Good impedance matching is required for the efficient transfer of signals between connected devices. Unfortunately a perfect match is rarely achieved. There are two common ways to express a device’s impedance mismatch: Return Loss and Standing-Wave Ratio (SWR). Both quantities describe the same phenomenon, but they have different interpretations.
When a transmission line or waveguide is driven by a perfectly matched signal source on one end, and terminated with an imperfect impedance match on the other end, reflections produce “standing waves”. This refers to the amplitude peaks and valleys that result from constructive or destructive interference at any given frequency. SWR is the ratio between the maximum and minimum signal amplitudes. Knowing this ratio is rarely helpful for design or analysis work. Then why is it frequently used as a figure of merit?
In the early years of RF and microwave engineering, a common way to measure a device’s impedance was to connect it to a signal source through a slotted transmission line or waveguide. A diode detector was connected to a sliding probe inserted into the slot. The probe was carefully positioned to determine the minimum and maximum signal amplitudes, as well as the distance between the load and the first amplitude minimum. A graphical tool or direct computations were then used to find the load impedance.
At sufficiently low frequencies, a lumped-element impedance bridge or coupler can be used to measure SWR if the source power is not too high. At high power levels, a directional coupler constructed from weakly coupled transmission lines is more commonly employed. Regardless of the devices used, SWR measurement is a required skill for tuning or trouble-shooting antennas. Many communication systems include SWR measurement devices in the signal path to continuously monitor antenna performance.
Although SWR can be measured using dual directional couplers and amplitude detectors, over a wide frequency range this method typically has poor accuracy unless an error-correction scheme is applied to compensate for the non-ideal performance of the components involved. Modern network analyzers and impedance analyzers automatically apply error correction in software. This type of equipment is usually preferred for measuring impedance mismatches at low power levels.
Network analyzers provide output data in the form of scattering parameters (S11, S21, etc.). For coaxial devices S11 is equal to the reflection coefficient computed from the measured load impedance and the system impedance. For waveguides, impedance refers to the ratio between the transverse electric and magnetic field strengths. These quantities are difficult if not impossible to measure accurately at microwave and mm-wave frequencies. To avoid such problems the magnitude of S11 is defined as the square-root of the ratio between incident and reflected power levels, which can be measured with good precision at frequencies well beyond 100 GHz.
│S11│ = Square Root ( (Reflected power) / (Incident power) )
The magnitude of S11 is called Return Loss. It indicates the amplitude of the reflected signal as a fraction of the incoming signal amplitude. When plotted as a function of frequency, Return Loss is usually expressed in units of dB and computed as 20 log(│S11│). A value of 0 dB represents total reflection. More negative numbers indicate less power lost to reflection. When expressed as a single value, Return Loss is usually provided as a positive number. To avoid confusion, specifications often say “return loss is better than X dB” so that its interpretation is independent of whether X is positive or negative.
Return Loss and SWR are mathematically related, so providing both specifications is somewhat redundant. SWR becomes less sensitive to S11 when S11 is smaller and highly sensitive to S11 when S11 approaches 1. Someone worried about the performance of an impedance-sensitive system may be more interested in Return Loss because it doesn’t obscure the effects of small impedance variations as much as SWR can. Return Loss is also more easily applied when computing the expected performance of combined devices. As a result, Return Loss is generally viewed as more helpful and revealing than SWR.
For those already familiar with SWR, it can be useful to remember a few reference points that relate Return Loss to SWR and other performance criteria:
│S11│ Return loss SWR Mismatch attenuation Performance
0.03 30 dB 1.06 0.004 dB Outstanding
0.1 20 dB 1.22 0.04 dB Excellent
0.2 14 dB 1.50 0.18 dB Good
0.3 10 dB 1.85 0.41 dB Marginal
0.5 6 dB 3.00 1.25 dB Poor
0.7 3 dB 5.67 3 dB Unacceptable
The table shows that Return Loss that is better than 14 dB will not induce more than 0.18 dB of peak-to-peak ripple in the frequency response, which is generally considered “good” for most applications.