.

I
4
li
Translated from ETZ-A, vol. 92 (1971) Nr. 1, p. 56. . .57
increasing the Trigger Range
of a Marx Generator
by means of Auxiliary Spark Gaps
By Arnold Rodewald
EMILEHAEFELY&COLTDBASEL
Swi tzerl and
Increasing the Trigger Range
of a Marx Generator
by means of Auxiliary Spark Gaps
By Arnold Rodewald
Translated from ETZ-A, vol. 92 (1971) Nr. 1, p, 56 ...57
I t is well known that the multiplying circuit according to
the Marx generator equipped with internal tail resistors R,
(f i g. l), possesses some favorable features as regards the
voltage distribution in the generator and the possibility of
parallel coupling of its stages. I f several of these are con-
nected in parallel, the time to half value remains practi-
cally unchanged as, parallel to the impulse capacitance
value increase, there is a decrease of the resultant ohmic
value of the tail resistor. This circuit is further particularly
adequate for generators having a high total charging vol-
tage, the reason being that voltage drops occurring during
the duration of the impulse voltage are distributed over all
stages, each of them equipped with a tail resistor R, and a
damping resistor Rd.
However, this circuit also presents a serious liability. It has
already been shown [l, 21 that the automatic firing of the
coupling spark gaps and therefore the actual feature of the
multiplying circuit are no longer insured if tail resistors
have an ohmic value of roughly less than 100 Q; this is
actually the case with high energy generators and/ or gene-
rators having a large number of stages with a low stage
charging voltage. Overvoltages which should lead to the
firing of spark gaps are, under such circumstances, too low
to trigger in a reliable manner. I t becomes moreover im-
possible to obtain a controlled triggering of the coupling
spark gaps below their static breakdown voltage, which
means that their firing range is very narrow. This range is
of the order of a few percents only, whereas it would be
desirable to have more than 10 O/ o of the static breakdown
voltage. This firing range at, is given by
at, = Vstat - W / %t
where Usttrt.
is the static breakdown voltage of the spark
gaps when no firing impulse is used, and U,, the voltage at
which the spark gaps can be fired with the aid of a firing
impulse and a certain dispersion utr. For practical purposes
of a high voltage laboratory, it is essential that the value
of at, should be larger than the normal dispersion range of
all coupling spark gaps connected in parallel; indeed, the ob-
ject under test 1 can only be tested with a previsible voltage
amplitude, by means of a controlled firing of the generator
at a predetermined charging voltage. A circuit is described
hereafter which possesses also a firing range of more than
loo/ o of the static breakdown voltage, while nevertheless
having the advantages of the one equipped with tail resistor
R, illustrated in f i g. 1, eventhough the tail resistors have
a very low ohmic value. The following considerations have
led to this new circuit:
1. Papers already mentioned above have shown that the
coupling spark gaps 2 to 4 encounter very poor break-
down conditions (at, = 0 5 O/ o) wi t h t ai l r esi st or s R,
having an ohmic value of the order of some 100 Q, while
these breakdown conditions become quite fair if the
ohmic values of these resistance reach a few thousand
ohms (at, =20 . *. 30 O/o).
2. Paper [l] explains the dependence on the values of the
tail resistors, in so far that each one of the coupling
spark gaps 2 to 4 fires due to an overvoltage which is
practically identical with the voltage drop on the corre-
sponding tail resistor. Voltage drops on high ohmic value
resistors are more important and last longer than those
on low ohmic value resistors. Fig. 2 illustrates, as an
-2
r:i
Rd
CS
4
RI
3
%
4
Re
4
2
Rd
R,
!i 00
I
TC
z-sv~A~~~~M& W/N/N/N/ /h m..,/x, SJ I
F i g. 1. Di agram of a 3-stage Marx mul ti pl yi ng ci rcui t. Resi stors
arranged accordi ng to Edwards and Stoles.
cs
i mpul se capaci tor
Re
tai l resi stor
5
chargi ng resi stor
Rd
dampi ng resi stor
uO
chargi ng vol tage
1 test obj ect 2 ... 4 coupl i ng spark gaps
b
l---c
F i g. 2. Vol tages US on the second coupl i ng spark gap 3 of a ZO-stage
generator as per fi g. 1. Spark gap 2 breaks down at tl i nstant; spark
gap 3 does not br eak down; Utr = dmrging vol tage at the i nstant of
triggering.
I with Re = 66 61 2 wi th Re = 8 kP
example, an overvoltage due to the firing of the second
coupling spark gap in a 20-stage, 4 MV, 400 kWs gene-
rator equipped either with R, = 66 B or 8 kQ tai l re-
sistors.
3. The probl em woul d be sol ved i f tai l resrstors coul d be
available, whose resistance value of some kiloohms would
remain constant as long as the coupling spark gaps have
not fired, and would then drop to such a value as it
should be necessary to form the tail of the impulse
voltage.
4. The questi on ari ses as to how each tai l resi stor can
real i ze whether i t di d or not accompl i sh i ts task of
firing its respective spark gap. The answer is given by
the vol tage drop al ong the resi stor. A s l ong as the
respecti ve coupl i ng spark gap has not f i red, onl y a
transient voltage appears on the tail resistor, this vol-
tage acting as an overvoltage to fire this particular spark
gap. Experience shows that this overvoltage may reach
about 40. . . GO O/ o of t he char gi ng vol t age U,, be-
fore the coupling spark gap fires. After firing has been
achieved, the tail resistor is connected directly in parallel
to the charging capacitor C,, i. e. the voltage on resistor
P i g. 3. Mars mul ti pl yi ng ci rcui t equi pped wi th auxi l i ary spark gaps
R(,,, l ow ohmi c tai l resi stor
Reh hi gh ohmi c tai l resi stor
5 ... ? auxi l i ary spark gaps
other i ndi cati ons - see f i g. 1.
largely on whether the distance between the coupling spark
gaps must be, or not, adjusted with high accuracy. Practical
experience gathered on a 20-stage 4 MV, 400 kWs generator
has shown that this new circuit is not at all influenced by
an inaccurate adjustment of the coupling spark gap distan-
ces. Both coupling and auxiliary spark gaps were coupled
in such a manner that distances between auxiliary spark
gaps were roughly one half of those of the coupling spark
gaps. Oscillograms in fig. 4 give an idea of the trigger
range of the new circuit. The coupling spark gaps were
adjusted, for the recording of a series of oscillograms, for
a static breakdown voltage of U,,:,, = 152 kV. The generator
was fired, at charging voltages between 150 kV and 105 kV,
10 times at each voltage value by means of an auxiliary
spark on the first stage coupling spark gap; high ohmic
tail resistors Rrl, had 8 kP each, and those of low
ohmic value GG IZ each. The ten oscillograms taken for each
value of the charging voltage were recorded one upon an-
other, without film transport, so as to show a time jitter.
I t can be seen that the trigger range lies between 150 kV
and 110 kV, i. e. it has a value of at,. = 27 O/ o if a time dis-
persion of utr < 50 ns is allowed. The improvement obtained
thanks to the new circuit is particularly evident with re-
ference to the convential circuit (fig. 1) with R, = 66 52
tail resistors, Fig. 5 shows oscillograms similar to those of
fig. 4. With a time dispersion of +, 2 120 ns, the trigger
range reaches only a value of ah. = 7 O/ o. On the whole, the
I: i g, -1. TII~~<~c~LIU!~ of a 4 hfV gcnelator wi th auxi l i ar y spar k gaps at di f f er ent bring vol tage val ues tit,. 10
osci l l ograms were recorded one
upon <lnrlth~r ior edcll Lit, val ue [ti me marks di stance 0.5 IIS).
u
atxt
= 152 KV, R = 66 12, RI,,, = 8 kQ.
4 Utr
= 150 kV. c) U,* = 140 kV,
4 Utr
=120 kV,
,,I,
e) LJtr = 110 kV.
generator fires much slower, this fact is brought in evidence
by a flat rise of the impulse wave curve.
Nowadays, numerous generators are not only used to pro-
duce 1.2,50 lightning impulses, but more and more frequently
a!so f or swi tchi ng surges of the order of 250;2500. I t i s
therefore logical to use in this new circuit a low ohmic tail
resistor R,,, for the forming of a 50 1s tail and, on the other
F i g, 5. Tri ggeri ng of a 4 MW generator wi thout auxi l i ary spark gaps
at di fferent fi ri ng vol tage val ues.
hand, the high ohmic tail resistor R,,, for the 2500 1~s tail.
u
s1:it
=152 kV Rr =66 Q =140 kV
To switch over from 50 Irs to 2500 1~s tai l , i t wi l l onl y be
other i ndi cati on - sek f i g. 4.
a) Utr = 150 kV b) Uitl
necessary just to increase the distance between the auxiliarv
spark gaps, so as to be certain that they will not fire
R,, (fig. 1) rises as rapidly as can be allowed by the
stray capacitances and self-inductances of the circuit;
this voltage has an amplitude of U. or possi bl y even a
higher value. With other words, the appearance of a
voltage > U, on the tail resistor can be taken as a proof
that the corresponding coupling spark gap has fired.
5. Ii. is therefore logical to start by creating, by means of
high ohmic value tail resistors R,, conditions favourable
for the firing of coupling spark gaps: subsequently, the
important voltage drop which appears on R,.,, immediately
after the spark gaps have fired, is used to fire an auxili-
ary gap (f i g. 3). Thi s auxi l i ary spark gap connects a
low ohmic value resistor R,,,, in parallel with R,,, and
thus reduces the active tail resistor to the required
ohmic value.
The distance between each auxiliary spark gap 5 to 7 must
be chosen so, that the overvoltage which is to fire the
corresponding coupling spark gap 2 to 4 does not lead to
the firing of the auxiliary spark gaps 5 to 3 breakdown of
the corresponding coupling spark gaps. I f the distance
bet ween t he coupl i ng spar k gaps i s modi f i ed, t he
distance must be naturally modified also between the auxili-
ary spark gaps. Practical use of this new device depends
Literature
111 Rodcwald, A . : Ausgleichsvorginge i n d e r Marxschen Vervielfachungs-
scholtung nach der Zi i ndung der ersten Schal tf unkenstr ecke. Bul l .
Schweiz. Elektrotechnischer Verein (Bul l eti n of the Swi ss El ectrotechn.
Assoc.), vol . 60 (1969) pp. 3?-44.
[7] Rodewnld, A . : D i e Ziindwahrscheinlichkieit d e r Schaltfunkenstredcen
i n der Marxschen Vervi el fachungsschal tung. Bul l . Schwei z. Elektro-
techn. Ver. (Bul l eti n of the Swi ss El ectrotechn. ASSOC.), vol . 60 (1969)
pp. 657-863.
Authors address:
Dr.-Ing. A. Rodewald, Emil Haefely & Cie AC,
CH - 4000 Base1 28, Postfach
571 1000