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.

TPS65217: AM335x based design reboots instead of shutdown

Part Number: TPS65217

We have an AM335x + TPS65217C based design and we are observing that the system shuts down but reboots after 1s when trying to execute "shutdown -h now".

Particulars of the circuit

1. Circuit has BAT pins shorted to GND via 1K resistor (battery not used)

2. Power provision is given from AC as well as USB pins.

Test conditions

1. Power is supplied from AC pins (5V)

2. USB cable is not connected

3. Communication to laptop is via Debug UART

4. Command entered "shutdown -h now"

Observations

1. PWR_EN goes low then has a spike of around 50ms duration and then goes back to LOW.

2. After 1 s of power down, the board reboots. VSYS rises again and power up sequence starts.

3. For the same test conditions, when USB is also connected, the board shuts down properly and permanently.

Diagram with no USB cable, only 5V external DC voltage powering the board. (reboots after 1s)

Diagram with USB cable connected and also 5V external powering the board. (permanent shutdown)

I have seen this issue being discussed in multiple threads, but couldnot get a conclusive solution on how to deal with this.

  • Hi Venugopal,

    Thanks for reaching out. I will review the details on this E2E and will provide an update on Monday. 

    Thanks,

    Brenda

  • Hi Venugopal,

    Would you be able to share the schematic for the PMIC? The issue you are seeing on this TPS65217 application with no battery could be related to a lockup of the PMIC due to a Brownout condition. This issue as well as the recommended solutions were documented in the following application report: Designing Robust TPS65217 Systems for VIN Brownout

    Thanks,

    Brenda

  • Hello Brenda, 

    I am afraid I am not authorized to share the schematic.

    To answer descriptively to your query, we have used the voltage supervisor + load switch solution as mentioned in the application note to take care of voltage brownout. Also, the BAT pin is grounded via a 1K resistor.

    As seen in the attached waveform, the AC pin voltage does not dip, and hence it may not be related to the brownout issue at all.

    I can gather any other waveforms which you may need to look at.

    To put the problem statement in a more concise way,

    1. The main source of power for the board is from AC pin (external 5V)

    2. BAT pin is grounded via 1K resistor.

    3. Alternate powering provision through USB is also provided.

    4. When the board is powered from external 5V at AC pin, if the USB is also connected, then the shutdown happens properly. 

    5. When the board is powered from external 5V at AC pin, if the USB is not connected, then the shutdown command leads to reboot after 1s.

    Thanks,

    Venugopal

  • Hi Venugopal,

    Thanks for providing the details! Here are is my feedback/questions:

    • How is VIO being supplied to the PMIC? Are you supplying VIO with an external regulator or you are using VLDO1 from the PMIC?
    • The reason why I asked the question above is because if you have an application without a battery but VIO is supplied with an external regulator then it should be treated as a "system with a battery" and the image below shows the circuit solution that should be implemented. 

    • If VIO is supplied using VLDO1 from the PMIC and you have an application without battery, then the solution circuit in the image below should be implemented. From the information you provided in the previous message, this scenario seems to match your conditions but please let me know if you have something different. 

    • The TPS65217 datasheet specifies specifies that TI does not recommend using both the AC and USB inputs when the battery is absent.  

    Thanks,

    Brenda

  • Hello Brenda,

    Thanks for the quick responses. Please find the details you asked for.

    • How is VIO being supplied to the PMIC? Are you supplying VIO with an external regulator or you are using VLDO1 from the PMIC?
      VIO is being supplied from VLDO1

    • The reason why I asked the question above is because if you have an application without a battery but VIO is supplied with an external regulator then it should be treated as a "system with a battery" and the image below shows the circuit solution that should be implemented. 
      We have implemented this solution with a voltage supervisor at 4.0V UVLO. Same part numbers as described in the application note are used.

    • If VIO is supplied using VLDO1 from the PMIC and you have an application without battery, then the solution circuit in the image below should be implemented. From the information you provided in the previous message, this scenario seems to match your conditions but please let me know if you have something different. 
      Yes, you are right. In our circuit, BAT and BAT_SENSE are pulled down to GND via 1K resistor. TS is floating.

    • The TPS65217 datasheet specifies specifies that TI does not recommend using both the AC and USB inputs when the battery is absent.  
      Surprisingly, using both AC and USB inputs, we are able to do proper shutdown. Using only AC input, we are not able to shutdown, and the board is rebooting.


    Thanks,
    Venugopal

  • Hi Venugopal,

    Thanks for providing the responses to my questions! Here are the recommended next steps for the debug:

    • What is connected to the PB_IN pin? Is it floating?
    • Are you supplying all rails with the SYS pin?
    • Can you short the USB pin on the TPS65217 PMIC to GND and repeat the test? It will be very helpful if you could try this and provide a capture including USB, AC, SYS and PWR_EN.

    Thanks,

    Brenda

  • Hi Brenda,

    • What is connected to the PB_IN pin? Is it floating?

    PB_IN pin is connected to a push button switch for momentarily shorting to GND if needed. This is planned to be "not populated" in the final version.

    • Are you supplying all rails with the SYS pin?

    - Yes, input (VIN_DCDC1,2,3 and VIN_LDO) to all PMIC rails is SYS pin.

    • Can you short the USB pin on the TPS65217 PMIC to GND and repeat the test? It will be very helpful if you could try this and provide a capture including USB, AC, SYS and PWR_EN.

    - I will provide this waveform in a day or two.

    Thanks,

    Venugopal

  • Hi Brenda, 

    Please find the waveform requested. In addition, I have also probed the PB_IN during the shutdown/reboot procedure.

    Test scenario : 

    - USB pin shorted to GND

    - Power supplied via AC pin using external 5V supply

    - Communication between device and laptop is via Debug UART port

    Thanks,

    Venugopal

  • Hi Venugopal,

    Thanks for providing the data and captures! There are just two more things I would like to try (listed below). If they don't fix the issue, it will be very challenging to do any further debug without having access to the schematic/layout which could be shared privately. If the schematic can't be shared at all then I might need to transfer this E2E to the processor team to help verify any potential software related issues. 

    • Can you leave PB_IN floating (disconnect the push bottom and 100nF cap), increase the Vin voltage to 6V and take the same capture again?
    • Just to make sure this is not happening on a single failure device, could you repeat the same debug on a second (brand new) PMIC? 

    Thanks,

    Brenda

  • Hello Brenda,

    Below are the waveforms you requested. To summarize things, 

    • Can you leave PB_IN floating (disconnect the push bottom and 100nF cap), increase the Vin voltage to 6V and take the same capture again?

    This doesnt make any difference to the reboot behaviour. (waveforms attached below for Vin = 5.0V and Vin = 5.8V)

    • Just to make sure this is not happening on a single failure device, could you repeat the same debug on a second (brand new) PMIC? 

    Tested 3 different boards, and the behaviour is consistent across boards

    Vin = 5.0V, PB_IN floating, Supply via AC pin

    Vin = 5.6V, PB_IN floating, Supply via AC pin

  • I have taken some additional waveforms too. Please see if this may be of use.

    Figure 1. BAT pin voltage (Vin = 5.0V, Supply via AC pin, USB not connected)

    Figure 2. Expanded the glitch in BAT pin voltage

    Figure 3. BAT pin voltage (Vin = 5.0V, Supply via AC pin, USB also connected)

    Figure 4. LDO3 and LDO1 during power down (Vin = 5.0V, Supply via AC pin, USB not connected)

    Some additional queries related to these graphs

    1. Figure 1 --> What is this glitch in the BAT pin voltage after 1s? Is this expected? (note that the BAT pin is grounded via 1K resistor)

    2. Figure 4 --> Why is the LDO3 and LDO1 voltage powering down at the same instant? Does the PMIC not follow power down sequence? (Note that during power up the waveforms are rising at different intervals)

    Our custom design follows the PMIC section as in BeagleBone schematic, with minor changes. However, the same linux image in Beaglebone doesnot exhibit this reboot phenomenon. 

    Notable deviations in HW are:

    1. We have implemented the voltage supervisor + load switch solution

    2. BAT pins and BAT_SENSE are grounded via 1K resistor(in BBB, they are floating)

  • Hi Venugopal,

    Thanks for sharing the latest results! I will be going through the information and will provide an update within the next 2-3 business days. 

    Thanks,

    Brenda

  • Hi Venugopal,

    Sorry for the delay on this reply, I got involved in some critical time sensitive activities and bandwidth was very limited. Just wanted to check if there has been any progress from your end on the root cause of this issue. Additionally could you please check what UVLO you are using? This information can be accessed by reading register 0x18 (UVLO Control Register), bit 1-0 (UVLO[1:0]). The TPS65217C comes with 3.3V UVLO as default but would like to check if this value has been changed. 

    Thanks,

    Brenda