The Vector Network Analyzer or VNA can best be considered as the "multimeter for the uW/RF engineer". It measures the S-parameters of RF/uw/mmWave active and passive devices. With the data the Engineer can start making a circuit design or he/she can conclude from the data if the sub-component meets the specs and will not degrade system performance once used in it.
At the minimum, the VNA measures S-parameters. This is a so-called "black-box" description of a component in terms waves: The component is stimulated with an incident wave and the resulting reflected and transmitted waves are measured. Since S-parameters are ratio-ed data, the absolute signal level is not (should not be) relevant.
The S-parameters can be converted to the perhaps more intuitive Y, Z or Chain circuit parameters that are well-known from classic electronic circuit design classes. The transmission coefficients (ratio of transmitted over incident wave) are a measure of the gain or loss in the device or component. The reflection coefficient (ratio of reflected over incident wave) are a measure of the circuit impedance. Please note that the S-parameter measurement is by definition done at 50 Ohm source and load impedance. So, the S-parameters by themselves describe very well the behavior of the DUT in the situation of 50 Ohm source and load impedance. It is up to the designer now to choose other source and load impedances (or: design the input and output matching networks!) so that the device performs even better e.g. has more gain and is better matched to the actual source and load.
The source is a signal generator that sweeps over a certain frequency range. With the aid of the switch, the device is sequentially stimulated at the input and output port while at the same time the output resp. the input port is terminated with a 50 Ohm termination. The couplers separate the incident and transmitted/reflected signals and the receivers amplify and detect the signals for further processing (ratio-ing) and presentation on the screen.
Over the years the algorithms for VNA calibration have evolved enormously into cryptic names that are often linked to the type of known (or partially known) standards like
- SOLT (Shot, Open, Load, Thru/through)
- TRL (Thru, Reflect, Line)
- LRM (Line, Reflect, Match).
This principle of the VNA is not limited to (just) 2-ports but is often extended to 4-ports. You can then measure the so-called "single-ended" 4-port S-parameters S11, S21, ... S44 where the simple, single source and loads have (by definition) a 50 Ohm impedance to common ground. Since the S-parameters are linear (they should be!) it can be calculated what the circuit behavior would be in case the excitation or stimulus would be e.g. a differential or balanced source, consisting of a floating source between 2 inputs or (the same) two sources each with same amplitude but with opposite phase of the sinusoidal signal. This would be called a differential or balanced signal stimulus and/or response. Since also ratios of single-ended and differential signals can be calculated, we talk in general of "mixed-mode S-parameters".
An important application area of the mixed-mode S-parameters is in the analysis of high-speed data transmission paths like router back-planes where the 'wiring' consists of complexly routed, differential or balanced transmission lines between daughter cards on the both ends. <link Signal Integrity> Since in SI we may be analyzing also cross-talk between several differential transmission lines, we need even an 8-, 12-, 32- or more-port S-parameter VNA or a Multi-port S-parameter test-set extension.
VNA's come in different flavors and from different brands! The frequency range is often key or limiting factor. The system bandwidth is determined by the coupler that separates incident and reflected/transmitted signals and of course by the signal generator and receiver capabilities. The typical working-horse is a single sweep <10MHz to 26, 50 or 67GHz VNA with coaxial connectors. There are broadband systems that go up to 110 or even 145GHz, also with a single coaxial connector, but in fact the sweep consists of 2 or 3 bands, stitched together.
Above 50GHz, but for sure above 110GHz, the measurements need to be done in waveguide because to date there exists no proper (read: low loss) other transmission line system for signals at those frequencies.
The waveguide system is highly divided in bands due to the coupling of wave propagation to the mechanical geometry and as a result of that the typical S-parameter measurement capabilities are also divided by send & receive extender modules per waveguide band. Based upon a 20 or 50GHz VNA the extenders allow generating and receiving the millimeter wave signals and perform S-parameter measurements in the given band.
Over the years the VNA has been used more and more in power characterization and source/load-pull measurements of RF power transistors <link> because of the better accuracy of the final measurement data. There the VNA receivers tend to be used also to measure the absolute value or the power in the signals in contrast to only doing ratio measurements as in standard S-parameter measurements. This has led to further improvements of the VNA like improved linearity of receivers and high spectral purity of the signal source in order to come to the Non-linear Vector Network Analyzer or NVNA. So with the NVNA you can measure all the incident, transmitted and reflected signals, typically at the fundamental aswell as at the harmonic frequencies in amplitude and phase. The phase is typically relative to the fundamental stimulus signal. From this frequency domain data in amplitude and phase, one can construct the voltage and current waveforms at the at the DUT reference plane. As with the linear VNA, an extensive calibration routine has to be performed with a calkit (for the ratio measurements), a power meter (absolute power level cal) and a phase reference (phase calibration).