Calibration X/Y axis

[chris 0]

Hello everybody,

I got a big problem with the accuracy of my parts.

My printed parts are always to small. For example a plate 50x70 is just 49,6x69,4.

So I do a little further tests and mounted a caliper to my machine, started Cura print run and

travel defined distances like 0,1, 1 and 10mm. The results affirmed my prits. The distance

at a travel from e.g. 10mm is just 9,65. And the travel from 1mm is just 0,89.

So I go and calibrate the hole machine. Loosen the pulleys, adjust the rods and so on. My machine is

100% straight (hope so ).

What can I do to get 10mm, when I say to Cura travel 10mm :evil:

Where is my mistake.

Is it possible, that the steper motors dont work correct (can i change settings anywhere?) and how many mm must be one rotation

Please help me to get accurace parts.

Thanks you Chris

[Daid 306]

Are you running the latest firmware? Very early firmwares had a wrong amount of steps per mm set.

You can set a different amount of steps per mm with the "M92" command.

For example "M92 X78.7402 Y78.7402" sets the default of 78.7402 steps per mm of X and Y movement.

[chris 0]

Thanks Daid, but what is the right value, that I should use?

[Daid 306]

78.7402 should be the right value...

[chris 0]

where can I see if the value is right. In a generated g-code, there is only M92 for the extruder motor, but not for the x and y axis. How to get sure i am running the newest firmware.

The only thing I can see, is the list in the printer interface. It says "Marlin 1.0.0 RC2" and M92 X78,74 Y78,74, but not 78,7402.

[joergen 2]

I got a big problem with the accuracy of my parts.

My printed parts are always to small. For example a plate 50x70 is just 49,6x69,4.

So I do a little further tests and mounted a caliper to my machine, started Cura print run and

travel defined distances like 0,1, 1 and 10mm. The results affirmed my prits. The distance

at a travel from e.g. 10mm is just 9,65. And the travel from 1mm is just 0,89.

it seems that your measurement procedure isn't completely kosher:

in your first item you are missing .4 and .6mm (or about 0.81 and 0.86% error)

in your second set of measurements you have .35 and .11mm (or about 3.6 and 12.3% error)

this means that if the XY steps would be wrong, you would see the same error rate (i.e. 1.5%) on all items. in addition, if you print a flat object with screw holes, the distance of the screw holes would vary also by the same error rate.

what you are seeing is a constant/absolut error, most likely due to shrinkage, under-extrusion and/or slicer error.

depending on your model/needs, try scaling up your model during slicing by 0.8% and see if that helps.

[chris 0]

Hi Joergen,

thanks for your reply. I don´t think, that only shrinking can be the problem, cause the measuring with

the caliper is nearly the same as print difference. Please don´t fix me on the values that I write above, cause

that was only right fast without looking to my records. The unaccuracy is not ever the same. Its not linear.

Thanks Chris

[chris 0]

sorry, but is there any other idea, what my problem is about?

[joergen 2]

sorry, but is there any other idea, what my problem is about?

a long shot, but there is a tiny chance that the belts don't mesh properly with pinions, and/or they are not aligned properly, and/or you still have slack in the system (belts not tight enough), or some of the set screws don't have proper grip on the D-shaped stepper axles. but all of this would rather manifest themselves in generally messy prints, not necessarily 0.2mm constant errors.

[joergen 2]

sorry, but is there any other idea, what my problem is about?

it would be interesting to learn how you ruled out thermoplastic shrinkage as the culprit.

[chris 0]

yesterday evening I started a test. I slowly increased the steps/mm value. Default was 78,74, like you said. I had a plate 30x30. At exact X80,15 and Y 79,85 my plate was really 30,00x30,00. Thats great. BUUUUUT then I printed with the same values a plate 50x50. It was much to big. At about 53x51.

So I don´t think I have to play with this value. Ok it was just atry :lol:

How exact are your prints for example. I can´t believe that I am the only one with such a problem.

Thanks Chris.

[Daid 306]

As joergen says, your problem is most likely shrinkage. My 20x20 boxes are 19.6mm, I just accept that plastic shrinks.

[chris 0]

as I wrote above, I did a test with a caliper. I measure the distances, the extruder went. And there were nearly the same tolerances as if I printed.

I don´t want to close my eyes from shrinking, but I don´t understand. Another thing is, that the y tolernaces are higher. When it´s a shrinking problem, would it not be the same on both axes?

[joergen 2]

yesterday evening I started a test. I slowly increased the steps/mm value. Default was 78,74, like you said. I had a plate 30x30. At exact X80,15 and Y 79,85 my plate was really 30,00x30,00. Thats great. BUUUUUT then I printed with the same values a plate 50x50. It was much to big. At about 53x51.

that is correct and totally expected. as I said before, changing the XY steps per mm introduces a linear scaling factor, in exact the same way we use the E steps per mm to calibrate the extrusion.

[joergen 2]

as I wrote above, I did a test with a caliper. I measure the distances, the extruder went. And there were nearly the same tolerances as if I printed.

I don´t want to close my eyes from shrinking, but I don´t understand. Another thing is, that the y tolernaces are higher. When it´s a shrinking problem, would it not be the same on both axes?

if it's not the same in both directions, this would point towards a mechanical problem, i.e. the set screw on the Y stepper not being perfect perpendicular.

In addition, shrinkage also depends on the cooling, especially the direction your fan is blowing from is probably show more shrinkage, since the fan cools it faster, so it shrinks faster, or the other side of the fan shrinks more because the plastic is setting slower, and has more time to affect the layers below... could be one or the other... I know ABS that is cooled harder shrinks more.

in addition, PLA goes through a significant change in density when heated up. below the glass transition temp of 60C, it is around 1.25g/cm3 for non-christaline PLA, when heated up to 210C, it expands to 1.0965g/cm3, comes out of the nozzle, and shrinks down to 1.25g/cm3 again. See http://www.jimluntllc.com/pdfs/polylactic_acid_technology.pdf, section 16.5.1.1. in other words, it shrinks about 14% in volume from molten to solid, which is enough IMO to bring it into the range of the error you are seeing.

[chris 0]

Joergen, first thanks for the detailed reply.

Sorry for my nescience, but what exactly is the "set screw" at the stepper motors?

In addition, shrinkage also depends on the cooling, especially the direction your fan is blowing from is probably show more shrinkage, since the fan cools it faster, so it shrinks faster, or the other side of the fan shrinks more because the plastic is setting slower, and has more time to affect the layers below... could be one or the other... I know ABS that is cooled harder shrinks more.

I am sure your explanation are correct. I agree with different shrinking factors cause of diverse cooler direction. Make sense, absolute. I think I will do further tests the next days.

Another thing I´d like to know is, what do all the people that download stl´s at e.g. Thingiverse. Are all the users affectedly to scale the stl´s, when they wanted the parts to be accuracy. For example to things that should filt together?

Thats not nice, isn´t it. That only works, if you know your machine (and PLA) very well.

Thanks in advanced for the replies.

Chris

[Daid 306]

When you are making parts, you account for inaccuracies in the final result. For example a 0.5mm margin is quite normal.

This is no different from milled parts, or molded parts, which are also not perfect processes.

[joergen 2]

Sorry for my nescience, but what exactly is the "set screw" at the stepper motors?

The set screw is the tiny screw that holds the pulley on the axle of the stepper:

http://www.flickr.com/photos/ultimaker/6244616864/

if the set screw is not perpendicular to the flat surface, you get a bit of slack at the end of each move.