• State Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 1 answer
  • Subscribers 82 subscribers
  • Views 368 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.

TMS320F28075: TMS320F28075: Uncorrectable ECC Error cause NMI watchdog time out when output power at a high level

Sheldon He
Expert 8585 points
Part Number: TMS320F28075


Hi expert,

My customer testing their digital power product with NMI watchdog reset happens on one of their several prototype machine.

In the failing machine, let's call it "model F", when it operating at a relative low power level - no problem happens. But once they add up to load to a high level, in second, a reset happens. They changed F28075 chip on their device once but the problem persists.

We checked the NMISHDFLG register in this case and find FLUNCERR bit is set which indicate some “FLASH uncorrectable errors” happens on the device.

They program their device using our debugger and they do store some channel parameter in flash sector to do calibration. 

I am now asking them to check below registers to further inspect this issue: UNC_ERR_ADDR_LOW, UNC_ERR_ADDR_HIGH, ERR_STATUS, ERR_INTFLG

Do you have any suggestion on this? For this project is already in final stage before MP. Please help me with any suggestions you have!

Sheldon

Thanks!

  • 0
    TI__Guru* 87155 points

    Sheldon,

    Do they use Flash API library in their application to program the calibration data?

    What is the format of the file that they are using to program using CCS - .out or hex file?  What is the CCS version that they are using?

    Thanks and regards,
    Vamsi

  • 0 in reply to Vamsi Gudivada
    TI__Expert 8585 points

    Vamsi,

    They are using flash api to program these data.

    They program .out file.

    I will check the ccs version for you.

    Sheldon

    Thanks

  • 0 in reply to Sheldon He
    TI__Guru* 87155 points

    Sheldon,

    Does the ECC error occur for the locations where they program using Flash API library?  OR for the Flash locations that got loaded by CCS?

    If they are Flash API programmed locations, could you check whether they programmed ECC correctly or not using Fapi_AutoEccGeneration mode?

    If they are CCS loaded Flash locations, then make sure they have "AutoEccGeneration" enabled in the GUI.

    Here are couple of links that can help you:

    1) Flash API Guide:  F2837xD and F2807x Flash API (V1.54) reference guide link: http://www.ti.com/lit/pdf/spnu629

    Note for F2807x users: The sample implementations (examples) provided in SPNU629 use 200MHz CPUCLK frequency since that document is primarily made for TMS320F2837xD devices. TMS320F2807x users should not use more than 120MHz of CPUCLK as mentioned in the data manual.

    2) Flash ECC and API wiki: processors.wiki.ti.com/.../C2000_Flash_FAQ

    Thanks and regards,

    Vamsi

  • 0 in reply to Vamsi Gudivada
    TI__Expert 8585 points
    Vamsi,

    They are using CCS 8.1. They have enabled "AutoEccGeneration" in project properties debug setting.
    I will update error location once I get feed back from them.

    Sheldon
    Thanks
  • 0 in reply to Sheldon He
    TI__Guru* 87155 points
    Sheldon,

    Thank you for the info.

    I don't need the address location - I asked so that they can isolate the issue. Since AutoEccGeneration is enabled in Plugin GUI, I guess the issue might be in the application's Flash API usage.

    Guide and Wiki that I shared in my previous reply should help the debug.

    Note that all the 64-bits of any 64-bit aligned memory should be programmed in one shot when programming ECC.

    Thanks and regards,
    Vamsi
  • 0 in reply to Vamsi Gudivada
    TI__Guru* 87155 points
    Sheldon,

    Thank you for validating my answer.
    Does it mean that the customer found the issue? Please post details that can help others.
    Can I close this post?

    Thanks and regards,
    Vamsi
  • 0 in reply to Vamsi Gudivada
    TI__Expert 8585 points
    Hi Vamsi,
    Thanks for your help. Since they have higher priority work to do and this problem never occurs on any other board, they didn't do the test till now. If I have any update I'll share the answer here.
    Please close the post.
    Sheldon
  • 0 in reply to Sheldon He
    TI__Guru* 87155 points

    Sheldon,

    Thank you for updating the details.

    It sounds like more of a board issue then.

    I am closing this post.

    You can open a new related post if needed.

    Thanks and regards,
    Vamsi