Shunt regulator question

[WB3JOK]

My 4 x 4-125A amp is nearing completion (missing only a vacuum variable for the pi-net loading). I smoke-tested it very gingerly today and only let the magic smoke out of one varistor and an already-bad 4-125A that flashed over. That 3500V supply with 54 uf filter cap scares me  :oops:

I have a question about the shunt regulator I am using for the screen voltage. I am using a 4-65A (a sweep tube would not have sufficient plate dissipation under zero or negative amplifier screen current) with a string of six 100V zeners from the plate to grid, returned through a resistor to a zener-regulated -100V. HV supply is from the CT of the plate trans (bridge rectifier), 8 uf cap, resulting in about 1600V to the 15K 100W dropping resistor, which works out to approx. 70 ma.

The question is, what should I do with the screen of the 4-65A? I temporarily tied it to the plate but haven't measured the screen vs. plate currents and am worried about exceeding the screen rating. With the triode connection, the output is a very well regulated 540V, (and of course this is also the screen voltage). The grid therefore must be at 540 - 600 = -60V which sounds like a reasonable operating point.

Should I use a suitable resistor to lower the screen to (say) 350V? Since in theory the plate voltage is constant, as it is also the regulated output, there should not be large fluctuations in screen voltage with a simple dropping resistor.

thanks
Charles

[WB3JOK]

I measured the screen current of the 4-65A regulator tube. At 540V and zero load it was 15 ma which is within the maximum 10W screen dissipation rating, but unnecessarily high. A 22K dropping resistor reduced it to only 4 ma and had no noticeable effect on regulation.

I also found the curves for 4-65A and 4-125A online. Even for running class AB1 it looks like 540V on the screens might be too high for lowest distortion. So I jumpered one of the 100V zeners in the regulator string and now have 450V. (It doesn't change exactly 100V since the screen voltage is also being reduced and so the tube doesn't conduct as hard for a given grid bias).

Now to find $129 for the last vacuum cap and I can put this beast on the air  :evil:
Top
Bottom

[WA1GFZ]

Triode connection would be fine if you get good regulation. I would try to
run in a mode that reduces the grid to grid voltage the most.