TM4C1294NCPDT: CAN peripheral not entering bus off state

Hi Jacob,

  Yes, you are correct that the TEC error counter needs to be above 255 to be in bus off state according to the CAN protocol.  You probably don't want to disconnect your node from the bus when testing bus off state but rather with the node connected. With the node connected, try to configure your node with different bit timing from the bus or try to create bit stuffing error. That might eventually lead into bus off state. I'm not an expert on CAN. If the suggestion doesn't work I will need to pass your question to our CAN expert who is current out of office. 

Hi Charles, thanks for the quick response.

Why wouldn't I want to test by disconnecting from the bus?  Seems like a very real possibility that could happen, and that it should generate a bus off state.

The specific reason I am testing that way is my board starts acting strangely when I'm disconnected from the bus - lots of errors when trying to communicate to other peripherals over SPI.  I eventually traced the problem to a constant flood of CAN status interrupts because of the automatic retransmission.  I need to be able to continue to run even if the CAN bus becomes disconnected so I'd like to detect the bus off condition and do something about it.  I guess I can detect the EPASS condition instead, but I'm still confused as to why the TEC error counter stops at 128.

Thanks again!

Hi Jacob,

  Perhaps you disconnect from the bus when you reach error passive state and hence the counter doesn't increment anymore to get to bus off. I will pass your question to our CAN expert for suggestions.

Thank you again for the reply, and for passing the question along to the expert.

Just to make sure I'm clear, in this test I am physically disconnected from the bus when powering up.  As far as I know CAN doesn't know anything about the state of things yet.  I then try to send a message and that's when the TEC counter starts incrementing and I get flooded with CAN STATUS interrupts telling me there's an AckError (which is expected), and once the TEC gets to 128 EPASS is set too.  At this point I don't change anything, no closing of the CAN peripheral or trying to cancel the message, so I would expect that the TEC would continue to increment, but it doesn't.

Thanks!

Charles is correct. Once you reach the error passive state, the CAN module stops trying to transmit. (It becomes passive, and will only listen.)  With no more transmission attempts, there are no more missed acknowledgements and therefore no additional errors.

The way the error count reduces, and the CAN module tries again to transmit, is if it sees good CAN messages on the bus. Since there are no other modules on your bus, the CAN module remains quiet.

To see "Bus Off", setup a network with at least three CAN nodes. Have the TM4C1294 at the wrong baud rate and the other two nodes successfully sending messages. The node at the wrong baud rate will send error frames until it goes passive. Then the other nodes will communicate, but the error count on the node with the wrong baud rate will continue to increase until it goes buss-off.

I see that you responded that my answer did not resolve your issue. This is the expected behavior of the CAN module when you try to transmit and there is no other node to acknowledge the transmission. Please let me know what additional questions you have. 

Hmm... that is an interesting point.  I would say that was the answer but I still get AckError interrupts continuously and forever until I reset the CAN peripheral. After I get the status interrupt I write 0x07 back to the CANSTS register but then the next interrupt will still have the AckError in it.  I think that's in part because of the automatic retransmission, but yeah, you'd think that would stop when in the error passive state.  Is it possible I'm not clearing an interrupt bit somewhere or something?

Thanks for the help.