• State TI Thinks Resolved
  • Locked Locked
  • Replies 5 replies
  • Answers 1 answer
  • Subscribers 16 subscribers
  • Views 1842 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

ISOW7842: How to implement CAN bus bias per ISO 11898 with isolated design?

Vitaliy Mosesov
Prodigy 110 points
Part Number: ISOW7842
Other Parts Discussed in Thread: ISO1042, ISOW7841

We are planning on using these isolation devices in our application to isolate CAN busses, however in reviewing detail implementation I came across design obstacle.  Per CAN standard 

ISO11898-2:2016 section 5.10 seem to require bus biasing to be implemented on CAN busses.  The bus bias is typically 2.5V in reference to common system ground (not floating ground of the isolator device) since all the downstream CAN devices share system ground not local floating ground of ISOW7842 chip.  In reference design for CAN busses http://www.ti.com/lit/ug/tiducv9/tiducv9.pdf  there's no indication how this is achieved.  But the waveforms in that document do show 2.5V common mode offset on the busses.  For CAN bus operation I don't believe bus bias is necessary to operate however that's not what seem implied by the standard.  

Unfortunately putting external bias using resistor dividers defeats the whole point of high voltage isolation since those resistors must be refereced to system ground and be very large value and handle the kV surge voltages.  Also, I believe it violates 50k max resistance specification from each CAN_H or CAN_L to GND specification, if those resistors need to be larger than 50k.  

Is this common mode bus bias necessary for industry application when doing CAN isolated design using ICs like ISOW7842 or can this bias be set to 0V?  How did others overcome this problem? 

Thank you for your feedback. 

  • 0
    TI__Genius 17100 points

    Hi Vitaliy,

    Thank you for posting to E2E! Since CAN is a differential signal communication standard, a bus bias is required however the reference design does not detail this since the CAN transceiver (TCAN1042) implements the 2.5V bias for the bus automatically.

    ISOW7842 is a general signal isolator, so it does not create a bias for differential communication, but, as shown in the reference design, it can isolate the digital COMMs signals like TX and RX from an MCU to the CAN transceiver.

    A solution to this problem can be to reproduce the reference design using the ISOW78xx with the TCAN1042 CAN transceiver. Additionally, have you considered using the ISO1042? This device is an isolated CAN transceiver and can be used in place of the ISOW78xx + TCAN1042 solution without the ability to generate an isolated power supply.


    Please let us know your thoughts.


    Thank you,
    Manuel Chavez

  • 0 in reply to Manuel Chavez
    Prodigy 110 points

    Hi Manuel,

    Thank you for your reply.  It is true that TCANxx and other similar CAN drivers generate their own 2.5V bias, however that is in reference to the local, isolated, ground that is not seen external to the unit and devices connected to the CAN bus will not see this offset since they are referenced to the system ground as I shown on one of the diagrams in previous post.  I just would like to know if in reference design and in all other applications by other users where these isolated ICs are used to isolate CAN busses the true bus bias was not implemented and had no issues? 

  • 0 in reply to Vitaliy Mosesov
    TI__Expert 6195 points

    Hi Vitaliy, 

    Just to clarify, the ISOW7842 is not interfacing with CAN signals directly. The ISOW7842 interfaces with the TX and RX of the CAN transceiver, which are just normal digital logic signals. In addition to the bias, there are other considerations when interfacing with a CAN bus. CANH and CANL are differential signals, and a dominant bit is driven by a CAN transceiver, and a recessive bit high impedance from the CAN transceiver (the termination resistor then allows CANH and CANL to converge to 2.5V). The ISOW7842 is not able to drive a differential signal or drive dominant or recessive bits in this way. 

    The TCAN1042 is what is interfacing with the CAN signals. 

    Typically in an industrial application the isolator is then inserted between the CAN transceiver and the MCU. The MCU is then isolated from the signals on the CAN bus. 

    Does this help answer your question? 

    Best regards, 

    Dan

  • 0 in reply to Dan Kisling
    TI__Expert 6195 points

    Hi Vitaliy, 

    I haven't heard from you in a few days. Were we able to answer your question about ISOW78xx being used in a CAN system?

    Best regards, 

    Dan

  • 0 in reply to Dan Kisling
    TI__Expert 6195 points

    Hi Vitaliy, 

    Just wanted to add a few additional thoughts here. I'll mark this post as "TI Thinks resolved" since we haven't heard from you in a few days, but if we haven't answered your question please feel free to reply to this thread or create a new one and I would be happy to further assist you. 

    CAN standard (ISO11898-2) asks for 2 types of bus biasing:

    • Normal Biasing (Required): 2.5V in normal mode and bias to ground in low power mode(sleep/standby)
    • Automatic Voltage Biasing (Optional): 2.5V in normal mode and if transceiver supports standby/sleep modes (low power modes)- bus bias to change automatically to 2.5V is traffic is there on bus and goes to 0 biasing (to save power) if traffic is idle for >1.2sec.

    Autonomous bus biasing can improve EMC because common mode does not need to go through steps of 2.5V every time dominant to recessive state change happens on bus (in dominant- master governs bus common mode, in recessive nodes which are in low power mode make common mode go to 0V- so this is avoided by Autonomous biasing). This is a feature of CAN transceiver and will be supported in new CAN devices. 

    Just to clarify, automatic voltage biasing is optional and only helps to improve EMC performance. For best EMC performance for Isolated CAN networks, connect floating grounds together for all nodes and have a shield- one end of shield to earth at one node. 

    I also wanted to point out we have an example of ISOW7841 being used in an isolated CAN FD network with other nodes (some using discrete CAN transceiver + isolator, others using integrated isolated CAN transceiver). I constructed a demo that was shown at a few tradeshows in the last year. You can find a brief video of it here: https://training.ti.com/interoperability-isolated-can-fd-nodes

    Best regards, 

    Dan