y slope with 3.5.0-rc1, not seen with 3.4.6

Adrian52

@gloomyandy not so far - can do. Didn't think they were that high for a delta.

Adrian52

@balajiramani hi - I just tried vase mode in orca slicer, and it printed vertically for me. I am using a recent 3.5rc1+ (from 2023-11-1), so that may be the difference .

Adrian52

@gloomyandy just tried with 900 x, y, and z jerk. The y slope remains the same. I originally tuned the jerk up to 2100 to control sharp corners bulging.

balajiramani

@Adrian52 Can you point me to the link where I can download the 3.5rc1+ firmware for Duet Wifi? I would like to try that out.

Adrian52

@balajiramani I think this is it
https://forum.duet3d.com/post/326478

balajiramani

@Adrian52 Thanks for the link. I will give that a shot and report back.

balajiramani

@balajiramani I tried the 3.5rc1+ firmware and the slope is no longer there.

Adrian52

@balajiramani is a cube with 15% fill vertical for you?

balajiramani

@Adrian52 Haven't tried that specifically. Let me try that and report back.

balajiramani

@balajiramani I just tried a cube with 15% infill (not sure how much that matters) with the 3.5.1rc1+ firmware. The slope is back. I am reverting back to 3.4.6 for now.

Adrian52

@balajiramani phew - glad it's not just me. I find zero fill (and no top) prints vertical - not tried other fill levels.

balajiramani

@Adrian52 Are you using vase mode or regular mode with zero infill and no top surface? When I use vase mode for a cube that is longer in the X direction as compared to the Y direction, I get sloped prints. If the cube is longer in the Y direction, then it prints fine.

Adrian52

@balajiramani I have been using regular mode rather than vase mode. Will check out varying x:y ratio with no fill on my setup.

Adrian52

@balajiramani
tried with unfilled cuboids, with 3 perimeter walls :
30x20x20 (x, y, z) - very pronounced y slope, about 1.3mm over the 20mm z height.
20x30x20 - y slope almost gone, but still clearly there; about 0.3mm over 20mm z.
20x20x20 - rechecked this with exactly the same slicing as above, printed consecutively, and it comes out vertical.
What??

Adrian52

@dc42 The problem has definitely returned. For me, the clearest demonstration of the y slope is using a 30x20x20 cube (x,y,z), printed with no fill and no top with 3 perimeters. Strangely, a 20x20x20 cube printed in the same way appears vertical. the 30x20x20 cube printed in vase mode has only a slight slope for me. I have just reverted to 3.4.6, and the same gcode file of 30x20x20 with 3 perimeters prints vertical.

dc42

@Adrian52 please try the firmware build at https://www.dropbox.com/scl/fo/gaxzegafmnmagaefo9mgc/h?rlkey=hk4wpttdje76oews8zkuh86q3&dl=0. It fixes a likely cause of this issue.

Adrian52

--@dc42 thank you - I have tried the new version. Firstly, the y slope seems to have disappeared - excellent.
However, this version does not seem to play well with input shaping as I have it set.
When printing 30x20x20 cuboid with no fill and no top, extruder movement was sporadically irregular, which left little loops here and there on the corners.
The 30x20x20 cuboid with fill and top gave a very strange result - there was a ~1mm -y layer shift at z=7.3mm, with a very slight shift -x. I set IS to none from the console, and the cuboid printed perfectly. When I switched back to ie3 from the console, the cuboid printed with the layer shift at 7.3mm again. With a magnifying glass, the two layer shifted cuboids appear identical.
The other thing I noticed was an occasional graunching noise during the ie3 prints. This was quite brief (1sec?), and irregular, happening at 1- 3 min intervals. Not present when IS off.
I am printing pla on my delta with stand alone duet2wifi at 60mm/sec.

Adrian52

@Adrian52 have revisited jerk settings, and find everything seems to work fine with 300 jerk. 3.5 seems to be much more sensitive to too high jerk.

dc42

@Adrian52 were you using the M593 L parameter? If so, there was a bug in the processing of that parameter in that firmware.

Adrian52

@dc42 I reported the L parameter problem in a separate thread. Currently I find prints look a little better with IS switched off - I suspect a slightly higher L parameter is needed, so looking forward to experimenting with rc2.