CAN Basic Architectures - CAN System Engineering - page 21

Fig. 1.9 Synchronizations on edge before and after the Sync_Seg

The advancement of the CAN protocol from version 1.1 to 1.2 increased the
oscillator tolerance and the option to synchronize on edges from dominant to re-
cessive became obsolete. Only the edges from recessive to dominant are used for
synchronization. CAN protocol implementations according to version 1.1 are no
longer in production.
Most synchronizations occur during arbitration. All CAN nodes synchronize
“hard” on the first node that sends a Start Of Frame bit, but their phases will be
shifted to each other, caused by differences in the signal propagation delays. Dur-
ing arbitration, there may be several transmitters (transmitters do not synchronize
on “late” edges) and the transmitter, which triggers the hard synchronization, does
not necessarily win the arbitration. Therefore, the receivers must synchronize them-
selves successively to different transmitters, whose edges arrive delayed by differ-
ent propagation delays.

1.2.6.2 Propagation Delays

The maximum signal propagation time in the CAN network (between the most
distant nodes, here called A and B) becomes relevant when both of them start CAN
frames at the same time. Supposing that A causes the hard synchronization of B ,
then B will operate with a phase shift of Delay A → B relative to A. Delay A → B
is the sum of the delay of A's bus driver, the propagation time on the CAN bus line
between A and B and the delay of B's bus coupling circuit. The critical point is
reached when A sends two consecutive recessive bits while B sends at the same time
a recessive and a dominant bit. A loses the arbitration at the dominant bit. However,
A sees that bit at the earliest 2 • Delay A → B ( Delay B → A assumed equal to