2018-03-12, 08:02 PM
Sounds to me like a classic case of a 'dirty' output from a switching converter/regulator. However it might be fixed or at least lessened by adding some additional filtering at the output of the converter and a little more at the Raspi ..... When i come across seemingly random glitches in a device and it uses a switching rather than a linear supply thats the first thing I start looking at. That's why I suggested using an o-scope in the AC coupled mode to look at the output while under load. A lot of them use a switching frequency of around 100 - 200 kHz. Some of the newer chips use higher frequencies into the low Mhz but those usually aren't as much of a problem which is why they went to the higher frequency because it makes the filtering easier and the components smaller and more SMD friendly