CAN Newsletter magazine
This article explains the CAN SIC XL transceiver approach and concept, the challenges in the networks, and how to combine the CAN XL protocol with the existing CAN FD transceiver, CAN SIC transceiver, and the CAN SIC XL transceiver.
The complete article is published in the December issue of the CAN Newsletter magazine 2020. This is just an excerpt.
CAN XL as an improvement of the well-established CAN FD protocol increases payload and increases the average bit-rate in a CAN network up to 10 Mbit/s. The CAN protocol was first time published more than 35 years ago and 25 years later, the first discussion about an improvement called CAN FD was started. After the successful release of the CAN FD protocol, the corresponding physical layer standard specifications and the availability of CAN FD transceivers and micro-controllers supporting CAN FD, it was time to initiate the next level of CAN called CAN XL. The main motivation was to increase the payload.
Starting from 8 byte in CAN and up to 64 byte in CAN FD,CAN XL is now doing a big step up to 2 000 byte in the payload. To reduce the transmitting time for such a big payload, higher bit-rates are needed to achieve acceptable transmitting times. Transmitting a CAN frame with 2 000 byte data with a bit-rate of 500 kbit/s needs 33 ms. A CAN FD frame transmitted with 500 kbit/s in the arbitration phase and 2 000 kbit/s in the data phase and a payload of 2 000 byte data needs more than 8 ms. For automotive applications, this transmitting time for CAN frames is too long and so the target was to achieve a transmitting time below 2 milliseconds. To achieve this transmitting time with 2 000 byte payload, a bit-rate of 10 Mbit/s and higher in the data phase is necessary. 500 kbit/s in the arbitration phase are set to allow the same distances between ECUs (electronic control units) in networks like CAN FD.
The CAN network is a serial bus system and allows more than two nodes connected on a network. In a serial bus system topology with a higher number of nodes, collisions are possible. To manage these collisions, CAN uses the CSMA/ CR (Carrier Sense Multiple Access/ Collision Resolution) concept. In the arbitration phase, one or more nodes can transmit a CAN frame on the network at the same time and the node with the highest priority wins the arbitration. To sup- port this CSMA/CR concept, transceivers controlling both levels (TxD=0 and TxD=1) on the network, cannot be used. In case of a collision, the level on the network will be not de- fined if one transceiver transmits level 0 and the other transceiver transmits level 1 at the same time. During this collision, the transceiver might be damaged. For that reason, a CAN transceiver controls only one level (TxD=0). This is called dominant level. During the recessive level (TxD=1), the transceiver output stages are high ohmic and the termination resistors are responsible for the recessive state on the network. In figure 1 this behavior is demonstrated. This concept allows a transmitting node to overwrite the recessive state on the network with a dominant level without the risk to damage a transceiver, transmitting a recessive level at the same time. With such a concept, collisions on the net- work can be supported. The disadvantage is that the transceiver’s output stages are changing from high impedance to low impedance and vice versa. This impedance change creates reflection on the network.
In the transmission line theory it is important that the wire impedance and the termination impedance at the end of the wire have the same value. If wire impedance and termination impedance matches, no reflection occurs. If the impedances between the wire and the termination are different, reflection is caused by the different impedances. The formula for reflection is as follows:
In table 1 some numbers for the reflection factors are shown. For termination impedances smaller than the wire impedance, the wave will be reflected at the end of the wire and changes the polarity. If the end of a wire is terminated with a transceiver only (100 kΩ), the wave will be fully reflected with an unchanged polarity. On star points, the impedance changes, too. On a star point with 3 stripes (1 line for the incoming wave and two lines for the outgoing wave) the reflection factor is -0,33. The two outgoing lines are in parallel with two times 120 Ω impedance and the overall impedance for the wave is 60 Ω.
These reflections are caused with every transition on a wire, independent if the network levels are changing from dominant to recessive or vice versa. However, there is one difference. In case of a recessive to dominant transition, the reflection will be damped by the low ohmic transceiver output stages. In case of the dominant to recessive transition, the network is high ohmic and the reflections fade. The length of the fading phase depends on the wire length and the number of stripes. A long fading phase limits the maximum bit-rate in the data phase because the sampling point has to be set after the fading is finished in order to get a reliable sampling. To realize higher bit-rates, the number of ringing must be reduced and thus the transition from dominant to recessive has to be controlled by the transceiver. This is the concept of the new CAN SIC (signal improvement capability) transceiver.
Two different solutions are available to support the SIC concept based on the specification CiA 601-4,
In the Tx-based solution, the transmitter controls actively the dominant to recessive transition and afterword’s up to 500 nanoseconds (ns) of the following recessive phase. In case of shorter recessive bits, the transmitter changes from active recessive to dominant directly. If the recessive bit is longer, the transmitter changes from active recessive to passive recessive (high ohmic) state like in standard CAN FD transceiver. With CAN SIC transceiver, up to 5 Mbit/s in star topologies and 8 Mbit/s in linear topologies are possible.
In the Rx-based solution, all nodes suppress the recessive signal after the dominant to recessive transition, triggered by the internal receiver. The suppression time depends on the product and is optimized for one bit rate. For example for 2 Mbit/s, the transceiver suppression time is up to 450 ns long.
To achieve higher bit-rates, concept independent, the symmetry parameter of CAN SIC transceiver are improved. The new parameters are shown in table 2. The maximum values are the same as in ISO 11898-2:2016, but all minimum values are reduced and allow higher bit- rates. Figure 3 shows the effect for 5 Mbit/s. The tailored parameters reduce the range of asymmetry dramatically and extend the range for network effects and the sample point position.
The main impact is coming from the reduction of the minimum limits for transmitted recessive bit width, changing from -45 ns to -10 ns and the reduction of the receiver symmetry minimum value, changing from -45 ns to -20 ns. A general disadvantage of the CAN FD and CAN SIC physical layer concept is the asymmetric distance between transmitter levels and receiver thresholds. The distance from the recessive level to the highest possible receiver threshold is 900 mV (millivolt) and the distance from the typical dominant level to the lowest receiver threshold is 1,5 V. This difference causes conceptual asymmetry for the timings. To achieve bit rates above 5 Mbit/s, another transmitter concept in the data phase is necessary. The main targets for the CAN SIC XL transmitter concepts are:
To cover all these requirements for the arbitration phase, the CAN SIC concept is used. In the data phase, an alternating network voltage concept is choosen based on the Flexray idea. The advantage of the Flexray concept is that the levels are symmetric to ground and the receiver thresholds. The impedances of both levels are close to the wire impedance (less reflection) and the timing asymmetries are very small. The new CAN SIC XL transceiver has now two modes instead of one mode like implemented in CAN and CAN FD transceiver. The new modes are:
News and reports