Details

The erratum is limited to the case when MCAN is in state "Receiver" (MCAN_PSR[4:3] ACT = "10") with no pending transmission (no MCAN_TXBRP bit set) and a new transmission is requested before the 3rd bit of Intermission is reached and this 3rd bit of intermission is seen dominant. This issue occurs only with disturbed CAN bus.

Under the following conditions, a message with wrong ID, format, and DLC is transmitted:

• MCAN is in state "Receiver" (MCAN_PSR[4:3] ACT = "10"), no pending transmission
• A new transmission is requested before the 3rd bit of Intermission is reached
• The CAN bus is sampled dominant at the third bit of Intermission which is treated as SoF (see ISO11898-1:2015 Section 10.4.2.2).

Under the conditions listed above, it may happen, that:

• The shift register is not loaded with ID, format, and DLC of the requested message
• The MCAN will start arbitration with wrong ID, format, and DLC on the next bit
• In case the ID wins arbitration, a CAN message with valid CRC is transmitted
• In case this message is acknowledged, the ID stored in the Tx Event FIFO is the ID of the requested Tx message and not the ID of the message transmitted on the CAN bus, no error is detected by the transmitting MCAN
• If the message loses arbitration or is disturbed by an error, it is retransmitted with correct arbitration and control fields.

Workaround

Workaround 1:

Request a new transmission only if another transmission is already pending or when the MCAN is not in state "Receiver" (when MCAN_PSR[4:3] ACT !=‚ "10").

To avoid activating the transmission request in the critical time window between the sample points of the 2nd and 3rd bit of Intermission, following can be done:

• Evaluate the Rx Interrupt flags MCAN_IR[19] DRX, MCAN_IR[0] RF0N, MCAN_IR[4] RF1N which are set at the last bit of EoF when a received and accepted message gets valid. Prevent the transmission of any message within 3 bit times after detecting the Rx interrupt flags.

Workaround 2:

A checksum covering arbitration and control fields can be added to the data field of the message to be transmitted, to detect frames transmitted with wrong arbitration and control fields.

Workaround 3:

Set MCAN_CCCR[0] INIT bit, add transmission request for the message and then clear MCAN_CCCR[0] INIT bit. This needs to be done for each message to be transmitted.

Workaround 4:

Keep the number of pending transmissions always at ">0" by frequently requesting a message so that the condition "no pending transmission" is never met. The frequently requested message may be given a low priority, losing arbitration to all other messages.

In case, where all the nodes present on the CAN bus have this workaround implemented, at least two nodes shall have this frequently sent low priority message with different priorities/identifiers. Rest of the nodes can have one of these priorities/identifiers (same) as the priority of this frequently sent message.

Comparison within available Workarounds is shown in Table 3-1.