Civil Engineering Reference
In-Depth Information
process and reduces risk early in the process by allowing designers to assure that
development is heading towards a reliable and robust network topology implemen-
tation. This helps to eliminate problems being discovered late in the development
process. However, simulation's benefits extend beyond the conceptual phase. Even
for network implementations that are already used in production vehicles, simula-
tion helps to ensure the required level of quality if changes to the topology need to
be applied. For example, suppose the on-board vehicle network department applies
changes to the wiring harness system due to routing considerations. Naturally, this
will also impact the electrical behaviour of the networking system (e.g. CAN bus).
Using a simulation-based approach, the network developer can quickly validate
how significantly the changes impact the networking behaviour and then work with
the on-board vehicle network department to reach a viable solution.
Before an executable specification (or virtual prototype) of a CAN topology is
built up, the following questions should be answered:
• What questions will be answered through simulation?
• What data need to be created to answer these questions (e.g. analog and digital)?
• What type of simulation is required to create the data (e.g. transient and AC)?
• In which format should the data (results) be prepared (e.g. signal curves and
tables)?
The importance of exploring and answering these questions becomes immediately
obvious when one tries to develop the topology model and characterize it with data.
The approach described in this chapter focuses on the signal integrity of CAN to-
pologies which partially answers the first question. To complete the answer to the
first question, it is necessary to use typical signal integrity evaluation criteria. This
encompasses quantities such as propagation delay which quantifies the timing delay
between the transmitting CAN node and receivers. Another important criterion is
the settle time after the transition from dominant to recessive state and vice versa.
In particular, the second criterion is closely associated with the robustness of the
system since a long settling time may lead to sampling errors by the CAN controller.
To validate the robustness of the network system in detail, it is desirable to deter-
mine the available timing margin. This means figuring out how close the sample
point of the CAN controller is to the critical ringing area after the transition of the
differential bus signal.
This automatically leads to the answer of the second question since the described
criteria above required a certain set of specific signals to be available after the simu-
lation. Evaluation of the settle time requires analogue signals. The propagation de-
lay is measured between the falling edges of the TxD signal at the transmitting node
and the RxD signal at the receiving nodes. Since both of these signals are digital, it
is understood that in order to model the entire topology both analogue and digital
models are required. Accurate simulation of the propagation delay also requires
inclusion of the propagation delay through the transceiver chip to complete an ac-
curate system behavioural model.
Having the first two questions answered, it becomes apparent that simulating
automotive networking topologies requires a holistic system level approach since
