltage of the inverter cannot be given to the multiplier directly,

since high dv/dt s of this pulsed shape waveform may cause inrush currents in

the multiplier’s capacitors. It is therefore necessary to reduce voltage stress

(dv/dt). An LC filter located at the output of the inverter eliminates high

frequency harmonics and delivers high quality voltage which has variable

amplitude with respect to the variation of duty cycles. Fig. 8.7 shows simulation

results for duty cycles of .05, 0.5 and 0.95, while output frequency is 50Hz.

215

0.82 0.83 0.84 0.85 0.86 0.87

-200

-100

0

100

200

(a)

Out

put

volta

ges

of in

vert

er a

nd f

ilter

(v)

Inverter

filter

1.82 1.83 1.84 1.85 1.86 1.87

-200

-100

0

100

200

(b)

Out

put

volta

ges

of in

vert

er a

nd f

ilter

(v)

Inverter

filter

2.81 2.82 2.83 2.84 2.85 2.86 2.87-300

-200

-100

0

100

200

300

(c)Time(s)

Out

put

volta

ges

of in

vert

er a

nd f

ilter

(v)

Inverter

filter

Fig.8. 7. Output voltage of inverter and filter for duty cycles of (a). 0.05 (b). 0.5 (c). 0.95.

The specifications of the simulated circuit are listed in Table 8.1.

TABLE 8. 1. CIRCUIT SPECIFICATIONS

Multipliers capacitors 1e-3 F Inverter’s dc voltage 200 V

Frequency of output voltage 1KHz

In Fig. 8.8(b), it is demonstrated that we can get different voltage magnitude in

the output of the filter and multiplier with variations in duty cycles of switching

in unipolar modulation. In this model, the duty cycle of switching changes from

0.1 to 0.9 and gives several voltage levels in the output. Inverter output voltage

in Fig. 8.8(c) illustrates the unipolar control method’s skill in providing the

multiplier with variable voltage levels (duty cycles of 0.1 & 0.5 & 0.9). Load

connections and their influence on system reply are shown in Fig. 8.8(d).

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

-200

-100

0

100

200

300

400

Time(s)

Inp

ut a

nd

cap

aci

tors

vo

ltag

es(

v)


0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16 0.18-300

-200

-100

0

100

200

300

400

500

600

Time(s)

Inp

ut a

nd

ca

pa

cito

res

volta

ge

s(v)

Input voltage of multiplier(v)Voltage over first capacitore of multiplier(v) Output voltage of multiplier(v)

(a) (b)

216

0 0.01 0.02 0.03 0.04 0.05 0.06-250

-200

-150

-100

-50

0

50

100

150

200

250

Time(s)

Inve

rte

r ou

tpu

t vol

tag

e(v

)

0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1

-300

-200

-100

0

100

200

300

400

500

600

Time(s)

Inp

ut a

nd

ca

pa

cito

rs v

olta

ge

s(v)


(c) (d)

Fig.8. 8. (a) Variable input voltage results in variable voltages in the output (b). Variable output voltage provided by an inverter under unipolar control method. (c). Inverter’s output waveform

with duty cycles of .1 & .5 & .9. (d). Load connections and voltage rehabilitation capability

8.5. Energy discussion for Plasma appl icat ions

Plasma generators have been recognized as one and probably the most

significant customers of pulsed power technology. In plasma applications, to

have the most efficient reaction supplying, the system’s input and output energy

should be almost equivalent, which means equal power exchange.

Lossoutin EEE += Lossoutin PPP += (8-5) The difference between Ein and Eout, which is known as energy loss, should be

minimized as low as possible. However, it could not be totally omitted due to

switching and delivery losses.

As is known in an inverter-multiplier dc power supply, the output capacitor is

responsible for delivering the output energy to the load. So the output energy can

be defined as

)(2

1 2min

2max VVCEout −= (8-6)

Whereas Vmax and Vmin are load voltages.

While the energy absorbed by load is defined as:

)).(.()).((2

R

VDTPtE R

sLoadonLoad == (8-7)

Whereas 2

minmax VVVR

+= and minmax VVV −=∆ . The load duty cycle is defined as:

Ron V

VCRt

∆= . (8-8)

8.5. Further analyses

The application of the close loop control technique to this inverter-multiplier unit

improves the accuracy, response time and quality of the output voltage. Fig. 8.9

217

illustrates the entire concept of inverter utilization in multiplier feeding including

control strategies.


The multiplier response to a square shaped input waveform in addition to the

capacitor’s current are demonstrated in Fig. 8.10(a). Thus the dv/dt of the

inverter output voltage required to be less sharp. It can be observed in Fig.

8.10(b) that how the current flowing through capacitor drops when the input

voltage deforms to a trapezoidal waveform for less dv/dt.

A first order low pass LC filter can either satisfy this aim or even detect

fundamental components of input voltage, based on its appointed cut off

frequency. Since having trapezoidal waveform in the output does not give the

flexibility of supplying multiplier with variable voltage magnitude it is preferred

to extract the fundamental components of input voltage via an appropriate filter.

However, taking this into account when the frequency of inverter output is

changed, the filter’s elements should be differed to adapt the cut of frequency to

the new conditions. It reveals that a digital filter and a control system need to be

installed to provide such adaptability. However, this brings complexity to the

system. On the other hand, feeding the multiplier with only a high frequency

does not incur such complexity, while it raises switching losses. A feedback

control for the system may considerably decrease the switching losses as well as

increase the system’s accuracy.

0 0.05 0.1 0.15 0.2 0.25

-300

-200

-100

0

100

200

300

400

Time(s)

Inp

ut a

nd

Cap

acito

rs v

olta

ges

(v)

0 0.05 0.1 0.15 0.2 0.25

-300

-200

-100

0

100

200

300

400

Time(s)

Inp

ut a

nd

cap

aci

tors

vo

ltag

es(

v)

218

0 0.05 0.1 0.15 0.2 0.25-2.5

-2

-1.5

-1

-0.5

0

0.5

1

1.5

2x 10

4

Time(s)

Firs

t ca

pa

cito

rs c

urr

en

t(A

)

0 0.05 0.1 0.15 0.2 0.25-120

-100

-80

-60

-40

-20

0

20

40

60

Time(s)

Firs

t cap

acito

rs c

urre

nt(A

)

(a) (b)

Fig.8. 10. (a). Multiplier voltages, and first capacitor current with pulse shape input waveforms (b). Multiplier voltages, and first capacitor current with trapezoidal input waveforms

8.5. Summary

This paper presents a combination of couple of previously known converters

releasing a useful configuration in high voltage with a number of the advantages

in various applications. Based on the variation of inverter duty cycles, adjustable

dc voltage level in the output have been found to be achievable. The transient

time is drastically shortened due to high frequency input voltage. Furthermore,

the efficiency of the system will be greatly improved by a feedback control. The

validity of the proposed system has been verified regarding acquired simulation

results.

8.6. References

[1] C. Chen, M. Baghaeinejad and L. R. Zheng, “Design and Implementation of a High Efficient Power Converter for self-powered UHF RFID Applications,” Proceedings of HDP’06

[2] F. Yuan and N. Soltani, “Design Techniques for Power Harvesting of Passive Wireless Microsensors,” 51st Midwest Symposium on Circuits and Systems, 2008. MWSCAS 2008, 10-13 Aug. 2008 pp. 289 – 293

[3] D. F. Spencer, R. Aryaeinejad and E. L. Reber, “Using the Cockroft-Walton Voltage Multiplier Design in Handheld Devices,” Nuclear Science Symposium Conference Record, 2001 IEEE vol. 2, 4-10 Nov. 2001 pp. 746 – 749

[4] L. Malesani and Roberto Piovan, “Theoretical Performance of the Capacitor-Diode Voltage Multiplier Fed by a Current Source,” IEEE Trans. Power Electronics, vol. 8, no. 2, Apr. 1993.

[5] Heeren, J. F. Kolb, S. Xiao, K. H. Schoenbach and H. Akiyama, “Pulsed Power Generators and Delivery Devices for Bioelectrical Applications,” Twenty-Seventh International Power Modulator Symposium, 2006. 14-18 May. 2006 pp. 486 – 489.

[6] D. Wang, T. Namihira, K. Fujiya, S. Katsuki and H. Akiyama, “The Reactor Design for Diesel Exhaust Control Usinga Magnetic Pulse Compressor,” IEEE Trans. Plasma Science, vol. 32, no. 5, Oct. 2004.

[7] T. Namihira, S. Tsukamoto, D. Wang, S. Katsuki, R. Hakam, H. Akiyama, Y. Uchida and M. Koike, “Improvement of NO removal efficiency using short-width pulse power,” IEEE Trans. Plasma Science, vol. 28, no. 2, pp. 434–442, Apr. 2000.

220

CHAPTER 9

Conclusions and Further Research

221

9.1. Conclus ions

The diversity of pulsed power applications is in a fast growing trend as a result

of the benefits offered by these pulses. Delivering high instantaneous power in a

compressed form (pulse shape) has military, industrial, medical and

environmental applications. Alongside this enormous extension in the range of

more sensitive and precise applications is the current and increasing demand for

more flexible, higher quality pulses in a more efficient process. Many research

studies being conducted in different areas need a flexible pulser to vary pulse

features and to investigate the influence of these variations on the applications.

Gas/magnetic switching technologies (such as spark gap and hydrogen thyratron)

have conventionally been used as switching devices in pulse modulator

structures because of their high voltage ratings and considerably low rising

times. However, they also suffer from drawbacks such as: low efficiency,

reliability and repetition rate, short life span, considerable bulk, weight and cost.

Recently developed solid-state technology is an appropriate substitution for these

switching devices due to their benefits to the whole process. Besides being

compact, efficient, reasonable and reliable, and having a long life span, their high

frequency switching skill allows for the repetitive operation of power supply.

The main concerns in using solid-state transistors are the voltage rating and the

rising time of available switches that, in some cases, cannot satisfy an

application’s requirements. However, there are several configurations and

techniques that make solid-state utilisation feasible for high voltage pulse

generation. Therefore, the proposal and development of novel methods and

topologies with a higher level of efficiency and flexibility for pulsed power

generators are the main scope of this research work. This aim is pursued through

several innovative proposals that can be classified into two categories with

following principal objectives:

• Developing and justifying novel solid-state based topologies for pulsed

power generation

• Improving available technologies that have the potential to accommodate

solid-state technology by revising, reconfiguring and adjusting their

structure and control algorithms.

222

Theoretical justifications have been carried out by analysing dynamic and steady

equations and calculations of stored and exchanged energies. The validity of

proposed topologies, along with applied operational techniques and the control

strategies adopted for their operation were then investigated through simulating

circuit models in MATLAB simulation platform and PSPICE. Further analyses

of simulation outcomes carried out and the drawbacks were detailed. Results and

Solutions considered with this respect. The simulation results were then

compared to the experimental results acquired from circuits implemented in the

laboratory in order to validate the proposed topologies. The outcomes are

summarized below.

9.1.1. Developing and proposing novel solid-state based topologies for

pulsed power generation

Chapter 6 presents the concept of supplying plasma applications with a current

source instead of a voltage source. Many plasma applications demonstrate

variable resistive-capacitive characteristics during the supply period. As proven,

a current source is a proper option to supply a capacitive load due to

compatibility reasons. Therefor, using a current source as the energy storage for

supplying plasma applications is more reasonable. A current source topology

composed of an inductor and two power switches inspired by dc-dc converters

was considered in this part of the research, and analyses were conducted to

investigate its feasibility in producing pulsed power. The main concept behind

this topology is to provide high dv/dt, regardless of the switching speed of a

power switch, and to control the current level to properly transfer adequate

energy to various plasma applications. Circuit analyses were carried out with

respect to energy conversion and compression equations and, accordingly, a

suitable control method to run the circuit was adopted. Switching transient

effects on produced dv/dt were investigated in simulations by delivering different

amounts of current to the load and comparing the rising times. According to

simulation results, achieved dv/dt s are relatively unrelated to the switching

velocity and can be increased by raising the current level. Outlines of this part of

the research were published in the proceedings of the 17thIEEE Pulsed Power

Conference 2009, 17IEEEPPC2009, Washington DC, USA, 28 June-2 July 2009.

223

Pursuing the idea of using a current source in energizing plasma loads, Chapter 2

considers the selection and development of appropriate topologies using solid-

state devices for high voltage pulse production. A combination of current and

voltage source structures with multi switch-resistor-capacitor units at the output

was designed to allow the high voltage at the output to be split equally and to be

shared by a series of switches. Supplying capacitive loads through a current

source brings compatibility benefits to the process. The design absolutely

eliminates the need for power diodes in this configuration. Having no diode

rectifier in charging and discharging paths can minimize the conduction losses.

Theoretical analyses followed by determination of principal switching states

have been undertaken in order to ensure the viability of the topology in satisfying

desired functions. To recognize the associated deficiencies, a two-stage model

was simulated in Matlab. As a result of this study, additional resistors were

connected in the common paths of the output units and a smart control algorithm

was design to discharge the residual energy of capacitors after each supply cycle

or (even) after an unsuccessful supply cycle. Ultimately, a prototype assembly

was implemented and tested in low voltage regime in the laboratory to prove the

true performance of the converter. An acceptable voltage sharing with less than

5% tolerance is performed by the output stages; this guarantees a similar

charging and discharging process for the capacitors. A comprehensive analysis

of this topology including simulation and experimental results, is published in

the IEEE Transactions on Plasma Science Vol. 38, pp. 2877-2887, in 2010.

To deal with the complexity of the smart control strategy in the former proposal

and to enable the feasibility of halting the supply process at any stage without

any concern regarding probable resonance due to the residual charge in the

capacitors and the inductor, a very small amendment was introduced to its

structure. Inspired by positive buck-boost topology, resistors were substituted

with diode rectifiers at common paths of output units to prevent backward

energy flowing. Although power diodes have been accommodated in this design,

the contribution they make to this topology does not include charging and

discharging paths. Consequently, the conduction losses will be unchanged. The

change not only secures the supply process but also extends the application of

this topology to two-stroke cases. Using asymmetrical capacitors at the output is

224

also considered in this topology. All possible switching configurations have been

derived from positive buck-boost structure, including multi output units. A

simplified model simulation was executed and the outcomes verified the security

of the process. The previous set up was modified to conduct the experimental

tests. The pulse amplitude was boosted to more than 1 kV in this test. The

outcomes of circuit analyses and simulations accompanied by experimental

achievements are reported in the form of a journal paper in the IEEE

Transactions on Dielectric and Electrical Insulation, Vol. 17, pp. 1901-1911, in

2010.

9.1.2. Improving conventional technologies that have potential to


adjusting their structure.

A preliminary concept of charging a series of capacitors through a resonance is

introduced in Chapter 7. A bidirectional two-leg diode-capacitor that uses

resonant phenomenon to generate a higher voltage is proposed in this chapter. In

this circuit, the voltage is boosted up four times the input voltage. This technique

can be utilized in the generation of fundamental voltages for a pulsed power

system. An inverter supplies the circuit with an alternative voltage. Symmetrical

charging of the capacitors and, consequently, produced voltage adjustability is

feasible through unipolar control of the inverter. Furthermore, the switching

losses are substantially decreased using this technique. The idea, the analysis and

the simulation results are addressed in a conference paper published in the

proceedings of the IET European Pulsed Power Conference, IETEPPC2009,

Geneva, Switzerland.

To develop the idea for producing higher voltages, a Marx configuration was

designed, including diode-capacitor units. The units are based on the

aforementioned resonant circuits that are connected through power switches. The

Marx topology supplied by an inverter in the front side, enables series

connection of charged capacitors and aggregation of voltages at the output. The

design was reported in a conference paper at the 20th Australian Universities

Power Engineering conference, December 2010, Christchurch, New Zealand.

225

The main contribution of charging capacitors through a resonance, where a

properly developed resonant Marx configuration allows the possibility of

producing a supreme high voltage pulse with a much reduced number of

components, it presented in Chapter 4. A precisely designed Marx topology

consisting of diode-capacitor units with proper accommodation that uses

alternate half a cycle resonant phenomena to charge two groups of capacitors is

introduced in this chapter. The units are composed of two diode-capacitor legs

with opposite direction and connected in cascade through two power switches. In

this way, each leg requires one diode instead of two, and half of the switches can

be selected as unnecessary fast switches (i.e. Thyristors). Therefore, there will be

a huge saving in initial cost, weight, volume, and system intricacy by reducing

associated driving modules. Each group of capacitors are charged up to twice the

input voltage level in a half a cycle resonate, so the number of stages needed to

generate similar voltage levels is reduced to half, in comparison with

conventional Marx stages. In this structure, a half-bridge inverter is utilized in

the entrance of the system to provide the alternating voltage waveform for the

resonant Marx. Symmetrical charge of the capacitors resulting in voltage

adjustability skill is given to this topology by connecting a bidirectional switch

path to the joint point of the inverter and the resonant Marx. The load supply

process will be achieved with fewer conduction losses and, accordingly, higher

efficiency due to reducing the number of contributing solid-state switches in the

discharging path to half. Another advantage of this topology is utilizing resonant

phenomenon as the operation method and triggering the switches at the instant at

which the current flowing through them is zero. This allows the switching losses

to be kept at the minimum level possible. The circuit analysis of this topology

and initial simulation results were presented as a conference paper at the 19th

Iranian Conference on Electrical Engineering, May 2011, Tehran, Iran. A

comprehensive report including circuit analysis, switching mode discussions,

and simulation and experimental results of this design is published as a journal

paper in IEEE Transactions on Plasma Science, 2011.

Another approach of this research work is a solid-state based design for MGs

that has the benefit of charging the capacitors through a current source topology.

This enables the charge of the capacitors to a desired voltage level regardless of

226

input voltage level. The new configuration is also designed based on diode-

capacitor units. Each unit comprise two diode-capacitor legs and a commutation

circuit that is functionalized to reverse the charge polarity of one of the

capacitors. An inductor and a slow switching device (such as Thyristor) form the

commutation circuit. The units are connected in cascade through a fast switching

device and a power diode. The number of diodes needed in this design is reduced

to 75% of the number in a conventional Marx. Additionally, half of the switching

devices in commutation Marx design are replaced by slow Thyristors that require

fewer driving modules. Simulations for both single shot and repetitive operation

have been performed and the attained results confirm the validity of the proposal.

A prototype set up has been fabricated in the laboratory to compare the

experimental approaches with the circuit analysis and simulation results. A

conference paper including the primary concept of this pulse generator and some

simulation results has been presented at the IEEE Energy Conversion Congress

and Exposition (ECCE2010), Atlanta, USA. Extra discussions on the topology

features, accompanied by more simulation and experimental test results, were

submitted as an paper to the IEEE Transactions on Dielectric and Electrical

Insulations that is published in Vol. 18, Issue 4, pp. 1181-1188, August 2011.

CDVM structures are considered in Chapter 8 of this thesis due to their merit in

boosting voltage, while employing no active components. Small size and weight

and high efficiency and reliability are among other advantages of these circuits.

A power converter to generate high voltage with an improved transient response

was designed in this part of the research. A simplified two-stage Cockcroft-

Walton VM that is supplied through an inverter was investigated in simulations.

Adjustment of output voltage level was possible as a result of variation of the

inverter’s duty cycle. The transient time was drastically shortened due to

supplying the VM circuit through a high frequency input voltage. Furthermore,

the efficiency of the system greatly improved by a feedback control. The validity

of the proposed system has been verified considering the simulation results.

These analyses and simulations, were published in a conference paper at the

proceedings of the 17th IEEE Pulsed Power Conference (17IEEE_PPC 2009)

held in Washington DC, USA.

227

9.2.3. A summary of features, advantages and restrictions of proposed

converters

A low-voltage switch-resistor-capacitor unit based topology for pulsed power applications

Switching Pulses

SS

S1

SL

D

R2Load

C1

-

+

+

-

L

S2

C2

+

-

Sn

Cn

+

-

R1LoadControl

Protocol

of

Series

Switches

Plasma Load Model


CLoad


RD1

RD2

AC-DC

ConverterVac

220 V50 Hz

VinCin



Load

Feature: A modified positive buck boost converter in the entrance Advantages: Is compromised of a charged inductor in the front side

that acts as a current source appropriate for a repetitive operation of the pulse modulator

Charges the output capacitors with a flexible and adjustable high level of voltage

Insulates the load side from the input side of power supply during pulse generation (controlling power flow during undesired arc phenomena)

Feature: A cascade combination of switch-resistor-capacitor units at the output

Advantages: Gives a possibility to utilize solid-state technology Fewer active and passive components used in

comparison with conventionally used pulsed power supplies

The charging and the discharging paths are free of additional insulation components that leads to a significant reduce in the conduction loss and a more efficient supply process

The generated voltage sharing enables utilisation of low/medium voltage switching devices, and the increases in produced dv/dt and generated voltage level

228

Restrictions Needs a high amount of current stored in the inductor The residual energy in the storing components has to be

discharged after each load supplying cycle in order to initialize the storing components for the next supplying cycle. This leads to an extra loss.

The smart switching process designed to discharge the residual energy increases the recovery complexity, reduces the reliability and the pulse repetition rate generation of the power supply

A new topology based on positive buck-boost converters for pulsed power applications

Switching Pulses

SS

S1

SL

D

R2Load

C1

-

+

+

-

L

S2

C2

+

-

Sn

Cn

+

-

R1LoadControl

Protocol

of

Series

Switches

Plasma Load Model


AC-DC

ConverterVac

220 V50 Hz

CLoad


D2

D3

D1

Dn

V inCin



Load


that acts as a current source appropriate for the repetitive operation of the pulse modulator



Feature: A cascade combination of switch-diode-capacitor units at the output

Advantages: Enables control over power flow (preventing power loss by stopping the load supply process at any stage)

Facilitates energizing specific loads that can be stimulated through a high voltage followed by a voltage stress in two steps (two-stroke)

Although diodes are connected between the capacitors and the switches, except D1, none of them contribute to the charging and discharging paths and merely have blocking functions

Gives a possibility to use solid state technology Fewer active and passive components used in

229

comparison with conventionally used pulsed power supplies

The generated voltage sharing enables utilisation of low/medium voltage switching devices, and increases in the produced dv/dt and the generated voltage level

Restrictions Practically requires considerable amount of current stored in the inductor that leads to an increase in losses

A new generation of Marx topology based on resonant converters

AC Grid

Rectifier

AC-DC Converter



Inverter, VSI

DC-AC Converter

Voltage Regulator

V inv(t)+

-

Feature: A single-leg one phase inverter connected to an inductor in the

entrance Advantages: Employs fewer switching devices and consequently

fewer driving modules and decreases the switching losses rather than utilizing an H bridge inverter

Charges the capacitors to a twice input voltage level with opposite polarities through two half a cycle resonants


Reduces the conduction loss to a lower level compared to a current source configuration

Feature: A new Marx configuration has been proposed through a new arrangement of capacitors, power switches and diodes.

Advantages: Half of fast switching devices, such as IGBTs, can be replaced by slower Thyristors, like SCRs, that significantly reduces the complexity of driving modules

The number of power diodes is decreased to half. The switching and conduction losses are reduced due to

utilisation of fewer active components Restrictions Due to relying on resonant phenomenon, the repetition

rate cannot be increased such as that in a current source configuration can.

Pulse repetition rate of this power supply is restricted by the size of the inductor and the equivalent capacitor

Charging capacitors in two distinguished modes increases the initializing process time and restricts the pulse repetition rate furthermore

Using a half bridge inverter necessitates two voltage sources in the dc link

230

A new family of Marx topology based on commutation circuits

D1 D4

C1 C4

L1

Vdc

D2

C2

D3

C3

S3

S4

+

-

+

-

+

-

+

-

D5

S1

S2

SCR1

Df

SCR2


L2 L3

+

-

+

-


that acts as a current source appropriate for the repetitive operation of the Marx modulator



Feature: The Marx configuration has been modified based on a new arrangement of the capacitors, power switches and diodes. Commutation circuits are used in order to inverse selected capacitor’s voltage polarity.

Advantages: The number of fast switching devices is substantially decreased, (half of the one used in conventional design)

The number of utilized diodes is decreased as well. Half of fast semiconductor switches are replaced by

slower Thyristors that require simpler driving modules The switching and conduction losses are reduced due to

utilisation of fewer active components Restrictions Changing polarity of selected capacitors through

commutation modes takes extra time and confines the repetition rate of the modulator

Exchanging energy through circulating current also wastes a portion of stored energy during the commutation process

9.2. Further research

This research study has focused on improving the efficiency of plasma

applications by either proposing new solid-state topologies or improving

available solid-state based technologies. Two joint current-voltage based

topologies which were inspired by positive buck-boost converter configuration

and extended to multi-output were justified in the first step. Subsequently,

considering Marx structure, two new configurations which use resonant and

231

commutation in their operation were proposed. Suggestions for further research

work in five specific areas are discussed below.

Developing insulated solid-state topologies for pulsed power

A number of transformer-less (non-insulated) pulsed power topologies have been

developed during this research study. Although transferring high power through

magnetic fields and fluxes can increase losses, a combination of power

electronics topologies and high frequency power transformers can result in

advantages such as using fewer switching devices with lower voltage ratings.

The exploration of new combined structures, including insulated topologies is

recommended for future research.

Using CDVMs as fundamental voltage boosters for an MG for

continuously high voltage applications

Developing an ac/dc converter based on CDVM circuits as basic units of an MG

can be useful in producing high voltage with a higher gain and a fast transient

recovery. According to this idea, either the output capacitors of several CDVM

circuits or the capacitors of one CDVM circuit can be considered as Marx

capacitors. That provides Marx with the benefit of a fast charging transient after

pulse supply.

Using PFNs as basic units of an MG

Ladder shape PFN circuits, having merit in energy compression and pulse

production, can be utilized either individually or in a combination of circuits as

primary units of an MG. The specific format of PFN including their simplicity of

structure and their composition of exclusively passive components make it

favourable for many pulsed power applications. The possibility of fabricating an

MG with joint PFNs can be considered as the subject of future research.

Applications

Using proposed and fabricated topologies to energize different applications and

then evaluating the efficiency and productivity of those applications can be

232

another task for future research. The efficiency of the power supply can be

evaluated individually or as part of the evaluation of the applications as a whole.

Applications such as lighting, ozonising, and biomedical can be considered in

this regard. Selecting of the most compatible and effective method to feed each

application can be explored, considering both load and power supply

specifications.

In addition, determining pulse train specifications, including pulse magnitude,

rising time and pulse repetition rate with respect to various applications will be

worthwhile. For example, a conventional MG can be utilized in order to change

pulse specifications. Subsequently, the most appropriate values of these

specifications can be distinguished for each application by measuring the

productivity in the load side.

zabihi sasan flexible high voltage pulsed power...

Documents