nRF52832 current consumption

[mikaelbdl]

Hello

I’m trying out the new nRF52832 (QFAA-AA0) with preview kit pca10036. I have S132 softdevice installed and running the beacon example from SDK 0.9.0 which appears to be identical to the nRF51 application from SDK 8.x.

Issue 1: The nRF52x consumes 12uA more current compared to nRF51x on the PCA10028 kit running the same application. In both cases I have modified the application for advertising set at 450ms with BSP/LEDs off. The exact figures are nRF51 = 49uA averaged (7mA pk) and nRF52 = 61uA averaged (6.2mA pk).

An extra 12uA is fairly substantial but I can’t see anything in the errata that would explain this.

Issue 2: Attempting to enable the DCDC converter with sd_power_dcdc_mode_set(NRF_POWER_DCDC_ENABLE) results in high current consumption of 252 uA though I can see that the peak current has reduced to about 4.1 mA. The device still functions.

On the PCA10028 the device runs at 42uA with DCDC enabled (IC rev 3, S110 v8.0.0).

Again, I can’t see anything in the errata that would explain this.

Any ideas?

PS: Current readings done with Keysight 34450A multimeter.

code:

test_beacon_nRF52.zip

test_beacon_nRF51.zip

[mikaelbdl]

Just overlooked this. Could it be related? infocenter.nordicsemi.com/index.jsp

[Øyvind Karlsen]

Hi,

It is possible that that issue is related, if you provide some code, I will attempt to replicate your results.

Best regards,

Øyvind

[mikaelbdl]

I've attached the Keil projects. Only modifications from provided examples are to main.c. To compare devices for issue 2 you can un-comment DCDC mode set line in main().

[Øyvind Karlsen]

Hi,

First off, sorry for the late answer, this proved to be more difficult than first thought.

I've done measurements now, this is my setup:

• Keysight N6705B DC Power Analyzer

• Keysight E3646A DC Power supply

• nRF 51 DK

• 2x nRF 52 DK

• nRF 52 Testbench

This testbench is the reference layout for the nRF52 as well as a small sensor circuit that draws current in the ~200nA range. It also has a capacitor bank for measurements, more on that later. I used the E3646A power supply to provide 3V with a current draw limitation of 12mA.

For the measurements before enabling DCDC-mode my results in the idle state were:

• nRF51 DK - 3,23µA

• nRF52 DK 1 - 9,24µA

• nRF52 DK 2 - 39µA

• nRF52 Testbench - 1,8µA

These results may seem a little odd, however it is the result for the testbench that is valid.

First there's the error that stems from the high dynamic range between current draw for TX mode and idle mode. Essentially this will make fluctuations in the idle state invisible to our measuring equipment. This will lead to an error in our measurements, but it will not be the main contributor.

The nRF52 has a power supply that will rapidly charge a capacitor at a voltage threshold, this causes high frequency "spikes" in the current drawn. These spikes are so high frequency that most measuring equipment is not able to measure them accurately, in fact aliasing will occur and you will see values that are incorrect. This is where the large values and variations likely stem from. To circumvent this issue and the high dynamic range issue a capacitor bank is added to smooth out our measurements, this allows us to get stable readings. We're working to address this issue, stay tuned for more information.

Also note that the leakage current from the preview DK will be larger than those for the testbench, due to a higher amount of components which lead to a higher chance of error.

For the DCDC-mode these were my results:

• nRF 51 DK - 3,25µA

• nRF 52 DK 1 - ~220µA

• nRF 52 DK 2 - ~220µA

• nRF52 Testbench - ~220µA

This is where we probably encounter the error from the errata, further testing shows that this is 220µA current draw will disappear when adjusting voltage. We're also working on this issue, so again, stay tuned for more information.

Best regards,

Øyvind

[mikaelbdl]

Hi Øyvind,

Thanks for the detailed reply.

I’ll admit the 34450A multimeter is not the best instrument for the job. I had been using this for relative comparisons between nRF51 firmware setups hoping that 190 Hz sampling would be enough to get reasonable comparisons without the need of an amplifier/oscilloscope, or power analyser. It sounds like this assumption well and truly falls over when comparing nRF51 to nRF52 so I will address this more carefully.

Do you have detail on capacitor bank sizing and attachment?

I have a low cost 80 kHz current amplifier on order that may mimic performance of your power analyser more closely. I will attempt to retest using this and capacitor bank averaging when I get the chance.

Kind Regards,

Mikael