FOAK: PC switches itself off

Working away, and this PC powered off. Tried to restart it but nothing. Took
it apart and waggled things. Nothing. Was about to swap out the PSU, but
first tried switching the supply off and back on again with the switch on
the back. Then I could restart it with the front switch. 30 minutes later it
went off again. No luck with the front switch, but toggle the rear switch
and all is well again. For now.

Before I rip it to bits again and swap out the PSU, anybody here seen such
behaviour before?

 

Mine did exact same thing - turns out it was over heating - quick blast with
the airline to clean all the shite out and normal service was resumed

 

Jim escribió:

I don't understand that. Once the green wire is grounded, isn't it only the
motherboard that can switch the PSU off?

I suppose I shouldn't wait until it happens again before changing the PSU.
I've got a 230W PSU in the shed, but I suspect that won't be up to a modern-
ish board with 2 hard drives, 2 CD drives and AGP etc.?

 

Buzby escribió:

Actually, that was the first thing I did. Hoovered all vents, fans etc., but
it happened again after that.

 

Whilst taking care to stop the fan (or all of them if you blast the whole
machine) rotating when you blast if with air. Those buggers can *really*
spin when they're driven by 150psi of air being shot through a needle sized
hole.

A screwdriver "just" inside the fan stops the blades turning, allows the air
to properly clean the blades and stops any chance of bollocksing the PSU
completely.

 

If you hold it down for a few seconds most motherboards are set to turn
the PC off.

I would buy a new one.

 

Is the processor heatsink making good contact? - I recently had similar
problems when two of the four plastic clips supporting my heatsink gave up
the ghost.

 

As well as the above, mine also needed re-seating. Clean the old
heat-sink paste and re-apply a tiny amount, spread as thinly as a thin
thing that's dieted, before re-seating the chip. One of mine the
heatsink had been put on 180 degrees out, so the core was only partially
contacting the actual heat sink so check for this as well!

 

AOL Had this on our own Dell desktop, needed a new PSU, secondhand off
ebay thankfully as new ones from Dell cost a small fortune.

Weird.

 

Be very carefull with the airline. I tried cleaning the shit out of my old
Dell laptop, the air made the fan go wheeeeeeeeee and its motor must have
behaved as a generator because it fucked the motherboard.
Clumsy ?
Moi ?

 

I ought to say, try plugging in the spare you have, you don't have to
even install it properly, you can swap the wires over with the side of
the case off.

That way you can find out if it really is the PSU at fault.

 

If it's an oldish mainboard it's worth checking for bulging caps
around the VRM area just in case.

Check the ATX12V CPU feed - little 4 pin square connector near the CPU
on the mainboard(2 black and 2 yellow wires), pull it off (ooer) and
give the terminals a quick squirt with contact cleaner, there was a
batch of these connectors produced by AMP a while ago that were just
out of tolerance, they would tarnish up very quickly and cause all
sort of issues like overheating messages at POST on certain desktop
boards.

As others mention though.
Re-seat everything you can.
Clean it out.
Check heatsink is actually cooling - hwmonitor is good for this.
Gpuinfo will also tell you if your gfx card is overheating.

 

Yes.

I turned it back on with the rocker switch at the back, and it did it again.
And again. Then on the fourth attempt it got halfway through booting and
went "BANG! Kapaow! >tink< >tink< >tink<" (magic smoke).

I opened up the power supply and there were some burn marks where a couple
of large resistors used to be. I replaced the power supply and the PC
did nothing. Then I looked closely at the motherboard and realised what
the >tink< noises were, it was the tops of various chips ricocheting around
the inside of the computer's case.

Both hard discs had melted ICs on it, the graphics card was killed,
the motherboard had several ICs blown up on it. The only parts that still
worked were the keyboard, mouse and CD-ROM drive. Everything else was fried.

Horowitz and Hill say of line powered switch mode power supplies in
_The Art of Electronics_ that they have been known to even "expire with
a loud explosion". They weren't wrong.

 

Jim escribió:

Of course, it hasn't done it since. Perhaps if I give it some work to do -
although it is 30C in here at the mo. I'll render an .avi.

 

80% Isopropyl for the cleaning stage is reccommended apparently.

Mine will be undergoing this process on both procs in a while as it keeps
doing that same "off by surprise" thing occaisionally and coughing a VCORE
error to the BMC. Paste and cleaner on my shopping list. I've already pissed
off the Mrs by hoovering it clean and remounting the fans.

Any advice on reseating the proc? Should do... Shouldn't do etc?

 

As others have said the PSU is the prime suspect - but while you have the
cover off, it can't hurt to inspect the electrolytic capacitors on the
motherboard just to make sure none have bulged tops.

 

Not to mention condensation in the compressed air hose, if that gets under a
BGA chip (the big square chips with no pins sticking through the board)
powering up while wet will cause leakage paths between the connections that
won't go away when it dries out.

 

Would you like the schematic for a thyristor crowbar overvolt protector?

 

I have just had the same problem - bought a new PSU and no difference.
Then I noticed the bulging capacitors on the MBD

 

Quite a common problem. The old PSU may well have had bad caps as well.