"Owen Duffy" wrote
The mechanism is not "that the reflected wave must travel back to
the PA anode and will be absorbed there causing overheating".
Take an example of the 50 ohms load discussed, and an electrical half
wave of 70 ohm line connected to a transmitter designed for a 50 ohm
load. The transmitter behaves exactly as if that line were 50 ohms.
Though there is a reflected travelling wave on the line, it does not
travel back to the PA anode where it is absorbed and converted to
heat.
____________

Really, the mechanism is there -- only the unique circumstance you describe
protects the PA from seeing reflected power in such cases. Other line
lengths in this scenario could stress PA components beyond their ratings.

And even if the PA saw no reflected power because of a fortunate length of
transmission line connecting it to a mismatched antenna/load, that reflected
power still exists in the transmission line, and may cause its failure.

Manufacturers of the rigid coaxial line used in broadcast stations (e.g.,
Dielectric) require derating its maximum power rating inversely by the value
of the SWR existing in it . A power derating factor related to SWR also
applies to Andrew Heliax® and RG-type coax line. Deliberately setting up,
or tolerating reflected power on a transmission line is not done without
risk.

RF