Controller Area Networks for Embedded Systems - Networked Embedded Systems - page 349

possible to a single line. Stubs are permitted for connecting devices to the bus, but their length should
be as short as possible. For example, at  Mbps the length of a stub must be shorter than  cm.
Several kinds of transmission media can be used:
• Two-wire bus, which enables differential signal transmissions and ensures reliable com-
munications. In this case, shielded twisted pair can be used to further enhance the
immunity to electromagnetic interferences.
• Single-wire bus, a simpler and cheaper solution that features lower immunity to interfer-
ences and is mainly suitable for use in automotive applications.
• Optical transmission medium, which ensures complete immunity to electromagnetic
noise and can be used in hazardous environments. Fiber optics is often adopted to inter-
connect (through repeaters) different CAN subnetworks. his is done to cover plants that
arespreadoveralargearea.
Several bit rates are available for the network, the most adopted being in the range from  kbps to
 Mbps (the latter value represents the maximum allowable bit rate according to the CAN specifica-
tions). he maximum extension of a CAN network depends directly on the bit rate. he exact relation
between these two quantities also involves parameters such as delays introduced by transceivers and
optocouplers. Generally speaking, the mathematical product between the length of the bus and the bit
rate has to be approximately constant. For example, the maximum extension allowed for a  kbps
networkisaboutm,andincreasesuptoaboutmwhenabitrateofkbpsisconsidered.
Signal repeaters can be used to increase the network extension, especially when large plants have
to be covered and the bit rate is low or medium. However, they introduce additional delays on the
communication paths, hence the maximum distance between any two nodes is effectively shortened
at high bit rates. Using repeaters also achieves topologies different from the bus (trees or combs, for
example). In this case, a good design could increase the effective area which is covered by the network.
It is worth noting that, unlike other field networks such as, for example, PROFIBUS-PA, there is in
general no cost-effective way in CAN to use the same wire for carrying both the signal and the power
supply. However, an additional pair of wires can be provided inside the bus cable for power supply.
Curiously enough, connectors are not standardized by CAN specifications. Instead, several com-
panion or higher-level application standards exist that define their own connectors and pin assign-
ment. CiA DS [DS], for example, foresees the use of a SUB-D connector, while DeviceNet
and CANopen also suggest the use of either -pin mini-style, microstyle, or open-style connectors.
In addition, these documents include recommendations for bus lines, cables, and standardized bit
rates, which were not included in the original CAN specifications.
15.2.1.2 Bit Encoding and Synchronization
InCANtheelectricalinterfaceofanodetothebusisbasedonanopen-collector-likescheme.Asa
consequence, the level on the bus can assume two complementary values, which are denoted sym-
bolically as dominant and recessive. Usually, the dominant level corresponds to the logical value 
whilsttherecessivelevelisthesameasthelogicalvalue.
CAN relies on the nonreturn to zero bit encoding, which features very high efficiency in that the
synchronization information is not encoded separately from data. Bit synchronization in each node
is achieved by means of a digital phase-locked loop, which extracts the timing information directly
fromthebitstreamreceivedfromthebus.Inparticular,edgesofthesignalareusedforsynchronizing
local clocks, so as to compensate tolerances and drifts of oscillators.
To provide a satisfactory degree of synchronization among nodes, the transmitted bit stream
should include a sufficient number of edges. To do this CAN relies on the so-called bit stuffing
technique. In practice, whenever  consecutive bits at the same value (either dominant or reces-
sive) appear in the transmitted bit stream, the transmitting node inserts one additional stuff bit at