continuous rotation servos should not be hard to get. There are a number of
magnetic angle sensors available now (I read an article yesterday that Tesla is
using them on their retractable door handles, so cheaper with better interfaces
now)
given how old the codebase is, I would suggest working from scratch.
quadralateral encoders are FAR more common than anything else, and much easier
to interface with. They give you a signal pulse on movement (you need a fast
enough system to be sure to not miss the pulse). Many motors are available that
have these encouders built-in (like the ones the maslow uses)
magnetic angle encoders like the makesmith used, have either PWM outputs, or you
have to query them and read the response, and Bar noted that the angle readings
are not actually that precise without calibration (which is not that hard to do
if you can rotate the shaft a known amount, say with a stepper, but if you are
doing that, is the magnetic angle encoder really giving you a lot of value?).
You also need special magnets with the magnetic field not going up through the
disk, but going across the disk.
If I was going to build a makesmith-like device today, hereās what I would be
looking at
start with a Pi Pico, it has lots of I/O to read encoders and PWM outputs to
drive hobby servos (open loop)
continuous rotation servos are readily available (and itās not hard to modify
standard servos to be continuous rotation, just remove the stops and set the pot
in the servo to mid-range instead of coupling it to the output shaft)
encoders are readily available, or they can be made with a couple optical
sensors. If you have access to older mice that have a ball instead of a optical
sensor, you have a couple sets of encoders in them that you can use (and the
disks to activate them). you may need to do some level shifting to couple 5v
sensors to the 3.3v pico, or you may be able to make it ājust workā
The pico can be programmed via the arduino IDE, and is significantly faster
than the stock arduino chip, so adapting the grbl gcode interpreter and motion
planner should not be that hard to do. The hard part would be changing the
output from driving steppers to having PID loops to match the measured position
and velocity to what the planner is looking for. Not a utterly trivial task, but
not rocket science either (and then not hard to adapt to more sophisticated
motors)
David Lang