Computer is shutting down - Please help!

[Twoyboy]

Having issues with my computer. It started months ago with the computer occasionally shutting down for no reason. I'd just be browsing the net and it would turn off and then restart. And not shut down nicely, shut down like the power cord was pulled. I work FIFO and it would happen maybe 4-5 times my first night back from work, and then very rarely for the rest of my break. I checked my CPU and MB temps and they were fine. I opened it up and cleaned out the dust. I ran disk checks, virus scans, everything and got no where. Eventually I just put up with it, mostly because it NEVER happened while I was playing a game, working on a file, watching a video, etc, only when I was browsing the net (or very occasionally doing nothing at all, like I wasn't even in the room).

Then a few days ago I decided to play Heroes of Might and Magic 6 again. Because I haven't played it in so long it had to update from v1.2 to v5.2 or something. At almost every stage of update, it shut down either just at the start of the install or just at the end of it. The game actually played a few times in between, but then next time I opened it it would start the next update and do it again. At about 5.2 it shut down and then wouldn't play the game, install the update or uninstall the game. Now, when I try to uninstall, it shuts down. Every time. During my trouble shooting I ran chkdsk, and it shut down. Chkdsk will make it shut down almost every single time, except in safe mode. It also occasionally finds errors and occasionally doesn't.

I've tried removing each RAM module, but it still did it. I tried turning off every service in the msconfig settings, but, while it seemed to be a bit more stable, it still did it.

Any ideas?

[White Haven]

The first thing you want to test in cases of unexpected cutoffs is the power supply. Most common cause for that sort of failure, although far from the only one. As an added bonus, it's a relatively easy thing to test.

[Twoyboy]

The first thing you want to test in cases of unexpected cutoffs is the power supply. Most common cause for that sort of failure, although far from the only one. As an added bonus, it's a relatively easy thing to test.

Is there any way to test it without buying another power supply and swapping them over? I would have done that already except the slightly non-random nature of the shut downs (ie every time I run chkdsk or try to update HOMM) led me to believe it was likely to be software related (especially since I have not been able to make it shut down in safe mode). But I will probably do that later in the week if the problem is not sorted by then.

[Vendetta]

Hard power offs are almost never software, they're either caused by heat or power supply. Software errors cause reboots not power off.

You could try swapping the power lead to your hard drive for a different one from the power supply, but if this is happening during disk access it's most likely to be power supply or hard disk related.

[Haruko]

Just to be sure, when you cleaned out the dust you also scraped out the dust embedded in the heat sink for the CPU? When I cleaned that out it got rid of my own restart problem.

[Executor32]

To test it without a spare power supply, you can get a power supply tester for rather cheap. We use this one at work, which even has an LCD screen to tell you the exact voltage being output on every rail. It hasn't been wrong yet.

[Edi]

Use Crystal Disk Info to check the hard drive for faults. If the warning comes up yellow and there is a reallocated sector count greater than 0 or uncorrectable sector count greater than 0, the hard drive is going. Unless you have automatic restart disabled in Windows, a blue screen crash will look like a hard power reboot and a dodgy hard drive will cause blue screens all the goddamn time.

I had that problem in the summer and the hard drive went from "might be a problem" to "will die inside a week" in just one month.

[Twoyboy]

Thanks everyone. I'm at work on site now but I'll have another look on Wednesday when I'm back.

Just to be sure, when you cleaned out the dust you also scraped out the dust embedded in the heat sink for the CPU? When I cleaned that out it got rid of my own restart problem.

I used a clean pastry brush and brushed it out as best I could. Should I take it off and clean it?

Use Crystal Disk Info to check the hard drive for faults. If the warning comes up yellow and there is a reallocated sector count greater than 0 or uncorrectable sector count greater than 0, the hard drive is going. Unless you have automatic restart disabled in Windows, a blue screen crash will look like a hard power reboot and a dodgy hard drive will cause blue screens all the goddamn time.

I had that problem in the summer and the hard drive went from "might be a problem" to "will die inside a week" in just one month.

Thanks. I'm still leaning towards other things before power supply, so I'll try this first and then test the power supply after.

[Twoyboy]

Use Crystal Disk Info to check the hard drive for faults. If the warning comes up yellow and there is a reallocated sector count greater than 0 or uncorrectable sector count greater than 0, the hard drive is going. Unless you have automatic restart disabled in Windows, a blue screen crash will look like a hard power reboot and a dodgy hard drive will cause blue screens all the goddamn time.

I had that problem in the summer and the hard drive went from "might be a problem" to "will die inside a week" in just one month.

The health status says good. The reallocated sectors count says Current: 100, Worst: 100, Threshold: 36, Raw Values: 000000000.

Uncorrectable errors says the same except the Threshold is 0.

Is that fine? If so, I'll grab a power supply or a tester later in the week.

[White Haven]

I do have to somewhat contest PSU testers, not because they're garbage or any such, but simply because of a personal rule I've evolved through years working in the business. I never trust a diagnostic that tells me something is good, not fully. If a diagnostic test flags something as bad, fine, it's found a problem. If it's 'good,' all that really means is that it hasn't found a problem, not that there isn't one. I've seen depressingly many power supplies that tested good but were...not.

[Edi]

Use Crystal Disk Info to check the hard drive for faults. If the warning comes up yellow and there is a reallocated sector count greater than 0 or uncorrectable sector count greater than 0, the hard drive is going. Unless you have automatic restart disabled in Windows, a blue screen crash will look like a hard power reboot and a dodgy hard drive will cause blue screens all the goddamn time.

I had that problem in the summer and the hard drive went from "might be a problem" to "will die inside a week" in just one month.

The health status says good. The reallocated sectors count says Current: 100, Worst: 100, Threshold: 36, Raw Values: 000000000.

Uncorrectable errors says the same except the Threshold is 0.

Is that fine? If so, I'll grab a power supply or a tester later in the week.

That's just fine. The raw values are the ones you're looking at, because if the reallocated, uncorrectable or pending sector count raw values are something other than zero, the program would give you yellow warning signs.

If you have very high read error rates and seek error rates, that may speak to some potential problems, but in and of themselves they mean nothing even if the value is high. Only if coupled with dodgy behavior and then only tentatively. So you should focus on other aspects first. Check that the machine isn't overheating, test the memory cards etc.

[Twoyboy]

Right, so I ran the scan, and it was fine. So the other day I thought I'd test the power supply idea by swapping the power supplies between my wife's computer and mine. I took mine out and went to take hers out and everything in her case is cable tied together, I can't take out the power supply without taking out almost every cable or cutting the cable ties. I decided I didn't want to do that so I put my computer back together, after taking the opportunity to take the CPU fan off and brush and clean it out. All I changed was the power cord, I thought I'd try swapping it with a spare.

And the problem was fixed! Game installed and updated fine. No issues at all! So, I assume either I had some loose cables inside and taking it out and putting them back fixed it, or the power cord was dodgy.

And then... it shut down as I was about to start playing the game. But this one was not a mystery, this one quite clearly told me my CPU was overheating. I think when I took the fan off I ruined the connection between the heat sink and the CPU because I cracked that thermal gel stuff. So then I had one of my mated help me put some of that back on and was back in business, no more shutdowns the past two days.

Of course, now my modem is playing up and crapped itself 6 or 7 times yesterday, which is why I didn't post earlier, but that seems ok now... so just waiting for the next thing to crap itself now. Thanks for everyone's help and if it starts happening again I will just replace the power supply straight up to save time.

[Executor32]

Yeah, if you remove a heatsink, you should always* clean off and reapply the thermal paste before putting it back on, if possible. Otherwise, as you found out, it can adversely affect the heatsink's ability to transfer heat from the CPU/GPU/whatevs.

*You might be able to get away with it if the thermal paste is fresh, but I'd rather be safe than sorry.