In the last revision, our bridge rectifier shorted negative capacitor voltage. A resistor softened this blow, so at least the diodes survived. Half-wave rectification can sidestep this issue. Negative capacitor voltage is still applied to the transformer secondary, through the raw-supply diode. And the transformer secondary is rated for the voltage. But we want to avoid reverse power flow (from exiting through the transformer to the power line). Instead we want to reclaim negative capacitor charge, ideally recycling stored energy to recharge the capacitor positively. 10 Amp fast diodes and a "Ringback Inductor" were added to do just this.
Resonant capacitor polarity-reversal follows each SCR firing. So we initiate capacitor polarity reversal while the transformer secondary is maximally negative, thus the raw-supply diode never conducts. Reverse capacitor polarity then discharges through 10 Amp fast diodes into the Ringback Inductor, investing stored energy into that magnetic field. Capacitor voltage collapses to zero again, but the Ringback Inductor isn't done yet. Our recovered energy is now stored in that magnetic field, which then collapses. Inductor polarity reverses; collapsing field now charges the capacitor positively. Transformer output polarity then goes positive, which continues trickle-charging the capacitor until our next SCR firing.
My first attempt to recycle pulse energy was simpler, but it proved lossy, just like Rev A., initially I omitted the ringback inductor. My hope was to let resonant action take one full cycle, delivering both polarities to the applicator coil in rapid succession. My SCR didn't recover from the conductive state fast enough. Large hockey puck SCRs could have a 1 millisecond recovery time. Forward voltage appeared again after only 100 microseconds. The SCR conducted again, so ringing continued until stored energy got dissipated. Ringback inductance leaves capacitor polarity negative long enough for SCR recovery. In this case, about 700 microseconds seemed adequate.