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
II
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
accommodate solid-state technology by revising, reconfiguring and
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
Fig.1. 56. Capacitor-Diode Voltage Multipliers (CDVM) (a). N-stage Cockcroft-
Walton Voltage Multiplier (b). N-stage Dickson charge pump (c). Another N-
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
Fig.2. 16. Laboratory prototype of pulsed power supply with double switch-
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
Fig.3. 11. Laboratory prototype of pulsed power supply with double switch-
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
Fig.8. 1. Capacitor-Diode Voltage Multipliers (CDVM) (a). N-stage Cockcroft-
Walton Voltage Multiplier (b). N-stage Dickson charge pump (c). Another N-
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.
2. Sasan Zabihi, Firuz Zare, Gerard Ledwich, Arindam Ghosh, Hidenory
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.
3. Sasan Zabihi, Firuz Zare, Gerard Ledwich, Arindam Ghosh, Hidenory
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
9. Sasan Zabihi, Firuz Zare, Gerard Ledwich, Arindam Ghosh, Hidenory
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
10. Sasan Zabihi, Firuz Zare, Gerard Ledwich, Arindam Ghosh, Hidenory
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
Load modelling control
AC-DC
ConverterVac
220 V50 Hz
CLoad
Current control method
D2
D3
D1
Dn
VinCin
Current Source Block Diagram Voltage Source
Current Source Voltage Source
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)
Fig.1. 30. Inductor current, capacitors and output voltages
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
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
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.
Fig.1. 56. Capacitor-Diode Voltage Multipliers (CDVM) (a). N-stage Cockcroft-Walton Voltage
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)
Input voltage of multiplier(v)Voltage over first capacitor of multiplier(v)Output voltage of multiplier(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)
Input voltage of multiplier(v)Voltage over first capacitor of multiplier(v)Output voltage of multiplier(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.
Fig.1. 63. An inverter supplying multiplier with variable frequency and amplitude
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)
Input voltage of multiplier(v)Voltage over first capacitor of multiplier(v)Output voltage of multiplier(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)
Input voltage of multiplier(v)Voltage over first capacitor of multiplier(v)Output voltage of multiplier(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
95
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
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.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.
Assuming the inductor current is constant during the capacitor charging mode,
the voltage stress can be calculated as follows.
)()( 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)
In repetitive pulse generation, a time interval is designated to the load to be
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)
Fig.2. 17. Inductor current, capacitors and output voltages
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.
120
121
I
Statement of Contribution of Co-Authors
The authors listed below have certified that:
I. 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 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.
Principal Supervisor Confirmation
I have sighted email or other correspondence from all Co-authors confirming their certifying
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
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 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. Conf igura t ion and analyses
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
Load modelling control
AC-DC
ConverterVac
220 V50 Hz
CLoad
Current control method
D2
D3
D1
Dn
V inCin
Current Source Block Diagram Voltage Source
Current Source Voltage Source
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.
Assuming the inductor current is constant during the capacitor charging mode,
the voltage stress can be calculated as follows.
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)
In repetitive pulse generation, a time interval is designated to the load to be
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.
3.5. Exper imenta l resul ts
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)
Fig.3. 12. Inductor current, capacitors and output voltages
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.
143
144
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 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
Principal Supervisor Confirmation
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*
*School of Electrical Engineering, Queensland University of Technology, GPO
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
Novel Marx topology(Bidirectional diode-capacitor units)
220 V50 Hz Single leg
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
Vin L C1,2,3,4 finv fr
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.
4.5. Exper imenta l resul ts
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
Vin L C1,2,3,4 finv fr
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
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. 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.
[4] 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.
[5] 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.
[6] 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.
[7] 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.
[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.
161
162
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 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
Hidenori Akiyama Provided us with general information about pulsed power supply specifications and its application demands.
Principal Supervisor Confirmation
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
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 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. Conf igura t ion and analyses
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
Two leg diode-capacitor unit 1 Two leg diode-capacitor unit 2Modified positive Buck-Boost Converter
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
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. 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.
[5] 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.
[6] 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.
[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.
[14] 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.
[15] 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.
[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.
180
181
I
Statement of Contribution of Co-Authors
The authors listed below have certified that:
I . 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:
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.
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
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 ""'
_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†
*School of Electrical Engineering, Queensland University of Technology, GPO
Box 2434, Brisbane, Australia
† Kumamoto University, Japan
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.
193
194
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 ~e 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 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
Principal Supervisor Confirmation
I have sighted email or other correspondence from all Co-authors confirming their certifying
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*
*School of Electrical Engineering, Queensland University of Technology, GPO
Box 2434, Brisbane, Australia
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
)
C1 voltageC2 voltageOutput 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 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)
C1 voltageC2 voltageOutput voltage
0 0.05 0.1 0.15-200
0
200
400
(d)Time(ms)
C1,
C2 &
Ou
tpu
t vo
ltag
es(V
)
C1 voltageC2 voltageOutput voltage
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
)
C1 voltageC2 voltageOutput voltage
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
)
C1 voltageC2 voltageOutput voltage
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
)
C1 voltageC2 voltageOutput voltage
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.
206
207
I
Statement of Contribution of Co-Authors
The authors listed below have certified that:
I. 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 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.
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
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.
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†
*School of Electrical Engineering, Queensland University of Technology, GPO
Box 2434, Brisbane, Australia
† Kumamoto University, Japan
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
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
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.
Fig.8. 1. Capacitor-Diode Voltage Multipliers (CDVM) (a). N-stage Cockcroft-Walton Voltage
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)
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)
Inpu
t an
d c
apa
cito
rs v
olta
ges(
v)
Input voltage of multiplier(v)Voltage over first capacitor of multiplier(v)Output voltage of multiplier(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)
Input voltage of multiplier(v)Voltage over first capacitor of multiplier(v)Output voltage of multiplier(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)
Input voltage of multiplier(v)Voltage over first capacitor of multiplier(v)Output voltage of multiplier(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)
Input voltage of multiplier(v)Voltage over first capacitor of multiplier(v)Output voltage of multiplier(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.
Fig.8. 9. An inverter supplying multiplier with variable frequency and amplitude
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.
219
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
accommodate solid-state technology by revising, reconfiguring and
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
Load modelling control
CLoad
Current control method
RD1
RD2
AC-DC
ConverterVac
220 V50 Hz
VinCin
Current Source Block Diagram Voltage Source
Current Source Voltage Source
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
Load modelling control
AC-DC
ConverterVac
220 V50 Hz
CLoad
Current control method
D2
D3
D1
Dn
V inCin
Current Source Block Diagram Voltage Source
Current Source Voltage Source
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 the 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-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
Novel Marx topology(Bidirectional diode-capacitor units)
220 V50 Hz Single leg
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
Insulates the load side from the input side of power supply during pulse generation (controlling power flow during undesired arc phenomena)
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
Two leg diode-capacitor unit 1 Two leg diode-capacitor unit 2Modified positive Buck-Boost Converter
L2 L3
+
-
+
-
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 the repetitive operation of the Marx 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: 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.