1999 Mil/Aerospace COTS Conference
Background
The 1999 Military/Aerospace COTS Conference was held in Berkeley CA on August 25-27, 1999. In this conference various aspects of the use of Commercial Off The Shelf (COTS) electronic components and systems in high reliability applications were discussed. There is continuing interest in the use of COTS in place of the former "high reliability" or "military" electronics. In some cases, procurement cost savings are realized while in others, the desired parts are only available in a COTS format. In low volume applications the procurement savings must be balanced against increased qualification and testing costs which are incurred when COTS components are utilized.
Summary of selected talks
Robert McCraig of Lockheed Martin (LM) Federal Systems talked about the use of COTS in systems for the new Navy Virginia class attack submarine. The USN goal is to reduce the total cost of ownership to 1/4th the cost in previous submarine classes. He stated that the USN sees COTS as the path to major cost reduction. The general scheme of using COTS Printed Circuit Boards (PCBs) in a submarine environment is to harden the cabinets and their mountings so that the PCBs can survive military levels of shock and vibration. The Virginia class will use a whole shock mounted platform for the electronics, including the deck and the operators in a "cocoon" like arrangement. A strategy of using standard Application Programmer Interfaces (APIs) is used for software so that software does not need to be regenerated when hardware is upgraded.
Robert stated that, in the past, DoD contractors such as Lockheed Martin made their profit from the military hardware that they provided. Now the profit is mostly in system integration since there is little direct profit realized from pass-through COTS. One major advantage they realize from using COTS is that their designers get to work with the latest versions of electronics and software and thus remain viable in the commercial hiring market. This enables LM to hire skilled engineers for work on military systems.
Eugene Hnatek of Compaq talked about the qualification of COTS technology and components for use in Compaq high reliability computer systems (the former Tandem). He emphasized the importance of accurately defining design and system performance and environmental requirements prior to component selection. They make a detailed technology assessment before choosing components for high reliability computing usage.
Francis Edwards from Raytheon discussed PEMs qualification for military systems. Their standard PEMs qualification is 96 h of HAST at 130°C/85% RH and 1000 temperature cycles, -55 to +125°C. They feel it is especially important to do preconditioning prior to HAST. They make a major effort to understand the supplier "Reliability Data," including how current the data are and what the supplier's failure criterion were. This information is frequently difficult to extract from supplier "Rel. Data."
Dave Locker from USA AMCOM discussed application and management of COTS for USA missile systems. He advocated "higher level" or PCB environmental stressing. They use an 85°C/85% RH, 60 h test and thermal shock, -46 to +71°C. At the component level they continue to recommend sequential HAST/temperature cycle stressing. They have seen "lot to lot" variations in PEMs and recommend qualifying each purchased lot. Their cost modeling suggests that high qualification costs are acceptable for high value systems such as missiles because any improvement in system reliability represents both a very large cost saving and a critical performance improvement in a military situation.
Luu Nguyen from National Semiconductor Corp. discussed the "microSMD" package, a Chip Scale Package (CSP). It is essentially identical to the Sandia "Mini-Ball Grid Array" (mBGA) of the early 90s, a bare chip with an array of solder ball interconnects which can be assembled with SMT equipment rather than the sophisticated flip-chip "vision" assembly systems. National sells the microSMD for 3-48 lead count devices with a 0.5 mm or larger pitch. These devices are intended for use in portable products such as cellular telephones. Luu discussed the environmental testing; temperature cycling, -40 to +125°C and THB testing, 85°C/85% RH for 1000 h. Through optimization of the chip solder mask and the solder ball geometry, acceptable THB and solder fatigue performance has been achieved for small to medium die sizes. The technology meets level 1moisture sensitivity and the parts are supplied in a tape and reel format.
John Fink from Honeywell discussed the current status of the Stack International organization, which is enabling competing commercial avionics suppliers to address electronics component manufacturers with one voice. In this way they achieve more responsiveness from the component suppliers than the individual companies can. Stack has developed a general procurement specification for integrated circuits (ICs) which is intended to replace the old MIL-STD-883 which applied only to hermetic parts. It can be reached from the publicly accessible part of the STACK website (http://www.stackinternational.com). Stack is also developing guidelines for use of components outside of the manufacturer's suggested operating range and a tool for reliability prediction that is intended to replace the old MIL-HDBK-217. John extended an invitation for other types of organizations, such as government agencies, to join Stack.
Bob Knoell discussed some work the Visteon Automotive Systems is doing with Pat McCluskey of CALCE on development of a "Virtual Qualification" tool for analyzing environmentally induced failure modes and the associated rates of components on a Printed Circuit Board (PCB). This work is in the same spirit as the Printed Wiring Analysis (PWA) tool that CALCE supplies to members for analyzing solder fatigue failures on PCBs in temperature cycling and/or vibration. For the new model they input details about the component package construction, including dimensions and materials. Using various Physics of Failure (PoF) models, the failure rates associated with thermal cycling, temperature, humidity stressing,... are derived. This work is in an early stage and Bob was careful to point out that the CALCE model would probably be more useful in pointing out potential weak links in a design rather than in providing quantitative failure rate predictions. The model is a potential "MIL-HDBK-217" replacement for prediction of board level reliability from the properties of the applied components.
Francis Edwards of Raytheon discussed the use of Parylene to protect PEMs against moisture induced damage. Parylene conformal coats are widely used for protecting PCBs against moisture and chemical induced damage and it is frequently claimed that the Parylene provides additional protection for the PEMs on the PCB. The diffusion analysis and experimental weight gain data that Francis presented showed that the additional protection provided by the Parylene was not significant for the parts he tested. For thin PEMs packages Parylene provides some transient retardation of moisture intrusion but it does not reduce the concentration of moisture at the die surface for steady state moisture exposure.
John Wall from the Defense Evaluation and Research Agency in England presented a very provocative paper on the enhancement of radiation induced damage in BJTs and MOSFETs caused by low dose rate exposure. He investigated the effect of packaging; ceramic hermetic(CHP), PEMs, Burn-In, with or without, and Co60 gamma dose rate; 0.002, 0.05, and 50 rad/s, on the performance of the devices. For BJTs he measured current gain (hfe) and for MOSFETs the threshold voltage (Vth). For both device types he observed an Enhanced Low Dose Rate Effect (ELDRE) in that, for a given total dose, the parameter degradation was larger in magnitude at the lower dose rates. However, in contrast to some other studies, John observed much larger degradation in the CHP parts than in the PEMs. He also observed a small improvement in parameter degradation in PEMs parts which had been burned in relative to those which had not.
The audience asked many pointed questions, particularly why the ELDRE had not manifested itself as electronics failure in the many deployed satellites which routinely receive low dose rate gamma exposure. John had no ready answers but he stands by his data and results.
Many additional interesting papers were presented at this conference. A book of the presentation materials will be available from the conference organizer, Ed Hakim, at the Center for Commercial Component Insertion
(ebhakim@bellatlantic.net).
-- submitted by Jim Sweet