Re: NEC-LIST: Modelling collinear coaxial array

David de Haaij wrote:
>
> Hello All
>
> Does anyone have an idea how to model collinear-coaxial arrays in NEC?
> Specifically in NECWin-Pro
>

I have worked with them in EZNec. The model is effectively a set of
wires of the OD of the coax, end to end. Between each adjacent pair
of ends is a source. The amplitudes and phases of the sources are
related by the transmission line formulas; in EZNec, this can be done
automatically for lossless line, which is a good starting point at
least, and could be "manually iterated" to a more accurate solution if
that's justified. In EZNec, the model for an N-element antenna is N
wires with gaps between their ends and divided into appropriate
numbers of segments, with N-1 single-segment wires, one between each
pair of ends of the N elements, used to simplify the bookkeeping of
introducing the sources (and transmission line connections). You
could do it with just a single wire with carefully chosen
segmentation, but my method preserved my sanity...

For example, I just brought up one such design I did for 440MHz, 10
elements. There are 19 wires. All the wires are at X=0, Y=0, and
have a diameter of 4.57mm. The bottom of the antenna is 0.5 meters up
from ground. So the first 5 wires, starting at the bottom, have Z
axis end points of (0.500m, 0.742m, 30 segs), (0.742m, 0.762m, 1 seg),
(0.762m, 1.003m, 30 segs), (1.003m, 1.023m, 1 seg), and (1.023m,
1.265m, 30 segs). There is a series of 14 additional elements in the
same relationship above these. There is a current source at the
center of the 2nd wire. There is a set of 8 transmission line
entries, from wire 2 to wire 4, from wire 4 to wire 6, etc., up to
from wire 16 to wire 18. The transmission lines are connected at the
midpoint of each (one-segment) even numbered wire. They are specified
at a length of "actual distance" between the connection points. Z0
and Vf are as appropriate for the line to be used. The connection is
"reverse" because the center conductor connects to the lower terminal
at the lower end and the upper terminal at the upper end of each line.

The results I've gotten from this model in general seem to agree quite
well with the performance I get from antennas constructed in this way.
Of course, the actual feed arrangement must accomodate impedance
matching and decoupling; decoupling is critical because antenna
currents on the feed line can seriously degrade the gain performance
of the antenna. I have not bothered to do the "manual iteration"
suggested above to account for transmission line losses, but the
modelling technique would be essentially the same, simply using a set
of sources instead of lines and reflecting the loading on each line
segment down the stack of elements. A back-of-the-envelope calc
convinced me this wouldn't be worth the effort for what little
adjustment it would make for the low loss line that I use, whose
segments are very close to 0.5 wavelenths long (accounting for v.f.).

Cheers,
Tom
tomb_at_lsid.hp.com
(also at k7itm_at_aol.com)
Received on Sat May 15 1999 - 06:51:27 EDT