Intel DX58SO high side VRM MOSFET keeps failing

[dougg3]

Hi there,

I have an Intel DX58SO motherboard that failed last year. One of the high side VRM MOSFETs failed shorted, and it killed my i7 980 CPU. This motherboard has a 6-phase VRM. The parts used on this motherboard's VRM are:

• 6 x High side MOSFET: NTMFS4841NH
• 6 x Low side MOSFET: NTMFS4834N
• 6 x Driver: ADP3121JCPZ

It was easy to figure out where the short was, because the bad MOSFET's gate was shorted to the drain/source too. Once I removed it, all the shorts were gone on the motherboard, and I confirmed on the removed MOSFET itself that the gate/drain/source were all shorted together. After figuring it out, I bought a new one from Digi-Key and soldered it on with my hot air station. I also got new thermal pads to apply to the VRM heatsinks and put everything back together with a Xeon W3680 (the less-expensive Xeon equivalent of the 980).

It worked fine for a while with very very light use, but earlier this year it failed again. I finally got around to looking deeper into it, and what I discovered is that the exact same MOSFET (the replacement I soldered on) failed in the exact same way again. I haven't tested yet if the Xeon was killed, but since it's a high side MOSFET, there's a high likelihood that it got fed 12V directly.

Anybody have any advice on what's going on here? Maybe I need better cooling on it? What would cause the replacement to fail in the exact same way?

When I read online about repairing a motherboard's VRM, I saw advice that you should typically replace the driver IC if a MOSFET shorts out, which I didn't do the first time because it was discontinued. I bought some from China this time and am waiting for them to arrive, so I can at least replace that chip this time as well.

What do you all think I should do this time around? My thinking is if I only replace the MOSFET again, it'll obviously fail again. It makes sense that the driver IC could be damaged since the gate was also shorted to 12V, so I'm guessing that's a good idea to replace this time too. I also bought a new diode, just in case. Are there any other best practices I should be following this time around? Should I also replace the low side MOSFET on the bad phase too? What about the MOSFETS on the other phases? I've seen some advice in the past in other threads about replacing other MOSFETs in the VRM at the same time. How far should I go? Any additional diagnostics I can perform? I've attached a picture of what the VRM circuit looks like, from the first time I repaired it.

One theory I have been floating around is that I have an aftermarket cooler (Cooler Master Hyper 212 EVO), and the fan on it doesn't blow straight down, so maybe the VRM isn't getting as cooled as it otherwise would with a stock Intel cooler.

This motherboard is old and not that great, but I think it is a fun project trying to repair it. It's definitely not worth the time it takes to fix, but I still want to give it a shot. Thank you for any advice!

[dicky96]

Re: Intel DX58SO high side VRM MOSFET keeps failing

When a high side mosfet goes short on a multiphase VRM it puts stress on the load (CPU in this case) the low side mosfets on all the phases and the mosfet driver on the faulty phase.

So you could try replacing all of them - if the parts are cheap

Possibly there is some cooling issue

Possibly the VRM controller has some issue and turns this phase on for too long

[dougg3]

Re: Intel DX58SO high side VRM MOSFET keeps failing

Thank you! That makes sense, I didn't think about the fact that it would also stress all of the other low side MOSFETs. Only the high-side MOSFETs were available on Digi-Key, so I did have to resort to utsource for the low-side MOSFETs and drivers. We'll see how it goes...

[dougg3]

Re: Intel DX58SO high side VRM MOSFET keeps failing

I sat down to look at the motherboard and realized how intimidating it would be to replace all of the MOSFETs. I really struggle to heat the board up enough with my cheapo hot air station, especially with the lead-free solder. I don't trust myself enough to replace them all without ruining something.

So instead, I tried replacing the blown high-side MOSFET again, but this time I also replaced the driver chip (ADP3121) with a brand new one too. It worked at first, and my Xeon seemed to be okay. But sure enough, the same high-side MOSFET blew *again*, in the exact same way. It actually blew very quickly this time, while I was still messing around with the motherboard before I even got a chance to try actually booting it into an OS.

This started making me wonder: why is it always this same phase that dies every time, even after replacing the driver? I guess like you said, it could be the VRM controller that is the problem. I tried to look for other reasons first. I grabbed my multimeter and compared resistances and diode drops on the different phases. They were all the same and seemed normal, except for one difference:

There's a 2.2 ohm resistor on each phase between the high side MOSFET's gate and the driver IC. On all of the other phases, the resistance (in circuit anyway) reads as about 2.5 ohms or so. That seems reasonable for a 2.2 ohm resistor. On the bad phase, it reads as about 4 ohms. That seems much too high for a 2.2 ohm resistor. Maybe the gate on the MOSFET doesn't like that. I took it out, and it indeed had a much-too-high resistance out of circuit as well. I just so happen to have 2.2 ohm resistors in the correct size, so I replaced it (and the high-side MOSFET again, for the third time). The motherboard seems happy so far. The MOSFET that keeps failing doesn't seem to be getting crazy hot or anything, at least not during my limited testing.

The bad MOSFETs did kill the Xeon though. The Xeon sort of works, but it hangs randomly in the boot process and sometimes fails to boot into the BIOS at all. This doesn't surprise me because it was exposed to 12V twice. My old i7 920 works perfectly though. I went back to the stock Intel heatsink that blows straight down on the board, just in case there is a VRM cooling issue.

Maybe the different resistance on the gate caused it to be out of sync with the rest of the VRM. I don't feel confident enough yet to say the problem is "fixed". I'm going to let it run for a while and see how long it lasts. I'm crossing my fingers in the hopes that the resistor is what's been causing this problem all along. On the other hand, maybe the resistor got screwed up *because* of the underlying problem that blows the MOSFET. Time will tell, I suppose!

[davidebaldini]

Re: Intel DX58SO high side VRM MOSFET keeps failing

Could anyone speculate whether this might be due to an open capacitor in the voltage regulator circuit that causes stress on its mosfet, as it happened on my power supply's APFC?

[PeppercornFox]

Re: Intel DX58SO high side VRM MOSFET keeps failing

Hi Dougg, I was wondering if you had an update on your dx58so board? Is it still kicking?

A high-side mosfet went short on my board a few months ago, I think from lack of cooling. Most of the resistors still read in spec (+.1) so if it only takes removing a few components I might give it a go. None of the gates are shorted though so

[dougg3]

Re: Intel DX58SO high side VRM MOSFET keeps failing

Hi PeppercornFox,

The DX58SO is still doing fine for over a year now, although I haven't been stressing it even close to any limits, and I don't really power it up super often either. I still have my i7 920 in it with the stock cooler. I really do think that this motherboard likes the stock cooler better because it does a better job of cooling off the VRM than the Hyper 212 EVO I was using before.