Reading your first S11 measurement

Return loss, VSWR, and reflection coefficient are three ways of writing down the same number. Here is how they relate, and what actually counts as a good result.

Almost every RF measurement session starts the same way: one cable, one device, and a trace on the screen that says how much of your signal is coming back at you. That trace is S11, and learning to read it well is most of the value a vector network analyzer gives you.

One measurement, three names

S11 is the reflection coefficient at port 1. It is a complex number, usually written as the Greek letter gamma, and it says what fraction of the incident wave returns from the device under test.

The magnitude of that number gets reported three different ways, and they are all the same measurement:

Reflection coefficient is the raw magnitude, from 0 to 1
Return loss is that magnitude in decibels, conventionally written as a positive number
VSWR is the standing wave ratio it would produce on the line

A reflection coefficient of 0.1 is 20 dB of return loss is a VSWR of 1.22. Nothing has changed except the units.

Why the decibel version is easier to live with

The linear scale compresses everything you care about into the bottom tenth of the range. A well matched antenna and a very well matched antenna sit at 0.1 and 0.03, which is hard to read on a linear plot but is a clear 10 dB apart in log scale.

If a number is going to span three orders of magnitude, put it in decibels and stop squinting.

What counts as good

The honest answer is that it depends on what you are building, but these are reasonable starting points:

Antennas are commonly specified better than 10 dB return loss across the band of interest
Filters in their passband should do better, often 15 to 20 dB
Connectors and cable assemblies should be better still, and a bad one shows up immediately

The reason 10 dB is such a common threshold is that it corresponds to about ten percent of the power being reflected. Below that, the reflected power is usually small compared to the other losses in the system, and chasing it further stops paying for itself.

At 20 dB return loss, only one percent of the power comes back. Going from 20 dB to 30 dB improves that to a tenth of a percent, which is real but rarely what limits a design.

Calibrate first, then believe the trace

An uncalibrated S11 measurement includes your cable, your connectors, and the analyzer's own imperfections. Calibration moves the reference plane to where your device actually starts, and removes the systematic errors in front of it.

This matters more than it sounds. A metre of cable can easily contribute more ripple to an uncalibrated trace than the device you are trying to characterise. Run a SOLT calibration with standardsAlmost every RF measurement session starts the same way: one cable, one device, and a trace on the screen that says how much of your signal is coming back at you. That trace is S11, and learning to read it well is most of the value a vector network analyzer gives you.