FPGAs tackle microcontroller tasks: Part 1 - Application growth strains architecture and ASICs

The number of microcontrollers (μCs) in automotive electronics systems continues growing with each model year. Microcontrollers power many electronics functions in today's automobile—the typical number in luxury models, for example, is well over 100 and counting.

Those numbers account for existing systems. But in the future, greater numbers will go toward improved GPS-based navigation systems, as well as in newer stability management, by-wire braking and steering, collision warning, voice recognition, Internet access, night vision, and collision avoidance systems, as well as many others.

Virtually every auto electronics application shown in figure below requires a μC, and those applications continue to expand dramatically. Each model year, automakers offer more sophisticated electronic systems to deal with vehicle safety, telematics, and infotainment. Increasing costs are paralleling these newer applications, as well. Auto electronics currently account for 22% of a vehicle's cost and are projected to increase up to 40% by 2010.

Semiconductor content essential for designing these auto electronic systems is also increasing and is projected to grow rapidly during the next five to 10 years. While vehicle growth is expected to be flat, semiconductor growth is projected to grow 9% compound annual growth rate (CAGR) to 2010, as shown below. This growth is fueled by multimedia-rich, in-cabin features, and safety applications such as lane departure warning systems.

Limits of growth
As auto electronics designers engage in these newer applications, they face a growing host of μC-related design issues. These problematic areas stem from traditional device architecture and the fact μCs are largely application-specific products. These limits mean a new μC with a different set of features is required for each application.

As a result, hidden design costs associated with μCs emerge. For example, if a particular μC doesn't have the right mix of features, system designers must augment it with external logic, software, or other integrated devices. Moreover, many μCs with specialized features and fixed number of dedicated interfaces often do not fulfill many application demands.

Semiconductor manufacturers have historically attempted to meet as much of the total market demand by marketing product device families from a single μC core architecture. This approach worked well with older process technologies and low manufacturing costs. However, development of new μC variants is significantly more costly as a result of today's advanced process technologies and higher levels of system integration.

Still, chipmakers continue to introduce new μC variants equipped with an increasing number of features to attract entire markets as they migrate to standard products rather than application-specific ones. Added features make these μCs more powerful, yet present more time-consuming design trade-offs for auto electronics designers.

At the same time, these newer μCs also dramatically increase product cost. Consequently, extra device costs make it more difficult for these μCs to serve cost-sensitive auto electronics applications. There is no solution to this issue without changing the root cause of the problem.