Re: [CAD_CAM_EDM_DRO] Re: More about motors?

ballendo@... wrote:

> Smoke wrote:
> >Looks like servo motors are IT then if I want an ACCURATE system and
> >steppers if I want less cost.
>
> Smoke,
>
> I Can't let this MYTH be perpetuated!
>
> There is no INHERANT difference beween the ACCURACY attained by
> steppers vs. servos. In fact, a case could be made (strongly) that
> with a stepper system you ALWAYS know the limits of your accuracy,
> while with servos the accuracy can be "all over the place", depending
> on system stability, etc.

I can't let this MYTH be perpetuated any longer!

Steppers run in an open-loop configuration (no encoders or other
position feedback system) are as accurate as the steppers and other
components are. What is the accuracy of the size of the steps on
a typical stepper motor? How about 5%? What is the accuracy of
the pulses on a typical shaft encoder for machine tool use? How about
better than 1%?

Now, let's say you micro-step that stepper with the 5% accuracy with
a multiplier of 8. Now, the (micro) step size is accurate to 40%
of a micro-step. But, as you micro-step in finer and finer divisions,
the torque the stepper puts out when moved one micro step gets
smaller and smaller, such that the stepper doesn't move at all when
stepped one single microstep. So, now, the error can be proven
to be +/- 100% of a (micro)step! Hmmm, that doesn't sound good at
all!

I built my own servo amps. I do not have a Doctorate in control science.
But, after working with it for a while, I got it to where I have great
difficulty measuring the error. It holds very solidly within +/-.0001"
position (as read by the encoder) and can accelerate while staying
well within .005" or less.

>
> There are stepper systems holding MICRON accuracy in the electronics
> and optics field. Servo's too.

Yes, and they are not using 5 TPI screws and moving loads up to 1000 Lbs
at 120 IPM or more!

>
> Accuracy is a function of the system design. It would be like saying
> propeller driven planes can't fly; only jets. Both fly. Both have
> pluses and minuses.
>
> As I said in my reply earlier, this IS a tougher question than it
> used to be. And it used to be tough. :-) I would rather have a solid
> well designed stepper system instead of a low-end servo, most days.
> Two years ago, I'd have said ANY day!
>
> Here's why:
>
> A stepper is a digital discrete device. As long as I work within its'
> torque/ speed limits, it WILL BE where I've told it to be, within
> its' step resolution. Nothing else determines this. I send a step. It
> takes a step. So I will always have an accuracy within ONE step.

No, you won't! If your motors have standard accuracy, the MOTORS
themselves can be off easily by up to a couple steps from the ideal
position computed by dividing a circle into 200 divisions (or
whatever number of steps / rev it does). If you split the steps
into finer divisions, the error just gets larger in terms of the new,
smaller step size. Finally, due to friction, the step motor is NOT
at the exact location it would have with no load. It is always behind
that position by some amount determined by the load. You just THINK
it is where you told it to be because you haven't measured it!

> A servo is a feedback analog device. It is, by definition, driven by
> the ERROR between where it "is", and where it "should" be. This error
> is determined by MANY components of the system; encoder resolution
> and accuracy, drive resolution and accuracy, "firmware"(programs that
> reside on chips, like your computer BIOS) or software equations and
> algorithms.

With suitable gain, the error is about 1 encoder count. In my case, that
is .00005" in the X and Y axes, .000025" in the Z. That is, of course,
before the lead screw errors come in, but that would be true with a
stepper, too.

> So, the Accuracy of a servo system is reliant on several things
> working together correctly, and well. A stepper system accuracy is
> dependant on the motor itself. Period.

Yes, so if there are errors, you have no idea what they are.

> (redundant items like
> leadscrews omitted, since BOTH systems are equally affected)
> Also, microstepping introduces potential rotational errors if the
> motor and drive aren't matched.

And, so a servo system MEASURES its' own errors, and corrects them
to the best of its ability. If it is unable to correct the error to within
a preset value, it stops the machine with a warning, and you can correct
the problem, like a crash.

> To sum up, servos' ARE being made to work to a higher degree of
> accuracy and reliability than in years past. And so ARE becoming
> a "viable" choice for lower-cost projects. Like yours.

Servos have been used since before steppers were even invented!
The first NC machine tools used servo motors in 1959 or so.

> The stepper just sits there, doing what it always has done, reliably
> positioning; not as fast maybe, but ACCURATELY.

Yes, it sits there, accurately reproducing the errors that were built
into it at the factory, ad infinitum. Yes, it can be plenty repeatable,
but repeatability is only one of the 3 'legs' of accuracy!

Does anyone have a spec for the typical or maximum permitted
position error on a typical stepper motor? This is a spec that
the manufacturers don't want to publish.

Well, nothing like a good flame war to keep us warm on these
winter nights!

Jon