Something didn't go according to plan with my latest project and destroyed a few components. The prime suspect is my recent modification to a MP1584 buck converter modules I bought on Amazon. By default, it would activate if input voltage is over 3V. This is far too low to deliver 3.3V output, so I modified it to defer activation until input voltage is nearly 12 Volts. A simple test with volt meter and simple LED was successful, but when I connected the ESP8266 microcontroller and INA219 sensor I released the magic smoke.

Before I connected those components I had checked the voltage output and it looked fine on a volt meter so I suspect there's something with a transition behavior. This calls for an oscilloscope and I have a cheap DSO 138 that is better than nothing. Since the time I bought the kit and put it together, something has gone wrong with its voltage reference and now absolute voltage readings are unreliable. However, I think the relative shape of the waveform still resembled reality.

I first checked typical startup behavior: what does the 3.3V output look like when I connect this modified buck converter to 13V input? I set the oscilloscope to hold data upon rising edge trigger, and captured this:

The shape looks reasonable. It corresponds to graph in the MP1584 datasheet and shows the advertised soft-start capability to avoid a voltage overshoot on startup, which was my first suspect. I then turned the oscilloscope to continuous scan mode and adjusted my power supply voltage up and down. When I stay well above the input enable voltage of ~11.8V everything looks good, the buck converter maintained a steady 3.3V output. (It reads 5.27V on the display due to the voltage reference weirdness I spoke of.)

But things started looking dicey as I dropped close to 12V. I started seeing a sawtooth pattern to the output voltage. Instead of holding at 3.3V, I started seeing the voltage drop then jump upwards, with the magnitude growing wider and wider as I dropped input voltage. The worst was when I dropped into range below the 11.8V activation voltage, but still staying above the shutdown voltage...

"Well, there's your problem."

Without accurate voltage reference I don't know exactly how many volts it whipped through, but at a guess it looks like it dips just below 2V before overshooting above 4V. This would definitely destroy components designed for 3.3V. It's great that I see what is happening, now I need to think about why it happens and decide what I will do about it.