An automatic optical testing device for computer motherboards

Abstract

The application relates to the technical field of automatic optical detection, and discloses a computer mainboard automatic optical testing device which comprises an excitation light source module, a multispectral sensor, a reference light power detector, a power supply module and a main control unit. The main control unit dynamically adjusts the driving current to stabilize the light source through a closed loop algorithm based on the measured light power; the core junction temperature of the light source is solved in real time through a discrete integral model, and the amount of red shift of the central wavelength is deduced; the instantaneous physical temperature of the target surface is deduced in combination with the ambient reference temperature and the target surface heat dissipation model. The device decouples and calculates the corrected double-band fluorescence intensity ratio by combining the target surface heat quenching effect and the amount of wavelength red shift. Finally, the local variance of the ratio is calculated by using a time sliding window, the instantaneous output light power is used to dynamically modulate the penalty weight, and a comprehensive defect recognition characteristic index is constructed to make a defect judgment. The application realizes dynamic decoupling of the interference of multiple physical fields, and significantly improves the robustness and precision of defect recognition.

Description

Technical Field

[0001] This invention belongs to the field of automatic optical inspection technology, and in particular relates to an automatic optical testing device for computer motherboards. Background Technology

[0002] In the electronics manufacturing industry, automated optical inspection technology is widely used for non-destructive evaluation of coating quality and microscopic defects on computer motherboards. Existing testing systems typically use fixed-power ultraviolet light-emitting diodes (UV LEDs) as excitation sources, acquire fluorescence images of the target surface using an industrial camera, and rely on preset static intensity thresholds or image template comparisons for defect determination. In terms of hardware architecture, most devices employ open-loop constant current driving or low-frequency photoelectric feedback loops, based on the assumption that the light source output characteristics and the target surface physical state remain absolutely stable during testing. However, this assumption has fundamental flaws in actual high-frequency continuous inspection scenarios. Prolonged operation of the excitation source leads to continuous accumulation of core junction temperature, causing high-frequency transient fluctuations in output optical power and a redshift of the center wavelength. Simultaneously, the physical temperature of the motherboard target surface rises rapidly after absorbing radiation energy, causing thermal quenching of the fluorescent material and nonlinear decay of radiation intensity. Existing technologies treat thermodynamic evolution and optical signal distortion as independent processes, lacking the ability to dynamically model multi-physics coupling interference and failing to establish a correlation mechanism between light source thermal drift and target surface temperature changes. This fragmented processing makes it difficult for the system to distinguish between heat-induced signal drift and real physical defects, resulting in frequent false alarms and missed detections during high-speed scanning on the production line, which seriously restricts the reliability of detection and production efficiency. Summary of the Invention

[0003] The purpose of this invention is to provide an automatic optical testing device for computer motherboards, which aims to solve the above-mentioned problems.

[0004] This invention is implemented as follows: an automatic optical testing device for computer motherboards includes: an excitation light source module, a multispectral sensor, a reference optical power detector thermally coupled to the excitation light source module, a power supply module, and a main control unit. The main control unit is configured to perform the following steps: based on the deviation between the measured instantaneous output optical power and a preset target optical power value, a drive current adjustment is generated through a closed-loop control algorithm to control the power supply module; the instantaneous forward current and voltage of the excitation light source module are collected, and combined with the measured instantaneous output optical power at the previous discrete moment, the instantaneous core junction temperature of the excitation light source module is calculated based on a discrete integral model incorporating steady-state thermal resistance, and the results are deduced. The instantaneous center wavelength redshift is used as the thermodynamic input. Combined with the ambient reference temperature, the instantaneous physical temperature of the motherboard optical test target is deduced based on the target surface heat dissipation model. The original radiation intensity of two independent characteristic fluorescence bands is collected. Combined with the thermal quenching effect generated by the instantaneous physical temperature of the target surface and the instantaneous center wavelength redshift, the corrected ratio of the two-band fluorescence intensity is decoupled and calculated. The local variance of the ratio of the two-band fluorescence intensity within the time sliding window is calculated. The penalty weight of the local variance is dynamically modulated by the instantaneous output optical power to construct a comprehensive defect identification feature index and compare it with a preset threshold to determine whether there is a defect in the test area.

[0005] A further technical solution involves iteratively calculating the instantaneous physical temperature of the target surface by constructing a first-order inertial model of target surface heat dissipation. Specifically: the temperature difference between the instantaneous physical temperature of the target surface and the ambient reference temperature at the previous discrete moment is exponentially decayed according to the equivalent thermal time constant of the test target surface region to obtain the residual temperature rise; the measured instantaneous output optical power at the current moment is combined with the coupling coefficient between the ultraviolet absorptivity and steady-state thermal resistance of the test target surface to calculate the additional temperature rise caused by photoinduced absorption; the ambient reference temperature, the residual temperature rise, and the additional temperature rise are added together to obtain the instantaneous physical temperature of the target surface at the current moment.

[0006] A further technical solution involves the main control unit adaptively acquiring the equivalent thermal time constant of the test target area and the coupling coefficient between the ultraviolet absorptivity and steady-state thermal resistance of the test target area through a pre-calibration program before formal optical testing. The pre-calibration program includes: driving the excitation light source module to continuously irradiate the test target area with a preset constant test light power to form a thermal step excitation and record the temperature rise response curve of the test target area; extracting the time consumed when the temperature rise reaches a specific response ratio threshold of the steady-state extreme temperature rise as the thermal time constant; and extracting the ratio of the steady-state extreme temperature rise to the constant test light power as the coupling coefficient.

[0007] In a further technical solution, the ambient reference temperature is a dynamically acquired real-time variable; the testing device also includes a background ambient temperature sensor arranged away from the radiation area of ​​the excitation light source module. The main control unit collects the raw data from the background ambient temperature sensor and introduces a digital low-pass filtering algorithm with a time constant greater than the thermal time constant for smoothing, so as to filter out high-frequency thermal disturbances during the test process, and uses the filtered output value as the ambient reference temperature to be extrapolated and updated at the current moment.

[0008] A further technical solution includes the following steps for decoupling and calculating the corrected dual-band fluorescence intensity ratio: calculating the ratio of the original radiation intensities of the first and second independent characteristic fluorescence bands collected by the multispectral sensor as the base ratio; constructing an exponentially mapped thermal quenching compensation factor based on the deviation between the instantaneous physical temperature of the target surface and the preset reference temperature, combined with the difference in nonradiative transition activation energy levels of the luminescent center and the Boltzmann constant; constructing a linearly mapped wavelength drift compensation factor based on the difference between the instantaneous center wavelength redshift and the nominal peak wavelength, combined with the spectral overlap coupling factor; and multiplying and weighting the base ratio, the thermal quenching compensation factor, and the wavelength drift compensation factor to obtain the corrected dual-band fluorescence intensity ratio.

[0009] A further technical solution involves establishing a discrete evolution model based on heat accumulation in the step of calculating the instantaneous core junction temperature: calculating the electrical power input corresponding to the product of the instantaneous forward current and the instantaneous voltage, subtracting the optical radiation dissipation corresponding to the instantaneous output optical power at the previous discrete moment, and the heat conduction loss obtained based on the difference between the core junction temperature and the measured temperature of the heat sink substrate, to obtain the net heat change rate; calculating the junction temperature change increment using the net heat change rate, equivalent heat capacity, and discrete sampling interval time, and superimposing it onto the instantaneous core junction temperature at the previous discrete moment; the step of estimating the instantaneous center wavelength redshift includes: based on the temperature difference between the calculated instantaneous core junction temperature and the preset reference temperature, performing linear drift compensation using the peak wavelength temperature drift coefficient of the excitation light source, and superimposing it onto the nominal peak wavelength.

[0010] A further technical solution includes the following steps for constructing a comprehensive defect identification feature index: performing a logarithmic transformation on the corrected dual-band fluorescence intensity ratio to obtain a baseline feature value; calculating the ratio of the measured instantaneous output optical power to the preset rated initial optical power as a dynamic attenuation modulation factor; multiplying the local variance of the dual-band fluorescence intensity ratio within a set time sliding window, the preset baseline variance penalty coefficient, and the dynamic attenuation modulation factor to construct a dynamic penalty term; and subtracting the dynamic penalty term from the baseline feature value to obtain the final comprehensive defect identification feature index.

[0011] A further technical solution is that the set time sliding window is implemented within the main control unit based on a first-in-first-out circular queue data structure; the local variance of the current time t is obtained using an online recursive statistical algorithm, that is, by using the local variance of the previous discrete time, combined with the latest dual-band fluorescence intensity ratio newly added to the queue at the current time t, and subtracting the oldest dual-band fluorescence intensity ratio that has slid out of the queue, incremental iterative calculation is performed.

[0012] A further technical solution involves employing a discrete positional proportional-integral-derivative (PID) control strategy in the step of updating the target drive current of the power supply module. Specifically, this includes: calculating the difference between the preset target optical power value and the measured instantaneous output optical power at the current moment to obtain the current feedback deviation; calculating the proportional adjustment component of the current feedback deviation, the integral adjustment component obtained by summing the feedback deviations at each historical discrete moment, and the derivative adjustment component obtained by the difference between the current feedback deviation and the feedback deviation at the previous discrete moment; and weighting and summing the proportional adjustment component, integral adjustment component, and derivative adjustment component according to their respective gain coefficients to obtain the absolute value of the target drive current output to the power supply module.

[0013] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0014] 1. This invention constructs a correction model that includes thermal quenching compensation and wavelength drift compensation by real-time simulation of the instantaneous physical temperature of the target surface and the instantaneous redshift of the center wavelength of the excitation source. This scheme can accurately separate the effects of environmental thermal disturbances and light source characteristic drift from the fluorescence intensity ratio signal, effectively overcoming the difficulty of traditional static detection in distinguishing between real physical defects and thermo-optical drift interference, thereby significantly reducing the false alarm and false negative rates during high-speed scanning on the production line.

[0015] 2. This invention constructs a comprehensive defect identification feature index, which not only performs a logarithmic transformation on the corrected fluorescence intensity ratio to enhance stability, but also incorporates local variance within a time sliding window to capture local abrupt changes in the signal. Simultaneously, it adaptively and dynamically modulates the penalty weight of the local variance using instantaneous output optical power, effectively suppressing instantaneous fluctuation noise from the light source, and significantly enhancing the judgment accuracy and anti-interference capability after multi-dimensional feature fusion.

[0016] 3. At the hardware control level, this device employs a discrete positional proportional-integral-derivative (PID) control strategy, which can precisely and dynamically adjust the drive current of the power supply module based on the microsecond-level measured instantaneous output optical power deviation. This control mechanism completely eliminates the steady-state error of the system at its source and effectively suppresses high-frequency oscillations in optical power, providing a highly stable and reliable optical excitation foundation for subsequent complex physical model calculations and high-quality fluorescence signal acquisition. Attached Figure Description

[0017] Figure 1 This invention provides an overall hardware structure block diagram of an automatic optical testing device for computer motherboards;

[0018] Figure 2 The flowchart of the multi-physics field decoupling and defect testing method for the main control unit provided by the present invention is shown. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0020] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0021] like Figure 1 and Figure 2 As shown, an automatic optical testing device for a computer motherboard according to an embodiment of the present invention includes: an excitation light source module, a multispectral sensor, a reference optical power detector thermally coupled to the excitation light source module, a power supply module, and a main control unit. The excitation light source module emits ultraviolet light of a specific wavelength onto the area of ​​the computer motherboard under test to excite the surface material of the motherboard to produce a fluorescence response. This module typically consists of one or more ultraviolet light-emitting diodes (UV-LEDs), whose output optical power and center wavelength are affected by their operating temperature. The multispectral sensor collects the fluorescence signal generated on the motherboard surface after irradiation by the excitation light source module. This sensor can simultaneously receive and distinguish the radiation intensity from different characteristic fluorescence bands, providing raw data for subsequent fluorescence analysis. The reference optical power detector monitors the instantaneous output optical power of the excitation light source module in real time. This detector is typically thermally coupled to the excitation light source module to ensure that its measurement results accurately reflect the actual output state of the light source. The power supply module provides the required driving current and voltage to the excitation light source module. The output parameters of this module can be precisely controlled by the main control unit to adjust the optical power of the excitation light source module. The main control unit serves as the control core of the entire testing device, responsible for tasks such as data acquisition, model calculation, algorithm control, and defect detection. This unit is typically implemented using a high-performance microprocessor or digital signal processor.

[0022] The main control unit is configured to perform the following steps:

[0023] Based on the deviation between the measured instantaneous output optical power and the preset target optical power value, a drive current adjustment is generated through a closed-loop control algorithm to control the power supply module. The closed-loop control algorithm adjusts the input parameters in real time according to the deviation between the system output and the target value, so that the system output approaches the target value. In this embodiment, the algorithm is used to adjust the output of the power supply module to maintain the stability of the output optical power of the excitation light source module.

[0024] The instantaneous forward current and voltage of the excitation source module are collected, and combined with the measured instantaneous output optical power at the previous discrete time step, the instantaneous core junction temperature of the excitation source module is calculated based on a discrete integral model that incorporates steady-state thermal resistance, and its instantaneous center wavelength redshift is deduced. The discrete integral model is used to calculate the instantaneous core junction temperature inside the light source by performing discrete time step integral operations on the instantaneous electrical parameters and optical power of the excitation source module. This model considers the thermal characteristics of the light source, such as heat capacity and steady-state thermal resistance.

[0025] Using instantaneous output optical power as a thermodynamic input and combining it with the ambient reference temperature, the instantaneous physical temperature of the motherboard optical test target surface is deduced based on the target surface heat dissipation model. The target surface heat dissipation model describes the absorption, conduction, and dissipation processes of the radiated energy from the excitation light source module on the motherboard optical test target surface, thereby deducing the instantaneous physical temperature of the target surface. This model considers parameters such as the ambient reference temperature and the thermal time constant of the target surface.

[0026] The raw radiant intensities of two independent characteristic fluorescence bands are collected. Combined with the thermal quenching effect caused by the instantaneous physical temperature of the target surface and the instantaneous center wavelength redshift, the corrected dual-band fluorescence intensity ratio is calculated decoupled. The thermal quenching effect describes the physical phenomenon that the fluorescence quantum efficiency of fluorescent materials decreases with increasing temperature. An increase in the instantaneous physical temperature of the target surface leads to a decrease in fluorescence intensity, affecting the accuracy of defect identification. The center wavelength redshift describes the phenomenon that the peak emission wavelength of the excitation light source module shifts towards longer wavelengths as its core junction temperature increases. This wavelength shift changes the matching degree between the excitation spectrum and the absorption spectrum of the fluorescent material, thus affecting the fluorescence intensity. The dual-band fluorescence intensity ratio serves to form a relatively stable fluorescence characteristic index by comparing the radiant intensities of the two independent characteristic fluorescence bands. This ratio is typically used to eliminate common-mode interference such as light source intensity fluctuations, highlighting the inherent fluorescence characteristics of the material.

[0027] The local variance of the dual-band fluorescence intensity ratio within a time-sliding window is calculated. The penalty weight of this local variance is dynamically modulated using instantaneous output optical power to construct a comprehensive defect identification feature index, which is then compared with a preset threshold to determine the presence of defects in the test area. The time-sliding window extracts a fixed-length subset of the latest data from a continuous data stream for statistical analysis. This window slides over time to ensure that the statistical results reflect the local dynamic characteristics of the data. The local variance measures the fluctuation of the dual-band fluorescence intensity ratio within the time-sliding window. The magnitude of the variance reflects signal stability or abnormal changes. The penalty weight dynamically adjusts the influence of the local variance on the comprehensive defect identification feature index based on the instantaneous output optical power. Through dynamic modulation, defect identification can be adapted to signal fluctuations under different optical powers. The comprehensive defect identification feature index integrates the corrected dual-band fluorescence intensity ratio with the dynamically modulated local variance to form a single value that is sensitive to defects and robust to thermal and optical interference. This index is ultimately compared with a preset threshold to determine the presence of defects in the test area.

[0028] The main features of the above technical solution will be explained in more detail below:

[0029] This application proposes an automatic optical testing device for computer motherboards. Its operation first involves precise control of the output optical power of the excitation light source module. Specifically, a simple proportional control strategy can be employed: the main control unit generates a drive current adjustment based on the deviation between the measured instantaneous output optical power and a preset target optical power value, and sends this adjustment to the power supply module. Upon receiving this adjustment, the power supply module directly changes its output current, thereby affecting the output optical power of the excitation light source module. For example, when the measured optical power is lower than the target value, the main control unit calculates a positive deviation and generates an adjustment to increase the current; conversely, when the measured optical power is higher than the target value, it generates an adjustment to decrease the current.

[0030] Furthermore, to accurately grasp the operating status of the excitation source module, the main control unit needs to calculate its instantaneous core junction temperature and deduce the instantaneous center wavelength redshift. One implementation method is for the main control unit to continuously acquire the instantaneous forward current and voltage of the excitation source module and combine it with the instantaneous output optical power measured at the previous discrete moment. Subsequently, the main control unit can use a simplified thermal model, for example, assuming that the thermal resistance and thermal capacity of the excitation source module are fixed constants, to estimate the instantaneous core junction temperature at the current moment through a simple difference equation. Based on this, the instantaneous center wavelength redshift can be deduced through a pre-calibrated linear relationship that directly maps the change in core junction temperature to wavelength shift.

[0031] Simultaneously, to compensate for changes in the fluorescence signal caused by heat on the motherboard's target surface, it is necessary to estimate the instantaneous physical temperature of the motherboard's optical testing target surface. Specifically, the main control unit can use the instantaneous output optical power as a thermodynamic input, combined with a preset fixed ambient reference temperature. Based on a simplified target surface heat dissipation model, for example assuming a constant thermal conductivity coefficient between the target surface and the environment, the instantaneous physical temperature of the target surface is estimated through a simple energy balance equation. This model can ignore the dynamic changes in the target surface thermal time constant, providing only an approximate temperature estimate.

[0032] After obtaining the aforementioned key parameters, the main control unit needs to decouple and calculate the corrected dual-band fluorescence intensity ratio. Specifically, after the multispectral sensor acquires the original radiation intensities of two independent characteristic fluorescence bands, the main control unit can first calculate a simple ratio of these two original radiation intensities. Subsequently, to correct for the thermal quenching effect caused by the instantaneous physical temperature of the target surface, the main control unit can introduce a preset fixed temperature compensation coefficient to linearly correct the original ratio. Similarly, for the spectral overlap effect caused by the instantaneous redshift of the center wavelength, another preset fixed wavelength compensation coefficient can be introduced for linear correction. In this way, the original fluorescence intensity ratio can be initially decoupled and corrected.

[0033] Finally, to construct a comprehensive defect identification feature index, the main control unit needs to calculate the local variance of the dual-band fluorescence intensity ratio within a time-sliding window and dynamically modulate the penalty weight of the local variance using instantaneous output optical power. One implementation involves the main control unit maintaining a fixed-length first-in-first-out queue as the time-sliding window and calculating the sample variance of the corrected dual-band fluorescence intensity ratio within the window in real time. For the penalty weight, a simple linear function can be used to map the instantaneous output optical power to a penalty coefficient; for example, the higher the optical power, the larger the penalty weight. Subsequently, the corrected dual-band fluorescence intensity ratio is simply weighted and summed with the local variance modulated by this penalty weight to form a comprehensive defect identification feature index. This index is then compared with a preset fixed threshold to determine whether a defect exists in the test area.

[0034] This application further proposes a step for extrapolating the instantaneous physical temperature of the target surface, which involves iterative calculation using a first-order inertial model of target surface heat dissipation. Specifically: the temperature difference between the instantaneous physical temperature of the target surface and the ambient reference temperature at the previous discrete moment is used to exponentially decay the residual temperature rise according to the equivalent thermal time constant of the test target surface region; the measured instantaneous output optical power at the current moment is combined with the coupling coefficient between the ultraviolet absorptivity and steady-state thermal resistance of the test target surface to calculate the additional temperature rise caused by photoinduced absorption; the ambient reference temperature, the residual temperature rise, and the additional temperature rise are added together to obtain the instantaneous physical temperature of the target surface at the current moment; the formula for calculating the instantaneous physical temperature of the target surface is as follows:

[0035]

[0036] in, This represents the instantaneous physical temperature of the target surface at the current time t; Represents the previous discrete time. The instantaneous physical temperature of the target surface; Indicates the measured or preset environmental reference temperature; This represents the discrete sampling interval time in microseconds. This represents the equivalent thermal time constant of the test target area; This represents the coupling coefficient between the ultraviolet absorptivity of the test target surface and the steady-state thermal resistance. This represents the instantaneous output optical power measured at the current time t. A formula for calculating the instantaneous physical temperature of the target surface is derived, aiming to dynamically track the temperature change of the test target surface under the illumination of the excitation light source module through a discrete integral model. This formula considers the heat dissipation characteristics of the target surface itself (through the thermal time constant). And the absorption and thermal conversion efficiency of the excitation light power (via the coupling coefficient) ), and combined with the ambient reference temperature By combining the target surface temperature with the previous moment's temperature, the instantaneous physical temperature of the target surface at the current moment can be predicted in real time and accurately. This involves a discrete sampling interval at the microsecond level. This ensures precise capture of dynamic temperature changes.

[0037] The equivalent thermal time constant of the test target area Coupling coefficient between the UV absorbance of the test target surface and the steady-state thermal resistance The pre-calibration is adaptively obtained by the main control unit through a pre-calibration program before the formal optical test. The pre-calibration program includes: driving the excitation light source module to a preset constant test light power. The test target surface is continuously irradiated to form a thermal step excitation, and the temperature rise response curve of the test target surface is recorded. The specific response ratio threshold of the temperature rise to the steady-state extreme temperature rise is then determined. The time consumed during this process is extracted as the thermal time constant. The steady-state extreme temperature rise and constant test optical power were compared. The ratio is extracted as the coupling coefficient. Test the equivalent thermal time constant of the target surface region. The coupling coefficient between the UV absorbance of the test target surface and the steady-state thermal resistance. These are key parameters characterizing the thermal properties of the test target surface. The adaptive acquisition of these parameters means that the system can be customized according to the actual motherboard material properties being tested, rather than using fixed values. The pre-calibration program records the temperature rise response of the target surface by simulating thermal excitation under actual working conditions, and then extracts these parameters. For example, when recording the temperature rise response curve, an infrared thermal imager or a high-precision thermocouple array can be used to measure the target surface temperature non-contactly or in contact.

[0038] The ambient reference temperature The test apparatus also includes a background ambient temperature sensor located away from the radiation area of ​​the excitation source module. The main control unit collects the raw data from the background ambient temperature sensor and introduces a time constant greater than the thermal time constant. A digital low-pass filter algorithm is used for smoothing to remove high-frequency thermal disturbances during the test, and the filtered output value is used as the current ambient temperature for prediction and updating. Ambient reference temperature The background ambient temperature is set as a dynamically acquired real-time variable to reflect the actual temperature fluctuations of the test environment. To accurately acquire this dynamic variable, the test setup is equipped with a background ambient temperature sensor. This sensor can be of various types, such as a thermistor, platinum resistance thermometer (RTD), or integrated digital temperature sensor. Its placement is far from the direct radiation area of ​​the excitation light source module to ensure that the measured temperature is the true background ambient temperature, rather than the temperature affected by localized heating from the light source. The main control unit processes the raw data acquired by the sensor, introducing a time constant greater than the thermal time constant. The digital low-pass filtering algorithm can employ moving average filtering, exponentially weighted moving average (EWMA) filtering, or Butterworth filtering, among others. Its purpose is to smooth sensor data, filter out high-frequency thermal disturbances or transient noise that may occur during testing, and thus provide a stable and representative environmental reference temperature value.

[0039] The proposed solution achieves accurate dynamic tracking of the instantaneous physical temperature of the target surface by constructing a thermodynamic model based on discrete integrals. This model incorporates a thermal time constant. With coupling coefficient By establishing a mathematical relationship between target surface temperature change and optical power input, the system can calculate the target surface temperature rise in real time based on the measured optical power, overcoming the limitation of traditional static measurements that cannot reflect thermal inertia processes. In the pre-calibration program, the thermal time constant is adaptively obtained using the temperature rise response curve under constant optical power excitation. With coupling coefficient This design ensures that model parameters can be customized for different motherboard materials and thermal properties, eliminating calculation errors caused by material differences. At the ambient reference temperature... For data acquisition, data is collected using a background ambient temperature sensor located far from the radiation area, and combined with a time constant greater than the thermal time constant. The digital low-pass filtering algorithm effectively eliminates high-frequency thermal noise interference in the test environment. This processing method ensures the accuracy of the data used as a model benchmark. The values ​​exhibit high stability and representativeness, avoiding the instantaneous impact of environmental fluctuations on the temperature extrapolation process, thus providing an accurate temperature input for subsequent correction of the fluorescence intensity ratio. Through the above technical means, the scheme of this application can provide precise instantaneous physical temperature of the target surface, providing a reliable physical basis for subsequent correction of the dual-band fluorescence intensity ratio, and significantly improving the system's environmental adaptability and detection stability under complex working conditions.

[0040] As a specific implementation method, the main control unit can employ a high-performance embedded processor, such as a microcontroller based on the ARM Cortex-M series architecture, to achieve microsecond-level discrete sampling and real-time calculation. During the pre-calibration process, the main control unit drives the excitation light source module to... ( The test target surface is continuously irradiated with a constant optical power. The equivalent thermal time constant of the test target surface region is obtained through calibration using an infrared thermal imager. for The coupling coefficient between the ultraviolet absorptivity of the target surface and the steady-state thermal resistance was tested. for For the ambient reference temperature The current stable value is obtained after passing through a digital sensor and a low-pass filter. To further illustrate the real-time extrapolation process of the instantaneous physical temperature of the target surface, the following calculation example is given: A discrete sampling interval time is set. ( The target surface temperature derived from the previous discrete moment. The measured instantaneous output optical power at the current moment The main control unit substitutes the above data into the derivation formula: at this time, the exponential decay term... The residual temperature rise is The additional temperature rise caused by photoinduced absorption is Finally, the instantaneous physical temperature of the target surface at the current moment is obtained. The microsecond-level iteration result will be stored and used as input for subsequent thermal quenching compensation.

[0041] Through the above technical solution, this application can provide accurate instantaneous physical temperature of the target surface, effectively compensating for the significant localized fluorescence thermal quenching effect caused by the rapid rise in physical temperature of the motherboard target surface after radiation absorption. This provides an accurate temperature input for the subsequent decoupled calculation of the corrected dual-band fluorescence intensity ratio. This enables the system to more accurately distinguish between real physical defects and thermo-optical drift interference, significantly reducing false alarms and missed detections that may occur in high-speed scanning detection, and improving the detection accuracy and reliability of the computer motherboard automatic optical testing device.

[0042] This application further proposes a decoupled calculation method for correcting the dual-band fluorescence intensity ratio, comprising: calculating the ratio of the original radiation intensities of the first and second independent characteristic fluorescence bands acquired by the multispectral sensor as the base ratio; constructing an exponentially mapped thermal quenching compensation factor based on the deviation between the instantaneous physical temperature of the target surface and the preset reference temperature, combined with the difference in nonradiative transition activation energy levels of the luminescent center and the Boltzmann constant; constructing a linearly mapped wavelength drift compensation factor based on the difference between the instantaneous center wavelength redshift and the nominal peak wavelength, combined with the spectral overlap coupling factor; multiplying and weighting the base ratio, the thermal quenching compensation factor, and the wavelength drift compensation factor to obtain the corrected dual-band fluorescence intensity ratio; and calculating the formula for the decoupled calculation of the corrected dual-band fluorescence intensity ratio:

[0043]

[0044] in, This represents the ratio of the two-band fluorescence intensities after correction at the current time t; and These represent the original radiative intensities of the first and second independent characteristic fluorescence bands acquired by the multispectral sensor, respectively. This represents the difference in nonradiative activation energy levels between the luminescent centers corresponding to two characteristic fluorescence bands; Represents the Boltzmann constant; This represents the instantaneous physical temperature of the target surface at the current time t; This indicates the preset reference temperature; Indicates the spectral overlap coupling factor; This indicates the instantaneous redshift of the center wavelength of the excitation source; Indicates the nominal peak wavelength at the reference temperature. Corrected dual-band fluorescence intensity ratio. This is the ratio of fluorescence intensity after correction for thermal quenching and redshift of the excitation source wavelength. Its purpose is to provide a characteristic indicator that accurately reflects the coating quality of the motherboard and is unaffected by dynamic changes in the environment and light source. The original radiant intensity of the first and second independent characteristic fluorescence bands... and This refers to the unprocessed radiation intensity signals acquired by a multispectral sensor within two preset, independent fluorescence bands. These raw intensity signals are the fundamental data for fluorescence analysis. Multispectral sensors can be implemented in various ways. For example, an array sensor integrating multiple narrowband filters and corresponding photodetectors can be used, with each detector responsible for acquiring the fluorescence intensity of a specific band. Alternatively, a spectrometer based on grating or prism dispersion principles can be used, obtaining intensity information for specific bands by sampling or integrating the spectrum. The difference in nonradiative activation energy levels corresponding to the luminescence centers of the two characteristic fluorescence bands is also relevant. This is a key physical parameter characterizing the difference in nonradiative transition probabilities of fluorescent materials at different temperatures. Determining this parameter typically requires experimental measurement; for example, the fluorescence lifetime or quantum efficiency of the target fluorescent material can be tested at different temperatures and then fitted using the Arrhenius equation. Alternatively, it can be theoretically derived through quantum chemical calculations or spectroscopic analysis. Instantaneous physical temperature of the target surface. This refers to the actual physical temperature of the motherboard target surface at the current moment during optical testing. This temperature is a key factor affecting the fluorescent thermal quenching effect. Besides obtaining it through deduction by the main control unit based on the target surface heat dissipation model, it can also be obtained by real-time monitoring of the test target surface using a non-contact infrared thermometer, or by contact measurement using a miniature thermocouple or thermistor without affecting the optical testing. The preset reference temperature... This is a reference point used for calibrating and comparing fluorescence intensity ratios. This temperature can be set according to the specific application scenario; for example, it can be set to the standard laboratory testing environment temperature, or the typical temperature under stable operating conditions for the excitation source and target surface. Spectral overlap coupling factor. This is a coefficient characterizing the effect of the overlap between the excitation source spectrum and the absorption spectrum of the fluorescent material on the fluorescence excitation efficiency. This factor reflects the degree to which the wavelength shift of the excitation source affects the fluorescence signal intensity. Its determination usually requires experimental calibration. For example, the excitation spectra of the fluorescent material at different excitation wavelengths can be measured, and convolution calculations and fitting can be performed in conjunction with the actual spectral distribution of the excitation source; alternatively, it can be determined by changing the wavelength of the excitation source and observing the change in fluorescence intensity. The instantaneous center wavelength redshift of the excitation source is also considered. This refers to the actual peak output wavelength of the excitation source at the current moment. This wavelength will drift (redshift) as the junction temperature of the source increases, thus affecting the fluorescence excitation efficiency. Besides calculating and extrapolating the instantaneous core junction temperature of the excitation source module using a discrete integral model by the main control unit, the peak wavelength can also be directly measured or indirectly inferred by integrating a miniature spectrometer or using a narrowband filter array with a photodetector to monitor the output spectrum of the excitation source in real time. The nominal peak wavelength at the reference temperature. This refers to the ideal or factory-nominal peak wavelength of the excitation source at a preset reference temperature. This wavelength can usually be obtained by consulting the excitation source's product specifications, or by measuring it using a high-precision spectrometer under laboratory conditions, after stabilizing the excitation source at the reference temperature.

[0045] The proposed solution achieves dynamic calibration of the fluorescence intensity ratio by introducing a multi-physics coupling correction model, thereby effectively eliminating measurement errors caused by thermal effects and spectral drift. Specifically, the solution first utilizes the original radiant intensities of the first and second independent characteristic fluorescence bands acquired by a multispectral sensor. and The original ratio was calculated. Based on this, to compensate for the fluorescence thermal quenching effect caused by changes in the target surface temperature, a thermal quenching compensation term based on the Boltzmann distribution was introduced. This compensation term utilizes the instantaneous physical temperature of the target surface. With the preset reference temperature The reciprocal difference, combined with the difference in nonradiative activation energy levels of the luminescence centers corresponding to the two characteristic fluorescence bands. and Boltzmann constant This provides real-time compensation for fluorescence intensity attenuation caused by increased target surface temperature. (Target surface instantaneous physical temperature) The acquisition of the data is based on the fact that the main control unit in the above device uses the instantaneous output optical power as a thermodynamic input, combines it with the ambient reference temperature, and derives it based on the target surface heat dissipation model. This provides accurate real-time temperature data for thermal quenching compensation.

[0046] Furthermore, to correct for the effect of excitation source wavelength drift on fluorescence excitation efficiency, this scheme introduces a spectral overlap coupling correction term. This correction term utilizes the instantaneous redshift of the excitation source's center wavelength. Nominal peak wavelength at reference temperature The difference, multiplied by the spectral overlap coupling factor This compensates for the change in excitation efficiency caused by wavelength drift of the excitation source. The instantaneous redshift of the excitation source's center wavelength is... The acquisition of wavelength drift information is based on the fact that the main control unit in the above device collects the instantaneous forward current and voltage of the excitation light source module, combines it with the instantaneous output optical power measured at the previous discrete moment, calculates the instantaneous core junction temperature of the excitation light source module based on the discrete integral model that introduces steady-state thermal resistance, and then deduces the result. This provides real-time wavelength drift information for spectral drift correction.

[0047] By integrating the aforementioned thermal quenching compensation and spectral drift correction into a single calculation model, this scheme achieves decoupled calculation of the fluorescence intensity ratio. This approach organically combines the spectral characteristics of the light source with the fluorescence response characteristics, solving the problem that a single intensity measurement cannot distinguish between thermal quenching and excitation efficiency fluctuations. The aforementioned device, through its main control unit, can acquire the instantaneous physical temperature of the target surface in real time. and the instantaneous center wavelength redshift of the excitation source The introduction of these dynamic parameters enables this scheme to dynamically observe and decouple multi-physics coupling interference, thus providing a high-precision corrected dual-band fluorescence intensity ratio even in complex thermo-optical environments.

[0048] As a specific implementation, in an automated optical testing device for computer motherboards, the multispectral sensor can be a module integrating two independent photodiodes and corresponding narrowband filters. For example, the first filter has a center wavelength of 450 nm and a full width at half maximum (FWHM) of 10 nm, used to collect the raw radiation intensity of the first independent characteristic fluorescence band. The second filter has a center wavelength of 520 nm and a full width at half maximum (FWHM) of 10 nm, and is used to collect the raw radiance of the second independent characteristic fluorescence band. The choice of these two bands is based on the characteristic emission spectrum of the target fluorescent material.

[0049] Before testing, relevant parameters can be determined through a pre-calibration procedure. For example, the motherboard material sample to be tested is placed in a precisely temperature-controlled experimental chamber, and its fluorescence emission spectrum is measured using a high-precision spectrometer at different temperature points. This measurement is then combined with theoretical model fitting to determine the difference in nonradiative activation energy levels between the luminescence centers corresponding to two characteristic fluorescence bands. Meanwhile, by changing the output wavelength of the excitation source (e.g., using a tunable laser) and measuring the change in fluorescence intensity, the spectral overlap coupling factor can be determined. Preset reference temperature It can be set to 25°C, while the nominal peak wavelength is at the reference temperature. The excitation light source can be determined according to its manufacturer's specifications, for example, 365nm.

[0050] During actual testing, the main control unit will acquire the instantaneous physical temperature of the target surface in real time. and the instantaneous center wavelength redshift of the excitation source Assuming the current instantaneous physical temperature of the motherboard target surface is... (Absolute temperature 313.15K), the center wavelength of the excitation source is redshifted in real time. Raw radiation intensity acquired by the multispectral sensor For 1000 units, The value is 500 units. The difference between the non-radiative activation levels at the luminescent center is known. Boltzmann constant Reference temperature (298.15K), spectral overlap coupling factor Nominal peak wavelength Substituting into the formula, the reciprocal difference in temperature is... The thermal quenching index is Wavelength drift compensation factor is The base ratio Multiplying by the compensation factor yields the corrected dual-band fluorescence intensity ratio. .

[0051] Through the above technical solution, this application can effectively solve the problems in traditional solutions, such as the thermal quenching effect of the excitation source on the motherboard target surface and the redshift of the excitation source wavelength, which cause the original fluorescence intensity ratio to fail to accurately reflect the motherboard coating quality and the system to have difficulty distinguishing between real physical defects and thermal-optical drift interference. This solution introduces a thermal quenching compensation term based on Boltzmann distribution, using the real-time acquired instantaneous physical temperature of the target surface to accurately compensate for the fluorescence intensity attenuation caused by temperature rise. Simultaneously, by introducing a spectral overlap coupling correction term, using the real-time derived instantaneous center wavelength redshift of the excitation source, it effectively corrects for the excitation efficiency change caused by the wavelength drift of the source. This multi-physics coupled dynamic calibration model ensures that the final corrected dual-band fluorescence intensity ratio... It can accurately decouple the interference of thermal effects and spectral drift, thus truly reflecting the physical state of the motherboard surface. This significantly improves the accuracy and robustness of defect identification, avoids false alarms and missed detections caused by thermal-optical drift in high-speed scanning, and provides high-precision feature input for subsequent defect identification.

[0052] This application further proposes that in the step of calculating the instantaneous core junction temperature, a discrete evolution model based on heat accumulation is established: the electrical power input corresponding to the product of instantaneous forward current and instantaneous voltage is calculated, the optical radiation dissipation corresponding to the instantaneous output optical power at the previous discrete moment is subtracted, and the heat conduction loss is obtained based on the difference between the core junction temperature and the measured temperature of the heat sink substrate, to obtain the net heat change rate; the junction temperature change increment is calculated using the net heat change rate, equivalent heat capacity, and discrete sampling interval time, and is superimposed on the instantaneous core junction temperature at the previous discrete moment; the step of estimating the instantaneous center wavelength redshift includes: based on the temperature difference between the calculated instantaneous core junction temperature and the preset reference temperature, linear drift compensation is performed using the peak wavelength temperature drift coefficient of the excitation light source, and superimposed on the nominal peak wavelength; the steps of calculating the instantaneous core junction temperature and estimating the instantaneous center wavelength redshift include the following calculation formulas:

[0053]

[0054]

[0055] in, and Representing the current time t and the previous discrete time respectively The instantaneous core junction temperature; This indicates the measured temperature of the heat sink substrate; This represents the discrete sampling interval time in microseconds. and These represent the equivalent heat capacity and steady-state thermal resistance of the excitation source, respectively. and These represent instantaneous forward current and instantaneous forward voltage, respectively. This represents the instantaneous output optical power measured at the previous discrete moment; This represents the instantaneous redshift of the center wavelength at the current time t; Indicates the nominal peak wavelength at the reference temperature; The temperature drift coefficient representing the peak wavelength of the excitation source; This indicates the preset reference temperature.

[0056] The steps of calculating the instantaneous core junction temperature and deducing the instantaneous center wavelength redshift aim to obtain the most critical thermal state parameter inside the excitation source—the core junction temperature—in real time and dynamically, and based on this, deduce the drift of its optical output characteristics (center wavelength). Its purpose is to provide accurate physical quantification for subsequent fluorescence signal correction, thereby eliminating optical distortion caused by the thermal effects of the source itself. This step typically relies on an embedded processor or digital signal processor (DSP) to perform high-speed calculations on the real-time acquired data. Instantaneous core junction temperature This refers to the actual operating temperature of the semiconductor PN junction inside the excitation light source at the current moment. It is a core factor affecting key parameters such as the light source's photoelectric conversion efficiency, output optical power, and center wavelength, and is fundamental to understanding and compensating for light source performance drift. The instantaneous core junction temperature at the previous discrete moment... Refers to the discrete sampling interval time at the microsecond level Previously, the core junction temperature of the light source was used as the starting point for iterative calculations in the discrete integral model. The measured temperature of the heat sink substrate... This refers to the actual temperature of the heat sink substrate in close contact with the excitation source at any given moment. This temperature can be measured in real time using thermocouples, thermistors, or infrared temperature sensors. The discrete sampling interval is in the microsecond range. The aim is to capture the rapidly changing thermal dynamics of a light source operating at high speeds. This involves exciting the equivalent heat capacity of the light source. This is a physical quantity describing the ability of an excitation source to increase in temperature when absorbing heat; its value can be obtained through experimental calibration or by consulting the device datasheet. The steady-state thermal resistance of the excitation source. This describes the resistance to heat conduction from the core junction to the heat sink substrate within the excitation source; its value can also be obtained through experimental calibration or by consulting the device datasheet. Instantaneous forward current. and instantaneous positive voltage These refer to the current flowing through the excitation source and the voltage applied across the excitation source at the current moment, respectively, which can be acquired in real time using high-precision current and voltage sensors. The instantaneous output optical power measured at the previous discrete moment... In the calculation of core junction temperature, it is used to subtract the portion of electrical input power that is converted into light energy. Instantaneous center wavelength redshift. The wavelength deviation of the light actually emitted by the excitation source at the current moment relative to its nominal wavelength is a key indicator for quantifying the optical output distortion of the light source. The nominal peak wavelength is at the reference temperature. As a reference for wavelength redshift. Peak wavelength temperature shift coefficient of the excitation source. The sensitivity of the peak wavelength of the excitation source to changes in core junction temperature is described, and is typically obtained through experimental calibration. A preset reference temperature is used. This refers to a stable temperature point set when calibrating the characteristics of a light source.

[0057] This scheme achieves real-time dynamic monitoring of the core junction temperature and wavelength drift of the excitation source by constructing a thermodynamic model, providing a physical basis for subsequent optical signal correction. The core of this scheme lies in the fact that the main control unit is configured to acquire the instantaneous forward current of the excitation source module in real time. With instantaneous positive voltage And combined with the instantaneous output optical power measured at the previous discrete moment and the measured temperature of the heat sink substrate By introducing steady-state thermal resistance and equivalent heat capacity The discrete integral model is used to accurately calculate the instantaneous core junction temperature of the excitation source module. Specifically, this discrete integral model considers the electrical input power of the light source. Subtract the instantaneous output optical power converted into light energy This yields power converted into heat energy. Simultaneously, the heat conduction between the core junction and the heat sink substrate is considered. By relating these thermal power terms to the equivalent heat capacity of the light source... Combined, and with discrete sampling intervals in the microsecond range By performing integral iterations, the thermal state changes of the light source within an extremely short time can be captured, overcoming the limitation of traditional static assumptions that fail to detect the internal thermal state of the light source. Based on this, once the precise instantaneous core junction temperature is obtained... The main control unit utilizes the peak wavelength temperature drift coefficient of the excitation light source. and the nominal peak wavelength at the reference temperature Through linear relationship Deducing the instantaneous redshift of the center wavelength at the current moment. This step directly correlates the electrical parameters of the light source with its optical output characteristics, enabling the system to quantify wavelength shifts caused by temperature fluctuations in real time. This lays the foundation for subsequent elimination of measurement errors caused by wavelength redshift. This scheme introduces equivalent heat capacity. With steady-state thermal resistance These parameters enable the model to accurately describe the thermal equilibrium process inside the light source, ensuring the stability and accuracy of junction temperature calculations under complex operating conditions. In this way, the optical testing device can fundamentally decouple thermal effects from optical signals, providing crucial input parameters for subsequent correction of the dual-band fluorescence intensity ratio.

[0058] As one specific implementation method, the above process can be implemented by a high-performance microcontroller or FPGA in the main control unit, at microsecond intervals. ( The process is repeated cyclically. Assume the system has a calibrated equivalent heat capacity for the excitation source. Steady-state thermal resistance Peak wavelength temperature drift coefficient , , At the current time t, a forward current is sampled. Forward voltage Measured temperature of heat sink substrate The optical power at the previous moment The temperature was set a moment ago. The heat loss through conduction is Net heat power is The increment of junction temperature change is The instantaneous core junction temperature is obtained after superposition. Deducing the instantaneous center wavelength redshift .

[0059] Through the above technical solution, this application can calculate the instantaneous core junction temperature of the excitation source in real time and accurately, and deduce its instantaneous center wavelength redshift. This overcomes the limitations of traditional solutions, which cannot sense the internal thermal state of the light source and are difficult to quantify the impact of wavelength drift on the fluorescence signal. By dynamically acquiring the true optical output characteristics of the light source, this solution provides a solid physical basis for subsequent decoupling correction of the thermal quenching effect and wavelength redshift of the fluorescence signal, thereby significantly improving the accuracy and reliability of the automatic optical testing device for computer motherboards in high-speed detection, and effectively avoiding false alarms and missed detections caused by the thermal and optical drift of the light source itself.

[0060] This application further proposes the following steps for constructing a comprehensive defect identification feature index: performing a logarithmic transformation on the corrected dual-band fluorescence intensity ratio to obtain a baseline feature value; calculating the ratio of the measured instantaneous output optical power to the preset rated initial optical power as a dynamic attenuation modulation factor; multiplying the local variance of the dual-band fluorescence intensity ratio within a set time sliding window, the preset baseline variance penalty coefficient, and the dynamic attenuation modulation factor to construct a dynamic penalty term; subtracting the dynamic penalty term from the baseline feature value to obtain the final comprehensive defect identification feature index; and constructing the calculation formula for the comprehensive defect identification feature index:

[0061]

[0062] in, This represents the comprehensive defect identification feature index; This represents the ratio of the two-band fluorescence intensities after correction at the current time t; This represents the preset benchmark variance penalty coefficient; This represents the local variance of the ratio of the two-band fluorescence intensities within a set time sliding window; This represents the instantaneous output optical power measured at the current time t; This indicates the rated initial optical power at the reference temperature.

[0063] The set time sliding window is implemented within the main control unit based on a first-in-first-out circular queue data structure; the local variance of the current time t. The data is obtained using an online recursive statistical algorithm, that is, by utilizing the previous discrete time step. Local variance The incremental iterative calculation is performed by combining the latest dual-band fluorescence intensity ratio of the newly added members at the current time t and subtracting the oldest dual-band fluorescence intensity ratio that has slipped out of the queue.

[0064] Comprehensive Defect Identification Feature Indicators This is a composite indicator designed to more robustly characterize the presence of defects on the motherboard's optical test surface by integrating multiple signal characteristics, including the corrected fluorescence intensity ratio, its local variance, and instantaneous output optical power. This indicator is calculated in real-time by a digital signal processor (DSP) or microcontroller (MCU) within the main control unit, involving basic mathematical operations such as floating-point arithmetic, logarithmic arithmetic, and multiplication and subtraction. The corrected dual-band fluorescence intensity ratio... This is the intensity ratio of two independent characteristic fluorescence bands after compensation for thermal quenching and the redshift of the excitation source's center wavelength. It provides pure fluorescence signal characteristics directly related to defect type and concentration, and has been stripped of the effects of environmental thermal disturbances and light source characteristic drift. This ratio is calculated by the main control unit based on the raw radiation intensity collected by the multispectral sensor, combined with the instantaneous physical temperature of the target surface and the instantaneous redshift of the excitation source's center wavelength, using a preset compensation model. A preset baseline variance penalty coefficient is also included. This is an adjustable parameter used to control the weight or influence of the local variance term in the comprehensive defect identification feature index. This coefficient can be set via a calibration program and stored in the non-volatile memory of the main control unit before the device leaves the factory, or it can be fine-tuned on-site via a user interface, for example, by testing and optimizing on known defect samples. The local variance of the dual-band fluorescence intensity ratio within a set time sliding window. This method measures the fluctuation of the corrected dual-band fluorescence intensity ratio over a recent time period, aiming to capture local instability in the signal. When defects exist in the test area, the local uniformity of the fluorescence signal may be disrupted, leading to increased ratio fluctuations. This change in variance indicates the defect. This can be achieved by maintaining a fixed-size circular buffer in the main control unit to store the N most recent corrected dual-band fluorescence intensity ratio data points. When a new data point arrives, it is added to the end of the queue, while the oldest data point is removed. Furthermore, the local variance can be calculated using an online recursive statistical algorithm, such as Welford's online algorithm or a similar recursive formula, by updating the accumulated value of the mean and sum of squares to efficiently calculate the new variance. Instantaneous output optical power. This is the actual optical power emitted by the excitation light source module at the current moment, measured in real time by the reference optical power detector. The measurement result is transmitted to the main control unit and used to dynamically modulate the penalty weight of the local variance to compensate for changes in signal variance caused by fluctuations in the light source power. The rated initial optical power at the reference temperature. It serves as a reference for instantaneous output optical power and is used to calculate the dynamic modulation factor. This value is determined and stored during the device design or calibration phase.

[0065] The solution proposed in this application constructs a comprehensive defect identification feature index. This method aims to overcome the insufficient robustness of traditional methods in defect identification during high-speed dynamic detection due to light source fluctuations and environmental disturbances. The construction of this index first involves compensating for the thermal quenching effect and the redshift of the excitation source center wavelength by the ratio of dual-band fluorescence intensity. A logarithmic transformation is performed to effectively compress the dynamic range of the signal and enhance its stability. Based on this, the local variance of this ratio within a set time sliding window is introduced. As a measure of signal volatility, when defects exist on the motherboard target surface, the uniformity of the fluorescence signal in the local area will be affected, leading to... Increased volatility, i.e. Increase. To further improve the accuracy of recognition, this scheme will increase the instantaneous output optical power. Rated initial optical power at reference temperature The ratio of [value] to [value] is used as a dynamic modulation factor to adjust the penalty weight of local variance in real time. This means that when the actual output optical power of the excitation source deviates from the rated value, the system can adaptively adjust its sensitivity to signal fluctuations, thereby avoiding misjudgments caused by fluctuations in source power. For example, when the source power is high, even small defects can cause relatively large signal fluctuations. In this case, by increasing the penalty weight, these defects can be captured more sensitively; conversely, when the source power is low, the penalty weight is reduced accordingly to avoid misjudging normal fluctuations as defects. At the implementation level, the main control unit uses a first-in-first-out circular queue data structure to manage data within a set time sliding window, ensuring the real-time performance and efficiency of the data flow. Meanwhile, local variance [is also considered]. The calculation employs an online recursive statistical algorithm, which uses the variance from the previous discrete time step, combined with newly added data at the current time step, and subtracts old data that has slipped out of the queue, to perform incremental iterative calculations. This calculation method significantly reduces the computational resources and time required for each variance calculation, meeting the stringent requirements of real-time detection in high-speed production lines. Through this multi-dimensional feature fusion and dynamic weighting mechanism, this solution can effectively distinguish between real physical defects and thermal-optical drift interference, significantly improving the anti-interference capability and judgment accuracy of defect identification.

[0066] As a specific implementation method, the main control unit uses its internal floating-point arithmetic unit to calculate the comprehensive defect identification feature index in real time. Assuming the local variance within the currently set time sliding window is obtained from the statistics of the circular queue inside the main control unit. The system presets a baseline variance penalty coefficient. Rated initial optical power The current time-corrected dual-band fluorescence intensity ratio Actual instantaneous output optical power The main control unit first obtains the baseline feature values: Calculate the dynamic attenuation modulation factor: Construct a dynamic penalty term: The subtraction operation yields the comprehensive defect identification feature index. This final indicator The system directly proceeds to the comparison with the judgment threshold (e.g., 0.500). Since 0.562 is greater than 0.500, it can be determined with high confidence whether there is a defect in the test area.

[0067] Through the above technical solution, this application effectively solves the problem that in high-speed dynamic detection, the fluorescence intensity ratio signal is mixed with a large amount of non-defect-related noise and dynamic drift due to instantaneous fluctuations in excitation light source power and environmental thermal disturbances. This makes it difficult to effectively distinguish between minute defects and system dynamic noise by relying solely on a single corrected ratio, resulting in insufficient robustness in identification. This solution introduces a comprehensive defect identification feature index, fusing the logarithmic transform value of the corrected dual-band fluorescence intensity ratio with the local variance of the ratio, and dynamically modulating the penalty weight of the local variance using instantaneous output optical power. This multi-dimensional feature fusion and dynamic weighting mechanism enables the system to capture defect features more comprehensively while effectively suppressing non-defect signal drift caused by light source fluctuations and environmental changes. In particular, the logarithmic transform compresses the signal dynamic range, enhancing signal stability; the local variance captures local discontinuities, separating defect features from background noise; and the instantaneous output optical power dynamically modulates the penalty weight, achieving adaptive compensation for light source power fluctuations and ensuring identification accuracy under different optical excitation intensities. Furthermore, the use of a first-in-first-out circular queue and an online recursive statistical algorithm ensures high-speed real-time detection with extremely low computational resource consumption, significantly improving the anti-interference capability, judgment accuracy, and overall robustness of defect identification, thereby effectively reducing the false alarm rate and false negative rate.

[0068] This application further proposes a discrete position-based proportional-integral-derivative (PID) control strategy in the step of updating the target drive current of the power supply module. Specifically, this includes: subtracting the preset target optical power value from the measured instantaneous output optical power at the current moment to obtain the current feedback deviation; calculating the proportional adjustment component of the current feedback deviation, the integral adjustment component obtained by summing the feedback deviations at each historical discrete moment, and the derivative adjustment component obtained by the difference between the current feedback deviation and the feedback deviation at the previous discrete moment; weighting and summing the proportional, integral, and derivative adjustment components according to their respective gain coefficients to obtain the absolute value of the target drive current output to the power supply module; and updating the drive current adjustment amount of the power supply module according to the deviation value calculated by the following formula:

[0069]

[0070]

[0071] in, and Representing the current time t and the previous discrete time respectively The optical power feedback deviation value; This indicates the preset optical power target value; This represents the instantaneous output optical power measured at the current time t; This represents the absolute value of the target drive current output to the power supply module at the current time t; , , These represent the proportional, integral, and differential gain coefficients, respectively. This represents the discrete sampling interval in microseconds; n represents the total number of discrete sampling steps corresponding to the current time; k represents the index of the cumulative discrete sampling step count from the initial time to the current time. This represents the optical power feedback deviation value at the k-th discrete time.

[0072] This scheme employs a discretized proportional-integral-derivative (PID) closed-loop control algorithm to dynamically adjust the drive current of the excitation light source module in real time and with precision, ensuring the stability of its output optical power. The core of this control mechanism lies in continuously monitoring the instantaneous output optical power of the excitation light source module. and compare it with the preset optical power target value. By comparing the values, the optical power feedback deviation at the current moment can be calculated. This deviation value represents the instantaneous output optical power measured at the current time t. Compared with the preset optical power target value The difference serves to quantify the deviation between the output optical power of the excitation light source module and the desired value, providing a basis for subsequent adjustment of the drive current. This deviation value can be obtained by the main control unit through direct subtraction or through differential amplification using analog circuitry.

[0073] The main control unit uses this deviation value, combined with the deviation value of the previous discrete time step. In addition to the historical accumulated deviation values, the absolute value of the target drive current required at the current moment is calculated using a PID control algorithm. This value is based on the optical power feedback deviation. The drive current, calculated using a proportional-integral-derivative (PID) control algorithm, is used to control the power supply module's supply to the excitation light source module. Its function is to dynamically adjust the drive current based on current, historical, and predicted deviations, so that the output optical power of the excitation light source module approaches and stabilizes at a preset target value. This calculation can be performed by a digital signal processor (DSP) or microcontroller (MCU) in the main control unit, or it can be implemented using a dedicated PID controller integrated circuit.

[0074] Specifically, the proportional gain coefficient Used to adjust the influence of the proportional term on the target drive current. The function of the proportional term is to immediately adjust the drive current according to the magnitude of the current deviation; the larger the deviation, the larger the adjustment, thus providing a fast response. The integral gain coefficient can be set based on empirical values, obtained through classic PID parameter tuning methods such as Ziegler-Nichols, or dynamically optimized during system operation using adaptive algorithms. Used to adjust the influence of the integral term on the target drive current. The function of the integral term is to accumulate historical deviations, eliminate steady-state errors of the system, and ensure that the optical power can remain stable near the target value over a long period of time. The setting of this value can also be based on empirical values ​​or obtained through system identification; its magnitude affects the speed at which the system eliminates steady-state errors and the amount of overshoot. Differential gain coefficient. Used to adjust the influence of the differential term on the target drive current. The function of the differential term is to make predictive adjustments based on the rate of change of the deviation, suppressing high-frequency oscillations in optical power and improving system stability. The setting can be based on empirical values ​​or obtained through system identification; its magnitude affects the system's ability to suppress disturbances and the smoothness of its response. The discrete sampling interval is in the microsecond range. The system defines the time period for optical power measurement, deviation calculation, and drive current update. The microsecond-level sampling interval ensures the system can capture and respond to rapid fluctuations in the output optical power of the excitation light source module in real time, thereby achieving high-precision dynamic control. This sampling interval can be precisely controlled by the main control unit's timer interrupt or the high-speed data acquisition module, for example, set to 1 microsecond, 10 microseconds, or 50 microseconds. The total number of discrete sampling steps *n* and the cumulative discrete step count index *k* are used for calculating the integral term, representing the cumulative number of sampling steps from the initial time to the current time. They ensure that the integral term accurately reflects the cumulative effect of historical deviations. These parameters are internally counted and indexed by the main control unit when executing the PID algorithm.

[0075] Calculated absolute value of target drive current This value is then sent to the power supply module, which precisely adjusts the current supplied to the excitation light source module based on this value. The entire process is performed with discrete sampling intervals on the order of microseconds. The cyclic operation ensures rapid and precise control of the output optical power of the excitation source module. This high-frequency dynamic adjustment enables the system to effectively address fluctuations in the output optical power of the excitation source module caused by factors such as rising core junction temperature and changes in ambient temperature, thus providing a highly stable optical excitation source at the source. Through this precise PID closed-loop control, the device of this application overcomes the limitations of traditional open-loop or simple feedback control in dealing with complex thermo-optical coupling effects. Working in conjunction with other modules in the device, it ensures that subsequent calculations of the instantaneous physical temperature of the target surface, the instantaneous center wavelength redshift, and the corrected dual-band fluorescence intensity ratio are all based on a stable and reliable optical excitation foundation, greatly improving the accuracy and robustness of defect identification.

[0076] The following is a concrete example to illustrate this. Assume the main control unit sets a discrete sampling interval time... The adjusted proportional coefficient Integral coefficient Differential coefficients At the current time t, the preset optical power target value is... Measured optical power Current deviation value Retrieve the deviation from the previous moment. The historical cumulative score is Calculate the three components separately: proportional component. Integral components ;differential component The absolute value of the target driving current is obtained by weighted summation. This value is then converted into a PWM duty cycle signal to drive the circuit to accurately compensate for the power of the light source.

[0077] Through the above technical solution, the device of this application can effectively solve the problem of nonlinear fluctuations in the output optical power of the excitation light source module caused by factors such as junction temperature changes in actual high-frequency optical testing scenarios, as well as the limitations of traditional control mechanisms in lacking refined and dynamic processing of optical power deviations. Specifically, a discrete PID closed-loop control algorithm is introduced, enabling the power supply module to accurately and dynamically adjust the drive current of the excitation light source module based on real-time, microsecond-level instantaneous output optical power feedback. The proportional term ensures a rapid response to optical power fluctuations, the integral term completely eliminates any potential steady-state errors in the system, ensuring that the optical power remains stable at the target value for a long period, while the derivative term effectively suppresses high-frequency oscillations and improves the dynamic stability of the system. This precise control mechanism fundamentally improves the stability and accuracy of the output optical power of the excitation light source module, enabling it to meet the stringent requirements of high-precision automatic optical testing for light source stability. When combined with other technical features in the device, this highly stable optical excitation provides a reliable foundation for subsequent complex physical model calculations. For example, when extrapolating the instantaneous physical temperature of the target surface, stable instantaneous output optical power... As a thermodynamic input, it makes the calculation results of the target surface heat dissipation model more accurate; when solving the instantaneous core junction temperature of the excitation source module and deducing the instantaneous center wavelength redshift, stability is achieved. This also ensures the input accuracy of the discrete integral model. Ultimately, this source-level optical power stability significantly reduces measurement errors introduced by light source fluctuations, resulting in a corrected dual-band fluorescence intensity ratio. It can more realistically reflect the defect information of the motherboard optical test target surface, thereby effectively improving the accuracy and reliability of the comprehensive defect identification feature index and reducing the risk of false alarms and missed detections.

[0078] 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, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An automatic optical testing device for computer motherboards, characterized in that, include: The system comprises an excitation source module, a multispectral sensor, a reference optical power detector thermally coupled to the excitation source module, a power supply module, and a main control unit; wherein the main control unit is configured to perform the following steps: Based on the deviation between the measured instantaneous output optical power and the preset optical power target value, a drive current adjustment amount is generated through a closed-loop control algorithm to control the power supply module; The instantaneous forward current and voltage of the excitation light source module are collected, and combined with the instantaneous output optical power measured at the previous discrete moment, the instantaneous core junction temperature of the excitation light source module is calculated based on the discrete integral model that introduces steady-state thermal resistance, and its instantaneous center wavelength redshift is deduced. Using the instantaneous output optical power as the thermodynamic input, combined with the ambient reference temperature, the instantaneous physical temperature of the motherboard optical test target surface is deduced based on the target surface heat dissipation model. The original radiation intensity of two independent characteristic fluorescence bands was collected, and the thermal quenching effect caused by the instantaneous physical temperature of the target surface and the instantaneous redshift of the center wavelength were combined to decouple and calculate the corrected ratio of the two-band fluorescence intensity. The local variance of the ratio of dual-band fluorescence intensity within a time sliding window is calculated. The penalty weight of the local variance is dynamically modulated by the instantaneous output optical power to construct a comprehensive defect identification feature index and compare it with a preset threshold to determine whether there is a defect in the test area.

2. The automatic optical testing device for computer motherboards according to claim 1, characterized in that, In the step of estimating the instantaneous physical temperature of the target surface, iterative calculations are performed by constructing a first-order inertial model of target surface heat dissipation; specifically: The residual temperature rise is obtained by exponentially decaying the temperature difference between the instantaneous physical temperature of the target surface and the ambient reference temperature at the previous discrete moment, according to the equivalent thermal time constant of the test target surface area. By combining the instantaneous output optical power measured at the current moment with the coupling coefficient between the ultraviolet absorptivity of the test target surface and the steady-state thermal resistance, the increase in temperature caused by photoinduced absorption can be calculated. The ambient reference temperature, residual temperature rise, and newly added temperature rise are added together to obtain the instantaneous physical temperature of the target surface at the current moment.

3. The automatic optical testing device for computer motherboards according to claim 2, characterized in that, The equivalent thermal time constant of the test target area and the coupling coefficient between the ultraviolet absorptivity and steady-state thermal resistance of the test target area are adaptively obtained by the main control unit through a pre-calibration program before the formal optical test. The pre-calibration program includes: The excitation light source module continuously irradiates the test target surface with a preset constant test light power to form a thermal step excitation and records the temperature rise response curve of the test target surface. The time consumed when the temperature rise reaches a specific response ratio threshold of the steady-state extreme temperature rise is extracted as the thermal time constant. The ratio of the steady-state extreme temperature rise to the constant test light power is extracted as the coupling coefficient.

4. The automatic optical testing device for computer motherboards according to claim 2, characterized in that, The ambient reference temperature is a dynamically acquired real-time variable; The testing device also includes a background ambient temperature sensor arranged away from the radiation area of ​​the excitation light source module. The main control unit collects the raw data from the background ambient temperature sensor and introduces a digital low-pass filtering algorithm with a time constant greater than the thermal time constant for smoothing, so as to filter out high-frequency thermal disturbances during the test process. The filtered output value is used as the environmental reference temperature to be extrapolated and updated at the current moment.

5. The automatic optical testing device for computer motherboards according to claim 1, characterized in that, The step of decoupling and calculating the corrected dual-band fluorescence intensity ratio includes: The ratio of the original radiative intensities of the first and second independent characteristic fluorescence bands acquired by the multispectral sensor is calculated as the baseline ratio. Based on the degree of deviation between the instantaneous physical temperature of the target surface and the preset reference temperature, and combined with the difference in nonradiative transition activation energy levels of the luminescent center and the Boltzmann constant, a thermal quenching compensation factor with an exponential mapping relationship is constructed. Based on the difference between the instantaneous center wavelength redshift and the nominal peak wavelength, and combined with the spectral overlap coupling factor, a wavelength drift compensation factor with a linear mapping relationship is constructed. The base ratio, thermal quenching compensation factor, and wavelength drift compensation factor are multiplied and weighted to obtain the corrected dual-band fluorescence intensity ratio.

6. The automatic optical testing device for computer motherboards according to claim 1, characterized in that, In the step of calculating the instantaneous core junction temperature, a discrete evolution model based on heat accumulation is established: The power input corresponding to the product of instantaneous forward current and instantaneous voltage is calculated, the light radiation dissipation corresponding to the instantaneous output light power at the previous discrete moment is deducted, and the heat conduction loss is obtained based on the difference between the core junction temperature and the measured temperature of the heat sink substrate, so as to obtain the net heat change rate. The incremental change in junction temperature is calculated using the net heat change rate, equivalent heat capacity, and discrete sampling interval, and then superimposed onto the instantaneous core junction temperature at the previous discrete moment. The steps for calculating the instantaneous center wavelength redshift include: based on the calculated temperature difference between the instantaneous core junction temperature and the preset reference temperature, performing linear drift compensation using the peak wavelength temperature drift coefficient of the excitation source, and superimposing it onto the nominal peak wavelength.

7. The automatic optical testing device for computer motherboards according to claim 1, characterized in that, The steps for constructing comprehensive defect identification feature indicators include: Logarithmic transformation is performed on the corrected dual-band fluorescence intensity ratio to obtain the baseline characteristic value; The ratio of the measured instantaneous output optical power to the preset rated initial optical power is calculated and used as the dynamic attenuation modulation factor. The local variance of the ratio of dual-band fluorescence intensity within a set time sliding window, the preset benchmark variance penalty coefficient, and the dynamic attenuation modulation factor are multiplied to construct a dynamic penalty term. The benchmark feature value minus the dynamic penalty term is used as the final comprehensive defect identification feature index.

8. The automatic optical testing device for computer motherboards according to claim 7, characterized in that, The set time sliding window is implemented within the main control unit based on a first-in-first-out circular queue data structure. The local variance at the current time t is obtained using an online recursive statistical algorithm. This algorithm uses the local variance at the previous discrete time step, combined with the latest dual-band fluorescence intensity ratio newly added to the queue at the current time t, and subtracts the oldest dual-band fluorescence intensity ratio that has been removed from the queue, to perform incremental iterative calculation.

9. The automatic optical testing device for computer motherboards according to claim 1, characterized in that, The step of updating the target drive current in the power supply module adopts a discrete positional proportional-integral-derivative control strategy, specifically including: The current feedback deviation is obtained by subtracting the preset optical power target value from the instantaneous output optical power measured at the current moment. Calculate the proportional adjustment component of the current feedback deviation, the integral adjustment component obtained by summing the feedback deviations at each historical discrete moment, and the differential adjustment component obtained by the difference between the current feedback deviation and the feedback deviation at the previous discrete moment. The proportional adjustment component, integral adjustment component, and derivative adjustment component are weighted and summed according to their respective gain coefficients to obtain the absolute value of the target drive current output to the power supply module.

Images