LeeBefore I start hacking away at things, how about a 1N4007 diode anode on each ground wire from pins 2 and 4 then combining those two wires at a frame ground point away from the buss bar. Also have an exact 4uf axial np set of caps coming instead of the ecq 4.7uf's. Spent quite a while going over boards today, could not spot any solder bridges or loose connections, interesting to me that I get a solid output waveform from the amp when NOT attached to the speaker relay and hash/noise that increases with a decrease in frequency starting around 240hz and lower.
During the testing I also have the meters disconnected as well, infact at one point had all the lamps out of circuit as well in case there was some sort of backfeed going on.
I cannot tell for sure how you have the bias transistors wired from a color code perspective. Is the transistor a 2N5088? Is its flat facing the chassis? If yes, then emitter is green wire? Base is black wire? Red is collector wire?