How to measure non-intrusive devices?

　　Due to the needs of microwave engineering, most microwave RF components have female ports, while the ports of cable connectors are one female and one male. These microwave RF components cannot be directly connected to the vector network analyzer for direct-through testing and are called non-insertion devices. In actual measurements, there are a large number of non-insertion devices, whose port connection types are opposite and whose male and female polarities are the same. Direct-through measurements cannot be performed directly during calibration. In order to connect non-insertion devices to the vector network analyzer, a double-female or double-male adapter must be added. In this way, the test value of the vector network analyzer is not the actual value of the device under test, but the value of the device under test plus the adapter.

　　In general measurements, the influence of adapters is ignored, and it is believed that the measured result is the actual value of the DUT. The loss and delay of the adapter are added to the DUT invisibly, so the accurate value of the DUT cannot be obtained. Or different calibration methods are used in the test, such as the undetermined straight-through method, the determined straight-through method, the equal adapter exchange method, the adapter removal method, etc., to reduce the influence of the adapter on the DUT. However, regardless of the quality of these methods, they cannot directly remove the influence of the adapter on non-inserted devices. This paper tests adapters with unknown characteristics (such as dual male and dual female connectors) through a single port. According to the algorithm and assumptions, the theoretical calculated value after cascading is close to the actual measured value, and then the value of the adapter is determined, and finally the accurate value of the DUT is obtained, thereby removing the influence of the adapter on the DUT during the test.

　　1 Principle

　　The signal flow graph (as shown in Figure 1) combined with scattering parameters is a simple and effective method for analyzing microwave networks and microwave measurement systems. The flow graph formula is also called Mason's non-touching loop rule, or Mason's formula for short. According to the Mason's formula, the transmission value between any two points in the signal flow graph can be directly calculated.

　　

　　Figure 1 Signal flow

　　Mersenne's formula:

　　From the Mason formula we know that:

　　

　　2 Verification Process

　　There are many types of adapters. For the convenience of research and universality, this experiment uses double negative and double positive for verification. The principles and processes of other types of adapters are the same.

　　The experiment requires a vector network analyzer (100 MHz to 40 GHz), two cables, a set of 2.4 mm calibration parts (AV31123, the calibration parts and cables should have high precision and small errors), and a programmable step attenuator. Set the linear frequency range from 1 to 21 GHz, take a point every 100 MHz, scan 201 points, and the intermediate frequency bandwidth is 100 Hz. Connect the cable to the two ports of the vector network analyzer, perform a full two-port SOLT calibration on the cable, and the cable port is the calibration end face to remove 12 system errors. After the preparation work is completed, different methods are used for testing. In method 1, the double-yin and double-yang values ​​are calculated by connecting a 50Ω matched load; in method 2, the double-yin and double-yang values ​​are calculated by connecting a programmable step attenuator. The following is the detailed process of the two methods.

　　2.1 Method 1

　　After port 1 is connected to a double male connector, an open circuit, a short circuit, and a 50 Ω matching load are connected respectively (as shown in Figure 2), and the reflection coefficients are measured as ΓAO, ΓAS, and ΓAL respectively. After port 2 is connected to a double female connector, an open circuit, a short circuit, and a 50 Ω matching load are connected respectively (as shown in Figure 3), and the reflection coefficients are measured as ΓBO, ΓBS, and Γ BL respectively. Then an open circuit, a short circuit, and a 50 Ω matching load are directly added to port 1, and the reflection coefficients are measured as Γ1O, Γ1S, and Γ1L respectively (as shown in Figure 4). An open circuit and a short circuit are directly added to port 2, and the reflection coefficients are measured as Γ2O, Γ2S, and Γ2L respectively (as shown in Figure 5).

　　

　　Figure 3: Double-female opener, short circuit breaker, and matching load

　　

　　Figure 4 Port 5 connected to an opener, a short, and a matching load

　　

　　Figure 5 Port 2 connected to an opener, a short, and a matching load

　　From formula (2), we can get the equations [SA] and [SB] of double positive and double negative respectively:

　　

　　Where: m11, m12, m21, m22 are the parameter values ​​of the double male connector to be tested, and n11, n12, n21, n22 are the S parameter values ​​of the double female connector. Assuming that the network is reciprocal, m12 = m21 n12 = n21, then the equation group can be solved to obtain the double male S parameter matrix SA and the double female S parameter matrix SB. They are:

　　

　　2.2 Method 2

　　Use a 20 dB attenuator as a load to replace the 50 Ω matched load in method 1. Use a vector network analyzer to measure the reflection coefficients ΓAL and ΓBL of the attenuator connected to the double-yin and double-yang. The reflection coefficients of the 20 dB attenuator are Γ 1L and Γ 2L respectively. The other steps are similar to method 1. Calculate S and compare it with that in method 1.

　　3 Verification

　　In order to verify the feasibility of the above method, two methods were used to verify the calculated values ​​of the double male and double female connectors.

　　3.1 Cascade double positive and double negative

　　First, calculate the S parameters of the double-yin and double-yang cascade. The S parameters of the cascaded two-port network cannot be calculated directly. The S parameters need to be converted to T parameters, and the cascade value [TAB] - [TA] [TB] is calculated, and then the [TAB] after cascade is converted to [SAB]. Cascade the double-yang and double-yin (as shown in Figure 6).

　　

　　Figure 6 Double positive and double negative cascade

　　The S parameters obtained by the above method are reference values, and these values ​​are compared with the S parameters calculated by the Mason formula. All test data are processed by MATLAB software, and the S parameter curve obtained by simulation after the final cascade is compared with the S parameter curve tested in the vector network analyzer.

　　From Figure 7, we can see that the difference between S12 and S21 is very small, but there is some difference between S11 and S22. The two curves of method 1 with a 50 Ω matched load and method 2 with a 20 dB attenuator are basically the same. After the double positive and double negative are cascaded, no matter what kind of load is connected, the cascade value is a fixed value. The reason for the difference is that the 20 dB attenuator is not a standard device, and there is reflection when the double positive and double negative are cascaded.

　　

　　

　　Figure 7 S11 and S22 calculated and tested by vector network analyzer for a 20 dB attenuator with dual male and dual female connections

　　The curve of the theoretical calculated value of S11 is very close to the test curve of the vector network analyzer. In the low frequency band, the curve of the theoretical calculated value of S22 is consistent with the test curve of the vector network analyzer. The curve of the theoretical calculated value is better at high frequencies because the reflection coefficient is below -40 dB and the isolation is high. In fact, the double-male and double-female connection ports are affected during the through test. Therefore, the theoretical calculated value is more accurate than the measured value.

　　In the experiment, the standard parts, adapters and attenuators were connected many times, and each connection would introduce human errors to varying degrees. The values ​​of the standard parts were obtained by testing after calibrating the vector network, and no higher-level calibration system was used for testing. The measurement of transmission parameters is indispensable for obtaining S parameters, but in the experiment, the transmission parameters were only obtained indirectly by measuring the reflection coefficient. There must be certain errors in the results caused by various factors. Correcting these errors can produce better results.

　　3.2 Programmable Step Attenuator

　　The following verification uses a programmable step attenuator, and the attenuation values ​​of the attenuator under test are 0, 10 dB, 20 dB, 30 dB, 40 dB, 50 dB, and 60 dB. The S parameters of the attenuator are calculated from the experimental data (as shown in Figures 8 to 10).

　　

　　Figure 8 Theoretical S21 value of attenuator

　　

　　Figure 9 Theoretical S11 value of attenuator

　　

　　Figure 10 Theoretical S22 value of attenuator

　　From the calculation results above, we can see that the S parameters of the programmable step attenuator calculated without the dual male and dual female connectors are the expected results. The calculated values ​​of the attenuator are verified to be consistent with the transmission parameters and reflection parameters given by Agilent.

　　4 Conclusion

　　This paper is a verification experiment, which removes the influence of dual-female and dual-male adapters on non-inserted devices in the test by calculating them. The problem of universal adapters is specialized and simplified, but it has a theoretical breakthrough for the measurement of other adapters and further research on port extension and de-embedding technology.