[TowerTalk] Analysis of phase errors for transmit phased arrays

[Jim Lux]

At 06:44 PM 1/15/2006, Rick Karlquist wrote:
>Jim Lux wrote:
> >  In numbers, if you control the phase to within 45
> > degrees, in 99.7% percent of the cases, you'll lose less than a half dB.
> >
> > The writeup and graphs are at:
> > http://home.earthlink.net/~w6rmk/antenna/phased/txtol.htm
> >
> > > Thanks
> > Jim, W6RMK
> >
>
>I didn't see in the report details of what the arrays were
>other than # of elements, but the 3rd paragraph says it all:
>"doesn't necessarily apply to superdirective arrays".
>There are enormous differences in phase sensitivity of different
>commonly used amateur arrays.  I can certainly give you plenty
>of examples where you need to control the phase to a lot better
>than 45 degrees to lose less than 1/2 dB.  Try end fire arrays
>with 3/16 wl spacing, binomial amplitudes.  Not too bad with cophasal
>currents, but as you increase the phase gradient to get more gain, (and
>enter the superdirectivity regime) the array gets extremely sensitive
>to phase and amplitude errors.  The gain you can get is directly
>related to accuracy.  On the other hand, broadside arrays will still
>work even with fairly large phase/amplitude errors, especially
>if you don't care about sidelobes.


That's the next step.. to look at superdirective arrays.  This analysis is 
only for arrays where you're using phasing that matches spacing (broadside 
arrays being a straightforward example).

A question about the superdirective array is whether the phasing 
criticality is the excitation, or the actual radiated phase from the 
element.  Superdirective arrays (in general) store a fair amount of energy 
in the near field, so the trickiness is in the mutual Z, which is highly 
sensitive to drive phases.

As it happens, the non-superdirective case was easy to run (since I had the 
models from work).  The superdirective case will require cranking up some 
new models.