CORRECTION: Forget my first proposed zener+resistor solution. Go with the last (triac) circuit.
Thanks for your updated schematic, I will try to breadboard it this weekend and compare them.
I did try to simulate the comparator circuit I posted in reply #4 on falstad but I failed to get it working.
(I could get it working with an op-amp but that has the wrong characteristics).
I still think the comparator is the nicest solution because it is easy to set the hysteresis also.
(Since we are using the capacitors to hold the relay it is important that it does not pull out if the voltage sags)...
Don't know if this has been mentioned, but a simple method would be to place a light bulb in series with the mains, then bypass the bulb with a proper sized relay that is driven by the supply output voltage. The inrush current won't be any higher than the bulb by itself, and the large amount of energy to charge the supply caps will get spread over the bulb since the bulb resistance increases as it heats up.
I actually used this exact method when testing the amplifier:
I brought it up first on a 60w then 300w incandescent light bulb and it worked fine.
Then when I bypassed the lamp I instantly blew the 10A fuses on the transformers secondary!
The reason was as I stated in my OP that one of the capacitors was shorted above ca 50v.
Though as crammed as the unit is fitting a light bulb inside it would not really be feasible.
Unless a really low wattage one would work, I may experiment because it is a fun way to do it!
And it has the added bonus of the thing not burning up if the relay fails to activate!
I've used this circuit (from Elektor 7-8/97) for many years for a subwoofer amplifier with a 330 VA toroid transformer. For some reason the R3 resistor has burned at some point. This resistor limits inrush current at switch-on, whereafter C1 limits current (C1 should preferably be X2 rated or similar). I changed it to a 2W resistor now to be on the safe side. Otherwise the circuit has worked fine. The circuit is meant for 230V and the relay coil is 24V (1200 ohm).
Thanks, this circuit is similar to SST-01 that I posted in my OP, the reason I prefer to measure the voltage on the output capacitors instead is to get rid of the time-delay logic.
And instead rely on the actual charge buildup of the capacitors to act as a timer, it may be overkill, but it is my goal with this circuit