Practical Design of a High-Voltage Pulsed Power Supply Implementing Sic Technology for Atmospheric Pressure Plasma Reactors

applied sciences

Article Practical Design of a High-Voltage Pulsed Power Supply Implementing SiC Technology for Atmospheric Pressure Plasma Reactors

Jacek Kołek and Marcin Hołub *

Faculty of Electrical Engineering, West Pomeranian University of Technology, 70-313 Szczecin, Poland; [email protected] * Correspondence: [email protected]

Received: 18 March 2019; Accepted: 2 April 2019; Published: 6 April 2019

Featured Application: Power supply construction for dielectric barrier or corona discharge, atmospheric pressure plasma reactors.

Abstract: Non-thermal plasma reactors offer numerous advantages and are used in a wide variety of applications. Exemplary ﬁelds of use include exhaust gas or water quality improvement, surface processing, agriculture, or medical processes. Two of the most popular discharge chamber constructions are dielectric barrier discharge (DBD) and pulsed corona discharge (PCD, barrierless) plasma reactors. Various power supply constructions are presented in the literature with varying complexity and operating principles. A simple, power electronic, pulsed power supply construction is presented in this work, implementing resonant power switch operation, modern silicon carbide (SiC) power semiconductors and a Tesla coil-like transformer design. The power supply construction is immune to short circuits in the reactor and can be used for both types of reactor constructions. The principles of operation, simulation results, and prototype test veriﬁcations are presented with main power supply characteristics.

Keywords: cold plasma power supply; high-voltage power supply; pulsed power supply; cold plasma sources

1. Introduction Non-thermal plasma discharge properties are used in a wide variety of processes. The main applications include agricultural use [1], medical processing [2], water and exhaust treatment [3,4], waste air quality improvement, and deodorization [5–7]. Further research is needed to provide application possibilities in different fields [4]. Regarding the plasma reactor constructions, dielectric barrier discharge (DBD, electrodes separated by a dielectric barrier) and corona discharge (CD, separated electrodes with no dielectric barrier) reactors are most widely used for cold plasma discharge generation under atmospheric pressure. The different properties of both constructions are an important factor in choosing a certain one for industrial applications and have a strong influence on power supply system designs. In order to efficiently supply a plasma module, many different power supply topologies and electrical waveforms exist. It is possible to use a simple, high-voltage sine wave source (from several to several tens of kV) with a frequency of 50 Hz to 50 kHz to generate the plasma in a DBD reactor. To generate plasma in CD reactors, it is necessary to have a high-voltage pulsed source (from several to several tens of kV), with a high-voltage steepness (typically in the range from several hundred to several thousand volts/ns [8]. Since the construction of CD reactors is much simpler and they have

Appl. Sci. 2019, 9, 1451; doi:10.3390/app9071451 www.mdpi.com/journal/applsci Appl.Appl. Sci. Sci.2019 2019,,9 9, 1451FOR PEER REVIEW 2 of2 12

smaller flow resistance than a typical DBD reactor, a highly efficient topology is required for a High a smaller flow resistance than a typical DBD reactor, a highly efficient topology is required for a High Voltage (HV) generator. Additionally, when using a HV pulse generator for DBD reactors, it is also Voltage (HV) generator. Additionally, when using a HV pulse generator for DBD reactors, it is also possible to significantly improve the energy efficiency of plasma generation [9]. Currently, in the possiblegeneration to significantlyof high-voltage improve pulse systems the energy of rotating efficiency gas of gaps, plasma gas generationgaps with trigger [9]. Currently, electrodes in or the generationthyratrons ofas well high-voltage as fast ionizing pulse dynistors systems of(FIDs) rotating are often gas gaps,used. gasFrom gaps the view withpoint trigger of the electrodes dynamic or thyratronsparameters as of well modern as fast power ionizing transistors dynistors (such (FIDs) as Si areC or often GaN used. MOSFETs) From the, the viewpoint implementation of the dynamic of HV parameterspulse gener ofators modern is possible. power transistors (such as SiC or GaN MOSFETs), the implementation of HV pulseTh generatorsis paper is is organized possible. as follows: In the first section, the description of proposed high-voltage resonantThis paper converter is organized with simul asation follows: results In the is firstprovided, section, followed the description by the design of proposed and construction high-voltage resonantdetails, a converternd suggestions with simulation are discussed results for isa prototype provided, power followed supply. by the Finally design, measurement and construction results details, of andthe suggestionsexperimental are setup discussed and practical for a prototype verification power of properties supply. Finally, with different measurement types resultsof loads of are the experimentalpresented. setup and practical verification of properties with different types of loads are presented.

2.2. MaterialsMaterials andand MethodsMethods AA new,new, simple simple topologytopology ofof thethe pulsedpulsed power supply is depicted in in Figure Figure 1.1. It consists of of three three mainmain conversion conversion stages:stages: (a)a) a DC voltage source source;; b) (b) a a resonant resonant charger charger circuit circuit constituting constituting the the initial initial pulsepulse energyenergy by the voltage voltage control control of of the the capacitor capacitor C2 C; and2; and c) an (c) out anput output pulse pulse shaper shaper with Tesla with coil Tesla- coil-likelike transformer. transformer. In order In order to investigate to investigate the basic the basicoperation operation,, role, and role, properties and properties of each of stage each, stage,they theywill willbe discussed be discussed separately separately,, in detail in detail,, in the in following the following subsections. subsections.

FigureFigure 1. 1. ProposedProposed pulsed pulsed power power supply for non-thermalnon-thermal plasma reactors. reactors. Three Three main main stages stages are are marked:marked: DC voltage voltage source source (blue (blue); );resonant resonant capacitor capacitor charger charger (red ();red and); andoutput output pulse pulseshaping shaping unit unit(green (green). ).

2.1.2.1. TheThe DCDC VoltageVoltage SourceSource TheThe firstfirst partpart ofof thethe proposedproposed high-voltagehigh-voltage power supply is is a a DC DC voltage voltage source source whose whose output output voltagevoltage value value may may be be fixed fixed at at several several hundreds hundred ofs of volts volts (using (using a passive a passive rectifier rectifier circuit circuit or anor activean active PFC stage),PFC stage) or variable, or variable up to up similar to similar values values (300–650 (300‒ V650 is theV is most the most useful useful and commonlyand commonly used used range range [10]). In[10 a]) specific. In a specific application, application this, this topology topology can can be be fed fed directly directly from from the the mains mains withwith a simple simple rectifier rectifier circuitcircuit with with capacitor capacitor bank. bank. In In other other cases—especially cases—especially when when galvanic galvanic isolation isolation from from the the power power grid grid is required—anis required—an isolated isolated AC/DC AC/DC power power supply supply should should be be used. used. InIn order order to verify verify the the laboratory laboratory prototype prototype properties properties for for different different input input voltage voltage values values,, the thelaboratory laboratory TDK TDK Lambda Lambda GEN600 GEN600-2.6-2.6 SMPS SMPS (switching (switching mode mode power power supply) supply) voltage voltage source source was was used.used. ThisThis providedprovided stable voltage with a a control control range range from from 0 0 to to 650 650 V V DC DC,, and and 2.6 2.6 A A of of average average outputoutput current.current. TheThe highhigh efficiencyefficiency and galvanic isolation o off the the used used power power supply supply were were additional additional advantageousadvantageous features.features.

2.2.2.2. TheThe ResonantResonant CapacitorCapacitor ChargerCharger TheThe presentedpresented powerpower supplysupply shouldshould generate high-voltagehigh-voltage pulses pulses with with adjustable adjustable ampli amplitudetude andand repetitionrepetition ratio. ratio. The The repetition repetition frequency frequency of the of output the output voltage voltage pulses pulses was set was in the set range in the of range250‒ of2000 250–2000 Hz. The Hz. role The of the role resonant of the resonantcapacitor capacitorcharger is charger to accumulate is to accumulate the energy theneeded energy to generate needed a to generatesingle voltage a single pulse voltage in pulsethe electric in the electric field of field a capacitor of a capacitor bank bank.. The The charging charging process process must must have have acceptableacceptable accuracyaccuracy andand hashas toto bebe fast enough to finish finish in the the available available time time (pause) (pause) between between output output pulses.pulses.A A simple,simple, resonantresonant constructionconstruction with two power switch switcheses w wasas chosen chosen to to reduce reduce the the switching switching lossloss of of the the used used powerpower semiconductors,semiconductors, andand toto make the charging process as as efficien efficientt as as possible possible.. Appl. Sci. 2019, 9 FOR PEER REVIEW 3 Appl. Sci. 2019, 9, 1451 3 of 12 The proposed topology acts in three repetitive intervals, as shown in Figure 2. The analytical descriptionAppl. Sci. 2019of this, 9 FOR part PEER of REVIEWthe topology and each separate interval were presented in [11]. 3 The proposed topology acts in three repetitive intervals, as shown in Figure2. The analytical description The of p thisroposed part topology of the topology acts in three and eachrepetitive separate intervals interval, as shown were presentedin Figure 2 in. The [11 ].analytical description of this part of the topology and each separate interval were presented in [11].

(a) (b) (a) (b) Figure 2. (a) The three intervals of capacitor charger operation and (b) simulated waveforms. PLECS® ® ® simulationFigureFigure 2. ( a2. platform) ( Thea) The three t hree was intervals intervals used for of of capacitor calculations.capacitor chargercharger Interval operation operation 1 (top) and and— (bthe ()b s) imulated simulatedenergy ofwaveforms waveforms. the input. PLECS source PLECS is simulationsimulation platform platform was was used used for for calculations. calculations. Interval Interval 1 1 (top) (top)—the—the energy energy of ofthethe input input source source is is resonantly transferred to capacitor (with voltage boost effect) C1. Interval 2 (middle)—the energy of resonantlyresonantly transferred transferred to to capacitor capacitor (with (withvoltage voltage boost effect) effect) C C1. Interval. Interval 2 (middle) 2 (middle)—the—the energy energy of of capacitor C1 is resonantly transferred to C2 and increases the voltage1 across C2. Interval 3 (bottom)— capacitorcapacitor C isC1 resonantly is resonantly transferred transferred to to C Cand2 and increases increases the the voltage voltage acrossacross CC2. Interval 3 3 (bottom) (bottom)—the— the resonant1 energy stored in the magnetic2 field of L2 is used to charge C2 using the freewheeling resonantthe resonant energy storedenergy instored the magnetic in the magnetic ﬁeld of field L is of used L2 is to used charge to charge C using C2 theusing freewheeling the freewheeling diode D . diode D3. 2 2 3 diode D3. The proposed resonant charger consists of two series of monoresonant RLC (resistive – inductive – The Theproposed proposed resonant resonant charger charger consist consists sof of two two series series of monoresonant RLC RLC (resistive (resistive – inductive – inductive capacitive) circuits connected in a cascade. In the real power supply construction, inductors L1 and – capacitive)– capacitive) circuits circuits connected connected in ina cascade.a cascade. In In the the realreal power supplysupply construction construction, inductors, inductors L1 andL1 and L2 were made of single MS-250026-2 micrometals cores with 15 turns each using litz wire windings L2 wereL2 were made 2made of single of single MS MS-250026-250026-2- 2m micrometalicrometalss corescores with 1515 turnsturns each each using using litz litz wire wire windings windings of ~5.8 mm of2 copper wire to obtain an inductance of 22 uH. R1 and R2 are the parasitic resistances of ~5.8of ~mm5.8 2mm of copper of copper wire wire to toobtain obtain an an inductance inductance ofof 22 uH.. RR11 andand R R2 2are are the the parasitic parasitic resistances resistances of inductors, wires, and printed circuit board tracks, in the real construction, the measured value of inductors,of inductors, wires, wires, and and printed printed circuit circuit board board tracks tracks,, inin thethe real constructionconstruction, the, the measured measured value value was was was 17 mΩ. The core material used is called SENDUST, and it is designed for high-frequency power 17 mΩ.17 mΩ. The Thecore core material material used used is is called called SENDUST SENDUST,, andand it isis designeddesigned for for high high-frequency-frequency power power applicationapplication inductors. inductors. The The C 1Ccapacitance1 capacitance hashas a measured value value of of 12 12 nF nF in inthe the prototype prototype and and is made is made application inductors. The C1 capacitance has a measured value of 12 nF in the prototype and is made of WIMAof WIMA FKP FKP capacitors capacitors connected connected in in series, series, duedue to the high high dV/dt dV/dt operation operation of that of that power power supply supply of WIMA FKP capacitors connected in series, due to the high dV/dt operation of that power supply stage.stage The. The C2 Ccapacitance2 capacitance is is a a 1 1 uF uF capacitor capacitor bank made made of of nine nine 1 1uF uF WIMA WIMA FKP FKP capacitors capacitors connected connected stage. The C2 capacitance is a 1 uF capacitor bank made of nine 1 uF WIMA FKP capacitors connected in ain mixed a mixed (parallel (parallel series) series) way, way due, due to to the the highhigh valuevalue of the the capacitor capacitor current current.. Figure Figure 3 shows3 shows the the in a mixed (parallel series) way, due to the high value of the capacitor current. Figure 3 shows the constructed resonant capacitor charger module and the C2 capacitor bank. constructed resonant capacitor charger module and the C2 capacitor bank. constructed resonant capacitor charger module and the C2 capacitor bank.

(a) (b)

Figure(a) 3. (a) The resonant capacitor charger and (b) the C2 capacitor(b) bank.

FigureFigure 3. 3. (a(a) )The The r resonantesonant capacitor capacitor charger charger and and ( (bb)) the the C C22 capacitorcapacitor bank. bank. Appl. Sci. 2019, 9, 1451 4 of 12 Appl. Sci. 2019, 9 FOR PEER REVIEW 4

For those values of the real L and C components, the charging of C1 capacitance takes 1.5 µs and For those values of the real L and C components, the charging of C1 capacitance takes 1.5 µs and the energy transfer between C1 and C2 takes 1.5 us. To fully charge the C2 capacitor, 30 repetitions of the energy transfer between C1 and C2 takes 1.5 µs. To fully charge the C2 capacitor, 30 repetitions three charging intervals are needed. The power switches S1 and S2 were 1700 V, 72 A, 45 mΩ SiC of three charging intervals are needed. The power switches S1 and S2 were 1700 V, 72 A, 45 mΩ SiC power MOSFETs C2M0045170D from Cree. The diodes D1–D3 were 1700 V, 10 A SiC power Schottky power MOSFETs C2M0045170D from Cree. The diodes D1–D3 were 1700 V, 10 A SiC power Schottky diodes GB10MPS17 from GeneSiC Semiconductor Inc. The diodes D1 and D2 prevent the reverse flow diodes GB10MPS17 from GeneSiC Semiconductor Inc. The diodes D1 and D2 prevent the reverse ﬂow of energy and,and, therefore,therefore, a low reverse recovery current feature ofof SiC diodes could be used.used. The role of the diode D3 is to block the negative current flow of the resonating tank to exclude negative of the diode D3 is to block the negative current ﬂow of the resonating tank to exclude negative voltages voltages on the capacitor C1. on the capacitor C1.

2.3. The O Outpututput P Pulseulse S Shapinghaping UnitUnit The most important part of the presentedpresented high-voltagehigh-voltage pulsed converter design is the output stage based on the TeslaTesla transformertransformer concept.concept. This allows for shaping of the output voltage pulse through the the resonance resonance of of the the circuit circuit’s’s L Land and C Ccomponents components,, as asschematically schematically depicted depicted in Figure in Figure 4. To4. Toobtain obtain the the highest highest amplitu amplitudesdes of ofoutput output voltage voltage pulse pulse without without mutual mutual oscillations oscillations between between pulses pulses,, the magnetic coupling coefﬁcientcoefficient kk shouldshould be equal toto approx. 0.6,0.6, and the rules of E Equationquation (1) must be fulﬁlledfulfilled [12,,1313]].. L ·C = L ·C (ω = ω ) (1) LL1∙C2 = LL2∙C3 (ω1= ω2) L1 2 L2 3 1 2 (1)

FigureFigure 4. 4. OutOutputput pulse pulse shaping shaping circuit circuit with T Teslaesla coil-likecoil-like cored transformer.transformer.

Most Tesla-basedTesla-based transformers are coreless (air coupling)coupling) but there are also constructions with magnetic cores cores [ [1144].]. Low Low permeability permeability m magneticagnetic cores cores (e.g (e.g.,., distributed distributed gap gap cores) cores) were were used used in the in thepresented presented design design,, instead instead of air of coupling air coupling,, which which allows allows for significant for significant redu reductionction in the in size the sizeof the of thetransforme transformer,r, although although there there is still is still a aproblem problem with with the the external external magnetic magnetic fieldfield (electromagnetic(electromagnetic interference,interference, i.e.,i.e., EMI) [[15]15].. A l lowow value in the transformer transformer’s’s magnetic coupling factor results in large values of leakageleakage inductanceinductance for the transformer windings. Magnetic field field lines of these two leakage inductances areare still still closed closed through through air, whichair, wh resultsich results in distortions in distortions of the magnetic of the fieldmagnetic in transformer field in surroundings.transformer surroundings The external. magneticThe external field magnetic of the transformer field of the means transformer that in addition means that to electromagnetic in addition to compatibilityelectromagnetic problem, compatibility there will problem also be, thereadditional will alsopower be lossesadditional since externalpower losses magnetic since fields external will interactmagnetic with field magnetics will interact elements, with e.g., magnetic transformer elements or converter, e.g., transformer enclosures, or etc. converter [1,16]. Inenclosure order tos, keep etc. the[1,16] magnetic. In order field to of keep theTesla the magnetic coil-like transformer field of the inside Tesla its coil magnetic-like transformer construction, inside a core-based its magnetic Tesla transformerconstruction design, a core- isbased proposed, Tesla astransformer shown in Figuredesign5 is. proposed, as shown in Figure 5. Proper selection of the number of cores and magnetic properties of magnetic cores led to a defined magnetic coupling. In the constructed prototype, seven cores were used in a 2–3–2 arrangement, which led to the desired k = 0.6 coupling coefficient. The term 2–3–2 means that two cores were used for primary side leakage inductance LL1, three cores were used for main part of the transformer T1, and two cores were implemented for secondary side leakage inductance LL2 of the design transformer. The different numbers and properties of the cores and their proper arrangement allows for reaching any defined magnetic coupling ratios, such as k = 0.6(6), k = 0.75, k = 0.8, k = 0.9, etc. This design idea could be easily implemented in LLC DC/DC converter design to simplify the magnetics. An important aspect of the design was to use the same magnetic core material for leakage inductances and the main transformer. This means that the system will be using the same part of the magnetic hysteresis loop of the material to maintain a constant coupling ratio during transformer operation. For the same reason, the number of primary windings, N1 (windings of the primary leakage

(a) (b) Appl. Sci. 2019, 9 FOR PEER REVIEW 4

For those values of the real L and C components, the charging of C1 capacitance takes 1.5 µs and the energy transfer between C1 and C2 takes 1.5 us. To fully charge the C2 capacitor, 30 repetitions of three charging intervals are needed. The power switches S1 and S2 were 1700 V, 72 A, 45 mΩ SiC power MOSFETs C2M0045170D from Cree. The diodes D1–D3 were 1700 V, 10 A SiC power Schottky diodes GB10MPS17 from GeneSiC Semiconductor Inc. The diodes D1 and D2 prevent the reverse flow of energy and, therefore, a low reverse recovery current feature of SiC diodes could be used. The role of the diode D3 is to block the negative current flow of the resonating tank to exclude negative voltages on the capacitor C1.

2.3. The Output Pulse Shaping Unit The most important part of the presented high-voltage pulsed converter design is the output stage based on the Tesla transformer concept. This allows for shaping of the output voltage pulse through the resonance of the circuit’s L and C components, as schematically depicted in Figure 4. To obtain the highest amplitudes of output voltage pulse without mutual oscillations between pulses, the magnetic coupling coefficient k should be equal to approx. 0.6, and the rules of Equation (1) must be fulfilled [12,13].

LL1∙C2 = LL2∙C3 (ω1= ω2) (1)

Figure 4. Output pulse shaping circuit with Tesla coil-like cored transformer.

Most Tesla-based transformers are coreless (air coupling) but there are also constructions with magnetic cores [14]. Low permeability magnetic cores (e.g., distributed gap cores) were used in the presented design, instead of air coupling, which allows for significant reduction in the size of the transformer, although there is still a problem with the external magnetic field (electromagnetic interference, i.e., EMI) [15]. A low value in the transformer’s magnetic coupling factor results in large values of leakage inductance for the transformer windings. Magnetic field lines of these two leakage Appl.inductances Sci. 2019, 9 ,are 1451 still closed through air, which results in distortions of the magnetic field5 of 12in transformer surroundings. The external magnetic field of the transformer means that in addition to electromagnetic compatibility problem, there will also be additional power losses since external inductancemagnetic fields and primarywill interact side ofwith the magnetic transformer), elements, should e.g., be equal. transformer Similarly, or N2converter (for secondary enclosure leakages, etc. inductance[1,16]. In and order secondary to keep side the of transformer) magnetic field should of bethe equal. Tesla coil-like transformer inside its magnetic

Appl. Sci. 2019, 9 FOR PEER REVIEW 5

Figure 5. (a) A cored Tesla transformer construction concept with leakage inductances LL1, LL2, and

the main magnetizing part T1; (b) the output pulse shaping unit simulation results, from the top:

switch S3 control signal, C2 capacitor voltage, switch current, C3 capacitor voltage and current.

Proper selection of the number of cores and magnetic properties of magnetic cores led to a defined magnetic coupling. In the constructed prototype, seven cores were used in a 2–3–2 arrangement, which led to the desired k = 0.6 coupling coefficient. The term 2–3–2 means that two cores were used for primary side leakage inductance LL1, three cores were used for main part of the transformer T1, and two cores were implemented for secondary side leakage inductance LL2 of the design transformer. The different numbers and properties of the cores and their proper arrangement allows for reaching any defined magnetic coupling ratios, such as k = 0.6(6), k = 0.75, k = 0.8, k = 0.9, etc. This design idea could be (easilya) implemented in LLC DC/DC converter(b) design to simplify the magnetics. FigureAn important 5. (a) A cored aspect Tesla of transformer the design construction was to use conceptthe same with magnetic leakage inductancescore material LL1 for,L L2leakage, and the inductancesmain magnetizing and the partmain T transformer.1;(b) the output This pulse means shaping that the unit system simulation will be results, using fromthe same the top: part switch of the S3 magneticcontrol signal,hysteresis C2 capacitor loop of the voltage, material switch to maintain current, Ca3 constantcapacitor coupling voltage andratio current. during transformer operation. For the same reason, the number of primary windings, N1 (windings of the primary leakageThe prototype inductance was and built primary with side seven of theMS-250026-2 transformer), micrometal should becores equal. with Similarly, a reduced N2 (for magnetic permeabilitysecondary leakage of µr = inductance 60. The primary and secondary winding side is madeof transformer) of 64 coils should connected be equal. in parallel. Each of these coils wereThe madeprototype as a was single built turn with of seven ~1 mm MS-250026-2 litz wire.2 micrometal The secondary cores windingwith a reduced is made magnetic of 50 turns of thepermeability same litz wire of μ thatr = 60. was The used primary in the winding primary is made side. of The 64 coils cores connected and windings in parallel. of the Each primary of these leakage 2 inductancecoils were and made primary as a single side turn of the of transformer~1 mm litz wire. were The insulated secondary by winding epoxy resin is made impregnation. of 50 turns of Figure 6 the same litz wire that was used in the primary side. The cores and windings of the primary leakage shows the transformer components before the ﬁnal assembly and the main transformer part mounted inductance and primary side of the transformer were insulated by epoxy resin impregnation. Figure with the primary leakage inductance and the primary windings. 6 shows the transformer components before the final assembly and the main transformer

(a) (b)

FigureFigure 6. (a) The6. (a) magnetic The magnetic transformer transformer parts parts and and (b )(b the) the transformer transformer with with mountedmounted primary primary windings. windings. Due to high current amplitudes on the primary side of the transformer, eight parallel 1200 V, 100 A IXA70I1200NADue to high IXYS current IGBTs amplitudes and eight on 1200 the primary V, 20 A side CoolSiC of the™ transformer,IDH20G120C5 eight parallel Inﬁneon 1200 antiparallel V, 100 A IXA70I1200NA IXYS IGBTs and eight 1200 V, 20 A CoolSiC™ IDH20G120C5 Infineon diodes were used. Alike C2, the C3 capacitance is made of 50 WIMA FKP 10 nF capacitors connected antiparallel diodes were used. Alike C2, the C3 capacitance is made of 50 WIMA FKP 10 nF capacitors in parallel, due to the high voltage amplitudes on the output of the power supply. The S3 switching connected in parallel, due to the high voltage amplitudes on the output of the power supply. The S3 module and the C output capacitor bank are presented in Figure7. switching module3 and the C3 output capacitor bank are presented in Figure 7. Appl. Sci. 2019, 9, 1451 6 of 12

Appl. Sci. 2019, 9 FOR PEER REVIEW 6

(a) (b)

FigureFigure 7. (a 7.) (Thea) The S3 Sswitching3 switching module module (8 (8 IGBTs IGBTs (Isolated Gate Gate Bipolar Bipolar Transistors) Transistors) and and 8 SiC 8 SiC antiparallel (a) (b) antiparalleldiodes arediodes under are theunder copper the co busbarpper busbar visible visible at the at top); the (top);b) the (b) C the3 output C3 output capacitance. capacitance.

Figure 7. (a) The S3 switching module (8 IGBTs (Isolated Gate Bipolar Transistors) and 8 SiC 3 TheThe powerantiparallel power supply supply diodes load arefor load undera DBD for the a or DBDcopper CD orplasma busbar CD visible plasmareactor at theis reactor to top); be (connectedb is) the to C be3 output connected in parallel capacitance. into parallelthe C to the capacitanceC3 capacitance;; hence, to hence, achieve to the achieve best energy the best transfer energy efficiency transfer, the efﬁciency, reactor capacity the reactor should capacity be taken should be intotaken account into Thein accountthe power power supply in supply the powerload design for supply a ,DBD especially or design, CD inplasma case especially of reactor DBD in reactorsis caseto be of connected that DBD exhibit reactors in largeparallel that internal to exhibit the C3 large capacityinternalcapacitance;. capacity. hence, to achieve the best energy transfer efficiency, the reactor capacity should be taken into account in the power supply design, especially in case of DBD reactors that exhibit large internal 3. Results3. Resultscapacity.

The3.The constructedResults constructed prototype prototype of of the the high high-voltage-voltage pulsedpulsed powerpower supply, supply implementing, implementing SiC Si technologyC technologyand the and Tesla the coil-like Tesla coil transformer-like transformer design, design is depicted, is depicted in Figure in Figure8. 8.

(a) (b) (a) (b) Figure 8. (a) The constructed high-voltage power supply prototype and (b) its main internal componentsFigureFigure 8. (equivalent(a ) 8. The (a) constructedThe circuit constructed component high-voltage high-voltage names) power. power supply supply prototype prototype and (b and) its ( mainb) its internal main internal components (equivalentcomponents circuit (equivalent component circuit names). component names). All of the waveforms presented below were recorded using the LeCroy WaveRunner 6100A All of the waveforms presented below were recorded using the LeCroy WaveRunner 6100A digital osAllcilloscope. of the waveformsThe voltage presenteds on C1 and below C2 were were measured recorded using using a passive the LeCroy Tektronix WaveRunner P5100A 6100A digital oscilloscope. The voltages on C1 and C2 were measured using a passive Tektronix P5100A probe,digital and oscilloscope.the currents on The L1 and voltages L2 were on measured C1 and C using2 were LeCroy measured CP031 using Hall effect a passive current Tektronix probes. P5100A probe, and the currents on L1 and L2 were measured using LeCroy CP031 Hall effect current probes. Theprobe, output and voltage the currentswas measured on L usingand L a passivewere measured Tektronix using P6015A LeCroy high-voltage CP031 probe. Hall effect The c currenturrent probes. The output voltage was measured1 using2 a passive Tektronix P6015A high-voltage probe. The current on the primary and secondary side of transformer was measured using a type 6585 current monitor The outputon the primary voltage and was secondary measured side using of transformer a passive was Tektronix measured P6015A using high-voltage a type 6585 current probe. monitor The current by Pearson. on theby primaryPearson. and secondary side of transformer was measured using a type 6585 current monitor byFigure Pearson. 9Figure presents 9 presents voltage voltage and current and current waveforms waveforms for C forand C Land elements L elements of the of resonantthe resonant capacitor capacitor charger chargerunit.Figure The 9unit. presentspresented The presented voltage waveforms waveforms and currentwere recorded were waveforms recorded for 400 for for V 400 Cof andinputV of Linput voltage elements voltage and of and under the under resonant high high load load capacitor conditions.chargerconditions. The unit. charging The The presented charging process process waveformsstarted start at timeed were at zero time recorded, zero,just after just for afterthe 400 previous the V of previous input output voltage output pulse pulse and generation undergeneration high load process was finished. During normal operation, the initial voltage across capacitor C2 did not drop to conditions.process was The finished. charging During process normal started operation, at time the zero, initial just voltage after theacross previous capacitor output C2 did pulse not drop generation to Appl.Appl. Sci. Sci.2019 2019, 9, 9 1451 FOR PEER REVIEW 7 of 12 7

zero because the unused energy from the output pulse generation process always recharge it. To process was ﬁnished. During normal operation, the initial voltage across capacitor C2 did not drop to charge the capacitor C2 from 0 V to its final voltage (400 V) 30 repetitions were needed, while in zero because the unused energy from the output pulse generation process always recharge it. To charge normal operation (under lighter loads), the typical number of repetitions was around 15–20. the capacitor C2 from 0 V to its ﬁnal voltage (400 V) 30 repetitions were needed, while in normal operation (under lighter loads), the typical number of repetitions was around 15–20.

Figure 9. The measured voltage and current waveforms of capacitor resonant charger (measured at mainFigure L and 9. The C components—refer measured voltage to and Figure current2). waveforms of capacitor resonant charger (measured at main L and C components—refer to Figure 2). In the worst case (the lowest initial voltage of C1 and highest amplitude of inductor L1 current) the efficiencyIn the wors of energyt case transfer (the lowest from initial DC voltage voltage source of C1 to and capacitor highest C amplitude1 was above of 98%. inductor With L the1 current same ) efficiency,the efficiency the energy of energy was transferredtransfer from from DC capacitor voltage source C1 to C to2. Thecapacitor overall C1 efficiency was above of 98%. the resonant With the capacitorsame efficiency charger,unit the was energy typically was transferred higher than from 96%. capacitor C1 to C2. The overall efficiency of the resonantFigure capacitor 10 shows charger voltage unit andwas currenttypically waveformshigher than of96%. the output pulse shaping circuitry. The waveformsFigure 10 showswere recorded voltage and for current300 V of waveforms input voltage of the (for output higher pulse voltage shaping values, circuitry energy. The consumptionwaveforms were in the recorded load significantly for 300 V of affects input currentvoltage waveforms).(for higher voltage The process values started, energy at consumption time zero, whenin the the load capacitor significantly C2 was affects charged current to its waveforms). initial value. The After process switching started S3 aton, time the energyzero, when stored the incapacitor capacitor C C2 wa2 wass charged transferred to its initial through value. the After transformer’s switching magnetic S3 on, the circuit energy to stored the capacitor in capacitor and C2 parallel-connectedwas transferred through load, either the atransformer DBD or PCD’s magnetic reactor. For circuit 400 V to of the input capacitor voltage, and the parallel amplitude-connected of the generatedload, either high-voltage a DBD or PCD pulse reactor. was around For 400 20 V kV, of due input to voltage the high, the transformer amplitude winding of the generated ratio (n = high 50). - Anyvoltage energy pulse which wa wass around not consumed 20 kV, due in theto the load high for gastransformer ionization winding backflowed ratio to (n capacitor = 50). Any C2 .energy which was not consumed in the load for gas ionization backflowed to capacitor C2. Appl.Appl. Sci. Sci.2019 2019, 9,, 9 1451 FOR PEER REVIEW 8 of 12 8 Appl. Sci. 2019, 9 FOR PEER REVIEW 8

Figure 10. The voltage and current waveforms of the high-voltage pulse shaping unit. Figure 10. TheThe v voltageoltage and and current current waveform waveforms ofs ofthe the high high-voltage-voltage pulse pulse shaping shaping unit. unit. For an ideal circuit, the current pulses should be quasi-sinusoidal. Distortions, especially visible in theFor currentFor anan idealideal of the circuit primary,, thethe current current side of pulses pulses the transformer, should should be be quasi werequasi-sinusoidal.- causedsinusoidal. by Distortions, the Distortions, parasitic especially capacitancesespecially visible visible of transformerinin thethe currentcurrent windings of the [primaryprimary17]. si sidede of of the the transformer transformer, we, were reca usedcaused by bythe the parasitic parasitic capacit capacitancesances of of transformer windings [17]. transformerThe energy windings transfer [17] efficiency. from C2 to C3 was above 84% and, therefore, the whole process of The energy transfer efficiency from C2 to C3 was above 84% and, therefore, the whole process of outputThe pulse energy generation transfer (energy efficiency transfer from fromC2 to C23 wato Cs 3aboveand back84% fromand, therefore C3 to C2), wasthe whole achieved process with of output pulse generation (energy transfer from C2 to C3 and back from C3 to C2) was achieved with an anoutput efficiency pulse of generation about 71%. (energy transfer from C2 to C3 and back from C3 to C2) was achieved with an efficiency of about 71%. efficiencyTo test of the about prototype 71%. in real application conditions, two types of loads were used. At first, a PCD To test the prototype in real application conditions, two types of loads were used. At first, a PCD reactorTo was test created the prototype as a matrix in real of application HV electrode conditions needles, two above types the of low-potentialloads were used. flat At metal first, plate.a PCD reactor was created as a matrix of HV electrode needles above the low-potential flat metal plate. The The electrical capacity of the reactor was 27 pF. Figure 11 shows the typical discharge formation reactorelectrical was capacity created of as the a matrixreactor ofwas HV 27 electrode pF. Figure needles 11 shows above the typicalthe low discharge-potential formation flat metal during plate. The during operation. electricaloperation. capacity of the reactor was 27 pF. Figure 11 shows the typical discharge formation during operation.

FigureFigure 11. 11.PCD PCD reactor reactor supplied supplied byby thethe high-voltagehigh-voltage pulse pulsedd power power supply supply proposed proposed in inthis this work work.. Figure 11. PCD reactor supplied by the high-voltage pulsed power supply proposed in this work. FigureFigure 12 12 shows shows the the typical typical voltagevoltage andand current waveforms waveforms of of a aPCD PCD reactor reactor discharge discharge and and calculated power and energy applied to the load. The corona discharges take place at the maximum calculatedFigure power 12 shows and energy the typical applied voltage to the and load. current The corona waveforms discharges of a takePCD place reactor at the discharge maximum and valuecalculated of the power voltage and (peak energy of ~20 applied kV); therefore,to the load the. The current corona waveform discharges has take two place components. at the maximum First, Appl. Sci. 2019, 9 FOR PEER REVIEW 9 Appl.Appl. Sci. Sci.2019 2019,,9 9, 1451FOR PEER REVIEW 99 of 12 valuevalue of of thethe voltagevoltage (peak ofof ~~2020 kVkV);) ;therefore therefore, the, the current current waveform waveform has has two two components components. First. F, irst, coming from the capacity of the reactor and capacitance of C3 (shaped by resonance of the output comingcoming fromfrom thethe capacitycapacity of the the reactor reactor and and capacitance capacitance of of C3 C (shaped3 (shaped by byresonance resonance of the of theoutput output circuit). Second, as the current peak connected with electrical discharges in the reactor [18]. circuit).circuit).Second, Second, asas the current peak peak connected connected with with electrical electrical discharges discharges in inthe the reactor reactor [18] [.18 ].

FigureFigure 12. 12. The measured output voltage and current waveforms for PCD reactor discharge and Figure 12. TheThe m measuredeasured output voltage and current waveforms for PCD reactor discharge and calculatedcalculated dischargedischarge power and and energy energy.. calculated discharge power and energy. AsAs aa secondsecond type type of of load load,, a DBD a DBD-type-type reactor reactor was used was as used show asnshown in Figure in 13 Figure. Cylindrical, 13. Cylindrical, metal As a second type of load, a DBD-type reactor was used as shown in Figure 13. Cylindrical, metal electrodes of the reactor are separated with dielectric tubes made of Al2O3 (aluminum oxide). The metal electrodes of the reactor are separated with dielectric tubes made of Al2O3 (aluminum oxide). electrodes of the reactor are separated with dielectric tubes made of Al2O3 (aluminum oxide). The Thedischarge discharge matrix matrix consists consists of 36 of identical 36 identical discharge discharge tubes. tubes. The total The capacitance total capacitance of the ofreactor the reactor was 65 was discharge matrix consists of 36 identical discharge tubes. The total capacitance of the reactor was 65 65pF. pF. pF.

Figure 13. The dielectric barrier discharge (DBD) reactor supplied by the proposed power supply prototype. FigureFigure 13. 13. TheThe dielectric dielectric barrier barrier discharge discharge (DBD (DBD)) reactor reactor supplied supplied by the proposed by the proposed power supply power supplyThe m prototype.easured waveforms for a DBD-type prototype.load are presented in Figure 14. The reactor current— plotted in red, such as in the case of PCD—consists of two components, namely, the reactor and the C3 capacitThe m measuredanceeasured current waveform waveforms. For a DBDs for for, thea DBDa dis DBD-typec-typeharge load current load are ispresented are typically presented inmade Figure inof many Figure14. The current 14rea.ctor Thepeaks current reactor of — current—plottedplottedshort timein red (nanosecond, such in red,as in such the range case as) in[19] of the .PCD The case— m ofconsistuch PCD—consists smallers of two distance components of two between components,, namely, electrodes namely,the reactor causes the and reactorthe the discharge to take place at much lower voltages (~8 kV) and its duration covers most of the time of the andC3 capacit the C3ancecapacitance current. F current.or a DBD For, the a DBD, disch thearge discharge current is current typically is typicallymade of many made ofcurrent many peaks current of peaksshortoutput oftime voltage short (nanosecond time pulse. (nanosecond range) range)[19]. The [19]. m Theuch much smaller smaller distance distance between between elec electrodestrodes causes causes the discharge to take place at much lower voltages (~8(~8 kV) and its duration covers most of the time of the output voltage pulse. Appl. Sci. 2019, 9, 1451 10 of 12 Appl. Sci. 2019, 9 FOR PEER REVIEW 10

FigureFigure 14. 14. TheThe measuredmeasured output output voltage voltage and and current current waveforms waveforms for for PCD PCD reactor reactor discharge discharge and and calculatedcalculatedpower power andand energyenergy applied to to the the load. load.

4.4. Discussion Discussion TheThe proposed proposed high-voltagehigh-voltage power power supply supply topology topology wa wass designed designed forfor capacitive capacitive types types of load. of load. ThisThis featurefeature makes makes it itsuitable suitable to supply to supply non-thermal non-thermal plasma plasma reactors reactors of various of construction various constructions.s. While Whilemost mosthigh- high-voltagevoltage power power suppl suppliesies are designed are designed to supply to supply only onlyone type one typeof load of load with with a specified a specified capacitycapacity [ 8[8,9],9],, thethe proposedproposed topology is is appropriate appropriate for for both both DBD DBD and and PCD PCD plasma plasma reactors reactors,, and and for for aa wide wide range range ofof reactorreactor capacities.capacities. The The c constructiononstruction is is fully fully resonant resonant,, with with all allthe the used used semiconductor semiconductor switchesswitches beingbeing commutatedcommutated in in ZCS ZCS ( (zerozero c currenturrent switching) switching) mode mode in inorder order to limit to limit switching switching loss. loss. SiCSiC technology technology was used used to to increase increase the the efficiency efficiency of consecutive of consecutive converters. converters. Usually Usually,, in thein case the of case high-voltage pulsed converters, control of semiconductor switches demands high precision and of high-voltage pulsed converters, control of semiconductor switches demands high precision and additional feedback (e.g., voltage or current sensing circuits) is often used [20]. An important feature additional feedback (e.g., voltage or current sensing circuits) is often used [20]. An important feature of the proposed converter is loss minimization due to soft switching of power switches regardless of of the proposed converter is loss minimization due to soft switching of power switches regardless of changes in the load, so there is no need to use additional feedbacks and/or additional snubbers or changes in the load, so there is no need to use additional feedbacks and/or additional snubbers or other switch protection circuits. The mentioned feature also significantly simplifies the power supply othercontrol switch algorithms. protection circuits. The mentioned feature also significantly simplifies the power supply controlThe algorithms. resonant capacitor charger unit allows for recharging of the energy storage capacitor with a highThe efficiency resonant and capacitor providescharger the additional unit allows possibility for recharging to adjust its of final the energyvoltage storagevalue by capacitor regulation with aof high the efficiencycharging interval and provides number. the The additional efficiency possibility of the charging to adjust process its final was voltage96%. value by regulation of theThe charging presented interval high number.-voltage transformer The efficiency design of the is scalable charging to processvarious wastransformer 96%. winding ratios andThe magnetic presented coupling high-voltage ratios through transformer proper designselection is of scalable the number to various and arrangement transformer of winding magnetic ratios andcores magnetic used in couplingthe design ratios. The p throughroposed propertransformer selection constr ofuction the number allows andfor obtain arrangementing low magnetic of magnetic corescoupling used between in the design. transformer The proposed windings transformer along with constructiona small external allows magnetic for obtaining field. The low e magneticnergy couplingconversion between efficiency transformer of the constructed windings transformer along with prototype a small wa externals similar magnetic or higher field. compared The energy to conversionsimilar constructions efficiency of[14 the,21] constructed, but the EMC transformer (electro prototypemagnetic c wasompatibility) similar or wa highers significantly compared to similarincreased constructions. [14,21], but the EMC (electromagnetic compatibility) was significantly increased. TheThe disadvantage disadvantage ofof thethe proposedproposed topologytopology is the high current amplitude amplitude o onn the the primary primary side side of theof transformer,the transformer however,, however this, this a typical a typical feature feature of of this this group group of of converters converters [14[14].]. TheThe output output pulsepulse waswas shaped by by resonance resonance of of the the high high-voltage-voltage transformer transformer leakage leakage inductances inductances and the capacitance of the load. In the case of small capacitance reactors (e.g., a PCD reactor with and the capacitance of the load. In the case of small capacitance reactors (e.g., a PCD reactor with wide discharge gaps) additional capacitance, C3, is connected in parallel to the load, and helps to wide discharge gaps) additional capacitance, C , is connected in parallel to the load, and helps to stabilize the resonance conditions. In the case of reactors3 exhibiting big capacitance (e.g., large area stabilize the resonance conditions. In the case of reactors exhibiting big capacitance (e.g., large area DBD reactors) capacitance C3 can be omitted, and only the capacity of the reactor will affect the DBD reactors) capacitance C can be omitted, and only the capacity of the reactor will affect the resonance frequency. To maxim3 ize the efficiency, the reactor capacitance should be taken into resonanceaccount in frequency. converter design. To maximize The power the efficiency, delivered theto the reactor load can capacitance be regulated should by both be taken the regulation into account inof converter output pulse design. amplitude The powers and delivered their repetiti to theon frequency. load can be The regulated maximum by bothfrequency the regulation of output ofpulse output pulsegeneration amplitudes of the andbuilt their prototype repetition was 2 frequency. kHz, and the The maximum maximum voltage frequency amplitude of output was pulse20 kV generationfor 400 of the built prototype was 2 kHz, and the maximum voltage amplitude was 20 kV for 400 V of input Appl. Sci. 2019, 9, 1451 11 of 12

DC voltage bus. The efficiency of the output pulse generation process for the pulse shaping unit was 71%, which is a good result considering that the applied high-voltage transformer had a large number of secondary windings and several magnetic cores. The output pulse shaping unit utilizes the principles of a Tesla coil transformer and the natural resonance of the circuit created by the transformer’s parasitic inductances and a capacitive load. The shape of the output voltage pulse was asymmetric, and its average value was equal to zero, which is beneficial for DBD reactors. After pulse formation, the energy unused for discharge generation backflows to the converter and results in low residual oscillations and low residual energy stored in the reactor dielectric barrier capacitance, which usually causes the rise of reactor temperature and decreases plasma generation efficiency. The operation of the power supply was successfully presented for both PCD and DBD, and the most common plasma reactor constructions. Interestingly, although both discharge regimes were different and took place at different voltage levels, the energy per cycle was similar, and slightly below 20 mJ.

5. Patents The power supply construction elements were patented with priority numbers P.427673 and P.429151.

Author Contributions: Conceptualization: J.K. and M.H.; validation: J.K.; resources: M.H. and J.K.; data curation: J.K.; writing—original draft preparation: J.K. and M.H. Funding: This research was partly funded by the European Union Interreg South Baltic Programme “Sludge Technological Ecological Progress—increasing the quality and reuse of sewage sludge” number STHB.02.02.00-32-0110/17; project acronym: STEP. Acknowledgments: The authors would like to thank Stanislaw Kalisiak for his inspiration and fruitful discussions as well as Ludwik Bojańskifor technical assistance during measurements and prototype preparation. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Ohta, T. Plasma in agriculture. In Cold Plasma in Food and Agriculture; Misra, N.N., Schluter, O., Cullen, P.J., Eds.; Academic Press: Waltham, MA, USA, 2016; Chapter 8; pp. 205–221. ISBN 9780128013656. 2. Metelmann, H.-R.; von Woedtke, T.; Weltmann, K.-D. Cold physical plasma for medical application. In Comprehensive Clinical Plasma Medicine; Springer International Publishing: New York, NY, USA, 2018. 3. Inhwan, H.; Jongku, J.; Taesuk, Y.; Jinmu, J. Water electrode plasma discharge to enhance the bacterial inactivation in water. Biotechnol. Biotechnol. Equip. 2018, 32, 530–534. [CrossRef] 4. Weltmann, K.-D.; Kolb, J.F.; Holub, M.; Uhrlandt, D.; Simek, M.; Ostrikov, K.K.; Hamaguchi, S.; Cvelbar, U.; Cernak, M.; Locke, B.; et al. The future for plasma science and technology. Plasma Process Polym. 2018, 16, e1800118. [CrossRef] 5. Schmidt, M.; Holub, M.; Jogi, I.; Sikk, M. Treatment of industrial exhaust gases by a dielectric barrier discharge. Eur. Phys. J. Appl. Phys. 2016, 75, 24708. [CrossRef] 6. Brandenburg, R.; Bogaerts, A.; Bongers, W.; Fridman, A.; Fridman, G.; Locke, R.B.; Miller, V.; Reuter, S.; Schiorlin, M.; Verreycken, T.; et al. White paper on the future of plasma science in environment, for gas conversion and agriculture. Plasma Process Polym. 2018, 16, e1700238. [CrossRef] 7. Wala, A.S.; Aymen, A.A.; Monia, G.; Bouzaza, A.; Aboussaound, W.; Soutrel, I.; Ouederni, A.; Wolbert, D.; Rtmini, S. Abatement of ammonia and butyraldehyde under non-thermal plasma and photocatalysis: Oxidation processes for the removal of mixture pollutants at pilot scale. Chem. Eng. J. 2018, 344, 165–172. [CrossRef] 8. Wang, D.; Namihira, T.; Akiyama, H. Pulsed discharge plasma for pollution control. In Air Pollution; InTech: London, UK, 2010. Appl. Sci. 2019, 9, 1451 12 of 12

9. Kolek, J.; Jakubowski, T.; Balcerak, M. Impact of voltage shape on efﬁciency of ozone generation. In Proceedings of the 19th European Conference on Power Electronics and Applications (EPE’17 ECCE Europe), Wasaw, Poland, 11–14 September 2017. 10. Brandenburg, R. Dielectric barrier discharges: Progress on plasma sources and on the understanding of regimes and single ﬁlaments. Plasma Sources Sci. Technol. 2017, 26, 053001. [CrossRef] 11. Kolek, J.; Kalisiak, S.; Holub, M. Resonant, high voltage power supply for non-thermal plasma reactors. WZEE 2018, in press. 12. Denicolai, M. Optimal performance for Tesla transformers. Rev. Sci. Inst. 2002, 73, 1–5. [CrossRef] 13. Liu, Y.; Lee, L.; Bing, Y.; Ge, Y.; Hu, W.; Lin, F. Resonant charging performance of spiral tesla transformer applied in compact high-voltage repetitive nanosecond pulse generator. IEEE Trans. Plasma Sci. 2013, 41, 3651–3658. [CrossRef] 14. Balcerak, M.; Jakubowski, T.; Kalisiak, S. Compact pulsed power supply based on Tesla transformer for plasma reactors. In Proceedings of the 19th European Conference on Power Electronics and Applications, Warsaw, Poland, 11–14 September 2017; pp. 1–9. 15. You, B.G.; Kim, J.S.; Lee, B.L.; Choi, G.B.; Yoo, D.W. Optimization of powder core inductors of buck-boost converters for hybrid electric vehicles. J. Electron. Eng. Technol. 2011, 6, 527–534. [CrossRef] 16. Kralj, L.; Miljavec, D. Stray losses in power transformer tank walls and construction parts. In Proceedings of the The XIX International Conference on Electrical Machines—ICEM, Rome, Italy, 6–8 September 2010; pp. 1–4. 17. Williams, D.M. Self-capacitance of high-voltage transformers. IEEE Trans. Power Electron. 2017, 22, 22–26. 18. Kim, Y.; Cha, M.S.; Shin, W.; Song, Y. Characteristics of dielectric barrier glow discharges with a low-frequency generator in nitrogen. J. Korean Phys. Soc. 2003, 43, 732–737. [CrossRef] 19. France, P.I.V.; Abb, E. Dielectric-barrier discharges principle and applications. Plasma Processes Polym. 1997, 7, C447–C466. 20. Pokryvailo, A.; Carp, C.; Scapellati, C. A high-power high_voltage power supply for long-pulse applications. IEEE Trans. Plasma Sci. 2010, 38, 2604–2610. [CrossRef] 21. SooWon, L.; ChuHyun, C.; Hong-Je, R.; Jong-Soo, K.; Geun-Hie, R.; Yun-Sik, J. Fabrication andoperation testing of a dual resonance pulse transformer for PFL pulse Charging. J. Korean Phys. Soc. 2011, 59, 3679–3682.

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).