Now apart from the fact this is running at about 25 kHz there's nothing complicated there is there ? The positive output feeds the input to a 7815 regulator (+15v) , the negative output feeds the input to a 7915 regulator (+15). Both via a 'choke' winding on T2. Each rail has an electrolytic capacitor of 68uF and a 1 nF polyester capacitor from rail to common. Now bear in mind BOTH regulators have been removed for testing. Also bear in mind that these regulators are each rated at one amp, and are mounted on small upright heat sinks, so presumably are expected to be delivering at least a 100mA or so.
Right, so there is no load on either of the outputs as the regulator is removed - voltage from common to+ve is about +42v, and to -ve is about -35v so in the ball park of where you'd expect. If I load either of them with a 22k resistor, so only drawing perhaps 2 mA, the voltage drops to 12volts

High resistance transformer winding perhaps? No it measures virtually zero ohms. Dodgy diodes perhaps. Well very possibly. They only bear the marking "4L" with the cathode marked with a blue band. They may be some exotic species - quite possibly schottky fast recovery ones? I tried replacing them with 1N4007's from stock but only got about 12v out, then I tried 1N5819's which are the only schottky ones I have in stock, with the same result.
So what the heck is a '4L' diode? Poor things have been in and out of the board four times now.
Here is a picture of the little darlings, out and then back in the board.
By the way I had the second transformer out of the board to trace it's windings and it seems it's being used as a choke, three windings, each in series with a rectified transformer output.