Recapped and cleaned. Some SMD ICs resoldered.

Analog board cleaned, dried, installed, and tested. It's now logic board turn.

I guess it was a bad solder joint somewhere.

I thought I found something. I put a second diode in parallel with D4, no change. I removed one leg of old D4, it started working. I fully replaced old D4 with new one and tried again, but the issue was back. I redid the solder joints in the mains area and it seems to work now.

Yep. HDD works and I have tried different HDDs. It works when warm enough. The 5V and 12V rails are not OK

It's just a guess since I can't measure it, but I do observe VCC worsen if CP4 area is cold and improve if CP4 area is warm. Testing reveals CP4 shouldn't be the issue. Main resistors measure OK. Maybe something is stealing power from that winding (red), or maybe DP4 is not giving enough power.

RP1 only provides like 4mA which is not enough to maintain TDA4605 switching, VCC needs power from an additional source. The TDA4605 starts switching, generating the secondary voltages, and generating a second VCC source, which comes from here highlighted in yellow. I hope I'm understanding this OK.

The capacitor for TDA4605 VCC is CP4 (47uF) but it doesn't seem to be the issue.
Unfortunately I can't peek at stuff on the primary side since its not referenced to ground.
If I understand the schematic correctly, VCC is generated from half mains via 22k resistor RP1 for the Start-up current.

It seems TDA4605's VCC is not stable enough. I believe the ripple is caused by TDA4605 VCC oscillating at 50 Hz. When left ON, TDA4605 VCC slowly rises up over time and ripple on 12 and 5 V decrease accordingly.. until Mac powers on. Heat is helping somewhere somehow.

Ok so i replaced most of the caps but it doesn't work. I get 12V and 5V but an HDD is not able to spin. The oscilloscope shows 2Vpp ripple on both at 50Hz with and without load. Voltages sag under load. I hope it's not the huge capacitor.

Btw it's from the same author as Sentenced to Be a Hero, IIRC.