Model early, model often, and pass the prototype around.
Paul Dvorak, Senior Editor
Despite the lifelike renderings of 3D CAD models, there are intangibles in designs that cannot be conveyed through a computer monitor. Just looking at an ergonomic handgrip, for example, cannot convey whether it's too big or too small, or ergonomic at all. That can only comes from handling one.
Hands on examinations and other advantages of rapid prototyping have been well reported. What's not well known is exactly what designers should do with RP parts. After picking it up and showing it to other designers, what comes next? Here's what a few RP pioneers had to say.
PART OF THE PROCESS
“First of all, engineers should make rapid prototyping a part of the design process,” says Ray Sander, head of the model lab at Battelle, Columbus, Ohio. “Designers are so good with solid-modeling systems that they whip up models fast sometimes a little too fast. If they are modeling a handgrip with switches and buttons, for example, they might shell out the parts, add ribs and bosses, and not even know if it fits a person's hand. To break this habit, we encourage modeling early and often,” he adds.
The best time to make a physical part for review depends on the engineer's experience and the part's complexity. However, late afternoon is often good because RP machines can be set to run at night. “Our engineers like to see their day's work on the following morning,” says Gary Steinberg, CAD and R&D records manager at Bissell Homecare Inc., Grand Rapids, Mich. “So when they come in, they can collect opinions on the design and make corrections before getting too far along.
Others agree. A 16-part insulin pen, for example, needed lots of internal details to ratchet properly. “The solid modeler could tell if we had interference in the assembly, but first you had to put it together on screen, and that was a brain teaser,” says Battelle's Sander. So at night during the project, new, slightly changed parts were built to maintain the mechanical designer's sanity.
“Actually, needs and concerns should tell when to prototype,” says Sander. “Generally, once we get CAD geometry, we build it. Then at meetings, we can put models into hands and let people do different things with the parts. That encourages the type of thinking that does not come from on-screen models.”
Prototypes should also force designers and engineers to resolve issues that might be easy to gloss over in computer models, suggests Michael Plesh, director, Technical Skill Center for the Art Center College of Design in Pasadena, Calif. “There are dozens of issues that usually aren't addressed until after the first prototype,” he says. “For example, how do pieces fit together? How will it be used? Does it work the way it is intended to? Does it have the right feel? Answers to these questions and others don't come without prototypes.”
You don't have to make the whole model. Cut out the area of interest and produce that. “One recent RP item had an estimated build time of 6 hours,” says Sander. “So the engineer put a window around the area of interest, cut it from the digital model, and built that in 45 minutes. This let the engineers sit with electrical people and others to check-fit batteries, a circuit board, and so on. Discussions got going early in the design process when the CAD model was easily changed.”
Don't worry too much about costs. “In the year before we brought RP machines in house, our engineers requested funds for 40 models,” says Jerry Lowe, a manager at Mitsubishi Digital Electronics America, Irvine, Calif. That's understandable because the service bureau they were using charged about $1,200/part and up. “But in the first six months of having machines in house, we built over 450 parts.”
Sander has noticed that engineers often want to compare SLA concept models against other RP technology. For instance, they compare surfaces because those on SLA models are excellent. That's ok. “But speed, not surface finish, should be driving the issue,” he says. “If models cost $300 or $400, regardless of surface finish, there will be few of them. A $300 part from a stereolithographic machine costs $12 from a 3D Printer, and maybe $17 to 25 in a different process,” adds Sander. When parts are so cheap it doesn't matter when they are made.
“We tried use-rating the 3D printer for a few years,” says Sander. “Accountants put a cost on the machine's time and we charged for each model. But we eventually realized it was so inexpensive, yet so valuable, that it was not worth the time to calculate costs. When the new year came up, accountants added a little to the hourly billing rate. But the RP machine is now treated as a color printer and its parts are part of our development process.
Use RP parts to provoke thought and improve collaboration. This is especially true for those without a technical background marketing and sales people. They get a lot more from a full-sized, well-painted prototype than by looking at a detailed isometric view on a computer. Some have a hard time seeing what's on screen. A part the size of a golf ball might look like a basketball. A real part shows physical size.
“People even draw on parts while asking, what do you think about this or that,” says Sander. “In fact, the parts become a 3D doodle.”
Mydea Inc., Orlando, Fla., developed a prototype for a surgical instrument that lets orthopedic surgeons manipulate and cut sutures during surgery. “Engineers at the medical company could immediately examine the tool and meet with surgeons for design feedback,” says said Michael Siemer, Mydea CEO and founder.
“The prototype was also used as a communication tool with plastic-injection molders to determine design changes for mass production,” he says. “They visualized the tool based on sketches and drawings, but even the best vision isn't as good as holding the object in your hands and seeing how it functions,” he explained.
Use RP parts to improve manufacturability. “Manufacturing problems that are less obvious on-screen come into sharp focus while examining a physical model,” says Plesh. “ Solidmodeling software generates almost any shape, but it may not be possible to manufacture or cost effective to make. RP parts should force engineers to think through different manufacturing steps and may spur design changes, even subtle ones, that make the final part easier and less expensive to build.”
Thick and thin walls on plastic parts, for example, are sometimes tough to spot on-screen, suggests Steinberg. Thick walls tend to warp or form sinks, and thin walls might be difficult to fill. “Thin sections are particularly easy to miss. But prototypes makes it quite clear because the wall is so flimsy.”
Make a prototype that can be used as a pattern. “For example,” says Plesh, “I built a model as a visual aid for a company's marketing group. They liked the model so much they ended up using it as a master. Production molds were created directly from it.
Build prototypes with moving mechanisms. The design that looks like a dream on-screen can be a nightmare on the assembly line, so check the design for fits and interferences. Building an assembly would tell whether or not there is wrench clearance, that special tools are needed to maneuver a small part into place, or it might suddenly become obvious that two separate parts work better as one. Combining parts is a cost-cutting tactic.
“Engineers might put 0.01-in. clearances on a couple parts that look fine on-screen, but you appreciate whether the clearance is enough or not when the part is in your hand,” says Steinberg.
In addition, when devices combine mechanical and electrical functions, include a mock circuit board, and check that the boss stand-offs are correct, and that the board does not interfere with the structure. “We might even put a 9-V battery shape on the circuit board to check for clearance,” says Sander. “And let the electrical engineers check it out for their concerns before the mechanical part gets too far along.”
But don't solid modelers check for interferences? “Most do. The CAD system here requires users to execute the function it's not automatic,” says Steinberg. “And then they have to understand results. Also, the modeler checks for interference only in one position. If parts rotate about one another, the software can trace an arc or envelope of motion, but the designer has to look for interferences or collisions.” He suggests using the mechanisms or kinematic features in modeling software to initially check for the model's range of motion.
Then build the assembly to make sure things fit, adds Lowe. “If a TV, one of our products, has a foldout door, we'll build an RP of it to make sure hinges are in the right place for aesthetics,-there are no gross interferences, and should the user press against the door, it does not flex too much or unintentionally come open. It should feel sufficiently solid. The assembly gives us the look, feel, and tolerances of the final product. And then production people get familiar with challenges the assembly might present, and a chance to voice their opinion.”
There are a few tricks to building RP assemblies adds Sander. “For parts that will be assembled, model makers might enlarge a hole and epoxy in a threaded fastener because most RP materials are not strong enough to press-fit other parts. Occasionally, too much line-toline interference keeps two RP parts from mating. Model makers, however, can tweak the CAD model for one part making is slightly smaller, then build it without saving the changes,” he says.
Don't ignore advantages of other RP methods such as carving Styrofoam, shaping cardboard, and using routing machines. “Early on, it's important to get something physical in your hand,” says Siemer. And it's just as useful to use cheap materials such as clay, play dough, Legos, gator board, whatever is handy. You might even pick up a similar part that costs a few dollars at the local store. Consider any type of physical modeling, not just RP.”
3D Systems Inc.,
Battelle Memorial Institute,
Roland DGA Corp.,
Z-Corp., (781) 852-5007,
THE RP LINE UP
Rapid prototyping now encompasses about a half-dozen different technologies. However, preprocessing solid models remains much the same for each. That is, RP software slices models into paper-thin sections that the machine can build layer-by-layer using a variety of materials. Companies may sell several different build technologies. Here's a thumbnail sketch of the most frequently encountered methods.
3D printing uses ink-jet heads to deposit a liquid binder that fuses powder into a required shape. A mechanism then lowers a build platform, making room for the next layer, and the process repeats. Finished parts are surrounded and supported by powder, which is shaken loose. Layer thickness is on the order of 0.1 mm. The process, one of the fastest, builds parts with a slightly grainy surface. Materials include a ceramic powder and a starch-based powder that is not as strong, but can be burned out for investment-casting applications. Three-dimension printing comes from Z-Corp., Burlington, Mass.
Fused-deposition modeling from Stratasys, Eden Prairie, Minn., builds movable 3D models from the bottom up, one layer at a time, with acrylnitrile butadiene-styrene (ABS) plastic. Accompanying software orients parts and creates necessary support structures. The software also plots a deposition path for the machine. ABS (in filament form and auto-loading cartridges) is fed into an extrusion head, heated to a semiliquid state, and deposited in layers as fine as 0.010-in. thick. The company makes several FDM machines ranging from fast concept modelers to slower, higher precision machines. Materials also include an elastomer (96 Durometer), polycarbonate, polyphenolsulfone, and investment-casting wax.
Multijet modeling is capable of depositing 600 300 dpi in layers down to 16 microns (0.0006 in.). The jetting head slides back and forth along the X axis depositing a single-layer of photopolymer onto a build tray. UV light cures and hardens each layer. The jetting block uses eight heads. Surfaces are said to be smooth and even. The machines build models with one material and use another, a gellike photopolymer, for support. The PolyJet process comes from Objet Geometries in Israel but is marketed by Stratasys Inc.
Stereolithography builds using a combination of laser, photochemistry, and software. An ultraviolet laser focuses on the surface of a vat of liquid photopolymer. The laser traces a cross section of a part, turning a thin layer of liquid plastic to solid. The cross section is lowered and recoated with liquid photopolymer and the laser traces the next slice atop the previous one. SLA machines come from 3D Systems, Valencia, Calif.
Selective layer sintering uses a laser beam to selectively fuse powdered materials, such as nylon, elastomer, even metals, into solid objects. Parts are built upon a platform which sits just below the surface in a bin of the heat-fused powder. A laser sinters the pattern of each layer onto the previous. SLS machines also come from 3D Systems.
Subtractive technology includes routers and milling machines that use CAM toolpaths to guide a cutter through a block of material. Manufacturers include Techno Inc., New Hyde Park, N.Y., and Roland DPA, Irvine, Calif.