Reflow soldering surface mounting process

Abstract

The invention discloses a reflow soldering surface mount technology, which belongs to the technical field of SMT (Surface Mount Technology), and comprises the following steps of: firstly, optimizing soldering flux activation through dynamic wind speed in an interval of 80-100 DEG C; infrared auxiliary heating is started at the stage of 100-130 DEG C, and the radiation intensity is accurately regulated and controlled based on the BGA position; the nitrogen flow is automatically adjusted within the range of 130-150 DEG C to maintain the low-oxygen environment; the reinforcement learning model is driven to rise to the peak value of 235 DEG C within 8-12 seconds, and precise welding of the QFN bonding pad is achieved through composite pulse hot air and infrared focusing; then, the temperature is reduced to 180 DEG C from 235 DEG C at the two-stage slope, the warping risk is predicted in real time through a laser displacement sensor, and deformation is controlled through a ceramic needle array dynamic support; the double-air-knife system works cooperatively, the oblique angle laminar flow of the upper air knife avoids a precise element, the vortex cooling rate of the lower air knife is increased to 6 DEG C per second, and the temperature uniformity is ensured to be + / -2 DEG C; the invention solves the technical problem of how to improve the low yield of the existing reflow soldering surface mount technology.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of SMT, in particular to a reflow soldering patch process. BACKGROUND

[0002] In recent years, with the rapid development of artificial intelligence and big data technology, the digitalization and intelligentization transformation of traditional industries have ushered in new development opportunities. Surface mount production lines are widely used in the electronic manufacturing industry, especially for high-performance electronic mounting production lines. Deep digitalization and intelligentization transformation of the production line is of great significance to the production quality and production efficiency of the product. The transformed electronic mounting production line can realize the pre-control of product quality and the accurate adaptation of production line process parameters. Ultimately, the purpose of reducing production cost and improving product yield is achieved.

[0003] Specifically, the quality prediction of the electronic mounting production line through artificial intelligence algorithm and big data technology is an effective means to realize the pre-control of product quality. Through the quality prediction method, important factors affecting product quality can be determined, and product production quality can be predicted in advance, so that process personnel can adjust the production line parameters in advance to minimize losses. Unreasonable production line process parameter settings will affect the production quality of the product. Through the process parameter optimization method, the accurate adaptation of the production line process parameters can be realized, thereby improving the production efficiency and reducing the production cost.

[0004] For example, the Chinese patent document with publication number CN107124835B discloses a reflow soldering patch process, which includes the following steps: printing tin paste, patching, intermediate inspection, reflow soldering and post-furnace inspection. By reducing the pressure during reflow soldering, the formation of large bubbles and cavities during the welding process can be avoided, or the escape of large bubbles can cause explosive exhaust, bringing out many small tin beads. By reducing the pressure in the welding area to form a negative pressure, small bubbles in the welding process can easily escape and are not easy to form large bubbles, and tin beads are not easy to splash.

[0005] However, the reflow soldering process disclosed above still has the technical problems of low quality prediction accuracy and low yield due to the dependence of process parameters on manual experience. Specifically, with the rapid development of high-tech industries such as artificial intelligence, new energy, aerospace and 5G, the demand for high-performance electronic mounting products is increasing, and high performance inevitably leads to high complexity and large size of electronic mounting boards. Due to the more complex processing technology of high complexity and large size PCB mounting, it usually has higher quality control requirements. The problems faced by high complexity and large size PCB mounting are on the one hand that the lack of available sample size leads to the inability to use traditional data mining and artificial intelligence algorithms for accurate quality prediction; on the other hand, the unreasonable setting of production line process parameters and the serious dependence on manual experience lead to the inability to achieve accurate adaptation of production line process parameters. SUMMARY

[0006] Therefore, it is necessary to improve the technical problem of low yield of the existing reflow soldering process, and provide a reflow soldering process.

[0007] A reflow soldering process, comprising the following steps: Step 1: Gradient preheating, temperature range 80-150℃: Substage 1, 80-100℃: Time: 30 seconds; Hot air mode: low-speed laminar flow, wind speed 0.5 m / s; Objective: Uniformly activate solder paste flux, avoid component thermal stress impact.

[0008] Substage 2, 100-130℃: Time: 20 seconds; Infrared assistance: short-wave infrared radiation, wavelength 3-5 μm, heat the bottom of BGA; Substage 3, 130-150℃: Time: 15 seconds; Introduce nitrogen to inert, make the oxygen concentration less than 500 ppm; Step 2: Double-peak holding, temperature range 150-190℃; Segmented holding: The first temperature range is controlled to be 150-170℃, and lasts for 70 seconds; The second temperature range is controlled to be 170-190℃, and lasts for 50 seconds; Pulse hot air control: 2Hz high-frequency pulse to clear the air gap below the component; Pulse interval period opens-5kPa vacuum adsorption to reduce voids; Step 3: Pulse reflow, temperature peak value 235℃, control within 10 seconds from 190℃ to 235℃; Composite heating mode: Pulse hot air, 1Hz, peak wind speed 8m / s; Synchronous infrared focusing, medium wave 4μm, aiming at QFN ground pad; Holding time: 15 seconds; Step 4: Gradient slow cooling, temperature range 235-180℃: Primary slow cooling: from 235°C to 220°C in 10 seconds, with a slope of -1.5°C / s; Secondary slow cooling: from 220°C to 180°C in 15 seconds, with a slope of -0.8°C / s; Dynamic substrate support: ceramic pin array conformal support PCB; Step 5: Forced air cooling, temperature range 180-50°C: Dual air knife system: Upper air knife: 45° oblique angle laminar flow; Lower air knife: vortex cooling, temperature uniformity reaches ±3°C; Cooling rate: 6°C / s.

[0009] Further, in another embodiment, the steps of the AI-enhanced reflow soldering process are as follows: Step 1: Intelligent gradient preheating stage, temperature range 80-150°C: Sub-stage 1: Temperature range 80-100°C, time 30 seconds; low-speed laminar hot air mode is used, with a dynamic adjustment range of wind speed of 0.3-0.8 meters per second; the amount of flux volatilization is monitored in real time by an infrared spectrometer, and the preheating time is adjusted based on feedback; Sub-stage 2: Temperature range 100-130°C, time 20 seconds; infrared auxiliary heating is enabled, using short-wave infrared radiation with a wavelength range of 3-5 microns; BGA positions are located in real time by image recognition technology, and the infrared radiation intensity is dynamically matched; Sub-stage 3: Temperature range 130-150°C, time 15 seconds; nitrogen atmosphere control is introduced, with automatic adjustment of nitrogen flow based on oxygen sensor data; the oxygen concentration is maintained stable within the range of 300-500 ppm; Step 2: Pulse reflow stage, temperature peak 235°C: The reinforcement learning model dynamically optimizes the temperature rise curve, completing the jump from 190 to 235°C within 8-12 seconds; a composite heating mode is used: the pulse hot air frequency dynamically ranges from 0.5 to 2 Hz, which is controlled in real time by a digital twin system; synchronous infrared focusing welding is performed for precise energy delivery to QFN ground pads; the holding time is 15 seconds; Step 3: Gradient slow cooling stage, temperature range 235-180°C: Primary slow cooling zone: temperature drops from 235°C to 220°C in 10 seconds; the cooling slope is controlled at -1.5°C / s; Secondary slow cooling zone: temperature drops from 220°C to 180°C in 15 seconds; the cooling slope is controlled at -0.8°C / s; Real-time scanning of PCB deformation by laser displacement sensor to predict warping risk areas; ceramic needle array support system supports PCB with dynamic response accuracy of plus or minus 0.1 millimeters; Step four: strengthen the air cooling stage, temperature range 180 to 50 degrees Celsius: Adopt double air knife collaborative cooling system: Upper air knife: 45-degree oblique angle laminar flow design, avoiding precision components through computer vision; Lower air knife: vortex cooling technology, vortex intensity adjusted in real time by PID controller; Temperature uniformity controlled within plus or minus 2 degrees Celsius; Cooling rate increased to 6 degrees Celsius per second.

[0010] Further, the operation logic of the digital twin system in step two is as follows: the sensor network collects real-time data of temperature, deformation, and gas concentration; the edge computing node performs data preprocessing and feature extraction; the digital twin decision center generates process parameter optimization instructions; the actuator dynamically adjusts the temperature zone power, air speed, and cooling intensity; and the quality detection results are fed back to the AI model for closed-loop optimization.

[0011] Further, the digital twin system in step two analyzes X-ray detection data through a long short-term memory network to predict the probability of solder voids in advance; and dynamically adjusts the patch pressure and hot air angle according to the nanoscale solder paste wettability monitoring results.

[0012] Further, the digital twin system in step two shares encrypted model parameters across production lines through a federated learning platform to continuously optimize the temperature control strategy; and analyzes the solder paste composition online to automatically recommend the weight percentage of nanoscale copper doping range between 0.08 and 0.12.

[0013] Further, the AI-controlled closed-loop process during the pulse reflow stage includes the following steps: Step 1: Real-time temperature field scanning: a. Use an infrared thermal imager to scan the PCB surface with a resolution of 0.1℃; b. Scanning frequency: collect full-plate temperature distribution data every 200 milliseconds; c. Key monitoring points: BGA solder joints, QFN ground pads, and heat-sensitive component areas; Step 2: Digital Lieng engine decision: Input data: current temperature field matrix, preheating stage historical thermal load data, and component layout coordinate information; and perform two core calculations in parallel: Task A: Predicting BGA solder joint real-time liquidus rate, based on transient thermodynamic model, simulating solder melting state, outputting liquidus rate deviation value, setting the target value of liquidus rate deviation value to not less than 99.5%; Task B: Identifying QFN thermal hysteresis area, comparing design temperature with actual temperature through convolutional neural network, marking high-risk area coordinates with temperature difference greater than 5℃; Step 3: Pulse hot air dynamic regulation: a. Receive liquidus rate deviation data stream b. Execute pulse frequency optimization algorithm: If the liquidus rate is less than 98%, increase the pulse frequency to 1.8-2.0Hz; If the liquidus rate is greater than 99% and the local temperature is greater than 232℃, reduce the frequency to 0.8Hz; Otherwise, maintain the default frequency of 1.2Hz; c. Control command is issued to the hot air generator, with response delay less than 50ms; Step 4: Infrared supplementary radiation activation: a. For marked thermal hysteresis area: coordinate mapping to infrared emission array by positioning system; automatically match wavelength, QFN area fixed at 4.5μm; b. Radiation intensity is dynamically adjusted according to the formula: supplementary power = basic value * 0.35 + (temperature difference - 5℃) x 0.02; Step 5: Temperature uniformity verification: a. Start secondary scanning after supplementary operation: focus on monitoring the original thermal hysteresis area; calculate the maximum temperature difference △T-max of the whole board; b. Decision branch: if ΔT-max≤5℃, it is judged to be up to standard, and enters the 15-second holding stage; if ΔT-max>5℃, return to step 2 for re-optimization; Step 6: Holding stage monitoring: a. Maintain peak temperature 235℃±3℃ b. Real-time detection of abnormal state: tin bead splashing monitoring: capture burst sound frequency signal through acoustic sensor; component overheating protection: when MLCC area is greater than 215℃, cut off local infrared source.

[0014] In summary, the present application discloses a reflow soldering process which adopts four-stage intelligent control. First, intelligent gradient preheating is performed: dynamic wind speed optimization is used to activate flux in the 80-100 Celsius range; infrared auxiliary heating is enabled in the 100-130 Celsius stage, and the radiation intensity is accurately controlled based on BGA position; nitrogen flow is automatically adjusted in the 130-150 Celsius range to maintain a low-oxygen environment; then, the pulse reflow stage is entered: a reinforcement learning model drives a jump to a peak value of 235 Celsius within 8-12 seconds, composite pulse hot air and infrared focusing are used to achieve accurate welding of QFN pads, and the holding time is shortened by 40%; next, gradient slow cooling is performed: two-stage slope is used to reduce the temperature from 235 Celsius to 180 Celsius, a laser displacement sensor is used to predict warping risks in real time, and a ceramic needle array is used to dynamically support and control deformation; finally, intensive air cooling is implemented: a double-air-knife system works cooperatively, the upper air knife uses oblique angle laminar flow to avoid precision components, and the lower air knife uses vortex cooling to increase the rate to 6 Celsius per second, ensuring that the temperature uniformity is within ±2 Celsius. The whole process of the reflow soldering process is dynamically optimized at the millisecond level through a digital twin system, thereby achieving adaptive adjustment of the soldering process, reducing tin bead splashing rate, and improving the failure rate of heat-sensitive components. Moreover, the reflow soldering process can improve the quality of soldering, such as reducing the void rate to <5%, which is better than the 15-20% of traditional processes. In addition, it can also make the warping degree of large-size PCB ≤0.3%, which is better than the 0.5-0.8% of traditional processes. Therefore, the reflow soldering process solves the technical problem of low yield of existing reflow soldering processes. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 The system architecture diagram of the digital twin dynamic optimization system applied to the reflow soldering process of the present application; Figure 2 The system architecture diagram of the AI-enhanced reflow soldering process implemented based on the reflow soldering process of the present application; Figure 3 The AI control closed-loop process schematic diagram applied to the reflow soldering process of the present application. DETAILED DESCRIPTION

[0016] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below. In the following description, many specific details are set forth in order to provide a thorough understanding of the present application. However, the present application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without departing from the scope of the present application, so the present application is not limited to the specific embodiments disclosed below.

[0017] In the description of the application, it should be understood that the orientation or positional relationship indicated by the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential" and the like is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the application and simplifying the description, and does not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the application.

[0018] In addition, the terms "first", "second" are only for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined with "first", "second" can explicitly or implicitly include at least one of the features. In the description of the application, the meaning of "a plurality of" is at least two, such as two, three, etc., unless otherwise explicitly specified and limited.

[0019] In the present application, unless otherwise explicitly specified and limited, the terms "mounting", "connection", "connection", "fixing" and the like should be understood broadly, for example, it can be fixed connection, or detachable connection, or integral; it can be mechanical connection, or electrical connection; it can be directly connected, or indirectly connected through intermediate medium, it can be the internal communication of two elements or the interaction relationship between two elements, unless otherwise explicitly limited. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0020] In the present application, unless otherwise explicitly specified and limited, the first feature "on" or "under" the second feature can be that the first and second features are in direct contact, or the first and second features are in indirect contact through an intermediate medium. Moreover, the first feature "above", "over" and "on" the second feature can be that the first feature is directly above or obliquely above the second feature, or only indicates that the horizontal height of the first feature is higher than that of the second feature. The first feature "below", "under" and "under" the second feature can be that the first feature is directly below or obliquely below the second feature, or only indicates that the horizontal height of the first feature is less than that of the second feature.

[0021] It is to be understood that when an element such as a layer, region or substrate is referred to as being "on" or "connected to" another element, it can be directly on or connected to the other element or intervening elements can be present. In addition, the term "connected" as used herein means the element that is connected can be directly connected to the other element or can be indirectly connected via intervening elements. As used herein, the term "vertical", "horizontal", "upper", "lower", "left", "right", and the like are merely used for the purpose of illustration and do not indicate an absolute orientation.

[0022] Referring to the drawings Figures 1 to 2 , A reflow soldering process according to the present application comprises the following steps: Step 1: Gradient preheating, temperature range 80-150℃: Sub-stage 1, 80-100℃: Time: 30 seconds; Hot air mode: low-speed laminar flow, wind speed 0.5 m / s; Objective: Uniformly activate solder paste flux, avoid thermal stress impact on components.

[0023] Sub-stage 2, 100-130℃: Time: 20 seconds; Infrared assistance: short-wave infrared radiation, wavelength 3-5 μm, heat the bottom of BGA; Sub-stage 3, 130-150℃: Time: 15 seconds; Introduce nitrogen to inert, make the oxygen concentration <500 ppm; Step 2: Double-peak holding, temperature range 150-190℃: Innovative design: segmented holding: 150-170℃, 70 seconds + 170-190℃, 50 seconds; Pulse hot air control: 2 Hz high-frequency pulse to clear the air gap below the components; Pulse interval period open -5 kPa vacuum suction to reduce voids; Step 3: Pulse reflow, temperature peak 235℃: Technical breakthrough: from 190℃ to 235℃ in 10 seconds, higher than the traditional 220℃; Composite heating mode: pulse hot air, 1 Hz, peak wind speed 8 m / s; Synchronous infrared focusing, medium wave 4 μm, aiming at QFN ground pad; Holding time: 15 seconds, 40% shorter than the traditional; Step 4: Gradient slow cooling, temperature range 235-180℃: Innovative slow cooling design, solve the technical problem of warping of large-size PCB: First-stage slow cooling: 235℃→220℃, 10 seconds, slope -1.5℃ / s; Second-stage slow cooling: 220℃→180℃, 15 seconds, slope -0.8℃ / s; Cooperate with dynamic substrate support: ceramic needle array follows the shape of the PCB; Step 5: Forced air cooling, temperature range 180-50℃: Using double air knife system: Upper air knife: 45° oblique angle laminar flow, which can avoid component displacement; Lower air knife: vortex cooling, temperature uniformity reaches ±3℃; Cooling rate: 6℃ / s, higher than the traditional 4℃ / s.

[0024] Specifically, the reflow soldering process of the present application compared with the traditional process, the difference between the two is as follows: Table 1: Technical innovation of the present application

[0025] Further, a digital twin dynamic optimization system applied to the reflow soldering process of the present application, the system architecture is as shown in Figure 1 The digital twin dynamic optimization system has a real-time monitoring layer, which scans the PCB temperature field with an infrared thermal imager with a resolution of 0.1℃, and monitors the substrate deformation through a pressure sensor; it also has a thermal field simulation layer, which can be based on the transient thermodynamic model of FEM, with an update frequency of 1Hz; moreover, this thermal field simulation layer can also predict the liquid phase rate of BGA solder joint and the warping trend of PCB; in addition, the digital twin dynamic optimization system has an optimization engine, which can dynamically adjust parameters: such as temperature zone power compensation value is ±10%; pulse hot air frequency is 0.5-2Hz, cooling air knife angle is ±5°.

[0026] Further, the reflow soldering process of the present application can improve the quality of soldering, such as reducing the void rate to <5%, which is better than the traditional process of 15-20%; in addition, it can also make the warping degree of large size PCB ≤0.3%, which is better than the traditional process of 0.5-0.8%.

[0027] Further, the reflow soldering process of the present application can optimize the efficiency of soldering, such as reducing the total welding time to 140 seconds, while the traditional process needs 180 seconds; energy consumption is reduced by 22%, because pulse heating reduces continuous power.

[0028] Further, the reflow soldering process of the present application can control defects, such as making the monument occurrence rate <0.1ppm, because the nano solder paste improves wettability; it can also make BGA pillow defects almost eliminated, because of infrared focused welding.

[0029] In summary, the reflow soldering process of the present application solves the traditional technical problems of warping, cold welding, voids, etc. in traditional reflow soldering of high complexity large size PCB through three innovations of dynamic temperature control, precise energy delivery and real-time optimization.

[0030] Further, in a reflow soldering process, a detailed step of an AI-enhanced reflow soldering process is as follows: Step one: intelligent gradient preheating stage, temperature range 80 to 150 degrees Celsius: Sub-stage one: temperature range 80 to 100 degrees Celsius, time 30 seconds; low-speed laminar flow hot air mode is adopted, with a dynamic adjustment range of wind speed of 0.3 to 0.8 meters per second; the goal is to uniformly activate the solder paste flux, avoiding the thermal stress impact on the components; the flux volatilization is monitored in real time by an infrared spectrometer, and the preheating time is adjusted by feedback; Sub-stage two: temperature range 100 to 130 degrees Celsius, time 20 seconds; infrared auxiliary heating is enabled, short-wave infrared radiation technology is adopted, with a wavelength range of 3 to 5 microns; the BGA position is located in real time through image recognition technology, and the infrared radiation intensity is dynamically matched; Sub-stage three: temperature range 130 to 150 degrees Celsius, time 15 seconds; nitrogen atmosphere control is introduced, based on oxygen sensor data to automatically adjust the nitrogen flow; the oxygen concentration is maintained stable within the range of 300 to 500 ppm; Step two: pulse reflow stage, temperature peak 235 degrees Celsius: The reinforcement learning model dynamically optimizes the temperature rise curve, completing the jump from 190 to 235 degrees Celsius within 8 to 12 seconds; a composite heating mode is adopted: the pulse hot air frequency dynamically ranges from 0.5 to 2 hertz, which is adjusted in real time by the digital twin system; synchronous infrared focusing welding is performed for precise energy delivery to the QFN ground pad; the holding time is 15 seconds, which is 40% shorter than the traditional process; Step three: gradient slow cooling stage, temperature range 235 to 180 degrees Celsius: First-stage slow cooling zone: temperature drops from 235 degrees Celsius to 220 degrees Celsius, taking 10 seconds; the cooling slope is controlled at -1.5 degrees Celsius per second; Second-stage slow cooling zone: temperature drops from 220 degrees Celsius to 180 degrees Celsius, taking 15 seconds; the cooling slope is controlled at -0.8 degrees Celsius per second.

[0031] The laser displacement sensor scans the PCB deformation in real time to predict the warping risk area; the ceramic needle array support system supports the PCB in shape, with a dynamic response accuracy of plus or minus 0.1 millimeters; Step four: intensive air cooling stage, temperature range 180 to 50 degrees Celsius: A double-air-knife collaborative cooling system is adopted: Upper air knife: 45-degree oblique angle laminar flow design, avoiding precision components through computer vision; Lower air knife: vortex cooling technology, with real-time adjustment of vortex intensity by a PID controller; The temperature uniformity is controlled within a range of plus or minus 2 degrees Celsius; The cooling rate is raised to 6 degrees Celsius per second, which is better than the 4 degrees Celsius per second of the conventional process.

[0032] Specifically, in the detailed steps of the AI-enhanced reflow soldering patch process described above, a digital twin dynamic regulation system is adopted, which fuses infrared thermal imager, vibration sensor and industrial camera data to construct a virtual process model; and based on a deep reinforcement learning algorithm, parameters such as pulse frequency, nitrogen flow rate and cooling slope are optimized in real time.

[0033] Specifically, in the detailed steps of the AI-enhanced reflow soldering patch process described above, the system architecture implemented is as shown in Figure 2 The "perception-decision-execution-feedback" closed loop is used to achieve millisecond-level dynamic optimization of process parameters and break through the yield bottleneck of high-complexity PCBs; this scheme deeply embeds artificial intelligence into the core process link, significantly improves the quality prediction accuracy and parameter adaptation capability, and is particularly suitable for large-size PCB manufacturing in high-reliability fields such as aerospace and 5G communication. Specifically, the running logic of the system is: the sensor network collects real-time data such as temperature, deformation and gas concentration; the edge computing node performs data preprocessing and feature extraction; the digital twin decision center generates process parameter optimization instructions; the actuator dynamically regulates the power of the temperature zone, the wind speed and the cooling intensity; the quality detection results are fed back to the AI model to realize closed-loop optimization. Thus, the process disclosed by the present application can achieve millisecond-level dynamic optimization through the perception, decision, execution and feedback closed loop, and break through the manufacturing bottleneck of high-complexity PCBs.

[0034] Further, in the detailed steps of the AI-enhanced reflow soldering patch process described above, it has a defect prediction and closed-loop control mode, which can analyze X-ray detection data through a long short-term memory network to predict the probability of solder voids in advance; and dynamically adjust the patch pressure and hot air angle according to the nanometer solder paste wettability monitoring results.

[0035] Further, in the detailed steps of the AI-enhanced reflow soldering patch process described above, it has a self-adaptive process evolution mode, which can share encrypted model parameters across production lines through a federated learning platform to continuously optimize the temperature control strategy; and analyze the solder paste composition online to automatically recommend the nanometer copper doping proportion range of 0.08 to 0.12 weight percent.

[0036] Specifically, the key performance improvements of the detailed steps of the AI-enhanced reflow soldering patch process described above are shown in Table 2: Table 2: Key performance improvements

[0037] Further, for the aforementioned step two, the pulse reflow stage is implemented as follows: stage target, complete the rapid jump of 190-235°C within 8-12 seconds, achieve complete liquid phase of solder, while avoid damage to heat-sensitive components.

[0038] Specifically, first, dynamic temperature rise control is performed, such as reinforcement learning temperature control model; that is, after input parameters: PCB layer number, component density distribution, solder paste heat capacity characteristics, and temperature field at the end of the preheating stage, output instructions: dynamic range of temperature rise rate: 18-25°C / s, traditional process ≤15°C / s; peak temperature fine-tuning interval: 232-238°C, which can be automatically adjusted according to the size of BGA. The decision mechanism of the dynamic temperature rise control is: infrared thermal imager data is collected every 200 milliseconds, and the optimal power distribution scheme is calculated through the Q-learning algorithm.

[0039] Further, the aforementioned pulse reflow stage also has a composite energy delivery system, such as a pulse hot air subsystem, the parameters of which are shown in Table 3: Table 3: Pulse hot air subsystem

[0040] Specifically, the working process of the pulse hot air subsystem is as follows: a. High-frequency pressure sensor monitors the boiling state of the solder paste; b. When local boiling lag is detected, increase the pulse frequency of the region to 1.5 Hz; c. When the QFN pad temperature > 230°C, switch to 0.8 Hz low-frequency mode to prevent overheating.

[0041] Further, the aforementioned pulse reflow stage also has an infrared focusing subsystem, the wavelength matching of which is: BGA area: 3-4 pm medium-short wave, with high penetration; QFN ground pad: 4-5 pm medium-long wave, surface thermal effect dominant; positioning accuracy: component coordinate mapping based on machine vision, spot positioning error ≤±0.05 mm.

[0042] Further, the physical effect optimization mechanism of the aforementioned pulse reflow stage includes solder joint liquid phase rate improvement and thermal shock inhibition. For solder joint liquid phase rate improvement, it has pulse hot air vortex effect: high-frequency airflow forms micro-scale vortex around the solder joint, accelerating the removal of flux residues, increasing the liquid phase rate from 92% to 99.5%; and infrared gradient heating: differential radiation for BGA bottom and QFN sidewall, eliminating the "shadow effect" caused by cold welding area. For thermal shock inhibition, it has pulse mutation buffer technology: when the temperature approaches 230°C, or within the window of 230±3°C, a ramp-down frequency program is started, which includes the following steps: If ΔT / Δt > 20℃ / s → pulse frequency reduced to 0.5Hz If local temperature difference of element > 8℃ → trigger infrared compensation heating Further, an AI-controlled closed-loop process is as shown in Figure 3 The key performance data is shown in Table 4 as follows: Table 4: Key performance verification data

[0043] Further, the failure protection measures provided by the reflow soldering patch process of the present application include cold solder real-time blocking and component overheating protection. For cold solder real-time blocking, when the solder joint temperature < 228℃ and the maintenance time > 1.5 seconds, the pulse frequency of the region is automatically increased by 20%. For component overheating protection, a hard threshold of 215℃ is set for the MLCC (multilayer ceramic capacitor) region, and the infrared radiation is immediately turned off when the temperature exceeds the threshold. Thus, through the synergistic control of pulse dynamics and infrared spectroscopy, the speed is increased by 40% while the risk of thermal damage is reduced to 1 / 24 of that of the traditional process.

[0044] Further, in the pulse reflow stage of the reflow soldering patch process of the present application, the AI-controlled closed-loop process specifically includes the following steps: Step 1: Real-time temperature field scanning: a. Use an infrared thermal imager to scan the PCB surface with a resolution of 0.1℃; b. Scan frequency: collect temperature distribution data of the whole board once every 200 milliseconds; c. Key monitoring points: BGA solder joints, QFN ground pads, and heat-sensitive component regions; Step 2: Digital Lieng engine decision: Input data: current temperature field matrix, preheating stage historical thermal load data, and component layout coordinate information; and perform two core calculations in parallel: Calculation task A: predict the real-time liquid phase rate of BGA solder joints, based on transient thermodynamic model, simulate the solder melting state, output liquid phase rate deviation value, target value ≥ 99.5%; Calculation task B: identify QFN thermal hysteresis regions, compare design temperature with actual temperature through convolutional neural network, and mark the coordinates of high-risk regions with temperature difference > 5℃; Step 3: Pulse hot air dynamic regulation: a. Receive liquid phase rate deviation data stream b. Execute pulse frequency optimization algorithm: If liquid phase rate < 98% → increase pulse frequency to 1.8-2.0Hz; If liquid phase rate > 99% and local temperature > 232℃ → reduce frequency to 0.8Hz; Default to maintain the reference frequency of 1.2Hz; c. Control instructions are issued to the hot air generator, with a response delay < 50 ms; Step 4: Infrared reinforcement radiation activation: a. For the marked thermal hysteresis area: the positioning system maps the coordinates to the infrared emission array; automatically match the wavelength, and the QFN area is fixed at 4.5 μm; b. The radiation intensity is dynamically adjusted according to the formula: reinforcement power (W / mm²) = basic value 0.35 + (temperature difference - 5℃) x 0.02; Step 5: Temperature uniformity verification: a. Start a second scan after reinforcement operation: focus on monitoring the original thermal hysteresis area; calculate the maximum temperature difference △T_max of the whole board; b. Decision branch: if ΔT_max≤5℃→ meet the standard, enter the 15-second holding phase; if ΔT_max>5℃→ return to step 2 to re-optimize; Step 6: Holding phase monitoring: a. Maintain the peak temperature at 235℃±3℃ b. Real-time detection of abnormal state: tin bead splashing monitoring: capture the burst sound frequency signal through acoustic sensor; component overheating protection: cut off the local infrared source when the MLCC area > 215℃.

[0045] Specifically, in the foregoing process, a double-channel parallel control mode is adopted, such as pulse hot air channel: master the global temperature rise rate; infrared reinforcement channel: treat local thermal hysteresis defects. Moreover, it also has three-level safety protection, as shown in the following Table 5: Table 5: Three-level safety protection

[0046] Specifically, for the self-learning optimization mechanism of the foregoing process, it is automatically executed after daily production ends: analyze the correlation between daily process data and defects; update the reward function parameters of the reinforcement learning model; optimize the pulse frequency-temperature response curve.

[0047] Specifically, the performance indicators of the foregoing process include: single cycle time consumption is 120-180 milliseconds; temperature uniformity control precision is ±3℃ for the whole board; liquid phase rate standard rate is 99.8%, more than 100,000 points statistics; thermal hysteresis correction success rate is 97.6%. As can be seen, the foregoing closed-loop system realizes precise control of the pulse reflow process through continuous iteration of "scanning-decision-execution-verification", and changes the passive response of traditional open-loop process to active prevention control.

[0048] In summary, the present application discloses a reflow soldering process which adopts four-stage intelligent control, first, intelligent gradient preheating is performed: dynamic wind speed optimization is used to activate flux in the range of 80 to 100 degrees Celsius; infrared auxiliary heating is enabled in the range of 100 to 130 degrees Celsius, and the radiation intensity is accurately controlled based on the BGA position; in the range of 130 to 150 degrees Celsius, the nitrogen flow is automatically adjusted to maintain a low-oxygen environment; then, the pulse reflow stage is entered: the reinforcement learning model drives the temperature to jump to 235 degrees Celsius within 8 to 12 seconds, composite pulse hot air and infrared focusing are used to realize accurate welding of QFN pads, and the holding time is shortened by 40%; then, gradient slow cooling is performed: the temperature is reduced from 235 degrees Celsius to 180 degrees Celsius in two stages, a laser displacement sensor is used to predict the warping risk in real time, and a ceramic needle array is used to dynamically support and control deformation; finally, intensive air cooling is implemented: the double air knife system works cooperatively, the upper air knife uses oblique angle laminar flow to avoid precision components, and the lower air knife uses vortex cooling to increase the cooling rate to 6 degrees Celsius per second, ensuring that the temperature uniformity is within 2 degrees Celsius; the whole process of the reflow soldering process is dynamically optimized at the millisecond level through a digital twin system; thus, adaptive adjustment of the soldering process can be realized, the tin bead spatter rate is reduced, the failure rate of heat-sensitive components is improved, and the like; moreover, the reflow soldering process can improve the quality of soldering, such as reducing the void rate to <5%, which is better than the 15-20% of the traditional process; in addition, the warping degree of large-size PCBs can be ≤0.3%, which is better than the 0.5-0.8% of the traditional process. Therefore, the reflow soldering process solves the technical problem of low yield of the existing reflow soldering process.

[0049] The technical features of the above-described embodiments can be combined in any manner, and to make the description concise, not all possible combinations of the technical features in the above-described embodiments are described, however, as long as the combinations of the technical features do not exist contradictions, they should be considered as the scope of the description.

[0050] The above-described embodiments only express several implementation manners of the present application, and the description is relatively specific and detailed, but it should not be understood as a limitation on the scope of the patent. It should be noted that for ordinary skilled persons in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A reflow soldering process, characterized by, It includes the following steps: Step 1: Gradient preheating, temperature range 80-150℃: Sub-stage 1, 80-100℃: Time: 30 seconds; Hot air mode: low-speed laminar flow, wind speed 0.5 m / s; Objective: Uniformly activate solder paste flux, avoid component thermal stress impact; Sub-stage 2, 100-130℃: Time: 20 seconds; Infrared auxiliary: short-wave infrared radiation, wavelength 3-5 μm, heat the bottom of BGA; Sub-stage 3, 130-150℃: Time: 15 seconds; Introduce nitrogen to inert, make oxygen concentration less than 500 ppm; Step 2: Double-peak holding, temperature range 150-190℃; Segmented holding: The first segment temperature range is controlled at 150-170℃ for 70 seconds; The second temperature range is controlled at 170-190℃ for 50 seconds; Pulse hot air control: 2 Hz high-frequency pulse to clear the air gap below the component; Pulse interval period opens -5 kPa vacuum suction to reduce voids; Step 3: Pulse reflow, temperature peak value 235℃, control within 10 seconds from 190℃ to 235℃; Composite heating mode: pulse hot air, 1 Hz, peak wind speed 8 m / s; Synchronous infrared focusing, medium wave 4 μm, aiming at QFN ground pad; Holding time: 15 seconds; Step 4: Gradient slow cooling, temperature range 235-180℃: First-level slow cooling: within 10 seconds from 235℃ to 220℃, the slope of the cooling curve is -1.5℃ / s; Second-level slow cooling: within 15 seconds from 220℃ to 180℃, the slope of the cooling curve is -0.8℃ / s; Dynamic substrate support: ceramic needle array conformal support PCB; Step 5: Forced air cooling, temperature range 180-50℃: Use double air knife system: Upper air knife: 45° oblique angle laminar flow; Lower air knife: vortex cooling, temperature uniformity reaches ±3℃; Cooling rate: 6℃ / s.

2. The reflow soldering process of claim 1, wherein: The steps of the AI-enhanced reflow soldering process are as follows: Step one: Intelligent gradient preheating stage, temperature range 80 to 150 degrees Celsius: Sub-stage one: temperature range 80 to 100 degrees Celsius, time 30 seconds; Use low-speed laminar flow hot air mode, wind speed dynamically adjusted range 0.3 to 0.8 meters per second; Monitor flux volatilization amount in real time through infrared spectrometer, feedback adjust preheating time; Sub-stage two: temperature range 100 to 130 degrees Celsius, time 20 seconds; Enable infrared auxiliary heating, use short-wave infrared radiation method, wavelength range 3 to 5 micrometers; Real-time locate BGA position through image recognition technology, dynamically match infrared radiation intensity; Sub-stage three: temperature range 130 to 150 degrees Celsius, time 15 seconds; Introduce nitrogen atmosphere control, automatically adjust nitrogen flow based on oxygen sensor data; Maintain oxygen concentration stable within 300 to 500 ppm range; Step two: pulse reflow stage, temperature peak value 235 degrees Celsius: The reinforcement learning model dynamically optimizes the temperature rising curve, achieving a 190-235℃ jump within 8-12 seconds; a composite heating mode is adopted: the pulse hot air frequency dynamically ranges from 0.5 to 2 Hz, which is real-time regulated by the digital twin system; synchronous infrared focusing welding is adopted to precisely deliver energy to the QFN ground pad; the holding time is 15 seconds; Step three: gradient slow cooling phase, temperature range 235-180℃: First-stage slow cooling zone: temperature drops from 235 to 220℃ in 10 seconds; the cooling rate is controlled at -1.5℃ / s; Second-stage slow cooling zone: temperature drops from 220 to 180℃ in 15 seconds; the cooling rate is controlled at -0.8℃ / s; Real-time scanning of PCB deformation by laser displacement sensor to predict warping risk areas; Ceramic needle array support system supports PCB with a dynamic response accuracy of ±0.1mm; Step four: intensive air cooling phase, temperature range 180-50℃: Dual air knife collaborative cooling system is adopted: Upper air knife: 45° oblique angle laminar flow design, avoiding precision components through computer vision; Lower air knife: vortex cooling technology, vortex intensity is real-time adjusted by PID controller; Temperature uniformity is controlled within ±2℃; Cooling rate is increased to 6℃ / s.

3. The reflow soldering process of claim 2, wherein: The operation logic of the digital twin system in step two is as follows: sensor network collects real-time data of temperature, deformation, and gas concentration; edge computing nodes perform data preprocessing and feature extraction; Digital twin decision center generates process parameter optimization instructions; Actuators dynamically regulate zone power, air speed, and cooling intensity; Quality detection results are fed back to the AI model for closed-loop optimization.

4. The reflow soldering process of claim 2, wherein: The digital twin system in step two analyzes X-ray detection data through a long short-term memory network to predict the probability of solder voids in advance; and dynamically adjusts the patch pressure and hot air angle based on the nanosolder paste wetness monitoring results.

5. The reflow soldering process of claim 2, wherein: The digital twin system in step two shares encrypted model parameters across production lines through a federated learning platform to continuously optimize the temperature control strategy; and analyzes the solder paste composition online to automatically recommend the weight percentage of nanocopper doping range between 0.08 and 0.

12.

6. The reflow soldering process of claim 2, wherein: The AI-controlled closed-loop process during the pulse reflow stage includes the following steps: Step 1: Real-time temperature field scanning: a. Use an infrared thermal imager to scan the PCB surface with a resolution of 0.1℃; b. Scanning frequency: collect temperature distribution data of the entire board every 200ms; c. Key monitoring points: BGA solder joints, QFN ground pads, and heat-sensitive component areas; Step 2: Digital twin engine decision: Input data: current temperature field matrix, preheating stage historical thermal load data, and component layout coordinate information; perform two core calculations in parallel: Calculation task A: predict the real-time liquid phase rate of BGA solder joints, based on the transient thermodynamic model, simulate the solder melting state, output the liquid phase rate deviation value, and set the target value of the liquid phase rate deviation to not less than 99.5%; Calculation task B: identify QFN thermal lag areas by comparing the design temperature with the actual temperature through a convolutional neural network, and mark the coordinates of high-risk areas with a temperature difference greater than 5℃. Step 3: Dynamic control of pulse hot air a. Receive liquid phase rate deviation data stream b. Execute pulse frequency optimization algorithm: If the liquid phase rate is less than 98%, increase the pulse frequency to 1.8-2.0 Hz; If the liquid phase rate is greater than 99% and the local temperature is greater than 232℃, reduce the frequency to 0.8 Hz; Otherwise, maintain the default frequency of 1.2 Hz; c. Control instructions are issued to the hot air generator, with a response delay of less than 50 ms; Step 4: Infrared reinforcement radiation activation a. For the marked thermal hysteresis area: the positioning system maps the coordinates to the infrared emission array; automatically match the wavelength, QFN area fixed at 4.5 μm; b. Radiation intensity is dynamically adjusted according to the formula: reinforcement power = basic value * 0.35 + (temperature difference - 5℃) * 0.02; Step 5: Temperature uniformity verification a. Start a second scan after reinforcement operation: focus on monitoring the original thermal hysteresis area; calculate the maximum temperature difference △T-max of the whole board; b. Decision branch: if ΔT-max ≤ 5℃, it is judged to be up to standard, and enters the 15-second holding stage; if ΔT-max > 5℃, return to step 2 for re-optimization; Step 6: Holding stage monitoring a. Maintain peak temperature 235℃ ± 3℃ b. Real-time detection of abnormal state: tin bead splashing monitoring: capture the burst sound frequency signal through acoustic sensor; component overheating protection: when the MLCC area is greater than 215℃, cut off the local infrared source.

Images