What's causing my buck converter's switching node to oscillate only some of the time?

Question

We got an order of 24 PCBs recently, and one of them doesn't work. I'm trying to track down the point of failure, and it seems to be our 3V regulator circuit:

At ~180ms intervals, the 3V output rail starts oscillating wildly, during which it drops below 2V and resets my microcontroller. It remains in this state for ~70ms, after which it returns to normal. Here's a scope capture. The green line is the 3V output and the yellow line is the switching node voltage:

During the failure event, the switching node voltage looks like this:

At other times, it looks perfectly normal:

I need to figure out what part is broken so we can rework the boards. My first guess is the IC itself -- the input voltage and enable signal look reasonable, the feedback lines up with the 3V rail, and the boost capacitor seems to be passing the output waveform, so what else is there? But I've never seen this behavior before, so I'd like to know if any other component failure could cause this.

(We have 30+ other boards working with this design, so I'm pretty sure it's a defect in a component or the assembly. I didn't see any bad solder joints.)

Answer 1

But I've never seen this behavior before, so I'd like to know if any other component failure could cause this.

All three scope shots depict your buck converter operating in discontinuous conduction mode (DCM) and pulse frequency modulation (PFM).

The sine wave you see mid-switch cycle is the near-zero current remaining in the inductor resonating with the parasitics on the sw-node (your scope probe being a large one). The converter opens the low-side NMOS near the zero-current crossing to prevent negative current conduction through the inductor.

Perhaps, you are seeing an abnormally light load on the 3V rail due to faulty down stream components, thus, causing the buck converter to behave differently on this board.

With good scope probing you may see your 3V rail rise sag a few 10's of mV, but it most likely isn't ringing at 1 MHz (that's an artifact of how you are probing your circuit).