1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

PSU PC not turning on reliably - is it a dodgy PSU? Maybe not

Discussion in 'Hardware' started by mikemorton, 3 May 2021.

  1. mikemorton

    mikemorton Minimodder

    Joined:
    5 Nov 2010
    Posts:
    777
    Likes Received:
    26
    I have a bit of an odd problem - which may be entirely PSU-related, but perhaps not.

    I've had my latest build up and running for the past three weeks. Every two or three days the PC will refuse to turn itself on in the morning, and over the past few days it's become more of a daily problem.

    If the PC is switched off for a couple of hours during the day it can be turned back on properly without any difficulty, it's only when it's left overnight that this problem has manifested itself so far.

    I've managed to work out a reliable, if temporary fix, which is to switch off and disconnect the power supply, push the "On" button on the PC and wait a couple of minutes for any residual electricity to dissipate and the lights to fade, then reconnect the power cable and turn it back on.

    It's not enough to just turn the PSU off and on again. I have to totally drain the system then reconnect. it. While this method works, I have no idea why it works, or more importantly what's going on in the first place.

    The PC was originally plugged into a surge-protected extension cable, then I tried plugging it directly into a wall socket, then I tried plugging it into a different wall socket in the same room - with identical results each time.

    The PSU is a Corsair 1000W model. It's six years old, but it had still been in its shrink wrap until this build.

    If the PSU is not actually dead, why is the PC behaving this way? Could it be to do with the wiring in the house? It's a 1930s build if that adds anything

    There's a bit of history with prior PSUs which I've used in the same room which makes me wonder.

    I'd used another Corsair 1000W PSU in a previous build which died completely after just a few days. That was then replaced by a Super Flower 1200W PSU and that also died within a couple of days.

    I was sent a replacement Super Flower and that one worked perfectly from the off without any problems at all. It's been used daily for the past several years in the same room, and is now in use elsewhere in the house.

    Have I just been unlucky or is the house wiring somehow harming my power supplies? Do you know what's been happening and why?
     
  2. modd1uk

    modd1uk Multimodder

    Joined:
    4 Sep 2006
    Posts:
    3,554
    Likes Received:
    447
    You got a spare PSU to test?
     
  3. mikemorton

    mikemorton Minimodder

    Joined:
    5 Nov 2010
    Posts:
    777
    Likes Received:
    26
    In the short term I can take one from another PC, then test it in my new build and then replace it - which is a bit of a faff, but may well be worth doing.

    I have a brand new Seasonic on pre-order, but that's at least a month away from being delivered. The idea was to have at least one PSU as a fully functioning spare.
     
  4. wyx087

    wyx087 Homeworld 3 is happening!!

    Joined:
    15 Aug 2007
    Posts:
    11,987
    Likes Received:
    706
    I had a very similar problem with a Corsair Ax850 PSU. But I only needed to power cycle the PSU input and things fired up again.I got fed up with it not Wake-on-LAN so swapped to a Seasonic 550w, and no problem ever since. That was a few years ago.

    Strangely, I recently sent the Ax850 as untested to another Bit-techer, and he said it worked flawlessly. So this buggy PSU can certainly be kept as spare and it may resume working after a few months unloved.

    Problems like this is hard to debug, because it could also be with the motherboard. Especially how you describe you needed to drain residual energy.
     
  5. mikemorton

    mikemorton Minimodder

    Joined:
    5 Nov 2010
    Posts:
    777
    Likes Received:
    26
    I think you're right - it could be a motherboard issue.

    I swapped the PSU for another unit this morning. The PC turned on OK, but having shut down the PC it then refused to turn back on again - just like with the original PSU - until I'd been through the residual power purge regime.

    I don't know what to do next. It's very frustrating.
     
  6. Midlight

    Midlight Minimodder

    Joined:
    6 Jun 2011
    Posts:
    351
    Likes Received:
    184
    Maybe try changing the CMOS battery. I know other people have had really random issues fixed by that.
     
  7. mikemorton

    mikemorton Minimodder

    Joined:
    5 Nov 2010
    Posts:
    777
    Likes Received:
    26
    Good idea. I'll give that a go now.
     
  8. mikemorton

    mikemorton Minimodder

    Joined:
    5 Nov 2010
    Posts:
    777
    Likes Received:
    26
    OK thanks - that seems to have worked - at least so far!

    I've refitted the original PSU and reset the CMOS. The machine has so far survived two shutdowns/restarts without me having to purge the residual electricity.

    I also found another person with a similar problem following a random web search. It's from 2013 but is quite interesting.

    The OP found a similar short-term workaround to me by dissipating the system. Here's one of the responses he got:

    "The reason that this works has to do with the "power-good" signal from your ATX power supply to the motherboard.

    "The power good signal is a +5v signal that tells the computer that the power supply is working properly. During regular operation, if the power supply outputs are not correct, such as during a surge or brown-out, the processor is reset. The processor reset continues to happen until the power-good signal is received again. If it is not detected at start-up, then the "won't boot" situation you described occurs. At start-up, when the power supply is turned on, it does some internal self-tests that normally take between 100ms and 500ms. If they are passed, the power-good signal is then sent to the processor timer chip which controls the processor reset. The timer chip expects to receive the signal between 100ms to 600ms after powering on. If the power-good signal is received outside of the expected time, too early or too late, the processor reset happens.

    "The reason your trick works is because the +5v signal is already there due to the capacitors mentioned in other posts that are located inside the power supply not discharging. By removing the power plug from the back of the computer (or battery on a laptop) and holding the power button in, you allow the current to flow from the capacitors through the motherboard +5V circuit which the timer and case fans are connected to. Once the fans discharge the capacitors and you plug the power in and hit the button, the power-good signal is sent and the PC boots.

    "An additional point to note, many cheaper power supplies do away with the self test and connect the power-good line to their +5v rail. The result is that the processor reset never occurs because a constant power-good signal is present regardless of what happens on the other voltages in the power supply. This can sometimes cause instability and parity errors in the processor and memory during surges and brown-outs. Many manufacturers use these cheaper power supplies in their desktop systems to save costs."
     
    wyx087 likes this.

Share This Page