Low corner with ABL

Discussion in 'Calibration, Help, and Troubleshooting' started by Kanedias, Jan 8, 2017.

  1. Kanedias

    Kanedias Well-Known Member

    Joined:
    Sep 12, 2015
    Messages:
    406
    Likes Received:
    124
    I've been using automatic bed levelling since I built my bigbox. My starting gcode does an autohome, followed by a a 5 point bed level - center and then each corner. Up untill recently this has worked perfectly and I've been able to print full bed sized selections of things no problem.

    Recently though the back left corner is ending up lower than the others. So If I have my Z gap tuned correctly for printing in the middle then that back left corner doesn't adhere.

    The active bed leveling is working, so I can see the bed moving as the head changes X and Y. Its just that one corner is wrong.

    I checked and corrected the bearing on the left hand Y carriage and thats helped a little but I don't know what else could be so consistently making it wrong. I've also blanked off the upper side of the sensor in case there was some light pollution affecting that one reading.

    Any ideas what could cause this? I'm hoping its not that the X carriage rods have somehow misaligned or something unfixable
     
  2. mike01hu

    mike01hu Well-Known Member

    Joined:
    Apr 22, 2014
    Messages:
    957
    Likes Received:
    166
    I've had inconsistent levelling due to the rod clamps becoming loose on both the X and Y carriages. Any looseness can cause slight movements of the rods as the carriage transits.
     
  3. Syko_Symatic

    Syko_Symatic Well-Known Member

    Joined:
    Sep 4, 2015
    Messages:
    348
    Likes Received:
    78
    I have just started suffering a similar problem but with a high corner, using the BLTouch with ABL. It is doing the same, touching all 9 points and calibrated to the center as well but seems to always have a high corner.
     
  4. Kanedias

    Kanedias Well-Known Member

    Joined:
    Sep 12, 2015
    Messages:
    406
    Likes Received:
    124
    The left side Y carriage bearing clamp screws were totally loose. I wonder would all of those benefit from locktite as well as the pulley grub screws. Having tightened it back up I now have an odd click sound from the right side as the Y moves past about half way which I don't like. I wonder if I should do the carriage commissioning again.
     
  5. Syko_Symatic

    Syko_Symatic Well-Known Member

    Joined:
    Sep 4, 2015
    Messages:
    348
    Likes Received:
    78
    Ahh the click, that can be the X-Axis rods slipping if they are not tight enough. Or the Y axis is out of alignment.
     
  6. Kanedias

    Kanedias Well-Known Member

    Joined:
    Sep 12, 2015
    Messages:
    406
    Likes Received:
    124
    Yep the y axis was out by 1mm.

    I can't get my head round how it can so consistently be misreading one location when nothing else has changed.
     
  7. Syko_Symatic

    Syko_Symatic Well-Known Member

    Joined:
    Sep 4, 2015
    Messages:
    348
    Likes Received:
    78
    Well mine is completely crazy? I have checked the bed (it's level), measured the difference between the bed and the frame on both sides (it's level), yet the probe is reading a 2mm difference across the bed! It has to be a similar issue as yours the axis has to be out!

    So checked all the bearings and the axis and it still is reading 2mm across the bed. Checked the sensor mount and it's all tight. Not really sure what else there is.
     
    #7 Syko_Symatic, Jan 10, 2017
    Last edited: Jan 10, 2017
  8. mike01hu

    mike01hu Well-Known Member

    Joined:
    Apr 22, 2014
    Messages:
    957
    Likes Received:
    166
    Don't Locktite the screws but do check them once in a while. It is worth checking all screws at least once after initial assembly, including the frame screws, but particularly all the carriage screws because the plastic does compress and distort over time until stability is achieved; I did this three or four times in the first six months of use and still do this now but less frequently as things have settled down. This is true for both kit and pre-assembled machines. The click, as @Syko_Symatic says, is due to loose rods.
     
  9. W1EBR.Gene

    W1EBR.Gene Well-Known Member

    Joined:
    Apr 10, 2016
    Messages:
    188
    Likes Received:
    26
    How are you measuring the levelness of the bed? For me, it is how closely the nozzle tracks the bed surface. I don't care if the bed is perfectly level relative to the frame because the frame can be off somewhat...

    So, when you say the bed is level, are you saying that the nozzle that has the same bed clearance at every location on the bed?
     
    #9 W1EBR.Gene, Jan 13, 2017
    Last edited: Jan 13, 2017
  10. Kanedias

    Kanedias Well-Known Member

    Joined:
    Sep 12, 2015
    Messages:
    406
    Likes Received:
    124
    I'm not measuring it at all. I know it's not level, or trammed to be completely accurate. The advantage of auto bed levelling is the 4 corners are measured and their skew recorded and the z stepper actively compensates. Or it should.

    As the ir sensor goes to each corner as part of auto bed levelling routine it should ensure that the z gap at each of those points is the same.

    But it isn't - when I print a single layer disk at each corner one of them is visibly different because the z gap is greater. I just can't understand how.
     
  11. W1EBR.Gene

    W1EBR.Gene Well-Known Member

    Joined:
    Apr 10, 2016
    Messages:
    188
    Likes Received:
    26
    Ah, ok. Which version of Marlin firmware are you running?

    The IR sensor may not be doing that because it doesn't measure distances per se. It looks for the distance from the bed at which the intensity of the reflection of the IR light source can be measured and is the same in both receiving sensors. For a given surface, this approach gives consistent results, but different surfaces will give different results. If the surface is different in the 4 corners of the bed, even for minor reasons, you might get different heights at which the sensor triggers.

    One way the surface can appear to vary is if the sensor picks up on the white grid lines on the heat bed. To see this effect, while the printer is powered on, after running Auto Home, set the z-axis height so that it almost triggers. Then with steppers disabled, slowly slide the carriage over the surface of the bed and watch the trigger indicator LED on the IR sensor flicker on and off. Since the white lines reflect the IR light source more efficiently than the black background, the sensor will trigger at a greater height over the bed than if the light source is reflecting off of the black background.

    Try putting a piece of black paper under the glass, and then run the bed leveling and see if that fixes your problem. Whatever the results are should help diagnose the problem.

    I hope this helps.
     
  12. Kanedias

    Kanedias Well-Known Member

    Joined:
    Sep 12, 2015
    Messages:
    406
    Likes Received:
    124
    I'll give it a whirl and report back. Is it worth cleaning the sensor itself with something like meths? I don't know if it's susceptible to dust or anything. I'm on RC6? I think... but I haven't changed it in quite a while, long before this issue occurred.

    One other thing that may be worth mentioning is the clearance of the bed is less than I'd like. When bringing the bed up to touch the extruder head the bed is pushing against the rod end covers and I can see the Perspex flex. I;m going to replace the 15mm standoffs with 20mm ones to compensate.

    I'm using the printer so much for things for work they've decided to buy one. I've recommended a Taz6 so it;l be interesting to see how it compares. I nearly went for a N2 Plus but was put off by the lack of bed levelling features and closed nature of it all.

    The auto bed levelling was so good when it worked, it's a real bummer that it's not working. I can still print but I was previously able to dial the z gap in so well so it stuck but kept the first layer dimensions from deforming.
     
    #12 Kanedias, Jan 13, 2017
    Last edited: Jan 13, 2017
  13. W1EBR.Gene

    W1EBR.Gene Well-Known Member

    Joined:
    Apr 10, 2016
    Messages:
    188
    Likes Received:
    26
    Let me know what happens with the black paper.

    I wouldn't use more than use a microfiber cloth to clean the IR LED and transistors.

    There is significant height allowance in the hotend position inside the carriage; I had the hotends up too high in the carriage and had the same problem you are having. You can loosen the adjusting screws and gently rock-twist-and-pull the whole assembly lower down toward the bed. You can also move the linear bearings downward in the bed so that they don't hit when the bed is at z=0mm
     
  14. Kanedias

    Kanedias Well-Known Member

    Joined:
    Sep 12, 2015
    Messages:
    406
    Likes Received:
    124
    Okay done and the results are interesting. The paper definitely made a difference. The deviation at one corner, while still there, appeared to be less in that the other 4 points weren't as close to the bed. A is before adding black paper, B is after.

    IMG_0176.JPG IMG_0177.JPG
     
  15. W1EBR.Gene

    W1EBR.Gene Well-Known Member

    Joined:
    Apr 10, 2016
    Messages:
    188
    Likes Received:
    26
    Do you think that all of the motion related parts run smoothly and are firmly held in place?

    My net from this is that with the IR sensor, Manual Mesh Bed Leveling makes the most sense. In that procedure, a single point on the bed is used by the IR sensor and a set of z offset values (offsets from the single measured point) for different x,y locations on the bed are set up manually, then must be saved, and then will be used by the firmware during printing for some number of initial layers.

    To consistently trigger at a given height above the bed without regard to the bed's surface reflectivity, a different type of probe is needed. fwiw- I bought the BLTOUCH probe since it should work regardless of the bed's surface or material from which the bed is made. It isn't a perfect solution but I think it makes Auto Bed Leveling a usable option.
     
    #15 W1EBR.Gene, Jan 14, 2017
    Last edited: Jan 14, 2017
  16. Kanedias

    Kanedias Well-Known Member

    Joined:
    Sep 12, 2015
    Messages:
    406
    Likes Received:
    124
    I was looking at how to enable mesh levelling and noticed this under the Auto Bed Leveeling settings

    Code:
     // Set the number of grid points per dimension.
        // You probably don't need more than 3 (squared=9).
        #define AUTO_BED_LEVELING_GRID_POINTS 2
    Hmm.... 9 point ABL


    If only I could remember which firmware I last uploaded, I _think_ it was Alex9779 Marlin-BigBox-Pro-RC which worked but I don't know if there is a way to tell.


    The longer bed standoffs are on, no issues and still have 300mm of Z travel and more room at the top end. Not really worth calling a mod as such but a nice little quality of life fix. Imo.
     
    mike01hu likes this.
  17. mike01hu

    mike01hu Well-Known Member

    Joined:
    Apr 22, 2014
    Messages:
    957
    Likes Received:
    166
    The problem with ABL, as you have found out, is that minor changes in reflectivity of the bed affect the results. The chances of all nine points being equal are not high and I would not use this. MBL is fine as the IR is not used but if I was going for ABL I would fit the BLTouch so that a true mechanical contact determines the level.
     
  18. Kanedias

    Kanedias Well-Known Member

    Joined:
    Sep 12, 2015
    Messages:
    406
    Likes Received:
    124
    Hmm, thats worth considering - do you know the best place to get one in the UK?
     
  19. Alex9779

    Alex9779 Moderator
    Staff Member

    Joined:
    Sep 4, 2015
    Messages:
    2,405
    Likes Received:
    731
    Just one thing to the BLTouch: it is nice though I am having one major issue with it. When moving at certain speeds like default travel of around 3000mm/min I get resonance. This results in the PIN rotating in the housing which make 1. a sound which is not so nice while printing and 2. when levelling it has an influence on the accuracy of the sensor. For the grid levelling with the DuetWifi I had to reduce the travel speed between points to 300-500mm/min to make the PIN not rotating. Then I get consistent results. This may very from motion system to motion system. So this might not be a problem for you but have a look at it when trying the BLTouch...
     
    mike01hu likes this.
  20. Kanedias

    Kanedias Well-Known Member

    Joined:
    Sep 12, 2015
    Messages:
    406
    Likes Received:
    124
    eugh, that sounds very slow. Disappointing since I've already started modelling a replacement carriage.
    I'm going to try 9 point levelling with the IR probe first, though I've ordered a bltouch. More points may average out the inconsistency.

    Well, maybe. Worth a go anyway. I'll also put a switch on the LED lights and turn them off during levelling in case they're having an effect.
     

Share This Page