Strange current spikes in nRF52 with RTC only

[Mateusz Gancarczyk]

Hi, currently I'm working on ultra low power device. This device need to be wake up in 333 ms interval. I use RTC for this. My question is about strange current spikes with i observed on Nordic Power Profiler Kir ( nRF6707 ). What is a source for this spikes?

This spikes have interval ~25ms. Average current looks quite OK: System on ( 1.2uA)+ RC ( 1uA ) + RTC ( 0.1 uA ) = 2.3 uA I'm not use any soft device. I use internal RC clock source for LFC.

My code is very simple: main:

int main(void)
{	
  NRF_POWER->DCDCEN = 1;
  lfclk_config();               
  rtc_config();
  
  while (1)
  {
		power_manage();	
  }
}

rtc_config:

static void rtc_config(void)
{
	uint32_t err_code;

	nrf_drv_rtc_config_t config = NRF_DRV_RTC_DEFAULT_CONFIG;
	config.prescaler = 64; // 32 768 Hz / 38 = 1024 Hz
	err_code = nrf_drv_rtc_init(&rtc, &config, rtc_handler);
	APP_ERROR_CHECK(err_code);
	nrf_drv_rtc_cc_set(&rtc,0, MEASURMENT_TIME_MS, true);
	nrf_drv_rtc_counter_clear(&rtc); 
	
	nrf_drv_rtc_enable(&rtc);
}

RTC handler, power manage and lfck config:

#define MEASURMENT_TIME_MS 333 
const nrf_drv_rtc_t rtc = NRF_DRV_RTC_INSTANCE(2); 

static void rtc_handler(nrf_drv_rtc_int_type_t int_type)
{
  if (int_type == NRF_DRV_RTC_INT_COMPARE0)
  {					
    nrf_drv_rtc_cc_set(&rtc,0, MEASURMENT_TIME_MS, true);		
    nrf_drv_rtc_counter_clear(&rtc); 
  }
}

static void power_manage( void )
{
	// WFE - SEV - WFE sequence to wait until a radio event require further actions.
	__WFE();
	__SEV();
	__WFE();
}

static void lfclk_config(void)
{
    ret_code_t err_code = nrf_drv_clock_init();
    APP_ERROR_CHECK(err_code);

    nrf_drv_clock_lfclk_request(NULL);
}

Does anyone know what could be the source of this spikes? RC temperature corecction? Nordic power profiler kit bug?

[MartinBL]

Hi,

Those short spikes are probably due to the nRF52's power supply working in ref refresh mode. Here is a quote from the current measurement guide:

The nRF52 power supply uses auto-controlled refresh modes to maximize efficiency. What this basically means is that a capacitor is recharged regularly to provide power to the chip. The recharge off the capacitor happens during a very short time interval, and results in a high frequency peak in current. This peak is measured to be about 10 - 15 us long. In order to get the correct average current during these peaks, a sampling frequency corresponding to half the peak length is required. This results in a minimum frequency of 200k samples per second (1 sample every 5 us).

BTW, aren't your RTC intervals closer to ~650ms? With a prescaler of 64 and a compare register value of MEASURMENT_TIME_MS = 333, you will get a f_RTC 32.768 / (64 + 1) = ~504 and hence, an interval of 333/504 = ~660ms.