[TowerTalk] Quagi Optimization

Jim Lux jimlux at earthlink.net
Mon Feb 20 12:06:49 EST 2006


At 08:05 AM 2/20/2006, Al Williams wrote:

>----- Original Message ----- From: "Jim Lux" <jimlux at earthlink.net>
>To: "Pat Barthelow" <aa6eg at hotmail.com>; <towertalk at contesting.com>
>>In general, elements closer to the driven element are more critical (they
>>have more current).
>
>For a REGULAR YAGI do the parasitic element's induced voltages cause 
>radiation which then cause currents in the neighboring elements and so on? 
>I am pretty sure that this is true but do these SECONDARY currents have 
>the amplitude and phase to have much effect on the total pattern?

You bet they do.  That's part of the challenge of yagi antenna design 
(using analytic methods): every element interacts with every other. You're 
essentially solving a NxN matrix equation to create a set of element 
excitations that gives you the desired pattern, with the elements of the 
matrix being the mutual coupling between the various pairs of elements.  A 
lot of the early Yagi work focussed on coming up with convenient 
formulations that would let you do this without too much complexity 
(imagine inverting a 10x10 matrix by hand!). Typically, you'd approximate 
the current along the element with something like a sinusoidal 
distribution, to make the calculation easier.

  And, if you want superdirective gain (which you do), adjacent elements 
are actually somewhat out of phase with each other, which has the effect of 
cancelling side and back radiation more than it cancels the forward 
radiation.  This has to be traded off with the IR losses in the elements 
(jacking up the coupling by putting elements close together may give you 
great directivity, but also huge losses, so the gain will be terrible). 
Relatively small changes in the current phase and amplitude will reduce the 
amount of cancellation, so the performance is degraded.


>Do antenna modelling programs take these secondary currents and radiation 
>into account when summing up the segments and elements?

yes, they do.  In fact, what they do is calculate the mutual coupling from 
each segment in the model to every other segment in the model, build up a 
HUGE matrix, and solve it.  So, for a 1000 segment model, you're talking 
solving 1000 equations with 1000 unknowns.  Once you know the current in 
each of those 1000 segments, you can calculate the far field by summing the 
contributions from each segment.




>K7PUC



More information about the TowerTalk mailing list