Aluminum electrolytic capacitor failure modes and ripple cur

[Ethan Petersen]

I am trying to get a handle on ripple current ratings for capacitors, starting with the ChemiCon KZN series. The data sheet has specifications for rated ripple current and maximum temperature. It seems I can reliably operate well beyond the ripple current rating, but I don't have a good understanding of what the real limitations are.

How do capacitor manufacturers come up with the ripple current rating for their parts? I had thought this was mostly a function of the capacitor's internal temperature, and the effect on the electrolyte. If it gets too hot the electrolyte might boil, or in a less extreme case the electrolyte permeates through the plug faster effectively reducing the life of the capacitor.. Does this sound correct? From the little bit of testing I have been doing, it seems more complicated than that. I would like to come up with a good way to measure this or otherwise put numbers on it.

Does anyone know of a good reference for the design of capacitors? I would like to get a better understanding of the chemistry. From what little I know, the chemistry in the electrolyte is pretty sophisticated. This is a mature technology.

Thanks,

Ethan

 

[Martin Brown]

On 28/08/2019 16:29, Ethan Petersen wrote:

I am trying to get a handle on ripple current ratings for capacitors,
starting with the ChemiCon KZN series. The data sheet has
specifications for rated ripple current and maximum temperature. It
seems I can reliably operate well beyond the ripple current rating,
but I don't have a good understanding of what the real limitations
are.

I suspect that they rate them conservatively as having them fail
catastrophically in service can be pretty horrible and messy.

This is one such estimate of MTBF vs operating conditions:

http://rubycon.co.jp/en/products/alumi/pdf/Life.pdf

How do capacitor manufacturers come up with the ripple current rating
for their parts? I had thought this was mostly a function of the
capacitor's internal temperature, and the effect on the electrolyte.
If it gets too hot the electrolyte might boil, or in a less extreme
case the electrolyte permeates through the plug faster effectively
reducing the life of the capacitor. Does this sound correct? From
the little bit of testing I have been doing, it seems more
complicated than that. I would like to come up with a good way to
measure this or otherwise put numbers on it.

The hotter it runs the shorter the lifetime. +10C roughly halving it.

Does anyone know of a good reference for the design of capacitors? I
would like to get a better understanding of the chemistry. From what
little I know, the chemistry in the electrolyte is pretty
sophisticated. This is a mature technology.

Mature technology with a twist and critical hidden trade secrets in the
electrolyte formulations that will catch out the unwary.

ISTR a decade or two back someone stole the electrolyte formula from a
well known Japanese manufacturer and made cheaper clones. The resulting
capacitors failed spectacularly after about 3 years in use but not
before they were installed on a heck of a lot of PC motherboards. eg.

https://www.geek.com/blurb/capacitor-failures-plague-motherboard-vendors-551780/

I had one motherboard of mine that failed this way. Fortunately I got to
it before the thing actually blew up. It showed symptoms of unreliable
ram when the ripple started to get a bit too high. The worst failing
capacitor had already bulged enough to be at a very rakish angle.

--
Regards,
Martin Brown

 

On Wed, 28 Aug 2019 17:02:18 +0100, Martin Brown
<'''newspam'''@nezumi.demon.co.uk> wrote:

On 28/08/2019 16:29, Ethan Petersen wrote:
I am trying to get a handle on ripple current ratings for capacitors,
starting with the ChemiCon KZN series. The data sheet has
specifications for rated ripple current and maximum temperature. It
seems I can reliably operate well beyond the ripple current rating,
but I don't have a good understanding of what the real limitations
are.

I suspect that they rate them conservatively as having them fail
catastrophically in service can be pretty horrible and messy.

This is one such estimate of MTBF vs operating conditions:

http://rubycon.co.jp/en/products/alumi/pdf/Life.pdf

How do capacitor manufacturers come up with the ripple current rating
for their parts? I had thought this was mostly a function of the
capacitor's internal temperature, and the effect on the electrolyte.
If it gets too hot the electrolyte might boil, or in a less extreme
case the electrolyte permeates through the plug faster effectively
reducing the life of the capacitor. Does this sound correct? From
the little bit of testing I have been doing, it seems more
complicated than that. I would like to come up with a good way to
measure this or otherwise put numbers on it.

The hotter it runs the shorter the lifetime. +10C roughly halving it.

Cooling, namely air flow, makes a big difference to temp rise from ESR
dissipation. Or being near warm parts.

ESR increases with temperature, sort of a runaway effect.

Several small caps in parallel is better thermally than one big one.

 

[piglet]

On 28/08/2019 16:29, Ethan Petersen wrote:

I am trying to get a handle on ripple current ratings for capacitors, starting with the ChemiCon KZN series. The data sheet has specifications for rated ripple current and maximum temperature. It seems I can reliably operate well beyond the ripple current rating, but I don't have a good understanding of what the real limitations are.

How do capacitor manufacturers come up with the ripple current rating for their parts? I had thought this was mostly a function of the capacitor's internal temperature, and the effect on the electrolyte. If it gets too hot the electrolyte might boil, or in a less extreme case the electrolyte permeates through the plug faster effectively reducing the life of the capacitor. Does this sound correct? From the little bit of testing I have been doing, it seems more complicated than that. I would like to come up with a good way to measure this or otherwise put numbers on it.

Does anyone know of a good reference for the design of capacitors? I would like to get a better understanding of the chemistry. From what little I know, the chemistry in the electrolyte is pretty sophisticated. This is a mature technology.

Thanks,

Ethan

I found the Illinois Capacitor lifetime calculators helpful:

<https://www.illinoiscapacitor.com/tech-center/life-calculators.aspx>

piglet

 

[Klaus Bahner]

On 28-08-2019 17:29, Ethan Petersen wrote:

Does anyone know of a good reference for the design of capacitors? I would like to get a better understanding of the chemistry. From what little I know, the chemistry in the electrolyte is pretty sophisticated. This is a mature technology.

Last time I checked - and this was almost a decade ago - there was not
much literature regarding the chemical/physical details of electrolytics
available. Two articles that give at least some but probably no
comprehensive insight into the topic are

Alwitt, R., Hills R. The Chemistry of Failure of Aluminum Electrolytic
Capacitors, IEEE Trans Parts, Materials and Packaging, Issue 2, Sep 1965

Stevens, J., Schaffer, J., Vandenham, J., Service Life of Large Aluminum
Electrolytic Cpacitors: Effects of Construction and Application, IEEE
Trans. Industry App. Vol. 38, No. 5, Sep/Oct 2002

I found especially the latter one quite interesting because it
contradicts the common conviction that "evaporation of the electrolyte"
is the root cause of capacitance loss and the ultimate failure of
electrolytic capacitors.

Klaus

 

On Wednesday, 28 August 2019 17:37:31 UTC+1, jla...@highlandsniptechnology.com wrote:

> ESR increases with temperature, sort of a runaway effect.

All the manufacturers' application notes I have looked at claim
that Al electrolytic capacitor ESR decreases as temperature increases,
so no thermal runaway.

John

 

[Phil Allison]

jla...@highlandsniptechnology.com wrote:

ESR increases with temperature, sort of a runaway effect.

** Nonsense.

The exact opposite is the case, ESR typically FALLS by a factor of 5 between room temp and max rated temp.

Anyone with a hot air gun ( ie hair dryer) and a basic ESR meter can verify this.



..... Phil

 

[Winfield Hill]

Ethan Petersen wrote...

I am trying to get a handle on ripple current ratings ...
It seems I can reliably operate well beyond ...

You do NOT want to operate beyond the ripple rating,
in fact you should stay at least 2x below the rating.
And 3x or 4x is even better.


--
Thanks,
- Win

 

[Tim Williams]

"Winfield Hill" <winfieldhill@yahoo.com> wrote in message
news:qk7aq5024ep@drn.newsguy.com...

Ethan Petersen wrote...

I am trying to get a handle on ripple current ratings ...
It seems I can reliably operate well beyond ...

You do NOT want to operate beyond the ripple rating,
in fact you should stay at least 2x below the rating.
And 3x or 4x is even better.

The limiting factor is core temperature. Which they conveniently don't show
you how to measure, of course*.

*Very rarely, you find a datasheet with RthCA (core to ambient) so you can
figure this out without having to special-order be-thermocouple'd capacitors
to measure it directly (which they can actually supply, if you ask nicely
enough).

Presumably, you can operate at higher ripple, at significantly lower
ambient, for the same lifetime. So, maybe a 2khr 105C part handles 1A at
105C ambient, say 1.4A at 85C, 2A at 65C, and so on.

It's not a lot of improvement for such a big cost in terms of lifetime, and
one much better served by buying the right part in the first place (say a
5khr 105C part that does 2A to begin with), or enough in parallel for the
same effect.

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/

 

On Wednesday, 28 August 2019 16:29:47 UTC+1, Ethan Petersen wrote:

I am trying to get a handle on ripple current ratings for capacitors, starting with the ChemiCon KZN series. The data sheet has specifications for rated ripple current and maximum temperature. It seems I can reliably operate well beyond the ripple current rating, but I don't have a good understanding of what the real limitations are.

How do capacitor manufacturers come up with the ripple current rating for their parts? I had thought this was mostly a function of the capacitor's internal temperature, and the effect on the electrolyte. If it gets too hot the electrolyte might boil, or in a less extreme case the electrolyte permeates through the plug faster effectively reducing the life of the capacitor. Does this sound correct? From the little bit of testing I have been doing, it seems more complicated than that. I would like to come up with a good way to measure this or otherwise put numbers on it.

Does anyone know of a good reference for the design of capacitors? I would like to get a better understanding of the chemistry. From what little I know, the chemistry in the electrolyte is pretty sophisticated. This is a mature technology.

Thanks,

Ethan

It seems to be a heat problem, but you won't notice caps getting hot in service, even when their service life is much reduced by the ripple current. You don't have to go anywhere near BP for problems to happen.

As others have said, stay well away from max ripple rating unless you don't want it to last. MTTF is the point at which 50% of your caps have failed, normally you want way more reliability than that.


NT

 

[Michael Terrell]

On Friday, August 30, 2019 at 1:54:18 AM UTC-4, tabb...@gmail.com wrote:

On Wednesday, 28 August 2019 16:29:47 UTC+1, Ethan Petersen wrote:

I am trying to get a handle on ripple current ratings for capacitors, starting with the ChemiCon KZN series. The data sheet has specifications for rated ripple current and maximum temperature. It seems I can reliably operate well beyond the ripple current rating, but I don't have a good understanding of what the real limitations are.

How do capacitor manufacturers come up with the ripple current rating for their parts? I had thought this was mostly a function of the capacitor's internal temperature, and the effect on the electrolyte. If it gets too hot the electrolyte might boil, or in a less extreme case the electrolyte permeates through the plug faster effectively reducing the life of the capacitor. Does this sound correct? From the little bit of testing I have been doing, it seems more complicated than that. I would like to come up with a good way to measure this or otherwise put numbers on it.

Does anyone know of a good reference for the design of capacitors? I would like to get a better understanding of the chemistry. From what little I know, the chemistry in the electrolyte is pretty sophisticated. This is a mature technology.

Thanks,

Ethan

It seems to be a heat problem, but you won't notice caps getting hot in service, even when their service life is much reduced by the ripple current. You don't have to go anywhere near BP for problems to happen.

As others have said, stay well away from max ripple rating unless you don't want it to last. MTTF is the point at which 50% of your caps have failed, normally you want way more reliability than that.

Years ago when vacuum tubes still freely roamed the Earth, half wave voltage doublers were common in low end consumer products. a 160uF, 250 electrolytic was used as the input capacitor. These would get hot enough to melt the tar inside the aluminum can, and blow its guts out of the can. Those capacitors were rated for ripple current, rather than 'Power Factor' or ESR. Some capacitor OEMs would sell crap that was supposed to be high current, but they were unmarked. The real techs would laugh at the penny pinchers who bought crap that wouldn'y last 30 days of heavy use.

 

[Ethan Petersen]

On Wednesday, August 28, 2019 at 11:14:39 PM UTC-7, Tim Williams wrote:

"Winfield Hill" <winfieldhill@yahoo.com> wrote in message
news:qk7aq5024ep@drn.newsguy.com...
Ethan Petersen wrote...

I am trying to get a handle on ripple current ratings ...
It seems I can reliably operate well beyond ...

You do NOT want to operate beyond the ripple rating,
in fact you should stay at least 2x below the rating.
And 3x or 4x is even better.

The limiting factor is core temperature. Which they conveniently don't show
you how to measure, of course*.

*Very rarely, you find a datasheet with RthCA (core to ambient) so you can
figure this out without having to special-order be-thermocouple'd capacitors
to measure it directly (which they can actually supply, if you ask nicely
enough).

Presumably, you can operate at higher ripple, at significantly lower
ambient, for the same lifetime. So, maybe a 2khr 105C part handles 1A at
105C ambient, say 1.4A at 85C, 2A at 65C, and so on.

It's not a lot of improvement for such a big cost in terms of lifetime, and
one much better served by buying the right part in the first place (say a
5khr 105C part that does 2A to begin with), or enough in parallel for the
same effect.

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/

Like I mentioned earlier, I still don't have a good understanding of the failure mechanisms, or the criteria the manufacturer uses to determine the ripple current rating. What I am hearing here contradicts experience from the field. Millions of devices shipped, over many years, and no field returns for capacitors, despite routinely operating at ripple current well over the rating. Competitors products are even more aggressive with ripple current ratings.

The Illinois Capacitor Lifetime Calculator looks useful. It seems to line up with other academic literature I have been reading. Thanks piglet.

There doesn't seem to be much published information on the chemistry in the capacitors. I guess everything is proprietary and held close. Some articles are useful such as the ones recommended by Klaus Bahner, thanks.

Since the failure mode seems to be excessive temperature, I have been trying to get a capacitor to fail due to excessive temperature and ripple current. Heating the capacitor to 160C for 30 minutes did not cause it to blow up, or change capacitance and ESR. I was monitoring this while running the test. Capacitance increased, and ESR decreased as temperature increased, then returned to normal when brought back to room temp. I was hoping to see a catastrophic failure such as the electrolyte boiling and the cap exploding, but no luck. That doesn't seem to be the failure mechanism.

So I tried again, except with 15 A rms ripple current, also at 160C. The capacitor is rated for 2.5 A rms ripple, so I am operating at 6x the rating. I had one capacitor fail because the leads fused. I re-soldered the leads much closer to the body of the cap, and it continued to run. Again no failures.

I suspect the ultimate failure mechanism is something more subtle, such as corrosion or some other electro-chemical reaction degrading the internal structure. This is probably going to require a big long life test, but it would be good to have a better understanding of the part before starting something like that.

The adventure continues,

Ethan

 

[Phil Allison]

Ethan Petersen wrote:

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

What I am hearing here contradicts experience from the field.
Millions of devices shipped, over many years, and no field
returns for capacitors, despite routinely operating at ripple
current well over the rating.

** Alluding to some mysterious example nobody is allowed to know about is FUCKING STUPID !!

Competitors products are even more aggressive with ripple current ratings..

** Really ?

The Illinois Capacitor Lifetime Calculator looks useful.
It seems to line up with other academic literature I have been reading.

** More mysterious allusions.

Some of their 500V electros are garbage.

Since the failure mode seems to be excessive temperature,

** Should be "a" failure mode, not "the".

I have been trying to get a capacitor to fail due to excessive temperature and ripple current. Heating the capacitor to 160C for 30 minutes did not cause it to blow up, or change capacitance and ESR.

** Another mystery example that goes against common experience.

160C is enough to make any electrolyte boil and an explosion is very much on the cards - or more accurately a sudden venting of lots of white smoke.

I was monitoring this while running the test. Capacitance increased, and ESR decreased as temperature increased, then returned to normal when brought back to room temp. I was hoping to see a catastrophic failure such as the electrolyte boiling and the cap exploding, but no luck. That doesn't seem to be the failure mechanism.

** I have *witnessed* electro caps venting violently from either excessive ambient ( inside and tube amplifier ) or from overvoltage causing excessive leakage current.

So I tried again, except with 15 A rms ripple current, also at 160C. The capacitor is rated for 2.5 A rms ripple, so I am operating at 6x the rating. I had one capacitor fail because the leads fused. I re-soldered the leads much closer to the body of the cap, and it continued to run. Again no failures.

** Bizarre.


> I suspect the ultimate failure mechanism is something more subtle, such as corrosion or some other electro-chemical reaction degrading the internal structure.

** Internal corrosion does kill electro caps by eating away the aluminium strips that connect to the terminals. This is more likely however with caps that have had little or no use.

Far and away the *main* failure mode is the cap going high ESR due to gradual loss of electrolyte over years. Poor sealing at the ends and or a high ambient are the main causes.

You are one colossal fool wasting your own time chasing non existent rabbits down holes.

Please go ahead, make my day.....



...... Phil

 

[Phil Allison]

Ethan Petersen wrote:

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

What I am hearing here contradicts experience from the field.
Millions of devices shipped, over many years, and no field
returns for capacitors, despite routinely operating at ripple
current well over the rating.

** Alluding to some mysterious example nobody is allowed to know about is FUCKING STUPID !!

Competitors products are even more aggressive with ripple current ratings..

** Really ?

The Illinois Capacitor Lifetime Calculator looks useful.
It seems to line up with other academic literature I have been reading.

** More mysterious allusions.

Some of their 500V electros are garbage.

Since the failure mode seems to be excessive temperature,

** Should be "a" failure mode, not "the".

I have been trying to get a capacitor to fail due to excessive temperature and ripple current. Heating the capacitor to 160C for 30 minutes did not cause it to blow up, or change capacitance and ESR.

** Another mystery example that goes against common experience.

160C is enough to make any electrolyte boil and an explosion is very much on the cards - or more accurately a sudden venting of lots of white smoke.

I was monitoring this while running the test. Capacitance increased, and ESR decreased as temperature increased, then returned to normal when brought back to room temp. I was hoping to see a catastrophic failure such as the electrolyte boiling and the cap exploding, but no luck. That doesn't seem to be the failure mechanism.

** I have *witnessed* electro caps venting violently from either excessive ambient ( inside and tube amplifier ) or from overvoltage causing excessive leakage current.

So I tried again, except with 15 A rms ripple current, also at 160C. The capacitor is rated for 2.5 A rms ripple, so I am operating at 6x the rating. I had one capacitor fail because the leads fused. I re-soldered the leads much closer to the body of the cap, and it continued to run. Again no failures.

** Bizarre.


> I suspect the ultimate failure mechanism is something more subtle, such as corrosion or some other electro-chemical reaction degrading the internal structure.

** Internal corrosion does kill electro caps by eating away the aluminium strips that connect to the terminals. This is more likely however with caps that have had little or no use.

Far and away the *main* failure mode is the cap going high ESR due to gradual loss of electrolyte over years. Poor sealing at the ends and or a high ambient are the main causes.

You are one colossal fool wasting your own time chasing non existent rabbits down holes.

Please go ahead, make my day.....



...... Phil

 

[George Herold]

On Wednesday, September 4, 2019 at 8:32:24 PM UTC-4, Ethan Petersen wrote:

On Wednesday, August 28, 2019 at 11:14:39 PM UTC-7, Tim Williams wrote:
"Winfield Hill" <winfieldhill@yahoo.com> wrote in message
news:qk7aq5024ep@drn.newsguy.com...
Ethan Petersen wrote...

I am trying to get a handle on ripple current ratings ...
It seems I can reliably operate well beyond ...

You do NOT want to operate beyond the ripple rating,
in fact you should stay at least 2x below the rating.
And 3x or 4x is even better.

The limiting factor is core temperature. Which they conveniently don't show
you how to measure, of course*.

*Very rarely, you find a datasheet with RthCA (core to ambient) so you can
figure this out without having to special-order be-thermocouple'd capacitors
to measure it directly (which they can actually supply, if you ask nicely
enough).

Presumably, you can operate at higher ripple, at significantly lower
ambient, for the same lifetime. So, maybe a 2khr 105C part handles 1A at
105C ambient, say 1.4A at 85C, 2A at 65C, and so on.

It's not a lot of improvement for such a big cost in terms of lifetime, and
one much better served by buying the right part in the first place (say a
5khr 105C part that does 2A to begin with), or enough in parallel for the
same effect.

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/


Like I mentioned earlier, I still don't have a good understanding of the failure mechanisms, or the criteria the manufacturer uses to determine the ripple current rating. What I am hearing here contradicts experience from the field. Millions of devices shipped, over many years, and no field returns for capacitors, despite routinely operating at ripple current well over the rating. Competitors products are even more aggressive with ripple current ratings.

The Illinois Capacitor Lifetime Calculator looks useful. It seems to line up with other academic literature I have been reading. Thanks piglet.

There doesn't seem to be much published information on the chemistry in the capacitors. I guess everything is proprietary and held close. Some articles are useful such as the ones recommended by Klaus Bahner, thanks.

Since the failure mode seems to be excessive temperature, I have been trying to get a capacitor to fail due to excessive temperature and ripple current. Heating the capacitor to 160C for 30 minutes did not cause it to blow up, or change capacitance and ESR. I was monitoring this while running the test. Capacitance increased, and ESR decreased as temperature increased, then returned to normal when brought back to room temp. I was hoping to see a catastrophic failure such as the electrolyte boiling and the cap exploding, but no luck. That doesn't seem to be the failure mechanism.

Well 30 minutes is not much time. If you put numbers into the
Illinois Capacitor Lifetime Calculator, what sort of time do you get?
Have you tried contacting the cap manufacturer?

George h.

So I tried again, except with 15 A rms ripple current, also at 160C. The capacitor is rated for 2.5 A rms ripple, so I am operating at 6x the rating. I had one capacitor fail because the leads fused. I re-soldered the leads much closer to the body of the cap, and it continued to run. Again no failures.

I suspect the ultimate failure mechanism is something more subtle, such as corrosion or some other electro-chemical reaction degrading the internal structure. This is probably going to require a big long life test, but it would be good to have a better understanding of the part before starting something like that.

The adventure continues,

Ethan