BARCN,wd,"" Fonnd,,', Day Special '"ne 2002

lI

I

An Ideal Thyratron-Less RepetitiveTE- Laser Pulser

Dhruba Jyoti BiswasLasee and Plasma Technology DivisionBhabha Atomic Reseacch Centre

Abstract

Lasee and Plasma Technology Division has developed a thyratron-less, transversely-excited (TE) laser pulser that has complete latch proof operation O-C resonantcharging. Thyratrons, a crucial component of the LIS programme of the OAE, are not manufacturedIn India and embargo applies to their sale. The heart of this novel TE laser pulser is a rotatingdielectric spark gap, which has been conceived, and operated In our laboratory. Itsunique geometry has been fully exploited to obtain operation of the pulser and also todrive two high repetition rate TE lasers either simultaneously or with a variable delay.

Introduction

The TE laser pulser performs the crucial jobof subjecting the gaseous medium to atran;verse electric discharge leading to theinversion of population. The function of apulser in the operation of a pulsed gas laserbegins with the drawing of energy from thesource and ends with the reallsation of most

of this as the internal energy of the (lasing)gas. Undoubtedly therefore, the overallefficiency of the laser depends quite strongly

on the performance of the pulser. In theoperation of pulsed gas lasers, energy is

initially stored in a condenser, which is thenrnade to discharge rapidly into the laser loadwith the help of a fast high voltage highcurrent switch. The performance of thepulser during the charging process thusdictates the wall plug efficiency of the laser.Lesser the energy expended by the pulserwhile drawing energy from the source, thebetter the efficiency of the laser.

A typical pulse generator for a TE laser is

shown in Fig 1. A DC high voltage supplycharges up a condenser through a chargingelement, normally a resistance or aninductance. The charging bypass providesa path for the charging current. Once the

IIVlIl' l'

v~1 S",leh~ L""d(,I"""",

h}Co;,

Fig.l , TVpleo/pu/>ogen,cato, roc e TE ge> fa,",

condenser is charged to the required voltage,the rapid closure of the high voltage andhigh current switch enables the condenser todeliver its stored energy into the laser loadbefore glow to arc transition can occur. Thecharging bypass must offer an impedancethat is many times more than that of thelaser load lest this should eat up a significantfraction of the energy stored in thecondenser lowering, thereby, the plug inefficiency of the laser. At the same time itsimpedance should be much less than that ofthe charging element so that the currentflowing through the conducting switch fromthe source followinga discharge can be keptlow for a given repetition rate. In the singleshot operation, the condenser is normally

BARC Naw,lett" Found,,', Day Spadal '"ua 2002

charged resistively and a spark gap istraditionally used as a switch. For repetitiveoperation, however, more efficient chargingby means of inductance is employed and athyratron replaces the spark gap. Followinga discharge when the switch is still in a stateof conduction, all the charging currentdelivered by the power supply gets divertedthrough this conducting switch. This currentmust be kept lower than the hold overcurrent of the switch to ensure its recovery.Low charging current limits the maximumachievable repetition rate from the pulser.An ideal pulser would be one that allowsrecovery of the switch however high thecharging current, and in turn repetition rate,is. We have conceived, developed, andoperated in our laboratory a so-called 'idealTE gas laser pulser' the repetitive operationof which is not hindered by the recoveryproblem of the switch. It is imperative thatan account of this work here follows a briefreview of the D-C resonant charging schemeon which a conventional repetitive TE gaslaser pulser is normally based. Interestedreaders are referred to a review article [1]for a deeper insight to the various aspects ofcharging and discharging processes of a TE-laser pulser.

Direct - Current (D-C) ResonantCharging

The main condenser can be charged to therequired voltage, ranging within tens of kVdepending on the type of the TE laser load,normally in two different ways: resistively orresonantly by a DCsource although resonantcharging of a condenser by an AC source isalso not uncommon. Resistive charging notonly suffers from poor charging efficiency butalso offers low charging frequency as therecovery problem of the switch assumesgreater significance here [1]. The chargingof the condenser through an inductance,commonly known as D-C resonant charging,finds wide application in the repetitiveoperation of TE gas lasers because of itsinherent high plug-in efficiency [2J and highrepetition rate capability [3].

Fig. 2 , a) Typical D-C re,onant chacging networkb) The chacging the voltage aceo"

the capacito. (V,), voltage arco" theinductance (VJ " a function of time it).

The Kirchoff's loop equations in a typicai D-Cresonant charging circuit (as shown in Fig2a), considering an ideal case where there isno resistance in the charging loop and a D- Csource (V,) charges the capacitance Cthrough an inductance L, can be written asfollows:

Vs-L(di/dt)-(q/C)=O (1)

Substituting i = (dq/dt), we obtain

(d'q/dt') + q/(LC) - Vs/L = 0 (2)

The solution of the above equation can beshown to be

q = V,C (1- cos(rot» (3)

where ro, the resonant frequency of thecircuit, is given by

ill = l/('J(LC) (4)

BARC NewsleU"

1

Found,,', Day Spedal "me 2002

The expressions for current (i) and voltagesacross the condenser (Ve) and theInductance (V,) can be worked out to be,

I ~ dqjdt ~ V, Cmsin(rot) (5)

Vc~V,(l-cos(rot))

V, ~ Vs cos(wt)

Fig. 3 a) Typical ce>onantly charged TE la,ec pul,ec

b) The charging w..ent and yoltage acro" theconden,ec in the aboye {,guce

The current and the voltages represented bythe above equations are illustrated in Fig 2b.Understandably, the sum of voltages acrossthe inductance and the condenser at anyinstant equals the supply voltage V,. It canbe seen from this figure that during a halfcycle when the charging current is in theforward direction, the condenser acquires avoltage twice that of the supply. During thenext half cycle, the current flows in thereverse direction and the condenser loses allits charge. If, however, a diode wereintroduced in the circuit (Fig 3a) to arrest theflow of reverse current the condenser wouldthen retain its voltage. This figure alsoshows the discharge path of the condenserthat includes the switch and the load. As the

(6)

(7)

switch closes, condenser discharges throughit into the load and the voltage across itdrops to zero. The current flows in theforward direction once again and thecondenser acquires twice the supply voltageand so on (Fig 3b).

,"""

'"-I )yes/] /l /l

"-~-'/R-

"\ YO"""~-hI~ /I "W"".'"""

Fig. 4 , Few charging and discharging wayefocms andthe charging w..ent (I) in a cesonantly chargedpulsec(a) OpecationatlOO%dutycycle(b) Operation at a lower duty cycle, the droop

in the yoltage across the condenser Isappacent here.

The time required for the condenser to getcharged to 2Vs is TI,J(LC),half the period ofoscillation. Thus at 100% duty cycle, I.e.,the condenser discharges as soon as Itsacquired voltage is 2Vs, the repetition rate isdouble the resonant frequency of the D-Cresonant charging network (FigAa). If,however, the network is operated at a lowerduty cycle, I.e., the condenser dischargeslong after acquiring the peak voltage, thevoltage across it would then droop (Fig 4b).This is because current, though small, mayflow for longer duration through the voltagemeasuring circuitry. The voltage may alsofall due to the flow of minority current in thereverse biased diode. This effect wouldassume significance in the operation at avery low duty cycle.

Immediately following a discharge, as theswitch is still in the state of conduction, thepower supply makes a short through it(Fig5a). However, unlike in the case of R-Ccharging, where the peak charging current

/

BARC New,"tt" Found,,', D,y Spedal hme 2002

(i=VsfR) appears immediately after thedischarge, the charging current here buildsup from zero value and can be expressedby the followingequation:

~i(t) = VsxM/L

L D

~ Conducting

I " //'1 1 ,,"

L, < L2Iin

Fig. 5 : aJ The powee ,upply making a sho,t thcough theconducting ,w"ch

bJ Shortciccu"cu"ent"afunctianaftimefo,two diffeeent value> of cha,ging indue"n" L

The build up of the short circuit current(shown in the Fig 5b for two values ofcharging inductance) should be such that itdoes not exceed the hold over current of the

switch until it's complete recovery. Higherthe inductance, slower the rise of the currentand better is the chance of the recovery ofthe switch although at the expense of themaximum achievable repetition rate, fm" .which is given by

fm" = 1/[2n,f(LC)]

As is seen from eqn.5, the same recoverycondition of the switch can also be met forsmaller values of Vs. However, Vs is fixedfrom the consideration of the laser load and

(8)

cannot be utilised to control rise of the short

circuit current. Thus for a particular load anda given switch, the value of L needs to beappropriately fixed to ensure recovery of theswitch.

The efficiency (") of resonant charging is

given by the following expression [3J.

"=1-n/(4Q), (10)

where the Quality factor Q is defined as

Q = 2nfm"L/R (11)

(9)

R being the total ohmic component of thecharging loop impedance and L the charginginductance. If R/(fm"L)«l, the conditiongenerally met while designing a resonantpulser, the charging efficiency approaches100%. Thus almost the entire energy drawnfrom the source is deposited into the laserload contributing to the increased plug inefficiency compared to the case of resistivecharging. As the condenser is charged totwice the supply voltage, the requirement ofvoltage from the source is lowered by afactor of half. The charging current arrivesat a low rate starting from a zero value if theoperatin9 conditions are chosen properly.This helps in the recovery of the switch that,in turn, allows high pulse repetitionfrequency (prf).

This scheme too suffers from few short-

comings. If the pIT is considerably lower thanthe resonant frequency of the network, i.e.,the pulser is operated at a low duty cycle,the discharge voltage may droopconsiderably. Secondly, the value of L cannotbe made arbitrarily small as then the rapidbuild up of the short circuit current wouldprevent the conducting switch to go into theoff state following a discharge. This effect,which limits the maximum achievable

repetition rate, has been the subject ofinvestigation in a number of studies [4,5Jaimed at enhancing the repetition ratecapability of a resonant charging network.The central point of these studies is to isolatethe power supply during a discharge so thatflow of short circuit current through theconducting switch Is prevented. The most

BARC New,lett" Found,,', Day Spedal "m 2002

effective means of controlling the rise ofshort circuit current without jeopardising therepetition rate capability of the pulsernetwork is to introduce a second switch in

the charging loop, [6-8] commonly known ascommand resonant charging

Command Resonant Charging:

Schematic diagram of a pulser based oncommand resonant charging is as shown inFig 6. As is seen, a second switch 5, is nowintroduced in the charging loop. Theoperation of the circuit can be explained asfollows. When a trigger pulse T, arrives in

("IJ S2 I

V I.

jr~

TJ, LolldSI~<T!

Fig. 6 : Typieol comm,nd "son,nt ohocged TE loseepulsee netwo;k

the charging switch 52, it closes and thecondenser C gets charged. As the chargingcurrent becomes zero, this switchautomatically turns off isolating the powersupply from the rest of the circuit. At thispoint, a second trigger T, arrives at thedischarge switch 5" which then closesenabling the condenser to discharge throughit into the load. During and following a

discharge,. the switch 52, which is in the offstate, forbids the fiow of any short circuitcurrent enabling 5, to recover within itsdeionisation time. Thus in this mode of

operation, maximum achievable repetitionrate is not limited by the switch latch upproblem. Simply lowering the value of Leanreduce charge-up time of the condenser.Few charging and discharging waveforms inthis mode of operation are shown in Fig 7.As the condenser is charged on commandhere, this network can be operated at anyduty cycle with almost no voltage droop. Thecondenser can be made to charge just beforeit has to be discharged. This increases thelife of both the discharge switch as well asthe storage capacitor as the hold off voltage

~,

Fig. 7: The oh';ging m;;ent (I) ,nd few ohocging ,nddi"hocging w,vefo;ms in , comm,nd ;oson,ntoh"ged pulsee netwo;k

appears across them briefly. The droop freeoperation of this pulser has been illustratedin Fig 8 for a duty cycle of 10%. Theresonant frequency shown here is 1kHz whilethe operating frequency is 100 Hz. 8 msafter a discharge the trigger arrives in 52allowing the condenser to be fully chargedwithin 1 ms. The condenser holds the chargefor 1 ms when the second trigger arrives in5, causing it to discharge into the load. Thevoltage droop in a similar operation withconventional resonant charging network canbe seen in Fig db.

"'~TI .. .

.

di",h~2'

"u=-,.,,'-'~'h~:iL

Fig. 8:

This method of charging too, however, is notdevoid of disadvantages. Firstly, both theanode and the cathode of the chargingswitch have to be maintained at highvoltages. High voltage isolation transformeris therefore required for the filaments. Muchart is also needed in the design of the gridand bias circuits. False triggering of thecharging switch also cannot be ruled outwhich can be caused by the electromagnetic

BARCN",,',"" Foood,,', Day Sp,da' "m, 2002

noise associated with the high voltage highcurrent discharge pulse initiated by theclosure of the discharge switch. This wouldmean that both the switches are

simultaneously in conduction once againbringing in the switch latch up problem witha much graver consequence as the value ofcharging inductance is kept low to make thecommand resonant charging high repetitionrate compatible.

An Ideal Repetitive TE LaserPulser

We have seen that short circuit proofoperation of a pulser based even on acommand resonant charging network cannotbe guaranteed. This makes the switches,normally two thyratrons, vulnerable todamage. Thyratrons are expensive, havelimited life, and embargo applies to their saiein India. Against this background we haveconceived, designed, developed, andoperated a novel, inexpensive, and simpleswitch that as a driver of a resonantlycharged TE laser pulser offers complete latchproof operation. By exploiting the uniquegeometry of this device we have achieved anumber of advantages that even a commandresonant charged pulser cannot match[9-14]. The heart of this device is a suitablyconfigured circular dielectric plate thatrotates between the electrodes of anordinary spark gap. Such rotationintrinsically isolates the power supply fromthe rest of the circuit during a discharge and,as explained below, is instrumental inmaking aTE-laser pulser driven by thisswitch a near ideal one.

Rotating Dielectric Spark Gap:

The rotating disc, the most important part ofthis switch, is shown in Fig 9. The disc is acircular plate of ~20 cm diameter with evennumber of equidistant holes (~6mmdiameter) drilled along a circle close to itsperiphery. By mounting the disc on a motor,it can be 50 rotated that the holes passsymmetrically between the electrodes of thespark gap. The operation of this switchcan be easily understood by considering a

",ccc"'"

Fig. 9, Sch,matic diawam of a ,otating dielect,k disc

+v,oc R

~I'-,

~- <9

.;t'

.

ELECTR"

,;,. . . p~~:~so' - ', ",'~ /

+-~ LOAU

Fig. 10, by a

resistively charged pulser as shown in Fig10. Within the travel time between two

adjacent holes the condenser C gets chargedclose to the supply voltage V" which is morethan the air breakdown voltage of the sparkgap, Every time a hole appears between theelectrodes of the spark gap, it closesallowing the condenser to discharge into theload. Following a discharge, the appearanceof the dielectric between the electrodes

truncates the flow of any charging currentthrough the switch causing thereby its forcedrecovery. The charging current thus canassume very high value reducing thereby thecharging time of the condenser. If therotation speed of the dielectric plate is madecompatible with the charge up time of thecondenser, the repetition rate would then beaccordingly enhanced. In contrast, therepetitive operation capability of

BARC New,Jetter Fouoder', Day SpedaJ )"oe 2002

conventional spark gap when used with thispulser is restricted, as the charging currenthas to be less than the holdover CUrrent ofthe switch. In the operation with rotatingdielectric spark gap as the switch, ~2kW ofpower was dissipated into a dummy load at arepetition rate of 300Hz [9). The fins on thedielectric disc, when rotated, blow air jet intothe spark gap and facilitate, thereby, suchhi9h repetition rate operation of the switch.

rCf--J;~;" ,;;-~'-l.:;''

r

~

t["~~~~j'~':~

""n ,;';n :.- I- -, .

., :"'=f\:.~:~

.LA

"'" 1::: :. .-'-- '~-'-:"-<:.,

Fig. 11 .- S,hematl, dlawam of the mtatlng dlele,t,;,,pa,k gap d,;ven ,..;"Ive/y ,harged pul,..along with opti,al ,en,o, ba,.d ";ggenng,'mu'a-y. 1 Rotating dlelectri, dl>c, 2. F;n"3. Moto', 4. Triggerable'pari<gap, 5. Tngg..,6. Load, 7. Pul,. lran>fom>er, 8. Optoele""'nlc ,.n,o" g. Cuffent amplifier,10.5CR

In the un-triggered mode of operation, thelarge jitter associated with the fluctuation ofthe breakdown voltage marred theperformance of the pulser. The triggering ofthe switch was accomplished by mounting alight emitting diode and a photo detectorface to face on either side of the rotatingdielectric in 5uch a way that whenever a holepassed between them another hole alsopassed between the electrodes of the sparkgap (Fig 11). At every such coincidence, thedetector received light from the emitter andgave out a pulse, which after processing wasused to trigger the spark gap. [n thetriggered mode of operation the jittersignificantlyreducedto ~ 25nsec [10].

As demonstrated in ref 11, a rotatingdielectric spark gap also has latch proofoperation capability when used withresonantly charged pulsers. However, suchan operation is not possible utilising a simplecircuit as of Fig 10 where the chargingresistance is replaced by an inductance. This

is because following a discharge the shortcircuit current (i) can build up during thetravel time of the hole (t) between theelectrodes of the conducting switch. For aTE laser load, this Current and hence theenergy stored in the charging inductor canbe appreciable [11). As the flow of thiscurrent is intercepted by the movingdielectric, the inductor releases the storedenergy causing the dielectric plate to ignite.The energy stored can be reduced byincreasing the value of L (as E = 'h (Li') =(V,'t2)/(2L), from eqn 6). This, however, isnot a practical soiution as it would be at theexpense of the achievable repetition rate.

.~oc

Fig. 12 , Short ,;~uit proof operation of the ,e,onantlych..ged pu/,e, when d,lven by a 'otatingdlelert,l, ,pa,k gap

An elegant solution to this problem is toincorporate a second spark gap in thecharging loop and rotate the same dielectricplate between the electrodes of both thespark gaps in such a way that holes appearin them exactly out of phase (Fig 12). Theoperation of this circuit can be explained inthe following manner. When a hole getsaligned with the charging spark gap SG2, itcloses allowing the condenser to be chargedto 2V, in a time (..J(LC) which is madesmaller than the travel time of the holeinside the gap by proper choice of L Afterthe passage of the hole, the appearance ofthe dielectric in the gap of this switchvirtually cuts off the power supply from therest of the circuit. A hole now appearsbetween the electrodes of the dischargespark gap (SG,) when it closes allowing thecondenser to discharge into the load. During

BARC Newsletter Found,,'s Day Special Issue 2002

and immediately following the discharge,which is normally very short lived for a TElaser load, the open switch SG, prevents anyshort circuit current Irom flowing throughSG.. The discharge switch thus readilyrecovers, as the Inductor now does not storeany energy during its conduction. Thecondenser would get charged once againwhen a second hole gets aligned with SG,and so on. In this mode of operation thepulser delivered -3.2kW of power at arepetition rate of 200 Hz into a dummy loadwhich resembled a typical TEA CO, laser.

l~ ~fl~

,moD

Fig. 13' Few cha'!]lng and dlscha'!]lng waveforms ofthe short drcult proof operation of theresonantly cha'!]ed pulser shown In F;g. 18.The Inset shows the cha'!]lng of the condenserIn an expanded "me scale.

Few charging and discharging waveformsshown in Fig 13 conform to the abovedescription. The Jitter in the operation ofboth SG, and SG, is apparent from thisfigure. When this pulser is used to drive alaser, the fluctuation in the closing ofcharging switch can be ignored, as it doesnot affect the performance of the laser.Therefore triggering of the discharge switchalone suffices which is accomplished usingthe optical sensor based mechanismdescribed in ref 9. Proper positioning of theholes with respect to the spark gaps suchthat the condenser discharges soon afteracquiring the peak voltage ensures droopfree operation with this device.

The rotating dielectric spark gap switch canalso drive simultaneously two high repetitionrate lasers [12]. The two lasers can beoperated synchronously or with a delay thatcan be as large as a millisecond. Schematicdiagram of the circuit is shown in Fig 14. The

tJi:i'. "..

~' '. <..

r'~ ""'--r ..0,",

r "'. ". ""'.. ".,0 <, h,.

i~.: U:r.-

Fig. 14 , Rotating dielectric spark gap as a d,;ver of tworesistively cha'!]ed high repetition ratelase~.Inset shows the positions of the holeswith respect to the d/scha'!]e gap SG, and SG,

same dielectric is rotated between the

electrodes of SG, and SG, in such a way thatholes appear In them simultaneously. SG,.triggered by the optical sensor basedtechnique, causes the condenser C, chargedto the supply voltage to discharge into theload R, while a pulse derived from this firstpulser triggers SG, causing C, to dischargeinto the load R,. A delay up to fewmicroseconds between the two dischargeshas been obtained by varying the value of I.The diodes D, and D, are required to isolatethe two discharge circuits. A larger delay,ranging from severai microseconds to morethan a millisecond, has also been obtainedby positioning SG, and SG, with respect tothe dielectric such that when a hole arrivesin SG" another hole is yet to arrive in SG,.The time interval between the arrivals of thetwo holes between the electrodes of their

respective spark gaps is the delay betweenthe two discharges. Triggering both SG, andSG, has considerably reduced the Jitter insuch operation. The voltage enhanced pulsefrom the optical sensor triggers SG, directlyand SG, after being delayed by a delaygenerator. The performance of this devicehas been tested by switching a total of 2.5kW of power at 200Hz into two identicaldummy loads resembling a typical TE laser interms of resistance. Though resistivecharging has been employed in thisoperation, resonant charging can be used forbetter efficiency. We note here that thistechnique can be, in principle, empioyed forsynchronisation of more than two lasers.

BARC New"e"" Found,,', Day Sp,,;al I"ue 2002

The unique geometry of the rotatingdielectric spark gap also allows diode-lessoperation of a command resonant chargingnetwork [13]. In a conventional switch,thyratron, spark gap or SCR driven resonantcharging pulser, the presence of diode ismandatory to arrest the flow of reversecurrent so as to maintain the voltage on thecondenser (refer to Fig 3a). These diodes,which should be capable of withstanding highvoltages and high currents, when form a partof the pulsers meant for repetitive operationof typical TE lasers are expensive and proneto damage, more so in the event of a shortcircuit. The principle of diode less operationcan be understood by referring to Fig 15.

.V,DC

F,g. 15, Diode-Ie» opeeetion of e ;osonently checgedpulsee netwo,k d,iven by e RD5G

SG, and SG, are so located that when thedielectric rotates, holes appear in themexactly out of phase. As a hole gets alignedwith SG,. it closes allowing the condenser tDget charged through L. If the time ofpassage of the hole between the electrodesDf SG, exactly equals the time taken by thecondenser to get charged fully (=n0(LC)) theappearance of the moving dielectric in thegap thereafter forces the switch to go intothe off state preventing the flow Df anyreverse current thus rendering the usage ofa Wade superfluous. As a second hole getsaligned with SG,. it closes and the condenserdischarges into the load. A pulser has beenoperated in this mode at a repetition rate of600 Hz with a dummy load resembling atypical TE laser. Few charging anddischarging waveforms at this repetition rateare shown in Fig 16. It would be seen that

the voltage Df the condenser has dropped toabout 90% of its initial value at the time of a

discharge. This indicates that some reversecurrent had flown through SG, before themoving dielectric appeared between itselectrodes and blocked it. Such a situation

can be overcome by adjusting the charge uptime of the condenser by making use of avariable choke or alternately by adjusting therotation speed of the motor.

I CHARGE

\DISCHARGE

//

/t

;:

--i'ms I- ,~

Fig. 16, Few che;ging end dische,ging wevefo,ms fo,the diode-less pulse,

The geometry of this switch allows an easyscalability of the maximum achievablerepetition rate. The repetition rate (f) in Hzhere can be written as

f = n x s (12)

where n is the number of holes on the discand s is the number of rotations per second.Increasing n or s or both, therefore, canincrease the repetition rate, however, up to acertain limit. If holes are taD close the

device no longer remains compatible withresonant charging. On the other hand, thespeed of rotation increases at the expense ofthe mechanical stability of the device.Further increase in the repetition rate ispossible by increasing the number ofdischarge gaps [14]. The schematic diagramof the circuit where a repetition rate of 1.2kHz has been achieved with a rotatingdielectric switch utilising two pairs ofdischarge gaps SG, and SG, is shown in Fig17. Reliable short circuit proof Dperation was

BARe New,'ett" 10 Found,,', Day Special I"ne 2002

<V '"'"

en

l

' 1J;""";~,,r, 0 0"

;)0I. IJ;~ 'no 0" 0 "

w, ~ 00 Of 00

Ei=UI T I. 1

-t:I-~~ '"~I",1-

[1J,0~

I~-~Ir,,""Fig. 17, Sd>ema"c diagram of the pulser used to ob~in

KHz repe""on ,,'e wnh ",~~ng dielectricspark gap. The in,el shows the >tagge,.dconfigura~on ofthe holes

achieved with resonant charging bystaggering the holes into an inner and anouter circle. The outer holes were alignedwith the charging gap (SG,) while the innerholes were aligned with the discharge gaps.The condensers e, and e, get chargedwhenever SG, conducts and e, dischargesthrough SG, and e, through SG, with a delaydetermined by the location of SG, and SG,with respect to the passing holes. DiodesD,and D, prevent the condensers fromdischarging through the same gap.

Conclusions

We have shown that the repetition rateoperation capability of an ordinary spark gapcan be greatly enhanced simply by rotating asuitably configured dielectric plate betweenits electrodes. Such a rotating dielectricspark gap has be.en shown to providecomplete latch proof operation of a repetitiveTE laser pulser with D-C resonant charging.This is indeed an achievement becausecomplete short circuit proof operation cannotbe guaranteed even with the popular methodof command resonant charging. Further inthis operation a rotating dielectric spark gapreplaces two (expensive) thyratronsmandatory for command resonant charging.The unique geometry of this switch has beenexploited to use it as a driver of i) a diodeless resonantly charged pulser, Ii) more than

one high repetition rate lasers synchronouslyor with desired delay and iii) of a repetitivepulser in the KHzrange.

Acknowledgement

The author acknowledges the keen interestshown and active support extended by N.Venkatramani, Head, Laser and PlasmaTechnology Division, BARCHe also gratefullyacknowledges the contributions of U.K.Chatterjee, former Head, L&PTDiv and hisco-worker ]. P. Nilaya. The author thanksB. S. Narayan for his coilaboration towardsachieving optical sensor based triggering ofthe rotating dielectric spark gap. He alsothanks his coileagues, at CATand BARC,N.S. Benerjl, A. Kumar, S. K. Sarkar, A. K.Nath, and U. Nundy for many usefuldiscussions and R. A. Nakhwa for excellenttechnical assistance.

References

1. D. ]. Biswas and ]. P. Nilaya, Prog.Quantum Electron 26, pp [-63 (2002)

2. P. K. Bhadani, Rev Sci Instrum 60, 605(1989).

3. ). V. Lebacqz and H. ). White, Pulsegenerators, ed: G. N. Glasoe and ]. V.Lebacqz , Vol 5, p-275 ( McGraw Hill,NewYork, 1948).

4. R. C. Sze and E. Seegmiller, IEEE] QuantElectron, QE-17, 81(1982).

5. S. Black and T. R. Borkes Digest ofTechnical papers, 2"' international pulsedpower conference, Lubbock, Texas, June1979.

6. K. R. Rickwood and ). McInnes, Rev SciInstrum 53,1667 (1982).

7. T. Kan, D. Ball, E. Schmitt, and ]. Hill,Appl Phys Lett 35, 676 (2979).

8. G.]. Scoles and B. P. Newton, Proc of11'" modulator symposium, Sept 1973(technical reprint no 78, English electricvalve co., Chelmsford, U. K).

9. ). P. Nilaya, D. ). Biswas, B. S. Narayan,and U. K. Chatterjee, Rev Sci Instrum,65, 3590 (1994).

BARC New,lett" II Found,,', Day Speeial "me 2002

10.J. P. Nilaya, Ph.D thesis, University ofMumbai (2001). D. J. Biswas, J. P.Nilaya, and U. K. Chatterjee, Rev SelInstrum 66,4813 (1995).

1l.D. J. Biswas, J. P. Nilaya, and U. K.Chatterjee, Opt Eng 36, 588 (1997).

12. D. J. Biswas and J. P. Nilaya, Rev SeiInstrum 72, 2505 (2001).

13. J. P. Nilaya and D. J. Biswas, Proe ofNational laser symsposium, PRL,Ahmedabad, India, p-48 (1998).

Dr Dhruha J. Biswas was conferred the Homi Bhabha Science & TechnologyAwardfor the year 2000 for his outstanding contributions in laser technology andrelated fie/ds.

About the author...

D, Dh,uba J. Biswas ,ecelved hi, ".Sc

e"twhile "DRS, BARC In 1979 afte, goaowoek on optical chao> fetched him the doctocate1986 whe,e he wa> on a two-yeac "bbatical.

technology of mld-infmed ga> la"" Includphy,lcal pmce"e,. He ha, to hi, coedit 64intemational phy,feal joumal,. He i, a

t of the INSA young "ienti"N. S.