---
title: "How a network analyzer separates the cable from the device"
description: "A reflected RF wave carries cable delay as well as load information. Follow its round trip to understand wave ratios, calibration and reference planes."
canonical_url: "https://sandboxphysics.com/stories/instruments/vector-network-analyzer/"
language: "en"
page_type: "story"
alternate_languages:
  zh-CN: "https://sandboxphysics.com/zh/stories/instruments/vector-network-analyzer/"
last_modified: "2026-10-28"
---

[How physics instruments measure](https://sandboxphysics.com/stories/instruments/)

Inside the instrument

# How a network analyzer separates the cable from the device

A network analyzer measures known standards to identify the instrument and cable response, then corrects its results to an agreed connection plane. The reported device begins at that plane. Include another length of cable and the same resistor acquires a different reflection phase.

Sandbox Physics · 2026-10-28

## The wave going out meets a wave coming back

A short wire connected to a battery can usually be treated as one electrical potential. At radio frequencies, propagation takes an appreciable part of an oscillation cycle. A disturbance launched now is still traveling toward the load as an earlier reflection returns toward the source.

The inner and outer conductors of a coaxial cable guide this traveling field. The forward voltage and current waves have a characteristic ratio called the line impedance. Many RF systems use fifty ohms. A load that matches the line accepts the arriving wave; a different impedance generally sends part of it back.

The word vector says that the instrument retains phase as well as amplitude. Phase records position within an oscillation cycle. Coherent receivers share an oscillating reference, distinguish incident and outgoing waves, and compare their complex amplitudes. Each frequency point then describes how much wave comes back and its timing relative to the incident wave.

Sources: [1] [3]

## A hundred-ohm load reflects one ninth of the incident power

Place the observation plane directly at a resistor connection. An ideal fifty-ohm line terminated by a hundred-ohm resistor returns a voltage wave one third as large as the incident wave. Power follows amplitude squared, so one ninth, about 11.1 percent, returns. The remaining power enters the resistor as heat. This power comparison uses a common real reference impedance.

Let $Z_L$ denote the load impedance and $Z_0$ the real line reference. The reflection coefficient $\Gamma$ is the outgoing-to-incident wave ratio. A load close to the reference makes its numerator small.

An ideal open admits no load current and has reflection coefficient plus one. A short forces the terminal voltage to zero, so its reflected voltage reverses sign and its coefficient is minus one. Both return all incident power, with a half-cycle difference in reflection phase.

$$
\Gamma=\frac{Z_L-Z_0}{Z_L+Z_0}
$$

This linear steady-state expression applies at the load connection. A hundred-ohm load and fifty-ohm reference give one third in amplitude. Square that magnitude to obtain the reflected-power fraction.

Sources: [1]

## Move the observation plane and the reflection travels twice

Move the observation plane twenty-five centimetres back toward the source. The incident wave travels down the added cable to reach the resistor. Its reflection travels back through the same cable. The resistor is unchanged, but the reflection phase includes two propagation paths.

For a chosen ideal matched cable with propagation speed two hundred million metres per second, a hundred-megahertz wave has a two-metre wavelength. Twenty-five centimetres is one eighth of a wavelength: forty-five degrees of lag each way and ninety degrees on the reflection round trip. A lossy cable also reduces the returning amplitude.

Calibration fixes a reference plane at a physical connection. The analyzer measures an open, a short and a known load there to determine leakage, wave tracking and source mismatch. The intervening cable response becomes part of that correction. Subsequent reflection results refer to the plane where those standards were connected.

Real opens have fringing capacitance and shorts have connection inductance. A calibration kit specifies these departures from ideal behavior and the frequency range of its definitions. Incorrect definitions can still fit the three observations. A separate known load tests the correction against an additional observation.

Sources: [3] [2]

## A filter has a second way for energy to leave

A filter normally has two measured ports. A wave incident on the first port can return there or emerge from the second. Exciting the other port measures reverse transmission and the other reflection. These four complex wave ratios form the scattering parameters.

A low-pass filter transmits lower frequencies. At higher frequencies, its inductors and capacitors change the port voltage-current relations, transmission falls and reflection often rises. Reduced transmission alone does not show how much power became heat. Reflection and dissipation must also be considered.

Each receiver channel has its own amplitude and phase response. Mismatch at either port feeds back through the device. A known through connection relates the two channels for transmission calibration. Commercial two-port calibration also accounts for changes introduced by switching the excitation between ports and for unwanted leakage.

If a known matched spacer remains beyond the calibration plane, its propagation model can shift the reported plane farther toward the device. Replacing or moving a test cable requires checking the calibration again. The corrected reflection and transmission then describe a clearly identified physical section.

Sources: [2] [4] [1]

## [Move the plane along a coaxial line](https://sandboxphysics.com/labs/vector-network-analyzer/)

Compare a filter, through and resistor, measure standards, and follow the effect of known cable length on complex wave ratios.

The experiment uses a reduced fifty-ohm two-port error model with retained relative-wave observations and independent checks. It does not establish absolute power or commercial calibration uncertainty.

## Sources and further reading

1. [Marks & Williams — A General Waveguide Circuit Theory, J. Res. NIST 97, 533–562 (1992)](https://nvlpubs.nist.gov/nistpubs/jres/097/jresv97n5p533_A1b.pdf)

   Read printed p537 §2.1 for positive complex time and RMS conventions, p538 §2.3 Eq16 for power, and p555 §4.2/§5 Eqs121–123 for ratio versus absolute-wave measurement and power waves. The present discussion fixes a real fifty-ohm reference.

2. [Keysight M9485A Help — Calibration Standards](https://helpfiles.keysight.com/csg/m9485a/s3_cals/calibration_standards.htm)

   Read About Calibration Kits, Standard Definitions and Class Assignments: offset delay, impedance and loss, open capacitance, short inductance, valid frequency limits, and distinct one-port/full-two-port standard classes. HTML snapshot read 28 October 2026.

3. [Keysight E5061B Help — Structural Return Loss, Theory](https://helpfiles.keysight.com/csg/e5061b/measurement_with_options/option_010_time_domain_fault_location_analysis/structural_return_loss_measurement/theory.htm)

   Read Techniques for Removing Connector Effects and Measurement Uncertainties for the SOL plane, connector electrical length and cable movement. Its particular seventy-five-ohm performance figures are not used here.

4. [scikit-rf — OnePort and EightTerm calibration documentation](https://scikit-rf.readthedocs.io/en/latest/api/calibration/generated/skrf.calibration.calibration.OnePort.html)

   Read the OnePort bilinear three-standard model and the separate EightTerm page on switch terms. Reviewed alongside release v2.2.0 (18 October 2026). The linked teaching model uses a derived reduced two-port correction with zero switch terms and isolation.

[How a vernier reads between millimetres](https://sandboxphysics.com/stories/instruments/length-metrology/) [Air helps hold up the object on a balance](https://sandboxphysics.com/stories/instruments/precision-balance/) [How Cavendish Measured the Attraction Between Spheres](https://sandboxphysics.com/stories/instruments/cavendish-balance/) [Which pressure does a liquid column measure?](https://sandboxphysics.com/stories/instruments/manometer/) [The cup absorbs heat too](https://sandboxphysics.com/stories/instruments/calorimeter/) [Does a harder turn mean a thicker liquid?](https://sandboxphysics.com/stories/instruments/rotational-rheometer/) [Why the air beyond a pipe changes its resonance](https://sandboxphysics.com/stories/instruments/resonance-tube/) [How a seismometer reads the ground through motion inside its case](https://sandboxphysics.com/stories/instruments/seismometer/) [Why does a meter change the circuit?](https://sandboxphysics.com/stories/instruments/electrical-meters/) [How a null detector reveals a resistance](https://sandboxphysics.com/stories/instruments/resistance-bridges/) [Why a capacitor can behave like an inductor](https://sandboxphysics.com/stories/instruments/lcr-impedance/) [Where does the slow wave on an oscilloscope come from?](https://sandboxphysics.com/stories/instruments/digital-oscilloscope/) [How a lock-in amplifier finds a signal buried in noise](https://sandboxphysics.com/stories/instruments/lock-in-amplifier/) [Where a small peak in a spectrum can come from](https://sandboxphysics.com/stories/instruments/spectrum-analyzer/)