Some of the most useful measurements in a machine are ones you’re not allowed to make by touching the thing. The clearance between a spinning tool and the workpiece. The radial wobble of a turbine shaft in its bearing. The position of a hydraulic spool buried in oil. Put a contacting probe there and you change the very thing you’re measuring — or it gets torn off. This is the home turf of the inductive (eddy-current) displacement sensor, and I spent a study digging into why it’s built the way it is.
The principle is small. Drive an alternating current through a coil and it sets up a magnetic field; bring a conductive target close and that field induces eddy currents in the target, which react back on the coil and change its effective inductance. Inductance becomes a stand-in for distance. Move the target a few microns and the coil’s inductance shifts measurably — with nothing mechanical crossing the gap.
Here’s the part that’s easy to skip past: you can’t read that with a DC voltage divider. A coil’s “signal” is its inductance, which is a reactance — it only does anything when the current is changing. So the sensor has to be excited with AC, and the inductance change is teased out as an impedance measurement, classically by sitting the coil in a bridge circuit. A balanced bridge turns a tiny inductance imbalance into a clean differential voltage you can amplify, instead of trying to spot a fractional-ohm change against the coil’s own ~200 Ω resistance. The choice of AC excitation isn’t an implementation detail; it’s forced by what you’re sensing.
What you buy with all that is robustness. The probe sees a magnetic effect, so oil films, dust, and metal swarf are basically invisible to it — the same grime that blinds an optical sensor. That’s exactly why these show up in machine-tool clearance monitoring, turbomachinery vibration and eccentricity, valve and spool position, and automotive suspension-height sensing — dirty, sealed, high-reliability spots. (Commercial eddy-current systems aren’t cheap, either: a new industrial unit runs well over a thousand dollars, which tells you where they earn their keep.)
The trade-offs are the flip side of the principle: the target has to be conductive metal, and the useful range is short, because the eddy-current coupling falls off fast with distance.
The takeaway that stuck with me is a small reframe. The moment you’re not allowed to touch the target, you stop thinking in volts and start thinking in impedance — and a surprising amount of instrumentation is exactly that move: pick a physical effect that crosses the gap on its own, then build the whole signal chain around reading it.