I've written a program to be used instead of pcb-gcode.
I've tested it on Mach3 (Inactive mode) but I have not run the resulting gcode on a real machine yet.
I just received my WidgitMaster (Great Machine!!) but I'm about 4 weeks from getting everything completed so I can test it fully myself.
I'm looking for someone that is interest in the added capabilities (over pcb-gcode) to test it and provide feed back.
It run in two stages. Stage one as a ULP on Eagle to acquire raw data.
Stage two is a Windows program that processes the data and produces the gcode.
There are two big differences in this program and pcb-gcode.
#1 - you can use up to three different mill bits to do the routing. Use a Big bit to do most of the work and the smaller ones for cleaning up the tight places.
I download a test board from the Internet and operated on it with two bits. Bit one was .015 with an isolation width of .03 and the second pass using a .005 bit giving a isolation layer of .02. Most of the work with the .005 bit will be air since the .015 did most of the work. The bottom side results are depicted below. Up to three different tools can be specified.
#2 - Optimize the routing path to minimize the number of tool moves. The first example below is the tool path obtained using pcb-gcode. The other one is optimized.
I got the cart in front of the horse on this one .... getting my software developed prior to having a machine to test it on.
If you are interested in testing and providing me feed back - please let me know.
The actual board is 4.0 x 2.0 inches. The pics below are shots from the program and are enlarged for viewing.