Burnt MOFSET - A2726

[Bitter]

I bought some cheap used hardware, 8G 1333 RAM, i5 750, and a P55 motherboard, the board has a smoked A2726 MOFSET on the VRM, at least two pins are shorted together when compared to another A2726 next to it. I'm confident I can replace the component but I don't know what else I should check on the board as I don't see anything else burnt. I suspect this was an overclocking failure as the VRM have no heat sinks and this particular board (without BIOS update) was noted for this issue, so I don't think there's another culprit but I can't verify as when the board is powered the burned part emits smoke.

 

I can find single components for about $5-8 so the cost isn't a big deal, I just don't want to order one and turns out I also needed to order some other parts. Once I de-solder the component and re-wet the pads should I check anything else while the part is out or just swap it out and try to power it on. I guess if it blows the new one right away the board is just trash.

 

Also, this part is labelled A2726 then U930 then M* Do I need exactly that part or is the U930 and M just specific to say that run of part or a particular maker and any A2726 is the same pinout and function? I may see if I can take one off an old board I have laying around rather than ordering one just for time sake and because cheap free good.

[Hackentosher]

If the FET cooked itself, you could probably just replace it and be fine. But if something up or downstream of it failed and took the FET down with it, then that could be a problem. I would buy two replacements, swap it in, maybe throw on a heatsink, and then see what happens.

[Bitter]

2 hours ago, Hackentosher said:

If the FET cooked itself, you could probably just replace it and be fine. But if something up or downstream of it failed and took the FET down with it, then that could be a problem. I would buy two replacements, swap it in, maybe throw on a heatsink, and then see what happens.

Thanks, I can get 5 for cheap from AE if I don't mind waiting a while, so I'll go that route if I can't scavenge one off an old board I don't care about anymore. I just tossed a couple boards out but I think I have some P4 era stuff from Dell that I'll look at to see if there's one on it, if not then it's time to wait 40 days and 40 nights and hope I can remember what I ordered the parts for when they come in. Even if the board isn't fixable $45 shipped for 8G 1333 and an i5 750 isn't terrible.

[CodeNova]

@Bitter, Hi, so I was doing some digging, and this is what I have so far.

company website

supplier website

datasheet

ebay link?

I guess the part is EOL (End of life), from what I can tell (by Renesas, formally NEC). As far as replacing the part, look for a 'UPA2726UT1A / UPA2726', if you can find one. I would recommended using a donor motherboard (same model), double check the chip numbers "A2762".

 

Edit: The ebay link is for a QFN8, so... idk, the part you need is an 8pin HVSON, I'm not sure what the difference is.

 

Edit: If there is any difference.

[Unimportant]

8 hours ago, Bitter said:

I bought some cheap used hardware, 8G 1333 RAM, i5 750, and a P55 motherboard, the board has a smoked A2726 MOFSET on the VRM, at least two pins are shorted together when compared to another A2726 next to it. I'm confident I can replace the component but I don't know what else I should check on the board as I don't see anything else burnt. I suspect this was an overclocking failure as the VRM have no heat sinks and this particular board (without BIOS update) was noted for this issue, so I don't think there's another culprit but I can't verify as when the board is powered the burned part emits smoke.

 

I can find single components for about $5-8 so the cost isn't a big deal, I just don't want to order one and turns out I also needed to order some other parts. Once I de-solder the component and re-wet the pads should I check anything else while the part is out or just swap it out and try to power it on. I guess if it blows the new one right away the board is just trash.

 

Also, this part is labelled A2726 then U930 then M* Do I need exactly that part or is the U930 and M just specific to say that run of part or a particular maker and any A2726 is the same pinout and function? I may see if I can take one off an old board I have laying around rather than ordering one just for time sake and because cheap free good.

Is the gate shorted to drain or source ? If it is then the gate driver is probably dead as well, those tend to be integrated into the PWM controller IC these days.

[p360991]

If you remove the damaged FET and its complementary one on that phase then you should be able to power up the board.But you might want to try with another cpu because if the high side one failed it might have killed the cpu with it.Also @Unimportantmentioned that the PMIC migh have died you should check that too.Here is what i would do:

1)Remove the damaged mosfet

2)check for other damaged mosfets and IC's

3)check for short circuit to ground

4)If there is nothing else damaged remove the other mosfet from the phase and test

[Bitter]

5 hours ago, CodeNova said:

@Bitter, Hi, so I was doing some digging, and this is what I have so far.

company website

supplier website

datasheet

ebay link?

I guess the part is EOL (End of life), from what I can tell (by Renesas, formally NEC). As far as replacing the part, look for a 'UPA2726UT1A / UPA2726', if you can find one. I would recommended using a donor motherboard (same model), double check the chip numbers "A2762".

 

Edit: The ebay link is for a QFN8, so... idk, the part you need is an 8pin HVSON, I'm not sure what the difference is.

 

Edit: If there is any difference.

So I do or don't need the exact model of A2726? I'll get a nice macro of it later today once I'm back home again so we can all see what I'm seeing, both damaged and the adjacent good component.

4 hours ago, James Evens said:

Ebay is a mixed bag. If you just want a mosfet buy on ebay. If you want this model don't buy on ebay or ali .

That's what I figured, but I wasn't sure I needed the exact part number or if one of the same function would work.

1 hour ago, Unimportant said:

Is the gate shorted to drain or source ? If it is then the gate driver is probably dead as well, those tend to be integrated into the PWM controller IC these days.

I honestly do not know at this point, I need to learn a little more and check this out more thoroughly. This is 'above my pay grade' so I'm using it as a learning experience. I have limited knowledge beyond component replacement and limited experience with SMT soldering as well, so I figured a dead board is a good place to start learning as it's lower risk.

2 minutes ago, p360991 said:

If you remove the damaged FET and its complementary one on that phase then you should be able to power up the board.But you might want to try with another cpu because if the high side one failed it might have killed the cpu with it.Also @Unimportantmentioned that the PMIC migh have died you should check that too.Here is what i would do:

1)Remove the damaged mosfet

2)check for other damaged mosfets and IC's

3)check for short circuit to ground

4)If there is nothing else damaged remove the other mosfet from the phase and test

I'm going to check the CPU in a known working board, is there a possibility that the CPU being dead would damage the working board as well? How would I tell the complementary component? There's the A2726 which is above two other components and near a choke, making a 'triangle' of SMT components, there's several 'triangle' clusters, would I be able to tell by checking continuity between some pins on the burned one and the next nearest one? Like I said, this is beyond my normal skill set but I want to learn and grow even if I don't succeed with a working board in the end I'll have still succeeded with learning something.

 

 

Also I want to say a huge thanks to everyone who replied here, honestly I didn't expect so much support and help. This is one of the most helpful forums I participate in and save for a rare bad interaction this place is great. Thank you so much, I truly do appreciate the help.

[p360991]

If the cpu has a short circuit then it would damage a working board.Posting the full motherboard name and a picture of yours would be very helpful.CPU vrm phases are usually indicated by the number of inductors.There should be 2 or 3 mosfets close to the inductor.Those mosfets and inductor form a "phase".Those phases use a Synchronous Buck converter topology.They use many of those phases to help spread the load and reduce ripple in high current applications.To check from Vcore to ground just probe one end of the idnuctor to ground(Use diode mode ie buzzer mode).If there is no short the value should start increasing.As for which components to test and where to probe I cant tell you because  no picture.Hope this helps

[Bitter]

2 hours ago, p360991 said:

If the cpu has a short circuit then it would damage a working board.Posting the full motherboard name and a picture of yours would be very helpful.CPU vrm phases are usually indicated by the number of inductors.There should be 2 or 3 mosfets close to the inductor.Those mosfets and inductor form a "phase".Those phases use a Synchronous Buck converter topology.They use many of those phases to help spread the load and reduce ripple in high current applications.To check from Vcore to ground just probe one end of the idnuctor to ground(Use diode mode ie buzzer mode).If there is no short the value should start increasing.As for which components to test and where to probe I cant tell you because  no picture.Hope this helps

That does help, I'll get some more info posted when I get home. Thank you.

 

Waiting somewhere,got the board model off the eBay listing.

Gigabyte GA-P55M-UD2 motherboard, Google says burnt VRM is common on these. I'll hold off on testing the CPU till I have a better idea of what happened to the board to cause the failure. Could the board damage a PSU powering it?

 

And if it's fixed someday I'll get some VRM cooling sorted out for it to try to prevent this in the future!

[p360991]

I've circled all the vrm mosfets.They are grouped by phase.Remove all of the mosfets that are in the circle that has the blown FET.I should warn you that this is very hard to do.I don't know what tools you have but this is hot air station territory.If you dont have that then I suggest don't bother.QFN-8 mosfets have a metal bottom which is soldered to the motherboard for better thermal dissipation.If this this is purely for learning purpose and have no expectation to work then you can try to preheat the board to 50 or60C in a oven and try to desolder with a soldering iron but that risks damaged traces and components.

[Bitter]

I do not have a hot air reflow station, but I've been wanting to buy one.

The burned MOFSET is the one nearest the CPU fan header below the cluster of 6 small SMT components next to the large metal body capacitor.

However on closer examination under a more color accurate light I can see all the VRM down the side of the board near the IO are discolored somewhat from heat and the back of the PCB is browned under them. I think this board is a lost cause, it's been 'ridden hard and put away wet' as the old saying goes. With the amount of heat it appears to have been subjected to I don't think I'll be confident using it.

 

How would I check the CPU for damage which could ruin another board that I put it into?

[Bitter]

Backside of the board showing the browning of the PCB

The front side of the browned area of the PCB from above

The top of the board, the burned MOFSET is to the left top next to the capacitor

Sorry the photos aren't better, the flash is pretty harsh so I had to try to soften it which pushed the ISO higher and made the outcome a little grainy/poorly focused.

 

The black spots look like splattered cola dried on the board and scrape off easily, don't smell burned, but are only seen around where the VRM's look like they were hot. I think this board is toasted.