Re: [CAD_CAM_EDM_DRO] Re: stepper vs servo

cadcamcenter@... wrote:

> Jon Elson <elson@p...> wrote:
> > Servos don't stall, unless the physical limits of the machine are
> reached.
>
> Trying to get exact meaning of terms:
> stall = stop?
> stall = stop temporily?
> Servos don't stall = there are no conditions, however severe, that
> they will stop/stop temporily when they are supposed to be moving?
> When torque is insufficient, only the feedrate fall below the
> commanded feedrate?

On many commercial machines, this is literally true. They are so powerful,
the will destroy the interfering object, or they will destroy the machine!
My system is designed to limit at about 1000 Lbs of linear force at the
table.
I'm a pretty conservative guy. I think the Bridgeport EZ-trak has a limit
set
somewhere around 5000 LBs! That's pretty scary.

I can't imagine any reasonable machining operation on a small mill or lathe
that would develop cutting forces anywhere near this level, and these kind
of forces could easily destroy a mid-sized machine.

So, if the such a servo ever slows down enough to cause a following error,
or indeed, stalls completely, there is something drastically wrong, and
causing
an emergency stop sounds like the most appropriate behavior.

> For steppers:
> stall = stop?
> stall = lose step?

When steppers are asked to accelerate too rapidly or run at too high a speed

under load, they will frequently stall, meaning they stand still and buzz.
Under
transient acceleration problems due to ragged step pulse trains, they may
just lose an occasional step (or 4). This is a real problem, because all
positions
will be off by an unknown amount until the machine is re-homed. Stalls
are pretty obvious, the machine could be inches off the correct position.
An occasional skipped step will not be obvious, and parts created after
that point will be off in some manner.

> > They can slow down, however. Some systems (the Allen-bradley 7320
> > I worked with before, for instance) had a feature called feed-rate
> > compensation.
> > Whenever one axis lagged more than some limit behind commanded
> position,
> > all axes cut their feedrate in half, hoping that would allow the
> lagging
> > axis
> > to catch up.
>
> Meaning servos alway act in unison? When Y slow down, X will slow
> down simultaneously, and there is no time-lag?

This is a function of the CNC control. Allen-Bradley has it, EMC doesn't.

> My feeble mind: The only way that X will know Y has slowed down is
> when Y has already slowed down and a signal is issued that this has
> happened, by which time X may already have proceeded some steps
> beyond Y? By which time there will already be some deviation, even if
> minute, from the planned path?

You must understand that the machine is NEVER where it is commanded
to be. there is ALWAYS some error between commanded position and
actual position. The job of a servo system is to make smooth motion, the
CNC control's job is to minimize the error. it is constantly comparing
actual
against desired position, and making adjustments in commanded velocity
to reduce the current error to a smaller value. In my CNC system, the CNC
control is making this comparison and computing a new velocity 1000 times
a second. It has a resolution of .00005" (that's 50 micro-inches). Keeping

following error down to a few counts of the 50 u-In encoder every
millisecond
keeps the machine quite close to the desired path at all times, especially
at
normal cutting feedrates.

> Is the answer below:
>
> > And, it is most likely that the lagging axis will leave extra
> material in
> > place,
> > rather than remove extra material, and a later pass can clean it up.

This helps, too.

Jon