Fritz Burkhardt, Senior Technical Engineer, Marketing, STMicroelectronics
Minimizing energy consumption has become a major factor for any new product design in recent years. Hardly any industry can afford to disregard this issue in the definition, development, or marketing of their products and the automotive industry is no exception. Manufacturers have made significant efforts towards reducing fuel consumption by efficient engine management and weight minimization. Meanwhile, developers are focusing on the vehicle's electronic functions as well. Until now, the use of low-power electronic-control units was particularly important for parked vehicles in order to achieve maximum standby times with existing battery capacities. In the meantime, however, current drain has become important for driving vehicles as well because the combustion engine must deliver the electrical energy with a direct influence on fuel consumption. Comparatively low fuel consumption is a key selling point because it immediately affects operating costs for the car owner. Today, however, another factor complements this perspective. The reduction of greenhouse-gas emissions, which are harmful to the earth's climate, has been the goal of many initiatives of the international community. Consequently, the European Parliament passed a regulation stating mandatory limits for the CO2 emissions of new vehicles. The regulation fixes severe penalties for violating these limits to ensure that car manufacturers assume their responsibility to reduce their products' energy consumption. The e-mobility trend may emerge as an additional motivation in the future: Minimizing energy consumption will become more and more important in electric vehicles whose operating range will be a significant factor for this technology's acceptance. Apart from factors including engine efficiency, vehicle weight, and aerodynamic drag, the power efficiency of the electronic control units is a wide field of activity for designers willing to save energy. Analyzing the electronics landscape in modern vehicles quickly yields several observations: The vehicle does not require all of the functions that the many control units offer all the time and in every driving situation. The continuous current drain these modules impose isn't, therefore, always necessary. This is particularly true for convenience functions including seat electronics, trailer-control units, or tailgate-control units because these functions only rarely operate or because they are not necessary all the time. Additional examples include door-control units, auxiliary heating, sunroofs, and rear-view cameras. On the other hand, it must be possible to activate these control units at any time in order to avoid any functional or convenience impairment. For example, assume an average current drain of 100 to 150 mA and a battery voltage of 14 V, potential saving amounts to 1.4 to 2.1 W for each idling control unit. Total energy savings for 20 CAN nodes capable of partial networking therefore amount to an average of 35 W without any negative impact on functions or convenience features. According to the established conversion formula, 40 W of electrical power represent 1.0g of CO2 emissions per kilometer. Thus, the introduction of partial networking leads to potential emission reductions of 0.85 grams of CO2 per km. The aforementioned EU regulation stating a violation penalty of â¬95 for each gram of CO2 per kilometer yields potential savings of approximately â¬80 Euros per vehicle for car manufacturers. But there are even more reasons why designers should consider opportunities for partial networking. These includes the charging of electric vehicles. Although this requires a communication link to the supervising control unit, most of the control units that connect to the bus are not necessary for this task and thus the vehicle can selectively power them down. The same is true for future application scenarios entailing data transmissions between a parked vehicle and mobile end devices. These future-use cases also result in increased requirements regarding the operating life of the components. Partial networking can compensate for this to a certain extent, resulting in reduced costs, which explains the industry's intensive efforts to exploit this potential. At the Fachkongress Automobil-Elektronik, which took place on June 8, 2011 in Ludwigsburg, Germany, German OEMs jointly announced starting volume production of CAN partial networking in the short and medium term. Although current CAN nodes already provide low-power modes, such as standby and sleep, they immediately wake up if any communication occurs on the bus. These low-power modes are thus only useful if all nodes on the bus disable simultaneouslyâ€"a so-called bus idleâ€" as is the case for a parked vehicle. When any message source transmits data on the CAN bus, transceivers wakeup all connected nodes. Consequently, individual control units cannot remain in sleep mode while communication on the bus is ongoing. One suitable approach divides the network into sub-networks and disconnecting specific controllers from the supply voltage. Apart from the restrictions regarding the network layout, using multiple power supplies leads to additional overhead. Nonetheless, designs of this kind are already in use today. Obviously, the most flexible approach is to wake up specific nodes using dedicated, pre-defined wake-up messages. To support selective addressing, CAN controllers in a sleep or standby mode must be able to monitor CAN messages for previously-negotiated wake-up messages and to respond only if necessary. Consider, for example, a network consisting of 12 nodes (Figure 1). Six nodes support partial networking while the remaining units use conventional transceivers without this capability. When the bus is active and a message addresses Node 7, Nodes 7-12 are active due to the ongoing bus communication while Nodes 1-6 remain in low-power mode because they did not receive a selective wake-up message.
Now consider a second scenario in which Node 5 receives a selective wake-up message (Figure 2). This node recognizes the wake-up request and enters active mode. Nodes 7-12 remain active also due to the ongoing bus communication while nodes 1-4 and 6 remain in low-power mode. Ideally, the detection of wake-up messages should occur in the transceiver because this is the only way to avoid activating the microcontroller with the resulting increased current drain. The strength of conventional transceivers is their bus-level translation capability with full signal fidelity and immunity to external noise signals and bus interference. Having only very basic logic functions for detecting simple bus errors, every bus edge activates them. This, in turn, precludes capturing and evaluating any incoming messages because this is the task of the MCU's on-board CAN controller, which has the precise reference clock necessary for evaluating the message. CAN transceivers capable of partial networking therefore need a highly precise internal reference clock in order to reliably capture and decode the incoming bitstream. In addition, this reference clock must be stable in the relevant temperature range. Considering the maximum tolerance of the transmitting node and common disturbances on the bus, including blurred edges, reflections, and EMI, the oscillator must provide precision of < 1% over the entire temperature range from -40 to +105 Â°C for the operating life of the component. The oscillator concept used in partial-networking transceivers therefore plays a primary role and represents the main challenge during the development of these devices.
The node must then extract the information payload from the bitstream according to the CAN protocol before the node can compare the data to its previously-configured wake-up message. The transceiver must thus provide an interface for configuring the partial networking mode and the dedicated wake-up message (Figure 3). As mentioned above, car manufacturers are working hard to bring partial networking into volume production. However, these efforts will only be successful if the industry can standardize transceiver features. For this purpose, a working group called SWITCH (Selective Wakeable Interoperable Transceiver CAN Highspeed) is preparing a standardization proposal based on a relevant requirement specification. The proposal is currently under discussion with the International Standardization Organization to define a supplement for ISO 11898 (Road Vehicles - Controller Area Network CAN). STMicroelectronics is contributing to the definition of this functionality and is working on the implementation of suitable transceivers. The company has worked closely with a major German car manufacturer, which is running in-vehicle tests on a first device. The company expects to be in volume production of the device in Q4 2012. www.st.com