Owen Duffy wrote:
It has been suggested that steady state VSWR measurements on a
transmission line are influenced by the source to line match, and that
the measurements are not valid unless the source is matched to the
line.

That seems in conflict with textbooks that suggest that the reflected
wave is entirely determined by the conditions at the load end of the
line, and that the conditions at any point x on the line can be
determined from x, Zo, the load impedance, and the line's propagation
constant.

So, I have performed a small experiment.

. . .

Comments?

If it isn't obvious from the math in the textbooks, it's simple to see
from an intuitive standpoint that the source match has no effect on the
SWR. Here's why:

Imagine a system where an AC voltage source, turned off, is connected to
a transmission line through some arbitrary resistance. The other end of
the transmission line is connected to some arbitrary load. When the
source is first energized, a wave will travel from the source end of the
line to the load, where some of it will be reflected. The fraction
reflected, and its angle, are determined by the relationship between the
load Z and the line Z0, and this ratio is in fact the reflection
coefficient at the load end.

We now have forward and reverse waves on the line, one of each, which
sum to form the standing wave, and from which we can calculate a
standing wave ratio. When the reverse traveling wave reaches the input
end of the line, some fraction of it will be re-reflected toward the
load, the fraction depending on the mismatch at the source. And here's
the key concept: No matter how much is re-reflected toward the load, the
same fraction of that new forward traveling wave will be reflected from
the load as was reflected when the original forward wave hit the load.
So the total of the two waves traveling forward divided by the total of
the two waves traveling rearward is the same as when we only had one
wave going each way. Consequently, the SWR is the same as it was for the
first two waves. You can continue this process until the magnitude of
each new wave is insignificant or, in other words, steady state is
reached. Each pair of new waves makes no change in the SWR. And all this
is true regardless of what fraction is reflected at the source end of
the line.

Let's just put some numbers on things as a simple example. Suppose the
source reflection coefficient (the source Z as seen from the line) is
0.5, and the load reflection coefficient is 0.3. (I'll assume a line
that's an even number of wavelengths long and reflection coefficients
which are purely real to avoid bothering keeping track of phase angles.)
And suppose the source has an amplitude such that the initial forward
voltage wave is exactly 1 volt. The first forward wave then has an
amplitude of 1 volt, and when it hits the load, a rearward moving wave
of 0.3 volt is created, resulting in a reverse/forward voltage wave
ratio of 0.3 and an SWR of 1.86. When the 0.3 volt rearward traveling
wave reflects off the source, it creates a new forward wave of 0.3 * 0.5
= 0.15 volt. And when this hits the far end, it sends back a wave of
0.15 * 0.3 = 0.045 volt. We now have 1 + 0.15 volts going forward and
0.3 + 0.045 volts going rearward, or a total of 1.15 volts forward and
0.345 volts reverse. The reverse/forward ratio is still exactly 0.3, and
the SWR still 1.86. You can continue this for any number of
forward-reverse pairs and always get the same result. And you can
substitute any value you'd like for the source reflection coefficient,
and also get the same result as long as you don't change the load
reflection coefficient.

A moment's thought reveals that if the cable is lossy, the same fraction
of each forward wave still returns to the source, so the same reasoning
can be used to get the same result -- that is, that the SWR is still
unaffected by the source match.

The problem with doing experiments to "prove" a well-known principle is
that it's often very difficult to get good experimental results. So when
the results don't agree with the established theory, too many
experimenters become convinced that they've disproved the principle.
What they should be doing is figuring out what went wrong with the
experimental measurements.

In this case, I'm relieved that the experiment does agree with the
theory which has been established, used, and proved for more than a
century rather than with amateur folklore. If it hadn't, it would have
proved only that the experiment was faulty.

Roy Lewallen, W7EL