RE: [CAD_CAM_EDM_DRO] anyone using jr-kerr servos

Hello,

Nice to see the interest in the pic-servo system! Lots of great questions.

I will try to answer as many as I can but I am just getting started. I have
been working with the pic servo CMC since it came out, and the v4.0 chips
for about 6months, and I think I have just scratched the surface.

My original intent was to build a stepper system, but I could not justify
building a machine that could get overloaded and not know it. I wanted the
machine to be able to automatically engage e-stop on axis fault, be able to
keep its location after an overload, and also go faster than 70-80 ipm.

my machine travels 9.5"-x 6"-y and 7"-z. I tried using the 5/8" ball screw
from mcmaster carr, but those things are ugly. they have a rough surface
finish, and lots of backlash. At 1200 rpm they make lots of noise. I am
trying to find another source for ball screws but its all expensive. For now
I am using 5/8" 8tpi nook acme screw for each axis. I have 33 watt 24 volt
brush servos, with 660 line encoders on each axis. This gives the machine a
resolution of .000047" that is not to say the system will repeat, or be
accurate to .000047", there is a difference between resolution,
repeatability, and accuracy. I would NOT design a pic servo system with less
resolution, in fact, if I had it to do again I would build in more
resolution at the cost of speed. I have emc running on a pent 233mhz and
have a step rate of 52khz which drives the motors at about 1200 rpm for a
rapid rate of aprox. 150ipm. The max I have achieved has been 190ipm, but I
don't have a counter balance on the z axis, so it would fault on every long
rapid move in the positive z-axis direction. Also at that speed the emc gui
is tooooo slow.

The following error I have been getting, of course, gets greater with
increased speed. But at a cutting speed of 10ipm I can set up a laptop
computer to continuously monitor the axis error, and I have been getting
about 10 to 20 encoder counts worth of lag. At rapid it peaks to as high as
400, and averages about 150.

So far I have been able to interpolate full circles in emc to within
.0007" roundness. I would like this to be better, but with screws that are
only made to .003 per foot accuracy, I think I am up against a wall. One
thing I have noticed about this system is that when taking heavy rough cuts
on a circular interpolation move the machine tends to overshoot when the
backlash compensation is activated. It overshoots quite a bit, like in the
area of .001, so the machine takes a noticeable gouge out of the part. This
tendency disappears on finish passes, so it is my theory that the
combination of following error and instant backlash compensation causes the
thing to overshoot but I don't know for shure, yet.

As for smoothness, well, there is a noticeable amount of added noise
coming from the servos when in a feed move. Rapids tend to sound the same,
whether on emc, or using the serial command interface. Circular moves sound
interesting on the emc. This noise can be seen in the finish of the part you
are creating, but so far I have not been disappointed. Moves are not at all
"jerky" but the emc does cause more vibration than the direct serial command
line.

I like the fact that the pic servos, driven with emc, unless power is
removed, can stay is synch with emc even after an axis fault condition. I
also like the speed. I don't like being forced to drive the system with
step/dir signals, but I am working on changing that right now.

In my opinion one major limiting factor in this case is the pic servos
ability to "zero" out. I have been experimenting with the Proportional,
Derivative, and Integral parameters but so far the best I can do is use P.
and D. and leave I. at zero. This causes the pic servo to be out of
position, after any kind of move, by between 4, to 6, encoder counts. I
think this can be worked out but I have not found a happy medium yet. I have
so far attributed this to my own inability, so I will keep trying.

I really like the new point buffer he has come out with in the new cmc
version. Its ability to reduce the demand for a realtime computing
environment is cool. I also like the fact that if you drive all your pic
servos from the same clock source, they will be able to "synch" with each
other extremely well. I have not looked closely at it yet, but my first
response is to think we will be wishing it had more point storage
capability, but we will see.


Hope this answered the questions that were asked and some that were not
asked.


Andrew Abken
Machinist
Spectra Physics Lasers

>From: "Carlos Guillermo" <carlos@...>
>Reply-To: CAD_CAM_EDM_DRO@egroups.com
>To: <CAD_CAM_EDM_DRO@egroups.com>
>Subject: RE: [CAD_CAM_EDM_DRO] anyone using jr-kerr servos
>Date: Thu, 1 Jun 2000 19:53:41 -0400
>
>Andrew -
>
>I've thought about using the JRKerr set up. How's the smoothness and
>following error with the step/dir input version? Also, what kind of speeds
>can you get, and what's the limiting factor?
>
>Thanks in advance,
>Carlos Guillermo
>VERVE Engineering & Design
>
>

________________________________________________________________________
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com