[TowerTalk] close spaced 4 square?

[Jim Lux]

At 01:53 AM 4/24/2006, Jim Jarvis wrote:

>Anybody have specific experience with an 1/8 wave spaced
>4 square array on 80 or 160?
>
>I'm interested in feeding/matching issues, along with
>bandwidth and pattern performance.
>
>Dealing with site constraints, including the presence of
>other towers, which may have to be detuned.
>
>It is likely that this will be an array of guyed tubular
>structures.  The existing towers don't permit the luxury of
>catenary support lines.  Target band is 80.
>
>Among the things I'm wondering is whether we'd be better off
>using 3 towers, which I believe we can space @ 1/4 lambda, and
>still get reasonable cardiods NE/SW and NW/SE, with a simple
>network.


So, you'd have the towers in an "L" shape? if x is EW and y is NS axis, 
then they'd be at something like (0,0), (1,1), and (1,-1) (scaled 
appropriately)

If the primary goal is to get the cardioid type pattern (a deep null), 
then, I think you can come up with, at worst, 4 simple (1 L, 1 C) networks 
to do the job, and a stack of relays.

The question would be whether the impedances are symmetric enough that one 
network can work for both directions.  Say you had identical elements, and 
the spacing is 1/8 wavelength, and you've constructed a network that gives 
you 1/8 wavelength phase shift, with equal amplitudes.  You could just 
interchange the two elements to reverse the pattern.  However, say one of 
the elements has twice the loss of the other.  Now, to get cancellation 
you'd have to feed twice as much power to the lossier element, but 
sometimes it's the lagging and sometimes the leading phase.


>As I see the problem, it's a tradeoff between a bit of forward gain
>and a more stable, lower Q system.   If I'm missing something,
>I'd appreciate a heads-up.

Not so much the forward gain.. it's the depth of null that's the tricky 
part.  If 10 dB is good enough, then it's not real, real critical (think in 
terms of arccos(1/sqrt(10)).  If you need 20 dB, then it gets more critical.


>And yes, I have Devoldere's book, latest edition, before you send
>me there.  Looking for some hands-on insights to augment that.


Jim W6RMK