Re: Re: EMC & Wire EDM? -  reverse toolpath

Dave

I've gotta say that my understanding of wire EDM is theoretical. I've never
programmed one and watching one cut is duller than watching paint dry.

(comments mixed in)

On Thursday 10 April 2003 07:45 am, you wrote:
> Message: 10
>    Date: Thu, 10 Apr 2003 11:40:18 -0000
>    From: "turbulatordude" <davemucha@...>
> Subject: Re: EMC & Wire EDM? -  reverse toolpath
>
> Hi Ray,
>
> If you think foam cutting as very similar to wire EDM, you can see
> that if you were cutting a curve and wanted to stop, you cannot just
> send an axis to home, it would cut across the part.

Right. But in most cases the wire need not travel back all the way along the
same path. If you cut the center out of the letter P you just need to be
certain that the cut leaving is along the same line as the cut entering. I'd
want to turn the heat off before I did that exit move so I'd program it using
a different block rather than using the same block in some sort of software
mirror.

> the only way to get the cutter out of the part would be to reverse
> the direction and follow the cut line out of the part.

Seems to me that with foam cutting you are really trying to cut something
out/away and once you've done this you can simply move the wire wherever you
want it. It does have to be a planned next move but this is true of most any
move you make with a common CNC.

> in EDM, the electrode cannot arc on the part, if it does, it is too
> close and needs to back off a little.  in wire EDM that toolpath
> might be on some part of some gear tooth.  Since the wire needs to be
> free from contact to spark, it is imperitive that it is in an open
> section.  this would be the last section of the cut.  so the software
> would have to be capable of not only reversing direction, but also to
> run the toolpath in reverse from any point in the program.

Right on. The tests that I've run here show that a simple position history
file should be sufficient to move the wire back along the cut path. I
wouldn't keep the history for much over a few thousandths. If we can't clear
a short in that distance we have bigger problems.

Again, it seems to me that if we wanted to back out of a long cut we would be
better off to write the code for that move after we have written the cut. An
example might be an arbitrarily shaped punch to be cut from a blank. Since
the outline of the punch is a closed path we would not need to back around
it. We just need to leave by the same path we entered or loosen the blank
and drag it out by hand.

> Probably not terribly difficult knowing it is strictly a 2 axis move,
> but it would need to tested.  

I believe that we can history as many axes as we have.

> AND zero comp would need to be dead on
> for the table.

You are right about needing to keep the location accurate. Servo or steppers
with encoder feedback would encourage known positioning. My impression is
that we will need to do quite a a bit of testing for diameter comp with the
reversing moves.

Ray Henry