Accelerometer for delta smart effector

  • An I2C accelerometer would allow the movement of the effector to be compared against the commanded move to provide diagnostic information for tuning.

    Potentially it could monitor performance allowing on-the-fly corrections to be made when an imminent overshoot or missed step is detected. (Optimistic, I know, but this is a wish list…)

    A simple surface mount 6 axis I2C device would provide both acceleration and tilt information whilst adding minimal weight.

    An equivalent add-on board could provide the same functionality for cartesian printers (possibly combined with the IR probe)

  • We've experimented with them (well Mike our consultant engineer did) and found for bed detection they were not an improvement on piezo or strain gauge. But for "inertial navigation" maybe they might be useful. Actual encoders on axis motors or linear encoders on axes would be much more accurate. If possibly unnecessary.

    I've seen quite convincing arguments that closed loop positioning maybe less useful than you might think, by the time you are out of position your print might be ruined.

  • administrators

    I am fairly sure that cheap accelerometers are nowhere near accurate enough to compare actual with requested motion.

  • Plus if you have stall detection working an impact with a curled edge or mishap during printing might well be detected and trigger a pause so you can rectify it. A closed loop system would just reposition itself possibly making the problem worse.

    There is an argument for using closed loop to enable cheap dc motors to be used for precise positioning thereby possibly allowing for a sub $100 3d printer.

  • I was thinking mainly along the lines of aiding tuning of maximum velocities and accelerations by commanding a move then using the accelerometer to see how quickly it stops. High precision is not too important in this scenario.

    I know you can calculate theoretical maximums but in my experience real machines never read the technical manuals.

    I have to agree though that on the fly corrections may be a little optimistic for cheap accelerometers.

  • I played with a gyro + accelerometer board to see if I could measure effector tilt, as a way to monitor for a magball disconnect. What I found was that at anything approaching typical printing speeds I got constant false triggers as the abrupt starts and stops were far too quick for the easy to get (cheap) accelerometers to deal with.

    It did work as a nice way to validate how square your effector's movement is across its total range of motion, but it was not great for anything over 20mm/s or so.

    It's actually on here in my post history if you want to look at it. It used a kalman filter with the gyro & accel outputs.

  • Nice to see somebody actually did it. Thanks, Kraegar!
    Good to know.

Log in to reply