Libuarte power consumption does not go down after uninit

[gaminn]

I use nrf_libuarte_async from nRF5 SDK v16.0.0, without softdevice on nrf52840. The idle power consumption of my board is several tens of uA. I call:

ret_code_t err_code = nrf_libuarte_async_init(&libuarte, ......);
APP_ERROR_CHECK(err_code);
nrf_libuarte_async_uninit(&libuarte);

and then the idle power consumption is several mA.

There must be some bug in nrf_libuarte_async. Can you please help me find it?

[Sigurd]

Hi,

From nrfx changelog, we have the following from version v1.8.2:

"Improved the UARTE driver to consume less current after the driver uninitialization. Now all clocks are disabled properly."

Note that SDK 16.0.0 uses nrfx v1.8.0, and SDK 17.0.0 uses nrfx v1.8.4

Can you check and see if you are able to reproduce the issue with SDK 17.0.0 ?

[gaminn]

Hi Sigurd,

power consumption with SDK16's nrf_libuarte_drv is 3.5 mA. When I replace nrf_libuarte_drv.c and .h with the one from SDK17 I get 2.8 mA. Obviously, there is some improvement but the problem still persists.

I don't want to switch to SDK17, that would mean heavy testing of all parts of my firmware.

[Sigurd]

Hi,

I have not able to reproduce these numbers you are measuring, so there could be something else happening here as well, but
could you try to replace the nrfx_uarte.c file with the one from nrfx v.1.8.4 ?
github.com/.../nrfx_uarte.c

[gaminn]

The problem is that libuarte doesn't deinitialize pins. This caused higher power consumption on our board.