lowest possible current consumption when system is running (nRF52810)

RE: lowest possible current consumption when system is running (nRF52810)

buja — Fri, 12 Mar 2021 15:55:55 GMT

Thank you for your reply Simon. I will try to get the measurement done on an ammeter as suggested and will let you know

RE: lowest possible current consumption when system is running (nRF52810)

Simonr — Fri, 12 Mar 2021 09:59:32 GMT

Okay, then this is related to the HW or your measurement setup it seems. I would like you to try using an Amperemeter as suggested, as that should give you a steady current draw of ~0.8uA while asleep. If not, that indicates an issue with the HW of your device.

Best regards,

Simon

RE: lowest possible current consumption when system is running (nRF52810)

buja — Thu, 11 Mar 2021 14:13:38 GMT

Hi Simon,

Thank you for your reply. I will confirm the STM32CubeMonitor-Power dependency.

I checked the current consumption in systemOFF and noise is there too. Please see the below given figure. The average current sometimes even go higher.

All the GPIOs are in the default state (I hope it is not floating).

int main(void)
{

// Initialize.
log_init();
power_init();
power_management_init();
enable_softdevice();

while(1) {
nrf_pwr_mgmt_shutdown(NRF_PWR_MGMT_SHUTDOWN_GOTO_SYSOFF);
}

The poser consumption graph is for the above code.

RE: lowest possible current consumption when system is running (nRF52810)

Simonr — Thu, 11 Mar 2021 12:57:06 GMT

Hi,

I'm not familiar with the STM32CubeMonitor-Power, and from its features I only see the expected accuracy for specific development kits. Can you confirm whether it is specified for measuring power consumption on custom devices as well?

From the graph this seems like noise/interference of some kind, and there could be quite a few things causing this. To exclude firmware for instance, you can try going to system OFF to see if there are any spikes when measuring there as well.

It could also be that one or more of your GPIOs are floating. It could also be the equipment you're using. If you measure using an Ampere meter for instance you can confirm whether this is the case, as an ampere meter is generally better at showing the average current consumption over time.

Best regards,

Simon

RE: lowest possible current consumption when system is running (nRF52810)

buja — Wed, 10 Mar 2021 19:14:53 GMT

Hi Somon,

Thank you for your reply.

Please find the snapshot from STM32CubeMonitor-Power. Device is always in systemON with a timer of 5s and a gpioTE enabled. GpioTE will not be activated at anytime.

You can see in the above figure that there is some irregularity(marked) in the current consumption. Sometimes it increases and the average current increases. What could be the reason for this ?

RE: lowest possible current consumption when system is running (nRF52810)

Simonr — Mon, 08 Mar 2021 08:11:51 GMT

1. For (the best) timer option during sleep, please check out the ION_RAMOFF_RTC_LFXO option in sleep, compared to ION_RAMOFF_EVENT (1.1uA VS. 0.6uA), so less than 1.1uA can't be expected. Can you provide a graph showing how this varying current consumption looks and is behaving?

2. Where exactly have you seen these answers? The sd_app_evt_wait() function uses RAM retention by default, and you'll have to disable which regions you'd like to not retain RAM in. Please check out the RAM Retention example and/or the answer by my colleague Vidar in this case to get more information on how RAM retention works. Please also note that you need to know what RAM sections you can disable without interfering with the application data.

Best regards,

Simon

RE: lowest possible current consumption when system is running (nRF52810)

buja — Sat, 06 Mar 2021 12:05:17 GMT

Hi Simon,

thank you for your reply.

1) I checked current consumption for various scenarios in systemON and it is same with timer/gpioTE and without timer/gpioTE.

clock source is XTAL

External clock accuracy used in the LL to compute timing is NRF_CLOCK_LF_ACCURACY_20_PPM

Actually the current consumption is not stable and is varying from 850nA to 1.1uA from time to time. Do you have any suggestion ?

2) I saw in many answers in forum that calling "sd_app_evt_wait" will put device in system ON without RAM retention. But when i read the register "RAM[n].POWER (n=0..7)" before calling sd_app_evt_wait(), the value is 0xFFFF which means that RAM will be retained according to the spec. Then how RAM doesn't retain data during system ON?

From the above table I am assuming that disabling RAM retention during systemON is only possible by disabling the power to the RAM. Is my understanding correct ? Also, if RAM is not retained, Will the code be moved from flash to RAM each time device is waking up from systemON ?

Best regards,

Nizar

RE: lowest possible current consumption when system is running (nRF52810)

Simonr — Fri, 05 Mar 2021 08:21:36 GMT

1. By default most of our examples are configured to not use RAM retention. If you're using a timer for wake up, that is why you're seeing ~1uA. Please check out the various sleep current consumption scenarios here.

Best regards,

Simon

RE: lowest possible current consumption when system is running (nRF52810)

buja — Thu, 04 Mar 2021 12:41:10 GMT

hi,

Thank you for your reply.

1) May I know how to configure systemON with no RAM retention. I dont know whether currently my system is using RAM or not during systemON ? I am currenly calling function "idle_state_handle() --> _WFE" to put the device in systemON. Also, I am using a timer/gpiote to wake up from systemON.

2) I am using softdevice and I hope clock configuration is made in the function nrf_sdh_enable_request().

.source = NRF_SDH_CLOCK_LF_SRC, --> 1 (NRF_CLOCK_LF_SRC_XTAL)
.rc_ctiv = NRF_SDH_CLOCK_LF_RC_CTIV, --> 0
.rc_temp_ctiv = NRF_SDH_CLOCK_LF_RC_TEMP_CTIV, --> 0
.accuracy = NRF_SDH_CLOCK_LF_ACCURACY --> 7

RE: lowest possible current consumption when system is running (nRF52810)

Simonr — Thu, 04 Mar 2021 07:35:25 GMT

The lowest expected current consumption in System OFF is 0.3µA, so I don't think you'll be able to get much lower than that. In system ON with no RAM retention, Wake on any event, you should be able to reach ~0.6µA, but there might be something else on your board drawing this extra current. Can you provide some more information regarding your system ON mode? What LF clock is your application using for instance, the LFXO crystal or the internal LFRC?

No, you can not reduce the clock frequency, as they are required to keep track of timing and staying inside the BLE spec.

Best regards,

Simon