If made from twinlead with insulation between the conductors (e.g.,
window line or 300 ohm flat or tubular TV twinlead), NEC based programs
can't model it exactly in a straightforward fashion for the same reason
it can't model patch antennas on a substrate -- the program can't
account for the effect of the dielectric between the conductors.

The two modes Ian speaks of can be separated by separating the two
currents into common and differential modes(*). The common mode current
is the source of all the radiation. It propagates as Ian says at a
relatively high velocity factor. This mode is modeled quite well with
EZNEC or NEC-4's insulated wire feature.

The differential mode current doesn't cause radiation. Its velocity
factor is determined by the dielectric between the wires, and NEC-based
programs don't have any way to account for the field modification which
the dielectric causes to bring this about. That's where the shortcoming is.

The net result is that the radiating properties of the stubs can be
accurately modeled, but their length which determines the "trapping" or
loading characteristics would be off. Of course, such an antenna made
from air-insulated twinlead could be modeled easily.

I've successfully modeled a folded dipole made with TV twinlead by
physically separating it into common mode and differential mode
structures. The common mode portion is a simple dipole, with diameter
equal to the effective diameter of the two conductors in parallel, and
with wire insulation. Then across the feedpoint I put non-radiating
transmission line models to model the differential mode transmission
line stubs. The lengths of these took into account the twinlead velocity
factor, and their impedances were 1/4 the impedance of the real stubs
because of the transforming property of the structure. The result was a
good model, provided that the feedpoint impedance was multiplied by 4.

A first look indicates that this approach wouldn't be practical with the
Lattin, because it would require large jumps in effective wire diameter
as you go along the antenna, which NEC doesn't handle well.

What you really need is a way to increase the differential mode length
of each of the stubs without impacting the common mode length. A few
quick sketches indicate that it just might be possible to insert a
transmission line model (which has no physical length or common mode
radiation) in series with a stub, which would accomplish the goal if its
length were made to equal the difference between electrical and physical
length of the real stub. But I don't have time to pursue it. Anyone
interested in doing so should begin with a single stub and carefully
observe its characteristics.

Whenever modeling close spaced parallel wires with NEC-based programs,
it's vital that the segment junctions be aligned on the wires. There's
more information about this in the EZNEC manual (available also with the
demo program). Look in the index under "Parallel Wires".

(*) Modeling programs don't treat the modes separately. But separating
them makes it easier to explain and understand what causes the problems.

Roy Lewallen, W7EL

Ian White GM3SEK wrote:
Cecil Moore wrote:
Roy Lewallen wrote:
But most often when you see an antenna inventor or seller claim that
his antenna "can't be modeled" by NEC, EZNEC, or other programs, it
just means that modeling fails to show the extraordinary performance
he claims for it. That's simply a failure of the program to include
the effects of magical properties and wishful thinking in its
calculations. I've come to regard such claims as a red flag
indicating a probable exaggeration of antenna performance.

I wish I could remember the correct spelling for the antenna
I tried to model. Something like "Lentine". It is a dipole
of sorts made from shorted and open sections of balanced
transmission line. I tried modeling it with wires in EZNEC
and got all sorts of errors. It looked something like this:

+--------+--------+--------FP--------+--------+--------+
+------ +------ +------ ------+ ------+ ------+

Anyone remember the correct spelling for that antenna?

Google for "Lattin antenna". (Too many "lentils", Cecil :-)

One of the first hits is http://www.g3ycc.karoo.net/lattin.htm which
shows a good sketch. The antenna is made from sections of 300-ohm ribbon
or tubular feeder, configured as a string of quarter-wave stubs that
progressively make the dipole shorter as the frequency increases.

The modeling challenge is that the ribbon operates in two different
modes at the same time: a radiating common mode with a velocity factor
of say 0.95; and a non-radiating "stub" mode with a VF of about 0.8. The
problem is to model both modes simultaneously, for the whole string of
stubs, without changing the physical dimensions of the real antenna. I'm
not sure if NEC can do this, but maybe Roy can comment?