maintained software to generate Voronoi toolpaths for PCB mechanical etching.

[mongo]

It has been a while sense I have had to make a PCB at home, and as I mechanical etch with a hand built CNC machine I tend to use Voronoi toolpaths to help reduce the time and to be kind to the tooling.

If you are not familiar here is a link to a paper, I won't link to my own work as I have a low post count.

https://groups.csail.mit.edu/drl/wiki/images/7/7d/Vona_Rus_2005_Voronoi_toolpaths_for_PCB_mechanical_etch_Simple_and_intuitive_algorithms_with_the_3D_GPU.pdf

Anyway the application I used (Visolate) to produce the toolpaths seems to be stuck in Java 6 land and I was hoping that the functionality may be common now but under another name.

Thanks in advance.

[awallin]

FWIW, for voronoi diagrams I wrote openvoronoi a few years ago:
https://github.com/aewallin/openvoronoi
more recently a Java port was made:
https://github.com/Rogach/jopenvoronoi

these are low-level geometry libraries and a complete vd-PCB solution would require a lot of work on top of this (reading pcb/footprint/netlist files, displaying them, input/output through the libraries, and output of gerber files)

Anders

[jeremy]

I never thought of or heard of this. Thank you for sharing it!

[mongo]

Thanks Anders, I'll add that to my admittedly too long to do list. I cannot promise time lines but if I can contribute back I'll send a pull request.  Python is great because of the ability to use numpy and python-usbtmc give and I have some great ideas.

Really if, when Intel opensources the quark RTOS (not sure if they stayed on ViperOS) this month the Edison it may be perfect for EMC2 to get away from parallel ports.  This would be great for small scale CNC and I think that mechanical etching will take off as the market is flooded with inexpensive CNC routers.

 

[jeremy]

Mongo, have you seen machinekit? It is almost exactly what you are describing.

[T3sl4co1l]

Interesting, but I'd like to see it with a positive-shape optimization rather than an absolute (and abstract) method.  That is, as shown, it flagrantly ignores the shape of the ground fill, which in my designs is as critical as the shape of the signal connections (which are also blown out by this method, but signal nodes are usually smaller so this need not be as big a deal).

To include this, one would have to define a set of points that belong to the polygon, and should remain fixed... but not so dense that little or no optimization is possible.  I expect it's not as easily defined as the original case, unfortunately...

Tim