Windows won't boot from cold (temperature-wise)

[silverman166]

Old PC, but hitherto very reliable, AMD Athlon XP 2000+, 2.5Gb RAM, 40Gb ide drive, clean XP Install on 10Gb partition (no progs loaded yet), dvd-rom drive, nVidia basic graphics PCI card, PCI sound card, PCI USB2 card. PSU renewed about a year ago. CMOS battery new.

This problem only occurs from cold. On power-up, the POST does its thing, and the Windows XP logo screen (with blue flashing dashes moving across a box) comes up. At this point, nothing happens, except the dashes continue crossing the box, for at least 5 minutes. If I use the reset button at this stage, the POST comes up again, and after the drives have been detected, a message appears saying that 'Windows did not start successfully', and asking me to select from Start in Safe Mode, or Start Normally. If I select Start Normally, XP indeed starts OK this time. But if I select to Start in Safe Mode, it doesn't start it.

Once Windows XP has started, and I shut it down, then power up again, XP proceeds to start successfully, first time.

So, could this be some sort of component problem or electrical spike from PSU or other?

(The 40Gb drive was originally used as a backup, and had no OS, and boot up was from an 80Gb hdd, partitioned in two equal parts, keeping OS and progs on one with Data on the other. This drive was in use when the problem first appeared, and has been completely disconnected for the purpose of eliminating it as a possible source of the problem. This drive had also undergone a clean install on the original OS partition, reformatting as part of the process, so no 'double OS' conflicts there).

All PCI cards have been removed and contacts cleaned, as has the interior generally.

I perhaps should also add that after this problem first appeared, I put up with having to reset for a while, until one occasion when, after a few minutes work the PC crashed, starting a lengthy investigation up a steep learning curve, eventually requiring the use of System Restore. I might also add that the POST restarted of its own accord even before the XP logo screen appeared until recently - it is only since yesterday that for some reason I've consistently reached the XP logo screen scenario.

I know this is a long post, but hope I've covered everything!

Hope someone can help with this.

Thanks.

[Xlphos]

hi,

If you are thinking heat problems did you reapply any thermal paste to the CPU? Perhaps new paste is required if all other components are good. The old paste may not be once what it was...

- Let us know how it goes

[silverman166]

Thanks for that. But in the event of poor themal contact with cpu, wouldn't the reverse situation apply ie wouldn't there tend to be more problems as the PC warmed up, rather than from cold?

[rubberman]

Thanks for that. But in the event of poor themal contact with cpu, wouldn't the reverse situation apply ie wouldn't there tend to be more problems as the PC warmed up, rather than from cold?

Probably. Is this from a thermally cold situation? Or is this also from normal room temperature?

[rubberman]

If from normal room temperature, then it is likely that there is a problem with some component where when "cold" there is a broken contact, but upon some increase in temperature, the expansion of material will make the contact again, resulting in a functioning system. Analysis? There is a physical problem with the system, and finding it may be next to impossible. My advice? Do what you are now, until it stops working altogether.

[silverman166]

Thanks, Rubberman - sorry not to reply sooner, but I've had a problem accessing the website from my laptop most of today for some reason.

The problem does occur at normal room temperature, but only when the PC has been shut down for quite some time - difficult to define exactly, but I've tended to leave it shut off for several hours before re-trying after making a change.

You've come to the same conclusion that I have reached, after a lengthy elimination process. I've still a nagging suspicion, just a gut feeling more than a reasoned approach, that the PSU may be at fault, so I may still try to have a diagnostic done, if not too costly.

At the moment, though, it's just an inconvenience more than a problem, which could continue for a long time or fail tomorrow. Only time will tell.

[gerbil]

Things I would suspect....
Memory. Get memtest86+ from .org, load it to some medium and boot cold from that. The site has informative reading. Run it for half an hour at least; you don't want even a single error.
You mention POST restarting by itself [before XP starts loading] - this points to a power problem - if some mb voltages are seen to go out of range [a warm-up voltage swing] a cpu reset signal is generated > your sys restarts. Check your PSU voltages from cold [it must be connected to a load [eg. the mb] to work at all; a tech will have a device which when plugged in will give readings of all voltages... a minute's work]. He won't see a spike with it, though. And it could be the monitoring chip on the mb hiccuping.. easiest is to swap out the PSU for a check.
But the PSU scenario doesn't sit well with the XP driver loading screen running on now.... and that SafeMode entry failure points to a registry problem, or failure to load/read the hive properly > hdd failure. Run a diagnostic [bootable] from your hdd manufacturer's site.
The most basic starting componentry you can go with is one mem stick, cpu, psu, mb [and video card if no internal graphics on the mb] - add pieces [with power off] from there to test : hdd, video, other pci, kbd.
Could be a chip on your mb or the cpu not happy when cold... warming up restores some broken internal contact [thermal shock], or else otherwise brings it inside its operating parameters - this sort of thing can last forever, or fail totally next week.

[silverman166]

Sorry not to have replied sooner, gerbil, but I did find your post very useful.

I took the PC to a specialist who eventually diagnosed a faulty motherboard - there are a number of capacitors which are showing signs of going/being faulty, in that the ends caps are swollen, where they should be flat (due to gas build up internally).

This could last a while, or completely fail at any time.

I may yet be tempted to replace the dodgy ones myself in due course, but its good to have a reason for the problem at last.

My thanks to you, and all other contributors, for your interest and suggestions.

[rubberman]

Yes, it seems that bad capacitors are a frequent source of strange system behavior these days. I think that is one place where system builders are scrimping on costs.