Navigation

    Duet3D Logo

    Duet3D

    • Register
    • Login
    • Search
    • Categories
    • Tags
    • Documentation
    • Order

    Mesh Bed Compensation and Bed Leveling

    Tuning and tweaking
    auto leveling mesh compensation
    4
    18
    2688
    Loading More Posts
    • Oldest to Newest
    • Newest to Oldest
    • Most Votes
    Reply
    • Reply as topic
    Log in to reply
    This topic has been deleted. Only users with topic management privileges can see it.
    • samlogan87
      samlogan87 last edited by

      Hi Everyone,

      I have a question about using these two functions. I have a CoreXY with 4 Z lead screws connected to 4 Z motors using a Precision Piezo Orion as the probe.

      Currently what I do is home all axis, and then conduct auto bed leveling in 4 points doing three iterations. Below is an example of the results

      2:13:47 PMLeadscrew adjustments made: 0.003 0.002 0.004 0.003, points used 4, deviation before 0.003 after 0.000
      2:13:18 PMLeadscrew adjustments made: 0.004 0.001 0.001 0.001, points used 4, deviation before 0.003 after 0.000
      2:12:50 PMG32
      Leadscrew adjustments made: 0.180 0.181 0.175 0.170, points used 4, deviation before 0.176 after 0.000.

      As you can see, I have a light lean in the bed from side to side when the motors are depowered that seems to be sorted using the 4 z axis motors, which is excellent. It works as it should when I test the zero height in all the corners.

      Now when I use mesh bed compensation and do a run, I get the below results

      0_1540605521274_898f4928-1e05-4f8e-a53c-8a0b40d4fcf0-image.png

      It shows that the 6mm tooling plate and print surface has some slight deviations. This may have been how I mounted the plate to the z carriage (having countersunk holes) maybe relieving internal stress in the plate. I guess that's what the mesh compensation is for however.

      What I don't understand however is the 4 corners that the auto compensation levels, is now out of level by more than what the deviation was during the auto bed leveling. After the mesh leveling was done, I then ran an auto bed leveling again and got below:

      3:02:20 PMLeadscrew adjustments made: 0.005 0.006 -0.004 -0.009, points used 4, deviation before 0.006 after 0.000
      3:01:52 PMLeadscrew adjustments made: -0.007 0.001 -0.000 -0.001, points used 4, deviation before 0.005 after 0.000
      3:01:24 PMG32
      Leadscrew adjustments made: 0.002 0.006 0.008 0.006, points used 4, deviation before 0.006 after 0.000

      This seemed to show that it is pretty normal and still less than what is being measured in the mesh compensation. Maybe not as good a before. Is the difference between the two because it clears any height transformations before it does the mesh compensation? The reason for asking is I seem to have issues getting my first layer to stick consistently. I am using PrintBite+ which is awesome. Things stick really well to it, however sometimes some areas of the part do not stick as well as other areas of the part and I am wondering if the bed compensation has something to do with it.

      Does anyone have any thoughts on this?

      Kind Regards,
      Sam

      1 Reply Last reply Reply Quote 0
      • samlogan87
        samlogan87 last edited by

        I also ran a second mesh compensation after and it is very very close to the above one.

        Kind Regards,
        Sam

        1 Reply Last reply Reply Quote 0
        • dc42
          dc42 administrators last edited by

          Do the 4 points you use for bed levelling fall within the area covered by the mesh?

          1 Reply Last reply Reply Quote 0
          • samlogan87
            samlogan87 last edited by

            Hi @dc42,

            Yes they do.

            Kind Regards,

            Sam

            1 Reply Last reply Reply Quote 0
            • samlogan87
              samlogan87 last edited by

              @dc42

              I forgot to mention I also re home the z axis after the auto bed leveling before I run the mesh compensation.

              Kind Regards,

              Sam

              1 Reply Last reply Reply Quote 0
              • dc42
                dc42 administrators last edited by

                That does seem odd. Can you arrange it so that the 4 points that you use for bed levelling correspond exactly to points on the mesh?

                1 Reply Last reply Reply Quote 0
                • samlogan87
                  samlogan87 last edited by

                  Hi @dc42

                  Sure can. Will give it a go tonight when I get home from work.

                  Kind Regards,

                  Sam

                  1 Reply Last reply Reply Quote 0
                  • samlogan87
                    samlogan87 last edited by

                    Hi @dc42

                    So I had a look at the actual maximum probing points for the mesh compensation and used them for the bed leveling.

                    Here was the probing for the auto leveling:

                    8:39:12 PMLeadscrew adjustments made: 0.001 -0.001 -0.000 -0.001, points used 4, deviation before 0.001 after 0.000
                    8:38:44 PMLeadscrew adjustments made: -0.003 0.003 0.001 -0.001, points used 4, deviation before 0.004 after 0.000
                    8:38:16 PMLeadscrew adjustments made: 0.156 0.148 0.154 0.151, points used 4, deviation before 0.153 after 0.000

                    Here is the mesh height map:

                    0_1540800861805_f6594bd9-10f1-48e2-92be-1a4d2d30121e-image.png

                    This was the results from doing auto leveling right after:

                    9:14:34 PMLeadscrew adjustments made: -0.000 0.004 -0.002 -0.002, points used 4, deviation before 0.003 after 0.000
                    9:14:07 PMLeadscrew adjustments made: -0.001 0.002 0.008 0.003, points used 4, deviation before 0.005 after 0.000
                    9:13:39 PMG32
                    Leadscrew adjustments made: 0.006 0.019 0.031 0.014, points used 4, deviation before 0.022 after 0.000

                    I guess the points do line up other than maybe the second that was 0.01mm out. I am guessing some of the issues I am having could be attributed to the fact that there would be a compounding error over 342 x2 z movements as it probes every point twice?

                    Kind Regards,
                    Sam

                    1 Reply Last reply Reply Quote 0
                    • samlogan87
                      samlogan87 last edited by samlogan87

                      Another quick question, do you have to specify a taper at all in the .config or does it default if you do not specify the M376 code? I have not had it in my config file.

                      Kind Regards,
                      Sam

                      1 Reply Last reply Reply Quote 0
                      • dc42
                        dc42 administrators last edited by dc42

                        If you probe repeatedly in the high-X high-y region (where the error appears to be greatest) using G30 S-1, do you get a consistent trigger height reading?

                        The default is that bed compensation is not tapered.

                        1 Reply Last reply Reply Quote 0
                        • samlogan87
                          samlogan87 last edited by

                          Hi @dc42
                          Thanks for that. I will give it a go.

                          So by not tapering, does that have an effect on the compensation? Or will it compensate over the entire height of the print?

                          Regards,
                          Sam

                          Phaedrux 1 Reply Last reply Reply Quote 0
                          • Phaedrux
                            Phaedrux Moderator @samlogan87 last edited by

                            @samlogan87 said in Mesh Bed Compensation and Bed Leveling:

                            Or will it compensate over the entire height of the print?

                            Exactly.

                            1 Reply Last reply Reply Quote 0
                            • samlogan87
                              samlogan87 last edited by

                              Hi @Phaedrux

                              Thanks for clarifying that. I just seem to have a few issues with consistency between mesh compensation results and auto leveling that I think is causing issues with my first layer being inaccurate.

                              Kind Regards,
                              Sam

                              1 Reply Last reply Reply Quote 0
                              • SupraGuy
                                SupraGuy last edited by

                                For almost any case like this, my first go-to for an explanation would be a machinist's straight edge. 0.1mm deviation over a couple hundred mm is practically impossible to see, but that straightedge should tell the real story. If you are getting incorrect reporting from the sensor, the first step is to determine if your sensor is wrong.

                                If your sensor is wrong, then the idea is to determine what rules (And there will be some) that determine HOW it is wrong. Sometimes certain portions of the bed have more compliance than others, particularly when your bed is spring mounted. Sometimes there is a sag or bow in axis rails, and though the bed isn't bent, the probe is correct. this is one of the things that a mesh probe can help correct for.

                                1 Reply Last reply Reply Quote 0
                                • samlogan87
                                  samlogan87 last edited by

                                  @SupraGuy

                                  I did one more run last night of the mesh leveling and I want to do another one again tonight to confirm, but some of it could be a slightly bent Y axis (That is my moving carriage axis). I am still getting inconsistency in the mesh leveling which I am not overly sure why, hence the post.

                                  Kind Regards,

                                  Sam

                                  1 Reply Last reply Reply Quote 0
                                  • samlogan87
                                    samlogan87 last edited by

                                    Well, I did another run and it is not a bent y axis. When I set it up, I had a few good nozzle digs into the bed and I am wondering whether somehow in doing so, it managed to put a slight bow in the center of the bed.

                                    @dc42 When I run G30 S-1 with the piezo probe, it does not lift after probing and progressively digs harder into the bed. Not sure what I need to change to stop this. Even so at X 340 Y360 according to the mesh height map, the z value should be .197mm. When I probe it is 0.02mm.

                                    Kind Regards,
                                    Sam

                                    1 Reply Last reply Reply Quote 0
                                    • dc42
                                      dc42 administrators last edited by

                                      You need to raise the head by 5mm between each pair of G30 S-1 commands.

                                      1 Reply Last reply Reply Quote 0
                                      • samlogan87
                                        samlogan87 last edited by

                                        Hi @dc42

                                        I am sure I was doing that. I will have another go at it tonight.

                                        Kind Regards,
                                        Sam

                                        1 Reply Last reply Reply Quote 0
                                        • First post
                                          Last post
                                        Unless otherwise noted, all forum content is licensed under CC-BY-SA