Probe calibration method and probe control device

Abstract

The application relates to the technical field of probe calibration, and discloses a probe calibration method and a probe control device.The method comprises the following steps: controlling a three-axis movement of a probe table and synchronously collecting laser interferometer interference fringe variation numbers and encoder displacement readings, and calculating a three-dimensional error compensation lookup table; collecting Mark point pixel coordinates on a standard calibration sheet image and performing affine transformation to obtain a coordinate transformation matrix; controlling a plurality of probes in a probe array to perform position deviation fitting on a calibration point according to the coordinate transformation matrix to obtain a system error function; querying the three-dimensional error compensation lookup table according to a target physical coordinate to obtain an encoder system deviation, and substituting the target physical coordinate into the system error function to obtain a probe geometric deviation; superimposing the encoder system deviation and the probe geometric deviation on the target physical coordinate to obtain a compensated control coordinate, and sending the compensated control coordinate to a motion controller to drive the probe table to move to a target position, thereby completing probe positioning and improving the probe positioning precision.

Description

Technical Field

[0001] This application relates to the field of probe calibration technology, and in particular to a probe calibration method and a probe control device. Background Technology

[0002] MEMS probe arrays play a crucial role in the electrical contact function of semiconductor wafer testing. As chip manufacturing processes continue to shrink, probe pitch has reached 55 micrometers or even smaller, placing stringent requirements on probe positioning accuracy at the sub-micrometer level. Existing probe stage positioning technology uses internal grating encoders for position feedback. However, during long-term use, the encoder is affected by guide rail wear, temperature changes, and mechanical transmission errors, resulting in accumulated errors of up to ±5 micrometers, which is far from meeting the positioning requirements of micro-pitch probes. More seriously, the probe array itself has installation errors, uneven tip spacing, and geometric deviations caused by thermal expansion. The combination of these factors makes it difficult for traditional single-error-source compensation methods to achieve high-precision positioning. Summary of the Invention

[0003] This application provides a probe calibration method and a probe control device, thereby improving the probe positioning accuracy.

[0004] A first aspect of this application provides a probe calibration method, the probe calibration method comprising: The probe stage is controlled to move in three axes and simultaneously acquires the number of changes in interference fringes of the laser interferometer and the encoder displacement reading, and a three-dimensional error compensation lookup table is calculated. The pixel coordinates of the Mark points on the standard calibration sheet image are acquired and subjected to affine transformation to obtain the coordinate transformation matrix; Based on the coordinate transformation matrix, multiple probes in the probe array are controlled to perform position deviation fitting with the calibration point to obtain the system error function; The encoder system deviation is obtained by querying the three-dimensional error compensation lookup table based on the target physical coordinates, and the probe geometric deviation is obtained by substituting the target physical coordinates into the system error function. The encoder system deviation and the probe geometric deviation are superimposed on the target physical coordinates to obtain the compensated control coordinates. The compensated control coordinates are then sent to the motion controller to drive the probe station to move to the target position, thus completing the probe positioning.

[0005] In conjunction with the first aspect, in a first implementation of the first aspect of the present invention, the control probe station moves along three axes and synchronously acquires the number of changes in interference fringes of the laser interferometer and the encoder displacement reading, and calculates a three-dimensional error compensation lookup table, including: The three axes of the drive probe station move uniformly from the initial position to the target position and synchronously record the number of interference fringe changes and encoder displacement readings of each axis at preset spatial intervals. The actual physical displacement of each axis is obtained by calculating point by point based on the number of changes in the interference fringes and the laser wavelength. The encoder displacement of each axis is obtained by subtracting the initial encoder reading from the encoder displacement reading. A three-dimensional error compensation lookup table is calculated for each spatial position based on the actual physical displacement and the encoder displacement.

[0006] In conjunction with the first aspect, in a second implementation of the first aspect of the present invention, the step of calculating the three-dimensional error compensation lookup table for each spatial position based on the actual physical displacement and the encoder displacement includes: Subtract the actual physical displacement of each measurement point from the encoder displacement of the corresponding measurement point to obtain the X-axis displacement deviation sequence, Y-axis displacement deviation sequence, and Z-axis displacement deviation sequence. Based on the spatial positions of each measurement point in the X-axis displacement deviation sequence, the Y-axis displacement deviation sequence, and the Z-axis displacement deviation sequence, the triaxial deviation data are correlated to form a three-dimensional error compensation lookup table.

[0007] In conjunction with the first aspect, in a third implementation of the first aspect of the present invention, the step of acquiring the pixel coordinates of the Mark points on the standard calibration sheet image and performing an affine transformation to obtain a coordinate transformation matrix includes: The probe array image is segmented by grayscale threshold and the centroid method is calculated to obtain the pixel coordinates of the probe tip. The standard calibration plate image is then subjected to circle detection to obtain the pixel coordinates of the Mark point. Substitute the known physical coordinates of each Mark point and the corresponding pixel coordinates of the Mark point into the affine transformation equation to form a system of linear equations. Represent the system of linear equations in matrix form and multiply the coefficient matrix by the constant matrix to obtain the affine transformation parameters. The affine transformation parameters are arranged in homogeneous coordinate form to form a coordinate transformation matrix.

[0008] In conjunction with the first aspect, in a fourth implementation of the first aspect of the present invention, the step of controlling multiple probes in the probe array to perform position deviation fitting with the calibration point according to the coordinate transformation matrix to obtain the system error function includes: Multiple probes in the probe array are selected, and the pixel coordinates of each probe are converted into physical coordinates through the coordinate transformation matrix to calculate the target position. Then, the probe stage is driven to move, and the Z-axis descends so that the probe tip contacts the surface of the calibration piece. The contact is determined by the contact resistance. The position deviation data of each probe is obtained by subtracting the actual position reached by the probe stage from the target position. A two-dimensional quadratic polynomial error function is established, which takes spatial coordinates as input variables and outputs the predicted value of position deviation. The two-dimensional quadratic polynomial error function includes a constant term, a coordinate linear term, a coordinate cross product term, and a coordinate quadratic term. The positional deviation data is used as sample points and substituted into the two-dimensional quadratic polynomial error function to construct a least-squares optimization equation and solve for the polynomial coefficients to obtain the system error function.

[0009] In conjunction with the first aspect, in the fifth implementation of the first aspect of the present invention, the step of substituting the position deviation data as sample points into the two-dimensional quadratic polynomial error function to construct a least-squares optimization equation and solving for the polynomial coefficients to obtain the system error function includes: The positional deviation data of each probe and its corresponding spatial coordinates are used as a set of sample points and substituted into the two-dimensional quadratic polynomial error function to form a prediction deviation expression containing polynomial coefficients. The difference between the actual position deviation of each sample point and the calculated value of the predicted deviation expression is calculated, and the squares of the difference are summed to construct a least squares optimization objective function with polynomial coefficients as the solution variables. The system error function is obtained by solving the polynomial coefficients of the least squares optimization objective function through matrix pseudo-inverse operation.

[0010] In conjunction with the first aspect, in the sixth implementation of the first aspect of the present invention, the step of querying the three-dimensional error compensation lookup table based on the target physical coordinates to obtain the encoder system deviation, and substituting the target physical coordinates into the system error function to obtain the probe geometric deviation, includes: Based on the target physical coordinates, find the M adjacent grid nodes surrounding the target physical coordinates and their corresponding three-dimensional position deviation vectors in the three-dimensional error compensation lookup table, and calculate the normalized coordinates of the target physical coordinates in the grid cells. The three-dimensional position deviation vectors of the M adjacent grid nodes are linearly interpolated along the three coordinate axes according to the normalized coordinates to obtain the encoder system deviation corresponding to the target physical coordinates; Substitute the target physical coordinates into the system error function for polynomial calculation to obtain the probe geometric deviation corresponding to the target physical coordinates.

[0011] In conjunction with the first aspect, in the seventh implementation of the first aspect of the present invention, the step of searching for M adjacent grid nodes surrounding the target physical coordinates and their corresponding three-dimensional position deviation vectors in the three-dimensional error compensation lookup table based on the target physical coordinates, and calculating the normalized coordinates of the target physical coordinates within the grid cells, includes: Based on the target physical coordinates, locate the grid cell in the three-dimensional error compensation lookup table and determine the spatial coordinate positions of the M adjacent grid nodes surrounding the target physical coordinates; Read the three-dimensional position deviation vectors stored in the M adjacent grid nodes to obtain the encoder deviation components corresponding to each grid node; The normalized coordinates are obtained by subtracting the lower boundary coordinates of the corresponding grid nodes from each axis component of the target physical coordinates and then dividing by the grid spacing.

[0012] In conjunction with the first aspect, in the eighth implementation of the first aspect of the present invention, the step of superimposing the encoder system deviation and the probe geometric deviation onto the target physical coordinates to obtain compensated control coordinates, and sending the compensated control coordinates to the motion controller to drive the probe station to move to the target position, thereby completing probe positioning, includes: The encoder system deviation and the probe geometric deviation are respectively superimposed on the target physical coordinates to obtain the compensated control coordinates, and the compensated control coordinates are sent to the motion controller as position control commands; The motion controller drives the probe station's X-axis motor, Y-axis motor, and Z-axis motor to move at a set speed and acceleration according to the received position control command, thereby moving the probe station to the target position corresponding to the compensated control coordinates and completing the probe positioning.

[0013] A second aspect of this application provides a probe control device, the probe control device comprising: The acquisition module is used to control the three-axis movement of the probe stage and synchronously acquire the number of changes in the interference fringes of the laser interferometer and the encoder displacement reading, and calculate the three-dimensional error compensation lookup table; The affine transformation module is used to acquire the pixel coordinates of Mark points on the standard calibration sheet image and perform affine transformation to obtain the coordinate transformation matrix. The fitting module is used to control multiple probes in the probe array to perform position deviation fitting on the calibration point according to the coordinate transformation matrix, so as to obtain the system error function; The calculation module is used to query the three-dimensional error compensation lookup table based on the target physical coordinates to obtain the encoder system deviation, and substitute the target physical coordinates into the system error function to obtain the probe geometric deviation; The drive module is used to superimpose the encoder system deviation and the probe geometric deviation onto the target physical coordinates to obtain the compensated control coordinates, and send the compensated control coordinates to the motion controller to drive the probe station to move to the target position and complete the probe positioning.

[0014] Compared with existing technologies, this application has the following advantages: The laser interferometer, as an independent nanometer-level displacement measurement benchmark, establishes a three-dimensional error compensation lookup table through systematic mapping of the entire workspace, completely eliminating the systematic deviation caused by mechanical accumulation in traditional encoders, and providing a traceable physical basis for subsequent compensation. Based on the affine transformation matrix of the Mark points on the standard calibration plate, a precise mapping relationship between the visual pixel coordinate system and the probe stage physical coordinate system is achieved, overcoming the limitations of traditional manual teaching or simple linear mapping in describing complex geometric relationships. Through multi-point contact touch testing combined with two-dimensional quadratic polynomial fitting, a continuous function model of the probe array spatial error is established, capable of predicting probe geometric deviations at arbitrary positions, overcoming the shortcomings of existing technologies that assume uniform error distribution. The encoder system deviation obtained by laser interferometry and the probe geometric deviation obtained by contact testing are algebraically superimposed to form a multi-source error fusion compensation strategy, achieving error elimination across the entire link from the measurement system to the actuator, improving the probe positioning accuracy from ±5 micrometers to ±0.5 micrometers, meeting the high-precision testing requirements of 55-micrometer micro-pitch MEMS probes. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] The structures, proportions, sizes, etc., shown in the accompanying drawings of this specification are only for the purpose of assisting those skilled in the art in understanding and reading the content disclosed in the specification, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0017] Figure 1 This is a schematic flowchart of the probe calibration method provided in an embodiment of the present invention; Figure 2 This is a schematic block diagram of the probe control device provided in an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0020] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0021] It should also be further understood that the term "and / or" as used in this application specification and the appended claims refers to any combination of one or more of the associated listed items, as well as all possible combinations, and includes such combinations. See also Figure 1 One embodiment of the probe calibration method in this application includes: Step 100: Control the probe stage to move along three axes and synchronously collect the number of changes in the interference fringes of the laser interferometer and the encoder displacement readings, and calculate the three-dimensional error compensation lookup table; Specifically, the probe stage's X, Y, and Z axes are set to their respective initial positions. The control system moves the probe stage at a constant speed along each of the three axes to its predetermined target position. A fixed spatial interval is set as the sampling trigger condition, so that at each preset distance increment (e.g., 1 mm for X and Y axes, 0.5 mm for Z axis), the total number of interference fringe changes of the laser interferometer and the displacement reading of the corresponding three-axis encoder are automatically recorded. The actual physical displacement at each sampling position is calculated point-by-point by multiplying the total number of interference fringe changes by half the laser wavelength (λ / 2, where the laser wavelength λ is 632.8 nm). This physical displacement is unaffected by encoder mechanical errors and hysteresis. The encoder displacement at each sampling point is subtracted from the initial encoder value to obtain the encoder displacement relative to the position change. Next, the actual physical displacement of each corresponding sampling point is compared with the encoder displacement to calculate the position error in the three axes. The difference is the system error vector of the current position. The system error vector is then organized and recorded according to three-dimensional spatial coordinates to construct a three-dimensional error compensation lookup table containing spatially distributed error data. The three-dimensional error compensation lookup table is represented in the form of a three-dimensional grid, where each grid node stores a position error vector [δx, δy, δz], thereby describing the encoder system error distribution within the entire working space of the probe station.

[0022] Step 200: Acquire the pixel coordinates of the Mark points on the standard calibration sheet image and perform an affine transformation to obtain the coordinate transformation matrix; Specifically, a top-view image of the probe array is acquired using a high-resolution industrial CCD camera. The image is an 8-bit grayscale image with a resolution of no less than 2048×2048 pixels and a pixel size of 3.45μm, ensuring that the lens optical axis is perpendicular to the XY plane of the probe stage. After image acquisition, grayscale thresholding is performed, with an upper limit of 180. Regions with pixel grayscale values ​​higher than the threshold are identified as probe tip regions, while those lower are treated as background and removed. Erosion is then used to remove edge burrs and noise, enhancing the clarity of the probe tip boundaries. Within each identified probe tip region, a grayscale weighted centroid algorithm is applied to extract the probe's center coordinates. By calculating the weighted average of all pixel grayscale values ​​and their positions, sub-pixel-level positioning accuracy is achieved. A standard calibration piece is placed on the probe stage's working surface. The calibration piece has a quartz glass substrate and its surface contains four circular mark points, each 100 μm in diameter. Their physical coordinates have been calibrated and fixed using high-precision measuring equipment to (0 mm, 0 mm), (50 mm, 0 mm), (50 mm, 50 mm), and (0 mm, 50 mm). The system controls the probe stage to move, allowing the camera to acquire an image field of view containing all the mark points. A circle detection operation is performed on the image, and the pixel center coordinates of the four circular mark points are accurately extracted using the edge gradient method or Hough transform method. A one-to-one mapping relationship between pixels and their physical positions is established with the corresponding four known physical coordinates. Substituting the four sets of coordinate pairs into the affine transformation equations constructs six equations, which express the mapping relationship between two-dimensional pixel coordinates and two-dimensional physical coordinates as a linear combination. The linear equations are rewritten as matrix expressions, where the coefficient matrix consists of pixel coordinates and the constant matrix represents the corresponding physical coordinate values. The coefficient matrix is ​​then inverted or pseudo-inverted, and multiplied with the constant matrix to obtain six affine transformation parameters. These parameters reflect the scaling, shearing, and translation relationships between the image coordinate system and the physical coordinate system. These six parameters are then reassembled into a third-order transformation matrix in homogeneous coordinate form, arranged in a 3x3 grid, with the third row containing (0, 0, 1).

[0023] Step 300: Based on the coordinate transformation matrix, control multiple probes in the probe array to perform position deviation fitting with the calibration point to obtain the system error function; It should be noted that multiple representative probes are selected as calibration sample points in the probe array. For example, five probes are selected from the lower left, lower right, upper left, upper right, and geometric center of the array, ensuring that their spatial coverage is sufficient to characterize the geometric error distribution of the entire array. Using a coordinate transformation matrix between pixel coordinates and physical coordinates, the pixel coordinates of these five probes in the image are converted into target position coordinates in the physical coordinate system, serving as the theoretical control position to which the probes should be driven. The control system issues motion commands to the X-axis and Y-axis linear motors based on the conversion results, moving the probe stage to the calculated target position coordinates. After reaching the planar position, the Z-axis descent process is initiated. When the probe tip approaches the surface of the standard calibration piece, real-time monitoring is performed by a contact detection circuit. When the detection resistance is less than the contact threshold of 100 ohms, it is considered a valid contact, and the Z-axis descent is immediately stopped, ensuring that the probe tip just contacts the calibration piece surface without overshoot. At this point, the actual positions of the X-axis, Y-axis, and Z-axis encoders are read to obtain the actual position reached by the probe stage at the moment of contact. This position is then subtracted from the preset target position to obtain the position deviation data of each probe in the X and Y directions at that point. Following this procedure, the calibration process described above is performed sequentially on five selected probes. Deviation samples are collected at preset physical coordinate points on the calibration plate, with typical points set as (10mm, 10mm), (90mm, 10mm), (10mm, 90mm), (90mm, 90mm), and (50mm, 50mm) to ensure uniform spatial distribution of sample points within the working area. Using planar coordinate positions as input variables and probe position deviations as output variables, two two-dimensional quadratic polynomial systematic error function models are constructed to predict the deviation trends in the X and Y directions at any spatial point. The systematic error function model includes constant offsets and linear offset terms, and also considers cross-interference between coordinate axes and nonlinear bending distortion, possessing fitting and global prediction capabilities. Substituting the five sets of deviation samples into the polynomial error function form, a least-squares optimization problem is constructed. By solving the coefficients of each term in the function through matrix calculations, a set of coefficients that minimizes the sum of squared deviations at the five sample points is obtained, completing the fitting modeling of the systematic error function.

[0024] Step 400: Query the three-dimensional error compensation lookup table based on the target physical coordinates to obtain the encoder system deviation, and substitute the target physical coordinates into the system error function to obtain the probe geometric deviation; Specifically, the system acquires the target physical coordinates (x, y, z) set by the user input or system commands, representing the desired spatial position where the probe will move to and complete its positioning. Upon receiving the target physical coordinates, it searches a 3D error compensation lookup table for the M nearest neighbors (M = 8) of the target coordinates. These neighbors are located at the eight vertices of the cube cell containing the target coordinates. The spatial coordinates of these nodes form a 3D grid cell. Each node stores a corresponding 3D position deviation vector in the lookup table, representing the encoder system's error data at that point. Based on the spatial relationship between the target coordinates and the coordinates of the eight nodes of the cube, the normalized coordinate ratios of the target coordinates in the X, Y, and Z directions are calculated to describe the relative position of the target point within the grid cell and serve as interpolation factors. Following the order from the X-axis, Y-axis to the Z-axis, the error vectors at these 8 nodes are first interpolated pairwise in the X-direction, then a second interpolation is performed on the intermediate result obtained from the X-direction interpolation in the Y-direction, and finally a third interpolation is performed in the Z-direction. This process of three one-dimensional linear interpolation operations is completed sequentially to obtain the encoder system deviation value corresponding to the target coordinate point in the three-dimensional error lookup table, represented as a three-dimensional error vector, which describes the cumulative offset of the encoder error in the X, Y, and Z directions, respectively. After completing the lookup table interpolation, the target coordinates (x, y) are substituted into the system error function model as input. The system error function model consists of multiple polynomial terms optimized by least-squares fitting. Algebraic calculations are performed on the input coordinates based on the coefficients of each polynomial to obtain the predicted geometric deviation values ​​in the X and Y directions, that is, the static offset of the probe array at the target position caused by factors such as uneven geometric arrangement, installation errors, or thermal drift.

[0025] Step 500: Superimpose the encoder system deviation and probe geometric deviation onto the target physical coordinates to obtain the compensated control coordinates, and send the compensated control coordinates to the motion controller to drive the probe station to move to the target position, thus completing the probe positioning.

[0026] Specifically, based on the target's physical coordinates, and combining the encoder system deviation obtained through interpolation using a 3D error compensation lookup table with the probe geometric deviation calculated by the system error function, error corrections are applied to the target coordinates in the X, Y, and Z directions. The encoder error and geometric error are then added as incremental terms to the target coordinate values ​​to generate control coordinates containing 3D compensation information. These control coordinates are encapsulated into standard position control commands according to a predefined motion command format and sent to the probe station's motion controller via a bus interface or the motion platform's control protocol. Upon receiving the control commands, the controller parses the target position values ​​in each of the three axes and, based on preset speed and acceleration limits, issues independent motion trajectory planning commands to the X-axis, Y-axis, and Z-axis motors. During execution, the controller dynamically adjusts the motor output based on the error between real-time position feedback and the target position, ensuring that each axis moves to the designated position according to the predefined smooth speed profile. After the X and Y axes complete planar positioning, the Z-axis continues its descent until the probe tip contacts the surface of the object under test. The contact detection circuit outputs a signal to confirm the positioning is complete, achieving a high-precision probe positioning process based on error compensation.

[0027] In one specific embodiment, the process of performing step 100 may specifically include the following steps: The three axes of the drive probe station move uniformly from the initial position to the target position and synchronously record the number of interference fringe changes and encoder displacement readings of each axis at preset spatial intervals. The actual physical displacement of each axis is obtained by calculating the number of changes in interference fringes and the laser wavelength point by point. The encoder displacement of each axis is obtained by subtracting the initial encoder reading from the encoder displacement reading. A three-dimensional error compensation lookup table is calculated for each spatial position based on the actual physical displacement and the encoder displacement.

[0028] Specifically, the X, Y, and Z axis motion modules of the probe station are initialized to zero, with each axis positioned at its mechanical origin or the set initial coordinate starting point. Simultaneously, the interference fringe counter of the laser interferometer is cleared to ensure its current reading corresponds to the physical displacement starting point. Uniform linear motion is then performed on each axis sequentially, using independent axis motion control to simplify data mapping. Taking the X-axis as an example, it is set to move uniformly from 0mm to 100mm at a speed of 5mm / s. During the motion, the spatial sampling interval is set to 1mm, meaning a data recording operation is performed every 1mm of movement. The cumulative changes in the interference fringes are sampled and counted at high frequency using the laser interferometer. For example, assuming a laser wavelength of 632.8nm, the displacement corresponding to each complete interference fringe is half its wavelength, i.e., 0.3164μm. The total number of collected interference fringes is multiplied by the displacement of a single fringe to calculate the actual physical displacement value corresponding to the current sampling point. Simultaneously, the current encoder value is read from the X-axis encoder and subtracted from the initial encoder value to obtain the cumulative displacement of the encoder since its initial position. The two displacement quantities mentioned above represent the actual physical displacement obtained through laser interferometry and the electrical displacement reported by the encoder feedback system, respectively. The difference between the two is the error term of the current point in the X-axis direction. Using the same method, the Y-axis is driven from 0mm to 100mm and the Z-axis from 0mm to 20mm. Equally spaced sampling points are sampled on each axis, with a sampling interval of 1mm for the Y-axis and X-axis, and 0.5mm for the Z-axis, forming a uniform sampling grid corresponding to the three axes. At each grid node, interference fringe counts and encoder readings on the X, Y, and Z axes are simultaneously collected, and point-by-point error calculations are performed to obtain error vectors corresponding one-to-one with the three-dimensional coordinate points. These vectors represent the displacement errors in the X, Y, and Z directions at that location. All error vectors are stored in a three-dimensional error compensation lookup table according to their three-dimensional coordinate positions. The lookup table is organized according to a spatial structure, with 100 points on the X-axis, 100 points on the Y-axis, and 40 points on the Z-axis, totaling 400,000 spatial points, forming a 100×100×40 three-dimensional error matrix grid, and the corresponding three-axis direction error is recorded at each node.

[0029] In one specific embodiment, the process of calculating the three-dimensional error compensation lookup table for each spatial position based on the actual physical displacement and the encoder displacement can specifically include the following steps: Subtract the actual physical displacement of each measurement point from the encoder displacement of the corresponding measurement point to obtain the X-axis displacement deviation sequence, Y-axis displacement deviation sequence, and Z-axis displacement deviation sequence. Based on the spatial positions of each measurement point in the X-axis displacement deviation sequence, Y-axis displacement deviation sequence, and Z-axis displacement deviation sequence, the triaxial deviation data are correlated to form a three-dimensional error compensation lookup table.

[0030] Specifically, during the movement of the probe station along its three axes, the physical displacement calculated from the changes in interference fringes and the corresponding grating encoder displacement readings are recorded point by point, and a sequence of measurement points is established with a fixed spatial sampling interval. Taking the X-axis as an example, the probe station moves uniformly from 0 mm to 100 mm along the X direction with a sampling interval of 1 mm, resulting in 100 measurement points. For each position point, there is a set of actual physical displacement values ​​calculated by the laser interferometer and a set of encoder readings. At this point, a difference calculation is performed on each measurement point, subtracting the corresponding encoder displacement from the physical displacement of the laser interferometer to obtain the displacement deviation data in the X direction. The same method is used for motion control and data sampling of the Y-axis and Z-axis, respectively. The Y-axis range is also 0 mm to 100 mm with a sampling interval of 1 mm, and the Z-axis range is 0 mm to 20 mm with a sampling interval of 0.5 mm, thus obtaining 100 Y-axis sampling points and 40 Z-axis sampling points, respectively. At each measurement point, the deviation values ​​in the Y-axis and Z-axis directions can be calculated. After completing all measurements and deviation calculations, the deviation data corresponding to the three axes are organized according to spatial location, constructing a three-dimensional grid data set. Each grid node corresponds to an actual measured spatial point, whose coordinates are determined by the X, Y, and Z sampling point indices. For example, (i=25, j=50, k=10) represents a position in space where X=25mm, Y=50mm, and Z=5mm. The X, Y, and Z direction deviation values ​​for the current point are stored at each node, forming a three-dimensional error compensation lookup table. The lookup table has a structure size of 100×100×40 and contains 400,000 three-dimensional spatial point error records. Each node contains a set of three-axis deviation vectors, and the data is managed in the form of matrices or nested arrays.

[0031] The process, after forming the three-dimensional error compensation lookup table, includes confidence assessment and local reconstruction steps: For each grid node in the three-dimensional error compensation lookup table, the probe station is moved multiple times to the corresponding position of the node, and the number of changes in the interference fringes of the laser interferometer is repeatedly collected. The standard deviation of the multiple measurement results is calculated as the measurement uncertainty of the node. The measurement uncertainty of each grid node is compared with a preset confidence threshold. When the measurement uncertainty of a certain grid node is greater than the confidence threshold, the node is marked as a low-confidence node. The distribution of all low-confidence nodes in three-dimensional space is statistically analyzed to form the region to be reconstructed. For the region to be reconstructed, the spatial density of the measurement points is increased, and the laser interferometry measurement process is re-executed to reduce the grid spacing in the region to be reconstructed to half or less of the original grid spacing, obtaining high-density displacement deviation data for the region. The high-density displacement deviation data is used to replace the original data of the corresponding low-confidence nodes in the three-dimensional error compensation lookup table. The lookup table after local reconstruction has higher interpolation accuracy and compensation reliability for regions with drastic error changes.

[0032] In one specific embodiment, the process of performing step 200 may specifically include the following steps: The probe array image is segmented by grayscale threshold and the centroid method is calculated to obtain the pixel coordinates of the probe tip. The standard calibration plate image is then subjected to circle detection to obtain the pixel coordinates of the Mark point. Substitute the known physical coordinates of each Mark point and the corresponding pixel coordinates of the Mark point into the affine transformation equation to form a system of linear equations. Represent the system of linear equations in matrix form and multiply the coefficient matrix by the constant matrix to obtain the affine transformation parameters. Arrange the affine transformation parameters in homogeneous coordinate form to form a coordinate transformation matrix.

[0033] Specifically, a high-resolution industrial CCD camera is vertically mounted above the probe array, with its imaging optical axis perpendicular to the XY plane of the probe stage, to acquire a top-view image with minimal distortion. After image acquisition, grayscale thresholding is performed on the original image. An appropriate upper limit for the grayscale threshold is set (e.g., 180). Pixel regions above the grayscale threshold are identified as high-reflectivity areas of the probe tip, while pixels below the grayscale threshold are treated as background areas and masked. Based on the segmentation results, a weighted grayscale centroid calculation is performed on each candidate region. According to the grayscale value distribution of all pixels within the region, a weighted average is used to obtain the sub-pixel coordinates of each probe tip in the image, i.e., (u n , v n The center position of the probe tip in the pixel coordinate system is defined as follows: A standard calibration piece is placed on the probe station working surface, and an image containing four known Mark points is acquired using a camera. These Mark points are designated as P1(x1,y1), P2(x2,y2), P3(x3,y3), and P4(x4,y4). From the acquired calibration piece image, a circular detection algorithm based on Hough transform or template matching is used to extract the pixel coordinates of the four Mark points, designated as Q1(u1,v1), Q2(u2,v2), Q3(u3,v3), and Q4(u4,v4). The four sets of physical coordinates and their corresponding pixel coordinates are then sequentially substituted into the standard two-dimensional affine transformation model x... w = a 11 ·u +a 12 ·v + t x y w = a 21 ·u + a 22 ·v + t y Construct two separate structures containing three variables (a, b, c, ...). 11 a 12 , t x and a 21 a 22 , t yThe linear equations of the above two equation systems are rearranged to form a matrix A·X = B, where A is a 6×6 coefficient matrix composed of pixel coordinates, X is a column vector of unknowns containing 6 affine parameters, and B is a constant vector composed of physical coordinates. The matrix A is then inverted or solved using a pseudo-inverse method to obtain the affine transformation parameter set [a...]. 11 , a 12 ,t x , a 21 , a 22 , t y The affine transformation parameter set is rearranged according to the homogeneous coordinate mapping principle to construct a 3×3 affine coordinate transformation matrix T = [[a] 11 , a 12 , t x ], [a 21 , a 22 , t y ], [0, 0, 1]].

[0034] In one specific embodiment, the process of performing step 300 may specifically include the following steps: Multiple probes in the probe array are selected and their pixel coordinates are converted into physical coordinates using a coordinate transformation matrix. After calculating the target position, the probe stage is driven to move. The Z-axis descends so that the probe tip contacts the calibration plate surface. The contact is determined by the contact resistance. The position deviation data of each probe is obtained by subtracting the actual position reached by the probe stage from the target position. A two-dimensional quadratic polynomial error function is established, which takes spatial coordinates as input variables and outputs the predicted value of position deviation. The two-dimensional quadratic polynomial error function includes a constant term, a first-order coordinate term, a cross-product term of coordinates, and a quadratic coordinate term. The positional deviation data is used as sample points to construct a least-squares optimization equation by substituting it into a two-dimensional quadratic polynomial error function and solving for the polynomial coefficients to obtain the system error function.

[0035] Specifically, several representative probes are selected from the probe array as calibration targets. For example, five probes are selected from the lower left, lower right, upper left, upper right, and center positions of the array. These probes spatially cover the entire probe working area, effectively capturing the overall geometric distortion trend. Based on the coordinate transformation matrix T, the image pixel coordinates (u...) of each selected probe are... n , v n1) The coordinates of the probe tip are converted into target coordinates in the physical coordinate system of the probe station through matrix multiplication. The target coordinates are the theoretical ideal position to which the probe tip is to move. After receiving the target position coordinate command, the motion control system drives the X-axis and Y-axis linear motors to move the probe station to the target plane position and starts the Z-axis to descend at a low and uniform speed. During the process, the contact resistance circuit composed of a constant current source, a resistance detection module and a comparator is monitored. When the probe tip contacts the surface of the calibration piece, the resistance drops sharply below the threshold (e.g., 100Ω) due to the closed circuit. The comparator outputs a high level to confirm the contact and immediately stops the Z-axis descent, ensuring that the probe completes the physical contact operation without lateral offset. The current position is read from the three-axis encoder to obtain the three-dimensional coordinates of the actual contact point. The actual coordinate values ​​in the XY directions are subtracted from the target coordinate values ​​to obtain the two-dimensional position deviation of the nth probe at the current calibration point. The above operation is repeated to collect spatial position deviation data for multiple probes at multiple corresponding calibration points to obtain an error sample set of multiple discrete sampling points. Using spatial coordinates (x, y) as independent variables and the corresponding error values ​​Δx(x,y) and Δy(x,y) as dependent variables, two-dimensional quadratic polynomial functions are constructed as systematic error models in the X and Y directions, respectively. The error function form includes a constant term c0, linear coordinate terms c1·x and c2·y, a cross-product term c3·x·y, and a quadratic term c4·x. 2 With c5·y 2 The constant term characterizes the fixed offset, the linear term describes the directional drift, the cross-product term reflects the coupling distortion between the X and Y axes, and the squared term fits the spatial non-uniform error caused by local nonlinear bending or thermal expansion. To solve for the polynomial coefficients, a least-squares optimization objective function is constructed, minimizing the sum of squares of the differences between the predicted and actual observations of all sample points under the error function. Specifically, this is achieved by substituting the deviation samples into the above function form to form a set of over-constrained equations, which are then expressed in matrix form and solved using the pseudo-inverse method or QR decomposition method to obtain a set of globally optimal polynomial coefficients. This forms a systematic error function describing the geometric distortion distribution of the probe array, which outputs the predicted deviation at any target location point (x, y).

[0036] The process includes obtaining the system error function, followed by establishing a regional error model: dividing the probe station workspace into multiple sub-regions according to error distribution characteristics; calculating the deviation gradient and variance for the probe position deviation data in each sub-region; determining the error complexity level of each sub-region based on the deviation gradient and variance; establishing polynomial error functions of different orders for sub-regions with different error complexity levels, using cubic or higher-order polynomials for sub-regions with high error complexity levels and first or second-order polynomials for sub-regions with low error complexity levels; associating and storing the boundary coordinate range of each sub-region with the corresponding polynomial error function to form a regional error model library; automatically selecting the polynomial error function corresponding to the sub-region when the target physical coordinates fall into a certain sub-region by querying the regional error model library based on the target physical coordinates; substituting the target physical coordinates into the selected polynomial error function for calculation to obtain the probe geometric deviation corresponding to the sub-region. Compared with a global single error function, the regional error model library can more accurately describe the non-uniform error characteristics of the probe array at different spatial locations.

[0037] In one specific embodiment, the process of substituting the position deviation data as sample points into a two-dimensional quadratic polynomial error function to construct a least-squares optimization equation and solving for the polynomial coefficients to obtain the system error function can specifically include the following steps: The positional deviation data of each probe and its corresponding spatial coordinates are used as a set of sample points and substituted into a two-dimensional quadratic polynomial error function to form a prediction deviation expression containing polynomial coefficients. The difference between the actual position deviation of each sample point and the calculated value of the predicted deviation expression is calculated, and the difference is squared and summed to construct a least squares optimization objective function with polynomial coefficients as the solution variables; The system error function is obtained by solving the polynomial coefficients of the least squares optimization objective function through matrix pseudo-inverse operations.

[0038] Specifically, after completing the touch test of multiple probes on a standard calibration plate, the positional deviation data in the X and Y directions between the target position and the actual contact position of each sample probe are obtained. The positional deviation values ​​are then paired with the corresponding spatial target position coordinates (x, y) to form a sample point set. The spatial coordinates (x, y) of the sample points are used as input variables of a two-dimensional quadratic polynomial error function, and the deviation value in each direction is used as the output target of the function. The deviation prediction value is expressed in the form of a quadratic polynomial structure with respect to spatial position, which includes constant terms, coordinate linear terms, cross terms, and coordinate quadratic terms, etc., and polynomial coefficients, thus constructing a prediction deviation expression containing several coefficients. Substituting the input coordinates of each sample point into the prediction deviation expression, the corresponding theoretical deviation prediction value is calculated. This theoretical deviation prediction value is then compared with the actual collected deviation value, and the difference is calculated and squared. Finally, the squared errors of all sample points are summed to form a least-squares optimization objective function with polynomial coefficients as variables. The objective function aims to minimize the total deviation prediction error by adjusting the coefficients to make all predicted values ​​as close as possible to the actual observed values. Based on the least-squares optimization objective function, all sample point data are organized into a standard matrix form. A design matrix expresses the relationship between the input coordinates and the polynomial coefficients, and the objective vector records the actual deviation observation values. The over-reduced linear equation system is solved using matrix pseudo-inverse operations to obtain a set of optimally fitted polynomial coefficient solutions. The polynomial coefficients are organized into a two-dimensional quadratic error function model to output the prediction deviation value under arbitrary plane target coordinates, constituting the system's geometric error function.

[0039] In one specific embodiment, the process of performing step 400 may specifically include the following steps: Based on the target physical coordinates, find the M adjacent grid nodes surrounding the target physical coordinates and their corresponding three-dimensional position deviation vectors in the three-dimensional error compensation lookup table, and calculate the normalized coordinates of the target physical coordinates in the grid cells. The three-dimensional position deviation vectors of M adjacent grid nodes are linearly interpolated along the three coordinate axes in sequence according to the normalized coordinates to obtain the encoder system deviation corresponding to the target physical coordinates; Substitute the target's physical coordinates into the system error function for polynomial calculation to obtain the probe's geometric deviation corresponding to the target's physical coordinates.

[0040] Specifically, the target physical coordinates are obtained, represented as three-dimensional real values ​​(x, y, z). Based on the target physical coordinates, a cubic cell surrounding the target physical coordinates is found in a three-dimensional error compensation lookup table. Each cubic cell consists of eight adjacent grid nodes, i.e., M=8 vertices. The spatial positions of these vertices are known, regularly distributed sampling points in the lookup table, and each point stores a three-dimensional deviation vector, representing the error generated by the encoder system at that node position. For interpolation, the normalized coordinates of the target point within the grid cell are calculated, i.e., the ratio of its relative position between two adjacent grid nodes is calculated along the X, Y, and Z axes, respectively, yielding three scaling parameters t. x t y With t z The value ranges from 0 to 1. The interpolation process is completed in three steps. In the X direction, linear interpolation is performed on four sets of relative nodes in the X direction among the eight points constituting the cube, generating four intermediate error vectors in the YZ plane. In the Y direction, the four intermediate results are divided into two groups and interpolated twice to obtain two intermediate vectors in the Z-axis direction. Finally, these two results are interpolated in the Z direction to obtain the three-dimensional deviation vector corresponding to the target point, which is the theoretical error position of the target point in the encoder system. The error vector is the encoder system deviation of the target point, covering the error components in the X, Y, and Z directions. At the same time, in order to obtain the geometric deviation of the probe tip, the X and Y coordinates of the target point are substituted into the two-dimensional quadratic error function model as independent variables. Algebraic operations are performed on the polynomials in the X and Y directions respectively. By substituting the coefficients and coordinate values, the geometric deviation at the target point is calculated. The geometric deviation reflects the positional error caused by non-encoder factors such as probe array arrangement, visual transformation error, and structural assembly error.

[0041] In one specific embodiment, the process of performing the step of finding the M adjacent grid nodes surrounding the target physical coordinates and their corresponding three-dimensional position deviation vectors in the three-dimensional error compensation lookup table based on the target physical coordinates, and calculating the normalized coordinates of the target physical coordinates within the grid cells, can specifically include the following steps: Locate the grid cell containing the target's physical coordinates in the 3D error compensation lookup table and determine the spatial coordinates of the M adjacent grid nodes surrounding the target's physical coordinates. Read the three-dimensional position deviation vectors stored in M ​​adjacent grid nodes to obtain the encoder deviation components corresponding to each grid node; Normalized coordinates are obtained by subtracting the corresponding lower boundary coordinates of the grid nodes from each axis component of the target physical coordinates and then dividing by the grid spacing.

[0042] Specifically, the target physical coordinates (x, y, z) are used as input for mesh positioning within a pre-constructed 3D error compensation lookup table. The lookup table uses a regular grid structure to cover the entire probe station's workspace. The X and Y axes are divided into 100 sampling intervals with a 1mm step, and the Z axis is divided into 40 sampling levels with a 0.5mm step. The entire lookup table consists of 100×100×40 cubic grid cells, each cell comprising eight spatial vertices, with each vertex storing a 3D error vector—the encoder error components in the X, Y, and Z directions. When the target coordinates are input, the minimum lower bound grid node index (i, j, k) of the target coordinates is quickly determined by dividing the target coordinates by the grid spacing of each axis and rounding the result down, thus clarifying the range of cubic cells containing the target point. After localization, adjacent upper boundary nodes are queried in the X, Y, and Z directions according to the index (i, j, k), forming eight 3D vertex positions. Each vertex has its own spatial coordinates, and the pre-stored 3D position deviation vectors of these eight vertex nodes can be read from the lookup table. These deviation data are systematic error information obtained by comparing the interferometer and encoder. For 3D linear interpolation, the target coordinates are normalized by subtracting the coordinate value of the corresponding lower boundary node from the component of the target coordinates on each axis to obtain the offset relative to the grid cell starting point. This offset is then divided by the grid spacing of each axis, i.e., X-axis divided by 1mm, Y-axis divided by 1mm, and Z-axis divided by 0.5mm, to obtain the normalized coordinates t in the three directions. x t y and t z The values ​​of these three normalization parameters are all in the range of 0 to 1, representing the relative positional ratio of the target point within the cubic cell.

[0043] In one specific embodiment, the process of performing step 500 may specifically include the following steps: The encoder system deviation and probe geometric deviation are respectively superimposed on the target physical coordinates to obtain the compensated control coordinates, and the compensated control coordinates are sent to the motion controller as position control commands. The motion controller drives the X-axis motor, Y-axis motor and Z-axis motor of the probe station to move at the set speed and acceleration according to the received position control command, so as to move the probe station to the target position corresponding to the compensated control coordinates and complete the probe positioning.

[0044] Specifically, the system takes the target physical coordinates set by the user or issued by the system as input, including three-dimensional coordinates of x, y, and z. These three-dimensional coordinates represent the spatial target position that the probe should ideally reach. After receiving the target coordinates, the system enters the error compensation module, which queries the three-dimensional error compensation lookup table to obtain the encoder system deviation value at the current target position. The encoder system deviation is a spatial error vector generated by the error distribution model established between the laser interferometer and the encoder, covering error components in the X, Y, and Z directions, representing the systematic drift of the encoder measurement at the current target position. Simultaneously, the planar components of the target coordinates, namely the x and y components, are substituted into the fitted two-dimensional system error function. The two-dimensional system error function outputs the geometric error prediction value of the probe array at the specified position in the form of a quadratic polynomial. The geometric deviation reflects the positioning error caused by probe assembly, visual transformation residuals, or structural nonlinearity factors. The encoder system deviation and the probe geometric deviation are superimposed in each direction: in the x-direction, the encoder deviation and geometric deviation are added to the x-component respectively; similarly, in the y-direction, they are summed; and in the z-direction, only the encoder deviation is added, since the probe geometric error mainly affects the planar positioning accuracy. This results in a new set of spatial coordinates, i.e., the compensated control coordinates. The compensated control coordinates are packaged into standard position control commands according to the probe station control protocol and sent to the motion controller of the probe station's lower-level machine via a communication bus such as RS-485, CAN, or Ethernet interface. Upon receiving the position control commands, the motion controller calculates the target position contained in the commands into independent target positions for the three axes. It then performs trajectory planning based on set acceleration limits, maximum speed thresholds, and deceleration curves, generating real-time speed and displacement control sequences for the X, Y, and Z axis motors. During execution, the controller employs a closed-loop feedback control mode, correcting position deviations in real time based on encoder feedback and driving each axis to operate synchronously or in stages, moving the probe station to the target position point containing the compensated information. Once all three axes have reached the target position and maintained their positions, the Z-axis contact process is initiated according to the contact determination mechanism, thereby completing the alignment and physical contact between the probe tip and the surface of the chip under test, achieving high-precision probe positioning control.

[0045] The process, after obtaining the system error function and before performing error compensation, includes the following steps: cross-validation and compensation strategy selection for the dual-reference measurement system. This involves selecting multiple cross-validation calibration points with precisely known physical coordinates on the surface of a standard calibration plate; controlling the probe to move to each cross-validation calibration point and obtaining the first actual arrival position through laser interferometer measurement and the second actual arrival position through contact detection; calculating the measurement consistency deviation between the first and second actual arrival positions; evaluating the reliability level of the two measurement references at each cross-validation calibration point based on the measurement consistency deviation; determining that the dual-reference measurement height is consistent at that position when the measurement consistency deviation is less than a first threshold; determining that there is a measurement deviation at that position when the measurement consistency deviation is greater than the first threshold but less than a second threshold; and determining that there is a serious conflict between the dual-reference measurements at that position when the measurement consistency deviation is greater than the second threshold. Finally, the cross-validation calibration points are... The reliability level of a point is associated with its spatial location. This is extended to the entire workspace using spatial interpolation to form a compensation strategy selection mapping table. This table records the compensation mode to be used in each area of ​​the workspace, including laser interferometry-dominated compensation mode, contact test-dominated compensation mode, dual-reference fusion compensation mode, and recalibration trigger mode. During subsequent probe positioning, the compensation strategy selection mapping table is consulted based on the target's physical coordinates to determine the corresponding compensation mode. When the laser interferometry-dominated compensation mode is used, only the encoder system deviation obtained from the 3D error compensation lookup table is used for compensation. When the contact test-dominated compensation mode is used, only the probe geometric deviation calculated using the system error function is used for compensation. When the dual-reference fusion compensation mode is used, the encoder system deviation and probe geometric deviation are superimposed for compensation. When the recalibration trigger mode is used, the area is prompted to re-execute the calibration process.

[0046] The process includes obtaining the actual physical displacement and encoder displacement of each axis, followed by a step to improve data reliability through repeated measurements: For each measurement point on the three axes of the probe station, the probe station is repeatedly moved from its initial position to that measurement point. During each movement, the number of interference fringe changes from the laser interferometer and the encoder displacement reading are simultaneously acquired, obtaining multiple sets of actual physical displacement and encoder displacement corresponding to that measurement point. Statistical analysis is then performed on the multiple sets of actual physical displacement at the same measurement point to calculate the arithmetic mean and standard deviation of the actual physical displacement at that measurement point. When the standard deviation is less than a preset fluctuation threshold, the arithmetic mean is used as the value for that measurement point. The final actual physical displacement is obtained by removing outlier data that deviate from the average by more than twice the standard deviation when the standard deviation is greater than the preset fluctuation threshold, and then recalculating the average value as the final actual physical displacement. The same statistical analysis method is used to obtain the final encoder displacement of the same measurement point by multiple sets of encoder displacements. The final actual physical displacement and the final encoder displacement obtained by fusing multiple repeated measurements of each measurement point are used for the calculation of the subsequent three-dimensional error compensation lookup table. Compared with the single measurement method, the multiple repeated measurement fusion method can reduce the influence of random errors and instantaneous disturbances in the measurement process and improve the reliability of the data stored in the error compensation lookup table.

[0047] The process, after obtaining the system error function and before performing target physical coordinate compensation, includes the step of establishing a generalized linear bias error model: performing residual analysis on the probe geometric deviation predicted by the system error function and the actual position deviation data obtained from the contact-based touch test; calculating the fitting residual for each probe at different spatial positions, where the fitting residual is the difference between the actual position deviation and the probe geometric deviation; performing correlation analysis between the fitting residual and the encoder system deviation at the corresponding spatial positions to establish a generalized linear bias error model. The expression of the generalized linear bias error model includes a probe geometric deviation term, an encoder system deviation term, and a deviation factor term reflecting the coupling relationship between the two. The deviation factor term is derived through the relationship between the probe geometric deviation and the encoder system deviation. The product of the deviations is obtained; the position deviation data, probe geometric deviation, and encoder system deviation are substituted into the generalized linear offset error model as a sample set, and the coefficients of the deviation factor are solved by the least squares method to obtain the generalized compensation model containing error coupling characteristics; when compensating for the target physical coordinates, the encoder system deviation obtained by query and the probe geometric deviation obtained by calculation are substituted into the generalized compensation model, and the interaction coupling amount of the two error sources is calculated by the deviation factor. The encoder system deviation, probe geometric deviation, and interaction coupling amount are superimposed on the target physical coordinates to obtain the compensated control coordinates. Compared with simple algebraic superposition, the generalized linear offset error model can compensate for the second-order nonlinear deviation caused by the interaction of encoder error and probe geometric error.

[0048] The probe calibration method in the embodiments of this application has been described above. The probe control device in the embodiments of this application is described below. Please refer to [link / reference]. Figure 2 One embodiment of the probe control device in this application includes: The acquisition module 11 is used to control the three-axis movement of the probe stage and synchronously acquire the number of changes in the interference fringes of the laser interferometer and the encoder displacement reading, and calculate the three-dimensional error compensation lookup table; Affine transformation module 12 is used to acquire the pixel coordinates of Mark points on the standard calibration sheet image and perform affine transformation to obtain the coordinate transformation matrix. Fitting module 13 is used to control multiple probes in the probe array to perform position deviation fitting on the calibration point according to the coordinate transformation matrix, so as to obtain the system error function; Calculation module 14 is used to query the three-dimensional error compensation lookup table based on the target physical coordinates to obtain the encoder system deviation, and substitute the target physical coordinates into the system error function to obtain the probe geometric deviation; The drive module 15 is used to superimpose the encoder system deviation and probe geometric deviation onto the target physical coordinates to obtain the compensated control coordinates, and send the compensated control coordinates to the motion controller to drive the probe station to move to the target position and complete the probe positioning.

[0049] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0050] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0051] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A probe calibration method, characterized in that, include: The process involves controlling the three-axis movement of the probe stage and synchronously acquiring the number of interference fringe changes and encoder displacement readings from a laser interferometer, then calculating a three-dimensional error compensation lookup table. Specifically, this includes: driving the three axes of the probe stage to move uniformly from an initial position to a target position and synchronously recording the number of interference fringe changes and encoder displacement readings for each axis at preset spatial intervals; calculating the actual physical displacement of each axis point-by-point based on the number of interference fringe changes and the laser wavelength; subtracting the initial encoder reading from the encoder displacement reading to obtain the encoder displacement of each axis; and calculating a three-dimensional error compensation lookup table for each spatial position based on the actual physical displacement and the encoder displacement. The process involves acquiring the pixel coordinates of Mark points on a standard calibration sheet image and performing an affine transformation to obtain a coordinate transformation matrix. Specifically, this includes: performing grayscale thresholding and centroid calculation on the probe array image to obtain the probe tip pixel coordinates; performing circle detection on the standard calibration sheet image to obtain the Mark point pixel coordinates; substituting the known physical coordinates of each Mark point and its corresponding pixel coordinates into the affine transformation equations to form a system of linear equations; expressing the system of linear equations in matrix form and multiplying the coefficient matrix by the constant matrix to obtain the affine transformation parameters; and arranging the affine transformation parameters in homogeneous coordinate form to form a coordinate transformation matrix. The system error function is obtained by controlling multiple probes in the probe array to perform position deviation fitting to the calibration point according to the coordinate transformation matrix. Specifically, this includes: selecting multiple probes in the probe array, converting the pixel coordinates of each probe to physical coordinates through the coordinate transformation matrix, calculating the target position, driving the probe stage to move, lowering the Z-axis to make the probe tip contact the calibration plate surface, and determining the contact by the contact resistance; subtracting the target position from the actual position reached by the probe stage to obtain the position deviation data of each probe; establishing a two-dimensional quadratic polynomial error function with spatial coordinates as input variables and outputting the predicted position deviation value, wherein the two-dimensional quadratic polynomial error function includes a constant term, a first-order coordinate term, a coordinate cross-product term, and a quadratic coordinate term; substituting the position deviation data as sample points into the two-dimensional quadratic polynomial error function to construct a least squares optimization equation and solving the polynomial coefficients to obtain the system error function. The encoder system deviation is obtained by querying the three-dimensional error compensation lookup table based on the target physical coordinates, and the probe geometric deviation is obtained by substituting the target physical coordinates into the system error function. Specifically, this includes: finding the M adjacent grid nodes surrounding the target physical coordinates and their corresponding three-dimensional position deviation vectors in the three-dimensional error compensation lookup table based on the target physical coordinates; calculating the normalized coordinates of the target physical coordinates within the grid cells; performing linear interpolation on the three-dimensional position deviation vectors of the M adjacent grid nodes sequentially along the three coordinate axes according to the normalized coordinates to obtain the encoder system deviation corresponding to the target physical coordinates; and substituting the target physical coordinates into the system error function for polynomial calculation to obtain the probe geometric deviation corresponding to the target physical coordinates. The encoder system deviation and the probe geometric deviation are superimposed on the target physical coordinates to obtain compensated control coordinates. These compensated control coordinates are then sent to the motion controller to drive the probe station to the target position, completing probe positioning. Specifically, this involves superimposing the encoder system deviation and the probe geometric deviation onto the target physical coordinates to obtain compensated control coordinates, and sending these compensated control coordinates as position control commands to the motion controller. The motion controller, based on the received position control commands, drives the probe station's X-axis, Y-axis, and Z-axis motors to move at set speeds and accelerations, moving the probe station to the target position corresponding to the compensated control coordinates, thus completing probe positioning.

2. The probe calibration method according to claim 1, characterized in that, The calculation of the three-dimensional error compensation lookup table for each spatial position based on the actual physical displacement and the encoder displacement includes: Subtract the actual physical displacement of each measurement point from the encoder displacement of the corresponding measurement point to obtain the X-axis displacement deviation sequence, Y-axis displacement deviation sequence, and Z-axis displacement deviation sequence. Based on the spatial positions of each measurement point in the X-axis displacement deviation sequence, the Y-axis displacement deviation sequence, and the Z-axis displacement deviation sequence, the triaxial deviation data are correlated to form a three-dimensional error compensation lookup table.

3. The probe calibration method according to claim 1, characterized in that, The process of substituting the position deviation data as sample points into the two-dimensional quadratic polynomial error function to construct a least-squares optimization equation and solving for the polynomial coefficients yields the system error function, including: The positional deviation data of each probe and its corresponding spatial coordinates are used as a set of sample points and substituted into the two-dimensional quadratic polynomial error function to form a prediction deviation expression containing polynomial coefficients. The difference between the actual position deviation of each sample point and the calculated value of the predicted deviation expression is calculated, and the squares of the difference are summed to construct a least squares optimization objective function with polynomial coefficients as the solution variables. The system error function is obtained by solving the polynomial coefficients of the least squares optimization objective function through matrix pseudo-inverse operation.

4. The probe calibration method according to claim 1, characterized in that, The step of finding the M neighboring grid nodes surrounding the target physical coordinates and their corresponding 3D position deviation vectors in the 3D error compensation lookup table based on the target physical coordinates, and calculating the normalized coordinates of the target physical coordinates within the grid cells, includes: Based on the target physical coordinates, locate the grid cell in the three-dimensional error compensation lookup table and determine the spatial coordinate positions of the M adjacent grid nodes surrounding the target physical coordinates; Read the three-dimensional position deviation vectors stored in the M adjacent grid nodes to obtain the encoder deviation components corresponding to each grid node; The normalized coordinates are obtained by subtracting the lower boundary coordinates of the corresponding grid nodes from each axis component of the target physical coordinates and then dividing by the grid spacing.

5. A probe control device, characterized in that, For performing the probe calibration method as described in any one of claims 1-4, comprising: The acquisition module is used to control the three-axis movement of the probe stage and synchronously acquire the number of changes in the interference fringes of the laser interferometer and the encoder displacement reading, and calculate the three-dimensional error compensation lookup table; The affine transformation module is used to acquire the pixel coordinates of Mark points on the standard calibration sheet image and perform affine transformation to obtain the coordinate transformation matrix. The fitting module is used to control multiple probes in the probe array to perform position deviation fitting on the calibration point according to the coordinate transformation matrix, so as to obtain the system error function; The calculation module is used to query the three-dimensional error compensation lookup table based on the target physical coordinates to obtain the encoder system deviation, and substitute the target physical coordinates into the system error function to obtain the probe geometric deviation; The drive module is used to superimpose the encoder system deviation and the probe geometric deviation onto the target physical coordinates to obtain the compensated control coordinates, and send the compensated control coordinates to the motion controller to drive the probe station to move to the target position and complete the probe positioning.

Images