PC crashes

[ppppenguin]

I'll give it a try.

Another intresting test might be to pull out the C drive fit a spare and instal (say) Ubuntu or even XP on it. Then see if that hangs like W7.

[ppppenguin]

Left PC off at mains for over an hour. Then left it on BIOS screen for half an hour. Didn't hang or crash. Went to exit BIOS and it hung with "BlInitialise library failed 0XC000009A" displayed in the top left corner of the screen. CTL-ALT-DEL didn't work but the power button on the PC did.

Am now in W7 again and after a few minutes it's OK. So far.

[ppppenguin]

Left PC off at mains overnight. Then on BIOS setup for about 15 minutes. When leaving BIOS it crashed though the keyboard still worked. See screen photo.

CTL-ALT-DEL worked OK and the PC booted into W7 without any trouble.

So there's a growing amount of evidence that the W7 hang problem is BIOS related. I might replace the battery anyway, even though it seems OK. Might be worth taking out the battery and leaving it out for half an hour just to make sure that anything it supports truly forgets what was stored.

Am I right in thinking that the standby 5V from the PSU supports the things that would be supported by the battery if the mains is off?

[ppppenguin]

Looks like others have had similar problems to me: http://forums.majorgeeks.com/index.php?t...up.229324/

Though whether they are standby power related is unknown. Maybe I should boot into safe mode to see if this helps but it looks like the real problem is at BIOS level.

[Mark Hennessy]

The CMOS battery powers the CMOS RAM and RTC.

The 5V SB rail powers enough of the motherboard to allow soft power on/off and wake-up events.

I don't know what you think you're proving by going into the BIOS setup - that is just a simple user interface to allow you to alter the settings. The BIOS proper is much, much more than that.

I assume you've tried a replacement PSU?

Having done that, I'd remove all the drives and install an OS on a spare drive, as you suggested in #121. Perhaps Ubuntu or similar at first, and then perhaps W7 if that went well. Booting a live CD/ISO is not the same. I'd use the most recent version of whatever distribution you favour as it will know more about your hardware. The final test would be to eliminate your SSD by reinstalling Linux then Windows completely on that, having first imaged the current build so you can go back to it.

I really wouldn't waste time messing about with the existing Windows installation until I was 100% convinced it wasn't a hardware fault. Been there, got the t-shirt.

[ppppenguin]

I've just changed the CMOS battery. The old one was about 3V, the new one is 3.3V.

What I've proven by going into BIOS setup is that it doesn't need Windows to cause a problem. The hardware and BIOS (either 1 of them or both) are capable of causing the problem.

The hanging problem has occurred with 2 different SSDs so Ithink I can assume they are OK. I will try installing another OS on a spare SSD or HDD but I'm pretty sure the real problem is below OS level.

I don't have a spare PSU for a PC that's this modern.

[ppppenguin]

When I replaced the battery I was hardly surprised that the BIOS reported that it had lost data. One of the options it offered was to start with optimised defaults so I did that. W7 booted happily and hasn't hung. The only other thing I had to do was reset the clock and calendar - I'm slightly surprised the BIOS didn't ask me to do that so I did it in Windows.

I probably won't know until tomorrow morning whether the battery change has fixed it. I'm not holding out any great hope as 3V really ought to be OK for a CMOS battery.

[Mark Hennessy]

3V is good for one of these: http://biz.maxell.com/en/primary_batteries/cr_coin.html - 3.3V is impressive!

A PC will boot and run with no CMOS battery (obviously you'll have to visit Setup before the OS tried to boot). It's a red herring.

The only way to test the hardware is by substitution. Sorry to state the obvious, but if you don't have a spare PSU, you need to find one. And if that leads nowhere, then you'll have to try a new motherboard, new RAM, new graphics card, etc. The order is dictated by what else you already have or can borrow, and cost of what you don't. It's a PITA, but what else can you do? It's not like you can do component-level diagnostics and maintenance on any of it (apart from the PSU, perhaps).

That's why it's nice to have more than one identical machine - easy in the workplace, not so easy at home (though I've largely achieved it).

Most of the time, it's power supplies IME. And also IME, the PSUs that come with "bit-boxes" like this tend to be pretty crappy unless the customer specified a decent upgrade at the time of ordering. Whenever I upgrade such power supplies, I see it as an opportunity to get a quieter model.

I had very similar symptoms to this on a P4 PC many years back after I upgraded the CPU. The PSU was OK in terms of basic condition, it just wasn't man enough to power the new CPU reliably. Prior to that, I'd be irritated by the unreliability of the optical drive, so had replaced it with another. But as an experiment, once I'd upgraded the PSU, I put the original drive back in and it was 100% reliable - turns out it was more power-hungry that its replacement, and was tipping things over the edge. I didn't put 2 and 2 together because I used the drive so infrequently. Obviously, P4s were very power-hungry. It was a massive step change from the PIII, and the industry sometimes struggled because people were asking for power supplies to actually provide what they claimed they could deliver...

What's interesting is that the PMR has grown a lot since the P3. The average power drawn by a PC used to be fairly constant, hardly changing with CPU activity, but the peak power drawn by a P4 system was much greater than the average idle power. Since the P4, things have got generally more efficient, so the average has gradually fallen, but the peaks caused by CPU activity are now greater compared to the average. All of which places greater demands on the power supply (both the thing in the box, and the power supplies sprinkled all over the motherboard).

What's also interesting is how the PC power supply is consolidating. Look inside Dave's HP machines - from memory, there is only a single 12V rail emerging from the PSU (ignoring the standby supply - also 12V IIRC, but it's been a couple of years). All the conversion is done on the motherboard, including the 5V for the drives. Obviously the HPs are proprietary so not the same as generic ATX machines, but the ATX12V standard is evolving all the time, and I can envisage a time when generic motherboards are powered from a single 12V (or higher) rail (plus standby).

Anyway, it might not be the PSU, but you have to start somewhere...

[ppppenguin]

Might have spoken too soon. Had a crash (BSOD, rather than a freeze) while I was away from the PC. Didn't see it happen, just saw a failed boot screen. But there is a minidump.

Not had a crash like this before on this PC. Hope it's a one off.

[Nick]

Inclined to agree with Mark. PSU is possible issue, and modular replacement is way forward. At work, as time is money, we module swap until problem goes away.

We're chasing shadows here.