On Fri, 07 Oct 2005 13:21:44 -0400, Bob Engelhardt
wrote:

Eric R Snow wrote:
... where a timing element, such as a toothed
pulley, are not used, belt creep occurs. ...
This is probably where the "backlash" comes from that I'm
seeing. ...

Are you saying that you do NOT have toothed pulleys? How hard would it
be to cut teeth?

Bob
BTW - why isn't is "pullies"?
Bob,
I wasn't using toothed belt pulleys for two reasons. First, getting
the diameter exactly 1.0000 in circunference would be tough. And they
are not available in this size. I need this because the encoder, for
now at least, is direct drive and the encoder max pulse output is
10,000 pulses per rev. which is my target resolution. Second, even
though all my CNC machines use toothed belt drives the setup is
different. The belt drives a screw with a pitch of .200. This means
that an encoder mounted on the driven screw can have 1/5 the
resolution for the same 10,000 pulses per inch. This screw business
also means that any rotational error from the motor is reduced to
1/5. If the encoder is on the screw then drive belt backlash also
doesn't matter as much. Finally, the CNC control has a backlash
parameter which can be changed any time. So any backlash in the
positioning system is accounted for by the software whenever an axis
is reversed. The counting display I have has no such compensation. It
does have a scale multiplication parameter that can be used to
compensate for slight errors. If I wanted lower resolution then the
display could compensate for larger errors.
Eric