Electronic component pin plating thickness detection and loss prediction system

Abstract

This invention discloses a system for detecting and predicting the thickness of plating on electronic components leads, belonging to the field of electronic component testing technology. The system consists of a sample screening module, an eddy current real-time non-destructive testing module, a multi-dimensional cause acquisition module, a loss and thickness prediction module, and a deviation calculation and early warning module. It screens valid samples by initially measuring the plating thickness of electronic components; acquires lead plating thickness data using eddy current non-destructive testing technology; collects multi-dimensional related parameters affecting plating loss and converts them into quantitative cause indicators; constructs a loss change model based on the quantitative relationship between thickness data and cause indicators, and generates predicted thickness values ​​by combining the trends of cause indicators; and achieves graded early warning by calculating theoretical deviation and comparing it with preset thresholds. This system balances the non-destructive nature and real-time performance of the detection, comprehensively considers the core influencing factors of plating loss, improves the scientificity and accuracy of loss prediction, and promptly identifies various risks.

Description

Technical Field

[0001] This invention relates to the field of electronic component testing technology, and more specifically, to a system for detecting the thickness of plating on electronic component leads and predicting losses. Background Technology

[0002] The thickness and wear status of the plating on the leads of electronic components directly affect the electrical performance, corrosion resistance, and service life of the components, and thus the stable operation of electronic equipment. Currently, the thickness of the plating on the leads of electronic components is mainly tested using destructive testing or conventional non-destructive testing techniques. Destructive testing can damage the sample and cannot achieve full life cycle monitoring. Conventional non-destructive testing techniques have problems such as low detection accuracy (error greater than 0.05μm), susceptibility to environmental interference during the detection process, and difficulty in real-time tracking of the plating thickness during the usage stage.

[0003] In terms of coating loss prediction, existing technologies mostly consider only a single factor (such as temperature or humidity), failing to comprehensively cover multiple core factors such as temperature fluctuations, environmental corrosion, and workload fluctuations. This results in low fitting of the loss prediction model and large deviations in the prediction results. At the same time, existing systems lack a scientific early warning mechanism, relying solely on a single thickness threshold for judgment. This makes it impossible to identify potential risks such as prediction accuracy deviations and accelerated loss in a timely manner, which can easily lead to untimely warnings of component failures, thereby causing malfunctions in electronic equipment and resulting in economic losses. Therefore, there is an urgent need for a system that combines high-precision real-time detection, multi-dimensional consideration of causes, accurate prediction, and scientific early warning functions to address the shortcomings of existing technologies. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an electronic component pin plating thickness detection and loss prediction system, which addresses the problems of insufficient detection accuracy, large prediction deviation, and untimely early warning in the prior art as mentioned in the background section through the following solutions.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a system for detecting the plating thickness of electronic component leads and predicting losses, characterized in that it comprises:

[0006] Sample screening module: used to determine the initial coating thickness of electronic components in the same batch and screen valid samples according to sampling rules;

[0007] Eddy current real-time non-destructive thickness measurement module: Used to detect the thickness of the pin plating of valid samples at different stages of use in real time using eddy current non-destructive testing technology, and output real-time thickness data;

[0008] Multi-dimensional cause acquisition module: used to synchronously collect multi-dimensional parameters that affect coating loss, and calculate multi-dimensional cause indicators;

[0009] The multi-dimensional causative indicators include temperature difference values, environmental corrosion degree indicators, and computational quantity difference values.

[0010] Loss and thickness prediction module: Based on the quantitative relationship between thickness data and multi-dimensional inducing indicators, a thickness loss change model is constructed and thickness prediction values ​​are generated;

[0011] The generation of the thickness prediction value first involves loss prediction. Specifically, the prediction causative index is obtained by extrapolating the trends of multi-dimensional causative indexes from the previous three time periods. In conventional scenarios, linear extrapolation is used. When the growth rate of the causative indexes in two consecutive adjacent time periods is ≥20%, exponential smoothing is used. The deviation between the prediction causative indexes and the actual values ​​in the most recent time period is controlled to be ≤30%. The prediction causative indexes are then substituted into the thickness loss change model to calculate the predicted loss amount. Finally, the real-time thickness in the i-th time period is used as a benchmark, and the predicted loss amount is subtracted to obtain the thickness prediction value for the (i+1)-th time period.

[0012] Deviation Calculation and Early Warning Module: Based on the thickness prediction value, the theoretical deviation is calculated, compared with the multi-dimensional early warning threshold, and early warning information is output.

[0013] The technical effects and advantages of this invention are as follows:

[0014] 1. This invention employs eddy current nondestructive testing technology, combined with dedicated testing components, multi-stage calibration, and anomaly data investigation mechanisms, to achieve full lifecycle thickness monitoring without damaging the pin plating structure. This effectively avoids the destructive or inaccurate problems of traditional testing technologies, ensuring the reliability and continuity of thickness testing.

[0015] 2. This invention comprehensively considers three core factors: temperature fluctuation, environmental corrosion, and computational fluctuation. It transforms multi-dimensional parameters into quantitative indicators and constructs a loss change model. Combined with trend extrapolation, it predicts the inducing factors and achieves a scientific prediction of coating loss. This overcomes the prediction deviation problem caused by considering only one inducing factor in the existing technology and improves the rationality and accuracy of loss prediction.

[0016] 3. This invention establishes a graded early warning response mechanism by setting multi-dimensional early warning thresholds covering prediction accuracy, actual thickness, and loss rate. This mechanism can not only identify the immediate risk of failure due to insufficient coating thickness in a timely manner, but also accurately warn of potential problems such as prediction deviation and accelerated loss. This provides precise guidance for the maintenance of electronic components and reduces potential equipment malfunctions.

[0017] 4. By establishing strict sampling rules, this invention measures and screens the initial thickness of electronic components of the same model in the same batch, ensuring that the initial state of effective samples is consistent. This eliminates the interference of initial thickness differences on loss analysis, provides high-quality data support for model building and predictive analysis, and further improves the stability and reliability of the overall system operation. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the overall system architecture of the present invention;

[0019] Figure 2 This is a schematic diagram of the sample screening process of the present invention;

[0020] Figure 3 This is a flowchart of the thickness detection process of the present invention;

[0021] Figure 4 This is a flowchart of the multi-dimensional cause acquisition process of the present invention;

[0022] Figure 5 This is a flowchart illustrating the prediction and early warning process of the present invention.

[0023] Figure 6 This is a schematic diagram illustrating the relationship between the predicted thickness loss and multi-dimensional inducing factors of this invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] As attached Figures 1 to 6 The electronic component pin plating thickness detection and loss prediction system shown includes:

[0026] Sample screening module: used to determine the initial coating thickness of electronic components in the same batch and screen valid samples according to sampling rules;

[0027] It should be specifically noted that the core function of the sample screening module is to provide samples with "consistent initial state and statistical representativeness" for subsequent coating thickness detection, loss analysis and model construction, to eliminate the interference of initial thickness differences on loss calculation, and to ensure that the loss data only reflects the coating consumption during the use stage.

[0028] It should be further explained that the sample selection method is as follows: select electronic components of the same production batch and model specifications, and ensure that the plating process (such as electroplating current, plating metal purity, electroplating time) and pin structure (such as package type, number of pins, pin size) of all samples are completely consistent; at the same time, the number of samples selected is not less than 30.

[0029] The initial pin plating thickness was determined as follows: a high-precision eddy current sensor (detection accuracy ≤ 0.01 μm) was used. During the detection, the sensor was fixed directly above the sample pin, maintaining a constant distance of 2 mm between the sensor and the pin surface. Thickness was measured three times for all pins of each sample. The arithmetic mean of the three detection results for a single pin was taken first, and then the average thickness of all pins in the same sample was calculated. This average value was taken as the initial plating thickness h0 of the sample.

[0030] The effective sample screening method is as follows: First, calculate the arithmetic mean of the initial coating thickness h0 of all preliminary samples, denoted as h avg Calculate the initial thickness deviation rate for a single sample using the following formula: Only samples with a deviation rate within the range of "-0.5% to +0.5%" are retained as valid samples, and samples with a deviation rate outside this range are removed.

[0031] Eddy current real-time non-destructive thickness measurement module: Used to detect the thickness of the pin plating of valid samples at different stages of use in real time using eddy current non-destructive testing technology, and output real-time thickness data;

[0032] The number of valid samples after screening must be ≥30; if it is less than 30, samples from the same batch must be added again for testing and screening to ensure that the amount of training data for the subsequent model meets the standard.

[0033] It should be specifically noted that the eddy current real-time non-destructive thickness measurement module uses eddy current non-destructive testing technology to accurately and in real-time acquire coating thickness data at different stages of use without damaging the coating structure of the effective sample pins.

[0034] It should be further explained that the selection and configuration of the core detection components relied upon by the eddy current nondestructive testing technology are as follows: a high-precision eddy current sensor is used as the core detection element, along with a sensor fixing mechanism, a signal conditioning unit, and a data acquisition terminal. The eddy current sensor uses a high-frequency eddy current probe with a detection accuracy ≤0.01μm and a response time ≤10ms. The probe diameter is adapted to the pin size of the electronic components (e.g., a 0.5mm diameter probe is used for 0402 surface mount resistor pins; a 1mm diameter probe is used for DIP package pins), ensuring that the probe sensing area completely covers the pin plating detection surface. The signal conditioning unit includes a differential amplifier circuit and a low-pass filter circuit to suppress external electromagnetic interference (such as current noise during component operation), amplifying the weak sensing signal output by the sensor to a 0~5V standard voltage signal. The cutoff frequency of the filter circuit is set to 1kHz to avoid high-frequency noise affecting thickness calculation. The data acquisition terminal uses a data acquisition card with a sampling rate ≥100Hz to ensure real-time capture of thickness data changes.

[0035] It should be further explained that the preparations before real-time detection include:

[0036] Sample fixing: The valid samples screened by the sample screening module are fixed on the insulation testing fixture. The fixture is equipped with a micron-level adjustment knob to fine-tune the sample position to keep the pin horizontal. At the same time, the detection points of all samples are unified (select the middle area of ​​the pin, avoid the uneven plating areas at the root and end) to ensure that the eddy current sensing area is consistent.

[0037] Environmental control: The detection environment is maintained at normal temperature (25±5℃) and normal humidity (45±5%RH), with the concentration of harmful gases ≤1ppm, and is far away from strong electromagnetic interference sources (such as high-power power supplies and high-frequency equipment) to avoid environmental factors interfering with the stability of the eddy current magnetic field and affecting the detection signal.

[0038] Sensor Calibration: Before starting the test, calibrate the eddy current sensor using standard plating test blocks (tin plating: 2.000μm, 3.000μm, 5.000μm; gold plating: 0.500μm, 1.000μm, 2.000μm). Align the probe with the plating surface of the test block to collect the signal. If the deviation between the detected value and the standard value is >0.005μm, adjust the gain knob of the signal conditioning unit until the deviation is ≤0.005μm to ensure that the detection accuracy meets the standard.

[0039] The subsequent steps of the real-time detection specifically include:

[0040] Testing timing matching: According to the preset usage time period of electronic components (such as 0~100h, 100~200h, 200~500h, etc.), key tests are carried out at the beginning and end of each time period. At the same time, random checks are conducted once per hour within each time period to balance the comprehensiveness and efficiency of the tests.

[0041] Probe positioning: The eddy current sensor is moved to the position directly above the effective sample pin by the fixing mechanism. The laser rangefinder is activated to monitor the distance between the probe and the pin surface. The mechanism is adjusted to keep the distance constant at 2mm (distance deviation ≤ ±0.05mm). After locking the mechanism, the detection is started to avoid distance fluctuations affecting the eddy current sensing intensity.

[0042] Thickness data calculation: For each sample, all pins are individually inspected using eddy current technology. Thickness signals are continuously acquired for each pin three times, and the real-time thickness of the pin is calculated using the following formula: Where h1, h2, and h3 are the three detection values ​​of a single pin; then the average thickness of all pins in the same sample is calculated as the real-time thickness data h of the sample at the current stage. i (i is the time period number).

[0043] It should be further explained that the accuracy of the real-time thickness data is ensured as follows: During the detection process, the distance between the eddy current sensor probe and the pin surface is continuously monitored using a laser rangefinder. If the distance fluctuation exceeds ±0.05mm, the system automatically pauses the detection and issues a prompt, continuing only after readjusting the positioning to a constant distance. For three consecutive thickness data acquisitions of a single pin, if the range (difference between the maximum and minimum values) exceeds 0.008μm, it is judged as a detection anomaly, and three more detections must be performed. If the anomaly still exists, the sample is marked as "pending verification," and probe contamination is investigated. Factors affecting the eddy current sensor include contamination and surface dirt on the pins. After testing every 50 valid samples, the sensor is recalibrated using standard plating test blocks (consistent with the initial calibration specifications: tin plating: 2.000μm, 3.000μm, 5.000μm; gold plating: 0.500μm, 1.000μm, 2.000μm). If the deviation between the detected value and the standard value is >0.005μm, the gain knob of the signal conditioning unit is readjusted until the deviation is ≤0.005μm, ensuring that the sensor's detection accuracy is always maintained within the range of ≤0.01μm.

[0044] Multi-dimensional cause acquisition module: used to synchronously collect multi-dimensional parameters that affect coating loss, and calculate multi-dimensional cause indicators;

[0045] The multi-dimensional causative indicators include temperature difference values, environmental corrosion degree indicators, and computational quantity difference values.

[0046] It should be specifically explained that the core function of the multi-dimensional cause acquisition module is to collect raw parameters through dedicated sensors and monitoring units for the three core influencing factors of electronic component pin plating loss (temperature fluctuation, environmental corrosion, and computational fluctuation), and convert them into quantitative multi-dimensional cause indicators through standardized calculation. Moreover, all data acquisition and calculation processes are completely synchronized with the "real-time thickness detection" of the eddy current real-time non-destructive thickness measurement module in the time dimension.

[0047] It should be further explained that the temperature difference value (ΔT) i The data collection and calculation methods are as follows:

[0048] Acquisition components and configuration: A contact temperature sensor with a measurement range of -50℃ to 150℃ and an accuracy of ±0.1℃ is selected. The sensor probe is fixed to the middle of the pin of the valid sample with thermally conductive adhesive to ensure that the actual temperature of the pin during operation (rather than the ambient temperature) is directly acquired, thus avoiding temperature transmission errors.

[0049] Data Acquisition Frequency and Synchronization: The acquisition frequency is consistent with that of the eddy current real-time non-destructive thickness measurement module. Temperature data is collected synchronously at the “start point”, “end point” and “hourly sampling point” of the preset usage time period (such as 0~100h, 100~200h). At the same time, real-time temperature is continuously collected at a frequency of 1 time / minute to ensure that the temperature fluctuations within the coverage time period are covered.

[0050] Temperature difference calculation: Taking a single preset usage time period as the calculation unit (denoted as the i-th time period, where i is the time period number, e.g., i=1 corresponds to 0~100h), the maximum temperature value (denoted as T) is extracted from all temperature data collected within this time period. imax (Unit: °C, representing the highest temperature of the pin during the i-th time period) and the minimum temperature (denoted as T). imin (Unit: °C, representing the lowest temperature of the pin during the i-th time period), calculate the temperature difference ΔT for that time period using the following formula. i ΔT i =T imax -T imin Where, ΔT i The temperature difference value (unit: °C) for the i-th time period reflects the fluctuation range of the pin temperature during that time period. The greater the fluctuation, the higher the risk of plating loss due to thermal stress.

[0051] It should be further explained that the environmental corrosion index (E) i The data collection and calculation methods are as follows:

[0052] Data Acquisition Components and Configuration: Raw parameters related to environmental corrosion are simultaneously acquired using three types of dedicated sensors. Component parameters and installation requirements are as follows:

[0053] Humidity sensor: Measurement range 0%~100%, accuracy ±1%, installed in a well-ventilated area of ​​the environment where the valid sample is located, avoiding direct contact with water vapor, to collect the ambient relative humidity (denoted as RH). i ,unit:%);

[0054] pH sensor: Measurement range 0~14, accuracy ±0.01, collects condensate (or simulated environmental liquid medium) from ambient air through a sampling unit to measure the ambient pH. i (No unit)

[0055] Harmful gas concentration sensor: Measurement range 0~100ppm, accuracy ±0.1ppm, for corrosive gases such as H2S and SO2, collecting the concentration of harmful gases (denoted as C). i (Unit: ppm)

[0056] Baseline parameter settings: Environmental parameters at the time of initial sample collection and testing are used as baseline values, including initial relative humidity (denoted as RH0, unit: %, reference value 45%) and optimal pH value for component operation (denoted as pH). opt Unitless, reference value 7.0, representing the pH condition with minimal coating corrosion; initial harmful gas concentration (denoted as C0, unit: ppm, reference value 0ppm); simultaneously setting the permissible environmental parameter limits for components: maximum relative humidity (denoted as RH). max (Unit: %, Reference value 85%), Allowable pH deviation from pH opt The maximum amplitude (denoted as ΔpH) max Unitless, reference value 1.0), maximum hazardous gas concentration (denoted as C). max (Unit: ppm, reference value 50ppm)

[0057] Calculation of environmental corrosion degree index: The weighted summation method is used to transform multi-dimensional corrosion parameters into a single quantitative index E. i (Value range [0,1], E) i The larger the value, the more severe the environmental corrosion (the formula is as follows): Wherein, ω1, ω2, and ω3 are the weighting coefficients for humidity, pH value, and harmful gas concentration, respectively, satisfying ω1+ω2+ω3=1 (reference values: ω1=0.4, ω2=0.3, ω3=0.3); the weight calibration process is as follows: select humidity (3 levels: 45%, 65%, 85%), pH value (3 levels: 6.0, 7.0, 8.0), and harmful gas concentration (3 levels: 0ppm, 25ppm, 50ppm) as test factors; conduct accelerated corrosion tests on the target coating material and measure the coating loss Δh under different working conditions; establish a multiple linear regression model with Δh as the dependent variable and the dimensionless ratio of each corrosion factor as the independent variable; normalize the absolute value of the regression coefficients to obtain the weights; the weights can be adjusted according to the corrosion sensitivity of different coating materials (e.g., gold coatings are more sensitive to harmful gases, so ω3 can be appropriately increased), and after adjustment, ω1+ω2+ω3=1 must be guaranteed; additional constraint: if E after calculation i <0, automatically corrected to 0; if E i >1, automatically corrected to 1, ensuring E i It always stays within the range of [0,1].

[0058] It should be further explained that the calculated difference value (ΔQ) i The data collection and calculation methods are as follows:

[0059] Data Acquisition Components and Configuration: Configure a computational monitoring unit, which connects to the control chip of the electronic components via SPI (Serial Peripheral Interface) or I2C (Integrated Circuit Bus) to acquire the operating current of the components (denoted as I) in real time. i (unit: mA) and clock frequency (denoted as f) i (Unit: MHz)

[0060] Real-time computation load conversion: Based on the hardware architecture of the components (such as the number of CPU cores and instruction cycles), a preset computation load conversion rule is used: the product of the operating current and the clock frequency is converted into real-time computation load in MIPS (millions of instructions per second) using a fixed coefficient (denoted as Q). i (Unit: MIPS), the conversion factor is determined according to the component model (e.g., the conversion factor for a certain MCU chip is 0.02, i.e., Qᵢ=0.02×I). i ×f i );

[0061] Calculation of computational load difference: Taking the i-th preset usage time period as the calculation unit, calculate the real-time computational load Q collected within that time period. i In the middle, extract the maximum computational cost (denoted as Q). imax (unit: MIPS) and minimum computational cost (denoted as Q) imin(Unit: MIPS) Calculate the computational variance ΔQ for this time period using the following formula. i ΔQ i =Q imax -Q imin Where, ΔQ i The computational load difference value (unit: MIPS) for the i-th time period reflects the fluctuation range of the component's workload during that time period. The greater the fluctuation, the more significant the plating loss on the pins due to Joule heating and electromigration.

[0062] It should be further explained that the synchronization mechanism of the multi-dimensional parameters is as follows: all temperature sensors, environmental corrosion sensors and computational monitoring units are equipped with a Beidou millisecond-level time synchronization module (automatically switching to GPS + local crystal oscillator dual-mode synchronization in indoor or obstructed environments). Based on the "real-time thickness detection timestamp" output by the eddy current real-time non-destructive thickness measurement module, the local clock is calibrated every 10 seconds to ensure that the timestamp of the original parameter acquisition and the calculation of the causative index deviates from the timestamp of the real-time thickness data by ≤10ms, so as to achieve complete time dimension matching of the "thickness-causative" data.

[0063] Loss and thickness prediction module: Based on the quantitative relationship between thickness data and multi-dimensional inducing indicators, a thickness loss change model is constructed and thickness prediction values ​​are generated;

[0064] It should be specifically noted that the core function of the loss and thickness prediction module is to receive real-time thickness data and multi-dimensional inducing factors, establish a quantitative correlation between "coating loss and multiple inducing factors", and output accurate thickness prediction values.

[0065] It should be further explained that the thickness loss (Δh) i The calculation method is as follows: taking a single preset usage time period as the calculation unit, based on the real-time thickness h of the (i-1)th time period. i-1 (h when i=1) i-1 =h0, where h0 is the initial coating thickness obtained by the sample screening module (h0 is fixed for the same sample) and the real-time thickness h in the i-th time period. i The coating thickness loss Δh during this time period is calculated using the following formula. i Δh i =h i-1 -h0 where Δh i ≥0 (because the coating thickness will only decrease and not increase); if the calculation result shows Δh i <0 indicates data anomaly: When i=1 (first time period), the temporary loss is set to 0 and marked as "data anomaly, testing equipment needs to be checked"; when i≥2, the temporary loss is set to 0 and marked as "data anomaly, sample status and testing accuracy need to be checked" to avoid abnormal data causing cumulative errors in loss.

[0066] It should be further explained that the specific method for constructing the thickness loss variation model is as follows:

[0067] A multiple linear regression model was used as the core loss model to quantify the combined impact of temperature fluctuations, environmental corrosion, and computational fluctuations on coating loss; the dependent variable was the thickness loss Δh. i (Unit: μm), the independent variable is the three major inducing factors obtained by the multi-dimensional inducing factor acquisition module—temperature difference value ΔT. i (Unit: °C), Environmental Corrosion Degree Index E i Unitless), calculation difference ΔQ i (Unit: MIPS)

[0068] The model formula is as follows: The meanings and constraints of each parameter are as follows: a, b, and c are regression coefficients (obtained by fitting sample data): a represents the coating loss corresponding to a unit temperature difference, in μm / ℃ (reference value: 0.002μm / ℃); b represents the coating loss corresponding to a unit environmental corrosion degree, in μm (reference value: 1.5μm); c represents the coating loss corresponding to a unit computational difference, in μm / MIPS (reference value: 0.001μm / MIPS); ε is the random error term (satisfying a normal distribution, with a mean of 0 and a variance σ²≤0.000011μm²), representing minor disturbances not captured by the model (such as minor sensor noise), and |ε|≤0.01μm within the 99.73% confidence interval to ensure model fitting accuracy;

[0069] The least squares method was used to fit the regression coefficients a, b, and c. The training set consisted of multiple preprocessed sample data sets, which had to meet the following requirements: ≥30 valid samples, and ≥5 time period data sets for each sample. If a single sample had fewer than 5 time period data sets, other sample data from the same batch and under the same operating conditions needed to be merged ("same operating conditions" is defined as the temperature difference ΔT during sample usage). i Environmental corrosion index E i Calculation difference value ΔQ i The fluctuation range is ≤10%, ensuring a total training set sample size of ≥30×5=150 groups (to guarantee statistical significance); by minimizing the "actual Δh" i Compared with the model prediction Δh pred,i The model was fitted using the sum of squares, and the goodness of fit R² ≥ 0.92, ensuring that the model can effectively explain the variation of coating loss.

[0070] It should be further explained that the predicted thickness value (h) pred,i+1The generation method is as follows: based on the constructed thickness loss change model, combined with the "predicted cause index of the (i+1)th time period", the predicted thickness for that time period is calculated, specifically including:

[0071] Obtain the predictive indices, including the predicted temperature difference value ΔT for the (i+1)th time period. pred,i+1 E, an index for predicting the degree of environmental corrosion pred,i+1 ΔQ, the difference in predicted calculations pred,i+1 This is obtained through extrapolation of historical data trends, specifically including:

[0072] Typical scenario: Using linear extrapolation, the values ​​of the next stage are predicted based on the average values ​​of the causal indicators in the previous three time periods;

[0073] Non-linear trend scenario: Calculate the growth rate of the inducing indicator in adjacent time periods. If the growth rate of two consecutive adjacent time periods is ≥20%, it is determined to be a non-linear trend. Use the exponential smoothing method (smoothing coefficient α=0.3) for prediction to reduce the error caused by trend abrupt changes.

[0074] Extrapolation error control: The deviation between the predicted value and the actual value in the most recent time period is ≤30%. If it exceeds this, a "prediction accuracy warning" will be triggered, prompting manual review of the trend's rationality.

[0075] Substituting the predicted cause index into the loss model, the predicted thickness loss Δh for the (i+1)th time period is calculated. pred,i+1 : ;

[0076] The real-time thickness h in the i-th time period i Based on the baseline, subtract the predicted loss Δh pred,i+1 The thickness prediction value h for the (i+1)th time period is obtained. pred,i+1 :h pred,i+1 =h i -Δh pred,i+1 According to the minimum allowable thickness threshold h of the coating min Supplementary risk assessment: If h pred,i+1 If <0, it is automatically corrected to 0, marked as "Predicted Risk (Complete Coating Loss)" and a Level 2 warning is triggered; if 0 < h pred,i+1 <h min Mark "Predicted Risk (Coating Thickness Below Minimum Allowable Value)" to trigger a Level 2 warning; if h pred,i+1 h min No risk, normal output of predicted values.

[0077] Deviation Calculation and Early Warning Module: Based on the thickness prediction value, the theoretical deviation is calculated, compared with the multi-dimensional early warning threshold, and early warning information is output.

[0078] It should be further explained that the theoretical deviation is calculated as follows:

[0079] Theoretical Deviation (D) Calculation: To evaluate the model's prediction accuracy, the theoretical deviation D for the i-th time period is calculated using the following formula, quantifying the degree of deviation between actual and predicted losses: Meaning of each parameter: h i h represents the actual real-time thickness in the i-th time period (from the eddy current real-time non-destructive thickness measurement module). pred,i Let be the predicted thickness value for the i-th time period (predicted based on data from the (i-1)-th time period); the value of D ranges from [0, +∞), with smaller values ​​indicating higher prediction accuracy, and when h i or h pred,i When the value is close to 0, it can accurately reflect the relative error and avoid the initial thickness from masking later accuracy issues;

[0080] It should be further explained that the multi-dimensional early warning threshold is set by: presetting two types of core early warning thresholds. The thresholds can be flexibly adjusted according to the model of the electronic component, the application scenario, and the coating material. The default reference values ​​are as follows:

[0081] Theoretical deviation threshold (D) th The default setting is 5%, representing the maximum allowable prediction deviation range when D ≥ D. th If the prediction model deviates significantly from the actual loss pattern, attention should be paid to data collection or model adaptability.

[0082] Minimum allowable thickness threshold for coating (h) min ): Classified by plating material - tin plating h min =2~5μm, gold plating h min =0.5~2μm, silver plating h min =1~3μm, this threshold is the minimum thickness requirement for the coating to ensure the electrical performance and corrosion resistance of components. When h i ≤h min At that time, there is a risk of component failure;

[0083] Loss rate threshold (Δh) rate,th (The default reference values ​​are set according to the material: tin plating ≤ 0.2μm / 100h, gold plating ≤ 0.05μm / 100h, silver plating ≤ 0.1μm / 100h. The wear rate refers to the plating wear amount standardized as "per 100 hours", and is calculated as follows: Let the duration of the i-th time period be Δt.) i (Unit: h, e.g., Δt for 0~100h) i =100h), then (Unit: μm / 100h), where Δh iThe amount of loss in the i-th time period is calculated by first calculating the loss per unit time (per hour), and then multiplying it by 100 to convert it to the loss per 100 hours. This indicator is used to identify the risk of accelerated wear of the coating.

[0084] It should be further explained that the method for determining the early warning conditions and the graded response is as follows:

[0085] Warning Condition Determination: An early warning is triggered if any of the following conditions are met. The determination logic is comprehensive and complementary: Real-time thickness condition: The real-time thickness hi ≤ hmin in the i-th time period (the minimum allowable thickness of the corresponding material); Deviation condition: Theoretical deviation D ≥ Dth (≥5%); Loss rate condition: Coating loss rate Δh rate,i ≥Δh rate,th (Exceeding the preset maximum allowable loss rate);

[0086] Early warning information is tiered and responded to: Early warnings are divided into two levels based on the severity of the risk, ensuring that the response is accurately tailored to the risk level.

[0087] Level 1 Warning (General Risk): If only the "deviation condition" or "loss rate condition does not exceed the threshold by 1.5 times" is met, it indicates that the prediction accuracy is insufficient or the loss is slightly accelerated, with no immediate risk of failure.

[0088] Level 2 Warning (Severe Risk): If the "real-time thickness condition" or "loss rate condition exceeds the threshold by 1.5 times" is met, it means that the coating can no longer guarantee the normal operation of the components, and there is an immediate risk of failure.

[0089] It should be further noted that some experimental data are shown in the table below:

[0090]

[0091] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0092] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A system for detecting the thickness of plating on electronic component leads and predicting losses, characterized in that, include: Sample screening module: used to determine the initial coating thickness of electronic components in the same batch and screen valid samples according to sampling rules; Eddy current real-time non-destructive thickness measurement module: Used to detect the thickness of the pin plating of valid samples at different stages of use in real time using eddy current non-destructive testing technology, and output real-time thickness data; Multi-dimensional cause acquisition module: used to synchronously collect multi-dimensional parameters that affect coating loss, and calculate multi-dimensional cause indicators; The multi-dimensional causative indicators include temperature difference values, environmental corrosion degree indicators, and computational quantity difference values. Loss and thickness prediction module: Based on the quantitative relationship between thickness data and multi-dimensional inducing indicators, a thickness loss change model is constructed and thickness prediction values ​​are generated; The generation of the thickness prediction value first involves loss prediction. Specifically, the prediction causative index is obtained by extrapolating the trends of multi-dimensional causative indexes from the previous three time periods. In conventional scenarios, linear extrapolation is used. When the growth rate of the causative indexes in two consecutive adjacent time periods is ≥20%, exponential smoothing is used. The deviation between the prediction causative indexes and the actual values ​​in the most recent time period is controlled to be ≤30%. The prediction causative indexes are then substituted into the thickness loss change model to calculate the predicted loss amount. Finally, the real-time thickness in the i-th time period is used as a benchmark, and the predicted loss amount is subtracted to obtain the thickness prediction value for the (i+1)-th time period. Deviation Calculation and Early Warning Module: Based on the thickness prediction value, the theoretical deviation is calculated, compared with the multi-dimensional early warning threshold, and early warning information is output.

2. The electronic component pin plating thickness detection and loss prediction system according to claim 1, characterized in that: The sampling rule is to select electronic components of the same production batch and model as initial samples, use eddy current sensors to measure the initial pin plating thickness of all initial samples, calculate the arithmetic mean of the initial plating thickness of all initial samples and the initial thickness deviation rate of a single sample, and screen samples with a deviation rate in the range of -0.5% to +0.5% as valid samples, and remove samples with a deviation rate exceeding this range.

3. The electronic component pin plating thickness detection and loss prediction system according to claim 1, characterized in that: The eddy current nondestructive testing technology uses a high-frequency eddy current sensor with a detection accuracy of ≤0.01μm and a response time of ≤10ms as the core detection element, and is equipped with a sensor fixing mechanism, a signal conditioning unit, and a data acquisition terminal with a sampling rate of ≥100Hz. Before testing, the sensor is calibrated with a standard coating test block. During testing, the sensor probe is kept at a constant distance of 2mm from the pin surface. Thickness data is collected three times for a single pin, and the arithmetic mean of the three data is taken as the thickness data of the pin. The average thickness of all pins in the same sample is taken as the real-time thickness data of the sample at the current stage.

4. The electronic component pin plating thickness detection and loss prediction system according to claim 1, characterized in that: The temperature difference value is calculated using a single preset usage time period as the unit. The temperature data of the pin within this time period is collected by a contact temperature sensor, and the maximum and minimum temperature values ​​are extracted. The difference between the two values ​​is the temperature difference value for this time period.

5. The electronic component pin plating thickness detection and loss prediction system according to claim 1, characterized in that: The environmental corrosion degree index is calculated using humidity, pH value, and harmful gas concentration. The sum of the weighting coefficients for humidity, pH value, and harmful gas concentration is 1, and the reference weighting coefficients are 0.4, 0.3, and 0.3, respectively. The value range of the environmental corrosion degree index is 0 to 1. When the calculated result is less than 0, it is corrected to 0; when it is greater than 1, it is corrected to 1.

6. The electronic component pin plating thickness detection and loss prediction system according to claim 1, characterized in that: The calculation difference value is calculated using a single preset usage time period as the calculation unit. The operating current and clock frequency of the components are collected by the calculation monitoring unit and converted into real-time calculation quantity according to a fixed coefficient. The maximum and minimum values ​​of the real-time calculation quantity within the time period are extracted, and the difference between the two is the calculation difference value for that time period.

7. The electronic component pin plating thickness detection and loss prediction system according to claim 1, characterized in that: The thickness loss variation model is a multiple linear regression model, with thickness loss as the dependent variable and multi-dimensional inducing indicators as independent variables. The least squares method is used to fit the regression coefficients. The training set must meet the following requirements: the number of effective samples must be ≥30 and each sample must have ≥5 data points for each time period. If there are insufficient samples, the data of samples from the same batch and under the same working conditions will be merged.

8. The electronic component pin plating thickness detection and loss prediction system according to claim 1, characterized in that: The theoretical deviation is obtained by comparing the real-time thickness data of the i-th time period with the predicted thickness value of that time period, calculating the ratio of the absolute difference between the two to the average of the two and converting it into a percentage. This method quantifies the degree of deviation between the actual thickness and the predicted thickness.

9. The electronic component pin plating thickness detection and loss prediction system according to claim 1, characterized in that: The multi-dimensional early warning thresholds include the theoretical deviation threshold, the minimum allowable thickness threshold for coatings classified by coating material, and the loss rate threshold.

Images