Add Metal Where the Probe Should Vanish

Waveguides don’t let you look inside. The electromagnetic field propagates through a rectangular metal tube—22.86 × 10.16 mm for X-band, precision-machined from aluminum or brass—and to measure what’s happening you must cut a slot along the centerline and insert a probe. But the probe itself disturbs the field. Metal in the electric field stores energy as capacitance, creating a reactive load that couples to the wave and changes the standing wave pattern you came to measure.

The solution is counterintuitive: add more metal. Specifically, an adjustable stub—a short length of transmission line attached near the probe—tuned to present an inductive reactance that exactly cancels the probe’s capacitance. The result is a high-impedance shunt: electrically present but minimally coupled, extracting just enough power to drive a Schottky diode detector while leaving the wave mostly undisturbed.

This is probe tuning, and it represents the central technical problem in slotted line interferometry. You cannot measure without observing, and observation requires interaction. The discipline lies in making that interaction small and predictable.

The Probe Coupling Problem

Insert a bare probe into a waveguide and you create a capacitor: the probe tip is one plate, the waveguide walls are the other, and the dielectric is air (or vacuum). At 10 GHz this capacitance is small—maybe 0.2 to 0.5 pF depending on probe diameter and insertion depth—but at microwave frequencies even femtofarads matter. Capacitive reactance at 10 GHz for 0.3 pF is about 53 ohms, comparable to standard system impedance.

The capacitance appears in shunt across the waveguide. Forward power divides: some continues down the line toward the load, some gets diverted into charging and discharging the probe capacitance. This diversion creates an additional reflection beyond whatever the load itself reflects. The standing wave pattern you measure now includes contributions from both the load mismatch and the probe mismatch. Without correction, the measurement includes the instrument.

Worse, the capacitive load is frequency-dependent (Xc = 1/(2πfC)), so the perturbation changes with frequency. A slotted line calibrated at 10 GHz reads incorrectly at 9.5 or 10.5 GHz unless the probe is retuned.

Tuning Out

Inductive reactance has the opposite frequency dependence: XL = 2πfL. At a given frequency, you can choose L such that XL + Xc = 0, creating a series resonance. The probe and stub together present infinite impedance (ideally), becoming electrically invisible to the waveguide while still extracting enough power for detection.

Practical implementations use an adjustable stub—usually a short section of coaxial line with a movable sliding contact. Lengthening the stub increases its inductive reactance. The procedure: set the source to your measurement frequency, insert the probe to operating depth (typically 1mm past the outer conductor for coax, 1mm into the waveguide for rectangular guide), adjust stub length while monitoring detector output, and watch for the peak sensitivity point where inductance cancels capacitance.

Sensitivity peaks because at resonance the impedance presented to the field is purely resistive (the power extracted by the detector diode) without reactive coupling. The detector sees maximum signal for a given insertion depth, and the field sees minimum perturbation.

Insertion Depth Trade-offs

Probe depth determines extracted power. Push deeper and the diode gets more signal, improving signal-to-noise ratio for weak sources. But deeper insertion increases capacitance (larger electrode area), requiring more inductance to cancel, and increases resistive loading (the probe extracts power that the detector dissipates as heat in the terminating resistor). Too much extracted power creates a measurable reflection from the probe itself, defeating the purpose of tuning.

Standard practice: insert just far enough to get clean nulls on the VSWR meter. For rectangular waveguide this is typically 1mm past the slot, positioned on the centerline of the broad wall where electric field strength is maximum for the TE₁₀ dominant mode. For coaxial slotted lines, the probe tip should just penetrate the outer conductor without approaching the centre conductor—usually 0.5 to 1mm depending on line diameter.

The slot itself must be narrow (typically 1mm wide) to minimize field disruption. Wider slots interrupt the wall currents that support waveguide propagation, creating reflections. The slot is cut where currents run longitudinally so that current flow isn’t blocked. In rectangular waveguide’s TE₁₀ mode, that’s the centerline. Cut the slot off-centre and you interrupt transverse currents, creating a discontinuity that reflects power back toward the source.

What You Measure

Once the probe is tuned, wavelength measurement is direct. Apply RF power from a source (signal generator, klystron, Gunn diode oscillator), terminate the far end with a short circuit to create a standing wave, and slide the probe carriage along the slotted line. The VSWR meter shows voltage amplitude: maximum at antinodes, minimum at nulls. Two consecutive nulls are separated by exactly λ/2.

At 10 GHz, free-space wavelength is 29.98 mm. Inside X-band waveguide, guide wavelength is longer: λg = λ₀ / √(1 - (fc/f)²), where fc is cutoff frequency (6.56 GHz for X-band). Calculate: λg = 29.98 / √(1 - (6.56/10)²) = 39.7 mm. Half that is 19.85 mm, easily measured with a 0.1mm-resolution vernier scale.

This is interferometry: coherent wave superposition creating a spatial pattern, measured mechanically instead of optically. The thermal expansion interferometer that measured everything expanding at once used 589nm sodium light and counted fringes shifting as temperature changed. Here the wavelength is 67,000 times longer, the detector is a Schottky diode instead of an eye, and the measurement is stable because the wavelength itself is what you’re characterizing. Source drift doesn’t corrupt the measurement—it becomes the data.

The Observation Problem at 10 GHz

This is measurement discipline at microwave scale: the act of observing extracts energy and introduces coupling. Heisenberg’s uncertainty principle applies to photons at 589nm, but at 10 GHz the constraint is practical rather than quantum. You can minimize the disturbance with careful design—tuning out reactance, limiting insertion depth, cutting slots where currents permit—but you cannot eliminate it entirely. The probe must couple to the field to detect anything, and coupling means perturbation.

Thermal expansion interferometry broke because the optical flat sat on the aluminum bar, both expanded together, and the measurement captured their relative motion without a stable reference. Fixing that requires thermal isolation, which converts optical measurement into mechanical engineering.

Slotted lines solve this differently. The probe is always present, always coupled, and you accept that the standing wave you measure includes its contribution. Tuning makes that contribution predictable and minimal, a known systematic error rather than an unquantified disturbance. Then you calibrate: measure a known load (short circuit, open circuit, matched termination), characterize the probe’s effect, and correct for it in subsequent measurements.

The HP 423A on the bench came from a university surplus auction, $180 including coax adapters and a 1970s VSWR meter that still works. X-band only, 8.2 to 12.4 GHz. No sweep, no display, no stored calibration. Just a slot, a probe, a tuning slug, and a scale graduated to 0.1mm. Signal source is a surplus Gunn diode oscillator module pulled from a police radar gun, powered by a bench supply set to 9V. Output frequency measured with a frequency counter: 10.525 GHz, stable to within ±5 MHz over thirty seconds.

Tuning the probe took four minutes. Adjust stub, watch meter, hunt for the peak, overshoot, back off, find it again. When tuned correctly the nulls drop sharply—from -20dB relative to antinodes down to -40dB or better. That 40dB null depth confirms the probe is coupling minimally. If tuning is off, nulls are shallow and broad, indicating significant reactive loading.

Two consecutive nulls measured 19.1mm apart. Guide wavelength: 38.2mm. Free-space wavelength: λ₀ = λg × √(1 - (fc/f)²) = 38.2 × √(1 - (6.56/10.525)²) = 28.5mm. Calculated frequency: c/λ₀ = 10.51 GHz. Counter reading: 10.525 GHz. Error: 0.14%, close enough given micrometer carriage backlash and the fact that null centres are estimated by eye.

This works because the probe was tuned. Without tuning, the measurement includes probe capacitance as an additional reflection, shifting null positions and corrupting the wavelength calculation. The stub turns observation into a controlled interaction rather than an uncontrolled disturbance.