so ive been messing around with this for a few weeks now and i think i finally have something worth sharing, or at least worth getting feedback on before i go too much further down the rabbit hole

basically i got tired of manually adjusting my yagi when im trying to track satellites and i had a pi4 just sitting in a drawer doing nothing so i figured why not. wrote a python script that pulls az/el data from gpredict via the rotctld interface and feeds it out through a GPIO expander board to an H-bridge i salvaged off an old printer motor controller. the rotator itself is a homebrew thing i built like 5 years ago with a windshield wiper motor, nothing fancy

it actually works pretty well but im having this weird issue where occasionally the az will overshoot by like 10-15 degrees and then hunt back and forth a couple times before settling. my feedback pot is a 10-turn bourns unit so i dont think its resolution, more like my PID loop isnt tuned right or maybe the motor is just too fast for the control loop to keep up with. anyone done something similar and dealt with the hunting issue

also thinking about adding an arduino nano as a dedicated ADC front end for the pot since the pi's GPIO is all digital obviously, right now im using an MCP3208 over SPI which works but feels a little clunky in the code

yeah the hunting thing is almost certainly your PID tuning, specifically your derivative term is probably too low or you're not using one at all. that wiper motor has a lot of inertia and the control loop needs to anticipate the overshoot before it happens, not react after. try backing off your proportional gain first and see if the oscillation amplitude changes, if it does you at least know the loop is responding which is a good sign

i did almost the exact same project a couple years back, also used a pi3 and gpredict, though i used an arduino mega as a dedicated co-processor for all the low level stuff — motor drive, encoder reading, limit switches. the pi just sends target az/el over serial and the arduino handles the closed loop control. works really clean that way and you dont have to worry about linux scheduler jitter messing up your timing sensitive stuff. the MCP3208 is fine btw, i used the same chip, just make sure your SPI clock isnt too high and that your pot wiper isnt floating when the rotator is at either mechanical stop

not overkill at all man, this is exactly the kind of stuff that makes the hobby worth it. i dont do satellites myself but ive been wanting to automate my HF antenna switching for a while and keep putting it off. whats the rotctld setup like on the pi side, is that pretty straightforward to get running or did you have to compile anything from source