In article ,
Joseph Gwinn wrote:
In article ,
Ade V wrote: (on 9 September 2010)
did gone and wrote:
[snippage]
Consider a circle rotating about an axis displaced from the center of the
circle. (This is all in 2D, and the various axes are perpendicular to the
plane of the circle.) Using RDM's nomenclature, the radius of the circle
is R, and the distance between rotation axis and circle center is X. In other
words, X is the runout.
[snippage]
By RDM, we compute 0.5*[(R+X)+(R-X)]= 0.5*[2R]= R, which is the radius of
the circle, regardless of the runout X. If we measure the diameter D with a
micrometer and compute R-D/2 as suggested, what we get is a measure of the
departure from roundness of the circle. We do not get the runout, which
has already cancelled out.
Unless I'm mistaken, RDM is _trying_ to cancel the runout? The runout is
getting in the way of determining how accurately the spindle is aimed
relative to the bed. Whilst runout is its own issue, it's not what RDM
is trying to fix...
Although the article doesn't say that, it may be that RDM is ultimately trying
to measure bed twist, but the method as published cannot achieve that. The
published method measures rod diameter despite runout, and there is no way to
deduce bed twist from rod diameter.
What the article claims to be measuring is the deviation of the spindle rotation
axis from parallelism with the ways, in two dimensions, horizontal and vertical.
The published method cannot do this, but changing only the math from summing
the runout max and min to taking the difference allows this deviation to be
measured.
What was done with the information was to cleverly shim the headstock where
it rests on the bed, to achieve parallelism.
Actually, with the sum, we don't get the full radius, we get the change from
some unknown constant, because we never zero the dial indicator at the unknown
center of rotation, we set the dial indicator up at some convenient offset,
and go from there.
On reflection, I think Ade V has put his finger on the answer.
Rollie's Dad (RD) was measuring using a dial indicator on the lathe carriage
sensing the spinning rod, the intent being to see how well aligned the headstock
was, using a random bit of round rod held in a chuck with random crookedness.
So, RD wanted to cancel the runout and crookedness, yielding the diameter of the
rod plus some constant.
The easiest way to visualize the runout is to imagine the spindle axis tracing
out a cone in space. To align the headstock, the axis of that cone is made
parallel to the bedway. The raw measurement is a combination of runout, actual
rod diameter, and deviation of cone axis from parallel. Cancelling the runout
yields the local apparent rod radius, which is the combination of actual rod
diameter and cone axis deviation. If the rod is a perfect cylinder, with
constant radius everywhere, then a constant dial indicator reading as the
carriage moves implies that the cone axis is parallel to the bedway. If the rod
radius varies with location, one must measure the actual rod radius and subtract
it to get the distance to the cone axis.
Now, by contrast, I'm currently interested in the runout that RD ignores, and
want to ignore the rod radius that RD uses.
So, to summarize (in the context of a vertical mill):
One half the *sum* of the the indicator measurements (corrected for rod radius)
yields the deviation from perfect parallelism between Z-Axis ways and/or quill
motion, ignoring runout and chuck crookedness. This is RD's Method (RDM).
One half the *difference* of the the indicator measurements yields the runout,
ignoring crookedness and imprecise parallelism.
One set of measurements can be used to compute both non-parallelism and runout.
In practice, the only part of non-parallelism that is adjustable in most
vertical mills is tramming.
One can also deduce crookedness of the chuck by running the test with the same
bar rotated into different positions, so bar crookedness can be separated
mathematically from chuck crookedness.
A practical note: I've found that R8 spring collets don't hold the rod quite
rigidly enough, probably because of a very slight mismatch between actual rod
diameter and actual hole diameter of the collet, so the rod is clamped in a ring
versus over an extended area. If one tugs on the rod, it will permanently shift
by a few tenths, and won't usually return to zero if the rod is plucked and
allowed to vibrate down to zero. A Jacobs Superchuck is somewhat more secure in
that while it also moves, the bar returns to ~zero when it is plucked. I will
next try an Albrecht keyless chuck, which is likely far more precisely made than
the superchuck.
What should work far better is a R8 to ER arbor, as ER collets have a far better
grip on a rod than a R8 spring collet. And ER collets have many more uses than
a test bar.
We may also be seeing the R8 arbor shifting in the spindle, but given the taper
it should return to zero when the bar is plucked. A test bar would have the
same problem with shifting in the spindle.
Joe Gwinn