The board does have six KZG 6.3V 1200uF capacitors. None of them are showing any signs of leakage. I removed one next to the ATX connector and tested it. It reported almost 1200uF and 0.02 ESR. The other board has four of these caps leaking, but on this first board I’m assuming they’re probably good.

I also ran some thermal imaging to see if anything lit up.

There is a diode connected between a voltage regular (?) 14N03LA MOSFET and FAN5067 ACPI controller. It registered 93C. I know these diodes can have operating temps in excess of 100C, so I don’t know if this is necessarily an issue. I tested the diode and there was no short.

The other thermal issue is the Winbond W83627HG I/O chip. It reported a hot spot of 75C after less than a minute of powering on. The operating temp for this chip is between 0 and 70C, so it’s already hotter than it should be. Again, I don’t know if it’s faulty and potentially causing the board to fail to boot, but I have concerns about that chip.

From what I can see in the photos, that diode, along with the integrated circuit right next to it, is related to the SATA/IDE/CF drive system.
It also appears that the diode's cathode connects directly to what looks like a voltage regulator located next to the DIMM sockets, and the other anode connects to a filter capacitor.

What you could do is check the part numbers of the supposed regulator and, if it is indeed the regulator, compare its voltage output with that of your other motherboard (even if the other motherboard isn't working, that regulator might still be functioning correctly).

that regulator is most likely FAN5067 Shponglefan was talking about. It makes 3.3V for PCI and whatever voltage DIMMs need. Diode is connected between ram supply and something esle.

> I tested the diode and there was no short

of course, If there was short it would just conduct with no power losses thus being cold 😀
What I dont understant is diode orientation, it loks like it provides power when U39 is not running? and being super hot means something else is trying to power ram thru this diode alone
measure resistance both sides of the diode to ground. Its possible ram is shorted, U39 detects that and refuses to start leaving poor Diode to try and power ram/short and cooks

post codes you are getting might be before ram initialization altogether

all in all I would start by verifying everything is getting good power first, cpu ram chipset slots

Bit of an update on diagnosing things.

First, I forgot to mention previously the power on behavior of this board. It will immediately power on after switching on the ATX power supply switch. It doesn't wait for the soft power ATX switch on the motherboard.

If I power off the board and immediately turn on the ATX switch on the PSU, it will be stuck with the reset line on and no POST codes. I have to wait about 5-10 seconds after powering down, before it will cycle through POST codes again.

I took OttoPS's advice in comparing the voltages on the two boards. The second board does wait for the soft power ATX switch rather than just powering up right away.

What I referred to as the voltage regulator previously is a 14N03LA MOSFET. It is connected from pin 1 to the FAN5067M ACPI controller chip on pin 14 (VADJOUT). The MOSFET also also connected from pin 2 to the diode (D9) which connects to pin 13 (VADJ) on the FAN5067M chip.

Here are pictures from both boards of the circuit.

I did end up replacing the KZG caps on the second board, since 4 of the 6 were visibly bulged. After replacing those caps, I got a POST of 13:12 on the second board. This is the same POST code I got on the first board with the missing inductor. And since the inductor is now removed from the second board, I expected that.

I measured voltages on both after the recap of the second board.

On the first board, I get 0V on the VADJOUT and 1.58V on VADJ. On the MOSFET I get 0V on pin 1, 2.9V on pin 2, and 1.53V on pin 3.

On the second board, I get 5.43V on the VADJOUT and 2.6V on VADJ. On the MOSFET I get 5.43V on pin 1, 2.9V on pin 2, and 2.6V on pin 3.

I also took some thermals of the second board. Neither the diode nor the Winbond I/O chip were overheating.

At this point, I may try to resurrect the second Advantech AIMB-742 motherboard. I ordered replacement parts from Digikey to replace the components I cannibalized for the first board.

If I can get the second board fully working, I have a better base of comparison with the first board.

Today I turned my attention to the second AIMB-742 motherboard to see if I could resurrect it.

I'd found a picture I'd taken when I first got it. It was pretty nasty.

It clean up okay, although there is still a bunch of rust on the chokes near the CPU. I think there is also some rust under the heatsink next to the chokes.

Four of the six KZG capacitors were bulging. I opted to replace all six capacitors with polymer caps.

In the process of replacing those caps, I accidentally knocked of a few of the SMD caps near the RAM slot while cleaning. The solder on these was corroded and weak. I cleaned them up and replaced them with new ones.

I also replaced the inductor and a few SMD caps on the bottom of the board that I had previously removed.

After doing the recap and replacement of other components, I tested the second board. To my surprise, it booted up just fine.

I ran some benchmarks and other utils and everything seems to be working. No lock ups or other issues occurred in my brief testing.

At least of the two motherboards, I managed to get one working. I'm surprised it was this one given all the corrosion, leaky caps, and general filth.

But at least with one working board, I have a baseline for comparing the other non-working one.

Congrats!

Yeah, when I saw the updated title I assumed you have gotten the first one working and had started on the second.
And ... yes: having a fully working one will help a LOT debugging the other.
Best of luck going forward!

Shponglefan wrote on 2026-04-21, 14:44:

FAN5067M ACPI controller chip

every time you call it ACPI controller I cringe 😀 Its Vreg for RAM and 3.3V

Shponglefan wrote on 2026-04-21, 14:44:

On the first board, I get 0V on the VADJOUT and 1.58V on VADJ. On the MOSFET I get 0V on pin 1, 2.9V on pin 2, and 1.53V on pin 3.
On the second board, I get 5.43V on the VADJOUT and 2.6V on VADJ. On the MOSFET I get 5.43V on pin 1, 2.9V on pin 2, and 2.6V on pin 3.

2.6V on pin 3 is healthy DDR1 ram supply
you have something shorting ram supply on first board

rasz_pl wrote on 2026-04-22, 05:41:

Shponglefan wrote on 2026-04-21, 14:44:

FAN5067M ACPI controller chip

every time you call it ACPI controller I cringe 😀 Its Vreg for RAM and 3.3V

https://mk.ic-components.com/files/4c/FAN5067M.pdf

Shponglefan wrote on 2026-04-21, 14:44:

On the first board, I get 0V on the VADJOUT and 1.58V on VADJ. On the MOSFET I get 0V on pin 1, 2.9V on pin 2, and 1.53V on pin 3.

1. Check the operation of the charge pump.
In Standby mode (+5V STB with ATX present, main voltages off), pin 16 should read at least 9–10 volts.
In active mode, pin 16 should read “ATX 12V – 1 diode,” i.e., approximately 11+ volts.
2. Remove mosfet, check Vadj must be around 11V ( ~ pin 16) at S0 (active) and S3 (Suspend to RAM, but you can`t change this option at the not working board)
0V at S5 (soft-off -standby)
In short, there is no control voltage for the MOSFET.
Either the MOSFET is faulty, or the FAN is faulty, or the FAN is not receiving the necessary signals to activate the memory regulator.
Consequently, the power supplied to the memory via the diode from the 3.3V Dual source is limited to the maximum current that the diode can handle.

Check the power supply to the heating SIO. If the motherboard powers up on its own, you must break Power-on signal circuit to the PSU to check the +3 V Dual in standby mode.
If it is working properly, the ACPI status will not change without a "live" SIO.
It is most likely controlling them via the them GPIO pin.

DaveDDS wrote on 2026-04-22, 02:25:

Congrats! […]
Show full quote

Congrats!

Yeah, when I saw the updated title I assumed you have gotten the first one working and had started on the second.
And ... yes: having a fully working one will help a LOT debugging the other.
Best of luck going forward!

Thank you! I'm hopeful to get it working!

rasz_pl wrote on 2026-04-22, 05:41:

every time you call it ACPI controller I cringe 😀 Its Vreg for RAM and 3.3V

The official datasheet calls it an "ACPI Dual Switch Controller" and also states, "The FAN5067 is a fully compliant ACPI controller IC".

¯\_(ツ)_/¯

2.6V on pin 3 is healthy DDR1 ram supply
you have something shorting ram supply on first board

That is something I need to figure out...

shevalier wrote on 2026-04-22, 06:13:

1. Check the operation of the charge pump. In Standby mode (+5V STB with ATX present, main voltages off), pin 16 should read at […]
Show full quote

1. Check the operation of the charge pump.
In Standby mode (+5V STB with ATX present, main voltages off), pin 16 should read at least 9–10 volts.
In active mode, pin 16 should read “ATX 12V – 1 diode,” i.e., approximately 11+ volts.
2. Remove mosfet, check Vadj must be around 11V ( ~ pin 16) at S0 (active) and S3 (Suspend to RAM, but you can`t change this option at the not working board)
0V at S5 (soft-off -standby)
In short, there is no control voltage for the MOSFET.
Either the MOSFET is faulty, or the FAN is faulty, or the FAN is not receiving the necessary signals to activate the memory regulator.
Consequently, the power supplied to the memory via the diode from the 3.3V Dual source is limited to the maximum current that the diode can handle.

Check the power supply to the heating SIO. If the motherboard powers up on its own, you must break Power-on signal circuit to the PSU to check the +3 V Dual in standby mode.
If it is working properly, the ACPI status will not change without a "live" SIO.
It is most likely controlling them via the them GPIO pin.

All good advice.

Right now my plan of attack is to map out the voltages/signals of the all the pins for the FAN5067 IC on both boards and see what differences there are.

From what I read in the datasheet, it's controlled by the PWR_OK, S3 and S5 signals. I want to see if I can figure out where those are connected to as well.

Prior to doing other testing, I decided to remove and test the KZG caps that were on the first board. One of the non-leaking caps from the second board tested a bit out of spec, so I wanted to rule out those potentially causing issues.

After removing all those caps, they did end up testing fine. I opted to replace them with the same polymer caps I used on the second board.

Shponglefan wrote on 2026-04-22, 18:26:

Right now my plan of attack is to map out the voltages/signals of the all the pins for the FAN5067 IC on both boards and see what differences there are.

+3.3V DUAL – that’s where I’d recommend starting.

If it’s too high (the regulator from 5V standby isn’t working), then everything powered by it will overheat.
This could include the SIO (which Winbond, ITE, Nuvoton...).
If the SIO is overheating in standby mode, unfortunately, it’s most likely dead, alas.

I'm not expecting that Winbond chip to have survived. I measured temps again and at one point it had hit 80C.

Good news is that replacements seem readily available.