This afternoon, after a break of a few weeks, I went back to my 10W 472kHz transverter design. The intention was to simply build a second sample to prove repeatability. As is often the case, things didn't quite work out that way: instead I discovered an MF oscillation when drive was removed in some situations. Despite trying all the usual tricks so far, I've not managed to fix it yet. It is not the PA: driving it and the squarer stage directly with a 472kHz signal all is well. It looks like a problem around the oscillator and mixer but so far I have not managed to tame it. Also, and this is probably the same issue, on RX there is a wideband noise resulting in desensitisation. For this evening I have stopped work and will take a fresh look tomorrow. Maybe the double balanced mixer in my original 2010 circuit was not such a bad idea after all.
This is a classic situation when a spectrum analyser would really help as I could see in an instant where the trouble was by poking around with an RF probe and observing the spectrum on the analyser.