by J.R. Wilson
Diminishing manufacturing sources (DMS) has been a major issue of concern for the military for more than four decades. Evidence of this is the 1959 creation of the Interservice Data Exchange Program. This organization evolved in 1970 into the Government-Industry Data Exchange Program — better known as GIDEP — which has a laser-like focus on government microelectronics quality and procurement.
With this increased military reliance on commercial microelectronics, DMS has become a looming problem with increasingly shorter runs between introduction and obsolescence of commercial-off-the-shelf electronics components. Now the United States is on a war footing and has rapidly depleted many of its weapons stores in the past few years in military actions in the Middle East, Africa, and Central Europe. With the depletion of U.S. weapons stocks, the situation has become all the more acute for DMS & Material Shortages (DMSMS) programs and GIDEP.
GIDEP receives funding by the U.S. government and management from the U.S. Navy. The organization is a cooperative effort of the U.S. and Canadian governments, as well as industry, to conserve resources by making maximum use of existing information. The program disseminates technical information that involves research, design, development, production, and operational phases of systems, facilities, and equipment. Since its inception, participants say they have prevented more than $1 billion in unplanned expenditures.
Now, however, DMS essentially has entered a new dimension, with new rules and requirements. The global war on terrorism is increasing the military's tempo of operations (opstempo), which intensifies pressure to find spare parts quickly. Military commanders might put certain operations aside for a few weeks in peacetime, but delay is no longer an option as the anti-terrorism war heats up.
"We're running into similar situations no matter what sector we look at," says Luis Garcia-Baco, chief of the U.S. Army Materiel Command's industrial base capabilities division in Alexandria, Va. "Microelectronic obsolescence has been very well categorized and understood. The problem now is in other areas, where the Army is trying to get support from other services and OSD (Office of the Secretary of Defense)."
From Kosovo to Iraq to other fronts of the war on terrorism, opstempo increasingly influences DMS, and conversely, DMS issues also influence an increased opstempo. Such traditional solutions as lifetime buys of discontinued components may prove inadequate if supplies diminish more quickly than military leaders expect, as has been the case with cruise missiles.
Other alternatives, such as remanufacturing or back engineering of out-of-production components, may now be too costly or time-consuming to meet the needs of a military on a war footing.
"For some applications, going to COTS (commercial-off-the-shelf) was a big mistake; how are you going to upgrade every 18 months? They (the military) are certainly heading for a much larger train wreck — and how spiral development answers that without a bigger budget is beyond me," warns Justine Corby, director-business development at Sarnoff Labs in Princeton, N.J.
"We will get to a point where a Sarnoff or even the aftermarket companies just won't be able to help you," Corby warns those who mandate the use of COTS components. "You're designing in technology that we will have trouble keeping up with and so will other companies. Then Congress will have no choice but to fund those redesigns, or you'll have no fleet.
Aftermarket component suppliers include Rochester Electronics of Newburyport, Mass., and Lansdale Semiconductor of Tempe, Ariz.
"At least we have a silicon processing base here," Corby continues. "If we are in a war and seriously need that, you can get it up to speed. But a lot of things require toxic chemicals and so forth that no one in this country handles that any more. Then what? You have to have a backup plan, including redesign; you have to have a plan and then another plan."
Aftermarket suppliers, and government-sponsored programs to manufacture components that private companies have discontinued such as the Generalized Emulation of Microcircuits (GEM) program at Sarnoff, may not be able to cope with widespread obsolescence and diminishing manufacturing sources, she says.
"If we know GEM technology is never going to manufacture this part, or Rochester is never going to inventory that part, then you better have a two- or three-tier plan," Corby says. The GEM program, developed by the Defense Logistics Agency (DLA) to help solve the DMS problem, is under management of Defense Supply Center Columbus (DSCC) in Columbus, Ohio, with Sarnoff as prime contractor.
GEM technology uses a set of gate arrays and single-line processing technology to build form, fit, and function replacements for non-procurable microcircuits. These GEM devices are supplied to existing system documentation such as source control drawings, military slash sheets, standard military drawings and/or M38510 specifications, and meet the specifications of the original devices.
Corby says military and industry leaders also may have to expand the definition of DMS to include "unreliable sources" — offshore COTS manufacturers whose governments may block future shipments if in disagreement with U.S. policies.
"We sometimes write mission-specific statements saying we don't want that to happen (foreign parts). That is usually written at the prime contractor level and is expected to flow down," says Tim Mawby, DMSMS program manager at ARINC Inc. of Annapolis, Md. "Now we're in a position where we're micromanaging to the piece/part level. And the folks at DLA and DSCC have responded 110 percent in that effort."
However, parts coming from an offshore manufacturing facility may not be coming for a foreign-owned company, further muddying the waters, notes Eric Brandriff, obsolescence management team lead on the Navy's Standard missile at Raytheon Missile Systems in Tucson, Ariz.
"We usually have multiple sources for our components. However, the majority of microelectronics come from manufacturers with many offshore manufacturing facilities," he says. "We recognize that just about anywhere you go, you will be dealing with someone who has an offshore facility. If we were limited to only Continental United States manufacturing, anybody in the military would be adversely affected — not just DMS, but regular procurements. That would have a much greater overall impact than just a technology change."
Still, the problem with COTS components is not limited just to availability, says DMS and defense standardizations consultant Joe Chapman of Midland, Texas.
Pushing the envelope
"Many contractors are buying commercial- or industrial-grade microcircuits, but still need higher or lower temps than the spec sheet says," Chapman explains. "It is common practice to go to a testing house and have them uprated (upscreened). And every test house engaged in this will tell you if they start with a standard 2,000 unit lot, they might get 1,500 that will meet the requirement. But in a few months, the same test on the same part, same supplier, may only yield 500, because the IC (integrated circuit) guys are tweaking these things daily to improve yields and take out margin they don't need and shrink them. That's not a problem for the commercial world, but it is for the military, Chapman says.
"So I'm more concerned about the future because there is so much evidence of these declining yields from the test houses," Chapman continues. "Even if they get 500 'good' ones, you have to wonder how good those really are. Maybe those 500 operate faster than the 1,500 did six months earlier — and who knows what impact that faster speed will have on the systems into which the engineers are putting them. The test house has no way to know what those changes may be if the IC house doesn't share it, which they may not. All these unknowns are big concerns — and getting bigger."
There also is the problem of available funds and the so-called "color of money" — or the different purposes for which Congress allocates money. There is no such thing as a DMS budget, per se; most money for that purpose comes from existing resupply or maintenance budgets. But in time of war, officials spend that money more narrowly for replenishing depleted stocks of missiles, for example, or to keep combat aircraft flying.
"We increase opstempo and we have no choice; we have to make the jets fly, no matter what we have to do. And a whole lot of people who still don't believe DMS is out there or understand the impact on their system will find out the truth," ARINC's Mawby says.
"The flip side, in a wartime scenario, is we need the funding to keep up the opstempo. But that funding goes for bombs and fuel and things needed to prosecute the war on a day-to-day basis, not on verifying new parts," Mawby continues. "You're supporting the warfighter while your repair pipeline stacks up and they will do whatever they have to do to get things out the door. Still, where we are now is a lot better than where we were two years ago."
Ongoing plans for a digitized battlefield, especially one in which Land Warrior-equipped soldiers essentially are clothed in integrated circuitry, will further influence DMS issues, although there is some debate about whether it will increase or decrease the problem.
Officials of the U.S. Department of Defense (DOD) also have been attacking DMS through a commonality and standardization effort. They are doing this in hopes of keeping suppliers interested in manufacturing obsolete parts. Reasoning goes that manufacturers will support old product lines if they have a sufficiently broad base of users throughout the military, rather than just one program that may only need a few hundred or thousand units to keep a system functioning. The standardization effort also has led to more attempts to port weapons across several different platforms. One example of this is mounting the Hellfire missile on the Predator unmanned aerial vehicle (UAV). Hellfire originally was designed for the U.S. Army's AH-64 Apache combat helicopter.
The rapid pace of technology is not always seen as a DMSMS catastrophe, however.
"There may actually be a trend to less DMS in the future," says Dr. Gary Gaugler, technical advisor on DMSMS and microelectronics obsolescence to the Defense Microelectronics Activity (DMEA) in Sacramento, Calif. Gaugler says DOD officials classify a part as DMS when the part's manufacturer quits making it because it is no longer profitable to do so.
"It's strictly a business decision," Gaugler says. "If you look at what has happened to technology over time, from the 1960s to today, the newer technologies are inherently more reliable than the older stuff. All things being equal, the newer technology has better reliability that one could extrapolate out to a reduction in DMS."
On the other hand, especially with respect to true COTS, he adds, forcing commercial-grade microchips into the kinds of heat-and-cold environments for which they were never designed can create whole new problems.
"The chips weren't made to work like that," Gaugler says. "They also now have whole different mechanisms of failure that earlier generations didn't have; they fail in different ways. The replacement of COTS elements will become extremely onerous when they become DMS, which they will. We're already seeing that in DSP (digital signal processing), for example."
DOD officials created DMEA to find ways of using advanced technologies to solve operational problems in weapon systems, increase operational capabilities, reduce operation and support costs, and reduce the effects of DMS. It also is the executive agent for DOD Integrated Circuit (IC) Microelectronics DMSMS. As the military increasingly relies on "smart" weapons, microelectronics obsolescence — which tends to run in 18-month cycles — has become the main factor driving DOD system degradation.
"The bulk of our work is not replacing a DMS part but to use advanced technology to replace the functionality of one or more chips, including some that aren't DMS yet," Gaugler says. "You just take some part of the box or board that is having trouble and redesign it to newer components. If it is a function that needs higher performance than you get from an FPGA, for example, then we build it in our foundry as a custom part. And in the future they just come back and we produce however many of those custom parts they want.
"An increasingly key element in this approach, however, is time," Gaugler says. "It takes about 16 weeks to do a new array, for example, but nine months to a year if a full custom job is required." Custom jobs, he explains, include underlayers that already exist in an array. Those times include required testing and validation, as, more often than not, the first attempt may not work. How quickly this can be done depends on a number of factors, including whether DMEA has access to the original dies or other production elements.
Sometimes U.S. systems integrators may want a foreign-manufactured part made domestically. For this, DMEA can perform two levels of reverse engineering. They can dismantle the part down to the silicon, and trace out the entire circuit diagram. For more complex chips, however, they may have to perform a functional decomposition. This looks not at how the part was designed or built, but only at its intended purpose. That requires developing a complete data set on the chip's functionality, then creating the necessary code and gates to verify the solution, then picking the silicon process to make the replacement.
How long either approach takes depends on the complexity of the chip. While the average time for a physical decomposition is about nine months, a functional could take anywhere from seven months to a little more than a year. Still, modern conflicts are unlikely to last that long, insofar as major stockpile expenditures and the need for replenishment are concerned.
All these factors have led the U.S. military services to become much more proactive with respect to DMS. In fact, service leaders try to look as far into the future as possible with respect to every part of every weapons system, including looking at obsolescence possibilities from the very first day of a new program. That includes a major dedication to standardization and commonality in new systems and in any upgrades or technology replacements that may be under consideration on old programs.
"We're involved with a lot of DMS teams in the services that are looking at obsolescence from a proactive standpoint," says Dave Robinson, DMSMS/GEM program manager at DSCC. "DLA works with them on that and looks, whenever possible, to a cross-platform commonality perspective. So from a teaming perspective, we try to show the platform PM (program manager)that the part is being used in a lot of other platforms and they perhaps shouldn't try to redesign or replace it because of the impact it will have on other programs. I think we're getting a lot better at doing that.
A closer look
"We tracked one component that was used in 189 spots in one Air Force platform [the F-15 jet fighter] — and we found it was used in 57 other weapons platforms, including other Air Force and Navy, plus a few Army and Marine applications," Robinson says. "If the F-15 program designed that part out, it would impact dozens of other programs. More than likely the redesigned part would not have been applicable to the other platform requirements because the Air Force would have had to redesign the card."
Robinson's office re-evaluates some 14,000 individual components each year from a DMS standpoint. About 90 percent of those are microcircuits, although he says the highest dollar value probably belongs to a relatively ancient technology — vacuum tubes.
"Tubes will be around for a long time and there isn't a weapons system out there that doesn't still require them in some way, some place, especially in avionics suites, radars, and shipboard systems," he says. "There's nothing out there that will carry the voltages and power those things carry and they are still required in many, many cases."
Because there are no specific DMS budgets to handle life buys or even just major stockpiling of parts, the DLA supply centers have become the military's warehouse.
"When someone comes in with justifiable requirements, we buy the items and put them on the shelf for them, even though there usually is no real requirement on them to come back and buy those items from us," Robinson explains. "If it is a risky buy, we do put some teeth into the agreement. We recently told the Air Force if they did not buy the number of one component they had predicted from us each year, we were going to bill them for the parts, anyway. We couldn't afford to put that much taxpayer money on the shelf without some kind of guarantee. In another instance, we required them to pay half the lifetime buy up front.
"Projections have gotten better. In the past, they might not have been what they should have been and when we do a lifetime buy, the customer is required to come in with exact requirements, replacement rates, when they expect surges and so forth," he says. "They don't have the funds to buy all those right now, so they come to DLA to do that."
Another solution is the aftermarket community, companies such as Rochester and Lansdale. Still, not everything falls into the aftermarket arena. Some parts also may be manufactured and tested within the services' own depot repair facilities. DLA has called upon them in the past to produce small quantities of some items urgently needed but no longer being made by the original manufacturer or no longer available from an offshore supplier.
In some cases, the government can order continued production of vital parts.
"In this type of situation, if there is a conflict with delivery of any component, we will invoke Defense Priority and Allocation Systems (DPAS), under Title I of the Defense Production Act, which gives us the opportunity to prioritize those components we need to support the global war against terrorism," Garcia-Baco says. "So if a component has an expected high demand, invoking DPAS means we could extract the quantities of that component we need and divert them to our needs. We have ways, under statute, to provide the warfighter with needed components if we have to."
However, adds AMC industrial operations director Harrell Barnett, there are limits to that authority that are much more important today, as more and more purely commercial components are designed into military systems, than when DPAS was first authorized during WWII.
"If it is a truly commercial item, DPAS is hard to enforce. It usually requires a defense contract or subcontract. If it is just a supplier item that the contractors buy off the shelf, it becomes difficult," he says.
While electronics continue to top the DMS "hit list", the increased opstempo also has created a demand for non-electronics that sometimes pushes them into the DMS mode.
"In the past, GIDEP reported 84 percent electronics-caused obsolescence and 16 percent non-electronic, but that 16 percent is getting a lot more attention," notes Walter Tomczykowski, DMSMS program director for ARINC. "DMS now includes more products — everything on the aircraft, for example, rather than just the electronics. The airframes are getting older and getting more use, so some of the non-electronic parts are into the wear-out phase and the original supplier may not be there.
"One of the things we're beginning to do for our programs is look at operational impacts, including predicting when they will run into a critical point where they may not have spares to support a system. The more you use equipment, the bigger the demand, so increasing opstempo will have an impact."
In a war environment, says Sarnoff's Corby, all DMS and related problems immediately turn critical as parts originally thought sufficient for 10 years are used up in six months.
"Then what? Then you have a big 'gotcha' if you can't find an aftermarket or we can't emulate it or there's nothing to cannibalize, which really only works on a large old fleet," Corby says. "You don't have that many JSTARS or Global Hawks out there. Then you get to the point of how to design out obsolescence right from the start," she says. "We can't emulate every part in the world. You have to know your parts up front and put together a two- or three-tier long-term plan. But will that be budgeted up front? Will the PM be around long enough? And then you have issues of the color of money, which is always a topic. What war does is accelerate all these so it hits the fan a lot faster."
Corby paints a chicken-or-egg picture for program managers trying to keep their systems up and running as DMS issues multiply rapidly and reaction time vanishes. Emulations, such as Sarnoff's GEM programs, are one tool, along with cannibalization, substitution, re-engineering, life buys, aftermarket suppliers, and so on, all of which have to be balanced out program by program.
"And that's where it becomes challenging, because the most proactive PM usually loses priority because other things come up that are more pressing, so by the time that proactive PM gets what he or she wants, you wind up being in reactive mode. And that can be very frustrating," she says.
"That will become even worse in time of war because the funding for your proactive DMS plan gets cut to deal with more urgent matters right now. Which means by the time they get their funding, it's a fire and requires a more expensive solution because they didn't react when it first came up. It may be too late then for a lifetime buy and you're driven into a redesign. Going to war only makes that worse by shifting priorities and accelerating depletions and, ultimately, exacerbating the DMS problem. But we all do it."
Planning for DMS
But Raytheon's Brandriff says being proactive, at least at the contractor level, may be the best line of defense in the long run, especially as a wartime environment pushes production demands.
"We have a group that goes out and looks at the nature and future of parts, which gives me the information I need to look at future alternatives in production. If I can predict a part is going to go obsolete two or three years ahead of time, that opens up a lot more options than if I waited for a lifetime buy notice from GIDEP. By contacting suppliers and using commercially available databases, we usually find out long before we are notified by GIDEP, probably 98 percent of the time," he says.
Rapid advances in technology pose some significant challenges to keeping up with some of those problems, especially given increased requirements stemming from the increased opstempo.
"With funding assistance from the customer, we have been able to further define within best practice what design engineering needs to do up front to allow for future parts changes down the road, more flexibility in the footprints they can accept on the board, the things necessary in design to allow us to swap parts when it becomes necessary," says Bill Hartshorn, who heads up the Raytheon team Brandriff described.
With America now on a war footing and coming off several years of major stockpile depletion, DMSMS is bound to influence budgets, schedules, and supplies in what, due to the accelerated timeframe, may quickly turn into a vicious circle. But as program managers, supply centers and contractors struggle to meet the new challenge, one concern remains uppermost.
"Our ultimate goal is to make sure the warfighter doesn't know what DMS is," Mawby says. "We're behind the scenes and all this should be transparent to them."
A technician at Sarnoff Labs in Princeton, N.J., uses modern chip-fabrication technology to copy form, fit, and function from obsolete electronic parts that are no longer available from their original manufacturers.
GEM technology uses a set of gate arrays and single-line processing technology to craft replacements for non-procurable microcircuits, using existing system documentation to meet the specifications of the original devices.