Method for on-line monitoring thickness of copper layer on PCB based on surface acoustic wave resonance

Abstract

The application provides a PCB copper cladding layer thickness online monitoring method based on surface acoustic wave resonance, relates to the technical field of electronic manufacturing, and comprises the following steps: step 1, design and preparation of a surface acoustic wave resonator: a stepped interdigital electrode structure or an asymmetric interdigital electrode array is prepared on the surface of a piezoelectric substrate of the surface acoustic wave resonator, the geometric parameters of the interdigital electrode are designed according to a target copper cladding layer thickness range and surface acoustic wave propagation characteristics, and the width, spacing and logarithm of the interdigital electrode are adjusted so that the working frequency range of the surface acoustic wave resonator covers the frequency shift range caused by the change of the copper cladding layer thickness; step 2, sensor installation and signal coupling; step 3, resonance frequency shift analysis and thickness mapping; and step 4, online feedback and process regulation. Through the precisely designed and optimized surface acoustic wave resonator and the efficient signal processing algorithm, accurate PCB copper cladding layer thickness monitoring can be provided in a complex environment.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic manufacturing, and in particular to an online monitoring method for the thickness of a PCB copper cladding layer based on surface acoustic wave resonance. Background Art

[0002] In high-end electronics manufacturing, such as high-frequency communication PCBs, thick copper PCBs, and automotive electronics PCBs, precise control of copper cladding thickness directly determines product performance and reliability. To avoid blind spots in inspection, thick copper PCB manufacturers are forced to excessively increase the design margin for electroplating uniformity, resulting in an increase in production costs of more than 15%. However, existing inspection technologies have significant limitations and are unable to meet the modern industry's demand for online, non-contact, high-precision monitoring, as shown in the following:

[0003] In existing technologies, the electromagnetic induction method is based on the eddy current effect principle and is only suitable for thickness detection of ferromagnetic metals (such as nickel and iron). It is completely ineffective for non-ferromagnetic pure copper layers widely used in the PCB industry.

[0004] X-ray fluorescence requires downtime for sampling and testing, disrupting production continuity. X-ray radiation protection requirements are high, increasing equipment costs and operational risks. The resolution of local thickness of multi-layer PCBs is insufficient (>±0.5 μm), making it impossible to identify uneven coatings in micro-areas.

[0005] Optical thickness measurement methods (such as spectral confocal) rely on analyzing the intensity of surface reflected light and are significantly affected by ambient light interference and the degree of copper cladding oxidation. The measurement error for rough surfaces (Ra>0.8 μm) or curved PCBs exceeds ±20%, and frequent calibration is required.

[0006] Therefore, developing a non-contact, high-precision, and highly anti-interference method for online monitoring of copper coating thickness has become an urgent need to break through the bottleneck of high-end electronic manufacturing. Summary of the Invention

[0007] In order to solve the problems of the prior art, the present invention provides an online monitoring method for the thickness of a PCB copper cladding layer based on surface acoustic wave resonance.

[0008] The technical solutions provided by the embodiments of the present invention are as follows:

[0009] The embodiment of the present invention provides a method for online monitoring of the thickness of a PCB copper coating based on surface acoustic wave resonance, comprising the following steps:

[0010] Step 1: Design and preparation of surface acoustic wave resonator:

[0011] A step-type interdigital electrode structure or an asymmetric interdigital electrode array is prepared on the surface of a piezoelectric substrate of a surface acoustic wave resonator, the geometric parameters of the interdigital electrode are designed and optimized according to a target copper cladding thickness range and surface acoustic wave propagation characteristics, and the working frequency band of the surface acoustic wave resonator covers the frequency shift range caused by the change of the copper cladding thickness by adjusting the width, pitch and logarithm of the interdigital electrode.

[0012] Step 2: sensor installation and signal coupling:

[0013] The surface acoustic wave resonator is fixed on the non-functional area of the PCB board by a high-temperature resistant adhesive material, the gap between the sensor and the copper cladding is adjusted by using a non-contact spacing calibration device, the mechanical vibration energy of the surface acoustic wave is effectively coupled to the surface of the copper cladding, and physical contact to the PCB surface is avoided.

[0014] A sweep excitation signal is transmitted to the sensor by a radio frequency reader, the frequency range of the sweep excitation signal is determined according to the design frequency band of the resonator, and the step interval and residence time are dynamically adjusted according to the signal signal-to-noise ratio requirement to capture the complete resonance response.

[0015] Step 3: resonance frequency shift analysis and thickness mapping:

[0016] The amplitude-frequency characteristics of the sensor reflected signal are collected, the frequency shift of the resonance frequency point is extracted by using a spectrum analysis algorithm, and the frequency shift is converted into a copper cladding thickness value by combining a pre-established frequency shift-thickness mapping model.

[0017] The frequency shift-thickness mapping model is constructed by combining finite element simulation and experimental calibration, the material properties, geometric structure and boundary conditions of the PCB substrate, copper cladding, adhesive layer and sensor are included in the finite element simulation, and the simulation error is corrected by experimental data to ensure the accuracy of the frequency shift-thickness mapping model under complex working conditions.

[0018] Step 4: online feedback and process control:

[0019] According to the deviation of the real-time thickness measurement result and the set target value, a closed-loop control algorithm is used to dynamically adjust the electroplating process parameters, the closed-loop control algorithm is based on the historical change trend and real-time rate of the thickness deviation for multivariate optimization to suppress process fluctuations and maintain the stability of the copper cladding thickness.

[0020] The technical scheme provided by the embodiment of the application has at least the following beneficial effects:

[0021] This method utilizes the principle of coupling surface acoustic wave resonators with mechanical vibration energy, avoiding potential PCB surface damage caused by traditional contact probes. By capturing the resonant frequency shift through a swept-frequency excitation signal and achieving micron-level resolution using a spectrum analysis algorithm, the sensor achieves a linearity of ±0.004mm, offering advantages over X-ray fluorescence (XRF) in continuous online monitoring scenarios. For multilayer PCB structures, directional interdigital electrodes and pseudo-random code frequency modulation (PFM) technology suppress interlayer signal crosstalk, addressing the depth-of-penetration limitations of eddy current testing methods in multilayer boards.

[0022] This method converts real-time resonant frequency shift data into thickness values ​​using a frequency shift-thickness mapping model, which then drives a closed-loop control system to dynamically adjust electroplating parameters. This method, which integrates a multivariable regression model with an adaptive PID controller, reduces copper thickness uniformity deviation from the traditional ±15% to ±6%, significantly outperforming the hysteresis control mode used in offline spot checks. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 The steps of the online monitoring method of the PCB copper coating thickness based on surface acoustic wave resonance provided by the embodiment of the present invention are as follows: Figure 1 ;

[0025] Figure 2 The steps of the online monitoring method of the PCB copper coating thickness based on surface acoustic wave resonance provided by the embodiment of the present invention are as follows: Figure 2 ;

[0026] Figure 3 A flowchart of the steps for determining process parameter weights through multivariate regression in the online monitoring method for PCB copper cladding thickness based on surface acoustic wave resonance provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0027] The technical solutions of the present invention are described below with reference to the accompanying drawings. It is also noted that, to provide a more detailed description, the following embodiments are best and preferred embodiments, and those skilled in the art may employ alternative implementations for certain known technologies. Furthermore, the accompanying drawings are intended only to provide a more detailed description of the embodiments and are not intended to limit the present invention.

[0028] It should be noted that references in the specification to "one embodiment," "an embodiment," "exemplary embodiments," "some embodiments," etc. indicate that the described embodiments may include specific features, structures, or characteristics, but not necessarily every embodiment will include such specific features, structures, or characteristics. Furthermore, when specific features, structures, or characteristics are described in conjunction with an embodiment, it is within the knowledge of persons skilled in the relevant art to implement such features, structures, or characteristics in conjunction with other embodiments (whether or not explicitly described).

[0029] In general, terms can be understood, at least in part, from their use in context. For example, depending at least in part on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending at least in part on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0030] like Figures 1 to 3 As shown, an embodiment of the present invention provides an online monitoring method for the thickness of a PCB copper clad layer based on surface acoustic wave resonance. The design of the surface acoustic wave resonator is the basis of the entire system. The key lies in the geometric structure optimization of the stepped interdigital electrode. By adjusting the width, spacing and logarithm of the interdigital electrode, the optimized design enables the resonator to generate an effective surface acoustic wave frequency shift within the target copper clad layer thickness range. Specifically, the operating frequency band of the surface acoustic wave resonator should cover the frequency shift range caused by changes in the copper clad layer thickness to ensure that the thickness changes can be accurately reflected.

[0031] The designed formula can be expressed as:

[0032] ;

[0033] in, is the frequency of the surface acoustic wave, is the speed of sound waves in the material, is the wavelength of the surface acoustic wave. The geometric parameters of the interdigital electrodes affect the propagation speed and wavelength of the acoustic wave, thereby affecting the frequency variation range.

[0034] In step 2, the SAW resonator is fixed to a non-functional area of ​​the PCB using a high-temperature-resistant adhesive. A non-contact spacing calibration device is used to adjust the gap between the resonator and the copper clad. This prevents damage to the PCB surface from physical contact while ensuring that the mechanical vibration energy of the SAW is effectively coupled to the copper clad.

[0035] The sweep excitation signal is transmitted by the radio frequency reader, and the frequency range is determined according to the design frequency band of the resonator. The step interval and dwell time of the sweep signal are dynamically adjusted according to the signal-to-noise ratio requirement. The purpose of this is to capture the complete resonance response and ensure that the signal has sufficient frequency resolution to accurately reflect the changes in the thickness of the copper layer.

[0036] In step 3, the frequency shift of the resonant frequency point is extracted by collecting the amplitude-frequency characteristics of the reflected signal using a spectrum analysis algorithm. This frequency shift reflects the change in the thickness of the copper layer. The spectrum analysis algorithm can extract the frequency shift information through Fourier transform, and the formula is as follows:

[0037] ;

[0038] Where, is the frequency shift, is the actual measured resonant frequency, is the reference frequency (i.e. the frequency without the copper layer).

[0039] Then, based on the pre-established frequency shift-thickness mapping model, the frequency shift is converted to a specific thickness value. This mapping model is constructed through finite element simulation and experimental calibration. Finite element simulation considers the material properties, geometric structure and boundary conditions of PCB substrate, copper layer, adhesive layer and sensor. Simulation data can provide preliminary frequency shift-thickness relationship, while experimental data can correct simulation errors and ensure the accuracy of the model under actual working conditions.

[0040] According to the deviation of the real-time measured copper layer thickness from the set target value, the closed-loop control algorithm starts to work. This algorithm dynamically adjusts the electroplating process parameters according to the historical thickness deviation trend and real-time change rate. Through multivariate optimization of historical data, the closed-loop control algorithm can suppress process fluctuations and maintain the stability of the copper layer thickness.

[0041] The specific control algorithm can be represented by the following formula:

[0042] ;

[0043] Where, is the adjustment amount of the electroplating time, is the thickness deviation, is the real-time thickness change rate,

[0044] and is the control coefficient optimized according to historical data. Through accurate adjustment of the electroplating time, the thickness of the copper layer can be corrected in real time to ensure that it always remains within the target range.

[0045] The online monitoring method for PCB copper cladding thickness based on surface acoustic wave resonance realizes highly automated thickness measurement and control, significantly improving the accuracy and stability of the production process, reducing the scrap rate, and improving overall production efficiency.

[0046] In one possible implementation, if the PCB is a multi-layer structure, further optimization of the sensor design and signal processing methods is necessary to suppress inter-layer signal crosstalk. Specifically, by designing directional interdigital electrodes, using pseudo-random code frequency modulation, and cross-correlation algorithms, signal accuracy and signal separation in the multi-layer structure are ensured, preventing signal interference between different layers.

[0047] In multilayer PCB structures, the design of directional interdigital electrodes is crucial to suppress crosstalk between layers. Specifically, an asymmetric interdigital electrode array is designed on the surface of the piezoelectric substrate. The interdigital widths of the transmitting and receiving electrodes are optimized based on the wavelength of the surface acoustic wave (SAW) and the spatial distribution of sensitive copper layer thicknesses. By adjusting the interdigital widths, the propagation characteristics of the SAW can be controlled, ensuring that the wave propagates within a specific area and preventing interference between different layers.

[0048] In addition, acoustic wave reflectors are installed at the ends of the electrodes. These reflectors constrain the propagation direction of the acoustic waves, ensuring they propagate only in the intended direction, thereby reducing crosstalk between interlayer signals. The design of the reflectors needs to be optimized based on the propagation direction and frequency characteristics of the surface acoustic wave. Typically, their spacing is determined based on the acoustic wave propagation characteristics and the material differences between the PCB layers.

[0049] To further suppress signal interference, pseudo-random code frequency modulation (PSFM) is applied to the carrier frequency of each sensor. This approach aims to make the signal more unique by varying the frequency modulation depth and code length of each sensor, thereby reducing interference when multiple sensor signals are superimposed. The code length and modulation depth of the Pseudo-random code are dynamically adjusted based on the interference intensity of the superimposed multi-sensor signals. This frequency modulation method enables the differentiation and decoupling of multiple sensor signals within the same frequency band.

[0050] After the signals are superimposed, a cross-correlation algorithm is needed to effectively separate the reflected signals of different sensors. The core idea of ​​the cross-correlation algorithm is to extract the target signal by calculating the similarity between the signals based on the frequency modulation characteristics of different sensors. Specifically, the cross-correlation function is used to describe the signal and The correlation between them is as follows:

[0051] ;

[0052] in, is the time delay, and By calculating the cross-correlation function of the reflection signals of different sensors, the reflection signal of each sensor can be accurately identified and extracted, thereby effectively separating the signals of each layer and avoiding crosstalk between layers.

[0053] The spacing of the reflector gratings is calculated based on the attenuation coefficient of surface acoustic waves in the dielectric between PCB layers. Specifically, the attenuation coefficient is determined by factors such as the signal propagation distance, the physical properties of the dielectric, and the frequency. By properly designing the spacing of the reflector gratings, the signal attenuation between adjacent layers exceeds a preset threshold. A larger attenuation coefficient increases the signal propagation attenuation, thereby reducing interlayer signal interference and improving signal independence and accuracy in multilayer PCBs.

[0054] Combining the design of directional interdigital electrodes, pseudo-random code frequency modulation, and the application of cross-correlation algorithm can effectively suppress signal interference in multi-layer PCB structures and improve the measurement accuracy and stability of the system in complex environments.

[0055] In one possible implementation, when the PCB surface is rough, additional measures are needed to address signal distortion caused by surface unevenness. To ensure the accuracy and stability of the monitoring signal, this method effectively addresses the rough surface issue by employing techniques such as an acoustic impedance matching layer, time-domain reflectometry, and wavelet threshold denoising.

[0056] Specifically, an acoustic impedance matching layer (AIM) is placed between the sensor and the copper clad to ensure that surface acoustic wave (SAW) signal transmission is unaffected by surface roughness. The AIM of the AIM is selected based on the roughness level of the PCB surface. Optimizing this AIM helps reduce acoustic wave reflections caused by the rough surface. The thickness gradient of the AIM is inversely proportional to the 3D scan data of the copper clad surface profile. This means that the thickness of the AIM is adjusted to the specific surface roughness of the PCB, resulting in a thicker AIM in rougher areas, effectively mitigating signal reflections and distortion caused by surface irregularities.

[0057] Furthermore, a smoothing function is used to connect the thickness transitions between adjacent micro-regions of the matching layer to ensure a smooth transition and avoid excessively abrupt thickness changes that could lead to unstable reflected waves. The design of the smoothing function must take into account the propagation and attenuation characteristics of acoustic waves in the matching layer to ensure a smooth transition of reflected waves between different regions.

[0058] After filling the matching layer, time domain reflectometry (TDR) is used to extract signal distortion components caused by surface roughness. TDR accurately analyzes the reflected signal from the sensor, extracting distortion characteristics caused by surface irregularities. These distortions appear as additional reflection peaks in the time domain waveform, affecting signal accuracy.

[0059] In order to effectively remove these distortions, the time domain waveform of the reflected signal will be convolved with the standard smooth surface reference waveform. The convolution operation is as follows:

[0060] ;

[0061] in, is the reflected signal, Is a standard smooth surface reference waveform, is the convolved signal. The convolution operation effectively matches the characteristics of the standard reference waveform with the reflected signal and extracts the distortion caused by surface roughness.

[0062] After processing the signal in the time domain, further noise and interference removal is required through frequency domain processing. To this end, a wavelet threshold denoising algorithm is used to remove interference from the frequency domain signal. Wavelet transform is an effective signal analysis method that decomposes the signal into different frequency components, thereby detecting and removing noise at different scales.

[0063] The basic idea of ​​the wavelet threshold denoising algorithm is to first decompose the signal into low-frequency and high-frequency components through wavelet transform, and then apply the threshold function to the high-frequency components to remove the noise. The steps of wavelet threshold denoising can be expressed as follows:

[0064] ;

[0065] in, are the coefficients after wavelet transformation, is the threshold parameter, is the coefficient after denoising, is a sign function of the coefficients. In this way, high-frequency noise components are suppressed while the effective components of the signal are retained, thereby improving the quality and accuracy of the signal.

[0066] The transition rate of the smoothing function is determined based on the attenuation characteristics of acoustic wave propagation in the matching layer. The attenuation coefficient of acoustic waves determines the ability of a signal to propagate through different media. In an acoustic impedance matching layer, higher attenuation results in faster signal decay. Therefore, the transition rate of the matching layer must be designed based on the attenuation coefficient to ensure smooth thickness variations and avoid excessive reflections. Properly designed transition rates help ensure that the interface reflection energy is below the set percentage of the total surface acoustic wave energy, thereby preventing excessive accumulation of reflections and ensuring system accuracy.

[0067] Filling acoustic impedance matching layers, time domain reflectometry analysis, wavelet threshold denoising algorithm and precise smoothing function design can effectively address the challenges brought by surface roughness and improve the accuracy and robustness of multi-layer PCB thickness monitoring methods.

[0068] In one possible implementation, when a system experiences vibration disturbances, specific measures need to be taken to reduce the impact of vibration on frequency shift. This can be effectively addressed by integrating topologically optimized porous damping elements and establishing a transfer function model between vibration acceleration and frequency shift.

[0069] Specifically, a topologically optimized porous damping element is integrated into the sensor package. Its primary function is to suppress interference from system vibration on the measurement signal. Vibration table testing and least-squares fitting determined the damping element's stiffness and damping coefficient to ensure its effective attenuation of vibration signals. Specifically, the damping element is designed to ensure that the vibration attenuation rate in the main interference frequency band exceeds a preset threshold, significantly reducing vibration interference with the surface acoustic wave signal and improving measurement stability and accuracy.

[0070] To identify and eliminate the frequency shift caused by vibration in real time, the system establishes a transfer function model between vibration acceleration and frequency shift. The basic assumption of this model is that there is a linear or nearly linear relationship between vibration acceleration and frequency shift, which can be described by a transfer function.

[0071] The transfer function model is of the form:

[0072] ;

[0073] in, is the frequency shift caused by vibration, is the acceleration of vibration, It is the transfer function between vibration acceleration and frequency shift, which reflects the influence of vibration frequency on frequency shift.

[0074] In practice, the transfer function is obtained by experimentally measuring vibration frequency, amplitude, and sensor position information and fitting it using the least squares method. This data is used to determine the degree to which vibration affects frequency shift, thereby building a more accurate model.

[0075] The sensor collects three-axis acceleration data in real time and inputs it into the transfer function model to calculate the frequency shift caused by vibration. This process can deduct the effect of vibration from the total frequency shift, thereby obtaining a more accurate copper layer thickness signal.

[0076] To improve the accuracy and timeliness of frequency shift predictions, the system incorporates a recurrent neural network (RNN) for dynamic prediction. RNNs process time series data and provide real-time predictions of the frequency shift caused by vibration. The network's training data covers the typical vibration spectrum found in production lines, ensuring the model can handle a wide range of vibration environments.

[0077] By processing the input acceleration data through a recursive neural network, the model can predict the frequency shift caused by vibration in real time and deduct it from the total frequency shift, thereby restoring the most realistic copper cladding thickness data.

[0078] By integrating porous damping elements, establishing a transfer function model and combining recursive neural networks for frequency shift prediction, the influence of vibration on surface acoustic wave resonance measurement can be effectively suppressed, significantly improving the accuracy and robustness of the online monitoring method for PCB copper cladding thickness.

[0079] In one possible implementation, temperature is a key factor affecting the propagation velocity of surface acoustic waves, so effective compensation measures are needed to eliminate temperature interference with measurement results. Step 1 further includes arranging a distributed array of micro-thermocouples within the sensor and combining it with a spatial interpolation algorithm to reconstruct the temperature field.

[0080] First, a distributed array of micro-thermocouples is placed within the sensor. The spacing of these thermocouples is determined by thermal conductivity characteristics. Thermocouples are typically placed at different locations on the PCB to ensure comprehensive sensing of temperature changes. The sampling interval of the thermocouple array should be chosen to ensure that the temperature distribution across the entire PCB area is covered. Avoiding excessive spacing that could cause errors in temperature field reconstruction.

[0081] The collected temperature data is interpolated by the spatial interpolation algorithm to reconstruct the two-dimensional temperature field of the entire PCB area. Spatial interpolation is to infer the temperature value of the blank position based on the known temperature data at different positions. Common interpolation methods include linear interpolation, bilinear interpolation or a higher precision method based on spline interpolation. Assume that the known temperature point is , spatial interpolation can calculate the temperature value of the unknown point by the following formula :

[0082] ;

[0083] in, is the interpolation weight, typically a function based on distance or spatial correlation. In practice, this function can employ techniques such as Lagrange interpolation, polynomial interpolation, or Gaussian process regression. The goal of the interpolation algorithm is to reconstruct the entire temperature field from a limited number of temperature measurement points, ensuring that the effect of temperature at each location on SAW propagation can be accurately estimated.

[0084] To eliminate the effects of temperature on the surface acoustic wave resonant frequency shift, a multidimensional compensation mapping table for the temperature-frequency shift relationship was established based on thermodynamic transient simulation. The simulation model incorporates the anisotropic thermal conductivity of the piezoelectric material and convection boundary conditions. Temperature gradients can cause changes in the acoustic phase velocity of the material, which can affect the measured surface acoustic wave frequency.

[0085] By calculating the change in acoustic phase velocity under different temperature fields, It can be expressed as a function of the temperature gradient:

[0086] ;

[0087] in, is the temperature sensitivity coefficient related to the piezoelectric material, is the spatial derivative of the temperature field, representing the magnitude of the temperature gradient. This relationship allows us to derive the effect of temperature on the propagation velocity of acoustic waves and generate a temperature-frequency shift compensation mapping table. The dimensionality of this compensation mapping table is determined by the number of temperature sensors used and the required accuracy. Generally, the more temperature sensors used, the higher the required compensation dimensionality.

[0088] Using the compensation mapping table, frequency shift compensation is performed using real-time temperature field data during the actual monitoring process. This effectively calibrates phase velocity changes caused by temperature variations, resulting in more accurate PCB copper thickness measurements.

[0089] The combination of the spatial interpolation algorithm and the temperature-frequency shift compensation mapping table effectively solves the impact of temperature changes on the surface acoustic wave frequency shift, and significantly improves the accuracy and robustness of the online monitoring method for PCB copper cladding thickness based on surface acoustic wave resonance.

[0090] In one possible implementation, first, a thickness deviation prediction model needs to be constructed, which uses the ion concentration and flow pulsation of the plating solution as feedforward variables. Ion concentration and flow pulsation are the main factors affecting the thickness of the copper layer during the electroplating process. By real-time monitoring of these variables, the thickness deviation under the current process state can be predicted. The prediction model can be established by methods such as multivariate linear regression or polynomial regression. Assume that the thickness deviation is , the ion concentration of the plating solution is , flow pulsation is , then the deviation prediction model can be expressed as:

[0091] ;

[0092] in, is the regression coefficient determined by regression analysis, are other factors that may affect the thickness. The purpose of this model is to predict the thickness deviation based on the feedforward variables, thereby providing data support for closed-loop control.

[0093] According to the output of the thickness deviation prediction model, the gain parameters of the adaptive PID (proportional-integral-differential) controller are dynamically adjusted. The role of the traditional PID controller is to adjust according to the deviation, but with the adaptive PID controller, the system can dynamically adjust the PID parameters according to the real-time thickness deviation changes to improve the control accuracy and response speed. In this method, the gain parameters of the PID controller include the proportional gain , integral gain , differential gain , they will automatically adjust as the thickness deviation changes to ensure that the PCB copper layer thickness is always within the set process tolerance range.

[0094] A deadband compensation module is embedded in the control loop to reduce control sensitivity when the thickness deviation is within the process tolerance band. The process tolerance band is dynamically updated based on the production line's historical stability data and reflects the allowable thickness fluctuation range during the production process. The introduction of the deadband compensation module weakens the controller's response when the thickness deviation is less than a certain range, avoiding over-adjustment and preventing frequent operation and unnecessary fluctuations caused by over-adjustment. The mathematical model for deadband compensation can be expressed as:

[0095] ;

[0096] in, These are the upper and lower limits of the process tolerance band. This module can effectively reduce the control response when the thickness deviation is small and within the tolerance band, thereby improving the stability of the system.

[0097] In the control strategy, the output priority of the PID controller is allocated according to the contribution weight of each control parameter. Through multiple sets of orthogonal experiments in the electroplating tank, the regression analysis method is used to determine the contribution of different process parameters (such as electroplating solution ion concentration, flow rate, etc.) to the thickness of the copper layer. The regression analysis of the orthogonal experiment can quantify the influence of each parameter and calculate the contribution of each parameter through the regression coefficient. The weights of each parameter are calculated and used to adjust the output priority of the PID controller. For example, for ion concentration and flow pulsation If the regression analysis results show If the influence on thickness deviation is greater, the PID controller will give priority to the change of ion concentration in the adjustment process.

[0098] Based on the combination of multivariate regression analysis, thickness deviation prediction model and adaptive PID controller, the online monitoring method of PCB copper layer thickness based on surface acoustic wave resonance has significantly improved the control accuracy, stability and efficiency.

[0099] In one possible implementation, first, the difference between the finite element simulation data and the experimental calibration data is decomposed in the frequency domain. Since material anisotropy, interface defects and other factors may cause measurement errors, frequency domain decomposition can help identify the systematic components of the error. For example, the anisotropy of the material will lead to different sound wave propagation speeds, while interface defects may cause local disturbances during the propagation of surface acoustic waves. Through frequency domain analysis, these error components can be separated and then quantitatively analyzed and compensated. After the error components are identified, the changing law of the surface acoustic wave resonant frequency shift under different environmental conditions can be understood more accurately, thereby providing a theoretical basis for subsequent compensation steps.

[0100] To further improve the accuracy of error compensation, a convolutional neural network (CNN) was used to analyze the discrepancies between finite element simulation data and experimental calibration data. The CNN model can capture complex nonlinear relationships and, supported by extensive experimental and simulation data, adaptively identify and predict the sources of measurement error. Specifically, the CNN inputs include simulation parameters, environmental variables (such as temperature and humidity), and material batch information, all of which can affect the propagation characteristics of surface acoustic waves and cause errors. The network outputs corrections for frequency shifts, which are used to compensate for errors caused by environmental variations and material batch factors.

[0101] In this way, CNN can dynamically predict and correct measurement errors caused by external changes based on the real-time environment and material status, greatly improving the accuracy of the frequency shift-thickness mapping model.

[0102] During experimental calibration, simultaneous comparative measurements with a standard thickness sample allow verification and adjustment of error compensation. The standard thickness sample provides a known and stable thickness value as a baseline for comparison. By performing simultaneous measurements with the standard sample, the effectiveness of the model compensation can be further verified, and compensation parameters can be fine-tuned based on the comparison results.

[0103] By combining frequency domain decomposition, convolutional neural network analysis, and comparative measurement of standard thickness samples, more accurate and reliable online thickness monitoring can be achieved, providing higher system stability and measurement quality, thereby promoting the refinement and automation of the PCB manufacturing process.

[0104] In one possible implementation, a nanoscale sawtooth structure is first designed on the edge of the interdigitated electrode. The depth and period ratio of the sawtooth structure are determined based on the wavelength of the surface acoustic wave (SAW) and the acoustic impedance matching requirements of the copper cladding. The propagation speed and characteristics of SAWs are affected by the physical properties (such as thickness and material) and acoustic impedance of the copper cladding. Therefore, by optimizing the depth and period of the sawtooth structure, the propagation characteristics of SAWs can be effectively adjusted, thereby improving the coupling efficiency of acoustic waves in the copper cladding.

[0105] The ratio of the sawtooth structure's depth to its period must be precisely calculated to ensure optimal matching at a specific surface acoustic wave frequency. If the sawtooth depth and period are properly set, they can enhance the shear coupling between the electrode and the copper cladding, making acoustic wave transmission more efficient.

[0106] The sawtooth angle is also a key design parameter. This angle needs to form a predetermined angle with the direction of acoustic wave propagation to maximize the shear coupling efficiency between the acoustic wave propagation direction and the copper cladding. By properly selecting the sawtooth angle, surface acoustic wave coupling between the electrode and the copper cladding is more efficient, thereby improving the sensitivity and accuracy of thickness measurement.

[0107] To further optimize the electrode structure and improve coupling efficiency, electromagnetic-acoustic field coupling simulation analysis was performed. This simulation process involves the coupling effects of multiple physical fields, including the PCB substrate, copper cladding, and adhesive layer. Through simulation, the propagation behavior of surface acoustic waves in different structures can be accurately simulated, thereby selecting the electrode structure with the highest energy concentration. Specifically, the simulation analysis can predict the effects of different electrode designs (including serrated structures, tilt angles, etc.) and ultimately select the optimal design by comparing the energy concentration of each design solution.

[0108] The design of the sawtooth structure, the optimization of the tilt angle, and the electromagnetic-acoustic field coupling simulation analysis can effectively improve the performance of the online monitoring method for PCB copper cladding thickness. This not only increases the measurement sensitivity and accuracy, but also ensures the stability and reliability of the system under complex working conditions.

[0109] In one possible implementation, a non-contact distance calibration device is introduced in step 2, and its working method mainly includes the coordinated operation of optical ranging and precision displacement control system, as well as the scientific selection and structural optimization of high-temperature resistant adhesive materials.

[0110] Specifically, the calibration device incorporates a high-precision distance measurement module that uses optical interferometry or laser triangulation to non-contactly monitor the real-time distance between the sensor and the PCB copper surface. This system achieves micron and even sub-micron distance measurement accuracy, ensuring the sensor is always in the ideal measurement distance position during dynamic production processes.

[0111] The measurement results are fed into the feedback control module in real time. This module calculates the displacement correction based on the error between the current spacing value and the preset optimal spacing value, and then controls the precision displacement platform for dynamic adjustment. The displacement platform is typically a multi-degree-of-freedom piezoelectric ceramic drive or a precision stepping mechanism, enabling high-speed response and high-resolution displacement adjustment, keeping the sensor in the operating range with the highest surface acoustic wave energy transfer efficiency.

[0112] Under the complex operating conditions of PCB production lines, such as high temperature, high speed, and vibration, spacing fluctuations are inevitable. This system achieves dynamic and stable control of spacing through continuous detection and feedback correction, greatly improving the transmission efficiency of surface acoustic waves between the sensor and the copper cladding, thereby enhancing the accuracy and stability of thickness detection.

[0113] The sensor is fixed to the bracket through an adhesive layer, and the bracket is connected to an adjustable precision displacement platform. The platform receives the ranging feedback signal and performs fine-tuning to ensure that the entire sensor system maintains the optimal distance from the PCB copper layer.

[0114] The relative position of the sensor and the PCB is precisely controlled through optical ranging and feedback control systems, and a stable bonding solution is used to provide a solid mechanical and signal foundation for the entire surface acoustic wave thickness detection system.

[0115] In one possible embodiment, in step 2, a working method is further proposed of applying ultrasonic vibration excitation and monitoring the change of acoustic impedance in real time during the curing process of the adhesive layer, while at the same time increasing the thermal conductivity of the adhesive layer and suppressing high-temperature creep by adding a nanoparticle reinforcement phase to the adhesive material.

[0116] Specifically, during the curing process of the adhesive layer, the system applies high-frequency ultrasonic waves to the adhesive layer via an ultrasonic vibration excitation source. This excitation source is typically an ultrasonic signal generated by a surface acoustic wave sensor or other ultrasonic transducer. The frequency is adjustable, and the appropriate excitation frequency is selected based on the bonding characteristics of the material. The ultrasonic vibrations promote the directional alignment of the molecular structure within the adhesive layer and influence the physical changes during the curing process.

[0117] As ultrasonic excitation is applied, the propagation of sound waves in the adhesive layer is affected by the degree of curing of the adhesive material. By measuring the changes in the acoustic impedance of the adhesive layer during the curing process, the system provides real-time feedback on the curing status of the adhesive layer. Acoustic impedance is determined by the density and speed of sound of the material. As the adhesive layer progresses from uncured to fully cured, its density and elastic modulus change, affecting the propagation characteristics of the sound waves. By monitoring changes in acoustic impedance, the real-time status of the curing process can be accurately determined.

[0118] Based on real-time feedback from acoustic impedance, the system dynamically adjusts the precision displacement stage's feedback control parameters according to the curing state. If the curing level does not meet expectations, the control system automatically adjusts the displacement stage's motion or pressure distribution to maintain the optimal spacing between the sensor and the copper cladding, thereby ensuring efficient surface acoustic wave transmission.

[0119] To improve the thermal conductivity of the adhesive layer and suppress high-temperature creep, nanoparticle reinforcement is added to the adhesive material. Commonly used nanoparticles include aluminum oxide, silicon nitride, and carbon nanotubes, which exhibit excellent thermal conductivity and high-temperature resistance. By carefully selecting the type and volume fraction of nanoparticles, the adhesive layer maintains excellent thermal conductivity during the high-temperature curing process, reducing stress caused by thermal gradients and thus minimizing creep effects.

[0120] The particle size distribution of nanoparticles is crucial for their dispersion and thermal conductivity within the adhesive layer. By using a centrifugal classification process, the particle size distribution of nanoparticles is controlled to ensure uniform dispersion within the adhesive matrix. This process effectively separates nanoparticles of varying sizes, preventing aggregation and ensuring that each nanoparticle is evenly distributed within the adhesive layer, thereby enhancing material performance.

[0121] By combining ultrasonic excitation with acoustic impedance monitoring, accurate information on the curing state of the adhesive layer can be obtained in real time, ensuring that the material properties during the bonding process are consistent with expectations. This in turn ensures the ideal spacing between the sensor and the copper cladding layer and improves the surface acoustic wave transmission efficiency.

[0122] The addition of nanoparticles effectively improves the thermal conductivity of the adhesive layer, enabling it to distribute heat more evenly in high-temperature environments, inhibiting creep at high temperatures, and improving the stability and reliability of the adhesive layer.

[0123] Combining ultrasonic excitation, acoustic impedance monitoring, and nanoparticle reinforcement techniques can significantly improve the accuracy, stability, and high-temperature resistance of this method, ensuring quality control of the adhesive layer during high-temperature curing and providing technical support for long-term, reliable PCB production.

[0124] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0125] This invention utilizes a precisely designed and optimized surface acoustic wave (SAW) resonator and efficient signal processing algorithms to accurately monitor PCB copper thickness in complex environments. Incorporating multiple compensation technologies (such as vibration, temperature, and surface roughness), it ensures real-time feedback and precise control of copper thickness during production, significantly improving process optimization and quality assurance.

[0126] The present invention encompasses any alternatives, modifications, equivalents, and solutions that fall within the spirit and scope of the present invention. While specific details are described in detail in the preferred embodiments to provide a thorough understanding of the present invention, those skilled in the art will be able to fully understand the present invention without these details. Furthermore, to avoid unnecessary confusion regarding the essence of the present invention, well-known methods, processes, procedures, components, and circuits have not been described in detail.

[0127] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for online monitoring of PCB copper coating thickness based on surface acoustic wave resonance, characterized in that: The following steps are involved: Step 1: Design and preparation of surface acoustic wave resonator: A stepped interdigitated electrode structure or an asymmetric interdigitated electrode array is fabricated on the surface of a piezoelectric substrate of a surface acoustic wave resonator. The geometric parameters of the interdigitated electrodes are optimized based on a target copper cladding thickness range and surface acoustic wave propagation characteristics. By adjusting the width, spacing, and number of pairs of interdigitated electrodes, the operating frequency band of the surface acoustic wave resonator covers the surface acoustic wave frequency shift range caused by variations in copper cladding thickness. Step 1 also includes: arranging a distributed micro-thermocouple array inside the sensor, with sampling intervals determined according to heat conduction characteristics, and reconstructing a two-dimensional temperature field using a spatial interpolation algorithm; A multi-dimensional compensation mapping table for the temperature-frequency shift relationship is established based on thermodynamic transient simulation. The thermodynamic transient simulation defines the anisotropic thermal conductivity and convection boundary conditions of the piezoelectric material and calculates the change in acoustic wave phase velocity caused by the temperature gradient. The phase velocity variation is proportional to the spatial derivative of the temperature field, and the dimension of the compensation mapping table is determined according to the number of temperature sensors and the accuracy requirements; Step 2: Sensor installation and signal coupling: The surface acoustic wave resonator is fixed to a non-functional area of ​​the PCB board using a high-temperature resistant adhesive material. The gap between the sensor and the copper layer is adjusted using a non-contact spacing calibration device to effectively couple the mechanical vibration energy of the surface acoustic wave to the copper layer surface while avoiding damage to the PCB surface caused by physical contact. Step 3: Resonance frequency shift analysis and thickness mapping: The amplitude-frequency characteristics of the sensor's reflected signal are collected, and the offset of the resonant frequency point is extracted using a spectrum analysis algorithm. Combined with the pre-established frequency shift-thickness mapping model, the frequency shift is converted into the copper cladding thickness value; Step 3 also includes: the frequency shift-thickness mapping model is constructed through the joint use of finite element simulation and experimental calibration. The finite element simulation includes the material properties, geometric structure and boundary conditions of the PCB substrate, copper layer, adhesive layer and sensor, and the simulation error is corrected by experimental data to ensure the accuracy of the frequency shift-thickness mapping model under complex working conditions. The error compensation method of the frequency shift-thickness mapping model includes: Perform frequency domain decomposition on the difference data between finite element simulation and experimental calibration to identify systematic error components caused by material anisotropy and interface defects; A convolutional neural network is used to analyze the differences between finite element simulation data and experimental calibration data, predicting and compensating for measurement errors caused by environmental variables and material batch factors. The network input includes simulation parameters, environmental variables, and material batch information, and the output is a frequency shift correction value. During the experimental calibration, a standard thickness sample is used for synchronous comparative measurement; Step 4: Online feedback and process control: Based on the deviation between the real-time thickness measurement results and the set target value, a closed-loop control algorithm is used to dynamically adjust the electroplating process parameters. The closed-loop control algorithm performs multivariable optimization based on the historical change trend of thickness deviation and the real-time rate to suppress process fluctuations and maintain the stability of the copper layer thickness. The closed-loop control algorithm in step 4 determines the process parameter weights through multivariate regression, specifically including: A thickness deviation prediction model is constructed by taking the plating solution ion concentration and flow pulsation as feedforward variables, and the gain parameters of the adaptive PID controller are dynamically adjusted. A deadband compensation module is embedded in the control loop to reduce control sensitivity when the thickness deviation is within the process tolerance band, which is dynamically updated based on the historical stability data of the production line; The output priority of the PID controller is assigned according to parameter contribution weights, and the weight values ​​are obtained by regression analysis of multiple groups of orthogonal experiments in the electroplating tank.

2. The method for online monitoring of PCB copper coating thickness based on surface acoustic wave resonance according to claim 1, characterized in that: When the PCB has a multi-layer structure, directional interdigital electrodes are used to suppress inter-layer signal crosstalk, specifically including: An asymmetric interdigitated electrode array is designed on the surface of the piezoelectric substrate. The difference in interdigitated width between the transmitting and receiving electrodes is determined by the spatial distribution of sensitive areas of the surface acoustic wave wavelength and the thickness of the copper layer. Acoustic wave reflection gratings are loaded at the ends of the electrodes to constrain the propagation direction of the acoustic wave. Applying pseudo-random code frequency modulation to the carrier frequency of each sensor, where the code length and modulation depth of the pseudo-random code are dynamically adjusted according to the interference intensity when multiple sensor signals are superimposed, and the reflected signals of different sensors are identified through a cross-correlation algorithm; The reflection grating spacing of the directional interdigital electrodes is calculated based on the attenuation coefficient of the surface acoustic wave in the PCB interlayer medium to ensure that the signal attenuation of adjacent layers exceeds a preset threshold.

3. The method for online monitoring of PCB copper coating thickness based on surface acoustic wave resonance according to claim 1, wherein: When the PCB has a rough surface, step 2 further includes: filling an acoustic impedance matching layer between the sensor and the copper clad layer, wherein the acoustic impedance value of the matching layer is selected according to the surface roughness level, the thickness gradient distribution of the matching layer is inversely proportional to the three-dimensional scanning data of the copper clad layer surface profile, and the thickness transition between adjacent micro-micro-layers is connected by a smooth function; The signal distortion component caused by surface roughness is extracted through time domain reflection analysis. The time domain waveform of the reflected signal is convolved with the standard smooth surface reference waveform, and the interference term is removed from the frequency domain signal using the wavelet threshold denoising algorithm. The transition rate of the smoothing function is determined according to the propagation attenuation characteristics of the acoustic wave in the matching layer, ensuring that the interface reflection energy is lower than a set proportion of the total energy of the surface acoustic wave.

4. The method for online monitoring of PCB copper coating thickness based on surface acoustic wave resonance according to claim 2 or 3, characterized in that: Step 1 also includes: integrating a topologically optimized porous damping element into the sensor package, wherein the stiffness and damping coefficient of the porous damping element are determined by vibration table testing and least squares fitting, so that the vibration attenuation rate of the main interference frequency band exceeds a preset threshold; A transfer function model of vibration acceleration and frequency shift is established, and three-axis acceleration data is collected in real time. The frequency shift caused by vibration is predicted using a recursive neural network and deducted from the total frequency shift. The input of the transfer function model includes vibration frequency, amplitude and sensor position information, and the neural network training data covers the typical vibration spectrum of the production line.

5. The method for online monitoring of PCB copper coating thickness based on surface acoustic wave resonance according to claim 1, characterized in that: The step-type interdigitated electrode in step 1 further includes: A nano-scale sawtooth structure is designed on the edge of the interdigitated finger. The sawtooth depth and period ratio are determined according to the surface acoustic wave wavelength and the acoustic impedance matching requirements of the copper cladding layer. The sawtooth inclination angle forms a preset angle with the sound wave propagation direction, maximizing the shear coupling efficiency between the sound wave and the copper cladding layer; Electromagnetic-acoustic field coupling simulation is used to screen the electrode structure with the highest energy concentration. The simulation includes the multi-physics field coupling effects of the PCB substrate, copper cladding layer, and adhesive layer.

6. The method for online monitoring of PCB copper coating thickness based on surface acoustic wave resonance according to claim 1, characterized in that: Step 2 also includes: transmitting a swept frequency excitation signal to the sensor through the radio frequency reader, wherein the frequency range of the swept frequency excitation signal is determined according to the design frequency band of the resonator, and the step interval and dwell time are dynamically adjusted according to the signal-to-noise ratio requirement to capture the complete resonant response; The working method of the non-contact distance calibration device includes: The distance between the sensor and the copper clad surface is monitored in real time using the principle of optical interference or laser triangulation. The precision displacement platform is driven by feedback control to dynamically adjust the distance so that the distance is stabilized within the range of optimal surface acoustic wave energy transfer efficiency. The selection of the high-temperature resistant adhesive material is based on the temperature curve of the PCB manufacturing environment and the chemical compatibility test results. The thickness of the adhesive layer is determined by rheological property analysis to ensure that the adhesive material does not experience flow deformation or bubble defects during the high-temperature curing process.

7. The method for online monitoring of PCB copper coating thickness based on surface acoustic wave resonance according to claim 6, characterized in that: The working method of the non-contact spacing calibration device described in step 2 also includes: applying ultrasonic vibration excitation during the curing process of the adhesive layer, judging the degree of curing in real time by monitoring the change in acoustic impedance, and dynamically adjusting the feedback control parameters of the displacement platform according to the curing state; at the same time, adding a nanoparticle reinforcement phase to the adhesive material, and the particle type and volume fraction are selected according to the high-temperature mechanical performance requirements of the adhesive layer to improve the thermal conductivity of the adhesive layer and inhibit high-temperature creep. The particle size distribution of the nanoparticles is controlled by a centrifugal classification process to ensure their uniform dispersion in the adhesive matrix.

Images