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.

AM4379: RTC problem

Part Number: AM4379
Other Parts Discussed in Thread: TMDSEVM437X, TMDXSK437X

Hi all,

I have a question that can not be backed up AM437x's RTC.

I tested RTC backup with the following two patterns.

(1) HW:TMDSEVM437X
SW:PROCESSOR-SDK-LINUX-RT-AM437X 03_03_00_04 & I use prebuild image.

I connected the battery(Coin Cell) to the EVM and turned off/on the AC adapter in that state.
While the AC adapter is not connected, VDDS_RTC keeps 1.8V.

With this configuration RTC is backed up as follows.

~#hwclock
Web Dec 14 23:26:29 2016 0.000000 seconds

------ Turn off/on the AC adapter ------

~#hwclock
Web Dec 14 23:27:05 2016 0.000000 seconds <-- "Time is progressing normally"


(2) HW:TMDXSK437X
SW:PROCESSOR-SDK-LINUX-RT-AM437X 03_03_00_04 & I use prebuild image.

As in (1), Coin-cell is connected.
While the AC adapter is not connected, VDDS_RTC keeps 1.8V.

With this configuration RTC does not back up as follows.

~#hwclock
Web Dec 14 23:26:29 2016 0.000000 seconds

------ Turn off/on the AC adapter ------

~#hwclock
Web Dec 14 23:26:12 2016 0.000000 seconds <-- "Time is back"


I think that the program for TMDXSK437X is overwriting the RTC register

How can I perform RTC backup on TMDXSK437X?

Best regards,

Sasaki

  • The factory team have been notified. They will respond here.
  • This sounds like an system level issue that has been reported and discussed by others on several E2E forum posts. If noise couples into the 32.768KHz crystal circuit, it may generate an internal glitch on the RTC clock which could cause the counters to jump back in time. In many cases, the problem was solved by selecting the internal RTC32KCLK source provided by PRCM. However, this may not be an option if you need the RTC to operate while the rest of the device is powered down. I recommend you search the E2E forums to understand what other people have done to solve this issue. If you must use the external 32.768KHz crystal circuit as your reference clock, you may need to find the noise source coupling into the crystal circuit and eliminate it.

    Regards,
    Paul
  • Paul-san,

    Thank you for your information.

    I was concerned about noise because the power supply configuration of EVM and SK are different.

    We will also refer to other E2E forums.

    Best regards,

    Sasaki

  • Hi peaves-san,

    In other words, if the noise of the SK board is removed, can I solve this RTC problem?

    Best regards,

    Sasaki

  • This question cannot be answered with a simple yes/no reply.

     

    The first step is confirming the problem is caused by noise coupling into the crystal circuit. The best way to confirm this is by temporarily modifying the board to replace the crystal circuit with a 1.8 volt 32.768 kHz LVCMS oscillator which will source a reference clock with very fast rise/fall transitions. The fast rise/fall time of the LVCMOS clock source will transition through the input buffer threshold so quickly it would be very difficult for noise to cause a problem. If this modification eliminates the problem, there is a very good chance the problem is caused by noise coupling into the crystal circuit.

     

    The next step for solving the problem would be finding the noise source and understanding how it couples into the crystal circuit. Once you understand these details, there is a chance you could change how you are using the board or make small modifications to the design that would eliminate the problem.

     

    Regards,

    Paul