Communications of the ACM

3d hard copy

Rapid Prototyping in Consumer Product Design

By Paul K. Wright
Communications of the ACM, June 2005, Vol. 48 No. 6, Pages 36-41
10.1145/1064830.1064856
Comments

Why are only certain consumer products the hot gifts to give and to get each holiday season? Last year the digital camera and the iPod portable music player (and its MP3 clones) were perhaps the leading contenders. One part of the answer is they are deliberately designed to deliver a high-quality consumer experience. Or as my colleague in Japan, Noriaki Kano of the University of Tokyo, has said about the best products, they "surprise and delight" the first wave of consumers to use them. The next part of the economic cycle is that the products are praised in design magazines and by word-of-mouth; before long they are must-have items for perhaps millions worldwide.

Figure

Rapid prototyping plays a critical role in their development. Somewhere in some back room are probably dozens of weirder-looking casings for cameras and MP3 players that did not feel quite right in the hands of focus-group participants or the earliest consumers, look right, or simply revealed to a design team that the inner electronics did not fit correctly in the external casing. One might think that with today's high-end computer-aided design (CAD) systems and high-resolution display screens, a product can be designed and assembled correctly in virtual space, then communicated to a factory and mass produced at the touch of a button. Indeed, many subsystems of the new Boeing 777 airplane were analyzed and designed virtually [1]. But experience designing, say, consumer electronics, computers, toys, home appliances, and automobiles seems to mediate against this virtual production facility for consumer products. In the Berkeley Manufacturing Institute's design work with electronics companies (such as Intel and Ford Motor Company), we create a succession of prototypes reflecting seven surprise-and-delight factors:

• Consumer in mind. Product development today always keeps the consumer in mind [2]. In focus groups, ideation groups, and ethnography studies, the invited participants prefer to see and feel real products, not just look at computer-generated images. Even though these evaluations rely on gut reactions, they are still a good way to evaluate and determine whether an emerging product generates a must-have feeling among consumers in its target market;
• Feel of a product. To refine the feel of a product, its ergonomic aspects must be evaluated, including the position and shape of hand grips, buttons, screens, dials, and ports;
• Physical prototypes. A physical prototype then structures the ensuing design interaction among the various types of designers, including electrical, mechanical, and industrial;
• Not through simulation alone. We have found that certain evaluations (such as antenna placement in relation to electromagnetic interference) are impossible to determine solely via simulation;
• Prototypes. Prototypes are the best way to demo an emerging product to an organization's nonengineering units (such as those involved in promotional image development, marketing, and sales);
• Development cycle. The overall development cycle is short. Creating a prototype—or rather a series of prototypes—is critical. Some markets change so quickly that toy makers in particular create two models—one that "works like" and one that "looks like"—before compressing all design considerations and iterations into a single footprint and launching into mass production in late summer. It is no wonder that the International Consumer Electronics Show is held annually the first week of January, usually in Las Vegas. The expectation is that the new gizmos being demoed might take the next 10 months to be fully refined, mass produced, shipped, and placed in stores well ahead of the next holiday season in early November; and
• Changing design. The electronic components themselves are being updated and changed on a regular, perhaps monthly, basis. The footprints of printed circuit boards change even more often. Thus, a quickly regenerated physical prototype of a mechanical casing makes all the diverse interests within an organization comfortable that the product will still be manufacturable in the final mass-production process before being shipped to distribution centers and stores. The figure shows a series of prototypes produced over six months of Intel's Personal Server, a small pager-like device for carrying personal electronic files from computer to computer. The sequence of images in the figure shows the casing designs that emerged, each tweaked to accommodate modifications in its electronic components.

---

In focus groups, ideation groups, and ethnography studies, the invited participants prefer to see and feel real products, not just look at computer-generated images.

---

Nearly a dozen rapid prototyping processes (such as those at www.boedeker.com/sla.htm) are available for producing first-look components. Three are especially relevant to consumer products: stereolithography (SLA), fused deposition modeling (FDM), and 3D printing. SLA was introduced commercially in 1987 by 3D Systems, Inc. (www.3dsystems.com). Any SLA design-to-manufacture process involves three key steps:

• Tessellation. The surfaces of a standard CAD file are tessellated, that is, they are approximated with an overlaying mesh of triangles. Tessellation is like throwing a fishnet stocking over the surfaces, converting them to a sea of ~10,000 triangles. In 1987 3D Systems called this file of triangles the STL file; like many de facto standards it has been used ever since, adopted by most if not all other processes, including FDM;
• Layers. The file of triangles is sent to the prototyping machine and sorted into layers. This is like slicing the original CAD model into a stack of very thin pancakes; and
• Laser curing. The physical process in SLA involves the curing of a photopolymer with a low-power laser. The photocurable liquid, resembling honey in appearance and consistency, is kept in a vat where the laser begins scanning the top surface. The scanning cures an initial layer that resembles ice forming on a pond. The layer rests on a mechanical elevator-platform just below the surface where it supports the first delicate layer. When the first layer is cured, the elevator jogs down a few tenths of a millimeter, liquid flows over it, and the laser begins working on the second layer that (with the correct controls) fuses to the first layer. The process is repeated many times over; layers accumulate over several hours until the object is formed. The elevator then rises, and like Excalibur's sword, the object emerges fully formed out of the "honey." Curing and hand finishing are subsequently needed and applied.

SLA retains a preeminent position in the rapid prototyping family because it is still the most accurate, though less so than a standard computer numerical controlled (CNC) milling machine. High-resolution laser-based SLA can manufacture parts with tolerances of from +/-0.002 inches to 0.003 inches (www.boedeker.com/sla.htm). SLA models are used as the master in the processes described later. So why was there a need to invent and use FDM and 3D printing? The simple answer is they are progressively cheaper and faster but involve a trade-off in terms of accuracy. 3D printing is fast enough to produce two or three prototypes the same afternoon for ideation groups. "Fail often fast, then do it right" is the mantra at IDEO (www.ideo.com), a leading consumer products design company. FDM usually requires an overnight run. Both FDM and 3D printing can be run by inexperienced students and other novice operators, needing no careful calibration or expensive photocurable liquid.

Both these processes use the same file format as SLA. FDM creates the layers with the equivalent of hot toothpaste extrusion of plastic or a cake-icing tool extruding hot plastic in rows (or roads) with a super-fine point. 3D printing is like a Xerox machine squirting epoxy resin layer-by-layer and row-by-row onto a ceramic powder resembling cornstarch.

Beyond the methods needed to create a single prototype it is worthwhile to review the scalability to more prototypes and to mass production. Although rapid prototyping was introduced in 1987, regular prototyping is hardly new. Five thousand years ago, to get the attention of some well-to-do patron of the arts or industrial baron—then, as now, the ultimate consumers—an early artisan from, say, Egypt or Korea would have used the lost-wax process to create one or more prototypes. Over a period of several days the artisan would laboriously carve wax models of, say, jewelry, ornamental tableware, or small statues. Sand or clay would then be hand-packed around the model. The wax would be melted out through a small drainage hole in the bottom to leave a hollow core. The small hole would be blocked, and molten metal—silver, gold, bronze, tin, or pewter—would be poured into the core through a wider hole at the top. After the metal had cooled and solidified, the prototype would be broken out of the sand, cleaned up, polished, and displayed for (it is hoped) purchase by the patron.

If it turned out to be a must-have item, the artisan might repeat the process. But perhaps the artisan would also invent shortcuts to speed the carving of the disposable wax models or simplify the original design to reduce the time-consuming carving phase. Thus, a few years later, the object might find its way into the equivalent of a mass market, in perhaps modified form.

In principle, little has changed. Today's modern computer-aided prototyping shops, receiving CAD files over the Internet, still produce small batches of aluminum molds for well-known consumer-product companies producing everything from video game controllers to water bottles to hand tools through a modified casting process in which the wax original is replaced with a prototype generated through SLA. Prototype engine parts for automobiles, tractor transmission parts, and medical devices are also commonly made in prototyping shops.

Unlike an early artisan's shop, where the expensive artwork/master is destroyed, today's prototyping shop employs a five-step process:

• Male master. A male master, or the same shape as the final or desired part, is made through SLA;
• Female master shell. An intermediate female master shell, or the part's inverted geometry, is created in hard, stable epoxy resin around the SLA part. Care is taken to maintain good geometry between the masters; the hardened epoxy female master is polished and corrected, as needed, for small errors in geometry and tolerance;
• Rubbery slurry. A rubbery slurry is poured into and around this intermediate master to create many reusable rubbery male submasters with cores;
• Many female molds. Many female molds are then created by pouring an aluminosilicate slurry into the rubbery molds. When dry, they are brittle and able to withstand the high temperatures of casting; and
• Final parts. Molten aluminum is poured into these silicate molds; after cooling, the destroyable molds are broken apart to reveal the final parts, still in need of hand grinding and polishing.

Plastic injection molding is an alternative manufacturing process, specially suited to very high-volume production. In it, a mold is made from aluminum (for short runs up to about 500 components) and/or hardened steel (for runs into the many millions of components). The aluminum molds can be cast as described earlier. Steel molds must be carved out on a CNC milling machine. The millions of components are created by shooting hot plastic into the central core of the molds—a challenging art form involving many design considerations in the original part. So it is again useful to first create an SLA or FDM prototype and show it to injection molding specialists for a series of evaluations involving three critical questions [3]:

• Draft. Do the sides of the part have a "draft" angle, or sloping sides (like a bucket)? It needs to be only a few degrees, but without it plastic parts will stick to the mold and be almost impossible to get free;
• Re-entrant surfaces. Is the part designed so thoughtlessly that its re-entrant surface requires that undercuts be created in the mold? Re-entrant surfaces of a product casing become locked in the mold wall after the plastic is shot into it, until a spring-loaded slider is activated for part removal, an unwanted added cost; and
• Thicknesses and bosses. Are the wall thicknesses and screw bosses designed so as to avoid unattractive sinkage in the external surfaces?

War stories abound of consumer products and toys that reach a mold maker (most likely in Taiwan) in mid-August without these questions being addressed. As a result, the product or toy is not ready for the holiday buying season beginning in November.

---

Physical prototypes allow the evaluation of subassembly fit, system aesthetics, and overall quality before taking on the huge and high-risk cost of manufacturing.

---

Back to Top

Conclusion

For consumer electronics, PC devices, household appliances, and toys, creating physical prototypes is central to survival of the fittest. In the early planning stage of a new or revised product, physical prototypes allow the evaluation of subassembly fit, system aesthetics, and overall quality before taking on the huge and high-risk cost of manufacturing. The physical prototype guides the design process, enabling ideation and focus groups to get a feel for the real product. Later on, the prototype can be used to help coordinate suppliers. Rapid prototyping has thus become, and will remain, an engineering tool, as well as a psychological tool, in the orchestration of product development and supply chain management.

Back to Top

Back to Top

Back to Top

Back to Top

Back to top

The Digital Library is published by the Association for Computing Machinery. Copyright © 2005 ACM, Inc.

---

No entries found