Old "Last Word" Universal Test Indicator (Starrett), CNC backlash

Hi to the lists,

I've been playing around with adding backlash to a controller program
I've been writing, and I THINK it's working! What's not working is my
antique "Last Word" indicator (possibly). I have it's 1/16" mount in a
collet in the spindle, and the ball on the end of the mount allows the
body of the indicator to still shift around a little. Nothing seems to
tighten up on the little ball swivel, even if I apply considerable
force. The "fingered cap nut" that holds against the ball doesn't seem
to hold (threads look fine). Or maybe it's not supposed to? I could
understand a little "safety" in allowing it to move if pushed too far.
Perhaps it's just too old (yeah, I could use a new one...).

I'm also curious how closely the .001" divisions on the indicator should
agree with movement? I THINK I've got the pressure square on, that is,
the indicator body is perpendicular to the body of the indicator. I
could understand if I was at an angle, but I shouldn't be. I don't have
any doc's on the indicator, as it was quite old when I bought it used.
I do have another "plunger" indicator that is also old and second
(third) hand, but it works sluggish, like the oil has dried up in it.
I've had it open, but everything looks clean. I suppose it could be in
the "spindle" part of it, but I don't think I'm going to take it
apart...

I'm not sure how "corrected" backlash should look. Oh yeah, I know what
backlash looks like, you go one direction for a while, then when you
reverse, it takes nn steps before the axis starts to move back. Then
the axis moves for each step that is issued to it. Fine. But what
happens (what is observable) with backlash correction? At the change in
direction, the controller "provides" the steps need to get the axis
moving again. Not sure what this looks like on the hardware. I suppose
the axis "delays" while the backlash is taken up, then moves the
required distance. If I turn backlash off, I "loose" distance compared
to when it is enabled (when moving back and forth between two points, my
test case). So I believe it is working. However I don't have a good
way to "know", because I don't read the full .010" move I commanded on
the indicator. Do I need a setup with two indicators, spaced a
specified distance apart? would that help?

This backlash now has me captivated, since I ran into the "casting
problem" a week or two ago.

Alan KM6VV

P.S. Oh yes, the new boxbed casting and some "sundries" (taps and dies,
and some steam pipe) came in from Stuart. £86+, more then the cost of
the 10V steam engine!