Nothing about being clever enough to do it, if a thing ain't possible it ain't possible! Almost certainly this one isn't, as you are trying to balance two different things at the same time. You can't, without increasing the AC impedance in the 'balance winding' circuitry over and above the DC resistance, and that implies adding a choke. And if you're going to add a choke, then you might as well use it for smoothing...
However, drawing 25% of the current through the balance winding and finding an improvement is a bit surprising. It's possible that ultrasonic oscillations exist (driving the valve highly non-linear) but drawing current and putting a load on the balance winding is applying sufficient damping to stop it, so you get a sudden big increase in quality. Terry's suggestion of grid stoppers is a good one: for maximum effect, the grid stopper resistor needs to be connected closely to the grid pin (10mm max of lead) with the other lead of the resistor being as long as you want (within reason). This won't hurt things, but can help. It's the work of a few moments to add 47kΩ or so of resistance directly in series with each of the two grids.
Looking forward to the next instalment!
However, drawing 25% of the current through the balance winding and finding an improvement is a bit surprising. It's possible that ultrasonic oscillations exist (driving the valve highly non-linear) but drawing current and putting a load on the balance winding is applying sufficient damping to stop it, so you get a sudden big increase in quality. Terry's suggestion of grid stoppers is a good one: for maximum effect, the grid stopper resistor needs to be connected closely to the grid pin (10mm max of lead) with the other lead of the resistor being as long as you want (within reason). This won't hurt things, but can help. It's the work of a few moments to add 47kΩ or so of resistance directly in series with each of the two grids.
Looking forward to the next instalment!
