[cz7asm]

The PWR_OK signal is used to signalize stabilized voltage rails *after* PSU powers up (PS_ON pin is asserted). The scenario I'm describing happens *before* PSU powers up - only the 5V AUX (standby) line is on. From this line the L1 controller is sourced plus all its peripherals like the L1 console, NVRAM + EEPROM or monitoring chips.
As I mentioned, I tried different PSU's that normally work with my other Fuels. Also to get L1 going one doesn't need to use a PSU, basically any 5V voltage source is enough.

However! Some moments ago I managed to try the NVRAM theory I mentioned in my previous post and it seems to help. It looks like this behavior is caused (or contributed) by NVRAM chip that has a dead battery. What I cannot explain why the power cycling helps though.

It's not easy to buy these NVRAM DS1742W-120 chips that still have healthy battery. I tried several pieces from eBay but they all seem to have dead battery or even none inside. I'm playing with an idea to try to build a circuit that would emulate it as there seems to be no replacement from Dallas available.

Anyways, now it needs some time to verify it really helped and it's not just a temporary glitch caused by today's snow calamity. Also many thanks for elaborating on the PSU PWR_OK issue - I'm sure I will encounter this problem sooner or later.