On Mon, 14 Sep 1998, John Payson wrote:

> One thing I was wondering about but didn't really see mentioned: if one
> is PWM'ing a motor while trying to measure its speed (via back EMF) what
> is the preferred way of handling inductive kick- back from the motor?

By getting rid of the kick first, then reading. A differentiator or a
micro-controller help a lot here. The differentiator can be used to detect
the end of the 'rise' after the kickback to actuate a sample & hold
mechanism.

> I'm particularly curious about a model-railroading application where
> there is an unknown and variable amount of resistance between the supply
> unit and the motor.

In remote controlled model railroading with active engines (i.e. smart
engines) one NEVER regulates the far away supply, one regulates the motor
directly in the engine from constant rail voltage. The variable resistance
(which is actually a pretty bad contact in practice)  does not normally
come into play.

> If the PWM outputs a constant current when on, and if inductive kickback
> can be filtered out so as not to skew the measurements, it would seem
> that gunk on the rails would be a 99.44% non-issue with regard to
> maintaining constant speed provided only that the current supply had
> adequate compliance voltage.
>
> Is such a design practical/feasible?

Mostly, no. Model railroad rails (and air lines) are among the noisiest
electrical contacts I know of. Someone has apparently suggested once, to
use this noise source for cryptographical purposes... ;)

Peter

PS: The gunk on the rails should not be there anyway. There are small
gizmos that can be mounted on the engines to clean the rails as they run.
There are also expensive rail-cleaning engines that look like the real
thing. Most use copper brushes, some have them turn...