Turn off UART on nrf52832

[genis]

I noticed that having the UART enabled on my application in a NRF52 (SD132, SDK 11.0.0) takes a lot of current consumption. If I completely disable any UART support on my application the current consumption drops dramatically. I can't do that, but considering my device will be 99% of the time idle, I need to disabled the UART before I put the processor to sleep.

I am using FreeRTOS so what I did is I enabled the ApplicationIdleHook and did as follows:

/* FreeRTOS application idle hook */
void vApplicationIdleHook(void) {   
    // Disable UART transmission and reception
    bool uart_disabled = false;
    if (!nrf_drv_uart_tx_in_progress()) {
        NRF_UART0->TASKS_SUSPEND = 1;
        //NRF_UART0->ENABLE = 0; 
        uart_disabled = true;
    }

    __WFI();

    if (uart_disabled) {
        //NRF_UART0->ENABLE = 1;
        NRF_UART0->TASKS_STARTTX = 1;
        NRF_UART0->TASKS_STARTRX = 1;
   }    
}

That theoretically should suspend the UART but unfortunately it increases current consumption compared to simply doing __WFI() and leaving the UART as it is. Notice that if I enable the lines where I do NRF_UART0->ENABLE = 0 and NRF_UART0->ENABLE = 1 the UART simply stops working.

I wonder that is the right way to disable the UART when you go to sleep.

[Susheel Nuguru]

You are mixing up using UART driver and accessing registers directly. This will cause the unpredictable behavior you are seeing now. I recommend you to uninit UART and init again using the driver. Since you said that it will 99% in idle the overhead for uninit and init everytime it wakes up is negligible.

/* FreeRTOS application idle hook */
void vApplicationIdleHook(void) {   
    // Disable UART transmission and reception
    bool uart_disabled = false;
    if (!nrf_drv_uart_tx_in_progress()) {
        nrf_drv_uart_uninit(uart_instance);
        uart_disabled = true;
    }

    __WFI();

    if (uart_disabled) {
        nrf_drv_uart_init(xxx);
   }    
}

I think the reason that writing directly to ENABLE not working is because the pins are still configured to be connected to UART . Anyways, let the driver take care of this.

[genis]

Sorry for the late response. That actually disables and reenables the UART but I still see a lot of current consumption that I don't see if I totally disable the UART. I use the UART through app_uart_fifo.c library.

[Susheel Nuguru]

that depends on how much your system is sleeping which totally depends on your application. How much current difference are you observing?

[genis]

Hello Aryan, now I enable and disable the UART every single time I need to write something in it so I only do the __WFI() on the idle task. Still, I see the same behaviour. We're talking of about 700uA to 170uA when I completely disable any UART. Though I do not discard a mistake in the code, it looks like when the UART is disabled some hardware is still draining current. I'll try to make a simpler code sample to reproduce that. Also, the cpu is waken every 500ms by and also by the BLE hardware which advertises.

[Susheel Nuguru]

UART should not take more than 55uA current, I would like to see this problem at my desk, so it would be nice if you can reproduce it and give me sample code to test. I could then assist you much faster.