Building experimental transmitters based on rare parts may be a risky adventure and you must keep alternative solutions in your design to solve potential future problems.
For instance I have a QB3/200 running at 1.2kV in my TX. May I switch to a 829B running at 900V or to a QE8/200 running at 600V in a fraction of a hour?
The answer is Yes... Why?
I have a multi-impedance modulation transformer and a multivoltage power supply, that can provide the full power on any of the outputs or a combination of them.
Here we use the last schematic of this paper:
http://www.ing.unitn.it/~fontana/Comparing%20Half.pdfMost interesting are full wave multipliers, that (starting from 230V AC) can provide any DC voltage stepped by 300V in any combination.
You will notice that all half wave multipliers (except the lowest voltage in the first schematic) are capacitive coupled to the power transformer. This solution avoids DC current flowing through the power transformer (DC current must be zero in commercial power transformers).
For half wave multipliers it is possible to see that starting from 230V AC it is not possible to produce 300V without any DC current in the transformer (first schematic). The soultion is to double 115V AC (last schematic).
In the last two schematics a transformer with 115-230V secondary is used (230V center tapped secondary).
Remember that with capacitive coupling you can also add negative voltages to the chain, but you must accept a "doubler".
With abundant capacitance provided by common PC power supply capacitors (330 - 660 uF), you could take a couple of amper from any output, if the transformer can provide full power on 115V and 230V output. You do not need to know in advance where you put the load.
This solution has solved all my problems, including the commercial source for the transformer...
Giorgio