Betting on CAN and CANopenby Olaf Pfeiffer This article was written shortly after the
|
|
||
Main benefits of CANOne of the more important features for industrial control implementations is "real-time". The real-time behavior of CAN compares reasonably well to other networks. In some applications high-priority messages are transmitted within a millisecond - all depending on configuration and busloads. To further enhance the suitability of CAN for applications with hard real-time requirements, TTCAN was recently specified. Time-Triggered CAN is an enhancement to the message triggering mechanisms of CAN and allows reserving a certain bandwidth for specified messages. Simplified, messages may only be sent within specified time slots after the occurrence of a synchronize message. This allows specifying guaranteed time slots for different types of messages. Today, the biggest advantages of CAN compared to other networks are the costs and the price/performance ratio. With CAN being implemented in standard microcontrollers from about 20 different chip manufacturers, the hardware costs are not much more than those caused by a regular serial line. However, the performance includes a secure implementation of a data link layer that ensures message transfers with a very strong CRC checksum and arbitrates by message priority (on collision, high priority messages get access to the bus). For applications where the price per node is not as critical as in high-volume applications, the main advantages come from the higher-layer protocols and the availability of off-the-shelf control and I/O components. With these components it becomes possible to build complex control applications without the requirement to develop all nodes of the network from scratch. In the past, one "knock-out" criterion against CAN was often the maximum speed/bandwidth versus maximum distance ratio. The maximum bus speed of 1Mbit allows for a data bandwidth of about 500kbit at a maximum bus length of about 100 feet. However speed and bandwidth decrease for longer network distances (over the thumb rule: to double the distance, cut the speed in half). Today, this criterion is weakened by the fact that more and more chip manufacturers produce microcontrollers with multiple CAN controller on-chip. On one hand this allows to multiply the CAN networks handled by a single node (and thus multiplies bandwidth). On the other hand, such chips can also be used to implement efficient bridges or routers that can be used to build larger CAN networks by combining several segments. [ Introduction | Traditional | CAN benefits | Recent | Outlook ] |
© ESAcademy, 2002 All materials |