By J.R. Wilson
1970
Federated Architecture
Avionics subsystems hard-wired together and operate independently
1980
Integrated Architecture
The MIL-STD-1553 data bus opens the era of integrated standard electronics
1990
The Era Of Commonality
The F-22 fighter and RAH-66 helicopter specify common integrated avionics
2000
Integrated COTS
The Joint Strike Fighter promises to use commercial off-the-shelf components
The so-called "DMS problem" — which includes rapid obsolescence and difficult maintenance — represents the dark underside of COTS, yet component suppliers in industry are starting to form plans on dealing successfully with DMS. Now, if the government would only get on board.
Obsolescence and diminishing manufacturing sources (DMS) have become the subject of considerable concern and discussion in recent years, but finding a solution has not been easy. The consensus among those attending the recent COTScon West 2000 in San Diego was the answer requires a combination of new business practices — by industry and government — and making sure that new technologies are designed with backward compatibility in mind.
"If I were a program manager, I would consider that the technology I put in will be around longer than the technology itself. No matter what tools they use, it will be ugly in five years," warns Dale Lillard, president of aftermarket supplier Lansdale Semiconductor Inc. in Tempe, Ariz.
"I see two things happening. First, the responsibility to support a weapon system beyond, say, 20 years will be on the shoulders of the contractor," Lillard says. "Second, the contractor will have to take into account military lifecycle management, which is considerably longer than the commercial markets. The integrator has ultimate responsibility to make sure they can provide that support, which means picking vendors and designs with lifecycle integrity. And that's the key focus — lifecycle integrity."
The lifecycle in question, however, is the military platform or system, not the individual components that go into it. Technology, especially as it relates to processors and other electronic components, is advancing at a rate of one new generation every six to eighteen months. But the platform into which that technology goes may well be in operation for half a century.
Spinning-cylinder hard drives, for example, almost certainly will not exist by 2020, points out Amos Deacon III, vice president for sales and marketing at rugged disk drive provider Phoenix International in Orange, Calif. Instead, Deacon says, these drives perhaps will have been replaced by a crystal cube smaller than a golf ball, something similar to what scientists IBM and others already are investigating. The problem will be to make certain that new technology can still work as a replacement for the original hard drives installed in today's modern aircraft such as the U.S. Air Force F-22 jet fighter or Joint Strike Fighter (JSF). That means compatible connectors and data transfer.
Further complicating the problem are those very internal communications systems. The JSF will be the first fully fiber optic aircraft, but in 20 years, even that may be ancient technology, with new platforms communicating with some form of short-range wireless RF system, such as Bluetooth.
New business practices
"We have to adopt business practices that allow us to incorporate those technologies and use them in the future," explains Duncan Young, director of marketing for single-board computer supplier DY 4 Systems in Kanata, Ontario. "It all starts at having a competent component selection procedure, with a number of gates in it. If something doesn't satisfy the criteria we need, we don't use it. We also look for long life and for suppliers who characterize their components properly.
"Then it's down to engineering design practices," Young continues. "As to the more business-related practices, if we offer a product with a five-year life cycle where the individual components have a three-year cycle, we have to put aside enough components in inventory to cover that five years. Where we know customers are going to have a longer cycle requirement, we adopt the principles of technology insertion. In that, we look for a product stream where we have successive generations of a product line that incorporate planned performance improvements but remain backwardly compatible with earlier products. Then our customers can, over the life of the program, change from one product to the next and be assured everything will work."
The military's heavy reliance on commercial-off-the-shelf (COTS) components has become a serious problem in an age where the technology drivers are telecommunications and entertainment, where products are disposable and long life and backward compatibility are rarely issues. About the only good news for the military on that score is the automotive and PC industries also have slipped into the "niche" market category occupied by military systems designers, giving the three, in concert, somewhat more influence than any one of them alone could exert.
To stay at the leading edge, especially to invest in developing new processors, requires a bigger market than the government or even the automotive and PC sectors. Building a new fabrication facility is extremely expensive, as is sustaining it for several years. Military leaders have recognized they have virtually no economic influence in those decisions, yet they have been slow to find adequate ways of living with it. For the contractor community, such solutions are literally a matter of life or death.
"The defense industry will find ways of satisfying the needs of the military or they themselves won't survive," DY 4's Young says. "To do that, they will work around what's available — and if that means a few casualties along the way, so be it. That's just the way economics work. The key in any of these things is to innovate. Every problem you come up against is an innovative solution waiting to be discovered. And so long as the governments are spending a lot of money on defense, there will be innovative solutions."
The real problem, he adds, is not getting industry to align its practices with the needs of its customers, but rather getting politicians to align their practices more with the needs of industry.
Political ramifications
"I don't really think government has come to terms with the idea that the role of the equipment is changing almost as rapidly as the technology is changing in terms of supporting it through the budget cycles," Young says. "There are two choices: One, the politicians step up and say this technology is imperative to the country or, two, industry looks for alternative ways to make money in meeting that requirement. But the government still has to provide an economic baseline for somebody to come up with an alternative solution.
Young cites the U.S. Navy as an example of how the services are attempting to look at the whole system-development cycle, from initial requirements all the way through the system's phaseout. But that planning effort is useful only if budgets that recognize the new reality can match it.
"The way you manage lifecycle today differs from how you would have managed it 10 years ago — and budget practices need to reflect that," Young says. "Today, budgets tend to be structured for procurement, with different money for operation and maintenance of equipment. The reality of the way systems are now put together and maintained means systems engineering is part of the process, running through the whole life of a system. The budgeting process doesn't really accommodate that paradigm shift."
Honeywell Aerospace Electronic Systems in Clearwater, Fla., is an example of how contractors attempt to retain control of their futures. Where DMS is concerned, Honeywell leaders are making sure their internal foundry remains operational to produce radiation-hardened components for use in space. There is some question within the industry, however, as to whether that commitment will survive Honeywell's upcoming merger with General Electric.
Richard Elmhurst, Honeywell's unit business development manager, argues that Honeywell's commitment to producing radiation-hardened integrated circuits will endure. "We have a lot of ongoing programs, and programs we are bidding on, that need the radiation-hardened components coming out of our foundry," Elmhurst says.
"There is a commitment on the part of DOD [the U.S. Department of Defense] to have that radiation hardened capability, which provides the basis for keeping it available long-term. That also keeps it available for others. That enables you to keep radiation-hardened parts rather than dealing with the whole radiation-tolerant problem and screening COTS for that," he says.
For space-based systems, designers generally think in terms of radiation-tolerant, which can withstand cosmic radiation up to a total dose of 30 kilorads. Weapons-level radiation will saturate that quickly, however, which is why such systems generally require the ability to withstand as much as 300 kilorads of total dose radiation.
"Rad-tolerant just doesn't work for weapons grade unless it is an extremely short mission, such as a missile," Elmhurst says. "You also have to make sure the product doesn't latch up, which is when a charged particle with sufficient energy breaks through a boundary layer and causes high current surges that eventually could burn a part out, causing non-recoverable damage.
"If a particular part is fabricated on an old technology, say a 1.2-micron technology — and our foundry lines are now at 0.35 microns — then our approach is to transition whatever products were being fabbed on 1.2 to the latest feature size. That's one way our customers will avoid the obsolescence problem. We also anticipate when such a change may be happening, probably one to two years in advance, so we can start that process."
To ensure their processors have "legs" to carry them into the future — a vital element of their effort to counter obsolescence — Honeywell scientists take an essentially COTS processor and port it to a rad-hard foundry. The current leading example of that is the PowerPC 603e, thus making Honeywell's next-generation processor compatible with Motorola's PowerPC 603e implementation. That allows 100 percent software compatibility with the commercially available PowerPC 603e processor.
"Because we have a processor that will execute the software just as the commercial product would, then people can use all of the available commercial tools for development of that software," Elmhurst says. "That also gives those tools a path into the future. And when those tools are upgraded, they are backward compatible. That means any software written in the future will have all the hooks and so forth needed to run on that processor.
Commercial viability
Elmhurst cautions, however, that it is extremely important to pick a processor that has a commercial future. "The PowerPC has been a very popular embedded processor and very popular in the space world and its use is very suitable for real-time operating system applications, which is why we chose it," he says.
Win Smith, chief technical officer at DNA Computing Solutions Inc. in Richardson, Texas, says he agrees: "One of the key reasons people are moving to the PowerPC is they believe the existing code will work even as you move to the next generation. That typically is not the case on DSP processors because in order to get the real power out of them you had to write assembly code. The PowerPC allows you to do a tremendous amount with higher level language, which is much more portable.
"The PowerPC is such a general purpose processor that, combined with the move DOD is encouraging toward standard libraries of signal and image processing code (VSIPL — vector signal image processing library — www.vsipl.org)," Smith says. "You don't have to worry about going into existing code and changing all the calls to these sophisticated algorithm libraries as you move from one board to another. Underneath the algorithm libraries, board manufacturers are getting things optimized in assembly language, but the user only has to break down a call to an API and it gathers all the optimized code underneath."
Such standardization is becoming an important part of the government's effort to combat DMS and obsolescence in the future.
"The COTS world is helping. Before everyone turned to COTS, they had their own form factors for boards and everything else," Smith says. "Now there are standard board sizes and standard interfaces, so regardless of what is on the inside, you still have to connect to the outside in some standardized way. The government is forcing COTS technology, which means standard form factors for boards and the use of the VSIPL library. There also are some major software initiatives, such as message passing between processors in a standard way. DARPA and other government organizations are trying to develop those standards.
For DNA and others producing boards for aerospace applications — and even more so for those producing ruggedized boards for the military — those military and aerospace customers continue to drive the future of their products, even where there may be some commercial applications, such as medical imaging. But they are still subject to commercial developments.
"What goes on those boards is driven by the commercial marketplace and sometimes we are constrained on what we can build because of the availability of those items," Smith says. "The [Motorola Alt: Vec chip] is a good example — we don't use nearly as many as Apple's and they are at the top of the list to get those that are available. So at the chip level, we face competition with commercial applications, but the primes who are buying boards from us don't face that problem directly because they are our market."
Maintenance strategies
Another new government business practice, involving maintenance strategy, also has taken a strong hold recently and is dramatically changing the contractor/military relationship, Smith says. In the past, he notes, the military had a three-level maintenance strategy — the first level was some kind of built-in test (BIT) while the hardware was operating; the second was a government repair facility to fix the boards or boxes; the third level was to ship it back to the original vendor for repair or replacement.
"More than half the expense of a program is the lifecycle maintenance and logistics support," he says. "The military has found that middle layer of maintenance is extremely expensive, so now they are driving vendors to have better BIT capabilities and the ability to replace units at lower and lower levels — and they are shipping back to the vendors to repair. By removing that middle level, they are saving a lot of money on people, facilities, test equipment, and spares. We are seeing that pretty consistently, at least on new programs. The old ones tend to have some inertia to overcome.
Smith cautions that government officials need to be careful about is making sure they require replacements to be form, fit, and function identical to the original. "Obviously, they still have to keep an inventory of parts because when they pull something out in the field, they have to replace it, Smith says. "That is another reason for making the building blocks smaller but also standardized, so if you have to replace a board, you may only need to keep one type in the replacement stores and not 20."
That reflects the government's continuing effort to adopt commercial business practices — but Lansdale's Lillard says he does not believe everyone — especially members of Congress — fully understands the concept of these "extended warranties".
"Military procurement as we know it will have to change, with Congress recognizing they will have to buy or lease years of service and take into account lifecycle costs," Lillard says. "And then the manufacturers will start planning for those lifecycles." The government, he says, "has to wake up" to the need to support its manufacturers with a volume of business and money in military grade components or become entirely dependent on COTS from manufacturers with no incentive to meet special needs or requirements."
Lee Mathiesen, Lansdale's manager of integrated circuits, says he sees the cost of contractor long-term liability and support as insurance for the military customer — that "extended warranty" so common in the commercial world. The problem in getting the government to accept that, he adds, is "they have to give up configuration control — and I'm not sure they will do that in my lifetime. But they either have to allow us to compete on a fair, level playing field or they have to subsidize us.
"The truth about DMS is there is no such thing as an obsolete part — there are only parts that are too expensive to buy," Mathiesen continues. "But if you're willing to pay the price, you can get anything built."
Air Force B-2 bomber program a model for dealing with DMS issues
Many experts in government and in industry say a model for success in dealing with diminishing manufacturing sources and component obsolescence is a U.S. Air Force program at Tinker Air Force Base, Okla., that involves the B-2 stealth bomber. As has often been noted with government efforts to deal with DMS, however, eventually it all boils down to funding.
"Our most challenging issue is getting the right color of money through the DOD [U.S. Department of Defense] budgeting process," says Air Force Col. Dave Gothard, B-2 systems support manager.
Gothard and his deputies have been briefing officials of the Aging Aircraft Program at Wright Patterson Air Force Base, Ohio, as well as other Air Force the laboratories and the Air Force Materiel Command to get their assistance and support, he says.
The key to the success Gothard's program has had had to date has been taking a proactive approach to DMS.
"The first thing we did was an upfront analysis on the B-2, focusing predominantly on what we knew to be our biggest problem — microelectronics," Gothard says. "We then felt it was critical to form a dedicated team within our integration shop in the program office, involving industry and government. We also created a business-based analysis that focused on a proactive rather than reactive approach."
Gothard says his experts are using all the key elements of that proactive approach within their team, which includes the prime — Northrop Grumman Corp. in Los Angeles — and several subcontractors. "We can actually forecast when an item is going to impact the readiness of the aircraft," Gothard says. "That's tough to do in a reactive mode, where it also would be difficult to support the budget requirements."
Members of the B-2 team have focused on the portability and life of the weapons system. They analyzed the aircraft by individual subsystems, ranked the drivers from the most critical down, and found a way to determine day-to-day when they are going to have problems.
"We continue to improve this over time," Gothard says. "We have put in place active programs to fix our obsolescence issues within available funding and try to get those problems out of the red and into a green condition, which means we are supportable for x-number of years. We have essentially automated all of this and we have a dedicated B-2 website for the team to interface."
So far the effort has been concentrated on the B-2's weapons systems, but Gothard says his experts will begin assessing other program systems, including off-aircraft elements, in the future. One of the first may be a detailed analysis of the engines and their support equipment.
The team also is continuing to refine its largely PC-based tools, which essentially are a combination of simple program spreadsheets, databases, and the website. They also have been developing a forecasting model that will identify by line replaceable units (LRUs) when a specific card in a specific box will get to the point where spares are no longer available.
"From a design and technology insertion perspective on the B-2 program, as we have identified problem issues and elements that may need to be upgraded in the future — such as communications or navigation — we make sure our team does a thorough review of what items will need to be procured — COTS, NDI (non-developmental items), whatever — so the team, in effect, has the ability to influence the design and support of those items to ensure we will be supportable in the out years," Gothard says.
"We try to make certain we don't go down a path toward unplanned obsolescence," he explains. "We rely heavily on our prime contractor to work with the suppliers, but we also have the latitude to deal with them directly. In addition to making recommendations to changes in design or working with the suppliers on what can be done, we also go through our own database and, if we find a useful substitute for a part, then we recommend that they consider it."
Gothard admits that looking into the future to prioritize weapons systems drivers is not easy, but says their integrated approach — with a broad-based team coming from industry and many parts of DOD — combined with a highly proactive philosophy is working.
"We have been very successful working more than 500 problems that could have impacted the weapon, coming up with solutions that will keep us highly supportable in the short term and have implemented multi-year buys, usually five or so years of inventory on items absolutely critical to the weapons system," Gothard says. "We've been pretty much on track as to what we predicted and what we've spent. Our biggest challenge is to make sure we get the appropriate level of funding to continue the analysis and implement the solutions in a timely manner. Just doing the analysis isn't enough."
Gothard says he believes the team approach and tools in use at the B-2 DMS Program can be adopted by any program.
"We've gotten pretty good support and have passed our information to the program executive officer for fighter and bomber programs, who has called our approach a best practice," he says. "What we've been doing also can help programs that are still in development — you don't have to be a legacy system to benefit from this approach." — J.R.W.
