A 277.5 kHz Spur: Which Path Put It in the Measurement?

An 18-bit evaluation signal chain showed a spur around 277.5 kHz at roughly −115 dBFS. The easy label was “power noise.” The useful question was: which path put it into the measurement?

The published investigation worked that question through. Shielding the switching-power area removed the spur. The eventual source was an 8.2 μH switching-power inductor whose emitted energy was coupling into an analog buffer input.

The source was switching activity. The susceptible node was an input stage, not automatically a supply rail. The artifact was a visible spur. Treating those as one undifferentiated “power problem” would have made the repair guesswork.

That is the diagnostic order I trust when a measurement chain develops a repeatable artifact: name the source, prove the coupling path, identify the susceptible node, then confirm how that node becomes the observed artifact.

A rail filter may be useful when the path is conducted. It is not the first answer when the path is radiated into an input node. Shielding, placement, filtering, or a component change can each be sensible, but only after the path has made one of them more than a guess.

The useful question is not “how clean is the supply?” It is “what path lets this switching energy become this measurement error?”

That question is specific enough to test, and it gives the remedy a chance to be specific too.