[SOLVED] Computer frequently powers down and back up immediately, no BSOD, temps OK, any suggestions?

[danjm99]

Over the last few days my computer has been restarting completely out of the blue. It happened once a few days ago, and I thought nothing of it, once yesterday, and today, it has happened three times. The computer is a new build, around 10 months old I think.

Specs are as followed:
Seasonic Focus+ gold, 750w
MSI Z390 PRO
i5-9600k w/ CM 212 evo cooler
2x8gb Corsair Vengeance
Samsung Evo 500gb/WDC 2tb
MSI RTX 2070

additional useful info:
CPU socket/core temps have not exceeded 64 celsius in the last hour, with a maximum recorded load of 100%. Hovering around 50-55 with a medium workload. GPU temps fine. Max 65 in Gears 5.
I have not seen any BSOD, and no memory.dmp file can be found in c:/windows indicating this was not a crash.
Disabling auto-restart on failure, and enabling permission prompts for updates does not alleviate the issue.
No recent hardware changes. I join the insider program yesterday and updated to 1898, but can't see it being this, as I encountered this issue once before then.

Any ideas or advice would be appreciated!

 

[Mrgr74]

Could it be something as simple as the reset button in the front sticking I wonder? Does the system restart after you've moved or bumped your desk/tower? What steps have you tried to duplicate the issue? Any pattern?

 

[Does it work]

It could be unstable overclocking. If you have overclocked anything, just bump it down a little. My 3600 doesn't like going above 4.05 (supposed to suit up to 4.2) nor does my ram above 3466 , it seems stable and working fine (good temps and all that) until it suddenly doesn't and it reboots (also auto reboot off, because when comp crashes i just go do something else, not bothered).

 

[danjm99]

Could it be something as simple as the reset button in the front sticking I wonder? Does the system restart after you've moved or bumped your desk/tower? What steps have you tried to duplicate the issue? Any pattern?

Reset switch was removed from the header yesterday, it seemed ok then this started happening again. I have also just removed the power switch in case this could also be the problem, I have noticed that sometimes it does not register and requires a very firm/multiple press(es) in order to trigger the power on cycle from cold.

I have not noticed any specific pattern to when this happens, if anything, it seems sporadic, I was laying in bed earlier after getting home from work and it happened twice in a space of 15 minutes while idle, which was when I decided to remove the power switch. I'm still waiting for results.

slightly off topic, do you know if I will be able to boot my computer by shorting the headers with a metallic object?

It could be unstable overclocking. If you have overclocked anything, just bump it down a little. My 3600 doesn't like going above 4.05 (supposed to suit up to 4.2) nor does my ram above 3466 , it seems stable and working fine (good temps and all that) until it suddenly doesn't and it reboots (also auto reboot off, because when comp crashes i just go do something else, not bothered).

No overclocks are present at current, so I have singled this out.

 

[Does it work]

No overclocks are present at current, so I have singled this out.

Just short your cmos and run that without any changes for a while, just for a feel around.

 

[danjm99]

Power header removal failed, computer just turned itself off again.

Event viewer shows an entry for kernel-power critical error. Still no memory dump.

 

[Mrgr74]

slightly off topic, do you know if I will be able to boot my computer by shorting the headers with a metallic object

I'd suggest against this. You can use the reset button in the power buttons place though if it appears to be OK itself. (should be as the reset button is used a fraction of the time the p/s is)

Have you done all of this? --> Kernel Power Error

In regards to @Does it work He means to reset the Cmos and looking at your mobo, I don't see a Cmos clear button so you can either just remove the battery for 30 seconds or use the jumper next to it if you have a cap.

 

[danjm99]

I'd suggest against this. You can use the reset button in the power buttons place though if it appears to be OK itself. (should be as the reset button is used a fraction of the time the p/s is)

Have you done all of this? --> Kernel Power Error

In regards to @Does it work He means to reset the Cmos and looking at your mobo, I don't see a Cmos clear button so you can either just remove the battery for 30 seconds or use the jumper next to it if you have a cap.

Set all of the power options & updated the bios. Drivers appear to be all up to date, Windows update seems to pick them up semi-regularly. Now it is just a waiting game to see if there is any changes.

Would the bios update eliminate the need for a cmos reset?

 

[Mrgr74]

Set all of the power options & updated the bios. Drivers appear to be all up to date, Windows update seems to pick them up semi-regularly. Now it is just a waiting game to see if there is any changes.

Would the bios update eliminate the need for a cmos reset?

No to the Cmos reset. Don't let the reset scare you as it's a very simple process that just resets all the settings in your motherboards cmos to their default settings. (Such as if your OC or set SSD/HDD/CDROM drive boot sequence.) In fact, resetting your Cmos is worlds safer than updating your Bios, because if done incorrectly, could possibly result in it bricking or being otherwise highly unstable. (as a general rule & not always the case as newer mobo's have built in safe guards)

Download CPU-Z and post a screen shot here. It'll show us your mobo's Bios ver. Perhaps it's an older version, whereas a new updated ver. fixes the issues you are having.

 

[danjm99]

No to the Cmos reset. Don't let the reset scare you as it's a very simple process that just resets all the settings in your motherboards cmos to their default settings. (Such as if your OC or set SSD/HDD/CDROM drive boot sequence.) In fact, resetting your Cmos is worlds safer than updating your Bios, because if done incorrectly, could possibly result in it bricking or being otherwise highly unstable. (as a general rule & not always the case as newer mobo's have built in safe guards)

Download CPU-Z and post a screen shot here. It'll show us your mobo's Bios ver. Perhaps it's an older version, whereas a new updated ver. fixes the issues you are having.

Like I say, I followed some of the steps in your previous link so hopefully the bios update provided a fix.
Here is that screenshot if you're interested.

https://imgur.com/9LOSxw1

View: https://i.imgur.com/9LOSxw1.png

 

[Mrgr74]

Please do keep us in the loop as to how everything runs. Glad to see you have the newest Bios Ver. too.

 

[danjm99]

Please do keep us in the loop as to how everything runs. Glad to see you have the newest Bios Ver. too.

No joy sadly. I'm thinking that I will return the power supply when I'm next off work as it sounds like it is probably the most likely culprit. I had high hopes for this seasonic unit, but I guess it's a high possibility given the nature of the issue I'm encountering?

 

[Mrgr74]

No joy sadly. I'm thinking that I will return the power supply when I'm next off work as it sounds like it is probably the most likely culprit. I had high hopes for this seasonic unit, but I guess it's a high possibility given the nature of the issue I'm encountering?

That is the next step yes. Sorry amigo. Really sucks you are having these ongoing issues. Do you know anyone that has a P/S you could try? Maybe buy one locally, try it & if you have the same problems, you know it's not the P/S and so don't have to send back your Seasonic. you can then return the P/S to the store and keep your current unit. Would really suck for you to send it off, only to get a new one and have the same issues. Maybe with the new unit, do a total & complete tear down of your PC. Down to the bare bones. Remove the mobo, GPU, CPU, Heat sink, RAM and reassemble everything bit by bit, testing as you go.

I forgot if you already mentioned, but have you tried removing the GPU and running on just the IGP? Your CPU has the on-CPU 630 GPU.

 

[Does it work]

Thats a shame if it is the psu, means a faulty one somehow got through testing, doesn't happen all too often with seasonic. On the plus side, their focus gold models come with an extremely long warranty, so if it is the psu, they should sort that out for you.

There is a slight possibility that there is a defect in the ram, causing it to sometimes short of its own accord, and without ram the computer will flick off and on, maybe all it needed was a slight under volt. Totally forgot I was having an issue like that in my system at first, under clocked it ever so slightly and the crashing stopped, as i previously mentioned (or thought I did, 3600>>3466 solved my problem, got it on sale at price of 2666 modules, so cant really complain)... really shoulda remembered that one lol...

If you already taken the PSU in on warranty though, hope its that (because then you might wanna rma the ram and thats always more of a pain than a psu) and you get a brand spanking new one, don't judge all seasonic psu's based on 1 that got through testing (very slim margin do, you just got unlucky if so).

 

[danjm99]

Thats a shame if it is the psu, means a faulty one somehow got through testing, doesn't happen all too often with seasonic. On the plus side, their focus gold models come with an extremely long warranty, so if it is the psu, they should sort that out for you.

There is a slight possibility that there is a defect in the ram, causing it to sometimes short of its own accord, and without ram the computer will flick off and on, maybe all it needed was a slight under volt. Totally forgot I was having an issue like that in my system at first, under clocked it ever so slightly and the crashing stopped, as i previously mentioned (or thought I did, 3600>>3466 solved my problem, got it on sale at price of 2666 modules, so cant really complain)... really shoulda remembered that one lol...

If you already taken the PSU in on warranty though, hope its that (because then you might wanna rma the ram and thats always more of a pain than a psu) and you get a brand spanking new one, don't judge all seasonic psu's based on 1 that got through testing (very slim margin do, you just got unlucky if so).

Ok. I haven't actually managed to RMA anything yet, such a busy week with starting a new job. I'll try a more intensive memory fault test before starting any returns because I would like to get it right the first time if I can.

Using memtest, would I be able to detect a fault such as the one you have described as a result of an undervoltage?

 

[Does it work]

Ok. I haven't actually managed to RMA anything yet, such a busy week with starting a new job. I'll try a more intensive memory fault test before starting any returns because I would like to get it right the first time if I can.

Using memtest, would I be able to detect a fault such as the one you have described as a result of an undervoltage?

Memtest will pick up pretty much any error that isn't stock, I'm saying to pull back on the frequency and timings just a little to (3600 to 3466, 3466 to 3200, 3200 to 3000 etc, or cas up 1 or 2 notches 15>>16 16>>18 sort of thing), it will let you see if its simply something inherent with the ram or not, run a memtest on stock for any issues though if you want, but running that test once you loosen the timings or cut the frequency will return errors though.

 

[danjm99]

Just an update that I've now swapped out the power supply. I took my system to the store where I purchased the parts from and they just done me a swap without any tests. Fingers crossed.

Hopefully the EVGA supernova 650 g+ is decent.

 

[danjm99]

Memtest will pick up pretty much any error that isn't stock, I'm saying to pull back on the frequency and timings just a little to (3600 to 3466, 3466 to 3200, 3200 to 3000 etc, or cas up 1 or 2 notches 15>>16 16>>18 sort of thing), it will let you see if its simply something inherent with the ram or not, run a memtest on stock for any issues though if you want, but running that test once you loosen the timings or cut the frequency will return errors though.

Nope. Still nothing, the RMA on the supply was a waste of time. I just got back from work and had the thing crap out on me again. Started up memtest86 straight away afterwards, left the computer for 10/15 minutes and came back to it on the windows login screen. The thing had powered off AGAIN. Completely rules out any software issues. Going to take the heatsink off in 10 minutes and look at the state of the thermal paste & mobo pins, but I'm sure it won't be this, the thing handles prime95 fine.

Any other advice? I'm fast getting tired of the thing now.

EDIT; heatsink & fan removed, cleaned with acetone and retreated with thermal compound. I probably put slightly too much on first time around.

Ram also reseated in first/third sockets. Bios warned me that this is not the optimal slots. Still early days. Who knows if it's a fix.

 

[Does it work]

Nope. Still nothing, the RMA on the supply was a waste of time. I just got back from work and had the thing crap out on me again. Started up memtest86 straight away afterwards, left the computer for 10/15 minutes and came back to it on the windows login screen. The thing had powered off AGAIN. Completely rules out any software issues. Going to take the heatsink off in 10 minutes and look at the state of the thermal paste & mobo pins, but I'm sure it won't be this, the thing handles prime95 fine.

Any other advice? I'm fast getting tired of the thing now.

EDIT; heatsink & fan removed, cleaned with acetone and retreated with thermal compound. I probably put slightly too much on first time around.

Ram also reseated in first/third sockets. Bios warned me that this is not the optimal slots. Still early days. Who knows if it's a fix.

I know the point of your cpu is to overcloc# but try undervolt/umdrrclock it too, my first cpu had issues running at stock and would only work stable with a power reduction, i rma it but still.

 

[danjm99]

Just an update, the cpu and motherboard were bench tested as faulty, with the cpu causing frequent failures and the motherboard causing intermittent failures. Both have been replaced and are awaiting my collection.