Perhaps you would like to join the small band of LF experimenters and apply for a special experimental licence. There is no problem in building a CW transmitter for LF transmission. It can be built much as one would build a transmitter for 1.8 or 3.5 NMz, except that you will need somewhat larger inductors.,
The biggest problem is achieving some sort of radiation efficiency in the antenna. A quarter wave Marconi antenna at 200KHz would have to be nearly 375 metres long, hardly the thing which would fit in the average back yard. Let's assume we hang up a long length of wire and achieve an effective height of 30 metres. This is 0.02 wavelength which would give us a calculated radiation resistance of 0.6 ohm. To resonate the antenna we need a series loading inductance with a reactance in the order of thousands of ohms. If the inductive reactance is 1000 ohms (800 microhenries) and the Q equal to 100, the loss resistance in the coil is equal to 10 ohms. It is not unreasonable to assume an earth loss resistance also of 10 ohms making a total loss resistance of 20 ohms. (Refer sketch). Antenna efficiency is therefore 100 x 0.6 / (0.6 + 20) percent or 2.9%. For 100 watts of power, we only radiate 3 watts.
We could reduce the loss resistance and increase the efficiency a little by improving the earth counterpoise system and raising the Q of the inductor. A high Q inductor of 800 microhenries or higher can be quite a massive unit. The John Adcock loading coil shown in July 1984 issue of Amateur Radio is 0.5 metre in diameter. Of course the way to really raise the efficiency is to increase the radiation resistance by increasing the length and height of the antenna, if that is possible.
Another thing to watch in these highly inductive loaded systems is the high voltage developed across the inductan ce. An antenna circuit of 20 ohms total resistance, loaded with say 2000 ohms of reactance, has a loaded Q equal to 100. A power of only 100 watts fed into this develops a potential of nearly 4500 volts. If you put your hand too near the coil, some nice RF arcs can be drawn.