Re: [CAD_CAM_EDM_DRO] Bridgeport BOSS ballscrews
Posted by
Jon Elson
on 2000-07-19 13:27:12 UTC
Stephen Tate wrote:
or the
bearing that constrains whatever rotates. In the Bridgeport X axis, the
screw
is normally fixed, and the nut rotates, so check that the screw is
rigidly fixed,
and the bearings the nut rotates on are not loose. Check the drive belt
to make
sure the teeth are not worn so that it can slip. Check that the belt
sprockets
are locked tightly to the shafts.
You don't mention which BOSS machine you have. If it is a stepper
machine,
there is a limit to how small a motion you can get out of a stepper.
The later
BOSS machines had a step size of .001", so that is only 2 steps of lost
motion. Don't assume it is all truly mechanical backlash. Most
importantly,
don't try to increase preload until all traces of backlash are removed.
You
might destroy the screws and nuts this way.
On a servo-equipped BOSS (I think that is the BOSS-7 and up), there may
be some difference between commanded and actual position, due to
finite servo gain. If you can display actual position instead of
commanded
position, use the actual value, as that is the true position indicated
by the
encoder, rather than the position the CNC control is attempting to
reach.
You may be able to increase servo gain to better null out positioning
error,
if actual is not equal to commanded at rest.
Finally, there is some lost motion between the screw and the table due
to
table rocking, since the screw/nut pulls from several inches below the
actual
bearing surface of the table. This gets progressively worse as the top
of
the saddle gets rounded off, allowing the table to tilt more and more.
If your ways are significantly worn, this can account for a good deal of
the lost motion you report.
What method are you using to measure the backlash? Are you using a
dial test indicator and just making very small moves, and observing how
much motion you need to command to see the pointer move? This could
be very misleading with some motion systems. Servo systems may have
deadbands to prevent 'hunting', which will defeat the attempt to measure
very small incremental moves. Making larger moves to approach a
specific
coordinate, and then changing the direction you approach that coordinate
from gives a better indication of true backlash and other lost motion
errors. The small moves also allows stick-slip friction to dominate.
One other good test is a repeatability test. Approach the same
coordinate
numerous times, and observe the scatter in final position with a dial
test
indicator. Do this at different speeds. You will almost certainly see
the
scatter increase at very low speeds.
Jon
> I have a BOSS with .0018 -.0019 backlash in the X axis which I amFirst, make sure there is nothing loose, either the mounting of the nut
> trying to get rid of. This backlash remains constant throught the
> full range of travel so the screw seems to be in good shape. Do I
> need to reload the nuts with bigger balls (no comments from the
> peanut gallery, please)or is there a preload adjustment? If I need
> bigger steel balls where can I buy them?
or the
bearing that constrains whatever rotates. In the Bridgeport X axis, the
screw
is normally fixed, and the nut rotates, so check that the screw is
rigidly fixed,
and the bearings the nut rotates on are not loose. Check the drive belt
to make
sure the teeth are not worn so that it can slip. Check that the belt
sprockets
are locked tightly to the shafts.
You don't mention which BOSS machine you have. If it is a stepper
machine,
there is a limit to how small a motion you can get out of a stepper.
The later
BOSS machines had a step size of .001", so that is only 2 steps of lost
motion. Don't assume it is all truly mechanical backlash. Most
importantly,
don't try to increase preload until all traces of backlash are removed.
You
might destroy the screws and nuts this way.
On a servo-equipped BOSS (I think that is the BOSS-7 and up), there may
be some difference between commanded and actual position, due to
finite servo gain. If you can display actual position instead of
commanded
position, use the actual value, as that is the true position indicated
by the
encoder, rather than the position the CNC control is attempting to
reach.
You may be able to increase servo gain to better null out positioning
error,
if actual is not equal to commanded at rest.
Finally, there is some lost motion between the screw and the table due
to
table rocking, since the screw/nut pulls from several inches below the
actual
bearing surface of the table. This gets progressively worse as the top
of
the saddle gets rounded off, allowing the table to tilt more and more.
If your ways are significantly worn, this can account for a good deal of
the lost motion you report.
What method are you using to measure the backlash? Are you using a
dial test indicator and just making very small moves, and observing how
much motion you need to command to see the pointer move? This could
be very misleading with some motion systems. Servo systems may have
deadbands to prevent 'hunting', which will defeat the attempt to measure
very small incremental moves. Making larger moves to approach a
specific
coordinate, and then changing the direction you approach that coordinate
from gives a better indication of true backlash and other lost motion
errors. The small moves also allows stick-slip friction to dominate.
One other good test is a repeatability test. Approach the same
coordinate
numerous times, and observe the scatter in final position with a dial
test
indicator. Do this at different speeds. You will almost certainly see
the
scatter increase at very low speeds.
Jon