I have lately been thinking about trying to setup a wire edm with emc2 control, but the details of the edm functionality really are sparse. One guy on cnczone has reported doing in on emc .. search for user 'mdynac'. Implementing it as proof of concept and implementing it functionally are two different things...
I think Andy is right, that the current limitation of emc is that it cannot go in reverse. I was thinking last night (a bit abstractly) that if you could output all the axis positions at a given interval (based on feedrate) and have the most recent ..say .. three positions retained, you could use those as destinations for a program that could reverse motion. Basically when the wire short is detected, the existing program pauses, and the "two steps backward, one step forward" could run, before resuming the original program. For purposes of a wire edm, the requirement to go in reverse is often only for very very small distances (of a very recently traveled path). I think the challenge would be to output positions fast enough, as while as record delete the oldest and reorder the newest into a small program. I hear computers are good at that sort of thing, so it may not be too bad. Toward a functional implementation, I would not know where to begin..... maybe it could even be done with variables for recent positions and a canned cycle.
If anyone has suggestions, I would be glad to hear them. I just happen to have a dead wire machine that I have been thinking about trying a retrofit on, and it would make a great test platform. It currently is complete with motors and a deionized water setup, so motor drives and of course the power supply are the challenges. For the power supply, I have stumbled upon the "Garden of EDM" design (easy to google) that seems really well thought out but sadly has no reports of success. Perhaps the whole edm subject is a hobbyist white whale, when attempted beyond simply 'proof of concept' levels ?
