Hi,
we are trying to use the PPK2 to supply and profile a custom device with audio amplifiers. During our tests a 400Hz (+harmonics) noise showed up our our audio path when the PPK2 was used to measure the current.
The same signal can also be extracted from the measured current samples themselves (using numpy FFT on the recorded CSV data).
It also can be reproduced with a simple setup shown below:
Simplified Setup:
- Connect the PPK2 to a PC with one USB cable to the Data/Power connector
- Connect a plain resistor (used a 680 ohm here) between Vout and adjacent Gnd pin on P1.
- Use nRF connect for PC to collect data (100kSamples, 60s duration, 1.2V output voltage, output supply ON)
- export and analyze the resulting samples.
Using this setup, a constant current would be expected.
Used devices/ SW: Windows 10, nRF connect 4.2.0, Power Profiler 3.5.5, PPK2 HW 1.0.1
Suspected Cause:
After some more investigation, it seems that this is caused by the onboard RGB leds pulsing during measurement (which is done with a 400Hz PWM signal, as seen on TP 37/38/39).
The noise at 400Hz pulsates with the LEDs fade, peaking during the fade's 50% point.
I tried to find the root cause, but it actually is hard to measure the noise on the output itself.
It is visible on some points on the PPK2 itself, e.g. the VREF_IA.
However, a quick test showed that the noise disappeared when the RGB LED is disconnected (cutting through TP 37/38/39, which cuts the trace from the LED control MOSFET to the LED itself).
It also is not visible in our full setup's audio path, when the PPK2 is not measuring (-> LED stay solid red, no PWM).
Measurement Data:
current samples from CSV, numpy FFT (raw current data is ~100MB but can be provided if needed)
voltage on VREF_IA:
--(below is another setup/measurement run)--
measurement on our custom device audio path, unmodified PPK2:
meaurement on our custom device audio path, PPK2 with LED traces cut:
note: there seems to be yet another noise at 200 Hz, but that is affecting our measurement much less, since the harmonics do not reach the 5kHz test tone, and another component which seems to be related to USB (carrier always visible at 8kHz, spectrum widening with data transfer)
Proposed Solution:
- For future HW revisions, the decoupling of LED control and measurement / reference supply could be investigated and fixed.
- For current devices, an easy fix would be to include a firmware fix to disable the LED fade (or the LED output completely).
Could you please add this as a setting to the firmware?
Best regards,
Patrick
