Auto Electronics

Net Gain

OEMs’ demand for speed and bandwidth is driving innovation in vehicle network technology.

For the PDF version of this article, click here.

Automakers' increasing use of electronics to differentiate their vehicles and stimulate demand is building the industry's appetite for network capacity, speed and bandwidth. Those demands are creating opportunities for makers of mainstream CAN and LIN components as the number of in-vehicle networks increases, and they are paving the way for next-generation networks, notably FlexRay, but also MOST for infotainment and, increasingly, IDB-1394, or automotive-grade FireWire.

“The amount of electronics going into cars these days — up to 70 ECUs in some instances — not to mention the amount of data being brought into the vehicle to work with IVN systems and infotainment systems, such as through USB and, potentially, through Wi-Fi, creates a challenge for designers to develop the simplest, fastest and most robust systems for their particular application,” said Toni Versluijs, business development manager for InVehicle Network Controllers and FlexRay at Philips Semiconductors (www.semiconductors.philips.com).

“From LIN and CAN to FlexRay, MOST, and others, the various bus protocols within automobiles are growing, due to the increasing number of electronic control units in cars,” added Willie Fitzgerald, director of product line marketing in Microchip Technology's Automotive Products Group (www.microchip.com). “These protocols contribute to a complex in-vehicle networking system, as no single protocol fits all application requirements,” he added.

“The effect of design complexity on the network is a rapidly growing communications burden,” said Larry Anderson, director of marketing for the Automotive Networking business unit in Mentor Graphics Corporation's System Level Engineering division. “Incorporating multiple ECU subsystems in the overall vehicle design can result in significant problems, including information loss, unwanted delays and intermittent errors that compromise functionality and performance.”

As the network expands, testing effectiveness can be compromised by the sheer number of possible combinations of signal interactions, according to Anderson. Mentor's Volcano Communication Technology (VCT), including network architect, in-vehicle software, and test and validation, is intended to avoid such compromise.

LIN, low-speed (125 kbps) CAN and high-speed (500 kbps) CAN are currently the dominant protocols, according to Fitzgerald, with LIN and low-speed CAN used for body control applications and high-speed CAN for engine control. Fitzgerald said Microchip offers flash-based microcontrollers, digital signal controllers and transceivers to support developers of CAN- and LIN-based ECUs. “The product families range from six-pin to 100-pin devices, all of which are compatible, which helps reduce development time and costs,” he noted, “and the devices are integrated to reduce the number of external components needed and eliminate the cost of those components.”

Texas Instruments' (www.ti.com) stand-alone LIN 2.0 transceiver, TPIC1021, eliminates the need for external protection components by providing up to 17 kV IEC and 12 kV HBM ESD protection — more than that of competing devices, according to Scott Monroe, a system architect in Texas Instruments' Mixed Signal Automotive products group. Packaged in an SO8, the TPIC1021 is said to survive transients per ISO 7637, and it has fault protection to handle -40 V to 40 V on the LIN bus.

The device can be used with resistance-capacitance (RC) oscillator-based MCU systems in next-generation LIN-based applications such as door modules, window lifters, seat controls, and intelligent sensors and actuators.

“LIN not only replaces switches and power FETs, and reduces wiring, but with increased intelligence in LIN nodes, and links from LIN to CAN, it also helps improve diagnostic capabilities,” Monroe said.

The V850E/Dx3 family of 32-bit automotive-grade microcontrollers that NEC Electronics America, (www.am.necel.com) launched in January features two CAN interfaces in addition to on-chip stepper motor drivers, LCD controllers/drivers, parallel LCD bus interfaces and sound generators. The devices also include up to 16 A/D channels, an I2C bus and 17 16-bit timers.

“CAN is a mature, established standard that will be around for a long time, and LIN is a growing standard that will be used even more in the future,” said David Stone, NEC Electronics America market development manager. “Once a network has been established, it takes a lot of effort to replace it.”

Renesas Technology America Inc. (www.renesas.com) is shipping R8C/tiny MCUs (Figure 1) that feature a 16-bit core and 8-bit or 16-bit peripherals. All of the devices in the family have a LIN interface and some also offer a CAN 2.0B interface with 16 mailboxes.

The R8C/tiny MCUs target current 8-bit applications, such as body control, that can benefit from the additional speed and/or application development headroom,” according to Paul Fox, director of marketing for Renesas' automotive business unit. The devices include a LIN hardware control circuit that reduces the number of interrupts required for synchronization. The LIN hardware module also offers bus collision detection and a wakeup function. The LIN controller can be run using the internal oscillator with 5% accuracy or can be trimmed via software to 1% accuracy. The internal oscillator eliminates the need for an external clock and frees two I/O pins.

Fox said Renesas is developing 32-bit CISC and RISC microcontrollers that will support Flex-Ray. Those devices are expected to be ready in mid-2007. But cost is a primary consideration for OEMs, especially in North America, he noted, and as a result, “Time-triggered, deterministic networking technologies like FlexRay and MOST are getting less traction in North America or Japan than in Europe.

Unless they really need the bandwidth, OEMs are delaying implementation (of high-speed networks) as long as possible,” he said. In the interim, some OEMs are adopting dedicated, peer-to-peer CAN networks with speeds of 500 kb/s or more.”

Targeting high-bandwidth automotive applications such as high-resolution TFT displays in navigation systems, Texas Instruments earlier this year introduced a 10-bit LVDS SerDes device in a 5 mm × 5 mm QFN package. Monroe said the device has a point-to-point throughput range from 100 Mbps to 660 Mbps.

Philips Versluijs said his firm has developed a family of fail-safe system basis chips that simplify ECU design by integrating LIN and/or CAN transceivers, LDO voltage regulators, an SPI, a watch-dog, and diagnostic features on a single chip. The SBC chips share a common pinning and software interface and are intended to reduce time to market for OEMs developing increasingly complex network applications.

Philips anticipates a growing number of connected nodes in vehicles as well as a need for more bandwidth. “FlexRay has the potential to solve the bandwidth problem and, ultimately, enable drive-by-wire,” Versluijs said. “The need for cross-protocol communication, such as between CAN and FlexRay, creates opportunity for gateway modules in which network branches of different protocols can come together. The gateway module facilitates communication between the outside world and a vehicle ECU for end-of-line programming, diagnostics and intermediate updates.”

Philips is currently sampling a FlexRay system consisting of a two-channel, SJA2510 FlexRay 2.1 controller (Figure 2) and a TJA1080 transceiver. The 80 MHz controller is based on a 32-bit ARM968 CPU with up to 1 MB of embedded flash memory and more than 48 kbytes of SRAM. It has 32 analog inputs, and 24 16-bit pulse width modulation outputs, and can support six CAN 2.0B controllers and eight LIN 2.0 master controllers.

“The TJA1080 is one of few transceivers able to operate in node and in star mode, so it can serve as a building block for any FlexRay topology,” Versluijs said, adding that several OEMs and module makers are currently evaluating both chips.

“BMW is implementing FlexRay in several models beginning this year. Meanwhile, LIN and CAN are as popular as ever,” Versluijs said. “FlexRay will start in high-end vehicles where it will replace CAN modules. In lower-priced vehicles, in-creased node count will be fully supported by CAN and LIN until FlexRay is proliferated through all vehicle classes. This will occur over the next decade.”

Jim Shockey, 16/32-bit automotive MCU product marketer at Freescale Semiconductor (www.freescale.com), said LIN networks are being added to CAN-based systems to extend networking capabilities further down into applications that use sensors and actuators. “For example, adding a LIN sub-network to allow the body control network to communicate not only with a seat controller, but effectively all the way down to each motor and sensor in a seat.” He said. “With LIN and CAN, we're talking about adding a significantly simpler network to an existing network structure.”

Migrating from event-driven to time-driven communication, which FlexRay provides, is a paradigm shift for in-vehicle communication, according to Shockey. “It requires all involved parties to be retrained, and that will take some time, but once this first step is taken, many new application areas will be discovered,” he suggested.

“FlexRay is not a simple protocol,” Shockey continued. “The topology configurations for waking up a cluster via the communication channel and for choosing the optimal frame size demonstrate its complexities, but FlexRay's deterministic and fault-tolerant characteristics, along with its 10 bps data rate, make it a much stronger choice than current protocols to enable brake-by-wire and other chassis, powertrain and safety-related applications.”

Freescale offers FlexRay devices, such as the MFR4300 stand-alone controller, based on its 16-bit HCS12 microcontroller family. In the near future, Freescale intends to release FlexRay devices based on the 32-bit power architecture. Freescale is working with a third party for licensing the FlexRay IP for use in custom implementations. “At this point, the emphasis for semiconductor manufacturers is driving down the cost of FlexRay systems by integrating the FlexRay controller into the MCU,” Shockey said. “The FlexRay consortium is considering various approaches to using FlexRay that would allow simpler and less expensive implementations.”

Last year, Freescale introduced integrated and stand-alone FlexRay controllers, MC9S12XFR and MFR4300, based on FlexRay version 2.1 protocol and Freescale's 16-bit, 40 MHz S12X core. The devices, which provide serial communication at up to 10 Mbps on each of two channels, target chassis control, body electronics and powertrain applications that require increased functionality and onboard diagnostics.

Last fall, Fujitsu Microelectronics America (www.fma.fujitsu.com) introduced the MB88121, a FlexRay controller initially featuring version 2.0 FlexRay IP developed by Robert Bosch GmbH (www.bosch.com). The firm plans to support FlexRay 2.1 IP and to introduce a FlexRay device this year based on its 32-bit RF core.

Mounted in a 64-pin LQFP package, the MB88121 delivers 10 Mbps over two channels, providing fault-tolerant, deterministic data transmission suitable for engine control and braking and steering subsystems. Internal speeds reach 80 MHz, with a 4 MHz, 5 MHz, 8 MHz and 10 MHz external oscillator or by external clock. The chip's parallel interface offers a maximum frequency of 33 MHz.

Akio Nezu, director of microcontroller products at Fujitsu Microelectronics America, said FlexRay is expected to be phased in over a long period of time, with the first applications expected by the end of 2007. “Initially, FlexRay will work in conjunction with the CAN bus,” he said, adding that CAN and LIN networks continue to perform well, with double and triple CAN architectures used for engine control, and CAN also used to link body control modules. LIN networks address specific 12 V applications such as air conditioning and door locks, he said.

MOST is also expected to expand slowly and its growth may be hindered by competition from IDB-1394. Complete standardization of MOST technology is expected to pave the way for deeper and wider penetration of infotainment systems in North American vehicles,” said Sandeep Kar, industry analyst in the Automotive and Transportation Group at Frost & Sullivan (www.transportation.frost.com), but he cautioned, “At this early stage of adoption of CAN, a rapid transition by North American automakers to the MOST protocol seems unlikely. There also are concerns regarding the fault-tolerance attributes of the MOST protocol, and automakers are concerned that a single faulty system in a MOST network could bring down the entire network.”

Nevertheless, Kar estimates that by 2012, 27.5% of all vehicles sold in North America will feature MOST networks for supporting infotainment applications. “The implementation of MOST over copper wires will significantly shift the focus from 1394 to MOST and establish it as the most pertinent protocol for multimedia applications,” he said. “The economies of scale resulting from volume deployment of MOST will facilitate penetration of the protocol in North American and Asian automakers' mid-market vehicles.”

Henry Muyshondt, director of business development for the Automotive Infotainment Systems group at SMSC (www.smsc.com) and technology coordinator for the MOST Cooperation (www.mostcooperation.com) estimated that 37 vehicles are equipped with MOST networks and nine more such vehicles are expected to be introduced this year. SMSC offers an intelligent network interface vontroller, OS81050 (Figure 3), which eases system development by allowing devices attached to the network to register with the network but otherwise function independently. Muyshondt added that faster MOST chips are also available.

Several firms, however, have introduced automotive-grade 1394 chips to address infotainment and other applications.

“Nissan and other automakers believe that new applications will require multiple connections to and from several media sources and sinkers,” said Ricardo Wong, manager of the Advanced Planning Group in Nissan Motor Co. Ltd.'s Electronics Engineering division (www.nissan.com). “For example: a display needs to be connected to several sources, including cameras, a navigation system, and DVDs, and a DVD source needs to be connected to several displays.” As a result, Wong said, migration to digital/multichannel-capable networks “is more important than before.” Wong, who worked on the AMI-C 1394 specification, added that 1394 has sufficient bandwidth and channels to support such networks.

“Nissan and other automakers have built demonstration vehicles with consumer-type 1394 products, but since Fujitsu came out with an automotive-capable 1394b PHY chip that operates just like 1394 chips for consumer or computer applications, the way is clear for carmakers and tier one suppliers to start developing complete systems,” said Max Bassler, president of Interactive Technology, a consulting firm (www.interacttech.net), and a founder and former chairman of the 1394 Trade Association (www.1394ta.org).

Bassler said Texas Instruments is in production with an automotive-rated 1394b device, the TSB-41BA3A, a three-port cable transceiver/arbiter. Companies, such as Yazaki (www.yazaki-na.com), Sumitomo Electric (www.sumitomoelectricusa.com) and Molex (www.molex.com) offer plastic optical fiber and/or high-speed copper connectors for IDB-1394 automotive networks. Some, like Yazaki, support MOST as well as IDB-1394.

Fujitsu Microelectronics America's MB88387 (Figure 4) is an IDB-1394-compliant (intelligent transportation system data bus-1394) device that combines a 1394b PHY, link, and digital transmission content protection (DTCP) in a single chip. Designed for in-vehicle audio-video multimedia entertainment systems, the MB88387 connects video and audio to vehicle displays from head units, DVD players, navigation systems, cameras, amplifiers or other devices and supports IEC61883 AV protocols for smooth audio and video stream playback.

The MB88387, in a 176-pin, 24-mm square LQFP, supports data transfer rates up to 400 Mbps and operates at 3.3 V (I/O) with internal voltage of 1.8 V. Its PHY layer supports two 1394b/S400 ports, and its DTCP function allows encryption or decryption of two datastreams at once. Systems networked using the MB88387 can transmit audio and video throughout the vehicle at distances up to 100 meters.

Texas Instruments' TSB41BA3A provides the digital and analog transceiver functions needed to implement a three-port node in a cable-based IEEE 1394 network. Each cable port incorporates two differential line transceivers with circuitry for monitoring line conditions as needed for determining connection status, for initialization and arbitration, and for packet reception and transmission. The TSB41BA3A is designed to interface with any of several TI link-layer controllers or it can be connected cable port to cable port to an integrated 1394 Link and PHY layer such as TI's TSB43AB2.

“It is early in the cycle, and 1394 controllers are just now being analyzed for their application,” said Fujitsu's Akio Nezu “They will see some cost reduction as designs are implemented and volumes are increased.”

ABOUT THE AUTHOR

John Day writes regularly about automotive electronics and other technology topics. He holds a BA degree in liberal arts from Northeastern University and an MA in journalism from Penn State. He is based in Michigan and can be reached by e-mail at [email protected].

Hide comments

Comments

  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Publish