Re: NEC-LIST: Using MiniNEC?

Tan Boon Phing <bptan_at_usa.net> wrote:
> Hi, I am trying to model a rectangular reflector using wire-grid
> method in MiniNEC(not the professional version). I will like to
> know how can I set up the geometry for the reflector using MiniNec.
> Thanks.

I've modeled wire grid reflectors using Roy Lewallen's
<w7el_at_teleport.com> ELNEC & EZNEC versions of MiniNEC & NEC-2,
respectively, and more recently using Jerry Burke's <burke2_at_llnl.gov>
NEC-2. Rather than enter all the individual wire coordinates by hand,
which would have been disastrous because I make too many errors, I've
written little ad-hoc computer programs to generate the wire lists.
For checking purposes you need to have a program for visualizing your
model in "3-D", and you also must have a way of catching any
duplicate/ coincident/ overlapping wires.

Commercial programs are available for doing most or all of these
things, although I don't recall any specifics. I do not recall seeing
any "freeware". Look and you may find something on one of the
NEC-related web/ftp sites, such as

  <http://dutettq.et.tudelft.nl/~koen/Nec/index.html>

  <http://www.dec.tis.net/~richesop/nec/index.html>

  <ftp.netcom.com in /pub/ra/rander/NEC>

  <ftp.emclab.umr.edu in /pub/aces/NEC>

I'm not certain that all these sites are still active. I should
mention that Burke's NEC-2 does have features that make it reasonably
easy to generate a simple rectangular or square, planar grid, and he
has also written some useful visualization programs.

My model grids have had square meshes with one segment per wire, or
have been hexagonal overall (to approximate a circle) with equilateral
triangular meshes, again with one-segment wires. Typical segment
lengths have been in the range of 0.05 to 0.1 wavelength (wl). Be
aware that a grid with 0.1-wl holes is pretty leaky; i.e., there's a
fair amount of radiation from the side that should be shielded. I
don't recall numbers, but you'll find tables and graphs showing the
amount of leakage to be expected theoretically, as a function of
segment length, in handbooks & textbooks on reflector antennas. I've
been pleasantly surprised that my model results have been reasonably
consistent with theory. Frequently I'll use shorter segments in the
part(s) of the reflector where the "surface" current density is
greater.

I've plotted the currents in all of the wires of my mesh-model
reflectors, and they have always looked physically "reasonable" and
smooth and artifact-free. Although I have read warnings about
potential numerical troubles in NEC-2 for models having lots of little
loops, I got very reasonable results from Jerry Burke's
single-precision version of NEC-2 for a mesh of equilateral triangles
having 4800 wires (4.8 thousand), each 0.05 wl long. (The wire radius
was 0.005 wl, if I recall correctly.) BTW, to model this reflector I
exploited its six-fold, or 60-degree, rotational symmetry to reduce
the amount of storage and computation required. This particular
4800-wire model solution executed completely in the 64 MB RAM of my
computer without disk swapping, which made a huge difference (an oder
of magnitude, perhaps) in running time. I don't recall whether
MiniNEC can exploit such symmetry.

73 -Chuck, W1HIS
Received on Thu Sep 25 1997 - 10:38:06 EDT