Once the switches were cleaned, I could see that the relays were shot to hell and doing very bizarre things to the signal. SAE used 48V relays, but I had no luck tracking them down. Not a huge deal...I just converted the circuit to use 24V relays instead. Required swapping out a few caps and resistors. The main dropping resistor went from 1K 3W to 750 ohm 8W (but is dissipating only about 4W). You can see the new relays in the pics, as well as the 8W open-center resistor, chosen to shed heat quickly so it doesn't desolder itself from the PC board.
The relay board was badly hacked, but once I removed the 48V relays I saw why...SAE used some bizarre relay configuration, and the last guy to replace these had no choice but to chop the shit out of the board to fit some standard 'LY2-style' relays in place. Some really creative soldering was necessary to get connection, but it'll work fine.
The DC offset on the amp settles so slowly that even with the 6 or 7 second power on delay, there still is a small DC 'thump' at the speakers. Not much I can do about that. The relay circuit does have provision to disconnect if excessive DC offset is present, but does not have a circuit to open the relays on power off...the relays just open when the circuit discharges. I don't like this, as you get a DC 'thump' (nothing terrible) as the relays open in addition to the 'thump' you get when they close. SAE knew all about this, and even put a 0.1µf 250V cap across the relay contacts!! to help prevent pitting of the contact surfaces. This makes for some real entertainment as the amp is powered down, as the 0.1µf caps act as a signal path for the strangest damn whine as the circuits discharge. I found this unacceptable, and changed the cap to a 0.01µf 250V cap, which dropped the whine down by a good 20db. You can still hear it, but it's only a curiosity now instead of the real eye-opener and embarrassing noise that it was beforehand.