Jim, I am not clear on what you said. Normally the grid choke would replace the grid leak resistor (usually 249K or 270K in Bottlehead power amps), from grid to ground. (Actually I mean from the grid stopper to ground; nothing but the stopper actually connects to the grid pin.) I suppose that, by analogy with how we describe power supply filters these days, you might properly call these C-R or C-L coupling rather than the traditional RC/LC terms.
There are two main technical effects of a grid choke:
1) Since it has inductance, it will resonate with the coupling capacitor. That resonance is damped mostly by the effective parallel resistance of the choke, which comes mostly from eddy-current losses. There is an argument for leaving the grid leak resistor in parallel with the grid choke so that a known minimum amount of damping is provided. This resonance means the bass response rolls off more steeply while remaining flat through the audio frequencies, so faster recovery from transient overloads is likely to occur.
2) At midband frequencies the impedance of the grid choke is higher than the resistor it replaced. This places a lighter load on the driver, possibly reducing its distortion. Leaving the resistor in parallel will eliminate that effect.
I have not done any measurement or listening to evaluate this tradeoff - which is why I wanted to know the capacitance that seems to work. If there are experiments with leaving the resistor in parallel, I would be most interested in the results as well!
With the longer chassis, and the DC heater supply, there may be a suitable location for the grid choke, near the driver and far from the power transformer.
