Re: [CAD_CAM_EDM_DRO] Re: Need G Code help

torsten98001 wrote:

>Tool radius compensation is a difficult function to implement
>I have not seen a Pc-Controller yet that was able to do this
>correct.
>

Don't listen to him! EMC does it right, as it works EXACTLY the same as my
old Allen-Bradley 7320 controller did it. besides compensating for some
other
differences in the G-code, my programs worked in identical ways. (It is
true
that the A-B would rarely give an error messages but would occasionally do
TRULY bizarre things like 10-turn spirals and head for the hills, where EMC
would give an error message, which is better.)

>Many claim they do but fall short usually sooner then later.
>

I am not aware of any test case that anyone has developed in the last 2
years or so that
shows EMC failing to follow the standard.

>Tip: You may be able to work around the problem using I and J
> convention instead of R values when interpolating circles.
> as in
> G1X8.9Y3.25
> G3X8.15Y2.5I0.0J-.75
> in the above example you can also omit the I entry entirely
> because the value is zero.
>
>

No, the I & J can give more troubles, due to limited precision in the
parameters entered.
At least with EMC. There is a way to relax the error limits on
start/end circle radius,
but it really gives no advantage in this case.

A negative radius, however, I think is just wrong. Your R radius is
clearly greater
than the tool radius, so it should be OK from that standpoint. You need
a "lead-in" move
immediately following the block with the G41 that is at least as long as
the tool radius,
for the cutter radius compensation to be interpolated in gradually over
the length of the
move. it appears you have a sufficiently long move (the G01) but it
probably has in
inside corner! I run into this all the time, and have to pick my
starting point (before the
G41) very carefully so this doesn't occur. You need to be sure that
with the largest
tool radius expected, the corner between the lead-in move and the next
move (your
G03, in this case) will NOT have an inside corner, but either be
perfectly straight or
an outside corner. This is easier to do in a CAD program than writing
the G-code by
hand, but you can still do it. Drawing it on graph paper may help.

Here's a writeup I did on this, when I finally figured out how to do it :
http://www.timgoldstein.com/cad_cam/diacomp.htm

It also took me a while to understand the concept of how one decides
whether it is "cutter
left" or "cutter right" of part, but think of it is as if you were
walking along behind the cutter as
it moved (relative to the part and machine).

Jon