Re: [CAD_CAM_EDM_DRO] Re: case histories
Posted by
Jon Elson
on 2000-04-05 16:22:05 UTC
Eric Keller wrote:
that may be
due to other errors in the machine. High resolution shaft encoders are
cheap. a
1000 line/rev encoder gives 4000 counts/rev in quadrature. If the
leadscrew is 5 TPI,
that is 20,000 counts/inch, or .00005" / count. This gives VERY smooth
motion
for CNC.
.050"! After
cutting and clamping the nuts, I still could do no better than .015" or
so, due to
differential wear in the Acme screws. The backlash on my entire system
from encoder
to table motion runs about .0013" on the worst axis. This is a
combination of encoder
lag, screw torsion, screw bowing, flex in the coupling from nut to
machine, thrust bearing
deflection, etc. etc. all added together. I'm still trying to do
better, but I'm not too
unhappy with what I'm getting now.
leave
setups on the machine for several days, and repeat operations many
times. I
would be worried about cumulative slip between the wire and drum.
Jon
> Jon,Yes, with a helical-slit shaft coupler.
> All this talk about encoders has me wondering what really is
> required. Are
> all of your encoders direct coupled to the ball screws on your mill?
> IRight. I have high-precision screws, about .00015" error per foot, and
> always figured i was going to get linear encoders, now i'm not sure.
> If you
> get prescision ball screws, you are going to have to go to the high
> accuracy
> linear encoders to beat the shaft encoder
that may be
due to other errors in the machine. High resolution shaft encoders are
cheap. a
1000 line/rev encoder gives 4000 counts/rev in quadrature. If the
leadscrew is 5 TPI,
that is 20,000 counts/inch, or .00005" / count. This gives VERY smooth
motion
for CNC.
> If you think about it, people use the indicator dials on the handlesBacklash! The backlash on my Acme screws were in the range of .030 to
> of
> their mills to make parts all the time, so what is the difference
> really
> between slapping an encoder directly on the shaft. Granted it's
> always nice
> to have more accuracy, but you pay for it.
.050"! After
cutting and clamping the nuts, I still could do no better than .015" or
so, due to
differential wear in the Acme screws. The backlash on my entire system
from encoder
to table motion runs about .0013" on the worst axis. This is a
combination of encoder
lag, screw torsion, screw bowing, flex in the coupling from nut to
machine, thrust bearing
deflection, etc. etc. all added together. I'm still trying to do
better, but I'm not too
unhappy with what I'm getting now.
> I've seen the discussion of the wire driven rotary encoder used as aI have doubts about the wire tracking the same every time. I sometimes
> linear
> encoder, but i never got that type of mechanism to work too well for
> me.
leave
setups on the machine for several days, and repeat operations many
times. I
would be worried about cumulative slip between the wire and drum.
Jon