Learn about multi-port and multi-site test optimization techniques

As the number of ports on components increases, testers need to perform accurate and fast multiport network analysis. Optimizing the configuration of a vector network analyzer (VNA) is key to minimizing test costs during the manufacturing phase. This application note provides an overview of multiport and multisite testing capabilities, compares different multiport test solutions, and finally discusses what to consider when configuring a multisite test station.

Reducing the cost of test is a major challenge in high-volume component production. As the number of ports on components increases, testers need accurate and fast multi-port network analysis. Minimizing operator intervention, reducing the number of connections and calibrations will have a great impact on measurement throughput, and the key to minimizing the cost of test is to optimize the configuration of the vector network analyzer (VNA).

Applying modern instrumentation, such as multiport PXI VNAs with advanced capabilities, can improve measurement throughput for high-volume production test. Keysight PXI vector network analyzers introduce many new features that increase throughput while providing more flexible test configurations than traditional benchtop solutions. For multiport applications, PXI vector network analyzers offer true multiport capabilities, allowing test engineers to configure up to 32-port VNAs in a single PXI chassis. In addition, its reconfigurable form factor takes up less space than many conventional VNA instruments for multi-site or parallel testing.

This application note provides an overview of multiport and multisite testing capabilities. What are the different types of multiport test solutions and what are their advantages and disadvantages? What issues need to be considered when configuring a multisite test station and what factors may affect the overall test throughput? This application note provides a detailed introduction to PXI vector network analyzers and techniques for optimizing test stations.

Increasing demand for multi-port testing

In the early days of network analysis, all measurements were focused on 2-port S parameters. As users increasingly needed to test power splitters, mixers, differential devices, and other devices, the capabilities of network analyzers (VNAs) continued to expand, eventually evolving to 4-port VNAs. Now, any device that requires more than four ports for network analysis is considered a multi-port device.

Many of today’s components integrate multiple functions on a single element. The number of ports in these components continues to increase, and so does their complexity. Examples include RF front-end modules (FEMs) that support multi-band operation in smartphones, multiple-input/multiple-output (MIMO) antennas, and passive interconnects such as RF connectors or cable assemblies in high-speed digital applications (Figure 1).

Figure 1. A wide range of applications is driving the need for accurate multiport VNA measurements.

Some multiport devices can be tested as just a series of 2-port measurements, while many applications require more thorough multiport testing of their devices. As multiport devices are measured with more and more parameters, high-volume manufacturers want to minimize the overall test time.

Types of Multiport Solutions

Different multiport measurement solutions can be provided according to different performance requirements, throughput requirements and budget.

Simple switch test socket

Early component manufacturers already had 2-port or 4-port VNAs, so it was natural to add a series of signal routing switches to handle devices with more than 4 ports. In this case, local mode measurements on the VNA were sufficient, requiring only RF switching to route the VNA ports to the pairs of ports on the device under test.

This type of multiport solution uses 2-port measurements for each common port path, so a 2-port network analyzer (VNA) with one common port and one switch port can complete all the required measurements. This approach is sometimes called a switch test set or simple switch tree.

A switch test socket consists of only RF switches that are organized into a matrix to provide the required measurement paths. Figure 2 shows an example solution and a block diagram of a simple switch tree test socket. Test sockets are typically composed of 1×2 or 1×4 RF switches to 1×6 RF switches. 1×2 RF switches are sometimes also used to provide RF loads to unused ports. 1x4 or 1x6 switches are usually mechanical switches and may not add loads to unused ports. If a path response is generated between two ports of a multiport device and this response is related to the load matching of the third port, then the switch matrix must provide loads on the unused ports. At frequencies above 40 GHz, it is usually not appropriate to use a large switch configuration with loads, and a 1×2 switch matrix should be used. 1x2 electronic switches can be used over a wider frequency range, but electronic switches usually have a small number of ports, so electronic switch test sockets are usually constructed with 1x2 RF switches.

Figure 2. Schematic diagram of a 2-port VNA with a 24-port test set.

The simple switch matrix in Figure 2 can be considered as consisting of a port 1 switch group and a port 2 switch group. Any path from the port 1 side to the port 2 side can be measured, but the ports on the port 1 side of the test set cannot be measured, and the ports on the port 2 side cannot be measured. Therefore, when there are 24 ports in the test set, only 12 paths can be measured from any of the 12 input ports. Therefore, this simple switch tree test set supports 144 paths, but the complete 24-port device actually has a total of 276 paths. There are 66 paths that cannot be measured on the VNA port 1 side and 66 paths that cannot be measured on the VNA port 2 side. To obtain a complete path matrix, a "full crossbar" switch matrix must be used.

Full crossbar switch test socket

Many multi-port devices require measurements from one port to other ports, and the path response often depends on the load or match applied to the other ports. Full crossbar solutions have further requirements: a full N×N port calibration measurement must be performed to correct for imperfect matches on each port. This requires not only a complete crossbar matrix, but also one that supports N×N calibration.

To complete a full crossbar test, a test socket configuration similar to that shown in Figure 3 can be used. In the conventional configuration, a 1×N switch tree is connected with a 1×2 switch crossbar on each port. This configuration allows any path to be measured, but the unused ports are terminated back to the 1×N switch, which is terminated internally at the load. If the 1×N switch is not internally terminated (but left open), then the 1×2 switch must provide termination for the unused ports.

Figure 3 shows a 1×2 port switch connected to a pair of 1×N switches to form a full crossbar switch. In this configuration, the ports not connected to the VNA are terminated in the switch load. However, it is difficult to perform a full N×N calibration using this type of switch matrix because the exact value of the load termination on a port changes with the switch settings of the other ports.

Figure 3. Full crossbar test socket example.

For example, if test socket ports 1 and 6 are the stimulus ports, ports 2 through 5 are terminated to the 1×6 switch on the left. If test socket port 5 is active, then port 6 can be terminated to the 1×6 switch on the right. The termination of the ports depends on the selected path, which makes calibration more difficult for ports other than the selected two. Custom switch test sockets may have a smaller number of paths, forming full crossbar combinations on some ports and simple switch trees on others.

Depending on the required measurement performance, either solid-state or electromechanical switches can be installed within the switch matrix. Solid-state switches are typically used for batch testing that requires fast switching times and long life, while electromechanical switches are used for high-power (i.e. > 1 W) network analysis.

VNA 与开关的组合是增加 VNA 端口数量的一种低成本解决方案。但是因为定向耦合器之后存在与 VNA 测试端口有关的开关损耗，所以相比独立的 VNA，系统性能在动态范围、轨迹噪声或温度稳定性方面会有所下降。尤其是对 10 GHz 以上的高频应用，性能下降的影响非常显著。下文中会对性能方面的权衡加以讨论。

Finally, switch-based solutions require switching of signal paths for measurements using a VNA receiver (typically with two or four test ports), thus requiring continuous measurements to obtain all S-parameters of a multiport device under test (DUT).

Extended test socket

Extended test sets are an improved design for full N×N calibration measurements because they include both directional couplers and switches. Extended test sets extend the source switch matrix of the network analyzer (VNA) to more outputs through source switches and the internal receivers to more ports through receiver switches. For each additional port, an additional test port coupler is required. Since the switching occurs after the VNA directional couplers, they can still be used as test ports. The test set ports are extended to all available ports. Figure 4 shows an example of a solution and a block diagram of the extended test set solution.

16-port PNA solution with U3042A E12 expansion test set

1. Schematic of a 4-port VNA with a 12-port test set

Figure 4. Example of an extended test socket.

A key point in the block diagram is that the test set is divided into a source loop and a receiver loop behind the test port coupler. Since any number of switch paths can be provided behind the test coupler, there is theoretically no limit to the number of ports that can be used. In addition, the block diagram allows for the addition of additional test sets, which can be stacked to provide any number of test ports. The typical configuration is to expand a 4-port VNA to 8 ports; a 10-port extended test set combined with a 2-port VNA for a total of 12 ports, and a 12-port extended test set combined with a 4-port VNA for a total of 16 ports.