Ray's Euro Visits

[Ray <rehenry-at-up.net>]

List

I will try to keep this post focused on EMC rather than wondering to far
off into memorable castles, eating establishments, delightful beverages,
and awesome legal driving speeds.

I am thankful that I was able to see a small part of what is being done
with the EMC in Europe.  I visited with three list members, Jan in
Antwerpen, Bert in Eindhoven, and Till in Stuttgart. I have found nearly
all of the list members that I have worked with to be open, friendly, and
tolerant.  These three went well beyond my expectations.  They invited
me in, were open, friendly, willing to share what they had, and knew.  They
each gave me valuable insights into their thinking about the EMC.  

I will divide this post and comment on each visit and then offer my
insights and conclusions at the end.


Belgium

Jan runs the mechanical and industrial end of a chocolate
manufacturing business.  That task includes the design of chocolates
and the manufacture of the molds that they are produced in.  I don't want
to call them candies because that seems to trivialize these wonderful
treats.  "If you can eat more than three," he said, "then it's not real
good chocolate."  And he was right!

I watched his EMC driven pick-and-place machine weld magnets onto chocolate
molds and listened to the plans that he has for three other machines.  He
has worked out some imaginative ways to connect a running EMC with a
programmable logic controller that work very well for him.  

He showed me g-code writers that he's developed in VB to handle much
of the repetitive task of milling an array of shapes.  These will even
handle the giant digitizer files that he produces with his Renishaw cmm. 
We talked about his need for additional probing and digitizing processes. 
And we talked about the need for customizing or internationalizing of the
interfaces, key bindings, and metrics.  

First conclusion from this visit.  Flemish chocolate and lunch are
awesome and Jan is a wonderful host and an innovative user of the EMC.  


Netherlands

Bert is a software developer/programmer for the Nyquist motion control
company.  One of his recent projects was controlling nanometer accurate
motions of pizzo linear motors.  He described these motors to me as being
like a saw blade with a tiny difference in distance from one tooth to the
next and the distances between pizzo crystals that the teeth sit on.  By
exciting adjacent crystals the saw blade can be made to move.  I'm not
sure I'd even recognize one of these as a motor if I saw it under a
microscope. <g>

Bert has fitted EMC to a very nice bench top EMCO mill.  His next
project is an EMCO lathe.  (for the US folk these are not ENCO!)  The mill
is set up with servo motors, tach velocity feedback and quadrature position
encoders for all three axes.  The EMC that he runs is his own NT version
that feeds cubic splines to Nyquist motion boards through a common VME
bus.  This is top of the line stuff.

He wrote his own gui using VB.  It includes most all of the features
available with tkemc and several things that are not available.  He has
included an awesome axis tuning routine.  He has compared motion using
velocity feedback from the tachs and position feedback using only the
encoders -- each method gives about the same quality of motion.  

His gui includes a unique g-code programmer.  He explained it as a
"front end" for Tom's interpreter.  With it he can write subroutines and
go-to's much more easily than you can with my subroutine script.  All
codes are sent to the EMC interpreter using the MDI mode.  He
highlights and works with the active line by capturing the serial numbers
assigned to and used by the NML.  

Bert also has a backplotter included that he wrote using visual c.  It
shows about the same tool path but redraws the plot considerably faster
then the tk versions that Paul and I have written.  

In addition to the future CNC fit to the EMCO lathe, Bert's plans include a
Linux install and a good deal of linux development.  He would like to see
us explore and adopt a different set of languages for the HMI.  He would
also like to see us speed up the development process.

I have two preliminary conclusions from the Netherlands visit.  Bert has
taken many of the ideas and parts of the NIST motion API and has
applied them within the constraints of the NT operating system.  I
consider his work to be an external validation of the worth of the NIST
standards effort. because the EMC concept is in there but the details,
the install, are very different.  In spite of those differences, we could
converse and understand how the software system works and our thinking
about development directions were very similar even though the operating
systems and specific code that we use are very different.

The second conclusions from the Netherlands is that Bert is a
fantastic cook of what he calls "extreme Dutch" cuizine.  We had a
wonderful visit with this family.


Stuttgart

I know something of the world of graduate education.  Sometimes I wish I
was still a part of it.  The number of "toys" that Till Franitza has at
the University of Stuttgart staggers my imagination.  When we arrived he
said that we were fortunate because the big machine was running that day.  

The big machine is BIG.  The concept was developed by a group of
professors and funded by some industrial partners.  It has been moving the
spindle platform for a few months now.  It is not as large as the initial
wave of hexapods like the ones at NIST and Sandia.  It uses fixed length
struts and moves the platform by dragging the far end of the struts with 
linear motors.  Since the industrial partners are sensitive to the release
of information about it, I won't say much here.  

It is a very interesting arrangement.  With it, Till is able to get regular
Cartesian motion and about 15 degrees of tilt on the cutting head.  It
is not run by the EMC program but uses NT software developed by a
software company related to the university there.  This project is a test
of that software and some of the mechanical design.  But the same kinds of
parallel motions have been built into Till's little hexapod, and it does
run with the EMC.  

Till built this tiny hexapod to prove to several of the skeptics around
him that free software and an old 486 could do the same job of running
hexapods as the very expensive and proprietary software and hardware that
had been used on previous university projects.  (There must be at least a
dozen different Stewart and related platform devices in the lab.  The
latest is a combination Stewart platform and robot machine that weighs in
at 30+ tons and was built by a Swedish firm.)  Till has a very good
understanding of the advantages of the EMC software for this kind of
machine tool environment.

I told Till that I would like to show off a hexapod like his at NAMES this
year.  He quickly started to disconnect it and said that I could take it.  
I'll arrange to have it shipped over if I don't make or find something
like it closer to home.  Actually, I was afraid to commit it to luggage
and afraid to try to explain it to security it I carried it onto a plane. 
And I can just imagine what might go through the head of a US customs
officer as he stared at a real live hexapod.

First conclusions from Stuttgart are that Till is making real progress
on the use of PC software to control his machines.  Unlike some on a
related list, most of us on the EMC list are aware of the difficulty of
integration of computers to produce systematic, predictable, and repeatable
motion.  Sometimes these programs seem to defy our best logic and effort. 
Till is working at the leading edge of this with linear motors and
non-Cartesian machines.  Each step forward is cause for celebration.


My EMC conclusions for what they are worth.

1. We need to be able to configure the language displayed in the
menus, buttons, and boxes of a gui.  These need to be variables that are
assigned values from a language file during startup.  The entire world does
not yet speak or easily read and work from US English.

2. We need to be able to assign different sets of keystroke bindings that
work more easily on some of these international keyboards.  Sure, I knew
before this trip that it would not be possible for me to operate an PC in
Japanese or Mandarin but the operatior problems caused by a "quertz" or
an "a?????" keyboard was a real revelation.  I was unable to sit down to
the pick-and-place machine and just type or run anything EMC!!!  With my
raging dyslexia and the totally rearranged keys it was impossible.

3. We must develop some international setup procedures.  A single
configuration line should get us from US settings (inch +) to Metric
settings.   This needs to include values in the ini, parm, and tool files
as well as jog increments and axis displays in the guis.  A significant
part of this is in there and just needs to be tested and documented and
the loose ends tied up.

4.  The conversational programming work that has been developing recently
thanks to Matt, Lawrence, and Terry needs to be expanded, consolidated, and
put into the release and onto sourceforge where more people can get
involved with it.

5. We need much more interaction between EMC related folk and much more
reporting of the kinds of projects being developed.  Our many different
primary languages make this difficult because many are not comfortable
with writing in English or American for natives who use those languages. 
We need to stress that what is most important on emc-at-nist.gov is what is
said not how well we are able to say it.  

6. We need to develop the EMC faster and better.  We can do part of this
if we expand the base of EMC development, improve the organization of
the development process, make documentation central to each step, and "get
the word out" about what the EMC is doing and can do.  I had thought that
the sourceforge repository might do this on it's own, but I now think that
we need to move back and implement Fred's notion of controlling or
advising committees in each of these areas.

7. I could not have been persuaded of this a year ago, but Tcl/Tk may not
be the best language to use as we develop more computation intense
operating environments.  Tcl/Tk is a quick way to develop and test Human
Machine Interface features but we may need to translate these features to
more efficient languages as we move ahead.  I remain convinced that we
need to select open source or freely available languages to use.

8.  I now firmly believe that the members of this list could build almost
any kind of motion machine.  We could apply the code files that we
loosely call the EMC to most any motion problem that could be imagined
and described.  I am certain that we collectively have the necessary,
programming, mechanical, electrical and electronic, math and physics
abilities.

Comments?
--

Ray

PS   I want to thank my son and all of my extended German family for tour
assistance.