Microstepping (interpolation)

PeterA

Thanks for the feed back. Very helpful

owlfab

Noise is the biggest thing for me.

Would you guys expect much of a noise difference between 16x with interpolation and 256x assuming the CPU can keep up? I'm on a Duet2 running 0.9 degree steppers and rapid movements get limited when I go to 256x. For whatever it's worth, I feel like the Prusa mk3 machines are quieter than mine for rapids, but I realize there are a lot of other variables as well.

deckingman

@owlfab said in Microstepping (interpolation):

.................... Would you guys expect much of a noise difference between 16x with interpolation and 256x assuming the CPU can keep up? .....................

No. Because 16X micro-steps interpolated (by the driver chip) into 16 smaller steps each = 16x16 = 256. So what gets sent to the motor is the same in either case.

matt3o

@owlfab you may try enabling stealthChop but with my motors it doesn't make much of a difference

Veti

@owlfab said in Microstepping (interpolation):

I feel like the Prusa mk3 machines are quieter than mine for rapids

duet 2 does not have stealthchop. the prusa does. that makes a lot of difference for the noise.

matt3o

@Veti ah right duet2... sorry, missed that. Not that on the duet3 I hear any difference, but could be my moons motors

A Former User

@PeterA

keep in mind that the more microsteps you introduce the more you divide your overall-torque into smaller incremental-torque. So basically this certainly gives you resolution but only maybe gives you accuracy. Of course if money is no objective to you, you can increase the microstepping endlessly and just buy a "stronger" stepper to have your "old" incremental-torque matched even with more microsteps Calculation-formulas for that are provided e.g. here https://www.faulhaber.com/fileadmin/user_upload_global/support/MC_Support/Motors/AppNotes/Faulhaber_AN015_EN.pdf

ctilley79

@LB said in Microstepping (interpolation):

@PeterA

keep in mind that the more microsteps you introduce the more you divide your overall-torque into smaller incremental-torque. So basically this certainly gives you resolution but only maybe gives you accuracy. Of course if money is no objective to you, you can increase the microstepping endlessly and just buy a "stronger" stepper to have your "old" incremental-torque matched even with more microsteps Calculation-formulas for that are provided e.g. here https://www.faulhaber.com/fileadmin/user_upload_global/support/MC_Support/Motors/AppNotes/Faulhaber_AN015_EN.pdf

Isn't this a little misleading if you're in interpolation mode? The driver is essentially running at 256x. The microprocessor is only generating 16x of them, but the driver is dividing those out into 256, so essentially the torque will be the same, the only difference is the computational load on the microprocessor increases on higher step rates right?

dc42

Microstep interpolation doesn't increase resolution or accuracy, because the MCU still commands the driver to 1/16 microstep positions; but it does make the motors quieter.

A Former User

@ctilley79

It is totally misleading Because in reality there is also: Friction, acceleration, jerk, driver-behaviour etc. pp. so the paper from faulhaber only gives you a hint of a ballpark where you end up if you use 1/16th microsteps as "possible-stop-positions" (versus the "interpolation/microply to 256" which is basically "smoothness" as dc42 pointed out, so basically quiteness but the stepper can not "stop"/"hold" on those positions. Basically as I understand it, the "interpolation"/"microplyer" is only that you set your voltage and ampere not in one step to the next 1/16 step, but go there via "quasi-sinusoidal" micro-interpolation LookUpTable for Voltage/Ampere so that it doesn´t shake that much)

I would roughly calc with the values in the paper and keep a safe 10-25% distance to it, because of all the other effects not yet in the whole-calculation

I usually use as little "possible-stop-position" hard-microsteps but of course use always the interpolation/microplyer microsteps for smoothnes - less workload for the cpu for the step-calculation, because the interpolation happens in the driver-chip. Also there are theories that it might be beneficial for accuracy if you use only full-&halfsteps as possible-stop-steps. Of course then your kinematics has to be adjusted to that

Hey - please share your thoughts, experiments and conclusions, might be of help for all of us we wouldn´t be here if there isn´t anything we wanna learn