Rapid prototyping moves toward rapid tooling.

No longer just for making prototype parts, 'rapid manufacturing' systems offer promising shortcuts to production tooling

Imagine if moldmaking were this easy: First, your part concept takes shape on a CAD/CAM system's three-dimensional solid modeler. Then the CAD program employs asoftware module to automatically create the resulting model's negative--essentially a digital version of the mold cavity once a parting plane is chosen. This computer-bound "mold" is then downloaded into a rapid manufacturing system, which builds a metal-casting pattern for the mold-cavity inserts in a matter of hours. Working with a foundry over the next few weeks, you have the cores and cavities cast in tool steel. Finally, you leave any finish-machining work to your trusted tool and die maker. From beginning to end,the injection mold takes less than eight weeks to create, instead of the 20 you might otherwise expect. Even better, the project costs only half of what you'd pay to cut steel the old-fashioned way.

If all this unbridled technological optimism sounds farfetched, think again. Right now, Ford Motor Co. is using an injection mold made this way and has two more in the works. Other auto companies also have begun "rapid-tooling" projects, as have some leading medical and electronics firms. To be sure, rapid-tooling methods still need plenty of fine-tuning and may never fit all applications. But so-called "desktop manufacturing" or "automated fabrication" systems--such as stereolithography (SLA) and selective laser sintering (SLS)--are evolving from their origins as automated model makers. Thanks to somerecent technical advances, these systems have entered the earliest stages

of    their most promising application--production tooling for parts that will
be     released into the field. "The long-term view is that rapid prototyping

will onlybe big if it finally addresses this manufacturing issue because the demand for prototypes pales in comparison with that for manufactured parts," says Dr. Paul Jacobs, research director at SLA supplier 3D Systems.

Yet even when the rapidly produced parts and molds remain in the low-volume class, desktop manufacturing systems offer benefits overlooked by many molders. "Injection molding operations are still an untapped market," says

Todd Grimm,   manager of Laser Prototypes Inc. This five-year-old
stereolithography service   bureau does more than 80% of its business in
plastics, but counts few molders   among its customers. The benefits from
"rapid tooling," however, can be         compelling for any molder who has a

stake in cutting tooling costs and helping customers through the mold-design process. Prototype and limited-run tools made with desktop manufacturing systems can be a way to achieve these world-class goals.

FORD IS FIRST

The growing network of rapid-prototyping service bureaus--or firms that hire outtheir prototyping systems and engineering expertise--may soon bring some rapid-tooling capabilities to companies far from the Fortune 500. For now, the truly leading-edge efforts remain the province of larger companies. According to3D's Jacobs, Ford Motor Co. is the first and only one so far to investment-cast injection mold tooling from stereolithography patterns.

At only five years old, SLA is already the most established rapid-prototyping system. It builds parts from layers of laser-cured photopolymer. Investment casting, meanwhile, is an ancient process. Using an expendable pattern as a guide, investment casting creates a ceramic shell into which molten metal is poured to form the final part. At Ford, the two technologies were coupled in Mayto make the tooling for an injection molded wiper cover that will go on one of the company's 1994 models. According to Karl Denton, engineering technologist inthe company's bumper and exterior plastics group, the mold measures about 6 x 8 x 6 in. for the core half and about the same for the cavity. Cast in A2 tool steel, each half weighs about 85 lb.

Before making the molds, Denton first produced prototype parts on Ford's SLA-250machine from 3D Systems. After three tries, the company was "satisfied with the geometry," Denton reports. Then, ProEngineer and ProMold software from

Parametric Technology in Waltham, Mass., were used to create computer files--in a format known as ".STL"--that describe the mold-cavity inserts. These files were fed into the SLA-250, which built the parts in

3D's "QuickCast," a new SLA "build style" based on a lattice structure and designed specifically for investment-casting patterns. A local foundry,

Howmet, cast the cavity inserts inA2 tool steel after building a ceramic shell around the QuickCast pattern. While the jury is still out on cast injection tool life, early indications are that the cast steel is harder than its machined counterpart. Perhaps it's even too hard--Denton reports that Ford's tool-and-die shop broke five carbide-tippedtools before making even 1/8 in. of progress in drilling out

water lines and    ejector channels on the cast inserts. "The cast steel was
incredibly hard,"     Denton recalls, "so the finish work was labor-intensive
and ate up the most     time." Other than the drilling and bench work, the

tool required only 30-40 hr of design and SLA work to make the casting pattern and another three weeks of foundry time.

Unfortunately, the surface finish imparted by the SLA pattern to the cast cavityinserts left the first round of molded wiper covers unacceptable for

release.   For the final version of the wiper cover, Ford nevertheless plans
to make the   tooling the same way--through SLA QuickCast and investment

casting. This time, Denton says, the surface-finish problem will be addressed at the pattern stage, either with some sort of coating or a light sandblasting. On the positive side, Denton says, the tool took only 10 weeks to complete rather than the 17 or more for similar tools made by conventional methods. This

"rapid" mold cost a little more than half as much as a conventional tool, with much of the price tag covering finish work that can be avoided on subsequent

efforts. In fact, Denton expects the next rapid-tooling efforts to cost only a quarter of the price of conventional tooling. "Rapid tooling will be an

important way of making molds due to the low turnaround time and cost savings," Denton says. "It doesn't make sense not to do it this way." Beyond speed and cost reductions, rapid tooling may offer other advantages. 3D'sJacobs points out that since cast steel is hard all the way through, it may allow moldmakers to bypass traditional hardening steps. And mold designers armedwith thermal-analysis software may be able to optimize cooling

beyond current   capabilities; cast-in cooling lines could take almost any
shape to conform to   the contours of a cavity, unlike the usual rectilinear

gun-drilled layouts. Besides using SLA, Ford will also make tools for the wiper cover from the selective laser sintering (SLS) process from DTM Corp. This process builds partslayer by layer from laser-sintered polycarbonate or nylon powders, or from investment-casting wax. Denton believes the porous finish of a sintered pattern may yield the desired texture for the wiper cover. And unlike stereolithography,SLS fabrications do not require support structures that must be removed by hand. For comparison's sake, Ford will also cut a tool for the wiper cover using conventional methods. A second rapid-tooling project in the works at Ford involves a more complex part with a sliding piece.

TECHNOLOGY ADVANCES

Today's opportunities for quick tooling have come on the heels of some technological advances in rapid-prototyping materials, hardware and software.

DTM, for instance, has developed a way to sinter metal powders in its SLS machine, opening the door to nearly instantaneous tooling. DTM's RapidTool process first produces parts, such as a cavity and core set, from a laser-sintered matrix of powdered metal in a polymer binder. Once removed from the SLS platform, the part is run through an oven to burn away the polymer component. A second oven cycle then clears the way for copper to be wicked into the voids left by the polymer, creating a metal alloy. At the recent SME AutoFact show in Chicago, DTM demonstrated injection molding of businesscard holders from a steel/copper tool. According to DTM marketing director Dan Girouard, the metal-powder technology will be commercially available by mid-year. In the meantime, DTM continues to experiment with different metals before picking a winner for the commercial release.

Stereolithography, too, has made technical strides over the past few months. 3D's Jacobs points to three key "enablers" that have allowed the SLA to break into production tooling for the first time: better accuracy, more dimensional stability, and the ability to create "negatives" of part models automatically ona CAD/CAM system. "These are the prerequisites of any rapid tooling," he says. The first two of these enabling factors got a big shot in the arm in July with the introduction of Ciba-Geigy Corp.'s new photocurable epoxy resin, designated XB5170. According to Jacobs, it allows parts as large as the SLA platform to be made to tolerances as tight as |+ or -~100 microns. The new material's effectivelinear shrinkage is 0.06%, while earlier epoxy resins, by contrast, often shrankas much as 0.4-0.8%. "As a consequence of this reduced

shrinkage, the epoxy     shows over an order of magnitude better dimensional
stability than previous     acrylate resins," Jacobs explains. The Ciba epoxy
also imparts structural       strength to the QuickCast lattice. And the

resin's low viscosity speeds drainageof uncured resin from the pattern, creating the void spaces crucial to a good investment casting.

On the hardware side, 3D has come out with upgrades to its process chamber for the SLA-250 and SLA-500 models. These include a new solid-state, laser-diode leveling device for a three-fold accuracy improvement over the old style leveling system.

Another upgrade addresses an earlier problem that affected the system's ability to produce sharp corners. Even when checked by scanning-speed

reductions or a   mechanical shutter, laser scanners would typically suffer
from errors when      trying to follow sharp corners. The result, says Jacobs,

would be overly thick corner regions from scanning-speed reductions or corners with an outer "tag" of polymer hardened during the time lag associated with shutter actuation. The new solution, an "Acousto-Optic Modulator (AOM)," is essentially a one-microsecond on/off switch for the SLA-500's laser beam. Now, the laser beam travels through the AOM solid-state crystal. Just before any sharp corners, an electrical pulse changes the crystal's index of refraction, deflecting the beam away from the resin and stopping exposure about 1000 times faster than mechanical shutters. Jacobs says the combination of the new resin and hardware upgrades have improved surface finish on up-facing planes by a factor of 20. NOT THERE YET

Despite all these recent strides, not even rapid-tooling's proponents claim it can completely replace conventional toolmaking methods. "We're not going to takeanyone's job away," Jacobs says. Bench work on rapid molds will always be neededto obtain the best surface finish. And Jacobs explains that all parts

from      layered fabrication systems suffer from a "stair-stepping effect" on
ramped     surfaces--though these have been minimized as material developments

drive down layer thicknesses. "Removing that last little bit of steel is

where the         toolmaker's real skill comes in," says Jacobs. "The
difference with rapid       tooling is that you start at that last cut."

DTM's Girouard is similarly reluctant to characterize RapidTool or investment casting from rapid-prototyping patterns as universal replacements for conventional methods, in part because rapid-prototyping systems do best in applications with "more forgiving" accuracy requirements. "We're not going to beas accurate as EDM or surface grinding," he says. "We're not so pretentious as to say we can fill those shoes--we can't." Yet Girouard does see rapid tooling as appropriate for parts without excessively tight tolerances--he

places the    current accuracy range for rapid-prototyping systems at
0.004-0.014 in.,        depending on part geometry and on how one defines

accuracy. "That may cover the requirements for half the molds ever created,"

Girouard says, listing           automotive, housewares and toy applications
as promising markets. "These may    have only a few tight tolerances out of a

hundred." And Girouard points out that simply because you can make tools good enough for commercial parts does not answer the question of tool life. He says some DTM customers in the medical and business-equipment industries use SLS to make "low-volume molds" for around 100,000 pieces. Girouard hesitates to call these sorts of efforts production tooling. DTM plans for its second generation of RapidTool powdered metals to include a metal alloy that can be hardened.

OTHER EFFORTS

Though 3D and DTM appear to be leading the charge, their systems may not be the only ones suited to rapid tooling. Cubital also reports investment casting efforts based on its Solider solid-ground curing technology, which builds layered parts from photopolymer. Curing on the Solider is controlled by a mask that defines each cross section.

Stratsys' Fused Deposition Modeling, an extrusion-like process that deposits resins in patterns corresponding to each part layer, has also been used for investment casting. One of the company's newer service bureaus, Proteus Inc., has chalked up the bulk of its business from making patterns for casting. "FDM doesn't fit every application," Proteus president Jack Herrington says. "But it works well in investment casting." For other rapid-prototyping projects,

this   service bureau has farmed out work to "partner" firms with SLAs or
Cubital      machines. Herrington also reports activity in making low-volume
epoxy tooling   from FDM patterns.

Still another promising effort comes from Soligen Inc. Its droplet deposition method builds layered prototypes from a powdered bed by selectively applying an adhesive that turns each cross section into a composite-like material. For really rapid tooling efforts, the machine can make ceramic shells directly, eliminating the need for patterns. At this point, the second generation of Soligen machines is in beta testing at a handful of sites. The company plans to launch a fully commercial version by mid-1994.

Finally, don't count out the service bureaus' role in making rapid tooling available. (For a list of rapid-prototyping service bureaus, see PT, Sept. '93, p. 17.) Many are already involved in making low-volume tooling from epoxy,

silicone and aluminum. 3D's Jacobs notes that several SLA service bureaus have already expressed interest in the QuickCast technology and the process-chamber hardware upgrades. "By the end of 1994, rapid tooling will be more reality than rarity," Jacobs predicts.

Check the Fit, Save a Mold

Even when rapid prototyping is kept to its more "traditional" role of fabricating single parts or patterns for low-volume tools, it still can help molders meet some world-class competitive goals--such as helping customers get their products to market quickly and cutting steel right the first time to save tooling costs.

A recent project at service bureau Laser Prototypes Inc. (LPI) demonstrates how rapid prototyping expedited a molding project. To come up with multiple plastic parts for a new pen-based data-collection computer, Telxon Corp. of Akron, Ohio,authorized LPI to employ four different rapid-tooling techniques and molding subcontractors. This effective husbandry of rapid-prototyping resources helped Telxon go from concept to 50 fully functional devices in only six weeks--and that time period included all engineering and design work, creation of CAD part models, and even mold-filling analysis for the prototype tools. Of the 12 x 8 x 4 in. computer's 20 plastic parts, 12 relatively simple ones went directly to production tooling. For these components, an SLA was used to provide quick "form, fit and function" part prototypes that not only provided opportunities for tweaking the design but also gave Telxon's marketers an early look at the product.

For the remaining key components, 50 copies of each were molded by the subcontractors, with LPI assigning different molding techniques based on material requirements and lead times. The device's "ultra-complex" upper and lower housings went into epoxy tooling after mold-filling analysis determined gating position. A 300-ton press then injection molded prototype parts in the same ABS/PC material that would be used for the release parts. Epoxy molds similarly produced the unit's battery covers.

The battery housing, meanwhile, was molded off-press with the help of a gun-likedevice to inject the plastic into an epoxy tool. This method was chosen due to time constraints--the 300-ton press was already busy making the battery covers. Yet another process, "thin-walled RIM" turned out the computer's accessory tray.The only other choice here, according to LPI manager Todd Grimm, would have beencasting into silicone rubber tooling. Grimm adds

that this method could only    produce two parts a day, sacrificing speed or
pushing up costs by relying on    multiple tools. By contrast, thin-walled
RIM, which takes place under vacuum    conditions, can make about 30 parts per
day in a single tool, Grimm says. Two   final components, a handle grip and a

complex gasket needing rubbery properties,went to conventional urethane casting. All the parts were made in five weeks, leaving the final week for assembly. Meanwhile, Xerox Corp.'s Modeling Shop has saved its internal and external customers from countless tooling errors through the use of SLA prototypes to check form, fit and function. Xerox model maker Rob Kendall recalls one customerthat ordered a prototype made of a copier-related electronic device

while the   mold was already being cut. Using that prototype to check the fit
of the        device's control buttons, the customer discovered an incorrect
draft that       prevented a proper fit. Luckily, the design error was caught

in time to save thecustomer an estimated $10,000 in retooling costs. "Once a mistake is in the toolit's difficult to get it out," says rapid prototyping manager Gene Grenek. "So we can really be an integral part of the tooling process." First, Do Your Homework

When it comes to evaluating the accuracy of different rapid-prototyping technologies, there's little agreement among suppliers on how to cite the numbers. "People can claim any accuracy they want," says 3D Systems' research director Paul Jacobs. To protect your investment when buying a rapid-prototypingsystem or hiring a service bureau, Jacobs strongly suggests that prospective users do the type of homework that should be familiar to anyone experienced in statistical process control (SPC).

First, demand part measurements--performed independently on a coordinate measuring machine (CMM). "Don't only try before you buy, measure before you buy," he says. Then, assure yourself that the rapid-prototyping system can provide acceptable repeatability. "Don't believe one isolated measurement is anyindication of accuracy," Jacobs adds. "Anybody can get lucky once." To validate repeatability, customers should ask to see error-distribution functions derived from statistically significant amounts of data. For example,

Jacobs can offer over 10,000 CMM measurements taken from many copies of a standardized SLA part. From that data, the company specifies SLA accuracy as |+ or -~100 microns at E90, or about 1.65 sigma if the data exhibit a normal distribution. That "E90" means that 90% of the measurements fall within 100 microns of specification.