Re: [CAD_CAM_EDM_DRO] re:Servo not stalling, a problem?

ballendo@... wrote:

> By definition, a servo is a "feedback" driven device.
>
> ANYTHING which causes the "loss" or "incorrectness" of this required
> feedback, CAN lead to axis "runaway". This could be the unplugging of
> a connector(or a wire breaking), a software glitch, or a servo
> drive "error" (not likely in most modern drives, but a "classic"
> problem). But the careful "balancing act" that IS a servo system can
> and does become unbalanced at times. Sometimes it is just a slow
> moving problem, or a simple adjustment... Or...
>
> If (when!?) the axis "runs away", the servo motor will be trying
> to "get somewhere it can't", either mechanically or electronically.
>
> Usually this means the FULL FORCE of the motor is being applied where
> it shouldn't be... Since the motor keeps trying to "get there from
> here", heat and/or pressure build up and... motor burns out/ drive
> fails/ machines(or people)get crushed/ tools break/ parts are ruined/

Since I developed my own servo amps, I've seen nearly all possible
failure
modes. Since the servo amp has a number of different sources of
feedback,
a failure of one of those generally does not cause a disaster, but a
controlled
fault trip.

Failure of the encoder means the CNC control has no idea where the table
is,
or that it is even moving. But, the servo amp has velocity feedback
from the
tach, so it will never move faster than the max design velocity, as the
velocity
dac can not supply a larger command signal than that value. Very
quickly,
the following error between where the table should be and the apparent
lack
of any motion will cause an E-stop.

Failure of the tach is more serious, but will generally cause a jerk of
maybe 1/4"
or so, and then a following error condition will cause an E-stop. The
velocity
control loop is not designed to be run open-loop, and so it may cause an

overcurrent trip.

Failure of the armature current feedback will cause a rapid increase of
motor
current. A good amp uses a completely independent method to check motor

current and shut the amp off if the threshold is exceeded. This
condition won't
likely cause much motion at all, just a loud click.

A properly designed servo system is not a disaster waiting to happen,
but
a reliable system. It IS possible for malfunctions to cause an axis to
try to
runaway at high speed, but fairly simple fault detection methods can
detect
most of those situations and cause an E-stop. Proper E-stop circuitry
should contain measures to remove servo power from the servo amps, such
that shorted transistors cannot continue feeding power to a motor. They

should also have energy dumping features to bring rapidly moving motors
to a quick but controlled stop.

Jon