Introduction/Ideas/Questions

Well I have just finished looking through all 180 digests, so now that
I have some idea what has been going on, I'll introduce myself and my
projects.

Basically, I was trained as an electrical engineer, but spent a lot of
my time in school impersonating a musician, and am finally finding a
happy balance of the two in the craft of instrument making. I've
finally got a reasonable shop setup to work on my french horn and
trumpet projects and am celebrating by taking this next week off from
my real job to try to make some progress on instrument tooling. A lot
of this would be a lot easier to do with computer assist, if not CNC.

Lathe DRO & CNC:

The main thing I need to control at the moment is a 13" South Bend
lathe used to make musical instruments, and more importantly, tapered
mandrels for trumpet and horn bells. Initially, I am trying to get
the US digital/HP linear strips up on at least the Z axis and maybe
the crossfeed as well. Then I plan to hack together software to tell
me where I am in relation to the calculated curves I want to cut.

Once I get tired of "manual numerical control" I really want to put a
drive on at least the crossfeed. I briefly experimented with a 100oz
inch stepper, toothed belt reduction, and L297/298 drivers on my
previous 9" SB lathe, and that seemed like it would work. I fear the
bigger machine will need something larger though. Fortunately, rapid
motion is not really necessary, at least on this axis, so I may just
be able to use lots of reduction. Also thinking of putting a really
crude drive on the Z axis - something like a power screwdriver motor
driving the leadscrew to enable "smart power feed" type operation for
multiple-pass cuts.

Alas the crossfeed screw has tons of backlash due to a worn nut. I'm
thinking of doing a conversion to 5/8-5 rolled ballscrews, but I
haven't yet decided if the nut will fit in the channel in the slide
(there is not much extra iron there to allow for machining it larger).
Avoiding the temptation to just order some screws and nuts, I guess
I'm going to machine a crude mock up of the ballnut that will slide
over the existing screw and get some idea of what intereferences the
return channels may run into. If I do go for the conversion, I will
probably make a new bearing assembly that screws into the casting in
place of the current one, so I'll just remove the screw, power
crossfeed gear, handwheel, and bearing as a single unit and pack them
away in a safe place.

I also have a rockwell knee-mill that I might convert at some point.
But for the moment, I think my primary need for milling might be
better served by bolting the Z axis of the sherline mill I have onto
the lathe cross-slide. With that setup, I might be able to turn a
tapered valve rotor or housing, then lock the spindle. The sherline
head would cut passages through the rotor by simply plunging with some
sort of end-cutting fly cutter, while the lathe cross-slide moved very
slightly on its ballscrew to create the necessary curvature to the passageway.

Steppers or Servos:

Was interested to read about the ideas of using the Kulaga/Mauch DRO
as feedback to EMC to close the loop, and also of the interest in
making a cheap replacement for the servo-to-go card. I looked at
their webpage, and the chips they use are all very standard. In fact,
they use the LS7166 which is related to the LS7266 that everyone seems
to be using. If the Xilinix FPGA isn't doing too much, software
compatability might even be possible - but I would assume that since
EMC is open source and its authors are online, even just getting a
similar programming model would be enough. The D/A outputs shouldn't
be expensive - when I built the line buffers and electrostatic
droplet-stearing DAC boards for MIT's 3d printing machine in 1994 we
used 12 bit DAC's that were less than $15 a piece, and I think prices
have fallen. I've also coded in verilog, so in theory I could help
hack up an FPGA if it is really needed - I think most of the companies
give away 1 year licenses for their development tools.

For my own experiments, I'm thinking of buying a prototype card that
comes with address decode in a pal, 8255 for parallel I/O, 8254 timer,
etc and then using that as a base for the Ls7266 encoder chips and
possible servo DAC's... in fact, I wouldn't be suprised if that's how
the servo-to-go developed. But my first bit of hardware (to be built
in the next 24 hours) will probably be a parallel port interface for
the 7266, to get basic DRO functionality without any need to open the
computer case.

Computer:

Current plan is to use a P120 ultimately under rtlinux/EMC, but
probably under dos/qbasic for initial DRO experiments. (We have linux
running a house network, and I did a test install of EMC on another
machine yesterday). I'm going to put it on the opposite side of a
partition wall behind the lathe with long keyboard/mouse/video cables.
Monitor is likley to be a little monoVGA that does only 640x480 - I
can buy a better one if needed, but it seems like that is the best to
sacrifice to the shop enviornment. I'll build a bracket on the wall
to hold it over the lathe headstock.

Keyboards/Mice:

It hardly seems worth it to spend 20 bucks on a type-through cover to
protect a $5 keyboard... Perhaps I'll experiment with clear plastic
bags. Or just buy lots of $5 keyboards. Mice may be an issue, too.
The MIT 3d printer had a standard mac mouse restrained with a wire tie
and sitting on a little fold out half-mousepad platform on the side of
the rack that the computer was in. I'm tempted to make a little
wooden platform that could sit on the lathe ways or something. I'm
also borrowing an X10 wireless mouse from work to experiment
with... would prove interesting if I could figure out how to decode
some of the other buttons.

Anyway, that is what I'll be up to in the near future.

Chris

--
Christopher C. Stratton, stratton@...
Engineer, Instrument Maker & Horn Player
(978)538-5179 work, (617)492-3358 home
30 Griswold Street Cambridge, MA 02138
http://www.mdc.net/~stratton