Probable Heat Problem

[finnrogers]

My PC has died on me and I was hoping for some advice. Here are the symptoms (not the yellow people off the telly):

1) I didn't change anything before it died, I just shut down as usual then when I went to boot it up the next day, it posted then froze just before getting to the Windows logo.

2) Restarted, and it wouldn't post.

3) Cleared the CMOS and tried again, nothing.

4) Cleared CMOS, took battery out and unplugged from mains, waited 20 mins then tried again, nothing.

5) Unplugged everything except one ram stick, gfx, cpu, still wouldn't post.

6) Gave up for the day then tried next day: Posted again (this is with just cpu, gfx, 1 ram stick still) - entered BIOS and check the PC health menu for temps - Seem absolutely fine, around 24 degrees for mobo, 32 for CPU. Then it froze again, in the middle of the BIOS, keyboard wouldn't respond at all. Restarted and it wouldn't post again.

7) Left it a couple of hours and it posts again but again only lasts about 30-60 seconds before freezing up.

So my conclusion would be a heat problem, but I can't for the life of me figure out where the problem lies, I've dusted off the heatsinks carefully, and tried reseating everything. The only thing I haven't done is reapply thermal paste to the cpu and gfx but why would they need doing it I haven't touched them since building the rig? Anyone had any experience of this before? Suggestions?

The PC specs are listed in my profile.

Your help will be much appreciated thanks...

[finnrogers]

bump

[Lucio]

With something as pervasive as your failure, it's got to be something fairly critical doing this. Unfortunately the only way to debug a failure this widespread is to replace things one at a time until the problem goes away.

I'd try in this order, ranked in order of most likely to fail

1) PSU
2) RAM
3) Motherboard
4) GFX card (though it seems unlikely to cause your problem)
5) CPU


Obviously there are ways to test each part, such as RMA'ing them to the manufacturer (if they're still under warrenty), but quite it takes time and you could get charged testing fees.

[finnrogers]

Right I've swapped the PSU and I've also taken all the thermal paste off the CPU, GPU and Motherboard Chip and reapplied Arctic Silver 5. I've also carefully dusted out everything (not that there was much tbh) and this has made no difference at all.

I also bought a replacement mobo on ebay (exactly the same board: MSI Neo2 Platinum) but unfortunately it is completely dead (doesn't even power on, no fan activity at all, nada) so wasted £18 there (it was sold as untested so I got screwed basically).

None of my parts are under warranty any more unfortunately.

I'm beginning to suspect that the gfx card may be at fault for the simple reason that there are some artifacts on the MSI boot logo (random black vertical lines all over the logo) and some of the letters in the BIOS screen seem brighter than others... unfortunately I can't get hold of a PCI or AGP gfx card at the moment to test. I may just have to swallow my pride and take it to a computer shop for them to diagnose the fault (then fix it myself).

I'll keep you posted (if anyone actually cares lol)

[Ferral]

Have you tried the one stick of ram at a time?

Does it boot into safemode at all? If it is booting into safemode could be something wrong with a driver or the Windows installation has become corrupted. You could also try disconnecting one of the hard discs and take out anything non essential for testing purposes.

Take out sound card, network card, disconnect additional Hard Disc and floppy drive etc. Literally take it down to barebones and see if it will start up.

[finnrogers]

Yes tried all the above. I've only got one ram stick in, the gfx card, cpu and psu - no other peripherals at all (except the keyboard). The problem is definately hardware related since it locks up randomly after a period of about 30 seconds (from cold). Once it has done this, it won't reboot i.e: there is fan activity but the monitor doesn't display anything and the normal confirmation "beep" doesn't happen. If I then leave the pc switched off for about an hour, it will switch on again but locks up again and usually a bit quicker than the previous time.

This is why I suspect a heat problem, but can for the life of me see why...

[Ferral]

Have you tried doing a bios update from a DOS environment, worth a shot to see if it does anything helpful.

Try under clocking things like the cpu and ram also from the bios.

Had something similar to this happening to me when I had my 6600GT card in few years back, turned out that it was the PSU on its way out and the voltage wasn't steady enough. Also last summer the fan I had on my processor was giving up the ghost and causing the system to shut down. Might be worthwhile replacing that before going any further. Boot it up and let it go off then restart the pc a few mins later and jump into the bios and check the temps on the CPU.

It could even be something daft like the bios battery and the cmos settings have changed so the system will shut down once the CPU reaches a certain temperature that is not set right.

[peterb]

Given that you didn't do anything to the machine and it just died, it suggests (catastrphic) failure of one component - rather than a heat/temperature problem.

You have changed the PSU (my first thought) but as you can get it to (sometimes) post with a minimal hardware configuration, I would look at the memory.

Download a copy of memtest86+ and burn it onto a CD. Boot from that and check out the memory.

[Ferral]

Another thing to have a closer look at on the board is the capacitors, check them all and see if any of them have started to split or have raised heads.

[finnrogers]

No this is all fine, I'm able to get into the bios to check the temps albeit only for about 30 seconds. All the temps are fine. The CPU cooler is definately working fine as I can feel a nice breeze of cool air going through it and I've tried reseating it several times (even reapplied thermal paste to it today).

I wouldn't be able to flash the bios since it doesn't stay alive long enough and I doubt that is the problem since I had been using this PC for about 2 years prior to it going wrong randomly.

I've cleared the CMOS a few times and then checked the warning settings in the BIOS - these are set quite high by default, when I'm checking temps they are never near these levels by the time it locks up.

I've just noticed something strange after a bit of tinkering though: If I unplug the power cable from the mains and wait for approx 15 seconds, then put it back in and boot up, it will beep once (as normal) and show the MSI boot logo full of artifacts and glitches some of which are flicking, but it won't let me into the BIOS. I've also noticed that if I gently push on the artic cooling freezer 7 on the gfx card, these on screen moving glitches change. Is this an indication that the gfx card is up the swanny?

Bearing in mind that I've already changed the PSU too.

[peterb]

Possibly - but test the memory. Do you have a spare graphics card you can use for testing?

[finnrogers]

I'm hoping a colleague is going to bring me one in to work tomorrow so I can test. I don't have any RAM knocking about but once the gfx has been tested, this will be my next port of call.