[BC] How would you build it?

[robertm]

 On Jul 31, 2014, at 9:38 AM, richardbrianjohnson at verizon.net wrote:
> 
> 
> "Modern" methods of generating an AM signal with a DSP and then amplifying it result is very low overall efficiency --scratch that.

That's not how it is done. The DSP generates a modulation control signal such as PDM or the on off control for PSM. Efficiencies AC to RF exceed 90%. 
> 
> I once proposed an AM transmitter that, using digital techniques and high-efficiency power-generators, to produce the following on-carrier RF __power__ supplies:
> 
> 1, 2, 4, 8 ,16, 32, 64, and 128 __volts__. these would be digitally-switched to produce a 8-bit modulated AM carrier. That would produce a theoretical distortion of 1/ 255 =  0.3 percent.
> 
> Of course, this was back in the '70s so it was immediately poo-pooed as being a pipe dream of someone who just discovered digital signal generation -- back when Harris came out with PSM and a conventional (for the time) RF stage.
> 
> I understand that someone is doing that now, but I don't know who. It would be a good method.

A lot of people do. Harris ( GatesAir) still does, Continental does with high power shortwave, Thompson does. They inherited the ABB designs which are very similar to Continental and totally DSP controlled. HCJB used this before any of the above names. I am sure I missed a few from the list. The only difference with some of the approaches is that with solid state RF modules are switched and not power supplies.