Modeling and optimizing method for inter-layer signal crosstalk of multi-layer PCB (Printed Circuit Board)

Abstract

The invention relates to a modeling and optimizing method for inter-layer signal crosstalk of a PCB (Printed Circuit Board). Frequency response data of all dielectric layers of the PCB under different temperature and humidity conditions are collected, and a mapping model with the temperature gradient and the humidity diffusion gradient as input and the dielectric constant as output is established. And calculating the dynamic distribution of mutual capacitance and mutual inductance under the environment change by combining the spatial layout parameters of the differential signal pair, and constructing a crosstalk sensitivity evaluation model with coupled frequency, environment and layout. And high-risk areas with crosstalk sensitivity caused by environment change are identified, and the crosstalk change rate of each area is quantified. For a sensitive area, a temperature and humidity compensation medium structure is introduced to stabilize local dielectric characteristics and reduce the influence of the environment on crosstalk. And according to the compensated medium distribution, adjusting a differential signal wiring path and impedance matching by adopting an optimization method with the aim of minimizing the crosstalk change rate, and generating a wiring layout with environmental robustness. According to the invention, the signal integrity and the transmission reliability of the multi-layer PCB in a temperature and humidity changing environment can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of PCB circuit boards, and in particular to a method for modeling and optimizing signal crosstalk between layers of a multi-layer PCB. Background Art

[0002] Multilayer printed circuit boards (PCBs) are widely used in high-speed signal transmission. To meet the demands of high-frequency communications, data centers, and advanced computing equipment, differential signal pair layouts are often used to enhance interference immunity. To mitigate crosstalk during high-speed signal transmission, existing methods primarily focus on adjusting trace spacing, controlling impedance, optimizing reference layer structures, and using electromagnetic simulation tools to model and predict crosstalk levels under fixed environmental conditions. These technical solutions effectively ensure signal integrity under standard temperature and humidity conditions.

[0003] However, with the diversification of operating environments, PCBs are frequently exposed to significant temperature and humidity fluctuations in real-world applications. This can cause the dielectric properties of the board's dielectric layer to dynamically change, leading to fluctuations in mutual capacitance and inductance, and consequently, unstable crosstalk levels between differential signal pairs. Existing crosstalk modeling and optimization methods generally fail to fully consider the dynamic characteristics of dielectric constant changes with temperature and humidity, and lack the means to predict and mitigate the risk of crosstalk degradation under environmental changes. This can easily lead to degradation or even failure of signal transmission performance under extreme environmental conditions.

[0004] Therefore, it is necessary to provide a new modeling and optimization method for signal crosstalk between layers of multi-layer PCB boards. Summary of the Invention

[0005] The present application provides a method for modeling and optimizing signal crosstalk between layers of a multi-layer PCB to improve the signal integrity and transmission reliability of the multi-layer PCB in an environment with varying temperature and humidity.

[0006] This application provides a method for modeling and optimizing signal crosstalk between layers of a multi-layer PCB, including:

[0007] Collect frequency response data of each dielectric layer of the PCB board under different temperature and humidity conditions, and establish a dynamic mapping model with temperature gradient and humidity diffusion gradient as independent variables and dielectric constant as dependent variable;

[0008] Based on the dynamic mapping model and the spatial layout parameters of each differential signal pair within a multi-layer PCB, the mutual capacitance and mutual inductance distributions under different temperature and humidity environments are calculated, and a crosstalk sensitivity assessment model is constructed that couples frequency, environmental conditions, and layout structure.

[0009] The crosstalk sensitivity evaluation model is used to identify highly sensitive areas where mutual capacitance and mutual inductance are prone to drastic fluctuations under temperature and humidity changes, and to quantify the crosstalk change rate in each area.

[0010] Based on the location of highly sensitive areas and the crosstalk variation characteristics, a temperature and humidity compensation dielectric structure is locally introduced to regulate the variation trend of the local dielectric constant and reduce the amplification effect of environmental changes on the crosstalk characteristics;

[0011] After the compensation structure is arranged, based on the updated dielectric distribution, an iterative simulation optimization method is adopted with minimizing the crosstalk change rate in highly sensitive areas as the objective function and with differential pair length matching error and transmission delay deviation as constraints. The differential signal routing path and impedance continuity are dynamically adjusted to generate a wiring layout with optimized environmental robustness.

[0012] Furthermore, the frequency response data of each dielectric layer of the PCB board under different temperature and humidity conditions are collected to establish a dynamic mapping model with temperature gradient and humidity diffusion gradient as independent variables and dielectric constant as dependent variable, including:

[0013] Collect frequency response data of each dielectric layer of the PCB board at different frequency points under multiple set temperature and humidity change paths;

[0014] Based on the collected data, the data under different working conditions are classified according to the temperature change rate and humidity diffusion rate, and weighted average processing is performed within the classified data group to generate the dielectric constant change trend of the corresponding frequency point;

[0015] Based on the generated dielectric constant variation trend, the dielectric constant value under the target environmental conditions is estimated by performing linear interpolation calculations between adjacent classification groups for any set temperature gradient and humidity diffusion gradient.

[0016] Furthermore, based on the dynamic mapping model, combined with the spatial layout parameters of each differential signal pair in the multi-layer PCB, the mutual capacitance and mutual inductance distribution under different temperature and humidity environments are calculated, and a crosstalk sensitivity evaluation model coupled with frequency, environmental conditions and layout structure is constructed, including:

[0017] The line width, line spacing, trace height, trace direction, and via layout parameters of each differential signal pair in a multi-layer PCB are used as the first input, the dielectric constant of each dielectric layer under the target temperature and humidity environment conditions is used as the second input, and the target frequency point is used as the third input;

[0018] Based on the first input, the second input, and the third input, the mutual capacitance change rate and the mutual inductance change rate of each pair of differential signal lines under the target temperature, humidity, and frequency conditions are calculated by table lookup reasoning combined with a local linear weighted method.

[0019] Based on the calculated mutual capacitance change rate and mutual inductance change rate, the corresponding near-end crosstalk level change amount and far-end crosstalk level change amount are derived;

[0020] Based on the derived near-end crosstalk change and far-end crosstalk change, the crosstalk change rate exceeding risk level of each pair of differential signal lines is evaluated according to the preset crosstalk change rate threshold, and the mutual capacitance change rate, mutual inductance change rate, near-end crosstalk change, far-end crosstalk change and exceeding risk level are used as input basis for identifying highly sensitive areas where mutual capacitance and mutual inductance are prone to violent fluctuations under temperature and humidity changes through the crosstalk sensitivity assessment model and quantifying the crosstalk change rate of each area.

[0021] Furthermore, the derivation of the corresponding near-end crosstalk level change and far-end crosstalk level change based on the calculated mutual capacitance change rate and mutual inductance change rate includes:

[0022] For each pair of differential signal lines, the electric and magnetic field coupling trends between the differential pairs are calculated based on the mutual capacitance and mutual inductance change rates calculated under the target temperature, humidity, and frequency conditions. The near-end crosstalk level change is primarily derived based on the mutual capacitance change rate, while the far-end crosstalk level change is primarily derived based on the mutual inductance change rate.

[0023] During the derivation process, the near-end crosstalk trend is approximately predicted by performing a forward weighted integration on the mutual capacitance change rate, and the far-end crosstalk trend is approximately predicted by performing a reverse weighted integration on the mutual inductance change rate. The weighting coefficient is then corrected based on the dynamic change amplitude of the ambient dielectric constant and permeability at each frequency point to improve the derivation accuracy.

[0024] Based on the derivation results, local normalization processing is used to standardize the near-end crosstalk change and the far-end crosstalk change of each differential signal pair to the same sensitivity scale, which serves as the basic input basis for subsequently identifying highly sensitive areas and quantifying the crosstalk change rate of each area through the crosstalk sensitivity assessment model.

[0025] Furthermore, based on the location of the highly sensitive area and the crosstalk variation characteristics, a temperature and humidity compensation dielectric structure is locally introduced to regulate the variation trend of the local dielectric constant and reduce the amplification effect of environmental changes on the crosstalk characteristics, including:

[0026] For the highly sensitive areas identified by the crosstalk sensitivity assessment model, the scope of the area requiring local compensation is determined based on the mutual capacitance change rate, mutual inductance change rate, near-end crosstalk change amount, and far-end crosstalk change amount of each differential signal pair in the area, combined with the routing density and local thermal distribution characteristics;

[0027] Within the determined compensation area, select a temperature and humidity compensation material whose dielectric constant matches that of the original dielectric material within the target frequency range and whose moisture absorption characteristics are less affected by the environment. Based on the characteristics of the surrounding environment of the differential signal routing, choose to use an embedded inlay structure or a surface-attached structure for local compensation;

[0028] The local dielectric parameters after the introduction of the temperature and humidity compensation medium are updated to the overall dielectric layer distribution model. After the compensation structure is arranged, the updated dielectric distribution is used as the input basis for dynamically adjusting the differential signal routing path and impedance continuity using an iterative simulation optimization method with minimizing the crosstalk change rate in the highly sensitive area as the objective function and differential pair length matching error and transmission delay deviation as constraints.

[0029] Furthermore, after the compensation structure is arranged, the differential signal routing path and impedance continuity are dynamically adjusted based on the updated dielectric distribution using an iterative simulation optimization method with minimizing the crosstalk change rate in the highly sensitive area as the objective function and with the differential pair length matching error and transmission delay deviation as constraints, including:

[0030] Taking the maximum crosstalk change rate of each differential signal pair in the highly sensitive area identified by the crosstalk sensitivity evaluation model under different temperature and humidity conditions as the optimization target, constructing a main objective function with minimizing the maximum crosstalk change rate;

[0031] During the routing fine-tuning process, the length matching error, transmission delay deviation, and local impedance continuity of the differential signal pair are used as optimization constraints to ensure that the changes in various parameters in each round of routing adjustment cannot exceed the preset tolerance range.

[0032] A simulation iteration method based on local routing path fine-tuning is adopted. For the differential signal pairs in the identified highly sensitive areas, the routing spacing, routing height or path curvature are gradually and slightly adjusted. After each fine-tuning, local electromagnetic simulation verification is performed to evaluate whether the crosstalk change rate is reduced and whether all constraints are met at the same time, until the objective function converges or the iteration stop condition is reached.

[0033] The beneficial effects of the technical solution provided by this application include:

[0034] (1) By establishing a dynamic mapping model of temperature and humidity and introducing environmental parameters, it can accurately reflect the actual characteristics of the dielectric properties of PCB boards as they change with temperature and humidity, providing basic data that is more in line with the actual working environment for subsequent crosstalk evaluation and optimization, and improving modeling accuracy. (2) Through the crosstalk sensitivity evaluation model that couples frequency, environmental conditions and spatial layout, it can identify high-risk areas that are susceptible to environmental changes in multiple dimensions, thereby achieving targeted optimization and avoiding design failure problems caused by ignoring environmental factors in traditional methods. (3) By locally introducing a temperature and humidity compensation dielectric structure, it can effectively suppress the crosstalk degradation caused by dielectric constant fluctuations in highly sensitive areas, and improve the signal integrity and anti-interference stability of the PCB under different environmental conditions. (4) Through iterative simulation optimization with the goal of minimizing the crosstalk change rate, combined with path adjustment and impedance matching, it is possible to achieve a wiring layout that optimizes environmental robustness while ensuring electrical performance, significantly improving the reliability and long-term performance stability of the PCB in complex application scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a flowchart of a method for modeling and optimizing signal crosstalk between layers of a multi-layer PCB provided in the first embodiment of the present application. DETAILED DESCRIPTION

[0036] The following description sets forth many specific details to facilitate a thorough understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the scope of the present application. Therefore, the present application is not limited to the specific implementations disclosed below.

[0037] The first embodiment of the present application provides a method for modeling and optimizing signal crosstalk between layers of a multi-layer PCB. Figure 1 , which is a schematic diagram of the first embodiment of the present application. Figure 1 A first embodiment of the present application provides a method for modeling and optimizing signal crosstalk between layers of a multi-layer PCB board, which is described in detail.

[0038] Step S101: collecting frequency response data of each dielectric layer of the PCB board under different temperature and humidity conditions, and establishing a dynamic mapping model with temperature gradient and humidity diffusion gradient as independent variables and dielectric constant as dependent variable.

[0039] During the specific implementation process, it is first necessary to select representative PCB material samples for the target multi-layer PCB board. Usually, this should at least include materials from different suppliers, different batches, or different dielectric stack structures to cover the material differences that may occur within the design application range. Subsequently, the samples are placed in a series of preset temperature and humidity environments for frequency response testing. The temperature range is usually set from -40°C to +85°C to cover the common temperature range of industrial applications, and the humidity range is usually set from 10% RH to 95% RH to simulate extreme environments such as dryness and high humidity.

[0040] Under each set of temperature and humidity conditions, a vector network analyzer is used to perform S-parameter testing on each dielectric layer of the board. The test frequency range covers at least twice the target application frequency band. For example, for a 10GHz application, the test frequency should extend to above 20GHz to ensure sufficient high-frequency characteristic information is obtained. During testing, Kelvin fixtures, microstrip lines, or coplanar waveguide test structures can be used. Combined with precise de-embedding technology, the test fixture can be removed from the measurement results to ensure that data that truly reflects the electrical properties of the dielectric layer is obtained.

[0041] Based on the above frequency response test results, the curves of the effective dielectric constant (Dk) and dissipation factor (Df) changing with frequency under each temperature and humidity combination are extracted. Furthermore, the temperature gradient (i.e., the rate of temperature change under different temperature conditions) and the humidity diffusion gradient (i.e., the rate of water diffusion or saturation degree inside the material under different humidity conditions) are used as independent variables, and the value of the dielectric constant at each frequency point is used as the dependent variable to establish a mathematical mapping relationship. Usually, multiple regression analysis, support vector regression (SVR) or neural network fitting are used to model the data to obtain a continuous and differentiable dynamic mapping function. The mapping function should be able to quickly predict the dielectric constant value of the corresponding frequency point based on the input temperature gradient and humidity diffusion gradient under any temperature and humidity combination conditions.

[0042] During the modeling process, to improve prediction accuracy, it is recommended to normalize the original test data and use cross-validation to evaluate the generalization performance of the fitted model. Furthermore, based on actual application needs, a frequency weighting term can be introduced into the mapping model to prioritize fitting accuracy in high-frequency bands, thereby better meeting the design requirements of high-speed signal transmission. The resulting dynamic mapping model is saved as a continuous function or high-order interpolation table for subsequent modeling and calculation of mutual capacitance, mutual inductance, and crosstalk sensitivity.

[0043] Through the above steps, the impact of temperature and humidity changes on the dielectric properties of PCB boards is accurately captured, and a dynamic mapping foundation that can be directly called in subsequent analysis steps is established. It has good real-time and adaptability, and can support the needs of crosstalk characteristic analysis and optimized design under various working conditions.

[0044] Furthermore, the frequency response data of each dielectric layer of the PCB board under different temperature and humidity conditions are collected to establish a dynamic mapping model with temperature gradient and humidity diffusion gradient as independent variables and dielectric constant as dependent variable, including:

[0045] Collect frequency response data of each dielectric layer of the PCB board at different frequency points under multiple set temperature and humidity change paths;

[0046] Based on the collected data, the data under different working conditions are classified according to the temperature change rate and humidity diffusion rate, and weighted average processing is performed within the classified data group to generate the dielectric constant change trend of the corresponding frequency point;

[0047] Based on the generated dielectric constant variation trend, for any set temperature gradient and humidity diffusion gradient, the dielectric constant value under the target environmental conditions is inferred by performing linear interpolation calculations between adjacent classification groups. The inferred dielectric constant is used as the input basis for the step of calculating the mutual capacitance and mutual inductance distribution under different temperature and humidity environments in combination with the spatial layout parameters of each differential signal pair in the multi-layer PCB.

[0048] During implementation, it's first necessary to set up multiple different temperature and humidity variation paths within the experimental environment to fully cover the temperature and humidity variations that the PCB board may encounter in actual applications. These temperature and humidity variation paths can include paths where the humidity remains constant during heating, paths where the temperature remains constant during humidity increases, and paths where the temperature and humidity change simultaneously but at different rates. Specifically, for example, a steady increase from 25°C 50% RH to 85°C 90% RH, or a rapid change from 30% RH to 80% RH under constant temperature conditions. For each set variation path, preset frequency points are selected, such as 1 GHz, 5 GHz, 10 GHz, and 20 GHz. At these frequency points, precision testing equipment such as a vector network analyzer is used to collect the frequency response characteristics of each dielectric material layer in the PCB board to obtain dielectric performance data under the corresponding temperature and humidity conditions. The frequency response data must include at least the effective dielectric constant and dissipation factor to ensure the accuracy of subsequent calculations.

[0049] After completing data collection, the large amount of raw data collected needs to be sorted. During the sorting process, the data is classified and grouped according to the temperature change rate and humidity diffusion rate corresponding to each operating point in the actual collected data. The temperature change rate can be obtained by dividing the temperature difference between adjacent time points by the time interval, and the humidity diffusion rate can be calculated based on the humidity change amplitude and the hygroscopic characteristics of the material. When classifying, data with similar temperature and humidity change rates are grouped together to reflect the regularity of the material's electrical properties under similar environmental dynamic change conditions. After classification is completed, within each classified data group, the data of different measurement samples at the same frequency point are weighted averaged. The weighting rules can be set according to measurement accuracy, data coverage density, or experimental confidence to ensure that the weighted average result can truly reflect the dielectric constant change trend under this rate category.

[0050] Through this weighted processing, we can generate dielectric constant trend curves corresponding to different temperature and humidity change rate categories at each frequency point. These curves reflect how the dielectric properties of the PCB material's dielectric layer change with temperature and humidity gradients during actual environmental changes, providing a basis for subsequent calculations.

[0051] After the trend is generated, for any new temperature gradient and humidity diffusion gradient condition that is not at the center of an existing classification group, linear interpolation can be performed between two or more adjacent classification groups. During interpolation, the dielectric constant change trends of the corresponding frequency points of adjacent classification groups are weighted and superimposed, using the temperature and humidity change rate as the basis for weighting, to infer the predicted dielectric constant value under the target environmental conditions.

[0052] The estimated dielectric constant values at each frequency point under the target environmental conditions serve as the input for subsequent calculations of mutual capacitance and mutual inductance distribution under varying temperature and humidity conditions, combined with the spatial layout parameters of each differential signal pair within a multilayer PCB. This continuous and dynamically responsive estimation system accurately reflects the impact of changing environmental conditions on dielectric and coupling properties in real-world applications, providing high-precision data support for identifying highly sensitive areas and optimizing routing.

[0053] When performing subsequent calculations of mutual capacitance and mutual inductance distribution, the inferred dielectric constant values at each frequency point under the target environmental conditions are combined with the spatial layout parameters of each differential signal pair in the multi-layer PCB. Specifically, for each pair of differential signal lines, based on the distribution of their line width, line spacing, trace height, trace layer, and surrounding reference planes, an electric field and magnetic field analytical model or a table lookup reasoning method based on existing simulation results is used to calculate the mutual capacitance and mutual inductance values between the differential signal pairs under the current temperature and humidity environmental conditions and target frequency. The inferred dielectric constant is input into the layout parameter solution process as a key material parameter, which directly affects the evaluation of the electric field distribution and magnetic field coupling strength, thereby ensuring that the changing trends of mutual capacitance and mutual inductance can be accurately captured under different temperature and humidity environmental conditions. In this way, synchronous modeling of environmental dynamics, material electrical property evolution, and signal trace spatial relationships can be achieved, providing a complete and precision-controlled basic electromagnetic parameter input for subsequent crosstalk sensitivity evaluation.

[0054] Step S102: Based on the dynamic mapping model and in combination with the spatial layout parameters of each differential signal pair in the multi-layer PCB, the mutual capacitance and mutual inductance distribution under different temperature and humidity environments are calculated to construct a crosstalk sensitivity evaluation model coupled with frequency, environmental conditions and layout structure.

[0055] After completing the acquisition of the dielectric layer frequency response data under temperature and humidity conditions and establishing a dynamic mapping model, it is necessary to further combine the actual spatial layout parameters of each differential signal pair within the multi-layer PCB based on this dynamic mapping model to calculate the mutual capacitance and mutual inductance distribution under different temperature and humidity environments, thereby providing accurate input for subsequent crosstalk sensitivity assessment.

[0056] First, key layout parameters for each pair of differential signal lines should be extracted from the PCB design data to be analyzed. These parameters include, but are not limited to, the centerline spacing between the two signal lines of the differential pair, their respective line widths, the vertical distance between the traces and the reference plane, the PCB layer number on which the traces are located, the trace bend radius, the via location and structure, the length distribution of the signal line segments, and the horizontal and vertical spacing between adjacent differential pairs. For signals traversing multiple layers of dielectric material, cross-layer trace information should also be extracted to ensure comprehensive modeling of interlayer coupling effects.

[0057] After extraction, the aforementioned dynamic mapping model should be invoked according to the specific geometric characteristics of the differential traces under each temperature and humidity combination to query the dielectric constant parameters of the corresponding dielectric layer. Based on this, a high-fidelity finite element model of the local structure of the multilayer PCB is constructed. When performing finite element modeling, it is recommended to use electromagnetic simulation tools (such as Ansys HFSS) for modeling and simulation, using adaptive meshing technology suitable for high-frequency, small-scale features to ensure accurate capture of electric and magnetic field changes at the boundaries.

[0058] Based on the aforementioned finite element model, the mutual capacitance and mutual inductance distribution data between each differential signal pair under unit excitation are calculated for each temperature and humidity environment. Mutual capacitance can be calculated by applying a fixed potential difference, analyzing the electrostatic field energy density distribution, and then integrating it. Mutual inductance can be derived by applying a unit current excitation and calculating the resulting magnetic field induced voltage. To improve solution efficiency, local subdomain modeling can be established for each differential pair and its adjacent traces, and symmetric boundary conditions can be used to reduce simulation computational complexity.

[0059] Furthermore, to accurately capture the impact of temperature and humidity on mutual capacitance and inductance, simulations must consider the frequency dependence of dielectric parameters as they change with temperature and humidity. Frequency-variable dielectric constants and loss factors should be introduced when modeling material properties and solved separately at different frequency sampling points. It is generally recommended to take at least 10 to 20 evenly distributed frequency sampling points within the design application frequency band and twice its range to ensure the continuity and accuracy of the frequency response curve.

[0060] The mutual capacitance and inductance data obtained at different sampling frequencies under various temperature and humidity conditions are organized into a three-dimensional matrix, with the three dimensions corresponding to the temperature and humidity conditions, the frequency points, and the differential pair combinations. This three-dimensional data matrix clearly reflects the changing trends in the electromagnetic coupling strength of each differential signal pair at a specific frequency under specific environmental changes.

[0061] Finally, based on the complete distribution data of mutual capacitance and mutual inductance, a crosstalk sensitivity assessment model is constructed that couples frequency, environmental conditions, and differential layout structure. This assessment model is typically built using statistical modeling or machine learning methods (such as principal component analysis (PCA), partial least squares regression (PLSR), and convolutional neural networks (CNN). The goal is to extract key variation characteristics from a large amount of mutual capacitance and mutual inductance data and establish an analysis engine that can predict the crosstalk sensitivity level after inputting layout parameters and environmental conditions. The crosstalk sensitivity level can be characterized by physical quantities such as near-end crosstalk (NEXT), far-end crosstalk (FEXT), and signal integrity indicators (such as eye opening), and compared and analyzed with the performance tolerance requirements under various environmental conditions.

[0062] To facilitate understanding and implementation, the following provides a specific example of a crosstalk sensitivity assessment model. When establishing a crosstalk sensitivity assessment model, it can be specifically divided into three functional modules.

[0063] The first is the input normalization module. The input of this module is PCB layout parameters, environmental condition parameters and frequency point data. Specifically, PCB layout parameters include the line width, spacing, height, direction, etc. of each pair of differential signal lines, environmental parameters include the current temperature and humidity conditions, and frequency parameters are several frequency values that need to be paid attention to in the transmission of the target signal. The function of the input normalization module is to process all input data into a unified scale, such as expressing all physical dimensions in millimeters, temperature in degrees Celsius, and humidity in percentage, and linearly stretching or compressing according to the overall distribution so that different physical quantities can be processed simultaneously by subsequent modules. The normalization method can be very simple, for example, by subtracting the overall mean from each input value and dividing it by the standard deviation to ensure that the data distribution is within a uniform range.

[0064] After normalization, the module enters the mutual capacitance and mutual inductance inference module. This module takes the normalized layout parameters, temperature and humidity conditions, and frequency points as input, and outputs the mutual capacitance change rate and mutual inductance change rate for each pair of differential signal lines under the corresponding environment and frequency. The mutual capacitance change rate reflects the change in the degree of electric field coupling under different environments, while the mutual inductance change rate reflects the change in the degree of magnetic field coupling. To simplify implementation, a simple lookup table can be established in the early stages of modeling based on existing simulation results or experimental measurement results. For example, the mutual capacitance and mutual inductance values corresponding to several typical layouts under different temperature and humidity environments can be pre-measured to form a small database. In actual use, based on the input layout characteristics and environmental conditions, the closest match is made in this database, or a simple linear interpolation approximation is performed to obtain the corresponding mutual capacitance change rate and mutual inductance change rate. This avoids complex on-site simulation during modeling and greatly reduces implementation difficulty.

[0065] After obtaining the mutual capacitance change rate and mutual inductance change rate, the data is fed into the crosstalk sensitivity prediction module. This module takes the mutual capacitance change rate and mutual inductance change rate as input and outputs a crosstalk sensitivity score for each pair of differential signal lines under different temperature and humidity conditions. To simplify modeling, a very basic multivariate linear regression method can be used. This method superimposes the mutual capacitance change rate and mutual inductance change rate with fixed weights to calculate a total change value, which serves as an intuitive score for crosstalk sensitivity. The weights can be set based on experience. For example, initially assuming that mutual capacitance changes have a greater impact on crosstalk, a higher weight can be assigned to them. In the absence of complex data training support, the weighted sum of the mutual capacitance change rate and the mutual inductance change rate can be used as the score to achieve rapid results.

[0066] Through the sequential processing of the above three modules, the complete process from PCB layout parameters, temperature and humidity environment conditions and frequency point input, to mutual capacitance and mutual inductance change rate calculation, and finally to crosstalk sensitivity score output can be realized.

[0067] Through the above steps, not only is high-precision modeling of mutual capacitance and mutual inductance achieved under multiple temperature, humidity, frequency, and layout conditions, but a crosstalk sensitivity analysis framework that can dynamically respond to environmental changes is also formed, providing a complete and reliable data foundation for subsequent sensitive area identification and optimization.

[0068] Furthermore, based on the dynamic mapping model, combined with the spatial layout parameters of each differential signal pair in the multi-layer PCB, the mutual capacitance and mutual inductance distribution under different temperature and humidity environments are calculated, and a crosstalk sensitivity evaluation model coupled with frequency, environmental conditions and layout structure is constructed, including:

[0069] The line width, line spacing, trace height, trace direction, and via layout parameters of each differential signal pair in a multi-layer PCB are used as the first input, the dielectric constant of each dielectric layer under the target temperature and humidity environment conditions is used as the second input, and the target frequency point is used as the third input;

[0070] Based on the first input, the second input, and the third input, the mutual capacitance change rate and the mutual inductance change rate of each pair of differential signal lines under the target temperature, humidity, and frequency conditions are calculated by table lookup reasoning combined with a local linear weighted method.

[0071] Based on the calculated mutual capacitance change rate and mutual inductance change rate, the corresponding near-end crosstalk level change amount and far-end crosstalk level change amount are further derived;

[0072] Based on the derived near-end crosstalk change and far-end crosstalk change, the crosstalk change rate exceeding risk level of each pair of differential signal lines is evaluated according to the preset crosstalk change rate threshold, and the mutual capacitance change rate, mutual inductance change rate, near-end crosstalk change, far-end crosstalk change and exceeding risk level are used as input basis for identifying highly sensitive areas where mutual capacitance and mutual inductance are prone to violent fluctuations under temperature and humidity changes through the crosstalk sensitivity assessment model and quantifying the crosstalk change rate of each area.

[0073] The implementation process begins with parameter extraction for all high-speed differential signal pairs within a multilayer PCB. For each differential signal pair, the line width, line spacing, trace height relative to the reference ground plane, trace direction, and trace length must be recorded. If cross-layer connections exist, the geometric layout and distribution of vias must also be recorded. These parameters together constitute the first type of input data for modeling, reflecting the geometric structural characteristics of the differential signal pairs within the PCB layers.

[0074] Then, for the selected temperature and humidity environment conditions, the dynamic mapping model established above, with temperature gradient and humidity diffusion gradient as independent variables and dielectric constant as the dependent variable, is used to determine the dielectric constant values corresponding to each PCB dielectric layer under the target temperature and humidity conditions. These dielectric constant values calculated by the dynamic mapping model constitute the second type of input data in the modeling process, accurately reflecting the impact of different temperature and humidity environments on the electrical properties of the board.

[0075] After determining the first and second types of input data, it is necessary to determine the target frequency points as the third type of input data based on the design requirements or actual application scenario. Generally, the frequency points selected should cover the design signal transmission frequency and its main harmonic components. For example, for a 10Gbps data channel, 5GHz, 10GHz, and 15GHz can be selected as representative frequency points to fully evaluate the coupling effects at high frequencies.

[0076] After the three types of input data are prepared, a table lookup inference method is used based on a pre-built standard lookup table to obtain basic mutual capacitance and mutual inductance data under different parameter combinations. Considering that the actual PCB layout parameters and environmental parameters may not fall completely on the existing data points, it is necessary to combine the local linear weighting method for estimation. Specifically, several neighboring samples closest to the target parameters in the parameter space are selected, and weighting coefficients are set according to the distance. The mutual capacitance and mutual inductance data of the neighboring samples are weighted averaged to obtain the estimated mutual capacitance change rate and mutual inductance change rate under the target conditions. When weighting, ensure that the weights are normalized, and appropriately increase the influence of samples closer to the target parameters to improve estimation accuracy.

[0077] After calculating the mutual capacitance and mutual inductance change rates, the near-end crosstalk and far-end crosstalk level changes for each pair of differential signal lines are further derived based on electromagnetic coupling theory under the target temperature, humidity, and frequency conditions. This derivation can be performed using table approximation, simple empirical formulas, or inductive estimation based on the degree of variation in mutual capacitance and mutual inductance and physical parameters such as signal path length, ensuring repeatability and engineering practicality.

[0078] After obtaining the near-end crosstalk (NEXT) and far-end crosstalk (FEXT) changes for each pair of differential signal lines, to further quantify the environmental sensitivity of the signal lines, each pair of differential signal lines is assessed for risk level exceeding the threshold based on a preset crosstalk change rate threshold. This assessment can be tiered, with a high risk rating defined, for example, when the crosstalk change rate exceeds a set baseline by 30%, a medium risk rating defined as between 15% and 30%, and a low risk rating defined as below 15%. Each pair of differential signal lines is assigned a corresponding risk level based on the assessment results.

[0079] Ultimately, the calculated mutual capacitance change rate, mutual inductance change rate, near-end crosstalk change, far-end crosstalk change, and corresponding risk level of over-limit will serve as input for the subsequent steps of identifying highly sensitive areas prone to dramatic fluctuations in mutual capacitance and mutual inductance under temperature and humidity changes using the crosstalk sensitivity assessment model and quantifying the crosstalk change rate in each area. This data will provide complete, accurate, and dynamically adaptable basic information for the precise location of highly sensitive areas, the design of local compensation structures, and the optimization of wiring layouts.

[0080] The calculated mutual capacitance change rate, mutual inductance change rate, near-end crosstalk change amount, far-end crosstalk change amount and over-limit risk level can be used as the input basis for the subsequent step of identifying highly sensitive areas where mutual capacitance and mutual inductance are prone to drastic fluctuations under temperature and humidity changes through the crosstalk sensitivity assessment model and quantifying the crosstalk change rate of each area. It can be applied in the following way. First, the various calculated results of each pair of differential signal lines correspond to the layout position of the PCB board according to the spatial position, and a crosstalk sensitivity distribution map is established, where the mutual capacitance change rate, mutual inductance change rate and crosstalk change amount can be used as sensitivity weight indicators, and the over-limit risk level is used as the judgment standard. For differential signal lines or signal areas with high sensitivity weight indicators and high over-limit risk levels, they are marked as highly sensitive areas in the crosstalk sensitivity assessment process, and further local aggregation analysis is performed to statistically analyze the concentration and change amplitude of the crosstalk change level in the area, thereby completing the accurate identification of highly sensitive areas. At the same time, quantifying the crosstalk change rate in each region allows us to form a regional crosstalk change rate indicator by taking a weighted average of the change rates of signal pairs marked as highly sensitive and their neighboring signal pairs, which serves as a priority for subsequent optimization design. This process ensures that the crosstalk sensitivity assessment model can quickly and accurately locate areas of potential reliability risks when dealing with complex temperature and humidity fluctuations, and provides accurate basic data support for subsequent local optimization and global layout adjustments.

[0081] Furthermore, the derivation of the corresponding near-end crosstalk level change and far-end crosstalk level change based on the calculated mutual capacitance change rate and mutual inductance change rate includes:

[0082] For each pair of differential signal lines, the electric and magnetic field coupling trends between the differential pairs are calculated based on the mutual capacitance and mutual inductance change rates calculated under the target temperature, humidity, and frequency conditions. The near-end crosstalk level change is primarily derived based on the mutual capacitance change rate, while the far-end crosstalk level change is primarily derived based on the mutual inductance change rate.

[0083] During the derivation process, the near-end crosstalk trend is approximately predicted by performing a forward weighted integration on the mutual capacitance change rate, and the far-end crosstalk trend is approximately predicted by performing a reverse weighted integration on the mutual inductance change rate. The weighting coefficient is then corrected based on the dynamic change amplitude of the ambient dielectric constant and permeability at each frequency point to improve the derivation accuracy.

[0084] Based on the derivation results, local normalization processing is used to standardize the near-end crosstalk change and the far-end crosstalk change of each differential signal pair to the same sensitivity scale, which serves as the basic input basis for subsequently identifying highly sensitive areas and quantifying the crosstalk change rate of each area through the crosstalk sensitivity assessment model.

[0085] During implementation, the mutual capacitance change rate and mutual inductance change rate data, calculated using a dynamic mapping model under target temperature and humidity conditions and specific frequency points, must be first used for each pair of differential signal lines within a multilayer PCB. These change rate data reflect the dynamic trends in the electric field coupling strength and magnetic field coupling strength under different environmental conditions. Based on this input data, the response characteristics of the electric and magnetic fields are modeled separately, with the electric field coupling change trend primarily derived from the mutual capacitance change rate, and the magnetic field coupling change trend primarily derived from the mutual inductance change rate. Since near-end crosstalk is typically dominated by direct electric field coupling, the mutual capacitance change rate should be used as the primary reference when deriving changes in near-end crosstalk levels. However, far-end crosstalk is typically dominated by inductive magnetic field coupling, so the mutual inductance change rate should be used as the primary reference when deriving changes in far-end crosstalk levels.

[0086] During the derivation process, in order to more accurately reflect the impact of dynamic changes in temperature and humidity environment on the crosstalk level, it is necessary to perform weighted integration processing on the mutual capacitance change rate and the mutual inductance change rate respectively. Specifically, for the mutual capacitance change rate, by performing forward weighted integration along the environmental change path, the contribution of the electric field coupling change under different environmental conditions to the near-end crosstalk level is accumulated; for the mutual inductance change rate, by performing reverse weighted integration, the impact of the magnetic field coupling change under different environmental conditions on the far-end crosstalk level is accumulated. The setting of the weighting coefficient not only needs to consider the change amplitude of mutual capacitance and mutual inductance at different frequency points, but also should be corrected in combination with the dynamic response characteristics of dielectric constant and magnetic permeability during environmental changes to avoid the reduction of derivation accuracy due to nonlinear changes in materials. When weighting, data points with larger change amplitudes within the frequency range with the most significant crosstalk sensitivity are given higher weights to highlight the environmental effects in areas with high actual signal integrity risks.

[0087] After completing the weighted integral derivation, in order to make the near-end crosstalk change and far-end crosstalk change of different differential signal pairs directly comparable, local normalization processing is required. The normalization processing can be based on the maximum crosstalk change derived from each differential pair within the target temperature and humidity environment change range or the set standard change for standardized conversion, so that the crosstalk sensitivity characteristics of each pair of differential signal lines are finally uniformly mapped to the same sensitivity scale. The normalized near-end crosstalk change and far-end crosstalk change can not only intuitively reflect the relative susceptibility of each signal pair under environmental changes, but also serve as the basic input basis for subsequent identification of highly sensitive areas through the crosstalk sensitivity assessment model and quantification of the crosstalk change rate of each area, providing accurate quantitative support for the introduction of local temperature and humidity compensation media and wiring optimization adjustment.

[0088] Furthermore, the crosstalk sensitivity assessment model coupled with the construction frequency, environmental conditions and layout structure further includes:

[0089] Based on the calculated mutual capacitance change rate ΔC(f, T, H) and mutual inductance change rate ΔL(f, T, H), the comprehensive crosstalk sensitivity index S(f, T, H) of the differential signal pair under the target temperature and humidity environment change conditions is calculated. The comprehensive crosstalk sensitivity index S(f, T, H) is defined by the following formula:

[0090] S(f,T,H)=α·(ΔC(f,T,H)) 2 +β·(ΔL(f,T,H)) 2 +γ

[0091] (ΔC(f,T,H) ΔL(f,T,H))

[0092] Wherein, ΔC(f, T, H) represents the mutual capacitance change rate of the differential signal pair under the conditions of frequency f, temperature T, and humidity H; ΔL(f, T, H) represents the mutual inductance change rate of the differential signal pair under the conditions of frequency f, temperature T, and humidity H; α is the mutual capacitance change rate weighting factor, which is used to adjust the contribution ratio of ΔC(f, T, H) to the overall sensitivity; β is the mutual inductance change rate weighting factor, which is used to adjust the contribution ratio of ΔL(f, T, H) to the overall sensitivity; γ is the interactive coupling term weight of the mutual capacitance change rate and the mutual inductance change rate, which is used to characterize the degree of the combined impact of the joint change of the mutual capacitance change rate and the mutual inductance change rate on the crosstalk sensitivity.

[0093] The values of α, β, and γ are adaptively set according to the signal frequency range and environmental fluctuation characteristics of the target application, satisfying the normalization constraint of α+β+γ=1;

[0094] The comprehensive crosstalk sensitivity index S(f, T, H) serves as an input basis for identifying highly sensitive areas and quantifying the crosstalk change rate of each area through the crosstalk sensitivity evaluation model.

[0095] In the specific implementation of this invention, when constructing a crosstalk sensitivity assessment model that couples frequency, environmental conditions, and layout structure, we further introduced a comprehensive crosstalk sensitivity index, S(f,T,H), derived from the rate of change of mutual capacitance and mutual inductance. This index is designed to more accurately characterize the susceptibility of differential signal pairs to crosstalk caused by changes in coupling effects under varying temperature and humidity conditions, providing a more quantitative and granular basis for the subsequent identification of highly sensitive areas.

[0096] In the specific calculation process, first, through the aforementioned steps of the present invention, the mutual capacitance change rate ΔC(f, T, H) and mutual inductance change rate ΔL(f, T, H) of each pair of differential signal lines under the conditions of target frequency f, temperature T, and humidity H are respectively calculated. Among them, ΔC(f, T, H) represents the percentage of the mutual capacitance change amplitude after the change compared with the reference environment (such as standard temperature and humidity conditions) under given environmental conditions, which is usually obtained by collecting frequency response data at different environmental points and analyzing the electric field distribution based on modeling. ΔL(f, T, H) represents the percentage of the mutual inductance change amplitude after the change compared with the reference environment under the same environmental conditions, which is usually calculated through magnetic field distribution simulation, equivalent inductance extraction or environmental disturbance testing.

[0097] After obtaining ΔC(f, T, H) and ΔL(f, T, H), this embodiment defines the crosstalk sensitivity index S(f, T, H) using the following formula:

[0098] S(f,T,H)=α·(ΔC(f,T,H)) 2 +β·(ΔL(f,T,H)) 2 +γ

[0099] (ΔC(f,T,H) ΔL(f,T,H))

[0100] In the above formula, α is the weighting factor for the mutual capacitance change rate, which is used to adjust the contribution of ΔC(f, T, H) to the overall crosstalk sensitivity. A value of α is generally recommended to be between 0.3 and 0.5, preferably for high-speed differential links with significant electric field coupling effects. β is the weighting factor for the mutual inductance change rate, which is used to adjust the contribution of ΔL(f, T, H) to the overall crosstalk sensitivity. A value of β is generally recommended to be between 0.3 and 0.5, and is suitable for high-speed applications where the impact of magnetic field induction on crosstalk is significant. γ is the weighting factor for the interaction between the mutual capacitance change rate and the mutual inductance change rate, which is used to characterize the modulation effect of the synergistic or antagonistic effects of electric and magnetic field changes on the overall crosstalk level under actual environmental disturbances. The value of γ is generally recommended to be between 0.1 and 0.3, and can be dynamically adjusted based on simulation statistics or experimental results.

[0101] In order to ensure that the comprehensive sensitivity index is comparable under different conditions, α, β, and γ must satisfy the following normalization constraints:

[0102] α+β+γ=1

[0103] This normalization requirement ensures that the comprehensive crosstalk sensitivity index S(f, T, H) remains within a unified dimensional system to avoid the problem that the sensitivity evaluation results of each differential signal pair cannot be directly compared due to different weight settings.

[0104] The comprehensive crosstalk sensitivity indicator S(f,T,H) proposed in this embodiment is essentially a quadratic weighted expression constructed with the near-end crosstalk change trend caused by changes in mutual capacitance as the primary contribution, the far-end crosstalk change trend caused by changes in mutual inductance as the secondary contribution, and the interactive effects of electromagnetic changes taken into account. The square term can highlight the impact strength of a single source of change; the interaction term can reflect the modulation effect of the synergistic effect of changes in mutual capacitance and mutual inductance on the total crosstalk susceptibility. When the mutual capacitance and mutual inductance changes are amplified in the same direction, the interaction term will further exacerbate the crosstalk sensitivity; when the mutual capacitance and mutual inductance change in opposite directions, the interaction term can moderately offset each other, thereby more accurately reflecting the crosstalk fluctuation risk in real signal transmission.

[0105] The technical effects of introducing the comprehensive crosstalk sensitivity index S(f,T,H) are mainly reflected in the following two aspects: First, compared with the analysis of mutual capacitance changes or mutual inductance changes alone, this index can comprehensively cover the combined impact of electric and magnetic field changes on crosstalk behavior, thereby improving the accuracy and completeness of identifying highly sensitive areas; Second, by introducing dynamic weighting and normalization mechanisms, the present invention can flexibly and adaptively adjust the sensitivity calculation model according to the specific frequency characteristics and degree of environmental changes, significantly enhancing the robustness and engineering practicality of crosstalk sensitivity assessment under dynamic changes in multiple frequencies and environments.

[0106] Therefore, by adopting the above-mentioned comprehensive crosstalk sensitivity index S(f,T,H), this embodiment can effectively identify the areas inside the multi-layer PCB that are most susceptible to crosstalk under complex temperature and humidity conditions, guide subsequent local dielectric compensation and differential signal routing optimization, and ultimately improve high-speed signal integrity and environmental robustness.

[0107] Step S103: using the crosstalk sensitivity evaluation model, identifying highly sensitive areas where mutual capacitance and mutual inductance are prone to drastic fluctuations under temperature and humidity changes, and quantifying the crosstalk change rate of each area.

[0108] After establishing a crosstalk sensitivity assessment model that couples frequency, environmental conditions, and layout structure, it is necessary to automatically identify highly sensitive areas where mutual capacitance and mutual inductance are prone to drastic fluctuations under temperature and humidity changes based on the model, and quantify the crosstalk change rate in these areas. To achieve this goal, the current layout structure of the multi-layer PCB, the corresponding temperature and humidity environment operating parameters, and the frequency range are first input into the crosstalk sensitivity assessment model. The model is based on pre-established simulation data or test data, and through table lookup reasoning or simple interpolation calculation, it can quickly deduce the mutual capacitance change rate, mutual inductance change rate, and corresponding crosstalk level change index of each pair of differential signal lines under different working conditions based on the input layout characteristics, environmental parameters, and frequency information.

[0109] The crosstalk sensitivity assessment model is trained based on finite element simulation data and incorporates electromagnetic behavior characteristics under multiple temperature, humidity, and frequency conditions. After inputting the current routing information, it can quickly calculate the crosstalk sensitivity score for each differential signal pair or area under temperature and humidity variations. This score typically considers the change rate of mutual capacitance, mutual inductance, near-end crosstalk (NEXT), far-end crosstalk (FEXT), and signal integrity indicators (such as eye opening height change), and summarizes them according to the set weights.

[0110] Based on the sensitivity score output by the model, the PCB layout is divided into several grid regions, each with a corresponding sensitivity score. By setting a sensitivity score threshold—for example, by taking the mean of all scores plus twice the standard deviation—the demarcation line is determined, and regions with scores above the threshold are identified as highly sensitive to environmental changes. For each highly sensitive region, the key influencing factors are further extracted, including the maximum mutual capacitance change rate, the maximum mutual inductance change rate, and their corresponding temperature, humidity, and frequency conditions, to provide a basis for subsequent optimization.

[0111] When quantifying the crosstalk change rate in highly sensitive areas, the model directly outputs the mutual capacitance and mutual inductance change rates, eliminating manual recalculation and ensuring consistency and repeatability in the evaluation process. For multiple operating condition results output by the model, the maximum change rate can be selected as the quantified value, or the average change within a specific temperature and humidity window can be used for evaluation, as required by design specifications.

[0112] In this way, not only is automatic and rapid identification of highly crosstalk-sensitive areas achieved, but it is also ensured that the identification basis always comes from a uniformly constructed crosstalk sensitivity assessment model, thus avoiding errors and inconsistencies caused by manual data processing. It ensures that subsequent compensation and optimization measures can be applied in a targeted manner to locations that are truly susceptible to environmental changes, thereby improving the signal integrity and transmission reliability of multi-layer PCBs in complex environments.

[0113] Step S104: Based on the location of the highly sensitive area and the crosstalk variation characteristics, a temperature and humidity compensation dielectric structure is locally introduced to regulate the variation trend of the local dielectric constant and reduce the amplification effect of environmental changes on the crosstalk characteristics.

[0114] After identifying highly sensitive areas where mutual capacitance and mutual inductance are prone to dramatic fluctuations under temperature and humidity changes through the crosstalk sensitivity assessment model and quantifying their crosstalk change rate, it is necessary to design and locally introduce temperature and humidity compensation dielectric structures based on the location and specific crosstalk change characteristics of these highly sensitive areas to achieve regulation of the local dielectric constant change trend, thereby reducing the amplification effect of environmental changes on the crosstalk characteristics.

[0115] During implementation, each identified high-sensitivity area must first be analyzed in detail to determine the routing layout of the differential signal pairs within the area, the stack-up structure, the type of adjacent reference planes, the parameters of the original dielectric material, and the dielectric constant variation curve under different temperature and humidity conditions. Based on the rate of change data provided by the crosstalk sensitivity assessment model, it is determined whether the sensitivity is dominated by mutual capacitance or mutual inductance, thereby determining the focus of the compensation design. If the primary cause is mutual capacitance, the compensation design should prioritize suppressing changes in the electric field distribution; if the primary cause is mutual inductance, the focus should be on stabilizing the magnetic field distribution.

[0116] After selecting the compensation strategy, the local temperature and humidity compensation dielectric structure is designed into an embedded or attached structure that fits the signal line direction based on the specific location of the highly sensitive area. Common methods include locally embedding a dielectric material with low humidity sensitivity in the sensitive area of the target layer, or adding a layer of insulating sheet with higher humidity stability above or below the sensitive trace. For example, you can choose a modified polytetrafluoroethylene (PTFE) material with a dielectric constant close to that of the original material but a moisture absorption rate of less than 0.1%, a low moisture absorption ceramic filler material, or a polymer film material with a low moisture diffusivity. When selecting materials, priority should be given to the material's low dielectric loss within the target frequency range and good process compatibility to facilitate actual manufacturing.

[0117] The shape and size of the compensation structure should be flexibly determined based on the size of the sensitive area. Generally, a safety margin can be extended in all directions, for example, by 1 to 2 mm, in accordance with PCB design rules, to ensure the complete coverage of the compensation effect. The thickness of the compensation structure should be determined to maximize the improvement of local dielectric stability without compromising the original stack thickness control and impedance continuity. If the compensation material is applied as a thin film, a local film lamination process can be used. If an embedded structure is used, a slot can be reserved in the prefabricated stack and filled with a patch, which can then be secured using a local flattening and lamination process.

[0118] After introducing the compensation structure, it is necessary to re-evaluate the variation curve of the local dielectric constant under various temperature and humidity conditions to ensure that the variation of the local dielectric constant is less than a certain percentage of the original uncompensated state, for example, the variation range is reduced by more than 30%, to confirm the effectiveness of the compensation measure. The evaluation method can simply update the dielectric parameters and re-call the original dynamic mapping model for prediction, or use a simplified simulation model for local verification.

[0119] When arranging local compensation structures, it's also important to maintain the overall impedance continuity of the signal traces. For differential pair traces, the dielectric constant variation of the compensation medium should be as symmetrical as possible around the two traces to avoid introducing new interference from differential-mode to common-mode conversion. If necessary, the spacing or trace width of the differential pairs can be fine-tuned within the compensation area to accommodate the impact of local environmental variations on impedance and ensure that the overall differential impedance meets design requirements.

[0120] Through the above method, under the premise of ensuring processing feasibility and electrical performance requirements, the local temperature and humidity compensation structure can be effectively utilized to suppress the dielectric constant fluctuations caused by environmental changes, reduce the crosstalk change rate in highly sensitive areas, and lay the foundation for subsequent global optimization of wiring layout and further improvement of PCB signal integrity.

[0121] Furthermore, based on the location of the highly sensitive area and the crosstalk variation characteristics, a temperature and humidity compensation dielectric structure is locally introduced to regulate the variation trend of the local dielectric constant and reduce the amplification effect of environmental changes on the crosstalk characteristics, including:

[0122] For the highly sensitive areas identified by the crosstalk sensitivity assessment model, the scope of the area requiring local compensation is determined based on the mutual capacitance change rate, mutual inductance change rate, near-end crosstalk change amount, and far-end crosstalk change amount of each differential signal pair in the area, combined with the routing density and local thermal distribution characteristics;

[0123] Within the determined compensation area, select a temperature and humidity compensation material whose dielectric constant matches that of the original dielectric material within the target frequency range and whose moisture absorption characteristics are less affected by the environment. Based on the characteristics of the surrounding environment of the differential signal routing, choose to use an embedded inlay structure or a surface-attached structure for local compensation;

[0124] The local dielectric parameters after the introduction of the temperature and humidity compensation medium are updated to the overall dielectric layer distribution model. After the compensation structure is arranged, the updated dielectric distribution is used as the input basis for dynamically adjusting the differential signal routing path and impedance continuity using an iterative simulation optimization method with minimizing the crosstalk change rate in the highly sensitive area as the objective function and differential pair length matching error and transmission delay deviation as constraints.

[0125] During implementation, the crosstalk sensitivity assessment model is first used to identify highly sensitive areas where mutual capacitance and mutual inductance are prone to dramatic fluctuations under temperature and humidity conditions. For each identified highly sensitive area, detailed electromagnetic performance data is collected for each differential signal pair within the area, including the rate of change of mutual capacitance, mutual inductance, near-end crosstalk, and far-end crosstalk. This data not only reflects the changing trends in the electromagnetic coupling characteristics between signal lines but also reveals the risk of signal integrity degradation within the area due to temperature and humidity changes.

[0126] After obtaining the aforementioned electromagnetic performance change data, a comprehensive analysis should be conducted in conjunction with the trace density of the differential signal pairs within the area and the local thermal distribution characteristics. Trace density can be determined by counting the number of differential signal lines or the average line spacing per unit area. The greater the density, the more significant the electromagnetic interference and coupling effects. Local thermal distribution characteristics can be determined based on PCB layer power consumption distribution or thermal simulation data. Areas with large temperature gradients are more likely to cause material performance changes. Therefore, by comprehensively considering the mutual capacitance change rate, mutual inductance change rate, crosstalk change, trace density, and thermal distribution, the specific areas where local temperature and humidity compensation is most urgently needed can be determined. When determining the compensation area, it should ensure that all signal paths with significant crosstalk changes and sensitive environmental responses are covered, and appropriately expand to surrounding areas to prevent local failures caused by boundary effects.

[0127] After identifying the area requiring local compensation, targeted dielectric materials for temperature and humidity compensation need to be selected. Material selection should prioritize dielectric constants close to those of the original dielectric material within the target frequency range to maintain impedance continuity across the differential pair. Furthermore, the material should exhibit low hygroscopicity to minimize the impact of changes in ambient humidity on its dielectric properties. It should also exhibit good thermal stability to minimize fluctuations in its properties with temperature fluctuations. The appropriate compensation structure should be selected based on the specific layout of the differential signal traces within the compensation area and the environmental characteristics. If the trace area is relatively regular and interlayer space permits, an embedded insert structure can be employed. This involves pre-reserving compensation slots within the PCB stackup, filling them with compensation material, and then performing localized lamination. If the trace area has more flexible surface space or requires simplified processing, a surface-attached structure can be employed. This involves attaching the temperature and humidity compensation material directly to the surface above or below the target area as a film or coating to ensure dielectric stability within the signal path's surroundings.

[0128] After the temperature and humidity compensation structure is arranged, the local electrical parameters of the newly introduced compensation medium need to be updated to the overall dielectric layer distribution model. During the update, the dielectric constant value, thickness change, and transition boundary characteristics with the original board material of each compensation area should be accurately marked to ensure that the subsequent electromagnetic simulation calculations can accurately reflect the actual structure. The updated dielectric layer distribution model will serve as the input basis for the subsequent wiring optimization steps. It will be used for dynamic adjustment of differential signal routing paths and impedance continuity using an iterative simulation optimization method with minimizing the crosstalk change rate in highly sensitive areas as the objective function and differential pair length matching error and transmission delay deviation as constraints. Through the above steps, while ensuring the stability of the local environment, the overall signal integrity and system reliability of the multi-layer PCB under complex temperature and humidity changes can be further improved.

[0129] Step S105: After the compensation structure is arranged, based on the updated dielectric distribution, an iterative simulation optimization method is used with minimizing the crosstalk change rate in the highly sensitive area as the objective function and with the differential pair length matching error and transmission delay deviation as constraints to dynamically adjust the differential signal routing path and impedance continuity to generate a wiring layout with optimized environmental robustness.

[0130] After the local introduction of the temperature and humidity compensation dielectric structure, the dielectric distribution and local electrical characteristics within the board layer have changed. Therefore, the wiring layout needs to be further optimized based on the new foundation to improve the robustness of signal transmission under environmental changes. To this end, it is necessary to use iterative simulation optimization methods based on the updated dielectric distribution to dynamically adjust the routing paths and impedance continuity of each differential signal pair within the multi-layer PCB, and ultimately generate a wiring layout solution that meets the environmental robustness requirements.

[0131] During specific implementation, it is first necessary to regenerate the local electromagnetic simulation model of each differential pair based on the dielectric layer parameters after the compensation structure is arranged. The simulation model should accurately reflect the new dielectric constant distribution of each area, and at the same time update the original position, line width, line spacing and interlayer via information of the signal traces to ensure that the simulation boundary conditions are consistent with the actual design. Subsequently, the crosstalk change rate is set to be minimized as the global optimization objective function. The crosstalk change rate can be measured by comparing indicators such as the change in the near-end crosstalk level and the change in the far-end crosstalk level of each differential signal under different temperature and humidity environments. The optimization goal is to reduce the crosstalk change rate in the sensitive area to a certain proportion below that before compensation, for example, to reduce it by more than 30%.

[0132] To ensure overall signal quality during the optimization process, the optimization strategy also requires the introduction of a series of constraints. These include ensuring that the path length error of the differential signal pair must not exceed the maximum allowable error specified in the design specification, for example, less than 100 microns, to ensure timing matching between the differential pairs. Transmission delay deviation must be controlled within the system's tolerances, for example, the relative delay difference must not exceed 5% of the target clock period. Furthermore, differential impedance continuity must be maintained at key nodes, for example, the differential impedance deviation in corners, vias, and layer transitions must not exceed ±10% of the design target. These constraints can be implemented as hard constraints during the optimization process or as penalty terms in the objective function, allowing for automatic trade-offs between various performance metrics during the iterative process.

[0133] The optimization method can adopt a local search strategy based on gradient descent, or more simply, a heuristic local fine-tuning approach. Specifically, the signal traces in the identified highly sensitive areas can first be subjected to minor path offsets, bend radius adjustments, local line width or spacing fine-tuning, and other operations. After each adjustment, the electromagnetic simulation module is re-called to calculate the mutual capacitance change rate, mutual inductance change rate, and corresponding crosstalk change indicators of the area under different temperature and humidity conditions. If the crosstalk change rate decreases after adjustment and all constraints are still met, the adjustment is accepted; otherwise, it is canceled and other adjustment paths are explored. Through continuous small-step iterations, the local optimal layout solution is gradually found in the design space.

[0134] During the initial iteration phase, adjustments can be prioritized for the most sensitive areas, gradually expanding to surrounding areas. In the mid- to late-stage optimization, the global crosstalk variation rate can be further reduced by introducing a wider range of path fine-tuning. After each iteration, a complete simulation verification of the entire PCB routing layout is required, including checking differential pair length matching, impedance continuity, transmission delay deviation, and crosstalk variation levels to ensure that the overall design meets design performance requirements under all target temperature and humidity environments.

[0135] Ultimately, the routing layout with the lowest crosstalk variation rate and optimal signal quality indicators, while satisfying the objective function and all constraints, is selected as the final solution after environmental robustness optimization. This solution should fully document the routing paths, lengths, layers, key parameters, and local dielectric compensation for each differential signal pair to facilitate subsequent PCB layout and production process preparation.

[0136] To facilitate understanding and implementation, the following provides a specific example of an objective function and constraints. The objective function can be set to minimize the maximum crosstalk change rate caused by environmental changes in all highly sensitive areas. Specifically, assuming the near-end crosstalk level of a differential signal pair under normal temperature and humidity conditions is the baseline value, if the near-end crosstalk level increases by 20% after re-simulation under high temperature and humidity conditions, the change rate is 20%. The highest change rate among each differential pair is used as the evaluation metric, and the optimization goal is to reduce the maximum crosstalk change rate from 20% to within 10% through routing adjustments.

[0137] During the optimization process, several design constraints need to be met simultaneously. The length matching error of the differential pair should be controlled within a limited range. For example, the length difference between the two signal lines should not exceed 0.1 mm to ensure the timing synchronization of the differential signal. The transmission delay deviation requires that the delay change of each pair of signal lines under all temperature and humidity conditions should not exceed 5% to prevent data desynchronization or timing errors in high-speed signals. At the same time, in order to ensure small reflections and good signal integrity, the characteristic impedance change of each pair of differential signal lines after optimization shall not exceed ±10% of the design target impedance. For example, when the target impedance is 100 ohms, the actual impedance range allowed is between 90 ohms and 110 ohms.

[0138] In actual optimization, the system prioritizes the differential pairs with the highest crosstalk change rate, attempting to fine-tune routing positions or partially change trace widths. After each adjustment, it immediately evaluates whether the objective function and all constraints mentioned above have improved or are still satisfied. Only when the new routing layout reduces the objective function and all constraints are simultaneously satisfied is the adjustment accepted and the next round of optimization proceeds. This iteration continues until the objective function cannot be further reduced or the preset number of iterations is reached, ultimately determining the optimal routing solution.

[0139] Through the above-mentioned iterative simulation optimization process, the differential signal routing path and impedance characteristics can be dynamically adjusted in a targeted manner, effectively reducing the risk of crosstalk degradation in highly sensitive areas under temperature and humidity changes, and improving the transmission reliability and long-term stability of high-speed signals on multi-layer PCBs under complex environmental conditions.

[0140] Furthermore, after the compensation structure is arranged, the differential signal routing path and impedance continuity are dynamically adjusted based on the updated dielectric distribution using an iterative simulation optimization method with minimizing the crosstalk change rate in the highly sensitive area as the objective function and with the differential pair length matching error and transmission delay deviation as constraints, including:

[0141] Taking the maximum crosstalk change rate of each differential signal pair in the highly sensitive area identified by the crosstalk sensitivity evaluation model under different temperature and humidity conditions as the optimization target, constructing a main objective function with minimizing the maximum crosstalk change rate;

[0142] During the routing fine-tuning process, the length matching error, transmission delay deviation, and local impedance continuity of the differential signal pair are used as optimization constraints to ensure that the changes in various parameters in each round of routing adjustment cannot exceed the preset tolerance range.

[0143] A simulation iteration method based on local routing path fine-tuning is adopted. For the differential signal pairs in the identified highly sensitive areas, the routing spacing, routing height or path curvature are gradually and slightly adjusted. After each fine-tuning, local electromagnetic simulation verification is performed to evaluate whether the crosstalk change rate is reduced and whether all constraints are met at the same time, until the objective function converges or the iteration stop condition is reached, without the need for global rerouting of the entire wiring.

[0144] After the compensation structure is arranged and the dielectric distribution is updated, in order to further improve the signal integrity of the multi-layer PCB in a complex temperature and humidity change environment, it is necessary to perform iterative simulation-based routing path and impedance continuity optimization for the differential signal pairs in the highly sensitive area. First, based on the highly sensitive areas identified by the crosstalk sensitivity assessment model, the crosstalk change rate data of each pair of differential signal lines under different temperature and humidity conditions is extracted, especially focusing on the maximum crosstalk change rate of each differential signal pair under all environmental change paths. With this maximum crosstalk change rate as the optimization focus, an optimization objective function is constructed, and the optimization goal is set to continuously reduce the maximum crosstalk change rate of each differential signal pair in the highly sensitive area during subsequent iterations to minimize the crosstalk level under environmental changes.

[0145] In actual wiring optimization operations, it is not a simple global adjustment of the PCB wiring structure, but rather a local fine-tuning of the differential signal path in the highly sensitive area. In the process of local fine-tuning, it is necessary to simultaneously consider the length matching error, transmission delay deviation and local impedance continuity of the differential signal pair as hard constraints to ensure that while optimizing the crosstalk change rate, no new signal integrity issues are caused. The length matching error is controlled within the allowable range to ensure the timing synchronization of the differential pair, the transmission delay deviation is controlled to prevent the path delay mismatch caused by temperature and humidity changes, and the impedance continuity control is to suppress the occurrence of reflections and secondary crosstalk. In order to maintain the accuracy and controllability of the optimization, in each round of wiring adjustment, all parameter changes must be limited to the preset tolerance range. For example, changes in line width must not cause the impedance deviation to exceed the standard, and adjustments to the routing bend radius must not lead to a significant increase in the reverse coupling effect.

[0146] Specific methods for local routing fine-tuning include slightly adjusting the line spacing of differential trace pairs, moderately changing the trace height or the relative distance between the trace layer and the reference plane, or introducing slight curvature changes in the trace path, thereby adjusting the electromagnetic field distribution locally and reducing the unevenness of mutual capacitance and mutual inductance. After each fine-tuning is completed, the updated mutual capacitance change rate, mutual inductance change rate, near-end crosstalk change, and far-end crosstalk change are immediately re-evaluated based on the electromagnetic simulation of the local area, and the constraints are checked to ensure that they are still met. Simulation verification must be performed based on the original updated dielectric distribution to ensure that the optimization process accurately reflects the changes in material properties after the introduction of the actual compensation structure.

[0147] If the local simulation results after the current fine-tuning show a decrease in the maximum crosstalk change rate within the target high-sensitivity area compared to the previous round, and all constraints are still met, the adjustment is accepted and used as the basis for the next iteration. Otherwise, the fine-tuning is withdrawn and other adjustment strategies or small changes to the path are tried. Through continuous small-step local fine-tuning and simulation evaluation, the optimal routing path distribution is gradually approached until the maximum crosstalk change rate reaches the preset convergence threshold or the maximum number of iterations is reached. The entire optimization process eliminates the need for large-scale rerouting of the entire PCB routing structure, effectively reducing process complexity and ensuring the overall consistency and manufacturability of the original design.

[0148] A second embodiment of the application provides an electronic device, comprising:

[0149] processor;

[0150] The memory is used to store a program, which, when read and executed by the processor, executes a multi-layer PCB board inter-layer signal crosstalk modeling and optimization method provided in the first embodiment of the present application.

[0151] The third embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, the method for modeling and optimizing signal crosstalk between layers of a multi-layer PCB board provided in the first embodiment of the present application is executed.

[0152] Although the present application is disclosed as above with the preferred embodiments, it is not intended to limit the present application. Any person skilled in the art may make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the scope of protection of the present application shall be based on the scope defined by the claims of the present application.

Claims

1. A multi-layer PCB inter-layer signal crosstalk modeling and optimization method, characterized in that: include: Collect frequency response data of each dielectric layer of the PCB board under different temperature and humidity conditions, and establish a dynamic mapping model with temperature gradient and humidity diffusion gradient as independent variables and dielectric constant as dependent variable; Based on the dynamic mapping model and the spatial layout parameters of each differential signal pair within a multi-layer PCB, the mutual capacitance and mutual inductance distributions under different temperature and humidity environments are calculated, and a crosstalk sensitivity assessment model is constructed that couples frequency, environmental conditions, and layout structure. The crosstalk sensitivity evaluation model is used to identify highly sensitive areas where mutual capacitance and mutual inductance are prone to drastic fluctuations under temperature and humidity changes, and to quantify the crosstalk change rate in each area. Based on the location of highly sensitive areas and the crosstalk variation characteristics, a temperature and humidity compensation dielectric structure is locally introduced to regulate the variation trend of the local dielectric constant and reduce the amplification effect of environmental changes on the crosstalk characteristics; After the compensation structure is arranged, based on the updated dielectric distribution, an iterative simulation optimization method is adopted with minimizing the crosstalk change rate in highly sensitive areas as the objective function and with differential pair length matching error and transmission delay deviation as constraints. The differential signal routing path and impedance continuity are dynamically adjusted to generate a wiring layout with optimized environmental robustness.

2. The method for modeling and optimizing signal crosstalk between layers of a multi-layer PCB according to claim 1, wherein: The method collects frequency response data of each dielectric layer of the PCB board under different temperature and humidity conditions, and establishes a dynamic mapping model with temperature gradient and humidity diffusion gradient as independent variables and dielectric constant as dependent variable, including: Collect frequency response data of each dielectric layer of the PCB board at different frequency points under multiple set temperature and humidity change paths; Based on the collected data, the data under different working conditions are classified according to the temperature change rate and humidity diffusion rate, and weighted average processing is performed within the classified data group to generate the dielectric constant change trend of the corresponding frequency point; Based on the generated dielectric constant variation trend, the dielectric constant value under the target environmental conditions is estimated by performing linear interpolation calculations between adjacent classification groups for any set temperature gradient and humidity diffusion gradient.

3. The method for modeling and optimizing signal crosstalk between layers of a multi-layer PCB according to claim 1, wherein: Based on the dynamic mapping model, combined with the spatial layout parameters of each differential signal pair in the multi-layer PCB, the mutual capacitance and mutual inductance distribution under different temperature and humidity environments are calculated, and a crosstalk sensitivity evaluation model coupled with frequency, environmental conditions and layout structure is constructed, including: The line width, line spacing, trace height, trace direction, and via layout parameters of each differential signal pair in a multi-layer PCB are used as the first input, the dielectric constant of each dielectric layer under the target temperature and humidity environment conditions is used as the second input, and the target frequency point is used as the third input; Based on the first input, the second input, and the third input, the mutual capacitance change rate and the mutual inductance change rate of each pair of differential signal lines under the target temperature, humidity, and frequency conditions are calculated by table lookup reasoning combined with a local linear weighted method. Based on the calculated mutual capacitance change rate and mutual inductance change rate, the corresponding near-end crosstalk level change amount and far-end crosstalk level change amount are derived; Based on the derived near-end crosstalk change and far-end crosstalk change, the crosstalk change rate exceeding risk level of each pair of differential signal lines is evaluated according to the preset crosstalk change rate threshold, and the mutual capacitance change rate, mutual inductance change rate, near-end crosstalk change, far-end crosstalk change and exceeding risk level are used as input basis for identifying highly sensitive areas where mutual capacitance and mutual inductance are prone to violent fluctuations under temperature and humidity changes through the crosstalk sensitivity assessment model and quantifying the crosstalk change rate of each area.

4. The method for modeling and optimizing signal crosstalk between layers of a multi-layer PCB according to claim 1, wherein: The derivation of the corresponding near-end crosstalk level change and far-end crosstalk level change according to the calculated mutual capacitance change rate and mutual inductance change rate includes: For each pair of differential signal lines, the electric and magnetic field coupling trends between the differential pairs are calculated based on the mutual capacitance and mutual inductance change rates calculated under the target temperature, humidity, and frequency conditions. The near-end crosstalk level change is primarily derived based on the mutual capacitance change rate, while the far-end crosstalk level change is primarily derived based on the mutual inductance change rate. During the derivation process, the near-end crosstalk trend is approximately predicted by performing a forward weighted integration on the mutual capacitance change rate, and the far-end crosstalk trend is approximately predicted by performing a reverse weighted integration on the mutual inductance change rate. The weighting coefficient is then corrected based on the dynamic change amplitude of the ambient dielectric constant and permeability at each frequency point to improve the derivation accuracy. Based on the derivation results, local normalization processing is used to standardize the near-end crosstalk change and the far-end crosstalk change of each differential signal pair to the same sensitivity scale, which serves as the basic input basis for subsequently identifying highly sensitive areas and quantifying the crosstalk change rate of each area through the crosstalk sensitivity assessment model.

5. The method for modeling and optimizing signal crosstalk between layers of a multi-layer PCB according to claim 1, wherein: Based on the location of the highly sensitive area and the crosstalk variation characteristics, a temperature and humidity compensation dielectric structure is locally introduced to regulate the variation trend of the local dielectric constant and reduce the amplification effect of environmental changes on the crosstalk characteristics, including: For the highly sensitive areas identified by the crosstalk sensitivity assessment model, the scope of the area requiring local compensation is determined based on the mutual capacitance change rate, mutual inductance change rate, near-end crosstalk change amount, and far-end crosstalk change amount of each differential signal pair in the area, combined with the routing density and local thermal distribution characteristics; Within the determined compensation area, select a temperature and humidity compensation material whose dielectric constant matches that of the original dielectric material within the target frequency range and whose moisture absorption characteristics are less affected by the environment. Based on the characteristics of the surrounding environment of the differential signal routing, choose to use an embedded inlay structure or a surface-attached structure for local compensation; The local dielectric parameters after the introduction of the temperature and humidity compensation medium are updated to the overall dielectric layer distribution model. After the compensation structure is arranged, the updated dielectric distribution is used as the input basis for dynamically adjusting the differential signal routing path and impedance continuity using an iterative simulation optimization method with minimizing the crosstalk change rate in the highly sensitive area as the objective function and differential pair length matching error and transmission delay deviation as constraints.

6. The method for modeling and optimizing signal crosstalk between layers of a multi-layer PCB according to claim 1, wherein: After the compensation structure is arranged, the differential signal routing path and impedance continuity are dynamically adjusted based on the updated dielectric distribution using an iterative simulation optimization method with minimizing the crosstalk change rate in the highly sensitive area as the objective function and with differential pair length matching error and transmission delay deviation as constraints, including: Taking the maximum crosstalk change rate of each differential signal pair in the highly sensitive area identified by the crosstalk sensitivity evaluation model under different temperature and humidity conditions as the optimization target, constructing a main objective function with minimizing the maximum crosstalk change rate; During the routing fine-tuning process, the length matching error, transmission delay deviation, and local impedance continuity of the differential signal pair are used as optimization constraints to ensure that the changes in various parameters in each round of routing adjustment cannot exceed the preset tolerance range. A simulation iteration method based on local routing path fine-tuning is adopted. For the differential signal pairs in the identified highly sensitive areas, the routing spacing, routing height or path curvature are gradually and slightly adjusted. After each fine-tuning, local electromagnetic simulation verification is performed to evaluate whether the crosstalk change rate is reduced and whether all constraints are met at the same time, until the objective function converges or the iteration stop condition is reached.

Images