nrf52 high current after sd_power_system_off

[yaron]

Hi I am trying to put the nrf52 device to system off state, in order to get minimal current consumption. When I don't use the soft device, and enter system off state by "NRF_POWER->SYSTEMOFF = 1", I see that CPU current goes down to a few uA. But when I enable the softdevice (using "SOFTDEVICE_HANDLER_INIT"), and than enter system off state by ""softdevice_handler_sd_disable" & "sd_power_system_off", I see CPU current of 5.83 mA!!!. If I add "NRF_POWER->SYSTEMOFF = 1" after "sd_power_system_off" , the CPU current goes down to 2.46 mA. I am currently using soft device version s132_nrf52_1.0.0-3.alpha. Is this a known issue with this version? If not, can some please explain? What is the correct way to enter system off state after soft device is enabled?

Thanks Yaron

[yaron]

Hi Visti I am calling SOFTDEVICE_HANDLER_INIT with "NRF_CLOCK_LFCLKSRC_RC_250_PPM_1000MS_CALIBRATION" option. (Like you, I also don't have a 32Khz xtal on my board). As I see it, if you don't call SOFTDEVICE_HANDLER_INIT, the soft device doesn't work, so no wonder that you get low current. I don't know what happens if you use a clock source which doesn't exist (do you get an error code from SOFTDEVICE_HANDLER_INIT?) Yaron

[Visti Andresen]

I don't think that calling the macro generates any errors (I haven't noticed any resets caused by it). But the blue tooth stack will not function with this setting, as there are no valid LF clock source.

But the differences i power drains caused by different settings seems to indicate to me that the cause of the high power drain is somehow clock related.

[yaron]

It is possible. Maybe someone from Nordic can try and test this scenario

[Visti Andresen]

When using LFCLKSRC some instructions after NRF_POWER->SYSTEMOFF = POWER_SYSTEMOFF_SYSTEMOFF_Enter; (but not for ever) LFCLKSRC_XTAL does not seem to execute commands after OFF (even with the SWD interface still attached)

[Visti Andresen]

I have tried to switch to nRF5_SDK_11.0.0-2.alpha_bc3f6a0 (I had a semi finished port to the alpha SDK 11 laying around)

And while it still is a semi dirty hack I have managed to get the softdevice up and running including advertisements.

But most importantly it now returns to ~ 1uA after a "NRF_POWER->SYSTEMOFF = POWER_SYSTEMOFF_SYSTEMOFF_Enter;" (While using the RC clock source for the LFCLK)

I strongly suspect that the non alpha version of the SDK/soft device also works as well.