Random crashes, no POST.

[koder]

Hello,

I'm experiencing very strange behavior with my box:

Epox 4sda5+
Intel Pentium 4 @2.4GHz
2x 512MB of Ram

It freezed with no apparent reason - and didn't reboot afterwards. Motherboard is equipped with POST-indicator (a two digit counter), and it won't even light up - just like machine was not running POST at all.

I have performed following:
* Checked if motherboard is getting power. Coolers spin, power led is on.
* BIOS reset. No change.
* Checked PSU - all voltages were ok.
* Replaced BIOS battery. No change (was almost dead, tough)
* Tried to re-boot after some hours, allowing whole system to cool down - no change.
* Examined condensators on motherboard - all looked fine.
* Performed bare-boot (motherboard out of chasis, on cardboard box, with only CPU and one RAM installed). No change.

...at this point i thought that CPU or Motherboard went down. However, i did perform an additional test:

* Removed Heatsink, cleaned, re-applied silicon.

...and machine went up normally. I re-inserted it into chasis, plugged everything, and got it to boot and it runned fine for few hours. Unfortunately, after that everything repeated itself - a freeze and no reboot.

It seems like machine works for a time - ranging from few seconds to few days - if i'll remove and re-install Heatsink. However, i did not notice anything unusuall with temperatures - it works at 40' in idle, 55' in full load, and load seems to have no impact whatsoever on crash frequency.

I did notice however, that motherboard is bending a little when i re-install heatsink. Also, even slightest movement of the chasis causes instant crash.

The fact that machine at times seems to work fine (i even got it to work for few days straight) seems to indicate that most of components are working fine. At this point i'm suspecting micro-fracture on motherboard (altough close examination did not reveal any visible damages) - that would explain why movement and applying force to it (clips on heatsink) have impact on stability.

But before that i wanted to get a second opinion. Any thoughts on what else to check? TIA.

[Donut]

Have you tried running it for a while outside the chassis? If it runs fine I'd be looking for a mobo standoff thats shorting out. Also what are you using for thermal grease on your heatsink? If my Pentium D hits 55', it's time to clean out some severe dust.


After reading a little further, what do you have for a case? Plenty of airflow? Also hat do you have for a heatsink?

[koder]

* Performed bare-boot (motherboard out of chasis, on cardboard box, with only CPU and one RAM installed). No change.
Yes i tried running it outside chasis. It does not help.

I do use a stock heatsink for P4, with cheap thermal grease and noname case, but this machine was running for some years without any problems. In fact, looking back at things, i think problems arose first time when i blowed the dust of heatsink. However, i have same problems when CPU is cool (~25', as measured with external thermomenter, after hours of downtime).

And it does not take any time - after power-on fans start spinning but machine remains idle. If this was an conituous behavior i would just assume that motherboard is broken - but at times (ie. after i move heatsink a bit) it boots fine and works, raging for few second to few days.

[Your Amish Daddy]

Sounds like the socket may be pulling away from the board...

[koder]

Sounds like the socket may be pulling away from the board...

Could you be please more vebose? How to verify that it is indeed the problem?

If i'm not mistaken, a loose CPU socket means "get a new mobo" :/

[Your Amish Daddy]

Well, That's all the verbosity there is for that. If it's pulling away, yeah it's time for a new board. Only way to really check is to get the machine running, and lightly lift the heatsink. Not much force, just enough to lift it off of the surface. If you can slightly turn it and the machine goes awry then that's your culprit.