Motherboard can't control fan - which component is bad?

[gintama]

My current motherboard, a Gigabyte GA-EP45-DS4, is failing. I have a spare, but it also has a problem: The Sys_Fan2 fan output is stuck at 12V. It is a 4-pin connector which could previously be controlled in software - until someone shorted two of the pins out while the system was on. I am hoping to restore the ability to regulate the voltage to a standard non-PCM fan.

Many of you know more a lot more about electronics than I do, and I was hoping someone could tell me which component would be damaged, or how I can find it beyond just tracing the circuit with a multimetre. If it's a fairly simple IC (i.e. not one with dozens of connections) then I'd like to try replacing it with the part from the other board.



As you can see from the image above, the user guide lists the pinout as being different than the PCM-capable CPU_Fan connector, but I don't know whether that's accurate. The board continues to reads the "Sense" line, the "Speed Control" line works (except for being locked at 100%/12V), and the CPU_Fan connector is still controllable, so my guess is that either the damage is localized or the voltage is controlled by a discrete component.


.

[Mad_Professor]

Re: Motherboard can't control fan - which component is bad?

do you know which of the pins were shorted out?

[cheapie]

Re: Motherboard can't control fan - which component is bad?

Does the BIOS have a "3-pin fan" setting? Alternatively, try cutting the 4th wire of the fan if there's some sort of autodetection going on.

[gintama]

Re: Motherboard can't control fan - which component is bad?

@Mad_Professor: Not with any certainty. This happened a few years ago, I believe I was testing something with a DMM (probably what the lowest setting's voltage was) and the probe slipped in between two pins. So probably Ground + Power, since in my mind Power + Sense should have killed the RPM reading as well... Depending on what the DMM allows to pass between its probes, it could also have been Power+Sense+Ground? since the other probe was still on the computer's case? But again, I have little knowledge of modern circuit design or the effects of shorting things.

@cheapie: I am using a 3-wire fan, so that should not be the issue. I did check the BIOS though: there is a "Smart Fan Control" setting [Enable|Disable], and a "Smart Fan Method" setting [Auto|Voltage|PCM], but only for the CPU_Fan connector.

[gintama]

Re: Motherboard can't control fan - which component is bad?

Hello again, I had time to fool around with this last night and here is what I found. Links are to component whitepapers. The image below shows the connections for the CPU_FAN header; while the SYS_FAN2 header is pretty much the same, it's at the opposite end of the board from the rest of the stuff. Again, I have pretty much no idea what I'm doing here so any advice or corrections will be welcomed.



1. The fan header's V+ pin is connected to both a nearby capacitor (the other side of which is grounded), and the Drain of an AP3310GH MOSFET.

2. The MOSFET Gate is connected to an output pin of an LM358 op-amp.

3. The MOSFET Source is unknown but the multimeter shows connectivity between it and the VCC pins of most nearby ICs.

4. The fan V+ pin is also connected to the LM358's IN+ and IN- legs through some other parts, but I don't know what this does.


From some brief reading, there are a lot of failure modes for MOSFETS but many of them involve destroying the driving circuit, so I might have to replace multiple components... Hopefully the damage doesn't go any further than the LM358? I also learned static from the soldapult will kill MOSFETs, so I'll have to avoid using that... Which is unfortunate because there are way too many tiny SMT parts nearby for me to feel comfortable doing this with hot air.

What's also fun is that the MOSFET for the SYS_FAN2 header is located underneath a PCI-E socket, but since there are no pins under that section I suppose I can just remove it:

[budm]

Re: Motherboard can't control fan - which component is bad?

What resistance do you get between Source pin and the Drain pin? Since it is P CH Mosfet, the lower the voltage on the Gate pin to ground the MOSFET will be on. What DC do you get at the Gate? Check the reading in AC Volt also. MOSFET usually failed in Source to Drain shorted out.

[gintama]

Re: Motherboard can't control fan - which component is bad?

Hi budm, Source to Drain on the SYS_FAN2 MOSFET is ~1 Ohm, which I'm guessing is bad since the one for CPU_FAN is ~30 KOhm.

As for your other question: I don't know if the board will turn on without a CPU and RAM in it, so I may not be able to test Gate voltage until I disassemble my current machine (it's water-cooled, so removing anything is a bit complicated).

Can a MOSFET fail with both Source-Drain and Drain-Gate short-circuited? If yes then I guess testing the output voltage of the LM358 while varying the speed control software will not conclusively tell me if only the MOSFET needs to be replaced?

[budm]

Re: Motherboard can't control fan - which component is bad?

Yes, it is bad. That means the Drain pin was shorted to the ground pin so the MOSFET basically was put across the power supply so it just blown short. I hope it did not damaged the IC that drives the Gate.
Since I do not see any burn up Gate drive resistor, the IC is more likely to be OK, the output of the IC is stuck lo, the MOSFET will be fully on but will not damage the MOSFET, if the out put stuck high then the MOSFET will be OFF. So replace the MOSEFT for now.

[gintama]

Re: Motherboard can't control fan - which component is bad?

Sorry, I just edited my last message before realizing you had replied. All except the last paragraph is the same though.

Thanks for the advice, I'll swap the MOSFET out as soon as I can. After examining the PCI-E connector again, it looks like it would actually be less difficult/messy to just desolder it. Since there are no SMT parts on the back of the board in that area I can use hot air without worrying about blowing anything away...

[budm]

Re: Motherboard can't control fan - which component is bad?

"Can a MOSFET fail with both Source-Drain and Drain-Gate short-circuited?" Drain to Gate short can take out the out put of the GATE drive IC.

[kaboom]

Re: Motherboard can't control fan - which component is bad?

"Can a MOSFET fail with both Source-Drain and Drain-Gate short-circuited?" Drain to Gate short can take out the out put of the GATE drive IC.

In a power supply, yes. In such instances, you'd have ~300v+ backfeeding into the driver IC.

In this case, there's an opamp driving the MOSFET, likely through a resistor. Depending on the opamp, it could be "output short tolerant."

[budm]

Re: Motherboard can't control fan - which component is bad?

Yes, may be may be not, and so far I do not see the damaged Gate resistor, so the out put may be OK, we'll find out when he puts in the new MOSFET.

[gintama]

Re: Motherboard can't control fan - which component is bad?

Getting ready to do the surgery this weekend, yay! Two more questions in the meantime:

- Where would I find the resistor mentioned above? Would it be like the one shown linked to the MOSFET Gate in the above photo (but leading to the correct MOSFET of course)?

- If the damage does extend beyond the MOSFET, any guesses whether only the OpAmp will be damaged, or the SuperIO chip as well?

I wish I could find a pinout for the ITE IT8720F chip this board uses, I might even be able to add a third controllable header. SpeedFan finds 3 fan controllers, but only two seem to be physically populated with the required components... The other 3 headers on this board are 12V-only.

[gintama]

Progress & Problems



Got the PCI-E slot trimmed without too much trouble. I first tried to remove it in one piece using the re-work station but had no luck - this ROHS solder is a pain.

I didn't spend much time mapping for connectivity once this was exposed, but the only thing the Gate appears connected to is the empty land highlighted in the image below - there doesn't seem to be a resistor running from it to VCC like the one for CPU_FAN, but maybe it's back at the top of the board somewhere...



More troubling, it seems as though the MOSFETs on the board are bonded with some kind of adhesive.

Prying up on the Gate and Source pins only succeeded in snapping them off (the solder was melted at the time so the lands underneath are fine). Holding the iron on the wide part of the Drain didn't do anything, the body of the chip didn't even wiggle. I'm thinking it may be better to try and grind it down with a dremel until only the pin is left, and then desolder it. It's a small space and there are so many plastic bits and tiny resistors that using hot air on it would probably make even more of a mess. The side-effect of this is, I will have to source a new MOSFET, as the motherboard this one will be replacing surely has them glued down as well...

[shadow]

Re: Motherboard can't control fan - which component is bad?

Man that is so annoying that the MOSFET's are glued down!

[gintama]

Re: Motherboard can't control fan - which component is bad?

I suppose the safe thing to do, would be to solder the new one overtop of the dead one. Since the old Source and Gate pins are gone, I'd just need to connect the two Drains together and it should work fine. Not that I'm one to do the safe thing...

[budm]

Re: Motherboard can't control fan - which component is bad?

To remove that MOSFET, I trim off the two pin with wire cutter, then you will add more solder to the big tab, the soldering iron has to be about 60 WATTS with wide tip.

[gintama]

In other news, I refuse to type MOSFET in all caps after this post.

Good call budm, I borrowed an old Weller soldering gun and brushed some flux paste on there, once I figured out which angle of attack wouldn't melt the nearby plastics everything turned out pretty well:



After seeing the underside of the MOSFET, I'm guessing that what looked like adhesive was also flux (or impurities displaced by it). I've never been in a situation where I needed a separate flux before - the one I got (Orange-Eco) is targeted at plumbing but is supposed to be safe for electronics as well, hopefully there are no nasty surprises.

Anyhow, it should now be trivial to pull one of the MOSFETs off the other board after tearing down the system... I wish I had a quick way to test it though - I only have one PCI-E video card and removing the waterblock and re-installing the air-cooler is definitely not quick!