Actually, several people (W8JI among them) have measured the output
impedance of common amateur linear amplifiers by at least a couple of
methods. The most credible measurements show, interestingly, a value
very close to 50 ohms when the amplifier is adjusted for normal operation.

[sotto voce] "and yet it moves" - updated to

Of course, it doesn't really matter, but people continue to make a big
deal out of it.

Roy Lewallen, W7EL


Hi All,

As offered in other postings, I challenged for some to differentiate
between themselves as Equipment Operators, Bench Techs, and
Cut-And-Paste Theoreticians. The "oomph" we all got from that
response was, as I also offered, the uncomfortable groan of avoiding
serious testing to instead indulge in eternal debates, the offering of
puzzles, and the recitation of dead white men's work.

So it follows that I am as responsible as others to differentiate
myself to the standards I set - it takes some time, but is not
seriously difficult. The quality of that differentiation is clearly
in the opprobrium above which maintains "it doesn't really matter" and
I can accept that, again, by my own terms of
"Does it make more than 1dB difference?"
which is more a quantitative test than the philosophical, qualitative
assertion that got us here.

The difference lies in a particular Engineering Note offered in the
1960's by Hewlett Packard: "Microwave Mismatch Analysis, Application
Note 56."

This note in turn references the work adopted by the National Bureau
of Standards published in IEEE Transactions of Microwave Theory and
Techniques, Vol. MTT-11, no. 3, May 1963, p. 179, by R.W. Beatty.

These references were then expanded in book form "Microwave Theory and
Applications," Stephen F. Adam, Prentice-Hall, 1969.

I took my training in the only school of Metrology in the US in 1972
and took these works to heart, by far and away not because they were
so much academic (even though they qualify in spades beyond all lesser
discussion), but because they were written for real engineers employed
in real measure employing real tools.

So much for Hubris, my own or others.

The simple fact of the matter is that if accuracy is not your cup of
tea, you wouldn't have read this far anyway. If you are concerned
with accuracy, and by that I mean making valid measurements of RF
power, then the Z of your source is exceedingly important in the face
of a mismatched load.

To wit I offer a simple verification of this observation and the works
enumerated above.

1. Take one Ham transmitter capable of supporting 5 to 10W into any
load without its ALC engaging and altering the power (this simplifies
matters but is not strictly necessary).

2. Obtain four 50 Ohm loads each capable of supporting that same 5 to
10W power without causing any of those loads to change Z (this is NOT
as simple as it may seem, but is not that difficult if you don't take
it for granted). Power will never be applied for more than 10 seconds
in any case during readings (or as long as it takes for your
transmitter to settle to a stable power).

3. Obtain a Wattmeter capable of reading both forward and reverse
power, where 5W is full scale deflection for the forward and reverse
power.

4. You will need an assortment of cables and T connectors (choose
your own connector style suitable for the frequency that you can
support these methods to be described, following).

5. Confirm ALL loads do not exhibit a reverse power reading for full
power (that same 5 to 10W) applied to them, one at a time, through
each and every cable.

5a. For the purposes of my demonstration, I employed a device of my
own design that I call my BVT, a Binary Variable Transmission line.
It contains an assortment of coaxial lines arranged such that they can
be inserted, inline, in 1 foot increments, for 0 through 32 feet. If
this sounds familiar, it was the subject of my very first posting to
this group some 8 years ago.

6. Apply one of those 50 Ohm loads in parallel, directly across the
transmitter output. Use a T connector to allow another line to be
connected in parallel as well. This port will be called the "Plane of
the Source."

7. Combine the remaining (3) 50 Ohm loads in parallel using two T
connectors such that one port remains open to accept a line to be
connected in parallel as well. This port will be called the "Plane of
the Load."

8. Connect a length of transmission line from the Plane of the Source
to the input of the Power Meter; and in turn, connect another line
from the output of the Power Meter to the Plane of the Load. The
center of the Power Meter (actually one port or the other depending on
which line you will vary in the future) is the "Plane of the
Measurement."

8a. I have chosen to use a 2 foot line to the input of the Power
Meter; and at the output of the power meter I have a 3 foot line
connecting to the BVT (which contains 1 foot of line in the zero
position) with another 3 foot line at the BVT output to the Plane of
the Load. As such, I had the capacity to change the position of the
Plane of the Measurement by 32 feet by one foot additions from a
minimum separation of roughly 8 feet. This also elongates the
distance between the Planes of the Source and Load from 8 feet to a
maximum of 40 feet.

9. I selected 24.9MHz by accident of meeting the Power Meter's
sensitivity and the transmitter's capacity to load this enormous
mismatch - it also happened to be eminently suitable for the data that
follows.

ALL data was taken by first setting the output power to force a
forward power reading of 5W (full scale) for EVERY variation, and then
taking the reverse power reading:

Initial reading 1.70W reverse
added 1 foot 1.45W
2 1.20W
3 0.95W
4 0.80W
5 0.60W
6 0.50W
7 0.45W
8 0.75W
9 0.80W
10 0.90W
11 1.00W
12 1.20W
13 1.30W
14 1.40W
15 1.45W

So, by now it may become apparent that:
there is only one mismatched system condition;
that condition is not impacted by any ALC action (either through it
not being engaged, or being nullified by maintaining a constant
forward power);
no change in the line conditions inject any additional mismatch (the
BVT is flat through all settings 0-15);
there is a substantial difference in what the apparent Power is.
there is a substantial difference in what the apparent SWR is.

The power determination for this system mismatched condition, i.e. a
power meter between two Planes of discontinuity, exhibits a low of
3.30W to a high of 4.55W depending on its position in, and the length
of the path between the Planes of discontinuity.

Hence the answer to the question of what difference does it make?
-1.395dB
Which also answers my standard: does it make more than 1dB difference?
Yes

It then follows that for a source that exhibits a 1:2 mismatch that is
feeding a 1:3 mismatch through a 50 Ohm system, you can always expect
some error, and that if you are ignorant of the dimensions, that error
can be as great as roughly 1.4dB.

What do the experts calculate? Well, I won't go into that math,
that is left to the student to find through research from any of the
three references cited. The graphs offered by each of them only allow
for a 2:1 mismatched source facing a 2:1 mismatched load which in turn
will yield up to a 1dB error.

Why the deliberated mismatch at the source? It refutes the myopic
discussion of source Z by rendering a known value in parallel with an
unknown (sic) which, when is 50 Ohms in itself, must render a source
that is NOT 50 Ohms (guaranteed in fact, and supported by the evidence
of data forecast by methods adopted by NBS).

73's
Richard Clark, KB7QHC