On 1 Feb 2015 03:31:17 GMT, "DoN. Nichols"
wrote:

On 2015-01-31, wrote:
On Sat, 31 Jan 2015 11:27:52 +0000, Baron
wrote:

prodded the keyboard

[ ... ]

To put a decent decoder on a wiper motor takes some hacking. Do you
put the encoder on the motor shaft, or on the output gear? On the
motor shaft gives much higher resolution, but requires a lot more
hacking (physically) to get to the motor shaft.

No just take the output from the DRO scale.

I'd rather just punch in a number - from this point go 100,231 counts,
knowing 10 counts is 1 thou, or whatever, and the control does the
rest. And you can program the feed speed by adding a delay loop
between steps on the stepper. Just using a PWM the motor slows dwn
under load so you could get a variable speed feed, which could affect
the look of the finish.

Depends. If you make it a *real* servo, there should be some
form of speed feedback, so the PWM is automatically adjusted to maintain
the desired speed.

A *real* servo motor has a tach generator on the shaft as well,
so this is easy to do.

If you've just got an encoder, you need to integrate the pulse
rate from the encoder to derive a speed value, and tuning it is a bit
trickier.

Standard modern found that out, in spades, on their NC lathes. They
attempted to use integrators to control the "servos" for the X and Y
axis, and repeatability was terrible.
After Standard Modern gave up on the lathe I spent hours on it
getting it useable (sort of, anyway)- we changed grounds and
sheilding, and put the PC on a dual conversion UPS (separately derived
power source) and got the failure rate down from over 30% to less than
3% on production runs of 600 parts.
It was the most useless damned piece of equipment I ever had the
msfortune of running across, My buddy Steve lost his shirt on it -
this was back at the time that Standard Modern failed and before
LeBlond took it over.
Enjoy,
DoN.