SLLA581 April 2022 TCAN1462-Q1 , TCAN1463-Q1 , TCAN1473-Q1
All trademarks are the property of their respective owners.
Signal improvement is an additional capability added to CAN-FD transceivers that enhances the maximum data rate achievable in complex star topologies by minimizing signal ringing. CAN SIC transceivers need to meet or exceed the specifications of the International Organization for Standardization (ISO) 11898-2:2016 high-speed CAN physical layer standard and the CAN-in-Automation (CiA) 601-4 signal improvement specification.
Figure 1-1 shows a regular CAN-FD transceiver where the CAN bus signal rings above 900 mV (the dominant threshold of a CAN receiver) and below 500 mV (the recessive threshold of a CAN receiver), resulting in receive data (RXD) glitches. With reference to CiA 601-4, Figure 1-2 shows how a CAN SIC capability transceiver attenuates bus signal ringing, resulting in the correct RXD signal.
In terms of electrical parameters, a CiA 601-4-compliant CAN SIC transceiver has a much tighter bit-timing symmetry and loop-delay specification compared to a regular CAN-FD transceiver, as shown in Table 1-1. The segregation of delays of transmit and receive paths can help system designers clearly calculate network propagation delay in the presence of other signal chain components. One thing to note is that the timing specified in CiA 601-4 is data rate-agnostic and holds true for both 2- and 5-Mbps operation.
CiA 601-4 Specifications | ISO 11898-2:2016 Specifications | ||||
---|---|---|---|---|---|
Parameter | Notation | Min [ns] |
Max [ns] |
Min [ns] |
Max [ns] |
Signal improvement time TX-based | tSIC_TX_base | N/A | 530 | N/A | |
Transmitted bit -width variation | ΔtBit(Bus) | –10 | 10 | –65 for 2 Mbps | 30 for 2 Mbps |
–45 for 5 Mbps | 10 for 5 Mbps | ||||
Received bit width | ΔtBit(RxD) | –30 | 20 | –100 for 2 Mbps | 50 for 2 Mbps |
–80 for 5 Mbps | 20 for 5 Mbps | ||||
Receiver timing symmetry | ΔtREC | –20 | 15 | –65 for 2 Mbps | 40 for 2 Mbps |
–45 for 5 Mbps | 15 for 5 Mbps | ||||
Propagation delay from transmitter data (TXD) to bus dominant | tprop(TxD-busdom) | 80 | Only loop delay, TXD to bus to RXD, is specified at 255 ns max | ||
Propagation delay from TXD to bus recessive | tprop(TxD-busrec) | 80 | |||
Propagation delay from bus to RXD dominant | tprop(busdom-RxD) | 110 | |||
Propagation delay from bus to RXD recessive | tprop(busrec-RxD) | 110 |
The first-generation CAN protocol, ISO 11898-2, also known as Classical CAN, was released around 1993. The protocol allowed only 8 bytes of payload data transfer, and a maximum specified data rate of 1 Mbps. These limitations were quickly realized in automotive applications, where vehicles have a number of electronic nodes that communicate with each other using the CAN bus.
The CAN-FD protocol specification was released around 2015, which increased the payload length to 64 bytes and the maximum signaling rate in the data phase to 5 Mbps. The arbitration phase signaling rate was still limited to 1 Mbps, however, for backwards compatibility with Classical CAN.
While CAN-FD brought the advantages of a faster data rate and a longer payload, it wasn’t sufficient to keep pace with the ever-increasing number of ECUs added to vehicle CAN bus networks. Designers realized that they could not harness the real potential of CAN-FD transceivers, as bus ringing resulting from complex star networks affected correct signal communication. Figure 2-1 is an example star topology.
In complex star topologies with multiple stubs, a signal traveling on the bus experiences impedance mismatch which causes reflections. These reflections distort the CAN bus and cause it to oscillate, resulting in an incorrect CAN bus level and RXD at the sampling point. Although these network effects were not specific to CAN-FD networks, at the lower-speed operation of Classical CAN the bit duration was longer, and the bus ringing diminished such that it was possible to sample the correct bit, as shown in Figure 2-2, resulting in correct communication.
For 5-Mbps CAN-FD operation, a 200-ns bit duration was much too small for the ringing in complex star topologies to disappear, hampering reliable data communication. This deterred system designers from using CAN-FD at 5 Mbps.
With an increase in the exchange of network data and faster throughput demands in modern-day vehicles, CAN SIC paves the way for a next-generation in-vehicle communication bus technology that is faster and provides more network flexibility and scalability.
The CAN bus has two logical states during normal operation: recessive and dominant, as shown in Figure 3-1.
A dominant bus state occurs when driving the bus differentially and corresponds to a logic low on the TXD and RXD pins. A recessive bus state occurs when the bus is biased to VCC/2 through the high-value internal input resistors (RIN) of the receiver and corresponds to a logic high on the TXD and RXD pins. A dominant state overwrites the recessive state during arbitration. The recessive-to-dominant signal edge on the CAN bus is usually clean, as it is strongly driven by the transmitter. The differential transmitter output impedance of the CAN transceiver during the dominant phase is approximately 50 Ω and closely matches the network characteristic impedance. For a regular CAN-FD transceiver, the dominant-to-recessive edge is when the driver differential output impedance suddenly goes to approximately 60 kΩ, and the signal reflected back experiences an impedance mismatch, which causes ringing.
Transmitter-based SIC detects the dominant-to-recessive edge on TXD and activates ringing suppression circuitry on the driver output. The CAN driver continues driving the bus recessive strongly until tSIC_TX_base so that reflections diminish and the recessive bit is clean at the sampling point. In this active recessive phase, the transmitter output impedance is low (approximately 100 Ω). Since the reflected signal does not see a huge impedance mismatch, ringing is attenuated considerably. After this phase ends and the device enters a passive recessive phase, the driver output impedance rises to approximately 60 kΩ. Figure 3-2 shows this phenomenon.
An important factor in the active recessive phase strongly driving the bus is that it should only last for a maximum of 530 ns (tSIC_TX_base, as listed in Table 1-1). The data phase of the CAN-FD protocol only lasts for 200 ns max (if operated at 5 Mbps), so this ringing suppression will be active for the entire recessive bit duration, resulting in correct CAN bus and RXD signals. For the arbitration phase, however – where the fastest bit duration is 1 µs for 1-Mbps operation, multiple transmitters can transmit simultaneously, and the dominant bit has to overwrite the recessive bit – the duration of ringing suppression may place some limits on the overall network length and arbitration speed. See the CiA 601-4 specification for more details.