Can you guide us through the calibration process that you did, and can you explain what you did? It is good to verify what you consider as 'right', in a perhaps new frame of reference, is actually 'right'.
It would also be interesting to print a cylinder of 20mm to see how it comes out. It could confirm or rule out some issues, and determine if perhaps the corners and accel/jerk have any influence on this offset.
Thank you very much for your quick response.
I just printed a 40mm cylinder and measured the cylinder in our QS Department with a laser measuring device. I measured a minimum diameter of 40.08mm and a maximum of 40.54mm. Im using the original Ultimaker Material and profile settings. Im just printing a knew cylinder after I tighten the long belts. The small belts were tightend enough. I will post the results when the print is finished.
Edited by GuestI just get the print finished. I still have the same dimensions. I tightend all belts (long and short). All axis are prober oiled and the extruder head is thight. There is now difference. I have the problem since unboxing the Ultimaker.
The cylinder issue might be caused by the axes not perpendicular. This much error is probalby visible if you slide the head around near the ends of travel and look down from above.
I'd rather concentrate on the cube for a bit which wouldn't be affected by axes not perpendicular (it would affect diagonal distances more which are more affected by overextruding corners anyway).
Did you use a caliper on the cube? Did the caliper touch the corners or only the flat faces? Because a difference of .05mm due to the corners is normal and easy to fix in cura. A difference of .05mm on the flat faces is most likely caused by backlash.
Note that similar to designing for injection molding - getting the last bit of error removed will necessitate modifying the part in cad to compensate. But this much error you shouldn't have to. Also you could just change steps/mm in the code.
Ooh - another thought - maybe your pulleys that the belt goes around are not drilled through the center such that they are off-axis. Maybe they wobble a bit causing one axis to speed up and slow down instead of going a constant speed and if you printed a 25mm cube the error would be the opposite. Usually this is visible if you look at a pulley and push the head around. This error I've never heard of on the UM3 but used to happen on the UMO years ago.
I just printed a cube after I retightend all belts again. The cube is 20.08 by 20.07 on the surfaces im fine with this result. The cylinders are still not round. I will check if the axes are perpendicular. Then I will print another cylinder. When I want to change the steps/mm, how can I do that.
One of the downsides of changing steps/mm is this affects your print volume. I forget the X volume on a UM2 but let's say it's 235mm. If you lie about steps/mm on X axis by 10% and say it's fewer steps/mm (to compensate for shrinkage) now it will be moving the X axis 10% further and when it gets to 235mm it will have long since hit something on the right side of the printer. But if you are changing by less than .5% this is not a concern.
Some versions of Marlin let you change steps/mm right on the control screen in preferences. I don't remember if you can do that on UM3. You can also change this with gcodes. Ideally I would ssh into the printer but you could just "print" the gcodes - put them on the usb stick and print the file.
You can put the um3 into developer mode and then use ssh to connect to your printer's IP address. username/password are ultimaker/ultimaker. This puts you into a powerful utility (type LIST to see some options).
sendgcode G28 X0 Y0 ; homes x and y axes
sendgcode G0 X100 ; moves to position 100 - you can measure how far it actually moved
There's a gcode to retrieve most settings including steps/mm for all axes. I forget what it is. Should be in here somewhere:
Or you can try "list" and that might display the value somewhere.
I believe the default value is 80.000 steps/mm. That's the value for UM2 and I think UM3 is the same.
sendgcode M92 X88.0 ; 72 steps/mm - should move 10% farther now test with G0 (move) commands - maybe home X axis first?
sendgcode M500 ; saves new steps/mm - if you don't do this you lose this setting when you power cycle
Doing a "reset to factory defaults" on the front panel will put the printer back to default steps/mm.
I haven't tried messing with steps/mm on my UM3 printers yet. Just UMO and UM2. But it should work fine. UM printers are designed to be experimented with. All the source code is also open source.
I haven't tried messing with steps/mm on my UM3 printers yet. Just UMO and UM2. But it should work fine. UM printers are designed to be experimented with. All the source code is also open source.
Isn’t possible (afaik) to save them since the um3.json (or was jedi.json?) inside share/griffin/griffin/machines will override any change done to the printerboard
Editing the jedi.json is the ‘easy’ way to change the steps.
I would highly avoid doing that since it could affect the switch mechanism. Is easier to use horizontal expansion values on cura/s3d to change the size. Ofc if the problem isn’t the same on x/y then there’s a mechanical problem somewhere. X/Y squareness could be the cause.
For a step by step manual of how to SSH to the um3 to edit the json file read this bondtech manual from page 12 and edit the steps of any motor, or even the purge amount, prime positions etc
https://drive.google.com/open?id=0B7A1MEUY39HCLTNPcUJNUndfUXc
Edited by Guest
1
Good point neotko - if you mess with the x or y steps/mm you also have to recalibrate the "switch nozzle movement". There is an entry for that in the calibration menus but I've never tried it.
Years later... it could also be the Y axis bearing. Push the nozzle left/right and then front/back. If it's much worse in front/back then it could be the Y axis bearing. You'll know because when the nozzle goes backwards a little, the top of the print head will come towards you a little.
Years later I still have no problem with my Y axis bearing but neotko had a loose bearing from day 1 with his printer. He fixed it with a shim.
1
ArunC posted a topic in UltiMaker Cura,
.thumb.jpeg.0b7a05eafc09add17b8338efde5852e9.jpeg)
Dustin posted a topic in Firmware,
Recommended Posts
gr5 2,167
That's not right. It should be better than that.
I can think of a few possibilities but the most likely I think is "play" also known as "backlash".
The most likely thing to cause backlash is loose belts. Check that both sides of all 10 belts (short belts and long belts) are similar tightness. Inside the side blocks there are belt tighteners (springs) that should make the tension the same but you can slip a tooth. To equalize the tension on the long belts loosen the pulley screw on one of the pulleys, redistribute the tension, make sure the axes are parallel (it helps if you push the head to one end) and then re-tighten. For the short belts, loosen the 4 motor screws outside the machine, push down firmly (about 5 pounds force) on the motor while retightening the screws.
Friction can also cause backlash - the problem will not be the same on both axes so this makes it easier to test. Push the head around and pay attention to the friction. one drop of light oil on all 6 axes can help.
The head could be loose. Push the nozzle around but not hard enough to move the head. Is it wobbly? it could be in the bearings or something loose in the print head. This is not likely.
It might not be backlash at all - did you caliper the corners as well or only the sides? If the corners are causing the problem then that's not backlash - that's overextrusion on the corners - make all printing speeds the same (for example infill same as shell) and keep the speed low for perfect corners - I recommend 20 to 25mm/sec if you want perfect cube corners.
Link to post
Share on other sites