Sounds like hardware failure, maybe GPU if I had to guess.

I would strip the pc down to bare minimum, so disconnect anything non-essential like printers, external HDDs, any cards other than GPU etc. and try again.
If you have multiple sticks of RAM, try one at a time to see if there's a change.
If you have any spare hardware handy like another GPU or PSU, swap them out and see.

ratfink wrote on 2025-07-23, 22:39:

Any ideas?
<snip>

Remove the cover of the power supply and check for bulging capacitors.

ratfink wrote on 2025-07-23, 22:39:

PSU was renewed a few years ago

Does that mean replaced with a new one?

Regardless of that though I would suspect the power supply as a potential issue or as was said another part is failing, potentially the old graphics card. Does that CPU have onboard graphics? If so pull that card and try it with the onboard.

Can be literally anything.
Bad mosfet, bad solder joint that comes good with temperature.

Personally, I'm a big fan of washing boards/gpus. You'd be surprised how many came good.
Most people look at me like a madman though, or argue that its bad, or feels wrong, or whatever.

So, it's up to you.

Ok thanks all.

I booted to the BIOS and left it there about 10 minutes, nothing went wrong. Tried the reboot button and it lost video and didn't reboot, on/off did nothing.

On/off at socket, went through this a few times. Then went into Windows, same as before.

Checked external cables, unplugged externals not needed, removed fan cover, removed side panel... nothing loose or out of place. I think there's an odd smell but can't be sure.

Went through the boot into Windows routine, I can't see a pattern to the freeze.

Turned off at the PSU and on again a few times. Now it won't start at all, a sort of err-errr-err attempt (case lights flicker) - no boot at all.

PSU turns out to be a CX750M, looks like I didn't swap it for new a few years ago. Pretty sure that CX750M is what I bought when my phenom board caught fire so... 2014.

PSUs often have short protection.

See if it has the same issues without gpu and other external things.
Obviously you won't get a display, but you can tell

And yes, intermittently dead mosfets on a gpu can cause intermittent shorts - but so can anything else.

Will do. Slightly sidetracked now by the entire surface of the case being sticky rubber, so now it's IPA and elbow grease to get the worst off.

Ok without any peripherals, seems to boot with a GT640 - fans spin, ther's HDD activty.

Connected standard peripherals, same behaviour as before. So this makes the GPU less of a suspect. Last crash of this type it was only sitting at the desktop.

The. on-off feels like it's sticking so that could be confusing things. But when the boot gets past a certain point the physical on/off doesn't work. Can't remember whether that's normal.

ratfink wrote on 2025-07-24, 09:18:

PSU turns out to be a CX750M, looks like I didn't swap it for new a few years ago. Pretty sure that CX750M is what I bought when my phenom board caught fire so... 2014.

"The electrolytic filtering caps are provided by Chemi-Con (105°C, KZE) and Su'scon (105°C, MF). We also spotted a single Taicon (105°C) capacitor."

Check the integrity of the Su'scon capacitors.

Rule out the GPU -- use a different one.

Took the PSU apart, everything looks good - no leaks or bulges.
Unfortunately I managed to drop the PC because... what I thought was a hand-hold for lifting it one end turned out not to be .... top of the case came off... rest of was only held at one end so it fell hard. Now it doesn't work at all. This is all beyond a joke right now. All I wanted to do was edit some pictures... Maybe it will just be the front panel connector came out, but these damn cases and short leads and everything buried and cramped... probably have to take a lot out just to see where that connector is. One step forward and three steps back.

GoblinUpTheRoad wrote on 2025-07-24, 00:26:

Sounds like hardware failure, maybe GPU if I had to guess.

Agreed, GTX560Ti is probably dying.
Crystal failure is very common for desktop GeForce 500 series due to bad IHS design.

I had pretty much the same thing happen just a few weeks ago with my own Ryzen 7 5800X setup (Gigabyte B550 X2 motherboard). It would either reboot itself, or more commonly just hard freeze - no apparent pattern, no error messages or BSODs, nothing in the event logs, no dependence on temperature or on what was/wasn't running. At first I could get a few hours of use between failures, but it got to the point where it'd often happen before Windows even started up.

The culprit turned out to be something I wasn't expecting... the CPU(!). Took the thing to a nearby store which also does repairs, and we tried a bunch of replacement parts. RAM, GPU, PSU made no difference, but swapping the CPU (for a 5700X3D) did the trick. Been stable as a rock since then.

So that's another thing to check for if you can.

VileR wrote on Yesterday, 15:52:

The culprit turned out to be something I wasn't expecting... the CPU(!). Took the thing to a nearby store which also does repairs, and we tried a bunch of replacement parts. RAM, GPU, PSU made no difference, but swapping the CPU (for a 5700X3D) did the trick. Been stable as a rock since then.
<snip>

SAFS - Sudden AMD Failure Syndrome...

My 2009 AMD still works. 😜 Are CPU failures still unusual in general?

Rare, but not unheard of