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.

TMS570LS0432: How to debug GLBSTAT bit 0: OSC FAIL

Part Number: TMS570LS0432
Other Parts Discussed in Thread: HALCOGEN

When GLBSTAT bit 0: OSC FAIL is set, does TI has any suggestion on how to narrow down and identify the root cause? Is the problem with the external crystal or with the internal oscillator?

Based on the datasheet, OSC FAIL is set under this condition: fHFLPO / 4 < fOSCIN < fHFLPO * 4, but measurement at OSCIN was consistent at 20MHz, so we need some help to find out what else could cause this failure.

Thanks,

-Kevin.

  • Hi Kevin,

    The OSC FAIL detection mechanism is a frequency bounds check as you describe in your post. As OSC FAIL can therefore be detected for two reasons:
    1) The external oscillator actually exceeds fHFLPO*4 or decreases below fHFLPO/4, OR
    2) The HF LPO changes such that fHFLPO exceeds fOSCIN*4 or decreases lower than fOSCIN/4

    If your OSCIN is consistent at 20MHz, please check whether the LPO trimming procedure is executed during the system module initialization (clock startup). Note that the default (untrimmed) fHFLPO can be from 5.5MHz to 19.5MHz. So if you have a part with an untrimmed fHFLPO of 5.5MHz, and say it gets trimmed down to 29.52% of this value, the actual fHFLPO aftre trimming would be 1.6236MHz. This will definitely cause an "oscillator fault" to be signaled.

    Hope this helps.

    Regards,
    Sunil

  • Hi Sunil,

    We do call the trimLPO() function, which program LPOMONCTL[15:0] with the value from OTP 0xF00801B4[31:16] as given by HalCoGen code. We also measured HFLPO has 9.6MHz when the failure occurred using ECLK pin and CLKTEST register.

    The LPO CLKDET mechanism, does it validated OSCIN input to the internal oscillator as well as the output from the internal oscillator? Since I was seeing 20MHz +/- 1% at OSCIN, and HFLPO also appeared to be running at 9.6MHz, could the failure be at the internal main oscillator?

    I also want to point out that the recovery sequence from the datasheet was able to re-enable the oscillator. The issue also happened after power cycle at a low temp which was within the operating temp range of TMS570.

    Thanks,
    -Kevin.

  • Hi Kevin,

    The LPO CLKDET mechanism checks the output from the internal oscillator (level-shifted, squared).

    A couple of things to check:

    1) Can you verify that the external oscillator is indeed starting up correctly on each power up across the entire temperature range? This is data that can only be provided by the provider of the external oscillator (crystal/resonator).

    2) How many systems do you see this behavior on? Can you check for cold solder joints that may affect connectivity at colder temperatures?

    Regards,
    Sunil
  • Hi Kevin,

    Do you have any update on this issue?

    Regards,
    Sunil
  • Hi Sunil,

    We have one failing unit and x-ray show no issue on solder joints. Further investigation is still on-going.

    Thanks,
    -Kevin.

  • Kevin,

    I am closing this thread for now. Please reopen it when you have more information after your investigation is completed.

    Regards,
    Sunil