System lockup...the PSU is the suspect

[tcharter]

I have the following system specs.
Athlon T-Bird 1200
Asus A7V133
Albatron GeForce TI4200P Turbo 128MB
512MB PC133 CAS 3
Creative SB128
40GIG Quantum Fireball ATA133
1RWCDROM
Floppy
and an old 1999 300W PSU
I am using Win2K
I was running the system ok with a GeForce2MX since Fall 2001. Updated my BIOS, Drivers and all. Then last fall I changed the GeForce2MX for the TI4200P Turbo. I Got a nice increase in games performances, but the system bottleneck became the CPU. So I started to overclock the FSB to 140MHZ and set the multiplier to 9.5 instead of 9. Again, I got a nice increase in performances to the expense of a little stability. (Just got some rare crashes to desktop when playing morrowind). However the CPU and northbridge were running very hot and the northbridge fan began to fail (producing a lot of noise when booting then returning to normal after 5-10 min). I used the system with those settings for almost 3 month without any trouble. I even got Unreal II running well and easy. Then I tried to set CAS Latency to 2 instead of 3 to see if I could get a little more speed out of this system. I could get in windows but NO games where running. I always got error messages from the games engines. So I set CAS Latency back to 2. Meanwhile the northbridge fan noise became worst and never stopped. This is where the trouble began. I would get in win2K but I experimented random lockup of the system usually 10-20 minutes after booting. I immediately suspected the northbridge wasn’t cooled enough anymore by the failing fan and was corrupting the data going trough it. So I bought new cooling for my system. Thermaltake crystal orb was installed on the Northridge, volcano7+ on the CPU and 80mm smart fan on the case. Now the Mobo temp is usually around 30C and CPU around 40C. But I still experiment random lockup of the system. Decreasing FSB back to 133MHZ seem to improve stability a little bit. Increasing CPU Vcore voltage from 1.5 to 1.6 also seem to improve stability. Here are the results of a little test a did.

FSB140/mult 9.5X/vcore 1.5 = can’t complete UT2003 benchmark; can’t complete 3Dmark2001SE benckmark; can’t complete PCMark2002
FSB133/mult10X/vcore 1.5 = can complete UT2003 benchmark; can’t complete 3Dmark2001SE benckmark; can complete PCMark2002
FSB133/mult10X/vcore 1.6 = can complete UT2003 benchmark; can complete 3Dmark2001SE benckmark; can’t complete PCMark2002
FSB133/mult9X/vcore Auto (all normal settings): Can sometime complete every benchmark but can sometime lockup in any of these.

Usually with any of these settings I don’t get any crashes if I let the computer unattended on the win2K desktop. However, if I am using any application I will get lockups.

I Know a bad power supply can cause system instability, especially if you got a lot of hardware/Fan/drives. Mine is an old 300W athlon certified one. I think it should normally be enough to power my system, but the fact that increasing CPU Vcore increase system stability make me think it could be a power issue. What do you think ? I feel this will be very hard to troubleshoot unless I can get spares parts to test everything independently.

Keep in mind that everything was going well until the northbridge fan noise increased and I tested CAS Latency 2. To correct what I thought was a heat problem, I installed 2 new powerful fan and replace the CPU one with a more powerful one. The frequency of the lockup did not decrease, but didn’t increase after that. The volcano and smart fan are taking their power directly from de PSU, not from the MOBO.

Also, what do you think of the other following hypothesis?

Me having damaged my SDRAM by playing with CAS latency. The ram would then send back corrupted data when a given broken cluster is used. I don’t know if it’s even possible to damage the ram by setting it to a CAS latency it can’t support.

The northbridge being damaged by heat when it’s fan was failing. If this had happened I would think the computer wouldn’t even boot.

 

[pmonett]

So you've been overclocking. I must say, 140 over 133 does not seem to be sufficient for burning a northbridge chip, but I know nothing in electronics anyway.
With your system specs, I would have gone for a 400W PSU a long time ago. I have switched over to that ever since my first Athlon Thunderbird, but I run with a full IDE bus (2 disks, a CD-R unit and a CD-RW).
I cannot imagine that the load from the graphics card could possibly damage the northbridge. I doubt very much that your RAM was broken by the CAS switch, they just hang and that is it.
Nope, for the crashes I'd suggest swapping out your PSU for a more powerful one (and try to get a twin fan one at the least - they stay cooler and live longer). But for the noise, I haven't got a clue.
How can a chip make noise anyway ? Isn't it the fan that makes noise ? If so, replace the fan immediately, or choose a passive heatsink of impressive size (if you can find one).

 

[tcharter]

Thank you for the advice pmonett

Sorry if you understood that the northbridge chip was making the noise. I am a French Canadian and my English is not that good. It was the fan over the northbridge that was making noise. I’ve already bought and install a heatsink with fan on the northbridge. I have upgraded all of my cooling stuff. So heat is not the problem. I have ordered a 420W PSU from thermaltake and should receive it any day. I will post here when it will be installed and tested so everyone will know if it solved my instability issue.