When I get the time (sigh) I want to try a co-linear for ADS-B, among other things.
Here’s what my experience with VHF/UHF antennas has taught me:
Ground planes: simple and effective. 5/8 wavelength can produce some gain. Big hole directly overhead. not very frequency selective
Co-linears: need to know velocity factor of line used. simple and can provide some gain. Hole overhead increases with gain. somewhat frequency selective
Turnstiles and eggbeaters: little gain, circular polarization, no hole overhead – great for sat work. frequency selective
Corner reflector dipoles: simple, effective, lots of gain in an identified sector. cheap. somewhat frequency selective
Yagis: effective, gain increases with number of elements, and so do sidelobes! very frequency selective!!
Discones: very wideband. takeoff angle increases somewhat with frequency. size of hole overhead decreases with frequency. hub is tricky, the rest simple.
That hole on the top – the radiation pattern of vertical antennas such as ground planes, co-linears, etc, all have a significant signal drop off directly above the antenna. As antenna “gain” increases, the antenna pattern goes more and more horizontal, decreasing pickup at higher angles. You end up with antenna which can hear really well at certain angles but is stone deaf overhead.
Remember that we are working with a system. Changing one part of the system can alter the performance of the rest of the system – that’s what we’re trying to do by changing antennas. Those changes can be positive, neutral, or negative. We need a way to measure the change in system performance, which also needs we need to measure baseline performance. PiAware gives us measurements: birds spotted and positions reported. We can use these measurements to track performance changes. We learn not just from changes that improve performance, but also from those which decrease performance, or don’t seem to help or hurt.
I’ve got a commercial discone on one end of the roof (and a 2-meter eggbeater on the other end). It works a treat for ADS-B, but needs filtering so as not to overload the poor little SDR. Mesured coverage and performance is pretty good, terrain limited on the West side, but plenty of (good) position reports in the 200nm range.
This suggests to me that antenna gain isn’t my big problem. From the sat work I’ve done, I’m interested in what I may be missing directly overhead. That strongly suggests an eggbeater or a turnstile as the next experimental antenna – no gain, but no hole overhead, either. Also somewhat frequency selective, so filtering requirements may be less.
Next on my list to explore is improving coverage of flight paths coming in from the East. Geometry/topology suggests I should have better results in that direction; I suspect the answer is traffic collisions. This suggests going sectorized with a corner reflector pointed East. Doing that will help overcome the collision problem. In other communications systems, a sectorized model allows support of more stations on the same frequency.
I’ll try a co-linear – but – while gain increases with the number of elements, the net increase goes down with that number, and the antenna pattern narrows further, increasing that hole on the top. The center frequency and net increase in gain is also dependent on carefully matching the length of each section. My scientific wild a** guess is 4 to 5 sections max, terminated to provide a DC path to ground.
I read with interest some of the antenna threads on this forum where someone goes from a multi-element co-linear to a ground plane and reports an increase in performance. How is that possible? Hypotheses include: (1) too much gain from the co-linear overloading the SDR, (2) poorly constructed co-linear making a total hash of things, (3) carefully constructed co-linear antenna pattern cutting out a significant amount of traffic outside that narrow pattern, or (4) Murphy’s favouriite, something else entirely.
But the fun of working with antennas at this frequency is that they’re small and relatively easy to make (except for the precision required at GHz frequencies). You need to run them for a while to collect comparison data, but that’s part of the scientific method.
Keep having fun–
bob k6rtm
