On Nov 25, 3:47*pm, Ned Simmons wrote:
On Wed, 25 Nov 2009 04:18:14 -0800 (PST), Jim Wilkins
...
Consequently, my understanding is that if you monitor runout with
enough resolution you'll see a component of the runout that's not in
synch with the RPM of the race. As a practical matter, for a
home-built spindle, I wouldn't worry about the effect with normal deep
row ball bearings. In the case of tapered rollers I don't have a good
SWAG one way or the other without doing more research.

Ned Simmons

My tentative plan for best accuracy is to cut center points on the
head and tail spindles and run them live until the work is almost to
size, then lock them down and make the final parallelism, size and
surface finish cuts between dead centers, driving the work directly
with a belt or rubber idler. I expect that the unhardened centers
might have to be refinished afterwards.

This applies more to the cylindrical grinding fixture than a homebrew
lathe, since I found an old South Bend headstock with the same spindle
thread as my lathe's to use for a temporary oversized wheel lathe.

jsw