of Keysight Technologies.
The switches may be either mechanical switches or solid‐state switches. Because all the switching occurs behind the test port couplers, the stability and performance of the measurements are much better than that of switching test sets, and loss in the switch, while it reduces the dynamic range, has no effect on stability of the measurements.
In some cases, an option may be provided to add a low‐noise amplifier (LNA) between the coupled port of the test coupler and the switch input. This improves performance as the gain of the LNA improves the dynamic range. Adding amplifiers in between the coupled arm and the switch also removes another source of error. In some cases, the source‐match of a port changes when the source and test port share the same VNA receiver, for example ports 1 and 3 in Figure 2.27. This error is typically small as the difference between the match of the VNA receiver and the match of the switch is small (on the order of −10 dB) and is further reduced by twice the coupling loss (32 dB) resulting in a typical source‐match error smaller than −40 dB. In most cases it has a negligible effect, but in some measurements, particularly circulators or couplers, it can become significant and is not removed in calibration, so adding an amplifier ensures that the match presented to the coupled arm is constant. The test ports also change load characteristics depending upon if they are terminated in a switch or the VNA internal load; however, the N‐port calibration methods characterize both these states and fully correct for the difference.
2.2.7.3 True‐Multiport VNAs
While the extension test set provides a directional‐coupler on each port of the test system, the reference coupler and the measurement receivers are shared, so the number of ports that can be measured simultaneously is limited to the number of receivers in the base instrument. Recently, improved integration has made it possible to include a full VNA test receiver on each port, so true‐multiport VNAs are now available. These come in a variety of form factors, but for the most part they are intended for manufacturing operations, where size and footprint are important.
One of the first offerings for a large‐port‐count true‐multiport VNA was the ZNBT from Rohde & Schwarz. It provides options from 8 to 24 ports, with a faceless instrument. In this configuration, it had six independent sources (one for each four ports) as well as receivers on each port.
A modular form of multiport VNAs has been introduced in a PXI format, which allows for configuring from 2 to more than 68 ports, potentially up to 100 ports, depending upon the number and model of VNA modules used. Figure 2.29 shows a modular system with eight 6‐port modules (Keysight model M9804‐006) and one 2‐port module (Keysight model M9804‐002), configured as a 50‐port VNA system. There is one source per module, but a full dual‐reflectometer and dual RF receiver for each port. Thus, the 2‐port modules have 1 source and 4 receivers; the 4‐port modules have 1 source and 8 receivers, and the 6‐port modules have 1 source and 12 receivers.
Figure 2.29 A 50‐port VNA system comprised of 6‐port and 2‐port modules.
Multiport VNAs in a modular format require the local oscillator to be shared across all modules to get the best trace noise performance. These systems provide a daisy chain approach to the connect the LO and the 10 MHz reference to each of the modules. The big advantage of a modular approach is the test system is easily reconfigured to support different test needs. For example, a 16‐port system, comprised of eight 2‐port modules, can be reconfigured into four sets of 4‐port VNAs.
While more expensive than a switched version of a VNA, the economics of a true‐multiport system readily become apparent when one considers the overall measurement time and number of sweeps needed to complete an N‐port calibrated measurement. Table 2.1 shows the number of sweeps needed to complete an N‐port calibrated measurement. From a strict sweep time point of view, a true multiport VNA greatly reduces the overall test time requirement.
Table 2.1 Sweeps needed for N‐port calibration
| Total Ports | Total Paths | Switched 2‐Port | Switched 4‐Port | True Multiport |
| 8 | 28 | 56 sweeps | 24 sweeps | 8 sweeps |
| 16 | 120 | 240 sweeps | 64 sweeps | 16 sweeps |
| 24 | 276 | 552 sweeps | 144 sweeps | 24 sweeps |
2.2.7.4 Calibration of Multiport VNAs
Calibration is often a concern with multiport test systems. Traditional S‐parameter calibrations require measurements between every path of a test system. However, new techniques have greatly reduced the total number of calibration steps to the point where a full N‐by‐N port S‐parameter calibration can be achieved with a single one‐port return loss calibration and N‐1 thru measurements using the quick‐short‐open‐load‐thru (QSOLT) calibration. More details of these new calibration methods will be discussed in Chapter 3. Some manufacturers provide high‐port‐count electronic calibration modules that can simplify the calibration process.
2.2.8 High‐Power Test Systems
Most VNAs have a maximum test port operating level on the order of 10–15 dBm, with a damage level on the order of +30 dBm. Beyond the operating level, the receiver will be in substantial compression, so the data is not valid. Many VNAs provide internal receiver attenuators that allow reducing the power to the receiver, providing operation to much higher levels. The maximum input power to test port couplers are often rated higher than the maximum level of other components behind the directional‐coupler so that with proper padding and isolation, the VNA can be operated to levels as high as +43 dBm, depending on the model. Operation above these levels is possible but requires substantial external components including external couplers to ensure the power level at the VNA components is below the power damage level. Details of high power test configuration are shown in Chapter 6.
Another common practice is to add fixed attenuators between the DUT output and the VNA test port. This works well so long as the total attenuation between the test port and the DUT is less than 10 dB. Adding attenuation after the test port coupler degrades the directivity by two times the attenuation (in dB), as will be shown at the end of Section 2.3.2. In practice, up to 10 dB of external attenuation can be added and compensated for with normal calibration techniques. If between 10 and 20 dB are added, the system becomes somewhat unstable, and for more than 20 dB of added attenuation, different techniques for calibration must be used, and the S22 measurements become unreliable.
For testing devices that require high‐power drives, it is common to add an amplifier to increase the power