Even Intel used bad capacitors

Leaking Caps wrote in news: snipped-for-privacy@anon.mixmaster.mixmin.net:

It's written on the other side. Take a few pictures, the camera flash and magnification will get you a reading, you just have to find an angle good enough to see it.

It's probably 10V 1000 µF. Panasonic make some good replacements. RS Components order code 526-1115, for example. Check the lead spacing, they're likely an exact match but maybe not.

You're looking at the wrong side !

That's pretty mild bulging and I see no leaking.

Graham

Any idea why such caps are very often 10V rated and not 6.3V ?

Graham

Probably to get the current rating that they need. They probably need a bigger package to get rid of the heat and the extra voltage comes along with that.

Eeyore wrote in news: snipped-for-privacy@hotmail.com:

Not really sure, but I guess it's a general principle of overrating being likely to improve life and performace if it's not significantly more expensive. Could argue for lots of stuff like higher ripple current, lower ESR, better heat dissipation if the can is bigger, better surge resistance, but in the end I doubt they calculate all that, they probably just test with one cap, then uprate to the next voltage for production on the grounds that it's cheaper if many are the same, and it probably makes less returns.

In short, crude uprating and using lots of equal parts costs less than trying to design their performance to precise limits. Same as it's always easier to fligh high and straight than to hug the ground to avoid radar, even though the initial cost of getting up there is greater.

OK, I took some more pictures and got their value. They are Nichicon brand, 560uF 6.3volt. Both of them show signs of oozing an amber liquid from the top.

The board was stable if I had one piece of memory installed. If I had two pieces installed, same brand and part#, I would get random crashes.

Leaking Caps wrote in news: snipped-for-privacy@anon.mixmaster.mixmin.net:

You could uprate to those 10V 1000µF caps I mentioned. Capacitance tolerance alone can change from nominal 560 to 20% more, it's not that critical. Adequate supply smoothing to memory modules IS though. So check the lead spacings. If they match, get some. Get several tens of them even, they'll replace other caps you haven't found yet, probably. Most firms like to stick to standard values, so you won't go far wrong buying between 20 and 100 of RS's part 526-1115. As far as I know, they only buy from direct sources where possible, I've not yet seen any problems with parts bought from them. Probably same with Mouser and Digikey, etc, I doubt they'd risk buying from loose sourcing that let in counterfeits easily, it would damage them too much if that happened.

One of those is a Nichicon part. I don't recall those being affected by the 'bad caps' issue so it's probably just simply 'worn out' from extended use.

Graham

the

id

or

Those are Aluminum Electrolytic Capacitors

I haven't heard of this issue for a few years now

Standard design pratice is to over rate the component value. The common term is derating;

It's pretty much fixed but rumour has it that some manufacturers are still shipping duff stock. But that's rumour.

Graham

Yes I'm aware of that but advice I've had from cap makers is that it's completely unneccesary. It affects ONLY leakage current and that's largely unimportant.

Graham

and

good

completely

Simply out of prudence neither would I normally but I have the following tale to tell.

I was designing an audio power amp and to meet the desired power output, the supply rails were ~ 67V off-load. That would have ruled out 63V caps for the psu but the manager was so concerned about cost that he spoke to the cap makers and asked for their advice.

They repsonded that there would only be a very mild increase in leakage current and that the caps were designed for at least a 10% 'overvoltage' anyway.

So we used 63V caps on a supply that was 67V off-load and we had ZERO reliability problems with that. In fact it was one of the most reliable products that the company ever made.

Yes but most kit isn't intended for military or aerospace use. Most deratings are historically based AIUI and modern components really don't need it.

Graham

are

That's why most consumer electronics dies within two years.

are

I'd expect electrolytics to last a lot longer if they are kept cool and/or run below rated voltage.

flash and

ngle good

ts.

pacing,

completely

rtant.

ale to tell.

he supply

but the

ked for their

rrent and

ability

the company

ngs are

I'm just saying that's where I got the data, not what I was designing for.

Cable company called the othet day, wanted me to buy VOIP.

I told them their service sucks, there was no way I would purchase anything. But you have not Complained since 2002?

How many people just move to another product?

deratings are

No it doesn't.

Graham

are

Correct. that's VITAL

Utterly WRONG. This is pure myth.

Graham

Eeyore wrote in news: snipped-for-privacy@hotmail.com:

I was given an amp once, burnt cap in the PSU. 7 other caps were fine, and the voltage was like yours, very slightly exceeding the cap spec. While most survived, the failure was spectacular. It's worth uprating to limit the odds to nearly zero because if it does fail, it might take out more than itself.