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.

Is CAN communication working between two different ground suppling cards?

Other Parts Discussed in Thread: SN65HVD232, ISO1050

I have two different system which has separate VDD and separate GND. I want to communicate between this two systems with CAN. I don't want to use ISO1050 i want to use regular transceiver like SN65HVD232, and i guess this is possible to communicate two systems which have two different ground, just using two line (CAN-H and CAN-L). Because CAN is a differantial base communication.

Am I right?

If i am not, why?

And if i am right, what is the possible problems in this type communication?

  • The answer is yes and no.  CAN is differential so yes it in theory will operate correctly as long as the two grounds don't become so far apart you take the CAN transceivers outside their common mode range both for the receiver and driver.  Ideally both nodes would still be referenced to a common ground.  CAN was design to deal with standard ground offsets due to ground shifts in the several volt range.  Will both nodes have the option to reference to earth ground? 

    -- Scott

  • First thank you for your answer.

    The nodes can not use common ground. This was my system's design:

     

    Nodes are have different ground and so the SN65HVDs' grounds are different. You say it works theoretically but in real this is probably doesn't work. And this system more is reliable:

    To use isolated CAN transceiver, and using power ground for ground of ISO1050s' CAN side.

     

    Am I understand you right?

    And can you tell me what you are thinking about these two systems?

     

  • I think you should use ADM3053 from analog devices. It has integrated isolated DC/DC converter for powering bus side of isolator. I have only the best experience with it. TI don't have any comparable solution. ;-)

  • We would recommend something like the below schematic.  Each CAN node would have it's own power supply from the main supply lines.  Then via high effenciency , low cost transformer driver you can get the power across the isolation barrier to the transceiver side.  This side would then have it's own "ISO GND" and could be capacitively coupled to protective earth.  This ISO GND or protective GND would provide the same GND potential for each node then either via routing the ISO GND wire or having the protective earth be the reference.

  • Thank you for your answer,

    I am using ISO1050 but in different way :) . I have a 5V power line too and i will use this line to supply ISO1050DUB. The system is something like that:

     

     But i will ask another question :). What is value of R1 and R2 and C7 and C8? Their values change according to bus speed, but can you give me an example?

    And can you tell me why we need 4700pF/2kv between two grounds?

  • This power concept will also work and cuts the number of supplies needed.  R1 & R2 and C7 and C8 are optional as shown in the box.  Depending on how harsh and EM environment you place the system in a combination of series resistance and bus filtering capacitance and / or also Common mode choke Lcm could be used to keep un-wanted extreme RF and transient noise out of the CAN system.  Normally they would not be populated for most CAN systems as the ISO1050 has very good EMC performance built in, but in extreme cases it may be necessary.  R1&2 need to be matched and should be as low as possible if used as they will have voltage drop across them taking away signal margin in the CAN network, thus reducing cable reach.  We have successfully used 5-10 ohm MELF resistors which are very good at transient currents / high power like in IEC Surge test.  C7/C8 should also be matched and again low as possible.  They will add to the CAN round trip loop time and thus start brining the maximum data rate possible down.  These are both trade-offs that have to be made vs the total network design, like how long is it, now many nodes, data rate desired, etc. 

    -- Scott