Originally by @thebugger. His version used a class-C RF amplifier, and I modified it to be class-E for maximal efficiency.
With a class-E instead of a class-C. I found that the 1mH inductor is plenty of filtering, and that a proper filter is actually quite unstable without a resistive load to soak up the oscillations. The class-E shown here is actually one of the circuits I’ve posted with slight modifications (slight rounding of the values). The 2.67nF capacitor is 2.2nF+470pF.
Edit: New MOSFET setup taking advantage of the fact that it will only ever have to provide current and never receive it. This new setup is complete with a driver stage that includes Schottky diodes to keep the fast changing voltage from going through the parasitic capacitance of the BJTs and charge the off one’s base causing both to turn on resulting in an unnecessary high through current.
Edit 2: I copied this over to a new schematic so that it could be made public instead of unlisted.
