Re: NIST EMC Software

Dan Falck wrote:

> From: Dan Falck <dfalck@...>
>
> This is directed to Jon Elson, who is using the Nist EMC CNC package, but
> should be of interest to all.
>
> Jon,
>
> Will you, theoretically, be able to sync the spindle of your mill to the
> Z-axis? Thread milling and lathe conversions are what I am interested in
> on this subject.

Thread milling doesn't require spindle sync. Any 3-axis machine should
be able to mill threads. I got a neat little thread mill for about $30, which
has a single row of 6 teeth. It looks sort of like a tiny double-angle cutter
mounted on a shank like a woodruff keyseat cutter. This can mill any
thread profile that isn't too coarse for the size of the tooth. I wrote a
program to write thread milling g-code programs, and it worked perfectly
the first time (a rare event).

Now, if you want to do hard tapping, you do need the spindle encoder, and
a prety tightly controlled spindle drive, so the CNC can stop and reverse
when the desired thread depth is reached. One problem with hard tapping
is that there will be no relief clutch to slip if the tap binds. I'm told that the
code is in EMC to handle a spindle encoder, and that hard tapping has been
done (I believe GM Powertrain tested this function). But, I don't believe
that anyone has a machine with that feature up right now.

> My next question is to Dan Mauch:
>
> Dan ,
>
> Have you looked at the Nist EMC package? They now have a step and
> direction version that would work with the servo system that you have (it
> takes step and direction pulses for input).

Yeesh! Why take EMC, set it up for steppers, and then convert back to use
servos? Why not use EMC set for servos, and have it drive servo motors
through analog-input servo amps?

>
> My last question is to the group:
>
> Who on the list is using the Nist EMC package?
>
> My interest in the package is this: I have used Dancam , Stepster, and now
> Maxnc software to control my small milling machines. These packages are
> fine for the price, but the Nist package is free and looks very powerful.
> It seems to take most G-codes that I am familiar with - many more than
> Maxnc and Stepster. The source code is available for the programmers in
> the group. It runs in RT Linux, which I am told is good (I am a newby to
> Linux).

It is TRUE, HARD REAL-TIME, unlike any of the systems you mention
above. So, feed rates (velocities) are accurate.

> Fred Proctor and the Nist group seem to be pretty quick to fix
> bugs and add features.

Yes, they'be done a good job of greatly increasing functionality, while
reducing bugs.

> It seems like you can run it via network, with
> another machine running the GUI, while the controller can run on another
> machine- so that you could use Windows or Xwindows to make it look good to
> the user.

Why use two computers? I do CAD design on a Windows PC, send the
files by network to the Linux machine running EMC, and then cut the part.
A ho-hum Pentium (I'm using a 100 MHz Pentium classic, with 16 MB
of memory) is PLENTY of power to run the real-time component of the
motion control and trajectory planning, as well as the non-real-time
user interface, and still be able to run the text editor, etc. at the same
time!
Actually, for simple tasks like circular borings, circular posts, rectangular
pockets, circular patterns of holes (optionally with peck drilling cycles)
and similar stuff, I wrote some programs that I run directly on the Linux
machine, in another window, to make the G-code.
If I'm going to make a number of parts with these sorts of features, I use my
programs to create the primitive operations, and then combine all the
G-code programs in the text editor, with whatever little changes are needed
to get optimum production from the machine.

As for looking 'good', see the 2nd picture in :
http://ascc.artsci.wustl.edu/~jmelson/EMC.html
I think you'll agree this looks just fine as a CNC control's display.

Jon