Part Wobble (Formerly: [need] Gates GT2 Belt[?])



  • Who supplies the genuine article in the UK?

    I think I'm getting bounce from my belts but not 100% sure.

    Jerk down to 450.



  • @doctrucker E3D sells them.



  • Thanks.

    They appear to be having a little problem with backlog! Taken over a week to get a stock item dispatched from E3D. There is warning on their website, but still seems odd.

    Do gates belts have ID marks on them?

    Just running through some temp-tower tests to see how low on acceleration I have to go to get rid of the post corner wobble/ghosting.



  • @doctrucker said in Gates GT2 Belt:

    Do gates belts have ID marks on them?

    I think so but that does not stop people from copying them. 😉 I will only switch to Gates belts once my no-name belt is... no longer what I am happy with. 😂



  • On reflection I think I've been barking up the wrong tree. The wobble appears belt related but not necessarily bounce, as I'd expect that to drop quicker.

    So maybe belt tension related...

    0_1550745890503_DSC_3060_red.JPG

    I'll let the acceleration test finish before tweaking anything else. May find a better value for features even if it doesn't fix my wobble.



  • 20t drive and toothed idler.

    No belt twisting.

    Seems to be worse when the slicer slows a layer down due to it being short.



  • @doctrucker That's not just infill showing through is it? The part looks thin so I guess it's solid with no infill but just a thought.....



  • Good check, that's had me before! 🙂

    PLA @180C, 0.4mm nozzle (E3D V6, std thermistor), 0.24mm layer, 0.42mm wide extrusion, 50mm/sec perimeter, 40mm/sec external perimeter, 20mm/sec small perimeter, 0% infill, three perimeters, 3 layer solid base.

    https://www.thingiverse.com/thing:1478258/files

    I'm ignoring the numbers on the part, just using it as something super simple (been burnt by test pieces trying to do too much at once) and currently trying to fix the simplest feature of all - vertical wall! Set Slic3r to tweak the settings for me based on z height eg:

    ; Layer [layer_num] jerk change gcode.
    {if [layer_z] < 10.0}M566 X1700 Y1700;
    {if [layer_z] >= 10.0}{if [layer_z] < 20.0}M566 X1500 Y1500
    {if [layer_z] >= 20.0}{if [layer_z] < 30.0}M566 X1300 Y1300
    {if [layer_z] >= 30.0}{if [layer_z] < 40.0}M566 X1100 Y1100
    {if [layer_z] >= 40.0}{if [layer_z] < 50.0}M566 X900 Y900
    {if [layer_z] >= 50.0}{if [layer_z] < 60.0}M566 X700 Y700
    {if [layer_z] >= 60.0}{if [layer_z] < 70.0}M566 X500 Y500
    {if [layer_z] >= 70.0}{if [layer_z] < 80.0}M566 X300 Y300
    {if [layer_z] >= 80.0}{if [layer_z] < 90.0}M566 X100 Y100



  • @doctrucker That looks like ringing to me. Changing the belt won't cure it. The motor, current, acceleration, and speed all affect it. Take a look here: https://forum.duet3d.com/topic/5951/periodicity-of-ringing



  • Cheers for the link. Clearly got a big chunk of reading and fiddling to get through. Slightly complicated by slower external perimeters, but fixed global jerk.

    Acceleration is currently 500 on the perimeters, jerk 1000, and speeds as before from 1.8 small steppers running at 1000mA.



  • @doctrucker You may want to try a wider extrusion width with a .4mm nozzle. General rule of thumb is 1.2x the nozzle width for the minimum line width so .48mm should be your base line. At .42mm you could be sacrificing some strength in layer adhesion. Think of it this way, you are laying down a molten tube of plastic at .4mm width and when it lies down on top of the previous layer it will want to squish outward and flatten a little bit.



  • I'll test the line thickness later.

    I noticed the worst of the problem is at line start, and when I did a test dropping extrusion multiplier from 100 - 80% on the 80% end of the scale it appear that the undulation was thickness, rather than position. To add yet another thing to the mix this maybe extruder related!

    I'm one filament guide short of a titan extruder (git it in a bundle of e3d parts on ebay) so I'll give that a go after printing a guide!

    Edit: current extruder is ormerod 2 style with printed helical gears.



  • Can I ask what steps per mm and microstepping you're using?

    I can't tell for sure from your picture, but it looks more like ghosting than vertical ribbing in that it fades away. I've had ribbing occur from belts meshing poorly with the toothed idler. I ended up using a twist in the belt to run the smooth side on a pair of flanged bearings. Ribbing gone.



  • It would be worth trying dynamic jerk control.

    If it helps it would support the fact if it is indeed ringing or extrusion issues.

    I used it on my machine, and the little bit of ghosting I was trying to eliminate is essentially gone.

    I used 40x40 XYZ calibration cubes with 1 perimeter and no infill with a print speed of 60mm/s on the walls.



  • I thought dynamic jerk was Duet 2 only? Will check though.

    Edit: Dynamic jerk is not available on v0.6 or v0.8.5 boards.

    16 microstepping, 80 steps per mm. Extruder is around 420.



  • Those belts do not look anything like gates GT belts.

    Gates GT is a curvilinear tooth profile. Your teeth look very much like trapezoidal teeth. These teeth profile are made for transferring materials, not precise locating.



  • They're also too cheep to be gates too! I was just using that one as a size guide to show the major ripple and the belt have a similar pitch. Not the same as the belts on the machine.

    Edit: My question about ID markings for the gates was so I could tell if I had recieved gates or normal if I ordered new belts.



  • just upgraded to e3d gates belts on my corexy (yes they are genuine) what also improved quality is replacing all china pulleys/idler with the gates from e3d. They have a 5mm inner diameter. I can see less/no wobble in the Y axis when moving in X, i did with the china ones iwth 3mm inner diameter. They were not true center, even the more expensive ones.



  • Yes, the off centre pulleys are a particular frustration of mine!



  • Some better pictures of current state. There is a little ghosting but these are mostly constant.

    First side (Parallel to Y). The section of the part with text seems to effect the side after. I'm assuming the tool is being moved slightly faster here.
    0_1550785866236_DSC_3065_red.JPG

    Second Side (Parallel to X)
    0_1550785883418_DSC_3066_red.JPG

    Third Side (Parallel to Y)
    ![0_1550785895082_DSC_3064_red.JPG](Uploading 100%)


Log in to reply