I've been blessed!

[old_tv_nut]

Changing the HV regulator had no effect.

I have determined that the delay line output (which feeds the high side of the brightness pot) doesn't drift, but the negative voltage for the low end of the pot as well as the wiper voltage does.

Now to do a sanity check on how much the drift is at the wiper vs the low end. The wiper should be a voltage-divided version of the negative supply drift. If it's larger, there's some variable load on the wiper.

If the drift is coming from drifting of the the negative bias generated by the horizontal output, it's going to be a problem to get at because all those components are under the chassis

[Penthode]

Is the HV not steady? I would expect the raster size to change with changing the brightness setting.

There are various clues to rule out the culprit. A major clue is the time the process begins afterswitch on and how log it takes to complete the issue.

My rule of thumb is Vacuum tubes heat the fastest and show complete final operation point is about three minutes. This suggests we can rule out the 12BY7. Next resistor heat from 3 to 10 minutes. Lastly capacitor complete heating from about 10 to 30 minutes. That is unless the capacitor is close to a vacuum tube or resistor. I have seen old paper and even film capacitors begin to increase leaking immediately after power up in a thermal runawy effect.

That said, the problem occurring over a 10 minute period suggests the bias voltages around the 12BY7 are driffting. This could include the plate supply, the screen supply as well as the grid bias.

You have determined the 12BY7 plate voltage is rising which suggests either the screen voltage is dropping, the cathode voltage is increasing or the control grid bias is dropping. I think careful monitoring and tabling the voltages on a timeline from cold start up to 15 minutes on will reveal what is going on.