A smt yield optimization method based on multi-source data analysis

Abstract

The application relates to the technical field of printed circuits, and particularly discloses an SMT yield optimization method based on multi-source data analysis, which obtains three-dimensional volume measurement values of each tin paste point and coordinate positions of the tin paste points on a printed circuit board after tin paste printing, collects actual measurement values of a pressing depth of a suction nozzle in a component mounting process, and pairs the two types of data according to product serial numbers and the coordinate positions to establish a space mapping relationship. Through traversing the mapping data, mounting positions with tin paste volume being too large and the pressing depth being too shallow are identified, corresponding pressing depth compensation values are calculated according to volume deviations of the positions, optimized pressing depth parameters are generated and written into a parameter storage area of a chip mounter, so that the chip mounter adopts the optimized pressing depth to perform mounting at corresponding coordinate positions when producing the same type of printed circuit board in the subsequent production, thereby realizing mounting parameter optimization based on production data, and the application is favorable for improving the yield level of an SMT production line.

Description

Technical Field

[0001] This invention relates to the field of printed circuit technology, and more specifically to an SMT yield optimization method based on multi-source data analysis. Background Technology

[0002] In surface mount technology (SMT) manufacturing, solder paste printing quality and component placement status are crucial factors affecting soldering reliability and overall production yield. In actual production, different equipment generates a large amount of inspection and operational data at different stages, such as inspection results after solder paste printing and operational status information fed back by the equipment during placement. This data is often recorded and stored independently by different equipment, lacking a unified correlation and analysis mechanism, making it difficult to effectively link data across different processes in production.

[0003] When soldering defects or mounting abnormalities occur in certain solder joint areas, production personnel can usually only make judgments based on data from a single process, making it difficult to accurately identify the interrelationships between factors in different processes. Furthermore, because the condition of solder joints varies at different locations on the same printed circuit board, relying on experience or adjusting overall parameters makes it difficult to precisely control localized problems. This leads to a degree of blindness in parameter adjustments during production, making it difficult to promptly identify potential process deviations and implement targeted optimizations, thereby affecting mounting stability and product yield. Summary of the Invention

[0004] The purpose of this invention is to provide an SMT yield optimization method based on multi-source data analysis to solve the above-mentioned technical problems.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A method for optimizing SMT yield based on multi-source data analysis includes the following steps:

[0007] After solder paste printing, test data is obtained, which includes the three-dimensional volume measurement value of each solder paste dot and the coordinate position of the solder paste dot on the printed circuit board.

[0008] Obtain the measured values ​​of the pressing depth during the electronic component placement process, and store each measured value of the pressing depth in association with the coordinate position of a solder paste dot;

[0009] Establish a spatial mapping table between the test data and the measured pressure depth. The spatial mapping table uses the product serial number of the printed circuit board as the primary key. The coordinate positions in the test data corresponding to the same product serial number are paired with the coordinate positions in the measured pressure depth and then stored.

[0010] Traverse the spatial mapping table. For each pair, when the three-dimensional volume measurement value of the solder paste dot is greater than the preset upper volume threshold and the corresponding measured pressing depth value is less than the preset lower pressing depth threshold, mark the pair as a pair to be optimized and extract the coordinate position of the solder paste dot of the pair to be optimized.

[0011] The average volume deviation is obtained by subtracting the preset standard volume value from the average of the three-dimensional volume measurement values ​​of all pairs to be optimized. The average volume deviation is multiplied by the preset depth compensation coefficient to obtain the depth compensation value. The depth compensation value is added to the reference pressing depth value at the corresponding coordinate position in the pick and place machine control table to generate the optimized pressing depth value.

[0012] The optimized pressure depth value is written into the parameter storage area of ​​the pick-and-place machine. When the pick-and-place machine subsequently produces the same type of printed circuit board, it uses the optimized pressure depth value to perform component placement at the corresponding coordinate position.

[0013] As a further aspect of the present invention: acquiring detection data includes:

[0014] A data export command is sent to the solder paste inspection equipment. In response to the data export command, the solder paste inspection equipment packages the inspection data of the current production batch into a data file and transmits it to the local server. After parsing the data file, the local server extracts the three-dimensional volume measurement value of each solder paste point. The three-dimensional volume measurement value includes the height measurement value, area measurement value, and volume estimate calculated by multiplying the height measurement value and area measurement value of the solder paste point. At the same time, the horizontal and vertical coordinate values ​​of the solder paste point on the printed circuit board are extracted to form the coordinate position.

[0015] As a further aspect of the present invention, the specific process for obtaining the measured value of the compression depth is as follows:

[0016] The position change curve of the nozzle during the placement process is read from the encoder feedback of the pick-and-place machine's servo motor. The moment when the nozzle contacts the solder paste surface is read from the sampling data of the pick-and-place machine's pressure sensor. The displacement difference between the starting point and the ending point is calculated as the measured value of the pressing depth. The coordinate position of the nozzle at the moment of placement is read from the motion controller of the pick-and-place machine. The coordinate position is matched with the measured value of the pressing depth and then written into the temporary storage area.

[0017] As a further aspect of the present invention: establishing a spatial mapping table includes:

[0018] Using the product serial number of the printed circuit board as the search keyword, all data records corresponding to the product serial number are queried in the test data storage area and the measured pressure depth storage area respectively. The coordinate positions in the queried test data are compared point by point with the coordinate positions in the measured pressure depth. When the difference between the horizontal coordinate and the vertical coordinate of the two coordinate positions is less than the preset positioning tolerance, they are determined to be the same solder paste point. The three-dimensional volume measurement value and the measured pressure depth value of the solder paste point are stored in the same row of the mapping table.

[0019] As a further aspect of the present invention: the preset upper limit threshold for volume is determined by statistically analyzing the three-dimensional volume measurement values ​​of all solder paste points in historical production batches, and the 85th percentile of the obtained three-dimensional volume measurement values ​​is determined as the upper limit threshold for volume. The preset lower limit threshold for pressure depth is determined by statistically analyzing the measured pressure depth values ​​of all solder paste points in historical production batches, and the 15th percentile of the obtained measured pressure depth values ​​is determined as the lower limit threshold for pressure depth. When traversing the spatial mapping table, the three-dimensional volume measurement values ​​and the measured pressure depth values ​​are read row by row and compared with the upper limit threshold for volume and the lower limit threshold for pressure depth, respectively.

[0020] As a further aspect of the present invention, the specific process for calculating the average value of the three-dimensional volume measurement values ​​of all pairs to be optimized is as follows: count the number of rows marked as pairs to be optimized in the spatial mapping table, sum the three-dimensional volume measurement values ​​of all pairs to be optimized to obtain the total volume, divide the total volume by the number of rows to obtain the average value of the three-dimensional volume measurement values, the preset standard volume value is the target volume value read from the solder paste printing process specification, and the preset depth compensation coefficient is the coefficient value obtained by fitting the mounting yield data of the same type of printed circuit board in historical production batches after regression analysis.

[0021] As a further aspect of the present invention, the specific process of writing the optimized pressing depth value into the parameter storage area of ​​the pick-and-place machine is as follows: the optimized pressing depth value and the corresponding coordinate position are combined into a parameter update record, the parameter update record is sent to the controller of the pick-and-place machine, the controller writes the parameter update record into a temporary parameter table in the memory of the pick-and-place machine, and when the controller detects that the product serial number of the printed circuit board belongs to the same product model as the product serial number of the previous production batch, the reference pressing depth value at the corresponding coordinate position in the pick-and-place machine control table is replaced with the optimized pressing depth value stored in the temporary parameter table before the placement program starts to execute.

[0022] The beneficial effects of this invention compared to the prior art are as follows:

[0023] This invention comprehensively analyzes data from different sources during the production process, enabling information from different processes to form effective correlations and thus providing a more comprehensive reflection of the actual production status at each location on the printed circuit board. Based on this, key factors that may affect placement quality during production can be identified, and targeted optimizations can be made for locations exhibiting abnormal characteristics, making adjustments to production parameters more evidence-based and targeted. This method reduces reliance on experience for parameter adjustments, resulting in more stable and consistent process control and facilitating the timely detection of potential process deviations. Simultaneously, this method continues to function in subsequent production processes, ensuring that production equipment maintains a more stable operating state during placement operations, thereby improving the consistency and reliability of the placement process. Through these methods, the placement quality during production can be effectively improved, reducing quality risks caused by localized process deviations and further promoting the stable operation of the overall production process, ultimately contributing to higher yield rates for SMT production lines. Attached Figure Description

[0024] The invention will now be further described with reference to the accompanying drawings.

[0025] Figure 1 This is a flowchart illustrating an SMT yield optimization method based on multi-source data analysis according to the present invention. Detailed Implementation

[0026] 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 embodiments of the present invention, and not all embodiments. 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.

[0027] Please see Figure 1 As shown, this invention is an SMT yield optimization method based on multi-source data analysis, comprising the following steps:

[0028] After solder paste printing, test data is obtained, which includes the three-dimensional volume measurement value of each solder paste dot and the coordinate position of the solder paste dot on the printed circuit board.

[0029] In a preferred embodiment of the present invention, acquiring detection data includes:

[0030] A data export command is sent to the solder paste inspection equipment. In response to the data export command, the solder paste inspection equipment packages the inspection data of the current production batch into a data file and transmits it to the local server. After parsing the data file, the local server extracts the three-dimensional volume measurement value of each solder paste point. The three-dimensional volume measurement value includes the height measurement value, area measurement value, and volume estimate calculated by multiplying the height measurement value and area measurement value of the solder paste point. At the same time, the horizontal and vertical coordinate values ​​of the solder paste point on the printed circuit board are extracted to form the coordinate position.

[0031] Specifically, after solder paste printing is completed, the printed circuit board (PCB) enters the inspection station of the solder paste inspection equipment. The conveying mechanism moves the PCB to the inspection platform, where a positioning mechanism performs edge calibration to ensure the PCB's position in the equipment coordinate system matches the preset program. The inspection equipment then scans each solder paste dot area point by point according to pre-stored pad position information, collecting surface height distribution data and corresponding planar contour data for each dot. The height measurement value is the vertical distance between the top of the solder paste and the base surface of the pad within the dot area. The inspection equipment can form multiple sampling points within the solder paste dot coverage area and use the highest point or a smoothed representative height as the height measurement value for that solder paste dot. The area measurement value is the coverage area of ​​the solder paste dot on the pad plane. Specifically, this is achieved by first cropping the area to be measured according to the pad design window, then identifying the outer contour of the solder paste according to grayscale edges and height abrupt change boundaries, and converting the number of pixels inside the contour into the actual area value. For example, when a single pixel corresponds to 0.01 square millimeters, accumulating 100 effective pixels within the contour yields the corresponding area measurement result. The volume estimate is obtained by multiplying the height measurement value and the area measurement value. It is used to characterize the three-dimensional volume measurement result of the solder paste point. If a solder paste point has edge collapse or uneven top, the inspection equipment first removes the local abnormal points, and then outputs the corrected height measurement value and area measurement value to ensure that the volume estimate matches the actual solder paste morphology. After completing the whole board scan, the solder paste inspection equipment generates a data file according to the current production batch. The data file contains the product serial number, inspection time, number of each solder paste point, height measurement value, area measurement value, volume estimate value, and corresponding horizontal and vertical coordinate values. The horizontal and vertical coordinate values ​​are recorded with the process reference point at the lower left corner of the printed circuit board as the origin. Upon receiving the data export instruction, the solder paste inspection equipment transmits the data file to the local server via industrial Ethernet. The local server performs field parsing and format verification on the data file, reads each parameter one by one according to the solder paste point number, merges the height measurement value, area measurement value, and volume estimation value into the three-dimensional volume measurement value of the solder paste point, and combines the horizontal and vertical coordinate values ​​into coordinate positions and writes them into the inspection data storage area to form a whole board inspection data record corresponding to the product serial number, which can be used for coordinate matching with the actual pressure depth measurement value in the subsequent mounting process.

[0032] Obtain the measured values ​​of the pressing depth during the electronic component placement process, and store each measured value of the pressing depth in association with the coordinate position of a solder paste dot;

[0033] In a preferred embodiment of the present invention, the specific process for obtaining the measured value of the compression depth is as follows:

[0034] The position change curve of the nozzle during the placement process is read from the encoder feedback of the pick-and-place machine's servo motor. The moment when the nozzle contacts the solder paste surface is read from the sampling data of the pick-and-place machine's pressure sensor. The displacement difference between the starting point and the ending point is calculated as the measured value of the pressing depth. The coordinate position of the nozzle at the moment of placement is read from the motion controller of the pick-and-place machine. The coordinate position is matched with the measured value of the pressing depth and then written into the temporary storage area.

[0035] Specifically, during the electronic component placement process, after the pick-and-place machine completes material handling and visual correction, the placement head, carrying the nozzle, moves above the target pad. The motion controller outputs a planar coordinate command for that placement point based on the current placement program, aligning the nozzle center with the corresponding solder paste point. Simultaneously with the start of the descent, position and pressure sampling are initiated. The servo motor's encoder continuously feeds back the nozzle's vertical displacement data. The controller records continuous position points according to a fixed sampling period, thus forming a position change curve of the nozzle moving downwards from a safe height until it stops descending. This position change curve is essentially a sequence of vertical coordinates arranged in chronological order; for example, within a certain placement cycle, height values ​​of 3.20, 3.05, 2.91, 2.80, and 2.72 can be recorded sequentially. The pressure sensor synchronously outputs data on the force change at the nozzle tip. Before the nozzle contacts the solder paste surface, the sampled value remains within the no-load fluctuation range. Once the nozzle tip makes initial contact with the solder paste surface, the sensor output continuously increases. The controller compares the current sampled value with the no-load baseline value. When the difference exceeds a preset contact threshold for multiple consecutive sampling points, the first sampling moment meeting the condition is determined as the contact moment, and the corresponding vertical position is read from the position change curve as the starting position. After the nozzle continues to descend and completes the component pressing action, the servo motor enters a stopped descent state. The controller determines the moment the nozzle stops descending based on the condition that the position change over multiple adjacent sampling cycles is less than a preset stop threshold, and the corresponding vertical position is read from the position change curve as the ending position. The displacement difference between the starting and ending positions is then defined as the measured pressing depth. This displacement difference reflects the actual downward movement of the nozzle from contact with the solder paste surface to the end of the mounting process. For example, if the starting position is 2.80 and the ending position is 2.46, the measured pressing depth for this mounting is recorded as 0.34. To avoid misjudgments caused by pressure fluctuations and mechanical vibrations, the controller performs a moving average on the pressure sampling sequence before determining the contact moment and removes isolated outliers in the position change curve, ensuring that the position data corresponding to the start and end points remain consistent with the actual motion state. After reading the measured pressure depth, the motion controller further extracts the coordinate position of the nozzle at the moment of placement. This coordinate position is represented by the horizontal and vertical coordinate values ​​in the pick-and-place machine's working coordinate system, and the coordinates corresponding to the component's center landing point are taken as the current placement point coordinates. The controller combines the product serial number, placement head number, nozzle number, placement point number, horizontal coordinate value, vertical coordinate value, and the corresponding measured pressure depth value into a placement process record and writes it to a temporary storage area. The temporary storage area is indexed according to the product serial number, so that each placement point record on the same printed circuit board can be paired and associated with the previously acquired solder paste dot detection data according to the coordinate position in subsequent processing.

[0036] Establish a spatial mapping table between the test data and the measured pressure depth. The spatial mapping table uses the product serial number of the printed circuit board as the primary key. The coordinate positions in the test data corresponding to the same product serial number are paired with the coordinate positions in the measured pressure depth and then stored.

[0037] In another preferred embodiment of the present invention, establishing a spatial mapping table includes:

[0038] Using the product serial number of the printed circuit board as the search keyword, all data records corresponding to the product serial number are queried in the test data storage area and the measured pressure depth storage area respectively. The coordinate positions in the queried test data are compared point by point with the coordinate positions in the measured pressure depth. When the difference between the horizontal coordinate and the vertical coordinate of the two coordinate positions is less than the preset positioning tolerance, they are determined to be the same solder paste point. The three-dimensional volume measurement value and the measured pressure depth value of the solder paste point are stored in the same row of the mapping table.

[0039] Specifically, after completing the storage of test data and measured depth values, the product serial number of the printed circuit board is read as the unique retrieval identifier for the current board-level data. Using this product serial number, all corresponding solder paste point test records are retrieved from the test data storage area, and all placement process records corresponding to the same product serial number are retrieved from the measured depth value storage area. Since the two types of data originate from the solder paste inspection equipment and the pick-and-place machine respectively, although both contain horizontal and vertical coordinate values, differences exist in the coordinate acquisition time, coordinate origin setting, and equipment positioning errors. Therefore, they cannot be directly matched based on identical values. It is necessary to first convert both types of records to a unified printed circuit board coordinate reference. In the specific processing, the coordinate position in the test data is maintained as the comparison reference. The coordinate position in the measured depth value record is corrected according to the board edge reference point corresponding to the product model, so that both types of coordinates are represented as planar position data relative to the same reference origin. After coordinate unification, the x-coordinate and y-coordinate values ​​of each solder paste point are read one by one from the detection data records. Then, the closest coordinate position is found in the set of push-down depth records corresponding to the same product serial number, and the difference between the x-coordinate and y-coordinate values ​​is calculated. The positioning tolerance is defined as the maximum planar offset range allowed to be identified as the same solder paste point. This tolerance can be pre-written in the adjustment records for the same product model. For example, the allowable lateral deviation can be set to 0.05, and the allowable axial deviation can be set to 0.05. When the difference between the x-coordinate and y-coordinate of a push-down depth record and the current detection record is both less than this positioning tolerance, the two records are considered to correspond to the same solder paste point on the printed circuit board. To avoid duplicate matching of a push-down depth record, after a valid pairing is completed, the push-down depth record is marked as used and will not participate in subsequent solder paste point comparisons. For each valid pairing, the three-dimensional volume measurement value and the corresponding measured depth of the solder paste point are extracted. The product serial number, solder paste point number, horizontal axis value, vertical axis value, three-dimensional volume measurement value and measured depth value are written into the same row of the spatial mapping table. The spatial mapping table uses the product serial number as the primary key index, so that all paired data on the same printed circuit board are stored centrally.

[0040] Traverse the spatial mapping table. For each pair, when the three-dimensional volume measurement value of the solder paste dot is greater than the preset upper volume threshold and the corresponding measured pressing depth value is less than the preset lower pressing depth threshold, mark the pair as a pair to be optimized and extract the coordinate position of the solder paste dot of the pair to be optimized.

[0041] In a preferred embodiment of the present invention, the preset upper limit threshold for volume is determined by statistically analyzing the three-dimensional volume measurements of all solder paste points in historical production batches, and the 85th percentile of the obtained three-dimensional volume measurements is determined as the upper limit threshold for volume. The preset lower limit threshold for pressure depth is determined by statistically analyzing the measured pressure depths of all solder paste points in historical production batches, and the 15th percentile of the obtained measured pressure depths is determined as the lower limit threshold for pressure depth. When traversing the spatial mapping table, the three-dimensional volume measurements and measured pressure depths are read row by row and compared with the upper limit threshold for volume and the lower limit threshold for pressure depth, respectively.

[0042] Specifically, after establishing the spatial mapping table, historical production batch data corresponding to the same product model is read as threshold statistical samples. This historical production batch data includes 3D volume measurements and measured compression depths that have been paired according to coordinates. After the 3D volume measurement set is formed, it is first sorted from smallest to largest. Then, the target value in the sorted results is extracted based on the 85th percentile position of the total number of records as the upper limit threshold for volume. If the number of records involved in the sorting is 1000, the value near the starting position of the last 15% is taken as the boundary value, indicating the judgment boundary for a larger volume. The set of measured compression depths is processed in the same way. After sorting all measured values ​​from smallest to largest, the value corresponding to the 15th percentile position is read as the lower limit threshold for compression depth, indicating the judgment boundary for a shallower compression. After determining the threshold, the spatial mapping table record corresponding to the current printed circuit board is retrieved according to the product serial number, and traversed row by row in order of row number. When reading each row, the 3D volume measurement value and the measured pressure depth value are extracted. The 3D volume measurement value is compared with the upper limit threshold, and the measured pressure depth value is compared with the lower limit threshold. When the 3D volume measurement value is greater than the upper limit threshold and the measured pressure depth value is less than the lower limit threshold, the solder paste point corresponding to that row is determined to meet the abnormal conditions in both the printing and mounting states. The pairing is added to the set of records to be optimized, and an optimization mark is written to that row. After writing the optimization mark, the horizontal and vertical coordinate values ​​of that row are extracted as the coordinate positions of the solder paste points to be paired, and stored together with the corresponding product serial number and solder paste point number. This is used to generate optimized pressure depth values ​​based on the specific coordinate positions later. Pairings that do not meet both of the above conditions are kept in their original state and are only retained in the spatial mapping table for subsequent traversal.

[0043] The average volume deviation is obtained by subtracting the preset standard volume value from the average of the three-dimensional volume measurement values ​​of all pairs to be optimized. The average volume deviation is multiplied by the preset depth compensation coefficient to obtain the depth compensation value. The depth compensation value is added to the reference pressing depth value at the corresponding coordinate position in the pick and place machine control table to generate the optimized pressing depth value.

[0044] In a preferred embodiment of the present invention, the specific process for calculating the average value of the three-dimensional volume measurements of all pairs to be optimized is as follows: count the number of rows marked as pairs to be optimized in the spatial mapping table, sum the three-dimensional volume measurements of all pairs to be optimized to obtain the total volume, divide the total volume by the number of rows to obtain the average value of the three-dimensional volume measurements, the preset standard volume value is the target volume value read from the solder paste printing process specification, and the preset depth compensation coefficient is the coefficient value obtained by fitting the mounting yield data of the same type of printed circuit board in historical production batches through regression analysis.

[0045] Specifically, after marking the pairs to be optimized, the spatial mapping table corresponding to the current product serial number is read, all rows marked as "to be optimized" are filtered, and the total number of such rows is counted. The three-dimensional volume measurement values ​​of each row are read out sequentially and accumulated to form the total volume corresponding to the pairs to be optimized. Then, the total volume is divided by the total number of rows to be optimized to obtain the average three-dimensional volume measurement of the current batch of pairs to be optimized. This average value is used to characterize the overall deviation of the abnormal positions on the current printed circuit board that require adjustment of mounting parameters in terms of solder paste volume. The standard volume value is read from the solder paste printing process specification, specifically the target volume value pre-set in the process file for this product model. The process file usually stores the target printing parameters in the manner corresponding to the pad numbers. When performing this step, the target volume field consistent with the current product model is read and used as the benchmark value for subsequent deviation calculation. Then, the average three-dimensional volume measurement is compared with the standard volume value. If the average three-dimensional volume measurement is higher than the standard volume value, the difference between the two is recorded as the average volume deviation, which reflects the degree to which the overall pairs to be optimized exceed the process target. The depth compensation coefficient is obtained by fitting historical production batch data of the same model of printed circuit board. Specifically, statistical data from multiple historical production batches of the same product model are retrieved. For each historical batch, the average volume deviation to be optimized in the spatial mapping table, the actual depth adjustment used during production, and the corresponding placement yield data are recorded. The data from each historical batch are compiled into a single sample record, ensuring that each record contains at least three fields: average volume deviation, depth adjustment, and placement yield. All sample records are then sorted from highest to lowest placement yield. Records with placement yields in the higher range (e.g., exceeding a preset threshold) are selected as the valid sample set to ensure that the data used for fitting comes from batches with relatively stable production conditions. After sample selection, the average volume deviation is used as the input variable, and the actual depth adjustment used in the corresponding batch is used as the output variable. Linear regression analysis is then used to establish the correspondence between the two. During linear regression processing, all valid samples are sorted from smallest to largest according to the average volume deviation. An optimal fitting line is calculated based on the distribution trend of the sample points in the two-dimensional plane. This fitted line describes the changing trend of the pressure depth adjustment as the average volume deviation changes. After fitting, the slope parameter is extracted from the fitted line as the depth compensation coefficient. This slope parameter represents the proportion of pressure depth adjustment corresponding to a unit volume deviation. Therefore, the depth compensation coefficient is a single constant value, and this constant value is uniformly used for a specific product model and does not change with coordinate position.For example, in the historical samples of a certain product model, as the average volume deviation gradually increases, the depth adjustment amount used to achieve a higher placement yield in the corresponding batch also shows a stable increasing trend. The slope obtained through linear regression can then be used as the depth compensation coefficient for that product model. This coefficient is then multiplied by the average volume deviation to obtain the corresponding depth compensation amount when calculating the depth compensation value later. After obtaining the depth compensation coefficient, the average volume deviation is multiplied by this coefficient to generate the depth compensation value shared by the current pairing to be optimized. Then, according to the coordinate positions recorded in the pairing to be optimized, the reference depth value at the corresponding position in the pick-and-place machine control table is read one by one. This reference depth value is added to the depth compensation value to form the optimized depth value at the corresponding coordinate position. The coordinate position and the optimized depth value are then paired and saved for later writing to the pick-and-place machine parameter storage area.

[0046] The optimized pressure depth value is written into the parameter storage area of ​​the pick-and-place machine. When the pick-and-place machine subsequently produces the same type of printed circuit board, it uses the optimized pressure depth value to perform component placement at the corresponding coordinate position.

[0047] In another preferred embodiment of the present invention, the specific process of writing the optimized pressing depth value into the parameter storage area of ​​the pick-and-place machine is as follows: the optimized pressing depth value and the corresponding coordinate position are combined into a parameter update record, the parameter update record is sent to the controller of the pick-and-place machine, the controller writes the parameter update record into a temporary parameter table in the memory of the pick-and-place machine, and when the controller detects that the product serial number of the printed circuit board belongs to the same product model as the product serial number of the previous production batch, the controller replaces the reference pressing depth value at the corresponding coordinate position in the pick-and-place machine control table with the optimized pressing depth value stored in the temporary parameter table before the placement program starts to execute.

[0048] Specifically, after obtaining the optimized pressure depth value corresponding to each pair to be optimized, each optimized pressure depth value is combined with its corresponding horizontal and vertical coordinate values ​​and product serial number into a parameter update record. This parameter update record includes at least the product model identifier, placement point number, coordinate position, original baseline pressure depth value, and optimized pressure depth value, enabling the controller to accurately write the updated content to the specified placement point based on the coordinate position. After the parameter update record is generated, it is sent to the placement machine controller via the communication link between the placement machine and the local server. Upon receiving the data, the controller first performs format and model verification on the record content to confirm that the coordinate position is consistent with the placement point definition in the current placement program. Then, it writes the verified parameter update record into a temporary parameter table in the placement machine's memory. The temporary parameter table is used to temporarily store the pressure depth adjustment data to be effective. Each record in the table corresponds one-to-one with a coordinate position in the placement machine control table, retaining both the original baseline pressure depth field and the optimized pressure depth field to be replaced. When a new printed circuit board (PCB) enters the pick-and-place machine, the controller reads the PCB's product serial number and retrieves the corresponding product model from the product traceability table. It then compares this model with the product model corresponding to the serial number of the previous production batch. If the comparison results indicate that they belong to the same product model, before the placement program begins, the baseline pressing depth value at the corresponding coordinate position in the pick-and-place machine control table is replaced with the optimized pressing depth value stored in the temporary parameter table. This ensures that the nozzle presses down according to the updated pressing depth when performing the component placement action at that coordinate position, thus completing the parameter application for subsequent production processes of the same PCB model.

[0049] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. A method for optimizing SMT yield based on multi-source data analysis, characterized in that, Includes the following steps: After solder paste printing, test data is obtained, which includes the three-dimensional volume measurement value of each solder paste dot and the coordinate position of the solder paste dot on the printed circuit board. Obtain the measured values ​​of the pressing depth during the electronic component placement process, and store each measured value of the pressing depth in association with the coordinate position of a solder paste dot; Establish a spatial mapping table between the test data and the measured pressure depth. The spatial mapping table uses the product serial number of the printed circuit board as the primary key. The coordinate positions in the test data corresponding to the same product serial number are paired with the coordinate positions in the measured pressure depth and then stored. Traverse the spatial mapping table. For each pair, when the three-dimensional volume measurement value of the solder paste dot is greater than the preset upper limit threshold and the corresponding measured value of the pressing depth is less than the preset lower limit threshold, mark the pair as a pair to be optimized and extract the coordinate position of the solder paste dot of the pair to be optimized. The average volume deviation is obtained by subtracting the preset standard volume value from the average of the three-dimensional volume measurement values ​​of all pairs to be optimized. The average volume deviation is multiplied by the preset depth compensation coefficient to obtain the depth compensation value. The depth compensation value is added to the reference pressing depth value at the corresponding coordinate position in the pick and place machine control table to generate the optimized pressing depth value. The optimized pressure depth value is written into the parameter storage area of ​​the pick-and-place machine. When the pick-and-place machine subsequently produces the same type of printed circuit board, it uses the optimized pressure depth value to perform component placement at the corresponding coordinate position.

2. The SMT yield optimization method based on multi-source data analysis according to claim 1, characterized in that, The obtained detection data includes: A data export command is sent to the solder paste inspection equipment. In response to the data export command, the solder paste inspection equipment packages the inspection data of the current production batch into a data file and transmits it to the local server. After parsing the data file, the local server extracts the three-dimensional volume measurement value of each solder paste point. The three-dimensional volume measurement value includes the height measurement value, area measurement value, and volume estimate calculated by multiplying the height measurement value and area measurement value of the solder paste point. At the same time, the horizontal and vertical coordinate values ​​of the solder paste point on the printed circuit board are extracted to form the coordinate position.

3. The SMT yield optimization method based on multi-source data analysis according to claim 2, characterized in that, The specific process for obtaining the measured value of the compression depth is as follows: The position change curve of the nozzle during the placement process is read from the encoder feedback of the pick-and-place machine's servo motor. The moment when the nozzle contacts the solder paste surface is read from the sampling data of the pick-and-place machine's pressure sensor. The displacement difference between the starting point and the ending point is calculated as the measured value of the pressing depth. The coordinate position of the nozzle at the moment of placement is read from the motion controller of the pick-and-place machine. The coordinate position is matched with the measured value of the pressing depth and then written into the temporary storage area.

4. The SMT yield optimization method based on multi-source data analysis according to claim 3, characterized in that, Establishing a spatial mapping table includes: Using the product serial number of the printed circuit board as the search keyword, all data records corresponding to the product serial number are queried in the test data storage area and the measured pressure depth storage area respectively. The coordinate positions in the queried test data are compared point by point with the coordinate positions in the measured pressure depth. When the difference between the horizontal coordinate and the vertical coordinate of the two coordinate positions is less than the preset positioning tolerance, they are determined to be the same solder paste point. The three-dimensional volume measurement value and the measured pressure depth value of the solder paste point are stored in the same row of the mapping table.

5. The SMT yield optimization method based on multi-source data analysis according to claim 4, characterized in that, The preset upper limit threshold for volume is determined by statistically analyzing the three-dimensional volume measurements of all solder paste points in historical production batches, and the 85th percentile of the obtained three-dimensional volume measurements is used as the upper limit threshold for volume. The preset lower limit threshold for pressure depth is determined by statistically analyzing the measured pressure depth of all solder paste points in historical production batches, and the 15th percentile of the obtained measured pressure depth is used as the lower limit threshold for pressure depth. When traversing the spatial mapping table, the three-dimensional volume measurements and measured pressure depths are read row by row and compared with the upper limit threshold for volume and the lower limit threshold for pressure depth, respectively.

6. The SMT yield optimization method based on multi-source data analysis according to claim 5, characterized in that, The specific process for calculating the average of the three-dimensional volume measurements of all pairs to be optimized is as follows: count the number of rows marked as pairs to be optimized in the spatial mapping table, sum the three-dimensional volume measurements of all pairs to be optimized to obtain the total volume, divide the total volume by the number of rows to obtain the average of the three-dimensional volume measurements, the preset standard volume value is the target volume value read from the solder paste printing process specification, and the preset depth compensation coefficient is the coefficient value obtained by fitting the regression analysis of the mounting yield data of the same model of printed circuit board in historical production batches.

7. The SMT yield optimization method based on multi-source data analysis according to claim 6, characterized in that, The specific process of writing the optimized pressing depth value into the parameter storage area of ​​the pick-and-place machine is as follows: the optimized pressing depth value and the corresponding coordinate position are combined into a parameter update record, and the parameter update record is sent to the controller of the pick-and-place machine. The controller writes the parameter update record into a temporary parameter table in the memory of the pick-and-place machine. When the controller detects that the product serial number of the printed circuit board belongs to the same product model as the product serial number of the previous production batch, it replaces the reference pressing depth value of the corresponding coordinate position in the pick-and-place machine control table with the optimized pressing depth value stored in the temporary parameter table before the placement program starts to execute.

Images