Total posts : 45366
The technique Neil described is called Progressive AM Modulation and is fascinating because it both increases the overall efficiency of the transmitter and allows peak power to be many times higher than achievable by standard AM modulation. If you are inclined to wade through patent-speak, Harris Corp. has a patent that describes the workings, patent #3898590. http://www.freepatentsonline.com/3898590.pdf
Power input to the final RF stage is defined as the DC voltage times the DC current. In normal AM modulation, the power is constant. It is the same through the modulation range of 0% to 100%. At 100% modulation, the average power is the same as it is at 0% modulation.
Things are different with progressive AM modulation because the power supply voltage to the final RF stage is increased in accordance with the peak level of the input audio signal, so the peak power can increase above the normally accepted 100% modulation limit of 4 times the 0% modulation level, and the resulting average power level increases.
There is no theoretical limit to the maximum power attainable with this scheme. A 100mW transmitter, measured with no audio input, could produce an average power of thousands of watts at maximum audio input level.
In a viable part 15 transmitter design, progressive AM modulation could be achieved by detecting the peak audio level, buffering it as required, and feeding the result to one of the many available voltage regulator ICs. The resulting voltage to the final RF stage would then increase proportionally to the peak audio input.
Legality is not likely in the end. When a manufacturer is going through the certification process, there is a chance that the certification lab will just measure the power input at 0% modulation and it will be certified, or maybe the lab will notice the trick. If the FCC happens by with their equipment, they may judge you as legal if you happen to have a dead carrier, or more likely, they will measure your field strength over the limit when you have program material playing.
As Neil said, I also recall a part 15 transmitter circuit that used this technique and produced an average power of 300 mW (if I remember correctly) at high modulation levels.