New PSU - +12V Rail shows 6V

Sulphur on 4/7/2016 at 18:21

I recently woke up to find my PC had gone into a coma while it was downloading something overnight. Like any good troubleshooter, I turned it off and on again, and ASUS' anti-surge protection gave me a BIOS warning about power surges and faulty PSUs. Fair enough, I thought, and checked the voltages in the BIOS. They were all off. This was an ~8 year old PSU, so I guessed that was its MTBF rate being hit.

So fast-forward to today: new PSU, same anti-surge message. BIOS shows 6v on the 12v rail, and now we come to the crux of the problem. My limited understanding of this scenario tells me I shouldn't even be able to turn on my PC under these conditions, yet here I am typing a post about it on the very same PC after overriding the protection message. So I'm probably killing my PC here... or not. Don't have a multimeter on me, so gonna be freaking hard to figure this one out. I suppose I'm better off getting the PSU replaced again?

Al_B on 4/7/2016 at 20:00

My first instinct is that it's a faulty reading from your BIOS and that the 12V rail (whichever one is being monitored) is very unlikely to actually be at 6V. It could well be the motherboard that is at fault and not the PSU.

One thing that would be worth doing is to power off your PC, unplug anything non-essential that uses the 12V rail (hard drives / DVD) and re-check the bios. If your motherboard has on-board graphics capability then it would also be worth disconnecting and unplugging the graphics card and trying again. This is just on the very slim chance that a peripheral is dragging one of the 12V rails down to that level.

Zerker on 4/7/2016 at 23:18

Do you own a multimeter? The best way to check is to actually just measure the thing :) Find a molex connector that isn't hooked up to anything and stick one probe into the yellow wire and the other into one of the blacks.

Sulphur on 5/7/2016 at 03:06

I was thinking about it being the BIOS at fault, but the other two voltages read on spec with a +0.08 differential in comparison, so I think it'd be unlikely that it's only getting the +12v rail wrong.

I forgot to mention: this is a 650W PSU, I'm running 4 HDDs, a core i5 6600, and a GTX 970 on it, same load and wattage as the last PSU.

I was pretty disappointed last night so I RMA'd the thing, but if the second one comes in with the same issue, that would mean a higher probability of it being a peripheral drain issue (which means the PSU model sucks) or mobo issue (which really would just suck, period, this is a fairly new mobo). I'm certainly going to get a multimeter. :)

Edit: the one thing that bugs me about this PSU is that it only has two six-pin PCI-E connectors, and they share one line. I'm assuming Corsair wouldn't be daft enough to make a design decision like that if it leads to this sort of issue

zoog on 5/7/2016 at 06:13

You may have 12V lamps to use as indicator but surely it's BIOS/south brodge problem.

Al_B on 5/7/2016 at 06:40

Getting a multimeter is the best way to be sure - however you'll need to make sure that you're measuring the same rail that the BIOS is measuring which could be trickier if there are multiple +12V rails in the system (i.e. it may not be as simple as measuring a molex connector). Some motherboards have testpoints explicitly for measuring the power supplies which some overclockers use to measure the actual voltages on the board rather than trusting the readings from the BIOS.

Quote Posted by Sulphur

There are certainly failures that could occur to make one rail measure incorrectly while the others are OK. It's such a drastic difference from +12V to +6V (and suspiciously a division by 2) that makes me suspect it may be either the circuitry or a conversion error on the motherboard. However, if you've already RMA'd the PSU then it's a moot point now until the new one comes in - and it may turn out to be a PSU problem after all. I'd suggest connecting it up with a light load first, just enough to get into the BIOS and work from there when you get it.

Quote Posted by Sulphur

Edit: the one thing that bugs me about this PSU is that it only has two six-pin PCI-E connectors, and they share one line. I'm assuming Corsair wouldn't be daft enough to make a design decision like that if it leads to this sort of issue

No, shouldn't be a problem - but you should be able to check the plate on the side of the PSU or look up its specifications to see what current output is given by the PCI-E connectors and cross-check with the graphics card.

If your problem still persists then it would be useful to know the make / model of the motherboard, PSU and graphics card.

Sulphur on 5/7/2016 at 13:08

You were right, Al. I got a replacement PSU - an (http://www.antecindia.co.in/product.php?id=NzA0MzUz) Antec VP650P. Same deal - voltages are dead straight on +5V and +3.3V, but the +12V averages around 6.5. So that neatly rules out the PSU being an issue, unless I received 2 PSUs from 2 different vendors with the exact same issue, which would just about require this PC to be aboard a spaceship with an infinite improbability drive to happen.

So now the hard part begins. Do I need to send the board in for a replacement? I'm hoping it's just a wonky onboard sensor and nothing major. I'll try throwing some heavy apps (read: games) at it to see if the system buckles. Anything else I should be looking at to run a damage assessment?

The motherboard is an (https://www.asus.com/Motherboards/Z170-P/specifications/) ASUS Z170-P. The GPU is a (https://www.zotac.com/us/product/graphics_card/gtx-970) Zotac GTX 970.

Edit: this is with minimal peripherals connected, and then with everything plugged in. No change to the 12V monitor reading in either situation.

Al_B on 5/7/2016 at 20:21

Thanks - as you say it's unlikely to be the same issue with two PSUs or at least it's unlikely that both would allow the system to start up under those conditions. One thing I'd be inclined to do is update the BIOS (or just refresh the version if you have the latest) - there's a slim chance that it's a bug they've fixed or flashing the bios will refresh some corrupt configuration settings that may have occurred when your PC crashed originally.

In terms of what to look out for - I'd expect crashes, instability or some of the peripherals to have issues if the power rail is at that level on the motherboard (somewhere) and it's not a sensor issue. Of course, it may be that the issue is only noticeable when using some obscure peripheral such as the legacy serial port or the HDMI output with the integrated graphics which you're not using. The safest option is obviously to get the motherboard replaced if that option is available to you but if it works correctly under stress conditions and you're not encountering issues then I would personally be tempted to chalk it up to a duff reading. However, I'd also save up for a new motherboard just in case it gives up the ghost - I wouldn't expect it to take anything else out with it based on what you've posted.

Sulphur on 6/7/2016 at 17:56

Thanks, Al. There has been one weird blip where Win10's System and Memory Compression process started eating around 30% of the available CPU cycles, which may be due to a driver failing to deal with a power transition state properly. Could only fix that one by setting the system to sleep and waking it up. After that, I noticed that the +12v rail reading showed as 3.4v on the BIOS and through monitoring software in Windows.

Fun times ahead, probably. I'll keep an eye out for more issues, as this may mean I'll need a new mobo eventually, if not immediately.

Thelvyn on 8/7/2016 at 14:22

Have you cleared the cmos?
I didn't see anyone mention this but it seems like the first option before replacing anything else or taking more extreme measures.

Sent from my LGMS631 using Tapatalk