Computer: Why is my computer turning itself off?

Sometimes, I think I’d be better off just getting a computer from an OEM. Today is one of those days, as I’ve been having (yet again) issues with my home-built system.

This time, the computer is turning itself off without warning. As in, I’ll have been happily puttering about for several hours, and all of a sudden the computer is completely powered off without going through a Windows shutdown. One second on, the next second, off.

Looking at the Event Viewer System log made me think this was a problem with the UPS, since errors generated by that were showing up saying that it wasn’t configured correctly with a time stamp immediately prior to system shutdown. I fixed that by moving the USB cable to a different port; you’d think it wouldn’t matter, but apparently the software prefers that the UPS be plugged into USB 1 or 2, rather than in USB 3 or 4. That UPS event isn’t showing up since I moved the cable.

The computer is still shutting down, though. Turning it back on usually but not always gives me several more hours. The exceptions are when it’ll power up, load the OS, and get partially through setting up my personal settings before powering off. I’ve noticed that when it’s done this, the POST messages will not list the CPU or RAM size. (Twice it even powered off before Windows finished loading; when I turned it back on, it went through the CHKDSK program (first on the NTFS boot drive, then on the FAT32 spare drive) and fixed some index entries.)

I thought I’d paddle through the BIOS to see what could be seen. I found a BIOS menu for “PC Health”, which lets me set an alarm for CPU temperatures exceeding 60, 70, 80, or 90 celsius. It had been disabled; I’ve been using a freeware program called Motherboard Monitor (MBM) to keep an eye on system temperatures and that seemed to indicate that everything was fine. Even under load, the CPU was, according to MBM, running around 40 celsius. The BIOS, however, is disagreeing. I turned the CPU Temperature alarm on and worked my way up the temperature points. The alarm is still going off even when it’s set to 90 celsius, which means that despite the liquid cooler system I put in, the CPU may be running dangerously hot. (I’ve since turned the BIOS monitor back off, since that was the only way to get peace.)

I’m not venturing a guess as to why the BIOS is claiming the CPU is running at a higher temperature than the software that is monitoring the CPU’s temperature sensor.

I haven’t been able to find info on safe operating temperatures yet. And, of course, there’s no guarantee that the problem IS an overheating CPU, since quite often I can simply turn the machine back on and have several more hours without issue.

Saturday, I replaced the power supply, going from a Thermaltake 560 watt to a Thermaltake 650 watt, on the hypothesis that a computer losing power may have a problem with the power supply. (Also, I seem to have problems with power supplies, anyway. They have to be the most frequently replaced part on any of my systems.)

This hasn’t helped, as the computer shut itself off again last night after about 2.5 hours of use. Note: it had run for 3 or 4 hours each on both Saturday and Sunday without any sign of a problem. Also, I shut the computer down completely when I’m not physically sitting at it since it is no longer in a room where I can shut the cats out (cats + keyboard = interesting internet experiences, sometimes).

I’m guessing a trip to a local PC shop is in order, as I’ve pretty much exhausted my limited store of hardware geekery. I’m not at ALL happy about this, so any suggestions of causes or fixes would be appreciated. Also, recommendations of good repair places in the Pottstown, PA area.

HW Info:[ul][]Gigabyte Tech GA-K8N Ultra SLI[]AMD Athlon 64 XP (Dual Core) 3200 or 3400 (I forget which)[]2 Hard drives[]2 CD/DVD R/W drives[]1 Floppy Drive[]Thermaltake Toughpower 650W Cable Management (the new power supply)[]Thermaltake Silent Water Liquid Cooling System[]2 (at the moment, until I remember the trick for putting in number 3) chassis cooling fansATI Radeon 8500 video card (unpowered)[/ul]Computer, monitor, and cable modem are all plugged into an APC Back-UPS RS.

The only times I have ever had problems like that have eventually turned out to be bad memory or overheating. I think most likely heat is the cultprit. Chkdisk is probably running just because windows had the plug yankd on it in mid-write.

Have you tried memtest or similar to test the memory, or replaced with known good RAM? Also try turning down your memory timings to something a bit slower than you have it currently.

As for heat, you mention liquid cooling. Are you sure everything there is working smoothly? An air bubble in the system or similar could cause occasional overheats as you describe. Nothing in life is totally reliable, and it may be that one of your sensors is broken, so you need to check yourself. Can you get your fingers in to check how hot things are actually, rather than relying on the sensors? If you burn your fingers, you definitely have problems. I have had a very hot waterblock before due to random coolant flow issues. I’d also suggest sticking in a bog-standard heatsink for a while to see if that stabilises things - that would isolate the problem to the water-cooling rig. How much arctic silver or whatever did you apply to the CPU?

Interesting. It might be a cpu overheating problem but that doesn’t normally cause a shutdown – normally the computer would hang or restart. Further I’m doubting your CPU runs at 90 – that’s pretty hot – but if the bios thinks it is that might be causing other problems. If it’s running at 90 you’ll be able feel the heat coming off the fans and the cooling system (or if they’re not working properly you’ll be able to feel the heat off the cpu if you pop the case once it’s shutdown – I don’t recommend getting to close to the cpu or the cooling system if it is running that hot ! )

So it sounds like your UPS or the computer if flicking the power switch for some reason. Certainly I’d take the UPS out of the system and try running it plugged directly the mains. *If it was me * I’d run it with the case open so I can check the cooling is working, get a feel for the heat it’s generating and see if pointing a desk fan at it makes any difference. However running the computer with the case open isn’t the most sensible thing you can do (especially with cats around) so it’s up to you. Also keep an eye of airflow and heat through the UPS on any other part of the system.

Some stupid questions, you say it shuts down after a few hours of use. It doesn’t seem to shut down at the same time every day ? Does it matter if you’re hammering the CPU (i.e. gaming, graphics) vs just surfing the web ?

I haven’t tried testing the memory, no. I don’t really know how to adjust the memory timings, though. I’ve never been an overclocker, so whatever the BIOS sets up is what it gets.

I’m not sure about the liquid cooling, no. The hoses are a bit bent in a couple places, and that’s worried me. They came that way out of the box. The copper sink on top of the CPU is hot to the touch, but it’s more of an uncomfortably hot water hot than a scalding AIEEEEE hot, if you know what I mean. I used enough thermal compound to thinly cover the top of the CPU; the directions specifically said not to apply it thickly.

The fan for the cooling system is not blowing out particuarly hot air, which is one of the reasons I’ve wondered if it was working correctly. On the other hand, I’ve had this system together since late February/early March, and until I moved it was in a room that I kept the cats OUT of, so it was always on. I’d have thought a problem like this would have shown up sooner.

I’ve had the cover off for the last week or so while trying to trouble shoot this. It’s in a basement, so it’s already pretty cool down there. There’s no perceptible heat being generated in the case; the only warm spot is the actual copper sink on top of the CPU, and the heat dissipates fast enough that you have to actually touch the sink to notice the heat.

While I’ll often have several programs open at once, they’re not usually CPU intensive (I think). The time seems to vary, but I’d have to start tracking it to give you a definitive answer on that. What is USUALLY happening is that I have Firefox open in the background and a game from Reflexive (Big Kahuna Reef 2) being played. I’d hesitate to say either individually or in combination is the cause, though. I think they’re just what I happen to be running when it happens because they’re what I run a lot.

I’ll see if I can dig up the heatsink that came with the CPU and try switching it out for a bit. I hate working with the CPU, it makes me terribly nervous (all those little pins that you Must Not Bend). sigh I really, really didn’t want to have to go through a gazillion boxes to find that heatsink.

So either you’re cooling system is working perfectly, or all the heat is remaining trapped in the CPU. I’d be inclined to trust mbm as I don’t think a CPU would survive running at 90 (someone will correct me if I’m wrong).

Apart from that I can only suggest the sensible things, double check all the internal connections, have a peer around and see if stuff looks dirty or if there are stray cat hairs. If you can move it again try it in a different room/place. My feeling is that it’s something physical – maybe something got shifted when you moved it, or there’s a stray screw in the case somewhere. It’s also worth looking at the power/reset switches in the to make sure they’re connected and look in one piece – and there’s nothing sticking. On one of my machines the power button can stick in various positions which could maybe cause your problems.

But really I think you’ll need to start testing the components, either yourself or by taking it somewhere to get looked at.

Gah. you’re = your in that last post. To prove I know the difference – all other mistakes are genuine.

Yeah, that was my thought. I’ve been trying to find information regarding safe operating temperatures, but no joy so far. On the one hand, 90 degrees celcius seems a bit extreme; on the other, the BIOS included it as its top safe operating temperature.

I suppose I could try moving the computer into the dining room for a while. My other choice is to find someone willing to disassemble and shift a computer workstation up two flights of stairs and then reassemble it. :eek:

Lots of dust accumulating on the fan blades, fer sure, and the compressed air is not getting it off. A damp paper towel, maybe? I’ll wiggle a few things to see if something falls out and check to make sure that the wires from the case to the motherboard are firmly in place (I hadn’t thought of that before).

'tis my darkest fear. I’ll try the other stuff first. Those little wires from the case to the motherboard are so fiddly, I wouldn’t be surprised if one **had ** come loose.

(Forgot to say in response to an earlier comment: The UPS is sitting on the floor with a good bit of airflow all around it.)

Hmm. Shame that system has no flowmeter. Hose kinks can cut the flow down a lot. 90c is OUCH hot, but if for some reason the heat isn’t transferring CPU-> waterblock then it’s conceivable a small throughflow of water would keep the system in a state where the CPU is smoking, but the block is at 50c-60c. I wouldn’t have expected that to happen if you have a good contact with a thin film of decent compund though. Odd. And yes, 90c is the kind of temp I’d expect to kill a CPU pretty comprehensively.

A well-working cooling system shift the heat to the radiator and out very efficiently, so the airflow won’t get particularly hot. Does the radiator get warm at all?

I second the suggestion of running it without the UPS - remove as many factors as possible and try to isolate the problem.

Oh, and Thermaltake is overpriced junk with poor build quality IME. I own(ed) one of their cases, some of their fans and one of their early water systems, which drained itself all over the bottom of my case after developing a hole in one of the radiator barbs. Fortunately no damage was done.

One more thing - a quick google shows people have odd problems like this occasionally. Have you flashed your BOIS recently?

The BIOS has never been flashed, but it’s also a more recent BIOS than the one referenced at the link. I’m a bit leery of doing so. Of the 3 times I’ve flashed BIOS in the past, 2 were successful and 1 was so unsuccessful I had to replace the motherboard.

BTW–I do say ouch if I touch that heat sink, but as I said it’s the “really hot tap water” ouch, not the “I’ve just touched a hot stove element” ouch. Mind you, once I identify “ouch” I remove my finger, so my perceptions may be slightly off.

Hmmmm. I seem to recall that tap water at about 60c or so will scald you in a few seconds, coffee is usually served at about 70-80c. If it’s too hot to keep your finger on for a slow count of ten, it’s probably too hot for your CPU to live at that temperature. How warm does your radiator get relative to the heat sink/water block? Do you have a thermometer or temperature gauge you can deploy to get an accurate reading of the water block?

I’ll have to check when I get home. If by radiator you’re talking about the pump and fan assembly attached to the case, that part is not at all warm. I can comfortably leave my hands on it. The copper heat sink on the CPU, not so much. Definitely no longer than a fairly fast count of 2, as I recall. Again, I’ll have to check.

I don’t think I have a thermometer I can use for this. What I have is a digital oral thermometer (ooo! kinky!) and a digital indoor/outdoor thermometer. I’ve been trying to think of something I could pick up that would work for this use, but no bright ideas so far. I may drop by the CompUSA just up the road from work this afternoon to see what they have to say.

I’m about ready to hand it over to them for repair, anyway. The frustration, it burns, as does the ignorance. I’m moderately savvy, but I have a feeling this has gone beyond my knowledge, skill set, and toolbox. sigh Dang it. Maybe I should take some PC repair classes . . .

That’s not good. This does sound like the CPU is too hot.

The tubing should not be kinked at all. Water should be able to flow freely through the system. The reason you can’t feel any heat coming from the radiator is because there’s not enough flow to bring the heat to the rad. Even if you have to take the watercooling system off and refit the tubing, proper water flow is essential - otherwise you may as well just put regular heatsinks on the CPU.

The hoses are basically crossed, sorta kinda, which has put bends in them. thinks You know, when I get home, I’m going to take a picture of the dang thing and post a link to it. It’s hard to describe. I was worried about that when I saw it come out of the box that way, but the only way I could see to fix it was to detach and reattach the hoses, something the manual specifically said not to do because of loss of fluid from/introduction of air into the system.

Meantime, the cover stays OFF the case while I do some back-up and stuff tonight; tomorrow I’ll take it into the shop. I’m not messing around with this; a new CPU/motherboard are too damned expensive.

I’d love to see pictures. Hope you can get it straightened out without the help of a shop.

I’ve watercooled for a few years, and whenever I had a problem with sharp corners in the tubing, I put a worm-ride hoseclamp over the kinked corner to smooth it out.

Looks like these: http:\www.plumbingworld.com/marinegradeclamps.html
I think they’re like 75 cents each at a hardware store. Just match the outside diameter to the range of the clamp.

Could there be some sort of blockage in your cooling system other than the kinks? Some water additives gunk things up over time (Water Wetter especially), and any loose particles can build up in the pump over time. Last time I cleaned out my water cooling system, there was some crud built up in the pump that was making it run a bit rough. I think it mostly was a combination of the water additive, metal particles from my crude DIY waterblock, and insoluble gunk like maybe some leftover flux from soldering.

Oh, and just to rule out bad memory, run Memtest86 like slaphead suggested. Like everyone else mentioned, when I’ve had problems like this it’s because I’ve overclocked too far or things are overheating. Hot waterblock but cool radiator also sounds bad; everything in my cooling system is lukewarm to the touch.

Could also be a spotty power supply. If you used a cheap one, or something that came with the case, there’s a good chance it’s dying.

I’ll run the mem test, just in case. But as I told slaphead, I don’t overclock at all. The power supply is brand-new, installed on Saturday, so unless it’s a lemon it should be ok. Hell, the old one may well be ok, for all I know. Replacing the power supply tends to be my default “fix” when I start having power problems.

Pictures (hopefully) tonight.

Many exotic otherwise untraceable errors can be tied to a bad PSU, no matter what the price tag. I use cheapos and have not had any problems/complaints/or warranty calls on the 20 or so I have installed.