[turretslug]

I do like your idea of optical isolation Hugo, I agree with the philosophy that an add-on buffer shouldn't degrade oscillator stability or affect calibration, and the opto- route would certainly be a good barrier to digital hash getting into the mixer and degrading reciprocal mixing and other drawbacks. That was my motivation for using a screen-grid valve with its very good anode-to-control grid isolation rather than semiconductor buffers with their relative reverse transparency. Possibly a FET/FET or FET/BJT cascode would improve on a single transistor initial buffer but I'm sceptical that it would be as effective as a pentode.

The other problem in my mind was excess additional heat- it's rare to find an oscillator box that isn't cramped, and/or hemmed in externally- "our radio is bigger than the competition" rarely appears as a USP! Even a B7g-size buffer is difficult to squeeze in, and even late-era subminiatures seem to be difficult to find with indirectly-heated cathodes needing less than a "hot watt" of typically 6.3V 0.15A. I believe that Nuvistors have something like a 70mA heater but these little valves get quite fiercely hot, which would make them difficult to place in a cramped and temperature-critical situation. Not to mention the fact that they are getting rarer and more expensive, and the holders even more so. The DF60/5678 struck me as an excellent candidate for this usage- there is a developed version with a different designation that takes 20mA at 1.25V filament power rather than 50mA, but it seems rather rarer and lacks the metallic screening coat.

I also like your point re. the simplicity and effectiveness of your valve oscillator pick-off approach- my thinking was that having an active device added within the set gives extra isolation consistency with sets that have a relatively highly-resolving scale up at 30MHz whether or not a DFM feed cable is connected. Again, aiming for something near to perfection!

Colin