I was having uefi issues recently and this problem appeared. The box would cut off and restart - like there was some relay clicking out. There is a relay. The power button is momentary, so it must be bridged by a relay. I had one of those cheap pc builder's psus, but it died so I got a decent 750W.

/Power died just now but I managed to recover the post. My cpu fan never stopped spinning, but there was a switching type sound as it died, just like when the cpu fan cuts out.

Where was I? Yes, I got a decent 750W replacement. This happened three times within 5 minutes each in the morning, then it lasted about an hour, now it might last all day.It was frustrating during the EFI madness to go for a break and find windows when you come back. I can't link this to load, and this computer never breaks into a sweat anyhow.

/Power died again, screen goes black and we go through the 'power up' routine. I had 20 minutes running then.

As you might understand, this pisses me off. The mains power is not dying. No work going on here, and my RazPi runs 24/7 because it's only a few watts and switching it is awkward for me. But if it's powered off, it sticks on a login screen. The mains is OK because it stayed on.

I don't know whether the switching sound I hear from the pc is a cause, or a by-product of the problem. The the CPU fan doesn't actually stop but that would take a few seconds.I haven't even noticed it slowing. It's on 12V, iirc. (It's been on for an hour now, and still going).

that can be an overheating issue too (anywhere).

Unlikely. It fails within 5 minutes of morning power-on. The longer it's on, the steadier it gets. It's the wrong way round for a heat problem. Cold, more likely, except it's above 4°C here and hasn't frozen over at all. The other possibility is damp, and it remains close to 100% humidity most of the winter because of the constant rain.

This is vought last year and it's a and well put together Box. The 750W Power supply I bought was a Cooler Master. There's an email from Amazon looking for a review!

If some power supply voltage went missing (say, 3.3V or -5V) and the 'power good' line toggled, what effect would that have on today's pcs? My real experience dates back to the 1990s/2000s.

My Cooler Master experiences haven't differed materially from other reasonably known brands. Maybe you just got an individual lemon, one that should not have gotten past inspector #6.

Right. But here's what's bugging me.

If the power went, as the power button is momentary, the pc would sit there until I pushed the button again. But it's like snakes & ladders, and I get a snake back to 0 for no reason, but it just carries on from there. Today it cut out as soon as I ran startx but it's fine since.

I was hoping would know the detail of how the innards work and could spot the problem. Power supply lines are all driven by the same transformer and it's very hard to imagine one dying without it staying dead.

What else tells the BIOS to start from 0x0000? An interrupt?

I assume the power issues started immediately after you made the psu replacement. Here are a few possible explanations.
1. As mrmazda suggested, you may have received a faulty psu;
2. When the cheapo psu died, it may have damaged other components on the motherboard;
3. During the replacement of the psu, something on the motherboard was damaged causing an intermittent short.

I think #3 is the most likely. In particular, check the 24 pin and 8 pin power connectors on the motherboard. See if any of the pins have been buggered or one of the wires in the harnesses look damaged. Also, some screw or metal bit may have been loose and wound up under the motherboard causing a short to the case. To troubleshoot, I recommend removing the motherboard from the case and place it on some non-conducting surface. Then connect the following to your motherboard: the psu power connections; keyboard and mouse; and a monitor. Basically, you are connected the minimum components to allow you to boot and see a display. Plug in your psu and momentarily short the power pins on the motherboard to start your system. See if you get the same intermittent power outages.

The power supply behaved well initially. Then I had an occasional monitor outage or the pc would go off, A couple of times a month. Mains sockets were suspected, I'd rattle them and press the power button. Pushing in stiff plugs with poor balance and only one hand is messy. That problem settled. Later, this started.

Point 3 is most unlikely. My son was trained up when I had an electronics workshop building PCs and fitted the new supply for me. Screws were not disturbed in most cases, and carefully accounted for. The leads were cut on the cheap PSU and replaced one by one. Plug out, plug in. Opportunity for mistakes was minimal.

I have yet to meet an intermittent short circuit on 5V stuff, without an inductor in the circuit. Everything is smoothed & damped anyhow. I've seen them on higher voltage boards sure enough. I've been fixing boards at component level since the 1970s until my stroke in 2015. There's no stray metal laying about where the job was done.

Now it's rebooting randomly with the long self test. Reboot has a shorter self test. It's restarted 4 times today, but I only pressed the power button once. The cpu fan works away, but it behaves as if I just powered it on. Yet smoothing should hold up the power longer.

The one certain symptom is that it reboots within 5 minutes of first being powered on. Everything else is unpredictable.

Progress! Of a sort.

I booted on a liveslak usb today and my box seemed fine. No random issues I thought it was good. The BIOS showed I had landed on an 'overclocking' profile and my 3.5GHz Ryzen 5 5600 was actually clocked to a max of 4.45(?) GHz. After two hours running I decided to load it, so I ran 'make clean && make -j12 all' while watching the cpu frequency on another terminal.

That lasted about 5 minutes before a fairly fundamental reboot. I went for the BIOS, and it was back at 3.5GHz, so I gather it had reset defaults, which is what I ran on, mostly. I exited, but never got as far as the boot menu again. It's constantly rebooting every 9 seconds.

From various reading, I've gathered that overclocking the Ryzen 5 5600 isn't necessarily wise. There is the faster Ryzen 5 5600XT. Apparently many dies that fail at the faster speed pass at the slower one.

So my intermittent fault is now permanent. Unless I get better advice, I'll try replacing

1. The Power Supply (returnable to Amazon)
2. The Motherboard.
3. The CPU.

Or maybe I'll hand it in and let some other dweeb try his parts in it without paying up in advance.

I'd agree with the other commenters - blame the PSU first and work up from there. But it's possible the motherboard has something wrong with it as well - cheap PSUs were (are?) notorious for taking components with them when they go. In my experience most CPUs are fairly resilient unless you're overvolting them profoundly, so while I wouldn't rule it out, I'd probably try a replacement PSU and if that fails, try swapping the motherboard out. If you have more than one stick of RAM, you can try each one individually on the off chance its a bad RAM stick.

And from the 'I wouldn't believe it if I hadn't seen it with my own eyes' file: I have seen instances of machines behaving very erratically, as if the power supply or motherboard was dead or damaged, because the cheap little momentary switch on the case (either for power or reset) is itself defective, and the motherboard is properly interpreting random 'reboot now' or 'shut down now' signals being sent (quite improperly) by the bad switch. You may try running with those headers disconnected (just short the pwr_on pins with something metal to start it up) and see if it works - probably this won't be the fix, but its free and easy to test.

I'm inclining towards the motherboard myself, and here's why.

The trigger factor for triggering the fault from intermittent to hard was building a kernel on an overclocked cpu with all cores working. That sounds more like motherboard than power supply, as the cpu & motherboard with nvme are sweating a lot, but the PSU is only sweating a little.

But the PSU gets replaced first, because I can. It's new from Amazon. The rest is out of warranty, so I pay. I'll probably hand it in to where they have stuff, and can do it.

There's no reset switch. Just a power one which isn't connected mostly

OK, I've repl;aced the power supply & the fault remains. It looks like the MSI B550M Bazooka M/B Here'a a video showing the lights beside the ram Sockets (Away from cpu; power supply; sata sockets.
This Video
Shows some lights flickering. The top one that's mostly on seems to be an 'on' led, and it reboots on the other one. Every time the screen begins to flash, it's gone for another reboot. I'd love those lights to be checked by someone with a WORKING M/B, because they could be something else entirely.

If

remove everything from every slot (ram, cards, whatever), clean them and put back to their original place. Probably helps.
Also check the documentation of the MB, the leds are definitely explained.
(You probably have a lighting capacitor, a transistor or resistor, and you need to replace that too)

I will check the documentation. The motherboard is just 18 months old, living in a clean place. The l;ikelihood of dirty contacts is remote. I can't do what you suggest because only one arm obeys me since my stroke. There's an nvme and a pricey CPU in that Motherboard that I definitely cannot touch. I had the pc built for me. So it's a professional repair guy next, if I don't get some sense out of the LEDs. Let's hope the LEDs throw up something.

Further on my last post, I'm now powered up and going somewhat unreliably. Zoom provokes the long reset within 5 minutes, but I built a kernel & modules on all 12 threads in about 12 minutes last evening. Nothing except the power supply was replaced.

It proved not to be:

• Overheating of any part (suggested in post #2)
• The psu was replaced a second time (post #4)
• Consequential Damage of previous misadventures (post #6)
• Casing Switchgear or obtuse M/B faults (post #9)

From what I can figure, my failed "Stress test" screwed up the BIOS contents teetotally, and the fix was to remove the CMOS battery and discharge the BIOS. Suddenly I had booting going on. It makes no sense whatsoever!

I bought a Ryzen 5 5600 for this box. There is a Ryzen 5 5600XT, with higher speed & turbo boost. What I didn't know was that a number of the 5600XT dies fail the 5600XT specs, but pass as lower speed 5600s. So they are sold as 5600s. The moral of that story is always buy the faster variety, or never overclock. I had overclocked a slower variety because I wasn't bothered figuring how to reset the frequency when the wrong BIOS Profile landed up one day. That BIOS is like a Mother-in-Law from Hell.

You can imagine my trials & tribulations as a hardware guy trying to earn a crust by fixing stuff, while faults like that were following me around to make life miserable!

I'm not out of the woods yet. I've got to get to the bottom of the Zoom shenanigans and make sure all else is OK.

This is not solved.

This box works all day, passes any stress test - EXCEPT ZOOM.

After 5 minutes in zoom, everything vanishes, and a complete restart ensues. On the Zoom meeting, my picture is frozen. But my box has reset. It's regular bang on 5 minutes. No more, no less. Of course I replaced the Zoom software, but that did nothing. I never actually replaced the motherboard. It wasn't needed.

The power supply is new and heat is not an issue, or cold.Zoom is an inconsequential load. But my box resets after 5 minutes every time, and every time I log back in.