PC abruptly losing power

[Soad]

I had a PC fault last year that took many months to figure out the problem. The faulty component turned out to be an AMD Ryzen 5900x.The system would just reboot itself out of the blue, sometimes getting stuck in the reboot cycle. Then it became far more of a nuisance with regular reboots and system crashes happening frequently, to cut a long story short the CPU was RMA'd and OCUK sent out a new one. The system was finally working again.

A month later and another issue has cropped up. The PC shutdown abruptly and wouldn't power back on. PSU switch turned off and back on again and the system came back on briefly before dying again. Plug socket changed, system came on again but died a few minutes later. No power yet again. Process kept repeating itself, it either happened seconds after powering the system on or between 5 and 15 minutes after booting into Windows.

What I have tried so far. New socket adapter, PSU power plug changed (then reverted back to the one provided with the PSU after it made no difference) the system taken apart and put together again in a different room using another wall outlet via a different surge protected adapter, mobo switched with an identical board to test. Everything was then left at it's default settings in the BIOS and Windows booted up and didn't shutdown. The monitor used was a 60hz 1440p Dell monitor from 2014 DVI cable connected maybe? All was good for the 40 minutes it was left there. Brought it back to the room it was originally in, hooked everything up and the system began powering down constantly again... the only differences between the rooms is the wall outlet + socket, DisplayPort + high refresh rate, the ethernet cable and a few peripherals aside from the wires mouse and keyboard that were moved between rooms. I have a sneaky suspicion it has something to do with the GPU or the ram modules. I thought my PSU was busted at first but maybe not? RAM profile is currently set to 2400mhz instead of the 3600mhz XMP Profile and the refresh rate has now been lowered to 60hz instead of 165hz over DisplayPort (I had it set to 120hz for a while because my GTX 1080ti runs at full power continuously otherwise) and the system hasn't shut off so far. Something else I noticed immediately during a fresh install a month ago is the GPU / monitor won't display a picture for the BIOS, it only works when the DP cable is removed and an HDMI is instead attached. That was never an issue on Windows 10 when the PC was up and running previously.

The system hasn't powered down since but it was only tested for a couple of hours last night. The mobo has indicated both VGA and CPU issues during the failed previous boots. GPU was also flashing white in one instance.

I guess the fault could be anything but what is the most likely sign given the information provided? Which components have the ability to trip the PSU into shutting off for protection? I was also suspecting the electrics in the room initially but everything else seems to be fine. I must add all of these shutdowns have been during boot or idly on the desktop. All seemed fine until it wasn't, I had also played RE4 Remake days prior. All of the system temps were and still are fine. No recent software changes, everything upto date, no Windows issues. The system managed to stay on last night so I ran checks on my storage drives through WD Dashboard all came back good. Ran some healthchecks with the cmd prompt for Windows all was good. I tried to run a Windows Memory Diagnostics but no signal... would need to attach the HDMI again.

Due to the previous CPU failing I have a spare PSU on hand and a spare mobo. Most of the system is new or barely used due to the previous CPU fault aswell as a few replacements ie PSU, new primary storage drive and an identical mobo that was kept.

I have a feeling it's either the GPU going faulty, a faulty DisplayPort Cable if it as the ability to trigger the PSU safety measures upon shutdown or the RAM, although the shutdowns happened with and without the XMP Profile enabled previously. PSU fault? I have no idea. What do all of you think is happening? Any suggestions on how to proceed to isolate? The monitor stays connected when the PC power died. I don't believe it as anything to do with the mobo at this point as both have been tested.

System specs

OS - Windows 11 23H2 (OS Build 22631.3007)
GPU - GTX 1080TI
Monitor - MSI MAG 274qrf-qd
CPU - AMD Ryzen 5900x
CPU Cooler: Alpenfohn Brocken 3 140mm
RAM - 32GB's Team Group Edition
MOBO - MSI MAG B550 TomaHawk
PSU - MSI MPG A850G PCIE5
Storage - WD Black SN850X Windows install Gen4
Case - Lian-Li Lancool II Mesh Performance

 

[Jamin280672]

I had a PC fault last year that took many months to figure out the problem. The faulty component turned out to be an AMD Ryzen 5900x.The system would just reboot itself out of the blue, sometimes getting stuck in the reboot cycle. Then it became far more of a nuisance with regular reboots and system crashes happening frequently, to cut a long story short the CPU was RMA'd and OCUK sent out a new one. The system was finally working again.

A month later and another issue has cropped up. The PC shutdown abruptly and wouldn't power back on. PSU switch turned off and back on again and the system came back on briefly before dying again. Plug socket changed, system came on again but died a few minutes later. No power yet again. Process kept repeating itself, it either happened seconds after powering the system on or between 5 and 15 minutes after booting into Windows.

What I have tried so far. New socket adapter, PSU power plug changed (then reverted back to the one provided with the PSU after it made no difference) the system taken apart and put together again in a different room using another wall outlet via a different surge protected adapter, mobo switched with an identical board to test. Everything was then left at it's default settings in the BIOS and Windows booted up and didn't shutdown. The monitor used was a 60hz 1440p Dell monitor from 2014 DVI cable connected maybe? All was good for the 40 minutes it was left there. Brought it back to the room it was originally in, hooked everything up and the system began powering down constantly again... the only differences between the rooms is the wall outlet + socket, DisplayPort + high refresh rate, the ethernet cable and a few peripherals aside from the wires mouse and keyboard that were moved between rooms. I have a sneaky suspicion it has something to do with the GPU or the ram modules. I thought my PSU was busted at first but maybe not? RAM profile is currently set to 2400mhz instead of the 3600mhz XMP Profile and the refresh rate has now been lowered to 60hz instead of 165hz over DisplayPort (I had it set to 120hz for a while because my GTX 1080ti runs at full power continuously otherwise) and the system hasn't shut off so far. Something else I noticed immediately during a fresh install a month ago is the GPU / monitor won't display a picture for the BIOS, it only works when the DP cable is removed and an HDMI is instead attached. That was never an issue on Windows 10 when the PC was up and running previously.

The system hasn't powered down since but it was only tested for a couple of hours last night. The mobo has indicated both VGA and CPU issues during the failed previous boots. GPU was also flashing white in one instance.

I guess the fault could be anything but what is the most likely sign given the information provided? Which components have the ability to trip the PSU into shutting off for protection? I was also suspecting the electrics in the room initially but everything else seems to be fine. I must add all of these shutdowns have been during boot or idly on the desktop. All seemed fine until it wasn't, I had also played RE4 Remake days prior. All of the system temps were and still are fine. No recent software changes, everything upto date, no Windows issues. The system managed to stay on last night so I ran checks on my storage drives through WD Dashboard all came back good. Ran some healthchecks with the cmd prompt for Windows all was good. I tried to run a Windows Memory Diagnostics but no signal... would need to attach the HDMI again.

Due to the previous CPU failing I have a spare PSU on hand and a spare mobo. Most of the system is new or barely used due to the previous CPU fault aswell as a few replacements ie PSU, new primary storage drive and an identical mobo that was kept.

I have a feeling it's either the GPU going faulty, a faulty DisplayPort Cable if it as the ability to trigger the PSU safety measures upon shutdown or the RAM, although the shutdowns happened with and without the XMP Profile enabled previously. PSU fault? I have no idea. What do all of you think is happening? Any suggestions on how to proceed to isolate? The monitor stays connected when the PC power died. I don't believe it as anything to do with the mobo at this point as both have been tested.

System specs

OS - Windows 11 23H2 (OS Build 22631.3007)
GPU - GTX 1080TI
Monitor - MSI MAG 274qrf-qd
CPU - AMD Ryzen 5900x
CPU Cooler: Alpenfohn Brocken 3 140mm
RAM - 32GB's Team Group Edition
MOBO - MSI MAG B550 TomaHawk
PSU - MSI MPG A850G PCIE5
Storage - WD Black SN850X Windows install Gen4
Case - Lian-Li Lancool II Mesh Performance

Without a doubt I would change the PSU first, MSI wouldnt be my first choice for a PSU, they wouldnt be my 6th choice either, it wont be made by MSI themselves, it will just be rebranded, it makes sense that it works a bit longer when you are running everything at stock, less power being drawn from the PSU, sounds like it tripping itself due to a fault, even dropping your monitor from 165 to 60 is less stress on the GPU therefore less power being pulled from the PSU.

 

[pastymuncher]

There is nothing wrong with the MSI A series psu's. They are high quality CWT built units and they build the majority of Corsairs psu's plus other brands as well and are easily up there with Seasonic in terms of quality. That is not to say that it hasn't developed a fault though which can happen to any psu. If you have a spare psu use it to do some testing (use it's own cables not the MSI's cables as the pinouts are likely to be different and something will get fried). Can you take your pc with it's monitor and anything else connected to it and set it up in another room to rule out that wall socket because it's strange that it worked fine in the other room but not back in it's proper place. If it ran alright at 60Hz then it could also be a gpu problem as the 1080ti is getting on a bit now. Are you powering the 1080ti from two seperate pci-e leads or just the daisy chained connectors from a single lead?

 

[Tetras]

the system taken apart and put together again in a different room using another wall outlet via a different surge protected adapter, mobo switched with an identical board to test. Everything was then left at it's default settings in the BIOS and Windows booted up and didn't shutdown. The monitor used was a 60hz 1440p Dell monitor from 2014 DVI cable connected maybe? All was good for the 40 minutes it was left there.

Brought it back to the room it was originally in, hooked everything up and the system began powering down constantly again... the only differences between the rooms is the wall outlet + socket, DisplayPort + high refresh rate, the ethernet cable and a few peripherals aside from the wires mouse and keyboard that were moved between rooms.

So, it works 100% fine in the other room?

Theoretically then, the problem has to be one of: the displayport cable and/or higher refresh rate, a faulty Ethernet controller, bad power that the surge protector somehow addresses, or a peripheral.

Any suggestions on how to proceed to isolate?

- A different DP cable.
- Change the refresh rate.
- Spare graphics card.
- Use a USB ethernet adapter, or just leave it disconnected.
- Leave the peripherals disconnected.

 

[Soad]

Without a doubt I would change the PSU first, MSI wouldnt be my first choice for a PSU, they wouldnt be my 6th choice either, it wont be made by MSI themselves, it will just be rebranded, it makes sense that it works a bit longer when you are running everything at stock, less power being drawn from the PSU, sounds like it tripping itself due to a fault, even dropping your monitor from 165 to 60 is less stress on the GPU therefore less power being pulled from the PSU.

The first MSI MPG A850G PSU I received had to be returned immediately as the system wouldn't even power on. Ths PSU was the first thing I reluctantly replaced to isolate the eventual CPU RMA. The EVGA Supernova G2 1000W a 9 year old purchase at this point still booted up fine. The second MSI MPG A850G to arrive powered up fine. Of course the then issue persisted as the CPU was faulty. I still have the EVGA PSU so it will be the next thing I test when I get time. As for stock settings I tend to leave everything at stock settings in the BIOS anyway, I only reset the BIOS's to rule things out. The only exception is the enabling of the XMP Profile 1. I have never touched PBO or anything else. I have MSI Afterburner installed on the software side of things for the GPU Fan Curve but that only ever kicks in during high intensive scenarios, GPU temperature is often chilling at 25-30c even with the case fans set to slow when sitting idly on the desktop or light browsing etc. Every other component is at a similar temperature.

There is nothing wrong with the MSI A series psu's. They are high quality CWT built units and they build the majority of Corsairs psu's plus other brands as well and are easily up there with Seasonic in terms of quality. That is not to say that it hasn't developed a fault though which can happen to any psu. If you have a spare psu use it to do some testing (use it's own cables not the MSI's cables as the pinouts are likely to be different and something will get fried). Can you take your pc with it's monitor and anything else connected to it and set it up in another room to rule out that wall socket because it's strange that it worked fine in the other room but not back in it's proper place. If it ran alright at 60Hz then it could also be a gpu problem as the 1080ti is getting on a bit now. Are you powering the 1080ti from two seperate pci-e leads or just the daisy chained connectors from a single lead?

Yes, swapping out the PSU temporarily is the next thing I will try, the spare I have is pretty old at this point, but it was still running fine previously.

I apologise for the confusion but the computer booting up and running fine in the other room for 40 minutes after the PC was taken apart and put together again with the mobo switched for testing was using a 60HZ Dell 1440p monitor with a DVI Cable, a different wall outlet of course, an ethernet cable that was only capable of 100MB's link speed due to the network switch limitations, it was just simpler to attach an already existing network switch than run a cable from the HUB all the way through the house. The wired mouse and keyboard were taken back and forth between each room. XMP disabled in the BIOS as I left everything at stock.

Back to the room the PC originated in the monitor attached is a MSI MAG 165hz 1440p display overvDisplayPort. In the Nvidia Control Panel I had the refresh rate set to 120hz though so the GPU Idles correctly. Peripherals / cables attached include an headset, ethernet 1000MB's (Internet speed is close to that on the downstream) mouse, keyboard, Razer Chroma Mouse mat, USB microphone and a USB plugged in for monitor firmware updates. I think that's it. DP cable, power cables etc obviously aswell.

The PC was back to shutting down until I lowered the refresh rate from 120hz to 60HZ. The only other thing I didn't reattach yesterday (2 hours) or today (3 hours 4 minutes before shutting the computer off manually) was the ethernet cable, currently plugged into a old PS4 instead just so I could access YouTube. Every other cable / peripheral are attached and the BIOS still at it's default configuration with no XMP Profile enabled.

Two separate PCIE leads for the GPU. I will try moving both the PC and higher refresh rate monitor elsewhere if all else fails to test.

So, it works 100% fine in the other room?

Theoretically then, the problem has to be one of: the displayport cable and/or higher refresh rate, a faulty Ethernet controller, bad power that the surge protector somehow addresses, or a peripheral.



- A different DP cable.
- Change the refresh rate.
- Spare graphics card.
- Use a USB ethernet adapter, or just leave it disconnected.
- Leave the peripherals disconnected.

A GTX 980 which is in another machine at the moment is the only other that could be tested, both could be swapped over temporarily.

DisplayPort Cable I would have to order as the other machine does have two DP's but I can't see him wanting me to mess with his monitors, his wiring and desk setup is overly complicated. Would any DisplayPort Cable do or do I need a specific one?

Refresh rate was already changed. Or I guess you mean increasing it again to see if the power goes out again, sure, I'll do that. Two different MOBO's with two faulty ethernet controllers all of a sudden? Is that even possible? All peripherals are connected at the moment and it's booting with the refresh rate lowered to 60hz, ram set to stock with no XMP and the ethernet not currently attached.

 

[Tetras]

DisplayPort Cable I would have to order as the other machine does have two DP's but I can't see him wanting me to mess with his monitors, his wiring and desk setup is overly complicated. Would any DisplayPort Cable do or do I need a specific one?

Since the standards have changed quite a bit over time, older/lower quality cables can definitely cause problems, though I'd expect one that came with the monitor to be fine. If the refresh rate is a problem, I'd suspect that a driver problem, or idle instability (due to the clocks/voltage switching) is more likely.

Refresh rate was already changed. Or I guess you mean increasing it again to see if the power goes out again, sure, I'll do that.

Yes.

Two different MOBO's with two faulty ethernet controllers all of a sudden? Is that even possible?

Not likely and I think the board has 2x LAN ports with different controllers. If it had the problematic Intel controller, maybe more so, but it uses Realtek.

 

[Soad]

Having some difficulty finding the fault. I have tried a bunch of stuff since my previous posts, making progress, but I still don't know the cause. It's looking as if this could be complicated, just my luck, it's never simple to identify or on the off chance it is, it's always the most expensive parts that need replacing.

So what have I tried or ruled out? A continuation from the above. One thing I forgot to mention in my previous posts is I did try hooking the rig up to a different bedrooms wall sockets with an extended adapter and the safety protection of the PSU still shut the PC down after a few minutes. All peripherals and cables attached. Monitor still connected to the sockets of the original bedroom.

Right so the RAM has been tested with and without the XMP Profile, it's fine. DP cable removed and HDMI attached, still shutdown.

GTX 1080ti put in another system with a 1440p 144hz dual monitor setup (Not my PC) worked fine. GTX 980 put into my system, DDU'd the drivers, went to install, PC powered off again. Successfully installed the drivers and it shut off again shortly after.

The DP cable supplied with the monitor is a 1.2a according to the specs. Replaced it with a 1.4 cable just to check something else out and it made no difference. Slightly off topic but ever since I formatted and installed Windows 11 I get no signal to the BIOS using DP. Something else that I found odd is the HDMI signal is found almost instantly, I even get to see a brief view of the MSI branded logo seconds after powering the system on. The DP signal always takes like 10s to find a picture, by which point Windows is already loaded. Is this normal behaviour with some monitors? This isn't a recent occurrence, it's been like this since day 1 of getting this monitor, The no signal to the BIOS however is recent since the Windows 11 install.

Next up was taking the PC downstairs and hooking it up to one of the two 144hz refresh rate monitors. A stress test application was done for 4 hours and the PC didn't shut down once. So it's looking as if the hardware is fine. The only things I can think of now is the PSU is behaving strangely with something in the other room causing it to trigger its safety protection. a dodgy peripheral, a fault or some kind of surge from the monitor or the USB thing I have attached to my monitor, it transmits power apparently.

Other than those or a bizarre PSU fault I have no clue. The old PSU I have still needs to be tested if it still shuts down once it's back to its original location with things added one at a time.

 

[Tetras]

The DP signal always takes like 10s to find a picture, by which point Windows is already loaded. Is this normal behaviour with some monitors? This isn't a recent occurrence, it's been like this since day 1 of getting this monitor, The no signal to the BIOS however is recent since the Windows 11 install.

Not normal as such, but not unusual. Was your old install CSM/Legacy instead of UEFI? I think nvidia have done firmware updates at various points for these problems.

NVIDIA has released a new firmware update for older Quadro and GeForce GTX graphics cards, that fixes a couple of major bugs that can occur when connecting to Displayport 1.3 or 1.4 monitors — including black screen issues. NVIDIA has supplied an automated firmware updater that will automatically install the firmware update if your GPU is affected.

NVIDIA’s new firmware update specifically fixes two major DisplayPort problems that originate from the GPU itself, including system freezes (or hanging) when a system first boots, and an issue surrounding blank screens before Windows loads on startup.

...

Even though these GPUs are old, there are still plenty of gamers on GeForce GTX 10 series graphics cards like the 1050 Ti and 1060 to make this a serious issue. NVIDIA does not explain why its newer GPUs are not affected, but we can presume that its more modern GPUs have these fixes baked in or has otherwise already deployed the fix for them. NVIDIA had to patch the same black screen issue out of the RTX 3080 TI and RTX 3060 two years ago, so it's anyone's guess if these problems are gone for good.

NVIDIA Releases A Firmware Fix For Black Screen Issue On These Older GPUs

NVIDIA has released a new firmware update for pre-RTX GPUs that fixes two major bugs surrounding DisplayPort 1.3/1.4 monitor functionality.

hothardware.com

I'd recommend you don't try those updates with an unstable stable, for obvious reasons.

 

[SuperElectric]

I had an intermitent power problem, killed my SSD, took months to figure it out. Turned out to be a faulty extension lead.

 

[ATIorNvidia]

The DP signal always takes like 10s to find a picture, by which point Windows is already loaded. Is this normal behaviour with some monitors? This isn't a recent occurrence, it's been like this since day 1 of getting this monitor, The no signal to the BIOS however is recent since the Windows 11 install.

Happens on my 24G2SPU, I'd say normal on some monitors

 

[Soad]

Not normal as such, but not unusual. Was your old install CSM/Legacy instead of UEFI? I think nvidia have done firmware updates at various points for these problems.

NVIDIA Releases A Firmware Fix For Black Screen Issue On These Older GPUs

NVIDIA has released a new firmware update for pre-RTX GPUs that fixes two major bugs surrounding DisplayPort 1.3/1.4 monitor functionality.

hothardware.com

I'd recommend you don't try those updates with an unstable stable, for obvious reasons.

Thanks, you're a lifesaver. I remember downloading some DP Nvidia Firmware 3 or so years ago. I couldn't access the M-flash screen when updating the BIOS back then, but it turned out to be unrelated as it was CSM Legacy that was causing it. My Windows 10 activation USB install from 2015 would never find UEFI on boot up, CSM Legacy install worked, I can't remember much about it from all the way back then, but I know for sure this current PC was installed on CSM Legacy previously with Windows 10. I had to temporarily set the BIOS to UEFI every time I wanted to flash the BIOS, then set it back to CSM to boot into Windows. For the Windows 11 install I downloaded it from the Internet, I was well aware of UEFI at this point, this time around I was able to locate and select my storage drive for the UEFI install.

That firmware you mentioned said I had an update. I risked it despite the on-going power delivery issues and I am now able to access the BIOS over DP, not only that though I can see rhe entire boot process for the first time in years, so thanks. I also risked updating my monitor firmware With that USB attached which could well turn out to be the cause behind the PSU tripping, I also updated the BIOS, the other identical MOBO was fully upto date, since switching to rule out the MOBO, that BIOS was one or two behind. I also downloaded some recently released chipset drivers from AMD's website and Realtek's audio from the MOBO driver website. Ran Windows scans again for possible faults with SFC and the DISM commands, all came back clear, Nvidia Drivers DDU'd and reinstalled. So the only thing left to try is what I mentioned last time around. Up the refresh rate then attach the extras one at a time, see if it powers down again in the room. As its time under every other condition, it has to be a component in this room or a strange fault with the PSU.

I had an intermitent power problem, killed my SSD, took months to figure it out. Turned out to be a faulty extension lead.

Thanks for the input, that however was the first thing I replaced, tested it in another room too with another extended socket, still powered down with everything attached. The system is working fine everywhere else when it's just the PC, mouse, keyboard attached to different setups (inc high refresh rates) with other cables and such from different rooms ie ethernet, DP or DVI cable with that older monitor. I just hope I'm able to definitively able to find out the answer to the fault with that setup in that room.