Quote:
Originally Posted by Craig Sawyers
You are looking for a capacitor that is still a capacitor at 40MHz  ie that is selfresonant frequency is greater than 40MHz. That means that its inductance has to be less than 16nH, including the leads that connect it to the cathode/screen/heaters. That is a tough ask...

I'd mildly disagree with that, actually!
Yes, at 40MHz, resonance occurs with 16nH and your 1,000pF. But who cares about 16nHworth of resonance? Even with 0.5uH stray inductance, at 40MHz you only have 120 ohms of impedance. And that's not going to be enough to cause instability. In the cathode, it might cause a bit of degeneration; on the screengrid, with a typical mutual conductance from grid to screen of 0.5mA/V, you'll have a signal on the screen of 1/20 of what you have on the grid. Hardly worth bothering about, even if it is phaseshifted the 'wrong way!'
Put it another way, suppose you have a superlowinductance 1,000pF capacitor with 1nF stray inductance. (Note that the SRF is 160MHz). At our frequency of interest, 40MHz, the impedance is 3.73 ohms. Now suppose overnight, a little elf comes along and swaps the dielectric in the capacitor for some ultrahigh permittivity stuff, so capacitance shoots up to 10F but nothing else changes. Impedance now drops to 0.25ohms, so better 40MHz decoupling than before, even though SRF is now only 1.6kHz!
Ultimately, the reason for lowish values of capacitance is that bigger values tend to be larger, so have MORE stray inductance. So, the initial choice could well have been based on that. Modern polyester caps are much better (the elves have started work!) so something of the same physical size is likely to be your best choice, and the extra capacitance is a bonus. Who knows, the original designers may have hidden some crafty neutralisation by careful component placement, to give most favourable response curve with what they had at the time!