Freezing BIOS even - any ideas why?

[Carol Haynes]

Got a client with a computer that freezes - even to the point of not allowing a forced switch off by holding the power button.

No errors showing in windows - it doesn't reboot ... black screen and everything dead but still with power LED on. You have switch off and on at the wall to reboot.

At first I thought it might be a windows XP standby issue but I have disabled standby and now the computer doesn't seem to crash once the system is up and running but after a switch on from cold if freezes during windows boot.

I have updated drivers (including graphics) and checked all error logs. I have also downloaded the Fujitsu diagnostics tool but that says everything is working fine. BIOS is at the final revision. I also did a check for any malware but couldn't find any.

I am a bit stumped - anyone got any ideas?

My thought is a chip on the point of dying but works when it has a chance to warm a little but would value some help.

Cheers

Carol

[Stoic Joker]

So a (warm) reboot does fine, but a cold boot tends to lockup? Main difference between them is the memory count/hardware check isn't done on the warm boot (iirc). <- Just thinking out loud here)

(No 4 sec power button shutdown) It's a definitely hardware level lockup, have you done a memory test?

Either that or a brown-out/power (PSU) issue, as those never leave anything in the logs about why/what happened.

[40hz]

Did you try replacing the mobo's battery?

[Carol Haynes]

Thanks - yes I did change the mobo battery and reset the BIOS to safe details.

I did run memory tests and got no errors.

Need to check the PSU - but it is strange that it runs well once everything is going.

[Stoic Joker]

Starting amperage spike from everything starting from zero at once?

Unplug (or move to alternate rail) a few nonessentials and see if the behavior changes.

[Bamse]

The bad chip you have as suspect could be a capacitor http://en.wikipedia....iki/Capacitor_plague See under symptoms. Don't know if problem is always visible. Fujitsu is well known as is problem, may be a search for model name + capacitor give useful hits?

[4wd]

My vote goes to the PSU - it sounds like the electrolytics are failing.

Other than that, usually when a piece of equipment won't start at switch on but will after it's had current flowing for a little while it indicates a hardware fault, (bad soldering, components gone marginal, etc), because the equipment has had a little time to warm up.

As per SJ, change the PSU first if that doesn't work then take it back to PSU, MB, RAM and CPU, see if that works then add things one at a time until it fails.

The only problem with temperature related faults is they can take a while to isolate since about the only time you can guarantee it'll happen is first thing on a cold morning, (and even then it won't if you're watching for it ).

EDIT: Pipped at the POST by Bamse

In one case, the reason for the manufacture of faulty electrolytic capacitors was industrial espionage gone wrong: several Taiwanese electrolyte manufacturers began using a stolen formula that was incomplete, and lacked ingredients needed to produce a stable capacitor.

 

More likely than exploding electrolytics is that they have gone high ESR^w due to aging and/or electrical ripple/spikes.  This will normally not be visible and you'd need to pull the equipment boards out and test each electrolytic capacitor with an ESR meter - not something to do if you're not electronically inclined.

[Carol Haynes]

Thanks for all the advice - mostly lines I was thinking along.

The problem is dealing with a system where I can't take it away to play with (they need it everyday) and whenever I see it I can't generate the problem, although I have seen it 'post freeze'.

Capacitor pushing towards failure or dodgy PSU sound like likely culprits to me but unfortunately I haven't got a spare PSU to try it out (I suppose I could borrow one from one of my systems).

I do have a PSU tester so I could give that a whirl.

[Shades]

Tantallic capacitors last a lot longer and eliminate bad spots inside themselves automatically, hardly affecting the capacity they had from the beginning.

From (jaded and mutilated) memory they are expensive to produce, really expensive to buy and not fit for very high frequencies. Then again, a PSU is not high frequency.

[f0dder]

Got a client with a computer that freezes - even to the point of not allowing a forced switch off by holding the power button.

-Carol Haynes (November 24, 2010, 02:03 PM)

Wow, don't think I've ever seen that before, even with pretty fubar boxes O_o

[4wd]

From (jaded and mutilated) memory they are expensive to produce, really expensive to buy and not fit for very high frequencies. Then again, a PSU is not high frequency.

-Shades (November 24, 2010, 04:46 PM)

All modern Switchmode PSUs are high frequency, starting from about 20kHz and going much higher.  This problem never, (well very, very rarely), appeared in older PSUs that only ran at mains frequency, (50/60 Hz).

Tantalums are unsuitable for this application because their ESR is much higher than electrolytics at these frequencies.

I do have a PSU tester so I could give that a whirl.

-Carol Haynes (November 24, 2010, 04:38 PM)

Unless it can provide an equivalent load at switch on to replicate inrush current, it might not do much except show if the output voltages are within limits.

Wow, don't think I've ever seen that before, even with pretty fubar boxes O_o

-f0dder (November 24, 2010, 04:49 PM)

You need to walk the Dark Side of electronics

[40hz]

Another possibility might be a bad trace, cold spot, or hairline fracture on the mobo. This is the other thing (along with iffy capacitors and PSUs) that can introduce weird thermal related system hangs. There's a very scary "flex test" you can do on a live system to check for board trace fractures. But I wouldn't recommend it unless you're already on the brink of packing something off to the community tip and don't care any more.  

Also a damaged slot (cold spot or damaged trace on RAM slots - or whatever is hosting the video card) are other likely culprits. If a defective signal path introduces enough chatter, timeouts or corrupt data, the CPU can go into la-lah land. Swap outs will help check for that.

Got a client with a computer that freezes - even to the point of not allowing a forced switch off by holding the power button.

-Carol Haynes (November 24, 2010, 02:03 PM)

Wow, don't think I've ever seen that before, even with pretty fubar boxes O_o

-f0dder (November 24, 2010, 04:49 PM)

Got an old HP Kayak workstation sitting in storage that does that. Sorta like a turtle - once it locks up, it stays locked up. Even after a replacement PSU and front panel switch. Only way to reboot it is to pull the AC cord.  

[Carol Haynes]

I am coming to the conclusion that they would be better off replacing the system. It's five years old so there is little point in wasting time and money keeping it alive - especially as it is so cranky and unpredictable.

Thanks for all the feedback - it is useful to have people to bounce ideas off when you work alone!