How do you simulate and model the S parameters of your PCB design?

Reflection coefficient S11 is the output at port 1 when driven by port 1. Transmission coefficient S21 is output of port 2 when driven by port 1. They are also referred to as return loss and insertion loss respectively.

By analyzing the S-parameters of a circuit network, one can gain insights into its behavior. Specifically, S-parameters are complex numbers that represent how radio-frequency (RF) signals are transmitted and reflected in a network. For instance, a low magnitude of S21, which denotes the forward transmission coefficient, suggests significant signal loss along the path. This loss could be due to dissipation into other ports or reflection back to the source. Furthermore, S-parameters can be utilized to calculate the characteristic impedance of a transmission path. This is critical for impedance matching, a technique used to minimize signal reflection and maximize power transfer in RF systems.

The VNA is actually measuring the S-parameters for the test fixture + the DUT, not just the DUT. The test fixture needs to be de-embedded from the fixture + DUT measurement. This means you will need to put the DUT on a test board with fixture that has known S-parameters. Once the combined measurement is taken, de-embedding is used to recover the DUT S-parameters. For a module that connects directly to the DUT (like an amp or isolator), the calibration procedure accounts for the connection to the module, so typically no de-embedding is needed. For a transmission line, de-embedding is required to eliminate the signal launch from the test board so that the DUT S-parameters can be extracted. This can be done by hand, or with MATLAB, etc.

Once I was measuring the input impedance of antenna … The reading I was getting from VNA accumulated the response of coax connector. The port extension method in VNA allows to measure the impedance at a point way beyond the connector.

S-parameters are typically measured, at microwave frequencies, with a network analyzer. All you need are a scope and function generator capable of measuring and producing frequencies in the desired range.

Calibration kit includes short, open, load and through standards.

The calibration process is crucial for accurate S-parameter measurements, particularly in laboratory environments. In software such as Ansys HFSS, ports can be conveniently defined and attached directly at the intended locations in the virtual model. However, in a laboratory setting, the scenario is more complex. Physical factors such as cables, connectors, and other fixtures introduce additional variables that affect the signals reaching the ports of a Vector Network Analyzer (VNA). Therefore, meticulous calibration is essential. This involves using a calibration kit that is compatible with the specific VNA being used. The kit must match the VNA in terms of dimensions, connector types, etc.

S-parameters can be calculated directly for some DUTs without a simulator. For example, S-parameters for circuits composed of cascading 2-ports can be calculated directly using ABCD parameters, and the cascaded ABCD network can be converted to S-parameters. 3D electromagnetic simulators like HFSS and CST can extract S-parameters from a 3D model of the PCB, and these are normally used in system-level simulations with additional components (driver/receiver, printed circuits, etc.). The final point about optimization involves simulating multiple S-parameters as a function of some system parameter, or it requires calculating multiple S-parameter curves for various parameter values. Sometimes, a quality metric (e.g., EVM) is used to optimize.

In modern prototyping, using a 3D electromagnetic simulator like Ansys HFSS is essential before creating a physical prototype. S-parameters are just one aspect of a multifaceted analysis. Take a smartwatch, for example: it requires examining how factors like temperature, stress, and strain influence components' S-parameters. Additionally, the antenna's performance, as represented by its S-parameters, can be impacted by variables like packaging, clothing, sweat, and skin contact. Ignoring these factors when measuring S-parameters can significantly delay a product's market launch, highlighting the importance of a comprehensive simulation approach.

S-parameters are usually saved in touchstone format. Software tools must have the capability to both generate and read touchstone files. Touchstone files are ascii text files saved with a .snp extension to the file name where “n” represents the number of ports. I.E. a .s4p file contains 4 ports.

For PCBs and in the view of dielectric stacks and PCB substrates and unless it is needed to calculate S parameters of a connector or a mounting on the PCB as a fixture, the optimal way is to use 2D solvers for such a purpose to extract S parameters. If the connectors are in place of analysis, better to use 3D solvers and extract those parameters especially if the purpose is to analyse EMC related issues. As an option, 2D solvers also provide considerable accuracy in extracting the s params of mounted connectors.

There are various software programs to measure S-parameters. One of them is RF PCB Toolbox provided by Matlab. Using RF PCB Toolbox, behavioral models can be created and results can be calculated analytically.

Verification of S-parameters is an extremely important issue. Otherwise, unexpected results may be encountered. The basics of measurement and analysis of S-parameters should be learned well. The results of mathematical calculations should be compared with simulation or test values.

S-parameters are verified with several methods. Because S-parameters are complex valued, RMSE for S-parameter data is actually taken over a complex valued frequency-dependent function, so it is really an L2 norm. S-parameter causality also needs to be verified by calculating an impulse response and examining for the presence of pre-shoot in the time domain. Causality for the S-parameter model can be enforced or ensured in several ways, such as windowing the S-parameter data at the high-end of the measurement bandwidth. For transfer functions, there is a technique involving the Hilbert trasnform, and there may be some analogous method for S-parameters as they are also essentially a linear transfer function.

The best way to become an Expert in moving from the Schema [Garden of Eden] to PCB/System [Garden of Eden full of Snakes [Parasitic Elements in practical PCB/System] So to get the 3 Pillars: Signal Integrity, Power Integrity, EMC/ESD is made of 3 steps: 1: to learn from Experts 2: to Simulate 3: to measure Especially for S Parameters!!!!

This section would make more sense if it referred to PCB interconnect rather than simply PCB. A printed circuit board will have many interconnects connecting components. PCB interconnects consist of mainly PCB traces but can also include any combination of additional wires, cables and connectors connected to the traces. In most cases S-parameters measured on PCBs we are interested in the performance of the interconnect between active devices and thus are passive measurements.

S-parameters are used as an evaluation metric to determine whether your system meets some specific design goal. Also there is no "gain and bandwidth" of a PCB. The devices on a PCB could have gain, and any frequency domain measurement can be used to determine a bandwidth for some measurement or device. The rest of this "how to use" section is nonsensical.

As operating frequencies increase, transmission problems may occur. In order to prevent back reflections, especially in amplifier designs, it is necessary to ideally determine the characteristics of the load and the line structure used. In order to verify the design, the design should be verified by performing S-parameter analysis in the RF PCB Toolbox application.

This article talks about "S-parameters of a PCB design" in multiple instances. The PCB does not have S-parameters, different elements in the PCB have S-parameters. This includes components, transmission lines, printed circuits, vias, etc.

S-parameters are behavioral models representing the behavior of a device or circuit over a set frequency range. It contains no other information of the device which makes them a practical way to share models while protecting intellectual property.