I'm hoping the comments below will answer this question. Your "edited to add" is definitely on the right track -- my example implicitly assumed that the smallest microsteps were smaller than I really cared about.ray_parkhurst wrote:I'm not sure why you downplay the missing of these microsteps.
I expect we agree on the physics of how microstepping works, so the difference is in how we think about it at a higher level.ray_parkhurst wrote:we seem to have a fundamentally different understanding of what a microstep is.
I think about microsteps in two different ways, depending on context. One way is to focus on what happens when the motor stops, trying to make that position as accurate as possible. The other way is to focus on what happens while the motor is moving, trying to make that movement as smooth as possible.
In the first case, position when stopped, I care about how close the device is to the position that I intended, or at least how close it is to the intended distance between two stopped positions whose absolute locations I may not care much about.
In the example that I gave, the context would be something like intending an interval of 20 microsteps, in a 200-frame focus stack. In that situation, being off by a few microsteps in an individual focus step is tolerable, but accumulating a high fraction of missed steps would be a disaster because then the start/end positions of the stack would be wrong.
In the second case, smoothness while the motor is moving, I would care mostly about "jerk" -- changes in acceleration. Even if the instantaneous motor position lags by a lot of microsteps, say 1/3 full step behind the commanded microstep position, I'm probably happy with that as long as the lag doesn't trigger some other issue like a nasty resonance. In particular I would have no concern about the low nominal torque of a 1/256 microstep, as long as in the end, when the motor is stopped, there's enough torque to reach and hold whatever final accuracy I care about. In those videos of silent motor motion, I would place a large bet that at the bottom of the hardware they're doing something equivalent to running with tiny microsteps and not worrying about whatever lag is introduced by the correspondingly tiny nominal torque.
Adding emphasis, exactly what I wrote was "From a standpoint of overall system behavior, this is a lot more like closed loop then open loop."ray_parkhurst wrote:I also don't understand your comment about this being more like closed loop than open loop, as I look at it exactly opposite. In fact it is more like a very bad open loop!
My comment was intended to underscore the point that microstep errors don't accumulate, while skipped full steps do.
The full steps are clearly open loop, since without adding feedback you can get arbitrarily large errors and nothing will correct for them.
But for the microsteps, if we stand back and squint it sure looks like something is monitoring the motor shaft position and somehow making adjustments so the shaft never gets very far away from the nominal microstep position. Described like that, the behavior sounds like "closed loop" or at least "proportional control". But I agree that using either of those standard phrases was probably a bad thing to do, because they have so many other connotations attached. Typically those phrases imply more hardware added specifically to provide the feedback, where in this case it's really just the "springiness" of the magnetic field doing the job.
--Rik
