On Sun, 7 Oct 2007 15:25:35 +1300, "~misfit~"
<misfit61nz@yahooligans.co.nz> wrote:

>Somewhere on the interweb "kony" typed:
>> On Sat, 6 Oct 2007 22:07:11 +1300, "~misfit~"
>> <misfit61nz@yahooligans.co.nz> wrote:
>>
>> > Somewhere on the interweb "kony" typed:
>> > <snip>
>> > > I suspect you have failed capacitor(s) in your power supply.
>> >
>> > I had that last week on one of my older machines. Scared the
>> > bejesus out of me:
>> >
>> > http://test.internet-webmaster.de/upload/1191392010.jpg
>>
>> Unless the fan fails, I seldom see those high-side caps
>> fail, it's almost always the ones after the transformer
>> output is rectified in the LC circuits, either the 3.3/5/12V
>> rails or the 5VSB.
>
>I knew those were the caps you were talking about, just wanted to share my
>explosive experience. So you've never seen those big ones fail? Maybe it's
>something to do with the fact that we have 220V mains in NZ? You're in the
>US are you not? 110V mains. /Most/ supplies I've seen are switchable so,
>when used with half the cap's rated voltage are probably less likely to
>fail. The fan was/is working fine.

Certainly any part is subject to a certain % failure rate,
but yes in the US we have 110V and in a PSU with voltage
doubling circuit, there are 200V caps leaving a 400V/340V
margin. On active PFC units, there should be same margin
regardless of input (so long as within spec'd tolerance).
I think it most likely you just came upon a defective cap,
or an extreme surge on the mains AC line.



>> Normally the PSU 5VSB and some parallel resistors drain
>> those caps down to a negligable level. I'm wondering if you
>> either had (have) another fault still, or if the remaining
>> voltage on the caps was low enough to be safe, even if a bit
>> startling.
>
>I hope the latter. It was only a small spark. However, I've heard others
>issue warnings about dangerous voltages being retained in PSU caps so
>thought I'd mention it.


I've heard them as well, but they never give specifics of
the circuit, so given that _all_ of those I've seen have
both the resistors and 5VSB sub-supply operating to drain
them to a far lower level, I wait for an example of one that
does't before presuming the voltage would remain high
enough to matter.

Even if a cap is completely unloaded, it will drain down
to a level where the retained capacity vs voltage may be
startling, but in caps of this volume, it does't take long
for them to drain to a safe level, besides that there are
two mechanisms for further discharge in an ATX PSU.

>
>> Certainly if a high voltage cap fails it has to be replaced,
>> but the more common failures are the caps around the
>> existing wiring harness connection to the PCB, or (still in
>> the same general vicinity on the PCB) one set of inductors
>> before that point.
>
>Yeah, I've heard that. Strangely, I've never had it happen myself. <touches
>wood>


You are either very fortunate, or avoid buying junk PSU.
Those last couple of stages output caps are pretty
common failure points in PSU powering modern systems. When
I first see signs of instability in a system formerly
working alright, the first thing I do is pull the PSU,
swapping in one known functioning properly, and examine it.
Over 1/2 the time, the PSU has failed caps in this scenario.

I should mention, this is mostly with original stock PSU.
I care not about warranty and immediately open a PSU to
inspect. If there are marginal caps, I can replace
them and reuse them on less demanding electronics projects.

IMO, tiny performance gains over time are not as important
as having any given example running for as long as
possible. Some aren't in the same situation, but I have
found a lot of people who would gladly pay $200 for older
parts so long as they're reliable, guaranteed to last for
awhile.