Although what follows is in regard to an inverted-L antenna, not a Tee, it
might be of interest.
My L is composed of telescoping aluminum tubing starting with 2 1/2" OD and
tapering to 1 3/8". The total height is 55' 10". From the top of that runs a 14
AWG copper wire that is 90' long, with the far end at ~45' above ground. The
vertical is insulated at the base and driven via a 6" 10 AWG pigtail against
(currently) twenty-four 55' long insulated radials laid on the ground and four
11', 6 AWG bare wires running to four 8' ground rods in a 16' square. All of
these connect to a DX Engineering ground plate about an inch and a half above
the dirt. The 1/2" Heliax connects to a type N connector mounted to the ground
plate. All in all, rather unremarkable, except for the surprising amount of
money invested, a good part of that the cost of the DX Engineering foldover
mount. (What happened to the beer-can-vertical with the Coke bottle insulator?)
Measuring the feedpoint impedance at the N connector on the ground plate using a
DG8SAQ VNWA shows Z = 29.2 j0 at 1850 kHz.
I created a model of this in AutoEZ invoking EZNEC+ V. There are some issues
with this however. Normally one would use the built-in stepped diameter
correction, but this only works when all segments are collinear. The horizontal
wire isn't. Placing a source on the 6" long wire is problematic since there is
a huge difference between the wire and the tubing diameters. Segment tapering is
a fix, but a 6" length is already too short to satisfy guidelines. The
compromise solution is to eliminate the 6" wire and connect the tubing to
"ground" and place the source at 0% from the end. Likewise, to simulate ground
loss with the Mininec type ground, a resistance is also placed on the bottom
wire at 0% from the end.
Using the AutoEZ optimizer with the simulated ground resistance and the length
of the "L" wire as variables, I let it adjust the variables to get the same Z in
the model as the measured data. By plotting this on the Smith chart and then
saving the result as an S1p file I was able to import that file into the DG8SAQ
program and overlay it on the measured data. For the limits on the 2:1 VSWR
circle (~1.8 to 1.9 MHz), the traces overlaid each other very nicely. The
simulated ground resistance to bring this about was 15 ohm..
The length of the horizontal wire in the model was 83.1 feet for a total length
of ~139 feet. However, the physical wire is 90 feet long for a total length of
~146 feet, a considerable difference. Both of these dimensions are longer than
the ~133 free space quarter wavelength perhaps indicating that the radial field
is still resonant and at a higher frequency. This is something that I can't
model with the NEC-2 engine. Severns mentions this in "Experimental
Determination of Ground System Performance for
HF Verticals Part 4 How Many Radials Does My Vertical Really Need?", QEX
May-June 2009. But he observed a change in resonant frequency depending on the
number of radials. I have not seen this, only a reduction in the real part of
the feedpoint Z with more radials.
The point of all of this is that modeling is a great tool and I'm a firm
believer in it, but it has its limitations.
FWIW,
Wes N7WS
On 2/3/2018 6:44 PM, Richard (Rick) Karlquist wrote:
You should really model this on EZNEC.
Having done this many times, I can say
that the length of the vertical plus the
length of one of the top wires should be
about 120 feet to hit exact 1830 resonance
without a loading coil. However,
there is absolutely no compelling reason why
this is necessary. A loading/matching coil
is far more convenient if you keep the wire
length to 25 feet. Therefore, it should
never be necessary to put it up and down a
bunch of times to get it "just right". There
is no "just right" with respect to the antenna
proper. Use the taps on the coil to do the trimming
to make the system just right. This has been
discussed many times on this reflector.
Rick N6RK
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|