S-parameter simulations are easy compared to measurements. The reason for this is that a measurement requires looking at the connectors, fixtures, fanouts, and probes used to collect a signal from a device on a PCB. Simulations allow a part to be isolated against probes and connectors, so only the DUT is being calculated in the S-parameter simulation.
This is never the case in an S-parameter measurement. The measurements performed with a VNA are always influenced by the cables, connectors, fan-in/fan-out, and pads on the PCB. The measurements from a test device always include these influences, so the S-parameters of the DUT need to be pulled out from the combined S–parameter measurement. Here are three ways this is done in your system analysis tools and with built-in functions in your VNA.
When taking an S-parameter measurement of a DUT with a test fixture, the measurements you are taking are really a combination of the DUT and the fixture. In reality, we only want the measurements of the DUT. How do we eliminate the influence of the test fixture?
Before S-parameter extraction, we normally calibrate the test fixture with a set of standards that define 4 situations at each port: short circuit, open circuit, perfect impedance match, and pass-through to the other port. But even after calibration of the device, the connecting ports and traces will be included in the measured S-parameters. In order to isolate the S-parameters of the DUT, there are three possible approaches:
The goal in these three methods is to move the calibration plane to the actual input ports of the DUT and therefore eliminate the influence of the test fixture as shown below.
Port extension and electrical delay move the calibration plane to the DUT. Source: Copper Moutain Technologies.
Port Extension (Applies to All S-Parameters)
Port extension is a mathematical adjustment that is applied to measurement of the S-parameters and it affects the data as though the measurement test ports were lengthened by a fixed amount. When the VNA calibration procedure is performed, the input port length is accounted for based on the measured phase shift during calibration. However, additional length exists on a test fixture due to lands, transmission lines, etc. and these can be modeled by extending the port.
Because the VNA knows the size of the port and its phase shift during calibration, it is a simple matter to add some phase to model a longer port. This is specified in terms of a time delay or distance delay. Either can be used because the VNA knows the propagation constant after calibration, so the VNA can convert this into the required phase value to extend the port.
Electrical Delay (Applies to 1 S-Parameter)
Electrical delay is very similar to port extension in that it artificially adds to the phase of the incoming/outgoing signals due to the test fixtures. The difference is that electrical delay models the additional distance to the DUT as a transmission line section rather than as a longer port. The extension of the transmission line distance assumes a lossless transmission line that only applies a phase shift. This makes the technique appropriate in situations where:
The connector to DUT distance is so short that transmission line losses are small
The connecting transmission line is impedance matched to the port
The connecting transmission line has low losses up to a certain frequency limit
De-embedding is a mathematical technique that is normally applied after the S-parameter data is exported from the DUT. It can be used in systems analysis software or in a scripting language like MATLAB or Mathematica. De-embedding requires knowing the S-parameters of the connections to the DUT on test fixture. As long as these are known or can be measured, they can be used in a de-embedding algorithm to extract the DUT S-parameters.
Read this article to learn more about S-parameter de-embedding.
Correspondence to Simulation
Measurements also need to be verified for accuracy, and one way to do this is by comparing with simulation. This is important because the S-parameters determined from measurement using the above methods will need to be compared with the DUT for accuracy sake. The quickest way to get some reference set of S-parameters is through calculation in an electromagnetic simulation.
The great thing about simulation is that there is no need to directly de-embed S-parameters from a particular structure in a 3D simulation. Instead, it’s possible to simulate the S-parameters for a particular structure directly. Take a transmission line embedded as a stripline as an example; any other nearby connectors, traces, vias, and pads will all affect the S-parameters in a measurement, but the S-parameters for the line can always be determined in isolation.
Whenever you need to simulate the electromagnetic behavior of components and circuits in your PCB, use the complete set of the system analysis tools from Cadence. Only Cadence offers a comprehensive set of circuit, IC, and PCB design tools for any application and any level of complexity. Cadence PCB design products also integrate with a multiphysics field solver for thermal analysis, including verification of thermally sensitive chip and package designs.