Parts in Autodesk Inventor

All,

I am working on building Mendel. My last step is that I really want to understand the mechanics of the entire process.

For this next bit I will admit I am an impatient reader and could not seem to find the answer that I am looking for. First question is about a compilation of all the required parts. They all seem to be in varied forms, yet not a single one for Autodesk Inventor! I couldn't even find 2d drawings that I could use to create the part files!!

The next question is if there is anyone else who is in the Baltimore area who would like to join forces and work on this together.

Thanks to all!!

Andrew

Why don't you post this message in the Baltimore/Wash. DC (and environs) RUG forum?

---

Bob Morrison
Wörth am Rhein, Germany
"Luke, use the source!"
BLOG - PHOTOS - Thingiverse

RepRap is an open source system, and aims to be both usable and modifiable using open source tools, so there is no hidden financial cost to being able to join the RepRap community and help to improve it, and so that the removal of a proprietary product from the market can never be a roadblock preventing the project from continuing in the future.

In the area of 3D CAD software, this ideal has been compromised because there was no "good" open source 3D CAD system that the mechanical design experts felt was appropriate for this kind of work, so people have used different proprietary closed source software tools for this. Early on Art of Illusion (AoI) was used to generate .stlfiles for the project, and so you will see .aoi files in the source tree. But when it was found to be insufficient (too buggy, I think), other proprietary tools were used -- whatever those involved had access to. Apparently the folks who did that work did not use Autodesk Inventor, which (I gather!) is your software tool of choice for this kind of work.

Current developments in this area include redoing the Mendel RP parts designs using OpenSCAD, which is an open source product, albeit one that probably appeals more to programmers than to visually-oriented graphics designer folks. In other words, there are signs of improvement in this aspect of things "coming soon". For now, getting a full set of 3D CAD files for RepRap RP parts in "your favourite" file format so you can use your chosen closed source proprietary 3D CAD program is challenging.

I don't have a great solution for you, I'm afraid, although there are some STEP files available in the reprap.org svn which Inventor might be able to import (I'm not familiar with Autodesk Inventor, as you can tell!).

Incidentally, the Gen3 electronics were designed in a "free as in beer but not open source" circuit design tool, Eagle, and this (mild?) compromise too is now being addressed by some folks working on electronics design using Geda.

Personally I'm thankful for those who are working to bring these aspects of RepRap back closer to the original vision, so that people can not just build themselves a low-cost 3D printer, not just replicate that for others, but can also become a part of improving or evolving our 3D printing system for all of us, without having to acquire expensive or potentially restrictive proprietary tools.

---

---

Jonathan

Jonathan,

Thanks for the reply.

Could you send me a more precise link for the *.step files?
I suspect they would work.

Thanks again!

Andrew

> Could you send me a more precise link for the *.step files?

Take a dive here: [reprap.svn.sourceforge.net]

Which file(s) precisely depends on which machine you're after and which variant you're keen on.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

> Current developments in this area include redoing
> the Mendel RP parts designs using OpenSCAD

Are these folks around somewhere here or in the Wiki? I had a look for a free CAD system myself and decided pretty quickly against OpenSCAD, because it's script based and - more importantly - does faceted objects only, the latter like Blender. A better choice would be BRL-CAD[1] or FreeCAD[2], IMHO, as these can use "true" rounded shapes (NURBS-based or parametric surfaces).


[1] [brlcad.org]
[2] [sourceforge.net]

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

I used to race Porsches and we had the saying "you have to pay to play". I think that drafting is a similar item. A quality drafting program costs money but it is ultimately worthwhile.


Andrew

Quote
Traumflug
Are these folks around somewhere here or in the Wiki?

A key member of that set of people is Timothy Schmidt. His git repo at http://github.com/timschmidt/parameterized-mendel is one collection of these files. I think I have seen this blogged about somewhere, but I failed to bookmark it and now I can't find it. I don't think it is on the wiki (yet?). Tim is timschmidt on these forums, but has not been active here for several weeks at least.

Quote
Traumflug
I had a look for a free CAD system myself and decided pretty quickly against OpenSCAD, because it's script based and - more importantly - does faceted objects only, the latter like Blender. A better choice would be BRL-CAD or FreeCAD, IMHO, as these can use "true" rounded shapes (NURBS-based or parametric surfaces).

Choosing a tool for yourself is different from choosing one for a specific task for a group of people, though.

I'm not enough of a CAD person to be able discuss the technical capabilities you mention in depth; let's just say that, from what I can gather, there is enough momentum towards OpenSCAD at the moment that if you want to seriously propose something else, you'll probably need to put a lot of time an energy into clearly showing the community why "something else" has clear practical benefits for RP part design. And soon. If a full (or almost full) set of Mendel RP parts design files in FreeCAD format were to suddenly appear, there would be a lot of interest, but otherwise... we're very likely heading towards OpenSCAD. OpenSCAD is open source, and has the ability to store parts designs in a compact and human-readable form that quite a few of us can understand and modify (scripts). And a large percentage of the Mendel RP parts have been modelled in it already, so whatever NURBS or parametric surface capabilities it lacks do not seem to have prevented that, so far at least. If you want to take the idea of our using something other than OpenSCAD further, I'd suggest you contact Tim, who actually knows something about CAD, unlike me

---

---

Jonathan

It's not like facets are going to limit the printer resolution any time soon. And you can increase the triangle count and detail in the stl file OpenSCAD exports, to insane numbers if you so please.

sphere(r = 2); // normal sphere with 2mm radius
sphere(2, $fn=100); // this will create a high resolution sphere with a 2mm radius

I have limited experience with OpenSCAD, but personally I like to make a variable so I can easily increase the resolution for the final export, while keeping the speed up during testing.

The strongest benefit of OpenSCAD is that it's script based, so you can easily change variables to fit whatever vitamins you have. Then again, I've been a POV-Ray entusiast for 10 years, so I love the workflow.

Edited 1 time(s). Last edit at 06/09/2010 05:52AM by Nudel.

Is OpenSCAD a mesh editor? I haven't used it but I thought it was a CSG editor.

---

[www.hydraraptor.blogspot.com]

From the OpenSCAD homepage:

Quote
OpenSCAD homepage
OpenSCAD provides two main modelling techniques: First there is constructive solid geometry (aka CSG) and second there is extrusion of 2D outlines. As data exchange format format for this 2D outlines Autocad DXF files are used. In addition to 2D paths for extrusion it is also possible to read design parametes from DXF files. Besides DXF files OpenSCAD can read and create 3D models in the STL and OFF file formats.

---

Bob Morrison
Wörth am Rhein, Germany
"Luke, use the source!"
BLOG - PHOTOS - Thingiverse

nophead Wrote:
-------------------------------------------------------
> Is OpenSCAD a mesh editor? I haven't used it but I
> thought it was a CSG editor.

My post seemed a bit confusing, so I edited it to clarify. OpenSCAD is a CSG editor, but exports a stl triangle mesh, in which you of course can control the level of detail.

It does this with no regard to nice edge loops or good mesh flow for subdivision though, but it's not really important for our use.

jmarsden Wrote:
> If a full (or almost
> full) set of Mendel RP parts design files in
> FreeCAD format were to suddenly appear, there
> would be a lot of interest, but otherwise...

Well, FreeCAD imports the available STEP file just fine, so it sort of appeared already.

Nudel wrote:
> It's not like facets are going to limit the printer resolution any time soon.

Perhaps not the current RepRap, but in Stereolithography a tesselating error of 0.02 mm is required already which creates files much bigger than NURBS based geometry. Milling is very accurate as well and I'm sort of tempted to equip a Darwin with a Dremel.

> The strongest benefit of OpenSCAD is that it's script based

This is common to most open CAD systems, including OpenSCAD, BRL-CAD and FreeCAD.

I'll ask Tim, perhaps he made experiences I didn't, yet.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

> in Stereolithography a tesselating error of 0.02 mm
> is required already which creates files much
> bigger than NURBS based geometry.

I don't know how what kind of input a stereolithography machine uses. But wouldn't gcode for a RepRap be the exactly same size whether you generate it from NURBS or a high detailed triangle mesh? The only file that would be bigger is the "temporary" STL file which you use to generate the gcode. Or am I misunderstanding some basic here?

> > The strongest benefit of OpenSCAD is that it's
> script based
>
> This is common to most open CAD systems, including
> OpenSCAD, BRL-CAD and FreeCAD.

Sorry, wasn't aware of that.

Personally I think Wings3D, Inkscape and OpenSCAD together forms a powerful tool. But then again, these are programs I'm familiar with, and the workflow might seem troublesome to others. And my Mendel is not printing yet, so I don't have any hands-on experience.

> But wouldn't gcode for a RepRap be the exactly same size whether you
> generate it from NURBS or a high detailed triangle mesh?

Yes, if you have this high detailed triangle mesh, the resulting gcode is the same.

> The only file
> that would be bigger is the "temporary" STL file which you use to generate
> the gcode.

What you can't do however is to generate a high detailed triangle mesh from a low detailed triangle mesh. Similar to how you can't get a high resolution pixel image from a low resolution pixel image.

As OpenSCAD features geometry building scripts you can re-calculate your model with higher detail, but you have to do this just for building and revert it after exporting the STL. These are extra loops. Additionally, once your geometry gets more sophisticated (lots of fillets, mold release angles, etc.) it becomes more likely re-calculations at different precisions won't work identically.

Triangle mesh vs. NURBS is similar to pixel image vs. vector drawing.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

> As OpenSCAD features geometry building scripts you
> can re-calculate your model with higher detail,
> but you have to do this just for building and
> revert it after exporting the STL. These are extra
> loops.

I would hardly say that changing a single predefined variable for the final compile is much of a hassle, but you have a point.
Then again, there will probably come a user defined final export option some time, if it's not already there and I missed it.

> Additionally, once your geometry gets more
> sophisticated (lots of fillets, mold release
> angles, etc.) it becomes more likely
> re-calculations at different precisions won't work
> identically.

I would see the precision and reliability of the printer to be a more likely problem than this calculation. And the gcode would be different for every printer anyway, right?

> Triangle mesh vs. NURBS is similar to pixel image
> vs. vector drawing.

Yes, but there's no reason to export a final 1200 DPI pixel image from your vector drawing, if the magazine print is only 150 DPI.

> I would hardly say that changing a single predefined variable for the final compile is much of a hassle

This isn't about the amount of work to redefine a variable, it's about the need to do it at all.

> there's no reason to export a final 1200 DPI pixel image from your vector drawing, if the magazine print is only 150 DPI.

Modern printers don't receive pixels, they receive vector commands (PCL, PostScript, PDF) and figure the best resolution on their own. This also allows for advanced technologies like edge smoothing which aren't even known by anything outside the printer.


Apologies if it looks like I questioned somebodies hard work. Just thought it'd be a good idea to turn to more modern geometrical technologies early, as the additional amount of work should be negligible, then. Professional CAD did this step about 15 years ago and they never turned back.

Undoubtedly it's better to actually do work with meshes instead of just talking like (currently) me.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

> This isn't about the amount of work to redefine a
> variable, it's about the need to do it at all.

That sir, I cannot argue against.

> > there's no reason to export a final 1200 DPI
> pixel image from your vector drawing, if the
> magazine print is only 150 DPI.
>
> Modern printers don't receive pixels, ...

True. Badly written by me. The point was why bother to use a 1200 DPI image in that case, when a 300 DPI will look just the same?

> Apologies if it looks like I questioned somebodies
> hard work. Just thought it'd be a good idea to
> turn to more modern geometrical technologies
> early

You have many fair arguments, and it's an important issue.
But I believe OpenSCAD has a truly future-proof language. It stores it in simple text files, and it contains all the necessary information to generate whatever resolution mesh we will need for several years to come.
If the RepRap community embraces OpenSCAD, it's open source nature will - and I probably have a childish optimism about this - be used to update and upgrade the program as necessary, to suit the new needs when they come. I see no reason why the gcode couldn't be generated directly from OpenSCAD for example? Or why repsnapper won't be able to read OpenSCAD files directly some time?

Also, my apologies, as I have a tendency to become a tad zealous when discussing themes around interactive modeling vs hand coding. Again, I blame POV-Ray.

"You know you have been raytracing too long when ... Your friends are used to the fact that you will suddenly stop walking in order to look at objects and figure out how to do them as CSGs.
-Jeff Lee"

> Undoubtedly it's better to actually do work with
> meshes instead of just talking like (currently)
> me.

Frankly, I don't know if we have a program to generate gcode from NURBS?

> But I believe OpenSCAD has a truly future-proof language.

Looks like an OpenSCAD script to FreeCAD script converter would be a good idea at some point. FreeCAD comes with a "Export to povray" function, btw.

> Frankly, I don't know if we have a program to generate gcode from NURBS?

Probably not, as it's faster to tesselate surfaces to a precision sufficient for the task at hand, first.

The CNC industry has done some steps into moving the toolhead directly from NURBS, though. This probably would be the ultimate goal as it saves both, triangle meshes and gcode and grants the maximum possible surface quality. Not to mention such a strategy can react to differences between geometry intended to create vs. geometry actually created. The LOM process, for example, calculates the next slice depending on how thick the previous slices/sheets came out.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY

I still don't understand this argument. Correct me if I am wrong, but my understanding is OpenScad is a CSG modeller and it only converts to meshes to render or make STLs. In that respect it is like all other solid modellers that render and export STL. It isn't like Blender, where the internal representation is a mesh. That being the case the only deficiency seems to be that you have to manually change the accuracy between rendering and exporting, whereas that is normally done automatically as two different preferences.

---

[www.hydraraptor.blogspot.com]

nophead Wrote:
-------------------------------------------------------
> Correct me
> if I am wrong, but my understanding is OpenScad is
> a CSG modeller and it only converts to meshes to
> render or make STLs.

A CSG isn't geometry in it's self, but a recipe to build the geometry. If you have a CSG, you can create NURBS based geometry from it just as well as triangulated geometry. High end CAD systems like Catia or SolidEdge use CSGs as well.

BTW., Tim answered my PM and he has a very strong point:

> OpenSCAD's human-readable
> text-based format can be modified by multiple
> people in incompatible ways, while still allowing
> the changes to be merged sanely back into mainline

Now that's a true advantage over all CAD systems I know so far.

---

Generation 7 Electronics	Teacup Firmware	RepRap DIY