Volvo_Promo_Autonomous

NXP Semiconductors Buttresses Automotive Ethernet Bet

Sept. 5, 2018
NXP Semiconductors Buttresses Automotive Ethernet Bet

No one knows what it will take to build the autonomous car of the future. But almost everyone knows that it will require many cameras, radar and other sensors to enable the vehicle to pinpoint its location and enormous computing power to make critical driving decisions. An unsolved problem is making sure that all the information inside the vehicle gets where it needs to go when it needs to get there.

To enable advanced driver assistance systems and infotainment systems—in addition to existing functions—cars will need to move information with lower latency and higher bandwidth. Many companies are trying to expand Ethernet’s uses for automotive applications, taking advantage of recent improvements to its timing and reliability. But before the standard takes over the car’s network, it needs to get faster.

NXP Semiconductors, the world’s largest supplier of automotive chips, is doubling down on the market. On Tuesday, the company announced its acquisition of networking IP startup OmniPHY to boost the bandwidth of its Ethernet solutions. The acquisition complements NXP’s other products based on existing standards including CAN, LIN and FlexRay. The financial terms of the deal were not disclosed.

“Traditionally, a lot of the data inside the vehicle was slow,” said Alexander Tan, NXP's vice president of automotive Ethernet solutions, a role created when he was hired last year after holding senior positions with Marvell and Texas Instruments. “The realization was that Ethernet is not just another component,” he said. “Now that you have all these connected domains there is the need to take all that data and aggregate and share it on a backbone.”

The CAN standard, which is used to connect door controllers and transmissions systems, ranges from bandwidths of less than one megabit to 10 megabits per second. LIN, which stands for local interconnect network, targets applications like adjustable side mirrors. Other vehicle networking technologies include MOST and LVDS. Ethernet is currently used inside cars to support 100 megabits to 1000 megabit per second networking.

Bandwidth requirements are mounting as more systems inside the vehicle connect to the same sensors and swap information with each other and the cloud. When pulling out of a parking spot, for example, the rear camera streams information not only to the dashboard but also to the collision avoidance system. That way, the car can brake automatically in case another car or pedestrian comes out  from a blind spot. This is where Ethernet can make the biggest difference.  

The acquisition of OmniPHY, which employed about 100 people and was founded by CEO Ritesh Saraf and CTO Claude Gauthier in 2012, is an attempt to augment its line of automotive Ethernet chips. The company said that it would focus on the development of 1.25 to 28 Gbps PHY technology and 10-, 100- and 1000BASE-T1 Ethernet in advanced processes. NXP currently has 100-megabit-per-second PHYs and switches in production.

The bandwidth requirements of cars are projected to skyrocket over the next decade, industry analysts say. "Cameras and displays will ramp the number of high-speed links in the car to 150 million by 2020 and by 2030 autonomous car systems will aggressively drive that number to 1.1 billion high-speed links," said Ian Riches, an automotive technology analyst for Strategy Analytics, in a statement.

Higher speeds are also required to support the car’s changing electronic architecture. Modern vehicles have miles of cables crammed inside door panels and behind the dashboards, connecting electronic control units that handle functions ranging from windshield wipers to speed gauges. Over time, many of these functions may be consolidated into central computers that act like the brains of the vehicle. And that will fundamentally change how cars are wired.

“The primary driver of this market is reducing the size and number of cables inside the vehicle,” Tan told Electronic Design. “What happens with the cables drives requirements for the silicon and what works and what doesn’t work in the vehicle. The wiring harness is one of the most expensive components in the car. It is also one of the heaviest.” The vehicle’s weight is a major factor on the range of electric vehicles.

The market for networking chips is intensifying. San Jose, California-based Aquantia recently introduced a new line of networking chips with Nvidia that transmit 10 gigabits per second of information over one pair of copper wires inside a single cable. The company, branching out from the enterprise sector, said that the Ethernet switches could be used to enable chip-to-chip communication on the same or separate boards.

Valens Semiconductor is targeting another standard, HDBaseT, which encodes rather than simply transmit information within the vehicle. The technology can transmit Ethernet, control, USB, audio and video and a hundred watts of power over a single copper cable simultaneously. This is the same type of cable Broadcom, Marvell and every other company is trying to pipe information through to cut costs and weight.

Networking silicon that can support bandwidths of over 100 gigabits per second are available for use in data centers. These chips are commonly used to test autonomous car prototypes but are too inefficient and too expensive to be installed in production cars. The other problem is related to the cabling: the higher the bandwidth, the more pairs of wires and the thicker the rubber shielding are required to protect the signal.

Most driverless cars have a data center in the trunk with switches and routers and then a huge harness with cables going to all the sensors around the vehicle,” Tan told Electronic Design. “Right now that’s not really a problem because the people building them are not trying to make them cheap or deploy them on the road.” He added: “You need to move away from these enterprise solutions with data center components to ones made specifically for automotive.”

That means building networking chips protected from electrical faults resulting from high temperatures, vibrations and other factors in automotive applications where high reliability is mandatory. That means using analog and digital signal processing to guard signals traveling down thinner cables from electromagnetic interference from radio frequency electronics inside the car. That also means lowering power consumption.

“The OmniPHY acquisition gives us the technical capability to take the next step in terms of the architecture of the cars,” Tan told Electronic Design. “There are not many teams in the world with that kind of experience. What was really important to me was bringing these capabilities in-house.”

About the Author

James Morra | Senior Editor

James Morra is a senior editor for Electronic Design, covering the semiconductor industry and new technology trends, with a focus on power electronics and power management. He also reports on the business behind electrical engineering, including the electronics supply chain. He joined Electronic Design in 2015 and is based in Chicago, Illinois.

Sponsored Recommendations

Comments

To join the conversation, and become an exclusive member of Electronic Design, create an account today!