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.

TMS320C6657: Processor Initialization issue

Part Number: TMS320C6657
Other Parts Discussed in Thread: TMDSEVM6657

Hi

I am running a C6657 at 1.25Ghz with an input clock of 100Mhz

Occasional the GOSTAT bit will never return ok when setting up PLL1. Sometimes the processor will lockup when setting PLL2 ENSAT

The problem happens as follows:

I power on, run the C6657 Gel global_default_init with PPL1_M = 24 and PLL1_D = 0 (100Mhz input -> 1250Mhz core)

If this is successful I can load my application and run without any problems for days. If I reset the processor with the debugger and run global_default_init there is a chance it will get stuck on the GOSTAT or ENSAT as described above. Sometimes I can reset, re-init and run many times before it occurs.

The failure occurs more often at lower temperatures. At 0C there is about a 50% chance of failure on reset.

We have tried this on a number of random boards and while they all fail at 0C they have different failure rates.

We tried increasing all the delays in the gel file by a factor of 1000 but that has no effect on the failure rate.

Please let us know if you have an suggestions on what is causing this issue.

Thanks

Dhar

  • Hi Dhar,

    Could you clarify what processor are you using:
    Commercial Temperature: 0°C to 85°C
    Extended Temperature: –40°C to 100°C

    Also do you by any chance have TMDSEVM6657? Can you try the same thin on the EVM running at 0°C to see if this problem will be replicated?

    Best Regards,
    Yordan
  • Hi

    We are using the commercial part (0C to 85C)

    The Eval boards we have are all 1GHZ C6657.

    Dhar

  • Hi Dhar

    The bits that it is stuck at typically indicates that something is not allowing the PLL to lock properly.

    Can you clarify a few things for us please

    1) When you say eval board - is this your custom board or a TI EVM?
    2) You mentioned different boards have different failures - can you quantify this further? How many boards built , tested , failing etc.
    3) Is the issue only when running at 1250 MHz? It seems like you are trying to transition from 1000 MHz to 1250 MHz with a reset or something?
    4) You mention a gel file being used - can you confirm that this is a gel file from TI SDK etc? If so, which revision of the SDK or emupack etc?
    5) What is your clock source? Have you confirmed that the clock source is working reliably at the failing temperature? Any signal dump for clock source on a passing vs failing case would be good.

    Regards
    Mukul
  • Hi Dhar
    One more critical point that my colleague reminded me -
    If you are solely relying on the gel file for these tests - I believe you may also be running into an artifact of using the GEL and reprogramming the PLLs after already initialized at full speed.
    It will be important for you to look at the PLL user guide for the steps for reprogamming PLLs etc, and make sure in your test you are using that sequence - the gel may not have been written with that use-case in mind - it is typically our way of getting users started quickly with chip level, board level , psc setup etc - and may not be the golden sequence for where you are trying to reprogram the PLL etc.

    Regards
    Mukul
  • Hi

    Thank you for the responses.

    We noticed there was an errata that covered this issue by removing the GOSTAT check and changing the sequence of programming slightly. This has resolved the issue.

    Thanks

    Dhar

  • Dhar
    Thanks for the update and confirming the issue is resolved. I will close the post.
    I will evaluate if the GEL file needs a corrective action to address this issue - let us know if you have any feedback.

    Regards
    Mukul