Different Current Readings with Different Sample Counts in NRF Power Profiler Kit 2

[Jameson]

HI TEAM,

I am using a custom board and measuring its overall current consumption using the NRF Power Profiler Kit 2 (PPK2).

Setup details:

• Power source: External 3.6 V C-size battery (9000 mAh) connected to Vin and Gnd in power profiler kit-2. And Vout and gnd connected to my custom board input supply.

• Measurement mode: Ampere meter mode (as recommended in the PPK2 user guide when using an external supply)

• Tool: NRF Power Profiler Kit 2

• Reference instrument: Fluke 17B+ multimeter.

Issue observed:

When changing the number of samples in the PPK2 software, I see different average current values:

• With 100,000 samples, current consumption is around 4–5 mA

• With 10,000 samples, current consumption drops to around 1–3 mA

• Changing the sample count to other values gives different current readings

This creates confusion about which reading represents the actual current consumption of my board.

Additionally, when I measure the same board using a Fluke 17B+ multimeter, the measured current is significantly different from the values reported by the PPK2.

My questions:

1. Why does changing the sample count in PPK2 affect the measured current?

2. Which sample count gives the most accurate and reliable current measurement?

3. How should the PPK2 readings be interpreted compared to a traditional multimeter?

4. What is the recommended method or configuration to obtain accurate and consistent current measurements using NRF Power Profiler Kit 2 for a custom board?

Any guidance or best practices for accurate power measurements with PPK2 would be greatly appreciated.

Thank you.

[Simon D-M]

Hi Jameson,

Why does changing the sample count in PPK2 affect the measured current?

The number of sample per second is the number of discrete measurement the PPK2 takes per second. So having a smaller sample rate is comparable of having a low-pass filter on your measurement. If you have current spikes (or drops) of duration smaller than the time between 2 samples, it will most likely not be registered.

Which sample count gives the most accurate and reliable current measurement?

The rule of thumb would be : The more sample per second you can have, the better it is.

A thing to be careful about is not having a sampling rate too close to a periodic event that is happening. This phenomenon is called "Aliasing" (Wikipedia). So, if you have a process that is going at a frequency close to 100 kHz, you might want to get a longer measurement for a better average.

How should the PPK2 readings be interpreted compared to a traditional multimeter?

These are not used for the same purpose.

The PPK2 is used to characterize current consumption or even debug stuff by looking at the current consumption. It can measure smaller current (sub-µA) with a higher precision, and has a sampling rate which is probably higher than a traditional multimeter.

Multimeters are more versatile. They are very good on measuring things "on-the-go". But (most of the time) cannot be trusted when the measured signal is too erratic. They are giving good measurement when the signal is clean (Fully flat or a clean Sinus). They, most likely, also don't have a good way to log the data automatically. 

What is the recommended method or configuration to obtain accurate and consistent current measurements using NRF Power Profiler Kit 2 for a custom board?

To get the best results, you should probably set the PPK2 to the highest sampling rate, and do the measurement over a relatively long period (it depends on what you are measuring).

However, a higher sampling rate also means a bigger file to store the data. Be sure to have enough space for the measurement on your computer.

If you have more questions, feel free to ask !

Best regards,

Simon D-M

[Jameson]

Hello team,

I am characterizing the power consumption of a custom embedded board using the NRF Power Profiler Kit 2 (PPK2) and the nRF Connect for Desktop Power Profiler app (v4.3.1).

Setup Configuration:

• Mode: Ammeter Mode

• Power Source: External 3.6V 9Ah Battery

• Hardware: Custom board (microcontroller-based)

Observation: I am seeing a consistent current spike for every 16 seconds. However, the reported peak magnitude of this spike changes drastically depending on the selected sampling rate in the app:

• 100k samples/s: Sudden Peak current = 8mA(every 16 seconds and peak lasts for 200microsecond)

• 10k samples/s: Sudden Peak current = 4mA(every 16 seconds and peak last for 1millisecond)

• 1k samples/s: Sudden Peak current = 0.9mA.

Questions:

1. Is this variance purely due to the sampling interval missing the peak of a high-speed pulse (aliasing/averaging), or is there an internal filtering mechanism in the PPK2 software that scales based on the sampling rate?

2. Given the 16-second regularity, are there known internal nRF Connect or PPK2 calibration routines that occur at this interval, or should I focus entirely on my firmware/hardware of  my custom board?

Any insights on how to properly interpret these discrepancies for a professional power report would be appreciated.For your referene i have attached image below for different sample test.

[Simon D-M]

Hi Jameson,

Jameson said:

Is this variance purely due to the sampling interval missing the peak of a high-speed pulse (aliasing/averaging), or is there an internal filtering mechanism in the PPK2 software that scales based on the sampling rate?

Yes, I strongly believe that what you are seeing here is due to under-sampling. You can most likely trust 100k sample/sec measurement.

You can see in this picture that if you highlight one sample every 10 samples you get a wave that is way closer to the one you measure at 10k samples/sec.

Jameson said:

Given the 16-second regularity, are there known internal nRF Connect or PPK2 calibration routines that occur at this interval, or should I focus entirely on my firmware/hardware of  my custom board?

No, you don't need to do anything on the PPK2. All the current spikes measured are caused by the thing you are measuring (in your case, it will be your custom board). You probably want to look at your code first, if anything is done every 16 seconds that is most likely the cause of those spikes.

Best regards,

Simon D-M