"And now for something completely different..."

[Bob Bruhns]

> why didn't they AM the signal when it was "small"
> and easy to do and then just use a linear final
> so all the muscle went into the signal instead
> of heating the modulation transformer?

There was a reason.  A linear amplifier is most efficient at maximum power, and the efficiency decreases for lower power, reaching
zero at zero output.  Theoretical maximum linear amplifier efficiency is PI / 4 (about 78.5%), but in the real world it is about
60% to 65%.  An AM signal is amplitude modulated and varies from carrier to peak to zero, and the carrier power is between 1/5 and
1/4 of the peak power because AM broadcast regulations allow some waveform asymmetry, up to 125% positive modulation if I recall
correctly.  As a result, carrier efficiency using a simple linear amplifier is not good, about 27% to a maximum of 32%. Even
counting the whole station, including the filament and plate power of the modulator, a plate modulated transmitter has
significantly higher efficiency, and the difference was enough to tip the financial scales.

In the old days, there were a few other alternatives to plate modulation that did get reasonably good efficiency.  One was called
AmpliPhase, it belonged to RCA and it combined two oppositely phase-modulated output stages.  Another was the Terman-Woodyard
amplifier, it combined two grid-modulated amplifiers in a special way so that one worked from carrier level down to zero, and the
other supplemented the first one in such a way that the first one then generated more power during the peaks as well.  Terman
designed a linear amplifier that worked on similar principles.  Then there was something called the Taylor system, that combined
two linear amplifiers such that one carried the carrier to zero power levels, and the other carried the peaks alone.  Ampli-Phase
was commercially viable, and some stations used it.  The inherent linearity of the Terman-Woodyard and Taylor systems were not so
good, but I understand that with enough negative feedback, they were OK.

The problem with those systems was complexity, and adjustment was difficult, and multi-frequency operation was tricky. That was an
issue with the old Conelrad requirements for stations to be able to transmit on 640 KHz or 1240 KHz, and also on shortwave where a
transmitter might be expected to change bands a few times a day.  So plate modulation and a class-B modulator were the usual
technology for many years.

  Bob, WA3WDR