When the grid goes positive to the cathode, its impedance drops and grid current flows...simple enough.

What about AC filaments? the whole thing isn't at the same potential, and it is changing. Part of the filament can go negative to the grid at any given time. How much of the cathode/filament can be negative to the grid before the grid impedance starts to change?

Like the 20vac Russian filaments? seems like there can be a large area of filament negative to the grid( which is unipotential ) at any given time...
regards,
Douglas