Re: [CAD_CAM_EDM_DRO] Re: Why is the push to such high step rates (Grex and Arc100)?

lcdpublishing wrote:

> So, if that is correct, then the only advantage
>of having 256 microsteps is perhaps a smother motion?
>

Have you found a Gecko G201 to not be smooth enough? I doubt it, unless
you are diamond turning telescope mirrors.

> Again,
>assuming microstep as a stepper motor driver, how many stepper
>motors are out there that can deliver any useable torque over 1000
>RPM?
>
>

I hooked some ancient NEMA 34 motors to a 150-Lb mini-mill and demoed it
at the 2003 NAMES show at 90 IPM. With 16 TPI leadscrews and 1:1 coupling,
the motors were spinning at 1440 RPM. That certainly is near the practical
limit with conventional motors. It required a VERY smooth pulse train
(which
I got with my Universal Stepper Controller board) and a very good
microstepping
driver (Gecko 201).

>As for servos, I can see where you can put on them very high
>resolution encoders, but where and how would you use something like
>that? Servos don't have much in the form of low RPM torque, so
>running something direct drive with them is unusual such as a timing
>belt pulley attached directly to the servo.
>
>

Not that unusual. In this case, you probably are not taking advantage of
the full power output of the motor, but for purposes of limiting rotational
inertia, using the motor without reduction may be the desirable. A high
belt reduction ratio may make the motor's inertia greater than the leadscew
plus table. Most high-end machine tools are built this way. it may be to
minimize inertia and maximize acceleration, or it may be to just eliminate
one more wear component (the belt).

>As for resolution and precision, in machine tools, it's pretty hard
>to find anything with a resolution better than .0001" or .001MM.
>
>

On my Bridgeport mill I have X and Y resolution of .00005" (50 uInch)
and a Z
resolution of .000025" X and Y have 5 TPI screws with a 1000 cycle/rev
encoder directly on the end of the screw. Z has a 10 TPI ballscrew with the
same type encoder on the screw.

My 150-Lb minimill has resolution of 128,000 encoder counts per inch, or
7.8 uInch. That has 16 TPI screws, 4:1 belt reduction and 500 cycle/rev
encoders
on the motors.

I have no illusion that either of these machines is ACCURATE to that
level, but
that is the resolution of the encoders.

>Beyond that is very impractical without taking some very special
>steps to insure the mechanicals can deal with any more resolution.
>
>Perhaps there is something in the production of silicon wafers or
>something that this sort of resolution and speed is utilized?
>

Absolutely. Wafer steppers move the wavers with astonishing precision,
gliding
on Dover air bearings, and measured with laser interferometers to a true
accuracy
of a couple of wavelengths of IR light over the entire surface of the wafer.

> I
>can't imagine anything in: metal working, woodworking, robotics,
>etc. so I am sort of stumped as to what applications this would be
>for?
>
>
>

The old, traditional velocity servo system used a DC tachometer to measure
velocity and keep the motion smooth, even when the CNC control had a very
slow update rate, like 60 Hz. After they got the incandescent bulbs out
of the
encoders, the next reliability thing that could be attacked was the
tach. If you
want the same smooth motion a DC tach can provide, you need something that
provides a good pulse rate even when moving agonizingly slowly. When
interpolating
large-radius arcs or moving linearly not quite parallel to the machine's
axes, it calls
for very slow movement of one axis. To avoid stair-steps, the motion
must be
kept smooth. For instance, on my Bridgeport mentioned above, (which has
a DC
tach) it can stay smooth down to about .01 IPM. At that rate the
encoder is only
providing 3 pulses a second. No way could a CNC motion control system
maintain
smooth movement with only 3 pulses a second. It would need 100+ pulses
a second
to keep it smooth, due to the quantized nature of the encoder.

Jon