>reradiation. Do some of you broadcastertypes have any experience with
>modeling gamma wires or other means of detuning? Does the closespacing
>limitation of NEC prevent this from being done accurately?
The closespacing limitation of NEC (2 or 4) usually is a problem only
when the wires are super close or when they are of different diameters. A
detuning wire/gamma of the same diameter as a guy or other wire should be
straightforward to model, with the proviso of making the parallel wire
segments match up as closely as possible. Close wires where the segments
do not start and end together tend to give erroneous results.
There are a number of techniques a modeler can use to capture a large
scale system of antennas and towers within a finite segmentation limit
(short of buying more expensive software). The usual limit is about 500
segments.
A truly accurate model of a 5 element HF Yagi that uses a collection of
tubing of different diameters may take up 150200 segments.
However, if you check the length and diameter of the substitute elements
used for Leeson corrections, then you can replace the large model with a
much smaller but still accurate one of uniform diameter elements.
However, you should ensure that the performance of the original and the
substitute are quite close before using the substitute. you can save up
to 75% of the segments this way and use them for other antennas and
metalic supports.
Accurate models of multiband beams that employ traps or even monobanders
that use linear loading are not easy to make. Developing monodiameter
wire equivalents is a useful technique. In some cases, loads can be
entered as series RLC loads, but they may not in all cases reveal all
possible interactions.
Modeling towers is somewhat tricky. You can use approximate equivalents
composed in the model of uniform diameter materialif you have reasonable
confidence that the substitute approximates the original. You can also
make triangular sections, but these require more segments per unit length.
Wherever a structure is complex, having some point of measured reference
for electrical parameters is useful when employing simplified models.
In all cases, beware of using too few segments on the models. For most
structures, the convergence test is a reasonable starting point for
adequacy of segmentation, but it may not in itself comprise a complete
test of adequacy. It only takes one inadequately modeled structure in a
field of structures to invalidate an interaction analysis. On the other
hand, if one is simply looking for likely candidates of interaction (in
contrast to close approximations of the degree of interaction) simplified
models can be useful, if each component is then run through some
variations of segmentation to ensure that the simplification is not
disguising a truly interactive source structure.
For a system of horizontal antennas, a good starting point is to model all
of the antennas firstincluding tests that rotate them. However, if
vertical antennas are part of the system, then all vertical structures
should be part of the initial test model.
Develop the full antenna farm model slowly, improving it as you uncover
different ways of improving it. Have a few variations on hand to test
each component revision. Start with the most critical structures (the
ones of greatest concern) and expand the model farm a piece at a time.
Even so, your house wiring, as well as telephone and TV cabling, my be
enough to drive you crazy. Whether you can ignore or delete structures
from a total model may depend on the current readings you find for the
pieces of the model. Wire structures with no segment having more than 1%
of the source current are generally safe to delete, but this must be true
for all frequencies used in the tests. I am not aware of any standard for
a threshhold of relevance in terms of currents, so how far above 1% you
can go and still delete a structure to simplify the model is subject to
your judgment.
In effect, with unlimited segmentation software, you can routinely (but
not necessarily easily or quickly) model all structures in the field of
interest. With limited segmentation, special care is needed to ensure
capturing the actual interactions.
These notes are not designed to discourage antenna farm modeling, but
instead to challenge individual ingenuity in developing finite models that
still do the job well. Hope they are useful.
73
LB, W4RNL
L. B. Cebik, W4RNL /\ /\ * / / / (Off)(423) 9747215
1434 High Mesa Drive / \/ \/\ /\ (Hm) (423) 9386335
Knoxville, Tennessee /\ \ \ \ / /  / (FAX)(423) 9743509
379384443 USA / \ \ \ \  cebik@utk.edu
URL: http://web.utk.edu/~cebik/radio.html

FAQ on WWW: http://www.contesting.com/antennawarefaq.html
Submissions: antennaware@contesting.com
Administrative requests: antennawareREQUEST@contesting.com
Problems: ownerantennaware@contesting.com
