Upload
phamxuyen
View
223
Download
0
Embed Size (px)
Citation preview
Weibull Grading and Surge Current Testing for Tantalum
Capacitors
Weibull Grading and Surge Current Testing for Tantalum
Capacitors
Alexander TeverovskyDell Services Federal Government, Inc.
Work performed for GSFC Code 562, Parts, Packaging, and Assembly Technologies Office,
NASA Electronic Parts and Packaging (NEPP) Program
1Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
OutlineOutline
2
Scintillation breakdown and self-healing in tantalum capacitors
Failures as Time Dependent Dielectric Breakdown (TDDB)
Weibull Grading Test (WGT) Surge Current Test (SCT).
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
Failures in Ta capacitors occur typically either under steady-state operating conditions, or at the first power turn-on.
To reduce the probability of the first type of failures, capacitors are screened/qualified by WGT.
SCT is used to mitigate the risk of the second type of failures.
What are these tests, how effective they are, and what can be done to improve the efficiency?
Failures in Ta capacitors occur typically either under steady-state operating conditions, or at the first power turn-on.
To reduce the probability of the first type of failures, capacitors are screened/qualified by WGT.
SCT is used to mitigate the risk of the second type of failures.
What are these tests, how effective they are, and what can be done to improve the efficiency?
Scintillation Breakdown and Self-Healing (SH) in Tantalum
Capacitors
Scintillation Breakdown and Self-Healing (SH) in Tantalum
Capacitors
3Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
Self-Healing in Wet Ta CapacitorsSelf-Healing in Wet Ta Capacitors
4
Scintillations are local momentary breakdowns terminated by self-healing.
SH occurs almost instantaneously and is due to additional electrochemical oxidations of Ta in the electrolyte (sulfuric acid).
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
Breakdown margin reduces substantially with rated voltage, VR.
SH is important to assure reliability of the parts, but it is not a panacea because of the pressure build-up due to H2 generation.
Breakdown margin reduces substantially with rated voltage, VR.
SH is important to assure reliability of the parts, but it is not a panacea because of the pressure build-up due to H2 generation.
0
1
2
3
4
5
6
7
0 25 50 75 100 125
VBR
/VR
rated voltage, V
Wet Tantalum Capacitors
standard, M39006high volumetric efficiency
0
1
2
3
4
5
6
7
0 25 50 75 100 125
VBR
/VR
rated voltage, V
Wet Tantalum Capacitors
standard, M39006high volumetric efficiency
0
20
40
60
80
100
120
140
160
180
0 5 10 15 20
volta
ge, V
time, sec
04005 wet Ta capacitors
210uF 125V, 2.5mA
870uF 60V, 10mA
0
20
40
60
80
100
120
140
160
180
0 5 10 15 20
volta
ge, V
time, sec
04005 wet Ta capacitors
210uF 125V, 2.5mA
870uF 60V, 10mA
Self-Healing in MnO2 Ta CapacitorsSelf-Healing in MnO2 Ta Capacitors
5
SH is due to local overheating of MnO2 and its transformation to a high-resistive Mn2O3, thus isolating the defective area of Ta2O5.
Tantalum can oxidize by oxygen released from MnO2.
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
Although the breakdown margin decreases with VR, it remains above ~2. Scintillations were damaging in ~30% cases. The larger the nominal capacitance and the lower the equivalent series
resistance, ESR, the greater the probability of damage. SH is a benefit of MnO2 parts, but scintillations should be considered as
failures during screening and qualification.
Although the breakdown margin decreases with VR, it remains above ~2. Scintillations were damaging in ~30% cases. The larger the nominal capacitance and the lower the equivalent series
resistance, ESR, the greater the probability of damage. SH is a benefit of MnO2 parts, but scintillations should be considered as
failures during screening and qualification.
1
2
3
4
5
6
7
8
0 10 20 30 40 50 60 70
VBR
/VR
rated voltage, V
Ceramic and Tantalum Capacitors
MIL Ta chip
COM Ta chip
1
2
3
4
5
6
7
8
0 10 20 30 40 50 60 70
VBR
/VR
rated voltage, V
Ceramic and Tantalum Capacitors
MIL Ta chip
COM Ta chip
051015202530354045
0 20 40 60 80 100 120
volta
ge, V
time, sec
MnO2 Ta 100uF 16V at 100uA CCS
SN6SN7SN8
051015202530354045
0 20 40 60 80 100 120
volta
ge, V
time, sec
MnO2 Ta 100uF 16V at 100uA CCS
SN6SN7SN8
0
10
20
30
40
50
60
0 20 40 60 80 100 120 140
volta
ge, V
time, sec
MnO2 Ta 47uF 20V at 100uA CCS
SN1SN2SN9
0
10
20
30
40
50
60
0 20 40 60 80 100 120 140
volta
ge, V
time, sec
MnO2 Ta 47uF 20V at 100uA CCS
SN1SN2SN9
Self-Healing in Polymer Ta Capacitors?Self-Healing in Polymer Ta Capacitors?
6
Contrary to MnO2 capacitors, where ignition is possible due to the exothermic reaction of oxygen that is emitted by MnO2 with Ta, polymer capacitors do not ignite.
For this reason, manufacturers “suggest” derating to 0.8VR compared to 0.5VR that is “recommended” for MnO2 capacitors.
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
Preliminary results show that all polymer capacitors failed in a short circuit mode – no self-healing.
There are other concerns that need to be addressed before using polymer capacitors in high-rel applications (e.g. effect of humidity).
Preliminary results show that all polymer capacitors failed in a short circuit mode – no self-healing.
There are other concerns that need to be addressed before using polymer capacitors in high-rel applications (e.g. effect of humidity).
0
5
10
15
20
25
30
0 5 10 15 20
volta
ge, V
time, sec
Polymer Ta 10uF 10V at 30uA CCS
SN11SN12SN13SN14SN16
0
5
10
15
20
25
30
0 5 10 15 20
volta
ge, V
time, sec
Polymer Ta 10uF 10V at 30uA CCS
SN11SN12SN13SN14SN16
0
20
40
60
80
100
120
140
160
0 5 10 15 20 25 30 35
volta
ge, V
time, sec
Polymer Ta 82uF 75V at 500uA CCS
SN12SN13SN14SN15SN16
0
20
40
60
80
100
120
140
160
0 5 10 15 20 25 30 35
volta
ge, V
time, sec
Polymer Ta 82uF 75V at 500uA CCS
SN12SN13SN14SN15SN16
Life Test Failures as TDDBLife Test Failures as TDDB
7Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
Scintillations during life testing should be considered as failures. Life test results can be simulated using the TDDB model, but
parameters should be adjusted.
Scintillations during life testing should be considered as failures. Life test results can be simulated using the TDDB model, but
parameters should be adjusted.
100uF 16V 144hr at RT 32V
1.E-07
1.E-06
1.E-05
1.E-04
0.01 0.1 1 10 100 1000
time, hr
curre
nt, A
100uF 16V 144hr at RT 32V
1.E-07
1.E-06
1.E-05
1.E-04
0.01 0.1 1 10 100 1000
time, hr
curre
nt, A
1
1ln pVBR
BRo
BRo V
VkTHt
VV
kTH
kTHtTTF 1expexpexp
100uF 16V at 125C 24V
time, hr
cum
ulat
ive
prob
abili
ty, %
1.E-2 10001.E-1 1 10 1001
5
10
50
90
100
Thermochemical model (TCM) of TDDB:
Time-to-failure, TTF, can be calculated based on distribution of VBR:
kTHEtTF o exp
H -energy for ions displacement, - field acceleration parameter, to - time constant.
simulation
experiment
Simulation at=40.2V, =5.8H=1 eV, t0=10-3 hr
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov. 8
Infant Mortality FailuresInfant Mortality Failures
35V capacitors with characteristic VBR=95V
breakdown voltage, V
cum
ulat
ive
prob
abili
ty, %
20 2001001.E-1
5.E-1
1
5
10
50
90
100
1.E-1
b=3
b=5b=10
b=20
b=40
Life Test Simulation 35V capacitors at 85C 52.5V
time, hrcu
mul
ativ
e pr
obab
ility
, %1.E-1 1.E+91 1.E+1 1.E+2 1.E+3 1.E+4 1.E+5 1.E+6 1.E+7 1.E+8
1.E-1
5.E-1
1
5
10
50
90
100
1.E-1
1.1 (40)
0.52 (20)
0.25 (10)
0.11 (5)
0.05 (3)
Experimental distributions of TTF for chip Ta capacitors in most cases can be described as 2-parameter Weibull function with < 1
Majority of failures can be considered as infant mortality (IM) failures. Monte-Carlo simulation of VBR
distributions at different Times to failure calculated based
on TDDB model
At the level of VBR that is typically observed for tantalum capacitors (5 to 20), simulations show that all failures, up to the end of life, are defect related IM failures.
At the level of VBR that is typically observed for tantalum capacitors (5 to 20), simulations show that all failures, up to the end of life, are defect related IM failures.
Weibull Grading TestWeibull Grading Test
9Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov. 10
Weibull Grading TestWeibull Grading Test WGT is a combination of two tests: burn-in (screening to remove IM
failures) and reliability qualification (to assess the failure rate, FR). If < 1 and can be monitored during testing, then the test can be
stopped when the necessary is achieved.
F(t)
t
63%
WGTnormalcond.
_WGT _norm
Original notion [Didinger’64]: “to pay for only as much grading as needed, without the risk of getting too little or overpaying for too much reliability”.
1
)(
tt
ttF exp1)(
WGTnorm AF
WGTnormWGTWGT tAFAFt )(WGTnorm
If all failures are due to IM, then any lot can be screened out to as high FR level as necessary.
If all failures are due to IM, then any lot can be screened out to as high FR level as necessary.
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov. 11
WGT ConditionWGT Condition WGT condition per MIL-PRF-55365:
oAt 85 oC for 40 hr and voltage from 1.1VR to 1.53VR.oA failure is defined as a blown 1A or 2A fuse.oFailures calculated at 2 hr and 40hr of testing.oFR is calculated using an empirical equation obtained ~ 40 years ego:
Considering that at V = VR accelerating factor, AF, is equal to 1:
B is the voltage acceleration constant (B = 18.77249321???)
Life test at 85 oC can confirm ~1%/1000hrs which is more than 100 times greater than is required for T-grade capacitors.
)exp( uBAF 1VRVu
RVVAF 77249321.18exp1003412025.7)(Factor on Accelerati 9
Long-term reliability is determined by a 40-hr accelerated testing! Long-term reliability is determined by a 40-hr accelerated testing!
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov. 12
WGT ProblemsWGT Problems Two groups of problems for FR assessment
using WGT: testing and calculations. Problems with test conditions:
o Verification of contacts in fixtures.o Only 300 samples are used for calculations.o Test results with different fuses might be different.o Short current spikes below ~10A (scintillations) will remain unnoticed.o Only two points to draw the F(t) line and calculate and . Problems with calculations:
o Early failures (before 15 min) are neglected.o Calculations do not account for the uncertainty in times to failure.o Due to insufficient sample size, small changes in the number of failures
might result in significant variations of FR.o If no failures observed, wear-out degradation (e.g. due to field-induced
crystallization or Vo++ migration) can be missed.
The most significant errors, up to 3 orders of magnitude, are due to incorrect voltage acceleration factors.
The most significant errors, up to 3 orders of magnitude, are due to incorrect voltage acceleration factors.
F(t)
t
63%
WGT
normalcond.
_WGT _norm
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov. 13
Historical Data on Voltage Acceleration Factor (AF)
Historical Data on Voltage Acceleration Factor (AF)
Experimental data from Navy Crane report, 1982, for various hermetic solid tantalum capacitors and data reported by KEMET in 1972 and 2006
1.E+00
1.E+01
1.E+02
1.E+03
1.E+04
1.E+05
1 1.1 1.2 1.3 1.4 1.5
acce
lera
tion
fact
or
V/VR
MIL-C-39003 capacitors
MIL-spec, B=18.77NC min, B=10.3NC max, B=24.4KEMET-1972, B=14.7Paulsen'06, B=8
1.E+00
1.E+01
1.E+02
1.E+03
1.E+04
1.E+05
1 1.1 1.2 1.3 1.4 1.5
acce
lera
tion
fact
or
V/VR
MIL-C-39003 capacitors
MIL-spec, B=18.77NC min, B=10.3NC max, B=24.4KEMET-1972, B=14.7Paulsen'06, B=8
The data confirm an increase of AF errors with V and wide range of variations of B, from 8 to 24.4.Note: Paulsen’s data for chip capacitors were recalculated from power [AF=(V/VR)n] to exponential [AF=exp(Bu) ] dependence on voltage.
The range of B is close to predictions of the TDDB model. The range of B is close to predictions of the TDDB model.
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov. 14
Manufacturers’ Reliability AFManufacturers’ Reliability AF
0.0001
0.001
0.01
0.1
1
10
100
1000
0 0.2 0.4 0.6 0.8 1 1.2 1.4
acce
lera
tion
fact
or
V/VR
MIL-PRF, B=18.77AVX, B=8.1KEMET, B=14.7NEC/TOKIN, B~7.2Hitach, B~6.8Vishay*, B=18.3Panasonic, B=3.4
0.0001
0.001
0.01
0.1
1
10
100
1000
0 0.2 0.4 0.6 0.8 1 1.2 1.4
acce
lera
tion
fact
or
V/VR
MIL-PRF, B=18.77AVX, B=8.1KEMET, B=14.7NEC/TOKIN, B~7.2Hitach, B~6.8Vishay*, B=18.3Panasonic, B=3.4
AF calculated at 85 oC using vendors’ and MIL-spec data
Manufacturer Ea, eVAVX 0.63
KEMET 1.17MIL-PRF-55365 2.1
HITACHI 0.62NEC/TOKIN 0.66
Vishay*/MIL-HDBK-217F 0.15Panasonic 0.16
Activation energy, Ea, calculated based on vendors’ and MIL-spec data
Manufacturers’ data on voltage and temperature acceleration factors vary substantially.
Inconsistency in the data require more analysis.
Manufacturers’ data on voltage and temperature acceleration factors vary substantially.
Inconsistency in the data require more analysis.
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov. 15
Highly Accelerated Life Testing (HALT) at Different Test Conditions
Highly Accelerated Life Testing (HALT) at Different Test Conditions
22uF 50V capacitors at different BI conditions
time, hr
cum
ulat
ive
prob
abili
ty, %
1.E-3 10001.E-2 1.E-1 1 10 1001
5
10
50
90
99
125C 80V
130C 60V
22C 100V
22C 87.5V
Ro nV
VkTH 1expModel prediction for characteristic life:
4.7uF 50V capacitors at different BI conditions
time, hr
cum
ulat
ive
prob
abili
ty, %
1.E-2 10001.E-1 1 10 1001
5
10
50
90
99
112.5V 22C 75V 125C87.5V 85C
100V 22C
87.5V 22C
75V 22C
Results of HALT for 22 F, 50 V and 4.7 F, 50 V capacitors at different voltages and temperatures
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov. 16
Statistical Model of FailuresStatistical Model of Failures
22110exp XX
00 ln kH
1 TX 1
1 RVnk
H
2
TVX 2
The characteristic life can be presented as a general log-linear relationship
PART F S 0 1 2 t0, hr H, eV n B_85C
4.7uF 50V 90 117 0.24 -15.06 12761.6 -68.61 2.9E-07 1.1 3.7 9.6
22uF 50V 40 41 0.40 -28.13 21026.1 -124.49 6.1E-13 1.8 3.4 17.4
22uF 63V 59 43 0.24 -28.45 25072.2 -145.03 4.4E-13 2.1 3.5 20.6
Based on TDDB model, the following conversion was used:
Results for 7 part types showed that the acceleration constant B varies from 10 to 20.
Results for 7 part types showed that the acceleration constant B varies from 10 to 20.
WGT SummaryWGT Summary
17Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
Errors in FR calculations related to the testing conditions sum up to an order of magnitude.
Errors related to incorrect AFs can change FR up to 103 times.Reliance on MIL-PRF-55365 method for reliability rating might
be misleading.TDDB model predicts an exponential dependence of AF on
voltage with the voltage acceleration constant B depending on temperature.
Temperature dependence of AF can be presented in the Arrhenius-like form with Ea varying with voltage.
AF can be obtained using HALT and approximation of test results with a general log-linear relationship at two levels of stress factors.
Surge Current TestingSurge Current Testing
18Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
Mechanisms of Surge Current FailuresMechanisms of Surge Current Failures
19
Sustained scintillation breakdown.If current is not limited, self-healing does not have time to develop.
Electrical oscillations in circuits with high inductance.
Local overheating of the cathode. Mechanical damage to tantalum
pentoxide dielectric caused by the impact of MnO2 crystals.
Stress-induced-generation of electron traps caused by electromagnetic forces developed by high currents.
In all models the rate of voltage increase is critical. In all models the rate of voltage increase is critical.
TaMnO2
Ignition due to exothermic reaction in tantalum capacitors. Prymak 2006.
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
Effect of dV/dt on Breakdown VoltageEffect of dV/dt on Breakdown Voltage
20
Scintillation breakdown voltage, VBRscint (dV/dt ~ 1 to 5 V/sec) is always greater than the surge current breakdown voltage, VBRsurge(dV/dt ~ 105 to 106 V/sec)
The rate of voltage increase changescharges and electrical field at the interface.
Accumulation of electrons on traps at the MnO2-Ta2O5 interface with time increases the barrier, the level of electron injection, and the probability of avalanching.
The rate of voltage increase is critical for breakdown. A resistor in series with a capacitor reduces dV/dt and failures. The rate of voltage increase is critical for breakdown. A resistor in series with a capacitor reduces dV/dt and failures.
Fast voltage rise Slow voltage riseFast voltage rise Slow voltage rise
TaMnO2
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
40
60
80
100
120
140
160
40 60 80 100 120 140 160
VBR
_3SC
T, V
VBR_scint, V
Effect of dV/dt on VBR for 50V capacitors
40
60
80
100
120
140
160
40 60 80 100 120 140 160
VBR
_3SC
T, V
VBR_scint, V
Effect of dV/dt on VBR for 50V capacitors
Surge Current Derating RequirementsSurge Current Derating Requirements
21
History of requirements for circuit resistance (Rac): In the 1960s: 3 Ω per each volt of operating voltage. By the 1980s: 1 Ω per each volt. From the 1990s: 0.1 Ω per volt or 1 ohm, whichever is greater. Manufacturers consider surge current failures as the major
reason for voltage derating. Do we need derating of currents in addition to voltage? The limit for acceptable surge currents is set by the SCT
conditions: the current during applications should not exceed the current during testing: Iappl.< Itest
Improvements in reliability and the need to increase the efficiency of power supply systems resulted in reduction of Rac.
At what conditions we can allow circuit designs without Rac? Need a closer look at how the Itest is specified.
Improvements in reliability and the need to increase the efficiency of power supply systems resulted in reduction of Rac.
At what conditions we can allow circuit designs without Rac? Need a closer look at how the Itest is specified.
“use as tested”“use as tested”
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
MIL-PRF 55365 SCT RequirementsMIL-PRF 55365 SCT Requirements
22
SCT per MIL-PRF-55365H: Number of cycles: Nc = 4 surge cycles. Energy storage capacitor: CB = 20×CDUT Test voltage: VR Charge time, tch, and discharge time, tdisch, ≥ 1sec. Total DC resistance of the test circuit, Rtc, including the wiring, fixturing, and
output impedance of the power supply should not exceed 1 Ω. Measurements after SCT: DCL, C, DF (still no requirements for ESR) The minimum surge peak current shall be: Itest ≥ VR/(Rtc + ESRspec), Rtc = 1 Ω. Failure condition: current = 1A after 1ms for C ≤ 330uF; 10ms for C ≤ 3.3mF,
and 100ms for C > 3.3mF. Before 12/1/2012: Nc = 10. tdisch = tch = 4 sec. Rc ≤ 1.2 Ohms. CB ≥ 50 mF. No Itest requirements.
New specifications recognize the role of ESR as a limiting factor for surge currents. However, no specifics for Itest verification.M55365 does not guarantee reliable
operation at VR: Vtest = VC_MAX = 0.95×VR, => the need for voltage derating.
New specifications recognize the role of ESR as a limiting factor for surge currents. However, no specifics for Itest verification.M55365 does not guarantee reliable
operation at VR: Vtest = VC_MAX = 0.95×VR, => the need for voltage derating.
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
Application vs. Testing ConditionsApplication vs. Testing Conditions
23
Typical application conditions: No limiting resistors, minimal inductance. Minimal contact resistance.
Resistance of the circuit, Rac, is minimal and the current is limited mostly by ESR.
Surge Current Test conditions: Unspecified contact resistance of fixtures. Limiting resistor can be up to 1 Ω. Relatively long wires and inductance. No clear requirements for Itest verification (when, how?).
Parts with poor contacts in the fixture can pass the testing.
Application conditions might be more severe than test conditions.There is a need in tightening test requirements.Application conditions might be more severe than test conditions.There is a need in tightening test requirements.
Example:• A 15F 10V CWR06 capacitor with specified ESR=2.5Ω and real ESR = 0.5Ω is used in a 5V line.• During application the part can experience a spike:
Iappl = 5/0.5 = 10 A.• During the testing it will be verified to the current:
Itest ≥ VR/(Rtc + ESRspec)Itest =10/(1+2.5) = 2.8 A
Example:• A 15F 10V CWR06 capacitor with specified ESR=2.5Ω and real ESR = 0.5Ω is used in a 5V line.• During application the part can experience a spike:
Iappl = 5/0.5 = 10 A.• During the testing it will be verified to the current:
Itest ≥ VR/(Rtc + ESRspec)Itest =10/(1+2.5) = 2.8 A
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
Specifics of Surge Current TestingSpecifics of Surge Current Testing
24
Theory and experiments show that the rate of the voltage increase is critical for SCT.
High current spike is a byproduct of the fast voltage rise, rather than the major cause of failure.
Even minor variations (~0.1 Ω) of Rc affect results of SCT.
Measurements of voltage level across the capacitor during testing that are used to assure that the part passed SCT are not informative without indication of timing.
The amplitude of the current spike is the most adequate characteristic to verify SCT conditions.The amplitude of the current spike is the most adequate
characteristic to verify SCT conditions.
0
0.2
0.4
0.6
0.8
1
0
1
2
3
4
5
0 100 200 300 400
V/Vo
I/Vo,
A/V
time, uS
I_0.1 OhmI_0.4 OhmI_0.7 OhmV_0.1 OhmV_0.4 OhmV_0.7 Ohm
0
0.2
0.4
0.6
0.8
1
0
1
2
3
4
5
0 100 200 300 400
V/Vo
I/Vo,
A/V
time, uS
I_0.1 OhmI_0.4 OhmI_0.7 OhmV_0.1 OhmV_0.4 OhmV_0.7 Ohm
Calculations for 220uF capacitor at Lc= 400 nH, ESR=0.1Ω Rc-var
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
Factors Affecting SCT ResultsFactors Affecting SCT Results
25
Rac includes resistance of wires and contacts.
The length of wires affects inductance. Type of switch. The rate of voltage increase in
case of using field effect transistor (FET).
To simulate worst case conditions Isp should be maximized. To simulate worst case conditions Isp should be maximized.
-30
0
30
60
90
120
150
0 10 20 30 40 50
curr
ent,
A
time, us
4" wires
12" wires
24" wires
-30
0
30
60
90
120
150
0 10 20 30 40 50
curr
ent,
A
time, us
4" wires
12" wires
24" wires
-30
0
30
60
90
120
150
0 10 20 30 40 50
curr
ent,
A
time, us
110 nH
400 nH
900 nH
-30
0
30
60
90
120
150
0 10 20 30 40 50
curr
ent,
A
time, us
110 nH
400 nH
900 nH
47uF R=0.25Ohm L-var. Simulation.
‐30
0
30
60
90
120
150
0 10 20 30 40 50
curr
ent,
A
time, us
0 kOhm
0.8 kOhm
1.6 kOhm
‐30
0
30
60
90
120
150
0 10 20 30 40 50
curr
ent,
A
time, us
0 kOhm
0.8 kOhm
1.6 kOhm
0
2
4
6
8
10
12
-15 0 15 30 45 60 75 90
gate
vol
tage
, V
time, us
0 kOhm1.6K Ohm.8 k ohm
0
2
4
6
8
10
12
-15 0 15 30 45 60 75 90
gate
vol
tage
, V
time, us
0 kOhm1.6K Ohm.8 k ohm
47uF 20Vexperiment
Effect of resistors at FET gate on Vg and Isp for 47uF 20V capacitors
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
Effective Resistance of the CircuitEffective Resistance of the Circuit
26
Isp increases linearly with voltage allowing for calculations of the effective resistance of the circuit, Reff,.
Reff corresponds to the impedance of the circuit and includes Rac, ESR, resistance of contacts, and circuit inductance, Lc.
Isp increases linearly with voltage allowing for calculations of the effective resistance of the circuit, Reff,.
Reff corresponds to the impedance of the circuit and includes Rac, ESR, resistance of contacts, and circuit inductance, Lc.
47uF 10V
10uF 16V
220uF 6V
0
40
80
120
160
200
0 20 40 60 80
curr
ent s
pike
, A
voltage, V
47uF 10V
10uF 16V
220uF 6V
0
40
80
120
160
200
0 20 40 60 80
curr
ent s
pike
, A
voltage, V
0
40
80
120
160
200
0 15 30 45 60 75 90
curr
ent,
A
time, us
6V 8V10V 12V14V 16V18V 20V
220uF 6V
0
40
80
120
160
200
0 15 30 45 60 75 90
curr
ent,
A
time, us
6V 8V10V 12V14V 16V18V 20V
220uF 6V
‐40
0
40
80
120
160
200
240
‐10 0 10 20 30 40 50 60 70
curr
ent,
A
time, us
20V 24V28V 32V36V
47u 20V
‐40
0
40
80
120
160
200
240
‐10 0 10 20 30 40 50 60 70
curr
ent,
A
time, us
20V 24V28V 32V36V
47u 20V Test anomalies, e.g. poor contacts, can be revealed by Isp(V) curves
good contacts
poor contacts0
50
100
150
200
250
300
0 5 10 15 20 25 30 35
curr
ent s
pike
, A
voltage, V
3SCT 220uF 6V
good contacts
poor contacts0
50
100
150
200
250
300
0 5 10 15 20 25 30 35
curr
ent s
pike
, A
voltage, V
3SCT 220uF 6V
after adjustment
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
Effect of Reff on Breakdown VoltageEffect of Reff on Breakdown Voltage
27
Different lot date codes of 22 F 35V capacitors.
Effect of temperature Parts with larger ESR (Reff) had greater VBR.
Temperature decreases ESR resulting in lower VBR.
An increase in Reff by ~0.1Ωresults in increase of VBR ~10%.
Parts with larger ESR (Reff) had greater VBR.
Temperature decreases ESR resulting in lower VBR.
An increase in Reff by ~0.1Ωresults in increase of VBR ~10%.
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24
0.2 0.22 0.24 0.26 0.28 0.3 0.32 0.34
ESR
, Ohm
Reff, Ohm
TBJD226K035LRLB0024
DC0513 DC0516
DC0824 DC0836
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24
0.2 0.22 0.24 0.26 0.28 0.3 0.32 0.34
ESR
, Ohm
Reff, Ohm
TBJD226K035LRLB0024
DC0513 DC0516
DC0824 DC0836
30
40
50
60
70
80
90
0.2 0.22 0.24 0.26 0.28 0.3 0.32 0.34
VBR
_SC
T, V
Reff, Ohm
TBJD226K035LRLB0024
DC0513 DC0516
DC0824 DC0836
30
40
50
60
70
80
90
0.2 0.22 0.24 0.26 0.28 0.3 0.32 0.34
VBR
_SC
T, V
Reff, Ohm
TBJD226K035LRLB0024
DC0513 DC0516
DC0824 DC0836
0
0.1
0.2
0.3
0.4
0
20
40
60
80
-60 -40 -20 0 20 40 60 80 100
Reff,
Ohm
VBR_
SCT,
V
temperature, deg.C
47uF 20V VBR220uF 6V VBR47uF 20V Reff220uF 6V Reff
0
0.1
0.2
0.3
0.4
0
20
40
60
80
-60 -40 -20 0 20 40 60 80 100
Reff,
Ohm
VBR_
SCT,
V
temperature, deg.C
47uF 20V VBR220uF 6V VBR47uF 20V Reff220uF 6V Reff
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
Correlation between ESR and ReffCorrelation between ESR and Reff
28
In an optimized set-up the difference between Reff and ESR is 0.1 - 0.2 Ω.
In an optimized set-up the difference between Reff and ESR is 0.1 - 0.2 Ω.
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Ref
f, O
hm
ESR, Ohm
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Ref
f, O
hm
ESR, Ohm
Effective resistancde during SCT and ESR
R, Ohmcu
mul
ativ
e pr
obab
ility
, %0.10 0.800.24 0.38 0.52 0.661
5
10
50
99ESR\TM8A106K016CBZNormal-2PRRX SRM MED FMF=16/S=0
Data PointsProbability Line
ESR\TM8T476K010CBZNormal-2PRRX SRM MED FMF=16/S=0
Data PointsProbability Line
Reff\TM8A106K016CBZNormal-2PRRX SRM MED FMF=16/S=0
Data PointsProbability Line
Reff\TM8T476K010CBZNormal-2PRRX SRM MED FMF=15/S=0
Data PointsProbability Line
20 lots of HV capacitors microchips
y = 0.6118x + 0.1963
0.3
0.35
0.4
0.45
0.2 0.25 0.3 0.35 0.4
Ref
f, O
hm
ESR, Ohm
47uF 10V uchips
y = 0.6118x + 0.1963
0.3
0.35
0.4
0.45
0.2 0.25 0.3 0.35 0.4
Ref
f, O
hm
ESR, Ohm
47uF 10V uchips
ESRRIVRR ac
speff
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.
SCT SummarySCT Summary
29
Results of SCT depend on the rate of voltage increase. To assure proper SCT conditions, current spike
amplitudes, Isp, should be measured for each part, and conditions VR/Isp = Reff < 0.5 + ESR should be verified.
Tantalum capacitors manufactured per M55365 might fail at first power turn-on due to non-adequate SCT conditions.
Additional testing and analysis (to compare levels of current spikes during testing and application) are necessary to use tantalum capacitors without limiting resistors.
Presented by Alexander Teverovsky at an iNEMI (International Electronics Manufacturing Initiative) WebEx, June 2013 and deliverable to NASA Electronic Parts and Packaging (NEPP) Program to be published on nepp.nasa.gov.