Obsolescence: every COTS designers bad dream

Feb. 1, 2000
Commercial technology continues to make the U.S. military the most advanced in the world at a competitive price. However, designers of defense platforms that need to last 20 to 30 years spend much of their time creating new ways to accommodate the powerful commercial technology that goes obsolete in 18 months

Obsolescence: every COTS designer`s bad dream

Commercial technology continues to make the U.S. military the most advanced in the world at a competitive price. However, designers of defense platforms that need to last 20 to 30 years spend much of their time creating new ways to accommodate the powerful commercial technology that goes obsolete in 18 months

By John McHale

The 1994 commercial-off-the-shelf (COTS) initiative brought military designers the powerful performance of commercial processors, the detailed graphics of computer gaming, and the ever-doubling capacities of commercial storage devices — but obsolescence, the monster under the COTS bed, continues to keep military designers up at night.

The latest high-performance COTS technologies go obsolete in 18 months or less while weapons systems still have five to 10 years in design cycles and service life spans of 20 to 30 years. Regardless of a COTS manufacturer`s technology-refresh plan, the new components still need to go through a qualification and validation process. It is a time-consuming process, which will see even more parts go obsolete.

The high turnover of technology is not a side effect of COTS but a fact of life in the commercial world, says John Dobroski, manager of display products for North America at Barco Display Systems division in Duluth, Ga.

To deal with the situation, military designers are implementing procedures such as technology insertion, design freezes, and lifetime buys. The goal is to prolong the life spans of commercial parts, says Richard Jaenicke, director of marketing at Mercury Computer Systems in Chelmsford, Mass.

Although embedded computers, which cover a wide variety of the military designs, have longer life spans than desktop technology, it is desktop technology that drives the market, Jaenicke points out. The desktop provides a vastly larger volume than embedded products, he adds.

The military is finally beginning to "accept the fact that things are changing," says Jackson Ho, program manager for vetronics systems at Computing Devices Canada in Nepean, Ontario.

The military is slowly getting the idea. Designers are using the most advanced cost-management techniques and are relaxing screening procedures to accommodate commercial technology, says Frank Ramos, strategic marketing manager for multichip modules at Aeroflex Inc. in Plainview, N.Y.

For years the military establishment considered plastic encapsulated parts taboo, says Karl Walter, Aeroflex`s multichip module product line director. However, during the mid-1990s, officials at the U.S. Air Force and at Boeing in Seattle needed the performance and price on the Boeing C-17 and C-130 cargo aircraft that they could get only from a plastic part, Walter continues.

They used the 6U VME microelectronic reduced-instruction-set computer (RISC) processor boards from engineers at the Aeroflex subsidiary, Aeroflex Laboratories, and Sanders, a Lockheed Martin Company in Nashua, N.H., Ramos says. The boards use the Aeroflex 64 bit MIPS RISC processor technology, he adds. Aeroflex Laboratories officials signed a license agreement with Sanders to manufacture the boards.

A small number of program managers are extra safety conscious and lean toward qualification, Ho says. These managers want to minimize risk, but "they are getting fewer and fewer." Space is one area where you have to requalify everything — sometimes three times before the customer is satisfied."

No matter what, designers must prepare themselves and their organizations for technology insertion, Ho says. Whether or not the customer wants to embrace new technology, the parts are going obsolete quickly, Ho warns.

Technology insertion

One of the keys to technology insertion is convincing military officials to embrace the concept.

"It is usually the primes that persuade the program offices, but where possible we push for technology insertion as well," Mercury`s Jaenicke says. Program managers should plan to insert new technology every five to six years, he adds.

The best example of combating obsolescence is the U.S. Navy`s Acoustics-Rapid COTS Insertion (A-RCI) program to upgrade the submarine fleet, Jaenicke says. Navy leaders have close relationships with software and hardware providers to keep close watch on which technologies are going obsolete to get the benefits of new technology as quickly as possible, he adds.

The program was launched with the Virginia-class (NSSN) attack submarines, and is expected to spread to 75 percent of the Navy`s submarines, including the Los Angeles (SSN-688) and Seawolf (SSN-21) classes, in a year and a half, naval officials say

ARCI keeps the upgrade process from standing still, Jaenicke says.

Computing Devices experts use modularity to provide for technology insertion in their field computers. They provide modular designs for their processors, embedded graphics boards, and removable memory, which enable them to upgrade their computers quickly with minimum cost and effort, Ho claims.

The Computing Devices Vulcan 6100 to 6300 workstations are all designed with modularity, he says. The 6300 is the command-and-control computer for the Canadian army`s Athena program, which will upgrade 350 of that army`s vehicles.

The program, which is only a couple of years old and has not begun actual design, already has gone from a 233 MHz Intel Pentium II microprocessor to a 266 MHz version. Today the program has moved to a 300 MHz Pentium, Ho says. The modular design will help Computing Devices accommodate these changes and any more that might come along, Ho explains.

The processors currently have life cycles of about 18 months, Ho says. However, Intel in Santa Clara, Calif., does support certain products in the embedded market for three to five years, he adds.

Hard drives are the toughest to keep up with because they change on such short notice, Ho says. The drive manufacturers notify their customers about two months prior to building the drives, then after six months of manufacturing often decide to move onto something else, Ho explains. Computing Devices has already changed hard drives twice on the Athena program; the current capacity is 6.4 gigabytes, he says.

Some systems also use smart batteries to enhance their power efficiency, Ho says. A smart battery tells the system what type chemistry it has, i.e. nickel hydride or lithium ion, Ho explains.

No designer can bring modularity to every component, Ho warns. "With a Pentium II you`d have 1,000 modules," he adds. There would not be enough connectors to balance it out.

Computing Devices officials also maintain relationships with certain providers so they can understand their providers` product roadmaps intimately, Ho says. This enables Computing Devices engineers to choose a product at the beginning of its lifecycle and plan for future upgrades when the technology becomes obsolete, he explains.

Barco engineers embrace standards as a way to discard obsolete technology and insert the new and improved versions without a hitch, says Jeff Malacarne director of engineering at Barco.

The U.S. Navy has already qualified three generations of Barco graphic boards under the AN/UYQ-70 Advanced Display System program, Malacarne says. The so-called "Q-70" started in 1994-95 and a new generation has been put in about every two to three years, he explains.

Four years ago military customers did not understand the need for technology insertion and industry standards, but they have made a great leap Malacarne says.

Video game technology has so overwhelmed the computer industry with its performance, that military designers are being forced to tap into the technology, Malacarne continues. This new technology goes obsolete very quickly, however, and the military must use industry standards as much as possible to make upgrading easier, he explains.

Still, Barco engineers work in a less integrated, higher-resolution digital environment than does the video game industry, Malacarne says. As such, Barco engineers can design a next generation and freeze it for five years. "The key is to obsolete yourself before someone obsoletes you," Dobroski says.

When designers provide the low cost, light weight, and solid performance with backward compatibility that technology insertion offers, they will maintain their customer base because for customers staying is easier than switching to another provider, Dobroski says.

The Q-70`s 20-inch cathode ray tube (CRT) design has become obsolete and the market for the others is shrinking, Dobroski says. "We`ve jumped ahead by offering liquid crystal displays (LCD) that are backward compatible to CRT technology," he continues. "This is an advantage over companies who only offer LCD displays without any CRT experience."

Barco`s display group also fights obsolescence through the use of basic commercial standards, scalability, and lifetime buys, Dobroski says.

The main idea is to minimize the cost to the customer when a part goes obsolete, says John Wemekamp, director of strategic marketing at rugged board manufacturer DY 4 Systems in Kanata, Ontario. Depending on the customer`s application, DY 4 engineers use lifetime buys, freeze designs, and apply technology insertion to combat obsolescence, he adds.

The technology insertion process involves "F3I or form fit function integration," Wemekamp says. The replacement board or card should be a subset of the original with improvements in power, processing speed, and other design considerations in mind, he explains.

DY 4`s DMV 179 single-board computer is a F3I replacement for the DY 4 178 board, which is a replacement for their 177 board, Wemekamp says. Replacing the DMV-177 was the DMV-179 on the U.S. Navy/Marine Corps F/A-18 Hornet jet fighter-bomber as part of the OSCAR (Open Systems Core Avionics Requirements) program.

Sometimes this is impossible to do because designers still must qualify the boards. There are times when a new DY 4 board, with the same specifications and tested by DY 4, will replace one currently installed and the manufacturers will waive some of the qualification and validation process, Wemekamp says.

This will happen more as the military continues to demand increased performance, he adds.

An alternative to providing technology insertion is initiating lifetime buys, which enable companies to take strong control of technology, Malacarne says. Opting for lifetime buys help companies make the most of their resources by freeing engineers to work on other projects, he explains.

Lifetime buys depend on the customer`s requirements, Ramos says. While they can solve an immediate problem, they also tend to be expensive. This is the reason that every program should provide for some type of technology insertion, Wemekamp insists.

Software compatibility

Software takes years to develop and often can be more expensive than hardware. Still, it does not go obsolete so quickly as hardware, says Richard Comfort, vice president of marketing at CPU Technology in Pleasanton, Calif. With that in mind, Comfort says designers must scale their hardware to work with old software.

Scalability, he insists, is the key to combating obsolescence. "Scalability means that a module is binary compatible with existing system software," Comfort explains.

Commercial processors, for example, have an average shelf life approaching 18 months while many computer systems may be in use for 25 years or more. The result, as the system ages, is increasing failure rates, high maintenance costs, and the inability to adapt to changing requirements.

CPU Technology uses its Validated Modernization process, which enables designers to insert the latest electronic technology into an existing system to increase reliability, safety, and performance with no ill effects on existing software, Comfort says. One or more circuit boards using system-on-a-chip technology comprise a module, he adds.

The core tools in CPU Technology`s process are automated verification (Behavioral Verification Technology, clock-cycle accurate system models (SystemLab), and reconfigurable design frameworks (CFRAMETM). The different processes provide extensive compatibility testing and complex system virtual prototyping, which can assure comprehensive design compatibility and performance, Comfort claims. The technologies combine to form CPU`s TriTech, which makes it possible to produce affordable, high-assurance, scalable modules, Comfort says.

The company is under contract to supply CPU1750A-60 microprocessors for the color-programmable display generator for the U.S. Air Force`s F-16 Common Configuration Implementation Program (CCIP) electronics upgrade program. Lockheed Martin Tactical Aircraft Systems in Fort Worth, Texas, is the prime contractor on the CCIP program. Engineers at Honeywell Defense Avionics Systems division in Albuquerque, N.M., are designing the display generator.

The CPU1750A-60 provides higher reliability and more performance than the MIL-STD-1750A from Performance Semiconductor in Sunnyvale, Calif., which it is designed to replace, Comfort claims. While retaining absolute binary compatibility with the existing legacy software, it provides a modern 64-bit engine to implement new capabilities. It is far less risky in time and money than redesigning the system around a new incompatible processor, CPU Technology officials say.

CPU`s process is technology independent because it is fabrication independent, Comfort says. For example during the development process for CSIP, National Semiconductor in Santa Clara, Calif., CPU Technology`s original foundry of choice, announced its exit from the digital business to concentrate on its traditional analog products, Comfort says. "They left us twisting in the wind," he adds.

However, because of foundry-process independent design, CPU Technology was able to switch to Epson America in Long Beach, Calif., with minimal impact on schedule and none on budget, Comfort claims. It is this same design technique that makes it possible for CPU Technology to guarantee its customers long-term support for its plug-and-play system-on- a-chip solutions, he adds.

"What makes us different is that we own all the intellectual property," Comfort says. Companies that use a PowerPC may get increased performance, but are stuck with the same size because Motorola Semiconductor Products Sector in Austin, Texas, owns the intellectual property, Comfort explains.

The military companies are at the mercy of the high-volume companies like Motorola, he continues, because they cannot take advantage of scalability as long as these companies hold all the intellectual property.

"We don`t obsolete at all, Comfort says, there is no need to because we support and design all our products."

DSPs

Users of digital signal processors (DSP) are also seeing obsolescence and compatibility issues due to the growth of the PowerPC.

Mercury engineers believe that RISC microprocessors such as the PowerPC are more forward compatible than floating-point DSPs because DSPs have longer life spans Jaenicke says. In addition, the SHARC 21060 and 21160 DSPs from Analog Devices in Norwood, Mass., are not forward compatible with the company`s new Tiger SHARC device, Jaenicke points out.

This is one of the reasons that Mercury has decided not to use the Tiger SHARC, Jaenicke says. Also, "the device`s performance per watt isn`t compelling enough for us to choose it," he adds.

PowerPC performance is starting to move ahead of the SHARC, Jaenicke says. Mercury will use RISC-based chips in a most of their new designs, he adds. The performance boost starts to make a difference "when you get a two-to-three-year difference in ease of use," Jaenicke explains.

Lovers of DSP devices from Texas Instruments in Dallas must see RISC have five times the performance before they even consider switching, Jaenicke says. Their loyalty is too strong, he adds.

In radar, sonar, and signal intelligence applications, RISC will continue to replace DSPs because they have more power, are easier to use, and have backward and forward compatibility, which helps combat obsolescence, Jaenicke says.

COTS backlash

Despite all the innovative solutions and technology insertion plans, there may be a COTS backlash among some military original equipment manufacturers that do not have the infrastructure to keep up with changing technologies, Aeroflex`s Walter says. These companies, he warns, may eventually go out of business.

When former U.S. Secretary of Defense William Perry announced his COTS initiative commercial technology was turning over every five to seven years and now the rate is 18 months, Ramos says. It is a constant battle for compatibility while technology life spans continue to shrink, he adds.

The military will never be able to stay abreast of the commercial technology, Ramos says. When the military lost its market share it lost its influence, CPU Technology`s Comfort says.

The large commercial vendors like Intel consider $190 million in volume a nuisance, which is why they abandoned the military market, Ramos explains. "The bottom line is the dollar," he says.

"We have not yet seen the full ramifications of COTS, Ramos adds.

CPU Technology`s Validated Modernization process to be part of Navy`s CPUP

Engineers at CPU Technology in Pleasanton, Calif., are developing prototype processors to help U.S. Navy systems designers solve electronic obsolescence issues as part of the service`s Compatible Processor Upgrade Program (CPUP).

CPUP is one way to solve obsolescence problems plaguing complex real-time computer systems such as avionics and communications systems. The Carderock Division of the Naval Surface Warfare Center in West Bethesda, Md., awarded the contract for the Naval Supply Systems Command (NAVSUP) in Mechanicsburg, Pa.

The software in these systems takes years to develop and perfect, CPU Technology officials say, but the hardware it runs on is becoming obsolete quickly. The aging systems end up with rising failure rates, high maintenance costs, and the inability to adapt to changing requirements, CPU Technology officials say.

CPU Technology engineers are using their Validated Modernization to decrease those costs and enable the new hardware to run the old software, and still have room to run new software programs.

"These vital programs address issues that are critical to the Navy`s ability to cost effectively maintain its readiness," says NAVSUP program manager, James Fitzgibbon. "CPU Technology`s advanced technology will help us solve some of the most difficult problems we confront today."

The technology makes it possible for CPU Technology engineers to ensure processor, circuit-board, or system compatibility with other processors, boards, or systems, CPU Technology officials claim. The same technique also enables CPU Technology engineers to guarantee their customers binary compatibility and life-cycle support for their ReNew plug-and-play system-on-a-chip solutions. — J.M.

New Aeroflex MCM offers backward compatibility

Aeroflex Circuit Technology engineers in Plainview, N.Y. offer backward compatibility with their new MIPS microprocessor multichip module (MCM) that performs better than their previous models, yet retains plug-in compatibility.

The new ACT-5271SC 64-bit superscalar microprocessor MCM with two megabytes of embedded secondary cache, packaged in a 280-lead ceramic flat pack, is backward compatible with Aeroflex`s previous-generation R4400-based MCM which has 1 megabyte of secondary cache.

The new RM5271-based part, a third-generation design, will fit in the same spot on the board as the company`s first- and second-generation parts, says Karl Walter, Aeroflex`s MCM product line director.

Aeroflex engineers plan to combat obsolescence by increasing performance with their products while maintaining their standard small footprint and providing a migration path to future products, Walter says.

At the heart of Aeroflex`s new part is the Quantum Effect Devices (QED) RM5271 dual-issue superscalar microprocessor chip that can issue one integer and one floating-point instruction per cycle. The 64-bit multiplexed system address/data bus offers high-performance write protocols to make the most of uncached write bandwidth.

To gain increased performance in slower legacy hardware, an FPGA-based FIFO structure allows the internal secondary cache bus to operate at two or three times the external system bus rate, Aeroflex engineers say. The CPU pipeline can run at 2, 2.5, 3, 3.5, 4, 4.5, 5, 6, 7, 8, or 9 times the secondary cache bus rate.

Aeroflex developed the ACT-5271SC at the request of several major military subcontractors to add computational power to the Bell-Boeing V-22 Osprey tiltrotor aircraft mission computer. Aeroflex`s MIPS microprocessor MCMs are currently designed into the mission computer on the V-22, the Navy Aegis program, and other military smart munitions and avionics platforms. — J.M.

Click here to enlarge image
Click here to enlarge image
Click here to enlarge image

The U.S. Air Force is using Aeroflex MIPS processor technology aboard their cargo aircraft.

Click here to enlarge image

The Vulcan 6300 workstation from Computing Devices Canada is part of the Canadian army`s Athena program.

Voice your opinion!

To join the conversation, and become an exclusive member of Military Aerospace, create an account today!