Re: [CAD_CAM_EDM_DRO] Yet more DRO

"D.F.S." wrote:

> > > I'll settle for the 10 microns, back when I was a kid We used a
> > > yardstick...
> > >
> >
> > For a DRO, this should be FINE. For a CNC, however, more resolution
>
> > might be nice.
> > I have resolution of .00005" (~ .00127mm, or 1.27 uM) on my CNC X
> and Y
> > axes right now, and it
> > is just about enough to get smooth circles interpolated. I
> DEFINITELY
> > would not want
> > to go any lower than that.
> >
> > Jon
>
> That was a joke, or at least 1/2 joke 1/2 irony.
>
> It just find it funny we went from 360 line/inch plastic scales all
> the
> way up to 10 microns and THAT was still not good enough for some
> people.
> You want nothing less than 50 millionths someone else thinks 1/1000 of
>
> that is good enough.
>
> Please don't take that wrong.
> You have a perfectly legitimate reason it's not.
> I just find the standards of what is acceptable here funny.

There are SO many sources of errors in machine tools, you need to stomp
on ALL
of them if you want good results. A good linear scale would be a great
help, it puts
about 5 potential error sources inside the feedback loop, most having to
do with
leadscrew problems. Also, for CNC to work properly, you need to have an
encoder
with a lot more resolution than your basic unit of measurement. If the
basic unit is
.0001" (~.0025mm) then you need a lot finer encoder resolution to
smoothly approach
and stop on one of these points.

I have a total system accuracy of about .002" on my Bridgeport right
now. Almost
all of that is backlash of some form, but not necessarily the most
obvious kind of
leadscrew/ballnut slack. I think a good part of it is bowing of the
leadscrew, and
flexing of a few of the lighter parts of the axis drive and ballnut
mount. I need to
get in there and do some work, but it makes nice parts as it is!

Jon