report: investigation into the effects of microsecond ...mil-std-462 mil-std- 1399 ssp 30237 ssp...

Final Report: Capacitor Transient Potential Tests

Report: Investigation Into The Effects Of Microsecond Power LineTransients On Line-Connected Capacitors

Ken Javor

EMC ComplianceP.O. Box 14161

Huntsville, Alabama 35815-0161


EXECUTIVE SUMMARY

An investigation was conducted into the effect of power-line transients on capacitors used by

NASA and installed on platform primary power inputs to avionics. The purpose was to

investigate whether capacitor voltage rating needs to be derated for expected spike potentials.

Concerns had been voiced in the past by NASA suppliers that MIL-STD-461 CS06-1ike

requirements were overly harsh and led to physically large capacitors.

The author had previously predicted that electrical-switching spike requirements representative

of actual power-line transient potentials, durations and source impedance would require no

derating. This investigation bore out that prediction. It was further determined that traditional

low source impedance CS06-1ike transients also will not damage a capacitor, although the spikes

themselves are not nearly as well filtered.

This report should be used to allay fears that CS06-1ike requirements drive capacitor voltage

derating. Only that derating required by the relatively long duration transients in power quality

specification need concern the equipment designer.

ACKNOWLEDGMENTS

The author wishes to thank Mr. Kurt Mikoleit of the Naval Surface Warfare Center, Dahlgren,

Virginia for the long term loan of a Solar 7399-2 spike generator. This equipment, capable of

delivering 8 MegaWatt pulses at 2 kV, was specially designed for the spike requirements of MIL-

STD-1399 and is not standard equipment in the typical EMI test facility.

Mr. Robert Kapustka of the Marshall Space Flight Center suggested rep_e, sentative bulk

capacitors he has employed as a designer of dc-dc converters. These were pulled from actual

parts bins used for space applications. Mr. Kapustka has been with NASA since the Apollo

program.

Mr. Mark Nave of Network Appliance suggested that an explosion container be built when using

the Model 7399-2 to stress tantalum electrolytics, since the electrolyte is sulfuric acid. Given

that evidence of explosive forces was found, this was a sound suggestion.

CS06

CUT

DIP

EUT

FET

LISN

MSFC

STS-1

power-line spike

462)cable-under-test

dual in-line package (integrated circuit package)

equipment-under-test

field-effect transistor

line impedance stabilization/simulation network

(NASA) Marshall Space Flight Center

switching transient simulator

ACRONYMS

susceptibility test requirement (MIL-STD-461/MIL-STD-


1.0 Introduction

This report documents the results of a test in which capacitors were subjected to spiked

potentials of varying amplitudes, durations, and source resistances. The work follows the

development of special transient generators which simulate electrical power bus switching

transients._ In addition to the use of these unique generators, two other generators which meet

the requirements of MIL-STD-461 CS06 and MIL-STD-1399 were employed to stress the same

capacitors subjected to switching transients earlier.

1.1 Purpose

The purpose of these tests is to demonstrate that capacitors connected across primary power

inputs in avionics need not have their voltage rating derated for switching transient-like potential

stresses. It will be shown that it is sufficient to derate capacitor voltage rating for power quality-

related surges, which are lower in amplitude, but have longer durations.

1.2 Scope

Six capacitor configurations were evaluated. All capacitors were selected as representative of

those used in spacecraft applications as hold-up or bulk energy storage or line-to-ground filters at

the input to switching power supplies (de/de converters). Specifically, these capacitors were

pulled from bins where power supply bulk capacitors are stored at MSFC. Capacitors used on

28 Vdc (Space Shuttle), 120 Vde (Space Station), and 270 Vdc (X-33) were selected, as well as a

capacitor rated for ust 30 Vdc, which was just marginally rated to operate on 28 Vdc.

Cap. Bus TypeA 28 Vdc

B 120 Vdc

C 120 Vdc

D 270 Vdc

E <28 Vdc

Nomenclature

M39006/+ 22-0600

M39006/25-0264H

AVX 87106-157 9622A

AVX 87106-237 9726A

M39006/ 22-0560

F 120 Vdc Sprague LC filter

Spike sources were as follows:

Potential Volts Source Impedance Ohms

0- 500 50

0- 600 1.2

0-2500 0.25

Cap Type

tantalum electrolytic

tantalum electrolyticceramic

C _tF140

WVDC

6O

125 2

200

ceramic 15.6 500

300tantalum electrolytic 3O

ceramic feedthrough

Generator Name

0.2 200

Manufacturer

STS-13

Model 8282-1

Model 7399-2

EMC ComplianceSolar

Solar

1.3 Background

The Space Station program imposed a specially tailored CS06 spike (SSP 30237). The tailoring

followed the practice of MIL-STD-461 of the spike potential being twice the nominal line

potential, but ignored the 100 or 200 Volt limit. For Space Station, this resulted in a 240 Volt

spike superimposed on a 120 Vdc bus. There were complaints from some equipment designers

about the severity of this requirement. In 1994, the author presented a paper at the IEEE EMC

IReference NASA report: NASA/CR--1999--209574

2 Two of these capacitors are used in series across the bus to get the proper derating. This results in 41 laF.3Switching Transient Simulator - 1


Number

MIL-STD-461

MIL-STD-462

MIL-STD- 1399

SSP 30237

SSP 30238

NASA/CR-

1999-209574

3.0 Results

symposium that was the result of investigations into this subject (the breadboard spike generator

described therein was developed into stand-alone test equipment as described in the footnote 1

report). A 50 _ spike source was developed which more accurately simulated switching

transients. 4 It was demonstrated that a small capacitor could effectively filter a 50 _ source

impedance 1 Its spike, but could not filter a 10 Its, 1 _ source impedance spike from a Solar

Model 8282-1 spike generator, which is the usual test equipment employed for CS06.

This report extends that simple observation into a full investigation.

2.0 Applicable Documents

Title

Electromagnetic Interference Characteristics Requirements for Equipment

Electromagnetic Interference Characteristics, Measurement of

Interface Standard for Shipboard Systems, Section 300A: Electric Power

Electromagnetic Emission & Susceptibility Requirements for EMC

Electromagnetic Techniques (Space Station)

Specification, Measurement, and Control of Electrical Switching Transients

Capacitors were tested as described above. Capacitance of each capacitor was measured prior to

and atter spike applications. Pass/fail criterion was the absence/presence of a change in capacity.

3.1 Summary _,,

No bulk or hold-up capacitor failed at any spike potential generated by either the STS-1 1 gs, 50

source or the Model 8282-1 10 gs, 1 f_ source. This included potentials of 500 V peak (spike

plus dc bus) liom the STS-1, and potentials of 600 Vdc superimposed on the bus potential

(resulting in peak excursions of 600 Volts above nominal bus potential). Ceramic feedthrough

capacitors were undamaged by the STS-1 generated transients, but failed at higher spike

potentials from the Model 8282-1. These spike potentials were, however, above those normallyimposed by NASA CS06 limits.

Capacitor E failed at a _olar Model 7399-2 spike source applied potential of 600 Volts for 20 Its.

Amplitude was 1.32 kV into a 0.5 f_ resistor. The spike source impedance was 0.25 f2. It took

almost four hours to fail the capacitor, with spikes applied at a rate of two pulses per minute.

The original purpose of this investigation was to show that capacitors do not have to be derated

for positive sense (turn-off) transients of microsecond duration if they are properly generated.

This means a source impedance reflecting that of real transmission lines, as opposed to the Ohm

or sub-Ohm source impedance of the Model 8282-1 CS06 spike source. The point here is that if

a platform mission does not subject it to lightning attachments, a low impedance high potential

transient on a power bus is difficult to explain.

4 50 _. was the source impedance for a positive going or turn-off transient. The negative-going or turn-on transientwas low impedance.

,f •


This investigation amply demonstrates this is the case. All the bulk capacitors except capacitor

E were able to load a 500 Volt (transient plus power bus) potential down to a level below their

voltage rating. None of the capacitors showed any degradation either short or long term after

such applications. In fact, the worst case application applied the 500 Volt spike to a 30 Vdc cap

(capacitor E) biased at 28 Vdc (essentially no derating, even for dc) and caused no effect

whatsoever after three hours of spikes at a repetition rate of one pulse per second.

But in addition to this fact, which it was the purpose of this investigation to demonstrate, it was

also found that the CS06 spike caused no damage to any of the capacitors tested. This was a

much more severe test, due to the low source impedance of the Model 8282-1 spike source.

While the spikes were not as effectively filtered, they caused no lasting degradation. All the

tantalum electrolytics got very hot, and capacitor E measured a higher capacitance just after the

test than before the test or after it had cooled off, but this could not be considered an operational

failure, since more capacitance would not cause a problem for a bulk capacitor function. Spike

potentials generated by the Model 8282-1 were capable of damaging the capacitor of Filter F, but

this occurred at potentials well in excess of normal CS06 limits (above 300 Volts spike potential).

Only a kiloAmp source was able to damage the other capacitors. Tantalum electrolytics failed at

kiloVolt/kiloAmp spike levels (20 Its duration) after repeated spike applications at two pulses

per minute (ppm). Ceramic bulk capacitors failed at the first kiloVolt/kiloAmp spike

application.

3.2 Test Methods

Three different test techniques were utilized. They were implemented in order of increasing

severity. The first tests used the EMC Compliance STS-1 switching transient simulators, which

generated a 1 _ts, 50 f_ source impedance spike. The next test applied spikes from the Solar

Model 8282-1 CS06 transient generator. The last set of spikes were applied fr0"Cffaa Solar Model

7399-2 borrowed from the Navy.

3.2.1 Application of Turn-Off Switching Transients

Capacitor value is measured prior to test start, and again at end of spike application. The test is

diagrammed in Figure 1. In Figure 1, CUT refers to capacitor-under-test.

o'sco UT gran_ iretrt r

Figure 1: Schematic of switching transient test

The oscilloscope measures potential from line-to-ground at the LISN output. The nominal spike

potential is measured without the capacitor in place, using a 1 or 10 M_ oscilloscope probe.

The spike is measured again with the capacitor in place. Spikes are imposed at a rate of at least 1

pps. Pictures of test set ups and capacitors are shown in Appendix A.

A characteristic of the STS-1 not previously discovered is that the switching FET has some kind

of protection circuitry which clamps applied potential to just above 500 Volts peak. Regardless

of switched load (Vspik e = 50 k'-2*Iswitched) or nominal bus potential, Vtota I (-- Vbusq-Vspik¢) is bounded


by just over 500 Volts. Thus the spike applied to the 28 Vdc bus was about 500 Volts, but the

spike applied to 120 Vdc was only a little over 400 Volts, and the spike applied to 270 Vdc wasabout 300 Volts.

In each case (for each capacitor at each bus potential), the initial spike potential applied was 100

Volts open circuit. The spike potential actually applied to the capacitor was much decreased in

amplitude from the 100 Volts measured open circuit. This spike potential was applied for

several minutes. The capacitor value was measured, found to be unchanged, and the spike

potential increased to 200 Volts. The same process was reiterated until the maximum 500 Volt

spike was applied. At this point, spikes were imposed at a rate of at least 1 pps for over one

hour. At least 3600 spikes are thus applied at the maximum potential to which the capacitors are

exposed and to which they are considered qualified. After this final application, the capacitance

is again measured (and found to be unchanged in all cases). Test data is shown below for the final

500 Volt spike application for each type of capacitor and bus.

3.2.1.1 Test Data

Data 1-1 and 1-2 show the STS-1 50 _ spike unloaded and loaded by a 140 pF, 60 V rated

capacitor (cap A). This capacitor is typical of one used as a bulk capacitor at the input of a 28

Vdc dc/dc converter. The spike was generated by switching a 2 f2 load off the 28 Vdc line. A 14

Amp switched current should have resulted in a 700 Volt spike (14 A • 50 f2), but protection

circuitry limited overall (spike plus bus) potential to 516 Volts. The capacitor so loads the spike

that peak spike amplitude (spike + nominal dc bus potential) is less than the de rating of the

capacitor. No capacitor failure is expected under such conditions. This is the case whenever the

50 f2 spike is generated across a capacitor. Therefore switching transient potentials should not

be used to derate capacitor voltages. The switched spike amplitude will not appear across the

capacitor. Demonstrating this was the major goal of this investigation. _

In order to show just how benign the switching transient is, the same 500 Volt spike as of Data

1-1 was generated across a 300 p.F tantalum electrolytic rated at 30 Vdc (Cap E), while it was

biased at 28 Vdc. The resulting spike is presented in Data I-3. It is almost identical to the spike

of Data 1-2, except here the peak spike amplitude (spike plus nominal dc potential) exceeds the

capacitor dc rating. Nevertheless, a three hour exposure at one pulse per second resulted in no

observable change in capacitance, and no observed heating.

A 500 Volt spike (switched load = 20 f_) was also imposed on capacitors B and C, which were

typical of those used dn 120 Vdc power. Capacitor B is two 82 _tF, 125 Vdc electrolytics in

series. Capacitor C is a 33.6 p.F, 200 Vdc. Data 1-4 is an unloaded spike injected on 120 Vdc.

Data 1-5 shows near complete filtering of Data 1-4 spike across capacitor B after over one hour

of one pulse per second application. The same result was obtained when the Data 1-4 spike was

injected into capacitor C for two hours at one pulse per second (Data 1-6).

A more visible spike appears in Data 1-7, which is the spike of Data 1-4 injected into filter F, a

Sprague threaded LC filter with capacitor value of 0.2 p.F. The capacitor was facing the LISN in

this test. Spike amplitude clearly exceeded the capacitor rating of 200 Vdc, but the spike was

applied for two hours at greater than one pulse per second with no change in capacitance.

Data 1-8 shows the 500 V, 50 f_ spike injected onto 270 Vdc. This same spike was injected over

one and one-half hour (at one pulse per second) into capacitor D, a 15.6 laF, 500 Vdc rated


ceramic. No visible spike was generated or recorded. Note that in any case, the unloaded

switching transient barely exceeds the capacitor D dc rating.

A!OO'/DC |O:IP_OBE O 1VOFFIO:IFROEE2jJ__-OIV s_Nr-_.E Tri,3 ;_J' --_OTvI

I , 3_.e

Ii t

"t " . I .4

i

I ' 5t6I I V

[ =_.0..... ..... V

"fl ( 1¢0--.L- ,x ' n_

I i HL'.'X.O

_,SFECIALFL_CT i

_L=N_AE HEA_UR_ PRllfr PRIrffEP.- _ FCeNAT* _ETUP¢

A leVOC 18"IP_OEE B IVOF_Ii):IPRO6E38_/DIVSIhSLE TrI#rAJ"-2DIV

I I

I ' 31.6! ..... L ..... ii : ___ il_

I '

a pS

'...... -_ 44.4

.... -'-- Itl_:" ' 30,@I I| -i ....

' ;'00I! ............. nsI

I iI

I .... !.... ; .....

NEASURE PR_HT PRrH'TERGEri., ¢ FORMAT; _ETLI=$

Data 1-1: 500 V spike injected on 28 Vdc Data 1-2: Spike of Data 1-1 on capacitor A

l(IVt¢ IO:|PROb'E 0 Ivr./FIO:IPR(IBE5Ol_s/DlV £/r_L£ T_..TLglAj"-$OIV

| I ..... C_JR__ _ --

I , 2EL4..... ._,i,.,,_l_ __ V

I_t_

.... , 3.10l.l=

I i

I

"L L ,' -.,v27,6

I y!

I

t J;I

i.,j / !'__1_',.'I_ :GF.J1ERATERF_._ISUREPR{HT PRIMER

t _ FORMAI't 5ETUP¢

r,_

A tO_°DC I0=IPROBE B IV OFF I8=IP_OL_Et_J'-.019 SIH_E Trig :h.r-201';

I , _ :t

t

--I:, ....... \

, \al I.... .,%

I

'l I ISRECIAL FUHCT:GELERATE _EA3UE

I.... L! ....

I

I

I

l

I

I

I

¢

!l

I

I

I

i

I

I

!

!

_RI_Ir P_It_rERFCeMAT_ SETUP*

V=zF I

44_V

V

96 o8V

FREQ:???

Data 1-3: Spike of Data 1-1 on capacitor E Data 1-4:500 Volt spike injected on 120 Vdc


A 5eVDC IO:IPR[_E B IVCIrFIO:IPROBE2GB.u_/OIV SIIIGI.£ ?Pl_tlAl -TDIV

L iI ! ,24

I '

i } i V

I ' -Pl..... -- 1@9

--1 I, V, .- ' MIH-Pt, i A 110

I ' RIVEti Id,,e

i

|I

rI-- ):_k_CIAL FI_ICT:_UIER_TE rqER_JRE PRINT PR£N'T£R

@ _ FOBIATi SETUP =-

HOL_

Data 1-5: Spike of Data 1-4 injected into cap

A 5GV_208_/D IV SI r_LE

I

_ I ,t

I

I

I

_1i

I

t_H

I

_¢I_ECIALFIJHG.Ti

5Er_RkTE I_Et_UR.EoH

I_',):IPP.OBE B IVOFFIO:IJ:RO._ET,.I_lA1 -2D { V

I

I

a

I

I

I

I

!

I

l

' I

!

Iir

l ....

,!II

dl::t.24ms

_IE_I :I27'Y

t]6

_11fl-P:

{I6,Q

B Data 1-6: Spike of Data 1-4 injected into cap C

n 5OVOC IO:I_R_SE B IVOFFIO:IPRgBIE201J_,O|V_II1_l-£ Tr.ig:A[ -20IV

1 _ 124

I t I...... , 12e

I

i , V' _ _II't-P !

.... VI

I...... l

II

.%'IECIAL ,,FUNCT:6EHEFATE PlEA_JRE F_,i'i'tr -FRiH'TER

, , FO NAr;S TL ,

RISE:<l ,_--0

J.l£ "

Data 1-7: Spike of Data 1-4 injected into cap F

AI(,eYOC I0:IPR_E B IV_FFIO.'IPI_OBE' lulA]IVSItlGLE T,,islA.r-_DIV

I I

I

i

| _ '_I ...... I

I

I ,I

I ......_I

I _ "" "Ira_rL_

I

I

I

I

i

I

I

I

I

dt :6.29

_EAM:304V

564

, V_IIN-P:

26_V

- RI_;EI

SRECIAL _t:LltJ_T =

_R_TE ME_SI_E PRINT PRniTER- I_ FOR_T¢ SETUP.*

Data 1-8:500 Volt spike injected on 270 Vdc

3.2.2 Application of CS06 - Like Positive Transients

The Solar Model 8282-1 Transient Pulse Generator was used in a MIL-STD-462 CS06-1ike

parallel injection technique (as diagrammed in Figure 2) to inject up to 600 Volt pulses into the

test capacitors. The 10 Its spike source impedance was very near 1 fL Data 2-1 and 2-2 show

spike waveforms into 5 f_ and 1 f_ loads respectively. Spike source resistance is given by:


R1 R2 (V1 - V2) 5f2, If2 (600V - 340V)

Rsource = V2 RI - V1 R2 = 340V,5f2 - 600V, lO = 1.2 f2

An unloaded source potential of about 700 Volts may also be deduced from this data. The test

set up was identical to CS06 parallel injection with two differences. A MIL-STD-462D LISN

was used in lieu of the MIL-STD-462 suggested 20 _tH coil, and the EUT (capacitor) was

connected directly to the Model 8282-1 parallel output banana jacks. Lead lengths were thus

minimized. Pictures of test set up may be viewed in Appendix A.

LISN

Figure 2: Test set up of CS06-1ike parallel spike injection

3.2.2.1 Test Data

The 600 V, 10 Its spike of Data 2-1 was injected on to capacitors B through E, with no test

failures. All tantalum electrolytics got very hot during this test. These tests were run for at least

one hour at one pulse per second. The ceramics were run at 2 pps with no noticeable heating.

One tantalum electrolytic, capacitor E, read 386 laF immediately at test end (very hot), but 316

_tF after returning to ambient temperature. This was not deemed a failure.

The most important result here is Data 2-3. This is the induced spike across the 300 laF, 30 Vdc

rated capacitor (E) biased at 28 Vdc. With absolutely no headroom to spare, this capacitor still

did not fail when the 600 Volt spike was injected. Note that about 600 Amps w_.. flowing at thepeak of the spike waveform in order to result in an applied spike amplitude of 114 Volts (greater

than 700 Volts source potential loaded through 1 f_ source resistance down to approximately 100

V). It is quite reasonable to extrapolate from this result that no properly derated tantalum

electrolytic bulk capacitor will be damaged by any low duty cycle CS06 - like spike application.

Because traditional CS06 test procedures mention spike repetition rates of 1 - I0 pps, spike

testing was also performed at 10 pps. An initial trial placed a 600 Volt spike setting across

capacitor A. The Model 8282-1 amplitude was turned all the way clockwise (maximum output).

This yields a spike po/;ential of greater than 600 Volts into 5 f2. Data 2-7 shows the resultant

spike across capacitor A. Spike amplitude actually dropped over the course of an hour long test

to a peak of 96 Volts. This was attributed to an increase in capacitance due to electrolytic fluid

heating. Capacitor A, nominally 140 _tF, rated at 60 Vdc, measured 168 I_F cold. After one hour

of 600 Volt spike application at 10 pps, the capacitor case was extremely hot and the capacitance

measured 320 _tF. This returned to the pre-test value of 168 _tF in about four minutes. Figure 3

shows measured post-test capacitance as a function of time, with a superimposed exponential

decrease. Close correlation shows that capacitance decreases exponentially with time:

C(t) = Co (1 + e-t), C = capacitance, Co = initial pre-test capacitance, t = time.

which is a solution to Newton's Law of Cooling:

0T

Ot = T, T = temperature, t = time.


Strong correlation between capacitance and predicted temperature as a function of time indicates

a very high probability of a causal relationship between electrolyte temperature and capacitance.

3-I I I I I I I I I

measured capacitance normalized to pro-test valtlt

- - • • • exponential model of empirical data

0

§

_- 2U

h"

D-•I. 1

hf

o

0-

0

I I !

3-_..-_,q3,..-__.E

2 4 6 8 10time (minutes)

Figure 3: Capacitance as a function of post-test time

Having completed a one hour application of a 10 pps 600 V, 10 _s spike to a 60 Vdc rated

capacitor (A), an identical test was performed on a 30 Vdc rated capacitor (E). Data 2-8 shows

the spike applied to capacitor E (recall that Model 8282-1 amplitude is at maximum, yielding

over 600 Volts peak into 5 f2). Both capacitors A and E were biased at 28 Vdc for these tests.

Capacitor E gave similar results to A post test, but hot capacitance exceeded the capacitance

meter's range and could not be measured until it was nearly room temperature. Post-test, the

clear plastic coating of both capacitors was discolored by the extreme heating. It is possible that

if the case of such a capacitor were in contact with a printed circuit board (PCB), that damage to

the board could result. However, recall that the applied spike in the above easel"was 600 Volts

peak. No CS06-1ike specification calls out such a spike at such repetition rates for an hour.

The same test was performed on capacitor B, the two series tantalums. Capacitor B was biased

at 120 Vdc (Space Station). The applied spike is shown in Data 2-9. Over two hours the

capacitors did not overheat, and immediate post-test capacitance was equal to the pre-test value.

When the 600 Volt, 10 its, 10 pps spike was applied to ceramic capacitor C, biased at 120 Vdc,

there was no effect. The test was repeated with the bias at 190 Vdc to most severely stress the

capacitor. There was t_o effect, other than mild temperature rise after a two hour application.

Data 2-10 shows the applied stress to a 5 Y2 load at 10 pps, while Data 2-11 shows the actualstress across capacitor C.

The above tests on the tantalum capacitors established a significant safety margin between

typical CS06 requirements and any resultant damage to bulk capacitors. However, the heating

issue may cause concern with respect to nearby circuit elements, including the PCB. Therefore

the above tests were rerun with a 240 Volt spike. This is the highest CS06 spike potential yet

levied on a NASA program. Data 2-12 shows the 240 Volt spike applied to a 5 f2 load. Data 2-

13 shows the result when the same spike generator setting as for Data 2-12 is applied to

capacitor E (300 itF, rated at 30 Vdc, biased at 28 Vdc). After one hour of a 10 pps application,

the capacitor was warm, but not too hot to handle. Initial post-test capacitance measured 380

itF, not much different than nominal 315 _tF.


Filter F was an L-filter rather than a discrete capacitor. Line-to-ground cap value is 0.2 _tF, other

specifications, taken from the RFI TM catalog, are given below.

SUBMINIATURE FILTERSM28861/5 200VDC 125VAC 400HZ

_B61m RFI R_NG

DA_-W PARTK C_ AbIP U_.'V A_V

10001 RE6905-1 L, 25 _2_0(] 125

[0002 RF6O0_ I,,, ,25 200 125

SPECIFICATIONSOPERAnNGTEMPER,_TUFIE_5"TO .t_S=CClELEC3RC TEST: 2.5 T_ES RATED _OLTAGE

It_SU!._TION RESISTANCE: 1000 klEC-,-,-,-_Ht,¢_

MI N,@ZS'C

C@NFOR_ TO "i3-tEAPPLICAI_E

RECUIREMENT9 0_: _1 L-'-Zc1861

RLTERSUPPLIEDWITHLOCk'WA.SHER&_ NUTRNISH.HOTSOLDERDIP

......... - .___

M_ MN IVUM INgER"IOH LIDG_ IN 13F-_FFt _ _ ..¢t'O.22_A p1j_ LOAD ]A=.O Ig VQROP I_K_ |._z ..,N_KH_ ll/H! 1UMHZ 1UH¢

.187 .375'_22 .....28 40 60 .60 70 I

,187 ,375_.22 28 40 60 60 70 INo manufacturer specifies filter series inductance, so this had to be measured. A series insertion

loss technique was used. The filter was placed in series between a 50 _ signal sotlrce and a 50 f2

spectrum analyzer. No connection whatsoever was made to the filter case, so the capacitor had

no effect on the measurement. A baseline measurement was made without the filter in line, so

that insertion loss could be ascertained. The equation which predicts the insertion loss is:

Insertion Loss (dB) = 20 log ( 50

q2500+(2nfL) 2 )

Test set-up schematic and equation is plotted below for L = 400 p.H, over a 100 to 1100 kHz

range. Test data shows.excellent agreement with theoretical insertion loss for 400 _H. Test dataupper trace A is a baseline measurement without the filter in the test circuit. The lower trace B is

due to the insertion of the inductor in series between signal source and receiver.

-10

choke-under-t est

-ICOOA

%%

m -20

C _--- -_-- -_ ,_

o -30 " "

-40

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1

Frequency MHz


Lir_azFrequency(IDiv = 100.O0_Hz)

The table below shows the response of the filter to CS06-1ike spikes. Filter F failed from a single

5 or 10 Its at about 450 V spike applied to the inductor end, when biased at 120 Vdc. It could

take the SSP 30237 10 Its 240 Volt pike at 10 pps indefinitely. It failed within two minutes of

350 V 10 Its spike 10 pps applications on the capacitor end. When spike duration was reduced

from 10 gs to 5 Its, damage threshold at 10 pps injection increased from 350 to 420 Volts (it took2 minutes to fail at the

spike duration ps

5

same rep rate at 420 Volts).

single pulse failure point Volts

464

10 450

n2811_OC IO:IPGOBE 8 5VOFFIS:LPROF_E5.us._O!'-pSIIIGLETr_g:AJ" -20IV

"T , dt :

I

[

J

)_PECIALGEMERATE MEASIJRE PRIIiT

@ , FOPlHAT_

10 pps, 2 minute failure point Volts

420

, 3,40

__! ......... _ ....... ._ It

' I/d_:: ,I

294

,_ _ ......... I:Hz i"- ",I.............. 6_)O

I v I '

' 8 _O ,!

i < 400 ,i i'_5 't

I "el '_

!

FU('ICT_ SPE xAL1FU[_TtPRINTER_TUP_

Data 2-1:600 V spike across 5

35O

AI@OVOC I@:IPROBE B 5VOFFIO:IPROBE5.,._s/I_IV&II4GLE 'I'_,.L::A J"-21}IV

/f

t/dr t294

_X-_ i

V

56.0V

RISE:1.20

_LD

GEHERATE_IEA3URE PRIHT PRIIfTER i_¢ _ FC_.MAT_SETUP,_

Data 2-2: Data 2-1 generator setting developsthis spike across 1 f_


A 20:/DC- Ig:]PROBE E: t;'OFFLQ:I F,°.OEE50.ul..OiYS]FIGI.E Trig l,_J"-201V

-' If', :LI3I , .us

' RI_IS|I31 .-_

i __.-L_a V

i -PlI •""-- i 14I ...._ i

I V " ' R_ISEI

I It _ .--!

|. I

_EI'ERATE _EI:_I.._E PR[DITt i F_MFiT¢

....._IItI-PI

5]VP_" 1,3:IPEOBI_ I; I_/OFFI,:,:IPRC_E2L_uJ.:D{VSINGLE T,.Igthl -2D[V

11

,!I

_ll,/

: 1 _u_-!_!

IJEAI%:i _._ 12.4I :,J

' r,lg_-P"_ ' , _,

I, V

\ -- l_[H-P'

j. E __ISE I' -:_ ,2eI .1.1_

I

t;

F'RIM'TERSEIUP_" _

/,

,/./.

--1U_B_tAL IEUH!_,"

eBtERQTE ilE_$URE PRIHT$ $ FOR_&QT$

Data 2-3: Data 2-1 generator setting develops Data 2-4: Data 2-1 generator setting develops

this spike across capacitor E this spike across capacitor B

A E_V DE

I

I

I

"-i-

I

-5-

I I

SI_CI_L Ul_£T___EDERATE 1,1EP,SLIRE

,= ¢.

I¢:iPBOBE BTrig IAJ"-?.DI'!

I

I

I

I

i I

!

i ,!

!

!

I

I

!

PRINT PRINTERF,]RMAT¢. SETUP_

IV OFF I0 _1PROEE:

dt :62Q

.us

r{Al_:

V

MI_,-P:221_V

l_IN-Pi

42 .Q

R I._E:

MOLD

QlOOVOC Ie:IF'_OBE B IVi_'FIQ:IPRO@E

Data 2-5: Data 2-1 generator setting develops Data 2-6: Data 2-1 generator setting develops

this spike across capacitor C this spike across capacitor D


Data 2-7: Maximum generator setting develops Data 2-8: Maximum generator setting develops

this spike across capacitor A at 10 pps this spike across capacitor E at 10 pps

A lt_3V OC 1(9=IPROBE5g_/I}lV _IH_LE

I

t

I

IlI

I

I

BT,'Is:A!-2OIV

i

!

.,,l

!

¢

!

!

!

IV OFF !O;I PROBE

I

. ..... J]I

% I!

i

I

I

GEIIER_TE MEASURE PRINT FRII_TER$ FORHAT_ SETUP¢,

dt:31_9lJs

[16V

308V

- ;:!_F.:<,t ._19

A .us.

Ig:l PROBE B IV OFF lO : ! PROBETrJ9 IAJ" -2DIV

• .... i

PRIIrrERFOF_IAT¢ 5EllJP¢

Data2-9: Maximum generator setting develops Data 2-10: Maximum generator setting

this spike across capacitor B at 10 pps develops 650 V spike across 5 f2 at I0 pps


LFJ:IPROBE B ]VOFF|g:IPP,OGE AltnOVDC IO:[FROBE B IV OFFIO:!PROBE20us_'DIV Trig ;A[ -2D.:V

I

, 0l I' ' VI

i !

,, i 244--l- ! , y

t _ 'I

_[I , V

, 'I'I I I T :u.

.c,t_CIF-4.FUI_T ! _ --L

GENERATE _F,_LIR.I_PP.IIM"PRRI_R." .'- FORMAI_ 9ETUPo

Data 2-11" Data 2-10 setting delivers this Data 2-12: Model 8282-1 generator set to

spike across capacitor C at 10 pps develop 240 spike across 5 _ at 10 pps

A ZOVOC IO:IPROBE B IVOFFI@:IPR_I_2Du_/DIU TriglA]" -3OlV

"L-_£1-

I

I

: ....

SPECIAL FUI'IF_T_.-GENERATE MEASUR_

¢ ¢

L:3

AJC)C-JVDCI_):!FROBE B IVOFFIO:IPRO_IBu--/OlV_I;IGLET_'Ig:AJ" -2OlV

, 124

, FIEAH:2_.e

, 9

' qP.X-P :I "

,..... 61.6

, 16.g

I

I

I

i

P.UII|

PRIHT_RSETuP¢ E

I

I

a.rI

!

I

I

I

I

I

;II

I

MF.AM:4,00V

_IAX-P_244Y

hl|N-P:- leClV

_I_E:BSOR ns

H_D

GENERATE rqEAcUJRE /_RI_IT¢ $ FORMAT:

PPIPIT PP,INtERFORIIAT¢ SETUP# B

Data 2-13" Data 2-12 setting delivers this Data 2-14: Model 8282-1 generates a 240 Volt

spike across capacitor E at 10 pps spike (SSP 30237 CS06) into 5 l')


AtOUT[_ ]O;1FROgE B |','OFFte3:tPROBE1,3.us_'DIV r r_ _ : nJ' --_D[ V

I , _IE_I1:, lIE

i = NAX-F...... t 53E

' MII't-FI

j V

!

,_,_L I ..... - - .-I

I

I

I

I sCECIAL FUHCT = .. ',GEItERftTE I_IER_'UF_ PRIri't PRIHTEE

AlOe/Or. IO:IF_RO!:E 8 I'¢OFFIO:IPR_n_t29p_-,'D]V [rJ,3 :_f -3OIV

I2_.

i :-f

"i- [ ' "i H lI'l-;I _' ..............' - t?_

IIt , V

, _--+_j i RI'._E'.

_.J....... ',...... -- -. _ .u_

I I',

I I L.• " I ...... _-"_"f

I I I

........j RUN

_RECI_L =r_'PCRATE _I_SURE PR'IOT PR+I_R.I_-EIR"'

Data 2-15: Filter F response to Data 2-14 Data 2-16: Filter F response to Data 2-14

spike injected into capacitor side spike injected into inductor side (measured

across capacitor

Filter F provides the best example of how correctly defining a switching transient provides design

relief. Data 1-7 is filter F response to a 50 _ spike of total (spike plus line potential) 540 Volts

(Data 1-4). The applied spike measured 420 Volts (540 - 120 V) open circuit, and would have

measured 210 Volts into 50 _. Spike amplitude in Data 1-7 is only 116 Volts (+,2,36V - 120 V).

Filter capacitor succeeds in loading the transient potential sufficiently that no damage is done.

Contrast this filter efficiency with what happens when a Model 8282-1 spike is applied to the

same filter F, as shown in Figure 3. Data 2-14 shows a 240 spike (from SSP 30237 CS06)

developed across 5 _. Data 2-15 shows the same spike applied to the capacitor side of filter F.

The peak potential has increased from that shown in Data 2-14 because the capacitor is a higherimpedance than 5 _. If the filter is turned around so that the inductor bears the brunt of the

spike, the potential across the capacitor is shown in Data 2-16. Spike potential is reduced from

that measured when i_jected directly into the capacitor, but it still exceeds the potential the

Model 8282-1 can deliver into 5 _. The inductor is incapable of significantly filtering the spike.

Figure 3"

LISN II

Test set up of CS06-1ike parallel spikeor inductor side face the

filter-under-test 8282-1

I-r- I

___ O'SCOp

injection for single filter F. (Either cap side

spike source.)

However, tlae test set-up of Figure 3 harkens back to the days when a MIL-F-15733 type filter

was installed in just such a line-to-ground configuration because power current was returned


through metallic platform structure. Nowadays it is much more common to return current on a

dedicated wire, at least until return current exits the equipment enclosure. The test set-up of

Figure 4 is a better model of modem power bus and filter topology.

filter-under-test

80pF, 125Vdc j _ I 8282-ILISN __ jfilterF_ J 6 j_-- ]

Figure 4: Test set up of CS06-1ike parallel spike injection for two filters F in modem power

bus topology.

In Figure 4, the filters F are paralleled by 40 gF of bulk capacitance. This should help to lower

the spike potential across capacitors in filter F, but filter F inductors interfere with that function.

Data 2-17 shows a 450 Volt spike calibrated across 5 D. This is the potential at which a single

application destroyed a stand-alone filter F. Data 2-18 shows the spike potential when the

setting resulting in Data 2-17 is applied as in Figure 4. For comparison, Data 2-19 shows spike

potential between filters F and the bulk capacitors. Almost the entire spike potential is dropped

across filters F. However in this test configuration the 450 Volt spike did not affect the filters F,

even applied at a rate of 10 pps for several minutes. When the spike potential was increased to

500 Volts (at a rate of 10 pps) it caused filter F capacitor failure in less than one minute.

A2ggVl_ 10:IFROBE B 1VOFFi0:IPR(IK A20DVD¢, IQ:I_'R0_E B I_'0trI_i_.'IPROL__-'_I_-_/fiIVSINGLE Tr_9 :k_ -2DIV 20$,I-/DCV_,I[IGLETei9 tC_.l"_-20lV

--I.... , " s I , RM_"

't VIg4 I I I9:"

j I t HQX-F

i _ 456 94_.

' I ' Hllt'l _

- ' RI_E: '.... ' RI_"I I, | 1.(_ --' F.............. " " 28,(

t---i " t L

....... ...... '- ""'-- --i -r!.l " ' " =

' t ', ' '_ ....... ii HOLD

__.PIECI¢-/= FUI_CT:/ L' ___ SF_CIQL. FL_ T:..... ]F;THPTBI{.IIIER,RIE HEASUREP.p..z.r_c_.;LRj_UT__., _[.Ef._r[ Mcn._u_ __.Url_r. _5_t_r_¢_

Data 2-17:450 V spike across 5 _ Data 2-18: Data 2-17 generator setting develops this

spike across filters F in Fig. 4 configuration

_',TPh,1


A 50YEF tO:IPRC_E B IVOFFI_3SlP_20_s,.'DtV SItiGLE TrJg.'Cd"-gB[V"I , RMS :

I I 122

I _ X-PI

V

:l,a_r_ _ ,_.__...,,__,..mm* .mm",.w....... '!...... E:

I _ _

I

I

1

J: ,JTJ) ,SPECI "_ 2

Data 2-19: portion of Data 2-18 spike potential dropped across bulk capacitors

3.2.3 Application ofMIL-STD-1399 - Like Positive Transients

The Solar Model 7399-2 Transient Pulse Generator was used in a MIL-STD-462 CS06-1ike

series injection technique (as diagrammed in Figure 5) to inject up to 2000 Volt pulses into the

test capacitors. The source impedance of this generator was measured to be 0.25 f_ (for the 17.5

its waveform). Data 3-1 and 3-2 show 17.5 _ts spike waveforms into 10 f2 ,_d 0.5 _ loads

respectively. The equation which gives the spike source resistance is:

R1 R2 (V1- V2)

Rsource/17.5 = V2 R1 - V1 R210_*0.5_ (1940V - 1320V)

1320V- 10O - 1920V-0.5_= 0.25

R1R2 (V1- V2) 10,0.5 (2460-2000)Rsource63 = V2R1 - V1R2 = 2000,10- 2460,0.5 -- 0.1225

An unloaded source po'_ential of about 2000 Volts may also be deduced from this data. The test

set up was identical to CS06 series injection with only the 7399-2 replacing the Model 8282-1.

The capacitor under test was placed in an explosion chamber in case it failed catastrophically.

Pictures of test set up may be viewed in Appendix A.

7399-2 cap under test

-L, ', ..z_I !

I 'SCO

1. I

cap large compared to CUT

Figure 5: Test set up of CS06-1ike series spike injection


3.2.3.1 Test Data

The 1320 V spike of Data 3-2 was injected on to capacitor E. Capacitor E is 300 ItF, rated for 30

Vdc. The power source bypass capacitor was 3300 pF in order to force the series injected spike

potential to drop mainly across capacitor E. Data 3-3 shows the spike potential as measured

across capacitor E. Note that it is very close to half the potential of the same spike source

applied to 0.5 EZ. Capacitor E presents a 0.25 f2 load impedance. In turn, this implies a peak

current flow of 5.2 kA in capacitor E. The spike potential of data 3-3 was applied for three

hours (at two pulses per minute, the maximum rate achievable with the 7399-2) with no failure.

At 3:40 into the test, a very large popping sound was heard, the applied spike was very large,

and capacitor E now only measured 1.5 pF. It measured 1.5 ItF a day later as well. The hot side

of the tubular capacitor was blackened from arcing. The ground side has a domed characteristic,

as opposed to the healthy capacitor whose fiat end cap is that of a cylinder. This is evidence of

high pressure developed in the capacitor. A picture in Appendix A shows capacitor E along side

a "healthy" capacitor A. Although heating must have occurred in order to generate the pressure,

at no time did capacitor E become hot or even warm to the touch, in contrast to the CS06-1ike

spike application, which injected much less current (600 A), but did so 30 to 60 times as often.

A single application of the same spike destroyed capacitor C while biased at 120 Vdc. A picture

and description of the damage is shown in Appendix A. Single applications of the 63 Its spike

were applied with successively greater amplitudes to capacitor A. Starting at 500 Volts (as

calibrated into a 0.5 f2 load) superimposed on 28 Vdc, with 100 Volt increments up to 1 kV, and

200 Volt steps thereafter, it was determined that the single application threshold of damage

occurred at 1400 Volts. At this potential, the post-test capacitance decreased to 1 ItF or less

(nominal was 140 I.tF, pre-test). In case the step-by-step incremental testing had weakened the

capacitor, a fresh capacitor A was also subjected to 1400 Volts, and also failed. Test data 3-6

through 3-15 show samples of calibrated and applied spikes. _,_

A 530Y 0C 18" 1 pR£_le B iV ceF' lcJ : 1PR£_ A _OV DC28u_,,:OIVSINGLE Tri3:AI -21]IV _oO_,_/OIVSIIEIL_

i i I. I

'i a 5"2t_....... , V I .

I

, lee--t-- I1, , v. 1.......

I_l - i I

.... ,\ , 1,9,1, I_V I

_ ' _IIlq-F ,

I " |

a.l'1

,, ..... l__.._ _ _ _ ---_I

1 ,

I I ISPEr_I@L FUIICTt

GEI_RArE M_SL_E _.P_R[U_'. PRJNT_ER

Data 3-1:1940 V, 17 Its spike across 10 f}

IO:IPRC@E B 1VOFFIO:IPROBE

,t I t_ECI At. FUNCT _

_-_'HEI_TZ HEA_JRE .FR_.[IJT_.FRJJ!t_ER

Data 3-2: Data 3-1 generator setting developsthis spike across 0.5

t30_

.... I i V

60 ,E" I V

I

' -- 1.3_f"

' - _IN-F

i VI

' RISE"

,_-- --- .... .U_I


A 58_ C_: |a._-"iPRC_E B IV OFF I0:iPROE_2G_s.'OT',)_tqGLE Tr._glh_ -2O[V

1

I , -

r\ ,I| ......

I l :- I -

I iI

I

l ,r

!

GENERATE M_SUR_ _r_ 1_ _ P.P.,.IN_TER

Data 3-3:

dr:124p5

?,PCV

MAX-f:2.4_

kV

RIgE.',t2Ells

P_LD

2460 V, 63 ps spike across 10

aOG,3YDC IO:IPROBE B IVOFFICI:IPROB@21_t_/DIVSINGLE Tr.ig:A.r-_-'_IIV

tI ..... i dt :

I ,, 12,1

I

i J 42E

.... v VI

I _J t ,_X-F.......... , _.OE

I _ , k'V

I l _s

all-i ..,,_ i'

L.

6{tIERATE I_EASURE PRIrIT PRI;IT_

HOt,m

Data 3-4: Data 3-3 generator setting developsthis spike across 0.5

Data 3-5:

")Om_IDIV SIh_I__

I

---h

I

!

I

I0 :iPROI_ B IV OFF i_: II_ROBET_q-_iAJ"-2DIV

I

Y

J

d

L

I

I

I

!.

i_= ..... i_ ""

, VI

I

_IIN43_V

__ p..=.

-IOLD

I

8L_IR_,,qT_ MDqe,_RE p__.I.F.,I_, p._..I_M..T__e

Data 3-I generatorsettingdevelops thisspike acrosscapacitorE, which caused

eventual failure


_26eV pc2B_/DIV SIIIGLE

I I

_I............. I

II.

I,

Il

\II

aI_- _ ...... r

N '"i "" "_- --:":_

J

l,J, ......t _OLD

SP_{:[AL F :

GE_IERATE NE_UR_ _RI:M!" t_R,.II/,,_,_ I_

Data 3-6:600 V 63 ps spike across0.5

IO:IPROBE B IV OFF IO :IFROBETr'ig:A.F-2air

dt:124ps

V

V

RI&E:- IIC

AIO@VOC I_:IPFgBE B IVOFFIS:IFROB_20_t"OlV SIflGLE Trio:Al -201V

-T-r__i_- [ j dt :124

I _ _.................. ' U_[t r,lE_l,l ,z

I _ , 6_ .[

L , V

' -Ze,_

"-I---'" - -,I- ;,I I'I

iSPECIAL F_ .! ...... ';.r,Er_:RCJE M_S_ PRIt_T PRLFIT_E_-

Data 3-7: Data 3-6 generator setting develops

this spike across capacitor A

2_VOC I@:IP_BE B20,u,_,_IV $_HSLE T_'ig:AI

F_r__:-I.... ....

Iii .....

I%

.L_%

I

- 7[ -- --

SFIECIAL FUI'£

-- I

i .

[1GEt.ATE MER_.WJRE

I

I

!

II

I' I

![_-_ -- mmm_ il I

-I,. I

%.,.', I

-- ,,_!

...... i

2(D_/DI V SIt_Li_

I

Data 3-8:800 V 63 _ts spike across 0.5 f_

IO:IPROBE B tV_FIO:IPROmT_,_9 :A2 -2OZV

I l dt:

I

_.,. O0d-- ,1 V

I ,I r_-;I

I " I 4_, V

] ps

"_'2-" _ . XZ,.,.I- -'-'I-..A"

tV (qrF tO :IROBE A2QOV DC

' d iLO

SI-_:C_AL " :

_EIIEI_TE I,IE'ASUI__P_R.[.H_T.P___II'_.I'_ER

Data 3-9: Data 3-8 generator setting develops

this spike across capacitor A


n2OOYOC A2"OOVDC. JO:tF_OEE B I'40FFIO:IPROBE2Eh,_..41IV gIfllSl.£ TrlgIA,I' -20I', e 20.ue..'DIV S/r'Ir, LI_ Trjg;C,.F -2DIV

I , [24

I I_AI| t j _IE_'J:2_ I 96 ,C

' ,| ....... ,_.... V --'P- - j , -° V

I[ '1 l'mX'; I.......... , t ._ ' 544

I i_ _ -- I, k_/ I V""I, ........... _ ' ' ' R_SE , l'_l_"

\ "I -l_.(_,, I' 2I i tJ:_..... L ..........

I0:IPROE:E B tV OFF I0 :IF'ROB|

.....{ "x' ',

_L FI._CT: | I

Data 3-10:1 kV 63 Its spike across 0.5 £2

i-_LB

AI

I

I

l ....... I

_ _ _ _ I II I 1I

I 'i i

SPECIAL Pt.!ItCTz 1 ,_.....GENERATE IEASURE _P._[r_T.. P.R_LI__[_'_

Data 3-11" Data 3-10 generator

develops this spike across capacitor A

setting

A3.30V _, 10:i PRO_ I_ 1VOFF 1$:iPR(_E2Gp_43I'_' $1NI'-_.E l"_,lp:Af -_0IV

I

It--

i

I

I. | , r=

I

I

\

iII

I

]I

I

t

,,!

._.... a_ :12_

,, p_'I

v

_-;t._

k',,,'

RI_E:-19.1

U_

_ I

I

I

I ' ',' _

-EFEC'r.¢.tL_E_IEPATE MEASURE PRIMT PRIBTER

Data 3-12:1.2 kV 63 Its spike across 0.5 f_

HOLD

A2_3OVDC IO:IPROBE B IVOFFIO:IFROBE20.us,43IVSIR_LE Tri_:sA] -}hIV

1 1241 : : _s

I ' rEAM II

--i.... , V

I __, 6&E

I , V

I

!

I

I

I

I

...-..pI

-t-•:FIE., I_ IELI/_-IT• i

C_MERA_E I_t',SL_.E PRIttT PRINTER

Data 3-13: Data 3-12 generator

develps this spike across capacitor A

HOLD

setting


9500'¢0C IO:IPFOBE 6 IVgFFIO;IFROBE R;'OC_DC IO:IFRO_E B IVOFFIO:IPROB[2B.t_/0IV _;h'IGLE Tr,i.g'A.r --_0Ilt _'gJ,If!l:l]'-/SIN!_LE Tr,i__IAJ" -2DIV

, 12_' @I

' r,le_n' : ,I.......... , .m3( --_t_- - ......... J-

r" '] , t ',

-,-4- X " It: i "I _'¢" I 1 'I L._:_ " T -,_.... ,\a-It _ '

• is a 1. _ _ - i . -- -- -- -- - -- l irqkl-1- m

i l t

SECIAL _FI#;'ICT_ " ' ,_ SFIECIAL '._EIER_'_TEHEAS/dRE PRi('iT PR{UT_E.R

Data 3-14: 1.4 kV 63 gs spike across 0.5 f_

L

16,<V

95_V

19E:

GF..tlEF_T_ _A.SURE _P_RrrLT _j!!T_E.R

Data 3-15: Data 3-14 generator setting

develops this spike across capacitor A.*

* Note that whereas at lower potentials the capacitor A spike potential was roughly half of the

potential applied to 0.5 f/, here the capacitor potential has risen to about 70% of the 0.5 t'l

potential. This was indicative of the onset of capacitor failure. After application of this spike,

the pre-test nominal capacitance of 140 gF was reduced to 1 gF.

-'r_÷

Appendix A

Test Setup Pictures

f


Picture 1: 50 f_ spike source test set up. Capacitor is connected to LISN power output and

brass bond strap to minimize potential drop across capacitor leads and connections. The power

supply hot output is connected to the back side (power input side) of LISN. k]_large capacitor

also bypasses the LISN power input to ground, to provide local energy storage. Switched

current is high in magnitude relative to what the power supply can deliver (also there is

inductance/resistance in leads between the supply and LISN). The electrolytic capacitor (on the

order of mF) has its high side connected to the LISN power input connector using a solder wick

bond strap less than one inch long, and a sheet of thin brass stock from the low side of the

capacitor to the ground plane, providing less than 3:1 length-to-width ratio in the ground path.

The picture above is schematically depicted below (Figure 1 from section 3.2.1):

switching

UT transiento'seo generator


o'scope hotconnection

tantalum

electrolyticcap-under-test

o'scopeground low :%

connection Impedancebond strapconnection

to ground

Cu tape for low impedance bond

Picture 2: Close up of tantalum capacitor connection to LISN output in 50 f_ spike test set

up. Capacitor lead length has been minimized by close connection to LISN port and use of brass

bond strap extension of ground plane.


MIL-STD-462D LISN 50 I_H, 50fL 50 A (used as power

supply protection filter only)

Solar Model 8282-1

CS06 spike sourceceramic DIP cap-under-test

Picture 3: CS06 parallel injection test set up. The LISN here is a filter protecting the power

supply output from the CS06 spike. Note that capacitor under test is connected directly to

Model 8282-1 parallel output jacks to minimize lead impedance. Figure 2 (from section 3.2.2)

below is a schematic of above test set up: _

LISN8282-1


Picture 4: Close up of ceramic capacitor under CS06 spike parallel injection. Minimized lead

length is apparent. Note that ceramic capacitor is configured as a DIP, but all leads on one side

are common. These were interwoven with solder wick and EMI tape to provide very short, low

impedance leads for connection to either LISN terminal and brass bond strap (50 f2 spike test) or

Model 8282-1 parallel output jacks.


0-270 Vdc power supply

oscilloscope

Power source

bypass capacitor

-- Solar 6220 CS01transformer

Solar 7399-2 spike source explosion chamber

Picture 5: MIL-STD-1399 series injection test set up. Explosion chamber holds capacitor

under test. Solar transformer is extra protection for power supply, recommended by Solar 7399-

2 instruction manual. For clarity, it is not shown in schematic below (Figure 5 of ffection 3.2.3):

7399-2 cap under test

.....--1-' LILI_ _ _t_ .......

I I 'SCOI I

I I

/I V._I

cap large compared to CUT


PVC perimeter

Lexan front and back

panel removable forcapacitor access

capacitors-under-test

Picture 6: Close up of explosion chamber with capacitor B (two series 125 Vdc rated

capacitors ganged for Space Station power bus derating) installed.


good cap A ground side

note end capdeformation

due to highinternal pressure

failed capacitor E

ground side up

Picture 7: End view comparison of failed capacitor E with good capacitor A (same package).

Capacitor E is seen to have undergone explosive compression, which, however, was contained by

its case. Capacitor failed open; its failure would not have directly caused other circuit

malfunction, except by its failure to perform its original function.

Picture 8: End view comparison of failed ceramic DIP capacitor C with good capacitor.

Capacitor C dielectric has broken into pieces. Capacitor failed short, blowing a 1 Amp fuse in

the power supply.