Re: Installing rotary encoders

> From: "Ted" <rtr@...>
>
> This procedure isn't difficult, but must be understood to successfully
> install a rotary encoder where none have been before.
>
>
> Then drill and tap for a couple of set screws at 90 degrees to allow you to
> indicate it in. Put the indicator near the end of the pin, because that's
> where the coupling will fit. The pin doesn't needto stick out more that an
> inch when you finish. You can cut it off and file the end when you have it
> in place, indicated, and clamped with the set screws. Do thiswork on the
> pin gently then indicate it again. Encoders don't like a lot of vibration
> so bring it within 3 thousandths or better. I'm always happier if it
> comes within one and a half thousandths.

Most encoder warranties are voided by NOT using an appropriate coupling.
I use couplings made from a single piece of metal, slotted helically to provide
radial and axial compliance, but no torsional compliance.

>
> You can connect it with an oldham coupler or similar low inertia couplers.
> Surprisingly, if you mount it so the shaft pin and the encoder pin come
> within a few thousandths beyond the slop in the system, you can use a piece
> of surgical tubing for the coupling. You can add an outer layer of heat
> shrink tubing if you think the surgical tubing is too compliant, but I've
> never had to. The extra stiffness will transmit more vibration to the
> encoder, a bad trade off. There is little friction in good encoder
> bearings so the surgical tubing wall thickness is adaquate for a coupling.
> Obviously, you can't do this if you need to drive the screw or brake it
> through the encoder shaft, a bad idea in any case.

Actually, many good encoders have a fair amount of drag in their
bearings. They use a pair of angular contact bearings with preload, to
stiffly constrain the shaft from any radial movement, which would
show up incorrectly as rotation. They also have at least one, usually
two shaft seals, to keep crud out of the bearings, and grease out
of the optics. That all adds up to inch-ounces of static drag.
I'm using 1000 line encoders, counting all transitions, so that is
4000 counts/rev, or more than a count for every tenth of a degree.
It would be foolish to waste that accuracy with a homemade coupling
that allowed twist to develop. Also, the metal coupling is good for
many years, what if the surgical rubber turned to gum? Do you know
what OIL does to LATEX? Yucck!

>
> Besides eccentricity in the rotation of the encoder coupling pin, the error
> that kills the most encoders is failure to allow enough room between the
> ends of these two shafts. All sorts of distortions occur when an axis is
> stopped after a rapid movement. Ten thousandths of longitudional slop is
> not unusual in a ball screw, more in an acme screw.

I don't know how a precision machine could possibly keep any accuracy
if there is .010" longitudinal movement of the screw! I have a total of
.001" of slack in my mill, and I'd like to find the sources, and reduce it.
But, that is the sum of ALL the sources of slack, like torsion of the
leadscrew, shaft couplings, bowing of the screw, slack in the angular
contact bearings, slop in the anti-backlash ballnut, etc. etc.

Jon