I saw G3YXM's low-HF design, and it looks promising for low frequency use, but he appears to have not gone that way and went for a switchmode version at LF/MF

Of course, I could just redesign that, or something similar, using suitable ferrite and a bit of feedback ,but it would be nice to avoid reinventing too many wheels.

Using switching MOSFETs as linear amps still doesn't feel quite right, but enough people seem have managed it now (including my own design for a low power one) that it looks a sound-enough route to follow up.    The one thing that does bother me about the technique is bias compensation with temperature.   It is really asking for active bias control, but in a class-B or -AB design, where bias just affects standing current, that gets more-than a bit tricky as a concept.

Andy  G4JNT