PC Random Shutdowns (No BSOD)

[RosieDreams]

I'm trying to find out what is causing my PC to randomly shutdown without warning before it ends up damaging something, but I'm having trouble finding the cause.

My build:

Dell Optiplex 9020 MT 

i7-4790

16gb DDR3

EVGA 650w GQ 80+ Gold

500gb SSD

4tb WD Blue 5400 RPM

 

I recently upgraded my GPU from an EVGA GTX 750 ti to a Zotac GTX 1070 Mini and since then my PC has randomly shutdown with no warning or BSOD/error codes. It did this three times before I sent the GPU (bought used on Ebay) back thinking it was a short in the GPU. I then bought a Zotac GTX 1070 ti Mini and the same thing just happened last night. Each time it's happened it's been after an entire day of usage. I work remotely so I'm on 8-10 hours working and then often 4-5 hours of gaming afterward. Out of the times it has happened, it's been after working all day and then at about the last (5th) hour of gaming. It seems to not matter how demanding the game is either. It has happened 3 times when playing Genshin Impact for roughly 30 minutes, after playing other games the rest of that time, but it also happened once playing Dead Rising 3 (a very low ask for this PSU, GPU and CPU.) This has also been with two different CPUs as I originally had the i5-4590 when this started happening, and it's still happening on the i7-4790. I've stress tested the GPU, CPU and PSU with MSI Kombuster, Superposition, GPUmem TEST, OCCT and nothing comes back in error and the temps an wattage stay within normal/safe levels. It's only when it's been a really long use day. I've ordered a PSU tester, but it seems like that's the likely cause. I just hope it hasn't damaged anything since it has OTP, OCP, OPP and OVP on it. The PSU is less than 3 years old so it's in warranty. My question is, does this sound like it's definitely the PSU, or is there something else I might be missing? Below is my case if it helps. I thought it might be my PSU overheating, but my case is pretty open and nothing else is overheating. I was also logging the system at the time of the last shutdown and noticed the GPUw had spiked right when the PC shutdown, but the wattage has been higher than this during normal gameplay with no issues. So maybe it is just a faulty PSU and the spikes trigger something in it?

[SilverbloodZero]

not really an expert at this but have you checked your windows event logs?

[RosieDreams]

Just now, SilverbloodZero said:

not really an expert at this but have you checked your windows event logs?

I have and unfortunately all it shows is 6008 (The previous system shutdown at 11:05:17 PM on ‎3/‎22/‎2023 was unexpected.) and then 41 (The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly.)

[SilverbloodZero]

9 minutes ago, RosieDreams said:

I have and unfortunately all it shows is 6008 (The previous system shutdown at 11:05:17 PM on ‎3/‎22/‎2023 was unexpected.) and then 41 (The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly.)

i guess it might really be your psu since its only on heavy or extended loads it shutsdown, well unless your cpu tempts are too high for too long it shutsdown.

[RosieDreams]

47 minutes ago, SilverbloodZero said:

i guess it might really be your psu since its only on heavy or extended loads it shutsdown, well unless your cpu tempts are too high for too long it shutsdown.

Seems like it's gonna be the PSU. The CPU never goes above 70 degrees Celcius. I guess I need to RMA the PSU since it's in warranty and the wattage is over the recommended for this system component setup.

[SquintyG33Rs]

it doesn't have to be the PSU overheating. I would personally bet on the motherboard VRM. this kind of motherboard tends to have 2 to 4 phases with no heatsink, 750Ti was very cool but now you're more than doubling the heat inside the case with the 1070 (assuming you keep the side panel on, since you didn't specifically mention it)

the i7 is about 20W more than the i5 so if it was happening with the less thirsty part it'll happen with the upgrade too.
there are a lot of things that can trigger protections and on office towers you don't have access to many sensors.

if it turns out to be cooling you can swap the case or add a fans or 2. maybe cut a bigger hole for 120mm exhaust.
Temps is a fairly straightforward thing to fix. you can even just leave the side panel off. if that's not enough put a box fans pointed at it...

[RosieDreams]

2 hours ago, SquintyG33Rs said:

it doesn't have to be the PSU overheating. I would personally bet on the motherboard VRM. this kind of motherboard tends to have 2 to 4 phases with no heatsink, 750Ti was very cool but now you're more than doubling the heat inside the case with the 1070 (assuming you keep the side panel on, since you didn't specifically mention it)

the i7 is about 20W more than the i5 so if it was happening with the less thirsty part it'll happen with the upgrade too.
there are a lot of things that can trigger protections and on office towers you don't have access to many sensors.

if it turns out to be cooling you can swap the case or add a fans or 2. maybe cut a bigger hole for 120mm exhaust.
Temps is a fairly straightforward thing to fix. you can even just leave the side panel off. if that's not enough put a box fans pointed at it...

I've got a PSU tester arriving tomorrow, so hopefully that can give some clarification. As far as the motherboard, you're saying that the extra heat from the 1070 ti could be overheating the motherboards VRM causing it to shutdown for safety? I do keep the case closed at all times, and I have indeed been noticing when I put my hand at the back of the PC while gaming with the new card the air being exhausted is warmer. I just though that since neither the CPU or the GPU were overheating, that everything was cool enough (save for if the PSU was defective since I can't check it's temps.) I could probably get a fan hub and adapter (the fan ports on this MOBO are proprietary) and add a fan or two to see if that remedies the issue. I'm honestly just afraid to even test it out of fear of damaging my other components (the new GPU especially) when the power cuts.

[SquintyG33Rs]

44 minutes ago, RosieDreams said:

I've got a PSU tester arriving tomorrow, so hopefully that can give some clarification. As far as the motherboard, you're saying that the extra heat from the 1070 ti could be overheating the motherboards VRM causing it to shutdown for safety? I do keep the case closed at all times, and I have indeed been noticing when I put my hand at the back of the PC while gaming with the new card the air being exhausted is warmer. I just though that since neither the CPU or the GPU were overheating, that everything was cool enough (save for if the PSU was defective since I can't check it's temps.) I could probably get a fan hub and adapter (the fan ports on this MOBO are proprietary) and add a fan or two to see if that remedies the issue. I'm honestly just afraid to even test it out of fear of damaging my other components (the new GPU especially) when the power cuts.

these safety triggers are... well... for safety. so usually it's a good thing they trip instead of a short happening and your house (or just the components) going up in flames.

I would try running with the sidepanel off. if it's actually internal temps of any kind this should be a pretty clear test. (and it's a very free test to do)

Assuming an average mATX case of 60L and about 200W (i5 and 750Ti would combined to 95W lower) of total powerdraw during the gaming period, with such poor airflow it's barely renewing so the 0.07kg of air in there would be about 24C (15C with old parts) above whatever your ambient temp is after 4h gaming.

a good PSU like yours wouldn't have any problems with that especially because it's not even being stressed in this scenario at less than half it's capacity. And if it was a really bad temp for it, it would only reduce it's lifespan, NOT trigger shutdown protections. and we're talking about something with 10y warranty.
 

but why i would think the VRM is potentially borderline... well we can't see them in your picture so they are small enough to fit behind the EPS cable. I also know that in that generation most of DELL's optiplex boards use 3 phase, thought I haven't seen much of them in that orientation.
if the 3 phase is near it's maximum capacity 10C worse cooling could totally trip it's over temp protection.

consumer motherboards building 16 phases is ridiculously overkill and it's bs marketing however this means they run well bellow their capacity and you'll find they're typically in the 70~80C external package temp with a spec rated maximum of 150C internal... So i think it's a likely culprit.

[RosieDreams]

On 3/23/2023 at 5:30 PM, SquintyG33Rs said:

these safety triggers are... well... for safety. so usually it's a good thing they trip instead of a short happening and your house (or just the components) going up in flames.

I would try running with the sidepanel off. if it's actually internal temps of any kind this should be a pretty clear test. (and it's a very free test to do)

Assuming an average mATX case of 60L and about 200W (i5 and 750Ti would combined to 95W lower) of total powerdraw during the gaming period, with such poor airflow it's barely renewing so the 0.07kg of air in there would be about 24C (15C with old parts) above whatever your ambient temp is after 4h gaming.

a good PSU like yours wouldn't have any problems with that especially because it's not even being stressed in this scenario at less than half it's capacity. And if it was a really bad temp for it, it would only reduce it's lifespan, NOT trigger shutdown protections. and we're talking about something with 10y warranty.
 

but why i would think the VRM is potentially borderline... well we can't see them in your picture so they are small enough to fit behind the EPS cable. I also know that in that generation most of DELL's optiplex boards use 3 phase, thought I haven't seen much of them in that orientation.
if the 3 phase is near it's maximum capacity 10C worse cooling could totally trip it's over temp protection.

consumer motherboards building 16 phases is ridiculously overkill and it's bs marketing however this means they run well bellow their capacity and you'll find they're typically in the 70~80C external package temp with a spec rated maximum of 150C internal... So i think it's a likely culprit.

Okay so the PSU tester was a bust. It doesn't check the PSU under load, which is what I would need (those are significantly more costly.) However, it seems you're on to something for sure. I have noticed with the new GPU and CPU in, the air coming out of the back of the case has been significantly warmer. On tope of that, the 1070 ti idles at just 6-7 degrees lower than the 750 ti at it's max (at least in a few hour gaming session.) I was running HW64 when the PC crashed and took a look at the GPU temps and as you can see, at the time of the shutdown the GPU was climbing, as was the CPU. I can only assume the ambient temp around the VRM were really high as well (which I believe I found right next to the CPU cooler) and they look pretty wimpy. Combined with the poor airflow as well. I tried gaming with the side-panel off using the 750 ti and it did keep the Temps around 6-8 degrees cooler than having it on. I'm ordering some fans and a small Noctua controller for them as this motherboard only has 2 pan connectors and they're both proprietary, so this should help I'm hoping a lot. We'll have to see. Thank you so much for breaking this all down, I hadn't even considered the Motherboard temps or the overall ambient temps. I was only checking that the CPU and GPU weren't going over their max.

[SquintyG33Rs]

19 hours ago, RosieDreams said:

Okay so the PSU tester was a bust. It doesn't check the PSU under load, which is what I would need (those are significantly more costly.) However, it seems you're on to something for sure. I have noticed with the new GPU and CPU in, the air coming out of the back of the case has been significantly warmer. On tope of that, the 1070 ti idles at just 6-7 degrees lower than the 750 ti at it's max (at least in a few hour gaming session.) I was running HW64 when the PC crashed and took a look at the GPU temps and as you can see, at the time of the shutdown the GPU was climbing, as was the CPU. I can only assume the ambient temp around the VRM were really high as well (which I believe I found right next to the CPU cooler) and they look pretty wimpy. Combined with the poor airflow as well. I tried gaming with the side-panel off using the 750 ti and it did keep the Temps around 6-8 degrees cooler than having it on. I'm ordering some fans and a small Noctua controller for them as this motherboard only has 2 pan connectors and they're both proprietary, so this should help I'm hoping a lot. We'll have to see. Thank you so much for breaking this all down, I hadn't even considered the Motherboard temps or the overall ambient temps. I was only checking that the CPU and GPU weren't going over their max.

 

yeah physics are pretty reliable at this scale though all the numbers i said are obviously ball park, but they give an idea of how fast and how much it heats up.

just fans might not be much help without holes that let them push air in/out the closed box. Some cases simply get better temps just from their design.

for the particular case of VRM, my own measurements on my motherboard read 80C external to the VRM when my CPU draws 90W sustained. the MSI B350M mortar is a cheap board... that's why i bought it.... Many people were extremely worried that the cheap B350 boards wouldn't be able to properly support all AM4 CPUs that would come out after 2017 because they built the motherboards to be much cheaper than the boards people bought for intel (they didn't know if they would really be selling these after intel dominated the market for 6 years)
most B350 motherboards had 6Phase VRM that where doubled 3 phase designs.

it'S obvious now people were overreacting, the engineering isn't that bad. these uber cheap boards can totally handle 160W sustained if you have a fan pointed directly at the board.

all this to say your motherboard is doing about the same amount of work as mine, with half the muscle... and mine is already considered wimpy. you do have a cooler that does move the air over the VRM which is good but they have no heatsinc on them for extra surface area, so even though it can do the work, it very likely that it is struggling when it's being hotboxed

[RosieDreams]

2 hours ago, SquintyG33Rs said:

yeah physics are pretty reliable at this scale though all the numbers i said are obviously ball park, but they give an idea of how fast and how much it heats up.

just fans might not be much help without holes that let them push air in/out the closed box. Some cases simply get better temps just from their design.

for the particular case of VRM, my own measurements on my motherboard read 80C external to the VRM when my CPU draws 90W sustained. the MSI B350M mortar is a cheap board... that's why i bought it.... Many people were extremely worried that the cheap B350 boards wouldn't be able to properly support all AM4 CPUs that would come out after 2017 because they built the motherboards to be much cheaper than the boards people bought for intel (they didn't know if they would really be selling these after intel dominated the market for 6 years)
most B350 motherboards had 6Phase VRM that where doubled 3 phase designs.

it'S obvious now people were overreacting, the engineering isn't that bad. these uber cheap boards can totally handle 160W sustained if you have a fan pointed directly at the board.

all this to say your motherboard is doing about the same amount of work as mine, with half the muscle... and mine is already considered wimpy. you do have a cooler that does move the air over the VRM which is good but they have no heatsinc on them for extra surface area, so even though it can do the work, it very likely that it is struggling when it's being hotboxed

I mean considering the case design and board age/initial intended purpose. I'm surprised it's doing as well as it has been. The current plan is to add one fan at the front of the case as it's entirely composed of a hole pattern as is the faceplate, so that should draw fresh air in and help push the hot air out as it will directly line up with that back exhaust fan. I'm also adding a fan on the side panel as there is a ventilation pattern there as well, right next the the GPU fan exhaust, so that fan in theory will help pull the GPU's hot air exhaust out of the case. I wish I could get rid of the HDD cage as I'm not using it at all and it would allow for a large enough fan to push air across the motherboard/CPU AND the GPU. Sadly I do not have the tools to remove it's rivets, nor the space to remove the entirety of the PC's innards and work on such a project. So hopefully this will be sufficient to mitigate the extra temps rising. Honestly, even down the line when we upgrade again, it'll be with a cheaper motherboard. Looking at going with a LGA1700 socket, so not the absolute bleeding edge of CPU's either. But we def don't have the money for that right now, so I'm really hoping these fans can solve the problem. Really hope it's not a defective PSU or I'll have to go back to the 290w stock PSU while we RMA this one. We'll see

 

Just read that you generally want side panel fans to be intake (I'm new to airflow setup clearly ha) so both new fans will be intakes then.

[SquintyG33Rs]

On 3/25/2023 at 10:38 PM, RosieDreams said:

I mean considering the case design and board age/initial intended purpose. I'm surprised it's doing as well as it has been. The current plan is to add one fan at the front of the case as it's entirely composed of a hole pattern as is the faceplate, so that should draw fresh air in and help push the hot air out as it will directly line up with that back exhaust fan. I'm also adding a fan on the side panel as there is a ventilation pattern there as well, right next the the GPU fan exhaust, so that fan in theory will help pull the GPU's hot air exhaust out of the case. I wish I could get rid of the HDD cage as I'm not using it at all and it would allow for a large enough fan to push air across the motherboard/CPU AND the GPU. Sadly I do not have the tools to remove it's rivets, nor the space to remove the entirety of the PC's innards and work on such a project. So hopefully this will be sufficient to mitigate the extra temps rising. Honestly, even down the line when we upgrade again, it'll be with a cheaper motherboard. Looking at going with a LGA1700 socket, so not the absolute bleeding edge of CPU's either. But we def don't have the money for that right now, so I'm really hoping these fans can solve the problem. Really hope it's not a defective PSU or I'll have to go back to the 290w stock PSU while we RMA this one. We'll see

 

Just read that you generally want side panel fans to be intake (I'm new to airflow setup clearly ha) so both new fans will be intakes then.

 

side vents are always very useful. yea I would probably use that as an intake, it'll help a lot.
the holes in the front of the case are very often not real in these office product lines. or even when they're technically real there's actually a pitiful amount of open space for it to get through. be careful.

If we think it's the VRM, you can get small aluminum heatsinks that you can either use thermally conductive glue or some other way of tying down to help in cooling the hot stuff. (just google "vrm heatsink" amazon has a ton of options)
heatsinks are fantastic because they multiply the effectiveness of better airflow.