Richard Fry wrote:
"If the source resistance of a tuned r-f PA stage was truly
non-dissipative, and the tx simply supplied less power into poor
matches, how would that explain the catastropic failures to the output
circuit components often seen when high power transmitters operate
without suitable SWR protection into highly mismatched loads?"

The PA is a switch. Almost no voltage across it when it is closed and no
current through it when it is open. Some of its impedance is dissipative
and some is non-dissipative.

A conjugate match to its total impedance is the way to deliver maximum
power from the transmitter to its load. Alexander H. Wing wrote on page
43 of "Thansmission Lines, Antennas, and Wave Guides":
"If a dissipationless network is insrted between a constant voltage
generator of internal impedance Zg1 and a load ZR such that maximum
power is delivered to the load, at every pair of terminals the
impedances looking in opposite directions are conjugates of each other."

An operating transmitter is normally adjusted for conjugate match with
its load.

Normal plate dissipation occurs when electrons strike the anode and
there is little damage to the tube when the current and cooling are
within limits.

Let an arc strike across the transmission line and it may effectively
become a short circuit which may impose an enormous mismatch in an
instant to the transmitter. That`s why a d-c supply is often connected
in series with a relay coil across the transmission line. The arc
completes the d-c circuit energizing the relay which breaks the
interlock circuit. The transmitter instantly is shut down until it is
manually restarted.

Tubes are often destroyed by internal arcs if overloads don`t act in
time.

Best regards, Richard Harrison, KB5WZI