Computer Mysteriously Dying

[TehMac]

Its an old AMD Dual Core, ATi x1900xt, 2GB DDR2, and a quality OCZ Power Supply Unit I had installed two and a half years ago (Spring 2008) after the old one had died.

For at least a year however, this computer has been acting strangely. If it idles for too long, it restarts, or even if it is in use, it can randomly restart, or sometimes, the hard drive shuts off, but the computer still continues to run. It doesn't restart until you hit the restart button on the comp tower.

Recently however, the symptoms have become more frequent and worse, so I decided to reformat the computer with Windows XP (it had been running on XP before as well) today. Well, I figured Windows had become corrupted, so that'd be a good idea. I'm about halfway done with the partition when the computer just shuts off...dead. No spinning fans, nothing...

Turn the computer back on, and it runs for maybe a good 12 seconds, then nothing. It doesn't even boot up the Windows screen before it completely shuts off.

So this makes me think it is a PSU issue. But whats weird is that the PSU is pretty new...two and something years old.

What do you guys think the issue is, and has anyone experienced this before? Thanks!

 

[KGB]

Clear the CMOS and look for bulging/burst caps on the MB.

 

[TehMac]

btw guys, I tried starting it up again recently, and the post-Bios screen gave me this notification:

NTLDR is missing. A quick google came up with these source of causes for this problems.

Computer is booting from a non-bootable source.
Computer hard disk drive is not properly setup in BIOS.
Corrupt NTLDR and/or NTDETECT.COM file.
Misconfiguration with the boot.ini file.
Attempting to upgrade from a Windows 95, 98, or ME computer that is using FAT32.
New hard disk drive being added.
Corrupt boot sector / master boot record.
Seriously corrupted version of Windows 2000 or Windows XP.
Loose or Faulty IDE/EIDE hard disk drive cable.
Failing to enable USB keyboard support in the BIOS.

I think the seriously corrupted XP is the most plausible, but that doesn't explain the random complete shut offs. Sounds like this computer needs serious help.

 

[TehMac]

Well, CMOS has been cleared, and no burst or bursting caps.

 

[TehMac]

So if you remember, I cleared the CMOS yesterday by removing the jumper, placing one pin over, waiting a minute, and then placing it back to its original position.

I booted the computer up this morning and the Asus gave me a POST analysis, stating there was a CMOS Error. I pressed continue, and of course got the NTLDR is missing notification.

When I restarted, I did not get any CMOS error, or any POST analysis, just the usual MOBO Loadup Logo + the NTLDR is missing notification.

I strongly believe now that the NTLDR is missing is a result of me trying to reformat my hard drive, being 51% done with the initial process, and then the computer randomly shutting off.

Does this sound like a PSU issue to anyone else?

 

[TehMac]

So if you remember, I cleared the CMOS yesterday by removing the jumper, placing one pin over, waiting a minute, and then placing it back to its original position.

I booted the computer up this morning and the Asus gave me a POST analysis, stating there was a CMOS Error. I pressed continue, and of course got the NTLDR is missing notification.

When I restarted, I did not get any CMOS error, or any POST analysis, just the usual MOBO Loadup Logo + the NTLDR is missing notification.

I strongly believe now that the NTLDR is missing is a result of me trying to reformat my hard drive, being 51% done with the initial process, and then the computer randomly shutting off.

Does this sound like a PSU issue to anyone else?

 

[KGB]

Could be a power supply.

Could also be a funky power and/or reset switch on the case.
Could be the MB.

I would try formatiing the HD and reinstalling the OS outside of the case. Do not have the case switches connected and see what happens.

 

[TehMac]

Could be a power supply.

Could also be a funky power and/or reset switch on the case.
Could be the MB.

I would try formatiing the HD and reinstalling the OS outside of the case. Do not have the case switches connected and see what happens.

I'm tempted to first do a multi meter test. I tried reformatting the HD and the damn computer shut off.

Only problem is that I'm reading this guide and it's telling me to stick certain probes in certain holes and my PSU's cables are all black which means it's kind of tricky to tell which is which.

 

[TehMac]

Ok so I got the multimeter sorted, BUT, I ran into some problems.

Below is a quote from the guy at Extreme Overclocking:

My multimeter, when reading at 20V DC, has a resolution of 10mV, with an error of 0.5 percent.

My multi meter is manual only has an option for:

2.5V DC, 50V DC, 100V DC.

I'm reading at 50V DC...

Measuring the 12V rails @ 50V DC I "only got" 3 Volts.
Measuring the 5V rails @ 50V DC I got 10 volts...


something doesn't make sense because this is what the chart for the minimum voltages specified by ATX says:

Now granted, I only was able to get the measurements of BOOT and POST because the computer will go no further than that, citing the NTLDR is missing error

So did I somehow screw up my multimeter measurements, or am I misreading the chart?

edit: This confusion could be because I misplaced the Red and Yellow Wire in the 4-pin Molex connector...my PSU's connectors are all completely black...

 

[KGB]

Well, you obviously swapped the measurements for the two rails.

Do you have another power supply to try?

 

[TehMac]

Could also be a funky power and/or reset switch on the case.
Could be the MB.

I would try formatiing the HD and reinstalling the OS outside of the case. Do not have the case switches connected and see what happens.

So, I disconnected the POWER SW and the Reset SW and the Speaker from the motherboard, used a flathead screw driver to short, and the computer started, but no POST Beep, even no Problem Beeps. Nothing.

Fans running and full capacity, nothing. Thoughts?

 

[TehMac]

Well, you obviously swapped the measurements for the two rails.

Do you have another power supply to try?

Not at the moment though. Unfortunately, the multimeter I am using is from the 1980's or something. This is what it looks like:
http://images.drillspot.com/pimages/1626/162682_300.jpg

it's just the settings are slightly different. So I'm not sure how accurate that is, especially since it isn't even at 20V, it's at 50V, (unless that just means how far up the Voltage is measured).

I know my BIOS is inaccurate and everything, but here's what it reports:
VCore Voltage: 1.39V
3.3V Voltage: 3.32V
5V Voltage: 4.83V
12V Voltage: 12.09V

These numbers smack right into the happy place of ATX's measurements, but again, the BIOS cannot be trusted. I'm not sure if my multimeter can be either.

 

[TehMac]

fixed it...or it fixed itself. I just plugged the speaker stub from the tower into that panel, and only removed the Power SW and the Reset SW.

 

[TehMac]

Ok, now I think I have isolated the problem.

I got the computer to sucessfully boot without the two MOBO pin/jumpers being connected to the motherboard. I used a screw driver to short the pins, causing the computer to successfully turn, POST, and Boot.

I then got the computer to sucessfully boot from the CD-Rom, initializing the Windows XP installation process. I got through all the hurdles, opting for a Format using the NFTS Method (Quick).

Around 30 percent into the installation process, the computer shut off completely. This is the Second time that has happened. Having eliminated the possibility of the POWER/Reset Pins randomly shorted, the evidence points ever more conclusively towards the Power Supply Unit being unable to sustain its voltages...


edit: Schneiderguy one of the most useful, intelligent, and helpful people on this forum has offered the idea that it may be the CPU that is causing this shut down. I recall going into the BIOS right upon startup and checking out the CPU's temperatures.

They ranged from 47-50*Celsius, and apparently 60*C is out of the CPU's operating range which might be around the temp the CPU reaches when the comp is on load.

 

[TehMac]

I think I have the problem fixed--at least for right now. I blasted the CPU area with lots of compressed air and vacuumed the entire computer.

After several boots I sucessfully postd and installed windows. Have not been having any problems with the bastard since then

 

[KGB]

Excellent news and good job!

Sorry I haven't been back to this thread for a while.

 

[TehMac]

thank you sir!

Apparently, no new PSU or anything of that nature was required. Just some good old fashioned compressed air.