• State
  • Replies 11 replies
  • Subscribers 34 subscribers
  • Views 122 views
  • Users 0 members are here
Support feedback
Options
Options
Related

TUSB7320: TUSB7320 Intermittent Startup Failure

hisashi yoda
Intellectual 650 points

Part Number: TUSB7320
Other Parts Discussed in Thread: TPS74401, TUSB7340

Hello,
We are investigating a startup failure issue with the TUSB7320.

[Issue]
A startup failure occasionally occurs with the TUSB7320.
The issue reproduces approximately 4 times out of 20 power cycles.
In failure cases, a periodic power fluctuation of approximately 5 to 6 kHz is observed.
This fluctuation begins when the 3.3-V rail reaches approximately 3.25 V during power-up.

[Power Measurements]
Normal condition:
- 3.3 V: 4.8 mVp-p
- 1.1 V: 1.6 mVp-p

Failure condition:
- 3.3 V: 9.9 mVp-p
- 1.1 V: 8.4 mVp-p

Please refer to the attached PDF for detailed waveforms.

TUSB7320_Startup_Failure_Waveforms.pdf 

[System Configuration]

- 3.3-V main power supply
- 1.1 V generated from 3.3 V using TPS74401
- VDDA is supplied through a ferrite bead

[Question]

At this point, we have not been able to determine the root cause, including whether there is a causal relationship between the startup failure and the observed power fluctuation.
Could you please advise what areas should be investigated first when analyzing this startup failure of the TUSB7320?

Best regards,
Yoda

  • 0
    TI__Guru**** 161946 points

    Hi Yoda:

           Is this new project or the same project as before?

         Where was the switching noise coming from? Did customer use LOD or DC bucks for VDD and VDD33?

    Best

    Brian

  • 0 in reply to Brian Zhou
    Intellectual 650 points

    Hello,

    This is the same project we discussed previously.

    For the power supplies:

    - VDD33 is generated by a DC-DC converter.
    - VDD11 is generated from VDD33 using TPS74401 (LDO).

    The approximately 5–6 kHz oscillation is observed only during failure conditions.
    The oscillation starts when the 3.3-V rail reaches approximately 3.25 V during power-up.

    At this time, we have not yet identified the source of the oscillation.

    Best regards,
    Yoda

  • 0 in reply to hisashi yoda
    TI__Guru**** 161946 points

    Hi Yoda:

        Was issue happened  only on few units?

        If yes, after replace new part, issue still excising?

      If  oscillation  freq is 5–6 kHz, it's probably not caused by DC-DC converter. DC-DC converter switching freq is normally 1-5Mhz.

    Best

    Brian

  • 0 in reply to Brian Zhou
    Intellectual 650 points

    Hello,

    The startup failure has been observed on two units so far.

    We have not replaced the TUSB7320 device yet.

    The package is difficult to rework, and we would like to better understand the failure mechanism before replacing the device. We are concerned that replacing the device at this stage could change the behavior and make root cause analysis more difficult.

     

    We have two additional questions.

    1.
    Is there any known mechanism inside the TUSB7320 that could generate oscillation or periodic behavior in the kHz range (approximately 5–6 kHz) during power-up?

    2.
    We reviewed the TUSB7320/TUSB7340 errata (SLLZ067B).
    www.ti.com/.../sllz067b.pdf

    The errata describes a condition where the WAKE# and CLKREQ# signals may oscillate due to the ESD structure when the signal rise time is too fast.

    Could this errata be related to the startup failure we are observing?

    Best regards,
    Yoda

  • 0 in reply to hisashi yoda
    TI__Guru**** 161946 points

    Hi Yoda:

     

    hisashi yoda said:
    1.
    Is there any known mechanism inside the TUSB7320 that could generate oscillation or periodic behavior in the kHz range (approximately 5–6 kHz) during power-up?

    There is no known other mechanism inside TUSB73x0 to generate noise in kHz range.

    hisashi yoda said:
    The errata describes a condition where the WAKE# and CLKREQ# signals may oscillate due to the ESD structure when the signal rise time is too fast.

    Did you pullup WAKE# and CLKREQ# pin? what is pullup resistor on these pins?

    Can you look at waveform on WAKE# and CLKREQ#?

    Regards

    Brian

  • 0 in reply to Brian Zhou
    Intellectual 650 points

    Hello,

    Regarding the pull-up resistors:

    WAKE#
    - Pulled up to +3.3VD through a 4.7-kΩ resistor.

    CLKREQ#
    - No pull-up resistor.
    - Left open.

    Regarding the waveforms:

    WAKE#
    - Please refer to the attached waveform.
    No significant difference was observed in the waveform between the normal and failure conditions.

    TUSB7320_WAKE_Waveform.pdf

    CLKREQ#
    - Measurement is not possible because the signal is not routed out and the pad is covered by the device package.

     

    We have one additional question.

    Could the approximately 5–6 kHz oscillation observed during power-up affect the internal POR (Power-On Reset) or startup sequence of the TUSB7320?

    Is it possible that such oscillation could prevent the TUSB7320 from starting up correctly and result in the startup failure we are observing?

    We have attached the waveforms observed during our investigation.

     

    Only during the failure condition, the VDDA supply current suddenly begins to fluctuate periodically after the VDDA voltage has completed its power-up ramp.

    This behavior is not observed during normal startup.

    TUSB7320_VDDA_Voltage_Current_Comparison.pdf

    Based on these results, we believe there may be a relationship between the internal operating state of the TUSB7320 and the periodic variation in the VDDA supply current.

     

    Best regards,
    Yoda

  • 0 in reply to hisashi yoda
    TI__Guru**** 161946 points

    Hi Yoda:

    hisashi yoda said:
    Could the approximately 5–6 kHz oscillation observed during power-up affect the internal POR (Power-On Reset) or startup sequence of the TUSB7320?

      Not sure how oscillation affect the internal POR , but it definitely  affect signal integrity  for PCIe and USB signal. The PLL inside host is low pass filter with bandwidth 5-10Mhz, that means the 5-6khz noise jitter from power supply  will 100% transfer to PCIE and USB signal by PLL and causing startup failure.

         Wake signal looks good to me, it may not the root cause for this issue.

         Since you have two units with issue, Can you replace one part so we can confirm it's PCB issue or device issue.

    By the way, even  for failed part, did it fail every time after power on reset?

    Best

    Brian

  • 0 in reply to Brian Zhou
    Intellectual 650 points

    Hi Brian:

    We have not replaced the TUSB7320 because reworking the device may change the failure condition and make further investigation difficult.

    The startup failure does not occur after every power-on reset. It is more likely to occur during the first power-up in the morning, after the board has been powered off for an extended period.


    We have three additional questions.

    1.
    The periodic power-supply fluctuations and supply-current variations are observed while GRST# is asserted.
    Which internal circuits of the TUSB7320 are powered and operating from VDDA while GRST# is asserted?
    For example, could internal circuits such as the POR, voltage-reference circuit, analog-bias circuit, or PLL be operating during this period?

    2.
    During the failure condition, voltage fluctuations with the same period are observed on both the 1.1-V and 3.3-V (VDDA) power rails.
    Are there any internal circuits or functions in the TUSB7320 that are associated with both the 1.1-V and VDDA power supplies?

    3.

    Periodic supply-current variations are observed on both the VDDA and 1.1-V power rails only during the failure condition.
    Based on these observations, we believe that the internal operating state of the TUSB7320 may be related to the periodic supply-current variations on both power rails.
    Could you please provide TI's view on this possibility?

    Best regards,
    Yoda

  • 0 in reply to hisashi yoda
    TI__Guru**** 161946 points

    HI Yoda:

        It's possible some block with high current was turn on or off during operation, but hard to understand it is on and off with constant frequency.

        I can't find the designer who know this device today, but I will try tomorrow to find him to discuss this issue.

    Best

    Brian

  • 0 in reply to Brian Zhou
    Intellectual 650 points
    Hi Brian,
     
    Thank you for your support and for checking with the device designer. We appreciate your continued assistance with this issue.
    We have one additional question from a software perspective.
     
    The issue was initially detected because the system watchdog timer was triggered.
    According to the software analysis, the watchdog timer was triggered because register access to the TUSB7320 became unavailable.
     
    Could the approximately 5–6 kHz power-supply oscillation observed during the failure condition be related to the loss of register access to the TUSB7320?
     
    Best regards,
    Yoda
  • 0 in reply to hisashi yoda
    TI__Guru**** 161946 points

    HI Yoda:

         I think it's due to oscillation  on power supply which causing loss of register access .

         After discussion with designer, we agreed that it's better to replace a new part on the board has issue first so we can conclude it's PCB issue  or the device issue.

       If the issue follow the device, we can decide how to continue the debug.

    Regards

    Brian