On Sat, 4 Sep 2004 06:23:28 -0500, "Richard Fry"
wrote:

|"Ian White, G3SEK wrote
|The meter measures nothing that involves the source, except
|the level of RF that it supplies. It does not respond in any way
|whatever to the source impedance.
|_____________
|
|Not that I said it did in my part of the thread, but nevertheless the above
|statement is not strictly true. In the case where the source Z of the tx PA
|does not match its load Z (which is typical), power reflected from the load
|mismatch will at least partly be re-reflected from the PA -- which then
|contributes to the power sensed by a "wattmeter" in the output path.

If you have a directional wattmeter that *perfectly* senses and
displays the ratio of forward to reverse power, what difference does
the absolute power make? Hint: It doesn't.

But let's avoid going any further here. This has been argued about in
at least 10,000 other threads and has absolutely no bearing on Ian's
or my statements regarding the unimportance of source Z in the
discussion of measuring reflection coefficient.

In a laboratory environment, the source is usually well matched at the
system impedance. This is for convenience and improves accuracy only
because is takes less mathematical horsepower to remove the effects of
the "re-reflection" you speak of. This is particularly true when
highly mismatched loads are measured; and two of the most highly
mismatched loads are an open and a short, both of which are often used
as calibration standards.

Re-reflection exists and is actually an undesired thing in laboratory
measurements, primarily because most measurements are swept frequency
and the changing phase of the reflections is a pain in the ass.
Reflections are a source of *measurement error*. Removing the effects
is called *error correction* in the laboratory setting.

But I can tell you that if I calibrate a high-quality network analyzer
using a 50 ohm generator and then add a tee and second 50 ohm load to
the generator output, making it a 2:1 mismatched source, there will be
*no* significant change in the answer presented when typical
mismatches are measured.

Any differences represent measurement *error*, not measurement
*reality*. Reality is the interaction of the line and load and the
source plays *no* role.