Is My PSU Dying?

D

deadman_uk

[H]ard|Gawd
Joined
Jun 30, 2003
Messages
2,036
For the last month, sometimes when I turn on my PC from cold, it powers up for 5-10 seconds, the monitor remains black (as if there is no signal), then the PC shuts itself down. It then turns itself back on again and either repeat the cycle 2-3 more times if not, will display the motherboard splash screen and then everything loads in fine. Also, it seems often when I reboot my PC, instead of rebooting, it shuts itself down completely once Windows closes.

This is the only problem I am experiencing. I don't have random shutdowns or crashes during normal PC activity or whilst gaming for hours on end. Around the same time, I upgraded from an MSI 7970 OC to an MSI 980 ti OC, however, I am pretty sure I have experienced this problem a few months ago but it happened so infrequently that I tended to not acknowledge it.

Is my PSU failing? I understand it's over 5 years old now and they don't last forever. Or could it be my motherboard? This too is getting old but newer than my PSU. My PC specs are in my signature.
 
I think the CMOS battery is toast.
If you have a meter, check if it's outputting at least 3 volts.
Sounds like the recurring reboot loops is the mobo re-learning proper memory timings and such.
 
I don't have a meter and wouldn't know how to use one if I did. Can I buy a new CMOS battery? are they cheap? and are they specific to each motherboard? or universal?
 
I googled your board and it looks like it uses the standard CR2032 type 'tablet' shaped battery. It's right under the top-most PCIe slot.

They are dirt cheap, but do get one of a reputable brand.

However, you need to be careful whilst removing it.

Definitely begin by disconnecting the power cord and then press the power button to discharge any stray power left in the capacitors.

Then, delicately push aside the little metal strip holding it in place. Use something which won't tear a hole in your board if your hand slips. Heck, a Q-Tip would do fine.

As soon as you simply move the little springy bit aside towards the socket edge, the battery should pop out of the socket.

To place a new battery in the socket, first land the side opposite to the springy bit. Then, the other half should simply slide itself in displacing the spring. The 'spring' will then simply pop into place securing the tablet.
Observe polarity - usually the stamped letters should be facing upwards.
 
Last edited:
There is a round battery on the motherboard, somewhere around PCI-E expansion slots. It will have a specific number on it, 2032, 2016, thats the size of the battery.

Testing it? Put it to your tongue, you might feel a bit of zing on the tongue if the battery is fine.
 
Do not use your mouth for measuring voltages. In this case the available current and voltage is too low to jolt you, but keep in mind the body's natural heart pacemaker nerve runs near the back of your throat.
Also, it's outright wrong because a direct short can damage the battery. And if you're unfortunate enough to try that with something like a lead acid UPS battery - you will get seriously hurt.
 
Yeah it looks like the right kind.

Though it'd be smart to rule out other stuff - like your initial suspect - the PSU - by swapping to a known good one.

While possible, a time period of under 5 years is a bit quick for a mobo battery to fail.
Do you turn off the power to the computer while it's not in use? I mean - switch off the power strip or disconnect the power cable?
 
i personally don't agree with it ONLY being the battery...Getting the new battery is not a bad idea regardless to be honest but it really points to a bad power supply to me. The part about the pc just shutting down points mostly to a failing supply. IF you dont have a meter to test voltages on the different connector leads, I would be curious what software says in the mean time. HWINFO64 is a pretty nice program for at least getting an unverified idea of what could be going on. What i recommend is running a combination of programs to try and draw max wattage at one time and see what happens. IF the supply has issues, it in my mind should shut down when trying to use max power. Anyways i would try running your choice of cpu utility like intel burn test or prime 95 and at the same time run something like furmark to throttle up the video card and see if it can last 5 or 10 minuets without issues. If it can survive that then we can feel at least a bit more comfortable with the supply. Lastly while its running go ahead and monitor the voltages with hwinfo just to see what it says. It goes without saying software measurements don't prove anything for sure but there better than nothing in the mean time.
Regarding the battery, we in the us can pick them up at a lot of local stores here, but its easier to order them maybe in the uk?
 
I would love to pop in another PSU and try it but I don't have another one to test. My PSU is over 5 years old but my motherboard is newer than that. I'll buy a new battery since it's so cheap to try. Buying a battery from a local store requires me to travel by bus since there are no local stores that sell them here (I don't drive) so it's a lot easier and cheaper to buy online.

Thanks for the information about software testing, I shall certainly try this over the weekend and see what happens. But just to make sure you read my post correctly, it doesn't just shut down randomly (this is the impression I get when I read your sentence), but it does when I attempt to reboot.
 
Primetime's suggestion to torture it with furmark, intel burn test, prime, memtest etc is still valid, because both a cold boot and a reboot are specific in that they're a 'zero to sixty' kind of moment, and some specific PSU failures like to manifest themselves during transients like that.
 
I will try this over the weekend but I have opened up HWINFO64 but not sure what exactly I should be looking at for now and when I stress test my PC.
 
It opens with a summary screen, and behind that (close it) there's a 'sensors' button. Basically wherever you see 11-13V that's where the 12V rail is measured, same goes for other major rails - 5V and 3.3V.
Then there's the cpu 'vcore' which tends to droop under load. The question is - how much variation is too much. So sticking to torture testing for extended periods of time should suffice.
 
Capture_zpsy12oxofn.png


In the sensor tab there is a motherboard voltage section. These numbers cant prove everything is ok, but really low numbers or fluctuations we give increased reason to investigate further. FOr somone like me who has already verified the numbers are pretty accurate then i can sorta skip breaking out the meter to test as often
 
I ran Furtest @2560x1440 and IntelBurn Test (standard mode) for over 5 minutes together at the same time. IntelBurn Test displayed a prompt at the 5 minute mark saying it passed. I left Furtest running for a further 5 minutes and experienced no issue. I would run it for longer but IntelBurn Test completes quickly. I had HWINFO64 creating a log in the background and I saved the CSV spreadsheet file. Can anyone take a look? (you may need to download the file as the online viewer doesn't allow you to zoom in enough) I've uploaded it to Google Drive here
 
I took a look at it and didn't see anything that would raise concern. This combined with the fact you were stable drawing max power is a good sign your power supply is good.
 
primetime said:
I took a look at it and didn't see anything that would raise concern. This combined with the fact you were stable drawing max power is a good sign your power supply is good.
Click to expand...

Thanks for taking the time to look at my spreadsheet and reply, I appreciate it. I do notice that when I turn my PC on from a cold boot and it doesn't load, the monitor remains blank. I know when my computer will successfully load in as the monitor would come on and I see the motherboard splash page.. During usage, the computer is solid, even when gaming demanding games like GTA V at very high settings. I am still waiting for the motherboard battery to arrive though.
 
Last edited:
I installed Windows 10 yesterday, not related to Windows 10 but when I shut down to go inside my case and turned on my PC again 5 minutes later, the PC refused to boot up. It was stuck in an endless loop of turn on, remain on for 10 seconds then turn off, remain off for 5 seconds, then turn on, over and over. Looking at the motherboard code, it read 55 which I know is a RAM code. I pulled out all 4 of my RAM sticks and reseated them and the PC then booted up fine.

I've had this happen before and it turned out to be a faulty RAM stick which I got a replacement for. Could this be faulty RAM causing my problem? I will run MemTest86 sometime this week. Also, I installed the new battery but it didn't help (didn't think it would but for £1, it was worth a shot!)
 
HCI memtest in my experience is far quicker at detecting marginal memory.
 
I've not heard of that but will check it out. I know windows has its own memory tester too, is it worth a go?
 
You must log in or register to reply here.
Back
Top