PSU or motherboard

[torin3]

I've got a problem and I want to see if I can narrow it down to PSU or motherboard.

This is a SuperMicro H8QG6+-F 4 processor motherboard that had 4 12 core 6100 series CPUs in it. Powered by a Corsair RM1000 PSU. It had a USB optical mouse, a USB keyboard, and 4 120mm fans plugged into the motherboard fan headers. 1 SATA HD running on a native SATA power cable (no adapters). And 16 1GB sticks of DDR3 ECC Registered RAM.

It ran for 3 days, apparently fine. Then it stopped working. Trying to figure out the problem, I saw what is in the picture above. 1 burnt pin and what appears to be some scorching. It was running on a new surge suppressor strip, but not a UPS.

Now I've been having a lot of trouble with a similar 4P board and RMA was finally refused on it. The only things I've recycled from that experience was 3 CPUs, which were tested in a 2P SM board, and the RAM also tested in the 2P board, and the CPU HS/fans.

I purchased the RM1000 new specifically for this system. And it still seems to power the 2P SM board, though I haven't run it for more than about 20 minutes.

I'm probably only going to be able to RMA the new board once, and considering they run almost $800 a pop, I want to make REALLY, REALLY sure this isn't going to happen again.

You can read my 4P folding thread in the folding sub-forum for the full tale of woe.

Can anyone help?

Thanks!

 

[Automata]

That is pin 9, which is 5v standby for the motherboard. Did this happen while it was on?

I'd start by checking the voltage of that pin on the power supply. If it is within spec (5%), then I'd guess the issue is with the board.

 

[Silver_Pharaoh]

Damn torin this sucks

Best place to start is thid's advice

 

[torin3]

I will test when I get home from work today.

It did happen when it was on. It was actually on and running for 3 days before this happened.

 

[Bobnova]

Probably the board, but the CPU is dead now too. That pin will have gotten hot and lost its temper in the PSU connector.

The RM series is not especially fantastic, as a note. There's a reason it's so much cheaper than other Gold PSUs. I hope it is, anyway.
It's not bad, but it's not that close to Corsair's other Gold units.
Think of it as a CX with better efficiency.

 

[ATMINSIDE]

Probably the board, but the CPU is dead now too. That pin will have gotten hot and lost its temper in the PSU connector.

The RM series is not especially fantastic, as a note. There's a reason it's so much cheaper than other Gold PSUs. I hope it is, anyway.
It's not bad, but it's not that close to Corsair's other Gold units.
Think of it as a CX with better efficiency.

+1

OklahomaWolf found LTEC capacitors in the 850W flavor, and wasn't overly impressed.

Not saying its the cause for sure, but I would suspect it over the motherboard personally.

 

[torin3]

Probably the board, but the CPU is dead now too. That pin will have gotten hot and lost its temper in the PSU connector.

I'm confused. Did you mean the PSU is dead now too? Because 2 of the CPUs from the board have been tested in another board and seem to be working. It was also powered by this PSU.

The RM series is not especially fantastic, as a note. There's a reason it's so much cheaper than other Gold PSUs. I hope it is, anyway.
It's not bad, but it's not that close to Corsair's other Gold units.
Think of it as a CX with better efficiency.

So if it was the PSU, and I can RMA it, I may sell it and buy a different PSU. I hear the 1.3KW EVGA PSUs are decent.

 

[Bobnova]

If the pin on the motherboard is torched and nasty the pin that mates with it in the PSU's side of that connector is torched and nasty as well.
If that pin has had issues on other motherboards, blame can be laid on the PSU for this current failure.

I would RMA the PSU.

 

[torin3]

Ok, the 5V rail was 5.1V

This PSU has only been on 2 SM board. 1 2P and 1 4P. I'll take a close look, but I didn't see any damage on the PSU side.

 

[Silver_Pharaoh]

5.1 seems alright...

 

[Bobnova]

Look for any blackening on that pin, or rainbow colors, or anything that makes it look different in any way than the other 23 pins in the connector.
Please note that I'm talking about the metal bit of it that connects electrically to the metal bit in the board, not the plastic bits.
I want to be really clear on this because I have no interest in being part of any confusion that slaughters another $800 board!

 

[torin3]

Look for any blackening on that pin, or rainbow colors, or anything that makes it look different in any way than the other 23 pins in the connector.
Please note that I'm talking about the metal bit of it that connects electrically to the metal bit in the board, not the plastic bits.
I want to be really clear on this because I have no interest in being part of any confusion that slaughters another $800 board!

Understood. My current plan is to contact Corsair and either RMA or send it in for testing. Then sell the verified / exchanged PSU and buy an EVGA 1300 watt Supernova PSU. I believe OW said he was pretty impressed with that model.

 

[torin3]

Here is a close up. Looks like there is some oxidation on the one side.

Since it is a modular cable, could just replacing it be an option, assuming the supply turns out to be good?

 

[Bobnova]

Yeah. That right there may be what killed the motherboard. Well that and the motherboard drawing a hell of a lot more 5VSB than normal. It's also possible the motherboard screwed up and draw a ton of 5VSB and the RM series doesn't have enough OCP on the 5VSB and it cause the oxidation.

In any event, definitely don't use that cable. Replacement cable would be fine, it's that pin specifically that is an issue, not something inside the PSU*




* Odds are excellent this is the case. There's always the statistical outlier.

 

[torin3]

So, would you consider it ethical to get a replacement cable, then sell it, with the note to not use it on a 4P motherboard?

If not, it looks like I'll have to go the full RMA route.

 

[Bobnova]

Sure, as long as you get 5v +/- 5% on that pin when the PSU is in standby mode.

 

[torin3]

Sure, as long as you get 5v +/- 5% on that pin when the PSU is in standby mode.

5.1V is what shows on the multi-meter.

 

[ATMINSIDE]

+5% would be 5.25V, so that's within spec.

 

[larrymoencurly]

Even without OCP, can a +5Vsb regulator put out enough current to burn or blue the connector pin?

Some +5Vsb regulators are badly designed and will briefly put out way too much voltage if a transistor in them fails.

 

[Bobnova]

I've seen some PSUs that connect the 5VSB rail to 5V when the system is on. No clue if that is one of them.
Most PSUs these days are rated for 3 amps of 5vsb, given time that'd be enough to fry something. It does seem slightly odd, though.