The increasing use of electronics in automobiles faces hurdles as the continuing need for more performance confronts the reality of the power required to achieve that performance. With 90% of the innovation in the automotive sector and 35% of the average cost of a vehicle already being attributed to electronics and software, different approaches to increase performance are required for the future.
Car companies and their suppliers all the way down to the silicon/semiconductor level have already started to address the issues. Body electronics is a specific area that adds to vehicle differentiation and continues to expand rapidly. Every new generation of vehicles has an increasing number of body electronics nodes.
Body electronics systems cover a broad variety of applications inside the passenger compartment, including comfort (HVAC, window lift), safety (advanced lighting, rain-light sensors), and vehicle networking (body control module (BCM), gateway). With many diverse subsystems involved, body electronics components connect to each other and other systems using several vehicle protocols including CAN, LIN, MOST, FlexRay, and even Ethernet. In addition to other functions, a BCM can address the transition between protocols and perform as a central gateway, where fast, reliable communication between vehicle subsystems is essential. However, even the most advanced, single-core 32-bit microcontrollers (MCUs) have insufficient performance to meet the increasing requirements of high-end, next-generation vehicles.
The historical approach to improved processor performance has been to increase the frequency. This well-known solution has two problems at today’s performance levels. With MCUs operating at 80 MHz, an increase of 20 to 30% would put the processor’s speed in the FM radio band of 87.5 to 108.0 MHz. For obvious interference reasons, this must be avoided in vehicles. A 50% improvement would put the operating frequency at 120 MHz. While this is achievable, the increased power requirement becomes prohibitive.
As shown below, because of the direct relationship to frequency, the 50% improvement in frequency results in a 50% increase in power. This power level would require a more expensive power supply and a more expensive packaging solution. The increased power requirements would mean a transition from lower cost linear regulators to a more expensive switched-mode power supply to avoid excessive heat dissipation in the power supply.
Increasing performance requires more power and a single central processing unit (CPU) would exceed the power limit to achieve next generation performance levels in body controllers.
Coping with the increased power consumption in the MCU would require both more expensive MCU packaging and a more expensive printed circuit board to dissipate the increased heat. Fortunately, there is an alternate solution to increase performance that stays below the critical power dissipation limit.
Rather than reinvent the wheel, a good place to investigate higher performance processing is the computing and networking areas that have already confronted this issue. The use of multi-core processors has been implemented in high-end desktop computers and servers for many years. Networking has also taken advantage of the improved performance from using multiple cores. For example, the QorIQ
communication processors from Freescale Semiconductor
run at frequencies up to 2.2 GHz and have as many as eight cores in a single package for base stations.
With a higher number of cores, the package must dissipate 30-40 watts compared to common automotive 1-2 watt packaging technology for MCUs. Of course, a lower number of cores consume less power. Also, 2.2 GHz is well above automotive requirements. In any case, multi-core architectures are the answer to manage power and increase performance.