Apparent discrepancy between online power profiler & measurements

[jose123]

Dear Nordic support,

We are trying to know the power consumption of nRF52840 SoC, by prediction and measurements. 

The Online Power Profiler for Bluetooth LE tool estimates during advertising at 0dBm, for a 23 Byte TX payload, 3V supply and DCDC regulator ON, there should be 13.28uC of charge consumed per event, for nRF52840.

I ran the HRM example (ble_app_hrs_pca10059_s140) on a nRF52840-DK (PCA10056) and measured current consumption, however I found there was ~20.4uC per event.

1. Do you know why the current consumption (20.4uC vs 13.28uC) is around 54% more than what the Online Power Profiler tool estimates?
2. Do you know why the measured current signal has fairly smooth transitions, unlike the sharp transitions shown in this Nordic blog post? nRF52 Online Power Profiler, Stian Røed Hafskjold 26 Apr 2016. Is it because there is a capacitor that is smoothing out the sharp current changes during the advertising event?

Measurement setup

SB40 track cut.

Digital ammeter in series of P22 'nRF current measurement' port of nRF52840-DK, with USB power. Settings: 1MS/s, 10mA range, 10nA resolution.

total charge 20.44 uC
average current 3.75202 mA
total duration 5.449 ms

Online Power Profiler for Bluetooth LE result

[run_ar]

Hi,

1. did you enable dcdc and did you make sure the TX output power is the same as what was configured in the power calculator? Based in the peaks in the actual measurements I would say something is different in the setup her.

2. Yes, this looks like the capacitors smoothing thing out, vs the calculator that uses the measured average for the each given event.

[jose123]

Hi run_ar, thanks for getting back to me!

The example is unmodified ble_app_hrs_pca10059_s140.

From checking sdk_config.h, there are some flags set to 0. Do you think this means DCDC is off?

#define NRFX_POWER_CONFIG_DEFAULT_DCDCEN 0

#define POWER_CONFIG_DEFAULT_DCDCEN 0

I will try with them on later when I have access to the high resolution ammeter. 

Regarding (2), I see. But do you know why in the nordic blog current measurement plots, nRF52 Online Power Profiler, Stian Røed Hafskjold 26 Apr 2016. it doesn't look like capacitive smoothing?

[run_ar]

This settings means the Dcdc is disabled.

2. In the power profiler, we are using the average value for an actual measured event. I.e. measurements for the calculator was done using a dcdc power analyzer, then taking the average value for different states, packet lengths etc.