Mini LED PCB solder mask layer quality regulation and control method based on intelligent detection feedback system

Abstract

The invention belongs to the technical field of printed circuit board manufacturing, and particularly relates to a Mini LED PCB solder mask layer quality regulation and control method based on an intelligent detection feedback system. The invention aims to solve the quality problems of serious apparent chromatic aberration, poor surface flatness, inconsistent ink thickness, deviation of solder mask windowing size and position and easy generation of scratches on the ink surface existing in the traditional Mini LED PCB solder mask. The method comprises the following steps: acquiring surface information data of the Mini LED PCB solder mask to be detected; intelligently analyzing the surface information data to generate a quality evaluation result and a deviation value; according to the technical scheme, the thickness uniformity and the surface flatness of the solder mask layer and the size precision and the consistency of solder mask windowing can be remarkably improved, surface defects are effectively prevented and reduced, excellent optical performance, electrical stability and long-term operation stability of a MiniLED display product are guaranteed, and meanwhile the production efficiency and the product yield are improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of printed circuit board manufacturing, and particularly relates to a MiniLED PCB solder resist layer quality regulation method based on an intelligent detection feedback system. BACKGROUND

[0002] With the rapid development of MiniLED display technology, it has been widely used in high-end display fields due to its significant advantages such as high brightness, high contrast, high color gamut and low power consumption. As the core carrier of the MiniLED display module, the manufacturing quality of the printed circuit board (PCB) directly affects its final performance. Among them, the solder resist layer, as a key process link in PCB manufacturing, plays a role in protecting the circuit, preventing short circuits, and defining the soldering area, which is crucial to the uniformity and stability of the MiniLED display effect.

[0003] Among them, the MiniLED technology puts forward higher requirements for the manufacturing precision and surface quality of the PCB solder resist layer. Due to the small size and dense arrangement of MiniLED chips, there are strict standards for the thickness uniformity, surface flatness and window size consistency of the solder resist layer to ensure the light transmission efficiency, electrical connection reliability and overall display uniformity. A high-quality solder resist layer is the basis for MiniLED products to achieve excellent optical performance and long-term stable operation.

[0004] However, the existing solder resist layer manufacturing process, especially the traditional liquid solder resist ink printing method, has significant technical bottlenecks. Due to the liquid coating characteristics of the ink, it is easily affected by the flowability, resulting in inconsistent ink thickness on the dielectric layer and copper surface, uneven solder resist layer surface, and difficulty in accurately controlling the overall flatness. Such thickness inconsistency and surface defects not only directly cause serious problems of MiniLED PCB surface color difference, but also further cause inconsistent and offset of the solder resist window size in the matrix arrangement, exacerbating the visual color difference phenomenon. In addition, the uneven ink surface will cause light scattering, reducing the amount of light reaching the display layer, thereby reducing display brightness, affecting energy efficiency, and causing differences in display effect in different areas, damaging the overall uniformity and quality of the picture. More than that, it may cause poor contact between the electrode and the circuit, increase the resistance, affect the current transmission, and thus cause instability of LED chip brightness and color, and even cause flickering or extinguishing failure. At the same time, the ink surface lacks effective protection, and is easily scratched or foreign object reverse sticking during manufacturing and operation, further reducing product yield and reliability. The above problems seriously restrict the high-quality manufacturing and performance stability of MiniLED display products, and need to be solved urgently. SUMMARY

[0005] In order to solve the problems of serious apparent color difference, poor surface flatness, inconsistent ink thickness, position deviation of solder mask layer of MiniLED PCB in traditional PCB manufacturing process, and the problem of easy scratch on the surface of solder mask layer, these problems not only lead to the decrease of brightness, energy efficiency, uneven display effect, instability of LED chip brightness and color, and even failure of MiniLED display products, but also affect the reflectivity and overall visual experience of the product, a MiniLED PCB solder mask layer quality regulation method based on intelligent detection feedback system is proposed. The method realizes the accurate identification, quantitative evaluation and real-time regulation of the manufacturing process parameters of the solder mask layer by introducing advanced intelligent detection technology and closed-loop feedback mechanism, thereby significantly improving the thickness uniformity, surface flatness, and size accuracy and consistency of solder mask layer, and ultimately ensuring the excellent optical performance and long-term operation stability of MiniLED display products.

[0006] According to an aspect of the present application, a MiniLED PCB solder mask layer quality regulation method based on an intelligent detection feedback system is provided, which comprises: obtaining surface information data of the MiniLED PCB solder mask layer to be detected; intelligently analyzing the surface information data to generate a quality evaluation result, the quality evaluation result including a deviation amount between the solder mask layer quality parameter and the preset quality standard; determining a regulation instruction of the solder mask layer manufacturing process according to the quality evaluation result and the deviation amount; executing the regulation instruction to optimize and adjust the solder mask layer manufacturing process.

[0007] Compared with the prior art, the MiniLED PCB solder mask layer quality regulation method based on the intelligent detection feedback system provided by the application can realize real-time and high-precision monitoring and quantitative evaluation of the thickness uniformity, surface flatness, anti-solder window size and position key quality parameters of the MiniLED PCB solder mask layer by constructing an integrated intelligent detection and feedback system. The method uses advanced image processing and machine learning algorithms to accurately identify various defects in the solder mask layer, including apparent color difference, local unevenness, window offset, surface scratches, and calculates the quality deviation. Based on these accurate detection and evaluation results, the system can intelligently generate regulation instructions to dynamically and adaptively optimize and adjust the key parameters in the solder mask manufacturing process, including ink viscosity, printing pressure, scraper angle, printing speed, curing temperature and curing time. Through this closed-loop feedback mechanism, the method effectively solves the problem of inconsistent thickness and poor flatness caused by ink fluidity in traditional processes, significantly improves the uniformity and flatness of the MiniLED PCB solder mask layer, eliminates the apparent color difference, ensures the consistency and position accuracy of the anti-solder window size, and reduces the probability of surface defects. Finally, the method greatly improves the optical performance, electrical stability and overall reliability of the MiniLED display product, providing higher quality visual experience for end users, while improving production efficiency and product yield.

[0008] The embodiments of the application will be described in detail below.

[0009] In the specific implementation of the method, first, the step of obtaining surface information data of the Mini LED PCB solder mask to be detected is performed. This step comprehensively and non-destructively scans and collects data on samples that have already applied a solder mask but have not yet fully cured or have cured on the Mini LED PCB by integrating a high-precision visual detection module and a three-dimensional topography measurement module. The visual detection module uses a high-resolution industrial camera combined with multi-spectral imaging technology to capture two-dimensional image data of the solder mask surface, which is used to analyze the apparent color uniformity, defect types including foreign matter, bubbles, scratches, and the two-dimensional geometric features of the solder mask window. The three-dimensional topography measurement module uses a laser confocal microscope or a white light interferometer non-contact optical measurement device to accurately obtain three-dimensional point cloud data or height map data of the solder mask surface, which is used to quantify the local thickness, overall flatness, roughness, and perpendicularity of the solder mask window edge. During the surface information data collection process, the stability and uniformity of the light source need to be ensured to avoid external environmental factors interfering with the detection results. The collected raw data usually includes high-resolution image files, three-dimensional depth maps, point cloud coordinate data, and spectral reflectance data. The surface information data is preliminarily converted and stored in data format, laying the foundation for subsequent intelligent analysis. The data collection frequency is adjusted according to the production line's rhythm to achieve real-time or quasi-real-time detection.

[0010] After obtaining the surface information data, the method enters the step of intelligent analysis of the surface information data to generate quality assessment results. This step is the key link of the method, which uses artificial intelligence algorithms to deeply mine and analyze massive surface information data. First, the original surface information data is preprocessed, including image denoising, contrast enhancement, color correction, three-dimensional point cloud data filtering and registration, to eliminate noise and errors in the data collection process and ensure the accuracy and consistency of the data. Then, the key feature quantities of the solder mask are extracted. For two-dimensional image data, a convolutional neural network deep learning model is used to automatically identify and segment the solder mask window area, defect area, and areas of different colors or brightness, and calculate their area, perimeter, centroid coordinates, and color distribution histogram feature parameters. For three-dimensional topography data, the root mean square roughness, peak-valley difference, thickness distribution statistics of specific areas, and local slope indicators are calculated to quantify the flatness and thickness uniformity of the solder mask.

[0011] The intelligent analysis step further utilizes machine learning or deep learning models to compare and evaluate the extracted feature parameters against pre-set Mini LED PCB solder mask quality standards. The quality standards are trained based on a large number of qualified and unqualified product samples and cover solder mask thickness tolerance, surface flatness requirements, anti-pad window size tolerance, position deviation range, and acceptable defect types and quantities. The intelligent analysis model is a multi-classifier or regression model, and its output is a quantitative evaluation result of the quality of the solder mask in various aspects. For example, the evaluation result explicitly indicates whether the local thickness of the solder mask exceeds the range, whether the surface has micron-level unevenness, whether the anti-pad window has millimeter-level deviation, whether there are invisible fine scratches or foreign matter attached, and quantifies the size of these deviations. The quality evaluation result not only contains specific numerical values of various quality parameters, but also contains deviation amounts from the pre-set quality standards, which explicitly indicate the degree and direction of the deviation of the current solder mask quality from the standard. For example, if the local thickness of the solder mask is detected to be five percent thinner, the system will explicitly report this deviation. The intelligent analysis process can achieve high-precision, high-efficiency, and automated identification and diagnosis of solder mask quality problems, overcoming the subjectivity and efficiency bottleneck of manual detection.

[0012] Next, the method executes a step of determining control instructions for the solder mask manufacturing process based on the quality evaluation result and the deviation amount. This step is the core of realizing closed-loop quality control. The quality evaluation result and the deviation amount output by the intelligent analysis module are input into the decision-making module. The decision-making module integrates an expert system or an intelligent control algorithm based on reinforcement learning, which has established a mapping relationship between the solder mask quality parameters and the manufacturing process parameters through learning and training on historical production data and process control experience. For example, if the intelligent analysis result shows that the solder mask is overall thin and the flatness is poor, the decision-making module will generate instructions to adjust the pressure or printing speed of the printing machine doctor blade and suggest increasing the ink viscosity according to experience or model calculation. If systematic deviation of the anti-pad window is detected, the decision-making module will instruct to adjust the printing alignment system or the installation position of the mask plate. If micron-level unevenness appears on the surface, the decision-making module may instruct to optimize the temperature curve or curing time of the curing oven to ensure that the ink is fully leveled. The specific content of the control instructions includes: printing pressure adjustment amount, doctor blade angle adjustment amount, printing speed adjustment amount, ink viscosity adjustment suggestion, alignment system compensation amount, mask plate calibration instruction, curing oven temperature setting, and curing time setting. The generation of control instructions is dynamic and real-time, and can be adjusted according to the latest detection data to ensure that the optimization of process parameters is targeted and timely.

[0013] Finally, the method enters the step of executing the control instructions to optimize and adjust the solder mask manufacturing process. This step converts the control instructions generated by the decision-making module into actual physical operations, which act on the relevant equipment on the Mini LED PCB solder mask production line. The actuator module connects with the printing machine, curing oven, automatic alignment system, and ink supply system production equipment through standard industrial communication interfaces, including Modbus, Ethernet / IP, and Profinet. When receiving the control instructions, the actuator module drives the corresponding device components to make precise adjustments. For example, the printing pressure adjustment amount in the instructions can precisely change the pressure of the printing doctor blade mechanism on the PCB through a servo motor; the doctor blade angle adjustment amount can precisely adjust the inclination angle of the doctor blade through a stepper motor; the printing speed adjustment amount can directly modify the running speed of the printing machine; the alignment system compensation amount can drive a high-precision motion platform to correct the micron-level displacement of the PCB or mask; and the curing parameter adjustment instructions can precisely control the temperature distribution and heating time in the curing oven through the temperature control system.

[0014] Optimization and adjustment is not a one-time process, but rather a continuous closed-loop feedback cycle throughout the production process. After the process parameters are adjusted, the subsequent produced Mini LED PCB solder mask will again be subjected to quality detection by the detection module, and the new surface information data will be input into the intelligent analysis module for evaluation. If the detection results show that the quality has not yet met the preset standards or there are new deviations, the system will generate and execute new control instructions again until the solder mask quality meets all requirements. This iterative optimization and adaptive adjustment capability enables the entire solder mask manufacturing process to respond to production fluctuations and environmental changes in real time, continuously approaching the optimal state.

[0015] Through the Mini LED PCB solder mask quality control method based on the intelligent detection feedback system, the manufacturing precision and quality of the Mini LED PCB solder mask can be significantly improved. Specifically, this method can: First, it significantly improves the thickness uniformity and surface flatness of the solder mask. By accurately measuring and intelligently analyzing the three-dimensional topography of the solder mask, and combining dynamic control of key parameters such as printing pressure and ink viscosity, the problem of inconsistent ink thickness on the dielectric layer and copper caused by fluidity in traditional liquid ink printing is effectively suppressed, and the unevenness of the solder mask surface is eliminated, achieving micron-level flatness control. This directly solves the problem of severe apparent color difference and reduces light scattering, increasing the amount of light reaching the display layer, thereby improving the brightness and energy efficiency of Mini LED display products and ensuring the consistency of display effects in different areas.

[0016] Second, precise control of the size accuracy and position consistency of the solder mask window. The intelligent vision detection module can accurately identify the two-dimensional geometric features and position information of the solder mask window. Combined with intelligent analysis algorithms, it can quickly find problems of inconsistent window size or position deviation. Through feedback control of the positioning system or mask, precise correction of the solder mask window can be achieved to ensure that its size and position meet the design requirements, thereby avoiding visual color difference caused by window deviation and ensuring stable connection and uniform light emission of the LED chip.

[0017] Third, effectively prevent and reduce surface defects. The intelligent detection system can identify various small defects on the solder mask surface, including scratches, foreign matter attachment, and bubbles. Through real-time feedback, process parameters can be adjusted in time or operators can be reminded to intervene, including optimizing the curing process to improve the surface hardness of the ink and reduce the risk of scratches, or discovering and handling ink or equipment cleaning problems at an early stage to reduce the generation of foreign matter or bubbles from the source. This improves the yield and reliability of MiniLED PCBs.

[0018] Fourth, improve the stability and reliability of electrical connections. A smooth and uniform solder mask and precise windowing can ensure good contact between the electrodes and the circuit, reduce contact resistance, and ensure uniform transmission and distribution of current. This directly avoids the problem of unstable brightness and color of LED chips, reduces the occurrence of flickering and extinguishing failures, and improves the long-term operation stability of MiniLED products.

[0019] Fifth, optimize the overall optical reflectivity of MiniLED products. A uniform and defect-free solder mask surface can achieve more uniform light reflection, avoiding excessive or weak reflection in some areas, thereby eliminating color distortion and making the displayed image or video brightness uniform and color pure, greatly improving the visual experience of MiniLED display products.

[0020] Sixth, realize automation and intelligentization of the production process. This method integrates detection, analysis, and control into one, forming an efficient and self-adaptive closed-loop control system, reducing dependence on human experience and human error, and significantly improving the automation level and production efficiency of MiniLED PCB solder mask manufacturing. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 is the overall technical scheme architecture diagram of the present application; Figure 2 is the core principle framework diagram of the MiniLED PCB solder mask quality control based on the intelligent detection feedback system in the present application. DETAILED DESCRIPTION

[0022] Please refer to Figure 1 and Figure 2The application provides a MiniLED PCB solder mask quality regulation method based on an intelligent detection feedback system. The method realizes accurate identification, quantitative evaluation and real-time regulation of the solder mask manufacturing process parameters by introducing advanced intelligent detection technology and a closed-loop feedback mechanism, thereby significantly improving the thickness uniformity, surface flatness, and size accuracy and consistency of the solder mask window, ultimately ensuring the excellent optical performance and long-term operational stability of MiniLED display products. The method constructs an integrated intelligent detection and feedback system that monitors and quantitatively evaluates the thickness uniformity, surface flatness, solder mask window size and position key quality parameters of MiniLED PCB in real time with high precision. The method uses advanced image processing and machine learning algorithms to accurately identify various defects in the solder mask, including apparent color difference, local unevenness, window offset, and surface scratches, and calculates the quality deviation. Based on these accurate detection and evaluation results, the system can intelligently generate regulation instructions to dynamically and adaptively optimize and adjust key parameters in the solder mask manufacturing process, including ink viscosity, printing pressure, scraper angle, printing speed, curing temperature, and curing time. Through this closed-loop feedback mechanism, the method effectively solves the problems of inconsistent thickness and poor flatness caused by ink fluidity in traditional processes, significantly improves the uniformity and flatness of MiniLED PCB solder mask, eliminates apparent color difference, ensures the consistency and position accuracy of solder mask window size, and reduces the probability of surface defects. Ultimately, the method greatly improves the optical performance, electrical stability and overall reliability of MiniLED display products, providing end users with a higher quality visual experience, while improving production efficiency and product yield.

[0023] According to one embodiment of the application, the MiniLED PCB solder mask quality regulation method based on an intelligent detection feedback system includes the following steps: Obtain surface information data of the MiniLED PCB solder mask to be detected; Intelligently analyze the surface information data to generate a quality evaluation result, which includes the deviation between the solder mask quality parameters and the preset quality standard; Determine the regulation instructions for the solder mask manufacturing process according to the quality evaluation result and the deviation; Execute the regulation instructions to optimize and adjust the solder mask manufacturing process.

[0024] In the implementation of the method, first, the step of acquiring surface information data of the Mini LED PCB solder mask to be detected is performed. It should be understood that the purpose of this step is to achieve comprehensive, high-precision, non-contact data acquisition of the surface quality of the Mini LED PCB solder mask, to provide accurate and reliable raw data for subsequent intelligent analysis and quality control. In the specific implementation process, this step comprehensively scans and collects data on the sample of the Mini LED PCB on which the solder mask has been applied but has not yet fully solidified or has solidified by integrating a high-precision visual detection module and a three-dimensional topography measurement module.

[0025] Specifically, acquiring surface information data of the Mini LED PCB solder mask to be detected includes the following detailed operations: First, the system startup and self-checking operation is performed to ensure that all detection devices and sensors are in normal working condition. This includes checking the lens focal length, aperture setting, and image sensor cleanliness of the high-resolution industrial camera; checking the laser output power, detector sensitivity of the laser confocal microscope or white light interferometer, and the accuracy of the Z-axis scanning platform; checking the connection stability of the data acquisition card and the processing unit. Any detected device abnormality will trigger an alarm and suspend the data collection process until the problem is resolved to avoid collecting invalid or incorrect data.

[0026] Second, the Mini LED PCB sample to be detected is accurately positioned and clamped at the specified position of the detection platform by the automated conveying system. The positioning system usually adopts a high-precision machine vision guided mechanical arm or linear module to ensure that the positioning accuracy of the PCB board in the X, Y, Z axes and the rotation direction reaches the micron level, to ensure the consistency and repeatability of subsequent detection. The clamping device uses a vacuum suction cup or a pneumatic clamp to avoid any mechanical damage to the surface of the Mini LED PCB.

[0027] Subsequently, light source regulation and environmental stabilization operations are performed. The light source system uses a multi-mode programmable light source, including a combination of ring light, coaxial light, and backlight, and combines multi-spectral imaging technology to optimize the lighting of different areas and features. The brightness, color temperature, and incident angle of the light source are dynamically adjusted according to the features to be detected. For example, low-angle diffuse reflection light may be used to highlight defects when detecting surface scratches, while high-contrast coaxial light may be used when detecting anti-solder window edges. To eliminate external environmental light interference, the entire detection area is encapsulated in a lightproof box, and environmental parameters are monitored in real time by temperature and humidity sensors to ensure that the temperature fluctuation does not exceed plus or minus zero point five degrees Celsius, and the relative humidity fluctuation does not exceed plus or minus five percent, to prevent thermal expansion and contraction or water vapor condensation from affecting the measurement accuracy.

[0028] After the light source and the environment are stabilized, the visual inspection module starts a high-resolution industrial camera combined with multi-spectral imaging technology to capture two-dimensional image data of the solder mask surface. The industrial camera usually uses a complementary metal-oxide-semiconductor sensor with a resolution of no less than 25 million pixels, equipped with a telecentric lens to eliminate perspective distortion and ensure measurement accuracy. Multi-spectral imaging technology can more carefully analyze the apparent color uniformity of the solder mask by collecting multiple narrow-band images in the visible light band (e.g., 400-700 nm) and near-infrared band (e.g., 800-1000 nm), and identify tiny color differences that the naked eye cannot detect. For example, by calculating the ratio or difference of reflectivity at different wavelengths, the consistency of the solder mask color is quantified. At the same time, this module is used to identify the two-dimensional geometry of foreign object attachment, micro-bubbles, and fine scratch surface defects, and accurately measure the area, perimeter, circularity, and barycenter coordinates of the solder mask window, with a measurement accuracy of sub-micron level.

[0029] Next, the three-dimensional topography measurement module uses a laser confocal microscope or a white light interferometer non-contact optical measurement device to accurately obtain three-dimensional point cloud data or height map data of the solder mask surface. The laser confocal microscope uses a scanning laser beam to scan the sample surface point by point, and removes out-of-focus light through a pinhole filter to achieve high-resolution three-dimensional topography measurement, with a Z-axis measurement accuracy of tens of nanometers. The white light interferometer uses the interference principle of a broadband light source to obtain three-dimensional topography data by scanning the sample height and analyzing the change of interference fringes. These devices are used to quantify the local thickness, overall flatness, root mean square roughness (e.g., below 100 nm), peak-to-valley difference, and perpendicularity of the solder mask window edge. For example, by comparing the height difference between the solder mask area and the adjacent copper foil area, the local thickness of the solder mask is accurately calculated.

[0030] After data acquisition is completed, the raw data collected will be transmitted and preliminarily processed. The raw data usually includes high-resolution image files (stored in TIF or PNG format), three-dimensional depth maps (stored in floating-point matrix format), point cloud coordinate data (stored in XYZ or PLY format), and spectral reflectance data (stored in multi-channel image format). These data are transmitted to the edge computing unit through high-speed industrial Ethernet (including Gigabit Ethernet) for preliminary data format conversion, compression, and storage. Data format conversion unifies the proprietary formats output by different devices into the system internal standard format, to facilitate subsequent analysis. Preliminary data storage usually uses a distributed file system or a high-performance database, and establishes detailed metadata index to record the acquisition time, PCB batch number, and detection station information, laying a foundation for subsequent intelligent analysis.

[0031] The data acquisition frequency is dynamically adjusted according to the production line's tempo to achieve real-time or quasi-real-time detection. For example, for a production line that produces one PCB per second, the data acquisition system needs to complete all image and three-dimensional data acquisition, transmission, and preliminary storage within one second. Through optimization of hardware configuration and software algorithms, the timeliness of data acquisition is ensured, and the production process is not bottlenecked.

[0032] After obtaining the surface information data, the method enters the step of intelligently analyzing the surface information data to generate quality evaluation results. It should be understood that this step is the key core link of the method, which aims to use advanced artificial intelligence algorithms to deeply mine and analyze massive, high-dimensional surface information data, convert the original data into quantitative quality evaluation results, and clearly indicate the deviation from the preset quality standard, thereby achieving high-precision, high-efficiency, and automated identification and diagnosis of MiniLED PCB solder mask quality problems. This step effectively overcomes the subjectivity and efficiency bottleneck of traditional manual detection, providing accurate and objective basis for subsequent process control.

[0033] Specifically, intelligently analyzing the surface information data to generate quality evaluation results includes the following detailed operations: First, the original surface information data is preprocessed. This operation aims to eliminate noise and errors in the data acquisition process and ensure the accuracy and consistency of the data. For two-dimensional image data, preprocessing includes image denoising using a non-local mean denoising algorithm or Gaussian filtering to suppress random noise and smooth the image; enhancing image contrast through adaptive histogram equalization or contrast limited adaptive histogram equalization algorithm to highlight solder mask texture, defects, and window edge; through color correction algorithm based on color calibration board or standard color block, compensate for color deviation caused by uneven light source or camera response, ensure the authenticity of color information. For three-dimensional point cloud data, preprocessing includes data filtering using statistical filtering (such as outlier removal) or voxel downsampling to remove noise points and redundant points and reduce data volume; through iterative closest point algorithm or principal distance registration algorithm, the three-dimensional data from different angles or different measurement modules are unified to the same coordinate system to ensure the integrity and accuracy of the overall topography information. The parameters of these preprocessing operations (including filter kernel size, threshold) will be adaptively adjusted according to data characteristics and application scenarios.

[0034] Secondly, the solder mask key feature extraction operation is performed. This operation aims to extract key indicators that can quantify the quality of the solder mask from the pre-processed surface information data. For two-dimensional image data, a pre-trained convolutional neural network (including U-Net or MaskR-CNN deep learning models) is used to automatically identify and accurately segment the solder mask opening area, foreign matter, bubble, scratch defect area, and areas of different colors or brightness. After completing the area segmentation, the system will calculate the geometric feature parameters of each identified area, including area (e.g., in square microns), perimeter (e.g., in microns), centroid coordinates (in pixels or physical coordinates), and color distribution histogram (e.g., pixel count at each of the 256 gray levels in the red, green, and blue channels). For three-dimensional topography data, a series of indicators are calculated to quantify the flatness and thickness uniformity of the solder mask, including root mean square roughness (e.g., the root mean square value of less than 100 nanometers within a 500 micrometer by 500 micrometer area), peak-to-valley difference (i.e., the height difference between the highest and lowest points on the surface, e.g., less than 1 micron), thickness distribution statistics (including average thickness, thickness standard deviation) of specific areas (e.g., inside the solder mask opening or on the copper area of the solder mask), and local slope (e.g., the maximum angle between the surface normal and the vertical direction within a certain sampling distance).

[0035] Next, the extracted feature parameters are compared and evaluated with pre-set MiniLED PCB solder mask quality standards using machine learning or deep learning models (including: Convolutional Neural Network (CNN) for image defect recognition, Support Vector Machine (SVM) and Random Forest for classification, XGBoost for regression prediction of deviation amount, U-Net for pixel-level segmentation detection, and Autoencoder for anomaly detection existing algorithms). The quality standards are based on a large number of qualified and unqualified product samples, which are strictly trained and verified, and cover the solder mask thickness tolerance (e.g. design thickness plus or minus 5%), surface flatness requirements (e.g. less than 3 microns of peak-to-valley difference on the entire board), anti-solder window size tolerance (e.g. design size plus or minus 10 microns), position deviation range (e.g. less than 20 microns in X and Y directions relative to the reference point), and acceptable defect type (e.g. foreign particle diameter less than 5 microns, scratch length less than 50 microns) and quantity threshold. The intelligent analysis model is a multi-classifier model (e.g. for judging defect type) or a regression model (e.g. for predicting deviation amount), and its output is a quantitative evaluation result of the quality of the solder mask in various aspects. For example, the evaluation result clearly indicates whether the local thickness of the solder mask exceeds the design range (e.g. detects a 8% thinning), whether there is micron-level unevenness on the surface (e.g. root mean square roughness reaches 120 nanometers), whether there is millimeter-level deviation of the anti-solder window (e.g. the center point is deviated by 30 microns in X direction), whether there is a microscopic scratch or foreign matter attachment invisible to the naked eye, and quantifies the size and severity of these deviations.

[0036] The quality evaluation result not only contains specific numerical values of various quality parameters, such as the average thickness of the solder mask being 23.5 microns, the root mean square roughness being 87 nanometers, and the length and width of the anti-solder window No. 1 being 98 microns and 102 microns respectively, but also explicitly includes the deviation amount from the pre-set quality standard, which accurately indicates the degree and direction of the deviation of the current solder mask quality from the standard. For example, if the local thickness of the solder mask is detected to be 5% thinner, the system will clearly report a deviation amount of -5%; if the area of the anti-solder window is 2% smaller than the standard area, the system will report a deviation amount of -2%. The intelligent analysis process can achieve high-precision, high-efficiency, and automated identification and diagnosis of solder mask quality problems, significantly overcoming the subjectivity, low efficiency, and lack of ability to identify small defects of manual detection, ensuring the objectivity, consistency, and traceability of quality evaluation.

[0037] Subsequently, the method performs a step of determining the adjustment instruction of the solder mask manufacturing process based on the quality evaluation result and the deviation amount. It should be understood that this step is the core link of realizing closed-loop quality adjustment, and its purpose is to convert the quantitative quality evaluation result and the accurate deviation amount output by the intelligent analysis module into specific and executable process parameter adjustment instructions, so as to correct the deviation in the solder mask manufacturing process and make the product quality return to the preset standard. The intelligent decision mechanism of this step can realize dynamic and self-adaptive optimization of the production process, avoiding the subjectivity and hysteresis of manual experience judgment.

[0038] Specifically, determining the adjustment instruction of the solder mask manufacturing process includes the following detailed operations: Firstly, the intelligent analysis module outputs the quality evaluation result and the detailed deviation amount as data input, which is transmitted to the decision module through a high-speed data bus. The decision module integrates a high-level expert system (including a rule-based reasoning engine (such as CLIPS, Drools), a knowledge graph, and a fuzzy logic controller existing technology, which realizes intelligent diagnosis and adjustment decision of manufacturing abnormalities by coding process expert experience and historical data rules) or an intelligent control algorithm based on deep reinforcement learning (including DQN (deep Q network), PPO (proximal policy optimization), and DDPG (deep deterministic policy gradient) existing algorithms, which learn the optimal control strategy by interacting with the environment). The expert system constructs a detailed rule base, which is extracted and summarized from a large amount of historical production data, practical experience of process engineers, and operation manuals of production equipment. For example, the rule base contains conditional statements such as “if the overall solder mask is 5% thinner and the root mean square roughness exceeds 100 nanometers, then suggest increasing the printing pressure by 3% and reducing the printing speed by 5%”. The intelligent control algorithm based on deep reinforcement learning learns to take what kind of process parameter adjustment action can maximize product pass rate and production efficiency by interacting with the production line digital twin model under different quality deviation states.

[0039] Secondly, the decision module deeply analyzes the received quality evaluation result and deviation amount to accurately identify the type, severity, location, and potential cause of the quality problem. For example, by analyzing the thickness distribution graph in the three-dimensional topography data, the system determines whether it is a global thinning or a local area unevenness; by comparing the centroid coordinates of the solder mask window, the system identifies whether it is a systematic deviation or a random deviation. Based on these detailed analysis results, the decision module queries its built-in knowledge base or executes the reasoning algorithm of the model to generate preliminary adjustment suggestions.

[0040] Subsequently, the decision module generates targeted regulation instructions according to the identified quality problem types and severity. The generation of regulation instructions is dynamic and real-time, which can be adjusted according to the latest detection data, ensuring that the optimization of process parameters has high pertinence and timeliness.

[0041] For example, if the intelligent analysis result shows that the overall solder mask layer is 4.5% thinner and the surface flatness is poor (for example, the peak-valley difference reaches 4 microns), the decision module will generate instructions to adjust the squeegee pressure of the printing machine by increasing it by 3%, and at the same time suggest increasing the ink viscosity by 5%. These parameter adjustments aim to increase the amount of ink deposited and improve its leveling performance.

[0042] If it is detected that there is a systematic deviation in the matrix arrangement of the solder mask window (for example, all windows are deviated by 30 microns in the positive direction of the X axis), the decision module will instruct to adjust the printing alignment system, such as by driving a high-precision motion platform to perform a reverse X-axis compensation displacement of 30 microns on the MiniLED PCB or mask plate, to correct the position accuracy of the window.

[0043] If the solder mask layer surface appears micron-level unevenness (for example, the local area root mean square roughness exceeds 150 nanometers), but the overall thickness is normal, the decision module may instruct to optimize the temperature curve of the curing oven, such as increasing the temperature of the pre-curing stage by 2 degrees Celsius and extending the curing time by 10 seconds, to ensure that the ink is fully leveled and the unevenness caused by surface tension is eliminated.

[0044] The specific content of the regulation instructions covers multiple key parameters in the solder mask manufacturing process, including: printing pressure adjustment amount (such as a pressure range of plus or minus 10%), squeegee angle adjustment amount (such as an angle range of plus or minus 2 degrees), printing speed adjustment amount (such as a speed range of plus or minus 15%), ink viscosity adjustment suggestion (such as a suggestion to increase or decrease the viscosity by 5% to 10%), alignment system compensation amount (such as X, Y direction displacement of plus or minus 50 microns), mask plate calibration instruction, curing oven temperature setting (such as pre-curing temperature 120 degrees Celsius, main curing temperature 150 degrees Celsius), curing time setting (such as pre-curing time two minutes, main curing time three minutes). Each instruction is accompanied by specific adjustment values, units and expected effects.

[0045] In the process of generating regulation instructions, the decision module will also perform instruction conflict resolution and priority sorting. For example, if multiple quality problems are detected at the same time that require different process adjustments, the system will optimize the combination of instructions according to the pre-set priority rules and mutual influence, ensuring the coordination and effectiveness of the adjustments, and avoiding the generation of new problems. Finally, the generated regulation instructions will be output through a standardized interface and recorded in detail for subsequent execution and traceability.

[0046] Finally, the method enters the step of executing the control instruction to optimize and adjust the solder mask manufacturing process. It should be understood that this step converts the abstract control instruction generated by the aforementioned decision module into actual physical operations, which directly act on the relevant equipment on the Mini LED PCB solder mask production line, thereby realizing precise and automated adjustment of the manufacturing process parameters to correct the identified quality deviations. This step is the execution core of the closed-loop feedback system, ensuring the landing of intelligent decisions and the actual optimization of the process.

[0047] Specifically, executing the control instruction to optimize and adjust the solder mask manufacturing process includes the following detailed operations: First, the executor module establishes real-time connections with various core devices on the production line through standard industrial communication interfaces, such as the Modbus protocol, Ethernet / IP protocol, and Profinet protocol. These devices include full-automatic printing machines, multi-zone curing ovens, high-precision automatic alignment systems, and ink supply systems. When receiving the control instruction transmitted by the decision module, the executor module will first analyze and verify the instruction to ensure the correctness of the instruction format and the legality of the parameters, preventing the execution of invalid or dangerous instructions.

[0048] Second, drive the printing machine parameter adjustment. If the instruction contains a printing pressure adjustment amount, such as requiring an increase of three percent in printing pressure, the executor module will drive the printing doctor blade mechanism through a servo motor to accurately change the vertical pressure acting on the Mini LED PCB board, with a control accuracy of Newton level. If the instruction requires adjusting the doctor blade angle, such as adjusting to eighty-five degrees, the executor module will accurately adjust the inclination angle of the doctor blade through a stepper motor, ensuring that the angle deviation is less than zero point one degree. If the instruction involves a printing speed adjustment amount, such as requiring a reduction of five percent in printing speed, the executor module will directly send the instruction to the motion controller of the printing machine to modify its running speed, with a response time in the millisecond level, ensuring the optimization of ink filling and leveling time during the printing process.

[0049] Next, drive the alignment system correction. If the control instruction contains an alignment system compensation amount, such as requiring a twenty-micron compensation to the X-axis positive direction, the executor module will drive the high-precision motion platform (usually using a piezoelectric ceramic platform or an air floating platform) to perform micron-level accurate displacement correction on the Mini LED PCB or mask plate. The repeatability positioning accuracy of the motion platform can reach zero point one microns, ensuring the position accuracy and consistency of the solder mask window.

[0050] Then, the ink supply system is adjusted. If the decision module suggests increasing the ink viscosity based on the test results, the actuator module will adjust the ink supply system, for example, by controlling the flow rate and mixing ratio of the filling pump, adding high or low viscosity diluent to the ink tank in precise proportion, adjusting the viscosity of the ink in real time to the target range (for example, twenty point five to twenty one point five Pascal seconds), and monitoring and feeding back in real time through the online viscosity sensor.

[0051] At the same time, the curing oven parameters are adjusted. If the control instruction contains the curing oven temperature setting or the curing time setting, for example, it requires the temperature of the first temperature zone of the curing oven to be raised by two degrees Celsius, and the curing time to be extended by ten seconds, the actuator module will accurately control the heating elements inside the curing oven through the integrated temperature control system to adjust the temperature distribution of different temperature zones (for example, temperature accuracy of plus or minus zero point five degrees Celsius) and the residence time of MiniLED PCB board in the oven. This aims to optimize the polymerization reaction of the ink, ensuring that it is fully cured and forms a smooth, hard and scratch-resistant surface.

[0052] After all the control instructions are executed, the actuator module will receive status feedback signals from the corresponding devices, such as the current pressure value of the printer, the scraper angle, the printing speed, the actual displacement of the alignment platform, the ink viscosity value, and the real-time temperature curve of the curing oven. These feedback information will be recorded and compared with the instruction content to confirm whether the control is accurately executed according to the instruction, and serve as auxiliary data for the next round of intelligent analysis.

[0053] Optimization and adjustment is not a one-time process, but a continuous closed-loop feedback cycle throughout the production process of MiniLED PCB solder mask layer. After the process parameters are adjusted, the subsequent produced MiniLED PCB solder mask layer products will again be subjected to quality detection by the detection module, and the new surface information data will again be input into the intelligent analysis module for evaluation. If the test results show that the quality has not yet reached the preset standard or there are new deviations (for example, although the thickness uniformity has improved, there is still slight deviation in windowing), the system will again generate and execute new control instructions until the solder mask layer quality parameters fully meet all preset requirements, reaching the optimal state of the process. The ability of iterative optimization and adaptive adjustment enables the entire solder mask manufacturing process to respond in real time to production fluctuations, equipment aging, raw material batch differences and environmental changes, continuously approaching the optimal state, ensuring the robustness of the production process and the consistency of product quality. At the same time, the log of the entire control process is recorded completely, providing valuable data support for long-term process optimization, fault diagnosis and product quality traceability.

[0054] Through the MiniLED PCB solder mask layer quality control method based on the intelligent detection feedback system, the manufacturing precision and quality of the MiniLED PCB solder mask layer can be significantly improved. Specifically, this method can: First, significantly improve the thickness uniformity and surface flatness of the solder resist layer. This method effectively suppresses the inconsistency of the ink thickness on the dielectric layer and copper caused by fluidity in traditional liquid ink printing by accurately measuring and intelligently analyzing the three-dimensional topography of the solder resist layer, and dynamically adjusting key parameters such as printing pressure, ink viscosity, and printing speed. The unevenness of the solder resist layer surface is eliminated. This micron-level flatness control directly solves the problem of severe apparent color difference in the solder resist layer, significantly reduces the scattering of light on the solder resist layer surface, thereby increasing the amount of light reaching the display layer, and improving the overall brightness and energy utilization efficiency of the MiniLED display product, and ensuring the high consistency of display effects in different areas, eliminating visual defects of uneven brightness and color.

[0055] Second, accurately control the size accuracy and position consistency of the solder resist window. The intelligent visual detection module can accurately identify the two-dimensional geometric features and precise position information of the solder resist window, with a recognition accuracy of sub-micron level. Combined with intelligent analysis algorithms, the system can quickly find problems such as window size not meeting design requirements or systematic positional deviation of the window. Through closed-loop feedback control of the installation position of the alignment system or the mask, this method can achieve accurate correction and compensation of the solder resist window, ensuring that its size and position meet strict design requirements, thereby fundamentally avoiding visual color difference caused by inconsistent or deviated solder resist window, ensuring that the LED chip can be stably and uniformly connected to the solder pad, and ensuring uniform light emission and long-term reliable operation of the LED chip.

[0056] Third, effectively prevent and reduce surface defects. The intelligent detection system can identify various small defects on the surface of the solder resist layer, such as foreign matter attachment with a diameter less than five microns, micro-bubbles, and fine scratches with a length less than fifty microns, and its detection speed far exceeds manual operation. Through real-time feedback, the system adjusts process parameters or reminds the operator to intervene in time. For example, when the system detects frequent small scratches, it instructs to optimize the curing process parameters to improve the hardness and wear resistance of the ink surface, thereby reducing the risk of scratches; or it discovers and handles the purity of the ink or the cleanliness of the equipment at an early stage, reducing the generation of foreign matter or bubbles from the source. This significantly improves the yield of MiniLED PCBs and improves the overall reliability of the product.

[0057] Fourth, improve the stability and reliability of electrical connection. The smooth and defect-free solder mask surface and precisely aligned solder mask openings ensure good and stable contact between the electrodes and the circuit, thereby significantly reducing the contact resistance. This optimization ensures uniform transmission and distribution of current in the MiniLED chip array. This directly avoids the occurrence of LED chip brightness and color instability, flickering, and local extinction failure caused by poor contact in traditional processes, greatly improving the long-term running stability and electrical performance of MiniLED products.

[0058] Fifth, optimize the overall optical reflectivity of MiniLED products. The uniform and defect-free solder mask surface can achieve more uniform and controllable light reflection, avoiding the problem of excessive or weak reflection in local areas caused by uneven traditional solder mask. This optimization fundamentally eliminates color distortion, enabling MiniLED display products to present images or videos with uniform brightness, pure colors, and delicate pictures, greatly improving the visual experience of end users and achieving higher display quality.

[0059] Sixth, realize automation and intelligentization of the production process. This method integrates high-precision data collection, intelligent analysis, decision-making generation, and device execution into one, forming an efficient and self-adaptive closed-loop control system. This system significantly reduces the dependence on human experience and reduces errors and instability caused by human operation, achieving higher level automation and intelligentization of MiniLED PCB solder mask manufacturing. This not only improves production efficiency and reduces operating costs, but more importantly, provides a solid technical guarantee for the large-scale and high-quality production of MiniLED products.

[0060] In summary, the present application introduces an intelligent detection feedback system, which fundamentally solves many quality problems of MiniLED PCB solder mask in traditional manufacturing processes, and provides strong technical support for the high-quality development of MiniLED display technology.

Claims

1. A method for controlling the quality of solder mask layer in MiniLED PCB based on an intelligent detection feedback system, characterized in that, include: The surface information data of the solder mask layer of the MiniLED PCB to be inspected is obtained. The surface information data is obtained by non-contact scanning and data acquisition of the solder mask layer through a high-precision vision inspection module and a three-dimensional morphology measurement module. The surface information data is used to identify the two-dimensional geometric features, defect types and three-dimensional morphology parameters of the solder mask layer. The surface information data is intelligently analyzed to generate a quality assessment result. The intelligent analysis uses image processing and machine learning algorithms to compare and evaluate the extracted feature parameters with the preset MiniLEDPCB solder mask layer quality standard. The quality assessment result clearly indicates the actual value of the key quality parameters of the solder mask layer and quantifies the deviation between the actual value and the preset quality standard. Based on the quality assessment results and the deviation, a decision module integrating an expert system or intelligent control algorithm is used to determine the control instructions for the key parameters of the solder mask manufacturing process. The control instructions are used to correct the deviation. The control command is executed to automatically optimize and adjust the solder mask manufacturing process, and the optimization and adjustment form a continuous closed-loop feedback loop.

2. The method for controlling the quality of the solder mask layer in MiniLED PCB according to claim 1, characterized in that, The surface information data of the solder resist layer of the MiniLED PCB to be inspected includes: Perform system startup and self-test operations to ensure that all detection devices and sensors are in normal working order; The MiniLEDPCB sample to be tested is precisely positioned and clamped at the designated location on the testing platform using an automated conveying system. Perform light source adjustment and environmental stabilization operations to eliminate external ambient light interference and ensure stable temperature and humidity; The high-precision visual inspection module captures two-dimensional image data of the solder resist layer surface; The three-dimensional topography measurement module acquires three-dimensional point cloud data or height map data of the solder resist surface; and The collected raw data is transmitted, undergoes preliminary data format conversion, and is stored.

3. The method for controlling the quality of the solder mask layer in MiniLED PCB according to claim 2, characterized in that, The high-precision visual inspection module captures two-dimensional image data of the solder resist surface, specifically including: A high-resolution industrial camera employing a complementary metal-oxide-semiconductor sensor with a resolution of no less than 25 million pixels, and equipped with a telecentric lens, captures two-dimensional image data of the solder resist surface; and Using multispectral imaging technology, multiple narrowband images are acquired in the visible and near-infrared bands to analyze the uniformity of the apparent color of the solder resist layer, identify the two-dimensional geometric morphology of foreign matter adhesion, micro bubbles, and minor scratches on the surface, and accurately measure the area, perimeter, roundness, and centroid coordinates of the solder resist openings, with measurement accuracy reaching the sub-micron level.

4. The method for controlling the quality of the solder mask layer in MiniLED PCB according to claim 2, characterized in that, The acquisition of three-dimensional point cloud data or height map data of the solder resist layer surface through the three-dimensional topography measurement module specifically includes: Using non-contact optical measurement equipment such as a laser confocal microscope or a white light interferometer, three-dimensional point cloud data or height map data of the solder resist layer surface are acquired, with a Z-axis measurement accuracy reaching tens of nanometers; and The local thickness, overall flatness, root mean square roughness, peak-to-valley difference, and verticality of the window edge of the solder resist layer are quantified, wherein the root mean square roughness is less than 100 nanometers in a 500-micrometer by 500-micrometer area.

5. The method for controlling the quality of the solder mask layer in MiniLED PCB according to claim 1, characterized in that, The intelligent analysis of the surface information data to generate quality assessment results includes: The original surface information data is preprocessed to eliminate noise and errors during the data acquisition process; Perform a key feature extraction operation on the solder resist layer to extract key indicators for quantifying the quality of the solder resist layer from the preprocessed surface information data; and Using machine learning or deep learning models, the extracted feature parameters are compared and evaluated with preset MiniLEDPCB solder mask layer quality standards with high precision, thereby generating the quality evaluation results and the deviation amount.

6. The method for controlling the quality of the solder mask layer in MiniLED PCB according to claim 5, characterized in that, The preprocessing operations for the raw surface information data include: For two-dimensional image data, nonlocal mean denoising algorithm or Gaussian filtering is used for image denoising, and adaptive histogram equalization or contrast-limited adaptive histogram equalization algorithm is used to enhance image contrast, and color deviation is compensated by color correction algorithm based on color calibration plate. For 3D point cloud data, statistical filtering or voxel downsampling is used for data filtering, and point cloud registration is performed by iterative nearest point algorithm or Pseudo distance registration algorithm to unify data from different perspectives into the same coordinate system.

7. The method for controlling the quality of the solder mask layer in MiniLED PCB according to claim 5, characterized in that, The operation of extracting key features of the solder mask layer includes: For two-dimensional image data, a convolutional neural network model is used to automatically identify and segment the weld shielding window area and defect area with high precision, and calculate their area, perimeter, centroid coordinates and color distribution histogram geometric feature parameters. For three-dimensional topographic data, the root mean square roughness, peak-to-valley difference, thickness distribution statistics of a specific region, and local slope three-dimensional parameters are calculated using three-dimensional analysis algorithms.

8. The method for controlling the quality of the solder mask layer in a MiniLED PCB according to claim 5, characterized in that, The comparison and evaluation using machine learning or deep learning models includes: The preset quality standards cover solder resist thickness tolerance, surface flatness requirements, solder resist window size tolerance, positional deviation range, and acceptable defect types and quantity thresholds. These preset quality standards are based on rigorous training and verification using qualified and unqualified product samples. The quality assessment results clearly indicate the actual values ​​of the local thickness of the solder resist layer, surface flatness, and the size and location of the solder resist opening, and quantify the deviation between the actual values ​​and the preset quality standards. The deviation precisely indicates the degree and direction of deviation from the standards.

9. The method for controlling the quality of the solder mask layer in a MiniLED PCB according to claim 1, characterized in that, Based on the quality assessment results and the deviation amount, the control instructions for the solder mask manufacturing process are determined as follows: The intelligent analysis outputs the quality assessment results and detailed deviations as data inputs and transmits them to the decision-making module. The decision-making module performs in-depth analysis of the received quality assessment results and deviations to accurately identify the type, severity, location, and potential causes of quality problems; and Based on the identified quality problem type and severity, the decision-making module generates targeted control instructions through a built-in expert system knowledge base or deep reinforcement learning model. These control instructions cover adjustments to printing pressure, doctor blade angle, printing speed, ink viscosity, alignment system compensation, mask calibration, curing oven temperature, and curing time.

10. The method for controlling the quality of the solder mask layer in a MiniLED PCB according to claim 1, characterized in that, Executing the control command to optimize and adjust the solder mask manufacturing process includes: The actuator module establishes a real-time connection with production equipment such as fully automatic printing machines, multi-temperature zone curing ovens, high-precision automatic alignment systems, and ink supply systems through a standard industrial communication interface. Adjusting printing press parameters includes precisely changing the vertical pressure of the squeegee on the MiniLED PCB board using a servo motor, precisely adjusting the squeegee tilt angle using a stepper motor, and modifying the printing press's running speed. Drive alignment system correction, including driving a high-precision motion platform to perform micron-level precise displacement correction of MiniLED PCB or mask; The system regulates the ink supply, including controlling the flow rate and mixing ratio of the dispensing pump, and adjusting the ink viscosity to the target range; and The parameters of the curing oven are adjusted, including the precise control of the temperature distribution in different temperature zones inside the curing oven and the residence time of the MiniLED PCB board in the oven through an integrated temperature control system.

Images