The mechanical receiver is interesting.
It set me thinking how you could improve the frequency stability ( a
problem Jim mentioned in his write up). A fly wheel was one idea but, while
that should reduce some frequency shift, it would also introduce tuning lag
effects. I suppose you could make it complex and have a move the mass
towards the centre somehow when tuning and towards the circumference when
the required frequency is found but that is probably overly complicated.
May be just having a variable governor and using that to control the
spreed, relying on the mass to provide the angular momentum to maintain a
constant RPM.