This is the basic spark gap transmitter. If you consider your train of damped oscillations, they can be thought of as a CW with some sort of exponentially decaying modulation. So in the output spectrum there will be the carrier frequency with sidebands that are multiples of the modulation frequency. Because the frequency of the modulation is so close to the carrier they will not be able to be filtered out.
Frank
Thanks Frank, I had it like this in my mind indeed. It is exactly like this, but there is also a gap (big one) between the carrier decaying pulses, as the Q of the LC can never be that high to sustain oscillations until the next (1KHz) modulating pulse comes in.
The diagram of the previous post helps visualizing it.
Here is an interesting experiment I did yesterday:
I made a tuned circuit for LW. Then I connected a crystal (one of these vintage higher power types) to one end of the tuned circuit and took the output from the crystal to the FFT analyzer.
All harmonics of the fundamental were disappeared! The crystal does a very good job in filtering.
Also, since the single tuned circuit can never have such a high Q as the crystal and it's bandwidth is much higher, even slight miss tuning (out of the crystal frequency) does not play a major role (if it is slight) and it seems the main frequency is determined by the crystal filter and it is stable of course.
The way the crystal filters reminds me this circuit
**broken link removed** but of course only the tuned LC is before the crystal.
However I doubt if the crystal can filter the so close to the "carrier" signals you refer to. A single crystal bilter is very narrow (a few 10s of Hz) but maybe not enough for these so close to the carrier signals.