There have been a couple of false starts here. First I ambitiously tried to make a pulse width modulated transmitter but couldn't get it working. Next I tried high level AM modulation using an audio transformer but I think it saturates and I could get only very small modulation with it.
Now I'm trying low level modulation of the IRF510 power stage. This looks distorted but seems to be working pretty well.
The RF board starts with parallel 7.122 crystals pulled as close to 7.125 as I can get, a buffer stage, a driver, an IRF510 final and a low pass filter.
Along the way, the bias on the IRF510 got out of control making a big bang and releasing its smoke. I now use a current meter at all times so I can spot thermal runaway before catastrophe strikes.
Happily they are cheap.
Lots of work still to do and I'm very much looking forward to having a contact with this tangled mess.