At the meeting today I mentioned the fact that the half-nuts DO NOT ALWAYS have to stay engaged when cutting a metric thread on a lathe with an Imperial (i.e., inch) leadscrew. Eldon asked me to expand on this idea at a future meeting.
However, on a matter like this it is always better to have a written
description of the idea that can be perused at one's own pace and saved for
future reference. Hence this note.
Let's assume the following scenario. My lathe has an 8 tpi (threads per inch)
leadscrew and I've set up the proper change gears to allow me to turn a (very
common) 1 mm metric thread. I've made my first pass and now it's time to
return the carriage to the right to begin the second pass.
We all know that, if I were turning an Imperial thread, I would simply
disengage the half-nuts, move the carriage to the right, and then use the
threading dial to determine when to reengage the half-nuts. But I'm turning a
metric thread and this won't work. (Why it won't work is left as an exercise
for the diligent student.)
Conventional wisdom at this point says that I must leave the half-nuts engaged,
withdraw the tool from the thread and then run the lathe in reverse to move the
carriage to the right. While this approach is foolproof (as much as anything
is foolproof), there is a quicker method that works IN CERTAIN CASES.
Everyone will agree that if I disengage the half-nuts and move the carriage to
the right 1 mm (the pitch of the thread I'm cutting), the tool will reengage
the thread I'm cutting perfectly. However the half-nuts will NOT reengage at
that point because 1 mm = 0.03937" is nowhere near the 0.125" pitch of my
On the other hand, if I move the carriage one inch to the right, the half-nuts
will, of course, reengage perfectly. But 1" = 25.4 mm so the cutting tool will
not reengage the thread - in fact that 0.4 mm difference means it will almost
perfectly split the thread that I've already cut - a disaster.
But suppose, I move the carriage to the right a distance which is BOTH an
integral multiple of 1 mm AND an integral multiple of the pitch of the
leadscrew. Obviously, the tool will reengage the thread I'm cutting AND the
half-nuts will reengage the leadscrew.
You can fool around with the math at this point (it's mind-numbingly simple)
and you'll quickly discover that:
5 inches = 127 mm
So, if I move the carriage precisely five inches to the right, I can reengage
the half-nuts and be assured that I'm still synchronized with the thread I'm
cutting. Integral multiples of this solution will also work. If my metric
thread is very long, the fact that:
10 inches = 254 mm
may be the way to go. Once we've got the smallest solution (5 inches in this
case), any integer multiple of that solution is also usable.
So what we do is, knowing ahead of time that we need to move 5 inches, before
starting to cut our metric thread we obtain or make a rod precisely 5 inches
long. When we want to move our carriage, we bring our carriage stop up against
its left side, disengage the half-nuts, move the carriage to the right, insert
the rod between the stop and the carriage, snug the carriage against the rod
and reengage the half-nuts. We're ready to make our second pass on the thread.
[Often one can find micrometer standards cheaply at tool liquidators. They
are, of course, very accurate in length and the longer ones are often cheap
because there's not much market for them. Buy them when you see them. They're
perfect for this application - either used singly or in combination.]
Now, finding the rod length for a given combination of metric thread and
Imperial leadscrew may be an overwhelming math chore for some people. Go to my
website (see signature below) and download the STICK program. It will
calculate the stick length needed.
[As an aside, this program will handle all the possible combinations of
metric/Imperial threads being cut on lathes with metric/Imperial leadscrews.
Read the text file included with the program for more information. Said file
is appended below for your convenience.]
I said this was a quicker method IN CERTAIN CASES. Sometimes the length of the
rod required is ridiculously long. As an example, consider cutting an 8 mm
metric thread on a lathe with an 8 tpi leadscrew. The STICK program tells us
we need a rod:
Move carriage 40.000 in = 1016.000 mm to right to recapture thread
Few of us have a lathe with a forty inch bed, much less a way of accurately
calibrating a stick to that length. Fortunately, this problem occurs mostly
with the very coarse metric pitches. The more common, finer pitches generally
yield manageable stick lengths.
Home Shop Freeware - Tools for People Who Build Things
In the 6/01 issue of "Machinist's Workshop", Peter F. Lott described an
interesting technique he uses when chasing metric threads on his Imperial
We all know that the thread dial on the lathe can't be used for this operation.
Conventional wisdom says that, to maintain registration, we need to leave the
half-nuts engaged and run the lathe in reverse to move the carriage to the
right prior to making the next cut on the lathe. Especially for long threads,
this can become quite tedious.
Peter quite correctly points out that, if we disengage the half-nuts and move
the carriage a distance to the right that is an integer multiple of the thread
pitch *and* is also an integral multiple of the leadscrew pitch, we can
reengage the half-nuts without losing registration on the thread we're cutting.
He goes on to supply a table of distances to move for various combinations of
metric threads and Imperial leadscrews.
I wanted to explore techniques for computing the information in the tables so I
wrote STICK. The name refers to the technique described in the article of
cutting a stick to the appropriate length and using it to accurately reposition
the carriage at the end of a pass.
While writing the program, I realized that Pete's process can be easily
extended to cutting Imperial threads on an Imperial leadscrew lathe and, for
the benefit of our metric friends, cutting Imperial threads on a metric
leadscrew lathe. While I was at it, I covered the last possibility, metric
threads on a metric leadscrew lathe. Metric on metric and Imperial on Imperial
aren't really needed but obtaining them was trivial after the metric on
Imperial code was written so what the heck - maybe someone would like to have
Update 5 December 2002
Jeff Sauer sent me a note describing an alternate approach where the threading
dial is calibrated and used as a measuring tool in lieu of the calibrated
stick. I've included his idea here for those of you who might like to try it.
I've found your software utilites very very useful.... Thanks
for coding these gems. I'm most grateful for CHANGE.EXE,
and lately have been trying STICK.EXE, although sometimes
the stick movement is more bothersome than reversing direction.
One possible improvement to STICK would be to translate the
required carrage motion into rotations of the threading dial.
For example, on my little change gear lathe with a 16 TPI
leadscrew, the threading dial has a 32 tooth gear that meshes
against the leadscrew (thus marking two inches of leadscrew
travel for each revolution of the threading dial). The threading
dial has four scribe marks on it so the feed nut can be engaged
at 2 inch, 1 inch, or 1/2 inch intervals. But if a paper "dial" is
attached on top of the threading dial (this new dial having as many
index lines as gear teeth below) then any integer leadscrew groove
could (with care) be engaged.
Perhaps STICK could print the paper dial, with as many index marks
as required for the selected cutting job, and print a table showing
a complete sequence of which index marks are used for each
For example... Suppose STICK says my modulo interval is 60 teeth
on my 16 pitch leadscrew. That's two complete revolutions of the
dial minus 1/8 turn (still using my 32 tooth thread dial gear). In this
example, a paper dial with 8 index lines labeled backwards from
0 to 7 would simplify the carrage movement. Each time I reposition
the carrage, I'd go a complete revolution, and then stop on the next
higher number on the dial.
Let me know what you think..... thanks.... Jeff Sauer