Is it a good idea to use nrf52840 as soft power latching?

[addast]

I plan to use nrf52840 in high voltage mode (connected directly to the power source via VDDH pin) with button.

And I plan to use voltage regulator with EN pin (with pull-down resistor,  so by default regulator is off) for external circuitry.

I want it to work as follows:

• When I will press the button, nrf52840 should awake and then send high signal to the voltage regulator EN pin, after that everything should be working.
• When I will press the button one more time, nrf52840 should go to system-off mode and disable voltage regulator, so overall power consumption should be around 1uA (which is literally nothing for 400mA battery)

Is it a good idea to organize power management like that?

What should I use as voltage regulator in my case, ldo or dc-dc (max power consumption of external circuitry will be 40mA in worst case scenario, input voltage 3-4.2v)?

[Mishka]

Hi,

the proposed setup leverages from the extended power capabilities of the nRF52840 (and also 833 and 820), and should just work.

There are a number of things to keep an eye on though.

The application is going to have two independent VDD rails. Please note, some load (for example, an accelerometer) may require separate VDD and VDDIO lines. In this case disabling the voltage regulator won't remove power from the load and it still draw current from the MCU via GPIO lines such as I2C or SPI. Also, the independent rails (i.e. different voltages) will likely require logical level converters between the MCU and the circuit behind the voltage regulator (an accelerometer).

Another issue may come from the input voltage range 3-4.2V. It looks like it's going to be a rechargeable lithium battery with cut-off voltage Vcut = 3V. Please note, a switching DC-DC may imply some limitations to the output voltage, and the practical limit of the DC-DC may be settled somewhere at Vin + 0.3V or a above (depends on the DC-DC model). Speaking pragmatically, this is not bad at all because it will force the application to extend the battery lifetime (by preventing overdischarge), but will force the application to 3.3V or above. An LDO has no such specifics, but is less efficient (i.e. less time on battery) and transfers excess of power into heat which may eventually turn into its own issue.

If the load can work from the same range as the nRF52840 it might be a better idea to look for a voltage supervisor instead of a regulator. Splitting the I/O lines may be required anyway in order to prevent powering the rest of the circuit through MCU GPIOs.

[addast]

I plan to use internal and external regulator with the same output voltage - 3.3v, so I don't need logic converters.
I also plan to leave nrf52 vcc output floating, so I don’t have to deal with two vdd rails.
IMU that I use in my project have same voltage range for both VDD and VDDIO lines (can accept 1.7-3.6v), so I will just connect both lines to external regulator.

Is it a good idea with this clarifications?

[Mishka]

As far as I understood, there are two 3.3V regulators: one for the nRF52840, and another one for an IMU (and possible other load). The idea is to shutdown the IMU regulator from the nRF52840, right?

In this case there is no need to covert logic levels indeed. This circuit should work, but please note, on shutdown all other I/O lines (if any) between the MCU and the IMU must be pulled low or disconnected. Otherwise there is a risk to supply power to the IMU through the nRF52 GPIO which may eventually destroy either IMU, MCU, or both.