[TowerTalk] Front-to-back / Front-to-rear

[Jim Lux]

Scott McClements wrote:
> Hi,
> 
> Since I am in the process of designing a new directional antenna
> (w/NEC2), I have started to think about which measure(s) of rearward
> rejection I would maximize on:
> 
> 1. Front-to-back - Difference in gain in main lobe direction and 180
> degrees opposite
> 2. Worst case front-to-back - Difference in gain between main lobe and
> highest gain in the quadrants opposite the main lobe.
> 3. Front-to-rear - Difference in gain between main lobe and average gain
> in the two quadrants opposite the main lobe.
> 
> I can easily understand the pros (and cons) of high front-to-back.
> But I am trying to understand which is more desirable for a
> directional antenna that will be used on 14Mhz-30Mhz -  a high worst
> case F/B or high F/R.  What would be the practical difference between
> the two?
> 
> In calculating F/R, to maximize performance,  would it make more sense
> to average gain over a smaller range, say +/- 45 degrees opposite the
> main lobe (instead of +/- 90 degrees)?
> 
> -Scott, WU2X
> 

You bring up some interesting questions.

First off.. Most of the benefit of directional antennas is not in the 
forward gain (as you've probably found, almost anything gets you to 
within 1 dB of the absolute best) but in the management of what the rest 
of the pattern looks like.

Not very much power, in percentage terms, is being radiated other than 
in the main lobe. Say you're radiating 1500W.. a halfway decent design 
probably radiates well over 90%, if not 95%, of that in the main lobe, 
so you've got 50-100W or so radiating in all other directions.  If, by 
some herculean efforts of optimization, you managed to get that 100W to 
squirt in the forward direction, you're looking at tenths of a dB 
improvement.  And, you'd likely have a design that was very, very picky 
about dimensions and mechanical stuff.

So, you probably want to think about what you want for receive 
performance.  I think there's sort of two factors to think about.  The 
first is the performance for seriously off beam signals (e.g. from the 
back).  If you have an interfering signal you want to suppress by 20dB, 
for instance.  The other factor is for signals that are slightly off the 
main beam (i.e. you're pointing from Southern CA at NY, and want to 
suppress the folks in Maryland).  For that, I think you want a pattern 
that falls off fast (or has a nice nulls to the side of the main lobe).

Sort of like for transmit, absolute forward gain isn't so important 
(because you're most likely atmospheric noise limited for SNR), but 
differential gain between the desired signal and the interferer, and the 
best place for that might be halfway down the side of the main lobe.


But, the other thing to seriously look at in your design is how 
sensitive it is to mechanical changes.  If your pattern changes 
radically when the element lengths change by an inch or three, that's a 
sign that  you might want to rethink your design.  Likewise if the 
element orientations change (angle relative to boom and other elements) 
a bit, and pattern changes a lot, that's a warning sign.  (Unless you've 
got some sort of way to guarantee that your antenna is going to be 
perfectly rigid and never move in the wind).  This, to me, is the most 
useful thing about modeling.  It answers the "what happens if X occurs" 
questions, because it's not too hard to systematically examine the 
behavior over a range of reasonable occurrances. This is especially so 
if you use a modeling tool that lets you put symbols and equations in 
the model, like 4nec2.  You can make the end points of the elements some 
function of an angle, and then vary it:

Say I've defined one of my Yagi elements as:

GW,101, 11, xpos, len*cos(theta), len*sin(theta), xpos, -len*cos(theta), 
-len*sin(theta), 0.01

which defines an element of length len, at angle theta relative to 
horizontal, at distance xpos along the boom.  I can then have the tool 
systematically vary theta from, say, -10 to 10 degrees in one degree 
steps and see what happens to the pattern.

Jim, W6RMK