Frank wrote:

As I understand NEC; large errors can be introduced by junctions of
dissimilar wire diameters, and in particular when the wires are at 90 deg.
Therefore, when you have designed your "GH" inductors, the rest of the
antenna should by constructed of the same diameter wire. This may be
difficult since Dan is using two coils of significantly different Qs. I
guess you could overcome this problem by varying the conductivity of the
inductor to obtain the desired Q. Also, since segmentation tends to be
relatively high in a helix, should segment length tapering be applied to
those segments adjacent to the helix?

Frank, VE6CB

It's difficult to give an absolute answer to these questions, but some
general comments and guidelines should help.

First, the error introduced by NEC-2 when wires of dissimilar diameter
are connected is generally small, unless the wires are grossly
different. This error can be minimized by making the segments as *long*
as possible adjacent to the junction, which of course is contrary to the
general principle that more segments are better. Even a small error can
cause major changes in the pattern when the dissimilar diameter wires
are in a parasitic element. EZNEC and a number of other programs have a
built-in method of avoiding this problem for certain antenna types, but
plain NEC-2 doesn't. NEC-4 is relatively free of this problem, but it's
quite expensive for hobby use.

The Q of an inductor is determined by the inductance and the loss. The
loss is a function of the dielectric, wire resistance, and radiation
(which isn't really loss, but lowers Q as though it were). NEC type
programs automatically account for the radiation, and it's easy to
include wire loss. So assuming negligible dielectric loss, the programs
should predict Q fairly accurately -- except for proximity affect.
Proximity effect could be modeled in NEC by increasing the resistivity
of the wires in the coil. EZNEC currently allows only a single wire
resistivity for the whole model (although this will probably change in
the next version). However, since the overall loss will be dominated by
the inductors, the higher resistivity could be specified for the whole
model without sacrificing significant accuracy. Alternatively, a number
of resistive loads could be inserted in the inductors.

Segment length tapering usually isn't necessary with NEC based programs,
unless there's a source near a place where the segment length changes.
An average gain check should be run to determine if there's a problem.
If there is, segment length tapering is one tool which can be tried in
improving the average gain.

Roy Lewallen, W7EL