After a few years of full immersion in construction of valves based transmitters for Shortwaves, I had accumulated a few components that might well be used for some experiments with solid state.
I have built a second SPY transmitter, with frequency range from 3 to 9 MHz, capable of delivering 20 W at 16 VDC supply. This radio has been a nice training ground on the importance of GOOD earthing when working with high frequency and high amplification devices. It includes also an 'L' network antenna tuner.
Despite the very 'crude' layout, ... it works. Only a relay is used: directly driven by the key, it makes the antenna switching and turns on/off on dot and dashes the Pierce oscillator (BF245A) and the driver (2N3553) together. The PA mosfet fed by the power supply continuously and a suitable resistive voltage divider determines its stand-by idle current by setting a proper Gate-to-Source voltage.
T1 is a 4:1 impedance ratio step down transformer, 6:3 turns on binocular 'VHF' type core. T2 is 1:4 impedance ratio step up transformer , 2:2 turns auto-transformer connected. The core is also binocular and it's a 'BN43-3312', type '43' mixture. You should play around a bit with the turns numbers, as they haven't been optimized. An idle Drain current of 50-100 mA, corresponding to a Gate-to-Source voltage in the range of 4 to 5 VDC approximately, should be set by the 470 ohm potentiometer. By doing this, the transmitter CW emission results extremely clean. Many other mosfet types could be used. IRF510 should perform better at same supply voltage as it has lower input capacitance. This circuit is not critical, then resulting open to easy experiments and adaptations with other semiconductors.
The impedance matching network is provided by two switches: the first allows to connect the variable capacitor up or downstream to the coils for better matching. The second switch puts in series L2 to L1 in case of short antennas (down to a few meters), as loading coil. L1 and L2 have same turns and diameter. L1 is tapped by an 11 poistions rotary switch in my prototype.
The construction of a chain made by an oscillator and some cascaded wide band high frequency amplifiers, is very hard to have self oscillations free: i.e., without power in the antenna with no Crystal in its holder !
Recently I've upgraded from a veroboard circuit support (see pictures below) to an unetched copper clad fiberglass plate and resoldered all the components 'manhattan style'. By doing this the transmitter has become much more stable, as I have reduced greatly all the parasitic feedback across stages due to relatively high impedance of the common paths to ground.
I have been very lucky to put the hands on a huge load of FT-243 crystals, from 3 to 8.6 MHz, at very low cost. And I am using them for some nice, peaceful QSO on CW with some friends in some remote and solitary portions of the Shortwaves. Only 3 or 4 watts, at reduced voltage DC supply, and a piece of wire hanging out of the window to the first tree ...
Radioman, Nov. 2014