From: Tom Rauch <w8ji@contesting.com>
>We also have to be very careful with models, because the soil we have
around
>an antenna is virtually never homogeneous, and almost always not even
>remotely close to homogeneous. It is a widely varying mix of moisture as
>well as soil types.
>
>Because of this, any model that treats the earth as a uniform media will
not
>be "accurate", even if we consider accurate + - 50% as "accurate".
-------------------------------------------
I certainly can't say in all cases that NEC-2, NEC-3, or NEC-4 will give
comparable accurate results (electric field magnitude and phase for example)
to actual measured data when using a uniform soil conductivity and relative
dielectric constant in the model (derived from say an average of some
measured values), but I have personally made measurements in the field and
compared them to a NEC-3 model and obtained surprising results for one very
specific HF example.
About 13 years ago, our group was looking at vertically polarized radiation
behavior of horizontal HF dipoles located near the ground. We had access to
HF antenna ranges, calibrated constant power sources, meters, calibrated
Rohde & Schwarz field strength meters, HP network analyzers, IBM mainframe
computers (for NEC), the whole ball of wax. SRI out of California provided
measuements for soil conductivity and dielectric constant (don't know or
remember their test method off hand) near the HF test antenna. Average
values of the SRI measured conductivity and dielectric constant were used in
the NEC-3 model. We measured far-field vertically polarized electric field
magnitudes off the ends of the dipole (yes, it is there and does exist!...
similar to beverage behavior) for various heights of the horizontal dipole
above ground ranging from 2 feet to 20 feet at 18.1 MHz.
Here is where NEC data must be carefully reviewed by users no matter what
the case or model. I don't know what EZ-NEC or any of the other graphics
based "consumer" products are spitting out, but we worked strictly with the
raw E-field data in the NEC output file before we plotted the mess!
Specifically the vector components of the Electric field (E-theta: the
vertical component, and E-phi: the horizontal component). Here is where
most hams will go wrong with comparison of measured data to NEC generated
data (I'll specifically address the vertical component here).
Unless specifically manipulated by another program or computation, NEC is
generating the spherical vector component E-theta (for all theta, phi, and
fixed r) of the electric field in one of its columns of the output file, not
an orthogonal (normal to the earth, E-normal) component that you would
measure with a vertical test antenna at some distance "r" from the source,
and some height "z" from the x-y plane. Granted, as the spherical vector
component E-theta approaches theta=90deg, E-theta would in effect be the
"normal" E-field value at exactly theta=90 deg (on the x-y plane). However,
is the field strength test antenna in question located on the x-y plane at
theta=90deg and z=0, or is it located at some other height above ground?
This is especially significant if lobes are changing rapidly. In some
cases, the actual E-field magnitude can change significantly (the pattern is
obviously changing anywhay) as an observer rises slightly above the x-y
plane (lossy interface border) a few meters, thus making it very important
to know at what angle theta from the z-axis the test antenna is located.
This leads to point #1:
1. When comparing a measured "vertical" electric field magnitude to a NEC
model generated "vertical" electric field magnitude, you must mathematically
extract the "normal" component of the NEC E-theta component at the
appropriate spherical point (r, theta, phi) in space where the field
strength test antenna was actually located! They may be almost equivalent
in some cases, but you must still do it to be mathematically correct. The
higher the field strength test antenna above ground (x-y plane) the more
significant this becomes! This is assuming your test antenna was normal to
the earth's surface, which is usually the case. Any tilt to the FS test
antenna, and this becomes more of a polarization match factor nightmare, and
measurements are even more difficult to compare.
The other issue is the radiated surface wave! Vertical antennas (and
horizontal wires near the earth's surface) have, as a part of their total
radiated vertical electric field, a surface wave E-field component (Zenneck
wave). Assuming you are not a great distance from the antenna, this
component is significant (depending on the antenna geometry, frequency, and
soil conditions) especially on MF and 160! If a person makes a field
strength measurement at some distance r from the transmit antenna, they are
probably measuring a vector sum of a surface wave and a space wave
component! Here again, if the receive test antenna is orthogonal to the
earth's surface, we only measure an orthogonal component of the E-theta
space wave component summed with the surface wave (which is usually
orthogonal unless some tilt has begun to occur).
The version of NEC we used (NEC-3 and others I presume) does not include the
surface wave radiated component in the NEC output file when you run a normal
card for all points of theta and phi and some distance r! You had to run a
seperate surface wave card (using the Norton-Sommerfeld ground
approximation) to get this data! This is a significant issue when comparing
E-field data measured near the earth's surface!!! NEC-3 would not, as far
as I know, allow the user to run both cards and merge the data into one
vector sum field...... Therefore.... We had to run them seperatly and write
Matlab programs to extract E-normal from E-theta in point #1 above, and then
add that vector component back to the surface wave component generated by
the surface wave card. Point #2:
2. The surface (zenneck) wave must be accounted for when running a NEC
analysis, and then must be added back to the E-normal component of the space
wave results! Then you get the total "normal" field (vertical component)
near the earth's surface! Then you can compare it to what you measure,
assuming sigma (conductivity) and epsilon-r (relative dielectric constant)
have been measured accurately. Here again, this is only important if you are
making measurements near an interface where a surface wave can be supported.
Way out in free space at smaller angles of theta (high elevation angles)
there is probably no surface wave component.
When we straightened out all this mess, we were able to generate NEC-3 model
electric field magnitudes that were within 1.5 dB (some within 0.5 dB) of
what was measured on the Rohde &Schwarz calibrated receiver for the specific
test case low dipole (for various heights above ground). I'm not saying the
SRI ground constant measurements were absolutely correct, nor am I an expert
at how they were made. Nor am I claiming that our test apparatus was
absolutely without fault. But for our small test case, the results were
surprisingly close, and we learned how significant the surface wave
component was in comparing NEC data!!! It wasn't even close until we
accounted for it!
I wished we had time to gather more data for other models (i.e. verticals)
and many other azmuthal directions around the test dipole where E-theta and
the surface wave is weaker. It took a lot of expensive equipment,
computers, and time to get the one set of small data, but it was
encouraging. The Matlab vector transformation programs we had to write were
a little tricky but not out of reach of most hams.
Anybody else out there working at it from this angle? I haven't seen one
word about this issue in any Ham mag or news group (with respect to
verticals, beverages, wires, etc where applicable), but I see lots of sloppy
NEC analysis, and resulting NEC negativity. How can we even talk about NEC
analysis (pro or con) without a detailed discussion/treatment of the NEC
surface wave analysis issue (everybody is leaving it completely out of the
vertical radiation analysis!) and the possible mistreatment/confusion of the
E-field output data (i.e proper coordinate system comparison/transformation
w/r to field strength measurements)?
Rick, WC4Q
|