Random lockups - PSU trouble?

[Dr. X]

Hey guys, yesterday (during the fresher's party of all things) my computer started locking up randomly, normally after 10 or 15 minutes.

Checked voltage levels this morning and the 3.3V line was up to 3.46V. It's a Chieftec PSU,which up until now hasn't shown any sort of fluctuations on that line. The +12V line is a little low, at 11.92V but has been a little low all along. The real blatant clue that it might be the PSU is that the CPU voltage hasn't been very stable. MBM caught it at 1.62V (stock 1.6V JIUHB), which while not enough to do any harm on it's own, is a real smoking gun that the PSU might be dying.

No, I don't have any mates I can borrow a spare PSU off and I'm in no position to buy myself a new PSU to test it out.

Specs:

AXP 1700+ JIUHB,1.6V, at stock. Hanging at about 30C with the side of the case off at the mo.
Chieftec Scorpio tower case
Chieftec 300W PSU.
Abit AT7, KT333 mobo
1 HDD, 1 CDRW, 1 DVD
5 case fans, 80mm fan over CPU
1 PCI NIC.

Help? Please?

[Austin]

Made any sw or hw changes recently? There is sometimes a setting on the mobo or in the BIOS which allows you to overvolt the 3.3v rail to 3.45v IIRC, worth checking. As for the voltages you report they don't seem to bad, a 2.60v CPU (JIUHB is 1.50v?) is fine running between 2.55v and 2.65v, some fluctuation is perfectly normal. 12v should be fine at 11.92v, there is always some variation in the spec anyway.

300W is cutting it a little fine in today's systems, esp if you run 5 case fans .... what are you running? You can try unplugging as much as possible just to see if you can alleviate some of the strain on the PSU. Unfortunately most of what you need to do is try your parts (1 at a time) in a known working PC (or their parts in yours) in order to find the problem part(s).

[Dr. X]

JIUHB comes in two flavours - DLTC3 and DUTC3, 1.5V and 1.6V respectively (could be t'other way around, mine's definitely 1.6V).

5 case fans could be excessive, but these are 'silent' YS Tech's, which are only 0.84W each. They were running at 7V but I've turned them up to 12V since the trouble started.

My computer's actually been running stably for the last 4 hours or so

No SW or HW changes at all recently. I'll check the BIOS thing as soon as I restart.

On another (possibly connected note) I was contacted (left a note) by the Keble College IT staff saying I have a virus and have been disconnected from the uni network.

However

I haven't actually been disconnected from the network, the provided tools (for Welchia, SoBig.F, Blaster) found nothing, AVG with latest updates found nothing, I had all the latest critical updates, and Kerio Personal Firewall running 24/7.

I'm thinking they got the wrong guy and this is unrealted with my (now gone?) problems.

[cliky]

try Checking the dust levels in your PSU. i had 1 that was just 6 months old and started playing up!! (admittedly it was run 24/7)

on inspection it was absolutely coated in dust!!

a qui10min clean and hey presto problem solved!!

i know regulary clean all my psu's every 3 months and also hold a spare in reserve just in case

[Dr. X]

[not a bump, but hey ;-) ]

dust levels in PSU were fine, I'd put a new fan in a few weeks back.

Couldn't find the AGP voltage thing in the BIOS, my AGP voltage is still running at 3.46V, but has been perfectly stable the last 4 days, since resetting the BIOS to it's enhanced defaults. I'll spend some time searching for my old peak settigns when I've got more time and fewer freshers to chase.

[unrealrocks]

OK - have you tryed the system with just the minimals, no case fans, no CD drives, nothing apart from MB, GFX, HDD, CPU, RAM!

A 300 Watt PSU should handle that fine, those slightly wrong voltages aint a problem BUT where did you read them from ... unless there read from a multimeter on the rails then I'd not trust them!

What makes the NIC?
Does high load make it freeze more often?
Does high network traffic make it freeze more often?

[Austin]

Originally posted by Dr. X
Couldn't find the AGP voltage thing in the BIOS, my AGP voltage is still running at 3.46V ...

What gfx card are you running? 3.3v would indicate it's a pretty old AGP2x card and that can often cause problems in more modern mobos like KT333, do you know if your slot is a 'Universal' slot? Probably not the cause, but worth checking out. Has the system been rock solid in the past and only recently been playing up?

[Dr. X]

Gainward GF3. MBM5 reported the 3.3V line (figured that must be the AGP voltage) at 3.46V and I therefore assumed it was out of whack.

unreal - was somehting in the BIOS causing the lockups - not had a chance to figure out exactly what was wrong.