Exploding power cap - Why?

[Tarasque]

I just finished buiding an Aleph 0. For the power supply I uesd Hitachi 22.000/100V industrial grade elco's.

The power supply connections were tripple checked and then again checked by a friend of mine. All was wired correctly and the amp fired up and working like a dream!

After about 20 hours of operation, one of the caps exploded. The whole plate with screw terminals came off the can.

So now I'm really puzzeled as to what happened here. I verified all connections again OK. I can not see or measure any problem with the amplifiers. After I cleaned the channel with the blown cap, I hooked it up to the other channels power supply and it again was working.

The problem is that I'm still hessitating to put a new cap in and operate in in my living room when I do not know why this cap exploded.

Suggestions...Anyone?

Thanks.

[Nigel Goodwin]

What voltage do you have across them?, usual causes for 'exploding' are connected wrong way round (unlikely as it wouldn't last 20 seconds, never mind 20 hours), or too high a voltage across them. Older electrolytics had a little reserve in their voltage rating, but modern ones are VERY narrowly specified, if you put 101V across a 100V electrolytic, even for a short time, it may well blow.

[Tarasque]

Hi Nigel,

The cap is specified as 100V. The power supply delivers 50V.

Thanks...Marcel.

[Maousse]

I may have an explanation if the caps were NOS, they might have need "reforming" before use.

Reforming elco is putting a current controlled tension to the cap, to rebuild the oxyde isolation of the aluminium foil inside. Because this isolation tends to dissolve into the electrolyte as time goes by.
The process is more sensitive with higher operating voltages.

When you use directly an old elco, losses are hifh and the cap heats up. When the electrolyte boils, pression builds up to the "explosion".

Do you intend to use elco from the same source? If yes, just power le power supply without load (amp) with a 2-5W 1k resistor inserted after the bridge and before the caps. If the current is low (less that, say 1V accross the resistor, that makes 1mA), it is OK.

hope this help.

[eduard]

Hello,
Be sure to disconnect all the caps and reform them all to be sure. The last post from France is right i think. Even if your working voltage is very far from the one stated on the cap it can get you into trouble after they have been on the shelf for several years. I remember that somebody told me that some surplus caps are coming from companies that are not allowed to use them anymore after they have been on the shelf more than 5 years just incase an accident will happen and the insurance will start looking for a cause greetings, Eduard

[Maousse]

Forgot to mention

If the losses are high, just let it on.
Check some time to time the tension across the resistor, it should lower little by little (may need few hours, even days).

If there is a short in the cap, its over, just throw it away.

Always stay around, il case of failure, if you have to leave, shut it down, all reforming already done is not lost.

[Tarasque]

Thanks Guy's...I also found something about Fake Hitachi caps.
Could this be a problem?

[Maousse]

Fake elcos can be the problem if they are NOT caps or low voltage caps. Can you put a picture of the surviving caps? just for information?

[cliffforrest]

Never forget the simple fact that it may be faulty.

It happens .....

[Babowana]

Originally posted by Tarasque



...I also found something about Fake Hitachi caps.
Could this be a problem?

I understand that the exploding of caps is due to heat generation,
and the heat generation is coming from three reasons:

1. Over-voltage

The over-voltage application provides excessive leakage current
across the cap. Then power dissipation happens according to I(leak) x V.

2. High ripple current

The high ripple current creates the power dissipation according to
Ir x Ir x ESR.

3. Reverse bias

The reverse bias creates heat due to the internal chemical reaction.

I guess when they made the fake, they did wrong printing of the
polarity indication on the plastic cover...
The most probable reason could be item 3...

I can't be 100% sure of it tho...