PPK2 missing samples depending on DUT

[Johannes Feldmann]

Hi all,

I'm using the PPK2 to measure the current of our IoT device. To get the data from my PPK2, I wrote my own software in order to integrate it into our test setup.

However, during the analysis of the measured data, I noticed that there are sometimes missing samples. Digging deeper I figured out that there is a dependency between the number of missing samples and the variation of the measured current.

In the following, I present the number of missing samples for several scenarios (each ~30 min).

Target voltage: 3V

• No current (open circuit) → Missing samples: 0
• Fixed current (Resistor 10k) → Missing samples: 0
• IoT DUT / Internal Voltage Regulator (SMU) → Missing samples 0.05% - 0.07%
• IoT DUT / External Voltage Regulator (AMP) → Missing samples 0.03% - 0.07%
• IoT DUT + Filter (220uF, 22uF, 2.2uF, 1uF, 10nF) / External Voltage Regulator (AMP) → Missing samples ~0.005%

Is this a general problem of the PPK2 or am I doing something wrong?

[Simonr]

Hi

What exactly does the software you've written do? Is there a reason you're not using the Power Profiler app in nRF Connect for Desktop, or are you? I don't see why you would be missing samples depending on what the application is doing. The only thing I can think of is that the PPK2 is prone to some current spikes when switching between the voltage shifters, which your software might write off as "missing samples"?

Best regards,

Simon

[Johannes Feldmann]

My software configures the PPK2, starts the measurement, and logs the raw data into the memory. At the end of a run, I go through the data and have a look at the counter which is part of each sample.

According to https://github.com/NordicSemiconductor/pc-nrfconnect-ppk/blob/main/src/device/serialDevice.js a sample is 32 bits long and consists of the following data fields:

31 : 24	23 : 18	16:14	13 : 0
Digital Inputs	Counter	Range	Value

While inspecting my data I see missing counter values in all variations from one missing sample to 63.

So, if I understood you correctly, you think that the MCU software is skipping samples if there is a change in the shunt value?

Regards,
Johannes

[Simonr]

Hi

I just spoke to one of the developers to confirm my suspicion, and yes, there are a few samples that will fall out as the voltage range is switched, as these samples will come out invalid. There is also a counter in the payload that should make sure the SW corrects itself if samples are lost between USB and SW handling.

Best regards,

Simon