trobbins
Super Member
As I see it, the worst-case time is when one anode of the 83 is conducting, and the winding tap connecting through to the other non-conducting anode of the 83 reaches a peak voltage of circa -530V. The voltage at that non-conducting 83 anode would depend on the voltage drop across the 6AU4GT in that leg, which is ill-defined as it depends on the leakage current through that 6AU4GT's reverse biased anode-to-cathode, and the leakage current from the 6AU4GT cathode-to-heater, and the leakage current through the 83 reverse biased anode-to-cathode. If the 6AU4GT anode-to-cathode leakage current was significantly more than the other two leakage paths then the 6AU4GT cathode could be approaching -530V, and so that 6AU4GT heater could be positive with respect to its cathode by up to 530Vpk. Leakage current versus voltage characteristics for reverse-biased diodes, and heater-cathode junctions, are not well-defined in that sample to sample variations could be huge and could vary with age of the valves, so imho it is problem as it is so ill-defined.
That said, I haven't seen any reverse-biased leakage characteristics of mercury vapour diodes, but could expect them to have a relatively high leakage when compared to a good hard-vacuum diode, in which case the reverse voltage applied to the non-conducting leg of the full-wave rectifier would mostly be applied across the 6AU4GT, and hence its cathode would be likely close to +530V and hence not a risk.
