As I cannot find anything wrong as far as broken wrong, I will assume this
is as designed. I cannot see why there is a low 4K7 dropper from 27V to the
4.7V zener that basically causes this 5V DC pulse, when only supplying 80uA
at most into the base of a TO92 transistor, Q13. I will try upping the
dropper and adding a C to give a time constant of 2 seconds and change the
main hold off time constant from 4 sec to 2 sec also but keeping 13V bias
point, both channels. A few strange noises in the first couple of seconds ,
if that and only if the phones are connected, is surely preferable to 5V DC
pulses fed , delayed, into the PA.
So both will rise approximately together and so no 5V surge is possible.