Have you tried running any diagnostics tools? Things like memory testers, I/O tests, etc.

I'd probably start there to see if those reveal anything.

Intermittent failure is one of the hardest things to diagnose in my experience. 😳

Not yet, only just got it back to the point where I can boot things. Will look for a suitable boot disk or cd with tools olI can try

How good is the "spare old PSU"? I was getting similar issues on my ASUS A7V, changed the PSU for a Thermaltake SPG-600D-BZ and it's working well.

I've found Memtest to be quite handy in isolating whether or not its a memory issue... Have you used it? It's something you burn onto a USB, CD (maybe floppy too? I've really only use it on windows 2000+ era machines), and it just runs separately before the bios boots to the OS. I typically let it run 5-7 passes to be super sure, but if it's obviously corrupted, you'll usually get multiple errors in the first couple passes.

In general, if you're out of ideas and don't have any obvious leads, I always just like to go component-by-component with diagnostic tools, like what "Shponglefan" said to help me regain some of my sanity and refocus my efforts. I've had those memory testers reveal some pretty bad (yet 'bootable') memory in the past, and that saved me a lot of headache.

Best of luck, and post back with updates!

- RubDub

I'd normally check caps (look for bulging) - if you have a scope, look for "rough" waveforms on them.. and CPU thermal paste - these you've already done.

Next, I'd look at RAM - if it has 4 or 6 sticks, I'd go down to two, if it still fails, try the other two.

Then probably look at power, I've had supplies fail in a way that they "work ok" for a few mins, then death.

Needless to say, remove and re-seat anything that's socketed (a couple to times to help clean pins).

And take anything out that's not absolutely necessary, esp. cards - if you have a sound card, you don't need that to boot. Go to a different video card... if you worry about "upsetting" your OS, boot DOS from floppy (assuming you have one) or USB (in this case you could remove all other drives). You can still runs DOS for hours, perhaps a simple benchmark (like SPEEDCOMP) to keep it working.

red-ray wrote on 2026-02-27, 19:33:

How good is the "spare old PSU"? I was getting similar issues on my ASUS A7V, changed the PSU for a Thermaltake SPG-600D-BZ and […]
Show full quote

How good is the "spare old PSU"? I was getting similar issues on my ASUS A7V, changed the PSU for a Thermaltake SPG-600D-BZ and it's working well.

Good question, tbh I didn't even know I had it. I've got another so will try that, but a brand new one might be a smart idea.

For new PSU, make sure to get something at least 80+ Bronze rated from a reputable brand. Cheap PSUs are generally junk. Seasonic my brand of choice but there are some other good brands as well.

How old are the PSUs you have been using? What brands and models?

I have some quite old Seasonic and other higher end brand PSUs and they are perfectly fine.

Could have sworn I replied yesterday but no idea where said reply went. The two spare PSUs are both cheap old units, one is a 250W HEC thing, and the other is a Hairong 320 unit which was apparently for P4 machines. The PSU I thought had died actually isn't dead, leading me to think there is some kind of intermittent short somewhere. That one is a 420W Thermal Master, which is supposedly capable of 30A on 5V and 30A on 3.3V. I've tried different RAM sticks that I found (with a single IDE drive and no PCI cards, just the AGP GeForce 2) and still no better. Trying to find some other graphics cards that I know I have somewhere hidden away. I found a Pentium 2 board while searching but can't find the CPUs for that so clearly there's somewhere I've not yet looked.

Something I forgot on the previous post: the 420W supply looks good inside, no obvious cap issues and it's "only" from 2009 so not as old as the rest of the machine.

As for graphics cards, well one card (TNT 2) gave me a BIOS beep code for no video card found, another has bulging caps of it's own so I left it alone (GeForce 3) and the third is a ATI Rage Pro Turbo - bit old for this machine but booted and gave the exact same behavior as the card I was running in there (GeForce 2).

So I guess now it's either buy a new PSU and hope that all 3 that I've tried have their issues, or assume the board has a fault I guess. I'll go-over the caps I replaced sometime this week, but not holding out hope there.

Try wiggling cards, connectors, RAM etc slightly while it's running. If it crashes it was due to oxidisation.

When HDDs crash you should also do a disk check on it to check file system errors...they can cause problems as well

MikeSG wrote on 2026-03-02, 13:53:

Try wiggling cards, connectors, RAM etc slightly while it's running. If it crashes it was due to oxidisation.

When HDDs crash you should also do a disk check on it to check file system errors...they can cause problems as well

Will try if I can get it to run for long enough. Defnitely leaning towards something power related, be it the PSU or power circuits on the motherboard, as it'll sit in the BIOS all day long, but as soon as it's doing work it freezes. Have a brand new PSU arriving tomorrow to help narrow it down, figured I could always return that if it doesn't sort the issue anyway.

New PSU made no difference so definitely something on the main board. I booted BeOS without any PS/2 devices connected and it ran for maybe 10 seconds longer, and I could move the mouse cursor around via a USB mouse but it was quite choppy which is very unusual for that OS. An LLM suggested prior to that an "interrupt storm" cause by something going bad, seems like a possibility given the choppy mouse movement.

Ran the motherboard on my desk to make investigation easier. The IC by the ram slots labelled ICS runs pretty hot, hotter than I expect it should. Looked it up and it's one of these: https://www.jotrin.com/product/parts/ICS94201DF_995

It's a programmable system frequency generator. Tried setting the bus speeds through the dip switches in case that relieved it of work but to no avail. I'm not the greatest with surface mount but this might be just within my capabilities, at least it has easy access. Anyone know anything about them? Is hot normal or an indication of a fault/short as it often is with chips?

that vintage Asus board caps almost never go bad
yes clock generators run hot, its normal. computer wouldnt even start with that chip dead
run Memtest from floppy

ASUS TUSL2-C boardview (its like diagram/schematic) Re: Keyboard not working

rasz_pl wrote on Yesterday, 05:49:

yes clock generators run hot, its normal. computer wouldnt even start with that chip dead

Bugger, was hoping I was onto something there. This morning I fired it up and it just gave me a siren like noise from the speaker and no video output which isn't listed in the manual. Reset the CMOS and it let me into the BIOS, saved the correct frequencies etc. again andwhen it restarted back to the siren and no display. This board is starting to drive me a bit nuts!

maybe It doesnt like something as is trying to protect itself 😀 AS99127F is ASUS asic containing sensors (temp, fan speed, voltages) and responsible for soft menu voltage config. I would look at its pins, try poking with a toothpick under a loupe checking if any move

Somehow I just got it to boot for a while... PS/2 devices stopped responding but USB input was good, it did lock at some stage while I was poking around with the multimeter but don't know when... then I rebooted and got the siren again. I'll check that chip tonight, along with another pass at the caps I changed. I might also remove C151 which has seeminly been known to cause PS/2 issues. The fact those devices dropped out first seems telling.

rasz_pl wrote on Yesterday, 12:01:

maybe It doesnt like something as is trying to protect itself 😀 AS99127F is ASUS asic containing sensors (temp, fan speed, voltages) and responsible for soft menu voltage config. I would look at its pins, try poking with a toothpick under a loupe checking if any move

Gave that a go, all seemed solid. Ran the Iron over them for good measure though.

Removed C151 and made no difference, reflowed a few other things around the PS/2 ports and also made no difference. I thought I was on to something earlier when the machine was running fine without the mouse attached, but then it started working with the mouse too after I measured voltage on the clock and data lines for the mouse port. I suspect the wire I jammed in there cleaned some of the contacts a bit. The system then seemed totally good in BeOS, before I realised the CPU bus was running at 66MHz not 100.

BeOS, Win 98 and Win 2000 are all behaving well with the bus at that speed, and given I don't need raw horse power from this box (using it for dev, but not compiling a lot of code) I think I'll run it like this for now because it's driving me bananas!

run Memtest
check VTT going to CPU, preferably on a scope so you can see if there is excessive ripple