Input Shaping working on Test Prints, but not on "real" Objects
-
Hello together,
at the moment I am struggeling with the tuning of my acceleration/ input shaper.
Currently I have installed the 3.5.0-rc.3 firmware on my Duet 2 Maestro. Printer is a self designed core xy printer (Hypercube/VzBot) Frankenstein.
Yesterday I recalibrated my Input Shaping using the Plugin an doing the Test Print on my printer.
And what can I say... it looked amazing considering the acceleration rising from 8000mm/s^2 to 12000mm/s^2
I was very happy with the result. I dialed in the acceleration to 10k.
Today I startet to reprint some of my printer parts and was shocked by the result.
What can I do here and why the test print looked so good an the "real" part so bad. For sure I can lower the acceleration, but the ringing tower looked promising -
@TrippZ those real world parts have short segments where RRFs current implementation of IS is not effective. There's been some improvements in the latest post rc3 builds but I can't guarantee they would help in your case.
-
@oliof Does Klipper a better Job for small segments than RRF does?
-
@TrippZ both Marlins ZV only and Klippers IS implementation do better than the 3.4.6 implementation; I don't think anyone did a comparison with the latest post 3.5rc3 fixes. I know only three people that have a machine that could run all three firmwares and gcode that would be identical except for start and end code (and only one that bothered to actually go through the trouble of setting up all three firmwares and slicer profiles), so a final verdict is still out.
-
@TrippZ You may want to test using the updated build of rc3 found here: https://forum.duet3d.com/post/334236
You may also find that less aggressive input shapers (like zvd and mzv) can be applied to shorter segments. Using a smaller value of the M593 L parameter may also allow more segments to have IS applied to them (the default value is 0.25 but you can go down to 0.01), however lowering this value may slow down the print. Lowering the print speed and acceleration may also allow IS to be applied to more moves.
-
@gloomyandy Thank you for the info. I installed the version (RC.3+), also that should fix my layer shift, that I got in a print yesterday. I applied now MZV instead of ZVD.
Can someone recommend to lower the damping factor (I have im mind, that I read that anywhere)
I do some tests and come back after.
Thanks! -
Make sure you recheck your pressure advance after setting your input shaper. I needed to apply both input shaping and pressure advance to get the best results.
I have a note in my config to retune pressure advance if I change the input shaper from MZV.
-
This post is deleted! -
@TrippZ currently Klipper applies IS to segments too small for RRF to apply IS too. The cost is more deviation from the path commanded by the GCode, which the Klipper documentation refers to as "smoothing". See https://www.klipper3d.org/Resonance_Compensation.html#after-enabling-input_shaper-i-get-too-smoothed-printed-parts-and-fine-details-are-lost.
The latest RRF 3.5 builds are able to apply IS to some shorter moves, but by no means all short moves. We already have work underway to support IS on all short moves, but this will be at the expensive of "smoothing".
