Re: [CAD_CAM_EDM_DRO] Re: stepper vs servo

cadcamcenter@... wrote:

> Hi,
>
> Wonder if this describe the difference between a stepper and a servo:
>
> Stepper: A ship where the OPTIMIST captain set a course, and doen't
> touch the steering wheel thereafter, assuming everything will be
> perfect. Everything OK as long as there are no unsurmountable
> obstacles, otherwise course will be off by an offset caused by the
> obstacle. Therefore a straight line will be a straight as long as
> there are no insurmountable obstacles
>
> A _______________________ B no obstacle
>
> A _________
> \_____________ B obstacle
>
> Servo: A ship with a PERFECTIONIST captain always trying to maintain
> a theoritically correct course, making macro/micro (more likely
> micro) adjustments to the steering wheel continuously. (anyone has
> compared steering a car and an ocean going liner?)
> _ ___
> A ___/ \ /\ / \_____ B
> \/ \/
>
> Final destination alway exactly, +- micron, at where it should be,
> course corrections may be very micro, but path will ALWAYS be wavy
> due to numerous course (micro) corrections.
>
> Theoritically then, as long as steppers are sufficiently powerful,
> maybe with very generous margins, the path cut will be more true to
> planned cut, whereas servos will alway be (microscopically) cutting
> a "wavy" path because of continuous (micron) course corrections.

No, absolutely wrong! Steppers have discrete steps, and can't position
in between them, without microstepping drives. Servos DO, indeed, cruise
smoothly between encoder points (at least true velocity servos with DC
tachs)! My servo system is totally smooth all the way down to .01 IPM!
That is about 3 encoder ticks per second! Because of the discrete steps
of a stepper motor, it is absolutely impossible to get smooth motion on the
slow axis of a coordinated move with plain steppers. Due to the sluggish
performance of microstepped steppers at very slow speeds, they can only
do a little bit better. The reason is that the torque produced for small
sub-step
moves is very small, and the friction and other dragging forces are high.
The stepper, when loaded with the typical machine tool, will develop more
and
more torque and then suddenly jump a substantial fraction of a step.

Typical servo systems have a digital to analog converter that produces
quite precise velocity commands to the servo amp. Typically, these go from
12 to 16 bits, or 1 part in 2048 up to one part in 65,536. This allows
extremely
fine adjustments in machine velocity, to make minute corrections in the path

being taken. My particular setup has encoder resolution of .00005" on the X
and Y
axes. Yes, that is NOT a typo, it is 50 Micro-Inches.

> Therefore, as long as there is sufficient margin on torque, a stepper
> more precise (pathwise) than a servo?

I can't believe these ill thought-out rumors are still going around! This
is total
FANTASY! If you want accuracy, talk to the guys who build the wafer
steppers
that print the patterns onto IC wafers. They use servos with air bearing
stages
and laser interferometers as their linear encoders. They position to
NANO-METER
resolution over a 12" travel, and while they don't do anything while moving,
they
could, as they are extremely smooth drives. (That's the high end.)

Don't ever think you are going to get something for nothing (except by
dumpster diving, which can be quite interesting). First, think of the
accuracy
of the magnet alignment in a stepper. The stated accuracy is generally +/-
5%.
Therefore, the positioning of any step is within 5% of a full rotation, or
18 DEGREES!
Assuming one full step = .001" with a 5 TPI leadscrew, then an 18 degree
error
equals .010" error. this will be a cyclical error that will build and wane
each .200"
of table travel, assuming a worst case stepper motor. Now, of course, some
of
them will be much better, but right there, you can see how much error you
are
getting, with no way to correct for it with an open-loop stepper.
Microstepping
will not be any help with this, without an encoder for feedback. Again,
variations
in the magnets and stator cores prevent the microstepping from finely
subdividing
the steps accurately.

Now, a servo system with a machine-tool quality encoder will have much
higher
accuracy of the encoder transitions. Good encoders are usually spec'd such
that
a transition is within +/- 1 cycle of its correct position all the way
around the
circle. For a 1000 cycle/rev encoder, where you get 4000 transitions
(encoder
counts) per turn, that is no worse than 1 part per thousand or .1 %, 50
TIMES
more accurate than the stepper. Here, my cyclical error will be no worse
than
4 counts, or .0002"!

With the velocity feedback from the DC tach to keep motion smooth, the high
encoder resolution and the CPU making 'steering' adjustments at 1000 times
per second, these errors are going to be MIGHTY small. I routinely bore
holes
with an end mill, making the machine's table follow a circular orbiting
path, and
the hole will come out VERY smooth! I can get it just a little smoother and
rounder
with a boring head, but milling it out is a LOT faster and more flexible.

> Hair splitter.

Well, it depends on what you are doing, but this is NOT splitting hairs.
For extremely
fine machining, which I sometimes do, the added precision and smoothness of
a
servo system CAN NOT BE BEAT! This is why ALL high end machine tools
use some form of servo, and NOT steppers.

Jon