Exceeding electrolytic cap rating

[Pierre]

Hello all.
I have recently experienced failure of one of my electrolitics in my homemade linear PSU.
It is a 85ºC 3300uF/63V one, and the largest voltage I have seen on it is around 64V.
The thing is that the supply was working ok but after one hour or so of operation, it shown clear signs of overheating and almost exploded (like if It had been boiling inside), spreading the acid around.

The capacitor was not exposed to a high temperature.

The thing is... can a voltage rating slightly exceeded produce this kind of failure, or what is the typical failure mode for electrolytics?

I know, I must reduce the voltage as thermal drift has produced this nasty result, but I am a bit concerned about such a catastrophic consequence with such a slightly exceeded voltage.

Thanks!

[guitar_joe]

I wouldn't recommend approaching / exceeding the ratings. Last linear psu i built was in an amplifier described in silicon chip. It had +/- 60v rails and the electolytics used were 80v.

Basically i come to the conclusion that once all mains variations and off load voltages are taken into account the maximum voltage the cap will see is maybe 10v below its rating in a well designed supply perhaps?

I'd say replace the electros with 75v or 80v types and they should never blow up.

[leadbelly]

Got a schematic? I have my doubts that it was actually in 64V service.

[Pjotr]

Hi Pierre,

It is recommended to stay 10 – 20 % below rated voltage for electrolytic caps, also depends on operating temperature. But more important is to keep in mind the current rating of the caps. Max. RMS current can vary a lot among different makes.

If you overstress them by ripple current they can become really hot and finally vent or even explode.

Cheers

[Conrad Hoffman]

As said above, ripple current is a big factor. If you just put DC on the cap, it would probably be fine quite a bit beyond the rating, but that's not how you use it. Also, if you have a diode breaking down, that will cause exactly the problem described. Look at the cap data sheet and factor in high line conditions, ripple current, temperature derating, and a safety factor! IMO, you should try to measure the ripple current if possible.

[Pierre]

Thanks!

[ashok]

You will also find from data sheets that the ripple rating of a capacitor does not double with doubling of capacitance. So two 10,000uF caps will have ( total ) higher ripple capability than one 22,000uF cap.

From one data sheet :
63 volt 10,000 uF ( per unit ) .....ripple current 9.1 A at 85 deg C 100Hz ( ESR = 0.036 ohms )

63 volt 22,000 uF.......................ripple current 11.2 A at 85 deg C 100 Hz (ESR = 0.017 ohms )

[Eva]

The chances of failure increase a lot as you approach or exceed the rated voltage.

Remember that everytime that an electrolytic capacitor is getting hot or drawing DC current something is going wrong and you are supposed to hurry up to switch power off before the thing explodes...

[AndrewT]

Hi,
I don't know how you measured the 64Vdc maximum, but most voltmeters read the average voltage.

One is told by the capacitor manufacturers that the critical voltage is the average plus the ripple voltage. I would have expected the peak voltage be Vdc + half Vripple, but Vpk is not what the manufacturer says.

What was the ripple voltage at the time of failure?

[AndrewT]

Hi,
if the cap di-electric has deteriorated whilst in storage, would the di-electric require to be reformed before being put back into full voltage duty?

What happens if the safe working voltage is exceeded (but still below the rated voltage) due to degraded di-electric? Is the leakage simply higher causing some internal heating, or does it cause a local hot spot that gradually deteriorates to the point of catastophic failure?