Re: Cutter Radius Compensation

[Jon Elson <jmelson-at-artsci.wustl.edu>]

Frank Hess wrote:

> Thanks Ray.  I _have_ read the section of "The NIST RS274/VGER
> Interpreter" that deals with this topic but I must be too dense to
> understand it.

Well, it is tricky.  The cutter radius compensation by itself is not all that
bad, but the sequence of ramping the compensation in at the beginning
of the pass, and ramping it back out at the end is what I always have
to make a few dry runs with.  It also took me forever to find out how the
'cutter left of part' or 'cutter right of part' is defined.  I finally learned
that
it is defined as if you were walking along the part, following the cutter as
it cut the part.  From that vantage point, the left/right is supposed to be
consistant.

Different controls handle the change from linear to circular interpolation
somewhat differently, especially with respect to gouge sensing.  Some
controls are more forgiving, or will bump the tool away from the surface
when a gouge would have ocurred with the defined path.  EMC is pretty
sensitive about gouges.  A trick I use when ramping the compensation
in (and out) is to preposition the tool so that it will approach the initial cut

at a slight angle, so that when the compensation is ramped in, the path
takes a slight convex turn, preventing EMC from complaining about a
gouge.  In other words, supposing I'm going to cut a rectangular block,
and the part of the toolpath where compensation will be active is defined
as the perimeter of the part's actual dimensions.  So, preposition the tool
an inch to the left of the part, say, such that the tool edge toward the
part would gouge is slightly if you made a +X move.  (In other words,
manually compensate just a little less than the tool radius.)  When the
cutter compensation is turned on from this approach, it needs to bring the
tool away from the work in the -Y direction, and that forces the path
to be convex, away from the part, thereby preventing EMC from sensing
a gouge.

Sorry about this convoluted description, it makes a lot more sense with
pictures.

Also, see my web page about this at :
 http://206.19.206.56/diacomp.htm

Jon