Power consumption using FreeRTOS

[shai.c]

Hi,

I'm using an application with FreeRTOS, nRF52833, SDK 16.

When I'm trying to get it into low power (sleep mode) the system still consumes ~4mA

In vPortSuppressTicksAndSleep (port_cmsis_systick.c file) I forced to use WFE instead sd_app_evt_wait

#if 0  
    #ifdef SOFTDEVICE_PRESENT  
    if (nrf_sdh_is_enabled())  
    {  
        uint32_t err_code = sd_app_evt_wait();  
        APP_ERROR_CHECK(err_code);  
    }  
    else  
#endif  
    #endif  
    {  
        /* No SD -  we would just block interrupts globally. 
        * BASEPRI cannot be used for that because it would prevent WFE from wake up. 
        */  
        do{  
            __WFE();  
        } while (0 == (NVIC->ISPR[0] | NVIC->ISPR[1]));  
    }  

After changing it the power consumption was reduced by ~3.5mA

What is the proper way to fix this problem?

[Susheel Nuguru]

I cannot think of anything.very specific.

Please check if you are hit by ERRATA  78 or 87 as mentioned here

[shai.c]

Hi,

I checked ERRATA 87 - CPU: Unexpected wake from System ON Idle when using FPU, according to the Workaround the following code need to be added before calling WFE 

#if (__FPU_USED == 1)
 _set_FPSCR(_get_FPSCR() & ~(0x0000009F)); 
 (void) __get_FPSCR();
 NVIC_ClearPendingIRQ(FPU_IRQn);
#endif

Does this code need to be added also before calling sd_app_evt_wait? ( I'm using SOFTDEVICE)

however, I tried to add this before calling sd_app_evt_wait but it didn't help

Any other suggestions?

[shai.c]

Hi Susheel Nuguru 

Any suggestions?

[Susheel Nuguru]

Hi Shai,

That seems to be a lot of current from a peripheral keeping the HFCLK ON. That is not normal and is not consistent with the testing I have done here on the examples.

Are you testing this on nRF52833 DK board or is it a custom board? Have you disabled all tasks in your application and let the idle task run only  to see how much current the idle application consumes?

[shai.c]

Hi Susheel Nuguru

I'm using nRF52833 DK running ble_app_hrs_freertos example.

1. when using the following code:

        if ( xModifiableIdleTime > 0 )
        {
#ifdef SOFTDEVICE_PRESENT
            if (nrf_sdh_is_enabled())
            {
#if (__FPU_USED == 1)
             __set_FPSCR(__get_FPSCR() & ~(0x0000009F)); 
             (void) __get_FPSCR();
             NVIC_ClearPendingIRQ(FPU_IRQn);
#endif  __FPU_USED
                uint32_t err_code = sd_app_evt_wait();
                APP_ERROR_CHECK(err_code);
            }
            else
#endif  // SOFTDEVICE_PRESENT
            {
                /* No SD -  we would just block interrupts globally.
                * BASEPRI cannot be used for that because it would prevent WFE from wake up.
                */
                do{
                    __WFE();
                } while (0 == (NVIC->ISPR[0] | NVIC->ISPR[1]));
            }
        }

the average power consumption is: 5.794mA

2. when using the following code (forcing to use WFE):

        if ( xModifiableIdleTime > 0 )
        {
#if 0
#ifdef SOFTDEVICE_PRESENT
            if (nrf_sdh_is_enabled())
            {
#if (__FPU_USED == 1)
             __set_FPSCR(__get_FPSCR() & ~(0x0000009F)); 
             (void) __get_FPSCR();
             NVIC_ClearPendingIRQ(FPU_IRQn);
#endif  __FPU_USED
                uint32_t err_code = sd_app_evt_wait();
                APP_ERROR_CHECK(err_code);
            }
            else
#endif  // SOFTDEVICE_PRESENT
#endif  // if 0
            {
                /* No SD -  we would just block interrupts globally.
                * BASEPRI cannot be used for that because it would prevent WFE from wake up.
                */
                do{
                    __WFE();
                } while (0 == (NVIC->ISPR[0] | NVIC->ISPR[1]));
            }
        }

the average power consumption is: 102uA

Which example did you test?

Thanks, Shai

[Susheel Nuguru]

I am sorry, I realized that my new nRF52833 board did not have the SB bridge cut to make the power measurement.

In your latest reply you say that you see a significant improvement with WFE, then this is something we saw before

I suspect that this is because of the use of MWU inside of sd, that causes this spike as seen in this thread. 

Try disabling the MWU before sd_app_event_wait and enabling just after wakeup in the port_cmsis_systick.c file like below

// replace sd_app_evt_wait with
/* MWU Disable */
NRF_MWU->REGIONENCLR
    = ((MWU_REGIONENCLR_RGN0WA_Clear << MWU_REGIONENCLR_RGN0WA_Pos) 
     | (MWU_REGIONENCLR_PRGN0WA_Clear << MWU_REGIONENCLR_PRGN0WA_Pos));

sd_app_evt_wait();

/* MWU Enable */
NRF_MWU->REGIONENSET
    = ((MWU_REGIONENSET_RGN0WA_Set << MWU_REGIONENSET_RGN0WA_Pos) 
     | (MWU_REGIONENSET_PRGN0WA_Set << MWU_REGIONENSET_PRGN0WA_Pos));

I have not tested this since myself, but if this did not work, then I can fix my board and try to dig a bit deeper into it tomorrow.