In article ,
"DoN. Nichols" wrote:

On 2008-05-06, Joseph Gwinn wrote:
In article ,
"DoN. Nichols" wrote:

On 2008-05-05, Joseph Gwinn wrote:

[ ... ]

Yep. I wanted a lathe with integral cabinet stand, on the theory that
it would all fit together better.

And you got that. (And got rid of the chatter problem with the
new gib. Congratulations.

Thanks. I've been reflecting on how I came to the conclusion that the
5914 needed new gibs.

The key was the gross tilting of the toolpost, which moved something
like a tenth or even an eighth of an inch, and yet nothing broke (except
for a tool bit now and then).

There is a reason people don't make springs out of cast iron, so the
castings could not be bending enough to explain such large deflections.
So it had to be the joints and/or slideways. The only joints were
firmly bolted (toolpost to slide) or wedged (tool holder to toolpost),
and no relative motion could be felt. And the headstock was firmly
bolted to the bed.

This left the slideways, of which there are three.

The first is between the bed ways and the carriage. This originally had
~0.003" wear-induced clearance, reduced to ~0.001" by removal of one
0.002" brass shim from each of the two rear hold-down plates. The
carriage is about 12" wide along the bed, so this 0.001" could account
for only about 0.001" of the tilt (measured at the end of the 5"
overhang). Only ~0.099" to go.

The second is the cross-slide dovetail (~2" wide), and the third is the
compound dovetail (~1.625" wide). The observed wear (~0.004" in both)
would easily explain the observed deflection, given the leverage from
the overhang plus the 2:1 amplification in each of the 60-degree
dovetails.

And -- your compound has less adjustment range for the gib
because the dovetail itself is shorter, so the compound would show the
problems sooner, given equal wear.

It was the narrower dovetail width that matters, more so than the
length. On a narrower dovetail the same amount of wear in thousandths
allows for greater angular motion.


And it didn't feel right when I tightened the gibs. The effect on
slideway drag was very gradual as I tightened the gibs; one would expect
the effect on drag to be far more abrupt. This pointed to misshapen
gibs springing under pressure, versus being in pure compression.

Yep! Good diagnostics.

Ya gotta have a feel for feel.

I did another experiment today, this time cutting 1018 steel with the
grooving tool with the flat nose 0.206" wide held upside down in a
toolholder and with the lathe in reverse, and the backgear engaged. I
can cut a wide groove without drama and in almost total silence (aside
for the noise of the gears), but the tool and toolpost rises visibly
under the strain.

Now this is a very severe test, yielding chips about -0.008" thick by
0.200" wide, and involves very large forces, so the toolbit rising is
not a surprise. One has to push pretty hard before the tool starts to
bite with an audible bump, and it then yields the 0.008" thick chip. If
one maintains the pressure, one can make a very long and curly chip.


[ ... ]

If you are not on Windows, why would this be a problem?

Because I don't depend on them *always* focusing on Windows
systems, just because they are the easiest target. There have been
experimental attacks which used javascript, or java, and I see lots of
information about security holes in the flash plugins (information from
the CERT mailing list). Anything attacking something other than the
basic native machine language can be made to work on other systems.

It's true that everything has vulnerabilities, but as a matter of actual
(versus theoretical) risk, once you leave Windows things get pretty
quiet, and the expense (dollar cost and time cost and lost opportunity
cost) of security soon outweighs the cost of cleaning up the occasional
problem.

I read in one of the Mac magazines that only 10% or 15% of Mac owners
use any kind of add-on anti-virus product, and the unprotected Mac
owners are none the worse for it, and have been for years.

Unix/Linux users have even less to worry about.

Yes -- though there is a tradeoff to being a user of the most
numerous unix version -- which I think is now Solaris 10. That makes it
the next likeliest target.

But it has not happened yet. And while Solaris isn't completely
hack-proof, it's orders of magnitude less vulnerable than Windows.


By contrast, an unprotected Windows machine on the web will last a few
days at most, and a dialup machine might manage months before getting
caught.

An unpatched Windows box, from first connection to the net in a
college dorm, has an expected mean-time-to-infection significantly less
than the time needed to download the patches via the net -- even with a
fast pipe. This is why at least some colleges (those who care about
security) will not connect a new box to the net until one of their staff
have come around with a DVD or CD containing all the patches and applied
them. This at least slows down the infection rate somewhat. :-)

A college dorm has to be some kind of worst case test.

One very large problem in Windows is that one does not have to provide a
password to install most software, while in MacOS and Unix, one does.
This sharply limits the growth rate of viruses as at ever step of the
way, a human is required. This prevention of automation is sufficient
to sharply reduce the prevalence of viruses in non-Windows platforms.

Now, Microsoft is slowly changing Windows so people don't have to run
their software from an administrator account, but this is like turning
the Queen Mary. But it five or ten years it will have been done.


[ ... ]

I have the gibs adjusted so there is much drag, but the cross-slide
backlash is about 0.020", and the compound slide backlash is about
0.006".

O.K.

I figured out how better to adjust things, and got this down to ~0.003".

Good!

I should say how I did this. It turned out to be simple. There is a
black steel collar upon which the dial rotates, the collar being fixed
to the screw shaft with a setscrew. Loosen setscrew. Tighten the cone
nut (or nylock nut) until the screw shaft binds, then back off just
enough that the screw shaft turns freely. Tighten setscrew.

Oh -- the thrust bearing was loose. On the cross-slide of mine,
the handcrank screws onto the leadscrew and then is locked by a cone
nut which goes into a countersink in the crank (IIRC). It is easy to
get this too tight. But it is also easy to tell that this is the source
of the play by feel. You feel a bit of a "thud" as the bearing play is
taken up, while the backlash in the actual leadscrew/net interface is
less sudden. :-)

The hand cranks are woodruff keyed to the screw shafts on the 5914, and
are kept on the screw shaft with a thread and cone nut as well.


Before doing this, it helps to disassemble, clean, and lubricate the
entire dial assembly, so things can turn freely.

Of course.

Mine turned OK albeit with drag, and I worried that people might not
realize how much better it would work after a cleaning, in particular if
one wants to move the dial to a new zero without disturbing things.


[ ... ]

Someday I'll have the room for a horizontal mill. I keep running into
jobs for which it would be perfect. But it sounds like a perfect
application for copious flood cooling, if only to control the swarf.

The Nichols horizontal mill takes a lot less space than many,
and (usually) offers a choice between leadscrew or lever feed through a
rack-and-pinion drive for the X-axis feed. And -- there are a pair of
micrometer stops which can be set up to restrict the X-axis travel
rather precisely. Nice for certain forms of production work.

I've seen pictures, but never met one in person. What do they weigh?


Joe Gwinn