Just blew up a motherboard...

[TheThirdMan]

Hey guys,

So I just blew up my first motherboard! Hurrah! But seriously, can anyone help identify the reason or what this chip is:

Backstory:

Purchase the motherboard (Supermicro X8DAi) on ebay after receiving a DOA Tyan board previously, arrives with an old bios that doesn't support the 56xx Xeon chips, only the 55xx chips.

I ran it with one E5504 whilst the new bios chip was ordered (the board is OEM with a locked bios so I can't upgrade it myself) and then it arrives, I install the new bios, boots up fine with the E5504, no issues. I swap it out for the two L5639s I have. Press the power button- the fans spin up for 1 second and then stop. Press power again- nothing. Press power a third time- boom (well not boom, put a sudden pop)! Small explosion of sparks and/or flame (surprised me too much for me to properly see what happened).

I'm not sure what chip that is in the picture, but a very strong burned/metallic smell came from it after it blew. I'm pretty sure I can't do anything about it, however I was hoping you guys might be able to help me figure out why it happened! I have a corsair HX1050 PSU so I can't imagine it's that.

The only thing that changed between the previously working setup with the E5504 and the new exploding setup was :

• I swapped out the CPUs
• moved half the ram into the ram banks for the second CPU as it was all in the first bank with one CPU running
• Attached two more fans- one for the back of the case and one for the second CPU cooler
• attached the HDD and Power LED connectors

As far as I can tell, nothing else changed between the two setups. I am using power cable extenders for both CPU power connectors and the main 24-pin power connector but I can't imagine this is the issue as I've ran it fine with the single E5504 with those cable extensions.


Any ideas what caused it?

 

[SOFTengCOMPelec]

Did you remove/disconnect/switch-off the mains wall supply to the computer, before making the hardware changes ?

Sometimes the problem can be if something shorts out, e.g. Something only partially plugged in, or the motherboard shorting against the metal of the case, etc.

My gut feeling is you don't stand a good chance of fixing it (my opinion/guess, maybe you can fix it) because it either has other faults now, due to the extremely high currents which blew the chip/transistor, and it can be very difficult to change surface mount components like that, if you are not into electronics and/or have the right equipment. (I.e. the delicate PCB tracks/pads can easily get broken, if you try to change it yourself).

Also the fault which blew the chip/transistor up, may still be present, and/or the chip(s) which connect to the blown part, may have also been damaged at the same time, due to the extreme overload (that component looks very badly damaged).
There may also be other components which have now been damaged, but are not showing visible signs of destruction.

 

[C101]

While I can't make any presumptions other than guesses as to why the fet blew up, I can point you to the part/replacement location.

Edit: Based on the pcb layout, it looks like it is the high-side fet for a buck regulator.

Part: http://ec.irf.com/v6/en/US/adirect/ir?cmd=catProductDetailFrame&productID=IRLR7821CPBF
Where to buy: http://www.digikey.com/product-detail/en/IRLR7821TRPBF/IRLR7821PBFCT-ND/812525

However, given the location and the type of part, I think that it will be a difficult part to replace. These parts have a large solder pad on the backside of the chip, which is difficult to desolder without a good hot air gun, as the motherboard acts as a heatsink. In addition, there is the very close proximity the plastic connector, which will melt if you use a hot air gun near it.

 

[TheThirdMan]

While I can't make any presumptions other than guesses as to why the fet blew up, I can point you to the part/replacement location.

Edit: Based on the pcb layout, it looks like it is the high-side fet for a buck regulator.

Part: http://ec.irf.com/v6/en/US/adirect/ir?cmd=catProductDetailFrame&productID=IRLR7821CPBF
Where to buy: http://www.digikey.com/product-detail/en/IRLR7821TRPBF/IRLR7821PBFCT-ND/812525

However, given the location and the type of part, I think that it will be a difficult part to replace. These parts have a large solder pad on the backside of the chip, which is difficult to desolder without a good hot air gun, as the motherboard acts as a heatsink. In addition, there is the very close proximity the plastic connector, which will melt if you use a hot air gun near it.

Thanks for that! I don't think there's much point in me trying to be honest. This was supposed to be a cheap build to use for rendering and risking the two CPUs (if they haven't been broken already) isn't really worth the risk if i can somehow find another cheap motherboard online.

Did you remove/disconnect/switch-off the mains wall supply to the computer, before making the hardware changes ?

Sometimes the problem can be if something shorts out, e.g. Something only partially plugged in, or the motherboard shorting against the metal of the case, etc.

My gut feeling is you don't stand a good chance of fixing it (my opinion/guess, maybe you can fix it) because it either has other faults now, due to the extremely high currents which blew the chip/transistor, and it can be very difficult to change surface mount components like that, if you are not into electronics and/or have the right equipment. (I.e. the delicate PCB tracks/pads can easily get broken, if you try to change it yourself).

Also the fault which blew the chip/transistor up, may still be present, and/or the chip(s) which connect to the blown part, may have also been damaged at the same time, due to the extreme overload (that component looks very badly damaged).
There may also be other components which have now been damaged, but are not showing visible signs of destruction.

Yeah i did remove the cable completely. I thought it might be shorting against the case but in my previous experience of this it won't blow stuff up! Though i suppose the two power ons before the explosion are similar to what would happen if it was shorting on the case.

I think i agree with you though. Not much point in trying a repair. As i said above it was supposed to be a cheap build to use as a headless rendering node and there's no point in risking the other components if they're still working.

Thanks for the tips guys.

 

[SOFTengCOMPelec]

Another thought comes to mind as well. You said the earlier board (1st one, Tyan) was not working. It could have blown your cpus and made one or both go short circuit (in theory).

I agree, if it (blown motherboard) was NOT too expensive, saving time/cost and hassle of a repair, which may not work, by not trying, is probably a good idea (in my opinion, others may disagree).

 

[SOFTengCOMPelec]

I thought it might be shorting against the case but in my previous experience of this it won't blow stuff up!

If you are saying you tried to turn it on, while it was NOT insulated from the metal base of the case, then that would be very likely to blow the motherboard up.
The insulating separators/stand-offs are ESSENTIAL really, even for quick tests.
Metal is way too good a conductor, to rest switched on stuff, against.

I.e. the gap between the bottom of the motherboard, and the bottom of the case, where the motherboard rests. E.g. Screws+washers + metal or plastic stand-off, etc. So that there is a gap between the motherboard and the metal case.

 

[TheThirdMan]

If you are saying you tried to turn it on, while it was NOT insulated from the metal base of the case, then that would be very likely to blow the motherboard up.
The insulating separators/stand-offs are ESSENTIAL really, even for quick tests.
Metal is way too good a conductor, to rest switched on stuff, against.

I.e. the gap between the bottom of the motherboard, and the bottom of the case, where the motherboard rests. E.g. Screws+washers + metal or plastic stand-off, etc. So that there is a gap between the motherboard and the metal case.

Well the case I use (Corsair 750D) has standoffs stamped into the motherboard tray so there's none of those brass standoffs you'd normally use for a mobo.

Another thought comes to mind as well. You said the earlier board (1st one, Tyan) was not working. It could have blown your cpus and made one or both go short circuit (in theory).

It could have, but the E5504 i have was also tried in that motherboard and then subsequently worked in the X8DAi without issue.

 

[C101]

If you are saying you tried to turn it on, while it was NOT insulated from the metal base of the case, then that would be very likely to blow the motherboard up.
The insulating separators/stand-offs are ESSENTIAL really, even for quick tests.
Metal is way too good a conductor, to rest switched on stuff, against.

I.e. the gap between the bottom of the motherboard, and the bottom of the case, where the motherboard rests. E.g. Screws+washers + metal or plastic stand-off, etc. So that there is a gap between the motherboard and the metal case.

Likely to be the case. I don't see any current sensing resistors on the board, which probably means there is no overcurrent protection. If the supply for the voltage regulator was even +5V, the maximum current that could flow through that fet would be ~400A, given the worst case Rds(on). At 12V this is much worse, at ~960A. These currents would very quickly create an over temperature condition in the part, which would then result in the damage you can see on your board.

 

[SOFTengCOMPelec]

Well the case I use (Corsair 750D) has standoffs stamped into the motherboard tray so there's none of those brass standoffs you'd normally use for a mobo.

Not according to this link, which says they MUST be used with it, despite the stamping. (The link seems to be for EXACTLY your case).

Standoffs are always required for installing a motherboard, if not used you have a big chance of causing shortcuts.

It might be, you only need 1 or 2 stand-offs, I'm not sure, but none might be a problem.

http://linustechtips.com/main/topic/62739-750d-no-standoffs-for-mobo-included/

If I'm wrong (and the link is wrong), then sorry.

But something has caused your motherboard to blow up, and a metal short circuit is a likely explanation.

EDIT: A very big motherboard like that, and with 2 processors, somewhat heavy heatsinks, flexes about a fair bit, so it needs a decent clearance, else it can short out. So maybe it was too close and/or touching the metal ?

Likely to be the case. I don't see any current sensing resistors on the board, which probably means there is no overcurrent protection. If the supply for the voltage regulator was even +5V, the maximum current that could flow through that fet would be ~400A, given the worst case Rds(on). At 12V this is much worse, at ~960A. These currents would very quickly create an over temperature condition in the part, which would then result in the damage you can see on your board.

Yes, it needs a HUGE overload current to blow up a power fet like that (low rds-on makes it harder to generate enough power, to overheat it).
It is not heatsinked enough to take much wattage, or even the right case style to be strongly heatsinked, so a high current overload, could easily pop it, as you have just said.

 

[TheThirdMan]

Likely to be the case. I don't see any current sensing resistors on the board, which probably means there is no overcurrent protection. If the supply for the voltage regulator was even +5V, the maximum current that could flow through that fet would be ~400A, given the worst case Rds(on). At 12V this is much worse, at ~960A. These currents would very quickly create an over temperature condition in the part, which would then result in the damage you can see on your board.

Thanks for the info! I've read around a bit about E-ATX and it seems although it's a standard used by multiple companies, some just add or remove a couple of mount holes for some reason, so even if they say "EATX" in their spec for the mobo, sometimes it won't line up with some holes in an EATX case.

Not according to this link, which says they MUST be used with it, despite the stamping. (The link seems to be for EXACTLY your case).

I think this is a little confusing. As far as I know, and from what I've read on the corsair forums from their own employees, the built-in standoffs should be fine with nothing else and they say they've "never had an issue" with this system. You can't add anything else on top of these as the back panel won't line up anymore.

It might be, you only need 1 or 2 stand-offs, I'm not sure, but none might be a problem.

http://linustechtips.com/main/topic/...mobo-included/

If I'm wrong (and the link is wrong), then sorry.

But something has caused your motherboard to blow up, and a metal short circuit is a likely explanation.

EDIT: A very big motherboard like that, and with 2 processors, somewhat heavy heatsinks, flexes about a fair bit, so it needs a decent clearance, else it can short out. So maybe it was too close and/or touching the metal ?

Thanks for the info mate. I suppose it is most likely to be a short. It's frustrating because the motherboard and case are both EATX.

I don't suppose that perhaps the reason it wasn't shorting before was that i had one CPU and was using one bank for ram? So that part of the mobo that was shorting on something wasn't affecting it because that part of the mobo wasn't "active"? Does it work like that or are all parts of the motherboard in use even with one CPU?

 

[SOFTengCOMPelec]

I am using power cable extenders for both CPU power connectors and the main 24-pin power connector but I can't imagine this is the issue as I've ran it fine with the single E5504 with those cable extensions.


Any ideas what caused it?

Just another thought. It is only a possibility, but may be worth checking out, with a multimeter (voltage meter), if you are confident enough to do it. But I'm not sure offhand what the pins and voltages are, but there should be guides on the internet.

If your 24 pin cpu power cable extender(s) were faulty, then (in theory) the wrong voltage and/or no voltage could be going to some pin(s).

So if previously you only used 1 cpu, but now tried to use 2 cpus, maybe the fault has shown itself.

Explanation (N.B. this is just a POSSIBILITY, many different faults are possible)

Wrong voltage (e.g. 5V and 12V swapped), could put far too much voltage on it, so it overheats and blows up, like in the picture.

No voltage, on a pin(s), might mean (if they are NOT paralleled up on the circuit board, I am not sure if they are, or they are NOT) that some of the other transistors (fets) had no voltage supply (if this is even possible, I don't know how they are wired, i.e. Parallel or separate circuits (series).

Therefore no voltage to pin(s) (if partly in series wired) means only some of the transistors (fets) would be used to supply the particular cpu, in which case they would probably get far too hot, and could blow up, like shown in your picture (in theory).

Some people recommend AGAINST using cheap e.g. ebay power extender/splitter cable things, etc etc. (Opinions vary on this, I'm not 100% sure myself, either way, but ideally it is best to NOT use them, as the currents can be VERY high with modern PC power supplies, so if the wires are too thin and can't take the current, then it can cause problems).

 

[TheThirdMan]

Just another thought. It is only a possibility, but may be worth checking out, with a multimeter (voltage meter), if you are confident enough to do it. But I'm not sure offhand what the pins and voltages are, but there should be guides on the internet.

If your 24 pin cpu power cable extender(s) were faulty, then (in theory) the wrong voltage and/or no voltage could be going to some pin(s).

So if previously you only used 1 cpu, but now tried to use 2 cpus, maybe the fault has shown itself.

Explanation (N.B. this is just a POSSIBILITY, many different faults are possible)

Wrong voltage (e.g. 5V and 12V swapped), could put far too much voltage on it, so it overheats and blows up, like in the picture.

No voltage, on a pin(s), might mean (if they are NOT paralleled up on the circuit board, I am not sure if they are, or they are NOT) that some of the other transistors (fets) had no voltage supply (if this is even possible, I don't know how they are wired, i.e. Parallel or separate circuits (series).

Therefore no voltage to pin(s) (if partly in series wired) means only some of the transistors (fets) would be used to supply the particular cpu, in which case they would probably get far too hot, and could blow up, like shown in your picture (in theory).

Some people recommend AGAINST using cheap e.g. ebay power extender/splitter cable things, etc etc. (Opinions vary on this, I'm not 100% sure myself, either way, but ideally it is best to NOT use them, as the currents can be VERY high with modern PC power supplies, so if the wires are too thin and can't take the current, then it can cause problems).

Thanks for this. The extension cables were off ebay but weren't cheap! I specifically went for the more expensive ones (£9 for the 24 pin and £6 for the CPU ones). It could still be this I suppose. I'm getting a replacement motherboard soon so I'm not sure how I can really test any of this without risking blowing the motherboard again!

I don't have a multimeter to hand so it's not something I can test myself. I suppose I could go without the extension cables but the PSU has annoyingly short power cables itself.

 

[PhIlLy ChEeSe]

Flee bay that board, let someone else risk the loss of hardware on it. I do notice the motherboard screw missing in the photo, we could be here for days speculating on what caused it. Or as you said, it simply did not show the short as you only had one CPU on the other side of it.
I bought a non working board of Flee bay, a socket 775 cheap. Same thing happened it blew a FET as soon as it was powered on. When I have some time I will try to replace the FET, but I will risk a CPU, PSU and memory to even try to power it on.