Motherboard died

[Tomvv]

Hi,

two days ago I smelled "the magic smoke" in my living room while I was working on another computer.

The next day my main desktop would not boot up.

I initially thought my power supply was to blame but after some thorough investigation I found out some component near the CPU died.

The questions I have now are:

• How could this have happened?
• Can it be fixed, and if yes how?
• Can I trust my power supply to safely work on another mobo?

Thanks in advance and kind regards,

Tom

P.s. The component in the picture looks like a voltage regulator to me.

[dr.diesel]

What brand/model power supply is it?  There is a huge plethora of ultra mega cheap PSs out there should be avoided.

[Tomvv]

XFXforce 550W ( pretty reputable I thought ).

I've tested the PSU using the paperclip method and no problem what so ever.

[dr.diesel]

My 10 sec Google search shows that XFX PSs are made by Seasonic, which is arguably the best in the industry, so you probably good there! 

[gaijin]

Still wouldn't hurt to test the voltages on the psu.

btw
http://www.irf.com/product-info/datasheets/data/irlr8726pbf.pdf

[wraper]

• Can it be fixed, and if yes how?

Change the faulty mosfet(s). If you are lucky, nothing else was fried.

[Tomvv]

Thanks for the replies.

I'm not very eager to replace the MOSFETS on this board any time soon as there is very little room available for my soldering iron.

While looking for some more information I stumbled on an interesting post.

http://www.hardwaresecrets.com/everything-you-need-to-know-about-the-motherboard-voltage-regulator-circuit/

[dr.diesel]

It's just a good excuse to upgrade to Skylake!

[wraper]

Thanks for the replies.

I'm not very eager to replace the MOSFETS on this board any time soon as there is very little room available for my soldering iron.

Desolder an inductor and there will be plenty of space.

[amyk]

Were you running the CPU at near 100% load? It could just be bad thermal connection to the heatsink causing the MOSFET to overheat.

[Tomvv]

Were you running the CPU at near 100% load? It could just be bad thermal connection to the heatsink causing the MOSFET to overheat.

I never ran the CPU at 100% or near to. I did however replaced my E5700 processor with a Q9550 which get warmer/hotter.

[Nermash]

Some time ago I had a Asus P4P800 MB that kept blowing one mosfet in the CPU supply. After two blown transistors, I pulled the DSO out to check the gate driver IC, and sure enough, it was stuck in ON mode. Replacing the gate driver and mosfet solved the issue. Hot air really helps when removing components in tight surroundings.

[Tomvv]

I've found some suppliers of said MOSFETS on AliExpress.

Would it be wise to buy ( quality wise etc ) to buy from them?

[rob77]

did you check the CPU ? it might have not survived if one of the the top mosfets of that multi-phase  synchronous buck converter blew up and got shorted

regarding your questions:

1. how and why ? because things get worn out (even ICs and mosfets do - mostly due to excessive heat)
2. yes it can be fixed if the PCB is not damaged - replace both mosfets in the particular blown phase and replace the control/driver chip if needed (check outputs with scope)
3. do a load test on your PSU and check the outputs with scope.

PSU is most likely not the root cause - the supply for the CPU is a multiphase synchro buck converter supplied by the 12V rail from the PSU - i can't believe your 12V rail went so high it blew a mosfet.. (but that's my opinion only - check the PSU under load to be sure)

p.s. actually there is one more dirty hack - disable that blown phase - de-solder both mosfets of that phase. sometimes the remaining phases are able to supply enough power with acceptable ripple.

[Augustus]

The Q9550 is a 95W TDP part which can pull a serious amount of power, especially if overclocked and driven with raised vcore... It's well known to be able to wreak havoc to budget boards in such situations 

[Shock]

Hot air and possibly preheating to get that off (and on).

[Tomvv]

It seems the processor survived.

I am running some CPU tests in my ASUS P5B at the moment.

[jitter]

The questions I have now are:

• How could this have happened?

Several possibilities. Overload or age seem most likely.

Overload: not likely if you've been using the Q9550 for a long time.

Age: given that you replaced an E5700, I would put this motherboard somewhere around 2007/2008? Right?
Especialliy if you're using it long and hard, the electrolytic capacitors may have aged by now. One of these going bad might have caused the mosfet to fail.

The photo shows signs of the right mosfet getting very hot, to the point it melted the solder and discoloured the metal. This was not a short sharp overload but a lingering one.
Maybe it's even a combo of the power used by the Q9550 and ageing components on the motherboard.

• Can it be fixed, and if yes how?

Yes. If the pcb is in good shape. Heat exposure may have damaged the adhesion of the traces to the base material. Lifting traces by accident then becomes all too easy.
Make sure all solder has melted before you try to lift the component off. If you have to apply force, then not all solder has melted. IMHO the only method of removal with a chance of succes is hot air. You may want to desolder some of the adjoining caps and coils first to prevent heat damage to them and to give you room to work. It looks like they're through hole devices.

• Can I trust my power supply to safely work on another mobo?

If all voltages measure OK, then yes.

P.S.  I just saw a Teapo cap bottom right in your photo, definitely not regarded as quality caps, esp. from that time period. Wonder what the brand of those next to the mosfets is...

[Tomvv]

Age: given that you replaced an E5700, I would put this motherboard somewhere around 2007/2008? Right?
Especialliy if you're using it long and hard, the electrolytic capacitors may have aged by now. One of these going bad might have caused the mosfet to fail.

This motherboard is from my Dell Inspiron 560, bought in 2010 ( but probably manufactured some years before ).

It was a low budget machine ( hence the crappy components ) in which I replaced the PSU, Video card and processor later on.

Five years is my machine update interval so I will probably opt for a new machine.

However I will try and see if I can resurrect this mobo by replacing the MOSFETS.

I just ordered a hot air rework station and some MOSFETS so all I need now is time. 

[jitter]

Age: given that you replaced an E5700, I would put this motherboard somewhere around 2007/2008? Right?
Especialliy if you're using it long and hard, the electrolytic capacitors may have aged by now. One of these going bad might have caused the mosfet to fail.

This motherboard is from my Dell Inspiron 560, bought in 2010 ( but probably manufactured some years before ).

It was a low budget machine ( hence the crappy components ) in which I replaced the PSU, Video card and processor later on.

Five years is my machine update interval so I will probably opt for a new machine.

However I will try and see if I can resurrect this mobo by replacing the MOSFETS.

I just ordered a hot air rework station and some MOSFETS so all I need now is time. 

Yeah that was a wild guess based on the fact that I bought an E8400 in 2008 for a system I assembled myself. 2010 was my second guess, I noticed that the purple caps have "951" and "952" printed on them which I presume to be datecodes for weeks 51 and 52 of 2009. That would put the manufacturing date of the motherboard in 2010.

Should your repair fail, the motherboard can always serve as a learning tool in developing your SMD (de)soldering skills.

My job is in industrial electronics, SMDs are no stranger to me and I don't understand why there are still people who prefer through hole. Provided you have the right tools (we use JBC) desoldering SMD is actually in almost all cases easier than desoldering through hole stuff, especially with leadfree solder.

Good luck with the repair (and don't forget to check the state of those electrolytic capacitors before you put in your new mosfets).

[amyk]

You can use one of these dummy CPUs to make sure the voltages are OK before putting a real CPU into the board after the repair:

http://www.ebay.com/itm/771-775-CPU-Dummy-Load-Motherboard-Repair-Tools-Tester/190512961764

http://www.ebay.com/itm/775-Intel-CPU-Dummy-Load-Motherboard-Repair-Tool-Tester-/200587406191

[fubar.gr]

The questions I have now are:

• How could this have happened?
• Can it be fixed, and if yes how?
• Can I trust my power supply to safely work on another mobo?

1) This happened because it is a badly designed voltage regulator. Some motherboards are notorious for blowing the low voltage side VRM mosfets.

2) Yes, you can repair it, but stick some heatsinks on the mosfets, including those in the other phases too. Don't bother heatsinking the high voltage mosfets, these handle smaller currents and rarely go bad.

Here's a repair I did on a motherboard with the same problem: http://fubar.gr/successful-gigabyte-ga-h55m-s2-motherboard-repair/

I only heatsinked the replaced mosfets and soon after the mosfets from another phase blew up. So make sure to heatsink them all.

3) Your power supply has nothing to do with the blown mosfets. It's all due to improper sizing/heatsinking of the mosfets.

[rob77]

Here's a repair I did on a motherboard with the same problem: http://fubar.gr/successful-gigabyte-ga-h55m-s2-motherboard-repair/

I only heatsinked the replaced mosfets and soon after the mosfets from another phase blew up. So make sure to heatsink them all.

gluing a heatsink to the plastic body of the mosfet is useless - you can achieve much better results via proper airflow around the board - the PCB is the mosfet's only heatsink because the heat is dissipated through the metal back of the mosfet (not the plastic part).

[Tomvv]

I believe the Q9550 ( and lack of cooling ) killed my mobo.

Here is a picture of the mosfets during the CPU upgrade.

[fubar.gr]

Here's a repair I did on a motherboard with the same problem: http://fubar.gr/successful-gigabyte-ga-h55m-s2-motherboard-repair/

I only heatsinked the replaced mosfets and soon after the mosfets from another phase blew up. So make sure to heatsink them all.

gluing a heatsink to the plastic body of the mosfet is useless - you can achieve much better results via proper airflow around the board - the PCB is the mosfet's only heatsink because the heat is dissipated through the metal back of the mosfet (not the plastic part).

Gaming mobos are factory equipped with heatsinks like this.

Heatsinks on plastic are not very efficient, but it is a stretch to call them useless.