Idea Box: Simplest Possible Transmitter Construction.

Here are some thoughts on the simplest possible transmitter designs.
How well (or not) they work will have to be tested some time. Simple designs are sometimes too simple.
Designed for CW only, single frequency.
A single canned crystal oscillator generally available for some amateur radio bands. Examples:
3.579MHz
3.686MHz
1.843MHz
14.318MHz
28.322MHz
I also have some on 14.296, 14.300MHz and 21.175MHz, but I am not sure those are easy to find. The first 5 are, sometimes found in old computer boards.
I have another one on 28.321, not sure where that one came from.
In a surplus shop in The Hague I found one on 28.339MHz.

The absolutley simplest design consists of simply connecting the canned oscillator output to 2 wires, acting as an antenna, and see if any range at all can be seen. The CW keying is simply done by keying the (5V) DC voltage of the oscillator.

Especially on the higher frequencies I would the resulting signal to have chirp, and maybe clicks, when keying this way, but maybe, just maybe the lower (1843 and 3579 kHz) will not have an excessive amount of chirp.

This is a simple enough experiment to make.

The output impedance of this type of oscillator is probably not correct for a 50 ohm antenna, but a simple impedance transformer made with a toroid core could be made, when used for testing it should have a fair amount of taps in the winding, but then, a 2 component transmitter could be fun, just trying to make a QSO.

The other part is the waveform from the oscillator. If it is a computer part, I would expect to see square waves, so harmonics should be filtered out. With the extremely low power, a quick experiment could probably be done, if a local station is available, but if more experiments are done, a simple low pass filter will have to be made. Also, I expect a decoupling capacitor for the power supply will be advantageous.

I have been talking to a local ham (distance 3km/2mi), about making some tests with extreme low power, so this could be one way to try this out.

I suspect that for a decent sounding signal a bit more effort has to be made, but maybe, just maybe, the component count can be kept down to 10.

Anyone out there having tried this ? If so, what is your simplest transmitter tested for a QSO?