Supermicro PWS-0060 *failure*, please advise

[willawake]

Re: Supermicro PWS-0060 *failure*, please advise

About the revision question:

I see where yours has "REV 3.0". My unit has nothing there, ie. just white space on the label.

means either it is revision 1.0 or someone forgot to put the revision. if it was fake it would not have those caps.

[Super Nade]

Re: Supermicro PWS-0060 *failure*, please advise

I was joking mate.

I think the connection may have been loose. I'll have to check it again. Currently burning in components. Man, do I miss AMD or what. Stupid Multi locked Intel.

[tiresias]

Re: Supermicro PWS-0060 *failure*, please advise

Currently burning in components.

Hehe - funny expression. Sounds scary.

[Super Nade]

Re: Supermicro PWS-0060 *failure*, please advise

O.K. Absolutely confirmed. The 12V line drops to 0.888 V and then I'm stuck in a reboot loop. The VCore I am applying is 1.55V. Now remember I have almost no load on the 5V and 3.3V lines.

I'll add a few HDD's and see what happens.

[tiresias]

Re: Supermicro PWS-0060 *failure*, please advise

O.K. Absolutely confirmed. The 12V line drops to 0.888 V and then I'm stuck in a reboot loop. The VCore I am applying is 1.55V. Now remember I have almost no load on the 5V and 3.3V lines.

I'll add a few HDD's and see what happens.

Hmm... That's very odd! Wouldn't Jonny have noticed such a disastrous issue during his tests though?

Also, hang on a minute...

Why should a reboot have that effect on the PSU?

[Super Nade]

Re: Supermicro PWS-0060 *failure*, please advise

I have to hurry home now. But here is a really bad video of what happens. Sorry I can't use lights as I'm working with IR lasers and I need to take data. I will take a better video when I get back from home.

This happens as memtest is loading, not when it reboots. It reboots because of PSU failure, I think. It reboots exactly when the voltages start dropping like crazy. I hope this unit does not take out my mainboard.


mvi_0048.avi

[tiresias]

Re: Supermicro PWS-0060 *failure*, please advise

I see it! Bizarre - it's as if the PSU goes out for a moment.

The strange thing is this:

On the one hand:

Many people, including a contact of Galvanized who owns a server farm, have seen these PSUs run dual Xeon servers 24-hours a day, with not so much as a hiccup. Supermicro themselves recommends these units for remarkably powerful and high-end workstations, such as FBDIMM equipped dual Woodcrest boards!

And yet, on the other hand:

I have a unit that fails after months of really light use (which could just be the odd lemon, OK).

And now you're seeing something extremely odd yourself.

.....

Could this be some kind of incompatibility? As in, the PSUs are very solid, but don't entirely fit the EPS standard, and are specially designed for Supermicro boards, and can have problems in other setups?

What do you think?

[Galvanized]

Re: Supermicro PWS-0060 *failure*, please advise

Minerva has not answered my PM of a couple days ago...She has not been on.

There is a minimum load spec for the 3.3 & 5V lines. I want to see what happens when that's met. I got 15 of those 2ohm@25W ceramic ballasts here, wish I could toss you a few. bgmicro.com for $0.49 each.

[jonnyGURU]

Re: Supermicro PWS-0060 *failure*, please advise

I think I had what could be considered at least minimum load on the 3.3V and 5V rails and saw that same drop in the 12V that Super-Nade is seeing.

[tiresias]

Re: Supermicro PWS-0060 *failure*, please advise

I think I had what could be considered at least minimum load on the 3.3V and 5V rails and saw that same drop in the 12V that Super-Nade is seeing.

All the way to zero? I know you mentioned there was a sharp 0.1V drop, but Super Nade's unit seems to be practically shutting off for a moment...

[jonnyGURU]

Re: Supermicro PWS-0060 *failure*, please advise

No. Not all of the way to zero. But a significant drop just the same.

Perhaps Super-Nade's hard drop was, as GY stated, due to even less of a 3.3V and/or 5V load as me. But I had a load on all rails none the less.

[Super Nade]

Re: Supermicro PWS-0060 *failure*, please advise

Well, I've collected 4 HDD's and 2 Opticals so far. I'm gonna load up the 3.3 and 5V till she explodes. I'll be on pretty late tonight, gotto finish up a paper, so hopefully I can keep you lads posted.

Hey Galvo, it would be great if you could toss me a few of those resistors. Do I need to actively cool them?

[Galvanized]

Re: Supermicro PWS-0060 *failure*, please advise

Sure! PM me an address and I'll drop some in the mail. Two in parallel would load the 5V by 5A and the 3.3V by 3.3A.

PCP&C sells a resistor to load the 5V line by 1A/5W and that helps stabilize a lightly crossloaded PSU. I'm going to make a lite-duty load tester for PSUs out of some of them, just to bench check voltages. One might go on the Zippy because the 5V rail puts out 40A and the 12V rail is 30A. Might need to do this if the 12V rail ever needs to be fully loaded.

They are 15mm in dia., 50mm long and are hollow tubes. Yes, I would activly cool them. Mount them over the case exhaust fan or over the PSU exhaust fan. Best would be inside the case...over the exhaust fan, I guess.
Oh! They take blade connectors...and I don't have any.

Update your profile. Where are you located? PM the address.

[Super Nade]

Re: Supermicro PWS-0060 *failure*, please advise

yhpm.

I'm trying to rip out as many HDD's I can from the computers lying waste in the department at my Uni. It's gonna take a while.

[tiresias]

Re: Supermicro PWS-0060 *failure*, please advise

I've discovered something else in the last hour that may be of interest...

The fan control board, on the 0060, is indeed more complex than you would expect. The functioning of the PSU itself depends on the presence of a connected fan - it's able to "notice" this even when it's using a two-wired fan (ie. with no tach output), probably by measuring current draw.

- With no fan connected to the two-pin plug, the PSU will turn on for an instant and immediately cut off again.

- If the fan is at any point disconnected, the PSU shuts everything off, except 5Vsb.

This makes fan modding considerably more difficult.

...

PS. I guess the designers of the PSU felt that it's better that the machine should cut off in case of fan failure, rather than continue running and risking damage due to an overheated PSU. It could be a source of PSU problems though, what with not following the "KISS" principle.

[Super Nade]

Re: Supermicro PWS-0060 *failure*, please advise

Mate, there is a 4 pin connector as well (thin wire), where should I plug that in? I have not plugged in any fan wires from the PSU to the MB. All the MB headers are 3-pin. Which one should I leave out? I could try forcing it into the CPU fan header (but the notches don't line up) and move the CPU fan power elsewhere.

[Super Nade]

Re: Supermicro PWS-0060 *failure*, please advise

Very very weird. Now this is worrying me a bit.

I have now attached, 1 Raptor 36Gb, 2 Opticals, 1 IDE. On the first attempt, I saw no crazy drop to 0V. On a subsequent Reboot, this phenomenon is repeating itself. I'm gonna try adding a couple of more HDD's, a few PCI cards and see what happens.

Interestingly, I witnessed NO such problems with the DFI setup I have:

DFI NF4 SLI DR
X1800xt
1 Gb BH5
2 x Raptors
1 x 120mm and 2 x 82 mm fans
1 SB Live PCI card

Is it the ASUS board which is messing up??

[tiresias]

Re: Supermicro PWS-0060 *failure*, please advise

Is it the ASUS board which is messing up??

You could perhaps check that (at least in part), by measuring potential between PS_ON and ground. Now while the motherboard and PSU are on, potential should be always low (zero, basically).

If the motherboard, for some reason would be 'tripping up' the PSU during a reboot, you might see PS_ON go high (usually 5V) just before the voltage drop.

About the little connector - the 0056 has it, but the single-fanned 0060 does not...

[Rainbow]

Re: Supermicro PWS-0060 *failure*, please advise

Yes, the board might turn off the PSU for a moment during reboot. Some boards do weird things...

[Super Nade]

Re: Supermicro PWS-0060 *failure*, please advise

You could perhaps check that (at least in part), by measuring potential between PS_ON and ground. Now while the motherboard and PSU are on, potential should be always low (zero, basically).

If the motherboard, for some reason would be 'tripping up' the PSU during a reboot, you might see PS_ON go high (usually 5V) just before the voltage drop.

About the little connector - the 0056 has it, but the single-fanned 0060 does not...

So do I go ahead and jam the connector into the header?

Anyway, I made a few more videos.

This is what I'm running:
http://www.zippyvideos.com/672126498...44/*snocforums

This is what happens:- [FF to the end to see the same effect]
http://www.zippyvideos.com/971291187...46/*snocforums

Now, I'm thinking, maybe the Presscot is overvolted a bit too much. But that does not explain why the hell the 12V line should drop to zero? Absolutely no probs with the DFI setup. I'm gonna unplug and reconnect everything again.