[TowerTalk] 80 4-square inside a 160 4-square? Pros and Cons please!

[Jim Lux]

At 09:58 AM 6/11/2004 -0700, Rick Karlquist wrote:
> > The function of that network is to ground the 160 elements
> > for 80 meter signals while parallel resonating on 160 so it
> > is open circuit on 160.
>
>If I understand what you are proposing, the 160 elements
>would be grounded halfwaves when operating on 80.  They would
>still distort the pattern (I modeled this on EZNEC).
>(The distortion is less that it would be if it were
>floating).
>
>I think you would need another trap halfway up, and in that
>case you want to float the 160 element on 80, not ground it.  This is like
>with guy wires:  < 1/3 wl if floating, < 1/6 wl if grounded.
>
>Rick N6RK

It might be tedious, but, given that the 160m elements (switched, trapped, 
whatever) are going to have some effect on the 80m patterns 
(traps,switches, etc. can reduce, but not eliminate the effects), is it 
possible that you could do an integrated design.  It's not as simple as 
just laying an 80m design on top of a 160m, but, fundamentally, I can't see 
why such a design cannot exist.  There's probably some arrangement of the 
elements that would produce perfectly acceptable results.

For example: This might not be practical for other reasons, but say you 
wanted to superimpose two 4 square type arrangements. You don't have to 
line up the elements.  One set of 4 could be 45 degrees from the other set.

There's also nothing that says that the center of the squares have to be 
collocated.  Maybe one element is a "dual band" 80/160 vertical.

Finally, if you have enough room (would that this is case!), one 
arrangement might be to put the higher frequency elements OUTSIDE the lower 
frequency ones.  Say, you built an 80 m 4 square (20m on a side).  Then, 
you put up the 40 m foursquare elements outside the 80m elements, but still 
spaced an odd multiple of 1/4 wavelength apart (i.e. 3/4lambda, or 30m or 
5/4-> 50m apart).


In any event, I find it hard to believe that you can't find SOME 
arrangement of the elements for the two bands (within a square defined by 
the larger elements) that doesn't have the right spacing and drive 
impedances and combined patterns that would work.

Clearly, this is a job for some sort of optimization program. There are a 
number of free general purpose optimizers out there that can run any 
modeling program (e.g. GENOPT).  The trick would be in defining an 
appropriate figure of merit to optimize.  With Yagi-Udas or Log Periodics 
for instance, there's a long history of manual optimization which helped to 
define useful figures of merit (F/B, VSWR bandwidth, etc.), so that when 
automatic optimizers came along, it was fairly straightforward to create 
appropriate "objective functions".  The same cannot be said for phased 
arrays, which have directivity and operational characteristics (and 
constraints) that are different from the "multiband beam on a tower" paradigm.

If anyone's interested in this kind of thing, I've got a bunch of programs 
that can parse the output of NEC, and other programs that can do things 
like measure beamwidth in the parsed output. They're in a combination of 
VB, C, and Matlab.

GENOPT can invoke a script/batch file which would have:

Modelbuilder -> takes parameters being varied and generates a NEC input file
NEC
Parser -> extracts data from NEC output
Evaluator -> uses parsed output to create figure(s) of merit

If someone were to suggest a suitable objective function/figure of merit 
for such antennas, I'll take a shot at doing the optimizing.  By the way, 
if someone has a little program that can do symbolic substitution and 
generate models into a NEC model file(like 4nec2 does with symbols), that 
would be really handy.  This would particularly handy if it can handle 
symmetry nicely.

Hmmm.. maybe something like an XML high level description of the antenna 
system that gets turned into appropriate NEC cards.