Summary

I did end up with a working PCB, but it took too long and was only just useable, I have gone back to Ferric Chloride.

Instructions

I bought a PCB Drill from RS, part number 547-616
I fitted it to my printer, wired to hot end supply (so I could control it).
I was using Eagle 6.3.0 Light
I haven't played with this since July 2013, so all this is from memory, hope it helps.

CarV2 was the Eagle project I was using
(basically a car voltage meter, inside zip)

ulp is the directory containg the code
inside ulp is pcb-gcode-3.5.2.11.zip containing original
see ulp\docs\readme.html
Originally I modified the ulp source, that version is in MyOld.
I cannot remember if I modified the current ulp.
I definately created profiles\MetricPrusa.pp
(choose GCode Style tab in PCB-GCODE and pick RepRap Metric Prusa)

I used double sided carpet tape to stick the PCB material to bed.
I only had shears to cut PCB, so was not flat enough.
A pain to mill with enough accuracy to mill copper without milling fibreglass.
I used 30 degree engravers bits, wore out one bit per PCB.
Maybe the drill was too light, maybe the drill was too slow.
Took a morning to route a 2" x 1" PCB.
Photos are of failed PCBs, I never took photos of final working unit and I have no idea what I have done with it.
Gave up and went back to good old ferric chloride.

FAILED RepRap PCB Routing