Re: Just How Accurate Is CNC?

Chuck, your math is off a bit. A 20MHz cycle is 0.05us = 50ns =
50000ps = 50000000fs. A standard PIC requires 4 cycles mimimum for
one instruction. That's 200000000fs (if I counted the zeros
correctly).

~petev

--- In CAD_CAM_EDM_DRO@yahoogroups.com, Chuck Rice <Chuck@W...> wrote:
> At 9:49 PM +0000 11/9/04, Major A wrote:
> > > I can well understand the cost issue, but motion measurement
has
> >> changed in the last few years. Much has been done with
inexpensive
> >> lasers. I do not know a whole lot about it, but it seems that
if I
> >> can buy a optical mouse for $5, there might be a more precise
version
> >> of a laser measuring device that could be mounted to the bottom
of a
> >> table to precisely read table movement. The optical mouse is
not that
> >
> >The motion tracking of an optical mouse is nowhere near linear, and
> >certainly not reproducible. You'll need a high-quality high-
resolution
> >reference which spans the entire length of the linear actuator, and
> >new detection software as well. It gets very tricky to say the
least,
> >especially if you're looking for resolutions of 0.001" or better.
> >
> >> surface could be much more consistent. Or you might use a laser
in
> >> the saddle reflecting off the end of the table and measure the
> >> trip-time per pulse.
> >
> >Not really. To resolve 0.001" of saddle movement, you'd need a time
> >resolution of about 150fs (1fs = 0.000001 ns). Good luck...
>
> So 150 fs = 0.000150 ns
>
> A crystal 20 MHz PIC can run at 5,000,000 instructions per second.
> Doesn't that give a headway of 75 instructions per required
resolution?
> Perhaps my math is off, but if not, with a PIC chip, you can do a
> lot in 75 instructions. If you wanted to drop it a notch tighter,
> 15 fs would give you 7.5 instructions per second. That might be
> pushing it.
>
> I doubt that I have enough knowledge to do it, but I think that
> someone will be doing it soon. Too many of the pieces are becoming
> available at low price points.
>
> Seems, even if you wanted to laser engrave a scale on the bottom of
a
> table, you could read it much the way we read a micrometer. You
> could have a set of LEDs that read the course scale and another set
> matched the fine scale. The two readings together would give you
> a pretty fine resolution.
>
> >You can use interferometry, but that gets very expensive because
you
> >not only need a very good laser for it (coherence length much
bigger
> >than saddle travel) but a complementary method for course
measurement,
> >unless you always want to move your saddle at about 0.1 IPM or
> >slower.
>
> Interferometry is outside my my knowledge, but it should be easy to
> reduce resolution. You can do a lot with PIC processors now and if
> you step up to something like a ColdFire processor you would have
> a lot of time to do measurement calculations. -Chuck-