[BC] Tube and transmitter design

[richardbjohnson at comcast.net]

When Eimac was making the tubes in San Carlos, and if you ran the
filament voltage within five percent of their rating, typical tube-life for
a set of 4-400s was about a year if you just throw them away. If
you would swap the modulators with the finals at about three-month
intervals, you could get two or more years out of them. If you were
at a site where the voltage varied considerably, you should have
installed  a optional Solar regulator for the filaments. Doing so would
have extended the life even more.

Radio station WERI in Westerly, RI had (still has) a 20V3 with no
filament regulator. Once the station went "unattended," I did a proof
of performance three years in a row without ever changing either the
modulators or the finals. However, the station was a daytimer and
the filaments were deliberately left on 24 hours a day. I don't know
if the cost of electricity, when factored into the cost of replacement
tubes, was a worthwhile investment.

 >  I think the
 > argument about fidelity is somewhat misleading.  The Iron in a plate mod
 > transmitter is the primary thing that determines fidelity.
 >

The modulaton transformer design has an effect, of course.
However, the primary thing that determines fidelity, except
at the extremes of the audio frequency range, is the gain
inside the feedback loop. When you have only two audio
stages, RC coupled, the phase-shift doesn't include anything
nasty within that loop as would be the case of an interstage
transformer required of 833 rigs. This meant that 4-400
rigs could include 18 to 20 dB of local feedback. Most 833
rigs didn't use any feedback although the Gates rig had a
feedback ladder that looked as though it should do some-
thing. After it would burn up, most were removed by
engineers in the field.

The last of feedback meant that distortion was acceptable
with new tubes and got progressively worse as the tubes
aged. Also the third harmonic distortion that you get with
Class B circuits, that makes a sine-wave turn into a
peaky triangle could only be smoothed out by running at
high bias currents, i.e., poor modulator efficiency.