RE: SLO-SYN 12.7 Amp motors

> -----Original Message-----
> From: Jim Fackert [SMTP:jfackert@...]
> Sent: Monday, January 10, 2000 6:44 PM
> To: CAD_CAM_EDM_DRO@onelist.com
> Subject: Re: [CAD_CAM_EDM_DRO] SLO-SYN 12.7 Amp motors
>
> Actually, it's another argument to get the accel/decel., step and
> direction
> stuff out of the pc. Use the pc for the user interface, and let it tell
> the
> (stepper OR servo) motor controller where to send the servos, and let them
> figure out how fast they can do it without losing synch. You are really
> begging for trouble trying to control real time non-linear stuff and
> interface stuff with the same processor. There is certainly a good
> argument
> for closed loop servo in place of open loop steppers, particularly where
> the
> motors are near unable to move the mechanism every step. Closed loop
> steppers area still a possibility though.
> imho...
>
> Jim Fackert
>

The biggest headache in a closed loop system (other than tuning) is
usually the encoder. They are expensive and prone to failure. Now add the
cost of an encoder to a step motor, the fact that large step/direction amps
are more expensive than servo amps and you have almost as much money in a
closed loop stepper system as you would a convertional servo drive. If you
try to make it a true servo system with PI or PID filtering then you also
have the tuning problems too - this in addition to the fact that you still
have a motor which, by design, is going to resonate.

Seems more sensible to stick with servo drives if you really can't
afford the possibility of a following error. However, I doubt most knee
mill applications really need it. Our old Bridgeport Boss 5 never skips
steps, even with mushy L/R drives. Many of the older step motor driven
mills are still running - probably more than the early servo systems which
were known for instability. True, the newer servo drives are more reliable
and easier to tune, but stepper systems have improved greatly and still
don't require tweaking of a transfer function.

The occasional lost step, if it occurs, can be corrected at the end
of a cycle with proper homing sequences. If the loads are predictable and
if (big if) the profiler doesn't stutter then lost steps should not be an
issue.

Doug Harrison