Ive been having an issue lately with my new machine
Ive printed a calibration cube and got the flowing dimensions
X = 19.89
Y = 20.02
Z = 20.01
as you can see my x axis is .1 off which is huge when you are trying to print mating parts.
i’ve tried to tension my belts 3 or 5 times but when i follow bambu labs guide it looks like when i loosen those 4 screws at the back of the machine the belt stays tight and i dont know if thats normal or not.
other things ive tried
done all the calibration methods in orca slicer including flow/PA
xy comp
different materials
cleaning the carbon fiber rods (no dirt as the machine is almost brand new)
Ive been trying to avoid scaling as that is a band aid fix
my old ender 3 and ender 3 s1 were getting within .05 consistently without any xyz step calibration
Rotate the part 90º on the build plate and print again. Does the short dimension move to Y or stay with X?
The geometry of the machine is “fixed”, they’re all going to be pretty much the same mechanically. So if this is a mechanically induced error, others should see it as well. And stepper motors don’t have variability in step distance. Tell one motor to rotate 0.1º and every motor given the same steps will move the same distance. You don’t need to calibrate steppers. I can’t think of a way that a CoreXY system could be wrong on one axis only unless it was never right to begin with (meaning the error is designed in and everyone should have the same issue with X). If the motor is losing steps and the machine isn’t detecting it, errors would be present in both X and Y.
Belt tension will result in repeatability problems. The effect of loose belts is pretty much the same as the effect of increased backlash in a jack screw. Each time the print head passes over a given point on a new layer, it’ll be at a slightly different position than the prior layer because of the belts. This effect will be random in nature (especially since this is a CoreXY design and both motors work together to achieve position in both the X and Y directions. A loose belt on either side will affect both X and Y positioning accuracy). Loose belts wouldn’t always cause one axis to be “short”.
Hey man thanks for your detailed reply. The part I’ve been printing is a 20 mm featureless cube so I doubt that rotating it would do anything but I’ll give it ago.
I have heard of other reports of other people having the same issue that they have compensated for by scaling the model in either the x or y directions. Which is something I don’t really want to do as it’s more of a band-aid fix.
You might be right in that my machine is defective. Might return it if bambu lab doesn’t offer any solutions as I think .1 is kinda unacceptable compared to an ender 3 from 2016
There is unquestionably a constant somewhere in the printer’s FW that specifies how many stepper motor steps there are per mm of motion. That constant needs to be tweaked. In the CAM software that runs the G-code for my milling machine, the stepper motor step constants are exposed and can be changed. BBL could almost certainly do the same thing (there may already be a G-code that can change these values).
I’m wondering whether the 0.1mm inaccuracy is an additive constant or a multiplicative factor. E.g., if you print a calibration cube of double the size does it stay 0.1mm or does it become 0.2mm? I believe the answer would tell something about where the error comes in.
I also noticed the same problem of difference in precision between the 3 axes. I printed a 60mm x 40mm x 5mm cross and the results were 59.90mm in x, 39.99 in y and 5.03 in z.
I redid several tests by changing different printing parameters and the gap was always substantially the same.
Then I printed a cross 3 times bigger, namely 180mm x 120mm. The result was 179.91 x 119.97. The error is therefore not proportional to the dimension but constant regardless of the dimension of the part. Like any manufacturing machine, the printer has limits of accuracy. It would be interesting to know the value given by the manufacturer, this one not appearing in any technical description. I would not be surprised wirh 0.1 mm for manufacturer specification.
Just for information : Creatily K1 and K1 Max specifications
The belt and gantry system used by most FDM printers has inherent limitations when it comes to positioning accuracy. Z tends to be more accurate because those steppers are driving ball screws. But X and Y use belts and belts stretch and flex (and when the stepper moves, it actually takes time for that “impulse” to propagate down the belt to the print head). It’s actually a repeatability limit. If you zoomed in to the surfaces you’re measuring, you’d see peaks and valleys. You’re measuring peak-to-peak since calipers can’t reach the layers that shifted inward a bit. So the average error is probably 1/2 of what you measure. Not that it helps any if you need better accuracy.
Still, these are very good numbers. The plastic is going to change size as a function of temperature, and that effect could easily make for larger errors. I think I remember reading that ABS shrinks by almost 7% from molten to solid (which is why it warps so easily). That temperature effect doesn’t go away after printing. Heat up the print, it’s going to get bigger. Cool it down, it’s going to get smaller.
It’s interesting to see if it’s talking about over all dimensional accuracy or just in a single axis. For example my true fully calibrated material actually requires a 0.07 offset to the xy compensation to get the diamensions I’ve gotten. The problem is that it seems to be happening only in the X axis which makes me thing there is a hardware fault somewhere
Where as a lesser machine a 300 dollar ender 3 is way more accurate within 0.03 consistently when printed in a controlled temperature environment
