Re: [CAD_CAM_EDM_DRO] how much motor cogging is OK?

Elliot Burke wrote:

>To use a motor for a servo, how much cogging is OK? My motor definitely has
>some cogging, it can be felt while rotating the shaft by hand. It hasn't
>been driven slow enough to see if this affects its rotation.
>
>As a figure of merit, is something like the cogging torque/moment of
>intertia reasonable?
>
>thanks for any help here, I'll be sure to share what is learned.
>
>
>

Cogging torque of the disconnected motor is not much of a problem, all
by itself.
The larger problems are not as easy to detect, but are related to the
same motor
design features that cause the palpable cogging. These are velocity
ripple and torque
ripple. Velocity ripple is the change in motor speed, when rotating
very slowly at
constant low voltage. A single D cell battery may be good for testing
this. Torque ripple
is the variation of torque at different shaft angles while the motor is
fed a constant
current. These two fluctuation confound a servo drive that is trying to
deliver a
constant velocity matching the commands coming from the CNC control.

Within the limit posed by the servo drive's finite gain, the error will be
small. As these ripples become larger, the instantaneous position error
rises until
it reaches a point that is unacceptable for your application. I know
this paragraph
really helps you a LOT!

What I'm trying to say is that the reluctance ripple you feel while
turning the motor
is not, in itself, any indication of motor performance. But, it IS a
sign that there MAY
be some significant torque and velocity ripple, due to the motor
design. The reluctance
ripple you feel has to do with the concentration of field flux lines by
the armature
poles. it is possible to wind the armature so it delivers pretty smooth
velocity and
torque DESPITE this magnetic cogging. One simple test is just count the
armatur
poles! If it has 7 or 9 poles, you have a mess. Cheap motors use as
few poles as
they can get away with, as any extra coils just adds cost. Servo motors
put in many
more poles to get rid of the ripple.

One way to fight these ripples is by having an encoder with lots of
cycles/rev.
1000 cycles/rev gives 4000 quadrature counts/rev, and in many simple
servo drives,
such as Gecko and Rutex (and my PWM drive system, too) servo loop gain
is proportional
to encoder resolution. So, doubling the encoder count will effectively
halve the angular
error. Combining a "lumpy" motor with a 250 cycle/rev encoder will
certainly lead to
mediocre performance. Good enough for a wood router, but may be very
marginal for
a precision milling machine.

(Did I drag this out enough?)

Jon