What is the time accuracy of data recorded with the Nordic Power Profiler Kit II?

[Alphanoob]

I am using the Nordic Semiconductor Power Profiling Kit II to measure the interval time of a periodic activity of a target device by monitoring the power consumption and measuring the time from one peak in power consumption to the next one. I found a lot of documentation on the accuracy of the current measurement but I did not find any information about the accuracy of the time trace. I did not find any statement about the clock accuracy/relative accuracy of the time stamps in the documentation. Of course there are contributions to the uncertainty due to the evaluation method (i.e. limited accuracy due to the smoothing selected in the advanced settings, the algorithm for finding the peak maximum or edge and so on) which I need to account for myself. Also I understand that the absolute accuracy will be worse because it requires accounting for latencies, delays and synchronization between the clock of the host computer and the data which might introduce additional inaccuracies. So what is the accuracy of the time difference between two current samples in the measurement data?

[tesc]

Hi,

I am afraid we do not have numbers for the accuracy of the time difference between two current samples in the measurement data. What we do have, is the ±20 % accuracy of average current measurement, of which the time domain would be a factor. From that it follows that expected time accuracy is also within ±20 %.

Please note that the Power Profiler Kit II, while a good tool for measuring power consumption during development, does not replace high-quality lab equipment. For timing measurements, the recommendation is to toggle a GPIO and use proper lab equipment (e.g. oscilloscope, logic analyzer) with the wanted time accuracy.

Regards,
Terje

[PaKa]

The timing of the PPK measurements are controlled by the crystal on the device so as long as you don't have any packet losses you should be in >>>1% accuracy. I would expect this to be in the PPM range but as there is SW on top there is always the possibility for skews. Note: this is with no packet loss so you should check that the number of samples you have match what the time period measured should contain if you feel the data seems unlikely.

[Alphanoob]

Hello PaKa,

thanks a lot for that hint. I will look into the number of packets in order to check for any lost data. The accuracy I need is in the range of about 5 %. So I will be able to trust the PPK 2 data (of course not to the same extent as calibrated lab equipment but in I should be able to see functional issues with the device under test resulting in a wrong timing).

tesc : Since toggling GPIOs/voltages is not an option for my measurement I do have to rely on current measurement. So far I was not able to get my hands on "high-quality lab equipment" for measuring currents less than 1 mA with a time resolution better than 50 kHz and a specified clock accuracy. Even the big-brand power analyzers I looked at so far don't specify their clock accuracies.

Thanks!
Alphanoob