3d file format for mold making

Hi All,

I always use IGES files when I send my final parts to China for injection molding. Only recently I had to use a different format because my file had a custom texture on it (embedded in the 3d file). IGES seemed to be too big and complicated so that’s why I sent a STL file (as well as other formats just in case).

The molder couldn’t deal with any other format that IGES. After a lot of back and forth I had to pick a crappy texture from a mold tech book.

Is this normal or is it the molder? Personally, I always thought that molders or manufacturers could process STL files for injection molding. In the toy industry for instance there must be a bunch of stuff made with freeform or Zbrush that don’t export in IGES.

As an additionnal information, I modeled the part in Solidworks and textured it in Zbrush (which wont export in IGES).

Any clues, answers or tips for next time?

Thanks in advance,

I would avoid STLs since their accuracy is so dependent on you (wrt chord lengths, angles, etc.) and in the end, you still exporting triangles… You can try STEP files, but I like sticking with IGES for the final export, regardless of how large they might end up. I Zip those suckers down…

Did you try a STEP file?

Do you also supplement with GD+T 2D Drawings?

That’s actually exactly what I do. I submit a IGES or STEP and ALWAYS include a control line drawing. Never send the 3D data without one.

What about a complex surface that is impossible to dimension? Do you just put in the critical dims? I’m fighting with our engineering group about this and would appreciate all opinions on the matter.

I’ve found that parasolid files in .XT format work well to send out solid assemblies for manufacturing - I would avoid mesh formats like STL.

It’s been an ongoing argument in our office lately about how detailed our dimensioned drawings need to be.
I believe it depends on the complexity and how critical the geometry being dimensioned is. –
My colleague always raises a good point - “we’re sending over the CAD – if they need a dimension they can just measure off of the cad themselves…” And while I understand this argument, it’s short sighted. It’s important to have recourse beyond just a solid file in case someone screws something up. If you receive non-conforming parts, it’s far easier to point at the drawing, date and page number and show clearly that you required a specific dimension\tolerance for that part.

This can become a legal thing too…

That being said - Our manufacturers no longer rely on the drawings to creating tooling paths - so it doesn’t need to be that detailed.

Very good question. No, I do not typically fully dimension the 2D drawings for the parts I design. The overall net shape and features is what the 3D file is intended to describe. The control-line drawing I’m referring to is intended to identify the critical dimensions for the part’s function like boss features that match up with other parts, overall dimensions, other mating features, etc. The drawing also contains information like material and finish specs.

The way I look at part drawings is this way: In conjunction with your 3D file (which will primarily be driving the cutter paths to create the mold), your drawing serves as the contract between you and the molder. Those dimensions explicitly define what features are critical to the part’s function. This is the drawing that the QC guy will use to check the part once it’s molded, and more importantly, the drawing you will reference if the part is delivered to you or the client out of spec.

There’s also a cost implication here. The more dimensions you add to a drawing, the more critical dimensions the QC has to measure and the more dimensions that part has to hold. All of that (can) drive up the part cost. Personally, I try to be pretty stingy with my dimensions because of this. Plus, like you alluded to, so features, like the organic outside surface that doesn’t mate to anything else, can very very difficult to measure accurately and serves no real benefit (Of course this all depends on your specific situation.)

The only time I do completely dimensioned drawings is for parts that are actually fabricated like sheet metal or weldments since they aren’t using a 3D database to cut a mold. The only quasi-exception here is that I give lay flats for sheet metal parts provided we (the sheet metal fabricator and I) agree of the proper stretch factor (k-factor) for the material bending.

I hope this helps. It’s really important for designers to understand the role that the 3D file vs the 2D drawing play in communicating with the manufacturer.


The file I have created was based on a model done in Solidworks (NURBS), exported to STL (triangles) format with very very little triangles for highest accuracy/tolerances, and then textured it using ZBrush. I have saved the textured file in STL format and theoretically this project should be done and over with.

However, as mentionned by vandebar, the molder simply cannot work with that type of file apparently, although another of my clients oftentimes uses the Freeform method to produce toy figurines (for injection molding I can only presume at the moment), and AFAIK Freeform produces triangles much like what my final is composed of.

So we have hit some sort of wall, being on one side, ZBrush/Freeform giving us really complete freedom of the final shape design, and on the other hand if we use some of the more traditional ‘texturing’ post-process techniques used in steel injection molds where a texture is ‘applied’ on top of a model, this forces us to use one of the molder’s palette of predefined textures, which does not come close to the result we wish to accomplish.

Is there someone to blame ? Has anyone ever done this before ?

I’m not sure about this one… If you’re doing a texture, those are typically applied chemically/photographically. Is the “texture” is more complicated or specific than that, then it should be integrated into the actual 3D model, but then I would still rely on IGES.

Anything more complicated than that and perhaps you should considering building the model an alternative way…