excrubulent

Distinguished
Sep 5, 2009
7
0
18,510
Okay, I've had a few hardware failures on this machine of late, and I'm starting to get nervous.

First, my CPU died, but I got it replaced on warranty. No harm, no foul. Thanks, Intel.

Then about a week or two later I had some weird GPU bottlenecking - the GPU was being underutilised and I was getting some mad lag. I tried a bunch of things to get it working again, but I couldn't. Eventually, after two weeks dwelling on this problem, I switched the RAM from my wife's computer into mine, and the problem went away. Obviously a RAM problem, then.

Then I put the original RAM back in an hour or so later, because I was curious and wanted to run Memtest. Then the computer wouldn't boot, and started cycling on and off every 5 seconds or so. For those keeping score at home, that's a new problem - the RAM wasn't stopping the computer from booting before this. So naturally I pushed the power button to stop it, and turn it off completely.

It turned off successfully, but then some tiny component on the motherboard SPONTANEOUSLY COMBUSTED. WHAT THE CRAP? After cursing prodigiously for a few moments, I found the location of the spot fire - some inconsequential looking thing between the audio input/outputs and the PCI-E 16x slot, where my GPU is, had some burn marks on it.

Anyway, feeling a bit balsy, I tried turning the computer on again, to see what it would do. One of the RAM sticks is busted custard, but the other one is fine. The computer turns on fine, and it loads XP without any problems. WHAT THE CRAP? Why does the computer still turn on? Was this little component just a nothing part?

I'm guessing that only a PSU could cause this kind of widespread damage. I have a multimeter at home, so is there any way to test whether the PSU is eating my parts? I also thought it could be the motherboard somehow, although I'd be surprised if that was the problem.

Of course, the other issue is, if I find out that it's the PSU, can I still use that motherboard? I've got a photo of the bit that went boom. Can I replace that one tiny component, or can I happily run the MOBO without it? I'd appreciate any uber boffins who have access to this kind of knowledge, as I'd like to only have to replace RAM and PSU, not MOBO too.

Also, I have a photo of the burnt bit on the MOBO, but I don't know how to attach it to a forum message, so if anyone can enlighten me, I'd be very appreciative.

System Specs:
CPU: Intel Q9450 @ 2.66GHz
MOBO: Gigabyte GA-EP45-DS4P
GPU: Sapphire Radeon HD4850
RAM: 2x2GB Generic 800MHz DDR2
PSU: Antec Earthwatts 500W
 
Cut and paste from one of my previous threads:

If you have a multimeter, you can do a rough checkout of a PSU using the "paper clip trick". You plug the bare PSU into the wall. Insert a paper clip into the green wire pin and one of the black wire pins beside it. That's how the case power switch works. It applies a ground to the green wire. Turn on the PSU and the fan should spin up. If it doesn't, the PSU is dead.

If you have a multimeter, you can check all the outputs. Yellow wires should be 12 volts, red 5 volts, orange 3.3 volts, blue wire -12 volts, purple wire is the 5 volt standby.

The gray wire is really important. It sends a control signal called something like "PowerOK" from the PSU to the motherboard. It should go from 0 volts to about 5 volts within a half second of pressing the case power switch. If you do not have this signal, your computer will not boot. The tolerances should be +/- 5%. If not, the PSU is bad.

Unfortunately (yes, there's a "gotcha"), passing all the above does not mean that the PSU is good. It's not being tested under any kind of load. But if the fan doesn't turn on, the PSU is dead.

You can check all of these voltages under load from the back of the main PSU power cable.
 
I've got a bad sense of deja vu here. About a month ago we had a similar report of a "small fire" somewhere near PCIe street. Turned out, there was also a "burn mark" on the video card's connector.

Could you please pull the vid card and check the card pins and the socket itself?

Anyhow, as jsc points out, testing can only detect some of the ways a psu can fail. But if you have a multimeter, there's no reason not to.

I'm gonna to take a chance sight unseen and say No, you won't be able to replace the burned part . . . even if the problem was your psu . . . and even if that part was the only thing affected.

Now would be a good time to see if you have warranty coverage on both of those devices. And to notify both Gigabyte and Antec . . . see what they want to do to make you whole.
 

excrubulent

Distinguished
Sep 5, 2009
7
0
18,510
Thanks, guys. I'll check the PSU, but I'm not really game to use it again. The real kicker is, between getting married and moving house and blahdey blah, there's no proof of purchase on any of my parts anymore. I'll be more fastidious in the future about that. Anyway, Intel replaced the CPU without proof of purchase, so I'll see what Gigabyte and Antec have to say for themselves. For the meantime, my HDD is taking refuge in my wife's case, with a liberal use of the F12 key.

Thanks again.
 

excrubulent

Distinguished
Sep 5, 2009
7
0
18,510
Also, in reply to Twoboxer, the fire was confined to one piece of... something on the board itself. It's about 4mm across, and the GPU wasn't touched at all.
 

TRENDING THREADS