Layer shift



  • Hallo,
    Have anybody ever had problems with a shift of the layers while printing?
    OK, let me explain the problem.
    I constructed a calibration cube 20mm edge length. After slicing, using Kisslicer, I started to print. The print went well without any warning. The result: a test cube with shifted layer. Each layer shows the values 20 x 20mm, but every layer is shifted about approx. 0.5mm to the prior one, all in the same direction.
    What I did next was checking the g-code from Kisslicer, but the data seems unremarkable to me. For cross-checking I unload the filament and fix a pen at the printer head instead. The sketch shows the explained behavior.
    Have you ever seen a printing behavior like that?
    Or is it a problem with Kisslicer after all?

    Greetings from Germany
    Nik



  • Sounds like loose belts or skipping steppers.
    1. Does it appear that it is catching on anything during a certain movement?
    2. Are your belts tight?
    3. What are your stepper motor currents and are your stepper motors making any weird noises and not moving freely?

    Bonus Points. What kind of movement system are you using?



  • What do you mean with movement system?
    The printer is a KosselXL. Obviously there is nothing the print head or the slider could be catching on. All Belts are tight, the motors can be moved freely and sound like they expected to.
    In the mean time I discussed the problem with a friend of mine and we have the idea to turn up the current a bit. The idea is based on the following fact: the shifting direction leads always to one of the pillar so the consideration was, that there might be something with the motor in this corner. Well, the failure and its symmetry points out a systematic error. If, theoretically, this stepper is losing some steps while going to the next layer (z-movement) the slider wouldn’t reach the correct height and the diagonal rod takes care that each layer is pushed in the opposite direction. So far the theory.
    What I try next is increasing the limitation of current for the motors (60% to approx. 70%) to insure the correct behavior of all motors.
    Thanks



  • I've had this with tmc2100 but they are very much inferior to tmc2660 on duet. Turn the current up, you can go as high as it takes so long as motors don't get too hot (max 60 deg C is my rule of thumb) 70-80% rated current is usual.

    It is likely to be a slipping pulley, belt, etc.. Its just the most likely cause. Do your motor shafts have flats on them? Are the grub screws aligned with the flats and tight?

    Common problems occur commonly.



  • Yes, there are flats on the motor shafts and surely all grub screws are tight and aligned.
    Let me see if the real world follows the theory or has its own physics…
    I`ll report later…
    Greetings
    Nik



  • That idea sounded just too simple, don’t you think?
    Increasing the current leads to …. nothing.
    The test cube is still shifted.
    What I’ll try next is changing the motor (pillar) assignment on the board. Due to the symmetric setup of the Kossel type printer that seems to be possible without any risk. Thereby I can encircle the source of my problem. If it’s a problem with the board, the shifting direction of the print also will move to the new assignment, if there is a problem with the mechanical setup (incl. motor) the failure won`t move.



  • good idea, dont forget to swap endstops



  • Lots of these in the last few years:)
    Loose grub screw / flowing pulley usually causes a wavy noisy print in one direction, loose belts also tend to make constant artifacts along the print.
    Layer shifts are often due to motor losing steps while moving fast on longer distance. This can be introduced by more friction on the shafts, e.g. bearing lip sucked in which I had a few months ago.
    The motor copes with the load until it encounters a long fast move.
    Another thing could be the heat built up in driver. Had this with allegros, half print was ok then it started to hiccup because the driver was cooling itself by switching off.
    Most of the problems I had with layer shifts were due to friction and bearings, both ball and sliding tubes like IGUS.
    One more thing to make sure is the wiring of the motors, loose connectors etc. They can introduce a mess also.



  • Hey folks, thanks a lot for your great hints, but I think I had a break through tonight.
    To say it with monty python: and now for something completely different!
    While tinkering at the Kossel, swapping the motor assignment, trying this and that suddenly the stepper at the suspicious pillar made a real bad sound. Great, finally there was something I could track.
    To make a long story short: it was the gardener . OK, no. What my problem was, is the tube, which leads the filament from the extruder motor to the hot end. The extruder motor is mounted at the “tricky” pillar as recommended in the manual. A PTFE tube act as a Bowden guide. Although this tube is installed in a gentle arch it acts as a spring, especially when filament is pushed through. This push impulse has influence on the balance of the Kossel print head.
    As the last test this night I printed my cube while relieving this “spring” by holding the tube so it going in the print head relaxed.
    And … surprise: the cube is really a cube. All the other actions and theories was just ghost-hunting.
    Now I have to think about installing the Bowden tube in a relaxed way.
    Nik



  • Glad you're making progress. As for the bowden tube how about getting rid of it altogether? I can see the point of them but they are a workaround for the problem of trying to get the mass off the effector (or carriage). There are numerous inherent reasons why they are bad, the filament is a loose fit in the tube and basically has to be or it would bind (the problem you've been having), because of this the filament can spiral inside the tube under pressure, exhibiting elastic properties, and it might even have some elasticity f its own. All of this adds up to far less precise filament control.

    Sure you can buy new fangled capricorn 1.8mm bowden tube with (and I struggle to believe this) even slipperier teflon, you can put cooking oil on your filament cleaning sponge, you can use pressure advance on RRF/Duet to offset the elastic effect caused by the bowden tube. These are all great bodges, but are blown entirely out of the water by two other approaches. The flying extruder, which shortens (by around 2/3), and straightens your bowden tube moving the extruder to a platform suspended on elastic above the effector, You can also achieve this like Deckingman has on a corexy with a passive or active second gantry for your extruder/s and the flexible drive extruder where a remote (and stationary) motor turns a flex-shaft to drive a direct extruder on the effector or carriage.

    I wouldn't have been able to see the difference if when I had a kossel mini, I didn't also have an i3. The difference in precision of laying down filament directly rather than via the tube was very apparent.

    If you are going to keep your bowden tube make it as short as possible, mount the extruder midway up a tower and eliminate as much bend as possible.


 

Looks like your connection to Duet3D was lost, please wait while we try to reconnect.