so ive been messing around with a raspberry pi zero w for a few weeks now trying to get it to talk to my old yaesu g-450a rotator controller. the idea is pretty simple, i just want to be able to point the antenna from my logging software without having to walk across the shack every time. sounds easy right

the rotator controller has that 6 pin din connector on the back with the analog voltage output for position sensing and i figured i could just read that with the pi and use a relay board to pulse the rotation inputs. but the pi doesnt have any analog inputs natively so i grabbed one of those little MCP3008 SPI ADCs from the parts drawer and got it wired up on a breadboard. its actually reading the voltage from the controller pretty well now, like +-2 degrees or so which is fine for HF

the problem im running into is the relay timing. when i send a rotate command from hamlib it sometimes overshoots by like 15-20 degrees especially if its moving fast, and i think its because theres some latency somewhere in my python script, maybe in the loop that checks the ADC. i bumped the polling rate up but then the cpu load on the zero gets kind of ridiculous for what its doing

has anyone done something similar and found a good polling interval that actually works, or is there some better way to handle this kind of closed loop control on a pi? ive seen people use arduinos for this kind of thing and wondered if the tighter timing makes a difference

yeah the pi is kind of the wrong tool for tight realtime control loops, thats honestly the issue. the linux scheduler just isnt going to give you consistent timing especially on something as limited as a zero. i did something similar with my T2X rotator a while back and ended up putting an arduino nano in between as basically a dedicated controller - pi sends target azimuth over serial, arduino handles the PID loop and relay timing locally, sends back position updates. works really well and the arduino can poll that ADC way faster without any of the scheduling jitter

the other thing worth looking at is a bit of deadband hysteresis in your control logic. my setup wont even activate the relays unless im more than like 3 degrees off target, and i ramp the braking early so it cuts power before it actually reaches the target bearing. takes a bit of tuning but once you dial it in the overshoot basically goes away

not sure about the polling stuff but i did this exact project last year with a g-650 and honestly hamlib rotctld is kind of finicky depending on what logging software youre using. what are you connecting it to on the pc side? i had a nightmare getting n1mm to talk to my pi setup until i figured out i had the rotctld model number wrong. totally unrelated to your timing thing but just something to check if you havent already