Cura ignoring small gap

[spumpsalive 0]

I'm trying to print a model that has two faces pretty close together, but not touching (about 0.3mm clearance). When I slice it in Cura, it "bridges" this gap, which messes up the part.

 

The part is available on Thingiverse here: https://www.thingiverse.com/thing:5228740 (0.5mm gap version). I've also attached a Cura project file in case there's any trouble reproducing.

 

I checked the STL file in Blender and FreeCAD; neither turned up any mesh errors. If I scale up the part to ~150% or more, the part slices correctly, which also makes me think the model is OK. It seems to be something about the small size of the gap.

 

Cura also seems to be using top/bottom layers to create this "bridge"—that doesn't seem right either.

 

I've tried toggling every setting I can think to try—mesh fixes, resolution/accuracy, stock presets, etc.—but nothing seems to have any effect. How do I get Cura to respect this gap, instead of ignoring it?

 

Un-sliced model; the "cup" and "ball" don't touch at any point:

 

Slicer output; two sides are "welded" together:

 

Iso view of a middle layer to show cross-section:

CE3PRO_Chip_Clip_.5mmgap.3mf

 

One more view; "welding" doesn't occur on outside 15-16 layers:

 

Environment: Cura 5.3.1 Flatpak on Fedora 38

 

Thanks!

 

[obewan 37]

FYI

I sliced your project file using cura 5.2.1 and it sliced perfectly, so it seems to be another problem with cura 5.3.*

 

[Torgeir 250]

Hi @spumpsalive,

 

Welcome in here.

 

Can confirm your issue with Cura 5.3.1, -but work Ok with Cura 5.3.0 Beta - Xmas version..

 

So appear it's a "thing/bug" in between those two versions?

 

Thanks

Torgeir

 

[GregValiant 1,330]

I can also confirm.

It's troubling in that this would appear to be the type of bug that indicates that 5.3.1 cannot be trusted.

This is the report on the model from formware.co online stl repair:

--> 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 7378ms ------
-> Vertex count changed from 29547 to 29547 (+0)
-> Triangle count changed from 59098 to 59098 (+0)

 

This is the first model I've submitted that got straight "A"'s.

 

EDIT:  Looking into this, toggling "Print Thin Walls" on and off will cause that cross-over to disappear and re-appear.  In particular the "Minimum Feature Size" seems to play a role.  When it is a small number the slice is fine.  Once it gets above 0.31 the slice is wrong.

 

Edited May 12, 2023 by GregValiant

[GregValiant 1,330]

It was only a question of time.  This morning This Bug Report showed up on GitHub.

 

This is a demonstration model.  The slit on the right is .15mm and the left slit is .50mm ( .05mm increments).

5.3.1

 

4.13.1

 

In my estimation, they aren't even close.  My earlier comment that 5.3.1 can't be trusted would appear to be right on.

Edited May 14, 2023 by GregValiant

[Torgeir 250]

Hi Greg,

 

right you are, -even ver 5.3.0 has this issue too.

To me it seems like when it come to small details it become worse.

So, if I'll like to try those new versions, I'll always use the monitor to inspect the sliced object before printing.

However, version 4.13.1 is the version I'm using and find very good.

I'm using one small test object, the "ctrl V 3D test" -and if a slicer can do this one well I'll know the UM2 print it fine.

This test object demand using a 0.25 nozzle to be perfect, but can be "some" acceptable using a 0.40 nozzle.

 

I'll just put up this picture for anyone (Greg knows this very well) to see how accurate Cura could be.

Here's how it's look in the monitor using Cura 4.13.1:

I'm using Cura standard profile, except for the "brim", as I'm using "skirt"

The minimum distance between walls: 0.5/0.4/0.3/0.2/0.1mm.

This test object can be found here: https://www.thingiverse.com/thing:704409

 

Torgeir

 

 

 

[GregValiant 1,330]

5.3.1 with a .4 nozzle

 

and 4.13.1 with a .4 nozzle

 

Different models, different continents, same result.  A user should not have to deal with that, or with this.

 

 

[spumpsalive 0]

Thanks all for the great response. The "Print Thin Walls" option is a helpful workaround for my case—if I run into further issues as a result, I'll roll back to 5.2.x. Let me know if there's any way I can assist further.