OSCON go BOOM!

[PCBONEZ]

Who says Solid Polymers don't fail?

So, the story..
I set up a dual Xeon skt 771 board on the workbench with one CPU to test some FB-DIMMs.
All the parts involved have been tested before but as the RAM is going into a final build I wanted to test it again.

First booted into the BIOS to check settings.
Clicked save but on restart Memtest [on a USB stick] didn't 'catch' and run.
Not usual for boot from USB to 'miss' on this particular board when doing a hot-restart.
For some reason this board only checks for USB-bootable on cold restarts.
- So I fully power down and reboot.
Comes up and memtest is loading then...

POW!! HISSSSsssssssss SMOKE!! - TURN IT OFF! - TURN IT OFF!

LOUD pop followed by the sound of escaping gas and a large puff of NASTY smelling black smoke.
I didn't see -exactly- where it came from.
At first I thought the south-bridge but here is no sign of a problem there.

Upon closer inspection one of the OSCON's [Vcore for the socket with NO CPU installed] is much 'taller' than it used to and the 'legs' are longer than they used to be.
-
Looks like the can tried to play missile leaving the 'guts' of the cap on the legs,, but the bung did not 'let go',, so my guess it vented along the path were one of the legs passes through the bung.

I haven't pulled the cap yet but I did get a couple snap-shots for your entertainment.
.

[Per Hansson]

Re: OSCON go BOOM!

Hmm, very interesting
One possible reason could be that with no CPU installed there was a fault in the CPU VCore regulator, since it got no feedback it could in theory increase the VCore indefinitely, all the way up to 12v, which would of course cook the 4v OSCONs.

I don't have any valid reason to think _why_ it would do that, especially on a multi CPU board (I presume) where it really must be taken into the design that a CPU slot might be unpopulated...

Only reason I'm entertaining the thought is I know what happens to a CPU when you volt-mod a mainboard to gain more voltage and mistakenly put the variable resistor at min value when you infact where meaning to put it at max, oh those poor Athlon XP CPU's

[PCBONEZ]

Re: OSCON go BOOM!

I think it's just a random cap failure.

Yes. Board is designed for single or dual operation.
Previously ran with Dual CPUs in actual service.

Vcore voltage was fine.
Gives audible alarm if not and that didn't happen.

The board has a lot of built-in over-kill.
Has third [different] VRM just for the RAM.

Not like any one Vcore cap is heavily loaded.
There are 38 of them [all OSCONs] in Vcore [19 per CPU].
That's at least double what is seen on consumer boards with comparable CPUs.

--
Point of posting is that -ANY- kind/brand of cap can fail on occasion.
Even Solid Polymer -and- even OSCON.
.

[Toasty]

Re: OSCON go BOOM!

Fallout from Japan....




Gonna pull the ones to either side to see? Closest to CPU/heat related?

.

[PCBONEZ]

Re: OSCON go BOOM!

Oh it's a good 2+ inches from the CPU.
It's by the empty socket.

Pulled bum cap and the board runs fine now.
I'll just leave it out and use single CPU until I can get a matching cap.
[Gotta keep the VRM color coordinated. They run better like that.]

Not sure if this one has that audible alarm as I thought.
As I recall the BIOS had a setting but I looked and it's not there.
That might have been in an older rev of the BIOS and now it's not there.

Pics of the pulled cap & empty spot attached.
Can see where the smoke left a 'jet blast' mark and a smoke trail on the board.
[Trail is between the two caps in back. Couldn't get enough light in there.]
Looks like the leak started at the leg like I thought.
Not sure if the split happened from the failure or from removal [RoHS solder and that one came out like pulling a tooth] so I may have caused the split after the fact.
I don't -think- I pulled that hard, but?
.

[Toasty]

Re: OSCON go BOOM!

>>I'll just leave it out and use single CPU until I can get a matching cap.
[Gotta keep the VRM color coordinated. They run better like that.]<<

lol

I'll put 2 out in the mail tomorrow.

[PCBONEZ]

Re: OSCON go BOOM!

Thank you!

[TheLaw]

Re: OSCON go BOOM!

oscon? OSCON? OSCON?

Of any of the polymer manufacturers to experience a failure with, Sanyo Oscon is the one I'd be sure that wouldn't blow.

That's really quite strange. I regard them as some of the best. I guess nothing is perfect though.

[PCBONEZ]

Re: OSCON go BOOM!

If you build a few million of something a year there's bound to be a few that don't come out just right.
.

[kc8adu]

Re: OSCON go BOOM!

i once got a panasonic fm with the jacket on wrong.long lead was on the neg stripe side.
wonder if that is what happened here?mismark?

[TheLaw]

Re: OSCON go BOOM!

i once got a panasonic fm with the jacket on wrong.long lead was on the neg stripe side.
wonder if that is what happened here?mismark?

Don't think it would have been working for such a long time like that.

[TBoneit]

Re: OSCON go BOOM!

Showing his ignorance he asks, Polys have polarity?

They're new to me since I was in electronics school. Then we had Electrolytics, Mica, & paper. Not to mention the variable capacitor for tuning the radio.
Of course back then a common voltage rating might be 400volt for filtering the B+ supply.

I even remember seeing some really old radios where the electrolytics were the wet type
"Mershon's solution was to simply fit the capacitor into an oversize aluminum or copper can, half-filled with extra electrolyte. These units are referred to as "wet electrolytics,"

[Wizard]

Re: OSCON go BOOM!

That kind of surprises bound to happen even with best one.

Electrolytic is really a 'wet' not in sense if it even is water based, the electrolyte can be any kind of formulations not just mixed with water.

Cheers, Wizard

[PCBONEZ]

Re: OSCON go BOOM!

Showing his ignorance he asks, Polys have polarity?

Yes
The half moon spot of color on the top [or sometimes it's a rectangle/stripe in the same place] marks the negative lead.
.

[Toasty]

Re: OSCON go BOOM!

Get the caps yet PC ?

[PCBONEZ]

Re: OSCON go BOOM!

No Sir, not yet.
Not a big deal though.
It will be like next week before I have time to do anything with that board again anyway.
.

[Toasty]

Re: OSCON go BOOM!

Wow! These have been 'in the wind' for a week, and I stuck more than enough stamps on it too.

[TBoneit]

Re: OSCON go BOOM!

Yes
The half moon spot of color on the top [or sometimes it's a rectangle/stripe in the same place] marks the negative lead.
.

What happens if they are reversed? Do they catastrophically fail like a Electroylitic does?

Thanks
RT

[PCBONEZ]

Re: OSCON go BOOM!

I would imagine so.
They are still electrolytic and still work by [and are polarized because] they have an oxide layer on the anode foil.
The oxide layer is the actual dielectric.
Sending electrons through the wrong way erodes [thins] the oxide layer and will ultimately cause the cap to short internally.
.

[Scenic]

Re: OSCON go BOOM!

polys will explode violently when polarized wrong. been there done that
no vent on top for the gases to get out.. so it'll blow the whole can across the room like a projectile (and you're usually left with the rubber bung and a tower of rolled cap innards on the board.. nasty stuff)