DEVELOPMENT OF A HIGH VOLTAGE ARBITRARY WAVEFORM GENERATOR

CAPABLE OF SIMULATING THE NATURAL ELECTRIC FIELDS ARISING FROM

STEPPED DOWNLEADERS

J.R. Gumley, F. D’Alessandro & C.J. Kossmann

ERICO Lightning Technologies, AUSTRALIA.

Abstract: In this paper we describe the mainfeatures of a new prototype, high voltage, arbitrarywaveform generator. The generator is capable ofprecisely simulating the rapidly escalating electricfield due to a lightning downleader, including thewell-known steps or pauses as the leader progressestoward its point of attachment near the ground. Itutilises a computer controlled interface that drives aset of series-stacked flyback transformers. Theresulting output is any desired (montonically)increasing waveform, downloaded from thecomputer.

1. Introduction

Research into direct-strike lightning protectiondesign and, specifically, comparative air terminalperformance, is usually conducted by one of threemeans: (i) “natural” field experiments where airterminals are placed in locations of high lightningincidence [1, 2], (ii) “artificial” field experimentswhere air terminals are exposed to lightningtriggered by rockets [3, 4], and (iii) laboratoryexperiments using high voltage impulse generators[5, 6, 7, 8].

The last of these is an attractive option becausetesting of air terminals on demand can provideresults much more quickly than having to rely on thevagaries of field testing. However, progress in thelaboratory has been limited for a number of reasons.One problem relates to the scaling of results from thelaboratory to the field, although some aspects of thislimitation can be overcome by performing a largenumber of systematic experiments. A more difficultproblem derives from the fact that the Marx-styleimpulse generator [9, 10], which is used in most highvoltage laboratories around the world, produces anRC-type waveshape. Hence, these generators areunable to simulate the temporal electric fieldwaveforms evident in natural lightning phenomena.The natural field at ground level has twocomponents: a “permanent” (or DC) and an“impulse” component. The latter component is a

rapidly escalating waveform for a nearby lightningstrike [11] which cannot be reproduced by thepresent generators. Researchers have attempted tocircumvent this problem by using switching impulsesand a peak voltage for a given terminal-plate gapsuch that the field risetimes are similar to thoseobserved in nature, namely ~ 1 kV/m/µs [6].However, this practice disregards the importance ofthe build-up to breakdown and the role it plays in airterminal performance. This is particularly so in non-conventional air terminals that rely on capacitivecoupling to a downleader in order to gain thenecessary energy for their operation. A comparisonof the basic waveforms is shown in Figure 1.

0

200

400

600

800

1000

0 1000 2000 3000

Natural e-field variation

250/2500µs switching impulse

Time (µs)

Ele

ctric

fiel

d (a

rbitr

ary

units

)

Figure 1: Comparison of waveform obtained fromMarx-style generator with that observed in naturefrom a progressing lightning downleader.

Figure 2 shows an active air terminal configurationdesigned to operate like the “Trigatron” in the firingmechanism of a Marx generator. The main differencein this configuration is that energy is derived fromthe approaching leader and not from a laboratorypower supply. In the pre-stroke period with electric

fields around 10-20 kV/m, the floating sphere willcollect random ions. These are passed to groundthrough the impedance which couples the sphere andearth rod. Under these conditions, and even in theearly stages of leader approach, the sphere remainseffectively grounded and presents a spherical surfaceof low field intensification. This acts to precludecorona and space charge formation.

Figure 2: Basic design of a corona reducing,capacitively coupled air terminal.

However when the field is increasing at a rateapproaching 1 kV/m/µs due to an approachingdownleader, the capacitive reactance due to thecoupling decreases and current attempts to increase.However, the presence of the impedance Z restrictsthe flow of displacement currents. This causes thesphere to rise in potential until a triggering arc iscreated across the gap between the sphere and theearthed rod.

Should triggering occur too early, a failed streamerwill leave space charge above the terminal, and thiswill act to reduce the local field strength. Testingwith a Marx generator can give very misleadingresults because the highest dV/dt occurs at thecommencement of the pulse and there is no triggerwhen peak voltage is being approached since dV/dtapproaches zero.

Figure 3(a) shows how testing with a Marx generatorproduces very early triggers when voltage isimpossibly low to form a streamer. The figure showshow the rate of pulse repetition progressively reduceswith time as the dV/dt reduces. At the current peak,dV/dt is zero and no triggering arc can occur.

Conversely, with the natural lightning waveform,trigger pulses hold off until dV/dt rises to anadequate value. Thereafter, pulse rate actuallyincreases with the increasing field strength, as shownin Figure 3(b). Streamer generation is actuallyretarded until the near field has adequate strength tosupport streamer formation. It is readily seen that theMarx generator waveform is totally unsuited fortesting this type of terminal.

MARX WAVEFORM

CURRENT PULSES

(a)

NATURAL WAVEFORM

CURRENT PULSES

(b)

Figure 3: (a) Early triggering due to a typical Marx-style switching impulse waveform; (b) Delayingtriggering that would actually occur under naturalconditions.

In the remainder of this paper, we describe the mainfeatures of a new prototype, high voltage, arbitrarywaveform generator. The generator is capable ofprecisely simulating the electric field due to alightning downleader, including the well-knownsteps or pauses as the leader progresses toward itspoint of attachment near the ground (Figure 4).

0

500

1000

1500

2000

0 200 400 600

TIme (µs)

Ele

ctric

fiel

d (

kV/m

)

Figure 4: Simulated electric field waveform at theground from a stepped downleader.

2. Generator Design

A block diagram of the mechanical design of our 10-stage, 200 kV prototype system is shown in Figure 5.The key to the whole concept lies in the combinedeffect of series stacking a number of speciallytailored transformers. The generator is capable ofaccurately simulating any monotonically increasingwaveform, such as the electric field due to a steppedlightning downleader.

- 200 kV PEAK

1.8

mA

PP

RO

X.

Figure 5: Mechanical design of the new high voltagegenerator.

A personal computer installed with a high speeddigital I/O card is used to output a 10-bit data word

which is a pulse width modulated (PWM)representation of the rate of rise of a point on thedesired waveform. Typically, the PWM frequency is100 kHz and the data rate is 2 MHz. This enables aPWM resolution of 5 %. Hence, it allows delays to beinserted between each bit in the data stream in 5 %steps in order to create a ripple effect. A simpleexample of this “interleaving” principle is illustratedin Figure 6.

O/P 1

2yy

SLOPE = 0

SLOPE = 0

SLOPE = x

SLOPE = 2x

O/P 2

O/P 3

O/P 4

O/P 5

O/P 6

O/P 7

O/P 8

O/P 9

O/P 10

Figure 6: Simplified example of the interleavingprinciple of the generator. The waveform slope isapproximately proportional to the duty cycle.

Each delayed bit of the output data is passed into oneof ten opto-driven cables as shown in Figure 7. Eachof these fibre optic signals is then passed into anisolated switched-mode power supply (SMPS) withits own floating DC power supply. The fibre opticsignals are converted back to electrical form insidethe SMPS’s.

The output stage of the generator uses transformersconfigured in a flyback topology to eliminate theneed for output inductors. When the transformerprimary winding switches are activated by theamplified SMPS outputs, the primary side of eachtransformer acts as an inductor due to the blockingaction of the output diode. When the switches are

deactivated, the voltage reverses and the inductiveenergy stored in the primary is released through thesecondary winding. The output diode then conductsso that a negative voltage appears on each output.

Risetimes > 1 kV/µs are achievable with this system.The advantage of series stacking the modulescomprising the generator is that each module onlyneeds to be able to output a voltage of Vout/n and,more importantly, output it at a rate of only 1/n of therequired slew rate, where n is the number ofmodules. An additional benefit of increasing thenumber of modules is that the ripple effect from theinterleaving is smoothed even further.

Each power unit in the prototype can produce up to20 kV with a series stack of 10 units reaching anoutput voltage of 200 kV after additional filtering.Also, each unit can maintain a constant voltageoutput, thus allowing the waveform to rise from apredetermined static level. At this stage, no attempthas been made to control the current waveform ofany subsequent air discharge. The prototype has beendesigned to simulate only a millisecond or soimmediately prior to the return stroke of a discharge.

Other advantages of this system include: (i) the testwaveform can be changed from concave, to linear, toconvex in a relatively short period of time (of theorder of minutes) so that empirical corrections forvariations in temperature, pressure and humidity arenot needed; (ii) computer control means that thewaveshapes can be stored and recalled at any time torepeat a test.

3. Preliminary tests

The prototype generator has a design objective ofproducing a variable, programmable wavefront with200 kV peak voltage. Accordingly, the authors haveno expectation of creating an upward leader in thelaboratory. This may come at a later time using a fullsize generator based on our experience with thisprototype. We do, however, expect to generatestreamers and to break down a gap with thesestreamers.

Notwithstanding these limitations, we expect someuseful testing to be performed. We currently proposeto:• compare sharp and blunt rods with and without

prior corona producing fields• test sharp and blunt rods with both concave and

convex waveshapes• test capacitively coupled air terminals under

different wavefronts to optimise their triggering

time with respect to absolute field strength andrate of rise of the electric field

• compare streamer generation from activeterminals and passive sharp/blunt rods.

The authors believe that a true streamer-to-leadertransition will not be observed with a generatorproducing less than 1 MV. Of course, the voltagerequired to observe this transition is not known withany certainty. All past testing with Marx generatorshas a progressively reducing rate of rise of electricfield from t = 0. At the time the generator hasreached the general area of the critical breakdownvoltage, the dV/dt is considerably reduced.

On the other hand, this generator produces thetypical waveform observed in nature and hencecauses a streamer to be launched into a sustainableelectric field strength, at a time when the rate of riseis rapidly escalating. It could quite easily be observedthat no change in the time to breakdown is recorded,but systematic tests are required to prove this.

4. Conclusions

This paper has described a new design of highvoltage generator that is capable of producing awaveform that faithfully replicates the type oflightning electric field waveform observed in nature.The generator tests presently in progress are part of abroad program of lightning protection research alsoinvolving computer modelling and field testing. Theresults will be published as soon as they becomeavailable.

We aim to increase the capacity of the generator to≥ 1 MV. If this is achievable, it will revolutioniselightning downleader simulation experiments ingeneral and, specifically, comparative air terminaltesting in the high voltage laboratory.

References

[1] Gumley, J.R.: “Lightning interceptiontechniques”, 20th International Conference onLightning Protection, Interlaken, Switzerland,paper 2.8, 1990.

[2] Gumley, J.R.: “Lightning interception and theupleader”, 22nd International Conference onLightning Protection, Budapest, Hungary, paperR 2-11, 1994.

[3] Moore, C.B., Rison, W., Mathis, J. & Paterson,L.: “Report on a competition between sharp and

blunt lightning rods”, preprint, 1997.

[4] Uman, M.A., et al: “1995 Triggered LightningExperiment in Florida”, 10th InternationalConference on Atmospheric Electricity, Osaka,Japan, pp. 644-7, 1996.

[5] Gary, C., et al: “Laboratory aspects regardingthe upward positive discharge due to negativelightning”, Rev. Roum. Sci. Techn. -Electrotechn. et Energ., Vol. 34, pp. 363-377,1989.

[6] Berger, G.: “The early streamer emissionlightning rod: laboratory simulation of theconnecting discharge from a lightning rodconductor”, 15th International Aerospace andGround Conference on Lightning and StaticElectricity, Atlantic City, published by US Dept.of Transportation, pp. 38-1 to 38-9, 1992.

[7] Berger, G.: “Formation of the positive leader oflong air sparks for various types of rodconductor”, 22nd International Conference onLightning Protection, Budapest, Hungary, paperR 2-01, 1994.

[8] Berger, G.: “Inception electric field of thelightning upward leader initiated from aFranklin rod in laboratory”, 11th International

Conference on Gas Discharges and theirApplications, Tokyo, Japan, 1995.

[9] Marx, E.: Deutsches Reichspatent no. 455933,1923.

[10] Kuffel, E. & Zaengl, W.S.: “High VoltageEngineering”, Pergamon Press, Oxford, 1984.

[11] Beasley, W.H., et al: “Electric fields precedingcloud-to-ground lightning flashes”, J. Geophys.Res., Vol. 87, pp. 4883-4902, 1982.

Address of Authors

J.R. (Rick) Gumley (Chief Research Engineer)Dr. F. D’Alessandro (Research Physicist)C.J. Kossmann (Electrical Engineer)

ERICO Lightning TechnologiesTechnopark, Dowsings Point, TAS. 7010GPO Box 536, Hobart, Tasmania. 7001AUSTRALIA.Tel: +61 3 62 373 200Fax: +61 3 62 730 399Email: rgumley@erico.com

fdalessandro@erico.comckossmann@erico.com

DELAY

DELAY

HIGH SPEEDDIGITAL

I/O CARD

PC

INTERLEAVEDDELAY

OPTODRIVER

FIBREOPTIC

OUTPUT

FIBREOPTIC

RECEIVER

D.C.SOURCE

SW ITCHDRIVER

SW ITCHDRIVER

-200 kVOUTPUT

MULTIPLEOUTPUTSTAGES

Figure 7: Diagram of the prototype 200 kV high voltage arbitrary waveform generator.