With a recent issue with my PC, I recently narrowed it down to PSU issue after seeing 400mV of peak-to-peak, so I figured why not find a way continuously monitor that in the future. I recall increased ripple is an early sign of certain failing components so I figure why not monitor that. While I intend to start with 12V, I do plan to expand that to other voltages, but just starting with 12V for now.I assume you are talking about an ATX power supply unit / PSU. Otherwise, if it's a power brick for say a Mini PC, it won't matter much because the 12V will still be stepped down inside the Mini PC.
Looking at 120Hz ripple due to bridge rect + caps? Not sure if it matters that much in terms of PC stability because the 12V is usually stepped down further. And failing caps can usually be seen to be bulging. I did change a single low ESR cap in an ATX power supply because it was probably in a poor location and got a lot of hot air due to nearby power FETs. That was back in 2013, after ~4 years of use and the AM2 PC was starting to exhibit instability. I want to blame that PSU's PCB layout for the failed cap because it was the only ATX PSU problem I ever had.
It doesn't really make sense to check the 12V line all the time -- I find it strange, to be honest. But lately there is news of Firefox crashing a lot possibly linked to Blue Team's parts (search for e.g. tom's hardware firefox crashing). Or maybe you are concerned about the 12V under heavy load. So I'm really more interested in the actual problem you are trying to solve... some sort of PC instability?
I think an oscilloscope will be better if you just want to look at the 12V shape. I might use a INA219 breakout board (I2C to Pico) to measure voltage instead, it's cheap and bus voltage LSB is 4 mV using a range of 4096 counts and sampling rate will be ~2 kHz. {Edited to add:) And INA219 is calibrated, accuracy is 0.2% typical. For a lot of use cases, modern parts like INA219 will outperform a bunch of discrete parts and a breakout board will get the job done with little fuss.
I was wrong about 120Hz though; I previously read the scope incorrectly and somehow made assumption that it came from rect. Took a closer look and it isn't 120Hz.
Statistics: Posted by ferrum — Mon Jul 21, 2025 2:51 am