LED testing circuit and testing method thereof

Abstract

This application relates to the field of online testing technology for LED manufacturing, and provides an LED testing circuit and its testing method. The method includes: providing a test power supply to a test board on which multi-color LEDs are mounted; sequentially illuminating LEDs of different colors on the test board according to preset colors; after illuminating each LED of a specific color, collecting three photosensitive parameters of the current emission through a photosensitive sensor, comparing the magnitudes of the three photosensitive parameters, and determining the actual color emitted by the LED based on the relationship between the magnitudes of the three photosensitive parameters according to a preset color determination logic; comparing the actual determined color with the currently illuminated preset color; if they match, the LED is determined to be working normally; if they do not match, it is determined to be malfunctioning; and sequentially completing the testing of all LEDs of the colors to be tested to obtain the final LED test result.

Description

Technical Field

[0001] This application relates to the field of online testing technology in the production and manufacturing of light-emitting diodes, and in particular to an LED testing circuit and its testing method. Background Technology

[0002] Multi-color integrated LEDs are widely used in display modules, smart signs, consumer electronics backlights and other fields. During mass production, the working status of each color channel of the LED board must be tested to avoid defects such as incorrect assembly, color deviation, and channel failure.

[0003] The existing multi-color LED testing solutions in the mass production stage are mainly divided into three categories: The first category is manual visual inspection, which relies on the human eye to judge whether the color is qualified. This method is inefficient, highly subjective, and has a high error rate, making it unsuitable for high-speed production lines. The second category is professional spectrometer testing, which collects the complete emission spectrum and compares it with the standard color. It has high accuracy but high equipment cost and long testing time per LED, making it unsuitable for large-scale online testing. The third category is conventional automated continuity testing, which only detects whether the LED channel is conductive and whether the brightness meets the standard, without judging the actual emitted color, and cannot detect misassembly or color deviation defects. A few automated solutions that can achieve color detection also require pre-calibrating the absolute threshold of each color for different models and batches. After collecting parameters, multiple rounds of threshold matching calculations must be completed. The process is complex, has high hardware requirements, and is prone to misjudgment due to batch-specific brightness deviations in LEDs. Summary of the Invention

[0004] This application provides an LED test circuit and test method, aiming to solve the problems of low efficiency, high subjectivity, high error rate, and inability to adapt to high-speed production line production in the current mass production stage of multi-color LED test schemes.

[0005] In a first aspect, embodiments of this application provide a testing method for an LED test circuit, the method comprising:

[0006] Provide test power to the board under test that is equipped with multi-color LEDs, and light up the LEDs of different colors on the board under test one by one according to the preset colors;

[0007] After each LED of the color to be detected is lit, the three photosensitive parameters of the current emission are collected by the photosensitive sensor. The three photosensitive parameters are compared and the actual color emitted by the LED is determined according to the preset color determination logic and the relationship between the three photosensitive parameters.

[0008] The actual determined color is compared with the currently lit preset color. If they match, the LED is considered to be working normally; otherwise, it is considered to be working abnormally. The test of all LEDs of the colors to be tested is completed in turn to obtain the final LED test results.

[0009] In some embodiments, the step of providing test power to the test board on which the multi-color LEDs to be tested are installed, and lighting up the LEDs of different colors on the test board one by one according to the preset colors, includes: lighting up only one LED corresponding to a color channel at a time according to the color lighting order corresponding to the preset colors, reserving a preset light stabilization time after the lighting operation is completed, and then performing the subsequent photosensitive parameter acquisition steps.

[0010] In some embodiments, after each LED of a color to be detected is lit, the three light-sensing parameters of the current light emission are collected by a photosensitive sensor, including: collecting the three original light-sensing values ​​of the current light emission by the photosensitive sensor, converting the three original light-sensing values ​​into decimal values, and obtaining the three light-sensing parameters.

[0011] In some embodiments, the three photosensitive parameters include a first parameter value, a second parameter value, and a third parameter value; the step of comparing the magnitudes of the three photosensitive parameters and determining the actual color emitted by the LED according to a preset color determination logic and the magnitude relationship of the three photosensitive parameters includes: if the first parameter value is greater than the second and third parameter values ​​and the difference between them is greater than a preset difference, and the first parameter value is greater than the second parameter value, and the second parameter value is greater than the third parameter value, the actual color is determined to be red light; if the second parameter value is greater than the first parameter value... If the difference between the numerical values ​​of the first, second, and third channels is greater than a preset difference, and the numerical values ​​of the second, third, and first channels are greater than the numerical values ​​of the third channel, the actual color is determined to be green. If the numerical value of the third channel is significantly greater than the numerical values ​​of the first, second, and third channels, and the difference between them is greater than a preset difference, and the numerical values ​​of the third, second, and first channels are greater than the numerical values ​​of the second channel, the actual color is determined to be blue. If the corresponding difference between the numerical values ​​of the first, second, and third channels is less than or equal to the preset difference, the actual color is determined to be white.

[0012] In some embodiments, comparing the actual determined color with the currently lit preset color, and determining that the light-emitting diode is working normally if they match, and that it is working abnormally if they do not match, includes: after each comparison, storing the current position of the light-emitting diode, the preset color, the actual determined color and the determination result in the test cache; after all tests are completed, organizing the determination results of all light-emitting diodes to generate test records, and marking the information of all light-emitting diodes that are working abnormally.

[0013] In some embodiments, the method further includes: before lighting all light-emitting diodes, pre-collecting three photosensitive reference parameters under ambient light; after collecting the photosensitive parameters of the currently lit light-emitting diode each time, removing the influence of the corresponding ambient light reference parameters from the three collected parameters to obtain corrected photosensitive parameters after eliminating ambient light interference; and then using the corrected photosensitive parameters to perform subsequent size comparison steps to avoid interference from ambient light in the production site with the test results.

[0014] In some embodiments, the method further includes: after each preset number of board tests are completed, automatically triggering a photosensitive sensor calibration process, controlling a preset standard color light source to emit light of a preset standard color in sequence, sequentially collecting photosensitive parameters corresponding to each standard color, comparing the collected parameters with the standard parameter range, adjusting the gain level of the photosensitive sensor to make the collected parameters fall within the standard parameter range, completing automatic calibration, and ensuring the accuracy of long-term batch testing.

[0015] In some embodiments, the method further includes: pre-collecting test data of all defective LED boards labeled with actual abnormal causes during the mass production process, and training an intelligent classification model for abnormal causes using the labeled test data; after all tests are completed and all abnormal results and corresponding test parameters of the current board under test are obtained, extracting the degree of deviation of photosensitive parameters of each channel of the board and the location distribution characteristics of abnormal channels, inputting the extracted features into the trained intelligent classification model for abnormal causes, and obtaining the specific abnormal cause corresponding to the current defective board through model reasoning, specifically distinguishing four different abnormal types: LED body color deviation abnormality, pin soldering contact abnormality, power supply channel fault abnormality, and control logic abnormality, and then sending the abnormal type and corresponding location information together to the rework management system, automatically pushing standardized rework guidance schemes for the corresponding abnormal type, so as to facilitate maintenance personnel to quickly locate the fault point.

[0016] In some embodiments, the method further includes: after a preset number of tests are completed on the same type and batch of test boards, extracting the photosensitive parameter data of each color for all boards that have been determined to be qualified, performing cluster analysis on the parameter distribution of each color, identifying the overall parameter offset of the current batch due to LED production consistency deviation, automatically adjusting the color judgment threshold of subsequent tests of the current batch according to the offset amount, replacing the general threshold stored in the parameter library, and using the adjusted new threshold to perform color judgment for subsequent tests of boards in the same batch, avoiding test misjudgment caused by inherent brightness deviation of LEDs in different batches, and improving the accuracy of batch testing.

[0017] Secondly, this application provides an LED testing circuit, including a power supply module, a photosensor, and a control unit; the power supply module is used to provide test power to a test board on which a multi-color light-emitting diode under test is mounted; the photosensor is disposed at the corresponding photosensor position of the light-emitting diode under test, and is used to collect three photosensor parameters of the light output by the light-emitting diode after it is lit; the control unit is electrically connected to the power supply module, the photosensor, and the light-emitting control terminal of the test board, and the control unit is used to implement the method provided in any embodiment of this application.

[0018] The testing method presented in this application has a simple logic, requiring no complex spectral calculations or pre-calibration of absolute thresholds for each color. It has low hardware performance requirements for the control unit, high testing speed, and can adapt to the high-speed online testing needs of mass production lines. The overall testing cost is far lower than that of traditional spectrometer testing solutions. This method can not only detect LED channel non-conductivity defects, but also accurately detect color errors and severe color shift defects, making up for the shortcomings of traditional methods that only perform continuity brightness testing and improving yield control capabilities. Through the judgment logic based on the relationship between the magnitudes of three photosensitive parameters, it has a high tolerance for normal production consistency brightness deviations of LEDs, is not prone to misjudgment, and its testing stability is far superior to traditional solutions based on fixed absolute thresholds.

[0019] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic flowchart illustrating the steps of a test method for an LED test circuit according to an embodiment of this application;

[0022] Figure 2 This is a schematic block diagram of an LED test circuit provided in one embodiment of this application;

[0023] Figure 3 This is a circuit diagram of an LED testing circuit provided in one embodiment of this application;

[0024] Figure 4 This is a schematic block diagram of the structure of a test system for an LED test circuit provided in one embodiment of this application;

[0025] Figure 5This is a schematic block diagram of the structure of a control unit provided in an embodiment of this application.

[0026] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Detailed Implementation

[0027] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0029] It should be understood that, in order to clearly describe the technical solutions of the embodiments of the present invention, the terms "first" and "second" are used in the embodiments of the present invention to distinguish identical or similar items with essentially the same function and effect. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and the terms "first" and "second" are not necessarily different.

[0030] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0031] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0032] Multi-color integrated LEDs are widely used in display modules, smart signs, consumer electronics backlights and other fields. During mass production, the working status of each color channel of the LED board must be tested to avoid defects such as incorrect assembly, color deviation, and channel failure.

[0033] The existing multi-color LED testing solutions in the mass production stage are mainly divided into three categories: The first category is manual visual inspection, which relies on the human eye to judge whether the color is qualified. This method is inefficient, highly subjective, and has a high error rate, making it unsuitable for high-speed production lines. The second category is professional spectrometer testing, which collects the complete emission spectrum and compares it with the standard color. It has high accuracy but high equipment cost and long testing time per LED, making it unsuitable for large-scale online testing. The third category is conventional automated continuity testing, which only detects whether the LED channel is conductive and whether the brightness meets the standard, without judging the actual emitted color, and cannot detect misassembly or color deviation defects. A few automated solutions that can achieve color detection also require pre-calibrating the absolute threshold of each color for different models and batches. After collecting parameters, multiple rounds of threshold matching calculations must be completed. The process is complex, has high hardware requirements, and is prone to misjudgment due to batch-specific brightness deviations in LEDs.

[0034] To solve the above problem, please refer to Figure 1 This application provides a testing method for an LED test circuit, which is applied to the control unit of the LED test circuit.

[0035] In some embodiments, such as Figure 2 As shown, the provided LED test circuit includes a power supply module 1, a photosensor 2, and a control unit 3. The power supply module provides test power to the test board 4 on which the multi-color LEDs to be tested are mounted. The photosensor is set at the corresponding photosensitive position of the LEDs to be tested and is used to collect three photosensitive parameters of the light output by the LEDs after they are lit. The control unit is electrically connected to the power supply module, the photosensor, and the light-emitting control terminal of the test board. The control unit is used to light up LEDs of different colors on the test board one by one, receive the three photosensitive parameters transmitted by the photosensor, compare the magnitudes of the three photosensitive parameters, determine the actual light emission color of the current LED according to the preset color judgment logic, compare the actual color with the preset color currently lit, determine the working state of the corresponding LED, and output the final test result after completing the test of all LEDs of the colors to be tested.

[0036] This LED testing circuit is a low-cost, highly integrated, and automated solution designed for detecting the correctness of light emission color and working status of independently emitting red, green, blue, and white LEDs. It addresses the industry pain points of traditional dedicated LED testers, such as high cost, complex operation, and high barriers to entry. Based on the VishayVEML3328 color photosensitive sensor, it achieves fully automated testing of the entire process, including power supply, light emission timing control, light signal acquisition, intelligent color determination, and working status of the LED under test.

[0037] This circuit consists of three core modules: a power supply module, a photosensor unit, and a control unit. It also interfaces with a test board (DBT) equipped with multi-color LEDs to form a complete test loop. Its core functions include: providing a stable power supply to the DBT and the test circuit; sequentially illuminating different colored LEDs according to a preset timing sequence; accurately acquiring the photosensitivity parameters of each LED after it is lit; logically determining the emitted color based on the three primary colors principle; comparing the actual emitted color with the preset color to determine the LED's operating status; and summarizing and outputting the final test results after completing all channels of testing.

[0038] The power supply module features dual-path isolated power supply capability, accommodating the power supply needs of both the test circuit itself and the board under test (DUT): the main power supply branch outputs a stable DC voltage of 2.6V~3.6V, matching the rated operating voltage range of the VEML3328 sensor, while also providing operating power to the control unit. It incorporates decoupling and filtering circuits to ensure power supply stability. The DUT power supply branch provides a programmable on / off test power supply, matching the rated operating voltage of the LED under test. On / off control is achieved by the control unit, and it incorporates overcurrent and overvoltage protection circuits to prevent damage to the test circuit from DUT failures.

[0039] The photosensitive sensor unit uses a Vishay VEML3328RGBCIR color sensor, which is the core component of this solution that replaces traditional testers. It adopts a 2.0mm×1.25mm×1.0mm surface mount OPLGA4 package and integrates five photosensitive units: R, G, B, Clear, and IR. It has a single-channel 16-bit resolution, an operating temperature range of -40℃ to +85℃, and multiple adjustable photosensitive gain and sensitivity levels. Extensive data verification has shown that it has excellent detection accuracy and reliability. It adopts a universal I2C (SMBus compatible) bus interface, which can easily complete register parameter configuration and photosensitive data reading without the need for dedicated software. Installed at the corresponding photosensitive position of the LED under test, it can convert the output light signal after a single LED is lit into three digital photosensitive parameters (R, G, and B) and transmit them to the control unit, providing a core data source for color determination.

[0040] The core of the control unit is an industrial-grade MCU (microcontroller unit) with an I2C communication interface and multi-channel GPIO control capabilities. It is the control and computing core of the entire test circuit. It is electrically connected to the light-emitting control terminal and power supply module of the board under test through GPIO. It can light up / turn off different colored LEDs under test one by one in a preset order, and only one single-color LED is lit at a time to avoid crosstalk of multi-color light affecting the acquisition accuracy. The sensor initialization configuration is completed through the I2C bus, and the raw hexadecimal R, G, and B values ​​output by the sensor are read in real time. Preprocessing operations such as decimal conversion and ambient light base value subtraction are performed. The built-in preset color judgment logic compares the amplitude of the three photosensitive parameters to determine the actual emission color of the current LED. After comparing it with the preset lighting color, the working status of the LED (pass / fail) is determined. After completing the test of all LEDs under test one by one, the final test results and fault information are summarized and output.

[0041] Based on the principle of visible light three primary color combination imaging, the system accurately determines the emitted color by comparing the amplitudes of the three photosensitive parameters R, G, and B. The preset determination logic perfectly matches the optimized rules verified by actual measurements, as follows: Red light determination: The acquired R value is much greater than the G and B values, and the amplitude order satisfies R>G>B; Green light determination: The acquired G value is much greater than the B and R values, and the amplitude order satisfies G>B>R; Blue light determination: The acquired B value is much greater than the G and R values, and the amplitude order satisfies B>G>R; White light determination: The acquired R, G, and B values ​​are relatively close, and the amplitude relationship satisfies B>R and G>R, while also satisfying G<(R+B) and B<(R+G).

[0042] For example, such as Figure 3 As shown, this embodiment is an 8-station parallel 4-color LED automated testing circuit designed for production line batch testing scenarios. The core acquisition unit is built based on the VEML3328 color sensor array, and the PCA9535RGERI2CIO expander is used to realize centralized program control. It is fully compatible with the LED color judgment logic and testing process described above. It can simultaneously complete the detection of the correctness of the light emission color and working status of 8 groups of red / green / blue / white LEDs. It solves the pain points of low efficiency of traditional single-channel testing, multi-node I2C address conflicts, and poor consistency of batch testing. The hardware cost is much lower than that of dedicated LED testers, and it is suitable for the automated batch testing needs of industrial production lines.

[0043] Figure 3 The circuit diagram shown includes: ① 8 independent VEML3328 color sensor acquisition units (U1~U8); ② I2C bus communication and interface unit; ③ PCA9535RGERIO expansion and power / LED control unit (U9); ④ system power supply and power distribution network.

[0044] The 8-channel VEML3328 color sensor acquisition unit (U1~U8) is the core photosensitive acquisition module of the test circuit. In the circuit diagram, U1~U8 are 8 VEML3328 color sensors of the same model. The single-channel circuit design is completely symmetrical to ensure the consistency of the test accuracy of the 8 stations. The single-channel circuit is explained in detail using U1 as an example. The circuit design of the other U2~U8 is completely the same as that of U1.

[0045] The core circuit of a single-channel sensor (taking U1 as an example) includes: Chip pins and power supply circuit: U1 is a VEML3328 RGBCIR color sensor. Pin 1 is the VDD power input, and pin 4 is the GND power ground. Pin 1 is connected to an independent power supply VDD1, and a decoupling capacitor C1 (100nF / 0402 package ceramic capacitor) is connected in parallel. The other end of the capacitor is connected to pin 4 and system ground. This decoupling circuit is used to filter out high-frequency ripple and noise from the power supply, ensure the stability of the sensor power supply, avoid power interference affecting the photosensitive acquisition accuracy, and fully match the power supply design requirements of the VEML3328 datasheet. I2C communication matching circuit: Pin 2 of U1 is SDA (I2C serial data pin), and pin 3 is SCL (I2C serial clock pin). Pin 2 is connected in series with a 33Ω resistor R1 to bring out 2SDA bus signals, and pin 3 is connected in series with a 33Ω resistor R2 to bring out 2SCL bus signals. The 33Ω series resistor is the I2C bus matching resistor, used to suppress signal reflection, reduce high-frequency interference, and ensure the stability of I2C communication when multiple nodes are connected. All sensors adopt the same impedance design to ensure consistent bus characteristics.

[0046] The SDA pins of U2~U8 are all connected in parallel to the 2SDA bus via corresponding matching resistors (R3, R5, R7, R9, R11, R17, R20), and the SCL pins are all connected in parallel to the 2SCL bus via corresponding matching resistors (R4, R6, R8, R10, R12, R19, R21), enabling 8 sensors to be connected to the same I2C bus. Through the independent hardware address configuration of VEML3328, the main control unit can independently address, configure parameters, and read data for each sensor.

[0047] U2~U8 are each equipped with independent power inputs VDD2~VDD8, and each power supply is equipped with an independent 100nF decoupling capacitor (C2~C8) to realize independent power on / off control of a single sensor, thus solving the problem of I2C address conflict of multiple sensors from the hardware level.

[0048] Eight VEML3328 sensors correspond to eight test stations. The photosensitive surface of each sensor faces the light-emitting surface of the LED at the test station. An 8mm standard photosensitive distance is set. Black light-shielding barriers are set between adjacent stations to prevent cross-light interference between stations and ensure that a single sensor only collects the output light of the LED at the corresponding station.

[0049] This unit implements the physical connection and bus level matching between the test circuit and the external main control unit (MCU / PLC / host computer). The circuit diagram includes two connectors, J1 and J2, as well as bus pull-up resistors R23 and R24. The specific details are as follows:

[0050] Connector circuit: J1 is an XH_4P_2.54mm / 90° right-angle connector. Pins 1, 2, and 3 correspond to SCL, SDA, and GND respectively. Pin 4 is unused and serves as the main interface to connect to the I2C bus of the external main control unit, enabling communication between the test system and the main control unit. J2 is an FC_2x3P_1.27mm / 90° connector. Pins 1, 3, and 5 correspond to SCL, SDA, and GND respectively. Pins 2, 4, and 6 are reserved pins, serving as backup interfaces to adapt to different specifications of main control terminal blocks, and also supporting the cascading expansion of multiple test boards.

[0051] A 10kΩ / 0402 package / 1% accuracy resistor R23 is connected in series on the SCL bus, and a 10kΩ / 0402 package / 1% accuracy resistor R24 ​​is connected in series on the SDA bus. The other ends of both resistors are connected to the PP3V3 system main power supply. The I2C bus is an open-drain output architecture, and the bus level must be pulled high through pull-up resistors. By using 10kΩ pull-up resistors according to the I2C standard and matching the 3.3V system level, the bus still has stable level transition capability and communication reliability when nine I2C slave nodes (8 sensors + 1 IO expander) are connected to the bus. The SCL and SDA signals from J1 and J2 are directly connected in parallel with the 2SCL and 2SDA sensor buses, and are also connected to the I2C interface of the PCA9535RGER, so that all I2C slave devices in the entire test circuit are connected to the same master control bus and uniformly addressed and controlled by the external master control unit.

[0052] The PCA9535RGER IO expansion and power / LED control unit (U9) is the core of the test circuit's programmable control. In the circuit diagram, U9 is the PCA9535RGER, a 16-bit bidirectional IO expansion chip with an I2C interface. It expands one I2C bus into 16 controllable IO ports, enabling power control of eight sensors and illumination control of eight LEDs under test. The specific circuit and implementation details are as follows:

[0053] Pin 21 (VCC) is connected to the PP3V3 system main power supply, and a decoupling capacitor C9 (100nF / 0402 package / 25V ceramic capacitor) is connected in parallel. The other end of the capacitor is connected to system ground to filter out power supply noise and ensure stable chip operation. Pin 9 (GND) is directly connected to system ground. Pins 12 (SCL) and 20 (SDA) are connected to the system I2C bus, respectively. Pin 12 is connected in series with a 33Ω matching resistor R16, and pin 20 is connected in series with a 33Ω matching resistor R17 to maintain consistency with the sensor bus matching design and ensure bus impedance uniformity. Pin 22 (SCLVCC) is connected in series with a 0Ω resistor R22 to the PP3V3 to provide a level reference for the chip's internal I2C interface. Pins 18 (A0), 23 (A1), and 24 (A2) are the chip's I2C address configuration pins. They are pulled up to PP3V3 via series with 10kΩ resistors R13, R14, and R15, respectively, meaning A0 / A1 / A2 are all high. This corresponds to a fixed I2C slave address of 0x27 (marked I2C4_ADDR:0x27 in the circuit diagram), completely isolated from the VEML3328 sensor address on the bus to avoid communication conflicts. Pin 10 (INT) is the chip's interrupt output pin. In this embodiment, it is connected to GND via jumper M1, reserving an interrupt function. It can be connected to the master control unit's interrupt pin as needed to achieve real-time reporting of IO status changes.

[0054] The PCA9535RGER's 16 I / O ports are divided into P0 ports (P00~P07) and P1 ports (P10~P17). In this embodiment, the function allocation is as follows:

[0055] Pins P00 to P07 (pins 1 to 8) correspond to outputs VDD1 to VDD8 respectively. P00 controls the power supply VDD1 for U1, P01 controls the power supply VDD2 for U2, and so on. P07 controls the power supply VDD8 for U8. The main control unit can configure the IO output status via the I2C bus, enabling independent power on / off for each VEML3328 channel. This supports time-sharing power-on to avoid address conflicts, reduces standby power consumption, and provides independent reset fault tolerance for each channel.

[0056] P10~P17 (pins 11, 13~17, 19) have a total of 8 I / O channels, which correspond to the LED light-emitting control terminals of 8 stations under test. Each I / O channel can be expanded to 4 sub-signals to control the lighting and extinguishing of the red, green, blue and white LEDs of the corresponding station, realizing the programmable timing control of all LEDs under test by the main control unit, which fully matches the requirements of the LED testing process.

[0057] This embodiment uses a 3.3V unified system power supply (PP3V3), which is fully compatible with the 2.6V~3.6V rated operating voltage range of VEML3328 and PCA9535RGER. The power supply is distributed as follows: PP3V3 is provided by an external main control unit or a dedicated DC power module, providing a stable DC power supply for the entire test system. The input terminal is reserved with an expandable self-resetting fuse and TVS diode to achieve overcurrent, overvoltage, and electrostatic discharge protection.

[0058] The PP3V3 directly provides operating power to the PCA9535RGER chip and the I2C bus pull-up resistors; the PP3V3 is connected to the P00~P07 ports of the PCA9535RGER, and generates controllable VDD1~VDD8 power through the chip's IO output, providing independent and controllable operating power to the 8-channel VEML3328 sensors.

[0059] Based on the aforementioned hardware circuit and combined with preset 4-color LED color determination logic, this embodiment achieves fully automated testing at 8 stations. The complete process is as follows:

[0060] 1. Pre-test preparation and system initialization: Fix the 8 sets of 4-color LED boards to be tested at their respective test stations, confirming that the LEDs correspond one-to-one with the photosensitive positions of the corresponding sensors, eliminating the risk of light crosstalk; connect the J1 interface to the I2C bus of the external main control MCU and connect it to the PP3V3 system power supply. After the main control MCU is powered on, address the PCA9535RGER through address 0x27 to complete chip initialization, configure P00~P07 and P10~P17 as push-pull output mode, and set all IO outputs to low level by default, powering off all sensors and turning off the LEDs under test; complete the I2C bus scan to confirm normal communication. The main control MCU enables each sensor sequentially through the PCA9535, reads and stores the base values ​​of the R, G, and B channels in a completely dark environment, and uses this data for ambient light subtraction in subsequent tests to eliminate ambient light interference.

[0061] The single-station, single-color LED testing procedure (taking red LED at station 1 as an example) includes:

[0062] Step 1: Enable station and sensor: The main control MCU outputs a high level through the P00 pin of PCA9535 to enable the power supply of U1, while keeping the other sensors powered off; at the same time, it outputs a control signal through the P10 pin to light up the red LED of station 1, while keeping all other LEDs off.

[0063] Step 2: Photosensitive data acquisition: The main control MCU addresses U1 via I2C, configures the sensor gain and integration time, and reads the raw hexadecimal R, G, and B values ​​corresponding to the current red LED light.

[0064] Step 3: Data preprocessing: Convert the raw hexadecimal values ​​to decimal values, subtract the ambient light baseline value, perform 5 consecutive samplings and take the average value to obtain the effective photosensitivity parameters.

[0065] Step 4: Color and Status Judgment: Input the valid parameters into the preset judgment logic. If the condition is met that "R value is much greater than G and B values, and R>G>B", then the actual color is determined to be red light, which is consistent with the preset color and marked as qualified. If the logic is not met or the value is close to the base value, it is marked as unqualified and the fault type (color error, no light emission, abnormal brightness, etc.) is recorded.

[0066] Step 5: Single-step test completion: Turn off the current red LED, turn off the power of U1, and wait for a 100ms delay to ensure that the light completely dissipates and avoid crosstalk.

[0067] Following the single-step process described above, the red, green, blue, and white LEDs at stations 1 to 8 are tested sequentially. Two testing modes are supported: time-sharing serial testing (testing each station and each color to ensure the highest accuracy) and parallel synchronous testing (simultaneously enabling all sensors and lighting up LEDs of the same color, improving testing efficiency by 8 times).

[0068] After all tests are completed, the main control MCU summarizes the results: if all LEDs at all workstations are qualified, it outputs "Test Qualified" and triggers a qualification prompt; if there are unqualified items, it outputs the unqualified workstation number, preset color, actual judged color, and fault type, triggers an audible and visual alarm, and can also upload the results to the host computer for archiving, realizing the traceability of defective products.

[0069] The provided LED test circuit test method includes steps S101 to S103. Details are as follows:

[0070] Step S101. Provide test power to the board under test that has the multi-color LEDs to be tested installed, and light up the LEDs of different colors on the board under test one by one according to the preset colors.

[0071] Specifically, this step is the pre-control stage of the entire testing process. It enables programmable and anti-crosstalk sequential lighting of the LED under test, replacing manual power-on and lighting operations and adapting to the automated testing rhythm of high-speed production lines. Through time-division single-channel lighting control, it ensures that only one monochrome LED emits light at a time, completely avoiding distortion of photosensitive data caused by crosstalk of multi-color light. It also enables controllable power supply to the board under test and the test circuit, preventing damage to the test equipment from board under test failures and ensuring electrical safety during the testing process.

[0072] The control unit implements two core controls through the IO expansion chip (PCA9535RGER): one is to control the power supply module to output a stable test power to the board under test, and the other is to output level signals to the LED light-emitting control terminal of the board under test through the GPIO port according to the preset timing, so as to realize the independent lighting and extinguishing of LEDs of different colors.

[0073] For example, in the 8-station parallel test circuit provided in the embodiments of this application, the single-channel test circuit can be directly simplified and reused. Before the test, the 4-color LED board to be tested is fixed on the corresponding station of the test fixture, and it is confirmed that the power supply interface of the board, the LED control interface and the test circuit are reliably connected. The control unit (MCU) sends a command to the PCA9535RGER through the I2C bus to configure the control IO of the power supply branch of the board under test to be in push-pull output mode, outputting a high level, controlling the power supply module to output a test power supply (usually 2.0V~3.3V) matching its rated voltage to the board under test, and at the same time monitoring the power supply circuit current. If the current exceeds the preset threshold (e.g., 200mA), the power supply is immediately cut off, the board is determined to have a short circuit fault, the test is terminated and an alarm is output. A fixed color lighting sequence is preset: red light → green light → blue light → white light. This sequence corresponds one-to-one with the subsequent color judgment logic to avoid logical confusion. The control unit outputs level signals to the LED control terminal of the corresponding station through ports such as P10~P17 of the PCA9535RGER. Only the LED corresponding to the current color to be tested is lit each time, while all other colors and LEDs at all stations remain off. After the lighting operation is completed, a 50ms light stabilization time is reserved (based on the light response characteristics of conventional surface mount LEDs, which can be adjusted according to the LED datasheet) to ensure that the LED brightness and colorimetry reach a stable state before proceeding with the subsequent photosensitive acquisition steps, avoiding test errors caused by unstable data at the moment of lighting.

[0074] For 8-station parallel testing scenarios, two control modes can be selected: High-precision serial mode: The entire process of testing 4-color LEDs is completed one by one in the order of station 1 to station 8. Only one LED of one station is lit at the same time, which minimizes the risk of crosstalk and maximizes the test accuracy; High-efficiency parallel mode: All stations execute tests synchronously. At the same time, all stations light up LEDs of the same color (such as all stations lighting up red light at the same time), which improves the test efficiency by 8 times and is suitable for high-speed mass production scenarios.

[0075] Step S102. After each LED of the color to be detected is lit, the three light-sensing parameters of the current emission are collected by the photosensitive sensor. The three light-sensing parameters are compared and the actual color emitted by the LED is determined according to the preset color determination logic and the relationship between the three light-sensing parameters.

[0076] Specifically, this step is the core computational link of the entire testing method. Its core positioning is relative value color determination based on the principle of visible light three primary colors, replacing the traditional absolute threshold calibration scheme. It does not require separate threshold calibration for different models and batches of LEDs. Color determination is completed only by the relative magnitude relationship of the three parameters R, G, and B, completely avoiding misjudgment caused by batch-specific brightness shifts of LEDs. It realizes the quantification and objective determination of emitted color, replacing subjective judgment by the human eye. At the same time, it does not require full-spectrum acquisition by a professional spectrometer, which greatly reduces hardware costs and testing time.

[0077] Using the R, G, and B photosensitive units of the VEML3328 color sensor, the visible light signal of the LED is converted into digital parameters. After obtaining valid data through preprocessing, the actual emitted color of the LED is accurately determined based on the relative magnitude relationship of the three parameters according to the preset quantization judgment logic.

[0078] For example, this step is triggered synchronously with the lighting operation of S101. This step is executed immediately after each LED under test is lit. The control unit addresses the VEML3328 sensor corresponding to the current lighting station via the I2C bus and first completes the sensor register configuration: according to the brightness of the LED under test, the appropriate gain level (VEML3328 supports four gain levels: 1x, 2x, 4x, and 8x) and integration time (normally set to 100ms to balance the signal-to-noise ratio and test speed) are set; the raw data of the registers of the three channels R, G, and B of the sensor are read: the R, G, and B channels of VEML3328 are all 16-bit resolution, corresponding to two 8-bit registers. The high 8 bits and low 8 bits of the register values ​​of each channel are read and combined into a hexadecimal raw photosensitive value; the raw values ​​of the same channel are continuously collected 5 times, the maximum and minimum values ​​are removed and the average value is taken to reduce random noise interference and obtain a stable raw photosensitive value.

[0079] The raw hexadecimal values ​​of the R, G, and B channels are converted into decimal integer values ​​to obtain the first parameter (R value), the second parameter (G value), and the third parameter (B value), which are the three photosensitive parameters mentioned in this step. The pre-stored ambient light reference parameters for the corresponding workstations (the reference parameters are obtained through a pre-calibration process) are subtracted from the three converted decimal values ​​to obtain the corrected photosensitive parameters after eliminating ambient light interference. All subsequent judgments are performed based on the corrected parameters.

[0080] The control unit has a built-in pre-calibrated quantization logic that compares the corrected R, G, and B parameters and determines the actual color according to the following rules: Red light determination: If the R value is ≥ 3 times the G value and the B value is ≥ 3 times the B value, and the amplitude order satisfies R > G > B, then the current actual emitted color is determined to be red light; Green light determination: If the G value is ≥ 3 times the R value and the B value is ≥ 3 times the B value, and the amplitude order satisfies G > B > R, then the current actual emitted color is determined to be green light; Blue light determination: If the B value is ≥ R... If the value of B is 3 times that of G, and the amplitude order satisfies B>G>R, then the current actual emitted color is determined to be blue light. White light determination: If the maximum difference between the three values ​​of R, G, and B is ≤ 50% of the maximum value among the three values, and the amplitude relationship satisfies B>R, G>R, and G<(R+B) and B<(R+G), then the current actual emitted color is determined to be white light. If the values ​​of the three parameters are all close to the ambient light reference value (difference ≤ 10% of the reference value), then the current LED is determined to be not emitting light and there is no effective color output.

[0081] Step S103. Compare the actual determined color with the currently lit preset color. If they match, the LED is determined to be working normally; if they do not match, it is determined to be working abnormally. Complete the test of all LEDs of the colors to be tested in turn to obtain the final LED test results.

[0082] Specifically, this step is the closed-loop output link of the entire testing process. Its core purpose is to achieve LED qualification judgment, accurate fault location, and full traceability of test results, solving the problems of missed fault detection, lack of traceability, and difficulty in troubleshooting in traditional testing.

[0083] The actual emitted color determined in S102 is compared with the preset color currently lit in S101 to determine the working status of a single LED. After testing all LEDs under test, all determination results are summarized to generate a traceable test record and output the final test results and defect information.

[0084] This step is executed immediately after the color determination of a single LED is completed in S102. The preset color of the currently lit LED is retrieved from the control unit's cache (e.g., if the currently lit LED is red in S101, then the preset color is red), and compared with the actual emitted color obtained in S102. If the actual color is completely consistent with the preset color, the LED is determined to be working normally (qualified). If the actual color is inconsistent with the preset color, or is determined to be not emitting light, the LED is determined to be working abnormally (unqualified). After the determination of a single LED is completed, the LED is immediately turned off, with a 100ms extinction delay to ensure that the light completely dissipates and to avoid crosstalk to the test of the next LED before the test process for the next LED is executed.

[0085] The control unit allocates a dedicated test cache array in its internal RAM. After each LED test is completed, the corresponding information is immediately written into the cache. The written information fields include: station number, LED channel number, preset color, actual judgment color, judgment result (pass / fail), R / G / B correction parameter value, and test timestamp. For unqualified LEDs, the fault type is additionally marked, including: color mismatch, no light emission, and abnormal brightness.

[0086] Following the preset testing sequence, after testing all colors and all workstations of the LEDs on the board under test sequentially, the control unit summarizes all test data in the cache and executes the final result output: If all LEDs under test are deemed qualified, "Test Qualified" is output through the OLED display of the test fixture, triggering a green indicator light and a qualified prompt sound. At the same time, the complete test record is uploaded to the host computer / production line MES system via UART serial port and stored in the local Flash chip; If there are unqualified LEDs, the unqualified workstation, channel, preset color, actual color and fault type are output through the display, triggering a red indicator light and an audible and visual alarm. At the same time, the defect information is uploaded to the production line system, marked as a defective product, and intercepted from flowing into the next process; After the result output is completed, the control unit cuts off the test power of the board under test, resets all IO ports, waits for the test trigger signal of the next board under test, and enters the next test cycle.

[0087] In some embodiments, the step of providing test power to the test board on which the multi-color LEDs to be tested are installed, and lighting up the LEDs of different colors on the test board one by one according to the preset colors, includes: lighting up only one LED corresponding to a color channel at a time according to the color lighting order corresponding to the preset colors, reserving a preset light stabilization time after the lighting operation is completed, and then performing the subsequent photosensitive parameter acquisition steps.

[0088] This embodiment optimizes the lighting control process of S101, and the core solution is to address the problems of unstable brightness / color at the moment of LED lighting and light crosstalk caused by simultaneous lighting of multiple channels, which leads to data distortion and test misjudgment.

[0089] Through strict single-channel interlock control, it is ensured that only one color channel of LED is lit at any given time. At the same time, a configurable light emission stabilization delay is inserted between the lighting operation and the acquisition operation to ensure that the LED is in a stable light emission state during acquisition, thereby improving the accuracy of the acquired data from the source.

[0090] In the firmware of the control unit, LED control interlock logic is written: the control IO of all color channels outputs a low level (off state) by default. Only when a certain channel is turned on, the IO of all channels is forced low first, and then a high level is output to the target channel, thus preventing two or more channels from being lit at the same time from the software level. For the 8-station parallel test scenario, in parallel mode, only the same color channel of all stations is allowed to be lit at the same time, and different color channels are interlocked to avoid cross-station crosstalk of different color light.

[0091] For standard 0402 and 0603 packaged surface mount LEDs, the preset stable lighting delay is 50ms; for high-power LEDs and backlight LEDs, the preset delay is 100ms, which can be flexibly configured through the host computer. After the control unit completes the IO level output for LED lighting, it calls the precise delay function in the firmware to wait for the preset stable duration. During the delay, no register reading or data acquisition operations are performed. After the delay ends, the sensor data acquisition process is triggered.

[0092] After the delay ends and before data acquisition begins, the control unit reads back the current level status of all LED control I / Os to confirm that only the target channel is high and the rest of the channels are low. Data acquisition is then performed only after the verification passes. If the verification fails, all LEDs are immediately turned off, the test is terminated, and a control anomaly alarm is output to prevent crosstalk data from entering the judgment process.

[0093] In some embodiments, after each LED of a color to be detected is lit, the three light-sensing parameters of the current light emission are collected by a photosensitive sensor, including: collecting the three original light-sensing values ​​of the current light emission by the photosensitive sensor, converting the three original light-sensing values ​​into decimal values, and obtaining the three light-sensing parameters.

[0094] This embodiment optimizes the parameter acquisition stage in S102, primarily addressing the issues that the raw hexadecimal register values ​​output by the VEML3328 sensor cannot be directly used for logical operations and that data dimensions are inconsistent across different gain levels. It achieves standardized conversion of photosensitive data, ensuring the universality of subsequent decision-making logic. Through register address mapping, high-low bit data merging, base conversion, and gain normalization, the non-standardized raw data output by the sensor is converted into decimal photosensitive parameters with uniform dimensions, ensuring data consistency across different gain levels and different batches of sensors.

[0095] The channel register mapping table for VEML3328 is predefined in the control unit firmware:

[0096]

[0097] The control unit reads the lower 8 bits and higher 8 bits of the corresponding channel register values ​​sequentially via the I2C bus, and combines them into 16-bit raw data through bit operations: uint16_traw_data=(high_byte<<8)|low_byte, thus obtaining the raw hexadecimal photosensitive value.

[0098] The merged 16-bit hexadecimal raw value is directly converted into a decimal integer value in the range of 0 to 65535. According to the current sensor gain level, the decimal value is normalized to convert it into an equivalent value under 1x gain. The conversion formula is: Normalized parameter = Decimal raw value / Gain multiplier. For example, if the current gain is 4x and the raw value is 40000, then the normalized parameter is 10000, ensuring that the same set of judgment logic can be used for parameters under different gain levels.

[0099] After the conversion is completed, a data validity check is performed: if the standardized parameter exceeds the range of 0~65535, or the value is 0, it is determined that the sensor acquisition is abnormal, and the acquisition process is re-executed; if there are 3 consecutive acquisition abnormalities, a sensor fault alarm is output, the test is terminated, and invalid data is prevented from entering the judgment process.

[0100] In some embodiments, the three photosensitive parameters include a first parameter value, a second parameter value, and a third parameter value; the step of comparing the magnitudes of the three photosensitive parameters and determining the actual color emitted by the LED according to a preset color determination logic and the magnitude relationship of the three photosensitive parameters includes: if the first parameter value is greater than the second and third parameter values ​​and the difference between them is greater than a preset difference, and the first parameter value is greater than the second parameter value, and the second parameter value is greater than the third parameter value, the actual color is determined to be red light; if the second parameter value is greater than the first parameter value... If the difference between the numerical values ​​of the first, second, and third channels is greater than a preset difference, and the numerical values ​​of the second, third, and first channels are greater than the numerical values ​​of the third channel, the actual color is determined to be green. If the numerical value of the third channel is significantly greater than the numerical values ​​of the first, second, and third channels, and the difference between them is greater than a preset difference, and the numerical values ​​of the third, second, and first channels are greater than the numerical values ​​of the second channel, the actual color is determined to be blue. If the corresponding difference between the numerical values ​​of the first, second, and third channels is less than or equal to the preset difference, the actual color is determined to be white.

[0101] This embodiment optimizes the color determination process in S102, and solves the problems of "far greater than" without quantitative standards, blurred determination boundaries in different scenarios, and easy misjudgment / missed judgment in the original determination logic. Through a configurable difference threshold, the color determination logic is fully quantized and adaptable, taking into account the testing needs of LEDs of different specifications.

[0102] Based on statistical data from a large number of qualified LED samples, the minimum difference threshold between the primary color channel and the secondary color channel is calibrated, transforming the qualitative description of "far greater than" into a quantitative numerical judgment condition. At the same time, a quantitative difference upper limit is set for white light judgment, realizing the full digitization of all color judgment conditions without subjective ambiguity boundaries.

[0103] For the target LED model, collect the standardized R, G, and B parameters of at least 1000 qualified samples, and statistically analyze the ratio distribution of the primary color to the secondary color in each color channel. Take 90% of the minimum ratio of the primary color to the secondary color in the qualified samples as the preset difference threshold. The general threshold is set to the primary color channel value ≥ 3 times the secondary color channel value (i.e., the difference between the primary and secondary channels ≥ 200% of the secondary color value). The preset difference threshold for white light determination is set to the maximum difference of the three channels ≤ 50% of the maximum value of the three channels. Store the calibrated thresholds in the Flash parameter library of the control unit, which supports configuration modification for different LED models via the host computer without modifying the firmware code.

[0104] In the control unit firmware, write the judgment logic with thresholds and define the following variables: R_val (red light parameter), G_val (green light parameter), B_val (blue light parameter), THD_main (main and secondary channel ratio threshold, default 3.0f), and THD_white (white light difference threshold, default 0.5f).

[0105] Red light detection code logic: if((R_val>=G_val*THD_main)&&(R_val>=B_val*THD_main)&&(R_val>G_val)&&(G_val>B_val)){actual_color=COLOR_RED;}; Green light detection code logic: elseif((G_val>=R_val*THD_main)&&(G_val>=B_val*THD_main)&&(G_val>B_val)&&(B_val>R_val)){actual_color=COLOR_RED;}; Blue light detection code logic: elseif((B_val>=R_val*THD_main)&&(B_val>=G_val*THD_main)&&(B_val>G_val)&&(G_val>R_val)){actual_color=COLOR_GREEN;}; _val>R_val)){actual_color=COLOR_BLUE;};White light determination code logic: c runs else{float max_val=max(R_val,max(G_val,B_val));float min_val=min(R_val,min(G_val,B_val));if(((max_val-min_val)<=max_val*THD_white)&&(B_val>R_val)&&(G_val>R_val)&&(G_val<(R_val+B_val))&&(B_val<(R_val+G_val))){actual_color=COLOR_WHITE;}else{actual_color=COLOR_INVALID; / / Invalid color}}.

[0106] If the initial judgment result is within the threshold boundary (e.g., the primary-to-secondary ratio is between 2.8 and 3.2), the sensor integration time will be automatically doubled, and data will be collected again for a second judgment. The result of the second judgment will be used to avoid misjudgment caused by the critical value.

[0107] In some embodiments, comparing the actual determined color with the currently lit preset color, and determining that the light-emitting diode is working normally if they match, and that it is working abnormally if they do not match, includes: after each comparison, storing the current position of the light-emitting diode, the preset color, the actual determined color and the determination result in the test cache; after all tests are completed, organizing the determination results of all light-emitting diodes to generate test records, and marking the information of all light-emitting diodes that are working abnormally.

[0108] This embodiment optimizes the result output stage of S103, and fundamentally solves the problems of traditional testing that only output pass / fail results, lack detailed process data, cannot trace defective products, and cannot perform batch quality analysis, so as to realize the traceability, queryability, and analyzability of the entire testing process data.

[0109] Through a layered caching design, real-time storage of test data for individual LEDs is achieved. After the entire testing process is completed, standardized test records are automatically generated, containing all process parameters and result information. The records support local storage and upload to a host computer, meeting the quality traceability requirements of the mass production process.

[0110] The hierarchical test cache design includes: Level 1 real-time cache: In the internal RAM of the control unit, a FIFO real-time cache of size 256 entries is allocated, with each cache entry corresponding to the test data of one LED. The data structure is defined as follows:

[0111] typedef struct { uint8_t station_id; / / Station number (1-8);

[0112] uint8_tchannel_id; / / Color channel number (1-4, corresponding to red / green / blue / white);

[0113] uint8_tpreset_color; / / Preset color;

[0114] uint8_tactual_color; / / Actual color to be determined;

[0115] uint8_tresult; / / Judgment result (0=pass, 1=fail, 2=abnormal);

[0116] uint16_tr_value; / / R normalization parameter;

[0117] uint16_tg_value; / / G-normalization parameter;

[0118] uint16_tb_value; / / B-normalization parameter;

[0119] uint32_ttimestamp; / / Test timestamp;

[0120] }test_data_t;

[0121] Secondary batch cache: A 16MB batch storage partition is allocated in the W25Q series Flash chip connected to the control unit, which can store complete test data of at least 10,000 boards under test, and ensure that the data is not lost when power is off.

[0122] After a single board is tested, the control unit automatically extracts all test data from the real-time cache and generates a standardized CSV test record. The record includes: board batch number, board serial number, test time, total number of workstations, number of qualified boards, number of unqualified boards, and detailed test data for each LED. For unqualified boards, a separate summary table of defects is marked in the record, including defective workstation, defective channel, fault type, and parameter deviation value.

[0123] After the test record is generated, the control unit uploads the record to the production line MES system and rework management system via UART serial port / Ethernet interface, and stores it in the local Flash partition. It supports querying the corresponding complete test record from the local Flash or host computer system by board serial number, batch number, and test time, realizing full-process traceability of defective products. For batch production scenarios, test data can be exported by day / batch for batch quality analysis, yield statistics, and abnormal trend early warning.

[0124] In some embodiments, the method further includes: before lighting all light-emitting diodes, pre-collecting three photosensitive reference parameters under ambient light; after collecting the photosensitive parameters of the currently lit light-emitting diode each time, removing the influence of the corresponding ambient light reference parameters from the three collected parameters to obtain corrected photosensitive parameters after eliminating ambient light interference; and then using the corrected photosensitive parameters to perform subsequent size comparison steps to avoid interference from ambient light in the production site with the test results.

[0125] This embodiment optimizes the ambient light interference problem in mass production testing. It focuses on solving the problems of distorted photosensitive parameters and test misjudgment caused by changes in ambient light in the production workshop and light leakage at workstations. By using pre-calibrated ambient light reference parameters, it achieves real-time elimination of ambient light interference and improves the adaptability of the testing solution to complex production environments.

[0126] Before the formal test, three reference parameters of the sensor were collected in a completely dark environment as ambient light baseline values. After each acquisition of the photosensitive parameters of the LED light emission, the baseline value of the corresponding channel was automatically subtracted to obtain the effective parameters after eliminating ambient light interference, ensuring that the test results are not affected by changes in ambient light.

[0127] The ambient light reference parameter pre-calibration process supports three trigger modes and can be flexibly configured: ① Automatic execution after each power-on of the test system; ② Automatic execution after every 100 board tests; ③ Manual triggering by the operator; The control unit, through the PCA9535RGER, forcibly turns off all LEDs at all workstations and shuts down the auxiliary light source of the test fixture to ensure that the sensors are in a completely dark environment without direct light; According to the workstation sequence, the VEML3328 sensor at each workstation is enabled sequentially, and configured with the same gain level and integration time as the formal test; The values ​​of the R, G, and B channels are continuously collected 10 times at each workstation, and the average value is taken after removing the maximum and minimum values ​​as the ambient light reference parameter for the corresponding channel at that workstation; The reference parameters of all workstations are stored in the Flash reference library of the control unit and simultaneously written to the real-time call cache of RAM for direct retrieval during formal testing.

[0128] Each time the raw decimal values ​​of R, G, and B of the LED under test are acquired, a correction calculation is automatically performed. The formulas include: Corrected R parameter = Acquired raw R value - Ambient light reference value of the corresponding station's R channel; Corrected G parameter = Acquired raw G value - Ambient light reference value of the corresponding station's G channel; Corrected B parameter = Acquired raw B value - Ambient light reference value of the corresponding station's B channel. If the corrected parameter is negative, it is automatically set to 0 to avoid negative values ​​affecting subsequent judgment logic. All subsequent size comparisons and color judgment operations are performed based on the corrected parameters.

[0129] During the calibration process, if the ambient light reference value of a certain workstation exceeds the preset upper limit (e.g., exceeding 5% of the range), it is determined that the test fixture is leaking light or the ambient light is too strong. An ambient light abnormality alarm will be output immediately to prompt the operator to check the light shielding structure and the on-site environment to avoid invalid calibration.

[0130] In some embodiments, the method further includes: after each preset number of board tests are completed, automatically triggering a photosensitive sensor calibration process, controlling a preset standard color light source to emit light of a preset standard color in sequence, sequentially collecting photosensitive parameters corresponding to each standard color, comparing the collected parameters with the standard parameter range, adjusting the gain level of the photosensitive sensor to make the collected parameters fall within the standard parameter range, completing automatic calibration, and ensuring the accuracy of long-term batch testing.

[0131] This embodiment optimizes the accuracy drift problem of sensors after long-term use. The core solution is to address the issues of gain drift, decreased photosensitivity, reduced test accuracy due to aging, and increased false positive rate that occur after long-term batch testing of the VEML3328 sensor. Closed-loop automatic calibration is achieved through a standard light source to ensure the accuracy and consistency of long-term mass production testing.

[0132] With a preset calibration trigger cycle, based on a programmable standard color light source, the sensor parameters under each standard color are collected and compared with the pre-stored standard parameter range. The sensor gain level and integration time are automatically adjusted so that the collected parameters fall within the standard range, achieving sensor calibration without disassembly and fully automatically, without manual intervention.

[0133] A programmable 4-color standard LED light source is installed at a fixed position on the test fixture. The color coordinates of the light source conform to the NIST national metrology standard. The installation position is completely consistent with the LED under test, ensuring that the sensor at each station can collect uniform light from the standard light source. The control terminal of the standard light source is connected to the reserved IO port of the PCA9535RGER, and the control unit realizes programmable lighting. In the new state of the sensor, the R, G, and B parameters corresponding to the four colors of the standard light source (red, green, blue, and white) are collected, and a tolerance range of ±10% is set as the standard parameter range, which is stored in the Flash calibration library of the control unit.

[0134] Two preset trigger conditions are provided. The calibration process will be automatically triggered when either condition is met: Count trigger: Automatically triggered after every 500 boards under test are tested; Time trigger: Automatically triggered after the system has run continuously for 72 hours; The calibration process is executed by default during material change and downtime on the production line, without affecting the normal production rhythm, and can also be manually triggered by the host computer.

[0135] The closed-loop automatic calibration execution process includes: 1. Pre-calibration preparation: The control unit turns off all LEDs under test and fixture light sources, executes the ambient light calibration process of Example 5, and updates the ambient light reference parameters; 2. Standard color acquisition: The control unit sequentially lights up the red, green, blue, and white light of the standard light source. After each color is lit, the R, G, and B parameters of all station sensors are sequentially acquired. After performing ambient light correction, the calibration acquisition value is obtained; 3. Parameter comparison and gain adjustment: The calibration acquisition value of each station is compared with the standard parameter range. If the acquired value is lower than the standard range... The lower limit automatically increases the sensor's gain level (gradually increasing from 1x to 8x); if the acquired value exceeds the upper limit of the standard range, the gain level is automatically reduced; after adjustment, the sensor is re-acquired until the acquired value falls within the standard parameter range; 4. Calibration result storage and verification: After all workstations are calibrated, the new gain parameters and integration time parameters are stored in the Flash parameter library for direct retrieval during subsequent formal testing; if the acquired value still cannot fall within the standard range after adjusting a workstation to the highest / lowest gain, it is determined that the sensor is aging / damaged, and a sensor fault alarm is output, prompting replacement.

[0136] In some embodiments, the method further includes: pre-collecting test data of all defective LED boards labeled with actual abnormal causes during the mass production process, and training an intelligent classification model for abnormal causes using the labeled test data; after all tests are completed and all abnormal results and corresponding test parameters of the current board under test are obtained, extracting the degree of deviation of photosensitive parameters of each channel of the board and the location distribution characteristics of abnormal channels, inputting the extracted features into the trained intelligent classification model for abnormal causes, and obtaining the specific abnormal cause corresponding to the current defective board through model reasoning, specifically distinguishing four different abnormal types: LED body color deviation abnormality, pin soldering contact abnormality, power supply channel fault abnormality, and control logic abnormality, and then sending the abnormal type and corresponding location information together to the rework management system, automatically pushing standardized rework guidance schemes for the corresponding abnormal type, so as to facilitate maintenance personnel to quickly locate the fault point.

[0137] This embodiment addresses the difficulty of locating defective products during rework. It solves the problem that traditional testing can only determine whether an LED is qualified or not, but cannot pinpoint the specific cause of the fault. Rework personnel need to check each board one by one, which is extremely inefficient and has a long rework cycle. By using a machine learning model, it achieves intelligent classification of defects and abnormalities, automatically distinguishes fault types, and pushes standardized rework guidance.

[0138] Based on the labeled defect data accumulated during mass production, a lightweight machine learning classification model is trained to extract the test parameter features of defective boards. Through model inference, four core fault types are accurately distinguished, enabling precise location of defective products and significantly improving rework efficiency.

[0139] The annotation dataset construction and model training involved collecting at least 10,000 sets of test data from defective boards from mass production historical data. Each set of data included: R / G / B parameters, abnormal channel distribution, power supply voltage data, and continuity test data, along with annotations of the corresponding actual fault causes, categorized into four main types:

[0140] 1. Abnormal color deviation of LED body: The color of the LED chip itself is deviated, while the rest of the circuit is normal;

[0141] 2. Poor contact due to cold solder joints on LED pads: Poor contact on LED pads or leads can result in low brightness or no light emission.

[0142] 3. Power supply channel malfunction: A power supply circuit malfunction in the board under test causes abnormal brightness in all channels;

[0143] 4. Abnormal control logic: Incorrect connection of the LED control circuit on the board or error in the program logic causes the illuminated color to be inconsistent with the preset color;

[0144] Feature engineering extracts core feature parameters, including: main-to-sub channel ratio deviation rate, single-channel brightness deviation rate, multi-channel parameter correlation, abnormal channel distribution characteristics, and correlation between power supply voltage and parameters;

[0145] Model training and deployment utilize a lightweight random forest classification model (suitable for edge deployment), with 80% of the dataset used for training and 20% for validation, ensuring a classification accuracy of ≥95%. The trained model is then exported as a lightweight inference model and deployed to the host computer of the test system.

[0146] The intelligent reasoning process for defects involves the following steps: After the test board is completed, if any defects are found, the control unit uploads the complete test data and power supply monitoring data of the board to the host computer. The host computer automatically extracts preset core feature parameters. These extracted feature parameters are then input into the trained classification model to perform reasoning operations, outputting the fault type with the highest probability and its confidence level. Based on the fault type obtained through reasoning, the system automatically retrieves the corresponding standardized rework plan from the rework guidance library. For example: for pin soldering defects, it pushes resoldering instructions for the corresponding LED pins at the workstation; for LED color deviation, it pushes replacement instructions for the corresponding LED; for power supply channel faults, it pushes inspection and repair instructions for the power supply circuit; for control logic defects, it pushes troubleshooting instructions for the control circuit and firmware. The fault type, workstation information, and rework guidance are simultaneously sent to the factory rework management system, allowing maintenance personnel to directly query the board serial number for accurate rework of defective products.

[0147] After each batch of repairs is completed, the actual cause of the fault is compared with the model's inference results, new labeled data is added to the dataset, and the model is incrementally trained regularly to continuously improve the model's classification accuracy.

[0148] In some embodiments, the method further includes: after a preset number of tests are completed on the same type and batch of test boards, extracting the photosensitive parameter data of each color for all boards that have been determined to be qualified, performing cluster analysis on the parameter distribution of each color, identifying the overall parameter offset of the current batch due to LED production consistency deviation, automatically adjusting the color judgment threshold of subsequent tests of the current batch according to the offset amount, replacing the general threshold stored in the parameter library, and using the adjusted new threshold to perform color judgment for subsequent tests of boards in the same batch, avoiding test misjudgment caused by inherent brightness deviation of LEDs in different batches, and improving the accuracy of batch testing.

[0149] This embodiment optimizes the problem of misjudgment caused by batch deviation of LEDs. The core solution is to address the issue that LEDs from different batches and wafers have inherent brightness and color deviations, and that a general fixed threshold will lead to batch misjudgment and falsely low yield. By clustering analysis of qualified samples in the same batch, the judgment threshold is automatically adjusted to suit the current batch, thereby improving the accuracy of batch testing.

[0150] Once a preset number of tests have been completed on boards of the same model and batch, the parameter data of all qualified samples are extracted. The parameter distribution characteristics of the current batch are obtained through cluster analysis. The color judgment threshold is automatically optimized and adjusted, replacing the general fixed threshold, so as to achieve batch-adaptive thresholding and avoid misjudgment caused by inherent LED deviations.

[0151] The threshold adaptive adjustment trigger mechanism automatically triggers the threshold adjustment process after 200 qualified tests are completed on the same model and production batch of test boards. Subsequently, the optimization adjustment is re-executed every 100 tests completed, realizing dynamic updates of the threshold. The board batch number entered by the host computer automatically identifies boards in the same batch, and the parameter data of different batches are counted independently without interference.

[0152] From the test record database, test data of all qualified boards in the current batch are extracted. For each of the four color channels (red, green, blue, and white), the corresponding standardized R, G, and B parameters are extracted to construct a sample dataset for each color channel. Outlier removal uses the 3σ criterion to remove outliers from each dataset, eliminating the influence of individual abnormal samples on the parameter distribution. Cluster analysis uses the K-means clustering algorithm to cluster the sample data for each color channel, obtaining the parameter distribution range of qualified samples in the current batch. The minimum ratio of the primary color to the secondary color channel and the maximum difference percentage of the white light channel are statistically calculated for each color channel.

[0153] The threshold calculation uses 90% of the minimum ratio of the primary and secondary channels of the qualified samples in the current batch as the new primary and secondary channel judgment threshold; and 110% of the maximum difference ratio of the white light channel as the new white light judgment threshold. The new thresholds must meet the preset upper and lower limits (primary-secondary ratio threshold range 2.0~4.0, white light difference threshold range 0.3~0.7) to avoid threshold anomalies caused by extreme data. If the range is exceeded, the original general threshold is retained and a batch deviation warning is output. After verification, the new thresholds are stored in the Flash parameter library of the control unit, replacing the original general thresholds. Subsequent board tests in the same batch will automatically call the new thresholds to perform color judgment. When a change in the board batch number is detected, the general thresholds are automatically reset, and the threshold learning and adjustment process for the new batch is re-executed to avoid mixing thresholds from different batches.

[0154] During cluster analysis, if the parameter distribution of the current batch deviates from that of historical batches by more than 20%, a batch quality warning will be automatically issued, prompting quality personnel to check the incoming quality of the LED batch and detect abnormalities in advance.

[0155] In some embodiments, in the traditional time-division lighting test scheme, a single 4-color LED board needs to complete 4 independent lighting-collection-extinguishing cycles, and the test cycle of a single board is ≥400ms. This cannot be adapted to high-speed SMT assembly lines and roll-to-roll flexible LED board production lines with a cycle time of <500ms. If multiple channels are forced to light up synchronously, it will cause the multi-color light to superimpose and mix, and the sensor will not be able to separate the independent photosensitive parameters of a single channel. The crosstalk rate is 100%, and the test accuracy is completely lost.

[0156] This embodiment is based on orthogonal frequency division multiplexing modulation + frequency domain signal separation technology. By loading non-interfering orthogonal frequency PWM modulation signals onto the four LED channels (red, green, blue, and white), all channels are synchronously lit with constant current, compressing the test cycle from the source. The VEML3328 sensor is controlled to perform continuous high-speed acquisition at a sampling rate much higher than the modulation frequency to obtain continuous light intensity data in the time domain. Through Fast Fourier Transform (FFT) accelerated by MCU hardware, the time domain signal is converted into a frequency domain signal, accurately separating the independent amplitude parameters of the R, G, and B channels corresponding to each modulation frequency, completely eliminating multi-color light crosstalk. The separated single-channel parameters are fully compatible with the original color determination logic, without modifying the core determination rules. At the same time, the DC component of ambient light can be directly filtered out, simplifying the ambient light calibration process.

[0157] Select an industrial-grade MCU (such as STM32H750) with hardware FFT acceleration and 8-channel high-speed PWM output, with a main frequency ≥400MHz, to ensure the real-time performance of FFT operation and high-speed sampling.

[0158] Drive circuit modification: The control terminal of each LED color channel is independently connected to the high-speed PWM output pin of the MCU. A constant current drive chip is used to replace the traditional switching transistor to ensure the linear stability of LED brightness and chromaticity during PWM modulation. The VEML3328 is configured in high-speed continuous sampling mode, the integration time limit is turned off, and the sampling rate is set to more than 10 times the highest modulation frequency (e.g., when the highest modulation frequency is 4kHz, the sampling rate is ≥40kHz) to ensure the integrity of time domain sampling.

[0159] Orthogonal frequencies without harmonic overlap are selected as the modulation frequencies for the four channels. The standard configuration is: 1kHz for the red light channel, 2kHz for the green light channel, 3kHz for the blue light channel, and 4kHz for the white light channel, with a frequency interval of ≥1kHz to avoid harmonic interference. The PWM ratio is uniformly set to 50% to ensure uniform light emission energy for each channel. Dead time is also configured to avoid overlapping of modulation signal edges. The PWM modulation signals for all channels are generated by the same timer of the MCU to ensure synchronous triggering with a phase difference of 0.

[0160] After the board under test is powered on, the MCU synchronously triggers the PWM modulation signal of the four LED channels through the timer. The four LEDs (red, green, blue, and white) are lit synchronously with constant current, without the need for time-division control. At the same time, the high-speed continuous sampling of the VEML3328 is started. A 20ms light-up and modulation stabilization time is reserved to ensure that the LED light-up and PWM modulation enter a steady state before the formal sampling is performed.

[0161] The sensor continuously acquires 1024 points of raw time-domain R, G, and B channel data. The MCU converts the 1024 points of time-domain data into frequency-domain data through a hardware FFT accelerator, while adding a Hanning window to suppress spectral leakage. In the frequency-domain data, the amplitude corresponding to the 1kHz, 2kHz, 3kHz, and 4kHz frequency points is extracted, which correspond to the independent R, G, and B parameters of the red, green, blue, and white light channels, respectively. At the same time, the DC component of ambient light at frequency 0 is directly filtered out, without the need for additional ambient light calibration. The four sets of independent parameters are then input into the preset color determination logic to complete the actual emission color determination of each channel, realizing color recognition of four channels in a single sampling.

[0162] The actual judgment color of each of the four channels is compared with the preset color of the corresponding channel to determine the pass / fail status of each channel. After the pass / fail status of all channels is determined simultaneously, all LEDs are immediately turned off. The test results are summarized, stored locally and uploaded to the host computer within 10ms, and a pass / fail signal is triggered at the same time. After the full process test of a single board is completed, the test trigger waiting state of the next board is immediately entered, realizing continuous testing on the production line without interruption.

[0163] To address power supply harmonic interference, a harmonic notch filtering algorithm is added during frequency domain separation to filter out interference signals from the 50 / 60Hz power supply and its harmonics. For critical value determination, the number of sampling points is automatically increased to 2048 points to improve frequency domain resolution and reduce the false judgment rate. After every 1000 boards are tested, standard light source calibration is automatically performed to correct gain drift under high-speed sampling of the sensor.

[0164] In some embodiments, traditional LED testing solutions can only determine the passability of light emission color and conductivity at the factory, but cannot identify potential early failures of LEDs (such as chip lattice defects, poor solder joints, poor package hermeticity, and phosphor aging). These defective products pass factory tests, but after hundreds of hours of use by the customer, they will quickly exhibit problems such as light decay, color shift, and dead LEDs. This is especially problematic in high-reliability scenarios such as automotive, medical, and industrial control, where zero-defect requirements are extremely high, and traditional solutions cannot meet these requirements.

[0165] This embodiment upgrades the core testing process from "static pass / fail judgment" to "full lifecycle health prediction." Building upon existing color testing, it simultaneously collects four core reliability physical quantities: LED forward voltage (Vf), junction temperature rise, luminous flux maintenance rate, and color coordinate drift. These are combined with R / G / B photosensitivity parameters to form a multi-physical quantity feature matrix. Based on the labeled dataset from LED accelerated aging tests, a fusion prediction model combining Weibull lifetime distribution and random forest classification is trained. Through these multi-physical quantity features, it accurately predicts the 1000-hour light decay rate, early failure probability, and lifespan of the LED. During mass production testing, it not only determines whether the current emission color is qualified but also filters out products with early failure risks, achieving an upgrade from "post-event repair" to "pre-event interception," completely preventing the outflow of defective products.

[0166] Without increasing the testing cycle, it can achieve factory prediction of LED reliability, intercept more than 95% of products with early failure potential, significantly reduce customer complaints and repair costs, and is fully compatible with mass production scenarios with automotive-grade and high reliability requirements such as automotive and medical applications.

[0167] A high-precision sampling resistor with an accuracy of 0.1Ω / 1% is connected in series with each LED channel. The two ends of the sampling resistor are connected to the 16-bit high-precision ADC pin of the MCU to synchronously collect the forward current and forward voltage Vf of the LED. An NTC thermistor (accuracy ±0.2℃) is mounted at the corresponding position of each test station, with a 1mm gap between it and the LED under test, to accurately collect the junction temperature rise data after the LED is lit. Using the Clear channel of VEML3328, a linear mapping relationship between the Clear channel value and the luminous flux is established through standard luminous flux source calibration to realize the real-time calculation of luminous flux.

[0168] The labeled dataset was constructed by selecting target LED models and completing a 1000-hour accelerated aging test. Full data such as Vf, temperature rise, R / G / B parameters, color coordinates, and luminous flux were collected before aging. Simultaneously, the light decay rate, failure type, and failure time after aging were labeled, constructing a labeled dataset of no less than 20,000 samples. Using multi-physical quantity features as input and "whether the light decay rate exceeds 30% after 1000 hours" and "whether early failure will occur" as output labels, a fusion model of Weibull distribution + random forest was trained. Five-fold cross-validation was used to ensure the model's prediction accuracy was ≥95%. The trained model was converted into lightweight inference code and deployed to the host computer of the test system, supporting real-time inference for single boards with an inference time ≤50ms.

[0169] After the board under test is powered on, the corresponding LEDs are lit up one by one in a preset order. At the same time as the LEDs are lit up, the ADC is started to collect the forward voltage Vf and the NTC is started to collect the ambient substrate temperature. Timing is started. During the lighting process, a 100ms over-stress pulse of 1.2 times the rated current is applied to the LEDs to accelerate the exposure of potential defects. After the over-stress ends, the rated current is restored and subsequent data acquisition is performed.

[0170] The system collects the R, G, B, and Clear channel parameters of the currently lit LED, performs ambient light correction, executes the preset color determination logic to obtain the actual emitted color, and calculates the color coordinates and luminous flux value based on the parameters. It also collects the temperature rise data 100ms after the LED is lit, calculates the temperature rise rate, and simultaneously collects the steady-state forward voltage Vf, comparing it with the rated range. The system combines the R / G / B parameter deviations, Vf deviations, temperature rise rate, luminous flux value, and color coordinate drift into a multi-physical quantity feature matrix, inputs it into a lightweight prediction model, and infers the current LED health score (0-100 points), the 1000-hour light decay prediction value, and the probability of early failure.

[0171] Only when both conditions are met simultaneously—the actual emitted color matching the preset color, a health score ≥ 60 points, and a predicted light decay rate ≤ 30% after 1000 hours—is the LED deemed ultimately qualified. If the color is qualified but the health score is not, it is classified as an "early failure potential product," and the corresponding potential problem type is marked (e.g., abnormal temperature rise due to poor soldering, phosphor aging potential, chip defect light decay risk). After completing all channel tests, the test results of the entire board are summarized, and the qualified / unqualified judgment, the location and type of potential products, and the health report are output and simultaneously uploaded to the production line MES system and reliability traceability system. For potential products, the corresponding failure analysis and rework guidance are automatically pushed to achieve accurate interception and location of defective products.

[0172] After each batch of mass production testing is completed, the defective product data and production line rework data returned by the client are added to the dataset to incrementally train the model and continuously improve the prediction accuracy. For different models and batches of LEDs, the corresponding model weights are automatically adapted without the need to retrain the full model.

[0173] In some embodiments, LEDs in products such as flexible OLED screens, automotive curved backlights, irregularly shaped smart signs, and wearable device ring backlights are distributed on a three-dimensional curved surface. The distance and emission angle between LEDs and sensors at different locations vary greatly (the distance difference between the edge and the center can be more than 10mm). Traditional fixed-station, fixed-parameter testing solutions result in LED light intensity deviations exceeding 30% in the curved surface edge areas, easily leading to color deviation misjudgments and missed brightness detections, making them unsuitable for mass production testing of irregularly shaped LED boards.

[0174] This embodiment utilizes 3D laser contour scanning, multi-degree-of-freedom adaptive focusing, and dynamic parameter matching technology to achieve high-precision testing of irregularly shaped curved LEDs across the entire area. A line laser ranging module scans the 3D contour of the test panel, constructing a spatial coordinate model of all LEDs and accurately calculating the relative distance between each LED and the sensor, as well as the angle of its emission normal. A three-axis stepper motor slide drives the sensor array to automatically move to the optimal focusing position for each LED, adjusting the pitch angle to match the LED's emission normal, ensuring that the photosensitive surface of each LED is perpendicular to the emission direction. Based on the focusing distance and emission angle, the sensor's gain, integration time, and sampling count are dynamically matched to ensure that the collected data for each LED is within the sensor's linear operating range, completely eliminating parameter deviations caused by spatial position differences. It is adaptable to testing irregularly shaped LED panels of any curvature and shape, with a full-area testing accuracy deviation of ≤3%, filling the industry gap of lacking standardized solutions for mass production testing of flexible / irregularly shaped LEDs. It is also compatible with conventional testing of planar LED panels, demonstrating strong versatility.

[0175] A high-precision X / Y / Z three-axis stepper motor slide is constructed, with a repeatability of ±0.02mm and a Z-axis travel of ≥50mm, adaptable to curved panels with different curvatures. The slide is controlled by an MCU through a stepper motor driver chip, supporting interpolation motion and precise positioning. A line laser ranging module is installed at the moving end of the slide, with a sampling frequency of ≥100Hz, a ranging accuracy of ±0.1mm, and a scanning area covering the entire area of ​​the board under test. By using a VEML3328 sensor with a 2×2 array and a micro servo gimbal with adjustable pitch angle, an angle adjustment of ±30° can be achieved to match LEDs with different light emission angles. A customized flexible vacuum adsorption fixture can fix curved and flexible LED boards, ensuring that the boards do not deform or shift during testing.

[0176] For irregularly shaped LED boards of the target model, a full-area laser scan is performed in advance to construct a three-dimensional digital model of the board, marking the spatial coordinates, normal angles, and optimal focusing distances of all LEDs, and generating a standardized test path file. For boards of the same model to be tested, a rapid feature scan is performed before testing to identify more than three positioning markers, automatically matching the pre-stored three-dimensional model, correcting the positional deviation caused by board clamping, and generating an accurate test path adapted to the current board.

[0177] After the board under test is fixed, the MCU controls the slide to drive the laser module to perform rapid feature scanning, complete the board positioning and 3D model matching, and generate the test path. According to the test path, the slide moves the sensor array to directly above the first LED under test. The servo gimbal is used to adjust the sensor pitch angle so that the sensor photosensitive surface is perpendicular to the LED's light emission normal. At the same time, the Z-axis is adjusted to the optimal focusing distance. After positioning is completed, the MCU outputs a control signal to light up the current LED under test, while all other LEDs remain off, allowing 30ms for light emission stabilization time.

[0178] Based on the current LED focusing distance, the gain level and integration time of the VEML3328 are dynamically adjusted: for every 2mm increase in focusing distance, the gain is increased by 1 level and the integration time is extended by 20ms, ensuring that the acquired R / G / B parameters are within the optimal linear range of the sensor (30%-70%). Based on the angular differences of the four sensors in the sensor array, the acquired parameters are angularly corrected to eliminate color shift errors caused by the emission angle, resulting in corrected effective R / G / B parameters. The corrected parameters are then input into a preset color determination logic to determine the actual emitted color of the current LED, while simultaneously calculating the brightness parameters to determine if they meet the specifications.

[0179] After completing the test of the current LED, the LED is turned off, and the slide moves to the next LED to be tested according to the test path. The above positioning-lighting-data acquisition-judgment process is repeated to complete the test of all LEDs on the board point by point. After the entire board is tested, the MCU summarizes all test results and generates a test heat map corresponding to the 3D model. This map visually displays the color, brightness, and pass / fail distribution of each LED and accurately marks the 3D coordinates of abnormal LEDs. The final pass / fail judgment is output, and the test data and 3D heat map are simultaneously uploaded to the host computer and stored in the product traceability system. For non-conforming products, the 3D coordinates, deviation value, and fault type of the abnormal location are automatically output to assist in subsequent rework.

[0180] Before each batch of tests, the laser ranging module is calibrated using a standard height block to correct the Z-axis distance error. For curved panels with large curvature, a zoned focusing test is adopted, with the focusing parameters of each zone adjusted separately to ensure consistent accuracy across the entire area. For clamping deformation of flexible panels, the test path is automatically corrected through real-time scanning to avoid test failures caused by clamping errors.

[0181] In some embodiments, a consumer electronics backlight production line typically has 50+ parallel testing stations. Each station has subtle differences in sensors, fixtures, and environment. Over time, issues such as sensor gain drift, fixture wear, and changes in ambient light can arise, leading to test results for the same standard board varying by more than 15% across different stations. This results in significant fluctuations in production line yield and extremely high difficulty in quality control. Traditional single-station offline calibration solutions require downtime for individual calibrations, with a single full-line calibration taking ≥4 hours, severely impacting production line uptime and failing to achieve real-time control of testing consistency across the entire production line.

[0182] This embodiment utilizes digital twin technology combined with traceable synchronous calibration for production line testing systems to achieve non-stop online synchronous calibration of all testing stations on the production line. A 1:1 digital twin is constructed for each testing station on the production line, mapping the station's hardware parameters, test data, environmental data, and calibration history in real time, thus building a digital twin platform for the entire production line testing system. Using the main standard station calibrated by relevant metrology units as the traceability benchmark, the system deviation between each station and the benchmark is calculated in real time through the twin model's deviation analysis algorithm. Standard calibration boards are circulated to achieve non-stop online calibration of all production line stations. The twin model automatically generates correction coefficients for each station, updating sensor parameters and judgment thresholds online without downtime or manual intervention. Through the twin model's predictive maintenance algorithm, faults such as sensor drift and fixture wear are predicted in advance, triggering calibration early to avoid yield fluctuations.

[0183] Based on the Industrial Internet platform, a 1:1 digital twin is constructed for each testing station on the production line. Each twin includes: a hardware attribute module (sensor model, calibration date, gain parameters, IO configuration, fixture specifications); a real-time operation module (real-time test data, yield, parameter distribution, ambient temperature / light intensity, equipment operating status); a calibration model module (historical calibration data, system deviation model, correction coefficient, threshold configuration); a fault early warning module (drift trend, wear prediction, maintenance reminders); and a data communication architecture (real-time data interaction between the control unit of each station and the twin platform via industrial Ethernet, with a data update frequency of ≤1s to ensure real-time mapping between the twin and the physical station). One station is selected as the master standard station, whose sensors, standard light sources, and calibration boards are all calibrated by the National Institute of Metrology, serving as the traceability benchmark for the entire production line, and its test data as the standard reference value for the entire production line. Customized standard 4-color LED boards, calibrated at the main standard workstation, are used. The color coordinates, brightness, and forward voltage of the LEDs on the boards are all metrologically calibrated with a parameter uncertainty of ≤1%, serving as the transfer standard for calibration across the entire production line. Dual trigger conditions are set, and simultaneous calibration across the entire production line is triggered when either condition is met: automatic triggering after every 10,000 boards are tested, or every 7 days; automatic triggering of calibration across the entire production line when the twin model detects that the test data at a certain workstation deviates from the main standard workstation by more than 3%; and automatic circulation of the standard calibration boards among various workstations via AGV carts on the production line, without manual intervention. The calibration process runs parallel to normal production without downtime.

[0184] The twin platform issues a calibration command, and the production line AGV transports the standard calibration board to the target test station. The station fixture automatically fixes the calibration board, and the station control unit lights up the red, green, blue, and white standard LEDs on the calibration board one by one according to the S101 standard procedure. Throughout the process, the twin platform monitors the lighting control timing and power supply parameters of the station in real time to ensure that the test conditions are completely consistent with those of the main standard station.

[0185] Following the standard procedure of S102, the workstation control unit collects the R / G / B parameters of the standard LED to complete the color determination. Simultaneously, it uploads the collected raw parameters and determination results to the twin platform in real time. The twin platform compares the collected parameters of this workstation with the standard parameters of the main standard workstation and calculates three types of correction coefficients for this workstation using a deviation analysis algorithm: sensor gain correction coefficient (correcting gain drift from long-term sensor use); ambient light compensation coefficient (correcting substrate deviation caused by differences in ambient light at the workstation); and determination threshold correction coefficient (correcting determination boundary deviation caused by differences in fixtures and optical paths). Simultaneously, the twin platform identifies hardware faults such as fixture wear, sensor aging, and optical path obstruction based on parameter deviation characteristics and outputs early warning information.

[0186] The twin platform automatically sends the calculated correction coefficients to the control units at the corresponding workstations. The control units update the parameter configurations in the Flash memory online, without requiring downtime or restart, and the changes take effect immediately. After the parameters are updated, the workstation immediately performs a secondary test on the standard calibration board and uploads the test results to the twin platform to verify that the deviation between the corrected test data and the standard value is ≤3%, confirming that the calibration is qualified. After the standard calibration board completes the calibration of all workstations in sequence, it returns to the main standard workstation for verification to ensure that the parameters of the standard board have not changed during the calibration process and that the calibration values ​​are valid. After the calibration of the entire production line is completed, the twin platform generates a calibration report, which includes the deviation before and after calibration, correction coefficients, and hardware status warnings for each workstation, and uploads it synchronously to the factory quality management system.

[0187] The twin platform analyzes historical data from each workstation to predict sensor drift trends and fixture wear, providing maintenance reminders 7 days in advance for predictive maintenance and preventing production line downtime due to sudden failures. For different product models, the twin platform automatically matches the corresponding calibration parameter library, enabling one-click calibration during product changeovers without the need for re-adjustment. Regularly updating the metrological benchmarks of the entire production line through metrological calibration at the main standard workstation ensures the traceability of test data across the entire production line and meets relevant quality system requirements.

[0188] Please see Figure 4 As shown, Figure 4 This is a schematic diagram of the structure of the LED test circuit test system 200 provided in the embodiments of this application. The LED test circuit test system 200 is used to execute the steps of the LED test circuit test methods shown in the above embodiments. The LED test circuit test system 200 can be a single server or a server cluster, or it can be a terminal, such as a handheld terminal, a laptop computer, a wearable device, or a robot.

[0189] like Figure 4 As shown, the LED test circuit test system 200 includes:

[0190] The sequential lighting unit 201 is used to provide test power to the board under test on which multi-color LEDs are installed, and to sequentially light up LEDs of different colors on the board under test according to preset colors.

[0191] The parameter acquisition unit 202 is used to acquire three light-sensing parameters of the current light emission through a photosensitive sensor after each light-emitting diode of the color to be detected is lit, compare the magnitude of the three light-sensing parameters, and determine the actual color emitted by the current light-emitting diode according to the preset color judgment logic and the relationship between the magnitude of the three light-sensing parameters.

[0192] The test completion unit 203 is used to compare the actual determined color with the currently lit preset color. If they match, the LED is determined to be working normally; if they do not match, it is determined to be working abnormally. The test is completed sequentially for all LEDs of the colors to be tested to obtain the final LED test results.

[0193] In some embodiments, the step of providing test power to the test board on which the multi-color LEDs to be tested are installed, and lighting up the LEDs of different colors on the test board one by one according to the preset colors, includes: lighting up only one LED corresponding to a color channel at a time according to the color lighting order corresponding to the preset colors, reserving a preset light stabilization time after the lighting operation is completed, and then performing the subsequent photosensitive parameter acquisition steps.

[0194] In some embodiments, after each LED of a color to be detected is lit, the three light-sensing parameters of the current light emission are collected by a photosensitive sensor, including: collecting the three original light-sensing values ​​of the current light emission by the photosensitive sensor, converting the three original light-sensing values ​​into decimal values, and obtaining the three light-sensing parameters.

[0195] In some embodiments, the three photosensitive parameters include a first parameter value, a second parameter value, and a third parameter value; the step of comparing the magnitudes of the three photosensitive parameters and determining the actual color emitted by the LED according to a preset color determination logic and the magnitude relationship of the three photosensitive parameters includes: if the first parameter value is greater than the second and third parameter values ​​and the difference between them is greater than a preset difference, and the first parameter value is greater than the second parameter value, and the second parameter value is greater than the third parameter value, the actual color is determined to be red light; if the second parameter value is greater than the first parameter value... If the difference between the numerical values ​​of the first, second, and third channels is greater than a preset difference, and the numerical values ​​of the second, third, and first channels are greater than the numerical values ​​of the third channel, the actual color is determined to be green. If the numerical value of the third channel is significantly greater than the numerical values ​​of the first, second, and third channels, and the difference between them is greater than a preset difference, and the numerical values ​​of the third, second, and first channels are greater than the numerical values ​​of the second channel, the actual color is determined to be blue. If the corresponding difference between the numerical values ​​of the first, second, and third channels is less than or equal to the preset difference, the actual color is determined to be white.

[0196] In some embodiments, comparing the actual determined color with the currently lit preset color, and determining that the light-emitting diode is working normally if they match, and that it is working abnormally if they do not match, includes: after each comparison, storing the current position of the light-emitting diode, the preset color, the actual determined color and the determination result in the test cache; after all tests are completed, organizing the determination results of all light-emitting diodes to generate test records, and marking the information of all light-emitting diodes that are working abnormally.

[0197] In some embodiments, the method further includes: before lighting all light-emitting diodes, pre-collecting three photosensitive reference parameters under ambient light; after collecting the photosensitive parameters of the currently lit light-emitting diode each time, removing the influence of the corresponding ambient light reference parameters from the three collected parameters to obtain corrected photosensitive parameters after eliminating ambient light interference; and then using the corrected photosensitive parameters to perform subsequent size comparison steps to avoid interference from ambient light in the production site with the test results.

[0198] In some embodiments, the method further includes: after each preset number of board tests are completed, automatically triggering a photosensitive sensor calibration process, controlling a preset standard color light source to emit light of a preset standard color in sequence, sequentially collecting photosensitive parameters corresponding to each standard color, comparing the collected parameters with the standard parameter range, adjusting the gain level of the photosensitive sensor to make the collected parameters fall within the standard parameter range, completing automatic calibration, and ensuring the accuracy of long-term batch testing.

[0199] In some embodiments, the method further includes: pre-collecting test data of all defective LED boards labeled with actual abnormal causes during the mass production process, and training an intelligent classification model for abnormal causes using the labeled test data; after all tests are completed and all abnormal results and corresponding test parameters of the current board under test are obtained, extracting the degree of deviation of photosensitive parameters of each channel of the board and the location distribution characteristics of abnormal channels, inputting the extracted features into the trained intelligent classification model for abnormal causes, and obtaining the specific abnormal cause corresponding to the current defective board through model reasoning, specifically distinguishing four different abnormal types: LED body color deviation abnormality, pin soldering contact abnormality, power supply channel fault abnormality, and control logic abnormality, and then sending the abnormal type and corresponding location information together to the rework management system, automatically pushing standardized rework guidance schemes for the corresponding abnormal type, so as to facilitate maintenance personnel to quickly locate the fault point.

[0200] In some embodiments, the method further includes: after a preset number of tests are completed on the same type and batch of test boards, extracting the photosensitive parameter data of each color for all boards that have been determined to be qualified, performing cluster analysis on the parameter distribution of each color, identifying the overall parameter offset of the current batch due to LED production consistency deviation, automatically adjusting the color judgment threshold of subsequent tests of the current batch according to the offset amount, replacing the general threshold stored in the parameter library, and using the adjusted new threshold to perform color judgment for subsequent tests of boards in the same batch, avoiding test misjudgment caused by inherent brightness deviation of LEDs in different batches, and improving the accuracy of batch testing.

[0201] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the LED test circuit test system and each module described above can be referred to the corresponding content in the various embodiments of the LED test circuit test method, and will not be repeated here.

[0202] The testing method for the LED test circuit described above can be implemented as a computer program, which can be used in various applications such as... Figure 4 It runs on the device shown.

[0203] Please see Figure 5 , Figure 5 This is a schematic block diagram of the control unit provided in an embodiment of this application. The control unit includes a processor, a memory, and a network interface connected via a device bus, wherein the memory may include a storage medium and internal memory.

[0204] The storage medium can store operating devices and computer programs. The computer program includes program instructions that, when executed, cause the processor to perform any testing method for the LED test circuit.

[0205] The processor provides computing and control capabilities to support the operation of the entire control unit.

[0206] The internal memory provides an environment for the execution of computer programs stored in non-volatile storage media. When the computer program is executed by the processor, it enables the processor to perform any test method for LED test circuits.

[0207] This network interface is used for network communication, such as sending assigned tasks. Those skilled in the art will understand that... Figure 5 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the terminal to which the present application is applied. The specific control unit may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0208] It should be understood that the processor can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among these, a general-purpose processor can be a microprocessor or any conventional processor.

[0209] In one embodiment, the processor is configured to run a computer program stored in memory to perform the following steps:

[0210] Provide test power to the board under test that is equipped with multi-color LEDs, and light up the LEDs of different colors on the board under test one by one according to the preset colors;

[0211] After each LED of the color to be detected is lit, the three photosensitive parameters of the current emission are collected by the photosensitive sensor. The three photosensitive parameters are compared and the actual color emitted by the LED is determined according to the preset color determination logic and the relationship between the three photosensitive parameters.

[0212] The actual determined color is compared with the currently lit preset color. If they match, the LED is considered to be working normally; otherwise, it is considered to be working abnormally. The test of all LEDs of the colors to be tested is completed in turn to obtain the final LED test results.

[0213] In some embodiments, the step of providing test power to the test board on which the multi-color LEDs to be tested are installed, and lighting up the LEDs of different colors on the test board one by one according to the preset colors, includes: lighting up only one LED corresponding to a color channel at a time according to the color lighting order corresponding to the preset colors, reserving a preset light stabilization time after the lighting operation is completed, and then performing the subsequent photosensitive parameter acquisition steps.

[0214] In some embodiments, after each LED of a color to be detected is lit, the three light-sensing parameters of the current light emission are collected by a photosensitive sensor, including: collecting the three original light-sensing values ​​of the current light emission by the photosensitive sensor, converting the three original light-sensing values ​​into decimal values, and obtaining the three light-sensing parameters.

[0215] In some embodiments, the three photosensitive parameters include a first parameter value, a second parameter value, and a third parameter value; the step of comparing the magnitudes of the three photosensitive parameters and determining the actual color emitted by the LED according to a preset color determination logic and the magnitude relationship of the three photosensitive parameters includes: if the first parameter value is greater than the second and third parameter values ​​and the difference between them is greater than a preset difference, and the first parameter value is greater than the second parameter value, and the second parameter value is greater than the third parameter value, the actual color is determined to be red light; if the second parameter value is greater than the first parameter value... If the difference between the numerical values ​​of the first, second, and third channels is greater than a preset difference, and the numerical values ​​of the second, third, and first channels are greater than the numerical values ​​of the third channel, the actual color is determined to be green. If the numerical value of the third channel is significantly greater than the numerical values ​​of the first, second, and third channels, and the difference between them is greater than a preset difference, and the numerical values ​​of the third, second, and first channels are greater than the numerical values ​​of the second channel, the actual color is determined to be blue. If the corresponding difference between the numerical values ​​of the first, second, and third channels is less than or equal to the preset difference, the actual color is determined to be white.

[0216] In some embodiments, comparing the actual determined color with the currently lit preset color, and determining that the light-emitting diode is working normally if they match, and that it is working abnormally if they do not match, includes: after each comparison, storing the current position of the light-emitting diode, the preset color, the actual determined color and the determination result in the test cache; after all tests are completed, organizing the determination results of all light-emitting diodes to generate test records, and marking the information of all light-emitting diodes that are working abnormally.

[0217] In some embodiments, the method further includes: before lighting all light-emitting diodes, pre-collecting three photosensitive reference parameters under ambient light; after collecting the photosensitive parameters of the currently lit light-emitting diode each time, removing the influence of the corresponding ambient light reference parameters from the three collected parameters to obtain corrected photosensitive parameters after eliminating ambient light interference; and then using the corrected photosensitive parameters to perform subsequent size comparison steps to avoid interference from ambient light in the production site with the test results.

[0218] In some embodiments, the method further includes: after each preset number of board tests are completed, automatically triggering a photosensitive sensor calibration process, controlling a preset standard color light source to emit light of a preset standard color in sequence, sequentially collecting photosensitive parameters corresponding to each standard color, comparing the collected parameters with the standard parameter range, adjusting the gain level of the photosensitive sensor to make the collected parameters fall within the standard parameter range, completing automatic calibration, and ensuring the accuracy of long-term batch testing.

[0219] In some embodiments, the method further includes: pre-collecting test data of all defective LED boards labeled with actual abnormal causes during the mass production process, and training an intelligent classification model for abnormal causes using the labeled test data; after all tests are completed and all abnormal results and corresponding test parameters of the current board under test are obtained, extracting the degree of deviation of photosensitive parameters of each channel of the board and the location distribution characteristics of abnormal channels, inputting the extracted features into the trained intelligent classification model for abnormal causes, and obtaining the specific abnormal cause corresponding to the current defective board through model reasoning, specifically distinguishing four different abnormal types: LED body color deviation abnormality, pin soldering contact abnormality, power supply channel fault abnormality, and control logic abnormality, and then sending the abnormal type and corresponding location information together to the rework management system, automatically pushing standardized rework guidance schemes for the corresponding abnormal type, so as to facilitate maintenance personnel to quickly locate the fault point.

[0220] In some embodiments, the method further includes: after a preset number of tests are completed on the same type and batch of test boards, extracting the photosensitive parameter data of each color for all boards that have been determined to be qualified, performing cluster analysis on the parameter distribution of each color, identifying the overall parameter offset of the current batch due to LED production consistency deviation, automatically adjusting the color judgment threshold of subsequent tests of the current batch according to the offset amount, replacing the general threshold stored in the parameter library, and using the adjusted new threshold to perform color judgment for subsequent tests of boards in the same batch, avoiding test misjudgment caused by inherent brightness deviation of LEDs in different batches, and improving the accuracy of batch testing.

[0221] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, causes the processor to implement the steps of the test method for the LED test circuit provided in any embodiment of this application.

[0222] The computer-readable storage medium can be an internal storage unit of the control unit described in the foregoing embodiments, such as the hard disk or memory of the control unit. Alternatively, the computer-readable storage medium can be an external storage device of the control unit, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., mounted on the control unit.

[0223] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A testing method for an LED test circuit, characterized in that, include: Provide test power to the board under test that is equipped with multi-color LEDs, and light up the LEDs of different colors on the board under test one by one according to the preset colors; After each LED of the color to be detected is lit, the three photosensitive parameters of the current emission are collected by the photosensitive sensor. The three photosensitive parameters are compared and the actual color emitted by the LED is determined according to the preset color determination logic and the relationship between the three photosensitive parameters. The actual determined color is compared with the currently lit preset color. If they match, the LED is considered to be working normally; otherwise, it is considered to be working abnormally. The test of all LEDs of the colors to be tested is completed in turn to obtain the final LED test results. The three photosensitive parameters include a first parameter value, a second parameter value, and a third parameter value. The comparison of the three photosensitive parameters, and the determination of the actual color emitted by the LED based on the relationship between the three photosensitive parameters according to a preset color determination logic, includes: if the first parameter value is greater than both the second and third parameter values, and the difference between them is greater than a preset difference, and the first parameter value is greater than the second parameter value, and the second parameter value is greater than the third parameter value, then the actual color is determined to be red; if the second parameter value is greater than both the first and third parameter values, then the actual color is determined to be red. If the values ​​of the third-path parameters are all greater than the preset difference, and the values ​​of the second-path parameters are greater than the values ​​of the third-path parameters, and the values ​​of the third-path parameters are greater than the values ​​of the first-path parameters, then the actual color is determined to be green light; if the values ​​of the third-path parameters are significantly greater than the values ​​of the first-path and second-path parameters, and the differences between them are all greater than the preset difference, and the values ​​of the third-path parameters are greater than the values ​​of the second-path parameters, and the values ​​of the second-path parameters are greater than the values ​​of the first-path parameters, then the actual color is determined to be blue light; if the corresponding differences between the values ​​of the first-path, second-path, and third-path parameters are less than or equal to the preset difference, then the actual color is determined to be white light. The method further includes: before lighting all light-emitting diodes, pre-collecting three photosensitive reference parameters under ambient light; after collecting the photosensitive parameters of the currently lit light-emitting diode each time, removing the influence of the corresponding ambient light reference parameters from the three collected parameters to obtain corrected photosensitive parameters after eliminating ambient light interference; and then using the corrected photosensitive parameters to perform subsequent size comparison steps to avoid interference from ambient light in the production site with the test results.

2. The method according to claim 1, characterized in that, The process of providing test power to the test board on which the multi-color LEDs to be tested are installed, and lighting up the LEDs of different colors on the test board one by one according to preset colors, includes: According to the preset color lighting sequence, only one LED corresponding to a color channel is lit at a time. After the lighting operation is completed, a preset light stabilization time is reserved before proceeding with the subsequent photosensitive parameter acquisition steps.

3. The method according to claim 1, characterized in that, After each LED of the color to be detected is lit, three photosensitive parameters of the current emission are collected by a photosensitive sensor, including: The three raw photosensitive values ​​of the current emission are collected by the photosensitive sensor, and the three raw photosensitive values ​​are converted into decimal values ​​to obtain the three photosensitive parameters.

4. The method according to claim 1, characterized in that, The step of comparing the actually determined color with the currently lit preset color, and determining that the LED is working normally if they match, and abnormally if they do not match, includes: After each comparison, the current position of the LED, the preset color, the actual judgment color, and the judgment result are stored in the test cache. After all tests are completed, the judgment results of all LEDs are organized to generate test records, and information of all LEDs that are malfunctioning is marked.

5. The method according to claim 1, characterized in that, The method further includes: After a preset number of boards are tested, the photosensitive sensor calibration process is automatically triggered. The preset standard color light source is controlled to emit light of the preset standard color in sequence, and the photosensitive parameters corresponding to each standard color are collected in sequence. The collected parameters are compared with the standard parameter range, and the gain level of the photosensitive sensor is adjusted so that the collected parameters fall within the standard parameter range, thus completing the automatic calibration and ensuring the accuracy of long-term batch testing.

6. The method according to claim 1, characterized in that, The method further includes: Test data of all defective LED boards labeled with actual abnormalities during mass production are collected in advance. An intelligent classification model for abnormality causes is trained using the labeled test data. After all tests are completed and all abnormal results and corresponding test parameters of the current board under test are obtained, the degree of deviation of photosensitive parameters of each channel of the board and the location distribution characteristics of abnormal channels are extracted. The extracted features are input into the trained intelligent classification model for abnormality causes. The model infers the specific abnormality cause corresponding to the current defective board, specifically distinguishing four different abnormality types: abnormal LED body color, abnormal pin soldering, abnormal power supply channel failure, and abnormal control logic. The abnormality type and corresponding location information are then sent to the rework management system, which automatically pushes the standardized rework guidance plan for the corresponding abnormality type, making it convenient for maintenance personnel to quickly locate the fault point.

7. The method according to claim 1, characterized in that, The method further includes: Once a preset number of tests have been completed on boards of the same model and batch, the photosensitive parameter data for each color of all qualified boards are extracted. Cluster analysis is performed on the parameter distribution of each color to identify the overall parameter offset caused by LED production inconsistency deviations in the current batch. Based on the offset, the color judgment threshold for subsequent tests in the current batch is automatically adjusted, replacing the general threshold stored in the parameter library. Subsequent tests of boards in the same batch use the adjusted new threshold to perform color judgment, avoiding test misjudgments caused by inherent brightness deviations of LEDs in different batches and improving the accuracy of batch testing.

8. An LED testing circuit, characterized in that, Includes a power supply module, a photosensor, and a control unit; The power supply module is used to provide test power to the board under test on which the multicolor light-emitting diodes under test are installed; The photosensitive sensor is set at the corresponding photosensitive position of the LED under test, and is used to collect three photosensitive parameters of the light output by the LED after it is lit. The control unit is electrically connected to the power supply module, the photosensitive sensor, and the light emission control terminal of the board under test, respectively, and the control unit is used to implement the method as described in any one of claims 1-7.

Images