so i've been messing with this for a few weeks now and honestly it started as a simple thing and has completely taken over my workbench. i wanted a way to automatically point my yagi based on APRS position data from a balloon launch our club was doing and one thing led to another and now i've got a raspberry pi running direwolf feeding position packets into a python script that spits out azimuth/elevation to an arduino mega which is controlling the H-bridge for my old Yaesu G-450A rotator.

mostly works. the problem im having is the feedback pot on the rotator is giving me noisy ADC readings and im not sure if its a grounding issue between the arduino and the rotator controller or just the pot itself getting old. ive tried adding a small cap across the pot terminals to smooth it out a little and that helped some but i still get like plus or minus 5 degree jitter when the thing is supposed to be sitting still. not a huge deal for HF but for 70cm EME eventually it'll matter.

curious if anyone else has built something like this or has dealt with the ADC noise thing. also wondering if i should just swap out the pot entirely, ive seen people put encoders on these but that seems like a bigger job than i want right now

yeah the pot noise thing is super common with these old rotators. before you start rewiring everything i'd check if you have a good star ground between the arduino ground and the rotator controller ground — i burnt probably two weekends on a similar project before i realized my arduino was floating relative to everything else. a 0.1uf cap right at the ADC pin to ground helped me more than putting it across the pot itself, worth trying if you havent already.

also you can do some software filtering on the arduino side, just average the last N readings and throw out anything more than X degrees from the running average. not elegant but it works fine for this kind of application. for a yagi on 70cm you'd probably want the window pretty tight but for now while you're testing just getting stable readings matters more than accuracy.

the encoder swap isnt as bad as it sounds btw if your rotator has a shaft you can access. i did mine with a cheap optical encoder from aliexpress and some 3d printed brackets, took an afternoon. much cleaner than dealing with pot drift and wear.

ive done something kinda similar for satellite work, rpi running gpredict with a hamlib rotctld daemon and then a custom arduino sketch handling the actual motor control. the direwolf to position data pipeline sounds really cool for balloon tracking though, never thought to do it that way.

one thing that bit me — make sure your python script is handling the case where packets come in faster than the rotator can actually move. i had a situation where i was queueing up commands and the rotator was still trying to catch up to a position from like 30 seconds ago while the target had moved completely across the sky. added a simple check where it just takes the latest position and discards anything older than a couple seconds and that fixed it.