As for the reality of the situation, answer me this:
1. How much power does your rig transmit?
2. How much power does your rig draw?
Correct me if the operation of dividing the first by the second does
not reveal an efficiency of roughly 40% and a power loss to heat of
roughly greater than that transmitted. Your rig has a massive heat
sink with a fan, n'est pas?

n'est pas? ? Anyway, the power lost inside the transmitter I thought
was due to the
inneficiencies of a class A or B amp configuration. That the amplifying
device (tube
or transistor) is metering out varying amounts of current from a
constant DC supply.
And the current has to pass thru the tube or transistor before it gets
to the load.
Assuming a sine wave at RF, zero crossing would be low wasted power, peak
also low wasted power, but at 0.707 peak max wasted power (class B). Class
C is more efficient except the power in the harmonics you get have to be
filtered
out, and you get a heated filter.

Now if the Thevenin impedance of the transmitter was 50 ohms, then half
of the remaining power you successfully converted to RF would heat the
transmitter some more. That 40% would become 20%. But if you design
the output right, making the Thevenin impedance very low, little RF power
is wasted in the transmitter (say 1%) and 39% is pumped out into space
(on the air)
by a matched antenna. Now that low Thevenin impedance will reflect
most of the reflected from the mismatched antenna power back to the antenna.
If the transmitter Thevenin impedance was 50 ohms, it would absorb the
reflections and get even hotter. Oh, this has its uses, like in a
signal coming
from a TTL line driver that passes thru a "source termination" resistor
of 100
ohms, then thru a 100 ohm impedance transmission line (like carefully
designed
traces on circuit boards in that GHz computer) and then to a single
destination
receiving TTL gate. which looks like a high impedance load. The reflected
signal gets absorbed by the source termination. A reason for not
destination
terminations is that this makes the source work harder pumping DC current
when the signal is a "high", vs essentially idling when just source
terminated.
No daisy chaining allowed, the signal looks like crap except at the very
end.