Printed circuit board resistance welding method

Abstract

The invention relates to a printed circuit board solder resisting method, which is characterized in that distinguishable miniature passive synchronous marks are preset in a PCB design stage, and optical phase coding is combined, so that multi-dimensional physical identifier embedding and automatic parameterization layout design are realized. Before and after exposure, a high-speed imaging detection and self-adaptive optical algorithm is adopted, phase response of multiple layers of synchronous marks is collected and normalized, multi-dimensional alignment reference features are extracted, a distribution alignment error model is established, overall rigid displacement and local stress deformation are distinguished, exposure equipment parameters are corrected in real time based on the model, and space error compensation is achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of "printed circuit board solder mask process, high-precision optical alignment and adaptive phase encoding", and particularly relates to a printed circuit board solder mask method. BACKGROUND

[0002] In the current field of printed circuit board (PCB) manufacturing, especially in the process of multi-layer PCB solder mask film exposure and photomask alignment, the alignment accuracy directly affects the product yield and reliability. With the development of electronic devices towards high density and high integration, the design structure of multi-layer PCB is becoming increasingly complex, and the line width and aperture are continuously shrinking, which puts forward higher and higher precision requirements for the exposure and alignment of solder mask film between multiple layers. The mainstream solder mask process generally uses photomask alignment and development integrated device, through the alignment target pre-set on the board or mask, with the help of high-resolution imaging and precise mechanical motion system, to realize the positioning and exposure of the solder mask film. However, with the increase of the number of layers and the complexity of the circuit, the sensitivity of the process to environmental disturbance is significantly enhanced.

[0003] The traditional technical route mainly relies on high-speed mechanical alignment device and target-based imaging recognition method. Most of the equipment uses double-color markers, two-dimensional codes or metal target points with shape recognition, through the visual system to introduce servo compensation algorithm, to realize the automatic alignment of the mask and the board. At the same time, some high-end manufacturing lines introduce local temperature and humidity sensing, real-time vibration monitoring and other environmental perception modules, to trigger the real-time fine adjustment compensation of the alignment equipment. However, the existing photomask alignment process usually has the following outstanding problems: (1) imaging recognition depends on the clarity of the marker, which is easy to cause target misjudgment when there is a small amount of dust or impurities on the board surface; (2) the response speed of the mechanical motion compensation of the equipment is limited, which is difficult to completely eliminate the influence of instantaneous micro-vibration or thermal deformation in the exposure and alignment process; (3) temperature and humidity changes will cause instantaneous expansion and contraction of the material, which is difficult to compensate consistently in mass production, and is easy to cause alignment error diffusion between batches; (4) for multi-layer stacked structure, due to the influence of stress, local deformation and other factors of each layer, single target or simple visual alignment cannot meet the high-precision multi-layer alignment requirements.

[0004] In recent years, some advanced manufacturing enterprises have tried to introduce digital image correlation (DIC) analysis, multi-spectral recognition or passive plug-in plastic markers to improve the alignment robustness, but still face the following technical bottlenecks: (1) the increase of active feedback and mechanical compensation modules greatly increases the system complexity and cost, which brings great pressure to equipment maintenance and production line scheduling; (2) the real-time environmental parameter acquisition and control precision is limited, which can only be analyzed and compensated afterwards, and it is difficult to realize the immediate and global perception and decoupling of the disturbance; (3) distributed multi-point monitoring and traditional target structure are difficult to balance the spatial resolution and anti-interference ability of high-density multi-layer area.

[0005] Especially in high-end multi-layer PCB solder mask process line, the following technical pain points often occur: during the solder mask film exposure process, as long as there is a slight vibration, temperature and humidity fluctuation or local dust fall, the traditional optical alignment method is easy to lose accuracy, and the correction of the traditional real-time mechanical compensation system is often difficult to completely eliminate errors due to the limitations of response and accuracy; on the other hand, due to the stacking difference of multi-layer structure, a single physical target or marker structure is difficult to reflect the overall interlayer space state, and the influence of the environment on the alignment parameters during batch production often shows unpredictable random drift, resulting in a significant increase in the rate of defective products in some batches. SUMMARY

[0006] The present application provides a printed circuit board solder mask method, which aims to solve one or more of the problems existing in the prior art mentioned in the background.

[0007] The printed circuit board solder mask method provided by the present application specifically comprises:

[0008] S1: Based on the PCB design data, a micro-passive synchronization marker is preset in the specified area of each layer of circuit board, and a distinguishable multi-layer alignment signal is formed through optical phase coding to realize the embedding of multi-dimensional physical identification.

[0009] S2: Before the solder mask film exposure, a high-speed imaging device is used to scan the multi-layer PCB and the synchronization marker on the mask surface globally, and the original optical response signal containing environmental differences and vibration interference is collected.

[0010] S3: The collected original optical response signal is preprocessed by denoising and normalization to eliminate invalid signal components generated by temperature and humidity, dust or slight deformation and other environmental disturbances, and the standardized phase response data of the multi-layer synchronization marker is obtained.

[0011] S4: The standardized phase response data is input into an adaptive optical processing algorithm, the optical phase difference of each micro-passive synchronization marker is extracted, and a multi-dimensional alignment reference feature set reflecting the working condition information of physical displacement, local deformation, etc. is extracted.

[0012] S5: Based on the above multi-dimensional alignment reference feature set, a distributed alignment error model of the multi-layer PCB under different environmental disturbance conditions is established, the spatial distribution characteristics of the overall rigid displacement and local stress deformation are distinguished, and the physical difference perception is realized.

[0013] S6: For the distributed alignment error model, an optical phase difference superposition algorithm is used to generate a virtual alignment plane of the current interlayer alignment series, and the difference compensation of the alignment data in the disturbance environment is realized.

[0014] S7: judge whether the current distribution alignment error model is in a preset threshold interval, if exceeding the threshold, automatically correct the alignment parameters of the exposure equipment and the mask pattern data according to the physical offset of each layer, and realize flexible parameter dynamic adaptation.

[0015] S8: in the solder resist film exposure process, the actual exposure operation is completed according to the corrected alignment parameters and mask pattern data, and the current batch working condition label and environmental parameters are recorded for subsequent tracing.

[0016] S9: after exposure, the synchronous mark area is detected by imaging again, the phase response of the synchronous mark after actual exposure is obtained, the final alignment residual error is calculated by comparing the initial multi-dimensional alignment reference, and dynamic feedback checking is realized.

[0017] S10: according to the dynamic feedback checking result, if the alignment residual error does not meet the process requirement, an abnormal alarm or a supplementary exposure process is triggered, and the residual error information is recorded to continuously optimize the distribution alignment error model and the adaptive optical processing algorithm, so as to improve the consistency and robustness of subsequent batch manufacturing.

[0018] The printed circuit board solder resist method provided by the application has the following beneficial effects:

[0019] (1) by embedding high-density and unique optical phase coding synchronous marks in the physical layer, the spatial reference distribution of multi-layer alignment is effectively realized. Different from the traditional mechanical positioning or single-point alignment mark, the present application can accurately perceive the whole rigid movement and local stress deformation under the influence of temperature and humidity fluctuations, equipment micro-vibration or dust particle interference, and other alignment error sources, by virtue of the multi-dimensional distributed optical phase signal. The adaptive optical algorithm can automatically distinguish and compensate the physical influence of various disturbances in real time. Experimental and batch data show that, compared with the existing process, the present application can reduce the alignment residual error standard deviation by 30% to 60%, and control the error main distribution within ±6um, which greatly exceeds the stability boundary relying on real-time servo compensation.

[0020] (2) the present application adopts a quantitative distribution error model and a virtual alignment plane technology, and reduces the complex working conditions under the coupling of "board-film-environment" three factors to data-driven flexible alignment compensation, realizes high consistency production without frequent shutdown calibration or manual secondary calibration. The whole process of automatic optical monitoring and compensation makes the average link time of the alignment process reduced by more than 20%, the batch yield fluctuation reduced to within 2%, significantly alleviates the batch instability problem caused by environmental changes, and effectively improves the size consistency and production of large batch products.

[0021] (3) As the system core relies on optical signal recognition and distribution modeling, the requirement for traditional high-cost real-time mechanical compensation is greatly weakened; and various disturbances are reduced to the corresponding distribution alignment error model for unified processing, which reduces the dependence on hardware such as equipment basic precision and environmental temperature control, and can achieve universal high-precision exposure in low-end automated exposure lines and various complex layers and material systems, significantly expanding the application range. Since error compensation is mainly achieved by data driving and optical algorithms, the conventional equipment iteration and maintenance period is extended, and the operation and maintenance cost is expected to decrease by 10% to 30%.

[0022] (4) The present application automatically collects batch environment tags, synchronously marks responses and exposure results throughout the whole process, and feeds back residual data to drive the distribution error model and optical processing strategy to continuously self-learn and optimize, establishing a traceable big data system for the whole link of process-environment-product. For abnormal batches and specific working conditions, the system can automatically attribute, trigger local compensation or process parameter evolution, support adaptive generalization of subsequent similar process platforms, different layers / material systems / mask designs, and effectively promote the upgrading of the solder mask exposure field to self-evolution and intelligent production mode. BRIEF DESCRIPTION OF DRAWINGS

[0023] FIG. 1 is a flowchart of a solder mask method for a printed circuit board. Figure 1

[0024] FIG. 2 is a sub-flowchart of the solder mask method for the printed circuit board. Figure 2

[0025] FIG. 3 is another sub-flowchart of the solder mask method for the printed circuit board. Figure 3 DETAILED DESCRIPTION

[0026] Embodiments of the present application are described in detail below with reference to the attached drawings, which show by way of example, embodiments in which the same or similar elements or elements having the same or similar functions are denoted by the same reference numbers throughout the drawings. The embodiments described below with reference to the drawings are exemplary and are intended only for the purpose of explaining the present application, and should not be construed as limiting the present application.

[0027] The disclosure below provides many different embodiments or examples for implementing different structures of the present application. For the purpose of simplification of the present application disclosure, the components and settings of specific examples are described below. Of course, they are only examples, and the purpose is not to limit the present application. In addition, the present application can repeatedly refer to numbers and / or letters in different examples, and such repetition is for the purpose of simplification and clarity, which itself does not indicate the relationship between the various embodiments and / or settings discussed. In addition, the present application provides examples of various specific processes and materials, but those skilled in the art can realize the application of other processes and the use of other materials.

[0028] ​​​As shown in the accompanying drawings Figure 1 The present application provides a printed circuit board solder mask method, specifically comprising:

[0029] S1: Based on the PCB design data, a micro-passive synchronization mark is preset in the specified area of each layer of circuit board, and a distinguishable multi-layer alignment signal is formed by optical phase coding to realize the embedding of multi-dimensional physical identification.

[0030] S2: Before the solder mask film exposure, a high-speed imaging device is used to globally scan the multi-layer PCB and the synchronization mark on the mask surface, and the original optical response signal containing environmental differences and vibration interference is collected.

[0031] S3: The collected original optical response signal is preprocessed by denoising and normalization to eliminate invalid signal components generated by temperature and humidity, dust or slight deformation and other environmental disturbances, and the standardized phase response data of the multi-layer synchronization mark is obtained.

[0032] S4: The standardized phase response data is input into an adaptive optical processing algorithm, and the multi-dimensional alignment reference feature set reflecting the working condition information such as physical displacement and local deformation is extracted by feature extraction of the optical phase difference of each micro-passive synchronization mark.

[0033] S5: Based on the above multi-dimensional alignment reference feature set, a distributed alignment error model of the multi-layer PCB under different environmental disturbance conditions is established to distinguish the spatial distribution characteristics of the overall rigid displacement and local stress deformation, and the physical difference perception is realized.

[0034] S6: For the distributed alignment error model, an optical phase difference superposition algorithm is used to generate a virtual alignment plane for the current inter-layer alignment series to realize the difference compensation of the alignment data in the disturbance environment.

[0035] S7: Determine whether the current distributed alignment error model is within the preset threshold range. If it exceeds the threshold, automatically correct the alignment parameters and mask pattern data of the exposure equipment according to the physical displacement of each layer to realize flexible parameter dynamic adaptation.

[0036] S8: During the solder mask film exposure process, the actual exposure operation is completed according to the corrected alignment parameters and mask pattern data, and the current batch working condition label and environmental parameters are recorded for subsequent tracing.

[0037] S9: After the exposure is completed, the synchronization mark area is detected again by imaging to obtain the phase response of the synchronization mark after the actual exposure, compare the initial multi-dimensional alignment reference, calculate the final alignment residual error, and realize dynamic feedback checking.

[0038] S10: According to the dynamic feedback check result, if the alignment residual does not meet the process requirement, trigger abnormal alarm or supplementary exposure process, and continuously optimize the distribution alignment error model and adaptive optical processing algorithm through the record residual information, to improve the consistency and robustness of subsequent batch manufacturing.

[0039] The step S1: based on the PCB design data, preset micro passive synchronization marks in each layer of the circuit board designated area, through the optical phase coding method, form distinguishable multi-layer alignment signal, to realize the embedding of multi-dimensional physical identification. Specifically includes:

[0040] S1.1: Structural analysis of PCB design data, extract the geometric layer structure, solder mask interval distribution and alignment accuracy requirement information of each layer in the target multi-layer PCB, to output the physical identification preset area list.

[0041] The complete original design data file of the multi-layer PCB is input, and the parameters such as the layer structure information, solder mask interval distribution and alignment accuracy requirement are structurally analyzed. The layer structure analysis algorithm (parameters: Gerber / ODB++ general PCB layout format) is used to realize the automatic layering and geometric information extraction of the multi-layer PCB design content. Further, through the rule-based solder mask interval identification method (parameters: solder mask area mark code, design annotation and design rule check table), the spatial distribution boundary and available area of each layer solder mask area are automatically extracted, and the distributed solder mask interval boundary coordinate data is output. Further, using the alignment accuracy requirement library (parameters: standard allowable tolerance, local high-precision alignment requirement, board layer stacking relationship matrix), the multi-layer structure alignment accuracy grading comparison algorithm is executed, the alignment accuracy grade label corresponding to different line density areas and functional modules is realized, and the grading label is output. Further, the space embeddable analysis algorithm (parameters: minimum safety distance rule between layers, solder mask area / conductor area repulsion relationship) is used to realize the idle space judgment of the physical identification embeddable area in each layer, to exclude the interference area related to electrical and mechanical functions, and to generate qualified physical identification candidate area. Through the data aggregation and conflict avoidance algorithm (parameters: minimum interval of adjacent layers, synchronization mark spacing threshold), the above candidate areas are screened to form the final physical identification preset area list that meets the alignment accuracy and safety specifications. Through the above structural analysis algorithm chain, the PCB design data is converted into layered, zoned and graded physical identification embedding basic data, realizing the bottom design data support capability of multi-layer PCB high-precision alignment.

[0042] For example, for a 12-layer high-density interconnection (HDI) PCB, its standard Gerber design file is imported, the layer structure parameters n = 12, the minimum layer spacing 100 μm, and the solder mask interval detection tolerance ± 30 μm are set in the parsing algorithm. Through automatic layer structure segmentation, the x-y dimension boundary of each layer and the z-axis stacking sequence are extracted. The solder mask coverage area and window area of each layer are located by using the solder mask interval identification rule, and a total of 108 groups of typical solder mask area coordinate blocks are output. The high-precision interval tolerance 10 μm and the standard interval tolerance 30 μm are set for the positioning accuracy database, and the regions such as BGA pins of the control chip are labeled as high-precision embedded regions. In the process of performing the embeddable analysis, the candidate regions overlapping with high-current traces (line width greater than 0.3 mm) and via regions (diameter greater than 0.2 mm) are filtered out, forming 8-12 groups of available identification embedding windows for each layer. Finally, the candidate region conflict filtering algorithm is used to remove overlapping candidate blocks with a cross-layer distance less than 200 μm or a same-layer distance less than 300 μm, and the final physical identification preset region list (data entries: 65 blocks in total for 9 layers) is output, laying a data foundation for the subsequent embedding and phase coding of the synchronous marker. After this step, the system can quickly locate the best physical identification embedding area under the complex, high-density multilayer PCB structure, significantly improving the physical foundation integrity and environmental adaptability of the positioning system.

[0043] S1.2: Based on the physical identification preset region list, a specific optical phase coding design algorithm is used to automatically assign phase coding parameters to each identification region of each layer, generate a unique set of synchronous marker parameters, and realize the generation of a multi-layer distinguishable synchronous signal pattern.

[0044] For each candidate region in the physical identification preset region list, the spatial coordinates, geometric dimensions, solder mask interval distribution of the layer where the region is located, and the positioning accuracy level label are input.

[0045] An optical phase coding design algorithm (parameters: coding type selection [binary phase / multi-level phase / wavelength multiplexing], minimum distinguishable phase step Δφ = π / 8, coding conflict minimum distance d min = 2, maximum assignable code word number M = region number x layer number) is used to realize the preliminary phase coding parameter assignment for each identification region of each layer.

[0046] Further, through a physical interlayer conflict avoidance algorithm (parameters: interlayer minimum code word Hamming distance h min = 3), combined with the sensitivity classification of the adjacent layer solder mask functional area, the phase coding that may cause cross-layer misjudgment or signal crosstalk is automatically excluded from the code word pool, the phase codes of adjacent layers and regions are spatially mapped and distinguished, and an interlayer safe code word distribution matrix is obtained.

[0047] Further, the multi-layer unique phase code generation strategy parameters are applied: regional center coordinates (Cx, Cy), level number L, encoding rule F(x, y, L), and the batch generation of unique synchronization marker parameters is realized according to the interlayer grouping and position sequence for all physical identification regions, and a synchronization marker parameter set is formed.

[0048] The conflict detection and concurrent allocation algorithm (parameters: same layer / cross layer minimum safety interval, effective signal window width Tw) is adopted to perform global traversal and allocation consistency verification on the generated synchronization marker parameter set, automatically identify and compensate the phase code allocation abnormalities caused by geometric proximity or functional overlap, and complete the stable mapping of the multi-layer distinguishable synchronization signal mode.

[0049] Through the above algorithm chain, the physical identification preset region list is converted into a unique synchronization marker parameter set which can be directly applied to the downstream synchronization marker layout design, optical integrated detection and phase decoding process, and the high consistency of the multi-layer PCB alignment parameters and the actual physical structure and the rapid recognizability under environmental disturbance are realized.

[0050] For example, in the identification embedding process of a 12-layer HDI PCB board, 8 groups of effective physical identification regions per layer are determined, and the actual number of synchronization markers to be allocated is 96. A multi-level phase encoding strategy is adopted, and the φ step is set to π / 8, so that 16 basic codes can be encoded in one phase cycle. For spatially adjacent regions, the minimum allowed Hamming distance is set to 4, the basic code is allocated through the spatial mapping algorithm F(x, y, L) = Mod(α·x+β·y+γ·L, 16), and the basic code is combined with the safety code word grouping cycle allocation to actually generate 12 groups of code word sequences. After conflict detection, for two adjacent identification regions with a geometric distance of 80 μm, the algorithm automatically increases the code word interval to avoid identification crossing. Finally, a set of synchronization marker parameter set with a length of 96 is output, each parameter contains level number, identification region coordinates and unique multi-level phase code. Under this parameter set, the laboratory imaging recognition test shows that the encoding and decoding of all regions have no cross and missing recognition, the misjudgment rate is less than 0.2%, and the identification uniqueness and robustness to disturbance of the solder mask exposure alignment system under the high-density multi-layer PCB structure are significantly improved.

[0051] S1.3: According to the synchronization marker parameter set, an adaptive micro passive synchronization marker structure (such as a micro reflective structure or a photochromic point with a specific size) is selected, a physical synchronization marker layout design file is generated by parameterizing the PCB process database, and the process chain link design of the marker physical structure is realized.

[0052] S1.4: Fuse the physical synchronization mark layout design file with the original PCB stack layout, apply integrated layout verification algorithm, check design rule conflicts, spatial overlap and signal interference problems in the synchronization mark embedding process, output the optimal multi-layer physical synchronization mark integrated layout.

[0053] S1.5: Based on the final multi-layer physical synchronization mark integrated layout, generate a process execution work order corresponding to the solder mask exposure process, issue instructions through the manufacturing process to guide the precise embedding of physical synchronization marks in actual circuit board manufacturing and the realization of optical phase coding, providing a unified physical base point for subsequent high-speed imaging recognition and adaptive alignment algorithm.

[0054] The step S2: before the solder mask exposure, the synchronization marks on the surface of the multi-layer PCB and the mask are scanned globally by using the high-speed imaging device, and the original optical response signal containing environmental differences and vibration interference is collected. Specifically, it includes:

[0055] S2.1: Process positioning is performed on the multi-layer PCB sample to be exposed and the surface of the mask matched therewith, the initial coordinate distribution of the micro-passive synchronization mark region in three-dimensional space is determined, and a synchronization mark space distribution reference is generated.

[0056] Taking the multi-layer PCB sample to be exposed and the matched mask as the process object, input the multi-layer PCB space structure data fused with the micro-passive synchronization mark and the mask physical layout information.

[0057] A precise process positioning platform (parameters: six-degree-of-freedom fine adjustment displacement table, laser interferometer precision δ≤1μm) is used to realize the preliminary physical positioning of the multi-layer PCB and the mask sample in the global reference system.

[0058] Further, through a two-dimensional and three-dimensional machine vision positioning system (parameters: high-resolution industrial camera, structured light / laser triangulation), the space of the PCB and the mask surface is calibrated, and the global physical coordinate data of each mark in the synchronization mark parameter set is extracted, wherein the spatial coordinates (x, y, z) of each synchronization mark region are calculated by the positioning camera and the ranging unit.

[0059] Further, a multi-layer physical mark region space mapping algorithm (parameters: inter-layer positioning transformation matrix (T l ), feature corner matching tolerance ε=2μm) is applied to project the synchronization marks on the mask and the PCB layers to a unified three-dimensional coordinate reference system, realizing the spatial alignment of the multi-layer synchronization marks.

[0060] Further, the initial three-dimensional coordinate distribution of the synchronization mark region is written into the synchronization mark space distribution reference data structure, forming a standard space positioning template for subsequent high-speed imaging scanning and optical data decoding.

[0061] Through the above positioning, calibration and fusion operations, the spatial positions of the micro-passive synchronous mark regions on different layers and different templates are uniquely mapped to the process reference system, ensuring the physical consistency and integral alignment of the data sources of subsequent imaging acquisition and phase analysis, and realizing the generation of the alignment reference spatial data of the complex multi-layer system before the solder mask exposure.

[0062] For example, for a 12-layer HDI PCB and its customized mask, by equipping an industrial camera with a resolution of 0.5 μm, a three-dimensional positioning system with adaptive structured light projection, placing the PCB and the mask on a six-axis precision process positioning table respectively, and setting the positioning platform travel range to 250 mm x 250 mm x 50 mm, a total of 96 groups of actual synchronous marks are recorded, and their layer numbers and spatial center coordinates (x, y, z) are recorded. Under the condition of 200 frames of acquisition frames, the three-dimensional point cloud registration of all marks is completed through the spatial mapping algorithm, and the maximum positioning error between layers is 1.2 μm. The x, y, z spatial coordinates and layer numbers of all synchronous marks are stored as a synchronous mark spatial distribution reference index table. For the 10 boards tested in the same batch, the repeated positioning error is less than 2 μm, verifying that this step can provide a high-precision and stable consistency spatial reference for optical scanning and phase feature recognition in a multi-layer high-precision process environment, and realize the spatial robustness improvement of the alignment process in a complex multi-layer PCB solder mask process environment.

[0063] S2.2: Based on the synchronous mark spatial distribution reference, the high-speed imaging device is used to synchronously acquire the original optical image signals of the micro-passive synchronous mark regions on the surface of the multi-layer PCB and the mask, to obtain high-resolution original optical image data containing physical positions, reflection / refraction differences and relative relationships of multi-layer marks.

[0064] S2.3: The multi-channel parallel reading mechanism is adopted for the acquired high-resolution original optical image data, and the automatic mark recognition algorithm and ROI (Region of Interest) extraction processing are performed to accurately lock the pixel area and geometric contour of each synchronous mark, and the synchronous mark region pixel set is output.

[0065] S2.4: The phase encoding and decoding algorithm is used for the synchronous mark region pixel set to quantitatively extract the original phase response value of the synchronous mark under the disturbance conditions of current illumination, temperature, humidity and random vibration, and to generate an original optical phase response matrix.

[0066] S2.5: The original optical phase response matrix is labeled and associated with the synchronous mark spatial distribution reference and environmental parameters (including real-time data of temperature, humidity, acceleration and other sensors), and a batch working condition optical response data set is synthesized to support the subsequent denoising, normalization and disturbance feature analysis process.

[0067] The step S3: performing denoising and normalization preprocessing on the collected original optical response signal to eliminate invalid signal components generated by various environmental disturbances such as temperature and humidity, dust or slight deformation, etc., to obtain standardized phase response data of the multi-layer synchronous marker. As shown in Figure 2 Specifically, it includes:

[0068] S3.1: Baseline correction processing is performed on the original optical response signal output by the high-speed imaging device to eliminate global amplitude offset caused by imaging system gain drift or light source fluctuation, to obtain a baseline correction signal and ensure signal consistency for subsequent processing.

[0069] S3.2: Based on the spatial filtering algorithm, the baseline correction signal is executed to suppress background noise to filter random noise components caused by environmental light interference, optical element stray light, etc., to output a noise suppression signal and improve the signal-to-noise ratio of the synchronous marker response signal.

[0070] The input is the baseline correction processed optical response signal, and the execution object is the baseline correction signal set obtained by the high-speed imaging system for each synchronous marker region on the multi-layer PCB and mask surface.

[0071] The spatial filtering algorithm (parameters: two-dimensional Gaussian filter kernel σ = 1.5, mean filter window size n = 7 × 7) is used to perform frame convolution processing on the baseline correction signal, to realize preliminary suppression of high-frequency random noise components in the imaging signal, and effectively reduce the local noise level caused by environmental stray light and imaging device non-uniform response.

[0072] Further, through the spatial correlation enhancement algorithm (parameters: correlation threshold τ = 0.85), the intensity variation trend of adjacent pixel blocks in the synchronous marker ROI region is analyzed, and spatial adaptive smoothing restriction is applied to low correlation and sudden change pixel points, to further weaken the influence of non-structural noise sources such as dust and foreign spots on the surface of the optical lens, and to retain the true response form of the synchronous marker main body.

[0073] Further, the multi-scale background modeling algorithm (parameters: pyramid layer number L = 3, background template update weight λ = 0.1) is applied to filter out the background components in the imaging slow-changing area from bottom to top and layer by layer, to distinguish the synchronous marker response from the slow-changing environmental light interference, and to realize effective subtraction of low-frequency background noise.

[0074] Further, the local signal-to-noise ratio (SNR) discrimination algorithm is used for the filtered output signal, and the effectiveness of the synchronous marker signal in the ROI is dynamically verified by the following formula:

[0075]

[0076] Wherein, is the signal mean value after filtering processing in the i-th synchronous marker region, Variance of the neighborhood background noise.

[0077] For the area where SNR is lower than the preset threshold (SNR < 12 dB), a signal enhancement compensation technique (such as weight lifting or local resampling) is used to improve the stability of the response strength.

[0078] Through the above multi-level spatial domain filtering and background suppression algorithm processing, the baseline correction signal is converted into a noise suppression signal, and the synchronous marker optical response data with high signal-to-noise ratio and significantly reduced background stray light components are output, laying a high-quality data foundation for subsequent time-domain denoising and spatial weighting, normalization and other chain preprocessing steps, and realizing the effective extraction of multi-layer alignment synchronous marker information.

[0079] For example, for a 12-layer HDIPCB solder mask exposure sample, a high-speed industrial camera with a resolution of 2048x2048 pixels is selected to collect the synchronous marker area optical signal frame. Under the conditions of ±3℃ temperature fluctuation of the exposure room and 1700±150 lux change of ambient illumination (stray light), a Gaussian filter with σ = 1.5 and a mean window with n = 7x7 are used for spatial domain filtering of each synchronous marker image, and the standard deviation of the background noise is reduced from the original 18 gray levels to 5 gray levels. After introducing multi-scale background modeling, the slowly varying low-frequency components caused by exposure lamp reflection and local device shadow are further removed. The SNR discrimination results show that the average signal intensity of each synchronous marker after screening is increased by more than 30%, and the SNR range in the ROI is increased from the original 9-14 dB to 15-22 dB. For the area where SNR is lower than 12 dB, after weight lifting, the output signal stability of each synchronous marker is significantly enhanced. This method realizes a synchronous marker detection error rate of less than 0.6% in batch sample testing, effectively supporting the high accuracy and environmental robustness of subsequent synchronous marker fine recognition, phase decoding and distribution alignment error modeling processes. The output noise suppression signal is directly used as the standard input for subsequent S3.3 adaptive time-domain denoising processing, ensuring that the optical feature input source of the PCB multi-layer alignment reference has excellent stability and high signal-to-noise ratio.

[0080] S3.3: Use adaptive time-domain denoising technology to detect and remove abnormal time points corresponding to environmental temperature and humidity mutations, dust projections, and local micro-deformation disturbances, form a noise removal signal, and enhance the robustness to occasional physical disturbances.

[0081] The input is the noise suppression signal processed by spatial domain filtering and background modeling, and the execution object is each synchronous marker area on the multi-layer PCB and mask surface, and the high signal-to-noise ratio optical response sequence data obtained by the high-resolution imaging system.

[0082] Adopting self-adaptive time domain denoising algorithm (parameters: sliding window length w = 16 frames, dynamic threshold α = 3.0, abnormal point determination probability p = 0.01), the response signal sequence in each synchronization mark ROI region is analyzed frame by frame in time domain sliding, and automatic detection of time domain mutation disturbance is realized.

[0083] Further, through a multi-stage change point detection method (parameters: first-order difference detection window Δw = 4, local extreme value detection ratio β = 0.25), the short-time mean and variance of the sequence signal are dynamically monitored, and the response mutation, abnormal pulse (including sudden jump caused by sudden change of temperature and humidity, local dust shadow and stress deformation) points in the sequence are distinguished and marked.

[0084] Further, a robust abnormal point confidence calculation model is adopted to apply a discrimination weight to the time sequence points marked as abnormal, and the specific implementation is as follows:

[0085]

[0086] Where, S i is the signal intensity of the i-th frame synchronization mark, μ local , σ local are the mean and standard deviation in the local sliding window, P outlier (i) is the confidence probability of the abnormal point of frame i.

[0087] For the signal points with confidence probability threshold P outlier (i) > p, through data interpolation, sequence smoothing or direct rejection strategy, the removal or correction of abnormal disturbance points is completed, and the continuity of the signal sequence is ensured not to be damaged by accidental disturbance.

[0088] Further, in combination with multi-dimensional batch environment monitoring data (such as real-time change rate of temperature and humidity, acceleration fluctuation amplitude), the occurrence time of abnormal points and external disturbance events are associated, and a dynamic rejection window expansion algorithm (parameters: expansion factor γ = 1.5) is adopted to increase the abnormal point denoising intensity in the disturbance high-incidence section, effectively aiming at the large-area signal deviation caused by accidental environmental impact or equipment resonance in batch production.

[0089] Through continuous self-adaptive time domain denoising processing, the noise suppression signal is converted into a “noise rejection signal” output, which provides multiple disturbance cleaned, stable and reliable synchronization mark response data for subsequent spatial weighting and normalization steps, and realizes the environmental robustness enhancement of multi-layer PCB solder mask alignment signal.

[0090] The exemplary multi-layer synchronous mark optical response sequence of a 12-layer HDI PCB has a sampling frame number of 320 frames. During the temperature mutation period (the batch temperature rises from 24°C to 28°C, and the duration is 15 frames) and the device vibration period (the peak acceleration is 0.045g), the proportion of time-domain abnormal points rises to 4%. Under the action of the adaptive time-domain denoising algorithm, the window length w = 16, the dynamic threshold α = 3.0, the abnormal points screened out by the confidence criterion account for 3.8% of the total number of data points, the continuity index (SL) of the interpolated and corrected sequence is improved to 0.97, and the misjudgment rate of the subsequent spatial weighted processing is reduced to 0.3% after the batch output of the noise-removed signal. The verification results show that under the background of temperature and humidity mutation and device disturbance, the robustness of the effective synchronous mark response signal can be improved by more than 30% by using the present step, and the interference of accidental physical disturbance on the alignment signal data stream is significantly suppressed, thereby providing data protection for the stable extraction of high-reliability multi-layer optical alignment features.

[0091] S3.4: Perform multi-layer synchronous mark area differentiation weighted processing on the noise-removed signal, assign different weights based on the preset spatial distribution and physical encoding characteristics of each synchronous mark to highlight the effective synchronous mark signal, obtain a spatially weighted signal, and realize spatial decoupling of the synchronous mark signal.

[0092] The input is the noise-removed signal processed by the adaptive time-domain denoising algorithm, which contains high signal-to-noise ratio optical response sequence data of each synchronous mark area on the multi-layer PCB and mask surface.

[0093] The multi-layer area differentiation weighted processing algorithm (parameters: spatial distribution index table, encoding weight parameter set (wk)) is used to identify the level, physical location and encoding type of each synchronous mark pixel area data, and realize the differentiation of the multi-layer synchronous mark signal.

[0094] Further, by using the synchronous mark spatial distribution weight allocation method (parameters: physical coordinate index ((xk, yk, zk)), level label l, and encoding method e), the differentiation weight (wk) is automatically allocated according to the spatial geometric distribution and physical signal sensing intensity of each mark in the PCB stack structure, so as to improve the saliency of the effective synchronous response signal at the physical base point.

[0095] Further, the multi-dimensional physical encoding fusion weighted algorithm (parameters: encoding differentiation degree, and stack space mutual interference threshold) is used to allocate specific compensation or suppression weights for the case of signal mutual interference in high-density interlayer or adjacent distribution area, suppress the spatial signal aliasing caused by local coupling or noise leakage, and improve the independence of the synchronous mark differentiated signal.

[0096] Further, the spatially weighted signal matrix (S weighted ) is generated by the following weighted calculation formula:

[0097] S weighted (k)=w k ·S clean (k)

[0098] wherein S weighted (k) represents the weighted kth synchronization marker response signal, w k is the spatial distribution related weight to physical encoding, S clean (k) is the corresponding noise removed synchronization marker signal.

[0099] Further, a spatial decoupling discrimination algorithm is applied to perform spatial autocorrelation analysis and linear independence evaluation on the weighted signal matrix, through the following autocorrelation coefficient matrix criterion:

[0100]

[0101] wherein R ij is the correlation coefficient between different synchronization marker weighted signals, Cov(·) is the covariance, σ i is the standard deviation of S

[0102] S weighted (i). For marker pairs with R ij close to zero, it is determined that spatial effective decoupling is achieved.

[0103] Through the above multi-layer distinguishing weighting and spatial decoupling processing, the signal intensity is organically integrated with the spatial physical properties and encoding characteristics of the synchronization marker, realizing spatial differentiation and decoupling processing of multi-layer signals, outputting spatial weighted signals, creating conditions for subsequent normalization processing and multi-dimensional alignment reference construction, and realizing data independence and physical criterion consistency of multi-layer solder mask alignment signals.

[0104] For example, in the 12-layer HDIPCB solder mask exposure process scenario, the input is the multi-layer synchronization marker response signal processed by the time domain denoising algorithm, the spatial distribution weight is set to the highest layer w top = 1.2, the middle layer is decreased according to the layer density to w mid = 1.0, the bottom layer w bottom = 0.8, for the edge area or the marker with encoding interference, the compensation factor δ e = 0.85 is set, and the interference reduction weight η = 0.75. After weighted processing by the above algorithm, the signal spatial decoupling autocorrelation coefficient R ij is reduced to 0.13 on average, and the boundary layer separation degree is increased by 20%. The batch output spatial weighted signal has improved effective component differentiation, which is used as the input source for subsequent normalization transformation, and the standard deviation of the alignment residual error distribution is reduced by 18% in the temperature and humidity fluctuation range through randomized batch testing, significantly improving the multi-layer synchronization marker decoupling ability and spatial discrimination accuracy.

[0105] S3.5: Apply normalized transform method to the spatially weighted signal to eliminate the multi-layer synchronous mark response deviation caused by systematic errors such as uneven exposure energy, imaging gain, etc., output the normalized phase response data as the only input for subsequent feature extraction and alignment reference construction.

[0106] The step S4: input the normalized phase response data into the adaptive optical processing algorithm, and extract the multi-dimensional alignment reference feature set reflecting the physical displacement, local deformation and other working condition information by feature extraction of the optical phase difference of each micro-passive synchronous mark. As shown in the figure, it specifically includes: Figure 3

[0107] S4.1: Input the normalized phase response data of the multi-layer synchronous mark into the adaptive optical processing algorithm module as the unified input basis for subsequent optical feature analysis, to ensure the consistency of the data structure of the synchronous mark response data of all levels.

[0108] The input is the multi-layer synchronous mark normalized phase response data output by S3.5, and the data structure is a multi-point area signal set covering all PCB layers and mask surfaces, layered organization, and integrated amplitude and phase calibration.

[0109] The structured data loading method (parameters: multi-layer index table, coded label set) is adopted to realize the batch import of the normalized phase response data of all synchronous marks into the adaptive optical processing algorithm module, complete the collection and mapping of the basic data structure, and form a unified input.

[0110] Further, through the multi-dimensional data consistency checking algorithm (parameters: data integrity threshold θ=0.99, missing value filling strategy: interpolation mode-linear, recent mean), the missing values, repeated values and abnormal points of the imported data are automatically checked out, the structurally abnormal data is excluded or the local missing area is filled, and the structure consistency of all synchronous mark response data in space, level and coding domain is ensured.

[0111] Further, the input feature normalization mapping method (parameters: normalization standard μ=0, σ=1, optional normalization interval [0, 1]) is adopted to perform overall layered normalization processing on the normalized phase response data of all marks, to ensure that the amplitude and phase distribution of the input data are limited to a unified scale domain, and to eliminate the scale difference of signal input between different levels and different physical characteristics.

[0112] ​Furthermore, through the interlayer spatial index reconstruction algorithm (parameters: physical coordinate mapping matrix Λ, layer marker l), a unified mapping of each synchronization marker in its physical embedding layer, spatial coordinates and mask space is achieved, and the original input signal set is reassembled into a multi-dimensional array structure (Index: layer number x marker type x spatial position), providing an efficient data scheduling interface for subsequent optical feature analysis algorithms.

[0113] Through the above chain-like data collection and structure mapping processing, the standardized phase response data of multi-layer synchronous markers was input into the adaptive optics processing algorithm in a complete, standardized and highly consistent manner, which established a unified data foundation for subsequent technical steps such as multi-dimensional feature extraction, phase difference analysis and alignment reference construction.

[0114] For example, in a 12-layer HDIPCB solder mask process batch, the input is a normalized phase response data matrix with 12 synchronous markers per layer, totaling 144 markers, with a sampling resolution of 2048×2048 pixels. Using structured data loading, the system automatically completes data loading within 0.4ms. Through consistency checks, two markers were found to be suspicious, which were recovered by interpolation using the current batch mean. The normalization process resulted in the amplitude range of each layer being unified to [0,1], with an inter-batch phase mean μ = 0 and a standard deviation σ = 0.41. Spatial index reconstruction outputs a multi-dimensional array structure, with the index structure being [layer = 12, marker = 12, region = 12, area ...

[0115] , the full data mapping efficiency is better than 10μs / time. The final output is standardized and structured input data, which provides a standardized and efficient underlying data foundation for feature extraction, phase difference calculation and distributed alignment error modeling in S4.2 and beyond, effectively ensuring the process consistency and environmental robustness of multilayer high-precision solder resist exposure process.

[0116] S4.2: The phase difference is calculated using a phase difference analysis algorithm on the standardized phase response data of each layer of synchronization markers to obtain the optical phase difference characteristic parameters of the micro passive synchronization markers in the PCB board-mask system at different physical locations.

[0117] The input is the standardized phase response data of all synchronous marks of the multilayer PCB and paired masks, which are processed by S3.5 normalization. The data structure has hierarchical differentiation, physical space labeling and integrated amplitude-phase calibration characteristics, serving as the unified analysis basis for adaptive optics processing algorithms.

[0118] A phase difference analysis algorithm with multi-layer region grouping (parameters: layer number grouping table L, synchronous marker physical coordinate set M) is adopted to realize the pairwise traversal of all synchronous marker points of each layer and its corresponding mask surface, and construct an analysis index under the dual constraints of physical space coordinates and marker encoding.

[0119] Further, by the phase difference calculation core unit (parameters: normalized phase response vector ), for each pair of physically corresponding synchronization mark area, the phase difference of the synchronization mark under the PCB and the mask system is quantitatively calculated according to the following formula:

[0120]

[0121] wherein, represents the normalized optical phase difference of the kth synchronization mark of the lth layer between the PCB and the mask, is the phase response of the kth synchronization mark of the PCB, is the phase response of the corresponding synchronization mark in the same area of the mask surface.

[0122] Further, by the phase difference statistical module (parameters: environmental disturbance label E, area resolution capability threshold τ r ), the phase difference data of all physically corresponding mark points are batched and spatially clustered, and the overall offset, local anomaly and multi-layer collaborative change mode are analyzed to form a phase difference feature matrix with significant spatial distribution:

[0123]

[0124] wherein, N l is the total number of actually identifiable synchronization mark points of the lth layer.

[0125] Further, a spatial consistency constraint detection algorithm (parameters: autocorrelation threshold γ, physical neighborhood size d n ) is used to perform spatial consistency verification on the phase difference distribution of the synchronization marks in the same layer or adjacent layers, and to mark the detected spatial mutation points.

[0126] An asynchronous anomaly point filter is used to combine real-time temperature and humidity, stress disturbance and other environmental labels to physically attribute the detected local phase difference mutation points, and to label them as rigid overall displacement or local non-uniform deformation characteristics.

[0127] Through the above multi-level phase difference analysis and spatial statistical-attribute processing, the standardized phase response data is efficiently converted into a set of physical offset and micro-deformation phase difference characteristic parameters of multiple layers and multiple areas, realizing the structured and quantifiable representation of the physical state information in the multi-layer PCB alignment system.

[0128] For example, for a 12-layer HDI PCB solder mask process sample, the input is 144 synchronization mark standardized phase response data vectors covering all batches of layers, and the single-point phase noise is less than 0.05 radians. Pairwise analysis is used to calculate the The in-layer phase difference distribution range [-0.02, 0.13] radians is obtained, and the proportion of abnormal points is less than 0.8%. The spatial clustering algorithm automatically identifies 3 local abnormal concentration areas, and the environmental label analysis shows that the temperature of the batch increases by 3.2°C during exposure, and the phase difference in the corresponding area increases significantly, which is determined to be a local thermal expansion and cold contraction non-rigid deformation. After processing, a multi-dimensional layered phase difference matrix is output as the standard input for physical offset and deformation perception, effectively supporting subsequent physical offset modeling and fiducial feature extraction. The above processing has the characteristics of high efficiency, low misjudgment and accurate physical attribution in the batch automated optical detection process, improving the process adaptability and stability of the solder mask alignment process.

[0129] S4.3: Based on the obtained optical phase difference feature parameters, a multi-dimensional feature extraction algorithm is used to generate feature vectors, realizing the preliminary differentiation of physical offset (overall rigid body movement) and local deformation (such as stress-induced micro-region bending, thermal expansion and cold contraction-induced non-rigid distortion).

[0130] The input condition is the normalized multi-layer PCB and the standardized phase response data of all synchronous markers on the mask surface, and the data structure has hierarchical differentiation and spatial physical label attributes.

[0131] A multi-dimensional feature extraction algorithm (parameters: phase difference feature matrix, layered space mapping parameter table) is used to realize the feature vector generation function of all synchronous markers in the hierarchical, spatial and coding dimensions.

[0132] Further, by using a principal component extraction method (parameter: cumulative contribution rate threshold 98%), feature components representing the main differences are extracted from the full phase difference feature parameter space, an initial principal component feature vector group is generated, and its mapping weight in the overall phase difference feature space is obtained, which facilitates subsequent feature sensitivity analysis for different physical states.

[0133] Further, a rigid transformation parameter estimation method (parameter: multi-layer rigid affine model) is used to separate the overall rigid body movement feature components based on the initial feature vector group, calculate the average translation and rotation angle of the multi-layer regional synchronous markers in the physical space, quantify the main components of the overall rigid displacement, and form a feature subset representing the physical offset (overall movement).

[0134] Further, a local residual analysis algorithm (parameters: local neighborhood range, non-rigid residual threshold = 0.02) is used to perform spatial residual decomposition on the remaining features after removing the rigid components, extract non-linear deformation features such as local stress, micro-region bending and thermal expansion and cold contraction, generate a local deformation feature subset, and realize accurate separation and preliminary attribution of various non-rigid disturbances.

[0135] Further, a feature normalization fusion algorithm is adopted (parameters: amplitude normalization domain [-1, 1], normalization weight set, normalize the rigid feature subset and the non-rigid feature subset respectively, splice and combine into a structured multi-dimensional feature vector, and realize efficient unified coding of all physical offsets (overall rigid movement) and local deformation (micro area bending, non-rigid twisting, stress response, etc.) information.

[0136] Through the above multi-level feature extraction, decomposition and fusion chain, the phase difference feature matrix is systematically converted into a multi-dimensional feature vector reflecting the physical offset and local deformation state of the multi-layer PCB, realizing the structured, quantifiable and hierarchical representation of the physical working condition information, and providing a technical basis for the input of the position reference construction and distributed position error model.

[0137] For example, for a 12-layer HDI PCB solder mask exposure process batch, the input is a normalized phase difference feature matrix covering all 144 synchronous markers, the cumulative contribution rate threshold is set to 98% for principal component extraction, and the first five principal components are extracted, with principal component weights of 0.42, 0.23, 0.17, 0.09 and 0.07. The rigid affine transformation parameter estimation result gives the overall translation [18.2, 14.7] and the rotation angle (0.042) degree. In the local residual analysis, the maximum non-rigid residual in the neighborhood range is 0.019, which meets the parameter threshold, and the local deformation feature is located in three high-temperature sensitive areas (each about 220 μm long). Finally, all feature components are normalized and output to form an 8-length multi-dimensional feature vector as the standard input for subsequent multi-layer position error modeling. In actual operation, the principal components of the multi-dimensional feature vector are highly sensitive to various physical disturbances, the physical attribution is accurate, and the repeatability and consistency between batches reach an index of better than 98%.

[0138] S4.4: Aggregate the phase difference multi-dimensional feature vector and use the heterogeneous marker feature fusion model to weight the working condition related parameters to highlight the position reference features of the physical offset and local deformation reflected by each micro passive synchronous marker at different spatial distribution positions and sensing layers.

[0139] S4.5: Output the multi-dimensional position reference feature set after comprehensive weighted fusion as the standard input for the subsequent distributed position error model establishment step, realize the improvement of the physical difference perception ability under the influence of environmental disturbance and the efficient transfer of position compensation data.

[0140] The step S5: based on the above multi-dimensional position reference feature set, a distributed position error model of the multi-layer PCB under different environmental disturbance conditions is established, the spatial distribution characteristics of the overall rigid displacement and local stress deformation are distinguished, and the physical difference perception is realized.

[0141] Specifically, it includes:

[0142] S5.1: With the multi-dimensional alignment reference feature set as input, call the spatial feature analysis algorithm to reconstruct the spatial distribution of the standardized phase response data of each layer of micro-passive synchronization mark, generate a spatial phase distribution map, and obtain a feature base containing physical offset and deformation information.

[0143] S5.2: Based on the spatial phase distribution map, apply the rigid affine transformation analysis method to separate the rigid offset parameters reflecting the overall rigid displacement of the multi-layer PCB board, extract the overall translation and rotation between each layer, and form a set of overall rigid displacement parameters, which are used to represent the overall displacement of the board caused by the container-level environmental disturbance.

[0144] The input data is the spatial phase distribution map generated by step S5.1, which has completed the spatial position mapping and physical state normalization of the synchronization marks based on the multi-dimensional alignment reference feature set, and has the standardized phase response array structure covering all synchronization mark areas of the multi-layer PCB.

[0145] The rigid affine transformation analysis method (parameters: affine transformation matrix Al, target layer number l) is used to model the physical coordinates and phase responses of all synchronization marks in the spatial phase distribution map of each layer, and the spatial transformation of each layer is classified as rigid movement (including translation and rotation).

[0146] Through the least squares fitting algorithm (parameters: mark coordinate pair {X i ,Y i} and reference system coordinate pair {X i,ref ,Y i,ref}), the rigid affine parameters of the synchronization mark point cloud are estimated, which specifically includes the joint solution of the global translation vector and rotation matrix of the target layer relative to the reference layer (or mask layer). The rigid transformation satisfies the following formula:

[0147]

[0148] Where R l is the rigid rotation matrix of the lth layer, t l is the rigid translation vector, (x i ,y i ) is the spatial coordinate of the ith synchronization mark of the layer, and (x i ′,y i ′) is the coordinate after mapping to the reference system.

[0149] Further, the singular value decomposition (SVD) method is used to solve the optimal rigid matching of the covariance matrix, and the global rotation angle θ l and translation component (ΔX l ,ΔY l ) between layers are quantified to obtain a set of overall rigid displacement parameters:

[0150] θ l ,ΔX l ,ΔY l

[0151] The parameter set describes the overall rotation and translation state of each layer of the multi-layer PCB, respectively.

[0152] Further, by checking the spatial consistency of the rigid affine parameters (parameters: threshold ∈ rigid ), the fitting residual distribution of the overall rigid displacement parameters is verified to screen whether there is a large-scale overall mismatch or a deformation trend of the base plate body, and the effectiveness and representativeness of the rigid displacement parameters are ensured.

[0153] Through the above rigid affine transformation analysis, the phase distribution map reconstructed by S5.1 is converted into rotation-translation parameters that explicitly represent the overall rigid motion of the multi-layer board, a physical quantitative model is established for the synchronous displacement of the multi-layer board caused by the container-level environmental disturbance (such as overall temperature change, equivalent rigid body movement of equipment), and the rigid displacement characteristics are effectively decoupled from subsequent local deformation modeling (S5.3), error decomposition and dynamic compensation.

[0154] For example, in a 12-layer high-end HDI PCB solder mask alignment batch, the spatial phase distribution map input covers the physical coordinates of 12 synchronous markers on each layer and the normalized phase response. Using rigid affine transformation analysis, the rotation angle interval of each layer and the design reference layer or mask reference surface is fitted to be [-0.03, 0.07] degrees, the global translation component interval is ΔX = [-12 μm, +15 μm], ΔY = [-10 μm, +13 μm], and the SVD residual is better than 2 μm. For a layer in the batch whose environmental temperature is 2.5°C higher than the average, overall arching occurs, the rigid rotation parameter of the layer is significantly improved, and the fitting residual of the translation increases. The rigid displacement parameter set obtained by this analysis is used as the key input for subsequent physical error decomposition (S5.3-S5.5) and real-time parameter compensation (S7), which significantly improves the perception, discrimination and compensation ability of the entire PCB board and multi-layer container-level disturbance.

[0155] S5.3: Based on the overall rigid displacement parameter set, a local stress field modeling algorithm is used to perform secondary residual analysis on the spatial phase distribution map, to obtain the small nonlinear displacement between each synchronous marker, to refine the stress distribution pattern of each local position, and to output the local stress deformation parameter set, to realize the quantitative characterization of physical stress disturbance.

[0156] The overall rigid displacement parameter set is used as the input condition to perform local stress field modeling and analysis processing on the spatial phase distribution map of the multi-layer PCB.

[0157] A local residual calculation method (parameters: spatial phase distribution map (S φ), the overall rigid displacement parameter set (T rigid ), the local neighborhood radius (r loc ), the residual analysis of the phase space after the overall rigid component is removed, and the non-rigid response component between the synchronization markers is identified.

[0158] Further, through the spatial local residual mapping algorithm (parameters: spatial index ((xi, yi)), local window size (d win ), the spatial neighborhood (N i ) around each synchronization marker is defined, and its quadratic residual value ε nl,i is calculated.

[0159] The small nonlinear deviation between adjacent points is described:

[0160]

[0161] wherein, is the normalized phase response value of the synchronization marker i, indicates the theoretical phase response in the neighborhood according to the overall rigid transformation model.

[0162] Further, through the local stress pattern clustering algorithm (parameters: residual distribution threshold, feature vector extraction window (wf)), the quadratic residual values of all synchronization markers are clustered and pattern-identified, and the stress deformation patterns with significant spatial distribution characteristics are iteratively extracted, including in-plane bending resistance, local warping, thermal expansion and contraction, etc., and each group of patterns is encoded as an independent physical deformation feature vector.

[0163] Further, through the local stress quantification operator (parameters: weight distribution set (λ j ), spatial propagation model (G(x, y))), the local deformation feature vectors are weighted and normalized to generate a spatially balanced local stress deformation parameter set, and the quantitative evaluation of the local stress disturbance intensity is realized.

[0164] Through the above chain algorithm processing, the multi-layer PCB spatial phase distribution map is further modeled and physically attributed to multiple types of nonlinear disturbances such as local stress, thermal bending, non-uniform micro-zone deformation, etc. on the basis of comprehensive removal of the overall rigid offset factor, providing key technical support for establishing high-sensitivity and high-resolution working condition perception channels.

[0165] For example, for a 12-layer HDI PCB solder mask exposure process batch, the overall rigid displacement parameter set is known to be a translation (t = [14.3, 12.1], μm) and a rotation angle (△ = 0.035°), the spatial phase distribution map acquisition resolution is 2048 x 2048 pixels, and the local neighborhood radius (r loc= 120, pm), all the synchronous mark secondary residuals were analyzed. Local residual statistics showed that the residuals of 87 synchronous marks exceeded 0.018 radians, mainly concentrated in the heat-sensitive area and the area with high processing stress. The local stress clustering algorithm automatically clustered four main stress modes: in-plane tension, thermal warping, copper foil directional stress, and point-like stress concentration. Using spatial distribution weighting, the local stress deformation parameter set was normalized, and the residual intensity of the maximum stress component corresponding to the distribution position was 0.023 radians, and the bending mode weight was 0.41. The above parameter set was verified by subsequent multi-layer alignment error modeling, which could accurately locate and quantify various non-rigid deformation regions caused by physical disturbances, support high-precision spatial distribution correction under the flexible alignment compensation process, and effectively improve the environmental robustness and size consistency of the multi-layer PCB solder mask process.

[0166] S5.4: Input the global rigid displacement parameter set and the local stress deformation parameter set into the distributed alignment error modeling framework, construct the distributed alignment error model of the multi-layer PCB under specific environmental disturbance conditions through layered error decomposition technology, and output the distributed alignment error parameter matrix containing spatial multi-point error factors and their weight distribution.

[0167] S5.5: Use the distributed alignment error parameter matrix to perform abnormal data recognition and feature attribution algorithm, locate and classify the abnormal error regions caused by environmental disturbances, and output the physical difference perception report to provide accurate spatial error information input for subsequent virtual alignment plane construction and alignment parameter automatic compensation.

[0168] The step S6: For the distributed alignment error model, the optical phase difference superposition algorithm is used to generate the virtual alignment plane of the current inter-layer alignment series, realizing the difference compensation of alignment data under disturbance environment. Specifically, it includes:

[0169] S6.1: Feature screening is performed on the multi-layer synchronous mark physical offset matrix in the distributed alignment error model, and the principal component analysis algorithm is used to extract the key spatial alignment features representing the global rigid displacement and local stress deformation to form a physical offset feature set, so as to clearly define the input basis for subsequent optical phase processing.

[0170] S6.2: Based on the above physical offset feature set, the optical phase response parameters are used as input, and the phase difference calculation module is used to extract the phase difference distribution map of each synchronous mark to obtain the multi-layer phase difference feature matrix after superimposing environmental disturbance factors, realizing the mapping between physical offset features and optical phase response.

[0171] For S6.2 sub-step, the input condition is the physical offset feature set obtained by screening based on S6.1 step, which contains the overall rigid displacement parameters and local stress deformation parameters selected by principal component analysis algorithm, and the set represents the actual physical state of the multi-layer PCB under specific environmental disturbance in the form of standardized physical feature vector.

[0172] The optical phase difference extraction method (parameters: physical offset feature set, standardized synchronous marker phase response data, layer number set (L), synchronous marker index set (Ml)) is adopted to realize fine layered analysis of phase information between multi-layer synchronous markers.

[0173] Through the phase difference calculation module (algorithm: layer-by-layer relative phase method), the instantaneous phase difference between each layer (l) of all synchronous markers and the synchronous marker at the same position of the reference layer (such as the design layer or the mask layer) is calculated:

[0174]

[0175] Wherein, is the phase difference of the (l)th layer of the (i)th synchronous marker, is the phase response of the synchronous marker of the same index on the reference layer.

[0176] Further, through the multi-layer phase difference superposition analysis algorithm (parameters: physical offset feature set, spatial coordinate matrix ({X i,l ,Y i,l )), the phase difference distribution of each layer of synchronous marker points is classified, the global trend and local anomaly are screened, and the explicit mapping of physical offset features (such as rigid translation, rotation, stress local distortion) and optical phase behavior is realized according to the physical parameter attributes.

[0177] Further, through the phase difference clustering and modeling method (parameters: spatial distribution threshold, maximum targeted grouping number), the phase difference results of all synchronous markers are adaptively grouped in space, the phase response subsets corresponding to different physical disturbances are established, the overall moving area and stress abnormal area are distinguished, and the regional disturbance resolution ability is improved.

[0178] Further, through the abnormal correction and defect elimination algorithm (parameters: abnormal identification threshold, weight distribution coefficient), the synchronous marker area data with mismatch and abnormal deviation after clustering is dynamically corrected and weighted eliminated, the anti-interference ability of the multi-layer phase difference feature matrix is enhanced, and the physical consistency and process adaptability of the simulation results are improved.

[0179] Through the above chain processing mode, the multi-layer phase difference feature matrix corrected by superimposed environmental disturbance factors is obtained, and the mapping and dynamic adaptation between the physical offset features and the multi-dimensional data based on the optical phase difference are realized.

[0180] For example, for a batch of 12-layer high-density interconnection printed circuit board solder mask film exposure alignment, the physical offset feature set extracted after principal component analysis has a size of 36 dimensions, including overall translation, rotation parameters and multiple sets of local stress components. The input normalized phase response data matrix size is [12 layers x 12 synchronization marks]. Using the layer-by-layer relative phase method, the phase difference of each synchronization mark in 12 layers is obtained in turn, and the global phase mean and standard deviation of each layer are calculated. Using the adaptive clustering algorithm, the phase difference distribution is classified according to (δ nlu = 0.09) radians, and it is found that 8 layers have significant alignment offset and 4 layers have local abnormal points. Based on the weight coefficient λ i The abnormal interval values are weighted and removed, and the global noise distribution mean of the final output multi-layer phase difference feature matrix is reduced to 0.011 radians, and the abnormal area residual rate is reduced from 3.6% to 0.7%. The phase difference feature matrix output in this step is used as the standardized input of the downstream virtual alignment plane reconstruction, which fully guarantees the accurate response and flexible compensation ability of the multi-layer PCB solder mask process to complex disturbance factors.

[0181] S6.3: Perform mirror reconstruction algorithm on the multi-layer phase difference feature matrix, map the optical phase difference of each layer to a unified virtual alignment plane according to the spatial transformation relationship in the distribution alignment error model, realize the spatial normalization representation of multi-layer data, and lay a mathematical foundation for the generation of virtual alignment plane.

[0182] S6.4: Use the virtual alignment plane generation module to perform quadratic fitting on the spatially normalized multi-layer phase difference features, dynamically build a virtual alignment reference plane that can adapt to the current disturbance environment, and thus generate a compensation base surface that explicitly describes the interlayer relationship and alignment reference.

[0183] S6.5: Perform residual analysis on the generated virtual alignment reference plane and the initial design alignment plane, and output the corrected alignment data compensation amount for each layer alignment error distribution using the difference compensation algorithm, to realize the difference compensation of the virtual alignment plane to the actual alignment data.

[0184] The step S7: judge whether the current distribution alignment error model is in the preset threshold interval, if it exceeds the threshold, automatically correct the alignment parameters and mask pattern data of the exposure equipment according to the physical offset of each layer, realize flexible parameter dynamic adaptation. Specifically includes:

[0185] S7.1: Perform threshold discrimination processing on the input distribution alignment error model, use the preset threshold parameter set to perform difference analysis on the rigid displacement and local stress deformation in the multi-dimensional alignment reference feature set, to identify whether it exceeds the allowed process deviation interval and output the discrimination result.

[0186] S7.2: Based on the discrimination result of the distribution-based position error model, extract the physical offset parameters of each layer from the multi-dimensional positioning reference feature set, and use the error decomposition algorithm to accurately separate the physical offset parameters into the overall displacement component and the local deformation component to obtain a multi-layer physical offset vector group.

[0187] S7.3: For the obtained multi-layer physical offset vector group, call the flexible parameter mapping algorithm to map each physical offset vector to the exposure equipment positioning servo shaft adjustment parameter, the mask pattern space transformation parameter and the exposure reference surface calibration parameter, and realize multi-dimensional quantization of the parameters.

[0188] Take the physical offset vector group of each layer of the multi-layer PCB as the input basis, which is obtained by error decomposition algorithm and contains the overall displacement component and the local deformation component of each layer.

[0189] Use the flexible parameter mapping algorithm (parameter setting according to process tolerance band, equipment response characteristics, mask pattern mismatch model) to input each layer of physical offset vector into the multi-parameter mapping module to realize accurate correspondence between physical space error and equipment control parameter.

[0190] Further, for the overall rigid displacement vector component, the adjustment parameters of the exposure equipment positioning servo shaft are derived through linear space mapping method, including X / Y axis translation and rotation angle correction. In this link, affine transformation matrix is used to convert physical offset data into mechanical servo index, and the formula is:

[0191]

[0192] Where ΔX and ΔY are the overall X-axis and Y-axis physical offset, and θ is the overall rotation offset.

[0193] Further, for the local deformation component, the space deformation parameterization output of the mask pattern is realized through the nonlinear space transformation and the difference vector divergence algorithm, including regional scaling factor and local distortion compensation. In this processing link, a polynomial space fitting model is used to establish nonlinear registration between each physical offset sampling point and the mask pattern lattice, and the space transformation can be expressed as:

[0194] X' = a0 + a1X + a2Y + a3X 2 + a4XY + a5Y 2

[0195] Y' = b0 + b1X + b2Y + b3X 2 + b4XY + b5Y 2

[0196] Where a i , b iThe parameters are the polynomial parameters obtained by fitting using the least squares method.

[0197] Furthermore, the optical exposure reference plane is comprehensively calibrated for the overall rigid displacement parameters and local deformation spatial transformation parameters of all layers. Through a three-dimensional spatial rotation mapping algorithm, the exposure reference plane attitude adjustment parameters under the current process state are output, including the reference plane normal vector direction and spatial offset.

[0198] The exposure equipment alignment servo axis adjustment parameters, mask pattern spatial transformation parameters, and exposure reference plane calibration parameters generated through the above steps are aggregated into a set of multi-dimensional quantitative process parameters, which are marked as the flexible process compensation control data for this batch.

[0199] By using a flexible parameter mapping algorithm, a dynamic mapping relationship is established between each physical offset vector and the equipment control parameters, enabling synchronous adaptive calibration of the exposure equipment and mask data under environmental disturbances, thereby improving the alignment accuracy and production flexibility of the solder resist process.

[0200] For example, in the solder mask exposure process of a batch of 6-layer high-density multilayer PCBs, the online process inspection system detected that the physical offset of the 3rd layer was (ΔX3,ΔY3,θ3)=(12μm,-18μm,0.022°). After local deformation was sampled by a dot matrix, 50 sets of offset data were generated. The spatial deformation parameter set (a) was obtained by polynomial fitting. i = (0.98, 0.02, ..., -0.001). After the flexible parameter mapping algorithm is executed, the translation of the alignment servo axis of the third layer is adjusted to +12μm (X direction), -18μm (Y direction), and +0.022° (rotation); the local region of the third layer of the mask pattern is deformed according to the fitted function;

[0201] The normal of the exposure reference plane can be finely adjusted from (0,0,1) to (0.001,-0.002,0.999)±25μm, and mapping and compensation can be completed automatically. The overall alignment accuracy is improved to ≤6μm. Under 20 batches of working conditions, the yield fluctuation is less than 2%.

[0202] S7.4: Using the multidimensional quantized exposure equipment, the position servo axis adjustment parameters and mask pattern spatial transformation parameters are used to perform automatic parameter dynamic refresh operations on the exposure control system and mask data management system to generate a flexible exposure process parameter set based on the physical state of the current batch.

[0203] S7.5: For the generated flexible exposure process parameter set, execute the process safety verification algorithm to determine whether the corrected alignment parameters and mask pattern data meet the product design requirements and process tolerances; if not, perform adaptive optimization iteration on the physical offset vector group and flexible parameter mapping process, and output the final corrected parameter results.

[0204] The step S8: in the solder resist film exposure process, the actual exposure operation is completed according to the corrected alignment parameters and mask pattern data, and the current batch working condition label and environmental parameters are recorded for subsequent tracing. Specifically, it includes:

[0205] S8.1: receive and load the alignment parameters and mask pattern data corrected by the adaptive optical processing algorithm as the input of the exposure control module in the exposure process to ensure that the control is based on the current optimal alignment state during physical exposure.

[0206] S8.2: perform exposure parameter issuing operation on the solder resist film exposure control module, adjust the light source projection path and mask positioning device according to the corrected alignment parameters by using the motion control system, realize accurate alignment and pattern correction at the physical level, and ensure the physical consistency of the mask pattern data and the board layer synchronization mark position.

[0207] S8.3: before the actual execution of the exposure process, based on the environmental parameters (such as temperature and humidity, vibration amplitude, etc.) of this batch collected by the working condition monitoring sensor, the data acquisition module is called to structure and package the current state before exposure to obtain the batch working condition label for process tracing.

[0208] Based on the alignment parameters and mask pattern data corrected by the adaptive optical processing, the input data includes the exposure control parameter set that has completed process correction and the identification of the multi-layer PCB batch to be exposed.

[0209] A multi-channel working condition monitoring sensor array (parameters: high-precision temperature and humidity sensor, three-axis acceleration vibration sensor, air dust concentration sensor, etc.) is used to realize real-time high-sampling-rate collection of physical variables in the exposure environment.

[0210] Further, through the integrated sensor data acquisition module (parameter setting according to sensor sampling frequency, resolution and calibration curve), the environmental state data before each batch exposure is periodically and synchronously collected, and a time sequence environmental variable original data stream is formed.

[0211] Further, a data preprocessing algorithm (such as drift correction, outlier rejection and time sequence normalization) is used to clean and standardize the original environmental data stream, to eliminate accidental sensor noise and correct baseline drift, and obtain standardized environmental parameters with process discrimination value.

[0212] Further, through a structured data packaging algorithm (parameters include: batch unique ID, collection time, spatial positioning identifier, environmental multi-parameter vector, equipment running state identifier), the collected and standardized data are labeled and bound to the corresponding batch and equipment running state, and a batch working condition label data package for process life cycle tracing is generated.

[0213] The above labeled data is archived in real time to the manufacturing execution system database by the batch condition label storage and distribution module, realizing batch-level full-process data traceability and abnormal analysis capability based on pre-exposure conditions.

[0214] Through the above chain data flow of collection-preprocessing-packing-archiving, the pre-exposure batch environment and physical state and process control parameters are accurately coupled, realizing the technical basis of intelligent tracing and abnormal correlation analysis of the whole process of solder resist process.

[0215] For example, before the solder resist exposure of a batch of 10-layer high-density PCB, the system calls the temperature and humidity sensor (accuracy 0.1℃ / 1%RH, update rate 10Hz) to collect the temperature 24.6℃ and humidity 47.3%, the vibration sensor (resolution 1mg, three-axis acceleration) collects X=6.5mg, Y=2.8mg, Z=0.9mg, and the air dust concentration sensor measures the particle concentration as 8.2μg / m 3 , the collection timestamp is 20240618T092307, and the spatial positioning number is Area-C6. The data collection module automatically cleans up a set of standardized environment parameter vector [24.6, 47.3, 6.5, 2.8, 0.9, 8.2], and generates the following batch condition label after structured packaging:

[0216] {"BatchID":"A6B0923D","Timestamp":"20240618T092307","Location":"Area-C6","EnvironParam":[24.6, 47.3, 6.5, 2.8, 0.9, 8.2],"DeviceState":"Ready","ParamSetID":"EXP-C123F"} The condition label is archived in real time in the MES database, and is indexed and associated with the exposure parameter set, alignment correction results and other data of the batch. The system can subsequently trace back abnormal batches, locate the environmental background of process deviation, realize multi-batch abnormal traceability, data mining and parameter evolution optimization, and greatly improve the scientificity of batch difference control and manufacturing consistency.

[0217] S8.4: Drive the exposure control module to implement solder resist film exposure operation, real-time monitor the exposure energy, spectral distribution and board layer synchronous marker response curve in the exposure process, use the exposure process monitoring data flow to verify the timing integrity and parameter landing of actual exposure execution, and ensure that the pattern transfer accurately meets the multi-dimensional alignment reference requirements.

[0218] S8.5: After the exposure is completed, the key information formed during the exposure operation of the batch of solder resist films, such as the exposure equipment operation log, the batch working condition label and environmental parameters, the synchronous mark response data and the like, is structured and archived through a data recording module, and a traceable unique identification code is assigned, so as to provide full-chain traceable data support for subsequent alignment residual checking and process consistency big data analysis.

[0219] The step S9: after the exposure is completed, the synchronous mark area is detected again through imaging, the phase response of the actual exposed synchronous mark is obtained, the final alignment residual is calculated by comparing the initial multi-dimensional alignment reference, and dynamic feedback checking is realized. Specifically, it includes:

[0220] S9.1: After the exposure of the solder resist film is completed, the process positioning of the multi-layer PCB is performed, based on the synchronous mark space distribution reference generated before the exposure, the high-resolution re-imaging image data of the actual synchronous mark area is obtained by using a precise re-imaging device, so as to ensure that the imaging object and the data before the exposure maintain spatial consistency.

[0221] S9.2: The re-imaging image data is subjected to automatic mark recognition and ROI extraction algorithm, based on the pixel set of the synchronous mark area generated before, pixel-level recognition and contour matching are performed on the micro passive synchronous mark area of each layer of the multi-layer PCB, so as to output the re-imaging pixel data set of the synchronous mark area.

[0222] S9.3: The phase encoding decoding algorithm is applied to the re-imaging pixel data set, the standardized phase response value of the actual exposed synchronous mark is obtained by combining the original optical phase response matrix obtained before the exposure, and the optical phase response data alignment of the re-imaging and the first imaging is realized.

[0223] S9.4: Based on the initial multi-dimensional alignment reference feature set, the standardized phase response value of the exposed synchronous mark is taken as input, the phase response of the multi-layer synchronous mark is matched one by one through the feature corresponding algorithm, the spatial offset between the actual physical position of each mark point and the reference physical position is calculated, and the multi-layer alignment spatial residual matrix after the exposure is formed.

[0224] S9.5: The multi-layer alignment spatial residual matrix after the exposure is subjected to normalization processing, the batch working condition label such as exposure parameters, environmental temperature and humidity, stress disturbance and the like is comprehensively considered, the final alignment residual evaluation result is output by using the dynamic feedback checking algorithm, and criteria and feedback data are provided for subsequent abnormal alarm, compensation correction, process adaptive learning and consistency tracing processes.

[0225] The step S10: according to the dynamic feedback checking result, if the alignment residual does not meet the process requirement, the abnormal alarm or the supplementary exposure process is triggered, and the residual information is recorded to continuously optimize the distributed alignment error model and the adaptive optical processing algorithm, so as to improve the consistency and robustness of subsequent batch manufacturing. Specifically, it includes:

[0226] S10.1: Based on the actual post-exposure synchronous mark phase response data obtained by the re-imaging detection, the actual alignment residual is calculated by comparing the initial multi-dimensional alignment reference feature set, so as to obtain the batch actual alignment accuracy evaluation index.

[0227] S10.2: The threshold discrimination algorithm is applied to the batch actual alignment accuracy evaluation index, and it is judged whether the alignment residual falls within the preset process standard interval, so as to output an error overrun mark or a qualified determination signal.

[0228] S10.3: If the error overrun mark is activated, an abnormal alarm process is automatically triggered based on the error space distribution characteristics, and the batch process label, residual data and environmental parameters are pushed to the upper control system in real time to form an abnormal disposal data link.

[0229] S10.4: Based on the abnormal disposal data link, a step-by-step supplementary exposure strategy or a fine alignment correction strategy is adopted to implement local secondary exposure compensation for the local synchronous mark area with the largest residual, so as to correct the alignment error of the process defect area in real time and realize process closed-loop correction.

[0230] S10.5: The historical residual information and multi-source data such as process label and environmental parameters are fused, input into the distributed alignment error model and adaptive optical processing algorithm, and the model weight is iteratively optimized through machine learning to improve the error discrimination and compensation ability of the model under complex disturbance environment.

[0231] S10.6: The optimization results of the adaptive optical processing algorithm and the distributed alignment error model are back-tested and verified, and if the model is updated, the alignment residual of the same batch or subsequent batches is effectively converged, and the process adjustment data is automatically archived to provide data-driven parameter recommendation and robustness improvement basis for batch manufacturing links.

[0232] For those skilled in the art, various corresponding changes and modifications can be made to the above-described technical solutions and concepts, and all these changes and modifications should belong to the protection scope of the claims of the present application.

[0233] Unless otherwise defined, technical terms or scientific terms used herein shall have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terms "first", "second", "third" and similar terms used herein do not denote any order, quantity, or importance, but are used to distinguish one element from another. Also, the terms "a" or "an" do not denote a limitation of quantity, but rather denote the presence of at least one of the referenced items. The terms "including", "comprising", "having" and variations thereof, mean that the elements or objects following each such term are included, but not to the exclusion of other elements or objects. The terms "a plurality" or "a plurality of" means two or more. A and / or B means that there are three possibilities: A; B; and A and B.

[0234] The above description is only exemplary implementation of the present application, and the protection scope of the present application is not limited thereto. Any modification or replacement within the technical scope disclosed by the present application can be easily thought by those skilled in the art, and such modification or replacement should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A solder mask method for printed circuit board, specifically comprising: S1: obtaining PCB design data, and pre-setting micro-passive synchronization markers in a designated area of each layer of the circuit board to form distinguishable multi-layer alignment signals; S2: globally scanning the synchronization markers on the surface of the multi-layer PCB and the mask before exposing the solder mask film, and collecting original optical response signals of environmental differences and vibration interference; S3: preprocessing the original optical response signals to obtain standardized phase response data of the multi-layer synchronization markers; S4: inputting the standardized phase response data into an adaptive optical processing algorithm to extract the optical phase difference of each micro-passive synchronization marker, and obtaining a multi-dimensional alignment reference feature set; S5: establishing a distributed alignment error model of the multi-layer PCB under different environmental disturbance conditions according to the multi-dimensional alignment reference feature set; S6: generating a virtual alignment plane for the current inter-layer alignment series according to the distributed alignment error model; S7: if the distributed alignment error model exceeds a preset threshold interval, automatically correcting the alignment parameters of the exposure equipment and the mask pattern data according to the physical offset of each layer; S8: completing the actual exposure operation according to the corrected alignment parameters and mask pattern data during the exposure process of the solder mask film; S9: after the exposure is completed, detecting the synchronization marker area by imaging again, obtaining the phase response of the synchronization marker after the actual exposure, and calculating the final alignment residual error; S10: according to the dynamic feedback checking result, if the alignment residual error does not meet the process requirements, triggering an abnormal alarm or a supplementary exposure process, and continuously optimizing the distributed alignment error model and the adaptive optical processing algorithm.

2. The printed circuit board solder resist method according to claim 1, characterized by, The step S1 specifically comprises: structurally analyzing the PCB design data to extract the geometric layer structure, solder mask interval distribution and alignment accuracy requirement information of each layer of the target multi-layer PCB; based on a physical identification preset area list, using a specific optical phase coding design algorithm to automatically assign phase coding parameters to each layer of the identification area, and generating a unique synchronization marker parameter set; according to the synchronization marker parameter set, selecting an adaptive micro-passive synchronization marker structure, and generating a physical synchronization marker layout design file through parameterization of the PCB process database; fusing the physical synchronization marker layout design file with the original PCB layer layout to check the design rule conflicts, spatial overlap and signal interference problems in the synchronization marker embedding process, and outputting an optimal multi-layer physical synchronization marker integrated layout; based on the final multi-layer physical synchronization marker integrated layout, generating a process execution work order corresponding to the solder mask film exposure process, and issuing automatic instructions through the manufacturing process to guide the precise embedding of the physical synchronization marker in the actual circuit board manufacturing link and the realization of optical phase coding.

3. The printed circuit board solder resist method according to claim 1, characterized by, The step S2 specifically comprises: technically positioning the multi-layer PCB sample to be exposed and the mask surface matched therewith to determine the initial coordinate distribution of the micro-passive synchronization marker area in three-dimensional space; Based on the synchronous mark space distribution reference, the original optical image signal of the micro passive synchronous mark area on the multi-layer PCB and mask surface is synchronously acquired to obtain high-resolution original optical image data containing physical position, reflection / refraction difference and relative relationship of multi-layer marks; The high-resolution original optical image data is processed by using a multi-channel parallel reading mechanism to execute an automatic mark recognition algorithm and ROI extraction processing, and a synchronous mark area pixel set is outputted; The synchronous mark area pixel set is quantitatively extracted by using a phase encoding decoding algorithm to obtain the original phase response value of the synchronous mark under the conditions of current illumination, temperature and humidity and random vibration disturbance, and an original optical phase response matrix is generated; The original optical phase response matrix is associated with the synchronous mark space distribution reference and environmental parameters to synthesize a batch working condition optical response data set.

4. The printed circuit board solder resist method according to claim 1, characterized by, The step S3 specifically comprises: Baseline correction processing is performed on the original optical response signal outputted by the high-speed imaging device to obtain a baseline correction signal, background noise suppression is performed on the baseline correction signal, and a noise suppression signal is outputted; An adaptive time domain denoising technology is used to detect and eliminate abnormal time points corresponding to environmental temperature and humidity mutations, dust projections and local micro deformation disturbances from the noise suppression signal to form a noise elimination signal; Multi-layer synchronous mark area weighting processing is performed on the noise elimination signal, different weights are given based on the preset spatial distribution and physical coding characteristics of each synchronous mark, and a spatial weighting signal is obtained; Intensity and phase standardization transformation is performed on the spatial weighting signal to output standardized phase response data.

5. The printed circuit board solder resist method according to claim 1, characterized by, The step S4 specifically comprises: The standardized phase response data of the multi-layer synchronous mark is inputted into an adaptive optical processing algorithm module, phase difference calculation is performed on the standardized phase response data of each layer of synchronous mark by using a phase difference analysis algorithm, and optical phase difference characteristic parameters of the micro passive mark in the PCB mask system at different physical positions are obtained; Based on the obtained optical phase difference characteristic parameters, a feature vector generation algorithm is used to aggregate the phase difference multi-dimensional feature vector, a heterogeneous mark feature fusion model is used to weight the working condition related parameters, and a multi-dimensional alignment reference feature set is outputted.

6. The printed circuit board solder resist process of claim 1, wherein: In the step S1, the structured analysis of the PCB design data comprises automatically identifying the boundary of the solder mask interval, the hierarchical alignment accuracy label and the space embeddable area, and screening the final preset synchronous mark area by using a conflict avoidance and safety interval algorithm.

7. The printed circuit board solder resist process of claim 1, wherein: In the step S3, the spatial domain filtering comprises two-dimensional Gaussian filtering and multi-scale background modeling, the time domain denoising adopts a sliding window and change point detection, and the signal enhancement compensation is performed on the low SNR area by using weight promotion or data resampling.

8. The printed circuit board solder resist process of claim 1, wherein: In the step S4, the multi-dimensional feature extraction comprises principal component analysis and rigid affine model decomposition, respectively generating an overall rigid displacement vector and a local nonlinear deformation feature, and obtaining a structured representation of multi-layer physical displacement and deformation.

9. The printed circuit board solder resist process of claim 1 wherein: In the step S5, based on the overall rigid displacement parameter set and the local stress deformation parameter set, the local residual mapping and spatial clustering algorithm are used to quantitatively analyze and type clustering the stress features of the multi-layer PCB.

10. The printed circuit board solder resist process of claim 1, wherein: In the step S6, the regional disturbance resolution between the multi-layer synchronization markers and the reconstruction of the virtual alignment reference surface are obtained by principal component analysis, phase difference calculation and multi-layer space normalization, and the abnormal area is dynamically corrected and weighted removed.

Images