ISO1050: Not add termination resistor resulting system reset

Hi Gabriel,

Could you please provide more information of the system?

Is DSP connecting to TXD and RXD?

In what condition did you see the DSP reset randomly? 

For example, when the input of TXD change form clock pattern to logic 1 (recessive for CAN bus).

Could you please provide the waveform of the CAN bus when the DSP is resetting?

Regards,

Alfred

Hi Gabriel,

Do you have any updates?

Regards,

Alfred

Dear Koteshwar

Sorry  to ask u again .

I got the wave form when not connected the terminal resistor .

Our DSP connected the ISO1050 Tx and Rx . 

We found these reset occure in normal oprtion.

Hi Gabriel,

Thanks for further update.

I couldn't really interpret the waveform that you have shared. What exactly is the waveform shown in the image? Is it CANH-CANL plot? Could you please also share a waveform when you have the termination resistor of 120Ω is added?

We do not expect the termination resistor to have any impact on the DSP resetting, it seems like the cause for DSP resetting is something else. The presence or absence of termination resistor only impacts CAN bus signal error rate especially when the bus has a very long cable used. To help you debug the issue better, could you please share a more detailed information about the issue? You can start helping me with answers to the below questions.

1. Share a larger schematic showing ISO1050 and DSP connection.
2. How many nodes are connected on the CAN bus?
3. How long is the CAN bus cable?
4. Where all are the termination resistors used / located?
5. What exactly is the issue? How do you know or identifying that the DSP is resetting?

Alternatively, to find out if there is any problem with ISO1050, you can apply a square wave from signal generator to TXD pin of ISO1050 and monitor the CANH/CANL waveform on the oscilloscope with and without termination resistor. This will clearly shows if there is any problem ISO1050 in transmitting data through isolation barrier onto CAN bus and vice-versa. Thanks.

Regards,
Koteshwar Rao

Hi Gabriel,

Do you have any update on the information that I have requested? Alternatively, did you ask customer to test ISO1050 separately using signal generator and oscilloscope to find out if there is any issue with ISO1050?

Let me know, thanks.

Regards,
Koteshwar Rao

Dear Koteshwar

Waveform at both ends of CAN without CAN resistance:

Waveform at both ends of CANH when 120R resistance is added to CANL:

1. Connection between ISO1050 and DSP:

The figure below shows that J509/J510 on the motherboard is the final CAN signal output, and L509/L510 parameter is the patch inductance - Ferrite -600 ω @100MHz-0.9 ω DC-0.2A-1206.

2. How many nodes are connected to the CAN bus?

Reply: as shown in the schematic diagram above.

3. How long is the CAN bus cable?

The length on PCB is about 15cm.

4. Where is the connected CAN terminal resistance?

Reply: The terminal resistance is connected to the J509 or J510 terminal. After connecting the CAN resistor, the fault disappears.

5. How to determine whether the DSP is reset?

We parse the software fault log for watchdog reset.

Hi Gabriel,

Thanks again for sharing the requested information, this is helpful. Please see my inputs below.

Gabriel Wang1 said:

I am assuming this is the waveform of CANH (also known as CAN+). The waveform looks quite noisy and we do not expect such a waveform when termination resistor is not used.

Gabriel Wang1 said:

This looks good and is expected waveform.

Gabriel Wang1 said:

The length on PCB is about 15cm.

I was looking for the length of the cable used between different CAN nodes during testing and also the number of nodes connected on the CAN bus. Please see below an example connection diagram of various CAN nodes connected onto one CAN bus cable.

As you can see above, all nodes connect to the CAN bus directly and only two termination resistors are used at the two extreme ends of the bus. It is important that only two termination resistors are used and at the extreme ends of the bus. This terminates the cable effectively and allows error free communication.
So do you know if customer used such multiple CAN nodes for testing? Did they have a cable connected to the CAN bus? How long was the cable and how many nodes were connected to the cable?

Gabriel Wang1 said:

Can you please ask customer to remove L509 and L510 and test again with and without termination resistor?

I still do not have a good understanding of the connection diagram during testing. It is unlikely that absence of the termination resistor causes such noise. It is possible that the overall connection diagram isn't following the standard guidelines and probably that's why just adding and removing termination resistor is making such a big difference.

Like I mentioned above, please see if you can share the overall system connection diagram of various nodes that is used during testing. Please do also ask customer to remove L509 and L510 and retest again with and without termination resistor. Thanks.

Regards,
Koteshwar Rao

Hi Gabriel,

Happy New Year 2022!

Do you have any update from customer on this?
If the issue is still unresolved and if you have a local TI contact, please do request them to reach out to us so that we can help you better. Thanks.

Regards,
Koteshwar Rao

Dear Koteshwar

Thanks for your attention about this case!

Below is customer's reply.

1. Currently, the customer uses this power supply on a single machine, and the CAN port on the power supply is not connected to the outside.

2. All CAN signal related connection diagrams have been sent, and there is no connection with other places.

3. L509 and L510 are under the radiator, which is inconvenient to disassemble. We added two jumpers on the reverse side of PCB to short-circuit L509 and L510, and simulated removing L509 and L510.

The test CAN waveform is as follows:

No short circuit L509 and L510, not add CAN resistance, CAN waveform

 short circuit L509 and L510, not add CAN resistance, CAN waveform

\

 short circuit L509 and L510, add CAN resistance, CAN waveform

 Not short circuit L509 and L510, add CAN resistance, CAN waveform

Hi Gabriel,

Thanks for the further update and sharing additional waveform as requested.

The above waveform look much better than the ones shared previously, even the ones without termination resistor. These waveform with and without termination resistor are matching what we expect, hence, please ask customer to continue using the termination resistor for optimum results.

I see removal L509 and L510 didn't make significant difference when the termination resistor is used, hence, I will let customer decide whether to keep them or not. Thanks.

Regards,
Koteshwar Rao