The current peak of 334uA during system on Low power mode

RE: The current peak of 334uA during system on Low power mode

Simonr — Fri, 16 Oct 2020 10:57:28 GMT

What does the APP_ERROR_CHECK return when you call ble_stack_stop()? Are you able to confirm that the SoftDevice is disabled as intended by this call? It might be that the SoftDevice isn't disabled successfully, and in order to see how, we need to see what values the ble_stack_stop() and nrf_sdh_disable_request() functions return.

Seeing as you seemingly have made some changes to your example, can you confirm that the nrf_sdh_disable_request() is identical to the default function found in nrf_sdh.c (see below), and that you haven't made any changes to the disable request?

ret_code_t nrf_sdh_disable_request(void)
{
    ret_code_t ret_code;

    if (!m_nrf_sdh_enabled)
    {
        return NRF_ERROR_INVALID_STATE;
    }

    m_nrf_sdh_continue = true;

    // Notify observers about SoftDevice disable request.
    if (sdh_request_observer_notify(NRF_SDH_EVT_DISABLE_REQUEST) == NRF_ERROR_BUSY)
    {
        // Disable process was stopped.
        return NRF_SUCCESS;
    }

    // Notify observers about starting SoftDevice disable process.
    sdh_state_observer_notify(NRF_SDH_EVT_STATE_DISABLE_PREPARE);

    CRITICAL_REGION_ENTER();
    ret_code          = sd_softdevice_disable();
    m_nrf_sdh_enabled = false;
    CRITICAL_REGION_EXIT();

    if (ret_code != NRF_SUCCESS)
    {
        return ret_code;
    }

    m_nrf_sdh_continue = false;

    softdevice_evt_irq_disable();

    // Notify observers about a finished SoftDevice enable process.
    sdh_state_observer_notify(NRF_SDH_EVT_STATE_DISABLED);

    return NRF_SUCCESS;
}

Best regards,

Simon

RE: The current peak of 334uA during system on Low power mode

Surya — Fri, 16 Oct 2020 05:29:18 GMT

maybe I added that function in the example. I am using SDK 17 (17.0.0_9d13099). anyway, I already mention I am calling "nrf_sdh_disable_request()" in that function.

the function goes as,

void ble_stack_stop(void)
{
uint32_t err_code;

err_code = nrf_sdh_disable_request();
APP_ERROR_CHECK(err_code);

ASSERT(!nrf_sdh_is_enabled());
}

RE: The current peak of 334uA during system on Low power mode

Simonr — Thu, 15 Oct 2020 13:00:39 GMT

l'm sorry, but what SDK version are you using? I'm not able to find any ble_app_pwr_profiling examples using the ble_stack_stop() function, please be more specific.

Best regards,

Simon

RE: The current peak of 334uA during system on Low power mode

Surya — Thu, 15 Oct 2020 06:38:07 GMT

Hi,

I called the function "ble_stack_stop()" provided in the power profiling example which calls "nrf_sdh_disable_request()".

but still, I am seeing spikes. if I do system-off the peaks will go but I want to do this in system_on mode(as per my requirement).

RE: The current peak of 334uA during system on Low power mode

Simonr — Wed, 14 Oct 2020 06:40:34 GMT

This peak is due to the SoftDevice waking up to calibrate the RC oscillator as I mentioned in my last reply. Stopping the BLE stack won't stop the SoftDevice from calibrating the RC oscillator. You can try disabling the SoftDevice altogether by calling the nrf_sdh_disable_request() to stop these spikes. You can also go to System OFF mode when putting the device to sleep, to achieve the lowest current consumption possible, but this will require an external trigger to wake the device up.

Alternatively you can edit the calibration parameters I mentioned to make these spikes occur less frequently.

Best regards,

Simon

RE: The current peak of 334uA during system on Low power mode

Surya — Tue, 13 Oct 2020 11:49:51 GMT

I am getting a current of an average of 1.9 uA to 2.1 uA with a peak of 334uA (a periodical of 4 sec).

before going to idle state by calling Pwr_mgmt_run() function, I am stopping the BLE stack, but still, the peak is coming.

RE: The current peak of 334uA during system on Low power mode

Simonr — Tue, 13 Oct 2020 11:40:51 GMT

The RC oscillator calibration can be adjusted by tweaking the calibration timer interval and calibration timer interval under constant temperature (NRF_SDH_CLOCK_LF_RC_CTIV and NRF_SDH_CLOCK_LF_RC_TEMP_CTIV in sdk_config.h). The SoftDevice should only wake up when it needs to calibrate the LFRC clock, but keep in mind that setting higher values than the recommended 16 and 2 (at least every 8 seconds within temperatures of 0.5°C and every 4 seconds if the temperature is changing) will most likely cause the oscillator to drift. Not calibrating the LFRC will likely lead to inaccurate timers in the low power mode for instance.

When the SoftDevice wakes up, the current peak will increase quite a bit, but the average current consumption during sleep when using the LFRC should not be much more than ~0.4µA according to the Electrical specification in the PS (see ION_RAMOFF_RTC versus ION_RAMOFF_RTC_LFXO for instance).

Best regards,

Simon

RE: The current peak of 334uA during system on Low power mode

Surya — Tue, 13 Oct 2020 06:35:45 GMT

hi Simonr, thanks for your reply.

as you can see my question is about if RC calibration is causing the current how to stop calibration. I don't want to change the clock option as I can't change the hardware now.

I want to know "how much current will be drawn for calibration of the clock?" if this is a known number.
I am seeing these peeks when I initialized and stopped the BLE which should not happen. is it like if once i configure i can't stop calibration?

RE: The current peak of 334uA during system on Low power mode

Simonr — Mon, 12 Oct 2020 13:18:27 GMT

As Turbo J says, the LF XTAL will reduce the overall current consumption significantly to any of the other LF clock options. You can check out the Online Power Profiler for BLE to see the differences between idle currents in a variety of configurations.

If you set the device to "system OFF" mode, the LF clock won't have to run, and the current consumption will be more similar for both instances, but in system ON mode an LF clock will be running, and if using the RC oscillator, it will have to wake up from time to time to recalibrate.

Best regards,

Simon

RE: The current peak of 334uA during system on Low power mode

Surya — Mon, 12 Oct 2020 11:33:28 GMT

hi, thanks for your reply.

can I know how much current is drawn during calibration?

do you mean to say it is not possible to stop calibration even when I am not using ble?

to do so there is no other way than going for a xtal crystal?

RE: The current peak of 334uA during system on Low power mode

Turbo J — Mon, 12 Oct 2020 11:18:37 GMT

RC calibration wakes just the HF crystal clock for a while in order to sync the clocks.

If you really care about idle power consumtion, just populate the LF XAL and use the crystal LF clock.