On a sunny day (22 May 2006 10:29:39 -0700) it happened "Telstar Electronics"
wrote in
.com:

Your statement about having the switching freq much greater then the
output freq I agree with. But if you wanted to make an amp for 27MHz...
what switching freq would you have? 2X, 5X, 10X ?\

Look at that example link, it uses 100kHZ (KILO Hertz) in the PWM AM modulator,
leaved the RF stage as is, but the PWM modulates it supply voltage.
that way the output.

In case of direct pwm to the RF, yes, I mentioned 1nS switching time,
could be done.
I like the digital idea better.....


Also, with a high
switching rate... with square-waves... you have tremendous harmonic
content. You also would require much higher frequency amplifers...
along with the quantum leap in complexity of combining all those
amplifiers,
no no, MOSFETS can switch that fast....

A/D conversion,

Any PC gives 16 bit audio, just some shift registers or some i2c io expanders,
no need for an AD converter, we are going to drive digital now :-)!!!

etc... I think your idea would work... it's
just not practical in my mind.

Remember the modulation frequency is in the audio range, at that speed you
have to change the PWM.
You need a pi filter lowpass in ANY case to get good harmonic surpression.

In the digital case (swithing 8 amplifiers), there are some more advantages.

Problem is adjusting the exact power level (so a ramp (sawtooth) waveform has
equal steps), adding (summing) the output takes inductors couplers...
impedance matching.
Hey, I have done that DA thing for up to 10 MHz or so....
the big boys can do it, why not we?

Probably will be worth a fortune on ebay 50 years from now.