Hardware Problem: Crash at startup - Help!

[Andi47]

Since the last days I experience a strange hardware problem with my desktop PC (a non-oc'ed P4 3400 MHz running Windows XP professional):

When I shut it down overnight and then start it up next morning it seems to start up properly, but suddenly it swithes off.

The switching off may occur everywhere starting from the initial hardware tests to the startup sequence of Windows or even a few seconds after windows has started and Prime95 started to crunch.
After this crash it won't startup again on hitting the power key on the keyboard or the power swith on the front side of the box - I have to switch off power (with the swith on the backside of the box or by unplugging the box) for half a minute and then in the most cases it starts up properly and is happily crunching (P95, ECM or QS) for the rest of the day.

One time the crash occured it sayed "CPU overclock or overvoltage failed" when I retried. I had NOT oc'ed the CPU or changed CPU voltage. Restoring default values in BIOS setup allowed me to start up the computer properly, but did not solve the crash problem.

The computer also sometimes (but not always) crashes when I use the Windows Standby function and "wake" the PC up again next morning - same symptoms, it suddenly switches off.

It doesn't seem to be a CPU temperature problem, since I have got a monitoring program (ASUS AiBooster) in the Autostart folder, and last time it crashed shortly after successful Windows startup, It showed a CPU temperature of ~45°C shortly before crashing. (currently the PC is running msieve since 2 hours at a cpu temperature of constantly 57°C, Vcore 1.39 Volts, CPU Fan 3270 RPM)

Does anyone have an idea what could be the fault?

[sdbardwick]

Some motherboards monitor the CPU fan speed and will spontaneously shut down (or fail to boot) if the CPU fan reports under a certain RPM. One quick and dirty test is to pull the CPU fan header off (instantly reporting 0 RPM) and waiting a few seconds to see if the computer powers down. A quicker and dirtier way is to block the fan from turning with a pen or similar stick-like instrument (but not a pencil - don't want conductive graphite floating about the computer innards) and seeing if that causes a shutdown.

[Andi47]

Originally Posted by sdbardwick

Some motherboards monitor the CPU fan speed and will spontaneously shut down (or fail to boot) if the CPU fan reports under a certain RPM. One quick and dirty test is to pull the CPU fan header off (instantly reporting 0 RPM) and waiting a few seconds to see if the computer powers down. A quicker and dirtier way is to block the fan from turning with a pen or similar stick-like instrument (but not a pencil - don't want conductive graphite floating about the computer innards) and seeing if that causes a shutdown.

IIRC it has crashed at least one time when the Fan was running at maximum speed (After the first crash I believed that it could be a heat problem and switched the fan to max. speed)

Is it possible that the fan sometimes reports a speed of zero even it is running quite fast?

[sdbardwick]

Shoot! That would have been a nice, easy solution.

It is possible for a fan to incorrectly report its speed or for a motherboard to incorrectly interpret speed sensor data. I have one board where the fan speed varies between 0 and 53000 RPM! I haven't investigated if it is the fan or the board; I set the board's BIOS options to ignore fain failures.

It could also be a faulty/failing power supply - I've had several off-brand power supplies behave as you describe, as well a one older Antec which did not like Pentium D processors. One had a bad fan that caused rapid overheating and thermal shutdown, the rest were just cheap ones with a short lifespan.

Sorry I can't be of more help right now; I'll check back later today after I get some sleep (2:35 AM here) and maybe inspiration will strike then.

[paulunderwood]

The temperature of the CPU looks okay'ish considering the time of year.

Can you look in BIOS to read all of your computer's voltages, for 3.3v, 5v, 12v, 12v(sb) etc?

[Andi47]

Originally Posted by paulunderwood

The temperature of the CPU looks okay'ish considering the time of year.

Can you look in BIOS to read all of your computer's voltages, for 3.3v, 5v, 12v, 12v(sb) etc?

Current voltages (PC running since 10 hours), using PC-Probe II to read the values:

Vcore: 1.39 V
3.3v: 3.33V
5v: 4.97V
12v: 11.51V

Using Ai-Booster to reed more values:

Memory: 1.80 V
PCIE: 1.50 V

[paulunderwood]

The voltages look ok.

What is the system temperature?

[Andi47]

Originally Posted by paulunderwood

The voltages look ok.

What is the system temperature?

Current temperature after 10 hours at maximum load:

CPU: 58°C
Mainboard: 46°C

P.S.: I have no problems after the computer is running for a few minutes, ALL crashes occured during or immeadiately after startup.

[paulunderwood]

The temperature difference between the mainboard and the CPU looks ok too.

At boot the system uses a lot of power, spinning up disks etc. It looks like a power issue to me, which could be caused by any number of faulty parts which would require a process of elimination to identify them, but first make sure all connectors from the power supply are in properly, and then, if necessary, start to disconnect things like the disks to eliminate the problem -- in the end it could be the power supply unit or...

Check the memory is seated well and run memtest to eliminate a memory problem...

[MatWur-S530113]

as he comes from Austria I try it in german language

probier mal das:
Systemsteuerung -> Leistung & Wartung -> System -> Tab 'erweitert' -> Einstellungen von 'Starten und Widerherstellen'
Da gibt es eine Option 'nach Systemfehler automatisch Neustart durchführen', stell diese Option mal ab und guck ob es was bringt.

S530113

[Andi47]

Thanks, I switched this option off. The next days will show if it works.