Hi Chris, thanks for your thoughts.
The low voltage caps have been replaced. One amp had been done already and I've done the second one. The one with the issues has 3300uF on the 36V rails and 4700uF on the 76V rails. Under normal running the +/- 12V rails are quite even.
It does look like the 12V regulators and the wierd resistors around them could well be causing the issues but I don't really want to start redesigning the power supply in case there was a very good reason for doing it that way.
Luckily the attack by the hamfisted tech was fairly localised and mainly involved replacing the Op-Amps badly, leaving solder bridges, wondering why it no longer worked and giving up on it!
As has been mentioned on other threads about these amps, the lower supply rails seem to collapse quickly then somehow be recharged somehow from the higher rails.
I have checked the service bulletins in the version of the manual I have but they many deal with sensitivity changes, noise pickup reduction etc. Nothing mentioning switch-on/off thumps.
Incidentally, replacing the 470uF electrolytic to ground in the feedback network cured the switch-on thump but does nothing to help the switch-off.
It's also curious that one channel seems to suffer much worse than the other which makes me think it's not so much a power supply issue.
Chris