app timer current consumption

[jdasilva]

Hello,

I would like to use app timer based on RTC1 in order to wake up the MCU every X sec. The app timer seems to work correctly when the MCU is working. I use the following code to initialize it :

void init_timer()
{
APP_TIMER_INIT(APP_TIMER_PRESCALER, APP_TIMER_OP_QUEUE_SIZE, false);
//APP_TIMER_TICKS(1, APP_TIMER_PRESCALER);
APP_TIMER_DEF(Avid_timer_id);
uint32_t err_code;

// Create timers
err_code = app_timer_create(&Avid_timer_id,
APP_TIMER_MODE_REPEATED,
Avid_timer_handler);
APP_ERROR_CHECK(err_code);

// Start timers
err_code = app_timer_start(Avid_timer_id, APP_TIMER_TICKS(1, APP_TIMER_PRESCALER), NULL);
APP_ERROR_CHECK(err_code);

Avid_ticks = ctr_tick;
Avidflag_PC.my_flag.app_timer_enable = 1;

}

With this the handler is called every 1 ms.

When I am in power down down mode without app timer (using app_timer_stop_all(); to stop it) the average current consumption is around 30µA. 

When I am in power down mode with app timer, the average current consumption is around 110 µA. 

Do you know what is wrong in my software and cause this high current consumption?

For information, I use the following code to enter in low power mode

static void power_manage(void)
{
//Used to stop interrupt flag of the floating co processor
__set_FPSCR(__get_FPSCR() & ~(0x0000009F));
(void) __get_FPSCR();
NVIC_ClearPendingIRQ(FPU_IRQn);

//Enter in low power mode
uint32_t err_code = sd_app_evt_wait();
APP_ERROR_CHECK(err_code);
}

By advanced, thank you for your help.

BR.

[AmbystomaLabs]

The RTC draws negligible current.  Only 0.1uA according to the data sheet.

I have used it on several projects and never had any issues with a power manage and getting the current to go to about 1.9uA for the entire device.

Based on your assertion that the power down number is 30uA, I would guess you are doing something wrong with your current measurement.  The spec shows about 1.2 to 1.5 for a system off.

Measuring microamps requires a really precise DMM with very little offset.  I have a stand alone bench unit just for that purpose. If that is not in your budget you can probably do decent measurements with a series resistance and an o-scope.  Some people in the DevZone use the PPK and seem to like it.

Also possible your choice of app ticks is too small.  This drives the minimum resolution of the app timer. Normally people set it large to keep the cpu from waking up too much.  Can't tell from your code since it's not there.

[jdasilva]

Hello,

Thanks for your reply.

To measure the current I have a board that can measure nA, so the issue is not due to the precision of the measure, I used it for severals other projects without any problem.

30µA is due to the rest of the board (the average current is 20µA due to the nrf52 plus 10 µA consumed by the rest of the board).

Effectively increasing the ticks the current decrease:

- 1 tick/ms => 90µA in sleep mode

- 1 tick/10ms => 14µA in sleep mode

- 1 tick/100ms => 6.5µA in sleep mode

- App timer off => 5.3µA in sleep mode

Thank you for your help.

BR

[AmbystomaLabs]

In addition to the sleep current being lower for longer period app ticks, the latency will be better too. 

If you use a soft device, the app ticks are not as high a priority as the SD ISR's so if you use a small period for the tick, each tick can be off by upwards of a msec depending on SD activity.  Whereas with a 100msec tick the same is true that it can be off by about 1 msec.

So, longer period for tick, means less latency and a more accurate tick time.  Most of the examples use 100msec tick time.