Probe positioning and pressing amount control method based on visual feedback and motion control

Abstract

This invention relates to a method for RF probe alignment and pressure control based on visual feedback and motion control, belonging to the field of semiconductor testing or module testing technology. The method includes: acquiring the geometric parameters of the RF probe and target pad through machine vision, mapping them to a unified spatial coordinate system after preprocessing to generate a three-dimensional relative positional relationship; driving a multi-axis displacement mechanism for visual servo alignment based on the relationship, and dynamically correcting the motion parameters in real time; controlling the RF probe to perform step-by-step, graded, uniform pressure, comparing continuously captured contact feature signals with a preset threshold, and performing a preset overpressure after contact is determined; adjusting the output torque and driving the RF probe to perform micro-displacement correction based on the deviation between the contact deformation signal and the standard value. This achieves high-precision, adaptive control of the RF probe alignment and pressure process, improving the repeatability, stability, and efficiency of the test.

Description

Technical Field

[0001] This invention belongs to the field of semiconductor testing or module testing technology, specifically relating to a probe alignment and pressure control method based on visual feedback and motion control. Background Technology

[0002] In the field of semiconductor and optical communication module testing, precise alignment and controllable pressure between RF probes and target pads are crucial for ensuring the reliability and consistency of electrical performance testing. Currently, the industry primarily relies on high-precision fully automated RF probe stations derived from wafer testing, such as the RF probe station disclosed in CN112992712A. This equipment, through the cooperation of a support body, feeding mechanism, and RF probe detection device, achieves automated module transfer and RF probe contact detection. It uses multiple sets of translation mechanisms to drive the detection stage and RF probes to move relative to each other in the vertical direction, and uses camera image acquisition to assist in positioning, which to some extent replaces manual operation, improves basic testing efficiency, and meets the initial needs of batch testing of some communication modules. However, when applied to testing scenarios of heterogeneous component assembly products such as optical communication modules, this automated RF probe station still has limitations and is difficult to adapt to the inherent assembly tolerances of the modules. The specific defects are as follows: The positioning logic relies on preset coordinates and lacks dynamic correction capabilities. The RF probe movement path of this device is based on a preset program and fixed coordinate planning. The camera is only used to determine whether the RF probe is above the module for basic positioning, and does not perform quantitative analysis on the real-time three-dimensional relative position of the RF probe and the pads. Due to the mechanical assembly of multiple components in the module, the spatial position and height consistency of the pads are much lower than that of the wafer, with assembly tolerances at the ±50um level. The fixed coordinate driving method is prone to RF probe alignment deviation, causing poor contact or the risk of probe collision.

[0003] The pressure control operates in an open-loop mode, lacking precise contact feedback and overvoltage regulation. The RF probe pressure process is executed according to a preset stroke, without a real-time capture mechanism for contact characteristic signals or a threshold determination mechanism, making it impossible to accurately identify the instantaneous contact between the RF probe and the pad. Furthermore, the device lacks a closed-loop overvoltage adjustment function; the pressure and depth rely entirely on program presets, which can easily lead to overvoltage damage to the RF probe or pad, or undervoltage causing unstable contact, affecting the repeatability of test results.

[0004] The integration of vision and motion control is low, and an adaptive control system has not been formed. The camera in the device only performs positioning and monitoring functions, and the collected image data is not linked with the motion parameters of the multi-axis displacement mechanism. Faced with positional deviations caused by individual differences in modules, it is impossible to dynamically correct parameters such as the movement rate and step size of the RF probe based on visual data. It is difficult to fundamentally compensate for the lack of workpiece precision, and the testing accuracy is highly dependent on the machining accuracy of the mechanical structure, thus limiting the application scenarios.

[0005] However, when applied to testing scenarios involving the assembly of heterogeneous components such as optical communication modules, this automated RF probe station still has limitations. It struggles to adapt to the inherent assembly tolerances of the modules, especially when the module assembly height tolerance is often greater than ±100μm. The fixed coordinate drive method easily leads to RF probe alignment deviations, causing poor contact or the risk of probe collision. Furthermore, the overvoltage of the RF probes is typically only 30–80μm. Achieving stable and reliable micron-level undervoltage control on modules with height fluctuations of ±100μm remains a challenge that existing technologies have not yet solved.

[0006] In summary, while existing automated RF probe stations have achieved preliminary automation in module testing, they have shortcomings in adapting to module assembly tolerances and achieving high-precision adaptive alignment and pressure control. The industry needs a technical solution that integrates real-time visual feedback, dynamic motion control, and closed-loop force control correction. This solution should replace fixed program-driven testing with intelligent control, improving testing efficiency while ensuring stability and reliability, and meeting the needs of large-scale testing of high-precision products such as optical communication modules. Summary of the Invention

[0007] To address the aforementioned problems in the existing technology, this invention provides a probe alignment and pressure control method based on visual feedback and motion control. The objective of this invention can be achieved through the following technical solutions: S1: Obtain the geometric parameters of the RF probe and the target pad of the RF module under test and perform preprocessing operations. Establish the RF probe-pad alignment space coordinate system and generate the three-dimensional relative position relationship between the RF probe and the target pad through coordinate calculation. S2: Based on the three-dimensional relative positional relationship between the RF probe and the target pad, drive the multi-axis displacement mechanism to move the RF probe relative to the stage carrying the module under test in space; collect the real-time positional deviation data between the RF probe and the target pad, and dynamically correct the motion parameters of the multi-axis displacement mechanism; combine the dynamically corrected motion parameters to move the RF probe to a preset safe height above the target pad; S3: Drive the multi-axis displacement mechanism to drive the RF probe to perform step-by-step uniform pressing along the preset pressing direction; continuously capture the contact feature signal between the RF probe and the target pad, compare the contact feature signal with the preset contact judgment threshold, and based on the contact feature signal that reaches the preset contact judgment threshold, control the multi-axis displacement mechanism to stop driving the RF probe to press down, and drive the multi-axis displacement mechanism to drive the RF probe to perform a pressing operation with a preset overvoltage amount; S4: Call the contact characteristic signal of the RF probe and the reference deformation parameter corresponding to the preset standard overpressure to perform difference calculation, generate the deviation value between the current overpressure and the standard overpressure, adjust the output torque of the multi-axis displacement mechanism based on the deviation value, drive the multi-axis displacement mechanism to drive the RF probe to perform micro-displacement correction, and output torque and RF probe micro-displacement amount.

[0008] As a preferred embodiment of the present invention, the preprocessing operation is specifically performed as follows: The original data corresponding to the acquired geometric parameters are converted to grayscale and denoised using a Gaussian filter template. Environmental interference and random noise are filtered out by setting a standard deviation threshold to generate the core contour information of the RF probe and the target pad. The contour edges of the RF probe and the target pad are extracted by setting a gradient double threshold to generate target contour data containing only the RF probe and the target pad. A combination algorithm of morphological dilation and erosion is used for contour optimization. The dilation operation fills the gaps between the contour edges, and the erosion operation removes the burrs on the contour surface and corrects the distortion of the contour edges.

[0009] Specifically, the method for establishing the RF probe-pad alignment spatial coordinate system is as follows: The upper surface of the stage carrying the module under test is set as the horizontal reference plane. The geometric center of the preset positioning hole of the stage is selected as the origin of the coordinate system. The direction perpendicular to the surface of the stage and upward is set as the vertical reference direction. The basic framework of the coordinate system is initially built. The accuracy of each reference axis of the coordinate system is calibrated by standard calibration parts. The scale coefficient of each axis is adjusted to correct the offset error and scaling deviation of the coordinate axis. The parameters of the coordinate system are locked. The coordinate system is used as the RF probe-pad alignment space coordinate system.

[0010] Specifically, the method for generating the three-dimensional relative positional relationship between the RF probe and the target pad is as follows: The geometric parameters of the preprocessed RF probe and target pad are mapped to the RF probe-pad alignment space coordinate system. Key spatial feature data of the RF probe and target pad are extracted. Through spatial vector operation, the distance difference between the center point of the RF probe tip and the center point of the target pad in each reference axis direction is calculated. The angle between the RF probe axis direction and the normal of the pad surface is obtained. Combining the calculated distance difference and angle data, a three-dimensional relative position relationship including relative distance and relative angle is generated.

[0011] Specifically, the method for performing relative spatial movement is as follows: The three-dimensional relative position relationship between the RF probe and the target pad is retrieved. Based on the reference axis directions of the RF probe-pad alignment spatial coordinate system, the directional movement path of the RF probe is generated. The multi-axis displacement mechanism is controlled to drive the RF probe to move in space independently. The single-axis movement stroke is planned according to the relative distance of each reference axis. The movement operation is performed step by step along the horizontal and vertical reference directions, and the RF probe is controlled to move smoothly at a preset rate.

[0012] Specifically, the method for dynamically correcting the motion parameters of the multi-axis displacement mechanism is as follows: The system collects real-time positional deviation data between the RF probe and the target pad, extracts the deviation values ​​between the actual and theoretical displacements of the RF probe in each reference axis direction, compares the deviation values ​​with the preset alignment accuracy threshold, calculates the displacement compensation amount corresponding to each reference axis based on the comparison results, adjusts the step amount and movement speed of each reference axis of the multi-axis displacement mechanism according to the compensation amount, corrects the displacement deviation of the RF probe in each reference axis direction, and updates the motion parameters of the multi-axis displacement mechanism synchronously.

[0013] Specifically, the method for performing the step-by-step, graded, uniform downward pressure is as follows: Based on the preset safe height above the target pad where the RF probe is located, the RF probe pressing direction is determined along the vertical reference direction of the RF probe-pad alignment spatial coordinate system. The graded step spacing of the RF probe pressing is planned, and the multi-axis displacement mechanism is controlled to drive the RF probe to make uniform linear motion along the pressing direction. The graded pressing action is executed based on the preset pressing rules.

[0014] Specifically, the contact feature signal is compared with a preset contact determination threshold, and the specific method is as follows: Extract the continuously captured contact feature signal acquisition data, retrieve the benchmark data of the preset contact judgment threshold, compare the acquisition data and the benchmark data in each direction according to the acquisition dimension of the contact feature signal, calculate the change of the contact feature signal acquisition data relative to the benchmark data of the preset contact judgment threshold, match it with the judgment range of the preset contact judgment threshold, generate the comparison calculation result according to the matching result, and make the contact feature signal compliance judgment based on the comparison calculation result.

[0015] Specifically, the method for performing the pressure reduction operation with a preset overpressure amount is as follows: Based on the contact characteristic signal compliance determination result, the pressure stroke parameters corresponding to the preset overpressure amount are planned, and the multi-axis displacement mechanism is controlled to drive the RF probe to move in a uniform linear motion along the original pressure direction. The pressure distance of the RF probe is controlled according to the planned pressure stroke parameters, and the multi-axis displacement mechanism is controlled to stop driving the RF probe to pressure down.

[0016] Specifically, the method for generating the deviation value between the current overpressure and the standard overpressure is as follows: The reference deformation parameters corresponding to the preset standard overpressure are retrieved. Based on the deformation detection dimension, the corresponding values ​​of the contact feature signal acquisition data and the reference deformation parameters are extracted, and the difference operation is performed to calculate the numerical difference results of each dimension. Combining the numerical difference results of each dimension, the deviation value between the current overpressure and the standard overpressure is generated.

[0017] Specifically, the method for adjusting the output torque of the multi-axis displacement mechanism is as follows: Based on the deviation between the current overpressure and the standard overpressure, and combined with the torque adjustment coefficient of each reference axis of the multi-axis displacement mechanism, the torque adjustment amount is calculated axis by axis. The torque adjustment operation is performed on each reference axis of the multi-axis displacement mechanism, and the matching degree of the torque output of each reference axis is calibrated simultaneously. Based on the calibration results, the torque output value of each axis is finely adjusted.

[0018] Specifically, the output torque and the minute displacement of the RF probe are determined by the following method: Based on the adjusted output torque of the multi-axis displacement mechanism and combined with the micro-displacement adjustment requirements of the RF probe, the micro-displacement parameters of the RF probe in each reference axis direction are planned. The multi-axis displacement mechanism is controlled to run according to the adjusted output torque, which drives the RF probe to perform micro-displacement along each reference axis direction according to the planned displacement parameters. The actual output torque data of the multi-axis displacement mechanism and the actual micro-displacement data of the RF probe in each reference axis direction are recorded simultaneously to form the corresponding data of torque and micro-displacement.

[0019] The beneficial effects of this invention are as follows: (1) By actively acquiring the geometric parameters of the RF probe and the module pads through machine vision and mapping them to a unified spatial coordinate system for calculation, the position and height deviations caused by the mechanical assembly of the module can be accurately identified and quantified. Subsequently, the multi-axis displacement mechanism is driven to move based on the generated three-dimensional relative position relationship, and real-time visual feedback is introduced for dynamic parameter correction. This overcomes the shortcomings of traditional wafer testing methods that rely on fixed strokes and cannot adapt to individual differences in modules. It also fundamentally replaces the inefficient manual alignment method that relies on human experience for accuracy. Furthermore, through real-time visual feedback and dynamic motion correction, this invention can still achieve precise overvoltage control of the RF probe at the 30–80 μm level, even when the module assembly height tolerance exceeds ±100 μm, thus improving the adaptability and reliability of the test.

[0020] (2) By step-by-step, uniformly pressed down, combined with real-time capture of contact characteristic signals and threshold determination, the contact state between the RF probe and the pad can be precisely controlled. After pressing down with a preset overpressure amount, micro-displacement compensation is further completed through deformation signal feedback and closed-loop adjustment of output torque. This process ensures the consistency of the pressing amount and contact force in each test, eliminates the fluctuation of test results caused by uneven force during manual operation, and effectively prevents damage to the RF probe or device caused by overpressure or poor contact, thereby improving the reliability of the test and the service life of the equipment. Attached Figure Description

[0021] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0022] Figure 1 This is a flowchart illustrating the radio frequency probe alignment and pressure control method based on visual feedback and motion control of the present invention. Figure 2 This is an architectural diagram of the radio frequency probe alignment and pressure control method based on visual feedback and motion control of the present invention. Detailed Implementation

[0023] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided.

[0024] Please see Figure 1-2 A probe alignment and pressure control method based on visual feedback and motion control, comprising: S1: Obtain the geometric parameters of the RF probe and the target pad of the RF module under test and perform preprocessing operations. Establish the RF probe-pad alignment space coordinate system and generate the three-dimensional relative position relationship between the RF probe and the target pad through coordinate calculation. S2: Based on the three-dimensional relative positional relationship between the RF probe and the target pad, drive the multi-axis displacement mechanism to move the RF probe relative to the stage carrying the module under test in space; collect the real-time positional deviation data between the RF probe and the target pad, and dynamically correct the motion parameters of the multi-axis displacement mechanism; combine the dynamically corrected motion parameters to move the RF probe to a preset safe height above the target pad; S3: Drive the multi-axis displacement mechanism to drive the RF probe to perform step-by-step uniform pressing along the preset pressing direction; continuously capture the contact feature signal between the RF probe and the target pad, compare the contact feature signal with the preset contact judgment threshold, and based on the contact feature signal that reaches the preset contact judgment threshold, control the multi-axis displacement mechanism to stop driving the RF probe to press down, and drive the multi-axis displacement mechanism to drive the RF probe to perform a pressing operation with a preset overvoltage amount; S4: Call the contact characteristic signal of the RF probe and the reference deformation parameter corresponding to the preset standard overpressure to perform difference calculation, generate the deviation value between the current overpressure and the standard overpressure, adjust the output torque of the multi-axis displacement mechanism based on the deviation value, drive the multi-axis displacement mechanism to drive the RF probe to perform micro-displacement correction, and output torque and RF probe micro-displacement amount.

[0025] Specifically, the preprocessing operation is performed using the following method: The original data corresponding to the acquired geometric parameters are converted to grayscale and denoised using a Gaussian filter template. Environmental interference and random noise are filtered out by setting a standard deviation threshold to generate the core contour information of the RF probe and the target pad. The contour edges of the RF probe and the target pad are extracted by setting a gradient double threshold to generate target contour data containing only the RF probe and the target pad. A combination algorithm of morphological dilation and erosion is used for contour optimization. The dilation operation fills the gaps between the contour edges, and the erosion operation removes the burrs on the contour surface and corrects the distortion of the contour edges.

[0026] Specifically, the method for establishing the RF probe-pad alignment spatial coordinate system is as follows: The upper surface of the stage carrying the module under test is set as the horizontal reference plane. The geometric center of the preset positioning hole of the stage is selected as the origin of the coordinate system. The direction perpendicular to the surface of the stage and upward is set as the vertical reference direction. The basic framework of the coordinate system is initially built. The accuracy of each reference axis of the coordinate system is calibrated by standard calibration parts. The scale coefficient of each axis is adjusted to correct the offset error and scaling deviation of the coordinate axis. The parameters of the coordinate system are locked. The coordinate system is used as the RF probe-pad alignment space coordinate system.

[0027] Specifically, the method for generating the three-dimensional relative positional relationship between the RF probe and the target pad is as follows: The geometric parameters of the preprocessed RF probe and target pad are mapped to the RF probe-pad alignment space coordinate system. Key spatial feature data of the RF probe and target pad are extracted. Through spatial vector operations, the distance difference between the center point of the RF probe tip and the center point of the target pad in each reference axis direction is calculated. The angle between the RF probe axis direction and the normal of the pad surface is obtained. Combining the calculated distance difference and angle data, a three-dimensional relative position relationship including relative distance and relative angle is generated. The key spatial feature data includes the position data of the center point of the RF probe tip, the direction data of the RF probe axis, the position data of the center of the target pad, and the normal data of the pad surface.

[0028] In this embodiment, based on the RF probe alignment and pressure control scenario for RF module RF performance testing, visual data processing and spatial position quantification are performed before precise alignment of the RF probe and the RF pads of the RF module under test. Considering the actual situation where the assembly height tolerance of the module under test is as high as ±100μm, the system adapts to the high-precision testing requirements of RF probe overvoltage of 30-80µm and RF module installation tolerance of ±50µm. This scenario is an automated production line station for RF module mass production testing. All hardware components are integrated into the automated testing equipment, completing the initial mechanical installation, electrical connection, parameter calibration, and debugging. The components work together to achieve visual data acquisition, processing, and spatial position calculation, providing a precise visual feedback data foundation for subsequent automated alignment, step-by-step pressure control, and contact determination of the RF probe. This ensures the accuracy and reliability of the contact between the RF probe and the pads, avoiding problems such as probe collisions and poor contact.

[0029] Implement hardware presets and component debugging In this embodiment, all the pre-installed hardware components have been fixed, their electrical communication debugged, and their initial parameters calibrated. The relative positions of each component remain fixed. The specific configuration and status are as follows: High-precision industrial area array camera: 2048×2048 resolution, 30fps frame rate, equipped with a 0.5x telecentric lens, with the lens optical axis vertically downward and facing the center area of ​​the stage. The field of view can completely cover the tip area of ​​the RF probe and the target pad area of ​​the RF module under test. The camera and vision processing unit establish real-time communication through Gigabit Ethernet, and the data acquisition and data transmission delay is ≤10ms. Vision processing unit: an embedded industrial controller with a built-in machine vision algorithm library and spatial computing module, supporting operations such as grayscale conversion, filtering, morphological operations, spatial vector operations, and coordinate transformation, with a computing accuracy of ≤1µm. It establishes bidirectional communication with the industrial area scan camera and the motion control unit of the multi-axis displacement mechanism. Precision vacuum adsorption stage: Made of marble, with a flatness of ≤0.005mm / m², equipped with a vacuum adsorption module for horizontally supporting and fixing the RF module under test. Two circular preset positioning holes are symmetrically arranged on the stage, with a diameter of Φ2mm and an accuracy of ±0.01mm. The geometric center distance between the two positioning holes is 50mm, providing a reference feature for establishing the coordinate system. High-precision standard calibration component: Ceramic calibration block with a cross-shaped feature point array of known three-dimensional coordinates engraved on the surface. The feature point coordinate accuracy is ±1µm and the feature point spacing is 10mm. It is used to complete the pixel-physical coordinate mapping and accuracy calibration of the coordinate system. RF probe: High-frequency RF test RF probe with a tip diameter of 20um and an RF probe axis straightness of ≤0.005mm / mm. It is fixed to the high-precision RF probe clamping end of the multi-axis displacement mechanism and suspended in the initial area above the stage. The tip is free from deformation and wear. RF module under test: RF communication module using surface mount technology. Its target pad is a rectangular tin-plated pad with a size of 2mm×1mm and a thickness of 0.05mm. The module is fixed to the center of the stage by vacuum adsorption and is within the effective acquisition field of view of the industrial area array camera.

[0030] Specific definition of geometric parameters In this embodiment, the geometric parameters of the RF probe and the target pad of the module under test, acquired by the high-precision industrial area array camera, constitute the original set of pixel domain and spatial domain data corresponding to the RF probe and the target pad within the camera's field of view. This data forms the basis for all subsequent processing and calculations and specifically includes two types of core parameters, both stored in the form of pixel coordinates / original analog quantities, as follows: (a) Geometric parameters related to radio frequency probes The pixel contour coordinate set of the RF probe tip and the pixel grayscale value distribution at the tip edge; The pixel-fitted line segment coordinates of the RF probe axis and the original analog quantity of the spatial extension direction of the axis; Pixel coordinates of the center point of the RF probe tip, and pixel data of the outer contour of the RF probe rod; The original analog quantity is the spatial height of the RF probe tip within the field of view.

[0031] (ii) Geometric parameters related to the target pad The pixel outline coordinate set of the target pad (including the pixel coordinates of the four vertices of the rectangular pad); Pixel grayscale distribution data of the target pad surface, and pixel contrast data of the pad edge; The pixel coordinates of the center region of the target pad and the original analog quantity of the spatial normal of the pad surface; The set of pixel position coordinates of the target pad on the stage surface.

[0032] (III) Geometric parameters related to the two The relative pixel positions of the RF probe and the target pad in the acquisition field of view, the raw data of the pixel pitch between the two, and the vertical pixel height difference between the RF probe tip and the pad surface.

[0033] The aforementioned geometric parameters are continuously acquired by an industrial area scan camera and transmitted in real time to the vision processing unit, where they are cached and stored in the form of pixel data files, awaiting subsequent preprocessing operations.

[0034] Preprocessing of geometric parameters The vision processing unit retrieves the original color image data corresponding to the geometric parameters of the cached RF probe and target pad, and performs operations in sequence according to the steps of grayscale conversion → Gaussian filtering noise reduction → gradient double threshold edge extraction → morphological dilation and erosion combined optimization, finally generating optimized contour data with continuous contours and smooth edges. The specific operation process is as follows: Grayscale Conversion: The vision processing unit performs a weighted average grayscale conversion on the original color image data. According to the formula Gray=0.299R+0.587G+0.114B, the grayscale value of each pixel (grayscale value range 0-255) is calculated, and the 3-channel color pixel data is converted into single-channel grayscale pixel data. Color information interference is removed, and only the brightness contour features of the RF probe and the target pad are retained to generate grayscale image data. Gaussian filtering noise reduction: The vision processing unit calls a 3×3 Gaussian filter template to perform a two-dimensional convolution operation on the grayscale image data. The standard deviation threshold is set to 0.8. Pixels whose grayscale values ​​deviate from the threshold range (interference pixels caused by ambient light fluctuations, lens noise, and slight reflections from the stage) are smoothed to filter out environmental interference and random noise, and only retain the core contour grayscale information of the RF probe and the target pad to generate the noise-reduced core contour information data. Gradient dual-threshold edge extraction: The vision processing unit performs Sobel gradient operation on the core contour information data to extract the gradient value and gradient direction of each pixel. The gradient dual threshold is set to a high threshold of 200 and a low threshold of 80: pixels with gradient values ​​higher than the high threshold are directly identified as contour edge points, pixels with gradient values ​​between the high and low thresholds and connected to edge points are identified as contour edge points, and pixels with gradient values ​​lower than the low threshold are directly discarded. Only the precise contour edge pixels of the RF probe tip, axis, and target pad shape are extracted to generate target contour data (including a precise edge pixel coordinate set) containing only the RF probe and target pad. Morphological dilation and erosion combined optimization: The vision processing unit calls a 3×3 rectangular structuring element to perform combined operations on the target contour data in the order of dilation followed by erosion. The dilation operation fills the gaps caused by missing pixels on the contour edge with pixels, so that the contour edge of the RF probe and the target pad remains continuous and uninterrupted. The erosion operation removes pixels from the contour surface due to noise (burrs), eliminating irregular protrusions on the contour surface. After the combined operation, the distortion shape of the contour edge is accurately corrected, and finally, optimized contour data of RF probe-target pad with continuous contour, smooth edges, and no burrs or gaps is generated, completing all the preprocessing operations of geometric parameters. This data is stored in the vision processing unit and awaits coordinate system mapping.

[0035] Establishment of the RF probe-pad alignment space coordinate system Based on the preprocessed optimized contour data, the vision processing unit, combined with the known reference features of the precision stage and the calibration parameters of the standard calibration parts, constructs and calibrates the RF probe-pad alignment spatial coordinate system according to the steps of basic framework construction → precision calibration → parameter locking. This serves as the unified physical reference for all subsequent spatial position calculations. The specific operation process is as follows: The basic framework of the coordinate system is initially established: the vision processing unit sets the upper surface of the precision vacuum adsorption stage carrying the module under test as the horizontal reference plane, which is defined as the XY plane of the coordinate system; the pixel contours of the two preset positioning holes of the stage are extracted by image recognition, and the geometric center pixel coordinates of the two positioning holes are calculated by the least squares circle fitting algorithm. The midpoint of the line connecting the two geometric centers is taken as the origin O(0,0,0) of the coordinate system; the direction perpendicular to the surface of the stage and upward is set as the positive Z-axis (vertical reference direction) of the coordinate system; the direction passing through the origin and parallel to the length of the stage is set as the positive X-axis, and the direction passing through the origin and parallel to the width of the stage is set as the positive Y-axis. Following the right-hand coordinate system rule, the basic framework of the XYZ three-dimensional orthogonal RF probe-pad alignment space coordinate system is initially established. The high-precision standard calibration component is vacuum-adsorbed and fixed at the center of the stage. An industrial area array camera acquires an image of the feature point array on the surface of the calibration component. The vision processing unit extracts the pixel coordinates of the feature points and, combined with the known actual three-dimensional physical coordinates (unit: μm) of the calibration component, establishes a linear mapping relationship between the pixel coordinates and the actual physical coordinates. Through this mapping relationship, the accuracy of each reference axis (X, Y, Z axes) of the initially constructed coordinate system is calibrated. The scale coefficient of each axis is adjusted to 0.5 μm / pixel (to realize the conversion from pixel scale to physical scale). The offset error (X, Y axis offset ≤ 2 μm, Z axis offset ≤ 3 μm) and scaling deviation caused by hardware installation are precisely corrected to ensure the calculation accuracy of each axis of the coordinate system. Locking coordinate system parameters: After precision calibration, the vision processing unit permanently locks all parameters of the coordinate system, including the origin position, the direction of each reference axis, the axis scale coefficient, the pixel-physical coordinate transformation coefficient, and the coordinate calculation precision. The calibrated and locked three-dimensional orthogonal coordinate system is used as the RF probe-pad alignment spatial coordinate system. All parameters are stored in the vision processing unit to provide a unified reference for the subsequent spatial mapping and calculation of geometric parameters.

[0036] Generation of the three-dimensional relative positional relationship between the RF probe and the target pad The vision processing unit maps the pre-processed RF probe-target pad optimized contour data to the established RF probe-pad alignment spatial coordinate system. Through coordinate transformation, key spatial feature extraction, spatial vector operation, and data integration, it quantifies and generates a three-dimensional relative positional relationship including relative distance and relative angle. The specific operation process is as follows: Coordinate system space mapping of geometric parameters: The vision processing unit calls the locked RF probe-pad alignment space coordinate system parameters, and converts the preprocessed optimized contour data (pixel coordinate set) into actual three-dimensional physical coordinates through a conversion coefficient of 0.5um / pixel, thus completing the accurate mapping of the geometric parameters of the RF probe and the target pad from the pixel domain to the physical space domain, and obtaining the actual physical contour data of the two in this coordinate system. Extracting key spatial feature data: Based on the mapped actual physical contour data, the vision processing unit extracts key spatial feature data of the RF probe and the target pad through feature fitting algorithms: The three-dimensional physical coordinates P(x1,y1,z1) of the center point of the RF probe tip are calculated using the center fitting algorithm, and the spatial straight line equation of the RF probe axis is fitted using the least squares straight line fitting algorithm to obtain the spatial vector a of the RF probe axis direction; The three-dimensional physical coordinates Q(x2,y2,z2) of the center point of the target pad are calculated using the centroid method, and the spatial plane equation of the pad surface is fitted using the physical contour data of the pad surface to obtain the spatial vector b of the pad surface normal. Spatial vector operations calculate relative position parameters: The vision processing unit performs spatial vector operations on the extracted key spatial feature data to calculate the relative distance and relative angle parameters respectively. Calculate the distance difference in the three axes: X-axis distance difference ΔX=|x1-x2|, Y-axis distance difference ΔY=|y1-y2|, Z-axis distance difference ΔZ=|z1-z2|, where ΔZ is the actual vertical height difference between the RF probe tip and the target pad surface, which is the core basis for the subsequent movement of the RF probe to a safe height; Calculate the angle between the axis and the normal: Calculate the angle θ between the RF probe axis and the normal to the pad surface using the dot product formula for spatial vectors. The formula is: ,in, Let be the spatial vector of the RF probe axis direction. The spatial vector is the normal vector of the pad surface. Let be the magnitude of the spatial vector of the RF probe axis. The magnitude of the spatial vector normal to the pad surface is given; the specific value of the included angle θ is calculated. In this embodiment, θ is controlled to be ≤5° to ensure that the RF probe pressing direction is basically perpendicular to the pad surface. The specific value of the included angle θ is calculated. In this embodiment, θ is controlled to be ≤5° to ensure that the downward pressing direction of the RF probe is basically perpendicular to the surface of the pad. Generating three-dimensional relative positional relationships: The vision processing unit integrates the calculated X / Y / Z three-axis distance differences (ΔX, ΔY, ΔZ) with the angle θ between the RF probe axis and the pad surface normal to generate a three-dimensional relative positional relationship data file containing relative distances (ΔX, ΔY, ΔZ) and relative angles (θ). This file is transmitted in real time via Ethernet to the motion control unit of the multi-axis displacement mechanism, providing accurate physical positional data for the relative spatial movement of the RF probe and the target pad in the subsequent S2 step.

[0037] Specifically, the method for performing relative spatial movement is as follows: The three-dimensional relative position relationship between the RF probe and the target pad is retrieved. Based on the reference axis directions of the RF probe-pad alignment spatial coordinate system, the directional movement path of the RF probe is generated. The multi-axis displacement mechanism is controlled to drive the RF probe to move in space independently. The single-axis movement stroke is planned according to the relative distance of each reference axis. The movement operation is performed step by step along the horizontal and vertical reference directions, and the RF probe is controlled to move smoothly at a preset rate.

[0038] Specifically, the method for dynamically correcting the motion parameters of the multi-axis displacement mechanism is as follows: The system collects real-time positional deviation data between the RF probe and the target pad, extracts the deviation values ​​between the actual and theoretical displacements of the RF probe in each reference axis direction, compares the deviation values ​​with the preset alignment accuracy threshold, calculates the displacement compensation amount corresponding to each reference axis based on the comparison results, adjusts the step amount and movement speed of each reference axis of the multi-axis displacement mechanism according to the compensation amount, corrects the displacement deviation of the RF probe in each reference axis direction, and updates the motion parameters of the multi-axis displacement mechanism synchronously.

[0039] Specifically, the method for performing the step-by-step, graded, uniform downward pressure is as follows: Based on the preset safe height above the target pad where the RF probe is located, the RF probe pressing direction is determined along the vertical reference direction of the RF probe-pad alignment spatial coordinate system. The graded step spacing of the RF probe pressing is planned, and the multi-axis displacement mechanism is controlled to drive the RF probe to make uniform linear motion along the pressing direction. The graded pressing action is executed based on the preset pressing rules.

[0040] Specifically, the contact feature signal is compared with a preset contact determination threshold, and the specific method is as follows: Extract the continuously captured contact feature signal acquisition data, retrieve the reference data of the preset contact judgment threshold, compare the acquired data and the reference data in one direction according to the acquisition dimension of the contact feature signal, calculate the change of the contact feature signal acquisition data relative to the preset contact judgment threshold reference data, match it with the judgment range of the preset contact judgment threshold, generate the comparison calculation result according to the matching result, and make the contact feature signal compliance judgment based on the comparison calculation result. The contact feature signal is the displacement change of the RF probe tip relative to the reference position and the morphological distortion data of the RF probe.

[0041] Specifically, the method for performing the pressure reduction operation with a preset overpressure amount is as follows: Based on the contact characteristic signal compliance determination result, the pressure stroke parameters corresponding to the preset overpressure amount are planned, and the multi-axis displacement mechanism is controlled to drive the RF probe to move in a uniform linear motion along the original pressure direction. The pressure distance of the RF probe is controlled according to the planned pressure stroke parameters, and the multi-axis displacement mechanism is controlled to stop driving the RF probe to pressure down.

[0042] This embodiment is applied to the RF probe depressurization station in the automated testing of RF module RF performance. Building upon the RF probe alignment spatial coordinate system construction and three-dimensional relative position relationship generation results completed in Embodiment 1, it focuses on the precise control of the RF probe depressurization amount. The specific operation is as follows: This embodiment pre-determines the hardware components based on the previous embodiment, and completes collaborative debugging. The core component parameters are added / refined as follows: multi-axis displacement mechanism (X / Y / Z three-axis orthogonal structure, step accuracy 0.1um, Z-axis (downward axis) uniform motion speed adjustable range 1-10um / s), motion control unit (real-time communication with vision processing unit, response delay ≤5ms, supports precise control of step amount and speed); industrial area array camera maintains a frame rate of 30fps continuous acquisition, vision processing unit adds a tip displacement and shape recognition algorithm module, which can capture the tip pixel position change and contour distortion in real time; the RF probe tip is tested and found to be unworn, the RF module under test is fixed by vacuum adsorption, the angle θ between the target pad of the module and the axis of the RF probe is 3°, which meets the vertical downward pressure requirement, and the RF probe has been moved to a preset safety height of 300μm above the pad through step S2 (significantly higher than the conventional setting, to avoid the risk of collision caused by the ±100μm height tolerance of the module).

[0043] Core parameter definition Graded step spacing: Combining module tolerance and RF probe accuracy, the single-step pressure spacing is set to 5um, and the pause after each step is 100ms to reserve time for visual recognition. Uniform pressing rate: The Z-axis pressing rate is set to 5um / s to avoid misjudgment due to needle tip deformation caused by excessive speed; Contact feature signal: defined as the X / Y axis crawling displacement data (accuracy 0.1um) of the RF probe tip relative to the reference position extracted by the vision processing unit and the tip contour distortion data (grayscale contour offset). Preset contact judgment threshold: needle tip crawling displacement ≥1um, or needle tip shape distortion grayscale offset ≥5 pixels, either condition is met to be judged as qualified; Preset overvoltage: Based on the characteristics of the RF probe, the overvoltage is set to 30um (≤80um safety threshold), and the corresponding downward stroke parameter is 30um.

[0044] Step-by-step, uniform downward pressing operation The motion control unit retrieves the stored RF probe-pad alignment spatial coordinate system parameters and safety height data, and performs graded uniform downward pressing according to preset rules. The vision processing unit is linked throughout the process to complete the probe tip status monitoring. The specific operation is as follows: The motion control unit locks the current safe height of the RF probe (Z-axis coordinate Z0=200um), determines the negative direction of the Z-axis (vertical downward) as the only downward pressing direction, and plans the total pressing stroke to the estimated contact position based on the preset 5um single-step spacing (Z-axis coordinate Z1=100um, reserving sufficient testing space to adapt to module tolerance). The motion control unit sends a command to the Z-axis driver of the multi-axis displacement mechanism to control the RF probe to perform the first step of pressing down along the negative Z-axis at a constant speed of 5um / s, moving from Z0=200um to Z=195um. After completing the 5um step interval, the multi-axis displacement mechanism is controlled to stop pressing down and stay for 100ms. During the dwell time, the industrial area array camera acquires real-time images of the needle tip and transmits them to the vision processing unit. The vision processing unit compares the current pixel position of the needle tip with the reference pixel position at the safe height, extracts the needle tip displacement data and shape contour data, determines that the needle tip has no crawling and no shape distortion, and feeds back a "can continue to press down" signal. After receiving the feedback signal, the motion control unit repeats the above step-down process. Each step is executed according to the logic of "5um constant speed down → pause for 100ms → visual recognition → feedback judgment". The step-down operation from Z=195um to Z=105um is completed in sequence, with a total of 19 steps and a total down pressure of 95um. The single-step spacing and speed remain constant at 5um / s throughout the process. By combining graded stepping and dwell recognition, the pressure is segmented and controllable. At the same time, it is compatible with the module's ±50um tolerance to avoid pin collision or contact misjudgment caused by one-time pressure.

[0045] Contact feature signal comparison calculation and compliance determination As the stepped pressure continues to be executed, when the RF probe tip approaches the target pad surface, the vision processing unit captures the contact feature signal in real time, compares it with a preset threshold, and performs a compliance determination. The specific operation is as follows: When the RF probe performs the 20th step of pressing down, it moves from Z=105um to Z=100um and then stops. The industrial area array camera quickly acquires 3 frames of tip images (interval 33ms) and transmits them to the vision processing unit. The vision processing unit extracts the core feature data of the tip from the 3 frames of images, removes image noise interference, and retains the valid acquisition data. The vision processing unit retrieves the preset contact judgment threshold benchmark data (crawling displacement 1µm, morphological distortion grayscale offset 5 pixels) and compares them according to the acquisition dimensions (displacement dimension, morphological dimension): Displacement dimension: The current X / Y axis pixel position of the needle tip is compared with the reference position (safe height). The pixel-physical conversion coefficient (0.5um / pixel) is used to calculate the needle tip X-axis crawling displacement of 0.8um and Y-axis crawling displacement of 0.3um, both of which do not reach the 1um threshold. Morphological dimension: Comparing the current grayscale distribution of the needle tip contour with the baseline contour, the grayscale offset is calculated to be 3 pixels, which does not reach the 5-pixel threshold; The vision processing unit generates a comparison calculation result, determines that the contact feature signal does not meet the standard, and feeds back to the motion control unit to continue pressing down. The RF probe performs step 21, pressing down from Z=100um to Z=95um. During the dwell period, the camera acquires images, and the vision processing unit extracts the data and compares them: the tip's X-axis crawling displacement is 1.2um, which exceeds the preset 1um threshold, and the morphological distortion grayscale offset is 4 pixels. The vision processing unit matches the displacement change with the preset judgment range, generates a comparison calculation result of "contact feature signal meets the standard", simultaneously extracts the needle tip crawling trajectory data, confirms that the needle tip crawls along the pad surface (non-random offset), and quickly transmits the standard judgment result to the motion control unit to trigger the pressing action switching.

[0046] During this process, the relative position of the needle tip and the camera remains unchanged before contact, and the needle tip is always in a stationary state as captured by vision; after contact, the needle tip crawls along the surface due to the resistance of the solder pad, and the vision processing unit captures this change through precise data comparison, completely abandoning the fixed height judgment logic and adapting to the position deviation caused by module tolerance.

[0047] Preset overpressure amount pressing operation After receiving the contact characteristic signal and determining whether it meets the standard, the motion control unit immediately switches to the overvoltage pressing mode and precisely controls the pressing distance according to the preset stroke parameters to ensure reliable contact between the RF probe and the pad. The specific operation is as follows: After receiving the compliance signal, the motion control unit immediately controls the multi-axis displacement mechanism to stop the graded step-down pressing action, and simultaneously retrieves the preset overpressure parameter (30um). Combined with the current radio frequency probe Z-axis coordinate (Z=95um), the overpressure pressing stroke parameters are planned: continue pressing down 30um along the original negative Z-axis direction, with a target coordinate Z=65um. The motion control unit sends a continuous pressing command to the multi-axis displacement mechanism, cancels the graded pause logic, and controls the RF probe to move continuously along the original pressing direction at a uniform speed of 5um / s without pause, thus avoiding contact instability caused by overpressure interruption. During the pressing process, the industrial area array camera continuously acquires data, and the vision processing unit monitors the tip shape in real time to confirm that the tip is not excessively deformed (the grayscale offset of shape distortion is ≤8 pixels, which is within the safe range). The data is then fed back to the motion control unit to ensure that the overpressure process is compliant. When the Z-axis encoder of the multi-axis displacement mechanism detects that the RF probe has moved to the target coordinate Z=65um (i.e., completed the 30um overpressure stroke), the motion control unit immediately sends a stop command to control the multi-axis displacement mechanism to stop driving the RF probe to press down and lock the current position of the RF probe. The vision processing unit re-acquires the tip image to confirm that the tip is tightly attached to the pad surface without loosening or excessive compression, and the overpressure is precisely controlled at 30um, meeting the technical requirement of ≤80um overpressure for RF probes, while also adapting to module tolerances to achieve reliable contact.

[0048] This embodiment successfully achieved overpressure control of only 30μm within a module height tolerance of ±100μm through a full-process operation involving graded step-by-step probing, visual recognition of probe tip crawling / deformation, and overpressure compensation. This ensures contact reliability and consistency, and enables dynamic and precise control of the RF probe's pressure. Step-by-step operation avoids the risk of probe collision, visual feedback judgment solves the fixed height failure problem caused by the module's ±50μm tolerance, and the 30μm overpressure ensures contact reliability. Furthermore, the data accuracy throughout the process is ≤0.1μm, fully meeting the high-precision requirements of RF module testing. The entire process is free of redundant actions, with rapid collaborative responses from all hardware components. The total time from contact determination to overpressure completion is ≤200ms, meeting the efficiency requirements of mass production testing and providing a stable contact foundation for subsequent torque adjustment and micro-displacement correction in step S4.

[0049] Specifically, the method for generating the deviation value between the current overpressure and the standard overpressure is as follows: The reference deformation parameters corresponding to the preset standard overpressure are retrieved. Based on the deformation detection dimension, the corresponding values ​​of the contact feature signal acquisition data and the reference deformation parameters are extracted, and the difference operation is performed to calculate the numerical difference results of each dimension. Combining the numerical difference results of each dimension, the deviation value between the current overpressure and the standard overpressure is generated.

[0050] Specifically, the method for adjusting the output torque of the multi-axis displacement mechanism is as follows: Based on the deviation between the current overpressure and the standard overpressure, and combined with the torque adjustment coefficient of each reference axis of the multi-axis displacement mechanism, the torque adjustment amount is calculated axis by axis. The torque adjustment operation is performed on each reference axis of the multi-axis displacement mechanism, and the matching degree of the torque output of each reference axis is calibrated simultaneously. Based on the calibration results, the torque output value of each axis is finely adjusted.

[0051] Specifically, the output torque and the minute displacement of the RF probe are determined by the following method: Based on the adjusted output torque of the multi-axis displacement mechanism and combined with the micro-displacement adjustment requirements of the RF probe, the micro-displacement parameters of the RF probe in each reference axis direction are planned. The multi-axis displacement mechanism is controlled to run according to the adjusted output torque, which drives the RF probe to perform micro-displacement along each reference axis direction according to the planned displacement parameters. The actual output torque data of the multi-axis displacement mechanism and the actual micro-displacement data of the RF probe in each reference axis direction are recorded simultaneously to form the corresponding data of torque and micro-displacement.

[0052] In this embodiment, based on the pressure calibration station after the RF probe contacts the RF module test, following the preset overpressure amount pressure operation results of Embodiment 2, the multi-axis displacement mechanism output torque adjustment and RF probe micro-displacement correction operations are performed to address the deviation between the current overpressure amount and the standard overpressure amount caused by the ±100μm installation tolerance of the module under test. This ensures that stable overpressure control at the 30–80μm level can still be achieved on modules with large height fluctuations.

[0053] In this embodiment, the preset standard overpressure is 30µm. The deviation value (ΔZ) between the current overpressure and the standard overpressure is calculated by the vision processing unit. The multi-axis displacement mechanism is equipped with independent torque sensors and adjustment modules for the X / Y / Z axes. The torque adjustment coefficient is a preset calibration value (Z axis is the main pressure axis, adjustment coefficient Kz=0.02N / µm, X / Y axes are auxiliary alignment axes, adjustment coefficient Kx=Ky=0.005N / µm). The micro-displacement adjustment accuracy is 0.1µm. All parameters are pre-stored in the motion control unit.

[0054] I. Operation of output torque adjustment for multi-axis displacement mechanism The motion control unit retrieves the overpressure deviation value and triaxial torque adjustment coefficient transmitted by the vision processing unit, and performs torque adjustment according to the core steps. The specific operation is as follows: The motion control unit extracts the deviation value ΔZ between the current overpressure and the standard overpressure, and calculates the torque adjustment amount axis by axis by axis in combination with the torque adjustment coefficients of each reference axis of the multi-axis displacement mechanism X / Y / Z. The calculation formula is: Torque adjustment amount = deviation value × corresponding axis adjustment coefficient, where the Z axis is the core axis of torque adjustment, and the X / Y axes are adjusted according to the needle tip alignment deviation. The motion control unit sends adjustment commands to the triaxial torque adjustment module of the multi-axis displacement mechanism. According to the calculated torque adjustment amount, it independently performs torque output adjustment operations on each reference axis and initially sets the triaxial torque output value. After torque adjustment is completed, the motion control unit initiates triaxial torque matching calibration. Real-time torque data is collected through torque sensors on each axis to determine the degree of coordination and matching of the three-axis torque output. If there is a torque output imbalance, the torque output value of each axis is finely adjusted based on the calibration results to ensure that the torque output is adapted to the actual overpressure deviation. After fine-tuning, the final torque output value of each reference axis of the multi-axis displacement mechanism is locked, and all torque adjustment operations are completed, providing stable torque power for subsequent micro-displacement correction of the RF probe.

[0055] II. Torque and RF probe micro-displacement output operation Based on the adjusted torque output value, the motion control unit, in conjunction with the micro-displacement adjustment requirements of the RF probe, plans the micro-displacement parameters and performs displacement correction, simultaneously recording and outputting the torque-displacement corresponding data. The specific operation is as follows: Based on the adjusted final torque output value and the contact state between the probe tip and the pad, the motion control unit plans the micro-displacement parameters of the RF probe in each reference axis direction of X / Y / Z. The displacement is controlled within 5um to avoid large displacement from damaging the existing contact state. The motion control unit sends a running command to the multi-axis displacement mechanism. The control mechanism runs at a constant speed according to the adjusted torque output value, driving the radio frequency probe along each reference axis direction to perform micro-displacement correction according to the planned displacement parameters, thereby achieving precise calibration of the contact pressure. During the micro-displacement process, the motion control unit synchronously collects and records the actual output torque data of each axis and the actual micro-displacement data of the RF probe through the position sensor and torque sensor of the multi-axis displacement mechanism, and the acquisition accuracy is consistent with the hardware accuracy of the device; After the micro-displacement correction is completed, the motion control unit will integrate the actual torque data and the actual micro-displacement data to form corresponding data of torque and micro-displacement. This data is transmitted to the equipment data storage module in real time to complete the data output, providing data support for torque parameter calibration and deviation correction in subsequent mass production testing.

[0056] This embodiment accurately compensates for overpressure deviations caused by module tolerances through coordinated operation of axial torque adjustment, matching degree calibration, and micro-displacement, ensuring stable contact pressure between the RF probe and the pad; at the same time, the output torque-displacement corresponding data realizes the traceability of the test process and adapts to the high-precision test requirements of RF modules.

[0057] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for probe alignment and control of the amount of pressure based on visual feedback and motion control, characterized by, include: S1: Obtain the geometric parameters of the RF probe and the target pad of the RF module under test and perform preprocessing operations. Establish the RF probe-pad alignment space coordinate system and generate the three-dimensional relative position relationship between the RF probe and the target pad through coordinate calculation. S2: Drive the multi-axis displacement mechanism to move the RF probe relative to the stage carrying the module under test in space; collect the real-time position deviation data between the RF probe and the target pad, dynamically correct the motion parameters of the multi-axis displacement mechanism, and move the RF probe to a preset safe height above the target pad; S3: Drive the multi-axis displacement mechanism to drive the RF probe to perform step-by-step uniform speed pressing along the preset pressing direction; The system continuously captures the contact characteristic signals between the RF probe and the target pad, and compares them with a preset contact judgment threshold. Based on the contact characteristic signals that reach the preset contact judgment threshold, the system controls the multi-axis displacement mechanism to stop pressing down on the RF probe, and drives the multi-axis displacement mechanism to press down on the RF probe with a preset overpressure amount. S4: Call the contact characteristic signal of the RF probe and the reference deformation parameter corresponding to the preset standard overpressure to perform difference calculation, generate the deviation value between the current overpressure and the standard overpressure, adjust the output torque of the multi-axis displacement mechanism based on the deviation value, drive the multi-axis displacement mechanism to drive the RF probe to perform micro-displacement correction, and output torque and RF probe micro-displacement amount.

2. The method of claim 1, wherein, The specific method for performing the preprocessing operation is as follows: The original data corresponding to the acquired geometric parameters are converted to grayscale and denoised using a Gaussian filter template. Environmental interference and random noise are filtered out by setting a standard deviation threshold to generate the core contour information of the RF probe and the target pad. The contour edges of the RF probe and the target pad are extracted by setting a gradient double threshold to generate target contour data containing only the RF probe and the target pad. A combination algorithm of morphological dilation and erosion is used for contour optimization. The dilation operation fills the gaps between the contour edges, and the erosion operation removes the burrs on the contour surface and corrects the distortion of the contour edges.

3. The method of claim 1, wherein, The specific method for establishing the RF probe-pad alignment spatial coordinate system is as follows: The upper surface of the stage carrying the module under test is set as the horizontal reference plane. The geometric center of the preset positioning hole of the stage is selected as the origin of the coordinate system. The direction perpendicular to the surface of the stage and upward is set as the vertical reference direction. The basic framework of the coordinate system is initially built. The accuracy of each reference axis of the coordinate system is calibrated by standard calibration parts. The scale coefficient of each axis is adjusted to correct the offset error and scaling deviation of the coordinate axis. The parameters of the coordinate system are locked. The coordinate system is used as the RF probe-pad alignment space coordinate system.

4. The method of claim 1, wherein, The specific method for generating the three-dimensional relative positional relationship between the RF probe and the target pad is as follows: The geometric parameters of the preprocessed RF probe and target pad are mapped to the RF probe-pad alignment space coordinate system. Key spatial feature data of the RF probe and target pad are extracted. Through spatial vector operation, the distance difference between the center point of the RF probe tip and the center point of the target pad in each reference axis direction is calculated. The angle between the RF probe axis direction and the normal of the pad surface is obtained. Combining the calculated distance difference and angle data, a three-dimensional relative position relationship including relative distance and relative angle is generated.

5. The method of claim 1, wherein, The specific method for performing relative spatial movement is as follows: The three-dimensional relative position relationship between the RF probe and the target pad is retrieved. Based on the reference axis directions of the RF probe-pad alignment spatial coordinate system, the directional movement path of the RF probe is generated. The multi-axis displacement mechanism is controlled to drive the RF probe to move in space independently. The single-axis movement stroke is planned according to the relative distance of each reference axis. The movement operation is performed step by step along the horizontal and vertical reference directions, and the RF probe is controlled to move smoothly at a preset rate.

6. The method of claim 1, wherein, The specific method for dynamically correcting the motion parameters of the multi-axis displacement mechanism is as follows: The system collects real-time positional deviation data between the RF probe and the target pad, extracts the deviation values ​​between the actual and theoretical displacements of the RF probe in each reference axis direction, compares the deviation values ​​with the preset alignment accuracy threshold, calculates the displacement compensation amount corresponding to each reference axis based on the comparison results, adjusts the step amount and movement speed of each reference axis of the multi-axis displacement mechanism according to the compensation amount, corrects the displacement deviation of the RF probe in each reference axis direction, and updates the motion parameters of the multi-axis displacement mechanism synchronously.

7. The method of claim 1, wherein, The specific method for performing the step-by-step, graded, uniform downward pressure is as follows: Based on the preset safe height above the target pad where the RF probe is located, the RF probe pressing direction is determined along the vertical reference direction of the RF probe-pad alignment spatial coordinate system. The graded step spacing of the RF probe pressing is planned, and the multi-axis displacement mechanism is controlled to drive the RF probe to make uniform linear motion along the pressing direction. The graded pressing action is executed based on the preset pressing rules.

8. The method of claim 1, wherein, The contact feature signal is compared with a preset contact determination threshold. The specific method is as follows: Extract the continuously captured contact feature signal acquisition data, retrieve the benchmark data of the preset contact judgment threshold, compare the acquisition data and the benchmark data in each direction according to the acquisition dimension of the contact feature signal, calculate the change of the contact feature signal acquisition data relative to the benchmark data of the preset contact judgment threshold, match it with the judgment range of the preset contact judgment threshold, generate the comparison calculation result according to the matching result, and make the contact feature signal compliance judgment based on the comparison calculation result.

9. The method of claim 1, wherein, The specific method for performing the pressure reduction operation with the preset overpressure amount is as follows: Based on the contact characteristic signal compliance determination result, the pressure stroke parameters corresponding to the preset overpressure amount are planned, and the multi-axis displacement mechanism is controlled to drive the RF probe to move in a uniform linear motion along the original pressure direction. The pressure distance of the RF probe is controlled according to the planned pressure stroke parameters, and the multi-axis displacement mechanism is controlled to stop driving the RF probe to pressure down.

10. The method of claim 1, wherein, The specific method for generating the deviation value between the current overpressure and the standard overpressure is as follows: The reference deformation parameters corresponding to the preset standard overpressure are retrieved. Based on the deformation detection dimension, the corresponding values ​​of the contact feature signal acquisition data and the reference deformation parameters are extracted, and the difference operation is performed to calculate the numerical difference results of each dimension. Combining the numerical difference results of each dimension, the deviation value between the current overpressure and the standard overpressure is generated.

11. The method of claim 1, wherein, The specific method for adjusting the output torque of the multi-axis displacement mechanism is as follows: Based on the deviation between the current overpressure and the standard overpressure, and combined with the torque adjustment coefficient of each reference axis of the multi-axis displacement mechanism, the torque adjustment amount is calculated axis by axis. The torque adjustment operation is performed on each reference axis of the multi-axis displacement mechanism, and the matching degree of the torque output of each reference axis is calibrated simultaneously. Based on the calibration results, the torque output value of each axis is finely adjusted.

12. The method according to claim 1, characterized in that, The specific method for determining the output torque and the minute displacement of the RF probe is as follows: Based on the adjusted output torque of the multi-axis displacement mechanism and combined with the micro-displacement adjustment requirements of the RF probe, the micro-displacement parameters of the RF probe in each reference axis direction are planned. The multi-axis displacement mechanism is controlled to run according to the adjusted output torque, which drives the RF probe to perform micro-displacement along each reference axis direction according to the planned displacement parameters. The actual output torque data of the multi-axis displacement mechanism and the actual micro-displacement data of the RF probe in each reference axis direction are recorded simultaneously to form the corresponding data of torque and micro-displacement.

Images