• State
  • Locked Locked
  • Replies 4 replies
  • Subscribers 98 subscribers
  • Views 244 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

AM62A7: Deep Sleep, Power off, Power consumption

user3775977
Prodigy 90 points

Part Number: AM62A7
Other Parts Discussed in Thread: SK-AM62A-LP

We are currently working on optimizing our (battery powered) product.
We are using SDK 11.01 now.

Unfortunately, we seem to be struggle with that.

1.Deep Sleep doesn't work  
    This has been documented in multiple post before and still doesn't seem to be fixed in SDK 11.01
    How do I send the device into deep sleep? (and yes, we are using the MJPEG encoder)
    Could you please provide a step-by-step guide how to do so?

2. Poweroff doesn't work
    The device is shutting down, but still consume the same power as before.
    Could you please provide a step-by-step guide how to do so?

3. The power consumption is too high
    From the documentation and the power calculation sheet, we would estimate ~600mW.
    (CPU 400MHz idle, C7 870MHz idle, R5 800MHz idle, VOUT enabled, LPDDR4 at 5% utilisation)
    Unfortunately, we are seeing much higher values (measured on 5V)
    ~270mV  -> Device is in reset (Baseline of our hardware)
    ~570mW -> Device is held in ROM-Bootloader (USB)
         -> This means, that the bootloader actively consumes 300mW - what seems reasonable.
     ~ 1450mW -> Device is held in tiboot3
         -> We are seeing an 880mW increase in power consumption, when SPL is initializing the system.
         -> According to all published information, we would expect 0.5W less than this and we are unsure, where this comes from. 
     Power consumption in Linux increases slightly, but not much and settles down to that, when the system is idle.
    Can you image a reason, why the power consumption is so high?
    We were suspecing the LPDDR4 or eMMC, but those should be idle.

Kind regards
Bernhard
    

  • 0
    TI__Mastermind 23895 points

    Hi Bernhard,

    Let me look into this and get back to you tomorrow.

    Best Regards,

    Anshu

  • 0 in reply to Anshu Madwesh
    TI__Mastermind 23895 points

    Hi Bernhard,

    Can you give some insight on your system use case?

    user3775977 said:
    Deep Sleep doesn't work

    There are some larger gaps in low power modes on AM62A to create a robust solution as it interacts with other system components. Please share a Deep Sleep log.

    user3775977 said:
    Poweroff doesn't work

    Are you trying to enter Partial I/O mode, just power off Linux, or power off the whole SoC?

    user3775977 said:
        From the documentation and the power calculation sheet, we would estimate ~600mW.

    Please share your filled out Power Estimation Tool.

    user3775977 said:
        Unfortunately, we are seeing much higher values (measured on 5V)

    Are you measuring on the supply of your PCB? If so, not just the SoC but all the other components on the PCB.

    Best Regards,

    Anshu

  • 0 in reply to Anshu Madwesh
    Prodigy 90 points

    Dear Anshu, 

    sorry for my late reply.

    We investigated deeper into that matter and it is all a complete mess.

    1. The firmware for the coprocessors was incompatible with deep sleep
        The firmware that TI delivers in the "ti-sci-fw"/"ti-dm-fw" package simply doesn't work with deep sleep.
        Instead you have to use the firmware from "ti-rtos-echo-test-fw"
        The manager or engineer, that made the decision to put the production firmware into a package named "echo-test", should just simply be fired.

    2. The SDK 10 "Analytics" release includes a kernel version that doesn't support the wakeup features
        The SDK10 "Analytics" release ships with Kernel 6.12.35, while the standard release ships with 6.12.57.
        So if you build a release with image processing support, you don't get wakeup capability, as described per the documentation.
        Our solution is to use the 6.12.57 kernel now (in the hopes, that there aren't any other problems)

    3. Power-Off simply doesn't work and leaves the system on

    4. The power consumptions is significantly higher than expected
         Unfortunately it is pretty hard to pinpoint what is the source of the problem, from the timing it might be an increased power draw, when the LPDDR4 is initialized.
         Please provide us with the exact steps to replicate the given current draw from the calculation sheet/app note with the AM62A-SK or any other platform you tested with. 

    Bernhard

  • 0 in reply to user3775977
    TI__Mastermind 23895 points

    Hello Bernhard,

    user3775977 said:
    Power-Off simply doesn't work and leaves the system on

    The linux command itself turns off the Linux OS, it doesn't shutdown the whole SoC. You can try to use Partial I/O mode if thats a requirement.

    user3775977 said:
       Please provide us with the exact steps to replicate the given current draw from the calculation sheet/app note with the AM62A-SK or any other platform you tested with. 

    On SK-AM62A-LP, there are current sensors across different SoC supply rails. The same rails are measured on the PET: https://www.ti.com/lit/an/sprad93b/sprad93b.pdf

    The PET is OS agnostic so it doesn't consider the boot process in the power consumption: https://www.ti.com/lit/an/sprad93b/sprad93b.pdf

    Best Regards,

    Anshu