Industry, DOD at odds over F-15 jet fighter parts reliability
An electronic parts aftermarket supplier is calling into question the way the U.S. government buys replacement components for the U.S. Air Force F-15 jet fighter. The supplier claims the F-15's replacement parts may be unreliable, while F-15 maintainers say they have experienced no problems.
By John McHale
TEMPE, Ariz. — An electronic parts aftermarket supplier is calling into question the way the U.S. government buys replacement components for the U.S. Air Force F-15 jet fighter. The supplier claims the F-15's replacement parts may be unreliable, while F-15 maintainers say they have experienced no problems.
Leaders at Lansdale Semiconductor in Tempe, Ariz., claim that some digital integrated circuits (ICs) on the F-15 are not form, fit, and function compatible with the original parts, and may cause errors in mission-critical systems aboard the U.S. Air Force F-15 aircraft during flight.
The replacement parts in question are from the Generalized Emulation of Microcircuits (GEM) program at the U.S. Defense Supply Center Columbus (DSCC) in Columbus, Ohio. GEM is in place to produce modern versions of obsolete components. Lansdale, on the other hand, owns the means to reproduce exact copies of ICs originally used on the F-15.
Potentially affected systems include the AN/APG-63 and AN/APG-70 radar systems from Raytheon in El Segundo, Calif., and the aircraft's identification friend or foe (IFF) systems, claims Dale Lillard, president and owner of Lansdale.
If one of those systems failed it could force the aircraft's pilot to abort the mission and potentially put other squadron members at higher risk; if the IFF falters, the pilot and his aircraft might face friendly fire potentially resulting in the loss of a multimillion-dollar aircraft and its crew, Lillard says. Raytheon officials did not respond to repeated requests for comment.
Lillard says so-called "soft" failures in GEM parts that have not been stringently tested and may work too slowly or too quickly for the existing system may cause catastrophic system failure. A soft failure describes a malfunction that subsequent ground tests cannot replicate.
"They also could be more susceptible to an electrostatic discharge that could affect other systems in the weapon causing them to misfire or fail," Lillard says. "However, there have been no confirmed failures on the F-15 due to this, but that does not mean it couldn't happen."
No soft failures have occurred in GEM devices in more than 10 years of board, system, and flight-testing of GEM devices in more than 2,000 F-15 applications, says Sam Calloway, an electrical engineer for avionics hardware for the F-15 Systems Support Engineering Group at Warner Robins Air Logistics Center at Robins Air Force Base, Ga.
Calloway, who says obtaining the original part may be more efficient but not always possible, has asked Lansdale to stop saying there will be soft failures on the F-15 because it is misleading.
Problem with the process
"The problem with the GEM parts results from a lack of behavior compatibility between parts made with different processes," Lillard contends.
The issue dates back to 1991 when officials at the Defense Logistics Agency (DLA), anticipating the discontinuance of various microcircuits, funded a research project to emulate relatively old BiPolar and CMOS devices using a newer BiCMOS process, Lillard continues. Changing the manufacturing process, he says, also affects configuration control or traceability of the parts.
Exacerbating the problem, Lillard says, is the refusal of DSCC officials to reissue part numbers after manufacturing devices with a different process than that used for the originals. Engineers at Sarnoff Corp. in Princeton, N.J., manufacture GEM ICs for DSCC. Sarnoff officals say they are comfortable with the way DSCC officials are responding to this issue and do not see any need to comment further.
"DSCC wanted to use a new technology that would be around for at least 20 years and BiCMOS was the latest technology at the time for producing integrated circuits while the BiPolar process was the original technology used to produce the parts," Lillard says. "However, parts produced using the BiCMOS process are not interchangeable with parts produced using the BiPolar process without consequences. Private testing efforts, using a DSCC-certified laboratory, have disclosed behavioral incompatibility between the parts, affecting performance and reliability."
Lillard cites the Russian military as an example of the potential consequences of this change. Many pieces of Russian military equipment failed because of the lack of configuration control and the inability to accurately trace part failure, Lillard says. Lillard warns U.S. military leaders not to follow the same path.
Lansdale and other aftermarket suppliers already manufacture parts with the original BiPolar process to make them compatible with the relatively old systems, such as the F-15, F-16, F-18, AEGIS, and the Patriot missile system, Lillard says. Nevertheless, parts from aftermarket suppliers are listed with the same part number as the GEM parts, he adds.
"Herein lies the error," claims Lee Matheisen, operations manager at Lansdale. "Although the parts are emulated using a different technology, they are being marked with the same part number as the original part even though they are not form, fit, and function the same as the originals. This practice is directly in opposition to standard configuration management practices. These look-a-like parts may work in the confines of one system in which they are qualified, but will they work in all the systems, which use the original part? Was a design review done? Was qualification performed? By keeping the original part number for a significantly different part, are we putting the other systems in jeopardy because the design constraints of these systems is different than the system which qualified the look-a-like?"
Lillard says that, when they see the same part numbers, technical and logistics managers and technicians are not aware that the GEM part is not an identical part with identical performance.
"These potentially nonconforming parts may not fail through normal testing and the degradation on the whole system may not be apparent until the system is pushed to its limit, as these systems often are in combat," Lillard continues. "This can have deadly consequences for the pilots and troops in the battlefield yet the weapon-system engineers and technicians will not have a way to trace or even find out about these failures."
This is not a simple problem, Matheisen points out. "The GEM program has built more than 40,000 parts, which are either in DSCC stores, repair depots, or military systems. These parts should be recalled and remarked to reflect that they really are not form, fit and function replacements of the original parts."
Officials at DSCC disagree. "GEM parts do meet all form, fit, and function requirements of the original items," counters Tom Beckstedt, equipment specialist for the GEM program at DSCC. "As to whether any problems will occur anyway, there is no way to predict all the possible idiosyncrasies upon which a system designer may have relied but did not include as a form, fit, and function requirement of his part specification."
While Lansdale's claims are theoretically possible, Beckstedt says a trend of statistical evidence must back it up before any changes are justified. "They have not provided evidence of that kind," Beckstedt says. "DSCC has nothing to hide; if we did we would not have provided parts to Lansdale to test."
The GEM parts and processes characterics can be traced back to the original manufacturer through the manufacturer's identification required by MIL-PRF-38585 or by the lot date code unique to each production run, he says.
According to Beckstedt, a GEM part not only meets the form, fit, and function required by the part specification, but also mimics, as closely as possible, the actual performance of the original part as obtained by characterizing samples of the original part. Further, the GEM program avoids doing emulations wherever possible, Beckstedt says.
"The GEM Program Office (DSCC-CCD) will only undertake an emulation when industry — including the original manufacturers, aftermarket, distributors, and part brokers — cannot meet the government's requirements," Beckstedt explains. "We agree with Lansdale that the original part should be used whenever possible. But even that is not absolutely without risk.
Support for Lansdale
Officials at the Joint Electronic Device Engineering Council (JEDEC) and the Government Electronics & Information Technology Association (GEIA) have joined Lansdale in requesting better configuration control of GEM parts.
Michael Cooper, of General Dynamics C4 Systems and Chair of the G-12 Solid State Devices Committee of the Government Electronics & Information Technology Association, and Brent Rhoton of Texas Instruments and Chair of JC-13.2 Committee for Government Liaison at the JEDEC Solid State Technology Association, outlined their concerns in a letter to Greg Saunders of the Defense Standardization Program Office regarding GEM parts.
"Although we see the benefit of the GEM program for supporting obsolete integrated circuits for weapon system logistics, we are very concerned about the re-use of existing standardized part numbers to identify GEM emulated devices," they wrote. "One of the basic principles of configuration management states that if form, fit, or function of an item changes, you must change the associated item number. In the case of GEM emulated devices, the form in the case of the device technology clearly changes and with that the potential for differences in unspecified or untested parameters.
"The decision to use the same part number for technologically different products being emulated by the GEM program needs to be stopped immediately," Cooper and Rhoton wrote.
Saunders says he has looked into the situation and "does not see enough evidence to do what Lansdale asks." The cost and effort involved in changing part numbers and the thousands of technical manuals are not justified by Lillard's claims, he says.
"Dale [Lillard] is a patriot," Saunders says, adding that the work that Lansdale and other aftermarket suppliers do for military weapons systems is important and "we want to continue to work with him in the future."
Some claim that Lansdale's complaints are for financial reasons; the GEM program, Lansdale officials believe, is unfairly competing with them. Critics say Lansdale is making these claims because they have not gotten anywhere with the competition argument.
"Hey, I'm not doing this for the money," Lillard counters. "If GEM wants to compete against us, TI, or National, let them. We'll win. What I'm concerned about is the potential loss of life if this problem is not fixed."
Lillard says he has talked to his lawyers but "you can't sue the government." Lillard does believe, however, that using a different process to produce the GEM parts violates the contract form, fit, and function requirement that Sarnoff has with DSCC — and he plans to have his lawyers advise Sarnoff so.
Lansdale has also enlisted the help of U.S. Sen. John McCain, R-Ariz., who sits on the Senate Armed Services Committee. A McCain source says the senator is interested in the issue and will continue to explore it, but declined to comment further until he learns more from the Department of Defense.
Sample parts in question
Engineers from Lansdale Semiconductor in Tempe, Ariz., tested GEM devices against the original parts to compare behavioral compatibility, and say they found errors with some GEM products.
"For all five of the device comparisons made, the original Lansdale parts passed all electrical testing," says Lee Matheisen, operations manager at Lansdale. "Lansdale electrically tested the GEM parts to the specification in the same manner as the original device manufacturer would have tested them and shown significant deviations from the original devices."
The parts they tested were M38510/30201BCA (QML#54LS40), M38510/02206BDA (QML#54H103), 351-7495-012 (QML#529), B77T0137-0123 (QML#54181), and M38510/06302BEA (QML#10525), Matheisen says.
Lansdale officials found part M38510/30201BCA (QML#54LS40) to be OK. Since the test, DSCC officials have ordered Part 10525 removed from the QML. For more information on test data go to Lansdale's Web site at www.lansdale.com.
"No recall has happened or is needed, since all production devices manufactured, except those provided to Lansdale, are still resident within the Sarnoff Corp," says Tom Beckstedt, equipment specialist for the GEM program at DSCC.
"The GEM contractor, Sarnoff Corp., and their subcontract QML test house, each independently developed a test program for M38510/06302BEA, and each independently made the same interpretation of the test set-up requirements for the parameter in question," he says.
"The qualifying activity, DSCC-VQC, investigated the issue and, working with all the aforementioned parties and the DSCC test lab, determined that the interpretations of Sarnoff and their test house did not meet the intent of the specification and as a result, the GEM parts did not meet the specification," Beckstedt continues. "Due to this evaluation, DSCC-VQC removed Sarnoff as a QML source of supply for this part. This is the first time that this has happened since Sarnoff became a QML supplier in 1996. This is an isolated case and accounts for less than 0.6 percent of the military device part numbers Sarnoff is approved to supply."
This is not, however, a serious indictment of the GEM program, Beckstedt says. "It would be erroneous to conclude from this one misinterpretation that GEM is inherently incapable of producing form, fit, and function parts."
Beckstedt counters that some of Lansdale's data "is potentially erroneous because the characteristics tested were not listed in the required specification marked on the GEM device. Four out of the five GEM devices are electrically compliant to the specifications they were intended to meet."