Re: emc-fpga.txt

[Ray <rehenry-at-up.net>]

Keith and Max

I am way out of my knowledge here so take my thoughts as tentative.

On Sat, 19 May 2001, Max wrote:
<s>
> Direct numeric interpolation of splines:
> I haven't seen anything about splines while I was looking through the rs274ngc
> interpreter source, so I guess splines are not supported by the interpreter now.
> In theory, limited by my somewhat restricted knowledge of EMC, adding support
> for splines would involve modifying the interpreter, adding a canonical function
> for splines to the posmath library and as well taking care of the usual dozens
> of places you did not even know before they existed ;-) Perhaps it is
> possible to have the splines split up in approximations of small tangential
> connected radii. But I don't know how the motion blending will deal with
> this. Does anyone have any experience with this ? 

In emc/src/emcmot there are two files cubic.c and cubic.h that get compiled
into cubic.o.  Cubic.o is linked into all of the motion modules and appears
to be an essential math process.  I seem to remember some discussion long
ago about fourth and fifth order math also but I don't see any of in the
code.

I think that Bert Eding <b.eding-at-chello.nl> in the Netherlands uses the
output of the cubic spline module to communicate between the EMC code that
he has compiled for Windows NT and the Nyquist motion boards that he runs
on his EMCO mill.  He could help you much more than I with the details here.

> The TkEmc GUI should be no big deal as it is only giving the NC code to the 
> rs274ngc interpreter and reading the interpreters status. Again correct me if I 
> am wrong.

This is exactly right.  In fact the GUI only passes the name of the run
file to the interpreter.  Beyond this initial message, all of what you see
of the running program is based on return signals overlaid on a copy of the
program file.

Ray