On Wed, 06 Sep 2006 10:36:52 -0700, Richard Clark
wrote:
On Wed, 06 Sep 2006 07:36:13 GMT, Owen Duffy wrote:
I have plotted theoretical forward and reflected power for a detailed
model of RG58C/U at 100MHz with a severely mismatched load to
illustrate the underlying thing that is being measured by a
reflectometer.
The plots are at http://www.vk1od.net/lost/NN7K.htm .
Comments?
Hi Owen,
Hi Richard,
Interpretative comment - You state here "theoretical" plots of a
model. At the reference you state "indicated forward power, reflected
power." To me this implies data taken, not generated from a model's
formula - but my interpretation may be a stretch.
For avoidance of any doubt, it is entirely theoretical. Line
characteristics are based on an RLGC model of the line derived from
Belden's published specs for 8262 (RG58C/U).
Firstly, I should not have said "indicated forward power, reflected
power" as the "indicated" value is a voltage or current, ie the square
root of power, it is the deflection on the indicator meter if you
like, the meter current on the reflectometer. I have changed the text.
The meaning "indicated" is that it is the voltage or current driving a
reflectometer meter, calculated from the sampler and line conditions.
Data source comment - However, you go to rather elaborate lengths to
specify a load of 2+0.5j (sic) and a line of 50.36571-j0.17872. What
was the point of injecting reactances here if you simply wish to
illustrate the degradation of SWR with feedline loss? They add to the
implication of data taken, not generated - except for 5 place
resolution at 100 MHz which is (excuse me) absurd.
The stated line Zo precision is a result of lazy cutting and pasting,
and it should not be seen to imply that level of accuracy. I have
changed it.
The load (2+0.5j) is relatively unimportant, but it was a load to
create a high VSWR, higher than would be practical for such a feedline
at that frequency, extreme to satisfy the people who assert strange
things can only be observed at extreme VSWR. The reactance is just a
non zero value for no other reason than to test the models behaviour
on a complex load.
What is significant, is the load VSWR is very high at ~25, I would
suggest higher than practical, and therefore extreme.
Elaboration of the implications of interpretation and data source
comment - Further implying this was data taken at the bench is the
declaration "THE REFLECTOMETER IS NOT CALIBRATED FOR THE LINE Zo."
Again, my presumption circuits peg on that one because the reactances
are so easily removed from emphatic concern if you are in control of
the "virtual" workspace.
A reality of lossy lines is that Zo is not purely resistive.
A source of instrument error is a sampler that is calibrated for a
nominal Zo (being Ro). Fig 1 shows the modelled VSWR (which is what
accounts for increased loss, impedance transformation etc), and Fig 2
shows the calculated indication for a reflectometer calibrated or
nulled at 50+j0. There is a small difference, and the difference
becomes minute at lower VSWR.
Informational comment - What is supposed to be the total length of
line? Was it matched at its input, or simply fed by a transmitter?
The x scale is in wavelengths (as noted on the page), Frequency is
100MHz, line length covered by the graph is therefore about 9m.
Picky comment - Graph font sizes are a squint too small.
Ok, it is a quick cut and paste from Mathcad. For a permanent article,
I would try to do better, for I too have aging eyesight!
Last, really picky comment - Your copyright notice at the bottom is
off by a year.
Thanks, I have fixed the footer (which is common for most of the
site).
The graphs are a quick and dirty presentation, but it is a very
detailed transmission line model underneath. I acknowledge there are
all sorts of error sources in real life, but keeping sight of the
underlying phenomena helps to flag whether measurements look invalid.
Back to the original proposition that asserted that changing feedline
length can vary measured VSWR dramatically, and that shortening a
feedline might improve VSWR, there are reasons why that might happen,
but it is not explained solely by what is happening on the inside of
the feedline. A possible explanation for example is that the feedline
is not decoupled and carries common mode current, and that changing
the feedline length changes the load impedance seen by the load end of
the transmission line.
So, enough from me, can someone produce details of a scenario with a
very high load VSWR where there are multiple sweet spots with 1:1 VSWR
on a practical line? ("VSWR" here means a property of the line at that
point, not what might be read by an unsuitable or defective
instrument, and an extremely long line is not "practical".)
Thanks...
Owen
--