micro wire-edm project (and software solution)

I have been interested in EDM for several years but recently after
thinking about building a small injection moulding machine gained
renewed enthusuasm for the project. I am aiming to build a small
machine designed for submerged cutting of small parts with wires in
the order of 0.05mm thick.

I am doing a little wiki on the project which can be seen here:

www.indoor.flyer.co.uk/edmwiki.htm

it is not an online editable wiki but if you follow the instructions
on the front page you can save it to disk and edit it (then email it
to me if you want).

The main part of the project that may interest people is the software:

Control of a wire EDM machines is difficult due to the need for
precise gap control based on the elecrical and physical conditions
between electrode and workpiece. The control software must be able to
increase/decrease the feed rate and on occasions reverse the machining
path a little.

My idea which I mentioned before was to record the step and direction
pulses out of the parallel port generated by Mach3/turbocnc. Once
recorded they can be "played" back and forth according to the detected
gap conditions. Origionally I was going to drip feed the pulses to a
PIC but after some thought decided that it would be better to store
the whole program at once. To do this I will use a multimedia card as
you find in your camera (up to 1Gb and cheap). It is quite easy to
interface a pic with such a card and it will also allow stand alone
operation of the EDM which will be handy because I doubt it will be
fast. A second pic will control the playing of the card to the drives.

The power supply is a work in progress, conventional u-EDM machines
seem to use RC supplies for simple sub microsecond pulse generation
with C's less than 1uF. However it may be nice to have faster cutting
with thicker wires by eiher a current limited (resisters) pulsed
supply or pulsed current-source switching inverter. I will start with
RC and then look at the alternatives. It seems however that the
inverter option is only suitable for longer pulse lengths as the
switching frequencies required would otherwise be well into MHz. The
second option is definately possible with the right mosfets but you
are building a heater.

I have looked into gap control rather a lot and it seems that although
not optimal taking the average gap voltage or average voltage accross
a shunt resister has been used quite sucessfully my several machines.

If you are interested in the project please drop me a line.

Cheers,

Graham

p.s. No lectures on the complexity of gap control and the use of fuzzy
logic etc etc please.