Did some voltage measurements on the FAN5067 chips on both motherboards. Below I've listed the voltages for the respective motherboards.

Voltages from the working motherboard:

Pin 2 - PUMP - 2.74V
Pin 4 - DUALOUT1 - 11.84V
Pin 5 - DUALOUT2 - 0.078V
Pin 7 - SLP_S3 - 3.28V
Pin 8 - SLP_S5 - 3.28V
Pin 12 - VADJFB - 1.25V
Pin 13 - VADJ - 2.61V
Pin 14 - VADJOUT - 5.43V

Voltages from the non-working motherboard:

Pin 1 - QCAP - 7V
Pin 2 - PUMP - 2.64V
Pin 4 - DUALOUT1 - 0.0V
Pin 5 - DUALOUT2 - 4.48V
Pin 7 - SLP_S3 - 0.0V
Pin 8 - SLP_S5 - 0.0V
Pin 12 - VADJFB - 0.76V
Pin 13 - VADJ - 1.58V
Pin 14 - VADJOUT - 0V
Pin 16 - VCCP - 11.9V

Voltages on Pin 1 (QCAP), Pin 2 (PUMP), Pin 3 (5VSTBY), Pin 6 (DUALFB), Pin 9 (PWR_OK), Pin 10 (SS), Pin 15 (5VMAIN), and Pin 16 (VCCP) are approximately the same on the two boards.

The lack of SLP_S3 and SLP_S5 is concerning. My understanding is those are generated via the ICH5 chip, though there could be various reasons why they are not present on the non-working board.

I'm also wondering if that Winbond chip overheating is a symptom of the issue or possibly a cause of the issue. Is it possible that a short in the Winbond chip itself could trigger SLP_S3 / SLP_S5 to remain low?

I could try removing the Winbond chip and even replace it. Though I wouldn't want to have the replacement get subsequently cooked if the issue lies elsewhere.

Shponglefan wrote on Yesterday, 16:04:

Voltages from the non-working motherboard:

Pin 2 - PUMP - 0.0V......
and Pin 16 (VCCP) are approximately the same on the two boards.

How's that? If the charge pump isn't working, there shouldn't be any voltage on pin 16. In standby mode, of course.

Main Power. Apply +12V through a diode on this pin to run the circuit in normal
mode. Bypass with a 0.1μF capacitor. When 12V is not present, this pin produces
voltage doubled 5V standby.

shevalier wrote on Yesterday, 16:29:

How's that? If the charge pump isn't working, there shouldn't be any voltage on pin 16. In standby mode, of course.

You're right, I must have mis-measured it the first time.

Just double checked it and I get the following (non-working board):

Pin 1 - QCAP - 7V
Pin 2 - PUMP - 2.64V

Pin 16 - VCCP - 11.9V

Did some more research into the SLP_S3 and SLP_S5 signals. From reading various threads at the BadCaps forum, it sounds like one of the potential causes could be a bad SIO chip. This could cause the ICH5 to not set SLP_S3 and SLP_S5 to high and hence not power up properly.

Since the Winbond chip is overheating I'm working on the assumption that chip is faulty and will need to be replaced.

I did do some probing of voltages on the chip and the 5V and 3.3V pins all read the correct voltages. So at the very least it looks like it's getting the correct power.

Also checked the SLP_SX GPIO pin and it reads 0V, which isn't surprising given the missing signals on the FAN5067 chip.

Shponglefan wrote on Today, 01:47:

Did some more research into the SLP_S3 and SLP_S5 signals. From reading various threads at the BadCaps forum, it sounds like one of the potential causes could be a bad SIO chip.

I don't know the current state of Power OK, but SLP being at "0" is a completely valid state for ACPI.
What can be seen right now:
-The charge pump is working, which is good. It's strange that it's 11.5V in Standby mode (the state without the PSU fan turned on, i.e., Standby). But that's not particularly important.
-3.3V DUAL seems to be working too.You have an issue with the memory power supply.
-Something is coming in from the outside to pin 13. It should be 0 there.

And let me ask you one more time, in what state are you measuring the voltages on this IC?
In Standby or working state?

shevalier wrote on Today, 11:25:

And let me ask you one more time, in what state are you measuring the voltages on this IC?
In Standby or working state?

The voltages are:

PWROK - 3.56V
SLP_S3 - 0V
SLP_S5 - 0V

Based on this, the FAN5067 chip is in the standby (S5) state.

My working theory is the bad Winbond IO chip may be interfering with the ICH5 chip from generating the SLP_S3 and SLP_S5 signals.

There is a post at the Badcaps forum which describes what is needed for proper SLP_S3 / SLP_S5 function. This post is in the context of laptops, so not everything applies: https://www.badcaps.net/forum/troubleshooting … 020#post1373020

I also did measure RSMRST at 1V on the Winbond chip.

I removed the Winbond W83627HG chip.

After doing so there was an immediate change in the board's behavior. It no longer starts up when switching on the ATX power supply. Of course, it no longer powers up at all (i.e. soft power doesn't work) but given the removal of the IO chip this is expected.

I also checked thermals and there are no significant thermal hot spots when the ATX PSU is switched on.

Next up will be to solder a replacement chip. I'm debating how I want to do this. Hot air is a bit of a pain to use given the proximity to the RAM sockets and other components. I may just solder it manually.