Faulty Bed Mesh Interpolation

[cottec]

Basic Information:

Printer Model: FLSUN V400 Delta
MCU / Printerboard: Robin Nano 2.0 - Stock
klippy.log (2.5 MB)

Might be more a cosmetic issue, but the interpolation of points that are not on the build plate (Diameter 300mm) is really really bad.
Is there a way I could change it to be more sutaitble?

[Sineos]

I doubt that this comes from the interpolation.
The “interpolation” simply takes the nearest probed point and uses its value. No magic involved.

[cottec]

any other idea? bug?

[mykepredko]

Is there any binding or movement stoppages when the Y axis approaches -150?

I have no idea where the three axis rails are but if there was one at Y negative, it might not be going the full travel down.

[Sineos]

@cottec

bug?

If you want to call your delta a bug then maybe
But seriously:

• Deltas are very sensitive to tilted effectors if not mechanically very carefully tuned and with highly precise tuning in the settings as well
• This tilt is less noticeable in the bed’s center and gets worse and worse the closer you move to bed’s edge
• As the probe is offset from the nozzle’s center, this will even further increase the effect of a tilt

If you look at your mesh (“default”) the lowest value -0.145629 and the highest is 0.436727.

[cottec]

hm really strange, maybe these corners had too much tension build ip in the spring between the two arms, dunno. took everything apart, put some fresh grease on there. did
Delta calibrate
probe offset
enhanced delta calibration
probe offset

then bed mesh

now
it’s getting more interesting…

is that the actual mechanical accurcy of the movement between probing lines?
or is that the roundness of the rollers on each arm?
there’s quite pattern visible. adjusting the mesh tension from 0.1 to 0.5 doesnt really flatten this, so that’s not “helping”

[Sineos]

Mind the scaling of the graphs. Your previous graphs had a scaling of -0.146 to +0.437 (over 0.5 millimeters!) and now your graph is scaled between -0.162 to +0.115, so:

• A considerable improvement - well done!
• If you used the same scaling as before, your bed would be dead flat

[cottec]

yeah I’m aware of that but it looked kind of cool and it’s showing that there is something going on…

buut, these spikes dont make much sense from a bed wavyness perspective.
That’s a mechanical “issue” or just the resolution, what do you think?

[Sineos]

Well, I cannot tell you for sure since on the one hand my experience with Deltas is severely limited and on the other hand this waviness could have been there before, just unnoticeable due to the scaling. My vote would be mechanical effects.

Just to highlight this point:
You produced a perfect learning example that most users tend to ignore:

• A 3D printer is first and foremost a mechanical thing and depends on the quality of the used components and the accuracy of its built
• There is an abundance of sources for introducing systematical imperfections:
  • Rigging, i.e. being parallel, perpendicular etc
  • Run-out of pulleys and bearings
  • Belt paths and tension
  • Center of Gravity
  • You name it
• Software (regardless if Marlin, Klipper, etc.) can only compensate for a fraction of such errors

Also, believing that a 200 USD consumer printer will take you anywhere in the range of microns is an illusion. A 200 USD linear rail is still to be considered cheap in this context.