Questions about positioning accuracy
Posted by
Michael Moore
on 2004-04-24 16:42:08 UTC
If I'm trying to do something fairly precise on my lathe or mill I'll set
up a dial indicator and use that to show movement instead of
trusting the dials on the handles (I don't have DRO on either
machine).
The stuff I've been reading about CNC makes it look like rotation of
the stepper or servo motors is used to determine movement of the
slide or table. Good rolled ball lead screws claim .003"/foot
accumulated error (I presume that is a maximum amount and they
probably run less than that) and ground ball lead screws are a bit
under that.
It seems like to be really accurate you'd need to have a high quality
DRO to cross check for motor/lead screw accumulated errors, but
that doesn't seem to be a feature of most of the photos of
machines that I've seen.
High-end CNC software appears to have provisions for machining
some known pattern over a fairly large area so that actual
measurements can be taken and fed back into the software,
whereupon the software will apply correction factors until things
come out spot on (I can do the same thing with the printer output
from my CAD program, printing a test item, measure the print, and
then tell the software what it actually came out as).
Is this accumulation of error a significant problem, or are the
components available for a DIY conversion sufficiently good that the
actual errors in the parts are well under the allowable max error
quoted by the manufacturers?
When people get their conversions up and running do you do test
pieces that you'll measure and then try and come up with some
sort of homebrew error table, so you know that if you want a 12"
long part you need to tell the machine to move and cut at 12.003"?
BTW, I got my copy of Smid's CNC Programming book yesterday
and it has been interesting reading. Going through the Visual Mill
"getting started" PDF file I downloaded from MecSoft's website was
also very helpful in giving me a better idea of how this CNC stuff
works.
thanks,
Michael
up a dial indicator and use that to show movement instead of
trusting the dials on the handles (I don't have DRO on either
machine).
The stuff I've been reading about CNC makes it look like rotation of
the stepper or servo motors is used to determine movement of the
slide or table. Good rolled ball lead screws claim .003"/foot
accumulated error (I presume that is a maximum amount and they
probably run less than that) and ground ball lead screws are a bit
under that.
It seems like to be really accurate you'd need to have a high quality
DRO to cross check for motor/lead screw accumulated errors, but
that doesn't seem to be a feature of most of the photos of
machines that I've seen.
High-end CNC software appears to have provisions for machining
some known pattern over a fairly large area so that actual
measurements can be taken and fed back into the software,
whereupon the software will apply correction factors until things
come out spot on (I can do the same thing with the printer output
from my CAD program, printing a test item, measure the print, and
then tell the software what it actually came out as).
Is this accumulation of error a significant problem, or are the
components available for a DIY conversion sufficiently good that the
actual errors in the parts are well under the allowable max error
quoted by the manufacturers?
When people get their conversions up and running do you do test
pieces that you'll measure and then try and come up with some
sort of homebrew error table, so you know that if you want a 12"
long part you need to tell the machine to move and cut at 12.003"?
BTW, I got my copy of Smid's CNC Programming book yesterday
and it has been interesting reading. Going through the Visual Mill
"getting started" PDF file I downloaded from MecSoft's website was
also very helpful in giving me a better idea of how this CNC stuff
works.
thanks,
Michael