matrix. To measure such a device, a switch matrix must be able to allow measuring every path of the device. Informally, these types of switch matrixes are called full cross‐bar switches, which implies that from the two ports of the VNA, any path of the DUT can be measured.
There is a further requirement on the Butler matrix; a full N‐by‐N port calibration measurement must be able to be performed to correct for the imperfect match of at each port. This requires not only a full cross‐bar matrix but one that supports N‐by‐N calibration as well. A third style of test set allows such N‐by‐N S‐parameters called an extension test set, which extends or adds to the number of test ports from a VNA, but these have largely been replaced with true‐multiport VNAs.
More recently, several vendors have developed high‐port‐count VNAs with up to 24 internal ports (R&S ZNBT) or configurable module multiport VNAs (Keysight M9875), in which PXI‐based VNA modules can be flexibly configured to large numbers of ports (more than 66 ports).
The various forms of multiport configurations are described next.
2.2.7.1 Switching Test Sets
Switching test sets contain only RF switches formed in a matrix to provide the needed measurement paths. Figure 2.25 shows the block diagram of a simple switch tree test set. These test sets are typically constructed from either 1 × 2 RF switches or 1 × 4 to 1 × 6 RF switches. The 1 × 2 RF switches are sometimes used as some versions provide for an RF load on the unused ports. The 1 × 4 or 1 × 6 are typically mechanical switches and may not load the unused ports. If a multiport device has a path response between two ports that depends on the load match of a third port, the switch matrix must provide a load on the unused port. Larger switch configurations that have loads are often not available above 40 GHz, so 1 × 2 matrix arrays are used. Electronic 1 × 2 switches are available over a wide range of frequencies, but there are few electronic switches with higher port counts, so electronically switched test sets are typically configured from 1 × 2 RF switches.
Figure 2.25 Simple switch tree test set.
The simple switch matrixes of Figure 2.25 can be viewed as having a port 1 switch set and port 2 switch set, and any path from the port 1 side to the port 2 side can be measured, but no measurements are available between ports on the port 1 of the switch set, nor between ports on the port 2 side. While there are 24 ports available in the test set, only 12 paths can be measured from any one of the 12 input ports. Thus, this simple switch tree test set can support 144 paths, but a full 24 port device actually has 276 paths. There are 66 paths on the VNA port 1 side that cannot be measured, and there are 66 paths on the VNA port 2 side that cannot be measured. To obtain a full matrix of paths, a so‐called full cross‐bar switch matrix is required.
To accomplish full cross‐bar testing, the configuration of the test set shown in Figure 2.26 is used. In the general configuration, sets of 1 × n switch trees are cross connected to 1 × 2 switches at each port. This configuration provides for any path to be measured, but the unused ports are terminated back in the 1 × n switches, which are terminated internally in a load. If the 1 × N switches are not internally terminated (rather, they are left open), then the 1 × 2 switch must provide a termination for an unused port. Figure 2.26 shows a full cross‐bar switch constructed of a 1 × 2 port switch connecting to a pair of 1 × n switches. With this configuration, every port that is not connected to the VNA is terminated in a switch load. However, it is difficult to use this type of switch matrix to perform full N‐by‐N calibrations as the exact value of the load termination of any port changes depending upon the switch settings of other ports.
Figure 2.26 Full cross‐bar switching test set.
For example, if test set ports 1 and 6 are the active ports, ports 2–5 are terminated in the 1 × 6 switch on the left. If test set port 5 is made active, then port 6 may be terminated in the 1 × 6 switch on the right. The fact that the termination of the port depends on the path selected makes calibration beyond the two ports selected more difficult.
Custom switching test sets might have a reduced number of paths, forming a combination of full cross‐bar on some ports and simple switch trees on other ports. For high speed and reliability, solid‐state switching is preferred. Mechanical switches have almost no loss, but solid‐state switches can have considerable loss at microwave frequencies. This loss is after the directional‐coupler and dramatically degrades the RF performance of the system. On the other hand, mechanical switches can have slight changes in return loss for each switch cycle, also leading to instabilities. Thus, this architecture of switches after the directional‐couplers of the VNA is a simple one, but at a cost of substantially reduced stability and performance.
2.2.7.2 Extension Test Sets
To satisfy the requirement for making full N‐by‐N calibrated measurements, often referred to as full N‐port cal measurements, a test set design has been developed that includes both directional‐couplers and switches. The original implementation of this style of extension test sets was configured to supply two additional ports to a two‐port VNA to create a 4‐port VNA for making the first balanced and differential measurements. The general idea of an extension test set is to essentially extend the source switch matrix of the VNA to more outputs through a source switch and also extend the internal receivers to more ports through a receiver switch. This requires that an additional test port coupler be provided for each additional port. Because the switching occurs behind the VNA directional‐couplers, they are still available as test ports: the ports on the test set extend the total number of ports available, which is why it's called an extension test set. Figure 2.27 shows block diagrams for a simple two‐port extension test set.
Figure 2.27 Extension test set block diagram.
One key point of the block diagram is that the test set breaks into the source and receiver loops behind the test port coupler. Since any number of switch paths can be supplied behind the test couplers, there is in theory no limit to the number of ports that can be used. Further, this block diagram allows additional test sets to be added so that any number of test ports can be created by stacking extension test sets. Common configurations are 4‐port extension test sets for a 4‐port VNA to extend to a total of 8 ports, 10‐port extension test sets for a 2‐port VNA to achieve a total 12 ports, and 12‐port extension test sets for a 4‐port VNA to achieve a total of 16 ports. Figure 2.28 shows a 4‐port VNA with two 4‐port extension test sets to create a 12‐port system.
Figure 2.28 12‐port system using a 4‐port VNA