[TowerTalk] Stacking Tribanders

[Lux, Jim]

On 6/22/22 12:20 PM, Billy Cox wrote:
> Hi Paul,
>
> Below is from pages 12 and 13 of Dean's document, perhaps
> this might help to explain what you have observed there?
>
> ACCURACY AND TESTING THE RESULTS
>
> What would I estimate as the “accuracy” of HFTA elevation predictions? I would say that I would trust the results within plus/minus 3 dB. In other words, take HFTA results with a grain of salt. Don’t obsess with changing the height of your antenna by fractions of a foot to see what happens!
>
> Having said that, now I must state that it is a good idea to compare elevation patterns in intervals of perhaps 1 foot to assess whether HFTA
> is generating reasonably smooth results. Often, the ¼steps used in the program don’t align exactly and artificial spikes (or holes) can be created. This is inherent in any ray-tracing program and can only be eliminated by using extremely small angular step increments —and doing so would slow down execution even more.


This is an interesting point.. There's a lot more processor HP available 
now.  There's no reason that one couldn't run 0.1 degree steps and 30cm 
increments.

Maybe that's what tools like HOBBIES (mentioned by Jim K9YC?) might buy 
you, although from the Wikipedia entry it's more a MoM code and doesn't 
deal with far field effects (terrain).

The scattering and diffraction in the far field is a more complex 
problem and requires different methods.  Jim Breakall, et al., in 1994 
"The modeling and measurement of HF skywave radiation patterns in 
irregular terrain" in IEEE trans on ant and prop (july 1994)

"The method of moments (MoM) was used in conjunction with the geometric 
theory of diffraction (GTD) for predicting the elevation-plane radiation 
patterns of simple high-frequency (HF) vertical monopoles and horizontal 
dipoles situated in irregular terrain. The three-dimensional terrain was 
approximated by seven connected flat plates that were very wide relative 
to the largest wavelength of interest. The plate length along the 
terrain profile was the longest possible that still adequately followed 
the shape of the path on the azimuth of the elevation pattern of 
interest and no shorter than 1 wavelength at the lowest frequency of 
interest. The MoM model was used to determine the antenna currents under 
the assumption that the terrain was planar (i.e., locally flat) over the 
distance pertinent to establishing the input impedance. The currents 
thus derived were used as inputs to the GTD model to determine the gain 
versus elevation angle of the antennas for HF skywave when situated in 
the irregular terrain. The surface wave solution for groundwave was not 
included since this does not appreciably contribute any effect to the 
skywave far-field patterns at HF in this case. The model predictions 
were made using perfect electric conducting (PEC) plates and using thin 
plates made of lossy dielectric material with the same conductivity and 
relative permittivity as measured for the soil. These computed results 
were compared with experimental elevation-plane pattern data obtained 
using a single-frequency helicopter-borne beacon transmitter towed on a 
long dielectric rope in the far field on a linear path directly over the 
antennas. The monopoles and dipoles were situated in front of, on top 
of, and behind a hill whose elevation above the flat surrounding terrain 
was about 45 m. The patterns of all of the antenna types and sitings 
exhibited diffraction effects caused by the irregular terrain, with the 
largest effects being observed at the highest measurement frequency (27 
MHz)"


>
> After I do an evaluation for a particular antenna height, I will often specify an overlay of three heights separated by one foot each. For example, if you are interested in a single antenna at a height of 80 feet on 14.0 MHz for the K5MA-330.PRO terrain, you might first compare three heights of 79, 80 and 81 feet, bracketing that height. The three curves overlaid on each other look relatively smooth, except there is a 1.4-dB “bump” for the 79-foot height.
>
> Now, run three heights of 80, 79 and 78 feet. Now, the curves for 78 and 79 feet look smooth, but the 80-foot curve has a noticeable dip. This means that spurious artifacts of the ray-tracing process are occurring at 80 feet in the program —but these would not occur in the real world. The solution: don’t use the 80 foot point in the computer analysis, but you would mount your real antenna at that 80-foot height if you like the response at 79 or 81 feet.


This gets to the famous quote from Hamming "The purpose of computing is 
insight not numbers"

Understanding the limitations and assumptions of the tools is important.