I'm debugging a problem on the negative output of a TPS65131 design. The output has intermittent oddly-shaped edges when the internal switch is turning on (transition from the negative voltage back to the supply voltage on the switching node). See the attached scope screenshot. These effects come and go over time and their severity varies. Most scope triggers do not include the "soft" edges and the rising edges of the waveforms are as straight as the falling edges shown in the screenshot. There also appears to be some variation of the duty cycle, as I can see the rising and falling edges do not switch at constant points in time.
The design uses a 5V source supply from a bench linear, output is -3V, input and output caps are 10uF and 20uF X7R respectively, diodes are MBRM120, inductors are MSS6132 4.7 uH. The positive side is also being used to make a +7V supply, and it appears to be okay with possibly a small amount of duty cycle variation (less than in seen in the negative supply). Both sides are followed with linear regulators: load is about 130 mA on the positive side and 180 mA on the negative side.
I've tried removing and changing the value of the CN cap (nominally it's 3.3uF, removing it makes the output go very unstable, setting it to 4700 pF as on the EVM board shows the same results as the screen shot). I've also tried removing the feedback speed up cap (C10 on the schematic on page 10 of the TPS65131 datasheet): this also didn't change the behavior.