Last year I got interested in the Radio Jove project that monitors
broadband RF emissions from Jupiter from about 15 MHz to 25 MHz. For
that purpose I built a 5 element wire log periodic antenna and supported
it from a 20 foot pole angled up at about 70 degrees from horizontal. I
modeled it in EZNEC and decided to try to verify the elevation pattern
with my drone. I put a small 9v battery powered 20 MHz crystal
oscillator on the drone and positioned it 200 feet away from the
antenna. The controller for my drone tells me both horizontal distance
and height above the launching point, so I did the calculations for
distance and height to keep the drone 200 feet away from me while I
stood under the antenna, and I had my wife in the shack take signal
strength readings every ten degrees in elevation angle as I flew the
drone up and overhead.
The log periodic is a horizontally polarized antenna so I just suspended
a short dipole below the drone, and using the camera on the drone I made
sure that the drone was always pointed directly toward me. The only way
I can think that it might be possible to do something like that to get
the azimuth profile of a vertical antenna would maybe to fabricate some
sort of 3-axis dipole (three dipoles all orthogonal to each other and
fed together) and suspend it far enough below the drone to minimize
asymmetric capacity interaction with the drone.
Anyway, when I plotted the relative received signal strengths for the
log periodic antenna, the profile almost exactly matched the calculated
pattern from EZNEC except for one data point.
It was only a very informal test and done only in one direction, but I
was kind of impressed with the capability.
Dave
p.s. I never managed to conclusively detect Jupiter emissions, but I
got some spectacular solar receptions. The free Radio Jove software
plots wideband results as a function of time, and the patterns can be
really interesting. If all you were interested in were solar emissions,
you could get some decent results with just a length of wire. If you
can hear the increase in noise, you can plot it.
On 1/15/2026 11:30 AM, Jim Lux wrote:
This, exactly.
There are people making measurements with drones - for instance, the pattern of
the 9 MHz antenna on Europa Clipper was measured with a drone (and with scale
models on a conventional range, etc.)
It is non-trivial.
Here's someone who did it in 2019 - HF radar antenna near field assessment
using a UAV
https://hal.science/hal-03306719v1/document
There are commercial companies doing this now for things like HF Broadcast antennas
(well, broadcast antennas in general) and UAVs have made it a lot easier to
get the raw data than hanging a probe off a helicopter like RELEDOP (1980s)
It's like any other near field measurement - you take a 3d grid of measurements in the near
field and then convert that to far field. The math is both simple and complex - it's
just a Fourier Transform in some sense, but in reality, the measurements are not on a
perfectly regular grid in a nice geometry, and there's all kinds of "probe
calibration" aspects.
There are full on modeling codes that take into account all azimuths, diffraction, etc.
As Brian points out, though, getting data to ingest is a challenge - Sure, you can get
a DEM from SRTM or Lidar data, but that doesn't tell you what vegetation is on that surface, nor
what the soil properties are. HFTA, being HPol, doesn't care much - it's all a reflection.
And there are people looking at doing Ground Penetrating Radar in a multistatic
sense to measure properties of the subsurface (The CADRE mission, headed to the
Moon eventually, has 3 little rovers carrying a GPR system based on the Ettus
E312 USRP and I believe there's a version flying them on UAVs)
https://www.hou.usra.edu/meetings/leag2025/pdf/5081.pdf
https://www.federico.io/pdf/DeLaCroix.Rossi.ea.AERO24.pdf
Here they are using it to measure snow depth using UAVs
https://rosap.ntl.bts.gov/view/dot/80070
Ultimately, though, it comes down to "do you want to spend thousands of hours
accumulating soil and environment data for limited improvement in the antenna system
performance"
Or, as Brian sort of points out - if you aren't going to have the input data
for the model, then why build a simplified version of the modeling code?
Interesting historical data here in N6BV's obit from last summer.
https://www.arrl.org/news/richard-dean-straw-n6bv-arrl-antenna-expert-silent-key
On Thu, 15 Jan 2026 04:23:57 -0800, Brian Beezley<k6sti@att.net> wrote:
K9YC wrote
"Now, reading that Brian had started it all, perhaps he might take it on."
TA, the terrain analysis program I wrote in the 1990s, handled vertical
as well as horizontal polarization.
I went to some effort to ensure that TA was accurate, including
comparing results against helicopter-borne radiation pattern
measurements. However, in the years since, I've become convinced that
ray tracing that considers only a single azimuth angle has serious
accuracy limitations that preclude its use in all but the simplest terrain.
Ground reflection and diffraction at any azimuth angle can wind up at
the angle of interest. Imagine what happens when radiation intersects
the slope of a hill off your target angle. Since even a directional HF
antenna has a broad forward lobe, it illuminates lots of ground away
from where it's aimed. Some of this power can come back to haunt you.
I've cautioned HFTA users with complex terrain about the limitations of
single-azimuth ray tracing. The response is invariably, "I know it's
accurate." When asked how they know, the answer is never satisfactory. I
think HFTA and TA blind users to their shortcomings by offering
fascinating and easily digestible results.
Ray tracing involves calculating power not only for direct reflection
and diffraction, but for reflection from reflection, diffraction from
reflection, reflection from diffraction, and diffraction from
diffraction. Then do it again for higher-order cascades. This must be
repeated over a dense elevation angle set to capture everything
relevant. The power of 1990s computers limited the speed of TA. I wrote
the time-consuming code in assembler to provide results in a reasonable
amount of time. Today's computers are much faster, have multiple CPUs,
and come with powerful vector instructions that can do eight
floating-point calculations simultaneously. Ray tracing over all
azimuths should be feasible today in a reasonable amount of time.
I've thought of writing a 2D (or is it 3D?) terrain analysis program.
But there's a showstopper: there's no empirical data to test it against.
Because the calculations are so complex, there's no way to ensure they
are correct without checking results against measured data for complex
terrain. As far as I know, none exists. I've searched for it and come up
empty.
I've thought about what it might take to make radiation pattern
measurements over complex terrain with a drone. But it's a complicated
problem with many hidden sources of error. When I was considering this,
each day I'd wake up with a new source of error that hadn't occurred to
me the day before. I think it would be easy to get in over your head
without ever knowing it. A computer program validated with fishy data is
not worth anyone's attention.
Brian
_______________________________________________
_______________________________________________
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
_______________________________________________
_______________________________________________
TowerTalk mailing list
TowerTalk@contesting.com
http://lists.contesting.com/mailman/listinfo/towertalk
|