"Ned Simmons" wrote in message
...
On Wed, 25 Nov 2009 00:18:26 -0500, "Ed Huntress"
wrote:


"Ned Simmons" wrote in message
. ..
On Tue, 24 Nov 2009 20:38:12 -0500, "Ed Huntress"
wrote:


In good lathe designs, I believe you'll find that the typical setup is
to
have a pair of preloaded bearings facing each other at the spindle-nose
end.
In classic designs, these were angular tapered-roller bearings in larger
lathes, and angular-contact ball bearings in smaller ones. All of your
Z-axis location is accomplished with these head-end bearings. Then the
tail
end of the spindle was held in a single- or double-row bearing that
allowed
linear movement -- either straight rollers, or ball bearings that allow
the
spindle to move.

It doesn't take much heat to make the spindle grow substantially, which
will
either overload your bearings or unload the preload on the head-end
bearings. Also, this lathe is no wimp. It would handle much higher loads
than my SB-10L and would be roughly the same size, although a little
wider
and probably a little shorter (mine has the 54" bed).

Hardinge uses a single pair of angular contact bearings, one bearing
at either end of the spindle, in the HLVH headstock -- a 3000RPM
spindle with 25 millionths runout.

Hmm. Is that right? I thought Hardinges had the classic two-bearing-front,
floating rear setup. But I don't know for sure. Maybe Gunner would know.

It's as I've described. I've been in there and have a picture here in
front of me.

Ok, then maybe it's just the HLVH? Because I used a cutaway drawing of a
Hardinge spindle once to illustrate exactly what we've been talking about
here, and, although I don't remember the drawing perfectly, I do remember
what I wrote for the caption. Are you familiar with the earlier Hardinges?



In any case, they use Class 9 bearings in their top-of-the-line, and those
spindles run exceptionally cool. They aren't a good example to follow for
any ordinary lathe. You'd never be able to duplicate the performance of
that
spindle.

The key to controlling the temperature rise is more closely related to
accurately setting the preload than the bearing class. In order to get
reasonable stiffness you'd probably not want to mount the nose
bearings directly back to back, and the most practical way to get the
required separation is with a pair of spacers. The length of those
spacers is critical to setting the preload, regardless of how long
they are.

I've designed quite a few spindles, and I think temperature effects
are way down on the list of things to worry about when thinking about
buliding a ball bearing headstock on the cheap with limited equipment.

Aha. Are you saying you'd run opposed, preloaded bearings at opposite ends
of the spindle on a lathe like we've described here? Because I was once
pretty familiar with bearing layouts in machine tools, and it just goes
against everything I was taught. I'm not saying you're wrong, but there is a
pretty fair amount of work behind the things I thought I learned when I was
writing about it. If it's all wrong, I have some work to do here in a hurry.



The preload is controlled by a pair
of spacers that introduce an offset between the inner and outer races.
The drive belt is outboard of the rear bearing.

One problem with tapered roller bearings, depending on how fussy you
are, is that the runout specs on standard bearings is pretty bad
compared to run of the mill ball bearings. And precision grade roller
bearings are horribly expensive and can be difficult to source.

--
Ned Simmons


--
Ned Simmons