so ive been messing around with this idea for a few weeks now and i think im finally close to something that actually works. background: i've got an old Yaesu G-450A rotator that i picked up at a hamfest for cheap, controller works but the interface is ancient and i want to be able to drive it from my logging software without sitting there manually cranking it.

the plan was to use a pi zero W to read the pot position voltage through an MCP3008 ADC over SPI, then drive the rotator motor direction via a couple of relays on the GPIO pins. wrote a little python script that polls the position every 500ms and compares it to a target bearing i send it over a simple socket connection. works in the house on the bench, rotator turns, stops when it gets close to the target, all good.

problem is i keep getting weird spikes on the ADC readings when the relay clicks. im guessing its inductive noise from the relay coil backfiring into the supply rail or something? i put a flyback diode across each relay coil but the spikes are still showing up. added a 100uf cap on the 3.3v rail near the MCP3008 and it helped a little but still seeing +-15 degrees of jitter sometimes when the relay transitions.

anyone gone down this road before? curious if an optocoupler between the pi and the relay board would help or if i need to rethink the power supply situation entirely. running both the pi and the relay board off the same 5v wall wart right now which is probably dumb in hindsight.

yeah separate supplies will probably fix most of your grief. relays are nasty on a shared rail, even with the flyback diodes. i had almost the exact same problem when i built a little sequencer for my preamp switching — the relay coil was dragging the voltage down just enough on transitions to cause all kinds of weird readings on the ADC. put the relay board on its own little buck converter fed from a 12v line and everything calmed right down.

the optocoupler idea is also solid if you want to be thorough, but honestly id try the power supply thing first since its the easier fix. also might be worth adding a small RC filter on the ADC input pin — something like 1k resistor and 100nf cap to ground right at the chip. smooths out a lot of the garbage before it even gets sampled. rotator pot signals dont change that fast anyway so a little filtering doesnt hurt anything.

I did something similar with an Arduino Mega and a G-1000 a couple years back, different hardware but same general idea. One thing that bit me was ground loops — had the Arduino and the rotator controller sharing a ground reference and it was causing all kinds of garbage. ended up galvanically isolating the ADC side with an ISO1541 I2C isolator chip and most of the noise went away.

also if youre polling every 500ms you might want to add some averaging on the ADC readings, like take 8 samples and average them before you use the value. cheap way to filter without touching the hardware. the MCP3008 is decent but its not doing you any favors in a noisy environment without some software smoothing too.