[TowerTalk] modeling interaction

Jim Lux jimlux at earthlink.net
Wed Jun 22 13:04:14 EDT 2005

At 04:53 AM 6/22/2005, Eric Scace K3NA wrote:
>Hi Dick --
>     Some thoughts:
>1.  Simply deleting the source from the unused antenna changes that
>element from a dipole to a continuous piece of metal -- which is NOT
>what you have.

Eric raises some very important points.  Bear in mind the difference 
between getting a feel for the sensitivity and trying to calculate exactly 
what's going on.  If you try several different impedances at the feed point 
of the second antenna that sort of span a wide range (don't forget reactive 
loads!  that coax into a short or open, for instance, presents a variety of 
potential impedances) and look at the changes in the 3d pattern (which is 
fairly easy with a front end like 4nec2), you'll quickly know if there's a 
qualitative problem.

If you want quantitative data (particularly for null depths, etc.), then a 
more rigorous approach as outlined by Eric is needed.  However, there IS a 
saving grace.  Recalculating the impedances and patterns is very fast if 
the only change is the loading impedances, because the interaction matrix 
doesn't have to be recalculated.  You just change the LD cards and do 
another XQ in the same deck. Changing frequency will trigger a 
recalculation of the interaction matrix, but even for a fairly complex 
model, that doesn't take too long on today's computers.

With a limited number of feedpoints, one could also use some other 
shortcuts, but they're conceptually complex, and probably not worth 
describing here.

The real challenge is in presenting the resulting data in a meaningful way. 
(maybe a 3d sensitivity pattern?)  A program like Octave (free Matlab 
clone) has some pretty powerful 3d plotting and matrix analysis that might 

More information about the TowerTalk mailing list