Determining quality of connection between test system and device under test

Abstract

The invention relates to determining the quality of a connection between a test system and a device under test. The present disclosure provides an exemplary system comprising: a tester configured to test a device under test (DUT); and a connection setting that is connectable to the DUT and disconnectable from the DUT. The tester is configured to transmit a radio frequency (RF) signal through the connection setting and capture a reflected signal from the connection setting. The reflected signal is based on the RF signal. The one or more processing devices are configured to use a trained machine learning model to determine a quality of a connection between the test system and the DUT based on the reflected signal.

Description

Technical Field

[0001] This specification describes example implementations of techniques for determining the quality of a connection between a test system and a device under test. Background Art

[0002] A test system is configured to test the operation of a device. The device being tested by the test system is referred to as a device under test (DUT). The test system may include test instruments to transmit test signals (such as radio frequency (RF) signals) and data to the DUT via a transmission medium (such as an RF transmission line or coaxial cable) for testing. The test system may also test the quality of the connection between the test system and the DUT, as the quality of the connection will affect the test results. Summary of the Invention

[0003] An exemplary system includes a tester configured to test a device under test (DUT); and a connection setup that is connectable to and disconnectable from the DUT. The tester is configured to transmit a radio frequency (RF) signal through the connection setup and capture a reflected signal from the connection setup. The reflected signal is based on the RF signal. One or more processing devices are configured to use a trained machine learning model to determine the quality of the connection between the test system and the DUT based on the reflected signal. The exemplary system may include one or more of the following features (alone or in combination).

[0004] The tester may be configured to capture first data when the connection setup is disconnected from the DUT. The first data may be based on a first reflected signal in the reflected signals. The tester may be configured to capture second data when the connection setup is connected to the DUT. The second data may be based on a second reflected signal in the reflected signals. The second reflected signals in the reflected signals may each be associated with a corresponding return loss. The corresponding return loss may include a return loss contribution from the DUT. Determining the quality of the connection may include determining the return loss of the connection setup when the connection setup is connected to the DUT minus the return loss contribution from the DUT.

[0005] The one or more processing devices may be configured to process the data based on the reflected signal in the first data and the second data so as to remove the return loss contribution from the tester. After processing, the one or more processing devices may be configured to resample the data based on the reflected signal in the first data and the second data. After resampling, the one or more processing devices may be configured to filter the data based on the reflected signal in the first data and the second data so as to attenuate the representation of the reflected signal in the first data and the second data. After filtering, the one or more processing devices may be configured to process the data based on the reflected signal in the first data and the second data so as to reduce the path loss associated with the connection setting in each signal and thereby generate combined data. The combined data may include magnitude data and phase data based on the reflected signal in the first data and the second data. The input to the trained machine learning model may be based on the magnitude data and the phase data.

[0006] The one or more processing devices may be configured to use a classifier machine learning model to identify the type of the DUT. The classifier machine learning model may select the trained machine learning model based on the type of the DUT. The one or more processing devices may be configured to use a classifier machine learning model to identify the type of the DUT. The classifier machine learning model may select a second trained machine learning model based on the type of the DUT for determining an electrical characteristic of a connection, the connection including the connection setup to the DUT. The one or more processing devices may be configured to execute the second trained machine learning model to determine the electrical characteristic of the connection. The electrical characteristic may include at least one of a capacitance or an inductance of the connection, the connection including the connection setup to the DUT.

[0007] An exemplary method includes operations for training a machine learning model to determine the quality of a connection between a tester and a device under test (DUT). The method may include the following operations: transmitting a first RF signal to a connection arrangement when the connection arrangement is in an open-end configuration; receiving a first reflection from the connection arrangement, wherein the first reflection is based on the first RF signal; transmitting a second RF signal to the connection arrangement when the connection arrangement is connected to the DUT; receiving a second reflection from the connection arrangement, wherein the second reflection is based on the second RF signal, and wherein the second reflection is associated with a corresponding return loss contribution from the DUT; obtaining information about at least one of the DUT or the tester performing the transmitting and receiving operations; and training the machine learning model based on the information, the first reflection, and the second reflection. The exemplary method may include one or more of the following features (alone or in combination).

[0008] The method may include performing the operation for at least one of: different types of DUTs or different instances of the same type of DUT. The method may include performing the operation for at least one of: different types of testers or different instances of the same type of tester. The method may include performing the operation for at least one of: different types of connection setups or different instances of the same type of connection setup. The method may include performing the operation for at least one of: different types of connections between the connection setup and the DUT or different instances of the same type of connection between the connection setup and the DUT. The machine learning model may include one or more machine learning models and may be further configured to classify the DUT and characterize the electrical connection including the connection setup to the DUT. The machine learning model may include one or more of a neural network model or a large language model. One or more non-transitory machine-readable media may store instructions that are executable by one or more processing devices to implement a simulator configured to perform the method with or without any one or more of the aforementioned characteristics.

[0009] Any two or more of the features described in this specification (including this summary) can be combined to form implementations not specifically described in this specification.

[0010] At least a portion of the devices, systems, and processes described in this specification may be configured, controlled, and / or implemented by executing instructions stored on one or more non-transitory machine-readable storage media on one or more processing devices. Examples of non-transitory machine-readable storage media include read-only memory, optical disk drives, memory disk drives, and random access memory. At least a portion of the devices, systems, and processes described in this specification may be configured, controlled, and / or implemented using a computing system consisting of one or more processing devices and a memory storing instructions that may be executed by the one or more processing devices to perform various control operations. The devices, systems, and processes described in this specification may be configured, for example, by design, construction, composition, arrangement, placement, programming, operation, activation, deactivation, and / or control.

[0011] The details of one or more implementations are set forth in the accompanying drawings and the detailed description which follows. Other features and advantages will be apparent from the detailed description and drawings, and from the claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 is a block diagram illustrating exemplary S-parameters for devices included in an exemplary test configuration.

[0013] Figure 2 The diagram shows different configurations with connection setup and device under test (DUT) Figure 1Block diagram of the test configuration.

[0014] Figure 3 is a block diagram of an exemplary test system.

[0015] Figure 4 is a block diagram of an exemplary test system in an open-end configuration.

[0016] Figure 5 is a block diagram of an exemplary test system in a closed-end configuration, where the test system is connected to a DUT.

[0017] Figure 6 is a block diagram of an exemplary analyzer configured to use machine learning techniques to determine the quality of a connection between a test system and a DUT.

[0018] Figure 7 is a block diagram of example signal processing blocks that may be included in an example analyzer.

[0019] Figure 8 is a graph illustrating an exemplary operation of a filter included in a signal processing block.

[0020] Figure 9 is a block diagram illustrating an exemplary simulation of the connection between the test system and the DUT.

[0021] Figure 10 is a circuit diagram showing exemplary connection points between a connection setup and a DUT.

[0022] Figure 11 is a circuit diagram representing an exemplary DUT.

[0023] Figure 12 is a block diagram illustrating components of a radio frequency simulator for training machine learning models.

[0024] Figure 13 is a flowchart illustrating operations included in an exemplary process for training a machine learning model.

[0025] Like reference numbers in different drawings identify similar elements. DETAILED DESCRIPTION

[0026] Test systems, such as automatic test equipment (ATE), are configured to test the operation of electronic devices, known as devices under test (DUTs). Examples of DUTs that can be tested by test systems include electronic devices, such as transmitters or microprocessors, and system-level devices, such as smartphones. Testing can include radio frequency (RF) testing for testing the RF components of the DUT. RF testing can be performed over wired media, such as RF transmission lines or coaxial cables.

[0027] The quality of the RF connection between the test system and the DUT is a factor that affects the accuracy and reliability of RF testing. The quality of the RF connection to the DUT can also affect RF test results. For example, in manufacturing, the quality of the connection to the DUT can affect Cpk (process capability index) and yield rate. Poor-quality connections can reduce RF test quality and / or increase RF test costs.

[0028] The quality of the RF connection to the DUT can degrade due to mechanical vibration or variations and wear of the RF probe, which can be difficult to measure and detect directly. In one example, the RF probe is an electrical connection between the DUT and an RF connection setup, which may be referred to herein as a "connection setup." In some implementations, the connection setup includes one or more media over which RF signals are transmitted, including, but not limited to, transmission media and connectors between the test system and the DUT. Connection problems are often only noticed when those connection problems begin to affect RF test results.

[0029] Return loss (RL) is a metric that can indicate the quality of an RF connection. In some examples, RL is a measure of the power ratio of the signal reflected by the DUT and / or discontinuities in the RF transmission line or cable that is part of the RF connection to the DUT. Scattering parameters known as "S-parameters" characterize the behavior of an RF connection. Among the S-parameters, the S11 parameter corresponds to RL. The S11 parameter is a complex number with a magnitude and a phase. RL is typically expressed as the negative of the S11 parameter's magnitude in decibels (dB). The smaller the S11 parameter's magnitude, the greater the RL, and vice versa.

[0030] Because of the RF connection between the connection setup and the DUT, the S11 parameter value for the RF connection between the test system and the DUT will be embedded with the S11 parameter value of the DUT. Thus, the S11 parameter value measured for the RF connection between the test system and the DUT will not be specific to the RF connection alone. In other words, the S11 parameter value measured for the RF connection between the test system and the DUT is not an accurate representation of the RL of the connection setup, but rather includes contributions from the DUT.

[0031] At this point, Figure 1 An exemplary test system 10 is shown having S-parameters 11, a connection setup 12 having S-parameters 14, and a DUT 15 having S-parameters 16. In some implementations, the S-parameters of the test system 10, the connection setup 12, and the DUT 15 all have different values. The S-parameters of the test system 10 may be known or determined. When a test system measurement Sm 19 made using the connection setup 12 connected to the DUT 15 is available, the challenge is to determine the value of the S-parameter 17 of the connection setup 12.

[0032] Reference Figure 2The exemplary systems and processes described herein determine the value of the S11 parameter of the test system 10 based on the measurement of the reflection of the signal output to the open-ended connection setup, as described below. The problem then becomes determining a circuit model for the connection setup and a fitted Sm 19 ( Figure 1 ). The mathematical calculations required to fit the circuit model to the SM can be computationally prohibitive because they require searching multiple parameters and circuit configurations for multiple connection setups 12a and multiple DUT configurations 15a.

[0033] Thus, the exemplary system and process performs circuit modeling and uses machine learning (ML) to determine the quality of the RF connection between the test system and the DUT (e.g., RL or S11 parameter values). Because it uses S11 data for different DUTs and trains the ML model, the process can quickly adapt to different types of DUTs. The system and process can utilize an RF simulator to generate a large and diverse training data set covering a variety of test systems, connection settings, DUTs, and connection qualities, thereby avoiding the need to manually collect and label real data, which may be impractical in some cases.

[0034] The processes described herein can be performed within a test system or on a local computer. However, at least a portion of the process, such as ML training, can be performed using services external to the test system or local computer. For example, ML training can be performed using cloud computing.

[0035] An exemplary system of the type described above includes a tester configured to test a DUT, and a connection arrangement (such as one or more RF transmission lines or coaxial cables) that is connectable to and disconnectable from the DUT. The tester is configured to transmit an RF signal through the connection arrangement and to capture a reflected signal from the connection arrangement based on the transmitted RF signal. One or more processing devices are configured to use a trained ML model to determine the quality of the connection between the test system and the DUT (such as an RL or S11 parameter value of the connection arrangement) based at least on the reflected RF signal.

[0036] Figure 3 Components 20 of an exemplary test system 21 are shown. In some implementations, component 20 includes a vector signal generator (VSG) 22, a vector signal analyzer (VSA) 24, and a local oscillator (LO) 25. Component 20 may form a vector network analyzer (VNA) and may be included in one or more test instruments 26, such as those described below. An exemplary VNA is a device configured to obtain the magnitude and phase of an input signal at a given frequency fi.

[0037] The exemplary VSG 22 is a hardware device configured to generate signals and output these signals to the DUT 33 via the connection setup 27. The exemplary connection setup may include, but is not limited to, RF transmission lines, coaxial cables, connectors, and / or other components that can be connected to and disconnected from the DUT. The signals may be or include RF signals used to test the DUT or to test the quality of the connection setup 27.

[0038] The exemplary VSA 19 is a hardware device configured to receive a signal from the connection setup 27 and identify the signal by measuring parameters (such as magnitude and / or phase) of the received signal. The received signal may be a full or partial reflection of an RF signal transmitted by the VSG 22. For example, the VSA 24 may be configured to compare the received signal with one or more reference signals or with one or more threshold values ​​to identify one or more parameters for the received signal.

[0039] An exemplary LO 26 is a hardware device that feeds up-conversion and down-conversion mixer circuits for up-converting and down-converting frequencies between the VSA or VSG and the combiner circuit 28. In some implementations, the LO 26 may be a separate component or may be incorporated into the VSG 22 or VSA 24.

[0040] In some implementations, the VSA 24, VSG 22, and LO 26 can be separate hardware devices. In some implementations, the VSA 24, VSG 22, and LO 26 can be combined into a single hardware device. In some implementations, the VSA 24, VSG 22, and LO 26 can be implemented using one or more semiconductor devices (such as transistors, diodes, and / or integrated circuits). In some implementations, the VSA 24, VSG 22, and LO 26 can be implemented using one or more processing devices (such as those described herein) that are configured to execute instructions stored in memory to implement VSA, VSG, and LO functionality. In some implementations, the VSA 24, VSG 22, and LO 26 can be implemented using a combination of one or more semiconductor devices and one or more processing devices.

[0041] Combiner circuit 28 or other device may be configured to route signals between VSA 22, VSG 24, and devices connected to and external to test system 21. The combiner circuit may be integrated into the same device or devices as VSA 24, VSG 22, and LO 26, or the combiner circuit may be a separate hardware device.

[0042] The VSA 24, VSG 22, LO 26, and combiner circuit 28 may be part of a test instrument 26 configured to test a DUT, such as those described herein. An exemplary test instrument is a hardware device that may include one or more processing devices 31, Figure 3 The test instrument 26 may further include test electronics such as a parametric measurement unit (PMU) and / or pin electronics (PE) 32.

[0043] The test instrument 26 may be configured (e.g., programmed) to output a test signal to test the RF connection setup to the DUT. For example, the processing device 31 may also be configured to control components 20 included in the test instrument 26 based on the execution of instructions 30 stored in the memory 29. In some implementations, the test instrument 26 may be configured to use ML techniques as described herein to determine the quality of the RF connection between the test system and the DUT. The test instrument 26 may also be configured (e.g., programmed) to output a test signal to test the DUT via the connection setup. The test signal used to test the DUT may be or include commands, instructions, data, parameters, variables, test vectors, and / or any other information designed to elicit a response from the DUT. The test instrument 26 may be configured to receive a response to the test signal and analyze the response to determine whether the DUT passed or failed the test. The test system 21 may include a control system 34. The control system 34 may include one or more processing devices 35 and a memory 36 storing instructions 37, which may be executed by the one or more processing devices 35 to perform various functions, including those listed below. In some implementations, control system 34 can control the operation of test instrument 26 (including component 20) as part of a testing process for determining the quality of the RF connection between the test system and the DUT. In some implementations, control system 34 can be configured to use measurements from test instrument 26 to determine the quality of the RF connection between the test system and the DUT using ML techniques.

[0044] In some implementations, a test system or component thereof that performs testing can be referred to as a “tester.” In some implementations, a test instrument or component thereof that performs testing functions can be referred to as a “tester.”

[0045] In exemplary operation, the VSG 22 can be controlled by the test instrument 26 or the control system 34 to transmit an unmodulated carrier signal 39 to the connection setup 27. The VSA 24 can be controlled by the test instrument 26 or the control system 34 to capture data representing reflections 40 of at least a portion of the unmodulated carrier signal for each of a plurality of frequencies f in the reflections 40. The captured data can include the magnitude and phase of the reflected signal at each frequency f. The captured data constitutes the aforementioned Sm measurement. The test system repeats these operations for a predetermined number of frequency points over a predefined frequency range and ultimately collects a data set as a frequency vector. In some implementations, the predetermined number of frequency points can be on the order of hundreds or thousands of frequencies. Thus, the vector includes data for one or more reflected signals at a plurality of different frequencies. The test system generates a data file based on the vector.

[0046] The VSG 22 and the VSA 24 are controllable to generate two data files using the above-described operations. Figure 4 and Figure 5 Different test system configurations 41 and 44 are shown for generating two data files, respectively. Figure 4 A test configuration is shown in which the connection arrangement 27 (eg, RF transmission line and / or coaxial cable) is open ended. In this context, open ended means that the connection arrangement is not connected to the DUT or any other device. Figure 5 FIG. 2 shows a test configuration in which a connection arrangement 27 is connected to the DUT. Figure 4 and Figure 5 As shown, the test system is calibrated to a calibration plane 45. In some implementations, measurements made by the test system 21 can be made relative to this calibration plane. In some implementations, calibration information for this plane is included in each data file and used to remove all or some of the connection quality contribution from the test system from the data in each data file.

[0047] To capture the first data file, referred to as "Data File #1", the VSA 24 and VSG 22 are controlled to use Figure 4 The test configuration operates as described above. The VSG 22 transmits first RF signals through the connection setup 27, and the VSA measures reflections of those first RF signals from the open end of the connection setup. Data file #1 includes magnitude and phase measurements of the reflected first RF signals at a plurality of frequency points fi. Data file #1 may also include the magnitude and phase of the transmitted first RF signals at a plurality of frequency points fi and an identification of the test system (such as its type, manufacturer, model, etc.). The data from data file #1 is used to detect Figure 4 The location of the open (or reference) point 64 in the time domain. In some implementations, the open (or reference) point is the connection point 46 ( Figure 5 ).

[0048] To capture a second data file, referred to as "Data File #2," the VSA 24 and VSG 22 are controlled to use Figure 5 The test configuration operates as described above. VSG 22 transmits second RF signals via connection setup 27, and the VSA measures reflections of those second RF signals. Some of the second RF signals may be identical to corresponding ones of the first RF signals. Data file #2 includes magnitude and phase measurements of the reflected second RF signals at multiple frequency points f. Data file #2 may also include the magnitude and phase of the transmitted second RF signals at multiple frequency points f, as well as an identification of the test system (such as its type, manufacturer, model, etc.). The data file may be processed by an analyzer to determine connection quality / RL / S11 parameter values ​​for connection setup 27.

[0049] exist Figure 6 Components of an exemplary analyzer 47 are shown in . Analyzer 47 may be implemented in software (eg, executable instructions) that is executed by a processing device on test instrument 26 , on control system 34 , or external to test system 21 .

[0050] In this example, the analyzer 47 includes one or more classifier ML models 49 and one or more regressor ML models 59. The classifier ML models 49 have been trained to classify the type of DUT with its RF connection in the test system configuration, such as Figure 4 and Figure 5 Examples include "open," meaning no DUT is connected to the connection setup; "any DUT," meaning the DUT is not a type known to the analyzer 47; and DUT #1, DUT #2, etc., where DUT #1, DUT #2, etc. refer to predefined DUTs whose identities are programmed into the analyzer 47.

[0051] The regressor ML models 50 include at least two different types of regressor ML models. Some regressor ML models 50 have been trained to determine the RL of the RF connection between the test system and the DUT, which removes the RL contribution from the DUT. These are referred to as RL regressor models. Some regressor ML models 50 have been trained to determine the electrical characteristics of the RF connection between the test system and the DUT. These are referred to as lumped circuit (LC) regressor models. In this regard, the electrical characteristics can be represented by inductance (L), capacitance (C), resistance (R), delay (D), etc., concentrated at a single point, and their behavior can be described by an idealized mathematical model. In some implementations, the electrical characteristics can be modeled using a circuit having one or more inductors, one or more capacitors, one or more resistors, one or more delay elements, and / or other passive circuit components.

[0052] Analyzer 47 is configured to receive data file #1 52 and data file #2 54. Analyzer can be configured to perform mathematical calculations 55 and 56 on each corresponding data file based on calibration data for the test system. The mathematical calculations can be based on a 1-port VNA calibration. For example, the data included in data file #1 52 and data file #2 54 can be processed to remove RL contributions from test system 21 itself from the data in each data file. Analyzer 47 is also configured to select one of ML models 49 to use as a classifier ML model 60 based on information about the identity of test system 21 from one of the data files (such as data file #1).

[0053] The signal processing block 57 of the analyzer 47 is configured to receive processed data from data file #1 52 and data file #2 54 and use the data in these data files to determine the magnitude and phase vector based on the reflected signal. Figure 7 The operation of the signal processing block 57 for determining these magnitude and phase vectors is described.

[0054] The output 59 of the signal processing block is passed to the classifier ML model 60. The classifier ML model 60 is executed to determine the DUT 33 (e.g., Figure 3 、 Figure 5 ) is "Any", "DUT#1", "DUT#2", etc. or there is no DUT (e.g. Figure 4 ), in which case the DUT is characterized as "open". This information 114 can be output from the analyzer 47 to the user. In some implementations, the analyzer 47 is configured to use the identity of the DUT determined by the classifier ML model 60 to have the highest probability of selecting at least two regressor models. In this example, the selected regressor models include the RL regressor model 61 and the LC regressor model 62. The analyzer 47 is also configured to use the magnitude and phase vectors output by the signal processing block 57 as input to the RL regressor model 61 and the LC regressor model 62.

[0055] The RL regressor model 61 has been trained to provide the output signal in the absence of data from the DUT 33 ( Figure 3 、 Figure 5 ) of the connection setting 27 at one or more predefined frequency points in the case of the RL or S11 parameter contribution of the connection setting 27. Figure 5In the test system configuration of FIG. 5 , RL regressor model 61 is configured to output RL or S11 parameter values ​​111 of connection setup 27 between test system 21 and connection point 46. LC regressor model 62 is configured to provide output parameters 112 (such as inductance, capacitance, resistance, etc.) that model connection setup 27 between test system 21 and connection point 46 to the DUT based on the magnitude and phase vectors output by signal processing block 57. In some implementations, LC regressor model 62 can be configured to output a textual definition or a graphical display of a circuit configuration that includes passive circuit elements (such as those described herein) that model the connection setup.

[0056] Figure 7 is a block diagram of an exemplary implementation of signal processing block 57. Signal processing block 57 may be implemented in software (eg, executable instructions) executed by one or more processing devices on test instrument 26, on control system 34, and / or external to test system 21.

[0057] The signal processing block 57 is configured to receive data file #1 52 and data file #2 54. The data from data file #1 52 and data file #2 54 are resampled along the frequency bins by resampler filters 66 and 67, respectively, to have the same characteristic size (e.g., frequency bin size) as the analyzer 47 and thereby generate resampled data 69 and 70. A filter bank 71 having K (where K is an integer greater than one) filters is generated, wherein each filter has a different time window size (resolution), such as Figure 8 More specifically, in Figure 8 In the example of FIG. 5 , exemplary filters 71 a, 71 b, 71 c from the filter bank have respective resolutions 72 a, 72 b, 72 c to attenuate data from data file #1 52 and data file #2 54 in the time domain. The filters attenuate data from data file #1 52 and data file #2 54 to focus the data from each data file at connection point 46 ( FIG. 5 ) to the DUT 33. Figure 5 )superior.

[0058] Return to view Figure 7 , in this example, each filter outputs a vector, resulting in a Figure 7 The six output vectors 74, 75 of the three-filter example shown in FIG. It should be noted that the signal processing block 57 may include more than three filters or less than three filters. These output vectors are Figure 7is referred to as filtered data. Each of the six vectors has the same size. To remove the effects of path loss on the data in vectors 74, 75, the elements of vector 74 based on data file #1 are divided (76) by the corresponding elements of vector 75 based on data file #2, or the elements of vector 75 based on data file #2 are divided by the corresponding elements of vector 74 based on data file #1. In this context, path loss may include the reduction in the power density of the signal as the signal travels along the connection arrangement 27. The resulting data 77 is combined data consisting of N vectors, each vector having M elements, where M and N are integers greater than one. The combined data 77 having N vectors (each vector having M features) is then passed to a magnitude calculator 79 and a phase calculator 80. The output of the magnitude calculator 79 is magnitude data 81 consisting of N vectors, each vector having M features. The output phase calculator 80 is phase data 82 consisting of N vectors, each vector having M features.

[0059] In some implementations, the ML model can be or include a 1-D multi-layer convolutional neural network. The network can be trained independently for each classifier ML model and regressor ML model, which can then be stored and loaded into the computing system as separate ML model files. The input data size can be 2N vectors, where each vector has M features, such as the magnitude and phase vectors described above. In some implementations, the ML model can be or include a large language model or any other neural network model.

[0060] Relative to Figures 9 to 13 Training of an ML model is described. Training can be performed on a computing system that is part of the test system, or it can be performed on a computing system external to the test system. For example, training can be performed on a cloud computing system. The resulting trained ML model can then be stored in any memory on the test system.

[0061] The techniques described herein can use an RF simulator to generate training data sets covering various types of connection setups and connection qualities. An exemplary RF simulator is a computer program that simulates the test system, the connection setup, the DUT, and the RF connections between these components. The RF simulator can be executed on a computing system that is part of the test system, or it can be executed on a computing system external to the test system. For example, the RF simulator can be executed on a cloud computing system.

[0062] Figure 9A block diagram illustrates an exemplary simulation that can be generated by an RF simulator for a test system configuration including a test system 84 and a DUT 85, which can be versions of the test system 21 and the DUT 33, respectively. In this regard, S-parameter files can be stored in the RF simulator for various types of testers and DUTs. For example, S parameter files can be stored for testers with different capabilities, for testers from different manufacturers, for different models of the same tester from the same manufacturer, for different instances of the same model of a tester, and so on. For example, S parameter files can be stored for DUTs with different capabilities, for DUTs from different manufacturers, for different models of the same DUT from the same manufacturer, for different instances of the same model of a DUT, and so on. Blocks 89, 90, and 91 represent simulated connection setups, simulated DUT connection points, and simulated arbitrary DUTs, respectively, all of which can be examples of the same elements described above.

[0063] In this example, the RF simulator is configured to randomly generate different combinations of test systems, connection setups, reference points, and DUTs, and train an ML model based on these combinations and stored S-parameter files for the test systems and DUTs. In some implementations, the connection setups are randomly generated to include different coaxial cables, transmission lines, attenuators, and / or connections. In some implementations, the connection setups include different types of connection setups or different instances of the same type of connection setup. The connection point 90 to the DUT can be modeled as follows: Figure 10 Shunt capacitor (C) 92 and inductor (L) 94 are shown, or any other suitable circuit configuration. The capacitance and inductance values ​​may be randomly selected from a distribution designed to provide a near uniform distribution of regression labels.

[0064] The DUT 85 may be defined by a user input S11 data file. The DUT 85 may be defined using a single S11 parameter value or multiple S11 data files. In the case of multiple S11 data files for a predefined DUT type, the RF simulator will randomly pick an S11 parameter value with equal probability for a generated data sample. An arbitrary DUT 91 is a model of a DUT type that is not classified as any of the predefined DUT types. The model 95 of the DUT may be a series of inductors (e.g., L1, L2, etc.) and capacitors (e.g., C1, C2, etc.) combined with additional losses and delays 96, 97, etc., as shown in FIG. Figure 11 As shown. Other suitable circuit configurations can be used to simulate the DUT. All parameters of the model can be randomly selected.

[0065] A training dataset can be generated independently for each classifier ML model or regressor ML model. For the classifier ML model, data labels are generated based on the DUT type used in the simulated connection setup 89. For the RL regressor ML model, data labels are generated based on the S11 data of the DUT connection point model 90. For the LC regressor ML model, data labels are generated based on the capacitance and inductance values ​​used in the DUT connection pattern 901.

[0066] Figure 12 An exemplary RF simulator 97 for generating a trained ML model is shown. The RF simulator 97 may be implemented in software (e.g., executable instructions) that is executed by a processing device (e.g., on a control system or external to the test system 21, such as in a cloud computing environment).

[0067] The RF simulator 97 may include a DUT database 99 containing S11 parameter values ​​for various DUTs and a tester database 100 containing S11 parameter values ​​for various test systems and information about the configuration of these test systems. In some implementations, a user may input S11 parameter values ​​101 for the DUT on which the ML model will be trained. In some implementations, a user may input information 102 to configure the DUT on which the ML model (such as a circuit model) will be trained. In some implementations, a user may input test vectors 104 for training. The user may also provide input 105 to start or "kick" training. A handler process 106 is configured to select information from the DUT database 99 and the tester database 100 based on the user inputs 101, 102, and / or 104. Code 103 (simulation platform) executed in RF simulator 97 generates training and validation data 107 based on data from DUT database 99, tester database 100, and user input 101, 102, and / or 104. The training data may include simulated measurements Sm of reflected signals, for example, as described herein with respect to Figure 3 Given a simulation test configuration, the verification data may include an expected S11 value.

[0068] The RF simulator 97 uses the training data to generate a trained classifier ML model 60, an RL regressor ML model 61, and an LC regressor ML model 62, and stores the classifier ML model 60, the RL regressor ML model 61, and the LC regressor ML model 62 in respective databases 104, 105, and 106. The classifier ML model 60, the RL regressor ML model 61, and the LC RL regressor ML model 62 can be distributed from the databases 104, 105, and 106 to the analyzer 47 or accessed from the databases 104, 105, and 106 by the analyzer 47. In some implementations, more or fewer than three ML models can be used. For example, functionality for two ML models can be combined into a single ML model, or functionality for a single model can be split into two ML models.

[0069] In some implementations, the RF simulator 97 uses the analyzer 47 and the validation dataset to test the trained classifier ML model, the RL regressor ML model, and the LC regressor model. Based on these tests, the system generates one or more training reports 109. The training report 109 can be output to the user to allow the user to determine the accuracy of the trained ML model before deploying it on the test system.

[0070] Reference Figure 13 , the RF simulator 97 may be configured to perform the operations of an exemplary process 110 to generate training data. The process 110 includes transmitting (110a) a first RF signal to the connection setup when the connection setup is in an open-end configuration. The operations may be performed by simulating a VSG and a connection setup (similar to Figure 3 The process 110 includes receiving (110b) a first reflection from a connection setup. The first reflection is based on the first RF signal; for example, the first reflection may include all or part of the first RF signal. This operation may be performed by an analog VSA and a connection setup (similar to Figure 3 The process 110 includes transmitting (110c) a second RF signal to the connection arrangement when the connection arrangement is connected to the DUT. This operation can be performed by simulating the VSG and the connection arrangement (similar to Figure 3 The process 110 includes receiving (110d) a second reflection from the connection setup. The second reflection is based on the second RF signal and is associated with a corresponding RL contribution from the DUT. This operation can be performed by simulating the VSA and the connection setup (similar to Figure 31 (e.g., the one shown) is performed to generate data file #2. Process 110 includes obtaining (110e) information about at least one of a DUT (e.g., DUT S11 parameters) or a tester (e.g., tester S11 parameters) performing transmit and receive operations. Operations 110a to 110e may be repeated hundreds, thousands, or millions of times to obtain training data for an ML model (e.g., a classifier ML model, an RL regressor ML model, and an LC regressor model). The results of performing operations 110a to 110e multiple times generate a training data set. Based on the training data set, the ML model may be trained (110f) for: different types of DUTs and / or different instances of the same type of DUT; different types of testers and / or different instances of the same type of tester; and / or different types of connection setups between the test system and the DUT, and / or different instances of the same type of connection setup between the test system and the DUT.

[0071] All or a portion of the systems and processes described herein (including but not limited to process 110), and modifications thereof, may be implemented, configured, and / or controlled, at least in part, by one or more computers using one or more computer programs tangibly embodied in one or more information carriers, such as one or more non-transitory machine-readable storage media. Computer programs may be written in any form of programming language, including compiled or interpreted languages, and they may be deployed in any form, including as stand-alone programs or as modules, parts, subroutines, or other units suitable for use in a computing environment. Computer programs may be deployed to execute on one computer or on multiple computers that are distributed and interconnected at one site or across multiple sites.

[0072] The actions associated with implementing, configuring, or controlling the test systems and processes described herein may be performed by one or more programmable processors executing one or more computer programs to control or perform all or some of the operations described herein. All or a portion of the test systems and processes may be implemented, configured, or controlled by dedicated logic circuitry (such as an FPGA (field programmable gate array) and / or an ASIC (application-specific integrated circuit)) or an embedded microprocessor localized to the instrument hardware.

[0073] Processors suitable for computer program execution include, for example, both general-purpose and special-purpose microprocessors, and any one or more processors of any type of digital computer. Typically, a processor will receive instructions and data from a read-only memory area or a random access memory area, or both. The elements of a computer include one or more processors for executing instructions and one or more memory devices for storing instructions and data. Typically, a computer will also include (or be operatively coupled to receive data from it or transfer data to it, or both) one or more machine-readable storage media, such as a mass storage device for storing data, such as a magnetic disk, magneto-optical disk, or optical disk. Non-transitory machine-readable storage media suitable for embodying computer program instructions and data include all forms of non-volatile storage, including, by way of example, semiconductor memory devices such as EPROM (Erasable Programmable Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), and flash memory devices; magnetic disks, such as internal hard disks or removable disks; magneto-optical disks; and CD-ROMs (Compact Disc Read-Only Memory) and DVD-ROMs (Digital Versatile Disc Read-Only Memory).

[0074] All examples described herein are non-limiting.

[0075] In the description and claims provided herein, the adjectives "first," "second," "third," etc. do not specify priority or order unless the context indicates otherwise. Instead, these adjectives are merely used to distinguish the nouns they modify.

[0076] Any mechanical or electrical connection herein may include a direct physical connection or an indirect physical connection involving one or more intervening components. A connection between two conductive components includes an electrical connection unless the context indicates otherwise. Signals described herein are electrical signals unless the context indicates otherwise.

[0077] Elements of the different described implementations may be combined to form other implementations not specifically described above. Elements may be omitted from a previously described system without generally adversely affecting its operation or the operation of the system. Furthermore, individual elements may be combined into one or more single elements to perform the functions described herein.

[0078] Other implementations not specifically described in this specification are also within the scope of the following claims.

Claims

1. A system, comprising: a tester configured to test a device under test (DUT); a connection arrangement connectable to and disconnectable from the DUT; wherein the tester is configured to transmit a radio frequency (RF) signal through the connection arrangement and capture a reflected signal from the connection arrangement, the reflected signal being based on the RF signal; and One or more processing devices configured to use the trained machine learning model to determine a quality of a connection between the test system and the DUT based on the reflected signal.

2. The system of claim 1 , wherein the tester is configured to capture first data when the connection setup is disconnected from the DUT, the first data being based on a first reflected signal among the reflected signals; and The tester is configured to capture second data when the connection setup is connected to the DUT, the second data being based on a second reflected signal among the reflected signals. 3 . The system of claim 2 , wherein each of the second ones of the reflected signals is associated with a respective return loss, the respective return loss comprising a return loss contribution from the DUT. 4 . The system of claim 3 , wherein determining the quality of the connection comprises determining a return loss of the connection setup when the connection setup is connected to the DUT minus the return loss contribution from the DUT. 5 . The system of claim 3 , wherein the one or more processing devices are configured to process data based on the reflected signals in the first data and the second data to remove a return loss contribution from the tester.

6. The system according to claim 5, wherein: After processing, the one or more processing devices are configured to resample data based on the reflection signals in the first data and the second data.

7. The system according to claim 6, wherein: After resampling, the one or more processing devices are configured to filter data based on the reflected signals in the first data and the second data to attenuate representations of the reflected signals in the first data and the second data.

8. The system according to claim 7, wherein: After filtering, the one or more processing devices are configured to process data based on the reflected signals in the first data and the second data to reduce path loss associated with the connection setup in each signal and thereby generate combined data.

9. The system of claim 8, wherein the combined data comprises magnitude data and phase data based on the reflected signals in the first data and the second data, wherein the input to the trained machine learning model is based on the magnitude data and the phase data.

10. The system of claim 1 , wherein the one or more processing devices are configured to use a classifier machine learning model to identify the type of the DUT; and Wherein the classifier machine learning model selects the trained machine learning model based on the type of the DUT.

11. The system of claim 1 , wherein the one or more processing devices are configured to use a classifier machine learning model to identify the type of the DUT; wherein the classifier machine learning model selects a second trained machine learning model based on the type of the DUT for use in determining electrical characteristics of connections, the connections including the connection setup to the DUT; and Wherein the one or more processing devices are configured to execute the second trained machine learning model to determine the electrical characteristic of the connection.

12. The system of claim 11, wherein the electrical characteristic comprises at least one of a capacitance or an inductance of the connection, the connection comprising the connection arrangement to the DUT.

13. A method of training a machine learning model for determining the quality of a connection between a tester and a device under test (DUT), the method comprising the following operations: transmitting a first RF signal to the connection arrangement when the connection arrangement is in an open-ended configuration; receiving a first reflection from the connection setup, the first reflection being based on the first RF signal; transmitting a second RF signal to the connection arrangement when the connection arrangement is connected to the DUT; receiving a second reflection from the connection setup, the second reflection being based on the second RF signal, the second reflection being associated with a corresponding return loss contribution from the DUT; obtaining information about at least one of the DUT or the tester performing the transmit and receive operations; as well as The machine learning model is trained based on the information, the first reflection, and the second reflection.

14. The method of claim 13, wherein the method comprises performing the operations for at least one of: different types of DUTs or different instances of the same type of DUT.

15. The method of claim 13, wherein the method comprises performing the operations for at least one of: different types of testers or different instances of the same type of tester.

16. The method of claim 13, wherein the method comprises performing the operations for at least one of: different types of connection setups or different instances of the same type of connection setup.

17. The method of claim 13, wherein the method comprises performing the operations for at least one of: different types of connections between the connection setup and the DUT or different instances of the same type of connection between the connection setup and the DUT.

18. The method of claim 13, wherein the machine learning model comprises one or more machine learning models and is further configured to classify the DUT and characterize the electrical connections including the connection setup to the DUT.

19. The method of claim 13, wherein the machine learning model comprises one or more of a neural network model or a large language model.

20. One or more non-transitory machine-readable media storing instructions executable by one or more processing devices to implement a simulator configured to perform the method of claim 13.

Images