[TowerTalk] 6M yagi

[Brian Beezley]

If you can manage it physically, one trick to lower receive noise is to 
vertically stack two shorter Yagis rather than using one long one. The 
spacing will direct a null toward ground. At half-wave spacing, the null 
will be directly below. This spacing is too small for best forward gain. 
For wider spacing, the null moves out. For example, at 0.7 wavelengths 
it is at 45 degrees. At one wavelength, it is at 30 degrees. The angles 
might put the null right at your noisy neighbor's place. A null directly 
below also reappears at one wavelength.

Note that the azimuth pattern of a horizontal Yagi is the free-space 
elevation pattern multiplied by the E-plane response of a dipole. So any 
adjustment to the elevation pattern to lower noise will affect the 
azimuth pattern. Optimization to reduce the response in a region 
centered directly below does the same thing to the azimuth pattern off 
the ends. This is where it is already small due to the E-plane dipole 
response. The additional drop is not likely to be of much benefit, and 
it may hurt the response toward the rear.

An automatic optimizer can trade off everything if you tell it your 
design preferences and constraints. It can even redirect its effort once 
a performance objective has been met. I find all this impossible to 
manage by manual optimization - it's been decades since I tried. The 
neat thing about a good optimizer, particularly one that uses a global 
algorithm rather than a local one, is that the final design is the best 
possible according to your trade-offs. If you tweak the design to 
improve one aspect, another will surely degrade. I always felt this was 
rather magical. The trick is telling it what you want. Often you don't 
know until you get some idea of the performance space and how one 
objective naturally trades for another. I once designed an optimizer 
that let you explore performance space using optimal designs. There was 
a triangle that represented gain, pattern, and SWR trade-offs. You could 
move a dot around inside the triangle with the mouse to dynamically 
change the trade-offs. While you did this, the optimizer was running. 
The idea is that all the designs you're exploring are the best possible 
given the priorities you've set and they are shown on the screen as you 
make trade-off changes. Computers weren't fast enough in those days to 
instantly optimize a design, but they are today. The last time I 
checked, which was eight years ago on single 3 GHz processor, my code 
was calculating 165,000 8-element designs per second. Things converge 
quickly to optimal at that rate! Try doing that with NEC. I've been 
toying with the idea of resurrecting YO and adding modern methods, such 
as multiple CPUs running in parallel. The problem of Yagi optimization 
is readily separable and can be handed off to multiple independent agents.

That was quite a ramble. And I was just getting started. I better cut 
this short and get some sleep. For some time now I've been getting up at 
midnight and programming for 18 hours straight. I think it's finally 
getting to me. This is what I've been working on:

https://k6sti.neocities.org/sp

Brian