Troubleshooting Cold Start Turn-Offs
A bit of a cryptic title, I admit! I am troubleshooting spontaneous shutdowns of a client's desktop PC that occur predominantly early in the boot-up sequence of a Vista-equipped PC when starting from cold.
The computer shuts down mostly during the first 5 progress bar animations of the Vista boot-up screen, but it can happen later in the boot-up process, up to and including the welcome screen. In almost all cases, though, this is only when starting the computer from cold (after it has been off for about 3 hours or more).
Basic hardware specs are:
- eMachines E4252 desktop (out of warranty).
- Motherboard: Abit NF-M2S / Acer EM61PM/EM61SM / ECS GeForce6100SM-M(2) / Packard Bell MCP61SM-AM.
- CPU: DualCore AMD Athlon 64 X2, 2100 MHz (10.5 x 200) 4000+
- Video: NVIDIA GeForce 6100 nForce 405 (128 MB).
- RAM: 2GB DDR2-800.
- Vista Home Premium SP2.
I have tried, all without success:
- Monitoring CPU temps: both cores run 35 - 45C (acc. to AIDA64).
- Replacing 300W PSU with new, 380W PSU.
- Testing with minimum peripherals (monitor, keyboard, mouse).
- Replacing RAM module, testing with MemTest86+.
- Disconnecting internal peripherals in turn: media card reader, DVD drive.
- Full HDD diagnostics (HDD, surface read tests, manufacturer diagnostics).
- Testing with PCIx video card (in place of the integrated nVidia).
- Updating Vista drivers for video card, LAN and WiFi NICs, etc.
- Different BIOS settings, particularly power=related (such as turning off AMD K8 Cool&Quiet Control).
I currently suspect a power-related issue with the mobo (faulty capacitor?), but I have admittedly not yet tried re-applying thermal paste to the CPU (since operating tempos are fine, and issue occurs on cold starts). Is there anything else you can suggest before I notify the client that the mobo needs replacing or otherwise advise investing in new hardware?
The computer shuts down mostly during the first 5 progress bar animations of the Vista boot-up screen, but it can happen later in the boot-up process, up to and including the welcome screen. In almost all cases, though, this is only when starting the computer from cold (after it has been off for about 3 hours or more).
Basic hardware specs are:
- eMachines E4252 desktop (out of warranty).
- Motherboard: Abit NF-M2S / Acer EM61PM/EM61SM / ECS GeForce6100SM-M(2) / Packard Bell MCP61SM-AM.
- CPU: DualCore AMD Athlon 64 X2, 2100 MHz (10.5 x 200) 4000+
- Video: NVIDIA GeForce 6100 nForce 405 (128 MB).
- RAM: 2GB DDR2-800.
- Vista Home Premium SP2.
I have tried, all without success:
- Monitoring CPU temps: both cores run 35 - 45C (acc. to AIDA64).
- Replacing 300W PSU with new, 380W PSU.
- Testing with minimum peripherals (monitor, keyboard, mouse).
- Replacing RAM module, testing with MemTest86+.
- Disconnecting internal peripherals in turn: media card reader, DVD drive.
- Full HDD diagnostics (HDD, surface read tests, manufacturer diagnostics).
- Testing with PCIx video card (in place of the integrated nVidia).
- Updating Vista drivers for video card, LAN and WiFi NICs, etc.
- Different BIOS settings, particularly power=related (such as turning off AMD K8 Cool&Quiet Control).
I currently suspect a power-related issue with the mobo (faulty capacitor?), but I have admittedly not yet tried re-applying thermal paste to the CPU (since operating tempos are fine, and issue occurs on cold starts). Is there anything else you can suggest before I notify the client that the mobo needs replacing or otherwise advise investing in new hardware?
Lukasz Chmielewski
Did you try safe mode ?
Sounds more like an O/S issue. Any blue screen? Safe mode as Roads suggested? Any suspicious errors in the system or program error logs? Tried a chkdsk/f?
ASKER
'tis the same in Safe Mode - but I have forgotten to turn on boot logging, which I have now done. Given the nature of the problem, I need to leave the machine off for a few hours before testing, so things move very slowly.
I have tried a couple of startups using a diagnostic WinPE build on CD, and they have been fine. Then again, the whole boot process is different for WinPE, with different hardware initialisation, so I'm not convinced it's a valid test.
It could be an OS issue, but I suspect not at the moment. There are no BSODs, chkdsk always comes up clean, and there is nothing of interest in the event logs. Vista's event logging doesn't get initialised until after the problem occurs, so I wouldn't expect it to be of use. On the rare occasion the machine turns off late in the boot process, there are no error events, just the "system shutdown unexpectedly" event on subsequent startup.
I have tried a couple of startups using a diagnostic WinPE build on CD, and they have been fine. Then again, the whole boot process is different for WinPE, with different hardware initialisation, so I'm not convinced it's a valid test.
It could be an OS issue, but I suspect not at the moment. There are no BSODs, chkdsk always comes up clean, and there is nothing of interest in the event logs. Vista's event logging doesn't get initialised until after the problem occurs, so I wouldn't expect it to be of use. On the rare occasion the machine turns off late in the boot process, there are no error events, just the "system shutdown unexpectedly" event on subsequent startup.
BSOD would rather indicate the issue with the hardware.
On the other hand I have never seen BSOD related to mobo. Seems like it might be an OS problem though.
On the other hand I have never seen BSOD related to mobo. Seems like it might be an OS problem though.
check for bulging or leaking capacators anyway. i posted an example
caps-800.jpg
caps-800.jpg
My first thoughts would be a faulty PSU, but you've got some pretty ugly capacitors. Replacing those should get you up and running again. It should cost you no more than $20 for caps and about 30 minutes of soldering.
ASKER
My first thoughts would be a faulty PSU, but you've got some pretty ugly capacitors.
Hey, let's not get personal now <grin>. Anyway, they're not mine they belong to nobus. I have made an initial examination of the capacitors and can't immediately see anything amiss. As I mentioned, I did swap out the PSU - the original is a 300W supply, which I replaced with a 380W PSU for testing, without success.
I have also now tried starting Windows in diagnostic mode (i.e., with a minimum of drivers and services), but the problem persists. It also persists with Safe Mode. Boot logging has been unhelpful, too - on failed startups no boot log is saved at all. In most cases, with boot logging enabled I can see the on-screen logging of the initial driver set, then the display switches to an OS Info page with a few seconds pause before a drive listing indicating that all drives are 'clean'. Mostly, the computer turns off during the mentioned pause, but sometimes it gets further in the boot process, up as far as the login screen before turning off spontaneously. In the latter case, the event logs show no errors (except, as I reported earlier, the event on subsequent restart that the previous shutdown was unexpected). Remember, all this happens only on cold boot. Sometimes it can happen a few times before successful start. Once started, the machine perfoms faultlessly.
ASKER
Further testing has confirmed the issue is definitely related to ambient temperature, so I remain convinced for now that it is a hardware problem. The errant compuuter is currently back with the client, who is using a warm air fan to warm things up in the morning to ensure a good start up!
could be a cracked wired, or bad solder point
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
that's fine - if it stays so; but at least, it is not to be considered a trustworthy PC
So be prepared to loose your data --> Backup !
So be prepared to loose your data --> Backup !
ASKER
that's fine - if it stays so; but at least, it is not to be considered a trustworthy PCIndeed, I have already made the client aware of this, and she is backing up regularly.
So be prepared to loose your data --> Backup !
ok - then all is fine..don't forget to close this Q then
ASKER
No proper solution found, but client has workaround.