Calibration question, large circle slightly not rould

Alexander Mundy

Just got my 400mm delta going and my second test print was a 312 mm OD circle.
0_1562210665206_312mm circle.jpg

Measuring the OD in line away from the towers I get:
X tower 311.12mm
Y tower 312.29mm
Z tower 311.71mm

I was very careful making my arms to the point of fine sanding each on a disc sander calibrated to 90 degrees checking against a fixture made with 2040 extrusion for go / no go to get each the same. After epoxying the ball cup ends on I made a fixture that attached to a tower of my ball cup 250mm delta with the rods and effector removed. Using the Z height I moved the carriage to where one ball cup engaged then down to where the other lined up. All rods were measured that way and were 471.0mm cup center to cup center.

The final delta radius in CAD was 226.8mm and auto calibration calculated 226.746mm.

I am curious if it's due to the XYZ variance from auto calibration?

M665
Diagonals 471.000:471.000:471.000, delta radius 226.746, homed height 860.853, bed radius 198.0, X -0.351°, Y -0.239°, Z 0.000°

I also followed the same sand, check go / no go with a 2040 fixture until all horizontal extrusions were the same. I tightly strapped the triangles before tightening. If there is a variance of the towers I wouldn't think it's from the machined metal corners. Best I can measure from top triangle / vertical extrusion corners to bottom ones diagonally checks out.

Best I can measure there isn't any effector tilt and the volcano nozzle is only 35mm below the ball cup horizontal plane. Rods are spaced 81mm center to center and adjacent rod center lines are 12mm at the ball centers. ( "b" at https://www.reprap.org/wiki/File:Delta_geometry3.png )

I do not understand the significance of the XYZ angles to print results. Do the XYZ angles correspond to my circle variance?

dc42

The XYZ angles are the tower position corrections, for when your towers are not exactly 120 degrees apart. If you think they are incorrect and affecting the scaling, you can try setting them back to zero and then run 4-point auto calibration.

Alexander Mundy

Best I can tell by printing various test objects, changing settings, trying S4, S6, S7, S8, and even S9 I do have geometrical error(s). Rotating the arms and effector does not change the results so I am relatively sure arm length and effector geometry is symmetric. At one early point I tried entering slightly different arm lengths (less than 0.5mm different) and got some really weird thing happening with auto calibrate where Z dropped during the first probe between the 2 probes (A5 S0.02) then scraping the bed moving to the second point and had to abort. I tried it a couple of times thinking my piezo system had a malfunction but as soon as I took the additional L values out it worked again. I am using 3.0beta5+4 so there might be something wonky when putting in different arm lengths.