Q714 is currently a Tektronix transistor marked 151-0606-00 referenced as Motorola SJE375 but I can't find any info on this device. The transistor that used to be there (broken) has no identification at all. I tried several MJE800 Darlingtons as listed in the manual but they fail because of rush-in current at startup. Can I add a small power resistor at the collector of Q714 to prevent this? Under normal operation, the current through Q714 seems to be about 1Amp (see below) so the MJE800 should work (4Amp rating)
If I set the cathode voltage at about -5000V (or a more positive value, max -4700), the device works find and oscillation doesn't stop anymore. The voltage drop across R714 in this mode is 1V or less (so 1 Amp current) and stays there. At -5000V, the image is clear but intensity control is limited. At -5500V, the specified value, the oscillation circuit eventually stops, after almost 30 minutes.
Still using 1R for R715 because my experiments show that a higher value here increases the oscillator run-time (limits current flow through Q714 is my guess, thus postponing the thermal runaway). Once I install the faster diodes (UF4004), I'll try to original 0.5 Ohm R715 again.
You say "The output will be shifted up if there is 1-ohm". I assume you are referring to U740 pin 6. Even at a limited cathode voltage of -5000V, the voltage here starts at 2V (with 1 Ohm R715).
Note that the screen image is crystal clear with -5000V cathode voltage but I would like to completely fix this device.
Should have more results in about a week.