Re the element to boom plates/clamps. I've found they make a noticeable
difference with NEC4. However, they are hard to model since the Lawson
substitute element diameters are so short to cause NEC 2 or 4 to not
like the segment length. They also have a big diameter step which NEC2
doesn't like. Lawson states his taper correction formula is based on
small diameter steps. I conclude that those who swear by insulated
elements do so to not have to swear at plates in the modeling.
Lawson (page 7-12) shows a 0.5% increase in resonant F for a 6x4"
element to boom plate on a 1" diameter 46Mhz dipole. Your measured
21375KHz measured vs 21140KHz modeled with no plates which is a 1.1%
change in freq with perhaps a larger plate for a 3" boom, but at half
the frequency. Direction ok, magnitude seems high. Does the plate
effect add up with 5 elements? Perhaps.
So if "relative results" (does that mean range vs model?) are better by
leaving them out using NEC2, it may be some combination of these
modeling problems.
I'm not as sure as you are that 2 out of the 3 variables (swr, gain,
f/b) fully define the performance of antennas over a band. Sure, f/b
and gain correlate, but the shape of all three curves vs frequency is
another matter, with many possibilities. Especially after seeing dozens
of end point "triple curve plots" that the AutoEZ optimizer has
generated after many thousands of trial designs.
Congrats on having set up a range - a significant accomplishment.
73,
Grant KZ1W
On 5/30/2016 11:10 AM, StellarCAT wrote:
agreed - but I have modeled my 6 element 15 and 5 element 20's. They
are conversions using a 15M6-125 to create a high strength 15M6DX (as
it models much better, especially in a stack) which I call 15M6DX-125
... and using a 20M5 to make a 20M5LGS-100 ... again better
performance version ... and I have range tested these. I don't have a
way to measure gain but using both SWR AND measureable peak FB points
I can see how and if it agrees with the model (and assume, I believe
correctly, that having these 2 variables means the gain will follow as
well).
So does it agree with the model? Generally it does not - or at least
not absolutely - it does relatively ... so the model is off in
frequency ... or stated another way using the dimensions from the
manufacturer produced results that were frequency-wise different from
the model - but the patterns are the same. So the only change is a
shift in where it all is. I'm sure the gain is also appropriately
placed when I've done this shift.
Also I'll note, and leave it up to others to debate, I built the 15
exactly (no worse than 1/8") to the M2 manual - the only differences
are the boom size going from the lighter weight 2" - 2 1/2" used on
the standard 15M6DX to the -125 boom (3") ... and the clamps are
different.
Now I've found that clamps seriously mess up the models - I find it is
more accurate in results (again relative results) to leave them out -
as well as the swaging (my first models included it) and I couldn't
find a clean way to model the boom with NEC2 but using Lawsons boom
compensation values it should be trivial.
The results? The 15M6DX showed a FB peak at 21375Khz! I had to
lengthen all elements by 3" each to get it to I believe 21140 where I
wanted it. The peak FB was ~48db b4 the change and about 44 db after.
So - are their dimensions actually producing high results? Or are the
clamps and/or boom causing THAT much of a change (I don't believe that
at all).
Finally the 20M5LGS made form a 20M5. Here the boom is the same, 3",
the clamps are different but because its 20 meters (lower frequency,
less percentage difference) I suspected they would have little effect
overall. I built exactly to the manufacturers dimensions ... the range
testing showed very close results re FB peak frequency and
distribution as well as an SWR curve that was nearly identical to the
model ---- with again the caveat that the model was at a different
frequency! [I was able to make one single change to the manufacturers
dimensions - I changed the hairpin moving it by 1" and that improved
the already very nice SWR curve making it no more than 1.30 across the
band).
So my take is that these tools are not useable as an absolute tool -
they are accurate in their results but the relationship of the results
to the actual frequency is relative and needs to be compensated for.
oh - and I'm told - if the elements were isolated and I was using NEC4
that the results would be accurate both in a curve by curve comparison
(model to actual) as well as absolute frequency placement... of course
starting with all M2 antenna I wasn't about to isolate them - and NEC4
is a pretty penny additional.
Gary
K9RX
-----Original Message----- From: Grant Saviers
Sent: Monday, May 30, 2016 1:45 PM
To: Jim Thomson ; towertalk@contesting.com
Subject: Re: [TowerTalk] [Bulk] Change in Frequency As Antenna Height
Rises
Ok, another cut at this with a much more complex model - 6 element 17m
beam on a 48' boom. (I happen to have that model in EZNEC Pro4 and
AutoEZ). Finding "resonance" isn't of much interest. What is of
interest to me is the change in swr, gain, and f/b over changes in
height for a design I've been optimizing.
Here is a verbal description of results:
1. Gain changes smoothly upward from 0dbi at 5' to 16.7dbi at the
optimized height of 100'.
2. SWR decreases from 2.6:1 at 5' to 1.3:1 at 15' and then decreases to
design height value 1.16:1 by 30' up.
3. Minimum f/b is 28db vs the optimized value 30db also at 15 ft.
However, maximum f/b increases significantly at selected frequencies as
the antenna is elevated.
A quarter wavelength on 17m is about 14ft, so this modeling of this beam
would indicate that a bit more than 1/4wl height is sufficient to
predict results at much higher elevations.
My modeling (as G3TXQ comments) shows that max gain, min swr, and max
f/b all happen at different frequencies. So, "tuning" a beam with swr
seems like a trip to Vegas re what the performance will be. I think it
is at best a quick check that there are no gross assembly errors. I've
had the AutoEZ optimizer generate beam models with 1:1 swr, but with
reverse patterns, straight up patterns, no gain, etc.
So for beams, I think good models/modeling provide the correct
dimensions. The above results were from EZNEC Pro/4, dbl precision,
driven with AutoEZ to generate 250 test cases, frequency x elevation. I
wouldn't try analyzing this without AutoEZ.
Grant KZ1W
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|