Fix Z banding with input shaping?

[TechNi]

Since input shaping is rather to compensate for vibrations at extreme speeds in X and Y direction, I wonder if it will be possible to introduce a Z-dependent compensation? Since lead screws are never 100% straight and thus shake the bed left and right if not properly kept in check with bearings, this would be neat.

The vibrations caused by uneven lead screws are very predictable and consistent, but they depend on the Z position. They usually increase the closer the bed gets to the motors.

Maybe somebody who has tested input shaping with Klipper or the RRF 3.4 beta can give some insight whether this is possible?

Thanks!

[T3P3Tony]

@techni to confirm you are seeing banding that is caused by the Z leadscrew(s) not being straight?

First to get this out the way: there are good mechanical designs that avoid this by ensuring the bed travel is constrained by rails/straight rods, and the leadscrews have enough play to not shake the bed.

That out of the way, input shaping is not the right mechanism to compensate for this sort of mechanical issue. Input shaping works by filtering acceleration speeds to avoid frequencies that excite resonance. For the issue you describe it is a "static", repeatable error, so changing the acceleration that the XY carriable moves wont change the amount the bed is offset in XY.

What would be required would be some sort of map of XY error by Z height (like a height map, but XY displacement in Z, instead of Z displacement in XY).

I will move this to the Firmware Wishlist however to fix it it may be easier to sort out the mechanical issue!

[oliof]

Side note: If you havent PID tuned your bed yet, do that too. The pulsing of a bang bang controlled heated bed can deceptively look like z wobble ...