[G0HZU_JMR]

I'm afraid I know next to nothing about valves (especially pentodes!) but it might be worth trying to work out what the DC startup conditions should be for your CV466 based BFO. I can try and kick this analysis off but I'm really a novice with valves.

At startup I presume the heater is already hot and the oscillator circuit is selected via the relay connection to the HT. This will instantly connect 100V to the top of R22 (150k) and R23 (220k). I presume that the R22 connection goes to the g2 connection (screen grid?) and this has a startup time constant of 150k + 1nF (C48) which means it will take over 150us for the screen grid to rise to a reasonable voltage.

During the start of this rise time at the screen grid I think that little anode current will flow and so I think there should be 100V at the anode of the pentode for maybe the first 20us after the relay clicks on. Then I'd expect to see the anode current begin to flow so the anode voltage will start to rapidly fall as the screen grid voltage at C48 starts to rise higher. As the anode current starts to increase then the DC voltage seen at the cathode and the grid should also start to climb because of the increasing anode current through R30 (10k).

I think the next bit is important: At some point during this risetime process the valve will cross through a region where it will have adequate anode voltage and current to generate a reasonable transconductance (gm) and this will be sufficient to start up the oscillation. I suspect that the time constant of 150k and 1nF at the screen grid will deliberately permit the anode current to rise fairly gracefully over the first 200us after startup. Hopefully this will allow sufficient time to begin oscillation at the sweet spot bias condition. However, if the 1nF cap C48 were to fail then the anode current would rise very quickly at power on and the valve might latch into a 'hard switched on' condition where it might not start as easily.

So I would check the 1nF cap C48. If you have a 2 channel scope you could watch the screen grid voltage and anode voltage during the first 200us after startup although I think you would need a storage scope or a modern digital scope to do this. Looking at your test point readings I'm not sure why the grid voltage (18V) is 5V higher than the cathode voltage (13V) in your diagram. Does this mean there is a (50uA) leakage path from the screen grid or the anode? Is it normal to see a 50uA leakage from the screen grid to the grid like this? It could be that the valve has a problem or maybe this is what happens when the valve latches into the on state with a very low anode voltage and no oscillation. I'm really just guessing.

You could try giving it a 'super graceful' power on condition by increasing C48 to maybe 3.3nF as this would allow the valve more time to cross the sweet spot zone where it can generate enough transconductance to start up. Once it starts it probably goes into a large signal condition where the valve will virtually switch on and off the anode current at 466kHz.

All the above is guesswork as I know very little about pentodes but it might kick off some conversation and something helpful might result from it...