ON7NQ, Danny Mees, has done extensive modeling work on the 3-element
5-band Cubex quad, optimizing it for separate feed to avoid some of the
problems associated with combined feed. His e-mail address is
Danny did much of his modeling work, later confirmed on the antenna
itself, on AO, and I did some confirmation modeling of his designs on
NEC-4, with no great differences. He managed to increase gain and F-B--as
well as getting a good match--on all bands. He even added a second
director to 10 meters, since the constant spacing of the design sets the
10 meter elements quite far apart. I think his recommended modifications
will eventually appear in the next Antenna compendium.
I have done some extensive modeling of common and separate feed systems on
2-element quads, using a spider design rather than constant spacing.
(QST ran a nice 2-element design with constant spacing back in the
early 90s--exact reference not at hand as I write this.) The
greatest common-feed interactions were 20-10 meters and 12-10 meters.
When you model, you cannot claim to have covered all possible
combinations; hence, I cannot say that there is no common-feed design that
will achieve the same results as a comparable separate feed model.
However, I found that it was much easier to achieve monoband results when
using a separate feed system and the unused drivers were closed loops.
Although it is relatively straight forward to get optimal performance from
a 2-element quad in a multiband array, more than 2 elements requires
constant spacing for a 2:1 frequency range--or the addition of high-band
spreaders to make the spacing on those bands optimal. Like a multi-band
Yagi, a multi-band quad is a compromise design in more than 2 elements.
When designing quads via modeling in NEC, do not skimp on segments and
keep them as aligned as possible. Using 7, 9, 11, 13, and 15 segments per
side per band from 10 to 20 meters achieves this quite closely, but the
resultant model is quite large. Auto-segmentation may or may not yield
correct results--it is best to do one's own convergence testing.
Of course, for those interested in modeling quads, Orr, Haviland, and
Koszeghy are the standard references for basic designs and design
L. B. Cebik, W4RNL /\ /\ * / / / (Off)(423) 974-7215
1434 High Mesa Drive / \/ \/\ ----/\--- (Hm) (423) 938-6335
Knoxville, Tennessee /\ \ \ \ / / || / (FAX)(423) 974-3509
37938-4443 USA / \ \ \ \ || firstname.lastname@example.org
FAQ on WWW: http://www.contesting.com/towertalkfaq.html
Administrative requests: towertalk-REQUEST@contesting.com