Why does my new computer randomly shut off?
December 1, 2007 8:32 PM Subscribe
My husband and I have put together a new computer for my father. The machine powers on, the BIOS detects all the hardware, and everything seems just fine. One problem: it randomly shuts down before we are even able to boot from the OS CD. What's going on here?
It's behaving exactly as if the PC is overheating except that it's not on nearly long enough for it to overheat! It shuts down within 60 seconds of being turned on. The CPU is barely even warm, so I'm not sure how it would be possible that it overheated. The PSU seems completely fine, and the motherboard is clearly getting power with a green light even when the machine is off. It's not shutting off at a specific point in time, so it's difficult to tie it to an action. IE: it's not shutting down every time we do a specific action. It shuts down randomly each time. Additionally, the time it takes to "overheat" gets progressively smaller each time we turn it on. Again, this seems to indicate overheating to me.
Other pertinent information: we have tried reseating the RAM in different combinations several times. This seems to have no effect at all. The BIOS is detecting the CPU at a slower speed than it should be. The card is a 2.4 and it's detecting at a 1.6. We get an error message at bootup: overclocking failed! please enter setup to change settings.
Specs:
580W PSU
Intel Core 2 Quad 2.4G
ASUS P5K-VM LGA 775
NVIDIA GeForce 7600 GT
500G SERIAL ATA HD
We don't have a more powerful PSU to test the system with, but I though the amount of power would be adequate for the system. Any help or insights would be appreciated as my parents plan to leave tomorrow.
It's behaving exactly as if the PC is overheating except that it's not on nearly long enough for it to overheat! It shuts down within 60 seconds of being turned on. The CPU is barely even warm, so I'm not sure how it would be possible that it overheated. The PSU seems completely fine, and the motherboard is clearly getting power with a green light even when the machine is off. It's not shutting off at a specific point in time, so it's difficult to tie it to an action. IE: it's not shutting down every time we do a specific action. It shuts down randomly each time. Additionally, the time it takes to "overheat" gets progressively smaller each time we turn it on. Again, this seems to indicate overheating to me.
Other pertinent information: we have tried reseating the RAM in different combinations several times. This seems to have no effect at all. The BIOS is detecting the CPU at a slower speed than it should be. The card is a 2.4 and it's detecting at a 1.6. We get an error message at bootup: overclocking failed! please enter setup to change settings.
Specs:
580W PSU
Intel Core 2 Quad 2.4G
ASUS P5K-VM LGA 775
NVIDIA GeForce 7600 GT
500G SERIAL ATA HD
We don't have a more powerful PSU to test the system with, but I though the amount of power would be adequate for the system. Any help or insights would be appreciated as my parents plan to leave tomorrow.
I've had very similar symptoms when the power supply was faulty. Open it up and see if any of the capacitors are fried (the tops of such capacitors will pop up and possibly emit dielectric in the form of crud).
posted by Krrrlson at 8:37 PM on December 1, 2007
posted by Krrrlson at 8:37 PM on December 1, 2007
(Here are a couple of busted caps next to an unbusted cap.)
posted by Krrrlson at 8:39 PM on December 1, 2007
posted by Krrrlson at 8:39 PM on December 1, 2007
Response by poster: Will opening up the PSU void the warranty? if it is faulty, I'd like to be able to trade it in for a new one. Unfortunately, I don't have another PSU I can swap because the connectors are different than on my PC. DMan: the green light on the MOBO is simply on to show that it is receiving power.
posted by theantikitty at 8:45 PM on December 1, 2007
posted by theantikitty at 8:45 PM on December 1, 2007
Oh, and test the system with another PSU if you have one available, even if it is less powerful. It may not run well, but it might get you through the boot sequence.
posted by Krrrlson at 8:46 PM on December 1, 2007
posted by Krrrlson at 8:46 PM on December 1, 2007
Best answer: It's quite possible you have a poor thermal bond between the CPU heat spreader, and the CPU heat sink. This could be due to many factors, including incorrect heat transfer compound or application, poor mounting of the heat sink/fan assembly, etc. The on-die CPU thermal diodes are capable of shutting down a CPU that overheats during boot-up as soon as the BIOS POST routine runs, typically within a few seconds of power on. Subsequent tries, without time for a complete cool down to ambient (more than an hour), will yield progressively shorter shutdown intervals, with a minimum time equal to that required for the POST routines to go far enough to permit thermal control by the processor - typically, a minimum of 3 to 5 seconds, although the processor, in a system with a big power supply, and a really bad thermal connection to the heat sink, can be way over temperature in about 1 second.
Many MOBOs provide a POWER READY LED function. Without checking the user manual for normal function, I would bet that a green light on the MOBO with AC present on the PSU, is perfectly normal.
posted by paulsc at 8:46 PM on December 1, 2007
Many MOBOs provide a POWER READY LED function. Without checking the user manual for normal function, I would bet that a green light on the MOBO with AC present on the PSU, is perfectly normal.
posted by paulsc at 8:46 PM on December 1, 2007
Whoops, too late. Yes, it might void the warranty, especially if there is a label or seal that will be broken if you open it. If there is no such label or seal, you should probably be okay.
posted by Krrrlson at 8:47 PM on December 1, 2007
posted by Krrrlson at 8:47 PM on December 1, 2007
I agree with paulsc. All the recent motherboards I've seen have a green LED on them that indicates that it's getting the proper standby power.
I'd expect the problem to be either a bad fan or a bad power supply. More likely the fan, although I replaced a power supply last week in a computer that would happily run for several hours to a couple of days before locking up. It's run fine for a week now.
Well, now that I reread the specs, it's probably not a fan, it's probably the PSU. A Core 2 system will throttle itself rather than lock up in the event of an overheating situation, so yeah, replace the power supply. ;)
posted by wierdo at 8:51 PM on December 1, 2007
I'd expect the problem to be either a bad fan or a bad power supply. More likely the fan, although I replaced a power supply last week in a computer that would happily run for several hours to a couple of days before locking up. It's run fine for a week now.
Well, now that I reread the specs, it's probably not a fan, it's probably the PSU. A Core 2 system will throttle itself rather than lock up in the event of an overheating situation, so yeah, replace the power supply. ;)
posted by wierdo at 8:51 PM on December 1, 2007
Apropos to paulsc, check that the heatsink is affixed to the CPU in place and with thermal compound. It sounds like it's overheating, and the reason it takes less time on subsequent boots is that you're booting it again before the CPU has a chance to cool down. The thermal compound is very important though, so be sure to check that.
posted by rhizome at 8:58 PM on December 1, 2007
posted by rhizome at 8:58 PM on December 1, 2007
One of my business partners has a PC that did this. He fixed it by bumping up the processor voltage in the BIOS.
YMMV.
posted by tomierna at 9:05 PM on December 1, 2007
YMMV.
posted by tomierna at 9:05 PM on December 1, 2007
Response by poster: Well, I'm back from the warzone! Even though the thermal bond had contact on the heat sink, it didn't have enough contact. We re-seated the heatsink against the CPU and it seems to be working fine. FWIW, googling that error produced a huge amount of results, so any future knowledge seekers should know that the error message might not indicate the exact same problem. It looks like it's a standard message produced whenever the system fails to initialize properly and is indicative of any number of problems. But, in this case, paulsc had it! Thanks for all of the help.
posted by theantikitty at 10:37 PM on December 1, 2007
posted by theantikitty at 10:37 PM on December 1, 2007
I realize this question has been answered, but I had a very similiar situation myself a few weeks ago when I put together my computer. I thought it was the CPU too, but in the end it turned out to be ONE bad stick of RAM out of two. I booted with one stick, it crashed, I booted with the other, worked like a charm. So if anyone is looking for an alternate solution when they find this question, try that as well.
posted by BlackLeotardFront at 11:58 AM on December 2, 2007
posted by BlackLeotardFront at 11:58 AM on December 2, 2007
« Older If I could remember more about the book, I coudl... | Help me send the Microsoft Word Thesaurus the way... Newer »
This thread is closed to new comments.
I'm not sure what you're trying to say here, but if a light is on when the computer is supposed to be off, there's something wrong in there. You sound like you know what you're doing though, so I'm guessing that's not the case, right?
Is the motherboard new?
posted by DMan at 8:36 PM on December 1, 2007