Last November I tried to analyze mods to the CM4251 Parabolic, but was frustrated by solid errors when I tried to run nikiml's Optimization Scripts for even simple one or two Variable searches (like Bowtie-to-Parabola and Bowtie-to-Reflector Separations).
I took another look at it and discovered that in 300ohm's "CM4251_Parabolic_CoarseModel" I needed to perfectly match some numbers that were slightly different in the 5th and 6th decimals (Wire numbers were reported by the nikiml's Opt. Script) and (very mysteriously) the FAT Element Radius numbers in the Parabolic Reflector (0.5 and 0.3125 inches) had to be changed to SYmbols before the Opt. Script would stop declaring a missing value for the first variable in the Target Function.
FYI: Geometry Editor is notorious for generating mismatched decimals. Many times I've methodically gone through and performed a Find/Replace with a value rounded off to only two decimals (TRICK: Only search for a number with the first three decimals...it will find ALL of the mismatches). This ALSO makes the 4nec2 File a LOT easier to read and facilitates aligning the fields into columns. [Of course, I much prefer using SYmbols, completely avoiding all of this numeric nonsense.....]
So I could then find the "optimum" Separation for the 9-El LPDA
from the Parabolic Reflector (see above)....it is a more "balanced" antenna, with generally flat frequency response, with about 1 dB higher Gain on lower channels, but loses 1-2 dB on the higher channels.
And the first of several runs T.B.D. for a Generic 2-Bay (Whisker) Bowtie with 9 Reflector Rods
(using 7 SYmbol driven Variables with a fixed Bowtie-Bowtie Separation):
I did not include this Generic 2-Bay in below Comparison Chart, since I don't think I've found the optimum...need to try different numbers of Reflector Rods to begin with.....and let the optimization run to completion, rather than aterminating after running for "only" one day & one night.....
I also analyzed three alternatives for a Bigger version of the current 2-Bay FEED
a) 20% bigger Solid-Triangle-Bowties, incl. the feedlines (ONLY)
b) 20% bigger Bowties and Grid Reflector, but maintain current Separation: Chapelruns' Example
c) 20% bigger Bowties, Grid Reflector and Separation (a COMPLETE 2-Bay Rescale)
I did NOT analyze a 20% larger Rescale for the ENTIRE Antenna (incl. Parabolic)
, which is what "SHOULD" be needed to truly shift the current Gain curve downward to NEW UHF Band....for obvious reasons... I hope this exercise illustrates that putting an Antenna with the highest Gain and F/B Ratio in front of the Parabola doesn't ensure good results....it is heavily interacting with the elements in the Parabola...so we need to do a JOINT Optimization on the entire structure to find the "best" FEED dimensions and placement.
I think a JOINTLY OPTIMIZED YAGI FEED
may be needed to finally improve on the original....runs T.B.D.....
BTW: My model for CM4251's 2-Bay Solid-Triangle-Bowties with Reflector Rods
(using SYmbol driven Variables) still refuses to be run by nikiml's Opt. Scripts, so I've been frustrated from determining the "optimum" Triangle Length, "Tine" Separation, Reflector Rod Length/Seperations, et.al.....Grrrrr....
Comparison Chart for these Parabolic modifications can be found here (I'll let the Chart do all of the explanations):