X & Y not perpindicular  

Page 1 / 2
  RSS
LaserGuru
(@laserguru)
New Member

Title says it; when I print two 160mm squares in a row, they stack perfectly the way they were printed, but not if I flip one over.  It's off  by over 1mm!!!  I don't see any adjustment for fixing this.  Do I just need to replace some parts on the printer?  Is it something I can adjust with firmware if I break the little tab off?

Posted : 03/03/2021 12:24 pm
fuchsr
(@fuchsr)
Honorable Member

Any pictures you can share? It's not totally clear to me what you mean. How heigh are the squares? This could be warping of the model, which could be caused by a number of issues, including lack of adhesion to the print plate, drafts, etc. Going from the issue you described to wanting to fiddle around with firmware seems like a pretty radical step unlikely to succeed...

Posted : 03/03/2021 2:38 pm
LaserGuru
(@laserguru)
New Member

When I try to print a square (or rectangle) I get a parallelogram.  It's slanted like in the drawing, only less noticeable.  They match up when stacked in the same orientation they were printed, but not when flipped over (and they should be the same either way).  I tried different filaments as well as rectangles with different internal cutouts and it's the same result.  It's just to repeatable to be warping.   The only thing I can figure it that the x axis has to be slightly out of alignment with the y axis.  Looking at the machine I don't see any way to adjust for this.

 

Gap

Posted : 03/03/2021 3:24 pm
Turro75
(@turro75)
Estimable Member

https://github.com/s-taylor/prusa-mini-skew/blob/main/README.md

Posted : 03/03/2021 6:14 pm
stag
 stag
(@stag)
Active Member

Yesterday I have realized I have the same issue after printing happily for almost a year now. The lack of official solution to this problem is troublesome. The thing is, the problem only appears when printing large objects, for example I printed 4 full-bed quarter-circles, and when I tried to assemble them, well, see for yourself below. I would like to see an official Prusa solution that addresses this problem! I believe most of us have this problem, just unaware of it.

Posted : 08/03/2021 12:36 pm
LaserGuru
(@laserguru)
New Member

@turro75 - GoSkew worked...mostly.  The other stuff didn't.

@stag - I too  would like to see an official solution, but I doubt we will since it appears to be an integral problem...the only adjustment is to torque some rods while tightening some screws and it doesn't really do anything.  GoSkew does work, but it's super janky.

 

Posted : 08/03/2021 12:44 pm
Javacer
(@javacer)
Eminent Member

It's a problem with printer cinematic...

Posted : 08/03/2021 5:25 pm
Turro75
(@turro75)
Estimable Member

I don't hink it's cinematic, it's just how it's assembled.

I created this object which takes roughly 1h and gives a perfect idea about skew on which axis and so.

https://www.thingiverse.com/thing:4735060

Posted : 08/03/2021 6:20 pm
stag
 stag
(@stag)
Active Member

GoSkew sort of works. But yes, it takes time to calibrate it, and then you need to add the post-processing of your gcode in every slicer profile. Still waiting for the official Prusa solution.

Posted : 09/03/2021 10:13 am
stag
 stag
(@stag)
Active Member

Also, another solution for me (printing the quarter-circles, see above) that I rotate every other quarter-circle by 90° on the bed, this way the skews are balancing each other. But it is easy to be smart after wasting 600 g of filament on those quarter-circles.

This post was modified 1 month ago by stag
Posted : 09/03/2021 3:06 pm
Diem
 Diem
(@diem)
Reputable Member
Posted by: @laserguru

Title says it; when I print two 160mm squares in a row, they stack perfectly the way they were printed, but not if I flip one over.  It's off  by over 1mm!!!  I don't see any adjustment for fixing this.  Do I just need to replace some parts on the printer?  Is it something I can adjust with firmware if I break the little tab off?

So: 1mm in 320, flipped to compound the error; a little over 0.3% error.  Not too bad.  Most of it is probably contraction as the part cools.

The coefficient of linear thermal expansion for PLA is approx 41 x 10-6 m/m·K

coefficient=dL/(L0*dT)

Therefore dL=coefficient*L0*dT

Assuming dT is the drop from nozzle temperature to ambient: 210C to 20C - that is 190C

substituting: dL=41*10^-6*160*190 = 1.25mm

We would expect a shrinkage of 1.25mm in 160mm or 0.78%

The heated bed does an excellent job of annealing the part and appears to reduce the shrinkage by more than half.

Shrinkage will not be even, the molecules are extruded and stretched along the lay of the filament and we can expect more shrinkage along the lay than between adjacent strands.  As they are built in alternating layers the smaller the Z dimension the stronger the eccentricity we can expect.  Different models in different orientations will react differently.

The answer, of course, is to make a trial print, measure it, and calculate a calibration into the original design.

Cheerio,

This post was modified 1 month ago by Diem
Posted : 09/03/2021 5:51 pm
Javacer
(@javacer)
Eminent Member

There is a solution, this is to flash with firmware with scew compensation enabled.
But the mini will have to "break" the board.
It took me a day to calibrate

 

Posted : 09/03/2021 7:42 pm
LaserGuru
(@laserguru)
New Member

@Diem - The problem comes from the physical axis being out of alignment; it's easy to see with an engineer's square; don't even need to print anything to see it.

@Javacer - Any alignment that requires voiding my warranty is less than ideal.  That warranty is the only reason I haven't modified this printer...and is probably also the only reason this printer still works.

Posted : 09/03/2021 8:41 pm
Turro75
(@turro75)
Estimable Member

@laserguru

a firmware with skew compensation enabled allows adjusting any loaded gcode on the fly.

Goskew doesn't void the warranty as it does the same as skew compensation but at slicing level modifying the gcode before loading onto printer.

In both cases consider it only if You are in a hurry to print something otherwise take Your time and rebuild the frame until able to produce reliable prints.

If all parts aren't bend following the assembly instructions should be able to providing You a squared frame. 
The most important thing is the order of the 3 screws tightening during Z to Y frame assembly, after that adjusting the XZ skew has to fix the remaining.

Posted : 09/03/2021 10:22 pm
Javacer
(@javacer)
Eminent Member

@turro75

In MINI this is not possible ... Assembling with a metrological device will not give best results. The Z axis is attached with a plastic box with a 0.1 mm manufacturing error(or more, who know 🙂 ). The angle will be about 90. But I can see the angle on the printouts is not 90 degrees.
Will programmatically improve prints. And it is not necessary to flash, you can use a builder. But it was not interesting to me 🙂

Oh, I'm not sure if the parts are printed on an accurately calibrated machine 🙂

This post was modified 1 month ago by Javacer
Posted : 09/03/2021 10:34 pm
Javacer
(@javacer)
Eminent Member

  - programm scew control...

Posted : 09/03/2021 10:42 pm
Diem
 Diem
(@diem)
Reputable Member
Posted by: @javacer

There is a solution, this is to flash with firmware with scew compensation enabled

Which will have to be recalculated each time you change filament and probably more often if you make very different classes of print.

If your part dimensions are so critical it's worth designing with tolerance, or fine adjustment, built in.

Cheerio,

This post was modified 1 month ago by Diem
Posted : 09/03/2021 11:36 pm
Javacer
(@javacer)
Eminent Member
Posted by: @diem
Posted by: @javacer

There is a solution, this is to flash with firmware with scew compensation enabled

Which will have to be recalculated each time you change filament and probably more often if you make very different classes of print.

No. It not needed.

Nozzle 0.25, linear advance calibration, flow calibration, prusament PLA with minimum dimension error.

Calibration scew.

Posted : 10/03/2021 12:50 am
stag
 stag
(@stag)
Active Member

@javacer

This is exactly what I expected from the Mini as well. (As a reply for your i3 XYZ calibration video)

This post was modified 1 month ago by stag
Posted : 10/03/2021 6:22 am
Diem
 Diem
(@diem)
Reputable Member
Posted by: @laserguru

@Diem - The problem comes from the physical axis being out of alignment; it's easy to see with an engineer's square; don't even need to print anything to see it.

You might have mentioned this in the fist place.  The answer of course is to reassemble, taking care with the alignment especially when tightening.  If the mount has been compressed you might need a shim or judicious sanding.

Cheerio,

Posted : 10/03/2021 7:18 am
Page 1 / 2
Share:

Please Login or Register