Allan Herriman wrote:

On Sat, 26 Feb 2005 04:53:03 +0000 (UTC), "Reg Edwards"
wrote:

The phrase "output impedance" in connection with amplifiers is ambiguous and
likely to result in arguments.

The correct description is "internal impedance" or "internal resistance" and
should always be used.

S22 is fairly well defined.

Just to bring back to the original discussion and reiterate:

S22 is a small signal (linear) parameter, by definition. It does not apply to
the large signal environment.

This idea is somewhat related to the idea that power amps should be tuned for
"maximum transfer of power," which is a small signal (s-param) issue, and
requires conjugate matching. The idea is incorrect because it ignores the
practical large signal non-linearity and *any* consideration of DC to RF
efficiency (which is prime for PA design). Linear parameters provide *no*
recognition of things like DC to signal power efficiency and therefore practical
issues like supply rails.

First order matching of an RF PA to a load involves transforming the load to the
optimum point on the AC load line (for example, more or less equal positive and
negative swing limits for class A). That's what "matching" is for an RF PA. It
makes no statement about actual "output impedance" of the source. What is said
is that "such and such RF PA will deliver X power into some specifed impedance
within some VSWR circle." That's all. The concept of output impedance begins
to break down for large signal devices.