Rock-Mite 80, further steps

The weak coupling from the oscillator to the buffer held the frequency comparably stable and nice, however, not much RF was radiated, my sigs made it into my neighbors RX and faintly some sort of trace.
Today, I increased coupling from 2p2 to 18p (orginally 47p), that again seems too strong, massive impact on the oscillator creating a really ugly but strong signal in my neighbor's grabber. Even the 2N2222's heat sink warms up now... Maybe reduce coupling again, to let's say maybe 10p ....

I further changes the capacitor terminating the crystal filter, from 33p to 22p, seems to have some impact too: