Tektronix SC502 scope brightness drifts

[CaptDon]

I have an SC502 15mhz analog scope (the modular series) and the brightness drifts. I noticed the comparator circuit made up of Q830 and Q820 is very temperature sensitive. I have tried freeze spray on all the different components in the circuit and none cause brightness drift EXCEPT the comparator transistor pair. I have physically bent them to touch each other and JB Welded them to each other. The reason I did that is heating one makes the brightness go up, heating the other makes the brightness go down so I tried to thermally bond them to neutralize the effect of temperature drift. It is better by a 'relative' eyeball quality factor of 50% less brightness drift, but it still has some drift. Anyone familiar with this anomaly? I had several SC502's in various workplaces and I don't recall this much drift.  Thanks for any leads!!

[ESTechnical]

Is the comparator just a long tailed pair? If the output is single-ended, is the load on the output perhaps higher than before leading to one transistor being too hot? Maybe the gain of one transistor has become very different to the other one? Is the supply voltage well regulated to this part of the scope circuit or is that drifting too?

[CaptDon]

both of the comparator bases are connected to +5 regulated via resistors and the commoned emitters are connected to +20 regulated via 7.68k. Q830 has its base driven from the intensity control which itself is a low voltage circuit, basically a pot connected from ground to +5 regulated, not high negative cathode voltage. Q820 is the other side of the comparator and its base is driven by the sweep unblanking signal, the chop signal when in dual trace chop mode, and external rear panel Z-Axis input which is unconnected. All voltages are correct and well regulated as is the high voltage. There is no blooming or defocusing and the C.R.T. is very sharp and bright as if it were new. The resistor values connected to the transistors are of high enough value that nothing is dissipating more than 10 to 20 milliwatts. When first turned on the trace is very bright, then it settles down to a 'normal' level during perhaps a 5 to 10 minute 'settling' period. I leave the intensity control alone knowing it will finally stabilize where I want it.