CAN bus cool knowledge-how does the edge step come from?

Do you know the waveform of the CAN bus? Do you know what factors cause the instability of the CAN signal? This article will take you to explore the culprit that affects the stability of the CAN waveform - the edge step.

Impedance matching refers to achieving a suitable match between the signal source or transmission line and the load. Impedance matching is mainly for adjusting the load power and suppressing signal reflection. However, the phenomenon of impedance mismatch can be seen everywhere in the CAN bus network. As shown in Figure 1, impedance mismatch will cause 7 phenomena, among which the most concerned are the steps on the rising and falling edges. The following will give a detailed introduction to the phenomenon of edge steps.

Figure 1 Impedance mismatch waveform

Explain how edge steps occur, how to eliminate them, and what impact they have on the bus;

1. The Origin of Edge Steps

In the network layout of the CAN bus, the hand-in-hand linear topology is the most ideal and conventional layout; however, in actual field, branches often occur. It is important to mention here that when calculating the length of the CAN bus, the length of the branch (from the transceiver end to the bus) must also be added. For this reason, we conducted an experiment on the excessive length of the branch. In the experiment, there are three CAN nodes in the CAN bus, the main line length is 15 meters, and the branch length of one of the nodes is 1 meter. The baud rate is 250k for communication . The figure below is the CAN waveform of the experiment. It can be clearly seen that there are steps on the rising and falling edges, which causes the baud rate to change, resulting in sampling errors (also known as bit width errors) at the receiving node.

Therefore, the source of the edge step is mainly the branch of the CAN node. If the branch is too long, the reflection will become stronger, which will cause the error of bit width misalignment. ISO11898 only stipulates that the branch should not exceed 0.3 meters at 1M baud rate, but does not make any statement in other cases. This depends on the experience of field engineers.

2. Eliminate edge steps

Edge steps are the main culprit for erroneous waveforms, so how can we eliminate edge step phenomena? The following will introduce some reliable and effective methods from the source and remedial measures.

1. Reduce branch length

The way to solve the problem at the root of CAN network layout is to reduce the branch length of CAN nodes, thereby reducing signal reflection and ensuring the stability of bit width. In the above experiment, other conditions remain unchanged, and only the branch length is reduced to 20cm; the figure below is the CAN waveform, and no edge steps are seen at this time. It can be seen that reducing the branch length is the most direct way to eliminate edge steps.

2. Add appropriate resistance to long branches

When the network layout cannot be changed and the signal reflection caused by the branch must exist. The most practical method is to add a resistor at the end of the long branch to eliminate the signal reflection. Similarly, in the above experiment, a 200Ω resistor is added at the branch node, and the communication experiment is carried out under the same conditions. The figure below is the CAN waveform of the experiment. At this time, it can be seen that the edge step has been eliminated, but the differential voltage becomes smaller after adding the resistor. Note that the differential voltage must not be less than 0.9V. It is worth mentioning here that the ability of a resistor with a resistance greater than 500Ω to absorb reflection is very weak, so when the resistor is hung at the end, it should be less than 500Ω.

3. Shorten the stump

As mentioned earlier, the branch length refers to the distance from the node transceiver to the bus. At the beginning of the node design, the TTL remote transmission method should be selected. Because the TTL level is not affected by the CAN capacitor, the transceiver should be placed close to the interface to reduce the length of the branch segment. It is recommended to control it within 10cm to ensure impedance continuity.

The most direct way to transmit TTL remotely is to place the CAN transceiver next to the CAN trunk line, so that there is no branch length. The optical cable star topology uses this method, as shown below; the CAN fiber optic transceiver is built into the box and uses TTL level to transmit remotely to another CAN fiber optic transceiver, solving the problem of random node changes (nodes can be powered on or off or plugged in and out at will).

4. Eliminate load concentration

In a CAN network with a more complex layout, in order to avoid reflection superposition caused by concentrated node placement, it is recommended that the distance between adjacent nodes should not be less than 2 cm, and the number of devices concentrated on a 10m cable should not exceed 4. Otherwise, capacitors should be added to absorb, and there should be at least 10m of cable distance between this concentration and the next concentration.

Similarly, in complex network layouts, networks with long and unequal branches often use hubs or repeaters for branching due to difficulty in impedance matching; hubs and repeaters have independent controllers and MCUs, forming independent straight-line topologies for each segment, as shown in the figure below.

5. Shielding layer segmented grounding

When the shielding layer is grounded at multiple points, attention should be paid to the potential of the grounding points to prevent ground return current from affecting signal quality. If the shielding layer is too long, segmented shielding and single-point grounding can be used, as shown in the figure below, which can effectively avoid the problem of ground return current.

3. Edge consistency test

Signal edge is an important indicator of signal quality. If the falling edge of the signal slows down, it will cause a certain degree of distortion of the CAN signal waveform, resulting in transceiver sampling errors. Referring to the edge test of mainstream car companies, the time from 10% to 90% of the edge is generally taken as the edge time, simulating the capacitive reactance that the DUT may be affected by when connected to the CAN network, so that the measurement results are more practical. The edge of the DUT is measured in the small capacitance and large capacitance load environments simulated by CANDT.

lTest purpose: Measure the rise or fall time of CANH, CANL and CANDIFF signals under small and large capacitance loads respectively;

l Test principle: The test principle is as shown in the figure below. When the DUT transmits data to the bus normally, the transmitted data frame is a sequence of dominant bits and recessive bits, that is, the transmitted data contains the rise and fall time information;

CANDT

The CANDT consistency test system released by ZLG Zhiyuan Electronics can automatically complete the consistency test of the physical layer, link layer and application layer of the CAN node. It is the only instrument in the current CAN bus test field that can perform complete physical layer automatic testing and export reports. It avoids the errors of manual measurement statistics, and at the same time, cooperates with the automated testing method to reduce the waste of test time, improve the accuracy of the test, and greatly save labor costs.

The CANDT consistency test system is based on the underlying analysis capabilities of CANScope and integrates necessary equipment such as oscilloscopes and power supplies. It can cover the CAN consistency test standards of OEMs and establish a CAN bus test and assurance system for OEMs and component companies.