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

The switches may be mechanical or solid state. Since all switching occurs after the test port coupler, the measurement stability and performance is much better than with a switched test socket, and switching losses (although they reduce dynamic range) have no effect on the stability of the measurement.

True multi-port solution

Today’s latest technology has enabled multiport network analyzer (VNA) solutions to make multiport measurements without the need for external switches or additional couplers. For example, the Keysight M937xA series PXIe VNA is a complete 2-port VNA that plugs into a PXI slot and can be configured as a multiport VNA when multiple modules are used (Figure 5). Up to 16 M937xA modules can be configured in a single PXI chassis, enabling measurement of 32-port DUTs and providing fully corrected capabilities up to 26.5 GHz.

(a) Full 2-port VNA (b) A single PXI chassis can accommodate up to 16 modules for 32-port measurements

Figure 5. Keysight M937xA PXIe VNA provides true multiport testing capabilities

Each PXI VNA module has an independent signal source, and each test port has an independent reference receiver and measurement receiver (Figure 5). The receiver is used to capture S-parameter data for all measurement paths simultaneously. Since there is no attenuation between the device under test and the measurement receiver in this multiport VNA solution, this true multiport capability can provide both accurate and stable measurements.

The M937xA multiport configuration uses jumpers to connect modules in the same chassis for signal synchronization (Figure 6). The 10 MHz frequency reference and trigger signals from the first module are distributed to all PXI VNA modules. The local oscillator (LO) signal from the first module is distributed to each additional module. In solutions with frequencies exceeding 20 GHz and more than 8 ports, it is recommended to distribute the LO signal output from the first module between the second and fifth modules.

Note: Unused LO port

Figure 7. Jumpers connecting 10 MHz reference and LO distribution in a multiport configuration.

The key advantage of PXI instruments is that they are very flexible, scalable and reconfigurable, allowing users to modify the test environment according to their needs and where they are used. In addition, if a PXI VNA module fails, users can easily replace spare parts to ensure normal production operation.

Compare Multi-Port Solutions

For the various multiport solutions described above, we compare them in several key aspects. Compared with switch-based solutions, true multiport has greater advantages in improving the throughput of multiport network analysis, which can improve performance by eliminating switch-related losses and achieve faster speeds by capturing data simultaneously with multiple receivers.

Number of measurement scans

A 2-port VNA requires forward and reverse sweeps from both test ports, followed by a full 2-port calibration, to achieve the most accurate measurements. If a 4-port VNA is used, multiport error correction can be achieved by sweeping four times from each test port.

图 9 显示了完整表征一个多端口被测器件所需要的测量扫描次数。由于真正的多端口PXI VNA 通过多个接收机（每个测试端口一个）捕获数据，相比以开关为基础的解决方案，扫描的次数更少，完成测量的速度更快。以一个 16 端口器件为例，采用全交叉开关矩阵的 2 端口 VNA 需要扫描 240 次才能获得 256 个 S 参数，而 16 端口的真正多端口 VNA 仅需扫描 16 次。扫描次数的显著减少，为基于 PXI VNA 的真正多端口解决方案缩短测试时间和提高测试吞吐量创造了明显的优势。

图 9. 真正的多端口解决方案与基于开关的解决方案相比，扫描次数显著减少，测量速度极大加快。

Dynamic Range

The system dynamic range is the difference between the maximum output power of the source port and the minimum input power that the receiver can measure. In a switch-based multiport solution, the switch attenuation will cause the dynamic range of the VNA to decrease. Figure 10 compares the dynamic range of a true multiport VNA with that of a switch-based test set. As shown in the figure, the dynamic range performance gap between a true multiport VNA that is not affected by the test set switches and a switch-based VNA increases with frequency.

Using a multiport test solution with a larger dynamic range has a great advantage because it allows the selection of a wider intermediate frequency bandwidth (IFBW) to achieve the same trace noise. For example, if the dynamic range is increased by 20 dB, a 100 times wider IFBW can be selected, resulting in a 100 times faster measurement speed with the same trace noise to achieve the same measurement results.

Figure 10. True multiport VNAs can provide greater dynamic range because they do not have the switch attenuation that affects performance as in switch-set based VNAs.

Temperature stability

If solid-state switches are installed on a switch-based test set VNA, the overall measurement performance is easily affected by changes in ambient temperature. Since the switches introduce drift errors, frequent calibrations must be performed to ensure calibration quality for multiport measurements.

Figure 11. By eliminating the temperature stability drift errors caused by the test socket switches, a true multiport solution can reduce or completely eliminate periodic recalibration.

Figure 11 shows the temperature stability performance of a true multiport VNA and a switch-based VNA solution. A short calibration was connected to each system test port, and the system was placed in a temperature-controlled chamber for reflection measurements. A baseline measurement was made at 25°C, and then the chamber was set to 18°C ​​and 33°C. The S11 amplitude data measured at each temperature was referenced to the first measurement at 25°C. The VNA with a switch matrix exhibited a drift of more than 50 mdB for every 1°C change in temperature, while the true multiport VNA had a maximum drift of less than 5 mdB.

A true multiport VNA solution eliminates external switches, which can reduce or completely eliminate periodic recalibration, significantly reducing overall measurement time.

Multi-port calibration

Calibrating a multiport VNA test system is more time consuming and complex than calibrating a standard 2-port or 4-port VNA. An N-port device has (N-1)(N)/2 possible paths, each of which requires a 2-port calibration.

There are two distinct types of calibration: mechanical and electronic (ECal). Mechanical calibration physically represents an “open,” “short,” “load,” and in some cases, a “thru.” For clarity, “thru” is used to describe the thru standard in a calibration kit, which is why it has historically appeared on the VNA menu.

These standards are often sold in bundles called mechanical calibration kits. Electronic calibration kits have built-in switchable standards that provide similar functionality to the open/short/load standards, as well as a thru state.

Mechanical calibration

For engineers who have done calibrations with mechanical standards before, performing a 10-port or 20-port mechanical calibration sounds tedious. QSOLT calibration is a very convenient technique that greatly simplifies multi-port calibration. Quick Open Short Load Thru (QSOLT) calibration is a different combination between SOLR (Unknown-Thru) and SOLT calibration. As the name implies, it requires an Open/Short/Load one-port calibration and a defined Thru, but it is fast because the one-port calibration only needs to be performed on one test port. In fact, you can use any one-port calibration method (such as bias short or even electronic calibration) on one test port, and then make a defined Thru measurement between port 1 and port 2. In this way, you can conveniently perform a full two-port calibration on an insertable path using a calibration kit that includes a set of standards.

If you have N ports on a multiport DUT, and all ports are of the same sex, you can create an N+1 test system using a flex cable on an extra port that matches the DUT connector. A simple single-port calibration on this extra port and a straight-through connection to the other ports one by one will create a full N+1 port calibration. This way, there is no need to move any of the other ports or even a corresponding calibration kit.

Figure 12. Mechanical calibration kits contain individual standards for characterizing system errors. Many kits include a test port adapter and a torque wrench to allow you to make the proper connections.

Electronic calibration (ECal)

Today, the new generation of electronic calibration uses custom GaAs IC switches to provide embedded nominal “open,” “short,” “load,” and “thru.” In addition to the “open” state, the custom IC may also include multiple “short” states to ensure that large phase differences between standards are maintained over the entire frequency range. Electronic calibration modules have very good repeatability and stability because they contain solid-state electronic switches.

From a specification or theoretical point of view, the best TRL mechanical calibration kits provide the highest calibration quality. Next is the ECAL standard with the best electronic calibration components, and then the SOLT with sliding load. Fixed load SOLT calibrations generally have the worst performance.

In fact, if RF cables are used, the error in cable curvature will undoubtedly cause serious errors in TRL calibration, making its calibration quality worse than that of electronic calibration. If possible human error and the repeatability of connectors in mechanical calibration are taken into account, the calibration quality of electronic calibration parts is undoubtedly superior to that of mechanical calibration parts in practical operation.

Keysight Technologies offers 2-port electronic calibration solutions covering the frequency range of 300 kHz to 67 GHz and up to

1. GHz 4-port electronic calibration solution. It should be noted that in multi-port calibration, whether using 2-port or 4-port electronic calibration, the number of connect/disconnect times remains the same, and the number of VNA button presses during calibration increases. Figure 13 uses 10 ports as an example to show how to perform multi-port calibration using 2-port and 4-port electronic calibration. Note that both methods require 10 connect/disconnect times.