Originally shared by René Jurack Let's see,

Originally shared by René Jurack

Let’s see, if a genuine GT3 belt does affect the ringing…

Holy crap!

@Griffin_Paquette Holy crap? Holy crap, a genuine belt? Or: Holy crap, that is baaaad ringing? :joy:

Bad ringing!

Yeah, that is what happens with 200mm/s and 10.000mm/s² :sunglasses:

That is very impressive though for the speed. Did you say you had a 500g total carriage weight? Is there room to shave that?

500g for the whole x-gantry including ALL moving parts, hotend, fans etc… one could go for titanium-sheets, titanium or aluminium bolts, ceramic bearings, strip the watercooling, mount a smaller hotend, switch to a smaller linear rail… so yeah, there is some room to shave of weight :+1:

I am not and expert on how the motion planning in smoothie works. But would that be a function of the acceleration value, or the Jerk (in smoothie the junction deviation). Since the ripples are in the direction it was previously traveling you would think controlling corner deceleration is the best method to mitigate the ringing.

I have always wondered how does the planner handle deceleration. Does it accelerate from zero starting at the corner… Then decelerate leading into the next corner until it reaches a speed that is in compliance with the junction deviation?

Sorry if I am phrasing this incorrectly… I have just always wondered how decel into a corner is planned for at the firmware level.

Assuming you had your accelleration and jerk settings perfect, would the accuracy of your extrusion keep up?

@Eclsnowman as far as I know, every firmware does it pretty much the same as Marlin. Which means, it decelerates towards the corner, and then it accelerates when it leaves the corner. It does this at the configured acceleration. The “magic trick” is the “jerk” setting (which has little to do with the official jerk, which is change in acceleration). This jerk is an instant change of speed. So at the corner, it’s not at 0mm/s. It will be at 0.5*[jerk-setting] and it will leave this corner at 0.5*[jerk-setting].
This is actually the main source of ringing, but setting the jerk to 0 will cause other effects. So you need to find the sweetspot for your mechanics.

Marlin uses a trapezoid algorithm where it starts with jerk (instant speed without acceleration), then accelerates to max acceleration speed then decelerates, with finally stopping with jerk.

Smoothieware is a little bit smarter about “jerk” than other grbl-derived firmwares, but I think in a 90 degree corner, the result is roughly the same.

Printrboard G2 uses a 6th-derivative acceleration regime that might resolve the problem but I don’t know of anyone that’s using it in projects yet and it would require reworking the DICE.

@Rene_Jurack a long shot - but does your belt path maybe have the teeth of the belt turning against smooth idlers?
i have had much success reducing (not completely removing as it’s a heavy moving bed) my Y axis ringing by changing from a smooth to toothed idler.

@Aric_Norine ​, try Marlin’s LIN_ADVANCE algorithm.
It really works.