While analyzing a problem with a small (150Kwatt oscillator) circuit, we discovered a faulty 200 pf 30VK vacuum cap.
This was found via a series of HI-POT testing, starting with the filament transformer (250 amps) which acts as the tube heater and tap for the cathode (-) voltage ( the + is grounded in this circuit)
First intentions were to determine if the transformer had a leaky problem due to higher than normal current in the tube and also to determine reasons why the resistor string for the High voltage return was getting annihilated at random times.
Any way, it appears that the Oscillator has been injected very high levels of sporadic bias on the grid, causing the circuit to generate excessive voltages but in narrow duty cycles. short enough to where the digital monitors and filtering of the circuit would null the effect but still at the resistors of course, hammering them.
It took a few times to determine this random effect.
It seems while you are hi-poting this capacitor and trying to find where the discharge is coming from, it heals itself and you can then put the full 30KV to it however, let it sit around and it goes back to breaking down starting at around 4Kv and slowly working it's way back to full level again as you do the retries with the HI_POT unit. We also notice this, the cap is behaving like a polarized neon bulb, in other words, it appears like there is ionized gas discharging randomly and the short isn't a low ohm short, it maybe no more than a1mA but it's enough to inject noise onto the grid of the tube. Also when I say polarized, I mean if I switch the polarity in the test it then seems to work, but that isn't how it's used in the circuit of course..
This cap is part of a parasitic circuit for the grid and has a resistor in series with it to the - side of the supply.
Another note I took is that if you look at the copper tube plates inside the glass they look like the surface of the moon and the envelope at the ends also looks like some deposits maybe collecting on them inside (white powderish)
This cap has been living in this circuit for many YEARS (like 35 or more) Is it possible the cap has gotten gassy from vaporization from arcs due to other failures over the years and now contains unwanted properties inside?
I remember replacing some vacuum caps in an amp I had years ago only because it seemed like people were hearing small popping noises in my signal, it was the caps breaking down but they were not actually doing anything visual damaging, does sort of sound like the same thing here.