Roman555 wrote on 2024-07-07, 20:02:

Before desoldering check that the resistor's pads are connected to the VCC pin and +12V P4 rail using a multimeter in ohm-meter mode.
Yes, I think it should be near ADP3168. It's a SMD resistor so it's better to use some magnifying glass or glasses to make desoldering proper.
Usually the SMD component is stayed soldered to one of its pads but it is just shifted or rotated to break chain. It's convenient as such tiny thing is easy to lose.

Checked to see if the resistor was connected to both VCC and 12V P4 and it was. Desoldered it, but still, connecting 12V P4 has the same effects as before. I guess this points to the ADP1368 being bad?

kotel wrote on 2024-07-08, 07:54:

Roman555 wrote on 2024-07-07, 20:02:

Before desoldering check that the resistor's pads are connected to the VCC pin and +12V P4 rail using a multimeter in ohm-meter mode.
Yes, I think it should be near ADP3168. It's a SMD resistor so it's better to use some magnifying glass or glasses to make desoldering proper.
Usually the SMD component is stayed soldered to one of its pads but it is just shifted or rotated to break chain. It's convenient as such tiny thing is easy to lose.

Checked to see if the resistor was connected to both VCC and 12V P4 and it was. Desoldered it, but still, connecting 12V P4 has the same effects as before. I guess this points to the ADP1368 being bad?

What do you mean "the same effects" ?
Where is the low resistance to GND now - on the VCC pin or on the 12V P4 rail ?

After desoldering it you can measure resistance between the VCC pin and GND and also resistance between the 12V P4 and GND separately, compare and watch the difference between them and early measurements. It was the purpose of desoldering the resistor.

kotel, try to understand how the components are connected. There're no schematic diagrams for any model of Intel mainboards. Frankly Intel mainboards were reliable enough to not have typical malfunctions that mainboards of other manufacturers had.
So print necessary pages of the reference schematic diagram and find relation between components on the PCB and on the diagram. Read datasheets. It'll help you to find where and what to measure and what to expect.
There's no step-by-step algorithm that guarantees a faulty place will be found and a mainboard will be repaired.

Roman555 wrote on 2024-07-08, 15:09:

What do you mean "the same effects" ? Where is the low resistance to GND now - on the VCC pin or on the 12V P4 rail ? […]
Show full quote

kotel wrote on 2024-07-08, 07:54:

Roman555 wrote on 2024-07-07, 20:02:

Before desoldering check that the resistor's pads are connected to the VCC pin and +12V P4 rail using a multimeter in ohm-meter mode.
Yes, I think it should be near ADP3168. It's a SMD resistor so it's better to use some magnifying glass or glasses to make desoldering proper.
Usually the SMD component is stayed soldered to one of its pads but it is just shifted or rotated to break chain. It's convenient as such tiny thing is easy to lose.

Checked to see if the resistor was connected to both VCC and 12V P4 and it was. Desoldered it, but still, connecting 12V P4 has the same effects as before. I guess this points to the ADP1368 being bad?

What do you mean "the same effects" ?
Where is the low resistance to GND now - on the VCC pin or on the 12V P4 rail ?

After desoldering it you can measure resistance between the VCC pin and GND and also resistance between the 12V P4 and GND separately, compare and watch the difference between them and early measurements. It was the purpose of desoldering the resistor.

kotel, try to understand how the components are connected. There're no schematic diagrams for any model of Intel mainboards. Frankly Intel mainboards were reliable enough to not have typical malfunctions that mainboards of other manufacturers had.
So print necessary pages of the reference schematic diagram and find relation between components on the PCB and on the diagram. Read datasheets. It'll help you to find where and what to measure and what to expect.
There's no step-by-step algorithm that guarantees a faulty place will be found and a mainboard will be repaired.

Got some time to work on this (sorry for not replying). By same effects I meant the VRM still shorts the PSU (my english can be sometimes hard to understand). Whats weird is that, after desoldering there weren't any connections to GND, 5V, 12V, 3.3V, 5VSB and 12V P4. After soldering it back on there still weren't any connections to the power lines. Ill look into the VRM schematics of an MSI MS-6850, which is also an socket 478. I still find this issue very weird since quite a lot of people had the same issue, but every post (I came across) died out or they bought a new board.

kotel wrote on 2024-07-15, 14:26:

Got some time to work on this (sorry for not replying). By same effects I meant the VRM still shorts the PSU (my english can be sometimes hard to understand). Whats weird is that, after desoldering there weren't any connections to GND, 5V, 12V, 3.3V, 5VSB and 12V P4. After soldering it back on there still weren't any connections to the power lines. Ill look into the VRM schematics of an MSI MS-6850, which is also an socket 478. I still find this issue very weird since quite a lot of people had the same issue, but every post (I came across) died out or they bought a new board.

Sorry, I doubt there was MSI MS-6850. Is it a typo in MSI MS-6580 ? Then a SM of MS-6580 says it's built using a very different component base.
That schematics of Intel 865 I provided under the link on the first page of the topic is much more suitable (IMO).

Yes, sorry. I always think its an MS-6850 but its MS-6580.

The P4 connector only supplies voltage to the VRM controller and that supplies voltage to VRM?

kotel wrote on 2024-08-05, 11:18:

The P4 connector only supplies voltage to the VRM controller and that supplies voltage to VRM?

Yes. The idea is to supply a CPU core from a dedicated rail.

Well... this is really bad. Went to test it today to see if something changed and... the board was powering off and on real fast and there was an unusual burning smell coming from the VRM mosfets. Nothing looks to be burned on the outside. There was also a weird "sizzling" noise too. I guess what probably happened is that the ADP3186 got disconnected (or something broke inside) and the mosfets started burning?

kotel wrote on 2024-08-23, 10:13:

Well... this is really bad. Went to test it today to see if something changed and... the board was powering off and on real fast and there was an unusual burning smell coming from the VRM mosfets. Nothing looks to be burned on the outside. There was also a weird "sizzling" noise too. I guess what probably happened is that the ADP3186 got disconnected (or something broke inside) and the mosfets started burning?

It's pity. It looks like the drivers (ADP3418) did't get voltage VCC (because the resistor was unsoldered) and couldn't close VRM MOSFETs so they burned off.

I didn't see any burn markings on both the board and mosfets, so I think there is still some hope left... I wonder what else could've died. I know for a fact some primary side mosfets are definitely toast (smell came from them) and maybe even the CPU too (I hope its not). Maybe even the mosfet drivers? Trying to see if its still worth (for me) to attempt repairs. I also wonder what would happen when I solder a good known 10 ohm resistor back into (what I assume) the broken resistor place. I don't think it would cause anymore harm than I've already done.

It's better to measure conductivity of every VRM mosfet and also power supply rails to the ground. Often shorted mosfets and their driver are replaced at the same time. It is necessary to check gate resistors. So It will be a lot of measure of conductivity to check many components before trying to turn on the system even without a CPU.
A CPU could be damaged if +12V P4 went through shorted or opened during long time a high-side mosfet (when a low-side mosfet was closed at the same time) that Vcore could become too high.

Roman555 wrote on 2024-08-25, 20:15:

It's better to measure conductivity of every VRM mosfet and also power supply rails to the ground. Often shorted mosfets and their driver are replaced at the same time. It is necessary to check gate resistors. So It will be a lot of measure of conductivity to check many components before trying to turn on the system even without a CPU.
A CPU could be damaged if +12V P4 went through shorted or opened during long time a high-side mosfet (when a low-side mosfet was closed at the same time) that Vcore could become too high.

+1

I had an Intel DG33TL board with a shorted upper MOSFET in one of the VRMs that kept burning over and over again, sometimes taking MOSFETs out elsewhere in the other phases. I ended up replacing all of the ADPxxxx (forgot exact part number) MOSFET drivers and all of the upper MOSFETs at once. Before soldering any of the new-used parts (I borrowed from a bunch of dead Xbox 360 motherboards) I checked all MOSFET Gate resistors, bootstrap capacitors, and other components around the MOSFET drivers, because I got tired of replacing parts here and there one at a time only to get more burned parts. That took care of the issue.
Surprisingly, my CPU didn't get damaged the 5 times an upper MOSFET shorted out and dumped 12V to V_core (it was a Pentium E2180 CPU, though I also had a Pentium 4 630 take a short to 12V one time on this board.) Then again, for such problematic hardware, I always try to use a known good PSU rated at 350 Watts or less so that it can shut down fast when it finds a short-circuit or overload.

So I suggest doing the same thing: replace all MOSFET drivers at once and all upper MOSFETs too (as they are the more likely ones to get damaged, though not always.) And put a known good CPU in it too. Don't want to do all of that work to possibly fry everything again with a dead CPU. Technically speaking, the CPU VRM should be able to detect a shorted/bad CPU and shut down properly (or over-load the PSU and make that shut down)... but you never know.
FWIW, Pentium 4 CPUs are still not that hard to find. So if you only have "nicer" higher-end P4's, maybe get something like a low-end 2.4 or 2.66 GHz Northwood/Prescott as a sacrificial lab rat.

Did measurements on the VRM mosfets and the only ones that seem to be damaged are Q1B3 and Q2B3 on the low side (both have around 0 ohms on gate to ground). I also saw that resistance droped and rose here and there on both high and low side of the VRM, which is weird... I guess i could replace that 10 ohm resistor connecting 12V P4 to ADP3168 cause it looks like it doesn't read anything anymore (might also be the probes cause they sometimes don't want to read SMD componnets). The CPU that was inside now was the unsupported Celeron D, so that might have also caused the VRM to malfunction.

Replaced the 10 ohm resistor but that didn't help. The same no POST codes stuff. Time to get searching on ali for a replacement ADP1368 then.