[weblacky]

Based on your response, I'd suggest monitoring the PWR_OK pin of the main PSU connector. https://wiki.preterhuman.net/Fuel_PSU_repair

My guess would be it's not being signalled fast enough due to PSU age, restarting a "charged" PSU is working...but not a cold start-up.. You may need a basic oscilloscope to see this signal. Most modern PCs need the signal within about 400ms (300ms is perfect), I've seen "zombie" PCs (fans start but board never posts) where the signal was at 800ms (but all voltages were GOOD!). Many boards need the signal within a specific power-on time period (they don't just wait forever then start at the signal). This indicates that the PSU is taking too long to stabilize its outputs for clean start-up

That's my basic guess, it's the PSU signalling. If you want to try faking this (dangerous as PSU really hasn't stabilized, which is what PWR-OK is for, you could remove the PWR_OK pin from your connector and use a bench PSU with a resistor on the connector trace or solder leg to fake a good signal immediately from power-on...likely you want to try VERY low current on this signal pin until it works...likely nothing greater the a few hundred MilliAmps to make the "signal" (pull "high" be present from the start).

However there is a reason for the signal. I'd start by doing a basic measurement and measure the time from powering on until the signal goes fully high. That time will be less than 1 second...and likely needs to be less than 500ms for the board to actually turn on correctly. You'll also get the max voltage you need to be at to tune a test-signal hookup.

Let us know what you find.