Controlled by a computer that is able to not only monitor what is happening inside the vehicle but the driving environment around and in front of it, the car of the future will be a fusion of electronic senses. And what it learns will influence much more than the car’s steering.
By Simon Duggleby, Technical Marketing Manager, Electronics, RS Components.
Imagine the scenario where you are driving a car downhill on a series of bends at reasonably high speed. Ahead and out of sight, a set of traffic lights sends the message that they will show red shortly and will hold the vehicle up. The message is relayed by another vehicle as it passes in the opposite direction. Having received the message, the car shifts to electric-only traction instead of allowing the combustion engine to charge the battery.
It does this once it has determined it has sufficient charge to keep going until the lights turn green and the main engine can kick in. Regenerative braking gradually slows the car, harnessing kinetic energy for temporary storage. As the car approaches the lights it slows. It waits for a short time and then restarts the combustion engine to pull away more easily.
Thanks to the advance information, the car’s systems cooperate to use the least energy possible to execute the manoeuvre. The car might not even have to wait for the green light – it changes as the vehicle approaches. But the car benefits from the savings made by not engaging the combustion engine unnecessarily and by taking advantage of the terrain.
Through such advance planning, vehicle makers expect to both improve the driving experience and make vehicles much more energy efficient than is possible using conventional engine-management techniques. To make this future possible demands cooperation between numerous electronic subsystems. The cooperation extends from external sensors to power transistors deep in the powertrain.
Powertrains are evolving rapidly as electrification takes hold in car design. One of the consequences is an increase in the voltage used to distribute electrical power around the vehicle. Having been stuck at 12V for decades, systems are moving to 48V and, in the engine, beyond. This increase in voltage reduces waste energy and allows the use of more efficient components in a wide variety of subsystems.
For example, High-Intensity Discharge (HID) headlamps and direct fuel-injection systems use supplies of 100V or more. Localised power electronics, such as high-voltage MOSFETs or IGBTs and their associated gate drivers, can be used to supply them. But a move to higher distribution voltages will reduce overall component count.
The incorporation of AC motors for traction will drive demand for power transistors able to block voltages of more than 1kV while they operate at several hundred volts. IGBTs, such as those made by Infineon Technologies and Semikron, offer the necessary breakdown-voltage and power-handling capability to support these new systems.
The key requirement is for a bidirectional AC/DC traction system that combines an AC electric motor with an inverter and battery charger circuit. In motor mode, an array of IGBTs can channel 100kW or more through the inverter module to the motor from the battery or the combustion engine.
When braking, the still-rotating AC motor acts as a generator, channelling power back to the battery. Other energy-recovery systems, in the exhaust, for example, deliver other opportunities for dealing with waste heat efficiently.
High-performance MCUs are required in the Electronic Control Units (ECUs) to handle complex signal processing inside the motor and inverter control units. Algorithms such as Kalman filters and the Park-Clarke space-vector transform help to ensure that the fluxes in the three phases of the motor are estimated correctly and balanced appropriately as the rotor turns. MCUs with Digital Signal Processor (DSP) engines provide the necessary horsepower for these functions.
For example, the TMS320F2806x Piccolo includes a complex-maths unit that accelerates the type of arithmetic needed to process the real and imaginary current components encountered in space-vector transformations. Through the automotive network, the MCUs take messages from higher-level computers in the vehicle. The messages contain information about road conditions ahead that will determine how energy will flow into and out of the motor.
The heart of the system delivering that information will be the Advanced Driver Assistance System (ADAS). The ADAS is a sophisticated multiprocessor platform. It pulls together data from multiple sensors around the car, transmissions from other vehicles and roadside devices as well as the car’s internal systems. Platforms such as the Renesas R-Car SoC family combine 32 or 64bit ARM processors with 3D graphics acceleration and video processing.
The graphics engine is important in applications such as 360° vision. Several cameras with wide-angle lenses mounted around the car deliver discrete images that need to be stitched together. The graphics processor provides the necessary image warping functions to provide coherent video of the environment around the car to the driver.
As cars become more automated, they need to perform functions such as localisation: the ADAS uses cameras and other sensors to marry what it sees with internal maps that allow it to plan routes down to centimetre accuracy. As well as supporting safe driving, localisation is a key contributor to the vehicle’s ability to minimise energy usage during a trip. It can look at the road ahead and make decisions about power requirements well in advance, predicting beneficial gearshifts and acceleration/deceleration profiles.
Linking these subsystems together will be a multi-tier, segmented network designed to support real-time control without adding excessive cable weight to the car. The top tier used for networking will be a form of Ethernet optimised for the harsh electrical environment inside the vehicle as well as real-time communication. Ethernet AVB, supported by devices such as the Renesas R-Car platform, has the ability to guarantee timeslots for important messages, such as those to and from the ECUs responsible for motor control.
To handle electrical interference, the 100Mbit/s automotive Ethernet protocol employs specialised signal conditioning based on the OPEN Alliance standard. As cameras and radar sensors become more sophisticated, it is likely that 1Gbit/s PHYs will be adopted over time.
However, the ability to use switching on a tree-like architecture will allow bandwidth to be deployed where it is needed in the short to medium term. This is in contrast to MOST, which demands devices attach to a single 150Mbit/s ring. To ensure signal integrity, the automotive networking interfaces employ connectors that are far more robust than those used for office-IT systems.
Gateways controlled by networking-optimised MCUs will pass commands from the core Ethernet backbone to more specialised networks that range from FlexRay through CAN to LIN. Enhancements to CAN in the form of the Flexible Datarate (FD) modification are helping to ensure the longevity of the real-time automotive control bus. CAN FD supports datarates higher than 1Mbit/s, letting automobile makers consolidate multiple CAN buses into just a few.
This change helps to reduce cabling complexity and improve connectivity between subsystems that will need to cooperate more closely. Transceivers such as the Microchip MCP25612FD have entered the market. To ease development and integration, protocol analysers such as the WS10 for LeCroy’s instruments now feature CAN FD support.
Tomorrow’s vehicles will depend on a carefully integrated suite of cooperating subsystems to deliver on the promise of a better driving experience with high fuel efficiency. But the technologies are now available to make it possible, from the high-performance compute engines down to the furthest reaches of the real-time control network.