Report from Westec

Report from Westec

I spent most of last week at Westec, the big tool show held every spring in
LA, sponsored by the Society of Manufacturing Engineers <www.sme.org>. It
took up three giant rooms at the Convention Center with manufacturers and
resellers of all sorts of manual and CNC machine tools, CAM software,
scanners and metrology equipment, water-jet and laser cutting rigs,
benders, EDM machines, marking devices, cutting tools and abrasives- in
short, anything that deals with parts manufacturing in one way or another.

Each room took a day to see, at least at the pace I was going. If one
refused to be fascinated by anything, never stayed for a demo, and
invariably rushed by instead of listening to a salesman's spiel, it would
be possible to see it all in a day, I suppose, but it wouldn't be as much
fun by a long shot. I lingered around the things I found most interesting,
of course, and doubtless sped by a lot of things others would stare at
longer, but there was plenty to keep me going for the whole three days.

Interesting examples of CNC mill design abounded: A bridge-type mill with a
x travel of 196 inches, y of 122 inches, and 47 inches of z ("less than a
million dollars" from www.correausa.com); a high-speed VMC in operation,
spewing inch-high blue chips like popcorn, a 5-axis mill-head with smooth
ball-joint motion within a hemispherical envelope, a mill with an arm
moving horizontally and in and out, holding a beam moving up and down in
its "fist", working on a piece of metal that stays still
(www.ops-gmbh.com).

All the big mill manufacturers were there: Bridgeport (they even had a
"Coke Classic" reissue of their old standby manual mill, but painted
black), Cincinatti, Fadal (showing off their big fat internally-chilled
ballscrews) Mori-Seiki, Milltronics, Okuma, Mazak, and Haas
<www.HaasCNC.com>. I guess I relate to small mills better, but I liked the
Haas Mini-mill - it seemed very competitively priced at $30k, having
travels of 16" x 12" x 10" a 6000 rpm spindle, enclosure, and a 10-tool
changer. By adding options, you can get up to a Super Mini-mill for $50k
with a 15,000 rpm spindle, better toolchanger, and faster control, for
high-speed machining and 1200 ipm rapids. Subtract options and money and
they'd sell you a "Toolroom mill" for $20k that had a 30" x travel, lacked
the toolchanger and enclosure, and came with a slower spindle and
handwheels as well as servos. MHO/Millright had a nice little mill too,
built on the bridge plan, with a 12-tool changer. And Servo, as well as
their larger models, had a small, fast, tight little mill for $17k, but the
Z-height was really low at 3.5".

There were smaller mills there too- MaxNC <www.maxnc.com> were showing the
closed-loop stepper driven MaxNC15 CL2 ($3206), and Roland Digital
<www.rolanddga.com>, demonstrating their new MDX-500 servo-controlled
machine ($21k) which has good precision but a rather low z-height, like
most of their machines. The representative was talking about a new
laser-scanner they'd just introduced from Japan, where Roland is based,
which would scan objects cylindrically using a turntable, but it wasn't on
view there at the show. The Graphitech <www.graphitech.com> folks, (who
make Cimagrafi) had a laser-scanner in operation digitizing relief objects.
This consisted of a laser head on an Align-right router base, which can be
converted back to a router by substituting a Porter-Cable head for the
laser- I'm not sure how much recalibrating it takes to put it back on,
though.

Most of the emphasis at this show was on Coordinate Measuring Machines,
rather than scanners, although one scanner maker, Capture 3d
<www.capture3d.com> was demonstrating how it made a test rocket-sled for
NASA with its dual-optical machine, which worked by projecting black bars
on the surface, then calculating 3d coordinates from the distortion of the
bars. Renishaw <www.renishaw.com> and Faro <www.faro.com> seemed to have
the biggest presence there, Faro selling large arm-like devices that
operate by hand, while Renishaw had the "Cyclone", which works more
automatically. One fellow there from High-Res Inc <www.Reverse-it.com> had
developed software plug-ins for programs like Pro-E, Solidworks, and
MasterCam to support digitizing probes directly (including the Microscribe,
at the low end plus Renishaw and Faro). Another company, Helmel Engineering
<www.helmel.com>, had a stand-alone software product called Geomet for
recognizing features on physical parts, and constructing solid geometry in
the computer- the rep there said he could get me a discount on a Renishaw
probe, but the coolest ones, with swiveling heads, cost $45k (although they
did have devices as low as $2.5k.) Delcam of the UK <www.delcam.com> ,
makers of MillWizard, Powermill, and ArtCam, also had a product in this
category called CopyCad, but it, like the rest, seemed fairly tedious to
use, since one has to reconstruct a part feature by feature, then knit them
all together with surface patches.

There weren't many CAD systems represented; Solidworks <www.solidworks.com>
had the place pretty much to itself. I got a demo of it, and it was
impressive. They had a Nordic-Trak machine (or its generic equivalent) as
an example part: when you pulled the handles, the footpedals would move.
The demo, conducted by a major reseller Hawk Ridge systems
<www.hawkridgesys.com> consisted of rebuilding the exercise machine's front
panel. I tried to throw the salesman for a loop by asking him to trim the
solid shell he created with a random wavy surface (it's hard to tell what a
program will really do if one lets a salesman stick to a canned
demonstration he's done a hundred times or more) but this was no problem.
The program costs $4k, but I could see it making sense for someone who had
to design machine parts on a regular basis. Unfortunately, training- vital
for a program this complex- cost $1200 per day.

Many competing high-end CAM systems were clustered in the same general
area, and I got demos one after another, so now I'm frankly having trouble
sorting them out in my mind. Actually, most of the programs at the high end
were fairly similar, using the same Parasolid kernel, and very similar
parts- with pockets, islands, fillets, and drilled holes - were used in the
demonstrations.

Mastercam <www.mastercam.com> seemed like a very nice program, and it did
some things I'd never seen before, like "morphing" toolpaths, which start
out, in the example, contouring a rectangular pocket, then as they spiral
inwards clearing the pocket changing the contour incrementally to mill
around a triangular island. The program was optimized for high-speed
machining, so interior clearing paths were rounded slightly in the corners,
and would loop around outside corners so as not to lose velocity.

At the Surfcam <www.surfware.com> booth, a guy named Jon Kammerer
<www.jonkammererguitars.com> was picking on his new patented guitar, which
had been milled in maple using Surfcam with all the sound-box's right
angles and flat surfaces replaced by smooth curves. He claims it's just as
loud as a traditional acoustic guitar- it was hard to tell about that at
the show, with all the noise from machinery, but it was definitely more
comfortable to hold. The software seemed capable, but it was hard to tell
what set it apart from the rest, except perhaps that it had the fastest
simulation. The z-level contour roughing for constant climb milling is a
nice feature. It costs $15k with the solid modeling option, and $18k if you
want the 5th axis module. Of course, as is common with programs in this
price-range, one is expected to pay 11% of this price to them every year as
"maintenance", in order to receive support and possible future upgrades.

GibbsCam <www.gibsCAM.com> had the best simulation- it was like watching a
color cartoon of the machining operation, in real time, and you could fly
around the part without freezing the action, just like a video-game.
Virtual Gibbs is less expensive than Surfcam, at $7500 (plus 10%/yr) for
milling capabilities up to 5-axis. It's largely wizard-based, with a "hole
wizard", for instance, that guides one through the steps involved in making
a hole. This may make it easier to grasp than programs based on icons or
nested menus, but it might also be annoying if it takes too long to process
a lot of holes. It optionally comes with a "Solidsurfacer" module that
writes code for milling NURBS directly, provided one has a controller, like
Fanuc and Seimens, that supports the requisite g-codes.

EdgeCAM <www.edgecam.com> offered a mold and die package for between
$8k-$11k, plus 7%/yr. with some advanced features like Constant Cusp
Finishing But if you wanted the Solids Machining package with that, for
importing solid geometry the price goes up $12k. I asked, but they said it
wouldn't find parting lines automatically, which I thought was important in
a molds package.

As it turned out, though, I only found one company there that made that
claim: Vero International. <http://www.vero-software.com>. Their
VISI-series mold design module has been optimized for plastic injection
molding, with standard ejection pins and other parts included in its
library. It also uses the Parasolid kernel (they say they pioneered its use
with surfaces and machining). It comes with a modeling program, rendering
engine, and high-speed machining, but doesn't support a live 5th axis. It
costs $14.1k for the mold software with simultaneous 4-axis capablity plus
the solid modeler; other modules like reverse engineering and 4-axis wire
EDM would be extra.

Esprit, from DP Technology <www.dptechnology.com>, seemed the one of the
more economical of the high-end packages, at $5k for the 3-axis version and
$7k for the 5-axis, although there's probably a yearly fee added. With a
parasolid kernel and NURBS-based output it seemed able to do at least most
of the things the other programs could do, and had a few tricks of its own,
like graphically comparing the simulated product of the machining operation
with the original CAD model and flagging the differences. It incorporates
Visual Basic for Applications in its interface, so one can write macros,
change the toolbars, and generally customize it to ones preferences. In
fact, they are cleverly taking advantage of this to harness their user-base
as developers, providing a place on their site for people to sell any
plug-ins they might make for their own use they feel might be useful to
others.

One company, One CNC, <www.onecnc.com> broke ranks with the rest on price,
as well as on the "maintenance issue", for which they deserve some
commendation. Their $2495 Mill Professional product (that's flat- no yearly
fee) while it doesn't support more than 3 axes, does machine NURBS
directly, and imports solids, surfaces, wireframes, and text, using their
proprietary kernel. They have their own hybrid solids-surfaces modeler, as
well as a 2 1/d package for half as much as the Pro version. They are
willing to sell a network-enabled version for the price of 2 seats, which
is certainly a better deal than buying ten separate licenses. Plus, they
offer competitive upgrades, in case you already have a CAM system.

There were three rapid prototyping systems on view, which took different
approaches to the problem of constructing physical objects from CAD data.
Z-corp <www.zcorp.com> has developed a system that resembles an ink-jet
printer for 3d. A jet head sprays a section of the model on a bed of starch
powder with a glue that consolidates it, then drops the bed a millimeter or
so, rolls out another layer of powder, and repeats the process, quickly
building up a part. Advantages of this approach are that the powder-bed
supports outlying portions of the part geometry, so no external support
structures are required, the removal of which can be a problem. Also, it is
faster than most other RP systems, although the surface smoothness is not
as great. Parts made this way are somewhat fragile, but their strength can
be enhanced by dipping in wax or super-glue. They could also dip them in a
flexible urethane to create flexible, rubbery parts. Another improvement
they've introduced since the last time I saw one of these was the ability,
in the model Z402C, to introduce integral color into the part by using
colored glue/ink, which they say is controlled by assigning colors in the
Cad program, much like conventional printing.

Another company, Stratasys, <www.stratasys.com> took a different approach
to the problem, using Fused Deposition Modeling (FDM) and their Maxum
machine to deposit ABS thermoplastic in superimposed layers, building up
forms that were much stronger than other sorts of RP objects. The size
envelope was also quite large at 600 x 500 x 600mm, and they had some
sizable engine blocks and manifolds on display done using this technique.
Most fascinating were the gear assemblies they had produced this way, with
the whole gear train, axles, enclosure, and all, built simultaneously. When
the soluble support material was removed, all the gears turned freely and
transferred motion from one to another.

The best surfaces of any RP process I've seen so far were produced by the
Solidscape <www.solid-scape.com> PatternMaster machine. This is an
development of the Sanders ModelMaker, under new corporate ownership, and
improved in its functionality. Although the build envelope was smaller than
Stratasys' at 12" x 6" x 8.5", it uses a waxy plastic material that burns
out easily, making it an excellent choice for producing lost-wax jewelry
models. They also had a metal mold they'd produced by building up the
inverse of the pattern and casting- the finish on the cavity was very good.
The process involves tracing along the vector contours of a part with their
hard wax, and filling in solid areas using a hollow grid. A soluble-wax
support structure is built simultaneously as a hollow grid as well, which
leaves no trace when it's dissolved away. Between each build-up pass, a
horizontal milling head moves across the part, planing down the
newly-deposited surface to flat, so any vertical irregularity would not
accumulate.

There was no shortage of components at the show for people like me who
dream about building their own CNC equipment, or retrofitting older
machines. Perhaps most impressive were the linear motor stages being
offered by various companies, most notably HiWin of San Jose.
<www.hiwin.com>. Instead of the conventional arrangement where a stage runs
on a linear guideway propelled by an encoder/motor attached to a ballscrew,
their new systems use linear motors, which are like motors with the
armature "unrolled", so instead of having to rotate and transfer rotary
motion to linear, the motion is directly linear from the start. They also
figured out a way to incorporate a magnetic linear encoder into a guideway/
truck system, so position is accurately and directly fed back to the
controller. Of course, these things aren't exactly cheap, so for the
"financially challenged" of us they offered a range of solutions, including
a selection of more conventional ballscrew-based linear slides.

Although I felt they were the stand-out, there were a number of other
companies offering similar items, like Lintech <www.lintechmotion.com>, and
INA <www.ina.com> who said they could probably come up with the obscure
metric fixed and angular-contact bearing sets to fit the surplus ballscrews
I've been collecting for a router project.

PC-based controls were hot, and a number of different systems were on
offer, from well-known brands like Centroid <www.centroidcnc.com> and
Milltronics <www.milltronics.net> to ones I'd never heard of before, like
Acroloop <www.acroloop.com>, MS-Tech <www.ms-tech,com>, Delta Tau
<www.deltatau.com>, and Omni-Turn <www.omniturn.com>. For some reason,
though, they didn't want to compete on price- it seemed like everybody was
trying to get at least $10k for a controller.

I could go on and on, I suppose (maybe I have already) -especially if I
knew the differences between the insert-type cutting tools that were common
as dirt there- but I think I've touched on the things that struck me as
most interesting. If anyone has questions about areas I may not have
mentioned, fire away- like I said, everybody notices different things, and
there was a lot I jsut had to ignore. But altogether, it was a busy three
days...

Andrew Werby
www.computersculpture.com


Andrew Werby - United Artworks
Sculpture, Jewelry, and Other Art Stuff
http://unitedartworks.com