This topic is somewhat related to the Belt Snapped During Calibration topic, but is also different enough that I thought it merited a separate topic.
Last night I ran across this video: https://www.youtube.com/watch?v=MwIBTbumd1Q which describes a 3D printed robotic actuator based on the friction of wrapping a rope around a cylinder. In the video the author presents discussions about the problems of fatigue-based failure of the rope as well as creep, the stretching of the rope over time. I thought this video might be of interest to the community due to the parallels with the M4. It is making me rethink the idea that one can calibrate and never have to recalibrate as long as the frame remains the same.
Last night I ran across this video: https://www.youtube.com/watch?v=MwIBTbumd1Q which describes a 3D printed
robotic actuator based on the friction of wrapping a rope around a cylinder.
In the video the author presents discussions about the problems of
fatigue-based failure of the rope as well as creep, the stretching of the rope
over time. I thought this video might be of interest to the community due to
the parallels with the M4. It is making me rethink the idea that one can
calibrate and never have to recalibrate as long as the frame remains the same.
it helps that we have fairly large bend radius around the spool, and the encoder
pulleys are also fairly large (especially in comparison to the tensioning
wires in the belt)
but we’ll have to see as we go along.
I’m hoping that when we get to the point of trying to save the belt length
across power cycles, we can also have a test routine that can move the sled to a
few places and check that the belt lengths match what’s expected. That can not
only be a sanity check that the saved belt lengths are plausible, but also be
used to check for belt stretch (or frame warping)