On the delta, the detail size is incorrect on the x-axis, on the g-axis the size is correct. Geometry checked all exactly. How can this be corrected? What parameters are regulated?

What amount of error are we talking about and does it increase or change with position?

Unfortunately delta printers are very sensitive to any geometric error and the default calibrations are mainly focused on getting nozzle to move planar over the bed so it sticks well. x-y-distortion is neglected here normally. One way to get x strech is that the extruder tils slightly when moving in x direction, also the amount would also depend on y depth so having a cube on high y might give other dimensions then on low y position. But I can not say this paramater causes it. I can say you can adjust variables going into position transformation by changinf geometry description like rod length, angles of towers. But there are also assumptions like towers are exactly perpendicualr to bed that must be fullfilles since there is no variable to adjust that. And already an error of 0.1mm can have signifacnt influence.

If you find x is a fixed percentage wrong all over bed I would suggest just scaling objects for that, but it normally is not a constant thing.

Ok 3mm is a lot. Maybe check belt tension. Steps per mm is same for all 3 belts and if they get stretched differently the result changes move distances. But of course other dimensions might be off as well. With 3mm it should be measureable.

## Comments

Unfortunately delta printers are very sensitive to any geometric error and the default calibrations are mainly focused on getting nozzle to move planar over the bed so it sticks well. x-y-distortion is neglected here normally. One way to get x strech is that the extruder tils slightly when moving in x direction, also the amount would also depend on y depth so having a cube on high y might give other dimensions then on low y position. But I can not say this paramater causes it. I can say you can adjust variables going into position transformation by changinf geometry description like rod length, angles of towers. But there are also assumptions like towers are exactly perpendicualr to bed that must be fullfilles since there is no variable to adjust that. And already an error of 0.1mm can have signifacnt influence.

If you find x is a fixed percentage wrong all over bed I would suggest just scaling objects for that, but it normally is not a constant thing.