Sometimes power, sometimes none.

[Emmon]

Hey all, been having a very strange problem over the last few days that i know is hardware related, but i cant for the life of me figure out how to narrow down the cause diagnostically.

System specs are as such:

AMD Piledriver 8350 @ 4.6Ghz 1.4V overclock
Asus Sabertooth 990FX R2.0
Corsair Carbide 540 Case (2 front fans 1 rear and 1foot air clearance all round, (it has its own little table))
Corsair H100 cooler (max temp cpu after a few hours gaming is 48C)
Seasonic G 750W "80 Plus Gold" PSU
Corsair Vengeance Pro Silver 2x4GB DDR3 1866Mhz Dual Channel

(All above components are 9 months old although H100 was a refurbished unit bought at that time)

Other components:

Boot Drive:
OCZ Vertex 3 60GB 2.2 firmware (3+ years old maybe)
Windows 7 - 64bit (100% legit) Though i've never installed any updates other than ones i select manually (about 3 in total, all in relation to the 8350 utilisation)

Storage / Game drive:
Samsung 840 EVO 120GB EXT0BB6Q Firmware (13 Months old)

GPU:
1 x Sapphire 7950 Vapor-x 3GB ram running 13.11 Catalyst drivers (13 Months old)


Now the problem is odd. I'm out of ideas which is why i'm posting here but it started 2 days ago when i'd just installed uplay then remembered i was going swimming so turned PC off, got ready then thought "ah i'll leave watchdogs downloading while im out, probably take 3 or so hours" and when i went to turn it on.. nothing. Nothing as if the power switch was broken and wasnt sending the message.

No fans spun (including PSU fan), no beeps just a single green light on the mobo, not even the power switch light went on.

So immediately i thought it to be a PSU problem. Turned the psu switch off at the back, took the kettle plug out the back and refitted it after 20 seconds, turned it back on.. still nothing but the single green light on the mobo.

Packed it up its original box with all its gubbinz and was planning to RMA it for testing but while driving it too the postoffice i went to a local computer shop (the kind that live off word of mouth but are really good and bend over backwards for you) and they had a voltage tester there so the connected my PSU up, 12V line showed stable (12.2V biggest measurement) and so i decided not to RMA straight away and do some more testing.

Today -
Reinstalled PSU.. reattached all cables, and re-seated jumpers on mobo.. It POST'd, couldnt read boot devices, then shut down after reaching 52 degrees on cpu

(I then facepalmed and attached H100 and HDDs to PSU)

Then it worked flawlessly, booted fine, it had even saved my overclock settings (4.6Ghz) so left it all alone, put it down to a loose jumper and carried on regardless.

Then around 2 hours ago while having my first big go at watchdogs.. it blue screens (its first ever) and when i try to power it up again.. nothing... same as before, so power, no fans spinning... nothing but the single green light on the mobo.

I go have a bath, come back.. case sides off, finger check all connectors on mobo and HDDs with power cable out.. re-attach and it boots and is stable (hence this post).

So i'm confident its going to happen again because i never found anything that made me go "ahhhhh it was this that was wrong" and to me blue screens mean some piece of hardware is faulty / on its way out.

So im running some HDD diagnostic tools at the moment but incase it dies again i thought i'd throw this post up to ask the Overclocker Oracles for any light they maybe able to shed on this confusing issue.

Thanks for reading, i know i've rambled on here but i felt its best to get as much info out as i could.

 

[stulid]

First thing is to put all clock speeds back to stock and try it again.

 

[Emmon]

Thanks Stulid, actions taken since post:

Mobo BIOS flash from version 2005 to latest (2501)
Overclock removed, Running at stock

Right after to BIOS flash it did refuse to power on again but after 5 mins (PSU unplugged during this time) it booted again

OCZ Vertex 3 (C: ) CHK DSK run at startup - was ready with camera incase of any bluescreens but went through fine

Samsung Evo 840 (Storage) CHK DSK run in windows - no errors.

 

[Adam_151]

Was the power supply checker

a)testing all the voltage rails
b)applying a ballast load while checking voltages

Unless it was doing both these things then its inconclusive, its probably inconclusive anyway as the problem is intermittant

It appears to be failing when the system is warm from what I gather*, I'd be tempted to wait until it fails to start again and try and measure what is happening at the time with a multimeter, i.e. are the standby rails correct, is the system pulling the signal line low to start full output and if so, are the rest of the rails correct and has the PSU given the power good signal back to the motherboard.

It certainly sounds like a power issue, but there will be circuits on the motherboard which are effectivly power supplies etc (to convert what is comming from the psu into appropiate voltages to run the CPU etc.

TD/TR - Probably the PSU, but cannot totally rule out the motherboard

*The failing to power up when warm sometimes points to failed smoothing capacitors in switching power supplies, the values drop significantly when its failed and being warm tends to reduce it rather. One of my LCD monitors would normally switch on fine (4 times out of 5), if it didn't you'd have to let it cool for half an hour before giving it another chance, switch it off when working, reboot the system or let windows put it into power save, and it wouldn't come back on. After 2 years of living with it getting gradullay more unreliable , stripped it down on thursday and soldered new capacitors into it (the old ones were clearly swelled up amnd measured about 10% of the value they should be) and it works fine

 

[stulid]

So is it working ok now after the BIOS flash and returning the speeds to normal? your post isnt 100% clear if it.

 

[deFiniLoGy]

I will try and RMA the PSU anyway.

If your PSU does not power up a single fan with ATX shorted, even occassionally, then there is clearly something wrong with the PSU and you have the right to RMA it if it is the case.

 

[Emmon]

Since i returned the system to stock and removed have updated all SSD firmware, graphics drivers and has been stable all day.

Consider this fixed, if it starts playing up again i'll find this thread again and bump it with new issues.

Was probably the overclock so thx stulid and i'll keep an eye on the PSU voltages over the next few days, see if i see any warning signs.

I'll also bookmark this because of your info adam_151 & deFiniLoGy, and if the PSU does start acting bas at least i know i've got 4 years of warranty left on my Seasonic (also yeah, they only tested the voltage rail at the shop)

Thanks all!

Edit - It almost seems as if my PC is giving me a "timeout" if i try to overclock, as if its saying "Oi! nope.. go and think about what you've done for 5 mins, then come back and use me sensibly " just tried to overclock the fsb a bit and give the core 1.4V again and it gave me the same treatment.