By Kelly Sewell
SCOTTSDALE, Ariz. - The COTS movement may be saving dollars in many military programs, but it`s a logistical nightmare for designers who are developing boards with commercial parts that may or may not be available in six months. The obsolescence problem "is 100 percent worse since the COTS mandate," claims Gorky Chin, director of engineering at Vista Controls Corp. in Valencia, Calif.
Chin`s comments are in reference to the June 1994 mandate by Defense Secretary William Perry to use commercial off-the-shelf (COTS) products whenever possible.
Ray Alderman, executive director of the VME International Trade Association (VITA), in Scottsdale, Ariz., has devised a formula to determine the mean time to obsolescence (MTTO) of commercial chips used on VME boards. Alderman admits his formula is unscientific and is drawn from his background in engineering, yet he claims it will help military leaders to define the problem.
Alderman`s formula has already caught the attention of engineers at the Naval Surface Warfare Center - Crane Division, in Crane, Ind. Jerry Braun, branch manager of Crane`s commercial technology applications engineering branch, says he and his staff are trying to come up with their own model to determine when to refresh technology on existing systems and which products to refresh. Braun`s model should be complete in the spring.
In the meantime, Braun says Alderman may be on to something with his formula. "To say it`s accurate, I`m not sure, but it sounds kind of reasonable. I`m not sure it`s the end-all formula for every product and technology. It seemed simplistic, but it made sense."
Alderman`s formula is based on a root-mean-square method. It takes one divided by the square root of the number of desktop chips on a board times the average life cycle of chip technology (see diagram).
"It seems rational to me," Chin says. "What [Alderman`s formula] is actually saying is `your obsolescence problem is as bad as your worst part.`"
Alderman builds his formula on the saying, `Identifying the problem is 90 percent of the solution?` "My formula identifies the problem. Now it`s up to the suppliers to come up with a solution," Alderman says.
Currently, users depend on vendors for information pertaining to production start and end times, as well as when the vendor will stop supporting a product.
Braun praises Alderman`s approach for its objectivity - anybody can take the formula and plug in his numbers to get the MTTO of a particular product. "Ray`s formula would take a lot of the bias out of the data," he says. "The vendor may just tell you what you want to hear. He`ll say we`re planning to support our product to 2020, but it all boils down to dollars and cents - if he`s not making money, he`ll quit making it and supporting it."
Engineers at DY 4 Systems in Kanata, Ontario, are addressing the obsolescence issue by incorporating suggestions from an American Institute of Aeronautics and Astronautics draft report entitled "Recommended Practice for Parts Management." The report suggests designers plan for obsolescence by looking at the technology roadmap of a component and accounting for it in the design, says Dave Pedley, vice president of quality assurance.
This is done by maintaining a close relationship with the supplier to maintain accurate technology roadmaps. By designing for obsolescence, engineers can avoid surprises later on. Pedley admits that suppliers are getting stung right now, facing obsolescence issues on products that are less than a decade old. "Five or six years ago, no one thought this would happen," he says. "So now we have to think, `How can we avoid this in the future?`"
Engineers at Vista Controls also have problems with parts obsolescence as vendors leave the field and parts fall from favor, Chin explains. But company officials take any of three approaches to combat obsolescence in future designs.
The first is to push for soft designs that incorporate field programmable gate array and application-specific integrated circuit technology, in addition to designs that use more universal languages such as Verilog and VHDL.
The second approach is to buy the rights from suppliers. That way, they purchase the die and store them until they are needed.
The third approach is to make a lifetime buy of the part. Chin says the third choice is the least favorable.
Alderman notes that his MTTO formula actually determines when components are halfway to disappearing from the market because of obsolescence. "If it works out to eight months, you actually have about 16 months, because there`s the time it takes when the chip maker announces its going away to when it`s actually gone."
