[morzh]

The distortion will be obviously present all across the output amp as the feedbacks from the outputs go to Q11/Q12.

This looks almost like the power voltage sags when you reach the peak, and it also with the load (without load you do not load the power).

Now, can you look at the actual +46V power when this happen with the scope on AC, and see if the "dip" (which will look like the "dip" in your output signal) is present there, right on the power bus, preferably closer to collector of either Q4 or Q1 (triangle marked as DD or maybe OD).
It should be present at the output for sure (both sides of the C29/C30 caps) but I am interested in the power.

If it sags, it will tell us something.


PS. Check every diode in the rect bridge and the good connection to F2 fuse, C12 cap (1000 uF) and good connection to the "DD" terminals of your output. If the wire to DD could be disconnected and the bridge loaded with say 2A load, you could measure the voltage and see if it sags under load.