Silicon Fuels the Automotive Industry

Jim Lipman

March 29, 2004

Jim Lipman

While hydrocarbon-based fuels make cars go, the automotive industry moves forward on a very different fuel—silicon chips. Advancements in chip technology are widely acknowledged as a major driving force in the communication, computer, and consumer-appliance market segments. These advancements, along with their convergence, are also transferring what you drive from simply cars to get from place-to-place to sophisticated personal-transportation vehicles providing extraordinary levels of efficiency, safety, comfort, and convenience.

This ongoing transformation is marked by several trends in automotive development, predominantly in the area of automotive electronics and, to a lesser extent, with the car's power source. The number of silicon chips in today's car is staggering. Ron Spence, Motorola's strategic marketing and senior systems engineer for the company's transportation standard-products group estimates as high as 30-70 microprocessors per vehicle, many of them 32-bit units. Microprocessor transistor count has shot up from about 1-million transistors in 1997 to as high as 30-million transistors for today's vehicles. The amount of memory associated with each microprocessor has also risen sharply in recent years—from around 256 Kbytes in 1996-97 to 2 Mbytes or higher today. If you think that the processing is just necessary to support today's highly sophisticated engines, think again. According to Spence, one microprocessor is dedicated to engine control—the others deal with all other facets of the vehicle. The value of the total semiconductor content the automotive industry uses is growing dramatically as semiconductors address new and old applications—from around $3B in 1992 to $11B in 2003.

In the ongoing drive for more fuel efficiency and lower pollution levels, the catalytic converter in the mid-1970s was a huge advancement. Processor-based engine-control subsystems go much further. Today's engine controllers try to balance fuel and air combinations to optimize gas mileage and minimize emissions using lookup tables with several thousand entries. According to Spence, there are many other factors that need to taken into account to truly optimize engine performance, including temperature, humidity, vehicle weight, speed, engine age, and gasoline quality. Future microprocessors will dynamically deal with all these contributing factors, adjusting engine parameters in real time. Spence notes another innovation in the wings for engine-operation optimization—variable valve control on a cylinder-by-cylinder basis. For example, such a subsystem could compensate for a sticky valve or out-of-spec fuel injector.

	Driving by-Wire The term "by-wire," also known as "X-by-wire," refers to the replacement of your car's mechanical and/or hydraulic subsystems by electric/electronic controls. Examples of subsystems where by-wire control development is occurring include throttle control, steering, and braking. Interfacing the driver and subsystem with a processor-controlled by-wire system results in less weight, more design flexibility, decreased manufacturing complexity, and higher reliability. Throttle by-wire puts an electrical connection between the throttle valve and gas pedal in place of a traditional cable. Electronic throttle control is more efficient than traditional throttle systems and has the added benefit of enhancing some of your car's safety systems, such as electronic stability and traction control. Steering by-wire replaces the steering column with a fault-tolerant controller and motors that connect to the steering rack to control direction. This type of system is safer than mechanical steering, and improves "road feel" and gas economy. Steer by-wire systems also give automobile manufacturers more flexibility in vehicle design. Some analysts claim that by 2010 1/3 of all new cars will have steering by-wire. A brake by-wire system, also called an electromechanical braking system (EMB), replaces a conventional hydraulic braking system with a non-fluidic electrical-component system that connects the four brake "corners" of your car to the brake pedal and to each other. This system provides better driver control during braking situations, features more uniform brake-force distribution, and improves traction and stability. Without the mechanical or hydraulic backup inherent in a hydraulic automotive braking system, brake by-wire requires very high reliability and, like a steering by-wire system, must be fault tolerant. Brake by-wire implementation requires a fault-tolerant communication protocol, such as FlexRay, a dependable power source, and some type of hardware redundancy. The TDMA-based FlexRay is attractive, since it comprises two separate 10 Mbps lines that supply redundant and fault-tolerant messaging capability.

There are many advantages in replacing your car's mechanical and hydraulic subsystems with electronic ones (See Sidebar, "Driving by-Wire"). Electric systems are inherently more reliable, more efficient, add less weight to a vehicle, and can offer more functionality than can mechanical systems. Lighter weight and more efficiency equate to better fuel economy—an average of 5% improvement over traditional systems, according to Texas Instruments—a factor on everyone's mind in these times of skyrocketing gasoline prices. However, the mechanical-to-electric shift is hampered by several factors, including the inertia of the automotive industry and your car's available power source.

A place where the mechanical-to-electric transformation is taking place, albeit slowly, is in your car's power steering. Some smaller cars in Europe appeared with electric power steering back in 1998, with volumes increasing steadily as European automotive manufacturers adapted the new technology. By using an electric motor to replace the traditional combination of hydraulic pump, fluid, hoses, and fluid reservoirs, you have a system that, while not significantly less expensive to produce, is smaller and lighter. In addition, the electric system is powered by the car's battery, not the engine, which adds to fuel-consumption efficiency and gives the automobile manufacturer more options in configuring the steering system (Figure 1).

Figure 1:  Replacing a traditional hydraulic steering system with an electric motor, as shown here, reduces the system's size and weight. The electric system also improves fuel efficiency and gives the car manufacturer more design flexibility. (courtesy Xilinx)

One barrier to adaptation of electronic power steering is the strain put on traditional 12V automotive power systems (See Sidebar, "When Will 42V be in Your Car?"). Jennifer Skinner Gray, TI's worldwide marketing manager for the company's C2000 digital signal controller, attributes the faster adaptation of electric power steering in Europe to the region's preponderance of smaller cars. Most current electric power-steering systems have relatively low output, which precludes their use in heavier vehicles, such as those loved by Americans. Powering higher-output steering systems requires a higher voltage, such as the developing 42V system.

Gray describes the C2000 as a 32-bit microcontroller with DSP attributes, ideal for automotive applications that need motor control, such as electric power steering. Other evolving automotive applications targeted by the C2000 include an integrated starter/alternator and a brushless fuel pump. The starter/alternator combination reduces part count, cuts fuel consumption, and lowers car emissions. Another advantage of this subsystem is fast vehicle starting—under 200ms—a key benefit for use in hybrid gas/electric and other vehicles that turn off the engine during stops to save fuel. A brushless fuel pump offers higher reliability, better efficiency, slightly better fuel economy, reduced current requirements during startup, and can use a cheaper motor than the traditional automotive fuel pump.

Figure 2:  A traditional copper-wire wiring harness is complex to assemble and install, takes a lot of room, and adds significant weight to a car. (courtesy Yazaki)

	When Will 42V be in Your Car? So—where is the higher voltage system automotive experts say we need to keep up with the many evolving automotive advancements? According to Joe Notaro, ST Micro's market development manager for the company's automotive business unit, the slowness of 42V adaptation is not surprising. Notaro indicates that the transition from 6V to 12V in the 1950s was the culmination of 10-12 years of effort. The technical difficulties of transitioning to a 42V system are huge—for example, what about existing 12V components, both electronic and electric, including electronic modules, lamps, switches, and motors? Another problem is the high-voltage surges that come with on/off engine cycles—as high as three times the supply voltage. Nevertheless, Notaro does feel that 42V will come, but in 2007 or later. So, why do we need a higher voltage system? The answer is simple—more demands on the car's power source from the engine, and from subsystems for improving safety (such as collision avoidance), providing communications and entertainment, and adding luxury (such as heated seats and power doors). Replacing hydraulic and mechanical subsystems with ones that are electric/electronic ("by-wire"), including steering, brake, and valve-control subsystems, also strain current 12V systems. Today's cars typically need around 3.5kW—a full by-wire car could require up to 20 kW in a 42V system. Despite having to overcome the huge obstacles of the existing 12V infrastructure, the need for a comprehensive 42V standard, and the inherent inertia to extensive change exhibited by the automotive industry, 42V systems will allow for more vehicle electronics and, in some cases, less expensive components. For example, transitioning from 12V to 42V results in a three-fold current reduction on the system, allowing the use of cheaper and smaller power semiconductors. It's not a question of if a 42V system will be implemented, but rather when.

Shed a few pounds and you run more efficiently—the same is true of your car. Fuel efficiency is directly affected by a car's weight—make the car lighter, other factors remaining constant, and gas mileage goes up. For a typical car, 300 additional pounds requires an extra gallon of gas on a 500-mile trip. The cable harnesses (including wire shielding) and associated connectors add significant weight to your car—several hundred pounds (Figure 2). Reduce this poundage, approximately 5-10% of the automobile's total weight, and you improve gas mileage.

Moving to in-vehicle networks along with newer buses and data-signal protocols, such as transitioning from CAN (Controller Area Network) to FlexRay, can shed significant pounds. The increased use of fiber-optic networks in place of copper wires also results in less weight. Optical data buses are already in widespread use in Europe, with over 5 million nodes installed in 2003. Other advantages of fiber optics over copper networks are less electromagnetic radiation, smaller size, and higher data rates.

Gaining a foothold in vehicles is Media Oriented Systems Transport (MOST)—a multimedia fiber-optic network developed by the automotive industry. MOST is optimized for telematic multimedia applications and complements other automotive networks, such as FlexRay.

A simple definition of telematics, provided by Brian Fortman, Texas Instruments' worldwide telematics marketing manager, is a means to enable data to and from the vehicle. What this means to the automotive manufacturer is providing what is needed to control data flow to and from the car. Traditionally, this covered mostly consumer entertainment and information appliances such as a radio and embedded cell phone and, more recently, the On-Star communication system, linking the car and driver to outside information and a variety of services. Future vehicles will sport telematics control over literally dozens of data channels, covering the human/machine interface, the automobile's internal buses, data interfaces, and location information (Figure 3).

The "brain" to control tomorrow's automotive telematics controller will have to be sophisticated, with medium-to-high compute power, depending on the feature set it will support (which usually tracks with vehicle cost), and lots of memory. Fortman explains that voice recognition will be an integral part of the controller's capability, since using voice commands helps keep a driver's focus on driving the car. Bill Fleck, Motorola's business development manager for the company's MPC5200 embedded processor, adds another important capability, video-processing support. Fleck sees a rapid proliferation of camera-based technology in future automobiles, used for applications such as collision avoidance, back-up assistance, and lane guidance.

Along with high-performance chips, telematics control will require equally sophisticated software, including high-level operating systems and application software that can run in real time. According to Fortman, only controllers utilizing RISC microprocessor and DSP combinations will be able to reach the level of support you'll need to make everything work.

However, Xilinx's Robert Bielby, senior director of strategic marketing, disagrees with Fleck's assessment. Bielby feels that the rapid introduction of new automobile-related services, shrinking vehicle-development cycle, and need for in-the-car upgradeability make FPGAs an ideal choice for several automotive applications, particularly those involved in engine control and multimedia-based infotainment systems. According to Bielby, programmable devices accelerate time-to-market and lower component development risk. The high flexibility and reconfigurability associated with FPGAs are ideal for addressing such issues as shifting and evolving automotive network standards, and the need for a common platform for a particular application that can be configured to match varying car types and models along with driver needs.

Figure 3:  A telematics controller handles all data flow into and out of the vehicle as well as oversees the car's internal data networks.

With future vehicles containing hundreds of microprocessors and DSPs, several high-speed buses, wired and wireless connectivity, and scores of high-complexity electronic modules, you might ask yourself if you are going to be driving a computer on wheels (Figure 4). Along with the very difficult technological barriers automotive OEMs and their subcontractors face is the obstacle of having to train automotive diagnostic and repair personal to deal with the high electronics content of tomorrow's cars. This is not an insignificant task, since a car crash caused by a component failure has far more serious consequences than when your computer crashes. It remains to be seen if tomorrow's driver, seeking repairs, will go to In-and-Out Auto Service or CompUSA.

Figure 4:  With future cars taking on aspects on a computer on wheels, automotive microprocessors are critical for making sure that all systems function properly and communicate with each other correctly. (courtesy Texas Instruments)