PC restarts under heavy load, faulty psu?

[qung3]

Spec:
Ryzen 3 2200G
Asrock b450m(bios updated)
Corsair lpx 2x8 3200
nvidia 1070 sc
evga 450BT (psu)


Been getting black screen / restarts for awhile when playing specific games, my suspicious is on the PSU because in the games where the system restarts are games that put a lot of load into the cpu, gpu and ram, for example I cant map in path of exile for more than 15 minutes without a crash, but I can play elden ring for hours and raid in wow without a single crash, I already ruled out the ram by trying with a different one with and without xmp.

So I ran some benchmarks while looking at hwinfo and noticed that while running heaven benchmark the gpu pcie +12V input voltage drop from 11.928V all the way to 11.450V, it may be within spec but its feel like on the very low end side and thats just running a gpu benchmark.

https://imgur.com/a/803dyrk

There is no bluescreens, minidumps, nothing on event viewer beside event 41, no artifacts.

Solved: Old psu was failing. had to replaced it.

 

[Bill_Bright]

You don't show your CPU temps - a lead suspect.

11.450V is a bit low, but as you noted, still within allowed limits. The allowed variance is ±5%. So 12VDC ±5% = 11.4 to 12.6VDC.

I don't think that is your problem but it is always wise to verify you are providing good, clean, stable power. So I recommend swapping in a known good PSU and see if the problem persists.

 

[GerKNG]

11.4V is really low and the PSU is basically a bottom of the barrel low end unit.
i'd say it's 99.9% the PSUs fault. replace it with whatever is a reliable brand and doesn't cost too much.

 

[DR4G00N]

Wiggle the power connectors slightly and see if the voltage changes. A bad connection can cause voltage the to drop excessively under load.

 

[Norton]

Running a GTX 1070 at load with a 450W psu may be triggering one of the protection circuits in the psu. I had a 500W psu that used to do that at full load due to it being a little too small for the config I was running at the time- are you hearing a click when it shuts down?

I would try a known good psu as @Bill_Bright suggested- note that most 1070's recommend a 500W psu

 

[thewan]

the many expert wannabies above me don't seem to realize that the pcie +12V input voltage refers to the power supplied by the PCIE slot. Not the GPU 8 pin, which is the one below it.

Wiggle the power connectors slightly and see if the voltage changes. A bad connection can cause voltage the to drop excessively under load.

Please don't wiggle your 24 pin motherboard connector, which is where the PCIE slot power comes from. One wrong move will bring down entire PC, not just the GPU.

11.4V is really low and the PSU is basically a bottom of the barrel low end unit.
i'd say it's 99.9% the PSUs fault. replace it with whatever is a reliable brand and doesn't cost too much.

The PCIE GPU 8 pin is fine, as noted by the values BELOW what the one OP is talking about

I'd argue that while the PSU is a cheap budget model. It isn't by all means a bottom of the barrel terrible choice to pick for a budget build, as verified and tested and praised by TPU's own PSU Tester (albeit for a rival website):

EVGA 450 BT PSU Review: Amazing Value At $25

Is it possible for a PSU that costs $40 (and can sometimes be found for as little as $25) to offer decent performance and reliability? EVGA's 450 BT promises both, appealing to enthusiasts on a budget.

www.tomshardware.com

Its more likely that your PSU is underpowered in general and therefore not distributing its load properly between the motherboard and the GPU, so therefore under full load it just restarts to protect itself and your components.

One very less likely issue is a faulty motherboard power regulation. Like you said, the voltage drop is suspiciously low, although still within spec. If you try a new more powerful PSU and still see the same issue, the most likely culprit is the motherboard. Again I repeat, OP is talking about the power supplied by the PCIE SLOT.

Dear ppl of TPU, is it so hard for you to read something properly and slowly and thoroughly? If you can't be bothered to read, don't try help and post. Posting the wrong information cost people hard earned income and maybe even their lives in the process.

 

[jsfitz54]

The 12V is low. 11.9V is about as low as it should go on a dip. Beyond that and you are under-powered.

But if the 5V and 3.3V dip extra low the PC will Black screen.

Please post the value of the 2 extra ranges.

Here's a comparison:

 

[Norton]

the many expert wannabies above me don't seem to realize that the pcie +12V input voltage refers to the power supplied by the PCIE slot. Not the GPU 8 pin, which is the one below it.

Dear ppl of TPU, is it so hard for you to read something properly and slowly and thoroughly? If you can't be bothered to read, don't try help and post. Posting the wrong information cost people hard earned income and maybe even their lives in the process.

If you have some information or want to provide guidance to the OP then go ahead and do so- the additional replies are far from necessary.

 

[phill]

Might not really help massively but have you a power meter that you could attach to your PSU power cable and see what the unit is actually pulling? If it's close to the 450w on the 12v line (which I'd very much doubt) then it would give reason to the issues. I have a 1070 and a 8086k running together and that pulls about 220w or so. In fairness the PSU is I think an 850w but I'd definitely like to presume that the PSU is the cause of the instability.

Another option would be, if possible, switching out a few things to see if the same issues carry on or not. Another CPU or GPU be great but if not, then maybe see if a 650w unit can be had? Suggest something Gold rated and a decent make, Seasonic, Super Flower, etc. I've been personally put off by some cheaper EVGA models (G+ 650w and 850w) as they seem to like quitting on me without much load on them....

 

[ThrashZone]

Hi,
Yeah contact evga they should give more diagnostics to run

I'd also first check the wall plug and see just how many other rooms/ plugs/ lights run off it as well.

Default cpu/ gpu turbo clocks and just running xmp profile on the memory should still be fine on 450w if it's working
Recommended psu spec's allow quite a bit of overclocking room.

One of these comes in handy to see what's being pulled from the wall

https://www.amazon.com/Power-Meter-Kill-Hx2-3-Wx1-5/dp/B00R03CVEI/ref=sr_1_4_mod_primary_new?keywords=kill-a-watt&qid=1677948125&sbo=RZvfv%2F%2FHxDF%2BO5021pAnSA%3D%3D&sr=8-4

 

[D007]

the many expert wannabies above me don't seem to realize that the pcie +12V input voltage refers to the power supplied by the PCIE slot. Not the GPU 8 pin, which is the one below it.

Please don't wiggle your 24 pin motherboard connector, which is where the PCIE slot power comes from. One wrong move will bring down entire PC, not just the GPU.

The PCIE GPU 8 pin is fine, as noted by the values BELOW what the one OP is talking about

I'd argue that while the PSU is a cheap budget model. It isn't by all means a bottom of the barrel terrible choice to pick for a budget build, as verified and tested and praised by TPU's own PSU Tester (albeit for a rival website):

EVGA 450 BT PSU Review: Amazing Value At $25

Is it possible for a PSU that costs $40 (and can sometimes be found for as little as $25) to offer decent performance and reliability? EVGA's 450 BT promises both, appealing to enthusiasts on a budget.

www.tomshardware.com

Its more likely that your PSU is underpowered in general and therefore not distributing its load properly between the motherboard and the GPU, so therefore under full load it just restarts to protect itself and your components.

One very less likely issue is a faulty motherboard power regulation. Like you said, the voltage drop is suspiciously low, although still within spec. If you try a new more powerful PSU and still see the same issue, the most likely culprit is the motherboard. Again I repeat, OP is talking about the power supplied by the PCIE SLOT.

Dear ppl of TPU, is it so hard for you to read something properly and slowly and thoroughly? If you can't be bothered to read, don't try help and post. Posting the wrong information cost people hard earned income and maybe even their lives in the process.

Ok, I’ll say it. No need to be a smug jerk to people.
Yes, that’s what you just came across as and it’s completely unnecessary.

My first step would be to revert the system to defaults in bios, then test it. It could just be an unstable overclock. Bad memory and CPU overclocks can cause a PC to shut down. If you reset bios defaults and it keeps happening, it’s most likely the PSU in my humble opinion.

 

[qung3]

Ok, I’ll say it. No need to be a smug jerk to people.
Yes, that’s what you just came across as and it’s completely unnecessary.

My first step would be to revert the system to defaults in bios, then test it. It could just be an unstable overclock. Bad memory and CPU overclocks can cause a PC to shut down. If you reset bios defaults and it keeps happening, it’s most likely the PSU in my humble opinion.

Already checked with xmp disabled and with another ram, tm5 all pass without errors

the many expert wannabies above me don't seem to realize that the pcie +12V input voltage refers to the power supplied by the PCIE slot. Not the GPU 8 pin, which is the one below it.

Please don't wiggle your 24 pin motherboard connector, which is where the PCIE slot power comes from. One wrong move will bring down entire PC, not just the GPU.

The PCIE GPU 8 pin is fine, as noted by the values BELOW what the one OP is talking about

I'd argue that while the PSU is a cheap budget model. It isn't by all means a bottom of the barrel terrible choice to pick for a budget build, as verified and tested and praised by TPU's own PSU Tester (albeit for a rival website):

EVGA 450 BT PSU Review: Amazing Value At $25

Is it possible for a PSU that costs $40 (and can sometimes be found for as little as $25) to offer decent performance and reliability? EVGA's 450 BT promises both, appealing to enthusiasts on a budget.

www.tomshardware.com

Its more likely that your PSU is underpowered in general and therefore not distributing its load properly between the motherboard and the GPU, so therefore under full load it just restarts to protect itself and your components.

One very less likely issue is a faulty motherboard power regulation. Like you said, the voltage drop is suspiciously low, although still within spec. If you try a new more powerful PSU and still see the same issue, the most likely culprit is the motherboard. Again I repeat, OP is talking about the power supplied by the PCIE SLOT.

Dear ppl of TPU, is it so hard for you to read something properly and slowly and thoroughly? If you can't be bothered to read, don't try help and post. Posting the wrong information cost people hard earned income and maybe even their lives in the process.

This is what driving me nuts, because the low voltage is on the pcie slot which is powered through the 12v in the 24pin connector, my mobo has 2 pcie slot so im going to move down the gpu and see if its make any difference.

The 12V is low. 11.9V is about as low as it should go on a dip. Beyond that and you are under-powered.

But if the 5V and 3.3V dip extra low the PC will Black screen.

Please post the value of the 2 extra ranges.

Here's a comparison:
View attachment 286378

Forgot to upload them

https://imgur.com/a/eC79V2g

 

[Bill_Bright]

As I noted above, the 11.45V is a "bit" low, but still within required standards. Your PSU may be the problem but it is NOT because the 12V is running at 11.45V

And while 450W is not big, it "should" be big enough because that Ryzen 3 and your graphics solutions are not power hogs.

So "IF" the issue is the PSU, it is because that specific PSU is not meeting its own published specs. So again, until you swap in a known good PSU, all you are doing is treading water, and getting tired.

As far as the card makers recommending 500w, it is important to remember two things. (1) They have no clue what the other hardware in the computer is and (2) no way do they want to recommend an underpowered PSU.

So they base their recommendations on mid to upper power hogs. Likely a decent i5 or even i7. The OP has an i3.

Even PSU calculators pad their results. They don't want to recommend a PSU that is underpowered either, so they all pad the results, some more than others. The eXtreme OuterVision PSU Calculator is and can be the most conservative for 2 main reasons. (1) They have a team of researchers on staff constantly researching components for us to keep their extensive databases accurate and current. And (2), it is the most flexible and has the most extensive databases of available options you can enter. This allows it to factor in all possible components to accurately calculate our needs rather than guess.

Plugging the OP components in and you will see 450W is plenty - assuming it is working properly and to specs.

 

[jsfitz54]

SEE post #14 above.

The new info for post # 13 looks good for all rails, 12V @11.933, 5V, 3.3V.
The 12V does seem to dip lower than it should. 11.563V on dip and it's not holding at the 11.9V.
So either under-powered, aging poorly, or some other cause in your environment.

 

[qung3]

SEE post #14 above.

The new info for post # 13 looks good for all rails, 12V @11.933, 5V, 3.3V.
The 12V does seem to dip lower than it should. 11.563V on dip and it's not holding at the 11.9V.
So either under-powered, aging poorly, or some other cause in your environment.

The picture is a bit missleading because I took the screenshot after heaven benchmark finished his run, thats why the 12v looks normal, under test the 12v in the mobo stay between 11.500v ~ 11.600v, the gpu 12v input voltage stay around 11.4Xv - 11.5Xv, the psu is near 5 years old.

I ordered a multimeter to confirm these readout in hwinfo, also with it I can check the psu while playing PoE (since is the only game that currently crash my system) with my phone camera pointing at the multimeter to see what was the last readout before shutting off, Im also downloading 3d mark demo to run some test tomorrow.

 

[AsRock]

11.4V is really low and the PSU is basically a bottom of the barrel low end unit.
i'd say it's 99.9% the PSUs fault. replace it with whatever is a reliable brand and doesn't cost too much.

Yeah company's like Seasonic would have you sending that in.

 

[GerKNG]

The PCIE GPU 8 pin is fine, as noted by the values BELOW what the one OP is talking about


the many expert wannabies above me don't seem to realize that the pcie +12V input voltage refers to the power supplied by the PCIE slot. Not the GPU 8 pin, which is the one below it.

i know that. i never said something else.
11.4v is really low AND the PSU is basically a bottom of the barrel low end unit.
this can still be a faulty PSU since this voltage comes the 24 Pin and should be 12V even under full load.
there is no VRM on the board generating power/voltage for the PCIe Slot...

 

[Bill_Bright]

There's a lot of misunderstanding going on here.

First, and again, the sensor on the PCIe slot should NEVER be considered an accurate voltage measuring device. At best, it tells us voltage is present.

As @jsfitz54 correctly noted, if you look at the figures in post #13, we can see the +12V is well within the allowable tolerance, with an average at a very acceptable 11.683VDC.

Even if you insist on using the PCIe sensor, look at the 12V input in post #1. It is averaging again, a very acceptable 11.619VDC.

It is NORMAL for voltages to vary a small amount as the load varies and in this case, we are talking a few 10ths of a volt - as determined by a very imprecise sensor.

11.4v is really low AND the PSU is basically a bottom of the barrel low end unit.

The PSU's 12VDC output is NOT 11.4V. It is nearly 11.7V. And while that PSU may be near entry level for EVGA, but it is far from being "a bottom of the barrel low end unit".

Let's remember, it is powering a little i3 and a graphics card that could hardly be called a power hog.

Again, I am NOT saying that PSU is good. I am saying the fact its 12VDC output is slightly low is not the cause of the OP's problems.

Remember, the reason why the ATX Form Factor standard allows the +12VDC voltage to be as low as 11.40VDC is because the connected components are required to tolerate and, if necessary, compensate for those voltages.

I ordered a multimeter to confirm these readout in hwinfo

That's fine but IN NO WAY will a multimeter tell you anything conclusive!

To properly and conclusively test a power supply unit, it must be tested under a variety of "realistic" loads then analyzed for excessive ripple and other anomalies that affect computer stability. This is done by a qualified technician using an oscilloscope or a dedicated power supply “analyzer” - sophisticated (and expensive!) electronic test equipment requiring special training to operate, and a basic knowledge of electronics theory to understand the results.

Most multimeters cannot measure anomalous AC components riding DC current, even if the user knew how to perform such tests, and understand the results. Therefore, conclusively testing a power supply is done in properly equipped electronics repair facilities, by qualified technicians.

Even many smaller shops with able technicians don't have the resources ($$$) to buy the necessary test equipment. This is why swapping in a known good power supply, a tried-and-true troubleshooting method used by amateurs and professionals alike for decades, is typically the best and easiest, conclusive alternative.

 

[qung3]

I borrowed a multimeter and checked the psu while shorting the power on + ground pins, everything seem to be "fine" the only oddity is that the -12v pin was at -11v, when I turn on the pc this readout go all over the place from -11.1 to -11.8, is not stable at all, the 12v is at 11.9 idle then during a benchmark it drop to 11.7, if the -12v is not used nowadays, why is that voltage not stable? I checked another PSU I have and that one reads solid stable -11.7, I cant run the pc with that unit because its an old one that doesnt even have a pcie connector and I would have to use a bunch of splitters.

I dont have the tools or the knowledge to fully disassemble the psu to check the capacitors, I still have my old mobo but I would have to buy some thermal paste to move the cpu over.

 

[Hankieroseman]

I borrowed a multimeter and checked the psu while shorting the power on + ground pins, everything seem to be "fine" the only oddity is that the -12v pin was at -11v, when I turn on the pc this readout go all over the place from -11.1 to -11.8, is not stable at all, the 12v is at 11.9 idle then during a benchmark it drop to 11.7, if the -12v is not used nowadays, why is that voltage not stable? I checked another PSU I have and that one reads solid stable -11.7, I cant run the pc with that unit because its an old one that doesnt even have a pcie connector and I would have to use a bunch of splitters.

I dont have the tools or the knowledge to fully disassemble the psu to check the capacitors, I still have my old mobo but I would have to buy some thermal paste to move the cpu over.

When my first PC went BANG! one day, I've been overkill on PSUs. I know they are not cheap like a lot of stuff lately. I gotta waste my pension check on something. You can't take it with you.

 

[Shrek]

I borrowed a multimeter and checked the psu while shorting the power on + ground pins, everything seem to be "fine" the only oddity is that the -12v pin was at -11v, when I turn on the pc this readout go all over the place from -11.1 to -11.8, is not stable at all, the 12v is at 11.9 idle then during a benchmark it drop to 11.7, if the -12v is not used nowadays, why is that voltage not stable?

Some supplies need a load to be stable, but I thought those days are over.

 

[D007]

Bro. Are you overclocking anything? I said revert ALL overclocks and load failsafe defaults in bios on everything... And you respond with "I set the ram to default"
Reset factory defaults on the ENTIRE system in bios if you have changed anything at all. Load failsafe defaults. Test..

 

[johnspack]

11.4v should not happen, period. My 12v never goes below 12v, and I have a power hog of a gpu and cpu. Please try a new psu, or borrow one from a friend.

 

[Deleted member 185158]

I borrowed a multimeter and checked the psu while shorting the power on + ground pins, everything seem to be "fine" the only oddity is that the -12v pin was at -11v, when I turn on the pc this readout go all over the place from -11.1 to -11.8, is not stable at all, the 12v is at 11.9 idle then during a benchmark it drop to 11.7, if the -12v is not used nowadays, why is that voltage not stable? I checked another PSU I have and that one reads solid stable -11.7, I cant run the pc with that unit because its an old one that doesnt even have a pcie connector and I would have to use a bunch of splitters.

I dont have the tools or the knowledge to fully disassemble the psu to check the capacitors, I still have my old mobo but I would have to buy some thermal paste to move the cpu over.

The PSU is junk. Crap. Garbage. A POS.

By no means should it dip below 12v if it had quality components inside.

Also, 450w (or any advertised wattage) is a peak output. The PSU is far from able to sustain this.

It's a whopping 35a 12v rail, actually only rated 420w peak.

That's only 5 amps more than the +5v rail of the PSU I'm using.... for a silly comparison.

 

[Dr. Dro]

My friend you are using (an old, but still) high-performance segment GPU on an $25 power supply, that is all you needed to say.... your PSU was simply not designed for this type of workload, it was fine when you were running the 2200G's integrated graphics but that's where it stops. Theoreticals really don't matter, you need to purchase a decent power supply, IMO, in the 700-watt range to acommodate for aging and a CPU upgrade in the future (that CPU is REALLY bad, to the point it will bottleneck even your old 1070).