A solder precision positioning and high-temperature welding process system for optical module packaging
By introducing a temperature field micro-sensing module and a temperature gradient benchmark model, combined with multi-field synergistic heating and adaptive optimization of process parameters, the problem of unknown temperature gradient in optical module packaging solder welding was solved, achieving precise control of temperature gradient distribution and improving solder positioning accuracy, thereby enhancing the stability and consistency of packaging yield.
Patent Information
- Application Number
- CN202611214880.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-11
- Publication Date
- 2026-11-06
AI Technical Summary
In existing optical module packaging solder welding technology, the temperature gradient distribution is unknown and uncontrollable, resulting in poor solder positioning accuracy, unstable packaging yield, difficulty in achieving online accurate sensing and quantitative characterization of temperature gradient distribution, and inability to establish a quantitative relationship model between temperature gradient and solder wetting behavior and thermal stress distribution. Empirical parameters obtained by trial and error are difficult to adapt to solder shape differences and batch dispersion.
By employing a temperature field microscopic sensing module and a temperature gradient benchmark model module, the spatial distribution information of the temperature field in the solder joint area is obtained through infrared microscopic thermal imaging and a micro-area temperature sensor array. The temperature gradient vector field is calculated, and a quantitative relationship model between the temperature gradient distribution and the solder wetting behavior and thermal stress distribution is established to achieve precise closed-loop control of the temperature gradient distribution. Differentiated power allocation is performed using a multi-field collaborative heating module and a precise heating power allocation unit. Combined with a process parameter adaptive optimization module and a machine learning optimization engine, automatic optimization and adaptive adjustment of process parameters are achieved.
It achieves online precise sensing and quantitative characterization of temperature gradient distribution, ensuring solder positioning accuracy and improving the stability and consistency of packaging yield. Through closed-loop control and adaptive optimization, it overcomes the shortcomings of parameter tuning in existing technologies and improves the accuracy and reliability of solder welding.
Smart Images
Figure CN123047266A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of optoelectronic technology, specifically a solder precision positioning and high-temperature welding process system for optical module packaging. Background Technology
[0002] Optical modules are functional components of optical communication systems. Their packaging quality directly determines the transmission performance of optical signals and the long-term reliability of the system. In the optical module packaging process, the optoelectronic device chip and the fiber array need to be precisely physically connected by solder. The solder welding quality directly affects the insertion loss, return loss index and service stability of the optical module. Therefore, the solder welding process is a key process that determines the packaging yield and overall performance level of the optical module. The thermodynamic behavior of the solder in the molten state is precisely controlled by the temperature field. The temperature gradient, as a spatial distribution parameter of the temperature field, directly determines the surface tension distribution, wetting and spreading path and internal thermal stress distribution of the solder after melting. It is a control factor that affects the solder positioning accuracy and the consistency of the solder joint morphology.
[0003] Existing optical module packaging solder welding technology generally adopts a closed-loop control scheme with single-point temperature feedback. The temperature value of a certain monitoring point in the solder joint area is used as the sole control target. The welding process and heating effect are judged by monitoring whether the temperature at this point reaches the solder melting point threshold. The power adjustment of the heating system is only used to change the overall temperature level of the solder joint area. It does not have the ability to target and control the spatial distribution of the temperature field. The temperature field distribution status mostly relies on offline infrared thermal imagers for sampling inspection or on process experience. In engineering practice, the heating process parameters are mainly adjusted by engineers through repeated trial and error to adapt to the production needs of solder welding of different specifications and batches.
[0004] This single-point absolute temperature control paradigm has long left the temperature gradient, a key variable determining solder melting behavior, in a state of uncertainty, uncontrollability, and inability to adjust. Single-point temperature data can only reflect the temperature at a local location and cannot characterize the spatial distribution of the temperature field. The flow direction and displacement of solder in the molten state are essentially driven directly by the surface tension gradient induced by the temperature gradient. When the temperature gradient distribution is asymmetrical, the solder will spontaneously migrate to the low-temperature region under the action of surface tension, causing the solder to deviate from the original design position. At the same time, due to the small size of the solder welding area, which is usually less than 1 square millimeter, conventional infrared thermal imagers have problems such as weak signal in the target area and insufficient spatial resolution, making it difficult to achieve online and accurate perception of the temperature gradient distribution. Consequently, it is impossible to establish a quantitative relationship model between the temperature gradient distribution and the solder wetting behavior and the thermal stress distribution of the solder joint. Empirical parameters obtained by trial and error are difficult to stably adapt to the influence of solder shape differences, batch dispersion, and environmental temperature changes, resulting in large fluctuations in packaging yield. Moreover, the existing heating control system has no directional adjustment capability for the temperature gradient, and even if the gradient distribution deviation is identified, it cannot be accurately corrected by power adjustment.
[0005] Therefore, the present invention provides a solder precision positioning and high-temperature welding process system for optical module packaging. Summary of the Invention
[0006] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.
[0007] The technical solution adopted by this invention to solve its technical problem is: a solder precision positioning and high-temperature welding process system for optical module packaging, the system including a processing link, the processing link being configured as follows: Acquire temperature field data in the solder joint area, and trigger precise allocation of heating power when the temperature gradient deviation exceeds the tolerance limit; The solder wetting angle deviation is obtained by performing surface tension gradient calculation on the actual temperature gradient distribution; Calculate the differentiated power allocation scheme for each heating unit based on the solder wetting angle deviation and output the heating power adjustment command; Based on the solder melting state monitoring data, a temperature gradient reshaping scheme was determined and the heating power of each heating unit was adjusted to change the temperature gradient distribution pattern. And an indication of the degree of deviation between the output solder joint morphology and the target specifications.
[0008] Preferably, the processing link acquires the temperature field data of the solder joint area by calculating the temperature gradient vector field in real time.
[0009] Preferably, the processing link performs surface tension gradient calculation on the actual temperature gradient distribution, including calculating the corresponding solder surface tension distribution based on the temperature gradient distribution, and predicting the flow direction and displacement trend of the solder in the molten state.
[0010] Preferably, the processing link also establishes a displacement trajectory prediction model based on the historical sequence data of the temperature gradient vector field and the thermodynamic parameters of the solder, so as to predict in advance the movement path and endpoint position of the solder under the current temperature gradient distribution.
[0011] Preferably, the processing link determines the temperature gradient reshaping scheme based on solder melting state monitoring data by using optical methods to collect the morphological feature data of the solder pool in real time and calculate the offset of the solder relative to the target position.
[0012] Preferably, when the solder offset exceeds a preset threshold, the processing link triggers a temperature gradient reshaping scheme, which first adjusts the heating power on the low-temperature side to adjust the temperature on that side, and then adjusts the heating power on the high-temperature side to optimize the overall temperature gradient distribution.
[0013] Preferably, the processing link further includes a process parameter database, which stores a set of process parameters for different solder types and different optical module packaging structures, and determines the initial power allocation scheme for each heating unit based on the process parameter database.
[0014] Preferably, the processing link establishes a nonlinear mapping relationship between temperature gradient distribution parameters and solder joint quality indicators by using a gradient boosting decision tree algorithm to achieve automatic optimization and adaptive adjustment of process parameters.
[0015] Preferably, the processing link adopts a cascade control architecture, with the temperature gradient deviation as the outer loop main control quantity and the solder offset as the inner loop main control quantity, and introduces feedforward compensation based on the thermal inertia of the solder to output heating power adjustment commands to each heating unit in real time.
[0016] Preferably, the deviation indicator of the solder joint morphology from the target specification output by the processing link includes real-time monitoring of solder surface curvature and sedimentation behavior, extracting solder joint morphology feature data, comparing the solder joint morphology feature data with preset quality standards, and generating a process quality score.
[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention introduces a temperature field microscopic sensing module and a temperature gradient reference model module, enabling online precise sensing and quantitative characterization of temperature gradient distribution in the field of optical module packaging solder welding. By acquiring spatial distribution information of the temperature field in the solder joint area through infrared microscopic thermal imaging and a micro-area temperature sensor array, the temperature gradient vector field distribution data is calculated, which can solve the problem of unknown temperature gradient distribution in the prior art and provide a data foundation for subsequent precise control.
[0018] 2. This invention establishes a quantitative relationship model between temperature gradient distribution and solder wetting behavior and thermal stress distribution, and constructs a temperature gradient benchmark model, breaking through the bottleneck of the lack of a temperature gradient benchmark model in the prior art. Through the synergistic effect of the solder wetting angle prediction submodule, the thermal stress distribution prediction submodule, and the solder displacement trajectory prediction submodule, it provides clear evaluation criteria and optimization objectives for the optimization of process parameters.
[0019] 3. This invention, through the cooperation of a multi-field collaborative heating module and a precise heating power distribution unit, can achieve precise closed-loop control of temperature gradient distribution. Multiple independent and controllable local heating sources can perform differentiated power distribution according to the instructions of the closed-loop control center to form a target temperature gradient distribution. When the solder deviates, the temperature gradient reshaping scheme can be automatically triggered to eliminate the temperature gradient deviation and ensure the solder positioning accuracy.
[0020] 4. By introducing a process parameter adaptive optimization module and a machine learning optimization engine, this invention can achieve automatic optimization and adaptive adjustment of process parameters. Based on historical process data and real-time quality feedback, the baseline values of process parameters are continuously updated, which can overcome the problem of failure of parameter debugging in different scenarios in the prior art through trial and error, and is conducive to improving the stability and consistency of packaging yield. Attached Figure Description
[0021] The invention will now be further described with reference to the accompanying drawings.
[0022] Figure 1 This is a structural block diagram of a solder precision positioning and high-temperature welding process system for optical module packaging in this invention; Figure 2 This is a schematic diagram illustrating the principle framework of the closed-loop precise control mechanism for temperature gradient in this invention. Detailed Implementation
[0023] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.
[0024] Example 1: This example is applied to the solder fusion process of optical module packaging, specifically targeting the high-temperature fusion positioning operation of solder between optoelectronic device chips and fiber arrays. This optical module packaging solder fusion process involves the application and precise control of heat energy in a tiny solder joint area. The area of the solder joint area is usually less than 1 square millimeter, and the behavior of the solder in the molten state is directly affected by the temperature gradient distribution.
[0025] like Figure 1 and Figure 2 As shown, the solder precise positioning and high-temperature welding process system for optical module packaging according to the present invention includes a temperature field microscopic sensing module, a temperature gradient reference model module, a multi-field collaborative heating module, a solder position real-time sensing module, a process parameter adaptive optimization module, a closed-loop control center, a heating power precise allocation unit, a process execution terminal, and a process quality real-time evaluation unit.
[0026] The system's overall architecture is built on a processing link, with each module forming a data path and control link through communication connections, enabling online accurate sensing of temperature gradient distribution, benchmark model comparison, precise allocation of heating power, and closed-loop control of temperature gradient.
[0027] The temperature field micro-sensing module is set near the solder joint area to obtain the spatial distribution information of the temperature field in the solder joint area. The temperature field micro-sensing module includes an infrared micro-thermal imaging unit and a micro-area temperature sensor array. The spatial resolution of the infrared micro-thermal imaging unit is better than or equal to 10 micrometers and is used to capture the temperature field distribution image of the solder area. The micro-area temperature sensor array contains at least 16 distributed temperature sensing nodes, which are evenly arranged around the solder joint. Each sensing node is connected to the temperature gradient calculation unit to collect temperature field data in real time and calculate two-dimensional temperature gradient vector field data.
[0028] The temperature field microscopic sensing module also includes a fiber-coupled infrared microscope and a micro-scanning galvanometer. The micro-scanning galvanometer performs grating scanning sampling on the solder joint surface, with a sampling interval of less than or equal to 50 micrometers. The fiber-coupled infrared microscope acquires infrared radiation signals from the solder joint area, and the micro-scanning galvanometer performs grating scanning with a sampling interval set to 50 micrometers. The temperature field micro-sensing module calculates the two-dimensional temperature gradient vector distribution across the entire field in real time. The temperature gradient calculation frequency is greater than or equal to 100 Hz, ensuring that the dynamic response speed of the temperature gradient distribution data matches the displacement speed of the molten solder. The temperature field micro-sensing module sends the calculated temperature gradient vector field data to the temperature gradient reference model module through the communication interface.
[0029] The temperature gradient reference model module is built into the storage unit of the closed-loop control center and is used to store the optimal temperature gradient distribution template. The temperature gradient reference model module has a built-in quantitative relationship model between the temperature gradient distribution and the solder wetting behavior. This quantitative relationship model stores the functional mapping relationship between the solder surface tension gradient and the temperature gradient. The temperature gradient benchmark model module receives actual temperature gradient distribution data from the temperature field micro-sensing module, compares the actual temperature gradient distribution with the optimal temperature gradient distribution template, calculates the temperature gradient deviation, and calculates the corresponding solder surface tension distribution based on the real-time temperature gradient distribution to predict the flow direction and displacement trend of the solder in the molten state.
[0030] The temperature gradient benchmark model module also includes a solder wetting angle prediction submodule and a thermal stress distribution prediction submodule. The solder wetting angle prediction submodule calculates the deviation between the target value and the actual value of the solder wetting angle based on the temperature gradient distribution. The calculation method is as follows: obtain the isotherm shape of the solder pool under the current temperature gradient distribution, calculate the actual wetting angle value based on the isotherm shape, and subtract the actual wetting angle value from the preset target wetting angle value to obtain the wetting angle deviation. The thermal stress distribution prediction submodule predicts the thermal stress distribution state of the weld joint area based on the temperature gradient distribution. The calculation method is as follows: extract the spatial derivative information of the temperature gradient distribution, convert the spatial derivative of the temperature gradient into a thermal strain distribution based on the coefficient of thermal expansion, and convert the thermal strain distribution into a thermal stress distribution based on the elastic modulus.
[0031] The temperature gradient baseline model module also includes a solder displacement trajectory prediction submodule. This submodule establishes a displacement trajectory prediction model based on historical sequence data of the temperature gradient vector field and solder thermodynamic parameters. It continuously receives and stores time-series data of the temperature gradient vector field, with the storage window length set to the data of the most recent 30 sampling periods. Based on the stored historical sequence data of the temperature gradient vector field, the submodule calculates the time change rate of the surface tension gradient, establishes a motion equation based on solder density and viscosity parameters, and solves the motion equation to obtain the solder displacement trajectory prediction result. This submodule can predict the movement path and endpoint position of the solder under the current temperature gradient distribution in advance.
[0032] The temperature gradient benchmark model module sends the temperature gradient deviation, solder wetting angle deviation, thermal stress distribution data, and solder displacement trajectory prediction results to the closed-loop control center.
[0033] The closed-loop control center is located in the system control architecture to receive the temperature gradient deviation and perform fusion calculations to generate heating power adjustment commands. The closed-loop control center adopts a cascade control architecture, with the temperature gradient deviation as the main control variable in the outer loop and the solder offset as the main control variable in the inner loop. The outer loop control loop of the cascade control architecture includes: acquiring the temperature gradient deviation as the outer loop setpoint, comparing the outer loop setpoint with the outer loop feedback value to obtain the outer loop deviation value, inputting the outer loop deviation value into the outer loop proportional-integral-derivative controller for calculation, and outputting the outer loop control quantity. The inner loop control circuit of the cascade control architecture includes: acquiring the solder offset as the inner loop setpoint, comparing the inner loop setpoint with the inner loop feedback value to obtain the inner loop deviation value, inputting the inner loop deviation value into the inner loop proportional-integral-derivative controller for calculation, and outputting the inner loop control quantity; The closed-loop control center introduces a feedforward compensation module based on the thermal inertia of the solder. The feedforward compensation module calculates the thermal inertia delay time of temperature change based on the heat capacity and thermal conductivity coefficient of the solder, and superimposes the feedforward compensation amount when outputting the control quantity to offset the impact of thermal inertia delay on the control response speed. The control cycle of the closed-loop control center is less than or equal to 1 millisecond, ensuring real-time output of heating power adjustment commands. The closed-loop control center sends the generated heating power adjustment commands to the heating power precision allocation unit.
[0034] The precise heating power allocation unit is communicatively connected to the closed-loop control center and is used to decompose the heating power adjustment command into the target power value of each heating unit; the precise heating power allocation unit receives the heating power adjustment command from the closed-loop control center, which includes the target power adjustment amount of each local heating source; The precise heating power allocation unit calculates the difference between the current power status of each local heating source and the target power adjustment amount, and converts the difference into the adjustment amount of the pulse width modulation signal duty cycle or current amplitude. The control accuracy of the precise heating power allocation unit is greater than or equal to 12 bits, which can realize the fine allocation of heating power. The precise heating power allocation unit sends the target power value of each local heating source to the process execution terminal through the control signal line.
[0035] The process execution terminal is electrically coupled to the multi-field coordinated heating module to receive the target power value and perform heating operations. The process execution terminal includes a power drive circuit and an interface conversion circuit. The power drive circuit receives the control signal from the heating power precision allocation unit and converts the control signal into the corresponding drive current or drive voltage and outputs it to each local heating source in the multi-field coordinated heating module.
[0036] The multi-field coordinated heating module is connected to the solder position real-time sensing module. The multi-field coordinated heating module includes multiple independent and controllable local heating sources and auxiliary airflow compensation units. Each local heating source is distributed around the solder joint area, and each heating source has independent power control capability. The multi-field coordinated heating module receives adjustment commands from the precise heating power allocation unit and differentiates the power of each local heating source to form a target temperature gradient distribution. Each local heating source is implemented using one or more combinations of laser arrays, induction coils, or micro-heating probes, and the heating power control accuracy of each local heating source is greater than or equal to 12 bits. The auxiliary airflow compensation unit is used to provide a thermal compensation atmosphere, and the airflow temperature control accuracy of the auxiliary airflow compensation unit is better than or equal to 5 degrees Celsius. The auxiliary airflow compensation unit includes an airflow heater and an airflow flow controller. The airflow heater adjusts the temperature of the compensation gas to the target value, and the airflow flow controller adjusts the flow rate and injection direction of the compensation gas to form a stable airflow field in the solder joint area, thereby assisting in adjusting the temperature distribution in the solder joint area.
[0037] The real-time solder position sensing module is used to monitor the solder melting state and position offset. The real-time solder position sensing module includes a solder melting state monitoring submodule and an offset calculation unit. The solder melting state monitoring submodule uses optical methods to collect the morphological feature data of the solder molten pool in real time. The solder melting state monitoring submodule includes an industrial camera and an image acquisition card. The industrial camera is set above the solder joint area to collect visible light images or near-infrared images of the solder molten pool. The image acquisition card preprocesses the images acquired by the industrial camera, including grayscale processing, contrast enhancement, and edge detection, and extracts the contour and morphological features of the solder pool. The offset calculation unit calculates the offset of the solder relative to the target position based on the morphological feature data acquired by the solder melting state monitoring submodule. The offset calculation unit performs template matching between the extracted solder pool contour and the preset solder target contour, calculates the difference in the coordinates of the center points of the two contours, and obtains the solder offset. The offset calculation unit transmits the offset data to the closed-loop control center and the real-time process quality evaluation unit in real time.
[0038] The real-time solder position sensing module triggers a temperature gradient reshaping scheme when the solder offset exceeds a preset threshold. The offset calculation unit compares the solder offset with the preset offset threshold in real time. When the solder offset is greater than or equal to the preset offset threshold, the offset calculation unit generates a temperature gradient reshaping scheme trigger signal and sends it to the closed-loop control center. The priority of the temperature gradient reshaping scheme is set as follows: The heating power on the low-temperature side is adjusted first to increase the temperature on that side. Then, based on the increase in the temperature on the low-temperature side, the heating power on the high-temperature side is adjusted to optimize the overall temperature gradient distribution. After receiving the temperature gradient reshaping scheme trigger signal, the closed-loop control center calculates the target power value that the low-temperature side heating source needs to increase and the target power value that the high-temperature side heating source needs to adjust, generates a new heating power adjustment command, and sends it to the heating power precise allocation unit.
[0039] The real-time process quality assessment unit is communicatively connected to the closed-loop control center, forming an auxiliary feedback channel for the temperature gradient closed-loop control loop. The real-time process quality assessment unit monitors the surface curvature and sedimentation behavior of the solder in real time. The real-time process quality assessment unit obtains the morphological characteristic data of the solder pool from the real-time solder position sensing module and the temperature field distribution data from the temperature field micro-sensing module. It processes and calculates the surface curvature parameters and sedimentation height parameters of the solder. The real-time process quality assessment unit extracts the weld joint morphology feature data and compares the weld joint morphology feature data with the preset quality standards. The preset quality standards are stored in the database of the real-time process quality assessment unit, including the allowable range of weld joint height, the allowable range of weld joint diameter, the allowable range of weld joint surface roughness, and the allowable range of weld joint void rate. The real-time process quality assessment unit compares the extracted actual weld joint morphology feature data with the preset quality standards item by item, calculates the deviation percentage of each indicator, generates a process quality score, and transmits the process quality score to the closed-loop control center as an auxiliary basis for control decision-making. At the same time, the real-time process quality assessment unit sends the process quality score and weld joint morphology feature data to the process parameter adaptive optimization module.
[0040] The process parameter adaptive optimization module is connected to the closed-loop control center and is used to update process parameters based on real-time data. The process parameter adaptive optimization module has a built-in process parameter database and machine learning optimization engine. The process parameter database stores a set of process parameters for different solder types and different optical module packaging structures. The process parameters in the database include the initial power allocation ratio of each local heating source, the heating power rise rate curve, the target range of temperature gradient distribution, and process time parameters. The machine learning optimization engine continuously updates the baseline values of the process parameters based on feedback data from the real-time process quality evaluation unit. The machine learning optimization engine establishes a nonlinear mapping relationship between temperature gradient distribution parameters and solder joint quality indicators. The engine employs a gradient boosting decision tree algorithm to achieve automatic optimization and adaptive adjustment of process parameters. The gradient boosting decision tree algorithm uses temperature gradient distribution parameters as input features and solder joint quality indicators as output labels. It constructs an integrated decision tree model through multiple iterations. In each iteration, the residual between the predicted value and the actual value of the previous iteration is calculated, and the residual is used as the training target of the decision tree in the next iteration. Finally, the prediction results of multiple decision trees are superimposed to obtain the optimized process parameter prediction value. The machine learning optimization engine continuously receives solder joint quality data from the real-time process quality assessment unit. After each solder joint quality assessment is completed, the decision tree model parameters are updated once, so that the baseline values of process parameters gradually converge to the optimal values as the production process progresses.
[0041] The system also includes a process database and a process knowledge graph module. The process database stores historical process execution records and corresponding solder joint quality data. The process knowledge graph module constructs the correlation between solder type, package structure, temperature gradient parameters, and solder joint quality based on historical data, assisting the process parameter adaptive optimization module in recommending parameters. The process knowledge graph module associates and models the solder type nodes, package structure nodes, temperature gradient parameter nodes, and solder joint quality nodes in the historical process data to form a knowledge graph topology. When the process parameter adaptive optimization module needs to determine the initial process parameters for a new solder type or a new package structure, it retrieves parameters from similar historical cases through knowledge graph queries as a reference starting point.
[0042] The workflow of this system and the collaborative operation of each module include: the process parameter adaptive optimization module determines the initial process parameter set for this solder welding process based on the correlation between historical data in the process database and the process knowledge graph. The initial process parameter set includes the initial power distribution ratio of each local heating source, the target temperature gradient distribution template, and the initial values of the control parameters of the heating power precision distribution unit. The process parameter adaptive optimization module sends the initial set of process parameters to the closed-loop control center and the heating power precise allocation unit.
[0043] After receiving the initial set of process parameters, the closed-loop control center loads the target temperature gradient distribution template into the temperature gradient reference model module. The temperature gradient reference model module sets the target temperature gradient distribution template as an internal comparison reference for subsequent calculation of the temperature gradient deviation.
[0044] The process execution terminal outputs initial drive current or initial drive voltage to each local heating source in the multi-field collaborative heating module according to the initial power allocation ratio of each local heating source in the initial process parameter set. Each local heating source begins to provide heat energy to the solder joint area, and the solder begins to heat up.
[0045] During the heating process, the temperature field micro-sensing module continuously collects temperature field data of the solder joint area, and the infrared micro-thermal imaging unit captures the temperature field distribution image of the solder area in real time with a spatial resolution of 10 micrometers. The 16 distributed temperature sensing nodes of the micro-area temperature sensor array synchronously collect the temperature values of each point around the solder joint. The fiber-coupled infrared microscope and the micro-scanning galvanometer work together. The micro-scanning galvanometer performs grating scanning sampling on the surface of the solder joint with a sampling interval of 50 micrometers. The temperature field micro-sensing module processes the collected raw temperature field data and calculates the full-field two-dimensional temperature gradient vector field data. The temperature gradient calculation frequency is set to 100 Hz. The temperature field micro-sensing module sends the calculated temperature gradient vector field data to the temperature gradient reference model module in real time.
[0046] After receiving the temperature gradient vector field data, the temperature gradient benchmark model module includes: The first step is to compare the actual temperature gradient distribution with the optimal temperature gradient distribution template at corresponding spatial points and calculate the temperature gradient deviation at each spatial location. The second step is to calculate the surface tension distribution of the solder based on the functional mapping relationship between the temperature gradient distribution and the surface tension gradient of the solder. The third step is to calculate the predicted value of the solder wetting angle based on the surface tension distribution of the solder, and compare the predicted value with the target wetting angle value to obtain the wetting angle deviation. The fourth step is for the thermal stress distribution prediction submodule to calculate the thermal stress distribution state of the weld joint area based on the temperature gradient distribution. The fifth step is to predict the solder displacement trajectory based on the historical sequence data of the temperature gradient vector field and the solder thermodynamic parameters.
[0047] The temperature gradient benchmark model module sends the processing results to the closed-loop control center. The processing results include temperature gradient deviation, solder surface tension distribution, solder wetting angle deviation, thermal stress distribution, and solder displacement trajectory prediction results.
[0048] After receiving and processing the results, the closed-loop control center performs cascade control calculations. The outer loop of the cascade control calculations uses the temperature gradient deviation as the main control variable. The temperature gradient deviation is input into the outer loop proportional-integral-derivative controller for calculation, and the outer loop control variable is output. The inner loop of the cascade control operation uses the solder offset as the main control variable. The solder offset is input into the inner loop proportional-integral-derivative controller for calculation, and the inner loop control variable is output. The closed loop control center introduces a feedforward compensation variable based on the thermal inertia of the solder in the feedforward channel, which is superimposed on the synthesis result of the outer loop control variable and the inner loop control variable to generate the final heating power adjustment command. The heating power adjustment command includes the target power adjustment value of each local heating source. The control period of the closed loop control center is set to 1 millisecond. The closed loop control center sends the heating power adjustment command to the heating power precision allocation unit.
[0049] After receiving the heating power adjustment command, the heating power precise allocation unit decomposes the adjustment command into specific power adjustment target values for each local heating source. The heating power precise allocation unit calculates the required output pulse width modulation signal duty cycle or current amplitude adjustment amount based on the current power status of each local heating source. The heating power precise allocation unit sends the target power value of each local heating source to the process execution terminal through the control signal line.
[0050] After receiving the target power value, the process execution terminal converts the target power value into the corresponding drive current or drive voltage through the power drive circuit and outputs it to each local heating source in the multi-field collaborative heating module. Each local heating source adjusts its own heating power output according to the received drive signal, so that the temperature gradient distribution in the solder joint area converges to the target distribution.
[0051] The solder position real-time sensing module synchronously monitors the solder position during the heating process. The industrial camera of the solder melting state monitoring submodule acquires visible light or near-infrared images of the solder pool in real time. The image acquisition card performs grayscale processing, contrast enhancement and edge detection on the acquired images to extract the contour and morphological features of the solder pool. The offset calculation unit performs template matching between the extracted solder pool contour and the preset solder target contour, calculates the difference between the coordinates of the solder center point and the coordinates of the target center point, and obtains the solder offset. The offset calculation unit transmits the solder offset data to the closed-loop control center in real time.
[0052] When the solder offset exceeds the preset offset threshold, the offset calculation unit triggers the temperature gradient reshaping scheme. The temperature gradient reshaping scheme includes: determining the direction of the solder offset and the low temperature side position, calculating the target power value that the low temperature side heating source needs to increase, and sending the low temperature side heating power increase command to the heating power precise allocation unit. Based on the increase in temperature on the low-temperature side, the overall temperature gradient distribution is recalculated, the target power value that the heating source on the high-temperature side needs to be adjusted is calculated, and the heating power adjustment command on the high-temperature side is sent to the heating power precision allocation unit. The multi-field collaborative heating module reallocates the power of each local heating source according to the adjusted heating power command, so that the temperature gradient distribution is optimized and the solder is driven to return to the target position.
[0053] The real-time process quality assessment unit monitors the surface curvature and settling behavior of the solder in real time. It obtains the morphological characteristic data of the solder pool from the real-time solder position sensing module and the temperature field distribution data from the temperature field micro-sensing module. The real-time process quality assessment unit processes the morphological characteristic data, calculates the solder surface curvature parameters and settling height parameters, compares the calculation results with the preset quality standards, generates a process quality score, and sends the process quality score to the closed-loop control center and the process parameter adaptive optimization module.
[0054] The machine learning optimization engine of the process parameter adaptive optimization module receives process quality score feedback data. The machine learning optimization engine adopts the gradient boosting decision tree algorithm, using the temperature gradient distribution parameters of the current process execution as input features and the current solder joint quality score as output labels to update and train the decision tree ensemble model. The machine learning optimization engine outputs the optimized process parameter baseline value prediction result, which is fed back to the closed-loop control center to update the initial value of the control parameter. The process parameter adaptive optimization module simultaneously updates the optimized process parameter baseline value to the corresponding record in the process database, realizing the adaptive adjustment of the process parameter.
[0055] The system continuously executes the closed-loop control cycle of the above workflow until the solder is fused and solidified to form the final solder joint. After the solder joint solidifies, the real-time process quality evaluation unit performs a final inspection of the solder joint morphology, extracts morphological feature data such as solder joint height, solder joint diameter, solder joint surface roughness, and solder joint void rate, compares the morphological feature data with the preset quality standards, generates the final process quality score, and outputs an indicator of the degree of deviation between the solder joint morphology and the target specifications. This indicator includes the deviation values of each morphological index and the final process quality score.
[0056] Example 2: At the system architecture level, the temperature field micro-sensing module in this embodiment increases the density of the micro-area temperature sensor array based on the configuration of embodiment one, increasing the total number of distributed temperature sensing nodes from 16 to 32. The node distribution covers the solder joint area and the cavity wall area within a 1 mm radius around it, so as to obtain more comprehensive temperature field spatial distribution data. The temperature gradient calculation frequency was increased from 100 Hz to 200 Hz to adapt to the faster temperature field changes in the cavity environment.
[0057] In the temperature gradient benchmark model module, the temperature gradient distribution template of this embodiment has been modified for the heat dissipation characteristics of the cavity structure. The influence factor of the cavity wall heat conduction boundary condition on the temperature gradient distribution has been incorporated. The displacement trajectory prediction model of the solder displacement trajectory prediction submodule has added the cavity wall constraint condition term, so that the prediction result is closer to the actual working condition of the cavity structure.
[0058] In the multi-field collaborative heating module, this embodiment increases the gas flow rate control accuracy of the auxiliary airflow compensation unit, improving the airflow temperature control accuracy from 5 degrees Celsius to 2 degrees Celsius. The airflow injection direction of the auxiliary airflow compensation unit is adjusted by an adjustable mechanism, which can adjust the coverage area and coverage intensity of the compensation gas according to the welding point position to compensate for the temperature gradient distribution distortion caused by heat dissipation from the cavity wall.
[0059] In the real-time solder position sensing module, the solder melting state monitoring submodule in this embodiment adds a near-infrared spectroscopy analysis function. By analyzing the spectral characteristics of the solder pool, the degree of solder oxidation and composition uniformity are determined, providing richer reference data for solder joint quality assessment.
[0060] In the temperature gradient reshaping scheme, this embodiment adds a cavity wall temperature compensation step to take into account the characteristics of the cavity structure. When the solder offset exceeds the preset threshold and triggers the temperature gradient reshaping scheme, the scheme execution process is modified as follows: adjust the power of the heating source located on the opposite side of the solder offset direction to increase the temperature on that side, and at the same time adjust the airflow parameters of the auxiliary airflow compensation unit located on the side of the solder offset direction to reduce the heat loss on that side, and comprehensively adjust the power distribution of each local heating source to form the corrected target temperature gradient distribution.
[0061] The adaptive optimization module for process parameters has been extended to include the set of process parameters with cavity structure. The airflow temperature and airflow velocity parameters of the auxiliary airflow compensation unit have been incorporated. The gradient boosting decision tree algorithm of the machine learning optimization engine is trained with the extended temperature gradient distribution parameters and auxiliary airflow parameters as input features, and a nonlinear mapping relationship between the multidimensional parameter space and the solder joint quality index has been established.
[0062] The workflow of this embodiment adds a parameter adjustment link for the auxiliary airflow compensation unit based on the first embodiment. When the closed-loop control center generates the heating power adjustment command, it simultaneously outputs the airflow parameter adjustment command for the auxiliary airflow compensation unit. The process execution terminal simultaneously drives the local heating source of the multi-field collaborative heating module to work in coordination with the auxiliary airflow compensation unit. The system continuously executes the closed-loop control cycle including the heat compensation link until the solder completes the fusion and solidifies to form the final solder joint.
[0063] Example 3: At the system architecture level, the spatial resolution of the infrared microscopic thermal imaging unit of the temperature field micro-sensing module in this embodiment is improved from 10 micrometers to 5 micrometers to meet the higher requirements for temperature field observation accuracy in the gold-tin alloy solder welding area. The sampling interval of the micro-scanning galvanometer is reduced from 50 micrometers to 25 micrometers, and the temperature gradient calculation frequency is maintained at 100 Hz to ensure the real-time performance of data processing.
[0064] In the temperature gradient baseline model module, the quantitative relationship model between the temperature gradient distribution and solder wetting behavior in this embodiment has been recalibrated based on the thermodynamic parameters of the gold-tin alloy solder. The correlation coefficient in the functional mapping relationship between the solder surface tension gradient and the temperature gradient has been corrected using measured data of the gold-tin alloy solder. The target wetting angle value of the solder wetting angle prediction submodule has been adjusted according to the process requirements of the gold-tin alloy solder. The displacement trajectory prediction model of the solder displacement trajectory prediction submodule updates the density, viscosity, and surface tension parameters of the gold-tin alloy solder to actual parameter values.
[0065] In the multi-field coordinated heating module, each local heating source in this embodiment is implemented using a laser array, and the heating power control accuracy is maintained at greater than or equal to 12 bits; Because of the high melting point of gold-tin alloy solder, the maximum output power of each local heating source has been increased accordingly to meet process requirements. The atmosphere composition of the auxiliary airflow compensation unit has been changed from ordinary nitrogen to inert protective gas to prevent the gold-tin alloy solder from oxidizing at high temperatures.
[0066] In the closed-loop control center, the parameters of the cascade control architecture in this embodiment have been specifically adjusted. The proportional coefficient, integral coefficient and derivative coefficient of the outer loop proportional-integral-derivative controller have been readjusted according to the thermal inertia and response characteristics of the gold-tin alloy solder. The control parameters of the inner loop proportional-integral-derivative controller were adjusted accordingly, and the thermal inertia delay time calculation parameters of the feedforward compensation module were updated to the heat capacity and thermal conductivity coefficient of the gold-tin alloy solder.
[0067] In the adaptive optimization module for process parameters, the process parameter database of this embodiment stores a set of exclusive process parameters for the gold-tin alloy solder type. The process parameters for gold-tin alloy solder include the initial power distribution ratio of each local heating source, the heating power rise rate curve, the target temperature gradient distribution template, the protective atmosphere flow rate parameter, and the process time parameter. The machine learning optimization engine uses the gradient boosting decision tree algorithm to establish a mapping relationship between the process parameters of gold-tin alloy solder and the solder joint quality indicators.
[0068] In the solder joint quality assessment stage, the real-time process quality assessment unit has established a dedicated quality standard database for the solder joint characteristics of gold-tin alloy solder. The quality assessment indicators for gold-tin alloy solder joints include the thickness of the intermetallic compound layer, the porosity of the solder joint interface, and the predicted value of the solder joint shear strength. The real-time process quality assessment unit extracts the solder joint morphology feature data and compares it with the dedicated quality standards to generate a process quality score for the gold-tin alloy solder process.
[0069] The workflow of this embodiment is basically the same as that of Embodiment 1. The functions and collaborative relationships of each module remain unchanged. Only the specific parameter configurations and model parameters have been adapted to suit the characteristics of the gold-tin alloy solder.
[0070] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.
Claims
1. A solder precision positioning and high-temperature welding process system for optical module packaging, characterized in that, The system includes a processing link, which is configured as follows: Acquire temperature field data in the solder joint area, and trigger precise allocation of heating power when the temperature gradient deviation exceeds the tolerance limit; The solder wetting angle deviation is obtained by performing surface tension gradient calculation on the actual temperature gradient distribution; Calculate the differentiated power allocation scheme for each heating unit based on the solder wetting angle deviation and output the heating power adjustment command; Based on the solder melting state monitoring data, a temperature gradient reshaping scheme was determined and the heating power of each heating unit was adjusted to change the temperature gradient distribution pattern. And an indication of the degree of deviation between the output solder joint morphology and the target specifications.
2. The solder precision positioning and high-temperature welding process system for optical module packaging according to claim 1, characterized in that, The processing link acquires temperature field data of the solder joint area, including real-time calculation of the temperature gradient vector field.
3. The solder precision positioning and high-temperature welding process system for optical module packaging according to claim 1, characterized in that, The processing link performs surface tension gradient calculation on the actual temperature gradient distribution, including calculating the corresponding solder surface tension distribution based on the temperature gradient distribution, and predicting the flow direction and displacement trend of the solder in the molten state.
4. The solder precision positioning and high-temperature welding process system for optical module packaging according to claim 1, characterized in that, The processing link also establishes a displacement trajectory prediction model based on the historical sequence data of the temperature gradient vector field and the thermodynamic parameters of the solder, so as to predict the movement path and endpoint position of the solder under the current temperature gradient distribution in advance.
5. The solder precision positioning and high-temperature welding process system for optical module packaging according to claim 1, characterized in that, The processing link determines the temperature gradient reshaping scheme based on solder melting state monitoring data, including using optical methods to collect the morphological characteristics data of the solder pool in real time and calculate the offset of the solder relative to the target position.
6. The solder precision positioning and high-temperature welding process system for optical module packaging according to claim 5, characterized in that, When the solder offset exceeds a preset threshold, the processing link triggers a temperature gradient reshaping scheme, which first adjusts the heating power on the low-temperature side to adjust the temperature on that side, and then adjusts the heating power on the high-temperature side to optimize the overall temperature gradient distribution.
7. The solder precision positioning and high-temperature welding process system for optical module packaging according to claim 1, characterized in that, The processing link also includes a process parameter database, which stores a set of process parameters for different solder types and different optical module packaging structures, and determines the initial power allocation scheme for each heating unit based on the process parameter database.
8. The solder precision positioning and high-temperature welding process system for optical module packaging according to claim 7, characterized in that, The processing link establishes a nonlinear mapping relationship between temperature gradient distribution parameters and solder joint quality indicators, including the use of gradient boosting decision tree algorithm to achieve automatic optimization and adaptive adjustment of process parameters.
9. The solder precision positioning and high-temperature welding process system for optical module packaging according to claim 1, characterized in that, The processing link adopts a cascade control architecture, with temperature gradient deviation as the outer loop main control quantity and solder offset as the inner loop main control quantity. Feedforward compensation based on solder thermal inertia is introduced, and heating power adjustment commands are output to each heating unit in real time.
10. The solder precision positioning and high-temperature welding process system for optical module packaging according to claim 1, characterized in that, The deviation indicator of the solder joint morphology from the target specification output by the processing link includes real-time monitoring of solder surface curvature and sedimentation behavior, extraction of solder joint morphology feature data, comparison of solder joint morphology feature data with preset quality standards, and generation of process quality score.