so ive been messing around with this for about three weeks now and i think im close but not quite there. basically i pulled an old prop pitch motor out of a junk pile and wanted to automate it with an arduino uno i had sitting in a drawer. the idea is pretty simple — read the pot on the rotator, compare it to a target bearing, drive an H-bridge to move the motor until theyre close enough, stop. sounds easy right.

the problem im running into is the feedback from the pot is noisy as hell. i got a 10-turn pot wired up and even with a 10uf cap across the wiper and ground the analogRead is still jumping around like 15-20 counts. that translates to maybe 5-6 degrees of error and for satellite work that's actually kinda significant. ive tried averaging 10 samples but then the response feels sluggish. wondering if anyone has done something similar or if theres a better way to handle the ADC noise on these boards. also using a L298N driver module which gets pretty warm even at modest speeds so maybe i need to look at something else there too.

running this with a raspberry pi as the front end eventually — plan is to have the pi run hamlib or maybe just a simple python script that talks to the arduino over serial and sends target azimuths. havent gotten that far yet but the hardware side is killing me first.

yeah the analogRead noise on unos is a known thing, the AREF pin is just not great for precision work. one thing that helped me a lot was switching to an external voltage reference — wire a clean 5v reference like an LM4040 or TL431 to AREF and connect it in software with analogReference(EXTERNAL). makes a noticeable difference. also make sure your pot wiper line isnt running near anything switching, even short runs near the H-bridge wires can inject garbage.

for the averaging thing, look at exponential moving average instead of a simple rolling average. its basically newValue = alpha * reading + (1-alpha) * lastValue and you tune alpha between 0 and 1. small alpha = smoother but slower, big alpha = faster but noisier. i use something like 0.1 or 0.15 for rotator stuff and it tracks well without being jerky. way less code than storing a ring buffer too.

the L298N getting warm is expected honestly, those things drop like 2v across the outputs so theyre burning off a bunch as heat. if you want something cleaner look at the DRV8871 or even just a pair of BTS7960 modules which can handle higher current and run way cooler. depends on what your prop pitch motor actually needs but worth checking.

did basically the same project last year except i used a pi pico instead of an uno because i had one spare and the pico's ADC is actually a bit worse than the uno in some ways lol so that was fun. ended up just going with a dedicated ADC chip over SPI, think it was an MCP3208, 12 bit and the noise was way more manageable. adds a little complexity but not much.

the hamlib + serial thing works fine once you get it going. i wrote a tiny arduino sketch that just accepts single-letter commands over serial like G for get position and S,270 for set target, kept it dead simple. the pi side was a python script with pyserial. honestly didnt even bother with hamlib in the end, just wrote my own little rotctld-ish thing, took maybe an afternoon. might be overkill for you depending on what software you wanna drive it from but if youre using gpredict for sat tracking it expects rotctld protocol so maybe worth doing it properly.