Real CAD, open source

Sifting through the Reprap forums, blogs and wiki brought some interesting links to CAD items. Because Reprap is open source, Adrian and his team tend to endorse open source and freely available CAD and drawing projects. This year has seen some released that are based on the OpenCascade platform. OpenCascade was started by Matra Datavision as a new platform for their Euclid product, then the project was spun off to open source when Matra sold Euclid to Dassault Systems. It's an attempt to create an open source 3D platform or kernel, similar in concept to Dassault/Spatial's ACIS or Parasolid's X_T. Three applications that utilise it are Dan Heeks' HeeksCAD, FreeCAD and NaroCAD. As with commercial programs that use ACIS and Parasolid, not all of the capacities of OpenCascade are utilised in each of these applications, but they exchange data with neutral file formats as well as proprietary ones. These programs have the capacity to perform some sophisticated functions that up until now have only been seen in commercial apps, like filleting/blending and chamfering and thickening/offsetting of surfaces. After installing NaroCAD, it wouldn't initialise for me, so although it READS like the most powerful, offering splines and revolved surfaces that the other two don't, I can't offer an opinion yet. Of the other two, I've been finding HeeksCAD more useable for modeling than FreeCAD, and more fun. Its GUI and constraints are more integrated and more friendly, whereas FreeCAD has a sort of modular interface, with different tool palettes for 2D drafting, 3D modeling, and working with meshes. Although there are toolbars for some functions, some of them aren't functional yet, and the way they're implemented ignores contextual relationships between functions. FreeCAD might be the more powerful of the two, though, as I've loaded some big model files created in other applications to test them both out, and FreeCAD loads them more quickly and is more responsive with them loaded. FreeCAD handles some mesh/facet formats that HeeksCAD doesn't, and has more sophisticated editing and analysis of mesh objects.

I'll post more about them when I can quantify the differences.

Hobby horse

I've decided to embark on building a rapid prototyping machine. Notionally they're 3D printers, commercial machines have been available since the late 1980s and they've been getting smaller, less expensive and cleaner since then. Desktopfactory.com had been promising a sub-USD5000 desktop model, but ran up against a finance shortfall in 2009 and were subsumed by 3DSystems, the originators of additive RP, who still want to release it. 3DSystems' current model lineup starts at USD10000. Too big, too complex, too expensive for me, which is the drawback of all additive machines apart from the recent open-source projects.
The other method of rapid prototyping is subtractive, ie carving or milling machines. The greatest drawback of subtractive machines is having to cut away lots of waste.
There's a technology that's halfway between the two: Layer Object Manufacture. This involves cutting profiles that are then laminated together out of a substrate. It's a process that has a lot of promise because its resolution can be much more precise if the substrate is stable. The waste is stripped away afterward, instead of having to be cut, and the waste is default support structure for overhanging features. A recently-developed LOM machine is the Mcor Matrix, an Irish-built machine that builds prototypes out of paper. 80gsm paper is .1mm thick, the z layer resolution. Clever, but still more than USD 20000, even if the running costs are minute in comparison to its competitors.
So, I'll draw on others' efforts. Reprap.org is motivated by the idea of a production robot that can reproduce itself, so their efforts are named after the scientists who founded our understanding of evolution, Darwin, Mendel, and so on.
I like what they've achieved, but I have some reservations about their machines. As a polymer deposition machine, it builds 15 cm³ volume per hour, which makes it a less-than-rapid prototyping machine in my view. The idea of finished parts dropping into your lap is attractive, but reprap's not a mass production machine either. The appearance, accuracy and strength of its output is dependent on the deposition nozzle's feed rate and temperature and the stated accuracy of .3mm-.5mm is less than what I want. But to improve that, and to increase the rate of deposition would probably take more mass, more power, and more complication.
So I'm thinking that I'll use the reprap Mendel basic structural design to implement paper LOM. I'm going to use a very-high-speed air-powered engraving tool called "Turbocarver" to route the sheets of paper instead of using a knife blade to slice them, which would require a cutting mat surface to bear against. The Turbocarver will handle a sheet of paper with next-to-no speed reduction (it's only .1mm thick, after all), I'm thinking, and the speed (400K rpm!) should shear the paper fibres with miniscule lateral force.
More about this project as it progresses......