• State TI Thinks Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 2 answers
  • Subscribers 83 subscribers
  • Views 947 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

Original question:

DCAN FIFO reading while receiving

TMS320F28379D: DCAN Rx fifo issue.

Quentin Gaspart
Prodigy 60 points

Part Number: TMS320F28379D
Other Parts Discussed in Thread: LAUNCHXL-F28379D,

Hello,

I encountered some issues with the DCAN pripheral within LAUNCHXL-F28379D.

I have made a test application which consist in a CAN repeater.

In a cycle the LAUNCHXL-F28379D receives several messages take them back from message RAM and resend it into the CAN bus.

Sometimes, after several thousands of packets, the reception order is altered.

My problem is very similar to the one described in this thread : https://e2e.ti.com/support/microcontrollers/hercules/f/312/t/476773#pi320995=8

Could you confirm that this is the same problem than this one : 

Best regards

  • 0
    TI__Mastermind 42780 points

    Hi Quentin,

    Currently looking at this issue.  Will keep you posted on the findings.

    Regards,

    Joseph

  • 0 in reply to Joseph Casuga
    Prodigy 60 points

    Hi Joseph,

    I would like to add some info.

    I have developed some basic functions to transmit and receive data in a FIFO order.
    In order to test the driver I have coded the simple CAN repeater as main function on the development kit.
    An external CAN module send continiously  messages at a 1kHz rate, and these messages are received on the LAUNCHXL board.
    The software is executing a task at a frequency around the 125Hz which retrieved the received data and send them back to the external module.
    I cannot guarantee that an access to the message RAM through interfaces and a reception of a new message is not happening at the same time.
    This will be the case in the foreseen applications of the microcontroller.
    During the Validation of the CAN repeater, I have observed that sometimes after hundreds of thousands messages, some data are not stored in the correct place.
    For your information I wait to have dump all the mailbox which contains new data before clearing the new data bits.
    I start with the mailbox with the highest priority and I verify that no new data has arrived during the reception process.
    I suppose that something goes wrong when the message RAM is read and a new message is arriving at the same time. (as mentioned in the errata) 
    Do you know if the DCAN IP is the same used for the other component (version,...)?
    If it is the case do you have a solution (on TMS320F28379D the DMA do not access to CAN peripheral if I am correct)?
    Best regards,
    Quentin
     
  • 0 in reply to Quentin Gaspart
    TI__Mastermind 42780 points

    Hi Quentin,

    Thank you for the additional information.  From your description, it is likely that the DCAN FIFO issue you cited from the Hercules thread is applicable to the F28379D DCAN IP.  The issue has been brought to our designers attention and they are currently looking into this.  If this is indeed the same issue, unfortunately there might not be a workaround since there is no DMA access to the CAN module.

    Regards,

    Josepj 

  • 0 in reply to Joseph Casuga
    Prodigy 60 points

    Hi Joseph,

    Thank you for your fast answer.

    I will wait for further updates.

    Thank you in advance for keeping me informed.

    Best regards,

    Quentin

  • 0 in reply to Quentin Gaspart
    TI__Mastermind 42780 points

    Hi Quentin,

    Sorry for the wait.  Our designers came back and confirmed that the DCAN IP in the F2837x product and rest of the C2000 family with similar IP have the same issue.  Other C2000 device families have DMA access to DCAN so the DMA workaround is possible.  Unfortunately for the F2837x family, there is no DMA access to the DCAN memory so the workaround is not possible.

    We are currently updating our errata document to reflect this advisory.

    Best regards,

    Joseph

  • 0 in reply to Joseph Casuga
    Prodigy 60 points

    Hi Joseph,

    Thank you for the support and your honesty. We will discuss internally to see how to use as is.

    Do you know if such issue can lead to a revision of the DCAN IP which will result to new C2000 revisions ?

    I have seen that the next batch of devices (F2838x) used also the DCAN IP. (even if there is surely another IP for CAN-FD)

    I would find it a pity to not update a brand new device.

     

    Thank you again,

    Best regards,

    Quentin

  • 0 in reply to Quentin Gaspart
    TI__Mastermind 42780 points

    Hi Quentin,

    Our design team does not have a near-term fix for the DCAN IP.  For the F2838x family though, DMA has access to the DCAN IF3 memory so a workaround for the FIFO issue is possible.  The errata sheet for F2838x will include this advisory for DCAN FIFO.

    Regards,

    Joseph

  • 0 in reply to Joseph Casuga
    TI__Mastermind 42780 points

    Hi Quentin,

              Closing this thread for now with the updated errata.

    Regards,

    Joseph