This article discusses it has become a basic assumption that prices do not rise but instead they should fall in the economics of electronic products. It was precisely to meet customers’ expectations of lower prices that a Canadian manufacturer went to an essentially mechanical redesign of a telecommunications device. Nortel’s work on the redesign followed the principles of a system called DFMA, or design for manufacture and assembly, a design analysis tool developed by Boothroyd Dewhurst Inc., Wakefield, RI. It enables product designers to pinpoint parts and assemblies that add unnecessary costs to a product, and then design to avoid them. DFMA is designed to give engineers a structured way to evaluate ease of assembly and the overall manufacturability of a product. Design for assembly requires the user to assess whether each part is necessary, and to consider the time and cost of assembling the product. Design for manufacture integrates information about manufacturing processes, allowing users to estimate manufacturing costs and make informed decisions about materials.

## Article

Public and private institutions—from universities to local cable television companies—are building up their communications networks. But even as demand grows for communications products, electronics equipment makers compete in a world where passing on costs to clients is not an option.

In the economics of electronics products, it has become a basic assumption that prices do not rise; they fall. It was precisely to meet customers' expectations of lower prices that a Canadian manufacturer went to an essentially mechanical redesign of a telecommunications device.

Nortel Networks of Ottawa, Ontario, a provider of communications equipment, software, and services, found itself with a successful design that had not reached the market quickly enough. The company believed that its product-called the S/DMS TransportNode OC-3 Express ("OC" is for "optical carrier")-offered more features and flexibility than the competitor's version and promised broader market penetration, but the competitor lowered the price on its own, more mature product.

The function of the TransportNode family of products is to carry voice, video, and data transmissions along an interconnected synchronous optical network, or Sonet for short. A compact unit contained in a single armloadsize shelf, the TransportNode OC-3 Express carries 2,016 calls per optical fiber at a speed of 155.52 megabits per second. It provides a low-cost way of extending Sonet capabilities to small and medium-scale applications, such as an office park, a school setting up for distance learning, or a neighborhood organization of teleconU11uters.

The OC-3 Express is packaged in what is essentially a sheet-metal box, measuring 14.75 xllx14 inches, including a removable air deflector and fiber storage tray. The original shelf had front and back covers, and welded shells for the sides, top, and bottom; each cost $276 U.S. to make. In response to the marketing problem, Nortel's product designers began to seek ways to reduce the OC-3's costs so as to be more competitive. According to a Nortel engineer, the purpose of the redesign was twofold. "We wanted to make cost reductions;' he said, "and we wanted to make the unit more environmentally friendly." Among the goals were eliminating eight feet of beryllium-copper gasket and finding an alternative to the zinc chromate plating used to control corrosion of the sheet metal. Engineers also had to discover a way to double the fiber-carrying capacity of the unit. The engineering team targeted the shelf mechanics, the box itself, for cost reduction. The project team disassembled the original OC-3 Express shelf and analyzed the function of each part. The original design included a hinged aluminum front cover that consisted of 53 parts and cost$78 to make. Most of the parts were fasteners.

As the analysis continued, designers also talked to vendors and customers, to work out a redesign strategy.

The decision: to tool the front cover in plastic and replace fasteners with snaps where possible. The redesigned cover consisted of only 17 parts and cost $26, a savings of$52 on the cover alone.

The new cover took about a quarter of the time, 95 seconds, to assemble, versus 378 seconds for the original.

The Nortel team discovered that a stainless-steel gasket used in the front cover was being cut with a laser at a cost of $20. Because the gasket was only 0.005-inch thick, the group decided to use a steel-rule die operation instead, which lowered the cost of the gasket to$5.

Redesign efforts for the remainder of the shelf resulted in similar savings. After learning from customers that access through the rear of the unit was not needed, Nortel engineers incorporated the separate back cover into the welded shell, which eliminated the beryllium-copper gasketing, no longer necessary for electromagnetic interference shielding. That saved $32. Much of the redesign was to reduce assembly complexity. The way the welded box went together was changed, so that it required less spinning around to assemble. It went from having welds on four sides to having welds on only two. The front cover also became all one-axis assembly. As a special project, the redesign team developed a plating of molybdenum phosphate to replace the environmentally undesirable zinc chromate used on the original product. A rearrangement of the shelves inside the box and the superior shielding provided by the new plating material let the engineers replace the remaining beryllium-copper gasket with foil-wrapped foam gasketing, which is more environmentally friendly and less costly. Peter Maheux, director of time-to-market implementation for Nortel, observed that "the mechanical engineer pulled in a materials expert and a fabrication expert, and they actually came up with a new cost-reduced plating process that's environmentally friendly. It wasn't even in the project plan. But they put it in because it was a good idea. In all, the redesign of the OC-3 Express shelf mechanics cut its total cost to$136, less than half the original cost of$276. Nortel expects to save about$700,000 a year in assembly and manufacturing costs.

Nortel's work on the redesign followed the principles of a system called DFMA, or design for manufacture and assembly, a design analysis tool developed by Boothroyd Dewhurst Inc. of Wake field, R .I. It enables product designers to pinpoint parts and assemblies that add unnecessary costs to a product, and then design to avoid them.

DFMA is designed to give engineers a structured way to evaluate ease of assembly and the overall manufacturability of a product. Design for assembly requires the user to assess whether each part is necessary, and to consider the time and cost of assembling the product. Design for manufacture integrates information about manufacturing processes, allowing users to estimate manufacturing costs and make informed decisions about materials. Because this approach requires engineers to step back and think about the basic design concept solution, it tends to encourage focused brainstorming and collaboration among project teams and with suppliers.

According to Maheux, adoption of DFMA methods required a mind-set change at Nortel. "Engineers have to get over a certain fear," he said. "Let's say it takes a year to design, develop, and release a product. Usually what happens with the old mind-set is that the engineer wants to get the front end done really quickly-the product definition, the customer requirements, all the engineering-and then get on with the design and fabrication. The change in the model is that the upfront work takes longer now. But the end result, the total end-to-end cycle, is a lot shorter."

Nortel applied DFMA techniques to another product to save money. The company's S/DMS TransportNode OC-l92 is a seven-foot, closet-size cabinet of equipment that provides highly sophisticated bandwidth management. The OC-l92 carries 129,024 calls on each optical fiber at a speed of 9.95 gigabits per second. With wave division multiplexing, the OC-192 can transport over 250,000 calls per fiber at speeds of up to 160 gigabits per second. The device is used to interconnect supercomputers in business and education, and to consolidate and direct huge volumes of transmissions.

Unlike the OC-3 Express, the OC-l92 emerged before competitors' versions and still captures a large share of market. It holds several shelves of components that control the device, manage the fiber optics, and contain fans for cooling. Installed on the transport shelf, the heart of the system, are a variety of circuit pack modules that function as transmitters, receivers, regenerators , or switches. Each transport shelf has space for 10 circuit packs, which fit into vertical slots. When an in stallation requires fewer than 10 active circuit packs, filler packs take up the extra slots.

A filler pack is more than just a dummy. It has a printed circuit board that plugs into the backplane of the unit, signaling to the system that a pack is in place and blocking electromagnetic interference that would otherwise radiate out the opening. It also helps manage airflow that cools the system.

Initially, filler packs were made of parts and castings from active packs. As sales of the OC-l92 grew, the design team identified an opportunity to lower the total cost of the OC-l92 by reducing the cost of the filler packs.

As part of its research into alternative materials, the systems-packaging team solicited help from plastics manufacturers and from vendors of injection-molded and die-cast parts. The main body of the redesigned filler pack,. which encloses the circuitry with two side shells, was constructed of polycarbonate structural foam plastic instead of die-cast aluminum. The design team's idea was to fabricate only one part, two of which could be brought together to form the left and right sides of the shell enclosure.

"The equation I like is one plus one equals three," says Maheux. "One person has an idea, and another person has an idea, and if you teach them how to communicate and team, they come out with a better idea."

The design tean1 reduced the cost of a filler pack from $410 to$65. The total number of parts was reduced from 59 to 32, and the time to assemble each filler pack was cut by two-thirds, from 15 minutes to five.

The entire redesign process lasted 10 months. The annual expected cost saving to Nortel was about \$3.45 million.