CRD300DA12E-XM3 – C2000 lock-up and CAN A communication failure during switching
Dear Wolfspeed Technical Support,
We are currently testing a CRD300DA12E-XM3 300 kW XM3 Three-Phase Inverter and we are experiencing a communication problem when the SiC power stage starts switching.
Our external ECU communicates with the C2000 controller through the isolated CAN A interface on J9.
When the inverter is not switching, CAN communication works correctly and the C2000 responds normally.
However, as soon as the SiC devices start switching, communication with the ECU is lost and the C2000 becomes unresponsive. Stopping the switching does not restore communication. The controller only recovers after removing the power supply from the control board and powering it again.
We have measured CANH and CANL with an oscilloscope in both operating conditions.
With the inverter stopped, the CAN signals are relatively clean. When switching is enabled, significant high-frequency disturbances appear simultaneously on both CANH and CANL. The disturbance appears mainly as common-mode noise, since similar switching-related transients are present on both CAN lines.
We have already tested ferrite cores around the communication cables in order to reduce the coupled noise, but the problem still persists.
We are also investigating the SPI communication between the EPU and ECU and checking how the communication behaviour changes as the DC-link voltage is increased.
Could you please advise us on the following points?
- Have you observed similar CAN communication failures or C2000 lock-ups with the CRD300DA12E-XM3 during switching?
- What grounding, shielding or common-mode filtering configuration do you recommend for the isolated CAN A interface?
- Could switching-related common-mode disturbances affect the CAN transceiver, digital isolator or C2000 controller even when using the isolated CAN A interface?
- Are there any recommended test points or measurements to determine whether the problem is related to the CAN interface, controller power supply, reset signal or ground bounce?
Thank you for your support.
Best regards,
Enrique Cruz
Comments
-
Thank you for your post, it has been approved and we will respond as soon as possible.
0 -
Hello Enrique,
I'm sorry to hear that you are experiencing issues with your CAN communication, and I hope that we can resolve them. I answered your questions below and provided some additional thoughts/suggestions.
- All CRD300DA12E-XM3 units are tested with an end-of-line (EOL) burn-in test before being shipped. This testing stresses the device at power and includes operating the inverters via CAN while switching at power. We don't experience the issues that you are observing during this EOL testing or during our internal testing.
- We use the default shielded cables included with the isolated CAN adapters to connect between the PCB and the CAN adapter. These cables are shielded but do not include any additional special grounding connections. Between the CAN adapter and the host computer, we route all the wires through a CM choke. You could also consider routing the cable between the PCB and CAN adapter through a CM choke. I also recommend limiting the length of these connections as much as possible while maintaining all safety requirements. What CAN adapter are you using? Is it a high-quality isolated CAN adapter? We do internal testing with these reference designs using the PEAK System PCAN-USB Pro FD and National Instruments USB-8473 isolated CAN adapters.
- The issue could be related to any of the devices that you mentioned. Again, however, this isn't an issue that we observe internally. A few ideas to consider. First, does your +12V power supply to the controller circuitry have enough power to supply the controller during operation? Perhaps the voltage rails are sagging due to an inadequate power supply when switching begins. Additionally, you could consider using an isolated +12V power supply for improved performance. Second, are you including sufficient deadtime to prevent shoot through in any of the switch positions? A shoot through event isn't a typical operating condition for this converter and causes safety circuitry to activate which deprioritizes all other aspects of the circuit. Third, what grounding are you employing for both the power and control circuitry? If you separate these grounds, this could break CM paths.
- The CAN transceivers are robust components which I would prioritize low in your troubleshooting. For troubleshooting, I would prioritize (1) checking that the operating conditions you selected are reasonable for the inverter, (2) checking that you are using a high-quality CAN adapter, (3) evaluating the controller power rail stability, and (4) adjusting the grounding scheme of the inverter.
Thanks,
Chris N.0
