Right, hardcore hardware nerds: If you're pretty sure you've isolated a PSU failure on a modern PC, do you even bother with the PSU-testing part?
-
What sucks is it's one of those situations where process of elimination says it, probably, has to be the PSU... unless it's the motherboard itself.
What happens is that the PC will randomly just shut off, completely, like a lightbulb. Then if you go to press the power button, nothing happens? But if you wait 20 minutes, THEN it'll restart, and be totally normal and fine for days.
Also, if you then hardpower again, power button works exactly as expected, nothing weird, it kinda gaslights you.
@glassbottommeg Definitely PSU, or a witches curse. I'd try the PSU first
-
Right, hardcore hardware nerds: If you're pretty sure you've isolated a PSU failure on a modern PC, do you even bother with the PSU-testing part?
Or do you just order a new PSU?
I figure it's a bit like a water pump on a car. By the time you've got the thing apart enough to test the potential failure point, even if it turns out it's still maybe ok, why not just... put a new one in there.
(like I might get a PSU AND a PSU tester, best case, I end up with a spare PSU? hrm)
@glassbottommeg In the past, PSUs have been cheap enough that it hasn't been worth risking the rest of the system with a questionable unit. Dunno if that's changed recently; I've been avoiding looking at part prices for the last six months.
If it's still under warranty, a tester is probably worth keeping in a drawer (or borrowing from a friend).
-
What sucks is it's one of those situations where process of elimination says it, probably, has to be the PSU... unless it's the motherboard itself.
What happens is that the PC will randomly just shut off, completely, like a lightbulb. Then if you go to press the power button, nothing happens? But if you wait 20 minutes, THEN it'll restart, and be totally normal and fine for days.
Also, if you then hardpower again, power button works exactly as expected, nothing weird, it kinda gaslights you.
It's that "somehow it registers the power button press, but doesn't do anything about it, for like 20 minutes" thing that has me skeeved out. I can't reason through a mode of failure that'd cause that behavior in a bad PSU other than uh, PSU overheat?
It's also POSSIBLE that it would have restarted on its own had I simply touched nothing for 20 minutes but, come on, even getting it to do this takes days cus it's just a random event, doesn't relate to loading at all, so that test ain't happening
-
What sucks is it's one of those situations where process of elimination says it, probably, has to be the PSU... unless it's the motherboard itself.
What happens is that the PC will randomly just shut off, completely, like a lightbulb. Then if you go to press the power button, nothing happens? But if you wait 20 minutes, THEN it'll restart, and be totally normal and fine for days.
Also, if you then hardpower again, power button works exactly as expected, nothing weird, it kinda gaslights you.
@glassbottommeg That definitely sounds like a dodgy PSU that's shutting off from transient overdraw or faulty thermal cutouts. I'd just get a new one at that point. But, if your budget can swing a PSU tester at the same time, may as well grab one at the same time so you can test the new one alongside the old.
-
@glassbottommeg Definitely PSU, or a witches curse. I'd try the PSU first
@sinbad wait what would you try if it turned out to be a witch's curse
-
It's that "somehow it registers the power button press, but doesn't do anything about it, for like 20 minutes" thing that has me skeeved out. I can't reason through a mode of failure that'd cause that behavior in a bad PSU other than uh, PSU overheat?
It's also POSSIBLE that it would have restarted on its own had I simply touched nothing for 20 minutes but, come on, even getting it to do this takes days cus it's just a random event, doesn't relate to loading at all, so that test ain't happening
@glassbottommeg At that point I usually give up and just order a whole new PC and refresh all the parts at once (but e.g. a cheap GPU and minimal drives - swap the old ones in). It's just not worth my time to go chasing part-by-part.
-
@sinbad wait what would you try if it turned out to be a witch's curse
@glassbottommeg try and find a really really nice make-up present for a witch because they're usually cool
-
Right, hardcore hardware nerds: If you're pretty sure you've isolated a PSU failure on a modern PC, do you even bother with the PSU-testing part?
Or do you just order a new PSU?
I figure it's a bit like a water pump on a car. By the time you've got the thing apart enough to test the potential failure point, even if it turns out it's still maybe ok, why not just... put a new one in there.
(like I might get a PSU AND a PSU tester, best case, I end up with a spare PSU? hrm)
@glassbottommeg what kind of price range were you considering for the PSU?
-
What sucks is it's one of those situations where process of elimination says it, probably, has to be the PSU... unless it's the motherboard itself.
What happens is that the PC will randomly just shut off, completely, like a lightbulb. Then if you go to press the power button, nothing happens? But if you wait 20 minutes, THEN it'll restart, and be totally normal and fine for days.
Also, if you then hardpower again, power button works exactly as expected, nothing weird, it kinda gaslights you.
@glassbottommeg I recommend you to check if it's not overheating.
-
It's that "somehow it registers the power button press, but doesn't do anything about it, for like 20 minutes" thing that has me skeeved out. I can't reason through a mode of failure that'd cause that behavior in a bad PSU other than uh, PSU overheat?
It's also POSSIBLE that it would have restarted on its own had I simply touched nothing for 20 minutes but, come on, even getting it to do this takes days cus it's just a random event, doesn't relate to loading at all, so that test ain't happening
@glassbottommeg could be PSU overcurrent protection.
There's a few PSU models out there which have really sensitive overcurrent protection which can be triggered by very brief (like, microsecond length) power draw spikes that modern GPUs can have in normal usage.
-
It's that "somehow it registers the power button press, but doesn't do anything about it, for like 20 minutes" thing that has me skeeved out. I can't reason through a mode of failure that'd cause that behavior in a bad PSU other than uh, PSU overheat?
It's also POSSIBLE that it would have restarted on its own had I simply touched nothing for 20 minutes but, come on, even getting it to do this takes days cus it's just a random event, doesn't relate to loading at all, so that test ain't happening
@glassbottommeg Another possibility: a failing power button. Possibly you press it and does nothing, because it's not making the connection. Then it does make it whenever its failing condition feels like it/decides to.
-
@glassbottommeg In the past, PSUs have been cheap enough that it hasn't been worth risking the rest of the system with a questionable unit. Dunno if that's changed recently; I've been avoiding looking at part prices for the last six months.
If it's still under warranty, a tester is probably worth keeping in a drawer (or borrowing from a friend).
@glassbottommeg Oh, my actual personal real answer is "grab a PSU from one of my old, retired PCs to confirm, then order appropriate replacement(s)". Maybe someone local has an old/spare PSU you can use for bench testing?
-
Right, hardcore hardware nerds: If you're pretty sure you've isolated a PSU failure on a modern PC, do you even bother with the PSU-testing part?
Or do you just order a new PSU?
I figure it's a bit like a water pump on a car. By the time you've got the thing apart enough to test the potential failure point, even if it turns out it's still maybe ok, why not just... put a new one in there.
(like I might get a PSU AND a PSU tester, best case, I end up with a spare PSU? hrm)
I would not trust the PSU again, I had this exact same issue on a server.
Took 20 minutes to get it to boot up again after shutdown as you say..
Also - things can go even more wrong over time, faulty PSU can be a hazard.
I'd just buy a new one to be honest.
-
@glassbottommeg Another possibility: a failing power button. Possibly you press it and does nothing, because it's not making the connection. Then it does make it whenever its failing condition feels like it/decides to.
@raulinbonn Nah, this is very explicitly further down the chain.
I know this because I have occasionally gotten frustrated and mashed it repeatedly to the same effect, heh.
It's acting like something in the chain has thermal protection engaged and is delaying the signal.
-
Right, hardcore hardware nerds: If you're pretty sure you've isolated a PSU failure on a modern PC, do you even bother with the PSU-testing part?
Or do you just order a new PSU?
I figure it's a bit like a water pump on a car. By the time you've got the thing apart enough to test the potential failure point, even if it turns out it's still maybe ok, why not just... put a new one in there.
(like I might get a PSU AND a PSU tester, best case, I end up with a spare PSU? hrm)
@glassbottommeg PSU testers are one of those things you have to spend a lot of money on to get one that will actually be effective at testing. The basic cheap ones only really test voltages and not current. So if you’re having a current problem then voltages can all test ok. I find a better tester is just having a spare known good power supply
-
Right, hardcore hardware nerds: If you're pretty sure you've isolated a PSU failure on a modern PC, do you even bother with the PSU-testing part?
Or do you just order a new PSU?
I figure it's a bit like a water pump on a car. By the time you've got the thing apart enough to test the potential failure point, even if it turns out it's still maybe ok, why not just... put a new one in there.
(like I might get a PSU AND a PSU tester, best case, I end up with a spare PSU? hrm)
@glassbottommeg Depends on the PSU.
-
Right, hardcore hardware nerds: If you're pretty sure you've isolated a PSU failure on a modern PC, do you even bother with the PSU-testing part?
Or do you just order a new PSU?
I figure it's a bit like a water pump on a car. By the time you've got the thing apart enough to test the potential failure point, even if it turns out it's still maybe ok, why not just... put a new one in there.
(like I might get a PSU AND a PSU tester, best case, I end up with a spare PSU? hrm)
@glassbottommeg Exorcism?
-
It's that "somehow it registers the power button press, but doesn't do anything about it, for like 20 minutes" thing that has me skeeved out. I can't reason through a mode of failure that'd cause that behavior in a bad PSU other than uh, PSU overheat?
It's also POSSIBLE that it would have restarted on its own had I simply touched nothing for 20 minutes but, come on, even getting it to do this takes days cus it's just a random event, doesn't relate to loading at all, so that test ain't happening
@glassbottommeg powering up happens when the PSU senses a closed circuit on some pins so it seems pretty reasonable that failure in the PSU control circuitry (either a protection circuit or just the magic smoke escaped from something) would cause that as well. I would just replace the PSU and go from there if it doesn't fix the problem.
-
@glassbottommeg PSU testers are one of those things you have to spend a lot of money on to get one that will actually be effective at testing. The basic cheap ones only really test voltages and not current. So if you’re having a current problem then voltages can all test ok. I find a better tester is just having a spare known good power supply
@scavello ah, damn, I was wondering about that
ah well, new PSU on order, so, hooray!
-
It's that "somehow it registers the power button press, but doesn't do anything about it, for like 20 minutes" thing that has me skeeved out. I can't reason through a mode of failure that'd cause that behavior in a bad PSU other than uh, PSU overheat?
It's also POSSIBLE that it would have restarted on its own had I simply touched nothing for 20 minutes but, come on, even getting it to do this takes days cus it's just a random event, doesn't relate to loading at all, so that test ain't happening
@glassbottommeg the 20 minute thing makes it sound like a purposeful thermal cut off to me. Like it's not just a circuit that's glitched and caused a power loss, it's literally waiting for the sensor to have cooled down again?
