Re: [CAD_CAM_EDM_DRO] Re: How to install motors into a Grizzly or Enco Milling CNC?

> I bought the drill/mill (or bench mill) from Grizzly. It
> is pretty sturdy and although kind of small, it will serve me for
> almost everything I have in mind. I am an electric engineer and I
> work with motion controls for photocopiers at Xerox so the machine
> stuff is only for hobby robotics.

Sounds familiar... I work on various equipment in a research lab at
MIT, which lately has meant trying to make a PMAC motion control board
do what we need. At the moment there is remarkable convergence and
crossover with my hobby CNC efforts - I keep thinking of how to solve
the problems I'm having at work while playing with stuff at home, and
vice versa.

> At this moment I am not sure which motors I am going to use, but most
> likely it is going to be a DC with an encoder, a brushless DC with
> also an encoder or a Stepper. The one I like the most is the
> Brushless DC because of its ease to control. I plan on designing my
> own controllers as it seems to be pretty easy to use a microcontroller
> and send the appropriate signals. The problem is how to interface it
> with the computer program. I don't have a problem with sending
> commands to the motors through the serial or parallel port. But how
> to take a 3D image and send it to the controller for interpretation,
> I still don't know how I will manage.

I would suggest that you design to an existing common interface such
as step and direction, or +/-10v analog velocity or the like so that
you can use existing computer software to control your machine. Even
if you end up writing your own in the end, being able to start with an
existing program will get you to the satisfaction of actually making
first chips sooner.

One thing you may want to consider if you have any experience with
unix-type platforms is the Enhanced Machine Controller (EMC) which
runs under a realtime version of linux. Although it takes a while to
understand, it's actually not too hard to write custom drivers to make
it talk to any kind of motor interface you might build.

These programs all execute machine command files called "G" code which
consists of simple instruction - cut from this point to this point
this fast, etc. If you want to take a 3D image and turn it into this,
you need CAM software that can do toolpath planning. This is no
simple task... If you really want to try to do it yourself, a
reasonable approach would be to start with a 3D format like STL, which
consists of triangular facets. Then you can slice that into a stack
of 2D layers, encoded as polygon outlines, each one either solid or
empty. Turn each polygon into commands for a mill spindle, making
sure to offset the endmill by half its diameter... and watching for
intersections with the other polygons. Then you move the cutter down
a small amount to the next layer and repeate. Obviously you get some
stair-stepping this way, but it breaks the problem down into a series
of manageable chunks, any of which you can buy commercial solutions
for.

> I do not plan on changing the ball screws on my machine as they seem
> to be quite expensive. I know there is a backlash on the ones my
> machine brought. At this time it is around 15 mils which is not good,
> but maybe I can compensate this with the code in the microcontroller.

The screws that came with your machine are simple leadscrews -
probably what is known as an 'acme' thread due to its squarish
profile. Nuts for ball screws use bearing balls in helical tracks to
provide rolling contact, which allows tighter tolerances, and hence
less backlash. Some sizes of ballscrew are a lot cheper than others
(5/8 and 1") but it still makes sense to start with what you have, and
upgrade later if you can find a way to make the bigger ball nuts fit.
15 mils is a lot for a small machine. Some people decrease it by
cutting a slit in the nut and building something to tighten it.
You'll notice the backlash most when milling circles - there will be a
jump when one axis reverses direction. Software can partially
compensate, but cutting loads can make the table move between the
limits of this play. Others have cast nuts out of low friction
materials like moglice, which can have less play and still turn
freely.

> And basically that is my idea at this time. If you guys need any help
> with the electronics, let me know as that is my field and motion
> control is my actual area of expertise (this doesn't mean I am very
> good, but the folks I work with are geniuses and I won't hesitate to
> question them). Hope to talk to the group really soon! Take care!

You'll find many eager ears for any practical designs you are willing
to share - stepper or servo drives, although there are some good
solutions for those out there now, there is always room for more
possibilities. You seem to have some experience with brushless
servos, something many of us do not - I'm sure many would be
interested if you posted a basic design for a driver for these.

Chris

--
Christopher C. Stratton, stratton@...
Instrument Maker, Horn Player & Engineer
22 Adrian Street, Somerville, MA 02143
http://www.mdc.net/~stratton
NEW PHONE NUMBER: (617) 628-1062 home, 253-2606 MIT