Help troubleshooting overheating or PSU issue

[twistedsymphony]

First off let me  say that I'm not entirely sure if it's an overheating problem or a PSU problem or something else...

Basically I've got a PC setup on my DDR cab running StepMania 3.9 it's run great for months without any problems.

2 days ago I was playing after about an hour the PC just turned off, like someone unplugged it.
Trying to turn it back on again right away it would power up, get through POST and then turn back off again... I tried this again and it did the same thing.

Yesterday I tried turning it on again and it ran fine for about 20 minutes, then once again turned off like someone cut the power.

I'm not getting any error messages, no warning, all the fans seem to be working fine.

----------- Machine spec:
This is a micro-ATX in a micro-tower, the case and power supply are new as of the time of the build as are all of the cooling fans. all of the fans appear to be working.

I forget what mobo and processor I'm running but it's got a 512MB Geforce 2 card, and an older IDE harddrive and 2GB of ram.

Windows XP SP2 (SP3 was incompatible with some piece of hardware). and the only application that runs is StepMania.


My suspicion is that SOMETHING is overheating since it shuts back off when attempting to reboot it... either the PSU is overheating and the lack of proper voltages is causing a shut down or the mobo is detecting an overheat and shutting down to protect itself... or something else altogether. The PC is currently sitting outside of the cab with plenty of ventilation and the room is very cool (~65 degrees) and dry.

The fans all appear to be working properly, and normally my method of troubleshooting is to swap out parts and try to narrow things down. unfortunately this is a specialized small form-factor case with a special small-form factor PSU and I don't have another one that I can swap in.

Does anyone have any ideas how I might troubleshoot this to determine what's causing my shut-down?

[cack01]

Here is what I would do.

Run a  selfbooting Prime 95 CD/USB.  Run this for at least 12 hours (should do a day....but time is money).  If this throws errors then it tells you that you have a hardware and not software problem....since it already has issues during boot I doubt it is software.

If you have errors in prime you most liekly need to look at your RAM or voltage.  The easiest way would be to download prime95 for windows (since we know windows is not the culprit and run it in conjuction with a hardware monitor like CPUID's Hardware Monitor.  There are probably better things out there, but you just need a way to see the voltage changes.  Watch your voltage and temps.  If everything looks good then I would suspect RAM or heat in the PSU.

Try testing with one stick of RAM at a time and see if things fail.  When all else fails, keep unplugging items and testing with the self booting Prime95 until things work.  I am also assuming you do not have a video card.

[twistedsymphony]

thanks for the tip... does Prime 95 store a report somewhere? I'm curious if I run it and my machine shuts off again how I'll be able to determine what the cause was any better than I can without it.

[lamprey]

Do you have another PSU you can  try?

I've never had that particular kind of issue with a PSU, but I have had a couple go bad on me.

Instead of trying to see if it is a heating issue, you could also try something like Memtest86 to see if a memory module has gone bad. Or you an try pulling them out or trying one at a time or something to see if memory is the issue.

[jamesjones626]

if you board or other components are dusty get a can of air and blow them off.  when was the last time you put new thermal grease on the cpu?  if its been a few years get yourself some artic silver and put new grease on it.

[twistedsymphony]

well I ran prime 95 and as expected the machine ran for about 20 minutes and then shutdown... during that time I didn't see any errors or warnings pop up on the screen so I'm assuming it didn't find any?


That doesn't matter though as I believe I've found the culprit... while running the prime 95 test I kept the lid off the case...

everything seemed fine then once I approached the 20 minute mark I noticed that the CPU fan came to a stop... then it kicked back on, then stopped again and then kind of clicked to make a revolution or two and stop, then another revolution or two and then stop and then soon after the machine shut down and while the fan itself felt warm the CPU heat sync was "burn your hand" hot.

Everything inside the case is very clean, I only built this PC a couple of months ago so there hasn't been any chances for dust or anything to get in there.... this fan was brand new when I built it.

I'm hoping the problem is with the fan itself and not the fan controller on the mobo.... though I don't know why the fan would work for 20 minutes and then stop... that's just weird.

[cack01]

Go into your motherboard bios and turn off anything that looks like an auto temp control for the fan (i.e. PWM).

or just connect the fan to a different header.  Most mother boards these days have fan headers for CPU, 2x case, 1 PSU.  You can you any of those.

[twistedsymphony]

I didn't have a spare 70mm 4-pin CPU fan laying around... I did have a couple of 3 pin 80mm fans though so I hooked one up to molex and set it on top of the CPU to keep it cool...

the other one I plugged into the CPU fan connector and set it outside the case just to see if it would stay running.

I ran Prime95 for about 2 hours with no problems and shut it down manually 

the CPU heatsync was cool to the touch the whole time.

Go into your motherboard bios and turn off anything that looks like an auto temp control for the fan (i.e. PWM).

or just connect the fan to a different header.  Most mother boards these days have fan headers for CPU, 2x case, 1 PSU.  You can you any of those.

good tip... I went into the bios and disabled the fan speed controller.

interestingly my 80mm fan apparently does an awesome job at cooling because the CPU temp was LOWER than the "system" temp.

After disabling the fan speed controller plugging the old fan into the connector it spun up but seemed slow. I stopped it with my finger and after doing so it went into "limp mode" like it did before just clicking over a rotation or two and stopping in between. Plugging one of my 3-pin 80mm fans into the connector and it spun up to full speed and stayed their strong.

So needless to say I've ordered a new CPU fan. I would have kept the 80mm but there's no good way to attach it. So I've taken the opportunity to upgrade the CPU fan to a beefy ball-bearing unit.

Thanks for your help everyone