by Phillip Southard, PDS Consulting, LLC - Plainsboro, NJ USAand Tabitha Ivey, Manufacturing Technology, Inc. - Houston, TX USAAbstract:
Outdated obsolescence solutions employed by the military and aerospace industries are described and discussed. New solutions are introduced that include smart printed circuit board (PCB) design, the use of obsolescence forecasting, and intellectual property (IP) based solutions. Included in the IP based solutions are recommendations to purchase RTL models to prevent architectural obsolescence, retargeting the silicon in question to a smaller geometry, and during system redesign updates use system-on-a- chip (SoC) strategies to combine peripheral components with main system functions.
Integrated Circuit (IC) component obsolescence continues to adversely impact the aerospace and defense industries. Depending on the program visibility and system priority the solution to IC parts obsolescence is seen as simple as conducting a life of time buy as opposed to initiating a risky system redesign. Current Obsolescence Landscape
IC component obsolescence, while an industry-old problem, still requires a more efficient solution. The mandate to use commercial off the shelf (COTS) technology has placed the aerospace and military system designer on a technology life cycle of about 2 4 years. This has compounded the problem of component obsolescence and not resolved the issue. Component purchases that fall into the COTS category continues to increase. In-Stat/Micro Design Resources (In-Stat/MDR, www.instat.com) and others predict that COTS purchases will grow to more than 35% of the worldwide military-aerospace IC market by 2005 .
The reliable discrete IC technologies of years past are being phased out. Transistor-Transistor Logic (TTL) introduced by Texas Instruments (TI) in the 1960s was phased out in the late 1990s. TTL technology was a reliable standard that designers could depend on for their system components of choice. TTL had a life cycle length of about 31 years.
As process engineers and scientist continued investigating ways to consolidate chips functions, complementary metal-oxide-semiconductor or CMOS was introduced to the market by RCA in 1968 (en.wikipedia.org). The use of CMOS discrete devices as a technology to include in new designs declined in the late 1990s. Its life cycle spanned approximately 29 years.
High-speed CMOS logic or HCT has had a useful new design life of 26 years. Some of the newest discrete devices use a Low Voltage Technology or LVT, which is projected to have a 13-year life cycle . As we can see from these references, the life cycles of COTS technologies are shrinking. The trend on the landscape of technology must be addressed.
Electronic devices are not the only components suffering from the obsolescence plague. Non-electronic devices cannot escape the problem created by new advances in technology. In the aerospace field, requests for components such as springs, valves, connectors, antennas and multiple gauges and instruments are coming up empty .
Obsolescence has been a challenge since the first transistor was packaged and is here to stay and will continue to spread into other segments of the military and aerospace systems. New Definition
The technology and commercial life trends mentioned above suggest that IC component obsolescence will continue to grow. One thing remains the same throughout the years observed above; there is a need for the logic functions of discrete components. It becomes apparent while evaluating the continued need for the logic functions and technology life-cycle trends; a new definition for component obsolescence is needed.
We will now consider IC component obsolescence to be composed of two halves:
- Silicon or technology obsolescence, the unavailability of fabrication wafers in a particular technology or geometry.
- Logic function or architectural obsolescence, the unavailability of the logic function that could be retargeted to newer technology, FGPAs or next generation ASICs.
This new consideration for component obsolescence points to separating the logic functions from the technology to which it has been implemented.
For example, the need for a discrete Hex Inverter IC has not vanished. While the technologies that have provided the logic execution have vastly changed. One could not readily find a 74S04 device, but could purchase a 74LVT04.Outdated Strategies & Solutions
Reactive response strategies have been used in the military and aerospace communities for many years. Reactive approaches include Life of Type buys and Bridge buys. Life of Type buys involves trying to predict and purchase the total number of components that will be used during the life of the platform. This is a risky approach as predicting component needs over the life of a program is difficult and shortages usually occur.
A bridge buy is commonly used to address architectural obsolescence. The total number of components needed during a prototype or development phase of a new platform design will be purchased. The bridge buy will support the number of parts needed until a system redesign for the production phase can be completed. Implementing this strategy means that the system being produced will be different from the system that was tested in the development phase. This will result in both schedule and funding impacts. Both of these reactive methods are quite expensive and possibly unpredictable.
Additionally, two engineering supported response strategies were used to defend against component obsolescence. In the 1980s, OEMs designers could respond to a component obsolescence issue by simply using two standard methods. One method to solve component obsolescence is part substitution and the other is to use an alternative source.
Both of these methods were feasible, at the time, due to multiple sources for military components. Sometimes, as many as, four vendors would supply devices that conformed to the same functional specification. This allowed for part substitution or using an alternative source.
As the number of sources of military components has shrunk, this is no longer a viable option and often substitution or other sources of supply are simply not available.
The military and aerospace communities found themselves dealing with obsolescence by initiating system redesigns. This is the most current method of addressing architectural obsolescence. System redesigns can include redesigning line replaceable units (LRU) or printed circuit card assemblies (CCAs) or other equipment.
Performing system redesigns have proven very costly for customers and have lengthy verification schedules. Along with high costs and long schedules, there exist many other system issues when trying to manage military systems and their sustainment. One major issue is maintaining system voltage vs. COTS declining chip voltage.
Borrowing formulas from economist, including New Present Value formulas, for determining when to redesign and when to make life-time-buys military system engineers and planners are aggressively trying to fight this ongoing battle .
Other programmatic responses found in the U.S. include developing government sponsored IC foundries. Over the past ten years, the Defense Logistics Agencys Generalized Emulation of Microcircuits (GEM) program has made a valuable contribution by emulating the functions of discontinued digital devices. However, in the future the shortened life cycle coupled with the increased complexity of devices may prove to be more challenging than a single government sponsored program can manage. An alternative for the aerospace and military industries is to follow successful commercial practices, not attempt to correct them.
Although silicon obsolescence could be defeated with a dedicated fabrication facility, the aerospace and military communities have not found this strategy affordable. Short of creating a dedicated fabrication for every type of IC, IP based system design is the most logical approach to address this issue.
As we have observed, dealing with component obsolescence is not a simple task. Initiating a board or system redesign cannot solve every occurrence of component obsolescence. Budgets and time constraints do not allow for this strategy. New Solutions
The need is clear. The military and aerospace communities need a new systematic approach for lowering impacts of both silicon/technology and architectural obsolescence as defined above.Assumptions
Before discussing new solutions, we need to allow for two assumptions. The first assumption is that silicon/technology obsolescence will continue. Second, PCB board redesign and modification will continue but can me minimized.
The two assumptions mentioned above create a framework for which new strategies can be constructed.New Strategies & Tactics
Obsolescence management is our first recommended strategy. By using obsolescence notification services and parts management strategies component obsolescence can be contained.
System designers can use notification services such as Product Watch from Farnell of the UK or PrecienceAlert of Precience MD, USA or Manufacturing Technology, Incorporateds AVCOM of the USA. All of the products can give managers and designers of military and aerospace systems advanced warning of obsolete, end of line, and end of stock items. Government-Industry Data Exchange Program (GIDEP) alerts has also proved to be a significant resource in monitoring component obsolescence.
For alert products like those mentioned above to be effective, OEMs or military & aerospace systems managers need to adopt a Whole Life Part management strategy. This dictates that system managers allocate money to have dedicated parts engineers on hand to manage parts list from initial parts selection to production and sustainment of the life of the system.
A second recommended strategy is to address architectural or logic function obsolescence by using IP based designs. By implementing this approach, architectural obsolescence can be defeated while monitoring silicon/technology obsolescence.
Designers can obtain register transfer level (RTL) models of soon to be or obsolete devices so the technology can be retargeted to either a next generation ASIC or FPGA. This essentiality creates a new source of supply while minimizing risks.
RTL models are commonly referred to as intellectual property (IP). There are 3 main types of IP. The first is OEM IP and this type is generally the most expensive IP to acquire. This refers to a behaviorally equivalent solution. There is also generic IP. This refers to the IP used by the OEM without custom modifications. This type is only functionally equivalent. There is also third party IP.
Examples of the types of IP being referred to are demonstrated by evaluating an Aeroflex RM7000 microprocessor. PMC Sierra manufactured the part and would be the OEM IP provider. PMC Sierras RM7000 was a customized version of a MIPS based microprocessor. MIPS would be a generic IP provider.
In addition to OEM and generic IP providers, there could be any number of third party IP providers. These IP models can be developed in house, but usually due to skill set deficiency and or time and money constraints, design groups look to third parties for IP solutions.
To be more specific, the models mentioned above are called Semiconductor Intellectual Property (SIP). Semico, a marketing and engineering research company located in Phoenix, Arizona, forecasts that SIP will be categorized into the following categories: Star SIP, Critical SIP, Purpose-Built SIP, Housekeeping SIP, and Commodity SIP .
Star SIP includes MCU and DSP Cores. Critical SIP includes mid to high performing analog blocks and specialty memory blocks. Purpose-Built SIP includes specialty blocks and communications interfaces. Housekeeping SIP includes Soft Error Correction and performance enhancement. The last category is commodity SIP and it includes standards based / building block libraries .
The use of IP in the form of RTL also allows for the retargeting of larger geometry layouts to smaller geometries. By having the RTL of a component, one can retarget a design to a new technology and reduce NRE costs. This also allows for some control over silicon/technology obsolescence and a choice in when to make process changes.
There are risks associated with buying and implementing any type of IP. The risks include an engineering risk, a contractual risk, and a patent or copyright risks.
While these risks can lead to complications during implementation and negotiations, using a pragmatic approach to SIP transaction can keep problems to a minimum. Buyers should be rigorous in their examination of data and may wish to look for licensors who document their SIP in accordance with industry-accepted data taxonomies such as the Virtual Socket Interface Alliances VCT specification .
A final strategy, based on lessons learned and new techniques, is to follow a systematic plan for new design starts and sustainment activities as they relate to combating IC component obsolescence.
Designers can begin by using smart PCB board design, i.e. partitioning of system components by chip complexity and obsolescence risk and not just function. Customers can request forecasting of potential obsolete components using Technology Road Mapping. This knowledge can then be used to implement a smart PCB design .
Technology Road Mapping is a process by which all ICs of a particular military or aerospace system are evaluated and recommendations are made on the best path ahead to mitigate architectural obsolescence. Recommendations vary depending upon the type of device being evaluated, the acquisition stage of the system, and availability of IP.
Devices like DSPs are evaluated for migration paths suggested by the IC vendor.
New PCB designs should be partitioned so at risk devices are located on special mini daughter boards. Daughter boards contain just a microprocessor or dsp with voltage regulators to support future voltage reference changes, and add spare pins to daughter interfaces to provide future expansion.IP SoC Consolidation
While time to market may currently dictate the use of COTS devices, keeping a focus on IP based system design will prove useful. When a system redesign is needed, the IP for a previously used COTS device may be available. Thus, providing an opportunity to shrink part count and increase reliability by consolidating the COTS device with other logic functions.
The consolidation of components into large blocks is a basic concept of IP SoC. System designs that use an IP SoC design approach would include FGPAs and ASICs. These devices will absorb peripheral components and IP blocks to create sub-systems on PCBs.
As IP SoC becomes more commercially adopted, the military and aerospace communities should also adopt this practice for faster time to market and cost savings.Conclusion
We believe adopting obsolescence containment & IP based strategies are crucial to building a systematic approach to aerospace and military system design and can reduce redesign costs and risk of downed or inoperable systems. Consistent with the objectives of Acquisition Reform, the aerospace and military industries should leverage the investment of the commercial industry in deployment of IP based design at chip and system level.
Manufacturing Science and Technology Division at Redstone Arsenal, Alabama is actively pursuing the use of IP in the aerospace and military environment to speed products to market and deter the effects of technology obsolescence. As of May 2003, funding was awarded to Manufacturing Technology, Inc and Honeywell Space Systems- Clearwater for an initial evaluation of star IP. References
 Prophet, Graham. Guarding against component obsolescence EDN 14 Nov. 2002: 63-73.
 ARINC, Component Obsolescence: Its Not Just for Electronics Anymore Presented at FAA/DoD/NASA Conference on Aging Aircraft Sept 16 2002.
 Porter, G. Zell, An Economic Method for Evaluating Electronic Component Obsolescence Solutions Boeing, May 1998, p. 4.
 Condra, Lloyd, Combating Electronic Component Obsolescence by Using Common Processes for Defense in Commercial Aerospace Electronics Presented at NDIA Conference.
 Popper, Nick. Risk reduction in SIP transactions Presented 2003 FSA Semiconductor Workshop, Oct 10 2003.
 Wawrzyniak, Richard: Semiconductor Intellectual property: Enabler of the Future Presented 2003 FSA Semiconductor Workshop, Oct 10 2003.