DRO thoughts
Posted by
Heuver, Brad (B.R.)
on 2000-04-13 07:25:35 UTC
I thought I'd offer up another opinion on this homebrew DRO issue.
First, I've built the Tom K./ Dan M. Dro board. I've got the US digital 360
lines/inch linear strips, and the Goldmine motors with the 1000 hole encoder
wheels. That said, I'm in the process of implementing the following:
The linear encoders will find a home on the lathe. For this, I
wanted the ACCURACY of the linear strips, and will happily live with the
RESOLUTION limited to 0.0007". I doubt that much of my work will require
sizes smaller than this, and if they do, I'll probably polish/file/grind the
last smidgeon and mic often to get there. The possibility of losing
accuracy with another style system is a bigger drawback than the lack of
fine resolution.
The encoders will live on the ends of the ballscrews that I salvaged
from the Cintimatic. These are currently being modified to be shoehorned
into my 1961 B'port. Here the resolution will be 1000holes>4000 counts/0.2"
lead = 0.00005". Accuracy will be limited to the quality of the Cinci's
ground ballscrews. (I don't really know what that is, but I'm sure it will
be adequate for my needs.
Despite all the wire/cable/rod ideas presented here, I feel that the
desire is to increase RESOLUTION, at the very real possibility of a lack of
ACCURACY. The linear strips purport to get around that issue, but do give
up RESOLUTION.
For the average HSM, I feel that accuracy is more important.
One potential solution to this is to add a ballscrew, as I'm doing
on the B'port. If the costs of a large screw are too great to justify, how
about adding a small screw, 1/4" - 3/8", and fixing it to the table at both
ends. Drive the nut from the table, allowing it to spin freely, with a
rotary encoder attached. Thrust washers on each side of the nut will be
needed, and will probably need spring loading to eliminate lash. A larger
helix than the typical 0.2" may be needed to effectively backdrive the nut,
but even up at 0.4"/rev you would have 0.0001" resolution. This type of
system could just as easily use a NOOK/KERK style anti backlash screw to
lower the cost. If we wanted to get into lead screw error correction, to
improve the accuracy, the fixed screw would allow computer compensation to
be incorporated (in theory) rather easily. (The programing is something I
know nothing about) If desired, bellows could be placed on the screw, or
just a better wiper system than the nut comes with added for chip
protection, but this would probably not be needed.
Brad Heuver
bheuver@...
First, I've built the Tom K./ Dan M. Dro board. I've got the US digital 360
lines/inch linear strips, and the Goldmine motors with the 1000 hole encoder
wheels. That said, I'm in the process of implementing the following:
The linear encoders will find a home on the lathe. For this, I
wanted the ACCURACY of the linear strips, and will happily live with the
RESOLUTION limited to 0.0007". I doubt that much of my work will require
sizes smaller than this, and if they do, I'll probably polish/file/grind the
last smidgeon and mic often to get there. The possibility of losing
accuracy with another style system is a bigger drawback than the lack of
fine resolution.
The encoders will live on the ends of the ballscrews that I salvaged
from the Cintimatic. These are currently being modified to be shoehorned
into my 1961 B'port. Here the resolution will be 1000holes>4000 counts/0.2"
lead = 0.00005". Accuracy will be limited to the quality of the Cinci's
ground ballscrews. (I don't really know what that is, but I'm sure it will
be adequate for my needs.
Despite all the wire/cable/rod ideas presented here, I feel that the
desire is to increase RESOLUTION, at the very real possibility of a lack of
ACCURACY. The linear strips purport to get around that issue, but do give
up RESOLUTION.
For the average HSM, I feel that accuracy is more important.
One potential solution to this is to add a ballscrew, as I'm doing
on the B'port. If the costs of a large screw are too great to justify, how
about adding a small screw, 1/4" - 3/8", and fixing it to the table at both
ends. Drive the nut from the table, allowing it to spin freely, with a
rotary encoder attached. Thrust washers on each side of the nut will be
needed, and will probably need spring loading to eliminate lash. A larger
helix than the typical 0.2" may be needed to effectively backdrive the nut,
but even up at 0.4"/rev you would have 0.0001" resolution. This type of
system could just as easily use a NOOK/KERK style anti backlash screw to
lower the cost. If we wanted to get into lead screw error correction, to
improve the accuracy, the fixed screw would allow computer compensation to
be incorporated (in theory) rather easily. (The programing is something I
know nothing about) If desired, bellows could be placed on the screw, or
just a better wiper system than the nut comes with added for chip
protection, but this would probably not be needed.
Brad Heuver
bheuver@...