Antenna Rotator/Controller, Azimuth-Elevation

This is a homebrew azimuth/elevation antenna rotator for amateur radio satellite use. It functions roughly in the same manner as a the commercially available Yaesu G5500 unit, although only for fairly light antennas. (My 70 cm/2 meter combination of crossed Yagis has a combined weight of less than 8 pounds.) It communicates via the widely used GS232 serial command protocol. One advantage is that it costs about 1/3 as much as the G5500. Another is that the cabling for the motors and control lines are simplified. Aside from the coaxial cables needed to feed the antennas it requires only one 12 volt power cable. It is entirely controlled by the WiFi LAN network through my home router.

The basic configuration was suggested by DO5TY in Germany ( https://tysonpower.de/blog/page/2 ). It uses two stepper motors, belt drives and worm gears and it has performed very well without mechanical problems. I made some adaptations including a vertical mast ball bearing support (a 3-inch lazy susan swivel from Amazon) and a more comprehensive microcontroller setup with open source software (K3NG Rotator, https://blog.radioartisan.com/arduino_rotator_controller/ ). It uses both an Arduino Mega and also a Raspberry Pi within the rotator cage itself. It works especially well in concert with the PSTRotator software package and the CNS S.A.T. box with the Icom IC-9700 transceiver, although these particular units are not required as part of the project.

1. 2x NEMA23 3 amp stepper motors
2. 2x TB6600 stepper motor drivers
3. 3X 6008 2RS bearings
4. 2x Aluminium tubes, 24 inch long, 1.5 inch OD
5. 1x 3-inch lazy susan ball bearing swivel plate (Amazon)
6. 2x Optical endstops
7. 18x 2020 Aluminium extrusion, each 20cm long
8. 8x inside 3 way corner bracket for 2020
9. T-Slot inside corners
10. M4 hammer nuts
11. M4x6mm or M4x8mm screws
12. 2x NEMA23 brackets
13. 4x 625 ZZ bearings
14. 2x M5x80m rods
15. M3 nuts
16. M3x10mm grub screws (18-20x)
17. 2x 40T GT2 pulley (5mm inner diameter for worm gears)
18. 2x 20T GT2 pulley (6.35mm inner diameter for motors)
19. 1x 20T GT2 pulley (6 mm inner diameter for potentiometer)
20. 1x 186mm 6mm wide closed loop GT2 belt for potentiometer
21. 2x 160mm closed loop GT2 belts for motors
22. potentiometer, 10-turn, 500 Ohm, 6mm shaft (3590S, Amazon)
23. Arduino MEGA
24. Raspberry Pi 4
25. 12 volt to 5 volt regulated step down DC power supply, 3 amp (e.g. PlusRoc, Amazon)
26. USB-A to USB-B cable (between Arduino Mega and Raspberry Pi)
27. numerous 3D printed plastic parts (see STL files)
28. 2x rubber automotive CV shaft boots cut to size for a measure of weatherproofing
29. a 3D CAD program like Fusion360 or FreeCAD to adjust part shapes slightly if needed
30. thread locker fluid to prevent screws from loosening

Most of the parts can be bought from Amazon, EBay, AliExpress and Home Depot. I modified and 3D-printed some of the DO5TY parts using PLA or ABS filament, and especially PETG for the higher stress parts like the worm gears. I adapted and designed some new parts using Autodesk Fusion360 CAD software, which has a free version for hobbyists. Expect to alter some of the part diameters slighty to account for subtle differences in plastic contraction and your particular build. Fractions of a millimeter count. Following DO5TY's basic project, the frame is based on 2020 aluminum extrusion struts in 20 cm lengths. It is held together with 3-way corner pieces. Additional cross pieces are mounted with 2020 L-brackets. These extruded construction pieces are wonderful but if you're new to the technique be prepared to tighten quite a few grub screws with hex wrenches! Once you’ve completed construction to your satisfaction remember to put medium strength thread locking fluid on the many grub screws. I used two NEMA23 (3 Amp, 6.35mm shaft) stepper motors. The vertical (azimuth) and elevation axes use 24 inch sections of 38mm (1 1/2 inch) aluminum tube. The top of the vertical tube is mounted on a 3 inch lazy susan swivel. The tubes require a thin 3D-printed bushing/adapter to make a snug fit to the bearings. The bearings, along with many other parts are connected to the cage using 3D-printed brackets, which are in turn bolted to the cage using hammer nuts and 4mm screws. The clever design of the 2020 extruded aluminum pieces allows the parts to conveniently slide along internal rails for precise positioning before securing tightly.

The stepper motors each use a 20T GT2 pulley (6.35mm shaft), a toothed timing belt (160mm closed loop, 10 mm wide) and a 40T pulley (5mm shaft) connected to the worm gear, along DO5TY's design. The stepper motors use TB6600 motor drivers. The aluminum tubes have 3D-printed worm gear collars mounted with M3 10mm grub screws. The mated worm gear shaft is mounted on a 80mm M5 bolt with the larger 40T pulley and smaller bearings at each end. I needed to use epoxy to secure the bolt to the plastic worm gear, but make sure the nuts and small bearings are already carefully secured appropriately first, and of course keep the bearings free of epoxy. I included optical end stops for each axis but the K3NG Arduino software has software limits that have been sufficient so far. A couple of lines of code allow the optical end stops to limit the rotator travel, if you so choose.

There is an extensive literature on controlling different satellite rotators with the open source K3NG rotator software ( https://github.com/k3ng/k3ng_rotator_controller ) and through the groups.io website ( https://groups.io/g/radioartisan/ ).

To those who aren't familiar with Az/El rotators it may be helpful to highlight the differences with this system from the Yaesu G5500. First, the Yaesu rotator uses ac motors to move the antennas and measures the axis positions using physical potentiometers. It uses a multiple conductor cable that runs to the shack where it connects with a proprietary controller that in turn connects to a PC running a rotator program. The PC knows where a given satellite will be in the sky at a specific time and date, gets information on the antenna's current position over a serial communication line from the microcontroller and issues serial commands back to the controller on where the antenna should point. The controller then translates these requests and turns on and off relays that feed ac power to the rotator motors. The potentiometers complete the feedback loop reporting on the new antenna position.

The K3NG rotator software is free, very flexible and can be used for other types of motors and quite a lot of different types of position sensors and command protocols. A wide range of different functions can be switched on or off depending on the user's needs. I chose to use stepper motors because they are now mass produced, cheap, highly accurate (they are used in 3D-printers, as is 2020 extruded aluminum) and have an impressive power to weight ratio. The Yaesu GS232A or GS232B command protocol (either one--they're equally useful and almost identical) works well because it's so simple and is widely used. Very important: a regular Arduino Nano or Uno does *NOT* have enough memory to run this software, even though the total amount of code isn't overwhelming. An Arduino with more memory is required, like an Arduino Mega, which works well. Some of the compile options I set in the K3NG rotator software are GS232B emulation, potentiometer azimuth sensor, ADXL345 elevation sensor with Adafruit library, and stepper motor option.

So why use a Raspberry Pi? Because it lets you communicate with the Arduino through a server running on the Raspberry Pi. Please watch the fine video regarding this by Anthony Good, K3NG ( https://www.youtube.com/watch?v=FUW9yghK944 ). You should also look at the other videos on the K3NG youtube channel ( https://www.youtube.com/channel/UC5o8UM1-heT5kJbwnJRkUYg ). In my adaptation, both the Arduino Mega and the Raspberry Pi are mounted inside the rotator cage at the masthead. While it may be possible to use a Raspberry Pi 3, I use a Raspberry Pi 4. The extra processing power of the Raspberry Pi 4 allows me to use the embedded WiFi and Raspberry Pi Connect (a sort of VPN) to log on to the desktop environment remotely from my shack. This is quite an advantage because it allows me to debug the Arduino software sketches and change, upload and test new sketches on the fly from my PC in the shack without climbing up to the masthead. The server on the masthead Raspberry Pi runs a linux hamlib ROTCTLD daemon. That also means that I can use SSH on my cellphone to test rotator positioning and calibration outside the house but on the ground using ROTCTL commands.

I use a simple, cheap accelerometer (ADXL345 with the Adafruit library) to measure elevation on the antenna boom. It plugs into the Arduino I2C bus easily with four pins, Vcc, Ground, SDA and SCL. I tried a similar I2C magnetometer for azimuth control but found it unreliable in the face of stray magnetic fields. After trying rotary and also incremental encoders I settled on a 10-turn 500 ohm potentiometer which gives surprisingly good results. I printed a 72-tooth timing gear that slips over the 38mm aluminum vertical azimuth tube and connected it to the potentiometer with a 20-tooth (20T) metal pulley so that 3.6 potentiometer turns amount to 360 degrees of antenna rotation.

Calibration of the rotator and mapping of potentiometer position to actual antenna position is done through the Arduino IDE via the Raspberry Pi Connect VPN on the PC desktop. (That's one reason it's so helpful to have the Raspberry Pi in the cage at the masthead.) At the serial monitor on the desktop with the IDE, clear the Arduino nonvolatile calibration memory with the command \X0 (zero). Then with the antenna in the base rest position (say, 0 degrees, due north, full counterclockwise CCW, and the potentiometer at or near full CCW) type O (letter "O") and hit return. That sets the full CCW position into memory. Then run the command to go full clockwise CW (say, 360, 450 or 540 degrees by typing "R" for right turn, then "S" for stop at the proper full CW position) and type "F" followed by hitting return. That sets the full CW position into memory. That's it!

The single power line under load to the stepper drivers runs at about 13 volts. The Arduino Mega and Raspberry Pi run off a step-down regulator at 5 volts. I printed a plastic rain shroud for the rotator which is sufficient for the occasional moderate rain here in Southern California. All connections are soldered and all exposed connections are sealed in tar tape. I have not seen any rotator malfunction from transmitter RF, nor have I noticed any motor hash from the steppers on receive.

How do you then get your PC to track a given satellite's position across the sky and tell the rotator software where to point the antenna? There are a number of products, some freeware, like Orbitron or GPredict. I prefer PSTRotator because it's easier to use, and it interfaces with the CNS S.A.T. box which coordinates everything with the IC-9700 transceiver--antenna and satellite position and tracking, upcoming satellite passes, band, frequency and multimode switching, Doppler shifts, and QSO logging all in the background. If your 70 cm and 2 meter Yagis don't weigh too much (about 3.5 pounds each in my case) this rotator works very well without having to spend upwards of $900-1000 for a new G5500.

This is a fun project that will help you use directional satellite antennas without having to buy an expensive commercial Az/El rotator. It handles both of my crossed Yagi antennas, 70cm and 2M, which themselves weigh less than 4 pounds each. The rotator itself with the wooden mount weighs 26 lbs. The antennas must be well balanced. High winds, ice and heavy rain may cause problems with this homebrew unit, but so far in Southern California with two minor storms it’s held up well so far. If you have some initiative, patience and time you may want to give it a try!