Wow, didn´t remember this old amp used such crazy high rails for *just* 300/4 .
For that power/impedance Industry
standard is +/-70V or so, sometimes +/-65V if you have a good power supply , hence my suggestion, which works perfectly in most any Ampeg (SVT3 Pro=+/-65V), Fender (BXR300=+/-70V) and so on and on and on.
You´ll find similar rails in similar power Harke, Ashdown, Laney, whatever, ... because .... um .... that´s what´s needed.
300/4 means 34.6V RMS , round that up to 35V RMS.
Which means: 49.5V peak.
Add slightly over 4V for transistor drop, + emitter resistor loss, you need +/-54V rails.
Now real World supplies drop under load, plus ripple appears; an average supply drop in a well designed commercial amp (DIYers and Audiophiles don´t count, they don´t have to work within a budget) is around 15%.
So no load rails in a Commercial 300/4 amplifier must be around (54/.85)=+/-63.5V
Call it +/-65V to have a little extra margin and we are talking.
Spec +/- 70V if you want even more margin andd/or use a cheaper/smaller/lighter transformer.
Or if you use MosFet outputs which often need somewhat higher voltage drop.
But +/-85V is
*crazy* , it means dropping (54/85)=36.5% !!!!
Only explanation I find, because GK are no fools, is that they do it on purpose, to get more "dynamic/musical" power for the same RMS rating.
FWIW more modern GK700 RB still uses +/-85V rails ... but it´s rated 480W RMS into 4 ohms !!!!
Back to the OP problem: pity he got stuck trying to repair such an unusual amplifier, with such a stressful supply (it´s definitely not easy on the amps it feeds) and to boot with a nuked PCB.
Still offer my "Plan B" as a practical solution, but now searching for a power module which can stand +/-85V ... even if those will drop *sharply* under load.
Plan C: get a
full ClassD power amp + integrated SPMS all-in-one-board, similar to ICE Power modules ... and in fact what G&K uses today.
What also do Fender, Aguilar, T&C Electronics and most others.