Back to the rack - nuts to the nuts

Well, back to square one. This thread pointed out the need to test one's
assumptions before proceeding to the next step.

I have been driving forward on the errant supposition that ballscrews are
"nautrally" just more accurate than rack and pinion. Here are my
observations.

Thermal growth was never overlooked - I simply failed to crunch the numbers
to determine its severity. 14 thousandths over ten feet for only 10C delta
is more than I had guessed.

The problem doesn't end there either. Gantry movement is rarely distributed
evenly over the total travel during an typical run. Thermal growth will be
greatest in the region of greatest activity. Thus it is possible to have
localized portions of the ballscrew that are thermally stretched, leading to
nonlinearity that can't be mapped for correction.

Thermal growth is practically a non-issue with rack and pinion. The rack is
bolted to a large thermal sink. I am still concerned about nonlinearities
manufactured into low cost rack as well as cyclic errors associated with the
pinion and its drivetrain.

We are looking for .005 inch total accuracy with cyclic errors less than
.003 inch. This may not be attainable with cheap rack and Browning doesn't
even want to discuss the subject as their product is not intended for this
purpose. I haven't priced precision ground rack and pinion yet.

Doug