Having used all three Output connectors under full power burst & random-output level testing on this amp, I have yet to experience any output level dropping. I did at least find Ch 2 had reverted to producing nasty 60hz Buzz from picking up the field from the AC mains wires to/from the Standby Switch, and cured that with a fabricated brass container, listed elsewhere.
I changed signal sources, using bass, but, still having clients around me, I had to do that driving a dummy load. Then changed to a special tone burst generator (Sound Tech 3100B), where you can generate two different levels (a 0dB level and an attenuated level, of whatever duration on both levels you want), and set it up for 40hz 8 cycles at 0dB (200mS) and 1.8sec @ -20dB. During that test, I was seeing the AC mains current going quite high during that 8-cycle burst, and dropped it to 4 cycles (100mS). Still had the same AC Mains peak current level. I found once the client next door had moved into one of the isolated rehearsal rooms down the hall, I changed to my 4 ohm test speaker (Ampeg BXE-115HL4), and drove that at full output for a couple hours. Still, never had the amp randomly drop in output level.
I talked to our general manager, who was at the show and both heard the amp dropping in level periodically, changed out the amp with another, and the problem went away, and he identified this amp (having a tape label on the rear vent grille) as definitely being the one in question. So, it's been fun in the shop while being hoodwinked by the amp refusing to reveal it's alternative behavior.
Running at 40Hz tone burst into the test speaker, it's no longer at the lowest impedance point on the speaker, so we're not seeing as much peak current flowing as in the burst driving the 4 ohm dummy load, though not by much. Probably has more to do with it being a reactive load than a pure resistance load. Those reference lines on the plot are those of the 8 x 4 ohm/500W Dale Load Bank.
At this point, if I go any further, it will be to pull the chassis out of the cabinet, extract the output PCB to look af the solder joints, as well as on the main PCB. Last fully serviced in May 2018. It did change in it's Ch 2 Buzz since then, so, who knows. Its' been more fun trying to induce it to start coughing than more surgery.