TCAN1042H: Terminal resistor consult for TCAN1042

Hi Lin,

It is strange to hear that loading the bus more than the two 120-ohm termination resistors appears to solve the data-loss issue. Normally, I would expect this larger load to make it more difficult for the transceivers to drive a Dominant value to the bus. I also would not expect TCAN1042 to struggle in the system you've described. 

I understand in this test the two nodes are placed 10 meters apart. During this test, is there more cabling present for future connections? My thought is that these longer unterminated lengths could be causing signal reflections which are dampened by an increased bus load. Would you also be able to confirm the values of the resistors being used? Finding the resistance between CANH and CANL should suffice. The device is characterized for 60-ohm and 50-ohm loads, so a value between these two can be expected.

If both of these appear in order, would you be able to share scope shots of CANH and CANL in both testing environments - Normally with packet loss and with the added termination that appears to fix the issue? This would help help us diagnose where the issue may be originating from.

Regards,
Eric

Hi Eric,

Thank you for your response here. I am thinking if this related to the total load capacitance value is too large. Customer used 24*TVS (SMBJ12CA) for the 12*CAN bus protection, I have let them do below test,

1. Ony 3*120ohm+12*TVS in the load, No data loss. See waveform below. Falling time=390ns;

2. See the data loss waveform, Falling time=500ns;

What do you think about this? For this is urgent case from customer, please help provide some suggestion for customer to identify the root cause. Thanks.

BR,

CL

Data loss CANH-CANL

Normal communication CANH-CANL (Ony 3*120ohm+12*TVS in the load)

Hi Lin,

It sounds like the use of many TVS diodes in this system may be providing too large of a capacitive load. Increasing the resistive load by adding extra termination resistors likely re-balances the RC timing to allow the quicker Dominant to Recessive edge. However, I would not recommend this as a permanent solution for the following reasons. 

When excess capacitance is added to a cable network, the characteristic impedance will change. Termination resistors are valued so that they match the impedance of the cable being used to best dampen signal reflections which are detrimental to signal integrity. By changing the characteristic impedance of the cabling and altering the resistive termination values, the two impedances will likely be mismatched and make the system susceptible to signal reflections. This may be what you have observed in the above screenshots as the small bump after the Dominant to Recessive transition. This could also be the result of unterminated or long stubs on the cabling network.
As more resistive load (lower overall resistance) is applied to a CAN bus, it increases the amount of current a transceiver will need to supply in order to drive a valid Dominant state. As this load increases, the resulting dominant magnitude will decrease. This will increase the system's susceptibility to differential noise and at some point will make any communication impossible. TCAN1042 is designed to drive a 50-ohm load (two 120 termination resistors plus receiver loads). Increasing the load (decreasing parallel resistance) further will likely have adverse effects on the system's ability to transmit data reliably. 

For the reasons above, I do not recommend adding extra resistors to solve this issue, but rather reduce the capacitance present in the system. I assume much of this capacitance is being introduced by the large number of TVS diodes in the design. I would advise the use of a TVS device with a smaller capacitive load or try to reduce the total number of devices present on the CAN bus. 

Looking at the sketch you included, we also noticed it appears that the system has a redundant CAN connection on each board. If only one CAN node were needed for each board, you would eliminate extra loading presented by the addition CAN transceiver and associated protection devices (if included for each transceiver). 

Would you be able to share the model of TVS you have implemented? I'm interested to see how much capacitance is being introduced to the system to produce the above waveform. Also, would it be possible to send me a schematic of this design? You can find my email by clicking on my E2E name. I would like to check if there are other load sources on the CAN network and how far each node is placed relative to the others. 

Let me know if you have any more questions or if you find something new with your testing.
Regards,
Eric

Hi Eric,

Thank you for your detailed response here. Customer now is using 12*TVS for total 12 nodes, for they have two CAN in one machine, using two TVS is ok for them. So far they can use 3*120ohm terminal resistor to solve this issue, the differential voltage is about 1.8V. Do you see any risk in this system? Thanks.

BR,

CL

Hi Lin,

I do not believe that adding extra 120-ohm resistors is a good solution to this issue. This solution is not scalable and likely will have other impacts on performance as other aspects vary. As mentioned my previous comment, these could be more complex attenuation that may not be apparent in the current test setup. 

It's good to hear that they were able to reduce the number of TVS diodes in use. It sounds like they were able to move to one TVS per board instead of one for every CAN transceiver. Would it be possible to test the setup with even less capacitive load to see if using a lower capacitance TVS would solve the issue? This could be simulated by further reducing the number of TVS diodes present in the setup. 

Regards,
Eric