Strange slicing behavior

JoeMickley

Repetier-Host V2.1.6, Cura slicer. Simple part, L shaped with holes thru in 2 axes. The STL file clearly shows the holes and that they are all of the way thru the part. After slicing, the Gcode model shows that the holes do not go all of the way thru the part. Further, the fill-in is on some intermediate layers. The hole starts at the foundation, closes off part way up in Z, then opens back up on higher layers. Only happens on the holes that are perpendicular to the build plate. Horizontal holes else where look fine. Went clear back and re-did the model from scratch, same result. Further, there are 3 holes side by side, but they fill in differently on the intermediate layers.

I have made similar parts on numerous occasions and have never seen this before. What in the world is going on here?

Repetier

Is the model manifold? Non manifold models get exactly such problems.
Also there are are some correction methods that can also influence handling of holes especially for non manifold models. They define how these are "fixed" which is not always what you want. In W10 there is a repair function for non manifold models.

JoeMickley

As far as I can tell the model is manifold. I have hit into that in the past and Repetier warns that it is not manifold at import. It did not do so this time, so I assume it is OK. Is there some independent method to check for that?

Some further experiments have produced unpredictable results. Sometimes it works, sometimes not (generally it is OK, the funny holes are the exception.) Orientation of the part (rotation about the vertical axis) seems to have some role in it, but I have been unable to define a rule set to use to avoid it.

Repetier

Is the outside color uniform? CuraEngine is very sensitive with face orientation. In host inside faces are purple and should not be seen. Wrong face orientation does not make a model non-manifold so that can also be a reason.

JoeMickley

I see a number of faces and parts of the STL model that I can accept as purple (pale, but contrasted with a blue on the other faces. but I am hard pressed to call these "inside faces".
Some of them are the vertical perimeter of the outside walls of the part. There are 2 vertical round posts that show partly blue and partly purple on their outside surfaces. They are split about into quadrants by color. The outside faces show as 2 blue and 2 purple.

The holes likewise show partly blue and partly purple, also split approximately in quadrants but I do not understand how they could be considered "inside" surfaces. (My definition of an inside surface would be the interior surface of the triangles that make up the STL structure. Is that the correct interpretation?)

I am quite willing to post the STL if that is possible. It is a simple part and not large.

Repetier

Click on the gear next to STL name and anaylse. It shows if normal orientation is wrong, but very much sounds like it. In first tab in that window is also a fix normals action - hope it works in your case. Does not always do as expected but since it is manifold you have good chances.

JoeMickley

This AM I did some additional tests, still trying to determine a rule set.
Trial 1 - Created a simple cube, shelled it out to a constant wall thickness, open at the top. single hole thru the bottom central to the part, single hole on the -Y face central to the face. Saved as STL, into Repetier. All faces show solid blue, Cura slicer produces a correct part with the holes thru.
Trial 2 - Part from trial 1, rotated 90 Deg in Z, side facing hole is now at -X. Out as STL, into Repetier. Repetier shows some faces in purple, quadrants in the hole as purple. Cura slicer fails to reproduce the bottom hole.
Trial 3 - Part from trial 1, rotated 90 Deg in Y, open face is now toward -Y, both holes are horizontal. Out as STL, into Repetier. Repetier shows some faces and hole parts in purple, Cura appears to slice correctly, reproduces part with both holes.
Trial 4- STL part from Trial 1, rotated, re-saved as STL. Repetier shows only minor streaks of purple in the bottom hole, all faces are blue. Cura fails to recognize the central bottom hole.

From all of this I deduce that any rotation of the part prior to STL export will produce a corrupted STL, This is a problem in my CAD system, not in Repetier or Cura. I have been accustomed to rotating parts to favor fabrication by the 3D printer. It would appear that I must orient the part in its original form when constructing it to avoid this problem. I am not fond of this solution but if it works, then so be it.

To your suggestion, I used the part from Trial 2 and anaylsed it. It shows that the part is manifold and normals are oriented. Nothing to "fix".

Question: Can Repetier rotate parts after STL import to orient them more favorably for printing?

Repetier

Yes, host can rotate. Just select the rotate icon (3 arrows in circle) and enter rotation angles.

JoeMickley

Excellent. Found the rotation stuff, tried it out, just what I was looking for. I will leave the warts in the CAD system for another time (Actually, now I don't care other than taking the precaution of not doing part transformations. The mystery of why that is true does annoy me but I have been annoyed before. Good results override annoyance).

Thank you for your assistance. I now have a solution I can work with to optimize orientation for better printer fab.