I have decided to shelve the Propeller Beacon project. Actually the project made the decision for itself. Bored with receiving spots for my 250mW WSPR signal on 30m I decided to try 20m where there are almost as many monitors. But after a handful of spots there was nothing. After a while I decided something must be wrong.
I checked that the beacon was still transmitting at the right times and it was. So I switched on the K3, ran the WSPR software and the reason for the lack of spots became clear. My signal was drifting the best part of 10Hz in each two minute cycle!
When I first ran the beacon on the bare Propeller board I had found the signal to be pretty stable after only a few minutes warm-up. After adding the LCD board - which plugs on top of the Propeller board - I noticed that my 30m transmissions were being quite consistently reported with -3Hz drift. The RF amplifier board with its heatsink is separate from the Propeller board and nowhere near its crystal oscillator. I hypothesized that the LCD board is trapping some heat in, enough to make the master clock oscillator drift.
The drift is probably a multiplication factor of the clock frequency so the higher in frequency you go, the worse the drift. Whatever the explanation, the Propeller + LCD combination is not usable as a WSPR beacon as it is.
It seems to me that there are two possible solutions. One would be to make my own Propeller processor board with a temperature controlled crystal clock. The other would be to use the Prop to control a Si570 synthesizer or something like that. Unfortunately I don't think either of those solutions are within my capabilities just at the moment. So I'm afraid the beacon project will be going on the shelf.