Microcontrollers, Sensors, Bus Networks, And RF Links To Invade Vehicles

Electronics technology is permeating nearly every aspect of the automobile. A key component is the microcontroller. We spoke with Adrian Kuzdas of Microchip Technology about the microcontroller's role in all of this.

ED:How pervasive will microcontrollers be in future automobiles?Kuzdas: Microcontrollers are replacing relays, switches, and traditional mechanical functions with higher-reliability components while eliminating the cost and weight of copper wire. Coupling an 8-bit controller with a sensor allows a designer to optimize system-partitioning decisions. Sensors typically provide an analog output that's difficult to transmit reliably in the noisy automotive environment. The controller-sensor combination enables preprocessing at the sensor location, and reliable transmission of a digital signal to a host controller.

An example of this can be found in airbag crash sensors. While accelerometers that provide a digital output are available, they're typically 20% to 30% more costly than the analog variety. Partitioning the remote sensor module to include the analog-to-digital conversion in an 8-bit controller not only reduces the cost of the accelerometer, but also builds in additional capability. Calibration and offset trim parameters can be stored in nonvolatile memory after the module is assembled, eliminating variation introduced when the sensor is mounted on the board.

ED:What trend do you see occurring in automotive bus configurations? How do CAN and LIN fit in?Kuzdas: High-bandwidth real-time-control applications like powertrain, airbags, and braking need the 1-Mbit/s speed of CAN, and their vehicle-critical nature justifies the associated cost. Some applications, though, can do well with slower, lower-cost subnetworks. In a typical driver's door, for example, various mirror control motors, switches, latch sensors, etc. can easily be managed with the 20-kbit/s LIN bus.

A vehicle could have several LIN networks. Aside from the door, LIN networks are ideal for seats, to connect switches, motors, position sensors, and heaters, as well as for environmental controls, to connect fan controls, displays, and vent doors. All of these LIN networks can be interconnected by a main CAN bus to interchange data between multiple subsystems.

A majority of active designs are based on distributed intelligence. Some manufacturers are exploring host-oriented systems using a very high-end central processor to control every subsystem. A host-oriented approach requires overcoming obstacles of complex software and reliability.

ED:Do you see embedded wireless RF technology adding a new dimension to automotive subsystems?Kuzdas: Adding tire-pressure monitors on each tire linked via RF to the automotive safety system is the first mainstream RF connectivity application. The challenge will be to come up with designs that meet the regulatory limitations imposed by the various countries' standards, like transmission frequencies and output power levels. Finally, to enable device enumeration, low-cost encryption technology is required. Imagine the potential problems of thousands of vehicles on the road together, each one rich in RF interconnectivity, and each one trying to interpret the others' RF data stream. Careful consideration must be given to encryption methodologies, to provide secure data transmission.ED:Can today's 8- and 16-bit controllers satisfy future automotive requirements, or will we need higher processing capabilities from DSPs?Kuzdas: Autos in general are getting much smarter. Airbags, for example, require extensive fast mathematical computation to address full or partial deployment and multiple-occupant detection. DSP performance is now exceeding 10 MIPS, placing advanced systems beyond the reach of 8-bit MCUs. Furthermore, complex sensor information can be assimilated using DSP technology that can push performance requirements beyond the capability of traditional 16-bit MCUs.

For most distributed nodes, eight bits will do just fine, but an increasing number of functions will require DSP functionality. What's needed is a full-function DSP with a full-function microcontroller, while meeting acceptable cost objectives. This is critical to enable the use of DSPs in closed-loop control applications, such as engine and transmission control, noise cancellation, voice recognition, active engine mounts, collision avoidance, vehicle navigation, and anti-skid braking.