In article ,
Karl Townsend wrote:

...
In spite of anything you may read, any diameter can be successfully knurled,
assuming you're not using a straight knurl. ANY DIAMETER!

Secret to success is all in how you start the knurl. I've written an
article on that subject in the distant past. Pick up on the technique and
you can start a successful knurl virtually 100% of the time.

While it's true that a knurl is, in a sense, a gear of sorts, it is also
true that it can wander side to side, which compensates for pitch anomalies.

Harold

I hate to hihack iggy's thread. Well,not really.

Knurling is one lathe operation I still F$%^ up and don't know what
was wrong. I did go to a scissor type tool so the machine doesn't
provide the force and its balanced on forces. But sometime it still
comes out double tracked and I don't know why.

The math is simple, and does parallel that for gears.

Let's start with a straight knurl (where the ridges are parallel to the
knurl wheel rotation axis).

If we have a knurl wheel in a holder with wheel axis parallel to the
rotation axis of the lathe spindle and bring the knurl wheel into
contact with the spinning workpiece, the knurl will impress a series of
parallel lines into the workpiece.

After a bit more than one rotation of the workpiece, the parallel knurl
lines will be all the way around the workpiece, plus a little overlap.
Ideally, the later knurl lines will land exactly on top of the early
knurl lines. If this is achieved, then as the knurl deepens, the knurl
wheel will by gear action be kept in exact step with the workpiece, and
there will be no double knurling.

For this to happen, the circumference of the work piece must be an exact
multiple of the spacing between knurl lines, and thus of the spacing
between ridges on the knurl wheel.

Said in math, (Pi*Dwork)/m = (Pi*Dknurl)/n

Simplifying the math and cancelling the factor of Pi, allowed workpiece
diameters are Dwork=(m/n)*Dknurl, where m is the integer number of
ridges to be pressed into the workpiece and n is the integer number of
ridges on the knurl wheel, and Dknurl is the outside diameter of the
knurl wheel.

The same formula works with angled-line knurl wheels where the axis of
rotation is parallel to that of the workpiece, so long as one always
uses the number of ridges in the circumference of the knurl wheel; this
does not change with line angle.

If the axis of rotation of the knurl wheel is not parallel to that of
the workpiece rotation axis, the optimum workpiece diameter will change
from that computed above; this allows one to adjust to a non-optimum
workpiece diameter, so long as the workpiece is securely clamped so it
won't creep under the resulting axial force. Lubrication is likely
necessary as well.

Joe Gwinn