Anda di halaman 1dari 18

Chapter-4

Characterization of Marx generator


4.1 Introduction

In chapter-3, the characterization of high voltage capacitors including those used in Marx
generators has been discussed. In this chapter characterization of a complete Marx generator
will be presented. The characterization of Marx generator will be in terms of the capacitance,
inductance & ESR of the generator as a whole.

4.2 System overview

BARC has a Marx generator (Marx-1) having 6 stages, each stage having two capacitors in
series, identified previously as Type-4 capacitors (of chapter-3), each rated 0.15µF, 50kV.
The Marx is placed in a closed cylindrical shell of internal diameter 53cm which is used as a
return path. This Marx can deliver approximately 300 kV to a matched load. The output
current (being a very important parameter) has a peak value of about 10 kA (IP), rise time of
200ns and Full Width Half Maximum duration (FWHM) of about 400 ns [28]. The desired
values were 20kA (peak) current or more with a FWHM of 200ns or lower. The rise time
would be correspondingly smaller.

One objective of this chapter is to characterize another Marx (also at BARC) of seven
stages with Type-5 capacitors (of chapter-3), each rated 0.15µF and 50kV, two in series per
stage. The aim is to assess possibility of obtaining 20kA with FWHM of a maximum of
200ns. The arrangement of capacitors in the Marx is shown in fig 4.1

Fig 4.1 Experimental setup of capacitors in the Marx-2


42
The second objective, if achieving the requirements is not possible, is to work out a
suitable configuration to realize the requirements of IP ≥ 20kA & FWHM ≤ 200ns. It may be
noted that except for seven stages, this Marx (Marx-2) is similar to Marx-1 which had six
stages.

Marx-2 experimentally gave an output current of 3.75kA at a charging voltage of 15kV/stage


through a load of 13Ω with rise time & FWHM of 150ns & 400ns respectively. The output
current requirement of 20kA was met but the FWHM of 400ns is twice the required value of
200ns. Thus, the work includes assessment of the lowest FWHM possible, the goal being less
than 200ns, and arrive at parameters of a Marx to achieve desired characteristics of 20kA
output current with a FWHM of 200ns (or lower).

The period of discharge of the Marx discharging into a short circuit [35] is given
approximately by equation (4.1)
√ ------ (4.1)

Where LT is the total inductance of the Marx circuit and CT is the effective
capacitance of the Marx during discharge. Here, the influence of ESR is neglected as shown
to be insignificant previously. FWHM can be estimated to be about 40% of the period of
discharge [35] given by equation (4.1). Therefore, from equation (4.1), there are two
possibilities to reduce FWHM

i) By selecting a Capacitor of lower capacitance. In the present case, the capacitance was
already fixed, therefore it cannot be brought down;
ii) Therefore, the only option was to study possible reduction in inductance of the Marx.
Also, by reducing the overall inductance of the Marx circuit, the characteristic
impedance, ZO will reduce as shown in equation (4.2) and when the Marx is discharging
into a matched load, the output peak current, given by equation (4.3), increases.
Now, the characteristic impedance of the Marx is

√ --------------------------- (4.2)

Therefore, peak output current into a matched load is


--------------------------- (4.3)

Where VO is the no load voltage of the Marx.

Possible reduction of inductance of Marx circuit is discussed in the next section.

43
4.2.1. Estimation of inductance of the Marx with the metallic cylindrical shell as return
path

As mentioned above, to achieve faster rise times & FWHM, reduction of inductance of the
circuit is crucial. To appreciate the significance of return path on inductance, assume the
forward path (representing the Marx column) to be a conductor of 3mm diameter and the
return path to be the metallic cylindrical shell of internal diameter 53cm.The inductance of
this configuration will be

( ) -------------------------- (4.4)

The basic principle in reducing the inductance is to reduce the volume occupied by the flux
between the forward and return paths. The coaxial cylindrical return conductor used presently
has a diameter of 53cm and results in a large volume (occupied by flux) and hence in larger
inductance.

Now consider a return path comprising a metal sheet at a spacing of 20mm from the centre of
a 3mm dia rod (considered in above example). The inductance would be

( ) -------------------------- (4.5)

This is a reduction of 36% from the previous value of 1.03 which can be very useful
in the limiting conditions that are almost always faced in such applications.

In the light of the above, it was decided to use a plane sheet of conductor as the return path
placed close to the forward path, consistent with the fact that there must be adequate
insulation to withstand the full level output voltage (on matched load). This insulation is
proposed to be in the form of a Teflon (PTFE) sheet of sufficient thickness, 12mm.The
Teflon sheet should have sufficient width to provide adequate surface creepage-distance to
avoid surface flashover. This exercise was conducted on the second Marx generator, Marx 2,
having 7 stages, each stage with two 0.15µF, 50kV Type-5(of chapter-3) capacitors in series.
A schematic showing arrangement of capacitor, Teflon insulating sheet & metallic return
plate is shown in fig.4.2

44
Fig 4.2 Schematic cross-sectional view of capacitor with return conductor

It may be seen that the spacing between the forward path (capacitors + electrodes + sparks
+ load) and return conductor is greatly reduced. For this Teflon sheet of thickness 12mm,
and hemispherical spark electrodes of diameter 10mm, this spacing would be around 17mm.
This thickness of 12mm is expected to comfortably withstand [36] a fast-impulse of 350kV
(output voltage across the matched load). The electric field analysis for the Marx set-up
shown in fig.4.1 is discussed in detail in chapter-5.

Marx 2 was built at BARC based on the above lines and using available capacitors, Type-5
(chapter-3). The capacitor is also of a different make than in Marx 1. A Teflon sheet, 12mm
thick, was placed 100mm above the plane of capacitors. A metal sheet, 350mm wide was
placed on top of the Teflon sheet as the return path. This generator is used to demonstrate the
principle.

4.3. Inductance of a Marx generator

The total inductance LT of a Marx Generator can be expressed as

LT = internal inductance of all the capacitors (LC) with respect to the return path

+ inductance of all the leads (Ll) with respect to the return path

+ inductance of the spark-paths (Ls) with respect to the return path

+ inductance of the load circuit (Llc), if any, with respect to the return path. ---- (4.6)

45
We will consider each of the above.

4.3.1. Inductance of the capacitor


The capacitor as discussed in chapter-3 is the most critical component. A HV capacitor is
built up of a number of capacitor elements [37] to achieve the required rated voltage. Each
element consists of very thin layers of dielectric material (usually Kraft paper or
polypropylene film of thickness 6µm or 12µm) and thin foils of aluminum/copper as
electrodes. The elements are interconnected by tabs, and stacked in series and parallel
combination to get the desired voltage and capacitance. Therefore the geometric design of
capacitor element, length and thickness of the internal lead connections, bushing structure
and insulation design contribute to the internal inductance of the capacitor. Again as
discussed in chapter-3 in addition to the internal inductance of a capacitor, there exists an
equivalent series resistance (ESR) of the capacitor [32]. For the capacitors used in the Marx
2, as shown in chapter-3, the ESR (per capacitor) is about 0.082Ω and nearly constant over
the current range of 250A to 3kA. As the intended output current is very much higher, 20kA,
it is reasonable to use a value of 0.08Ω per capacitor. Thus, for all the 14 capacitors, ESR
comes to 1.12Ω.

The inductance of the capacitor including its external discharge path is 140nH. Details of
estimation are given in appendix 1.Thus, the total internal inductance of the 14 capacitors is

LC = 14 x 140nH = 1.96 µH ------------------------- (4.7)

4.3.2. Inductance of leads with respect to return path


In the present case, the current return path of the Marx was a metal sheet at a height of
112mm above the capacitor, shown in fig.4.3 (a). Width of the sheet was 350mm. The leads
linking the capacitors are copper strips, 25mm wide shown in Fig.4.3(b).

46
Fig 4.3 (a) Side view of capacitor configuration

Fig 4.3 (b) Top view of capacitor configuration

47
Fig 4.3(c) Current flow path of capacitor configuration

The capacitors have rail electrodes of length 100mm. From the configuration of fig. 4.3(c),
it can be visualized that more current will flow through the rail electrodes in the portion BC,
closer to the spark gap, than the portion AB which is farther away (from the spark gap).
Therefore, for calculating the inductance, an effective length le, less than la (actual length of
rail electrodes), has to be used. In the present work, it is assumed that

(4.8)

As each capacitor has two rail electrodes, the effective length for 14 capacitors is 140cm.

The length of the spark gap electrode & its shank is 5.5cm. These electrodes comprising of
a shank of width 25mm & a hemisphere of diameter 25mm, are equivalent to strips of same
width as the rail electrodes. The spark gap electrodes, one for each stage, together have a
length of 38.5cms.

Further, the links between the two capacitors of a stage (one per stage) are also strips of width
25mm and length 3cm. These links, seven in number, have a total length of 21cms.

Thus, the total length of all leads is 140+38.5+21≈ 200cm. Therefore, we need to estimate
the inductance of 200cm long, 25mm wide strip with respect to the ground plane, the
separation being 100mm. Since separation (between the strip & ground plane) is much

48
greater than the width of strip, we cannot apply the usual formula( ). Heuristically,

we represented the strip by a cylindrical conductor, diameter d, such that

in this case.

This gives a value of 19.7mm for diameter. This method of finding the radius is valid only for
strips of width small with respect to spacing.

An alternative and better approach is to treat the conductor as a bundled conductor. The
25mm width can be segmented into five equal parts i.e. each of 5mm width. The geometrical
mean radius (GMR) of this system is given by

It may be seen that the two values are about 16% different from each other. This is not
significant as the dependence of inductance on radius is logarithmic. Using the value of
8.2mm for the equivalent radius, the inductance of the leads (length of 2m) with respect to the
return path is given by

( ( ) ) ------------------------- (4.9)

4.3.3. Inductance of spark path with respect to return path


The Inductance of the spark channel is estimated considering the arc as a conductor of
diameter 0.1mm. Length of each spark channel is taken as 5mm, the spacing between the
spark gap electrodes. For 8 spark gaps, the length is 40mm; the spark (assumed to be between
the centers of the hemispheres) is approximately 110mm from the return path. Thus,
inductance of the arc channels

( ( ) ) ------------------------ (4.10)

4.3.4. Inductance of Load circuit with respect to return path


The load circuit comprises of a tubular copper sulphate resistor (diameter 50mm & length
150mm) connected to the Marx by copper strips of 25mm width, thickness 2mm and length
of 25cm. Considering an effective length of 22.5cm for the strips, the inductance of the two
parallel strips as under 3.2, is given by

49
π
-------------------------- (4.11)
π

Fig 4.4 Schematic of load circuit

Inductance of the copper sulphate resistor (diameter 50mm) is a little tricky. The return path
is at the beginning (LV end) of the Marx. Distance is about 250cm from the copper sulphate
resistor. The return path will be treated as a conductor of same diameter i.e.50mm.
Considering the logarithmic dependence, this assumption is not of significance. Thus, the
inductance will be

π
--------------------- (4.12)
π

4.3.5. Total Inductance of Marx circuit


From sections 3.1, 3.2, 3.3 and 3.4, the total inductance LT of the Marx circuit is obtained
by adding all the inductances mentioned above

Thus, LT = 1.96 + 0.64 + 0.033 + 0.13 + 0.14

= 2.9 µH ------------------------ (4.13)

It may be seen that the internal inductance of the capacitors plays the dominant role while
the inductance of the leads is only next in importance. The latter can be reduced to some
extent by bringing the return path closer to the surface of the Marx, to 17mm, as discussed.
For this reduced spacing, the value reduces to 0.28 µH which reduces the total inductance to
2.54 µH (a reduction of about 12.4%).

50
4.4. Marx discharge into short circuit
The current waveform for discharge of the Marx into a short circuit is shown in Fig. 4.5(a).

Fig 4.5 (a) Experimental current waveform for discharge of the Marx into a short circuit

Fig 4.5 (b) PSPICE current waveform for discharge of the Marx into a short circuit

51
Period & frequency of discharge wave are 1.14µs and 0.87MHz. From the known value of
CT ≃0.011µF, using equation (4.3) the value of LT is 3µH. This value is quite close to the
estimate of 2.9µH, the variation being less than 5%. This validates the method of estimating
the inductance of the Marx.

As discussed earlier in chapter-3, the value of ESR has been estimated for the Marx by
making PSPICE simulation (fig 4.5(b)) & experimental current waveform match closely.

Table-4.1 Experimental negative peak currents and corresponding resistances of Marx-2


generator
L=3µH 1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th 11th 12th 13th 14th 15th
peak peak Peak Peak Peak peak peak peak peak peak peak peak peak peak peak

Experimental 3500 2800 2250 1850 1500 1200 1000 850 700 500 400 250 200 150 100

Simulation

ESR=1.15Ω 3500 2800 2250 1850 1500 1200 1000 850 700 500 400 ----- ----- ----- -----

ESR=1.25Ω --- ----- ----- ----- ----- ----- ----- ----- ---- ----- ----- 250 200 ----- -----

ESR=1.3Ω ---- ---- ---- ----- ----- --- ---- ----- ----- ----- ----- ----- ----- 150 100

From Table 4.1, the value of ESR is 1.15Ω (for the higher currents) and this is for whole
Marx comprising of 14 capacitors. The ESR per individual capacitor comes to 0.082Ω which
is almost exactly the same as the previously obtained value of 0.08Ω of Type-5 capacitor in
chapter-3. It was verified by simulation that this difference in ESR does not significantly
influence the current magnitude or the rise time/FWHM. This is obviously due to the load
resistance of 13Ω being much higher than ESR.

4.5 Marx discharge into a load of 13Ω

Marx-2 was charged and discharged into a load (copper sulphate solution) of resistance of
13Ω. Together with ESR of 1.15Ω (obtained by tallying experimental & simulated
waveform), total value of resistance in the circuit is 14.15Ω.

52
X-axis – 500ns/division
Y-axis -2.5kA/division

Fig 4.6(a) Discharge current waveform of Marx with load

Fig 4.6(b) PSPICE discharge current waveform of Marx with load

53
The experimental discharge current waveform obtained is shown in Fig. 4.6(a) and with
estimated value of inductance (2.9 µH) using PSPICE, the discharge current waveform has
been obtained and given in Fig. 4.6(b). The two waveforms agree extremely well and give
FWHM values of 400ns and 385ns by experiment & simulation respectively. Thus, the
achievable FWHM is 385ns which is much higher than the required value of 200ns.

Experimental peak current magnitude was 3.75 kA at 15kV/stage, while from simulation the
peak current is 3.76 kA. It is desired to obtain 20kA or higher at the full rated voltage. From
the experimental & simulation data i.e. Fig.4.6(a) and Fig. 4.6(b) the estimated current at a
load voltage of 350kV (no-load charging voltage of 700kV) are 25kA and 25.1kA which
meet the requirement. This very close agreement further validates the method of estimating
the total inductance of the Marx which is crucial.

4.6. Approach to obtain desired output current & FWHM

As discussed above, Marx 2 would not be capable of meeting the important requirement of
FWHM less than 200ns. In this section, a configuration will be worked out to meet the
requirement. The desired output current pulse characteristics are a minimum current of 20kA
& a maximum FWHM of 200ns. The generator will have 7 stages, as above, 2 capacitors per
stage, load resistance would be 13Ω. ESR, as per above experiments would be 1.15Ω giving
a total circuit resistance, ZT to be 14.15Ω. Now, considering ZO to be 14.15Ω,

Further, Ω √ ---------------------- (4.14)

From the work of Carey and Mayes [35], FWHM is about 40% of To, the period of discharge
of the Marx into a short circuit. Present experiments (supported by simulation) gives FWHM
to be about 35% of To i.e. FWHM = [ √ ]. This is fairly close to the relation
given by Carey and Mayes [35]

Thus, for the requirement of FWHM to be less than 200ns,

[ √ ] (4.15)

Solving equations (4.14) and (4.15), we get

54
CT = 6.43nF and LT = 1.29µH

Thus, individual capacitor has to be rated 6.43nF * 14= 90nF. If we assume individual
capacitors of rating 90nF, 50kV, . Today capacitors with internal

inductance of short external discharge path of 40nH are available. Considering the effect of
return path (as explained in appendix-1, for a different capacitor) the sum of the true internal
inductance and the contribution of the external discharge path comes to about 48nH, say
50nH. If such capacitors are used, total inductance of capacitors would be 14*50nH =0.7µH.
This means that the inductance due to the leads, arc path and load circuit should be limited to
(1.29µH -0.7µH) = 0.59µH. We will now examine the feasibility of this.

Considering a return metallic plane 20mm above the plane of capacitors, for leads of
25mm width totaling a length of 200cm as before, using equation (4.9), total inductance of
the leads would be:

[ ( ) ]

( ) ---------------------------- (4.16)

It may be noted that a capacitor of 90nF will be physically smaller than the capacitor of
0.15µF (used in Marx-2). Therefore, the length of the Marx and hence that of the return path
will also be lesser giving rise to a corresponding reduction in the inductance.

For spark path of length 40mm & diameter 0.1mm, the spark (assumed to be between the
centers of the hemispheres) is approximately 25mm from the return path and using equation
(4.10), inductance would be

[ ]

( ) ------------------------ (4.17)

If a pressure of 3 atmospheres is used to house the Marx, length of spark gaps would reduce
by a factor of 3 giving Ls =0.01µH

55
If the load circuit is assumed to have the same configuration as before, its inductance
would be 0.27µH. Thus, the total inductance of the Marx would be 1.3µH, which is close to
the required level of 1.29µH.

From PSPICE simulation with CT = 6.4nF, ESR = 1.15Ω & load resistance = 13Ω, with each
capacitor internal inductance(Li) of 50nH and 60nH, the peak current, rise time (10% to 90%)
and FWHM are estimated and shown in table- 4.2.

Table- 4.2 Characteristics of Marx with new assumed parameters

Li =50nH Li =60nH
Total Marx inductance LT, µH 1.3 1.44
Peak current, kA 26.9 26.2
Rise time (10% to 90%) , ns 66.7 71.6
FWHM, ns 198.3 208.3

Further improvements are possible by reducing the inductance of the load resistance
circuit by using 50mm wide strips (instead of 25mm width) of reduced length of 150mm
(instead of 225mm). Also, in principle, it is possible to reduce the inductance of the load
circuit further by reducing the height of the copper sulphate resistor to 100mm.

Considering the leads that connect the load circuit as a 50mm wide strip instead of 25mm and
strip length of 150mm long instead of 225mm. The geometrical mean radius (GMR) of this
system is estimated by segmenting the width into five equal parts i.e. each of 10mm width
and the GMR is given by

√ -------------------- (4.18)

Considering these leads connecting the load circuit as two parallel strips, the inductance of
this parallel strips is given by
π
-------------------------- (4.19)
π

Inductance of copper sulphate solution with tube length of 100mm instead of 150mm is given
by
π
------------------------- (4.20)
π

56
Therefore the total inductance of load circuit comes down from 0.27 to 0.16
(0.07 +0.09 ). From PSPICE simulation with CT = 6.4nF, ESR = 1.15Ω & load
resistance = 14.35Ω, with each capacitor internal inductance(Li) of 50nH and 60nH , the
peak current, rise time (10% to 90%) and FWHM are estimated with the revised load circuit
inductance of 0.16 H and are shown in table-4.3.

Table- 4.3 Characteristics of Marx with revised load inductance values

Li =50nH Li =60nH
Total Marx inductance LT, µH 1.19 1.33
Peak current, kA 27.5 26.7
Rise time (10% to 90%) , ns 62.8 67.7
FWHM, ns 190.5 201.5

Considering the effect of return path with short discharge length on the true internal
inductance of the capacitor, the capacitor inductance as 50nH. The desired goal is met but, as
the inductance of the capacitor varies with respect to the spacing between the capacitor
terminals (forward path) and the return path, the inductance of the capacitor with respect to
return path will be more than the specified value, hence considering Li =60nH, peak current is
met and FWHM value of 201.5ns is very close to the required 200ns. By stringent control it
may be possible to obtain the required values and the simulated waveform is shown in fig 4.7.

Fig 4.7 PSPICE discharge current waveform of Marx with optimized value of inductance

57
As the required value of peak current is only 20kA and we were able to achieve peak current
of 26.7kA and there is a margin of about 30%. If there is compromise in the energy of Marx,
an effort can be made to reduce FWHM further by considering the capacitors of 80nF, 70nF,
60nF and 50nF instead of 90nF. As capacitance reduces there will be an increase in the
characteristic impedance and therefore peak current reduces and FWHM also gets reduced.
Table- 4.4 gives the characteristics of Marx with revised capacitor values and complete Marx
inductance 1.33µH remaining same.

Table- 4.4 Characteristics of Marx with revised capacitor values

C =80nF C =70nF C =60nF C=50nF


Peak current, kA 25.9 25 23.9 22.7
FWHM, ns 189.4 175.5 162.6 148.2

From the above table, based on the energy requirement, the capacitors of different ratings can
be selected to get better reduction in FWHM value.

4.7 Summary and conclusions

In this chapter, characterization of a complete Marx generator has been studied. An attempt
is made to bring down the total inductance of Marx generator to achieve faster rise time,
reduced FWHM of 200ns and peak current greater than 20kA. On the whole, by extremely
stringent control of the inductances, it is possible to realize required characteristics of the
output pulse. In particular, recall that a 3mm rod with reference to the cylindrical shell of
53cm diameter has an inductance of 1.03µH/m giving about 2.1µH for the inductance of 2m
length of leads. Obviously, return path in the form of a cylindrical shell of 53cm (inner
diameter) cannot give desired results.

It has been shown by experiments & simulation that by proper estimation of the
inductances of a Marx circuit, it is possible to get a good estimate of the output current
characteristics: peak magnitude & FWHM. The relative criticality of the various inductance
components – those of the capacitors, leads, load circuit & spark gaps- can be clearly
appreciated by estimation. The estimated value of Marx circuit inductance and inductance
obtained from experimental discharge waveform are 2.9µH and 3µH respectively and the
difference is less than 5%. With this configuration, one of the objectives, obtaining the peak

58
current greater than 20kA is met but the second objective of obtaining FWHM less than
200ns was shown to be not achievable. This is because the total inductance of the Marx was
too high i.e. 2.9µH. The internal inductance of the capacitors contributes a major portion in
the total inductance of Marx and is 68% of the total Marx inductance. Obviously, the internal
inductance of the capacitors plays the major role while that of lead inductances comes only
next. The latter can be reduced by choosing a closely spaced return path in the form of a
metallic sheet.

An approach to obtain the desired peak current and FWHM has been discussed. With the
availability of high voltage capacitors with internal inductance of 40nH and with very tight
control it may be possible to obtain the desired values. Also if there is a compromise in
energy requirement, low rating capacitor values can be selected to get better reduction in
FWHM value.

Clearly, the performance of the electrical insulation between the Marx and the metallic
sheet return path, considered to be 12mm thick Teflon sheet presently, is of vital importance.
This aspect will be discussed in chapter-5.

-------------------

59

Anda mungkin juga menyukai