This is the second CAN Quiz Question on the CAN protocol. This one is quite hard and not only tests knowledge of the dark corners of the CAN protocol, it also highlights a couple of important system design issues.
Question
There is a CAN device that’s acting very strangely on a 500kbit/sec CAN bus. A logic analyzer trace is taken from the CAN TX pin (the output from the CAN controller that goes into the CAN transceiver). The trace shows periodic bursts of weird pulses that don’t contain a CAN frame. The device hasn’t been hijacked by malware (and so isn’t trying to do a CAN protocol attack) and the device drivers are working as they were designed.
The question is simple: “What is happening?”
The specific timings are very important clues. A single burst of pulses consists of:
- A dominant pulse of 34 microseconds
- Followed by 15 cycles of 50 microseconds recessive / 2 microseconds dominant
The time between the end of the last 2 microsecond dominant pulse in a burst and the start of the 34 microsecond dominant pulse of the next cycle is 2.837864 milliseconds.
Supplemental questions
- What could be causing this scenario to happen?
- What could be the consequences?
- How could the situation be detected?
- How could it be mitigated?
Super-Expert Level Supplemental Question
- Which CAN controller is this?
