NRF52832 low power mode / sleep mode [ZephyrsOS]

RE: NRF52832 low power mode / sleep mode [ZephyrsOS]

Simonr — Thu, 06 Nov 2025 13:47:35 GMT

Hi Robert

The 32.768kHz crystal connected to P0.00 and P0.01 will also not be counted by the nRF52832 while in system OFF mode, as the CPU is turned off entirely.

1.44mA sounds very high if all peripherals are turned off, so I think something hasn't been shut down correctly, that tthe sys_poweroff() is taking care of for you. It's hard to say based on this what exactly, so I suggest you create a new DevZone ticket where you describe in detail what you're doing, what you're trying to do, and what goes wrong exactly.

Best regards,

Simon

RE: NRF52832 low power mode / sleep mode [ZephyrsOS]

Robert de Brum — Wed, 05 Nov 2025 20:29:59 GMT

I have a similar situation that is making me arrive at this same conclusion:

I need to maintain the RTC even in "low-power" mode so that I can maintain the day/time on an nrf52832-based custom board. I am understanding I will likely need to implement this in software without entering system off mode.

I have seen in several blogs that I cannot use sys_poweroff(), as this will halt all clock sources. However I do have an external crystal hooked up to p0.00 and p0.01. Is it true that I still can't use sys_poweroff() even if I have an external clock source?

When I use sys_poweroff() I can boot back up on button press and I see 46uA drawn while in system off mode. This is good enough for now and I think I can get it lower. However when I replace sys_poweroff() with k_sleep(K_FOREVER) I see 1.44mA drawn. I know my hardware is different but if I am still powering down all peripherals the same way, what would be the reason for the difference between the sys_poweroff()drawing 46uA and k_sleep drawing 1.44mA?

RE: NRF52832 low power mode / sleep mode [ZephyrsOS]

Simonr — Wed, 05 Feb 2025 07:34:37 GMT

Hi Fernando

Glad to hear a lot here is resolved on your end with testing.

Regarding system ON mode, k_sleep should be sufficient I believe. The RTC should be used if you need a counter to keep track of time or similar while in sleep mode, but should not be necessary in otherwise.

Best regards,

Simon

RE: NRF52832 low power mode / sleep mode [ZephyrsOS]

Fernando Fontes — Tue, 04 Feb 2025 15:10:40 GMT

Hi everyone,

Following up on this investigation, I’ve conducted several tests:

Unsoldered the TDC1000 from the PCB
- No improvements. The average current remained around 42µA.
- However, I no longer needed to configure cs_gpio as INPUT to achieve this current.
- This suggests that the TDC1000 is responsible for drawing 6mA when the pin is not set as input, likely due to leakage.
Unsoldered the LIS3DH from the PCB
- Significant improvements. The average current dropped from 42µA to 2.5µA, which is close to the expected minimum when not using the DC-DC converter.
Unsoldered the Quectel modem from the PCB
- No improvements.

Since the LIS3DH’s expected consumption in low-power mode is about 1µA, the 42µA reading didn’t make sense. With the LIS3DH removed, the current consumption was as expected, confirming that the I2C pull-ups were not the issue. However, if LIS_INT is at logic level 0, a current of 33µA is expected—adding this to the 2.5µA explains the measured increase.

To further investigate, I re-soldered the LIS3DH and removed R30, which resulted in an average current consumption of 3.26µA.

At this point, I believe the issue is mostly resolved, and I don't expect to get much lower consumption. I also attempted to set the LIS_INT pin to OUTPUT_HIGH before going to sleep, expecting it to resolve the issue, but it didn't. Since the correct current consumption is observed with the LIS3DH removed, it seems that the LIS3DH itself is pulling INT1 low. I’ll continue investigating potential configurations within the LIS3DH to address this (change the interrupt polarity for instance).

This brings me close to a final solution. However, I have one remaining question:
I can't use system off since I need periodic wake-ups. What is the best approach to optimize power consumption in system on mode? I assume calling pm_device_action_run(..., PM_DEVICE_ACTION_SUSPEND) is the correct approach, but what’s the best way to sleep until the next wake-up? Is k_sleep() sufficient, or should I rely on the RTC for better efficiency?

Best regards,
Fernando Fontes

RE: NRF52832 low power mode / sleep mode [ZephyrsOS]

Fernando Fontes — Thu, 30 Jan 2025 16:06:29 GMT

Hi Simon,

What do you suggest to check if the current is actually going to the nRF52? Cut the CS line and measure there?

The TDC1000 which is where the CS pins connects to is being powered via two different pmos:

Since the pmos is off there's no voltage going to the supply pins (VDD1, VDD2, VIO). There is always some leakage but not enough to power the TDC1000. That's why it's super strange the 6mA. Shouldn't the gpio_pin_configure_dt(&cs_gpio,GPIO_OUTPUT_INACTIVE) make the pin to low logic level? Meaning, 0V? That would make almost impossible for current to be flowing trough the CS pin.

devzone.nordicsemi.com/.../ppk2_2D00_6mA.ppk2 devzone.nordicsemi.com/.../ppk2_2D00_50u.ppk2

I've uploaded the 6mA (1 min) version and the 50uA (3 min) version. Sorry, reduced to 1 min when I saw the file size :).

Best regards,

Fernando Fontes

RE: NRF52832 low power mode / sleep mode [ZephyrsOS]

Simonr — Thu, 30 Jan 2025 12:14:39 GMT

Thank you. I discussed this with a colleague today, and have some more suggestions here. Have you checked that the 6mA actually goes to the nRF52. That this current is actually drawn by the nRF52 here and that it goes over the P0.18. Please also make sure that the sensor P0.18 is connected and powered by the same VDD as the nRF52, as powering the board through the CS pin might result in something like that.

Can you export and upload the .ppk files here so we can take a closer look at the power consumption you're seeing. I'm glad to read you've found some sort of a workaround, but the 6mA floor current definitely shouldn't be there, so there is something causing this.

Best regards,

Simon

RE: NRF52832 low power mode / sleep mode [ZephyrsOS]

Fernando Fontes — Wed, 29 Jan 2025 15:07:13 GMT

Ok, so I've been playing around with the gpio configuration by calling the gpio_pin_configure_dt() and this was what I got:

gpio_pin_configure_dt(&cs_gpio, GPIO_INPUT):

gpio_pin_configure_dt(&cs_gpio,GPIO_OUTPUT_INACTIVE):

I've measured with the multimeter and oscilloscope and when the P0.18 is configured as GPIO_OUTPUT_INACTIVE the voltage is around 2V and with GPIO_INPUT is 0V. It doesn't make sense for GPIO_OUTPUT_INACTIVE to make the voltage around 2V right?

CS line is only connecting the MCU and the TDC1000. Nothing else connected on that line.

I also don't have any clue from where does this 1mA pulses are coming from. At the beginning I thought it was the LIS3D accelerometer, but that are the 400uA. Disabling the accel by setting ODR to 0 gives me this:

Is the CPU waking up?

With the following code, I can fix the 6mA issue, since I force the pin to go to input mode before sleeping.

/*
 * Copyright (c) 2019 Nordic Semiconductor ASA
 *
 * SPDX-License-Identifier: Apache-2.0
 */

#include <inttypes.h>
#include <stdio.h>

#include <zephyr/device.h>
#include <zephyr/drivers/gpio.h>
#include <zephyr/drivers/sensor.h>
#include <zephyr/kernel.h>
#include <zephyr/pm/device.h>
#include <zephyr/sys/poweroff.h>
#include <zephyr/sys/util.h>
#include <zephyr/pm/device.h>

#define SPI_NODE DT_NODELABEL(spi1)

static const struct device *spi_dev = DEVICE_DT_GET(SPI_NODE);
static const struct gpio_dt_spec cs_gpio = GPIO_DT_SPEC_GET_BY_IDX(SPI_NODE, cs_gpios, 0);
const struct device *const sensor = DEVICE_DT_GET_ANY(st_lis2dh);

int main(void)
{
	struct sensor_trigger trig;
    int rc;

    trig.type = SENSOR_TRIG_DATA_READY;
    trig.chan = SENSOR_CHAN_ACCEL_XYZ;

	// turn off
    struct sensor_value odr = {
        .val1 = 0,
    };

    rc = sensor_attr_set(sensor, trig.chan,
                         SENSOR_ATTR_SAMPLING_FREQUENCY,
                         &odr);

	pm_device_action_run(spi_dev, PM_DEVICE_ACTION_SUSPEND);

	gpio_pin_configure_dt(&cs_gpio, GPIO_INPUT);

	sys_poweroff();

	return 0;
}

I'm also attaching the zephyr.dts and zephyr.map of this build.

devzone.nordicsemi.com/.../7725.zephyr.map devzone.nordicsemi.com/.../1057.zephyr.dts

Best regards,

Fernando Fontes

RE: NRF52832 low power mode / sleep mode [ZephyrsOS]

Fernando Fontes — Tue, 28 Jan 2025 14:07:36 GMT

Hi Simon,

Thank you for the ongoing support.

The example prints a message every time it goes into sleep mode. I was trusting that. Then I moved to my original setup and I’ve disabled the logs in the conf.proj file.

I’m using the power profiler kit 2 in voltage mode to measure the power consumption. The programmer is not connected. I’ve this difference from 50uA to 6mA when the SPI is enabled with the CS pin on the P0.18 port.

Tomorrow I will be on the lab and I will measure the pin to try to get more insights on what is happening when it goes to sleep mode.

I will also try to upload the files later today or tomorrow.

So, there is nothing obvious regarding the P0.18 that could be causing this?

Best regards,

Fernando Fontes

RE: NRF52832 low power mode / sleep mode [ZephyrsOS]

Simonr — Tue, 28 Jan 2025 12:12:29 GMT

How do you verify that you're in an idle mode on your end here. How do you measure current consumption, and can you confirm that a debugger isn't attached and running or that debug mode isn't enabled, as that would cause pin P0.18 to be active.

Have you checked the signal on P0.18 when seeing this high current consumption to see what's happening on that pin. Try setting the P0.18 to an idle level and see if that affects the current consumption.

Can you upload the zephyr.dts file of these two builds here so we can see what the differences are. Please also upload the memory map for reference.

Best regards,

Simon

RE: NRF52832 low power mode / sleep mode [ZephyrsOS]

Fernando Fontes — Mon, 27 Jan 2025 15:50:40 GMT

Ok, I did some follow up investigation and the problem is not the target board, both nrf52dk_nrf52832.overlay or nrf52kbd_nrf52832.overlay work the same on the system_off example.

The problem is actually on the proj.conf file. More specifically I did one by one enable until finding the issue and noticed the following:

CONFIG_SENSOR=y will increase the power consumption from 40.6uA by around 12uA. Later on, when enabling the CONFIG_LIS2DH_ODR_1=y and CONFIG_LIS2DH_OPER_MODE_LOW_POWER=y there was a reduction of 6uA. So, I guess this increase is resulting by the sensor callbacks being triggered?

Anyway, the big issue is the CONFIG_SPI=y. This makes the power consumption to go for the 6mA. I did some investigation and I think the issue has something to do with this line: cs-gpios = <&gpio0 18 GPIO_ACTIVE_LOW>. If I use any other gpio from my debug pins, it works just fine. It goes to sleep mode with the 50uA.

P0.18 on the nrf52832 is the SWO pin. Maybe, the sleep mode on the SPI puts this pin in some state that conflicts with the SWO definitions? What can I do to fix this? Changing the hardware is not really an option at the moment.

(No, external pull-ups on the pin).

Best regards,

Fernando Fontes