DC offset ± 48V situation on Yamaha AX 750

As an answer: -35.4 - (-38.5) = 3.1 V, close enough to 2.2 but huge negative offset comparative to normal ±1.1V
No, in this case that is not close enough. Note the position of VR601. Rotate it through it's range while monitoring the voltage between the bases of Q635 and Q636.

The offset is going to be meaningless as long as the amp is open loop.

I suggest 1...2...3 kohm resistors instead of Q640 and Q642 to avoid "floating".
That is one way to close the loop without connecting the outputs. But I would suggest lifting R673 and connecting the emitters of the drivers through 120 ohm resistors to the output line (the center connection on either of the 0.22 x2 resistors). But check the voltage range you see adjusting VR601 first.
It is my first time on such a "transient troubleshooting". It seems to be more difficult comparative to a static situation.
Open loop, any troubleshooting will be difficult and more luck than skill.
 
Getting better with 2x100 ohms instead of outputs. Stabilised.

It is far more better now : ± 1.5 V instead of ± 40V.
Should be ± 1.1 V.

Still high idle current if outputs connected.

R470 operational (1 V up and down comparative to A,B)

I'll try to figure how to decrease idle current now.
 
Confirm Q632 is good. If you still have the original part and it test good (diode tests) then you might want to reinstall it. Otherwise, check R660 & R665.

Q632 is superior equivalent. Original part found cracked.R660 & R665 are original parts and OK. I will figure out idle current adjusting without changing them.


At this point I would like to see 2.0 volts between the bases of Q635 & Q636.

Me, too.