On advice from a forum member I fired the Soundcraftsmen RA7502 up minus the 5A transistor on the B channel, under instruction to avoid playing anything loud through it. The following pic is a long shutter shot, in dark while the LEDs went through their start cycle (protect LED is not lit once the amp is on line) then a bit of light on everything for a few remaining seconds.
Just sitting at idle there is evidence that the missing transistor on the B channel is throwing the PNP and NPN rails out of balance, as that heatsink is not cold to the touch. After about 10 minutes of idle Channel A shows an offset of 48.7 millivolts while the ailing channel shows 148.7. I don't see more than just one trim pot (which I take to be bias), and on this amp it's imperative I don't knock stuff out of whack. The class H ranges to a higher voltage supply when headroom on the standard supply is approached, so blindly tinkering invites a BBQ. This essentially means I'll be securing a copy of the service data and giving it a proper go-through before putting it to work. Low likelihood I'll be bringing this to ENF, which is fine as I plan to only bring half the PA anyway.
I've decided I'm going to replace all the actives in both sides of the amp with either the On-Semi MJ21193/94 or MJ21195/96 complementary pairs. The standard ratings run is very similar, identical in fact except the SOA (Safe Operating Area) of 2.5A,80V,1 second for the 93/94 and 3A,80V,1 second for the 95/96, the real departure from similarity is in comparing the curves.
I'm a bit out of my depth in terms of relating power transfer to acoustical properties, but for the moment I am favouring the softer transfer curves of the 95/96 pair, not to mention the higher SOA. If anyone has an ailing RA7502 they want to cut free for cheap I'd love to A/B the different pairs.