Is this a motherboard fault?

Meridian · 21 Jul 2019 at 07:33

I have a second computer which only gets used a couple of times a week, and spends the rest of the time switched off. Yesterday, I powered it on and and it started up, but the screen never came on. I shut it down, switched off the mains, cleared the capacitors, then tried again. This time it booted without issue. It needed to update a driver, so I told it to reboot. Again, the screen failed to start up. I also noticed that HDD light was off.

After a series of reboots, none of which got beyond the BIOS screen, and most of which failed to get that far, what I noticed was that

a) Sometimes everything is running, but the boot has clearly stalled. Again, the HDD light does not flash, except briefly when it powers on. This board does not have LEDs to tell you the boot stage.

b) Sometimes the fans on the GFX card and the water pumps slow down and stop after about five seconds, but everything else stays on. This includes the lights on the GFX, and all the case fans. A couple of case fans drive off the motherboard, the rest off off a molex-connected fan controller.

c) I cleared CMOS (in case it was an o/c issue) and managed to get into BIOS. The trouble is, the boot will only go into BIOS: when you tell it to continue, it actually reboots, which means another failed boot. But I seriously doubt this was anything to do with the overclock. And this trick often fails, with the boot stalling long before the BIOS screen.

The other likely cause, the PSU, is about five years old, and more than enough power for the job. But in my experience PSUs tend to just go bang.

Specs are:

Intel i5820K - was at @ 43 x 100 (4.3GHz), now at BIOS default
Swiftech H20-320 w/c kit, with Apogee GTZ block
MSI X99A Krait Edition
4 x 4GB Corsair Vengeance PC4-25600 3200MHz RAM
EVGA GTX980Ti Superclocked
500GB Samsung 860 EVO SSD
2 x Seagate Barracuda 2TB 7200 HDD in RAID 1
Corsair Obsidian 800D, Corsair HX1200i PSU

MY feeling is that this either PSU or motherboard. Unfortunately I don't have a spare of either I can easily test with, and both are expensive items to buy if they turn out to be un-needed.

The PSU is modular, so I'll try different sockets and see what happens. That at least I can doo without expense. But my feeling is that this is motherboard. Does this seem reasonable?

Dirk Diggler · 21 Jul 2019 at 09:26

Sounds like a rogue USB device.

Meridian · 21 Jul 2019 at 10:38

That's easy enough to test as there are only about three plugged in. I'll try it.

chrcoluk · 21 Jul 2019 at 11:12

yeah try unplugging usb, if no go try unplugging any i/o device like hdds or ssd even if boot device just to see if post is fixed.

If all that is a no go, my prime suspect would be the board.

Ram could cause this tho. So putting it at jedec settings is also worth a go. I dont think a cpu overclock would cause this, especially as also most boards dont activate the overclock now until in the OS.

Meridian · 21 Jul 2019 at 12:09

OK, so I unplugged all the USB devices, including the internal connections from the motherboard to various readers. Nope. There's not much scope to move PSU connections. I'll try the HDDs at some point, although all are less than a year old.

JarppaK · 22 Jul 2019 at 07:17

I give few might try things.

Try 1st if your monitor cable is broken to use other cable to test that. Then do test another graphic card as well if change a cable do not work. Because monitor stay offline it tell somehow that it do not get any signal there.

Unstable OC usually boot computer and then you have to enter bios at 1st to see oc parameters again. This in many occasion notify at start at OC instability issue.

Does it give any notify after bios screen, that operating system missing or not found? This I really doubt because of the screen not come on. If HDD is gone you can only enter bios not more than that.

Meridian · 22 Jul 2019 at 21:24

The machine doesn't get as far as the GFX card as far as I can tell. The thing is deffo powered up, but the boot seems to stop before the GFX are enabled. I plan to swap it out to test - that's something I do have a spare of - but I'll bet money it's not the cause. Also, I said earlier that the overclock has been removed, and it's trying to boot at defaults. It either never gets as far as BIOS notification, or on rare occasions it happily boots into BIOS. It just then reboots, and stalls.

JarppaK · 23 Jul 2019 at 10:39

Also one what came in my mind. If there is a bios what is some how corrupted. This is a long shot, but not what I haven't seen before. Because you also got OC settings earlier. Later you put bios back to defaults. At that point from OC --> to default settings might crash bios or write incorrect.

My AMD Threadripper setup got few months back corrupted bios some how. 1st OC settings work fine long time. OC settings was stable even high load of testing. One day Windows restart with out bsod. After that I could only enter bios because OC settings failed. So this point I need to enter bios what was working only default settings. I need to update bios what fix issues at once.

Some how my case bios was corrupted so computer only working default bios settings.

Its a long shot, but worth a try maybe.

Edit. Many motherboards should give fault light on at cpu, hdd, memory etc. on startup when testing system. If system finds error one of these it won't shut down that light.

orbitalwalsh · 23 Jul 2019 at 10:44

mobo battery ?

chrcoluk · 23 Jul 2019 at 14:08

Yeah I had a NUC that wouldnt post, but I knew it was powering up as the internal LED lit up, I swapped the battery and it came alive.

Meridian · 24 Jul 2019 at 20:52

I'll give it a go. I can't do much before the weekend, but if I get time I'll add it to the list - thanks. Motherboards are generally far more reliable than they used to be, so it's rare for me to need to diagnose one these days.

Meridian · 27 Jul 2019 at 10:38

Update. I finally had a chance for a better play. The CMOS battery is fine (3.15V). I tried a different GFX card, same result. I unplugged everything except the RAM and the GFX. Nothing. I'm pretty certain it's the motherboard. Assuming I'm right this is my fourth motherboard failure in twenty years (2 x MSI, 1 x Asus, 1 x EVGA), but the only other component I've had fail is a gfx card, and that was my fault.

Thanks for suggestions everyone.

freddie64 · 31 Jul 2019 at 05:50

If you can get in to bios, check what the 3.3v is at.

Had a similar problem and mine was only on 2.8v. All i had to do was remove my 24 pin and GPU extension leads and it went back to 3.29v and rock solid again. Not sure if it was the extensions or just the fact i reseated them but it is worth a shot.

Meridian · 1 Aug 2019 at 21:00

I can't get into BIOS any more, and haven't been able to for a while.The trick of resetting it to get in no longer works.

freddie64 · 1 Aug 2019 at 21:33

Have you tried the second bios?

Meridian · 3 Aug 2019 at 07:07

No - I've been a way for a few days. I'll give it a try, but I'll bet you £5 to a charity of your choice that it doesn't get me into a full boot.

Update: no, doesn't work. The board just went into a boot cycle, with each cycle only about five seconds long. I then discovered that of the four RAM sticks, the two on the left (nearest the backplane) were warm, but the other two weren't. I took out the two hot ones, and now the boot cycling stopped, but it didn't finish the boot either. My money is still on motherboard.

Meridian · 25 Aug 2019 at 08:45

More updating.

After trying everything else I could think of (different PCIe slot for the GFX, different card, one RAM stick at a time in different slots etc) I gave up. The main thing that I noticed on all these attempts is that within seconds of booting up, the RAM would be insanely hot. I meant "you can't touch this" hot. I am guessing that the RAM voltage regulation has failed.

So I bought a clearance Socket 1511 motherboard, which meant new clearance CPU as well. Installed them yesterday using the old RAM. No boot. Tried one stick on its own - no boot. I then fitted a stick of RAM out of my main machine instead, and then it booted fine. It looks like the motherboard failed, and took out the RAM as well. I still need to test each stick in turn to see if any of them work, but it's not likely. It's annoying enough that the motherboard blew, but it's four years old so it got a fair turn. But it's much more annoying that it blew the RAM as well.

LuckyBenski · 25 Aug 2019 at 11:11

Did you try your main system's RAM in the setup before swapping motherboard/CPU?

Meridian · 25 Aug 2019 at 11:57

Sadly, no!

But I still can't think of any reason why RAM would exceed 70oC when not overclocked or over-volted (BIOS was reset, if you read above) if the motherboard which was working properly. Also sadly, the mucking about of putting the old board back in isn't worth the bother (even the water block had to be changed) especially given the high probability it will bork the good RAM. Which was why I didn't try it before.

I've fallen for a similar miscalculation in the past. I thought the PSU of a system had blown (it had), so I replaced it. The faulty motherboard then blew the new PSU as well. In my experience it's always the motherboard.

Nah, I'm just of on a RAM hunt.