You are on page 1of 68

• DECEMBER 2019

Evolution of Electrolytic
Capacitors- Why a Reliability
Engineer Should Know This

ASQ Tech Talk


May 13, 2021
Abstract
• Electrolytic capacitors have long been identified as a weak link for long term high
reliability applications. However, capacitor manufacturers have made significant
improvements to the materials and manufacturing processes to enhance their
reliability. This webinar will discuss those changes, provide insight into the various
failure mechanisms for electrolytic capacitors and describe appropriate accelerated
tests to validate performance.

• We will take a deeper dive into the methodologies utilized to improve capacitor
performance, e.g. foil purity and electrolyte volume. We will also discuss, from a
reliability perspective, the impact of changing to a higher temperature electrolyte
(from ethylene glycol to DMF, DMA and GBL) and also changes in the bung material
(from butyl to EPDM).

• There are several environmental factors involved in the aging of electrolytic


capacitors. Electrolyte loss due to drying out and leakage current due to oxide
degradation are thermally related as is the self-heating associated with ripple
current. The impact of the applied voltage level is also a driver as it can cause
leakage current increases as well. All of these issues result in a capacitance
decrease, an increase in ESR, and a change to the dissipation factor. Many other
failure mechanisms associated with manufacturing will also be discussed.

• Examples of calculations for life expectancy will be shown to demonstrate the effects
of applied voltage, rated temperature, ripple current, and the endurance factor
coupled with the application usage profile.

2
CHANGES IN E-CAPS
CHANGES IN E-CAPS TO FIX PAST RELIABILITY ISSUES

• Improved quality control (foil purity, electrolyte volume, etc.)

• Migration to higher temperature electrolyte (from ethylene glycol to DMF, DMA,


and GBL)

• Changes in bung material (from butyl to (ethylene propylene diene monomer


rubber (EPDM). These chemicals have given this rubber compound a greater
resistance to ultraviolet rays, some chemicals and it is able to operate in
moderate heat.

• Some issues still exist


- Private-labeling of lower cost capacitors
- Empirically-based life calculations
- Balance between high capacitance (thin dielectric) and high voltage (thick dielectric)

4
LIST OF FAILURE MECHANISMS
LIST OF FAILURE MECHANISMS
• Major Factors for Aging of Electrolytic Capacitors
- Temperature
• Electrolyte Loss/Dry Out
• Leakage current >> Oxide degradation
- Ripple Current
• Self heating >> Electrolyte Loss/Dry Out
- Voltage level
• Leakage current >> Oxide degradation

• These Effects Result In;


- Capacitance Decrease
- Increase in ESR
- Dissipation Factor change
Electrolytic Capacitor Failure Modes
• Short Circuit
- Between Electrodes
- Crack in Separator All result in an increase in internal pressure
- Crack of Oxide Layer
• Open Circuit
- Disconnection to Terminals
- Bad Terminal Contact
- Electrolyte Dry Out
• Capacitance Drop/ESR Increase/DF change
- Electrolyte Vaporization – results in an increase in internal pressure
- Oxide Layer Degradation
- Electrolyte Dry Out
Root Causes During Production
• Impurities of metal particles
• Contamination by chloride
• Deficiency in oxide layer
• Burred foil
• Poor sealing – can, lead or crimp issue
• Poor connections

• Can lead To:


- Short circuit
- Leakage current Increase

8
Root Causes within Application or by Aging
• Short and Open Circuit
- Mechanical stress
• Open Vent/Electrolyte Leakage
- Thermal Stress
- Wrong Polarity
- High ripple Current
- Overvoltage
- Rapid charge/Discharge

• Leads to Capacitance Drop, ESR Increase, DF Change

9
LIST OF FAILURE MODELS
ELECTROLYTIC CAPACITORS (cont.)
oEvaporation prediction has been based on a widely held standard
aging relationship

oDoubling of lifetime with every 10C drop in temp (note: This is not
Arrhenius!)
L x = Lo x 2 (To-Tx) /10
oHowever, there are variations from manufacturer to manufacturer

11
CAPACITOR LIFE PREDICTION (NICHICON)
o Lr is rated lifetime
o Tr is rated temperature Miniature w/o ripple
o T is ambient
temperature
o Ir is rated ripple current
o I is actual ripple current
Miniature w/ ripple
o Dtr is the temperature
rise due to rated ripple
current
o Dt is the temperature
rise due to actual ripple Large can
current
o a and K are coefficients
CAPACITOR LIFE PREDICTION (NIPPON CHEMI-CON)

o Lr is rated lifetime
o Tr is rated temperature Miniature w/o ripple
o T is ambient temperature
o Ir is rated ripple current
o I is actual ripple current Miniature w/ ripple
o Dtr is the temperature rise
due to rated ripple current
o Dt is the temperature rise
Large can
due to actual ripple
current
o A and Kn are coefficients
CAPACITOR LIFE PREDICTION (RUBYCON)
o Lr is rated lifetime
o Tr is rated temperature
o T is ambient temperature Miniature
o Ir is rated ripple current
o I is actual ripple current
o Dtr is the temperature rise
due to rated ripple current Large can

o Dt is the temperature rise


due to actual ripple current
o Vr is rated voltage
o V is actual voltage
CAPACITOR LIFE PREDICTION (PANASONIC/SANYO)

o Lr is rated lifetime
o Tr is rated temperature
All
o Ta is core temperature
E-CAP RELIABILITY PREDICTION -
WHAT IS THE RIGHT APPROACH?
FOUR STEP PROCESS
• Parameters

• Acceleration Transform

• Use Environment

• Statistics

17
WHICH PARAMETERS
(TEMPERATURE/LIFETIME)?
• Supplier Ratings

• Extrapolate based on weight loss

• Extrapolate based on parametric change (capacitance / ESR)

• Test to Failure

18
WHICH PARAMETERS
(TEMPERATURE/LIFETIME)?
• Supplier Ratings

• Extrapolate based on weight loss

• Extrapolate based on parametric change (capacitance / ESR)

• Test to Failure

19
WHICH PARAMETERS
(TEMPERATURE/LIFETIME)?
• Extrapolate based on weight loss

20
WHICH PARAMETERS
(TEMPERATURE/LIFETIME)?

• Extrapolate based on
parametric change
(capacitance / ESR)

21
WHICH PARAMETERS
(TEMPERATURE/LIFETIME)?
• Test to Failure
0.00%
Capacitance, µF

-5.00%
% Change in Capacitance

-10.00%

-15.00%

-20.00%

-25.00%
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Time (Hours)

22
WHICH ACCELERATION TRANSFORM?
• Supplier Formula

• Simplified Formula

• Reliability Physics ???

23
RELIABILITY PHYSICS
oReliability Physics approach derived from
- Electrolyte vapor pressure
- Volume of electrolyte
- Critical volume
- Evaporation/leak rate
- Embrittlement rate of rubber/plastic stopper

oUnfortunately, most capacitor manufacturers consider this


information ‘proprietary’
RELIABILITY PHYSICS

Determine
ESR

Calculate
Calculate
Loss of
Core Temp
Electrolyte

Calculate Gasperi
IEEE IAS
Vapor 1997
Pressure

25
RELIABILITY PHYSICS

Determine
ESR

Calculate
Calculate
Loss of
Core Temp
Electrolyte

Calculate Gasperi
IEEE IAS
Vapor 1997
Pressure

26
RELIABILITY PHYSICS

Determine
ESR

Calculate
Calculate
Loss of
Core Temp
Electrolyte

Calculate Gasperi
IEEE IAS
Vapor 1997
Pressure

27
RELIABILITY PHYSICS

Determine
ESR

Calculate
Calculate
Loss of
Core Temp
Electrolyte

Calculate Gasperi
IEEE IAS
Vapor 1997
Pressure

28
RELIABILITY PHYSICS

Determine
ESR

Calculate
Calculate
Loss of
Core Temp
Electrolyte

Calculate Gasperi
IEEE IAS
Vapor 1997
Pressure

29
RELIABILITY PHYSICS – STEP 1

oDetermine ESR at test/application temperature


- Typically ESRT/ESR25C = A x (B – (C x TANH((T-T0)/D)))
- Empirically determined by Ansys
• A=6
• B = 1.8
100
• C = 1.7
ESR1
• T0 = 240
ESR2
• D = 25 ESRT / ESR25C 10 ESR3
TANH

0
220 270 320 370
Temperature (K)

30
RELIABILITY PHYSICS – STEP 2/3

• Calculate Core Temperature


×
- (H is heat transfer per surface area)
×( )
- Add to ambient temperature to determine core temperature

• Calculate Vapor Pressure [ ]


(Antoine Equation, mmHg)
- Ethylene Gylcol (EG) / 99%H2O:
A = 9.19 / B = 3103 / C = 309.7
- Dimethyl Formamide (DMF) and g-Butyrolactone (GBL)

31
RELIABILITY PHYSICS – STEP 4
oCalculate Loss of Electrolyte
- Vt0+Dt = Vt0 – (k x P x Dt)
- k is leak rate based on vapor pressure (empirically determined, ml/mmHg/hr)
- Dt is the time step

oESR is dependent on electrolyte volume


- ESRt0+Dt/ESRt0 = (Vt0+Dt/Vt0)2

32
WHICH ENVIRONMENT?

• Constant Temperature

• Temperature Variation

33
WHICH ENVIRONMENT?

• Constant Temperature
65C

• Temperature Variation

34
WHICH ENVIRONMENT?

• Constant Temperature

• Temperature Variation
o o
Month Cycles/Year Ramp Dwell Max. Temp ( C) Min. Temp. ( C)
Jan.+Feb.+Dec. 90 6 hrs 6 hrs 20 5
March+November 60 6 hrs 6 hrs 25 10
April+October 60 6 hrs 6 hrs 30 15
May+September 60 6 hrs 6 hrs 35 20
June+July+August 90 6 hrs 6 hrs 40 25

35
WHICH ENVIRONMENT?

• Constant Temperature

• Temperature Variation
90
On, stopped 80 On, driving
state 70
state
Temperature (˚C)

60
50
40
30
20
10
Off 0
state 0 5 10 15 20 25 30 35
Time (Minutes)

36
WHICH STATISTICS
• Require Margin

• Assume Weibull

• Perform Monte Carlo Simulation

37
Weibull Plot
100.0%

90.0%

80.0%

70.0%

60.0%
% Failure

50.0%
Beta = 10
40.0% Beta = 3.5
30.0%

20.0%

10.0%

0.0%
0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000
Time (hrs)

• Based on previous testing, Beta values for electrolytic


capacitor failures can range from 3.5 to over 10

38
E-CAP LIFE PREDICTION - SUMMARY
• Industry tends to be too reliant on supplier ratings
- Failure definition can vary
- Lifetime can be with or without ripple current
- Probability of failure after lifetime is never defined (test to 0 failures)
- Different approaches to extend lifetime
• Large volume of electrolyte
• Higher boiling point electrolyte / lower vapor pressure
• Better seal
• Ability to operate at lower electrolyte volumes
- Degradation of seal/bung not addressed

• Best approach is dependent on available resources and level of risk

39
CALCULATING TIME TO FAILURE
Calculating Life Expectancy
- The life of an aluminum electrolytic is determined by
• Lr = rated lifetime of the capacitor at rated ripple current
• Tr = rated temperature of the capacitor
• TE = actual temperature of the electrolyte
 Tr TE 
 
L  Lr  2  10 

• This formula is based on the evaporation of a liquid

• Note: Panasonic and Sanyo use this formula

41
Calculating Life Expectancy
• Field conditions
• ~10C rise due to ripple current
• Peak temperature of 85C in worst case scenario
• Assume 65C for 12 hours each day
• Check also at 55C
• Voltage applied is 80% of rated.
• Sample size of 28 pieces for a test

42
Capacitor LifeTime Calculations (Nichicon)

43
Calculating Life Expectancy
• Case Study – Nichicon PW series electrolytic capacitors
• The parameters remain the same with the only change
being the endurance testing done at the supplier. It
has a significant impact on life expectancy. Potential
annual usage factored in.

44
Calculating Life Expectancy
• Lowering the operating temperature by 10C and using more typical annual
usage

45
Calculating Life Expectancy
• Case Study – Su’scon aluminum electrolytic capacitors
• Have lower endurance ratings so their life expectancy is considerably
lower. Also using more typical annual usage.

46
Calculating Life Expectancy
• Lowering the operating temperature 10C

47
Calculating Life Expectancy
• Controlling the parameters
• The Ripple current in the capacitor is derived by:
- 𝑐𝑎𝑝= 𝑝𝑣 𝑝𝑣 √2*n
• Where Icap is the RMS capacitor ripple current in A, Ppv is the PV Power in W, and Vpv the PV
voltage in V and n is the number of capacitors in parallel.
• Note that this equation is valid because there is no significant power
conversion induced high frequency ripple current in the electrolytic
capacitors.

48
RIPPLE CURRENT AND TEMPERATURE RISE
• Two ways to calculate thermal conductance (and
therefore temperature rise due to ripple)

• One, per EIA-ECA-797, is to assume applying the rated


ripple current results in a 5C temperature rise
- UCC: Power = I2R = (3.8)2 x 0.014 = 0.202 W
- Nichicon: Power = I2R = (3.4)2 x 0.02 = 0.231 W

• Thermal conductance can therefore be determined


- UCC = 5 / 0.202 = 24.7 °C/W
- Nichicon = 5 / 0.231 = 21.6 °C/W

49
RIPPLE CURRENT AND TEMPERATURE RISE (cont.)
• The second way, also per EIA-ECA-797, is to assume a convection
coefficient of 0.001 W/°C/cm2

• Thermal conductance is the product of the convection coefficient and the


surface area
- Length = 40mm
- Width = 18mm
- Area = 2pr2 + 2prh = 509 + 2260 = 2769 mm2 = 27.7 cm2

• Thermal conductance is therefore 36.1 °C/W

50
CALCULATED TEMPERATURE RISE

Capacitor UCC Nichicon

Applied Power (I2R) 0.35 W 0.5 W

Thermal Conductance (Approach #1) 24.7 °C/W 21.6 °C/W

Temperature Rise (Approach #1) 8.6 °C 10.8 °C

Thermal Conductance (Approach #2) 36.1 °C/W 36.1 °C/W

Temperature Rise (Approach #2) 12.6 °C 18.0 °C

51
CALCULATED TEMPERATURE RISE (Caveat)
• The analytical equations being used to calculate temperature rise assume
heat loss primarily through convection

• However, with some designs, heat loss could be primarily through


conduction

• Actual heat rise will likely need to be through more sophisticated


modeling or measurement

52
LIST OF ACCELERATED TESTS
Novel Weight Loss Test Approach
Life Test – Traditional (& Accelerated: Rate of Weight Loss)

Apply rated Oven and Turn off


ripple and bias ripple current electricity and
e- e-
at rated heating causes cool to room
temperature. electrolyte temperature.
e- evaporation. e-
PCB PCB PCB

No. Continue
testing. Weigh.

Weight: 1.432 g
Did ESR have a
>200%
Yes. Testing Measure ESR Electrical
increase from
is complete. characterization
the initial
.
value?
Life Test– Accelerated: Critical Weight Loss

Puncture a Oven heating Turn off heat


hole in the causes and cool to
capacitor. electrolyte room
evaporation. temperature.

No. Continue
testing. Weigh.

Weight: 1.432 g
Did ESR have a
>200%
Yes. Testing Measure ESR Electrical
increase from
is complete. characterization
the initial
.
value?
Life Test – Accelerated Wear-Out Failure
Mode

Dried out, off Grey paper


white paper indicative of
indicative of electrolyte
electrolyte saturation.
Accelerated test evaporation. As-received
Background: Traditional Electrolytic Capacitor
Testing
• Tests capacitors for thousands or tens of thousands of hours until failure occurs
at desired test conditions
• Weight loss, ESR, capacitance, and DC leakage are periodically monitored to
assess capacitor functionality
• Failure often defined in datasheets as:
- 200% increase in ESR
- 20% decrease in capacitance
- DC leakage current above threshold specified in data sheet
• Experience has indicated that until failure the data shows a linear rate of weight
loss
- Physical failure typically involves catastrophic eruption of electrolyte from the capacitor, which
results in significant weight loss
• % increase in ESR shows an exponential increase over time
- Since weight loss is proportional to time, % increase in ESR also has an exponential
relationship to weight loss
Background: Accelerated Electrolytic Capacitor Life
Testing
• Once the % increase in ESR compared to weight loss has been
characterized, and the rate of weight loss at a given condition is known, a
calculation can be made to determine the time to a given % increase in
ESR
- This allows evaluation of the lifetime of caps in a shorter time period than standard
life test that runs for entire duration of capacitor life
• Critical Weight Loss Test
- Determines the relationship of % increase in ESR to weight loss
- Varies between different capacitor types and manufacturers
• Rate of Weight Loss Test
- Capacitors are tested under desired conditions for 3-4 weeks to determine the rate of
weight loss over time
Experimental Procedure: Critical Weight Loss
Testing
• Puncture small hole in 8 capacitors for each population
- This increases the rate of weight loss during baking, shortening the test
- Electrical characterization is performed before and after puncturing the hole to
ensure the electrical functionality of the capacitor was not damaged from the
puncture
• Depending on capacitor size, place population in an oven at 85°C (for
smaller capacitors) or 105°C (for larger capacitors) for a given amount of
time
- Bake times vary between 0.5-8 hours, depending on test data and degree of definition
desired
• Weigh capacitor and measure ESR at room temperature after baking
• Repeat baking and measuring process until data indicates at least 200%
increase in ESR from the initial value at room temperature
Experimental Procedure: Critical Weight Loss Testing Assumptions

• Failure defined as 200% increase in ESR measured at


room temperature
• Primary wear-out mechanism is electrolyte evaporation
• Each population of capacitors has a critical weight loss
value
- Once the capacitor losses the necessary critical weight it does
not have enough electrolyte to function and is deemed failed
Experimental Procedure: Rate of Weight Loss
Testing
• Measure rate of weight loss for 18-20 capacitors from each provided
population of capacitors at half-rated voltage and half-rated ripple current
- Capacitors were wired to sample holders and placed into a test coupon, with voltage
and ripple applied via a Chroma 11800 Ripple Current Tester
• Testing performed in a thermal chamber at 105 ˚C
- Samples were placed inside a cardboard box within the chamber to reduce
convection, simulating the field environment
• Capacitors weighed multiple times over 500 hours to determine rate of
weight loss for each population
• Rate of weight loss is linear until failure
• Rate is temperature dependent
- Testing performed at 105 C
Results: Analysis of Experimental Data
• Critical weight loss data analyzed
- Required weight loss for 200% increase in ESR with 95% confidence
- Rate of weight loss with 99% confidence
- Estimates provided of expected life at high and low values of critical weight loss and
rate of weight loss
Critical Weight Loss Results: United Chemi-Con 450V -
EKXG451ESS680MMN3S
Rate of Weight Loss Results: United Chemi-
Con 450V - EKXG451ESS680MMN3S
Expected Life Predictions: United Chemi-Con 450V -
EKXG451ESS680MMN3S
• The table above shows calculations for expected life in hours of capacitors
for the best and worst case of the confidence intervals for critical weight
loss and rate of weight loss
- Required weight loss for 200% increase in ESR with 95% confidence
- Rate of weight loss with 99% confidence

Predicted Test Life (hours)


Weight Loss Rate of Weight Loss (g/hour)
Threshold (g) 1.61E-05 1.79E-05
0.379 23,600 21,130
0.468 29,140 26,090
THANKS
Greg Caswell
Lead Consulting Engineer
443-834-9284 (cell)
301-640-5825
Greg.caswell@ansys.com

You might also like