I'm interested in knowing if anyone has modeled, built, and/or tested the
W5GI multiband antenna described in July 2003 CQ magazine. I'm especially
interested in any experience mounting the antenna in inverted-V fashion,
rather than as a flat-top T.

For those who haven't seen the article, the W5GI is constructed using two
physical 1/4-wave lengths of coax, center-to-braid shorted on the far ends,
as the middle 1/4-wave segment in each leg of the antenna. These coax
segments are supposed to provide the phase reversal necessary for the
collinear design, and also act as part of the collinear radiator set. (The
phasing stubs often used when building a wire collinear are not employed
here.) Since the coax segments are not phase-reversing on other bands, the
antenna is supposed to act as a G5RV on bands pther than 20M. I'm
principally interested in its performance on 20M, since the G5RV's theory
and characteristics of operation have been extensively discussed elsewhere.

Supposedly, people attempting to model this antenna using available software
tools have come up with disparate results. The author claims that numerous
people have built and used the antenna on 20M and have been impressed with
its performance. The major lobes supposedly occur broadside to the wire, as
one would expect with a collinear.

The "mystery" here, as I understand it, is whether the coax segments
actually do act as 180-degree phase-reversing elements, and if so, why the
antenna seems to work correctly only when the coax elements are cut to a
physical quarter weavelength (16.5 ft) rather than to an electrical quarter
wavelength. (It's easy to see why this is true when the antenna is operated
as a G5RV on others bands -- but what about when operating as a 20M
collinear?)

Bill / W5WVO