NRF52840 too high power consumption

RE: NRF52840 too high power consumption

jonas.woerner — Wed, 10 Apr 2019 13:27:59 GMT

I will mark this thread as answered and submit a sperate question for the cryptocell problem, as I think, others might be interested too.

RE: NRF52840 too high power consumption

jonas.woerner — Wed, 10 Apr 2019 09:53:47 GMT

[quote userid="14926" url="~/f/nordic-q-a/45912/nrf52840-too-high-power-consumption/181072"]I would expect mbed to use software cryptography for MBEDTLS by default[/quote]

I verified it, mbedTLS definitly uses the Cryptocell hardware, you even have to initialize it, but you can turn it off and use software only if you wanted to.So I guess I will have to somehow disable the cryptocell310 before entering systemOFF.

RE: NRF52840 too high power consumption

Jørgen Holmefjord — Tue, 09 Apr 2019 13:08:43 GMT

I would expect mbed to use software cryptography for MBEDTLS by default, are you sure it is using the cryptocell? There is an errata related to FPU usage, but this should give you considerably higher current draw.

1.9 uA sounds a bit high for System OFF, but it could be possible with the right configuration. See the sleep current consumption in the electrical specifications. The PPK might also have difficulties measuring lower currents.

RE: NRF52840 too high power consumption

jonas.woerner — Tue, 09 Apr 2019 12:54:07 GMT

OK. I just verified it, its indeed the pullups and the LED, if I use P012, I get a poweroff current of 1.9uA.

however, I just noticed another problem, in my main program, I still see a constant current of about 38uA, but only if I used cryptographic functions prior.. Do I have to manually turn off the cryptocell hardware?

anyways, I will also ask this in the mbed forums, as its related to MBEDTLS.

one question for you though, Is 1.9uA an acceptable systemOFF current? I feel like it should drop even lower.

thanks for pointing me in the right direction!

RE: NRF52840 too high power consumption

Jørgen Holmefjord — Tue, 09 Apr 2019 11:40:19 GMT

In general, we do not provide much support on mbed, as it is being developed by 3rd party. I do not know how InterruptIn is implemented in mbed. Have you tested with other pins? P0.13 is connected to VDD through LED and pull-up by default on the nRF52840 (P)DK.

RE: NRF52840 too high power consumption

jonas.woerner — Tue, 09 Apr 2019 11:35:12 GMT

I did not configure any wakeup source for testing to eliminate that issue.

I can give you a minimal project, BUT it uses MBED to configure a Interrupt pin, as I will use mbed for my main project, the functions to power off the chip is pure nordic SDK though.. is this ok for you, or is mbed a no-go?

also, I just found out that if I just make an instance of the InterruptIn the current will drop to 1.8uA, but if I configure its mode (pullUp/down/none) the current will stay at about 40uA.

#include "mbed.h"
#include "PinNames.h"
#include "nrf_soc.h"


InterruptIn irt(P0_13);
void take_nap(){
    NRF_POWER->SYSTEMOFF=1;
    sd_power_system_off();      //consumes slightly less power
    while(1);
}

int main(void)
{ 
    NRF_POWER->DCDCEN = 0x00000001;
    NRF_UART0->TASKS_STOPTX = 1;
    NRF_UART0->TASKS_STOPRX = 1;
    NRF_UART0->ENABLE = 0;
    
    irt.mode(PullNone); //when using this, it will consume more power.

    while (true) 
    {
       take_nap();
    }
}

thank you for your help,

Jonas

RE: NRF52840 too high power consumption

Jørgen Holmefjord — Tue, 09 Apr 2019 11:23:36 GMT

Hi,

How did you configure the wakeup pins? You should make sure that the pins are not configured in a way that will draw current through pull-ups, etc.

Can you provide a minimal project that can be used to reproduce this issue on the DK?

Best regards,
Jørgen