Just some "food for thought".
Remember that I am not a master builder or EE, so my probing techniques may not be
"textbook perfect" and somewhat brutish and primitive, but they get me to where I want to go.
1)Did you take steps to correctly insulate the heatsinked HV sand to avoid an arc path across the mounting screws?
2) Did you cut all of the traces (or lift/remove all of the associated parts) that I suggested before
when you went with the zener string in lieu of the battery circuit?
3) Sounds possibly like some leaky capacitance at work here. ("output voltage raises slowly")
If it were me, I would cut some traces and/or remove Q1, Q2 and Q8 and anything else that connects across the rails
in order to isolate the filtered input stage and the bias circuit.
I would then verify that the filtered DC was correct entering the bias circuit, and recheck the bias output for 580 volts.
Divide and conquer. If the bias nows reads correct, test then replace the items you removed previously. (Or reconnect the cut traces)
4) As for the main power rails, how do they compare to one another? (static test and under power)
I am assuming that #2 above was completed, because I have no way of knowing what is going on with the batteries.
This could shed some light on different faults in the two curcuits.
