Satellite Antennas for Ham Radio With Switchable Circular Polarization

This is how I built my right or left switchable, circularly polarized crossed Yagi antennas for amateur radio satellite work, for 2 meters and 70 cm. The pair works well. In six months I’ve already reached 41 states. I use a small relay in each to change from right handed circular polarization (RHCP) to left handed circular polarization (LHCP). If you keep it simple by just using RHCP without the relay and the extra phase delay lines you should do well. Most satellites use RHCP, and most commercial Yagis are built only for RHCP, but there are times when switching to LHCP improves the signal. (It can make a huge difference, and the signal can change in less than a minute!)

I designed the plastic parts with Autodesk Fusion 360, with their limited free version for interested hobbyists.

I started with an inexpensive 9 element Yagi from China on Aliexpress for $29. I can’t find it any longer, so I took measurements that you can use to make your own. It uses a 3/4” square aluminum boom, about 46” long. It uses 9mm (0.35 inch) aluminum tubing for the elements. You can get a pack of 12 tubes, 36” long, on Amazon, for under $30. (I used these to make my 2 meter Yagi with the spare boom.) If you look around online you can find a 48” square boom piece for about $11. I bought two of the cheap 430 MHz Yagis and cannibalized the element pieces from one to use as the cross elements on the other, and drilled secondary holes at 90 degrees on the boom 1/2 inch in front of each primary element. The driven element came with a standard adjustable gamma match, but I used a different driven element configuration for the 70cm, for reasons I’ll describe later, without a gamma match, and to simplify construction.

The 2 meter Yagi uses the spare boom, 46” (117cm) long, and the elements are 9 mm outer diameter. I used mmana-gal basic, a free program you can find online, to model a Yagi. That was a starting point. The numbers I got are close to the actual dimensions I finally used, except for the driven elements. The main reason for that has to do with how I mounted the driven elements for both antennas: insulated from the boom with plastic mounting brackets, made on a 3D printer. (See my stl files.) The reflector and director elements can remain shorted to the boom and work perfectly as in any Yagi, if you like. But the reflector on the 2 meter Yagi has to be longer than 36” (92cm), so I used two shorter tubes mounted to the boom with a plastic bracket, and a jumper wire to make the entire element longer than 36”. For me, finding a reasonably priced tube longer than 36” was difficult. Of course the electrical length of two 9mm tubes with a 2” jumper wire is probably different from a long 9mm tube, so the mmana-gal numbers may not be perfect, but a careful estimate yielded good results, as you will see from the antenna patterns of the finished antennas.

Insulating the driven elements from the boom and each other is necessary if you want to generate circular polarization. Circular polarization of an rf wave propagating through space occurs when two dipoles in the same plane but at right angles emit the same sinusoidal wave but with one dipole lagging 1/4 wave behind the other. It’s pretty simple to do that by having the second dipole fed by a longer feed line than the first—1/4 wavelength behind. But if you use coaxial cable you have to make sure that both of the legs of the second dipole are delayed—not just the center coaxial conductor, but the shield as well. So you can’t have all the shield connections shorted to the boom or each other. Just as in a turnstile antenna, the two dipoles have to have the shields separated beyond the Y-split from the common feed. That’s why the driven elements need insulating brackets. I used PETG plastic for printing them. In the 70cm Yagi the driven elements are mounted along an “X” bracket. In the 2m Yagi I built a gamma match, but using a plain, balanced dipole might have worked well too.

The brackets with jumper wires for the 2 meter reflector worked so well that I used them also for the two 2 meter director elements. One advantage is that the brackets slide easily forwards and backwards on the boom so that you can use that feature for final adjustment before securing the bracket positions. Naturally the driven elements’ anterior-posterior positions affect the performance the most. For the 2 meter Yagi I made my own gamma matches, but I used a different mounting bracket to keep each driven element with its own gamma match insulated from the boom and also from each other. It turns out that RG8X coax slides snugly into the 9 mm tubes, which is great for making the tube capacitor for each gamma match.

For the 70cm Yagi, the elements are as follows:

Element Full Width

Reflector 38.4cm

Driven Element (see below)

Director 1 29.2

Director 2 26.2

D3 23.2

D4 20.3

D5 17.0

D6 14.0

D7 11.1

Depending on the type of wire or tubing used for the driven element, the velocity factor may make the resonance different from what you might calculate. For example, I first used RG8X for the driven elements of the 70cm Yagi, because it’s easier to trim and replace than an aluminum tube, but the polyethylene dielectric reduced the velocity factor appreciably. It shortens the elements. I switched to a copper conductor along a PETG scaffold. PETG doesn’t change the velocity factor as much as polyethylene. A longer element has a bigger receiving cross section. Just use a conductor that’s easy to trim, fix it to the plastic mounting bracket which fits snugly on the boom, start long, and trim carefully. One other thing: tune each dipole in isolation apart from the other elements and off the boom. I set mine for 437 MHz. The reflector and directors will passively interact with the driven element (that’s why they’re there) and make pruning much more difficult. Then put the driven element bracket back on the boom. Here are the spacings for my 70cm Yagi:

Distance forward from the previous element:

Reflector 0

Driven element 14.3cm

D1 11.7

D2-D7 11.7

Then, with the second boom of the same length, I constructed the 2m Yagi, with aluminum tubes, centered on 145 MHz. In this case I made the driven elements with the 9mm tubing, and built gamma matches for each. Again I used printed plastic brackets to keep both primary and secondary dipoles electrically isolated from each other beyond the Y-branching split away from the antenna feed point, and to keep the dipoles 90 degrees apart in their phase. And each dipole is insulated from the boom. The element lengths, including the 2” (5cm) center jumpers, are as follows:

Element Full Width

Reflector 99cm

Driven Element 91.4

Director 1 90

Director 2 89

And the distances forward from the previous element:

Reflector 0

Driven Element 35.5cm

Director 1 40.6

Director 2 35.5

Resonances of the dipoles hanging in free space between trees, were tested and adjusted (approximated by looking at SWR with a nanoVNA). In the 2m Yagi, hanging suspended in free space, the spacing of the elements was adjustable with the plastic brackets sliding along the boom. This regime is a compromise, but the antennas work. When each was completed and mounted on the rotator the graphs show good patterns, and they perform well in practice.

Now the next task is to make it a Yagi with circular polarization. As mentioned above, to make it a permanently right hand circular polarization antenna, which is fine, you can simply measure the distance from the branch point where the feed goes to each dipole separately, and just make sure that the electrical feed length to the secondary dipole is 1/4 wavelength longer than the length to the primary dipole. Remember to correct for the velocity factor of the particular coaxial cable you’re using. Now you have to pay attention to which leg of each dipole is connected to which connector of the coaxial feed. Here is the convention: from the rear of the dipole, look forward in the direction of the major (forward) lobe of the Yagi. Look at the primary dipole, the one that has the shorter length of coax from the main split point of the feed line. Find the leg that is connected to, say, the center conductor. Then if the leg connected to the corresponding center conductor feeding the secondary dipole (the one with the longer delay line) is 90 degrees counterclockwise to the center conductor of the primary dipole, that antenna will have right hand circular polarization (RHCP). If it’s 90 degrees clockwise from the primary dipole, that’s left hand circular polarization (LHCP). Almost all satellites are built for RHCP.

There are a few operators who switch the polarization of both uplink and downlink antennas from RCHP to LHCP and back to RHCP on the fly. I have heard the difference on the downlink and decided to try it myself. The following is a snapshot of how I did it. It’s more difficult but it paid off with impressive results.

In short, you have to switch the primary dipole from being 90 degrees ahead of phase compared with the secondary dipole, to making it 90 degrees behind the secondary dipole. Of course, the secondary dipole already has a hardwired 90 degree delay soldered in. So you can use a relay to switch in a 1/2 wave extra delay to the path of the primary dipole. This flips the phase relationship from right to left circular polarization. Also, since one leg of the primary dipole is fed by the center coaxial conductor and the other leg is fed by the shield, you have to switch in a full delay for both the shield and the center conductor. This means you have to use a 4PDT relay, half for the center conductor and the other half for the shield. See my schematic diagram. I have seen schematics from a commercial antenna manufacturer for a relay kit to do this, but only with a DPDT relay, and only for the center conductor. You really need a 4PDT relay. Without it you’ll introduce a big phase imbalance between the legs of the primary dipole when you switch in the LHCP delay line.

One criticism might be that you need an RF rated relay. I experimented with a small 4PDT unit, two for less than $7 from Amazon, rated at 5 amps for DC. I run 100 watts on 2m and 70 on 70cm. Another possible issue is potential problems with isolation and crossover currents. I use one for the 70cm and another for the 2m Yagi. It has worked well so far.

The phase delays have to be fairly precise, especially on 70cm. How long is the phase delay contributed by the path through the relay? It can’t be ignored. Another problem is that you can’t always trust the published velocity factor of your coaxial cable. You can’t just measure with a ruler if you use a relay. You have to measure it with a nanoVNA, and at each step of the construction of your delay line harness. I found through many different measurements of different coax lengths that my brand new RG59 had a velocity factor of about 0.71. (The published values are 0.66 for solid PE and about 0.84 for foam. Mine must be slightly foamy.) And the true electrical length through two poles of a double throw relay connected through a short stub is more than 6 inches, so careful attention is important.

Here’s how to measure and prune the coax: first, make sure the nanoVNA is calibrated for every test lead and every frequency sweep for each type of test. Just store them, but if you change your test lead you must recalibrate. Now, instead of testing in the SWR format, use the PHASE format. Connect the nanoVNA to the length of coax you’re testing, center to center conductor and shield to shield, and with the far end of the coax left open. One oversimplification is to imagine that the nanoVNA sends out pulses and measures the time it takes for reflections to return. Reflections will return from every point along the line where there is an impedance discontinuity, whether it’s from the open end of the coax, a length of coax with a different impedance, or a simple relay from Amazon. The nanoVNA will display a vertical jump at every sweep frequency corresponding to a 1/4 wavelength of cable at that spot on the sweep. A longer length of coax represents a longer wavelength and so the jump on the screen will read at a lower frequency for that distant reflection, compared to the screen trace jump corresponding to a reflection from a spot closer to the nanoVNA test point. You can map out the lengths of each bit of coax, where the relay is, how long the path is through the relay and the length of the next bit of coax, and so forth. See my nanoVNA photo. Write these frequencies down. Then get a calculator, take 29980 (adjusted speed of light) divided by the measured frequency in MHz. Then divide that by 4. The result is the electrical length in cm of the spot along the line where the nanoVNA’s signal bounced off. Now if you trim that coax to where the jump on the nanoVNA reads, say, 437 MHz, you know that that length of cable is 1/4 wavelength at that frequency, regardless of the true velocity factor of the coax or relay or whatever the wave is traveling through.

I recommend that you work backwards starting from the point of connection of the primary dipole itself (without the dipole connected yet) to the where the relay will be mounted, but without the relay yet connected. This is constricted physically along the span of the boom between the driven element and the position where the relay will be mounted. Use the nanoVNA to measure the true electrical length of that stretch of coax. Then attach the relay but only that end, and in the unenergized mode (short RHCP path). Measure the path through the coax and the relay. The difference will be length of the path through the relay. Then attach the next length of coax, heading back towards the transmitter, to where it will connect to the feed line to the secondary dipole in a Y-split, but don’t yet connect it to anything else, leaving that coax end open. Measure that combined length with the nanoVNA. Whatever that length is (the path from the common feed point through the relay to the primary dipole), now you know also what the electrical length must be for coax going straight to the secondary dipole. It must be 1/4 wavelength longer at 437 MHz. And you really should measure and trim that new length of coax to that precise length also using the nanoVNA. You can’t just get a tape measure and use the published velocity factor for the coax.

Then finally you have to measure and cut the length of coax that is switched in to the make a longer path to the primary dipole for the LHCP state when the relay is activated. Cut a length of coax longer than 1/2 wavelength at 437 MHz with a best guess for the real velocity factor you’re dealing with (you should have a good idea by now), and connect one side to the relay. Then activate the relay so that the 1/2 wavelength stub is connected into the path to the primary dipole. Little by little trim that length down while then temporarily attaching that other end of the long coax stub to the relay, both shield and center connections, and measure it with the nanoVNA to where the total true electrical length from the future Y-branch point (again without the line to the secondary dipole yet connected) to the feed point of the primary dipole, is precisely 1/2 wavelength longer than the short path in the deactivated (RHCP) mode. It helps to have a pad, a pen and a nearby calculator in one hand, and a good non-dominant hand to hold the coax to the relay. A nanoVNA with a decent sized screen is also a plus. Once you’ve got all that done, complete the soldering at the relay and don’t forget the heat shrink tubing. Connect the proximal side of the coax-relay-coax system going to the primary dipole, to the proximal end of the 1/4-wavelength-longer line of coax headed towards the secondary dipole. Then connect the distal ends of each coax line to its respective dipole. You’re just about ready to connect the common feed branch point to the 50 ohm feed line to the shack. But wait…

It turns out that if you use 50 ohm coax for the phasing harness, it won't present a 50 ohm match for the main feed from the shack. And it’s not just a simple question of two 73 ohm dipoles in parallel cutting the impedance to 36.5 ohms. If the dipoles are not in phase the impedance can be different, and if they are in the same plane there may be coupling between them reducing the impedance. It turns out that a better match for this system is to use 93 ohm coax (I used RG62, which available at my local electronics shop) for all the connections within the phasing harness, and feeding it with 50 ohm coax from the shack. Please see my SWR charts for my 2 meter antenna. This gives an impedance very close to 1:1 SWR near resonance. If you can't get RG62, an alternative is to use 75 ohm coax for the harness (RG59), and with a 1/4 wave coax transformer between the 50 ohm main feed line from the shack and the starting Y-point split between the feed to the primary dipole and the feed to the secondary dipole. The transformer is 37.5 ohms, made by putting two 1/4 wavelength pieces of 75 ohm RG59 coax in parallel. This should theoretically yield an SWR around 1.2 to 1.4 near resonance, which is in fact what I have observed. I have tried both versions of the harness for my 2 meter antenna, and I may switch to the RG62 also for my 70 cm antenna, but I currently use 75 ohm coax for the latter. Both versions of the harness are effective.

You can control the relays by running a couple of extra cables to the shack with toggle switches. But I used WiFi. My rotator already has +12 volts and a WiFi Raspberry Pi 4 at the masthead, so I installed a LAMP stack and built a two channel 2N2222 switch to get two of the GPIO pins to drive each relay independently. The web page shows two buttons, one reading “2M R” in green, which switches with a mouse click to “2M L” in red, and the other which can toggle “440 R” green to “440 L” in red. See my index.html, toggle.php and toggle_gpio.py files, which go in /var/www/html. Make sure the file ownerships and permissions are appropriate.

(note: there is a problem —a deliberate restriction on this website— with uploading two of the files, index.html and toggle.php. The python file is fine as is. For the other two, I was only allowed to upload them as Rich Text Format files, with the .rtf extension. You need to download those two files, then open them in a text editor, and save them as plain text files without the .rtf extension. Make sure they are saved without a hidden .txt extension, which many file browsers hide. It’s meant to be helpful to hide the .txt extension but it causes more problems than it solves. Therefore index.html must not be index.html.txt and toggle.php must not be toggle.php.txt. )

There. Easy, no? Now do the same process for the 2 meter Yagi. All you need is time, patience and and understanding wife. All told, this was much faster than constructing my hand made Az-El rotator. You can find my article on that on Instructables.com at

https://www.instructables.com/Antenna-RotatorController-Azimuth-Elevation/

If you love designing and printing 3D parts that really do something, I suggest you take a look.

Lastly, you should put slip-on ferrite beads over your coax feed coming into the Yagi to block common mode currents down to the shack. I use two long mix 31 ferrites for the 2 meter Yagi, and three shorter mix 61 ferrites for the 70cm Yagi.

How well does it work? First, take a look at the gain pattern polar plots for each antenna. 2 meters is in blue, and 70cm is in red. I measured received field strength in dBm of local repeaters in 10 degree increments, measured with an SDRPlay SDR receiver.. The values were very similar between RHCP and LHCP, and so I averaged the readings at each declination. The received signals were adjusted in Excel so that signal strengths are plotted on a relative scale with the weakest signal is about 1 and all other signals are in dB above that. I am satisfied with the result.

For 5 months I used these antennas in fixed RHCP configuration before adding the phasing relays and had already made almost 1000 contacts, mostly SSB and some FT4 and FM, and even a few satellite contacts on CW, in 41 states. How much better is it being able to switch on the fly from RHCP to LHCP and back to RHCP? In most cases it does make a difference, often a tremendous difference, and what is really remarkable is that the relationship can flip on a single pass of a single satellite within 45 seconds and then back again. This is true sometimes of my uplink and sometimes my downlink. It can make the difference between good copy and no copy, RHCP or LHCP, and unpredictably. I suspect that the polarization of the tumbling satellite and the changing polarization from propagation of the signal through the ionosphere in a magnetic field is occurring, which Faraday observed in polarized light almost two centuries ago.

See you on the next satellite pass!