Automakers asked to can CAN
PARK RIDGE, Ill. A consortium of the world's biggest auto and chip makers moved last week to make a safety-critical data bus the main electrical backbone for tens of millions of future vehicles.
Members of the FlexRay Consortium said they will push to use their time-triggered architecture to control powertrain, chassis, airbag and drive-by-wire systems, in addition to having their technology serve as the primary cable that ties all vehicle data buses together. The move to FlexRay would represent a startling change for the auto industry, which in the past few years has invested tremendous time, money and effort in developing the controller-area network (CAN) buses for vehicles.
"It's going to be tough to make this massive of a change," said Benjamin Baker, director of the Electrical Center for General Motors' North American Car Group (Warren, Mich.). "But eventually CAN will run out of horsepower, and we'll need a replacement."
Consortium members estimated it would take about a decade to effect the changeover.
Although the consortium's plan is considered far-reaching and ambitious, industry analysts said it must be taken seriously, mainly because of the enormous influence of FlexRay's member companies. FlexRay includes the world's three biggest automakers General Motors, Ford and DaimlerChrysler along with the largest automotive semiconductor supplier, Motorola Inc. FlexRay also counts such industry heavyweights as BMW, Philips Semiconductors, Bosch Automotive Group and Texas Instruments Inc. among its members.
Nonetheless, the move to FlexRay is seen as a costly, uphill battle for the auto industry. FlexRay, launched in 2000 as a fault-tolerant solution for drive-by-wire technologies, is more expensive than CAN. Automotive engineers, for whom a few pennies can make a difference in high-volume production decisions, are expected to balk initially at FlexRay's higher costs.
The move to FlexRay is also expected to be especially difficult, given how much the industry has already invested in CAN chips and other CAN hardware. Many automakers have just begun putting CAN-based systems into production and a broad array of vehicles will incorporate their first CAN systems in the next few model years.
"A lot of engineering work will have to be dumped in order to move to another new network," said Paul Hansen, publisher of The Hansen Report on Automotive Electronics. "Automakers have already spent tens of thousands of man-years bringing CAN out."
Still, automakers said last week, they foresee a need for a more powerful data bus with inherent fault tolerance and higher bandwidth. Data buses, which pass electronic signals between a vehicle's microprocessors, are growing more important as automakers incorporate more processors and develop more interdependent systems. Some luxury vehicles now incorporate as many as 90 microprocessors. Moreover, many of the subsystems, such as powertrain, brakes and suspension, must be able to "talk" over a central electrical backbone.
The need for such communications will be accentuated in coming years as automakers migrate toward steer-by-wire, brake-by-wire, throttle-by-wire and suspension-by-wire systems, which replace mechanical components with electronic control. Those systems are also expected to work with radar devices and controllers that do automatic lane-keeping, adaptive cruise control and crash sensing.
Because of the critical nature of those functions, engineers said they will need a central data bus with high bandwidth and, in particular, fault tolerance.
FlexRay is considered a strong candidate for such safety-critical applications because its time-triggered architecture ensures that there is always a slot for important messages. It thus provides a level of redundancy for steering, brakes and other systems that will no longer have the inherent redundancy of traditional hydraulics.
FlexRay also offers greater bandwidth than most existing systems. CAN has a data rate of 1 Mbit/second; each of FlexRay's two channels offers 10 Mbits/s.
The real surprise in FlexRay's announcement, however, was not that the time-triggered bus would be used for drive-by-wire apps but that the automakers plan to employ it in powertrain, chassis control and passive safety. CAN is today's bus of choice for powertrain and chassis. And most passive safety applications are discretely wired, so they use no central data bus.
The move to FlexRay may be seen as high-priced overkill in some quarters. "But over the long haul, the question is: Will we need a backbone like this? And the answer is, yes, we will," said Baker of GM.
Engineers see a need for a fault-tolerant, high-bandwidth bus, even in the operation of engines and transmissions. Over the next decade, many expect engine designers to move toward a system of electronic valve control, in which traditional mechanical camshafts will be replaced by electronic controllers and electric motors. Such applications will require higher-bandwidth, fault-tolerant data buses.
"If you simply compare a point-to-point CAN connection to a point-to-point FlexRay connection, then FlexRay is more expensive," said Andreas Both, FlexRay business and technology manager for Motorola Inc. (Munich, Germany). "But there will be applications where higher data bandwidth or fault tolerance is needed and this will be the most cost-effective solution."
Striving for simplicity
Engineers also said broader use of the FlexRay architecture would simplify development for vehicle engineers who must now deal with a broad array of buses, including J-1850, Byteflight, LIN, MOST, IEEE 1394, Safe-by-Wire, CAN and others.
By replacing CAN, J-1850 and Byteflight in some applications, engineers say, they could cut the time and development costs and eliminate a number of gateways.
"Vehicles have evolved with multiple data buses," said Baker of GM. "With so many data buses, it becomes too expensive and too difficult for automakers to make changes. We'll probably still end up having more than one data bus per vehicle, but our goal is not to have five, or six, or seven anymore."
Engineers said they expect the low-speed LIN bus to continue to control windows, seats and mirrors. They also forecast a continued, but diminished, use of CAN in chassis and powertrain and a continued big role in telematics and automotive multimedia for the fiber optic MOST bus.
But BMW engineers said they plan to replace Byteflight with FlexRay in passive safety apps.
"This is an advantage for all of us automakers, semiconductor manufacturers and tier-one suppliers," said Harald Heinecke, manager of systems design at BMW's research and development facility (Munich).
Analysts expect the suppliers of FlexRay chips and hardware to be some of the same companies that now supply CAN chips. Bosch, which pioneered the development of CAN, is part of the FlexRay Consortium, as are Motorola and Philips Semiconductors, which sell CAN hardware.
If FlexRay does meet resistance, observers said, it will likely come from those engineers who have already invested thousands of hours in the development of other data buses.
"This is the same situation we had six years ago, when CAN was introduced," said Heinecke of BMW. "Everybody said CAN was too expensive . . . but now most automakers are using it."