The class-C amplifier in question used 6L6s IIRC. They where driven by a transformer circuit that I never quite figured out at the time. The output was a transformer that sent a nominal 100V signal to where the speakers were. The design seemed to be optimized to use as few tubes as you can to make as much noise as you can. I doubt that much time was spent on sound quality.
[....]Yes it is a nice idea.
These days, I have to use MOS parts in the output and the RF power is not for radiation. The biggest issue I had was making the RF have as little near carrier noise as I could without making things run too hot. Class C with a very strong gate drive and a very quiet power supply turned out to work the best. With bipolars it seemed that a large value of bias resistance and running the part short of bottoming worked the best. The noise that matters is from about 0.01Hz to
500KHz offset from the carrier. I can't trust the high Q to clean it up because I just can't seem to get inductors with a Q over a billion or two.