Limiting no-load output voltage of a tube amp power supply

I am designing a power supply for my tube amp. Here's what I have come up with:

Duncan's PSU designer shows that it should work OK (and provide 256V at 320mA). However, should one or both output tubes fail, the voltage would rise to 550V, which the other tubes may not like. Also, since the rectifier is directly heated, it will warm up much faster than the output tubes, probably resulting in too high voltage for a brief time after turn on.

So, I need to somehow limit the output voltage. However, I do not want to use any semiconductors anywhere in the amp. So far, I have come up with one way to do it and want your opinion on it or other suggestions:

simulate this circuit – Schematic created using CircuitLab

R3 is made up of 10x 100Ohm 10W resistors, R2 value calculated so the relay turns on when the voltage is too high and turns off when the voltage gets too low (output tubes warmed up etc).

This would probably oscillate if R2 is not calculated precisely enough or with line voltage variations. Also, it would result in an overload of the power supply when the relay is active and the output tubes start working.

Are there any better ways of doing this?

• « the voltage would rise to 550V, which the other tubes may not like». The maximum voltage at cut off is clearly indicated on tube's datasheets. 550V is not that high for vacuum tubes. Applying sudden HT on warm tubes is worst.
– greg
Commented Jun 12, 2015 at 15:27
• @greg, some tubes are specified at 500V (6N2P) or 250V (6AF6 magic eye). Applying sudden HT is bad, that's why I think I need a way to limit the voltage (which would happen with the relay). Also, what if the heaters in the output tubes fail? Then again, maybe I am worried about nothing... Commented Jun 12, 2015 at 21:02
• Yup, this is a shunt regulator. You can find imho better schematics using silicium.
– greg
Commented Jun 15, 2015 at 13:13