Help! My PC keeps shutting down

[spartan3d]

Hi all,

It's been weeks that my PC keeps shutting down randomly, sometimes while working in Windows, sometimes in Linux, sometimes while checking BIOS settings, even sometimes just several seconds after turning on this PC in the morning.

No BSOD generated. I'd also set Windows not to automatically restart computer when BSOD occurs.
I never overclock this PC.

The specs:
Mobo: Abit IS7, with latest BIOS version 24
CPU: Intel Pentium 4 HT 3.2 GHz Northwood
RAM: 2 Vgen + 2 Kingston DDR1, 2 Gbytes total
Graphics: Winfast Geforce 6600 GT AGP
HDD: Western Digital WD800BB Caviar 80 Gbytes IDE
DVD-drive: LiteOn DVDRW IDE
PSU: Simbadda AL-B530E PFC 530W
Case: Enlight EN-7238 ATX
Automatic voltage regulator: Matsugawa Stavolt 1000 VA, with servo-motor control system

I firstly bought this machine in 2004.

I'd tried:
- resetting CMOS from a jumper setting on the mobo
- replacing CPU with Intel Pentium 4 HT 2.4 GHz Northwood
- replacing thermal paste with Arctic Silver
- swapping RAM slots
- replacing graphic card with Elsa Synergy4 380XGL Quadro
- replacing PSU with FSP Saga 420W
- cleaning mobo, CPU socket, RAM slots, Graphic card, PSU from dust

Here's the stress test result:

Firstly I thought it's because of excessive temperature, since the PC is located in a large room without air conditioning, with room temperature about 86F. So, I add an additional intake 12cm fan like this:

This fan takes air inside the casing, through grids on below part of side cover.

Someday I brought this PC to a smaller room, about 4x6 sq meters, air conditioned, with room temperature about 77F. But it still kept shutting down.

When suddenly shut down:
- Num Lock on the keyboard is still on, but not responsive when some keys stroke
- CPU fan is still rotating
- GPU fan is still rotating
- PSU fan is still rotating
- DVD-drive is still working, the tray still can be opened and closed

What happened to my PC?
Any solution really appreciated.

Thanks in advance!

 

[engiNURD]

That's some aged hardware you've got there! What was once a badass system may be suffering from old age.

How long does Prime95 run before the system crashes?
What's the exact model number of the FSP PSU you replaced the Enlight with?

Here are the next steps I'd take...
Inspect each part for any bulging capacitors.
Replace the thermal interface material for the northbridge/southbridge coolers.
Taken the system apart, clean everything, test it outside of the case, then put it back together if it works.
Try a different PSU.
Run memtest overnight.

 

[spartan3d]

That's some aged hardware you've got there! What was once a badass system may be suffering from old age.

How long does Prime95 run before the system crashes?
What's the exact model number of the FSP PSU you replaced the Enlight with?

Here are the next steps I'd take...
Inspect each part for any bulging capacitors.
Replace the thermal interface material for the northbridge/southbridge coolers.
Taken the system apart, clean everything, test it outside of the case, then put it back together if it works.
Try a different PSU.
Run memtest overnight.

Indeed, it's an aged machine.
I also inspected the physical appearance of the mobo, I didn't see any bulging capacitors.
FSP PSU I tried as replacement was FSP Saga AX450-PN.

Northbridge tiny fan had died, I just uninstalled it, just mounted the heatsink without the fan. From the first time, Southbridge doesn't have any HSF installed.

I'm doubting whether it can survive an overnight test. In just a few minutes (no longer than 15-30 minutes up to some few hours), it crashes, whether it's loaded with lots of running programs in Windows or Linux, or even when idle status or when starting up BIOS, it still crashes.

 

[Deleted member 88227]

Does the Windows event logger say anything?

 

[enieves137]

Notice on your screen shot that the negative dc volts are out of spec, their is a 10% tolerance. If you replaced the PSU then it's got to be your motherboard. You don't have a heat issue but a voltage regulation problem on the mainboard. That would cause crashes!

 

[m_isom]

I would not rule out the power supply, it has been discontinued. And judging by the pictures on newegg it seems to be a low quality part with thin wires and probably does not have the weight of the better budget power supplies.

http://www.newegg.com/Product/Product.aspx?Item=N82E16817139026
coupon code EMCNHHB34 takes off $8
$37 with free shipping and has a $20 rebate
bought 4 of these now and have not had a problem.

 

[spartan3d]

I just replaced the Simbadda PSU with FSP Saga again. It still crashes like before.
Weird, in CPUID Hardware Monitor, -12V is still read as -7.01V and -5V as -3.69V.
On http://hardforum.com/showthread.php?t=756864 it's said:

-12 V: This voltage is used on some types of serial port circuits, whose amplifier circuits require both -12V and +12V. It is not needed on some newer systems, and even on older ones not very much is used, because the serial ports require little power. Most power supplies provide it for compatibility with older hardware, but usually with a current limit of less than 1 A.

-5 V: A now archaic voltage, -5 V was used on some of the earliest PCs for floppy controllers and other circuits used by ISA bus cards. It is usually provided, in small quantity (generally less than 1A), for compatibility with older hardware. Some form factor power supplies such as the SFX no longer bother to supply it (systems using the SFX power supply are intended not to have ISA bus slots).

I don't have ISA cards nor FDD on this machine. Do these low negative volts really matter?

 

[PersonalJ]

Sounds like it's your motherboard at this point. I'm assuming you've already tested the RAM.

 

[Draxanoth]

Odds are it's going to be the motherboard or PSU. Almost everything else would likely cause it to stop working without causing it to power off.

I assume when you say it "shuts down randomly" you mean it goes off instantly, as if someone yanked the power cable from the back of your machine. That's a bad symptom to have.

Only real way to narrow it down is to connect another PSU and see if the problem goes away. If it still persists at that point you probably have a motherboard that has died of old age.

 

[engiNURD]

Try a better quality PSU.

 

[spartan3d]

I assume when you say it "shuts down randomly" you mean it goes off instantly, as if someone yanked the power cable from the back of your machine.

No, it's not like that. Instead:
- Num Lock on the keyboard is still on, but not responsive when some keys stroke
- CPU fan is still rotating
- GPU fan is still rotating
- PSU fan is still rotating
- DVD-drive is still working, the tray still can be opened and closed

It's more like the monitor lost video signal. But also the PC got frozen (keyboard not responsive anymore).

Sounds like it's your motherboard at this point. I'm assuming you've already tested the RAM.

I also replaced the PSU with another Simbadda 430W PSU.
This is the 3rd PSU I installed on this machine.
CPUID Hardware Monitor showed the same result: -12V becomes -7.01V, -5V becomes -3.69V.

I just ran SpeedFan test and it showed GPU temperature as well:

This screenshot was taken after 2-hour stress test with Prime95, VideoStability Tester, Hardware Monitor, GPU-Z running simultaneously.

When running VideoStability Tester, GPU temperature raised up to 87C.

After that, I booted and ran memtest86+ (which is an OS-independent software run from a boot CD), it crashed again after only 2-3 minutes running.

I must make sure whether it's PSU or motherboard that's faulty, before purchasing a new one.

 

[Quartz-1]

Have you checked the capacitors on the motherboard?

Try running memtest again and see if it crashes in the same place. If it does, then you likely have a faulty memory stick. Remove each in turn and try again.

 

[spartan3d]

It's been 1 hour running Memtest86+ now and it hasn't crashed yet. Memtest doesn't show any Errors either.
Update: 1 Pass completed, no errors.

I'd already checked the capacitors on the mobo; I didn't see any bulging capacitors.

 

[Paragon]

What's with the voltage regulator?

 

[enieves137]

Low voltage reading is indicative of bad motherboard regardless if the voltage is being used or not, it's beside the point. The point is that you’re having issues with voltage regulation. Memtest is lower-level function test, meaning only part of the motherboard is being used.

Multimeter would be useful at this point to determine how valid the voltage readings in the software.

 

[fredgarven]

What is the exact purpose of the voltage regulator for your specific system?

I do agree that based on the information provided, your motherboard is most likely the problem.

Replace the whole system and be through with it.

 

[spartan3d]

AVR is a device commonly used in my country, due to unstable voltage, often damaging electronic devices and computers.

Btw, I guess this is time to say goodbye to my old mobo.
Thanks everyone for the suggestions.