PrusaSlicer 2.3.0 slicing errors
 
Notifiche
Cancella tutti

[Risolto] PrusaSlicer 2.3.0 slicing errors  

  RSS
CraigRK
(@craigrk)
Active Member
PrusaSlicer 2.3.0 slicing errors

Hi there,

I have drawn an object in openScad (probably irrelevant as I draw all my items in that) which I am attempting to print, but PrusaSlicer is showing issues (holes) with the object when sliced.

I have exported two projects as 3mf files in the attached (Anet) zip file.

 

The first (Anet_RibbonCableClamps_xmotor.3mf) contains one copy of the object and as you will see it shows a slice missing from one of the screw holes to the edge between layers 21 and 22.

 

In the other I loaded another copy of the object (which I had exported rotated 90 degrees from openScad) and the defects come at the same place if I drop the object onto the bed (to the same orientation as the 1st) and defects at a different location if the object is rotated to a different angle.  The second object has defects at layers 155-156 & 160-161.

Note that while I exported these two objects separately at different orientations, when they are the same orientation in PrusaSlicer they exhibit the same defects.  But the defects are different depending on the orientation, so seems to be with PrusaSlicers interpretation rather than the objects themselves.

 

Please advise.

Best Answer by Diem:

The problem almost certainly lies in the openSCAD file, and it's not your fault.

The issue arises when computed differences (one shape cutting bits out of another) come to values that differ in the umpteenth decimal place leaving an impossible to print micro-thin, invisible, remnant. The workaround when this happens is to make the cut-out bigger than it needs to be so that it fully penetrates.

This is a physical manifestation of the difficulty in translating from one number base to another, some decimal values cannot easily be represented in binary.

So increment the cut depth by a tiny amount, render and slice again.  If the problem persists come back and let us take a look at the openSCAD script.

Hth, Cheerio,

Postato : 08/02/2021 4:32 pm
Neophyl
(@neophyl)
Illustrious Member
RE: PrusaSlicer 2.3.0 slicing errors

I'll take a look when I get home after work but I can tell you now that the majority of posts we see with 'xxxx part of my models are missing' have all come from Openscad.  Its notorious for not generating manifold models unless you are very careful.

Postato : 08/02/2021 4:46 pm
CraigRK hanno apprezzato
Diem
 Diem
(@diem)
Illustrious Member
RE: PrusaSlicer 2.3.0 slicing errors

The problem almost certainly lies in the openSCAD file, and it's not your fault.

The issue arises when computed differences (one shape cutting bits out of another) come to values that differ in the umpteenth decimal place leaving an impossible to print micro-thin, invisible, remnant. The workaround when this happens is to make the cut-out bigger than it needs to be so that it fully penetrates.

This is a physical manifestation of the difficulty in translating from one number base to another, some decimal values cannot easily be represented in binary.

So increment the cut depth by a tiny amount, render and slice again.  If the problem persists come back and let us take a look at the openSCAD script.

Hth, Cheerio,

Postato : 08/02/2021 5:07 pm
AndrewF329, CraigRK e bobstro hanno apprezzato
CraigRK
(@craigrk)
Active Member
Topic starter answered:
RE: PrusaSlicer 2.3.0 slicing errors
Posted by: @neophyl

I'll take a look when I get home after work but I can tell you now that the majority of posts we see with 'xxxx part of my models are missing' have all come from Openscad.  Its notorious for not generating manifold models unless you are very careful.

Thanks.  I'll go experiment a bit.  I noticed that the "Fix in Netfabb" is not available, so I assumed that this meant that PrusaSlicer thought the object was manifold.

I've been using OpenScad & Slic3r for years and not seen this before, but of course these are newer versions of everything.

Postato : 08/02/2021 5:58 pm
CraigRK
(@craigrk)
Active Member
Topic starter answered:
RE: PrusaSlicer 2.3.0 slicing errors
Posted by: @diem

The problem almost certainly lies in the openSCAD file, and it's not your fault.

The issue arises when computed differences (one shape cutting bits out of another) come to values that differ in the umpteenth decimal place leaving an impossible to print micro-thin, invisible, remnant. The workaround when this happens is to make the cut-out bigger than it needs to be so that it fully penetrates.

This is a physical manifestation of the difficulty in translating from one number base to another, some decimal values cannot easily be represented in binary.

So increment the cut depth by a tiny amount, render and slice again.  If the problem persists come back and let us take a look at the openSCAD script.

Hth, Cheerio,

Hiya,

Thanks for the detail, what you are saying makes sense, so I will have a tinker.  As noted before, these days I exclusively use openSCAD for my modelling (as I have done for at least 8 years or so) and this is the first time I've encountered this issue, hence reporting it here.

The script is not very elegant, so I'll have a tweak and see if I can make it more elegant which at the same time might deal with the issue.  (Specifically I was being lazy on some of my rotations, etc). 

In terms of the cut depths, I learnt a lot from nophead's Mendel 90 openSCAD files and leverage some of his methods, a key element here was trying to ensure that one doesn't co-planar faces between differenced objects (not sure that makes sense), but basically ensure the 2nd object is always larger (he, and I, use an eta variable which I tend to always add or subtract as appropriate).  Not sure if this is the characteristic you were talking about.

Postato : 08/02/2021 6:04 pm
CraigRK
(@craigrk)
Active Member
Topic starter answered:
RE: PrusaSlicer 2.3.0 slicing errors

Hi all,

Thanks to the input provided I have managed to solve my issue!!

At the highest level the model is made up of two elements:

  1. The ribbon cable clamp (directly referencing nophead's Mendel90 design), unioned with
  2. The corner bracket (with buttresses) for fitting to a NEMA motor. (with two mounting holes)

The issue for those who looked at the model was that some layers dissappeared orginating at the holes, so I thought the guilty party was the holes.  But with some trial error I discovered that the issue occurred when I unioned the ribbon clamp and the corner bracket.  By adding a "render()" statement to when calling the ribbon clamp module I was able to render an object without defects.

Regards,

Craig

 

 

Postato : 08/02/2021 6:40 pm
Neophyl
(@neophyl)
Illustrious Member
RE: PrusaSlicer 2.3.0 slicing errors

Cool glad you got it figured out. 
2.3 seems less tolerant of the common openscad pitfalls than earlier versions, although I get the impression from various responses on GitHub that they are going to take a look at that in the future. 

Postato : 08/02/2021 7:13 pm
CraigRK
(@craigrk)
Active Member
Topic starter answered:
RE: PrusaSlicer 2.3.0 slicing errors

@neophyl

Thanks (PS: I did debate which of the answers to vote best answer as you both steered me in the right direction).

Now that I know about the issue I can possible work around it when it happens.  At least it means I can continue with my design work.  That part is now printed and I'm designing the next piece. 

I'm updating an Anet printer I got essentially for free (well I paid less than the cost of the E3D Hot end and BLTouch) to incorporate features of the Mendel90 and Prusa designs.  (Keeps me out of mischief).

Postato : 08/02/2021 7:38 pm
Condividi: