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.

  • TI Thinks Resolved

ISO1050: Operation without CANbus voltage reference

Part Number: ISO1050

My application is an update to an existing design.  This device is a 200W class electric motor actuator for use on a 12VDC vehicle using CAN.  The legacy connector between the device and the vehicle has four pins, CAN_H, CAN_L, +12VDCin, and GND.  I don't have the freedom to modify this connector.

I am concerned about this configuration.  Under full power, ~20A is pulled off the vehicle.  If the GND path has much more than 0.100 ohms resistance, the CAN lines will be appear to have violated the common mode specification from the perspective of the CAN transceiver in our product.  We could add isolation around the CAN transceiver in the new design, however it still needs some reference for the CAN bus (?).  The floating supply and the transceiver need to have a 0v reference to properly locate its own output relative to the vehicle bus.

I've seen the reference designs from TI and others, however they don't answer this question conclusively.  A split termination with Vsense tied to the center tap might be a solution if this device was one of the bus terminators.  Unfortunately, I can't assume that this device will always be allowed as a bus terminator.  Could large value resistors be used instead of 60 ohms to avoid loading the bus, yet still supply a pseudo-reference?

Certainly, this isn't the first time this has come up.  Is there an industry accepted practice to solve this challenge?  

  • Thom,

    First off, thank you for the great question.

    I think you are getting the internal common mode specification confused with the external common mode specification.

    It is true that the bus output voltage specification has to be between 2 and 3V. So 20A at 0.100 ohms would be about a 2V drop, appearing to not meet the specification. But the standard actually does allow this. The 2-3V specification is the internal common mode. So the output of the device has to be between 2 and 3V with respect to the output ground (on the ISO1050 that would be GND2). See Bus output voltage (Recessive) in the ISO1050 datasheet in section 6.6 Electrical Characteristics: Driver on page 7 of the ISO1050 datasheet. 

    The CAN standard also has an external common mode specification. This allows the CAN signal common mode to go between -12V and +12V without violating spec as an input. This is where that 20A and 0.100 ohms would come into play. And here 2V is an acceptable amount. In section 6.3 Recommended Operating Conditions - Voltage at bus pins (separately or common mode) is specified at -12 to 12

    To your question about split termination and biased split termination, this can be done, but is usually done in the context of improving EMC performance. 

    Split Termination

    Biased Split Termination

    To your question about split termination and biased split termination, this can be done, but is usually done in the context of improving EMC performance. The 60 ohm resistors cannot be changed, but R1 and R2 in the biased split termination could be sufficiently large.

    Please let me know if that answered your question or if you need me to clarify anything!

    Best regards,

    Dan

  • In reply to Dan Kisling:

    Hi Dan, thanks for the reply!

    Actually, I wasn't looking at the internal common mode requirement, but when you mentioned it I looked. That one doesn't bug me like the external one does. (BTW, I should have said 0.175 ohms as this would net 3.5v CM, punching through the operational CM limits for CAN_H; CAN_L would still be OK, I get the nuance).

    My understanding from reading (maybe an out of date version of) the CAN electrical specification was that the allowable CM for CAN_L could run -2 to 7 volts, not -12 to +12 volts. The ISO1050 may enjoy the wider range, but how do I ensure that my transmitter wouldn't drift up/down 12v and be functional, yet the rest of the nodes may not? Stated differently, if the ISO1050 swings down to -7 volts, it would be fine but the rest of the receivers would see a standard violation of 5 volts magnitude.

    The whole bit about split termination was a path towards a potential solution (pun intended). But maybe we focus on the problem a bit first. IS this actually a problem- am I reading this right?

    Thanks again, I appreciate the quick turn!
    Thom
  • In reply to Thom Kreider:

    Hi Thom,

    Thanks for the pun; it put a smile on my face.

    The CAN standard changed in December of 2016 to -12V to +12V. You are correct that it was -2V to 7V previous to that.

    To the question of "IS this actually a problem," I don't think it is, but we should of course do our due diligence to ensure it isn't.

    If you would like to share a schematic our team would be happy to review it. You can post here if you are comfortable posting the schematic on a public forum, or we can move over to private messages to share sensitive information. I've sent you a friend request on E2E so we can use the messenger feature.

    Looking forward to hearing from you!

    Dan

  • In reply to Dan Kisling:

    Dan,
    I've accepted your Friend request.

    For the public record: while I understand the change to +/-12v in 12/2016, there are still a lot of heritage equipment in the field that complies only to the older specification. It is that older specification that I'm held to as we upgrade older ECU's on those busses. We need a solution or a ruling that those busses would still work.

    Looking forward to working through this with you!
    Thom
  • In reply to Thom Kreider:

    Hi Thom,

    Thanks for accepting my friend request.

    You do bring up an important point: making this work with heritage equipment complying with an older specification is very important.

    Perhaps I can explain how I see the problem and you can correct me if my understanding is off:

    • Worst case common mode swing: 20A * 0.175Ω = 3.5V
    • Worse case driver output of ISO1050 as per section 6.6 of the datasheet: 2V, 3V referenced to GND2
    • Allowable input to other devices on bus as per old standard: -2V, 7V

    For a first pass, let’s assume we can just add these voltages together:

    • Worst case for low voltage: 2V output -3.5V ground voltage swing: -1.5V seen at input of next device. This meets the input specification as -1.5V > -2.0V
    • Worst case for high voltage: 3V output +3.5 ground voltage swing: 6.5V seen at input of next device. This meets the input specification as 6.5V < 7.0V

    Best regards,

    Dan

  • In reply to Dan Kisling:

    This is a systems engineering task where I'm trying to find where things break and what we would do if they did. I don't have a hard requirement to meet - I'm the author of that requirement based on analysis.

    At some point, there is a maximum resistance in the power return path before we break the CAN specification (0.2 ohms). What do we do when that limit is hit to avoid bus failure? If the answer is $0.50 worth of parts to get another 12v of headroom that will be one thing. If the answer is $50 to get another 1v, that will be another.

    What could we do with any part, not just IOS1050 and how much would it cost to get to the next level?

    Thanks,
    Thom
  • In reply to Thom Kreider:

    Hi Thom,

    I think I understand your question. Let me take this to the wider team and see what their thoughts are. As today is a holiday in India and MLK day in US, please give us some time to get back to you on this one. I will give you a response by Wednesday.

    Best regards,
    Dan
  • In reply to Dan Kisling:

    Hi Thom,

    The team had a discussion on this thread today and we had a few thoughts:

    0.2Ω would be the limit for the data path if there was a 20A pull on the data lines. We should also investigate the 20A on the data path to make sure we fully understand the problem.

    Let’s take the ISO1050 as an example, though we do want to apply this to any CAN system.

    Short-circuit steady-state output current is limited to a max of 105mA according to the Electrical Characteristics: Driver of the ISO1050 datasheet (see page 7 of the datasheet). If I recall correctly, the CAN specification limits this to 115mA for any CAN transceiver.

    We have to consider what is driving side 2 of the isolator such that it would be able to sustain such current. Typically (though I cannot say every time) side 2 isolated power supply is derived from side 1. For example, let’s use the Typical Application circuit in section 9.2 on page 22 of the ISO1050 datasheet.

     

    In this case the secondary power supply comes from the SN6501 paired with a transformer and LDO. The absolute maximum rating for peak output switch current of the SN6501 is 500mA – far from our original problem statement of 20A.

    So at this point it seems appropriate for us to question the 20A pull on the CAN data path that would push us outside of our common mode limit.

    Here are some clarifying questions we had:

    1. Could you supply a block diagram for a system that you are considering? Perhaps this can explain how 20A (or something on that order of magnitude) can occur on side 2. 
    2. How long would this 20A occur? Is this DC or a short burst?

    With this information we should be able to explore what limitations the system will have and how to account for them. 

    Please let me know if I can clarify any parts of my post!

    Best regards,

    Dan

  • In reply to Dan Kisling:

    Dan,

    I see the root of our communication issue.  The issue is 20A through the +13.8v bus and power return, not down the CAN lines.  I took your picture and doctored it up a little.  The isolation boundary isn't interesting to this discussion - the power supply to the isolated side is removed because I'm not worried about the iso boundary.

    Notice that the very large load in the star shape (made it ludicrously large to get the idea across).  That load is sharing the GND return line with CAN.  There's 0.2 ohms stuck in that shared line on the way back to the battery and SPG.

    This will drive CM onto the CAN lines.  It will NOT draw 20A down those lines, just assert a bunch of apparent voltage onto the two lines from the perspective of the other CAN bus nodes.  There's nothing special or unusual about the CAN bus.  The nodes will see a bunch of CM and there's no way to solve that other than to bring in a dedicated CAN reference to float the PHY against OR some other mystery solution that I'm trying to get figured out.

    Thom

  • In reply to Thom Kreider:

    Hi Thom,

    I think we are on the same page now.

    One thing I would like to bring up is that VCM is internally driven on the ISO1050, and virtually all CAN transceivers do this as well..

    While VCM may be exposed externally on some chips and not on others, it is still present internally.  Driving the bus to VCM externally (for example, through a biased split termination as we discussed earlier) could help with things such as EMC, but are not really expected to have a large impact on common mode. Furthermore, this technique can only be used at terminating nodes.

    So this brings us to the critical questions: how do we solve this common mode problem when we get outside of the bounds we solved for earlier – and how much is that going to cost?

    The good news is there are parts that far exceed the -2V to +7V of the old CAN standard. As I mentioned earlier the ISO1050 can operate from -12V to +12V. As another example, the TCAN1042 can exceed +/-20V Common Mode Voltage (See VIT+ and VIT- on page 9 of the TCAN1042 datasheet). Here is a list of CAN transceivers on ti.com that have a high voltage common mode voltage.

    Admittedly CAN transceivers from other semiconductor companies can also have a large common mode voltage range. I’ll leave those to a google search as an exercise for the reader.

    I am not at liberty to discuss specific pricing on the forum, but I think we can still talk to this. As you put it “If the answer is $0.50 worth of parts to get another 12v of headroom that will be one thing. If the answer is $50 to get another 1v, that will be another.” The TCAN device I listed earlier are in the ballpark of $1 and will give you large common mode headroom.   

    As always, let me know if I need to clarify anything or if I've answered your question. 

    It is pleasure discussing this with you. 

    Best regards, 

    Dan

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.