RE: EMC vs. Galil Motion card (was: Will emc work for me?)

["Craig Edwards" <cedwards-at-ceinetworks.com>]

One thing to consider with regard to using a "dedicated" or proprietary
multi-axis motion control card is that you won't be able to take advantage
of what I think is really one of the biggest strengths of EMC; it's look
ahead and trajectory planning algorithms.  Not that proprietary solutions
can't have these features, but the beauty of open sourced EMC is simply that
a lot of really talented folks are now able to contribute on-going
improvements to algorithms and coding, where as most commercial products are
not upgradeable at their core.

I'm a relative newbie to CNC, but have been researching a wide variety of
motion control solutions for small multi-axis machine applications.  For
motion control applications which do not require very high performance
synchronization between axis, there are many advantages to a distributed
motion control approach, whereby each axis is controlled by it's own control
loop, reducing the "real time" demands on the centralized CPU.  There are
many approaches to this, ranging in functionality from "dumb" step frequency
generators to stand-alone microprocessor based control modules with higher
level interfaces such as USB, CAN, or RS-485.  Boards such as the Galil
receive their commands across the internal ISA or PCI bus, but are still
somewhat "isolated" from the real time code executing on the primary CPU.

Personally, my opinion is that since primary processors, memory, etc. have
declined so much in price and increased so much in power, one high speed CPU
running all the low level (PID loops, etc) code for all the axis makes the
most sense. Both from a performance (meaning keeping the cutting tool where
it's supposed to be at all times while moving it at maximum speed) and cost
perspective.  3 or 4 years ago this may not have been the case, nor is it
today with non real-time OS like Windows.  But assuming a real time variant
of Linux.......  Of course, there would still need to be a small amount of
dedicated hardware in the form of DACs and encoder/counters (if running
servos), but there are low cost solutions available and more in development.

I apologize if this is getting somewhat off topic from the thread, but I've
really been putting some effort into understanding the tradeoffs with regard
to motion control architectures for small multi-axis machines.... and this
question seemed a good place to bring out this issue.  I think EMC has
enormous potential in the arena of coordinated highly optimized multi axis
control, (despite all these 4th axis issues...) and I wanted to highlight
what I think was a great design decision.  Since I really believe that
software architecture should drive the hardware requirements I think it
would be great to hear from the real experts on this list as to what their
"ideal" hardware environment might look like!

just my 2 cents, hope it stimulates some further discussion.

-Craig Edwards

-----Original Message-----
From: emc-at-nist.gov [emc-at-nist.gov]On Behalf Of Ray Henry
Sent: Saturday, July 06, 2002 1:51 PM
To: Multiple recipients of list
Subject: Re: EMC vs. Galil Motion card (was: Will emc work for me?)




Scratch might be easier but EMC's interpreter will run stand alone and
produce canonical motion commands from a GCode file.  If you translated
these
commands into Galil rather than starting from scratch you might gain the
benefit of the machine world view and verifying of GCode that are done by
the
interpreter.  The cost/benefit over starting from scratch would depend upon
the Galil language.

Tickle/Tk is available for Mac and for MS-Windows so the tkemc gui can be
moved.  Several list members worked on fixing tkemc and emcsh to fit a
MS-Windows environment and demonstrated it at NAMES in April.  We might be
able to write a new extended wish shell like emcsh that would communicate
with the Galil.  The advantage of that is that we could retain the present
look and feel and keep the Galil stuff current with further development of
that gui if that is important.

We could also write a tickle script that would serve as a setup program for
the Galil card.

Ray


On Saturday 06 July 2002 07:53, Carl wrote:
> On 07/05/02, at 07:16 PM, "Markus Meyer" <meyer-at-mesw.de> said:
> >So the Galil card follows a very different
> >concept, because all critical calculation takes place on the board rather
> > than on the main CPU of the computer, as it does with EMC.
>
> Which is why it works even with Windows running on the CPU. ;-)
>
> >AISE it would be far easier to write a frontend for the Galil card that
> > shows some nice buttons and additionally write some script that
> >translates GCode into the native Galil card language than hacking EMC to
> >support the card, effectively throwing away (1) and (2).
>
> Would the emc frontends be useful as building blocks for such an effort?
>
> If this is getting to be a FAQ here, then there is obviously some demand.
> Is anyone working on it already?