[mictester]

A little further information:

The original prototype was built "Manhattan"-style on a piece of copper-clad PCB material. I used about 4½" X 2½", and put the completed board into an aluminium box 6" X 4½" X 1½" that I had lying around for years. The audio socket(s) were phono type, and the RF out was through a BNC. I used a zener diode on the oscillator supply at first - 560Ω from the 12V supply to the top of a 5V6 zener worked quite well, but the temperature coefficient of the zener diode combined with the thermal drift of the oscillator to make things worse! The 78L05 was a quick fix.

My oscillator coil was initially an axial choke - it oscillated OK, but drifted badly, so I dug out an Amidon FT37-61 ring core, and calculated that I needed 29 turns for 47µH. I put mine a bit higher up the band, by winding 26 turns, and the 40p trimmer would allow me to tune from about 1500 - 1650 kHz. 32 turns put me down to roughly 650 - 800 kHz - as the American cousins would say "YMMV"!. I wound the coil with (roughly) 30swg enamelled copper wire that was actually from the secondary of a dud transformer. Once the turns were on, I held them - and the whole coil - in place with candle wax. The oscillator is surprisingly stable considering its simplicity.

It's fully modulated with a few hundred millivolts of audio - however, you can scale the gain of the modulator by changing the feedback resistor. A fairly crude (transistor and FET-based) audio limiter was scaled for the 0.775V nominal output, and this worked well with the little transmitter. I also did a little audio filtering before the modulator, rolling off the top end at 6.5 kHz, but introducing a (sort of) pre-emphasis HF lift from about 1 kHz upwards, which made the overall sound "brighter".

I was considering designing a PCB for this thing, but haven't found the time. Between my brother and myself, this circuit has been replicated about half-a-dozen times now. None of them are exactly this circuit - the component values aren't "set in stone" - I'd just use the first "approximately right" part that I found on my bench!

Improvements:

Is it worth spending much time on something as crude as this? The oscillator stability could be improved by careful selection of components or (better) by using a crystal. The output matching and filtering could be optimised, and a crude ATU could be constructed to load it into the "dangly bit of wire" more efficiently. You could polish this thing endlessly if you wanted to - the circuit shown is just the starting point!