Layers disappearing when using raft + print one at a time mode

[Izionn6 0]

Hi,

When slicing my models, some intermediate layers on one of the model simply disappear. It occurs when I use at the same time a raft and the "print one at a time" setting.

I'm using the latest Cura version, the 5.7.0-beta.1 but the bug also occured on the 5.2.1 version.

No problem occurs when :

-slicing each model separately

-not using raft (skirt and brim looks fine)

-using the "print all at once" setting

 

I attached the cura file if needed 

The part with missing layers :

The other two looks fine :

The Xray view reveals nothing : 

layersMissing.3mf

[GregValiant 1,410]

The models have errors.  Mesh Tools picks out open seams really well (non-watertight) but finding "flipped surfaces" and/or duplicate surfaces is a challenge.

 

This is the file opened in MS 3D Builder.  All three models have errors.  Flipped surfaces and duplicate surfaces will cause a slicer to get confused and skip layers.

 

This is in PrusaSlicer.  You can see that the layers are missing or incomplete.

 

Meanwhile, back in Cura, once the model is repaired (MS 3D Builder) it slices as you would expect.

Edited March 30 by GregValiant

[GregValiant 1,410]

This is odd.

I opened the 3mf in MS 3D Builder and it showed errors.  Cura and PrusaSlicer both sliced the models as if they had errors.

I exported the models from Cura as an STL file and surprisingly the exported models don't show any errors in either MS 3D Builder or when uploaded to https://formware.co/OnlineStlRepair.

In addition, after clearing the build plate and then importing the "exported" models they sliced fine.

 

The bottom line here is that once again I'm clueless.  I'll guess and say that something happened to the models during translation which caused them to slice incorrectly and re-translating the files fixed the error...but I don't know if that is true.

 

This is the report from the Formware site regarding the "exported" STL file.  Everything is copacetic.

-> Analysed your file:
--> 0 Naked edges (?)
--> 0 Planar holes (?)
--> 0 Non-planar holes (?)
--> 0 Non-manifold edges (?)
--> 0 Inverted faces (?)
--> 0 Degenerate faces (?)
--> 0 Duplicate faces (?)
--> 0 Disjoint shells (?)
-> Repairing: 100.00%
----- Repair completed in 2123ms ------
-> Vertex count changed from 2391 to 2391 (+0)
-> Triangle count changed from 4794 to 4794 (+0)

 

The surfaces bothered to show up.

Edited March 30 by GregValiant

[Izionn6 0]

So, is the CAD (in my case FreeCAD) badly export the models or is it the translation/rotation that's messing up my pieces ?

Here are the models I exported from FreeCad and their files :

joint3mmDeep.obj servoGuider.obj servoSubGuider.obj FreeCadFiles.zip

[Slashee_the_Cow 482]

FreeCAD is usually pretty good on the export, though nothing is perfect - it could be that Cura doesn't like your .obj files and would work if you exported them as .stl (I always export my files from FreeCAD as STL). What (I think) @GregValiant is saying is that this worked for him:

1. With your starting file in Cura, for each object select it and go to File > Export and save as an STL.
2. Clear the build plate (delete everything in the scene).
3. Open the STL files you just exported to put them on the build plate.

 

[jaysenodell 18]

Fellow freecad user. 
 

I have had much better success using mesh workbench to create and STL from a body then export the STL for cura. Just use the default mesh workbench settings. Everything’s is much smoother with that workflow. 

[Izionn6 0]

Unfortunaly, nothing worked out. It seems like the print "One at a time" setting is the one messing up my pieces :

- Here, with "All at once" setting there is 12 layers in the top part

- But, with "One at a time", there are only 10 layers

Furthermore, the piece that is behind (same but with bigger top part) got his 13th layer (of the top part) removed.

 

For now, I will simply print using "All at once" until maybe someone find a solution or the bug is patched. Thank you all for your responses.

[GregValiant 1,410]

I don't think the problem here is in One-at-a-Time but rather the "Slicing Tolerance" being set to "Exclusive".

 

@rburema the original project file in this thread will show the problem.  I found that setting the slicing tolerance to "Middle" made the problem go away.  (The "Layer Height" of the project is 0.2 and the Raft Air Gap is 0.3).

 

Exclusive and Raft Air Gap of 0.3:

 

Middle and Raft Air Gap of 0.3,

 

 

Edited April 1 by GregValiant

[Izionn6 0]

Setting it to middle doesn't totally solve the issue because when in "All at once" mode (in middle tolerance), there is 14 layers on the top circle. 

In "one at a time", there is only 9 (still middle tolerance). So 3 just simply disappear and one get send inside the raft (?) (printed at the start of layer 81) :

[rburema 26]

I think there are at least 3 different issues on display here, which is part of what creates the confusion.

• Like people already mentioned: Cura expects models themselves to be 'watertight'. (I think this is a reasonable requirement for anything working volumetrically, like a slicer.) This isn't really guaranteed out of the box by some modelling software. Cura will do its best nonetheless but might mess up even in seemingly simple cases when things aren't watertight. (Conversely, it might also largely succeed for what seems like a pigs breakfast of a model...)
• In the initial preview visualisation of Cura (so right after slicing, as opposed to re-loading in the gcode), there is a hard to track-down bug that sometimes occurs causing the appearance of missing layers. We have had the hardest luck reproducing this, even though we very occasionally see it ourselves, and thus have had no luck fixing this. (And as a visual only thing that doesn't actually mess up prints and mostly goes away on its own, it doesn't get a ton of priority...) -- A tell-tale sign is a should-be higher layer seemingly appearing on the build-plate though!
• Thirdly, there may or may not be an actual big with the slicing tolerance. (This is one of those waterbed features that's hard to balance in the sense that if you fix one thing, another is likely to regress...) Hence why it's still in experimental after all these years. (Or at least one of the things that actually has a good reason to still be in that category.)

You can (apparently) fix the first issue by exporting and reimporting in Cura, in this case at least. You can see if the 2nd thing is an issue (if it keeps happening), by reloading the already sliced .gcode in Cura -- it uses a slightly different method of visualisation then (with some info missing), but it doesn't have the bug either. The last item is 'fixed' (well, workaround'ed) by setting the slicing tolerance to middle (as said).