Model won't create valid STL file

I’ve been helping Andrew Staton (@Quadhurst) to model (among other things) mid 20th C lighting fixtures.

It was remarkably tricky to figure out the geometry, but eventually we managed to draw this one.

Andrew wants to make a few 3D prints of some of his models at 1:10th scale, but this one is proving very difficult.

It’s called a Stanton 7B bracket - a reinforced concrete bracket that sits on top of a reinforced concrete column, and carries a lantern suspended from the top on the spigot shown (at least, in the 3D model - in the original, the lantern was bolted on).

The overall height is approximately 3ft, and the sideways ‘reach’ between the centre of the round part of the base and the lantern attachment is 1’ 9".

When exported to STL at full size, in inch units, the STL on re-import into SU is no longer solid. Solid Inspector 2 reports 1 stray edge, 5 Surface border issues, and 384 internal face edges.

I’ve changed the model units to metres, and scaled the model up 100x, to export in metres, but import into the slicer in mm to create the desired 1:10 scale version.

The scaled up model also reports as solid (in Solid Inspector 2) but not in SU.

I exported it to STL with metres as the unit, and reimported it to SU using mm as the import units, as well as having the model units set to mm.

It comes in at the correct size, but the STL is not solid.


I attach both the original model, made of a mixture of solid components and non-solid surfaces (as groups) created by Fredo’s Curviloft/Skin Contours, and the exploded and cleaned up model, created with a mixture of Outer Shell for the solid components, and manual editing of the exploded result combined with the exploded group surfaces.

The edges for the Curviloft/Skin Contours were copied and pasted in place from the underlying components, but it looks from the STL import that the problem surface border issues are at edges where the Curviloft surfaces are located.

Why does SU think the result of the original exploded model is not solid, when SI thinks it is?

Why is the exported STL from what SI thinks is “solid” not still a solid?

Is there any way of fixing this?

It behaves the same in both SU2021 and 2022, both the latest versions at the time of writing.

I’ve saved the models back to SU2017 to make it easier for more people to review the model and possibly help.
Stanton 7B 13 Mar - problem.skp (2.7 MB)

Interestingly, when opened in SU 2017 Solid Inspector reports over 2700 ‘Short Edges’ - mainly in and around the base, but with a few at the upper end of the arm.

I’ve tried the STL in a Cura slicer, but the slicer isn’t creating layers, though it doesn’t show errors either.

Hide some of the outer surfaces. You’ll see there are a bunch of internal faces and edges. Cleaning those up should get you sorted out.

Thanks for such a quick response, @DaveR.

We did manage to print one of a different model of Stanton bracket.

For that one, we added faces to the Curviloft skinned fillets, to make them solid, and printed the result without any outer shell or exploding.

Many of those added faces match the ones showing as ‘internal’ when you hid the outer ‘filleted’ surfaces.

We’ll try directly following your advice on this one.

Why does SI2 think the model is solid, then?

[Few mins later]
Trickier than it looks to delete internal faces in the base area. Haven’t time tonight to work it out, but the principle seems sound.

Thanks, Dave. More work tomorrow. 1:30am here - way past bedtime!

I fixed the one at the origin. Compare it to the other one and to yours.

Solid Inspector 2 isn’t perfect. There are some things that can escape it.

There are places in your model where you need to run Intersect Faces before erasing the internal junk. The area around the top of the round base can’t be right in your model. I modified that.

Stanton 7B 13 Mar - problem.skp (1.2 MB)

Yes, I got as far as seeing that, and ran Intersect faces on the curved base face, then trying to delete the ‘internal’ parts of the round base. But that didn’t immediately fix things, though it started to I think.

But I really must stop for the night now.

Will look further tomorrow (well, later today, now).

Tx again.

It’s similar in that you’ve both created models with a bunch of internal faces that need to be removed. It’s less work to avoid the internal faces from the beginning.

It depends on the model. In the case of your screw thing, if you follow the video I did exactly you won’t end up with any internal faces. That requires making the profil so it doesn’t overlap. When I modeled your screw thing I did the same thing I showed in the video. I set up the profile the same way making sure it’s the same height as the screw pitch.

Then the geometry in themiddle was deleted before adding the flat ends.

As I wrote in your other thread, if you want a hole down the middle, add that after you’ve got a solid screw component.

I see something that I can change.

I changed the vertical height of the profile to get it working. Also, I ran CleanUp3 on it and ignored Solid Inspectors’ advice. Next, it was imported into the slicer, also ignoring the warnings. the print is what I need. Unfortunately, the pitch is unknown as I am copying an existing Shop-Vac fitting.
Thanks for the Help.