No buck. One half of the bridge is shifted in phase against the other, so the effective drive voltage ramps while the bus stays flat. There are fewer parts in the power path, but the timing is harder. Loneoceans' tinyQCW page explains the method. Three of the longest arcs on the wall use it.
APWM is the half-bridge version. The top and bottom switches run different, complementary duty cycles, so the drive ramps without a buck converter and without a full bridge. Anders Mikkelsen’s own list of reasons to use it is “simplicity, a reduced parts count, less gate drivers, less transistors”. He reports, as a surprise, that ZVS still holds, provided the phase lead is increased slightly as the duty cycle falls. A self-oscillating or PLL driver tracks that by itself. The frequency need only move on the order of 1/Q, “something like 10 % at most”. That is far less than with frequency modulation, the other way to ramp a half bridge. ZCS is lost, since one transistor now switches before the current crosses zero. The asymmetric conduction periods give the other transistor more of the conduction loss. On the SiC parts he was using, the two came out close to balanced.
A variant balances the two perfectly by alternating which transistor interrupts the load current on successive cycles. It applies the waveforms of alternate phase shift modulation, as used on full-bridge QCW coils, to a single bridge. He did not use it. At low modulation index it puts out a strong second subharmonic, and at zero output the bridge runs at half the operating frequency, which nobody has yet tried against a resonator. The implementation turned out to be cheap. The self-oscillating driver already has a ramp across its oscillation capacitor. Comparing the power ramp against that one ends the drive pulses early, and that comparison is the modulator. The ramp arrives over fibre as a PWM signal unrelated to bridge switching. The receiver turns it into both the analogue ramp and a START signal that resets the overcurrent latch once the ramp nears zero.
Sources
- loneoceans: tinyQCW (phase shift explained)
- highvoltageforum.net: Anders Mikkelsen on what APWM is, and why ZVS survives it (Nov 2024)
- highvoltageforum.net: the fibre receiver, the START signal and the modulator’s dead band (Oct 2025)
This is the card from the driver tree, written from what the builder published. Send corrections through the forum.