As powertrain engineers race to improve mileage by about 50% over the next decade, linking many sensors and controllers together so they operate in coordination shines the spotlight on the networks that provide these links. Though bandwidth requirements will soar, the ubiquitous CAN network should meet powertrain requirements for several years.
Panelists at SAE 2013 World Congress agreed that while CAN may give up some of its dominance over the long term, it will dominate this segment of vehicle design for the foreseeable future.
“From a powertrain perspective, CAN does an excellent job. It will be with us for years,” said Philip Yuhasz, Director of Controls & Calibration at Ford Motor Co.
Other members of the panel, entitled “New Fuels and High-Tech Innovations: Are Engine Controls and the CAN Bus up to the Challenge?,” agreed with that assessment. Though FlexRay and ethernet are being discussed as technologies that boost bandwidth, panelists said that cost is one of several roadblocks that have prevented many automakers from replacing CAN’s 1 Mbit/s data rate with the 10 Mbits/s rate of FlexRay.
“Speed costs money,” said David E. Helton, Chief Engineer, Gasoline Engine Management Systems for Delphi Powertrain Systems. “If it was cheap to go to 100 Mbits/s, we would do it.”
An emerging extension, CAN Flexible Data-Rate (CAN FD), takes that up to 10-12 Mbits/s, depending on how it’s implemented. It can communicate with existing CAN nodes, so fast nodes can be linked to nodes that run at slower rates without software or hardware revisions. Helton is bullish on its potential acceptance.
“We feel it will be ready for production in vehicle year 2018,” he said. “CAN FD could be a viable solution for the next 10 years.”
Inertia is another factor that weighs heavily in CAN’s favor. Onboard diagnostic considerations will also help the network remain dominant. Regulators now specify CAN connections for OBD test equipment, which means there’s a huge installed infrastructure that will be difficult to change. As regulations evolve, CAN may be able to address growth that occurs as systems get more complex.
“Today, the number of ECUs that can talk to CAN diagnostic tools is limited to eight, which was done to reduce the chance of problems,” said Daniel Nicholson, Executive Director of Global Powertrain Drivability and Calibration & Verification at General Motors. “We’d be supportive of increasing that number; it would allow more flexibility for us. Today, more complex vehicles like the Chevy Volt have to connect secondary ECUs to a central controller to communicate with diagnostic equipment. That can be limiting.”
Looking at the long term, many panelists predicted that the coexistence between CAN and LIN will be the model for a networking scheme that also includes faster networks like FlexRay and/or ethernet.
“We’ll see CAN, multiple CAN networks, FlexRay and ethernet, we won’t see one bus architecture that will meet all needs,” said Gregory Weber, Director, Powertrain Controls Engineering at the Chrysler Group LLC. “You need multiple buses so one point of failure won’t shut the vehicle down.”
That may come as the role of software grows. Distributed computing may evolve to the point that processors throughout the powertrain run pieces of code.
“As we’ve gone to distributed control, there’s been a need for more interaction between the engine and transmission,” Helton said. “When you have a bank of ECUs and algorithms are seeking time on each of them, demand for data rate will continue to go up.”