• TI thinks resolved

ISO1050: Low resistance failure between CANHI, CANLO, GND

Part Number: ISO1050

Hello all,

I have recently had 3 units come across my desk which were described as 'not communicating on the CAN bus'.

These units are potted so I began by measuring R's & C's across CANHI, CANLO, GND and VCC against a reference unit. I immediately noticed that the resistance between the CAN lines and GND, and every permutation thereof, was reading below 100 ohm.

I then removed the potting to see what was causing this - I removed the input protection circuit and measured again, with the same result.

After replacing the ISO1050DUB package the unit functioned again.

I repeated this on the second unit, however this time it was just the CANHI and GND pair that exhibited a low resistance.

I verified that the power supply to each side of the transceiver was reading 5.05V and the CAN lines are protected by a dual CAN-BUS ESD protection diode and a common mode choke.

What could possibly be the cause of this and is there anything else we can incorporate into the design to prevent this from happening again?

All 3 units that have failed appear [ in removing the potting the markings were removed - but all the units were built at the same time, so components would have been off the same reel ] to be built from the same batch of ISO1050's so is it possible that this was a batch issue?

I have 2 extracted ISO1050DUB's if anyone would like to take a look also. I can extract the 3rd if necessary and try not to remove the package marking.

Many thanks.


  • Hi Ryan,

    Thank you for using TI E2E.

    • The 100ohm resistancereading on the CAN lines does seem low.
      • On a good unit, the differential impedance between CANH and CANL should be between 30k and 80k-ohms; 
      • The single ended impedance reading should be ~half of this.
    • Curve tracing the CANH and CANL pins w.r.t GND2  would be a good predictor of any potential EOS/damage inside the chip.  If possible, these pins could be curve traced to +/- 25V with a current limit of  <50mA.   This will help us look for signs of early/premature turn-on (if any) of the components inside
    • On the second unit you mentioned, it looks like this is constrained to just CANH. So the sanity check experiments proposed above may further confirm this as a trend or provide a different data point. It will be helpful either way.
    • Supply values: voltage values look healthy and should be fine. Do you see any abnormally high current consumption on either side 1 or side 2 of the chip? (when compared to a good one)
    • Power -up default value on CANH, CANL: Typically, when provided with a 60ohm termination resistor between CANH and CANL, and with TXD input = high; the default value on the CAN lines must be CANH=CANL = Vcc2/2 = ~2.525V  [ = Recessive] in your case. If one or both of these pins are off, then that will offer additional clues on what may have gone wrong.
    • Could you please tell us which ESD diodesare being used? PESD1CAN and PESD2CAN are commonly used, but I wish to double -check and not assume anything at this point. 
      • The key thing here is to position the TVS/ESD diode breakdown voltage. 
      • It must be above the Operational range of the part (+/- 12V per the data sheet) so as to not interrupt normal operation, and also must be below the Abs Max value ( = damage rating) of the part, so that the chip is not damaged.
      • When positioned this way, normal operation is uninterrupted. But when an ESD or other high voltage event happens, then the diode will trigger first and still protect the device. 
      • If the diode breakdown voltage is too high, then the ISO1050 may trigger first - which we don't want to happen.

    • Choke details would also help as a sanity check. Chokes with inductances between 11uH - 100uH are common.

    Possible cause(s): Still early to say for sure, but one of the CAN pins may have been damaged due to an EOS event or during handling/processing. Admittedly it is hard to figure out which system EOS event may have caused this.  This also depends on which ESD/TVS diode is chosen. We are not aware of any issues that might be caused or made worse by the potting process per se, but need to check. The curve trace data will help on this for sure.

    Sorry for the broad list. Please feel free to let us know what you think and if it's possible to look into the above as time permits. We can also do a schematic review for you if you'd like - although that would be best done over email or private message(PM) to keep it off a public forum. We can then circle back and summarize the key points here later as needed.

    We hope this helps. 

    Best regards,


  • In reply to Abhi Aarey:


    Thank you for your in depth and well considered reply.

    The tests you have outlined above I will perform and provide results within the next 2 weeks.
    I am happy to provide information w.r.t. the choke and the ESD diodes used over email.

    I am of the same opinion that it is likely caused by an EOS event - it would be nice to identify this and be able to put in place appropriate protection to prevent this in the future.

    I have also taken the time to create a basic ESD simulator, to observe the effects seen by the ISO1050 to see if this provides any insight.

    I would like to take you up on the offer of a schematic review if possible please - I have sent you a request on here with my email address.

    Best Regards,
  • In reply to Ryan Pannell:

    Hi Ryan,

    You are welcome, the plan sounds good.
    Yes, we can definitely review components (chokes, diodes) and schematic over PM/email and then post a summary here later.

    Thanks and regards,

  • In reply to Abhi Aarey:

    Hi Ryan,

    Just summarizing the key points of our email discussion here:

    1) Reviewed schematics offline
    2) Choke looks OK.
    3) TVS can be optimized to reduce risk.

    • Can pick unidirectional devices if we are confident that negative common mode voltages do not occur in the system but bidirectional devices do give more flexibility.
    • Some example solutions that break outside the Recommended region (to not interfere) but safely clamp before the Abs Max region (to protect the device) are listed here and here.
    • Based on the above links, we also discussed the PESD12VL2BT device that you mentioned. (Same footprint as existing option). 

    You are always welcome to contact us with any additional questions or comments.

    Best regards,