Google+ post by D Rob on 2014-03-11 04:56:09 UTC

All the relationships are mathematically linear, so going from 10 to 8 is just a factor of 0.8 difference.

The problem with microstepping is that the stepper is not able to hold a microstep position as well as a full step position. This leads to a cyclical error in layer heights and banding in the print. So we need layer heights to be increments of a full step, even if the actual layers don’t fall on a full step. Which is an illustration that the position error when using microstepping is significant and different from the whole step error. Depending on the load, the actual shaft position can sag as far as halfway to the next whole step position without losing position entirely. So we can’t set bed elevation within 0.001 by using 1/32 microstepping. The position error could be as large as 0.016, depending on which microstep we’re attempting to hold. (If we operate on the verge of losing steps. We probably run at something more like 80% of that.) Microstepping smooths things and makes the system quieter, but does not improve accuracy by the same factor as the microstepping fraction.

You can get 20 tooth GT2 pulleys for 8mm shafts, what that would do to a belt holding up a heated bed I have no idea, but it helps the numbers a lot :smiley:

Agreed that we should not rely on microsteps, full step per 5 microns is my vote!