Keeping it square.

[Jeff Short]

What is the best way to square up a printer upon assembly and how often does everyone check to insure it stays that way?  Has anyone seen issues after tightening up the belts?

Any tips, tricks or hints.

-Jeff

[mark.r...@gmail.com]

I built a square subframe for the XY stage using 40x40 t-slot and verified squareness using a machinist's square and by measuring diagonals corner to corner.  I aligned the Y axis linear guides to the XY stage frame members using metal spacers and verified with a machinist's square.  When tensioning the belts you can pull the X axis out of square if you aren't careful.  I tension one belt leaving it a little less taut that I ultimately want it (because tensioning the second one also tightens the first one), then pull the other tight and check for similar tension by plucking the belts and checking the "notes" they produce.  I check the X axis for squareness with Y by checking it's parallelism with one of the frame's cross members, and adjust the tension a little until it's parallel to the cross-bar.

The final check is to print a rectangle on the bed and measure the diagonals.  If they are the same, the X and Y axes are square.

My printer's Z axis also has it's own frame.  I align that frame with the printer's frame using a metal spacer to set the corner of the Z axis frame a fixed distance from the back of the printer's frame.  I verify squareness by first checking the motion without any belts- just run the bed up and down manually to check smoothness.  I verify squareness to the XY stage by measuring diagonals from the Z axis subframe to the XY subframe.

If the printer is relatively small you can use this to check squareness: https://www.youmagine.com/designs/alignment-and-calibration-cube

Be sure to DL the spreadsheet with it.  It will calculate the angular error based on the difference in diagonal measurements.

[Brandoan]

Its great that you can understand it is not aligned with this test but how do you implement the correction? There seems to be a disconnect on what do you physically do to correct a misalignment in a CORE XY setup, or maybe you should address in the firmware?

-B

[Mark Rehorst]

The spreadsheet is probably best for i3 type mechanisms.  CoreXY has many places where things can go wrong and cause strange results.

Both motors turn equal amounts to move the extruder in X or in Y, and a single motor runs when you move at 45/135 degrees.  If you change the steps/mm of one of the motors, you screw up all those relationships.  

In an ideal printer mechanism, the belts will be parallel to their respective axes, the axes will be perfectly orthogonal, and nothing will flex.  Relative belt tensions shouldn't matter.  You can get close to that, but not perfect.

In a real mechanism, you want it to be square (X and Y axis guide rails have to be orthogonal) and the belt tensions pretty close to equal (otherwise, they will skew the X axis relative to Y).  Adjusting the tension of the second belt affects the tension of the first belt, so you have to go back and forth until you get them about equal.  The guide rails have to be rigid- if the rails flex, belt tension may bow the X axis rail(s) (depending on how the rails are mounted) slightly and pull the Y axis rails inward.  The belts segments that vary in length depending on the extruder carriage position all have to be parallel to their respective guide rails, otherwise belt tension and mm/step will vary with extruder carriage position.  Prints will be distorted and the size and shape of the print will depend on where it prints on the bed.  If you're using stacked belts/pulleys, tilting pulley axles will probably contribute to print inaccuracies, too.

So, you have to look at the printer and figure out where any non-ideal conditions exist and try to fix them, and realize that getting it perfect may be impossible.