RE: [CAD_CAM_EDM_DRO] I'm curious

On your first comment; most modern microsteppers start at full step and
range up to 50,800 with the new stuff pushing 100,000.
Axis 1 and 2 are the RA and Dec respectively with a max speed of 4 rps, Axis
3 is the secondary focuser, and Axis 4 is the Dome the can move up to 19-20
rps.
Motors on axis 1,2 are Zeta4's with S83-135 (450 oz-in), 3 is also Zeta 4
with S57-51 (overkill at 65 oz-in), 4 is an S8 (6 or 7 amps I think) driving
a S102-250 (1500 oz-in).
Encoders on Axis 1,2 are 2500 lines which gives 10000 quadrature, 3 is 5000
line, 4 is 2000 line. With the exception of 3rd axis, encoders on steppers
are for stall detection and a somewhat low precision position maintenance.

All driven by AT6400
(http://www.compumotor.com/products/Controllers__3752__30_32_80_567_29.html)
.

Second one concerns repeatability, compared to a telescope a 3 or 4 axis
router would be a walk in the park. It's not a question of would a router
benefits from a resolution of 1/40000 of an inch, it's what you do with it
to increase your accuracy and repeatability.
If you want high accuracy and repeatability over a few thousands then you
must include environmental elements, it's easy to allow for expansion and
contraction in the linear system as either a simple coefficients
(temperature, pressure and humidity) or empirical data. I prefer the former
because it a bit less of a black box.
Cyclical errors due eccentricity are the simplest to eliminated with "ratio
* sin(angle of the shaft)" or some such derivation of, just plot the error
and create a equation to model it.

-----Original Message-----
From: CAD_CAM_EDM_DRO@yahoogroups.com
[mailto:CAD_CAM_EDM_DRO@yahoogroups.com] On Behalf Of Erie Patsellis
Sent: Wednesday, July 20, 2005 7:20 PM
To: CAD_CAM_EDM_DRO@yahoogroups.com
Subject: Re: [CAD_CAM_EDM_DRO] I'm curious

I have used Compumotor and Pac-Sci drives for several years, True the
compumotor drives do offer some silly microstepping (somewhere around 16,000
steps per rev IIRC, been a while) but it's not likely to be seen in the
final drive. Frankly, anything over 10 microsteps is flat out non repeatable
by 99% of the modern motors. It's quite simply a hollow number, now tell me
more about your telescope drive, the last few I built didn't need fast
slewing, and (so far) have faired quite well, using Gecko drives, too!. The
gecko servo drives ARE closed loop (as you dont' specify where to close the
loop, I pick at the drive) tell the drive to move x units, it either will or
will issue a fault. those are the only two options with any servo drive, not
just the Geckos.

Funny thing, I retrofitted a router a few months back, was basically built
for a stepper drive system, about .5" per rev of the pinion gear.
Then in a fit of 'specmanship' was fitted with AC servomotors with some
ridiculously high encoder count, ~2048 spr (read in quad) or ~20,000 pulses
per inch. Question, do you think that this router can resolve 40,000
discrete steps per inch? When retrofitted with a Gecko, 300w DC servomotors
and some (badly needed) gear reduction, it will now resolve ~10,000 spi,
however the actual resolution is swamped by several cyclical errors, due to
the non cocentricity of several of the pulleys.
I was able to minimize it by a magnatude or two by careful tightening of the
hubs (clamping hub L series belts), but it is still the dominant error mode
in that machine. Now, do you honestly thing an error of half a thou is that
critical on a machine routing MDF 16 hours a day?
Customer said yes, paid accordingly; reality says I can induce 2-10 thou of
error by increasing the humidity of the material post cutting. Hell, I can
get the machine to grow nearly 1/8" over the 12' length by measuring in the
early morning and then again in the scorching afternoon......


erie