Progress in the reduction of inductance in the standard 100 kV energy storage capacitor

Abstract

The basic metal case low profile bushing energy storage capacitor design has changed little from the 1.85 μF, 60 kV capacitor developed for the LANL SCYLLAC program in the late 1960's. Their enduring use testifies to a robust design. Today energy storage capacitors having a lower equivalent series inductance (ESL) will contribute to increasing the power capability of new or revised pulsed power machines. Lower ESL coupled with an improved terminal configuration for better integration with the system design, will produce faster discharge times and lower driver impedance, making higher power systems more sensible and energy efficient. A lower ESL capacitor that is compatible with existing proven hardware will also make upgrading more cost effective. This paper discusses the establishment of standardized test methods for determining the inductance of different SCYLLAC style energy storage capacitors; for example Maxwell Type C capacitors, which are now manufactured by General Atomics Energy Products, a part of Sorrento Electronics, and units from Aerovox. The effects leading to imprecision in inductance measurements will be noted. The inductance of existing designs will be compared with new hardware compatible prototype configurations with a goal of reducing the inductance up to 50%.

Full text

Page 1

GENERAL ATOMICS ENERGY PRODUCTS
Engineering Bulletin



PROGRESS IN THE REDUCTION
OF INDUCTANCE IN THE
STANDARD 100kV ENERGY
STORAGE CAPACITOR

R.A. Cooper, J.B. Ennis, F.W. MacDougall, and J.F. Bates

Presented at:
IEEE Pulsed Power Conference
June 2001



© 2001 IEEE


www.ga-esi.com

Page 2

PROGRESS IN THE REDUCTION OF INDUCTANCE IN THE
STANDARD 100kV ENERGY STORAGE CAPACITOR

R.A. Cooper, J.B. Ennis, F.W. MacDougall, and J.F. Bates.
General Atomics Energy Products, Sorrento Electronics.
4949 Greencraig Lane, San Diego, Ca. 92123

A.R. Miller
Titan Pulse Science Division
4855 Ruffner St., San Diego, Ca 92111


Abstract

The basic metal case low profile bushing energy storage
capacitor design has changed little from the 1.85uF 60kV
capacitor developed for the LANL SCYLLAC program in
the late 1960’s. Their enduring use testifies to a robust
design. Today energy storage capacitors having a lower
equivalent series inductance (ESL) will contribute to
increasing the power capability of new or revised pulsed
power machines. Lower ESL coupled with an improved
terminal configuration for better integration with the
system design, will produce faster discharge times and
lower driver impedance, making higher power systems
more sensible and energy efficient. A lower ESL
capacitor that is compatible with existing proven
hardware will also make upgrading more cost effective.
This paper discusses the establishment of standardized
test methods for determining the inductance of different
SCYLLAC style energy storage capacitors; for example
Maxwell Type C capacitors, which are now manufactured
by General Atomics Energy Products, a part of Sorrento
Electronics, and units from Aerovox. The effects leading
to imprecision in inductance measurements will be noted.
The inductance of existing designs will be compared with
new hardware compatible prototype configurations with a
goal of reducing the inductance up to 50%.


I. SCYLLAC STYLE CAPACITOR

The basic high voltage, low inductance energy storage
capacitor was developed for the Los Alamos National
Laboratories’ SCYLLAC program, in the late 1960’s.
This 1.85uF 60kV, 3.3kJ design set the standard for metal
case capacitors with a case size of 11 x 14 x 25 inches and
a dish shaped bushing with coaxial terminal. See Figure 1.
As technology advanced to meet demand for higher
energy densities, the Scyllac design evolved through the
2.8uF 60kV(mid 1970’s), the 1.3uF 100kV(1980) and the
3uF 100kV design in the mid- 1980’s. An important
Sandia National Laboratories life test program [1] to
validate capacitors for new, larger, pulsed power systems
helped drive the 3uF design.
Today the operating parameters, required life and
reliability drive the capacitance and energy density.
During this time the basic capacitor has changed little and
numerous Marx generator configurations were developed
using the same Scyllac style capacitor, making it a
standard component in pulse power systems.















Figure 1. GAEP’s Scyllac Style Capacitor


II. LOWER CAPACITOR INDUCTANCE

Today energy storage capacitors having a lower
equivalent series inductance will contribute to increasing
the power capability of new or revised pulsed power
machines. A lower capacitor ESL with improved terminal
configuration for better integration with the system design
will produce benefits. These benefits are reduced system
inductance, lower driver impedance, faster discharge
times and higher peak currents, making higher power or
more reliable systems more sensible and energy efficient.
While not based on type C capacitors the recent LANL
Atlas program is an example of the performance
obtainable from a low inductance cap and integrated
switch design. 96 parallel 240kV Marx generators having
two railgap switched dual capacitor stages each, deliver
28 MA in < 5 u sec.
We asked ourselves three questions: What is the
inductance of the current Scyllac units? If lower ESL is
important, can it reduced? What is best way to measure

Page 3

the ESL? Many existing systems already use this style
capacitor; therefore, a lower ESL version would be more
cost effective than a complete new design.

A. Capacitor Inductance
Intrinsic capacitor inductance is due to windings,
internal connections, case, insulation design and output
terminal; fundamentally the magnetic flux produced in the
capacitor by the discharge current. Inductance
measurements therefore require a complete circuit for
current, including external connections and a switch
which add inductance. Capacitor inductance is defined as
that part of the circuit within the capacitor envelope or
with the closest fitting short circuit across the terminals.
Elements external to the capacitor usually dominate the
circuit inductance making accurate measurements of the
small internal components difficult. Most of the capacitor
inductance is associated with the terminal (or header) and
the connections to it. Various approaches to inductance
reduction (many proprietary, most empirical) are
constrained by the standard terminal design. Inductance
reduction generally increases electrical stress and can
decrease reliability. Control of inductance by design
requires knowledge of current density, its distribution and
analytic models for calculation. Current distribution
(electronic and displacement) can be difficult to determine
making inductance predictions uncertain. Our approach to
inductance reduction is to reduce the magnetic flux inside
the capacitor. This reduction was measured by the
methods described later.

B. Benefits of Lower Inductance
In Marx generator driver water transfer capacitor based
pulse power systems, Marx capacitor inductance
reduction makes sense as part of an overall effort to
reduce system inductance. By itself, this reduction can
also have a non-negligible affect. In a typical high energy
Marx 60 capacitors can contribute 1.8uH to the ~12uH
total. A reduction of 10nH per capacitor is ~5% of this
total. Water breakdown stress scales like the inverse 1/3
power of the charge time or the 1/6 power of the
inductance and the increase in a predicted breakdown
stress is only about 1%. However, the increase in
reliability is such that the failure rate due to water
breakdown decreases ~15% for a system operated at 80%
of the predicted critical stress and this is non-negligible.


III. PROTOTYPE TEST CAPACITORS

GAEP manufactured two dry capacitors for these tests.
One, X32081, in the standard 100kV configuration and
the other, X32080, used a number of inductance lowering
techniques. Both were insulated for 100kV operation and
used a dielectric system of paper and polypropylene film
to allow the discharge testing without impregnation.
X32081 is a reference capacitor to benchmark variations
in size, dielectric materials or impregnation affecting the
inductance. The shorter length also permitted easier
handling and assembly. Table 1 shows standard full size
2uF to 3uF 100kV capacitors from both General Atomic
Energy Products (ex-Maxwell) and Aerovox used for
inductance comparison.


IV. INDUCTANCE TEST METHODS

The two methods used to measure inductance were a
ringing discharge setup built by Richard Miller of Titan
and a standing wave method by Bob Cooper of General
Atomics. Both require a test plus a calculation. The
standing wave setup is simple and quick enough for
production testing while the ringing discharge requires
moderate voltage (5 to 10kV). Bruce Hayworth [2]
discussed various test methods in his paper on the subject
and concluded the standing wave was difficult for
inductance lower than ~50nH. He suggested a variable-
inductance multiple discharge method. Tests comparing
these two methods at General Atomic on other low
inductance capacitors had shown that a comparable result
was obtained. One of the goals of this paper is to show a
standing wave test with proper equipment is comparable
to a ringing discharge when measuring at low inductance.

A. Ringing Discharge Method
The basis for this method is that if the capacitance is
known then circuit inductance may be determined from
the period of the an oscillatory discharge waveform. If
the external circuit inductance is small compared to the
capacitor and exactly calculable (e.g. coaxial) it may be
subtracted from the measurement leaving only the switch,
which initiated the discharge, to be accounted for. To
maximize signal to noise ratio and to test a representative
current level, 5 to 10kV is used. A solid dielectric switch
is used to minimize inductance, contributing <½ nH in
32827
32865
SX210E23
3.0 uF
2.4 uF
2.1 uF
STD
STD
STD
11 x 14 x 25
11 x 14 x 25
11 x 14 x 25
Standard
Standard
Standard
All Paper / Castor Oil
All Paper / Castor Oil
Paper / PPL / DOP
Table 1. Test Capacitors.
100 kV Insulation
STD
Mixed
Model No.
X32081
X32080
Meas. Cap
3.47 uF
3.264 uF
Rated Cap
Case Size
11 x 14 x 14.5
11 x 14 x 14.5
Construction
Standard
Lower Inductance
Dielectric
Paper / PPL / Dry
Paper / PPL / Dry

Page 4

these tests. Experience with gas sparkgaps shows that
they add inductance and switch resistance contributes to
imprecise determination
Essentially this is a one shot determination of circuit
inductance, similar to the multi-shot method described by
Hayworth [1], where circuit inductance is plotted against
external inductance with the zero external inductance
value being that of the capacitor. Here the external
inductance is that of a coaxial “top hat” or “shorting
cap” with a solid dielectric switch between the shorting
cap and an extension of the “hot” center terminal. A
Rogowski coil picks up the discharge waveform, see Fig.
2.

of capacitor resistance.
Rogowski
Coil
Solid
Dielectric
Switch
Shorting
Cap

Figure 2. Ringing Discharge Setup – Shorting Cap off.

Since the conductor dimensions are easily measured,
and if we define the capacitor inductance as beginning at
the plane of the ground ring, the external inductance may
be calculated as:

L ext. = 2 l ln ro / ri nH, (1)

where the length l of the center conductor is in cm, ro and
ri are the outer and inner radii of the coaxial terminal.
Subtract the value from that determined from the
capacitance and period. The solid switch contribution can
be estimated from the same formula taking an arc channel
diameter of ~.001 cm and length of <.013 cm (5 mils)
giving ~0.25 nH. A sharp point driven through copper
tape mechanically deforms the dielectric to produce an
intrinsic strength breakdown with a very short resistive
phase. See Fig. 3. The key to this measurement is the
solid dielectric switch and calculable external inductance.
For the hardware used here the external inductance is
4.83nH.

B. Standing Wave Method
The standing wave method consists of a signal
generator and vacuum tube voltmeter connected by
coaxial cables to a copper hat formed over the capacitor’s
Scyllac bushing. See Fig. 4. One adjusts the frequency
from the signal generator until one finds the minimum
voltage response. At this resonant frequency the
inductance can be calculated (See Eq. 2).

L= (4πʝfʝC) -1 (2)

Shorting Cap
Manual Initiator

Figure 3. Ringing Discharge Setup

Signal Generator
Vacuum Tube Voltmeter
Coax Cable

Figure 4. Sanding Wave Inductance Setup.

The copper hat is form fitted over the bushing to
minimize external inductance. The braids of the coax
cable are soldered to the hat near the center and the
coaxial center conductors soldered to a brass washer.

Coax Braid
Soldered to Hat
Coax Center
Soldered to
Washer

Figure 5. Copper Top Hat and Coax Cable.

Page 5

These connections are 180ʔ from each other to minimize
cross coupling. A layer of 5mil polyester insulation
separates the hat from the capacitor center electrode,
while the outside of the hat bolts to the ground ring. A
special brass bolt provides contact while holding the brass
washer in place as seen in Fig. 5. This configuration
provided very stable and reproducible measurements.
External inductance here is the added volume created by
the high voltage barrier. The dimensions of this volume
are measurable and inductance calculable and in this case
the external inductance is .78 nH.


V. INDUCTANCE MEASUREMENTS

Measurements using both methods were made on four
different capacitors. Two additional 100kV designs used
only the Standing Wave Method. The results in Table 2
show that the two methods closely agree after correction
for the external inductance. For the GAEP capacitors, the
difference varies from about .5 to 1.3nH. The 2+nH
difference in the Aerovox measurement can be attributed
to poor fitting of the copper hat. GAEP Models 32827,
32865 and 32864 are all standard 11 x 14 x 25” 100kV
capacitors with the only significant difference being their
capacitance. All of the GAEP capacitors have nearly the
same measured inductance, 27 to 29nH. The Aerovox
unit measured about 30nH.
The inductance measurements show a reduction of
~5nH between the reference Model X32081and the low
ESL Model X32080. X32081 had a 2+nH lower
inductance than the completed capacitors, perhaps due in
part to the drying process where the paper dielectric
shrinks, increasing the headspace under the Scyllac
bushing or the shorter case length.

VI. SUMMARY

Lower equivalent series inductance in the standard
Scyllac style energy storage capacitor provides a cost-
effective means to upgrade existing pulse power machines
and performance advantages in new systems. We have
demonstrated a reduction of ~20% and are working on
further reductions.
The inductance of General Atomics Energy Products
standard 100 kV Scyllac style capacitors measured 27 to
29nH. A similar Aerovox unit measured about 30nH.
Compared to the reference capacitor (X32081), the
X32080 prototype design exhibited a 5nH ESL reduction.
The measurements using the both the Ringing Discharge
and Standing Wave methods
comparable after subtracting the external inductance. A
better fitting copper hat can reduce the greater difference
were satisfyingly
Table 2 Inductance Measurements.
INDUCTANCE MEASUREMENTS OF 100kV INSULATED SCYLLAC TYPE CAPACITORS
Manufacturer Capacitor
Model No.uFDischarge
Meas.
NH
Cap.Raw
External
Inductance
Correction
nH
Inductance
Value
nH
Raw
Standing
Wave
Meas.
NH
28.7
33
26
20.2
28.7
29.1
External
Inductance
Correction
nH
Inductance
Value
nH
Maxwell /GAEP
Aerovox
GAEP (STD)
GAEP (Low L)
GAEP
GAEP
32827
SX210E23
X32081
X32080
32865
32864
3.14
2.097
3.47
3.264
2.53
2.058
32
34.9
30.53
25.56
4.83
4.83
4.83
4.83
27.17
30.07
25.7
20.73
0.78
0.78
0.78
0.78
0.78
0.78
27.92
32.22
25.22
19.42
27.92
28.32

in the Standing wave measurement of the Aerovox unit.
The measured prototype inductance of about 20nH is
probably the minimum for this style of capacitor
representing a reduction of about 30%. A 50% reduction
will require a redesigned terminal. Production of such an
ultra-low inductance capacitor entails resolution of a
number of manufacturing issues.
System inductance is still a function of the integration
of capacitor and switch. Improvements in this area could
take better advantage of a lower inductance Scyllac or, as
in the case of LANL Atlas program, the development of
an integrated capacitor and switch.


VII. REFERENCES

Published conference proceedings:
[1] L. X. Schneider, S. R. Babcock, G. e. Laderach,
“Lifetime Testing of Commercially Available 3.0 uF,
100 kV Pulsed-Power Capacitors,” in Proc. IEEE PPC
’89, 1989, p. 902.

Periodicals:
[2] Bruce Hayworth,
“How to Tell A Nanohenry from a Microfarad,”
EID – Electronic Instrumentation, April 1972,
Pages 36 – 39.