nRF52832, nRF5340 current consumption comparison

[John Lee]

It is an extension of the ticket below.
devzone.nordicsemi.com/.../nrf5340-current-consumption-issues
I'd like to find out what causes the huge difference in current consumption on the two boards.

An online power profiler (OPP) was used to estimate the radio current, and PPK2 was used to measure the actual current consumption of the board.

<52832>




OPP: 77uA
Measurements: 72.23 uA
Measurement_base: 3.04 uA

<5340>




OPP: 56uA
Measurements: 85.87 uA
Measurement_base: 6.76 uA

OPP simulations suggest that the 5340 will have lower current consumption, but measurements do not.
Comparing the currents over one connection interval clearly shows the difference. Both radio/idle the 5340 has high current consumption.

<52832>



Interval (60 ms): 70.28 uA
Peak (5 ms): 809.6 uA
Base (55 ms): 3.06 uA

<5340>



Interval: 78.53 uA
Peak: 870.45 uA
Base: 6.54 uA

I'd like to know where this difference comes from.
Is it simply because the 5340 has two cores?
52832, IDDFLASHCACHEDCDC: 3.7 mA
5340, IAPPCPU3 + INETCPU2: 3.6 mA + 2.6 mA = 6.2 mA

I tried to compare the current consumption of the two mcu peripherals in the datasheet, but the 5340 doesn't provide enough information. (gpiote, pof...)
I want to make sure I've achieved sufficient current optimization.

[Hieu]

Hi John,

I would like to think the Online Power Profiler has already taken into account the consumption of both cores.

I have asked internally on what could be going on here and will update you when I get feedback.

Meanwhile, may I ask if you have performed power optimization of both the nRF52832 and nRF5340 to the same degree in your tests?

Hieu

[Hieu]

Hi John,

Unfortunately, we don't find anything that immediately stands out. Could you please give us some details about what application you used to get the nRF5340 measurement? We can try to replicate it.

Also, am I right to understand that you don't have a concern about the estimation and actual measurement of the nRF52832?

Hieu

[John Lee]

The 52832 board is in production and assumes proper power optimization has been made.
We are using 5340 for new project and we are optimizing towards 52832 boards.

It's not clear if the same optimizations were applied.
Rather, it's something I'd like to check out.
I applied power optimization by referring to the Nordic blog,
I'm new to NCS so maybe there's something I'm missing.
(As noted in the previous ticket, the 52832 uses nRF52 SDK 17.1.0 and the 5340 uses NCS 2.3.0.)

Is it possible with the 5340 to achieve the same level of power consumption as the 52832?
I need to check if it consumes higher power because of the physical (electrical) characteristics of the MCU or if there is an optimization that I did not apply.

Provides the config file from the 5340 project build.
Please make sure I have configured proper optimizations.
config_230816.zip

[Simonr]

Hi

By "the Nordic blog" I assume you refer to the optimizing power on nRF5340 SoC designs guide, correct? In your config file I see that CONFIG_SERIAL is set to Y, which is a common cause for increased current consumption as it is part of the configs required to use serial logging. Try disabling this to make sure logging is properly disabled to see if that affects the current consumption on the nRF5340. Other than that I didn't spot anything on first glance that should be causing this.

Also make sure all unused peripherals are disabled on both the APP and NET core in your nRF5340 project. This can be done with the DeviceTree GUI in the VS code nRF Connect extension for example, where you get an overview of what processes/peripherals are enabled in a specific project and makes it easy to see what's enabled and not.

Best regards,

Simon

[John Lee]

Yes, I configured power optimization by referring to that blog.

I have configured CONFIG_SERIAL=y, CONFIG_UART_ASYNC_API=y only for app core because my project contains uart serial communication.

Disabled all peripherals not used by the app and net core.

John Lee said:

Is it possible with the 5340 to achieve the same level of power consumption as the 52832?
I need to check if it consumes higher power because of the physical (electrical) characteristics of the MCU or if there is an optimization that I did not apply.

Can you answer the above question?

Simonr said:

Other than that I didn't spot anything on first glance that should be causing this.

If so, I'll conclude that I've achieved maximum optimization.