Prints rotated about the z axis
Posting this question out of curiosity rather than it being a current issue. I'm in the process of rebuilding my printer so I have no way to test suggestions right now and I suspect the problem will go away once the rebuild is complete. That said….
I have a delta printer, based on the Griffin but heavily modified, using Duet WiFi + IR Z probe and Magnetic joints/arms. Calibration (6-factor) resolves to an error of around 0.015, which seems reasonable but I don't have a reference. My heightmap.csv also shows a high degree of flatness to the bed, albeit very slightly tilted (I don't have the details to hand, but all of the probe points are green).
Despite this, XY moves (and, by definition, prints) are rotated 10-ish degrees about the z axis. See photo:
I'm guessing that a mechanical misalignment is causing the problem but my knowledge of delta kinematics is way too poor to reason it out. My best guess is that one or more of the PLA (I know, I know) corners has deformed slightly and twisted the frame or is tilting the bed. Because the error is consistent, the calibration routine can account for it. The only issue i'm noticing is that the first layer is very tricky to lay down cleanly at the moment, I'm wondering if the corrected path of the nozzle tip is to blame if it's no longer perpendicular to the bed.
Does this sound reasonable? Part of the rebuild work is to give the printer metal corners (Robotdigg, rather than Kossel 25K), so plastic deformation won't be a problem any more. Just wondering if there is some other potential cause that I should look out for during the rebuild.
Just to be clear, using normal delta geometry the X axis is parallel to a line between the X and Y axes, and the Y axis is parallel to a line from bed centre to the Z tower.
The only things I can think of that would cause a 10 degree rotation are:
1. You specified a rotation in your slicer when you sliced the part;
2. Your X, Y and Z tower angle corrections are all about 10 degrees. Auto calibration will adjust the X and Y corrections if you ask for 6 or greater factor correction, but not Z.
Strange, I thought I'd posted a reply earlier, must have previewed but not submitted.
Your X Y geometry agrees with mine!
- It's not the slicer, as the same happens with jogging from Paneldue
- The towers are not out of line by that much.
I've checked and the calibration delivers the following:
M665 L288.280 R82.110 H270.360 B125.0 X6.141 Y6.048 Z5.830
I should have measured the angle of rotation rather than eyeballing it, but 6 degrees is about the right order. If the bed was tilted at a line that intersects the centre of the XZ horizontal and the Y tower, I'm guessing that would have a similar impact on apparent tower positions. If that is the case I suspect one of the horizontals (that the bed rests directly on) has moved somehow.
Thanks for all your help by the way David. It never ceases to amaze how active you are on all sorts of different forums and groups and still find time to develop the hardware and software. I've got a Duet WiFi, PanelDue and IR Probe and they are all top-notch and come with great documentation. I see a future with the PCB carriage and effector added to the collection too, just don't tell my wife….
The problem is the Z5.380 that you have in your M665 command. Calibration never changes that value. My guess is that you mis-copied it at some stage. I suggest you set that Z value to zero, also subtract that value from the M665 X and Y parameters, and re-calibrate.
That's so strange, I must have made a mistake when copying it, as you say.
As soon as I get the printer back together I'll zero it and retry.
Thanks for your patience!