The burgeoning electronic content in automobiles has allowed numerous innovations to help improve the safety and comfort of cars and at the same time optimize the overall cost of automobiles. This has given rise to the design of chips with high levels of integration and increasing embedded memory content, permitting a multitude of functions within a single device.
However, the rapid growth in automotive electronics has presented multiple challenges in the manufacture, maintenance, and support of such systems. Because of harsh environment and physical space restrictions, automotive electronic components should be designed for very high reliability, and easy support and maintenance. Yet another challenge is to achieve lower economies of scale for these high-volume products, by generating significant improvement in manufacturing yield, which is dominated by the embedded memory within the chip.
According to SAE (Society of Automotive Engineers) reports, the electronic content in automobiles is growing, and by year 2010 close to 40% of automobile costs will be in the electronics. Typical automotive electronics comprise multiple electronic control units (ECUs) to address various functions ranging from engine management, emission and safety controls to driver information systems, such as dashboard instrumentation and navigation controls. Safety and reliability of the electronics are of key importance since they operate in harsh environmental conditions-40 to 300C. This performance demands a robust design and manufacturing flow to ensure operation in such extreme conditions.
The automobile industry faces multiple challenges in the design, manufacture and service of electronic subsystems. Design and manufacturing engineering teams face conflicting requirements of minimizing the cost of design and manufacturing test while maintaining the highest quality of the product. Limited access to several of the electronics modules poses severe challenges to service technicians and increases maintenance costs.
Over the past decade, chip designers have started to adapt structured test methodologies to address some of these challenges. Built-in-self-test (BIST) is one of the key test methodologies deployed in many large designs to test both logic and embedded memory within the chip. It involves embedding BIST logic in the design to automatically generate the test and verify the design against manufacturing defects, thus avoiding completely external test generation.
Economical testing at every phase of the automotive electronics life cycle is critical and with BIST, test costs can be reduced considerably during manufacturing. BIST also helps in the field in terms of minimizing the maintenance overhead. For example, the ECUs can be programmed to run self-test (powered by the embedded BIST) periodically, which can be monitored on a control panel. This avoids the need to access each ECU, some of which could be in hard to reach areas of the automobile.
As the sophistication of automotive electronics increases, more and more intelligence is added, bringing more information to the dashboard. This means increasing embedded memory content in the electronics. Because embedded memories impact the overall yield of the chip quite dramatically, it is necessary to have a robust test and repair methodology to be able to produce high quality chips with optimal yield.
One way to obtain yield improvement in memories is to use redundant, or spare, elements during manufacturing repair. Historically, embedded memories have been self-testable but not repairable. Embedded memories can be repaired using adequate amounts and appropriate types of redundant elements for a given memory. However, determining what is adequate and appropriate for a given memory requires both memory design knowledge and failure history information for the process node under consideration.
This alone is a challenge, but providing the right redundant elements does not entirely solve the problem. Knowing how to detect and locate memory defects and allocate redundant elements requires manufacturing knowledge of defect distribution. With embedded logic, we can both test and repair memories with redundancy to improve the manufacturing yield.