Gigabyte GA-EP45C-UD3R VRM fault

[dicky96]

Hi guys
I have a Gigabyte GA-EP45C with a VRM problem

When attaching an ATX power supply the motherboard trips out the PSU

If I disconnect the 8 pin 12V connector it will power up

I was seeing about 36 ohms from 12V 8 pin connector to ground, and 0 ohm across the high side MOSFETs drain/source

There are 6 supply phases. I removed the High side MOSFETs one at a time.

Logic said to start with the three at the side of the PCB as they were easiest to desolder as there were less surrounding components

Sods law said that the short circuit MOSFET was the last one I unsoldered near the back panel connectors

The other five all test OK on my multimeter diode mode. I can turn the FETs on and off by touching +ve or -'ve to the gate, so they look just fine.

Having removed all the high side MOSFETs, I now have about 1Mohm from 12V 8 pin connector to ground, but I have about 48 ohm from the drain of the low side MOSFETs to ground. It starts a few ohms lower and increases like a capacitor charging. This is with no CPU fitted. So it's not a short just a bit of a low resistance.

I thought it was a good guess to unsolder the two low side MOSFETS on the phase that had a short circuit high side MOSFET.

However once i did that those also test OK on my multimeter, I now have 52.6ohms from Drain of low side MOSFETs to ground, again starting lower like a capacitor is charging up.

I could proceed to unsolder all the low side MOSFETs but I'm not sure if this really is a fault and don't want to be wasting my time unnecessarily (and I'm feeling lazy lol).

Maybe Vcore supplies something else like North Bridge and that is what I am reading?

I'm also thinking maybe to refit the five good high side MOSFETs are power the board up with 5 phases running just to see what happens?

High side MOSFETs A2726 and low side are A2724. They are not expensive.

Rich.

[piernov]

Re: Gigabyte GA-EP45C-UD3R VRM fault

There's a good amount of capacitance on CPU VCore so it'll indeed start at low resistance and climb up. It should climb to the kiloohms range after several seconds, if it stays low (without CPU) there may be an issue.
CPU VCore itself should not be going to the northbridge.

While it may work, I don't recommend trying to run without all phases. If ever you don't mind risking damaging something else, you have to also remove the low-side MOSFET or the inductor of the incomplete phase, and better have the first phases working correctly (ie. disable phase 6, not phase 0).


As usual, for a reliable repair I'd recommend replacing all high-side MOSFET and their drivers (or the controller if drivers are integrated) at the same time. And low-side if one is bad too. And coming from a good supplier (no random stuff from China eBay/Aliexpress), all from the same batch. It's more expensive but it's common for it to fail again when replacing only the failed MOSFET with a random one from Aliexpress.

[dicky96]

Re: Gigabyte GA-EP45C-UD3R VRM fault

OK thanks
I will try remove all the other low side FETs to see why I have a low resistance (about 50 ohms) from Vcore to ground as that sounds too low.

I was thinking of trying to run it with 5 phases just to prove the motherboard can actually boot and there is not some other fault, not as a permanent fix. I did already remove the low side mosfets from the bad channel. Is it still a bad idea as in it may damage something else?

I'm not sure what you mean by 'no random stuff from China eBay/Aliexpress' I can get the correct part number devices from Aliexpress, 10 of them for a few euros. Ebay was more expensive.

[dicky96]

Re: Gigabyte GA-EP45C-UD3R VRM fault

I removed all the low side MOSFETs. Interesting, as I removed each pair the resistance from Vcore to ground went up a couple of ohms

In the end, with all low side MOSFET removed, I have 71.8ohms from Vcore to ground.

All the low side MOSFET test OK on the multimeter, I can turn them on and off in diode mode.

Rich

[cpt.charlie]

Re: Gigabyte GA-EP45C-UD3R VRM fault

Maybe you have a bad VRM controller IC.

[dicky96]

Re: Gigabyte GA-EP45C-UD3R VRM fault

Funnily enough I woke up in the night thinking that the Controller IC phase pin would be connected to the junction of the High and Low side FETs and maybe that is the resistance I can see as, effectively, the Inductors connect all the phases together at that point; as far as my meter reading DC resistance would be concerned. Possibly I have a bad controller IC on the same phase I had a bad high side FET.

I don't go to bed thinking about these things but it is surprising how often I wake up about 4am with a light bulb turned on in my head in my head. Am I the only one that does that? The only other person I know who does the same thing is a computer programmer.

I did try to remove the inductors rather than the low side FETs but they are through-hole and my 60W soldering ZD916 Iron, even with some help from the hot air, just wasn't up to the job. I must by something better.

[dicky96]

Re: Gigabyte GA-EP45C-UD3R VRM fault

I removed the ISL6609 from the same phase as the faulty high side MOSFET but that didn't make any difference.

According to the typical application on the datasheet the only connection directly from the controller ISL6336 is on the VSEN pin which comes form the other side of the inductors.

I really need to remove the inductors so I can isolate the phases and load side to find where the low resistance actually is. Unfortunately my soldering iron just won't do it, even with a bit of help for the 858D

I'll order some FETS, drivers and a controllers anyway as they will take a while to come from China

I realise now I need to buy some decent kit to continue to work on motherboards. I'm thinking Quick 861DW and a Pace ADS200 would be up to the job?

Rich.

[cpt.charlie]

Re: Gigabyte GA-EP45C-UD3R VRM fault

I would recommend to remove the VRM controller and check again.