7

My computer is playing some strange joke at me, and I'm at a loss.

Everybody knows that hot temperatures causes many components to become unstable and even cause damage. When I built this machine, I was careful to do a bunch of experiments with different airflow configurations, with CPU-and-GPU intensive benchmarks running, and measuring the temperature at different points to find the best configuration.

Generally, the internal temperature of this machine is very close to ambient temperature (with 2 exceptions: the GPU always runs ~10°C above ambient, and the CPU goes to at most ~12°C above ambient when running CPU-intensive applications).

Now comes the problem: at really cold mornings. At cold mornings (~10°C (50°F) or below), the system shows really strange general instability. Beeps from the BIOS, the operating system displays strange messages about memory corruption and PCI devices that failed to initialize, and then, everything hangs. The computer hangs to the point that all USB devices turn of. Hitting the Reset button blanks the screen but the mainboard fails to proceed to the boot.

I never thought that low temperatures could cause computer instability. And 10°C is not that cold, it is still within operating temperatures of all components as far as I checked.

I'm quite sure it is related with the temperature. There were too many cold and warm mornings that demonstrated that the problem only showed in the cold ones. I'm working around this problem by letting the computer sit at the GRUB menu for some 5-10 minutes before letting the OS start, or blowing hot air inside the case if I'm in a hurry.

I don't know what should I approach next to find the source of this problem. Since it is difficult to reproduce (I need a cold morning, and the computer must be cooled down (powered off for many hours)), I need to plan ahead the experiment, since it stabilizes after a few minutes on.

Machine is an ASUS P5VDC-X equipped with Intel P4 64-bit 3GHz, 2GB of DDR2 RAM, a 500W PSU and an nVidia GeForce 9600GT. Everything else was already disconnected/removed before and the problem persisted, so the problem should be in one of these remaining parts.

Any help is appreciated!

Improve this question
7
  • This is indeed odd. However, you don't need to leave the machine sitting for so long. A CPU will get to its required temp after about a minute, and the disks will warm up out of sheer momentum after around the same time. On a side note, I try to avoid letting hard drives dip below 10°C.
    – user3463
    Oct 14, 2010 at 2:30
  • @Randolph initially, I let the computer sitting for 3 minutes and it was enough. But one time, it crashed even after these 3 minutes, then I raised it to 5 minutes and no crashes since. Since I power it on and go to breakfast, it usually ends with something withing the 5-10 minutes range.
    – Juliano
    Oct 14, 2010 at 2:34
  • Well it's an odd one, definitely. You're probably right about the temp. I've marked your question as a favourite and voted it up.
    – user3463
    Oct 14, 2010 at 2:35
  • I think it may be not related to CPU or the GPU... I'm thinking perhaps some passive components like capacitors or transistors? The mainboard doesn't have any swollen capacitors, but I didn't check the PSU yet. Maybe?
    – Juliano
    Oct 14, 2010 at 2:36
  • 2
    I'll concur that this is weird. I also want to know HOW ON EARTH YOU CAN LIVE WITH A PC IN A ROOM THAT COLD. I am shivering just thinking about it! I think Jared's got the right idea, though...I certainly can't think of any other mechanisms unless this area is fairly damp and you're actually getting condensation, but that would probably cause more problems than just this...
    – Shinrai
    Oct 14, 2010 at 14:25

1 Answer 1

Reset to default
5

I almost wonder if you have a bad solder joint somewhere on your computer. As I'm sure you know, metal contracts when cold, and expands when it warms. I'm wondering if there is a bad solder joint that is pulling away from the motherboard, causing an intermittent connection when cold, but as it warms up, the joint is pressed against the board, creating a stable connection.

Unfortunately, even someone with experience could have difficulty finding the possibly single point on something as large as a motherboard.

If you're wanting to take a look yourself, this guide will give you some idea of what you're looking for.

Improve this answer
1
  • It seems that the problem may be in the power supply. I replaced it but now I didn't have any more sub 10°C mornings to put it at real testing conditions. I disassembled the original PSU and it have some ugly solder joints, although I can't tell how bad they are. I'm still waiting for another cold morning to continue the test. For now, you get the accepted answer. Thanks!
    – Juliano
    Nov 3, 2010 at 3:51

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .