High-speed precise surface mounting system for electronic components

Abstract

The invention relates to the technical field of precise mounting systems for electronic components, and discloses a high-speed precise mounting system for electronic components, which comprises a first vibration sensing unit and a second vibration sensing unit which are used for acquiring vibration signals of a mounting head and a substrate table; the logic processing module is used for calculating a relative vibration signal, predicting a zero point window and synchronously triggering a Z axis in a first mode, switching to a second mode after triggering, defining the two signals as a near-field excitation source and a far-field response signal, inputting a self-adaptive filtering algorithm to calculate a current characteristic coefficient representing a dynamic transfer characteristic, and outputting the current characteristic coefficient to the second mode; according to the method, the relative vibration zero point is predicted, mounting is triggered synchronously, traditional stable time waiting is replaced, speed and precision restrictions are decoupled, dynamic transfer characteristics are calibrated through a multiplex sensor and an adaptive filtering algorithm, and the accuracy of the dynamic transfer characteristics is improved. And the anti-interference online diagnosis of the Z-axis execution process quality is realized.

Description

Technical Field

[0001] This invention relates to a high-speed precision mounting system for electronic components, belonging to the technical field of precision mounting systems for electronic components. Background Technology

[0002] High-speed precision placement systems are currently key equipment in electronic component manufacturing. Their core task is to drive the placement head to place electronic components at the target position on the substrate with high speed and precision. The efficiency and accuracy of this process directly determine the manufacturing quality and production capacity of electronic products. To achieve high-speed movement, existing technologies require the gantry and placement head to undergo drastic acceleration and deceleration processes. According to physical laws, drastic deceleration inevitably leads to transient vibrations and ringing phenomena in the mechanical structure near the target position. To ensure that the placement accuracy is not affected by this vibration, the common practice in the field is to enforce a fixed stabilization time after the movement stop command is issued, until the vibration amplitude decays to within a safe threshold before allowing the Z-axis actuator to start the placement action. This passive stabilization time constitutes a constraint on manufacturing efficiency and forces the system to make trade-offs between pursuing higher speed and ensuring higher precision.

[0003] To shorten this stabilization time, industry attempts to improve the system have focused on two main directions: one is to physically suppress the generation and amplitude of vibration by using heavier bases, more rigid gantry structures, or more expensive linear motors, but this increases the manufacturing cost and maintenance complexity of the equipment; the other is to attempt to actively dampen or rapidly attenuate vibration through more advanced control algorithms, but this places higher demands on the control system. However, in addition to improving the mechanical hardware structure, existing technologies also have fundamental shortcomings in software control strategies to improve placement efficiency. For example, Chinese utility model patent CN204507453U discloses a high-speed, high-precision electronic packaging robot based on machine vision and laser detection. Although this technology uses a CCD machine vision system... While using integrated laser inspection systems to locate and measure components, claiming to achieve high repeatability, the core idea remains the same: relying on a vision system to acquire still images and transform coordinates of components and target positions, combined with motion control algorithms to guide the robot's movement. This detection mechanism, which relies on still images and absolute coordinates, does not fundamentally resolve the inherent contradiction between mechanical ring-down caused by high-speed movement and placement accuracy. In other words, it still requires waiting for the motion platform to reach a relatively static or acceptable stable state before precise visual measurement and packaging operations can be performed. Essentially, it has not completely escaped the dependence on settling time, and its positioning accuracy is highly susceptible to the external common-mode vibration noise commonly present in the manufacturing environment, making it difficult to maintain high stability in harsh industrial environments.

[0004] Therefore, the technical problem to be solved by this invention is how to establish a new mounting triggering mechanism that no longer passively waits for vibration to be eliminated, but can actively identify the relative positional relationship between the mounting head and the substrate during vibration, and accurately capture the zero-error moment of instantaneous alignment between the two to execute the mounting action. Summary of the Invention

[0005] This invention provides a high-speed precision mounting system for electronic components. Its main purpose is to solve the problem in the prior art that the need to wait for stabilization time in order to avoid the influence of mechanical vibration leads to limitations in manufacturing efficiency and movement speed.

[0006] To achieve the above objectives, the present invention provides a high-speed precision mounting system for electronic components, the system comprising: Placement head; A substrate stage for supporting a substrate; Z-axis actuator is used to drive the placement head to perform placement actions on the substrate stage; The first vibration sensing unit is used to acquire the real-time head vibration signal of the mounting head; The second vibration sensing unit is used to acquire the real-time vibration signal of the substrate stage. The logic processing module is used to: in the first operating mode, calculate the relative vibration signal between the head vibration signal and the substrate vibration signal in real time; based on the relative vibration signal, predict one or more zero-point windows in the future where the relative vibration signal crosses the zero error threshold; and issue a synchronization trigger command to the Z-axis actuator only during the zero-point window; immediately switch to the second operating mode after the synchronization trigger command is issued; in the second operating mode: use the real-time head vibration signal as the near-field excitation source signal and the real-time substrate vibration signal as the far-field response signal; input the near-field excitation source signal and the far-field response signal into an adaptive filtering algorithm to calculate the current characteristic coefficients that characterize the dynamic transmission characteristics between the two in real time; compare the current characteristic coefficients with a preset benchmark characteristic coefficient template representing a healthy execution state, and generate an execution quality criterion for the current operating state of the Z-axis actuator based on the comparison result.

[0007] Preferably, the system further includes a gantry for driving the placement head to move horizontally above the substrate stage; the logic processing module is used to immediately start a first operating mode after the gantry's movement stop command is issued. The first operating mode includes: subtracting the head vibration signal from the substrate vibration signal through a digital differential to generate a relative vibration signal; and predicting the zero-point window and issuing a synchronous trigger command based on the relative vibration signal.

[0008] Preferably, in the first operating mode, the logic processing module is further configured to: lock the resonant frequency and phase of the relative vibration signal through a digital filter; establish an attenuated sine wave model representing the relative vibration signal based on the resonant frequency and phase; and calculate the zero-point window based on the attenuated sine wave model.

[0009] Preferably, the reference characteristic coefficient template is a set of reference coefficient values; the logic processing module is also used to enable the Z-axis actuator to operate in a known healthy state in calibration mode, and during this period continuously acquire near-field excitation source signals and far-field response signals, run an adaptive filtering algorithm until convergence, and store a set of stable state coefficient values ​​after convergence as reference coefficient values.

[0010] Preferably, the logic processing module is used to process the current feature coefficients. With reference characteristic coefficient template All are considered dimensional feature vectors, where The dimension of the feature vectors; the logic processing module calculates the Euclidean distance between two feature vectors. To generate execution quality criteria, where Euclidean distance Determined in the following ways: ,in, and These are the current feature coefficient vector and the reference feature coefficient template vector, respectively. Each component; the logic processing module is also used to process the Euclidean distance. Compare with the preset maintenance threshold.

[0011] Preferably, the system also includes an equipment maintenance interface, and the logic processing module is further used to output a predictive maintenance alarm through the equipment maintenance interface when the execution quality criterion indicates that the Z-axis actuator is in an abnormal operating state.

[0012] Preferably, both the first vibration sensing unit and the second vibration sensing unit are MEMS accelerometers, and are used to acquire real-time head vibration signals and real-time substrate vibration signals at a sampling frequency of 1kHz to 10kHz.

[0013] Preferably, in the second operating mode, the logic processing module is used as the input of the near-field excitation source signal to the adaptive filtering algorithm and the far-field response signal as the desired output of the adaptive filtering algorithm; the adaptive filtering algorithm is used to adjust a set of internal filter coefficients in real time so as to minimize the mean square error between the input signal after filtering and the desired output signal; the current characteristic coefficient is the filter coefficient of the adaptive filtering algorithm at the end of the preset calculation window.

[0014] Preferably, the logic processing module is implemented by a field-programmable gate array (FPGA).

[0015] Preferably, the zero error threshold is preset based on the inherent mounting accuracy tolerance of the high-speed precision mounting system for electronic components.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. By configuring vibration sensing units on the placement head and substrate stage respectively, the real-time focus is not on the absolute vibration of each, but on the relative vibration signal obtained by subtracting the two, which characterizes the true positional relationship between the nozzle and the pad. Based on the analysis of this relative vibration signal, the system actively predicts the precise timing when it will cross the zero-error state in the future, and selects the timing to synchronously trigger the Z-axis placement command. This operating mode transforms the mechanical ringing generated after high-speed movement from a constraint that consumes manufacturing cycle time and must be passively waited for to decay into a trigger timing reference that can be accurately predicted and utilized. As a result, the execution of the placement action no longer depends on the traditional fixed settling time, and the movement speed and placement accuracy are decoupled in the manufacturing process.

[0017] 2. Logically, the placement cycle is divided into two closely linked stages: the zero-point triggering stage and the execution verification stage. The system utilizes the same set of first and second vibration sensing units and processing modules. After completing the triggering task in the first stage, it immediately reuses its function and switches to the second stage to perform in-situ verification of the physical movement process of the Z-axis actuator. This time-division multiplexing mechanism allows the same set of simple physical components to undertake two different types of technical tasks in a single placement process. Without adding new sensing components, the system monitoring is expanded from single trigger timing precision control to physical health diagnosis of the subsequent execution process, thus improving the completeness of the equipment's functions.

[0018] 3. During the quality verification phase, the technical approach does not rely on comparing the currently acquired signal waveform with historical reference waveforms. Instead, the Z-axis actuator is treated as a dynamic physical system. The head vibration signal and the substrate vibration signal acquired at the same time are redefined as the near-field excitation source signal and far-field response signal of the system's internal excitation, respectively. The system uses an adaptive filtering algorithm to calculate the current characteristic coefficients that characterize the dynamic transmission characteristics between the two in real time. Since the environmental vibration or equipment residual vibration that is common in the electrical component manufacturing site is a common-mode noise that is unrelated to the Z-axis execution action, it is suppressed in the algorithm's operation logic. Therefore, the execution quality offset generated by the final criterion can more realistically reflect the changes in the mechanical transmission characteristics of the Z-axis actuator itself, avoiding the problem of false alarms caused by the interference of on-site noise in traditional diagnostic methods. Attached Figure Description

[0019] Figure 1 This is a system functional architecture diagram of the present invention that combines dual-mode processing and dual vibration sensing; Figure 2 This is an iterative curve of the adaptive filter characteristic coefficients under different health states of the present invention; Figure 3 This is a timing flowchart of the relative vibration zero-point prediction and synchronous triggering of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be described in detail below with reference to specific embodiments, but the present invention is not limited to these embodiments.

[0021] This invention provides a high-speed precision mounting system for electronic components, which addresses the problem in electrical component manufacturing where fixed settling times are required to avoid mechanical vibrations caused by high-speed movement, thus limiting manufacturing cycle time and accuracy. The system includes a first operating mode for dynamic triggering and a second operating mode for process diagnostics. Its system architecture mainly includes a mounting head, a substrate stage for supporting the substrate, a Z-axis actuator for driving the mounting head to perform mounting actions, and a first vibration sensing unit and a second vibration sensing unit for sensing vibrations. All of these units are connected to a logic processing module. During the mounting cycle, a high-speed gantry drives the mounting head to move horizontally above the substrate stage. When a stop command is issued for the gantry movement, its mechanical structure will produce a ringing phenomenon. The first vibration sensing unit is fixed to the moving part of the mounting head, and the second vibration sensing unit is fixed to the base of the substrate stage. All sensors are capable of high-frequency sampling, such as MEMS accelerometers, with a sampling frequency set between 1kHz and 10kHz. This sampling frequency is sufficient to collect the harmonic components of mechanical vibration, providing real-time head vibration and substrate vibration signals to the logic processing module. After the gantry's first movement stop command is issued, the logic processing module immediately activates the first operating mode to determine the Z-axis triggering timing in the presence of vibration. In this mode, the logic processing module calculates the relative vibration signal by taking the real-time head vibration and substrate vibration signals as inputs, subtracting the two signals point by point using a digital differential converter, and obtaining... The system outputs a relative vibration signal, which cancels out the common-mode vibration noise acting on both components and characterizes the relative positional error between the mounting head nozzle and the substrate pads. Next, the logic processing module predicts one or more future zero-point windows based on this relative vibration signal. The algorithm steps are as follows: First, the relative vibration signal is input into a digital filter, such as a high-Q bandpass filter or a digital phase-locked loop, to lock the resonant frequency and phase information of the signal. This frequency is usually determined by the gantry structure and can be an 80Hz main mode. Second, based on the frequency and phase, and the attenuation trend of the signal amplitude, an attenuated sine wave model characterizing the relative vibration signal is established. Third, based on this mathematical model, the time intervals during which a series of future signal curves will cross a preset zero-error threshold are calculated. These intervals are defined as zero-point windows. Here, the zero-error threshold is a process parameter, and its value is determined according to the calibration procedure: During equipment debugging, the inherent mounting accuracy tolerance that the high-speed precision mounting system for electronic components can achieve under static, vibration-free conditions is measured, for example... The tolerance is set as the zero error threshold. Finally, the logic processing module selects the reachable zero window from the multiple predicted zero windows. This selection can be determined based on the response delay of the Z-axis actuator, such as selecting the first or second window, and issuing a synchronous trigger command to the Z-axis actuator only during the period of that window.

[0022] To monitor the physical health of the Z-axis actuator, which is a high-frequency moving component and whose operating state can affect placement quality, the logic processing module immediately switches to a second operating mode after the synchronous trigger command is issued. This mode utilizes the same set of sensing hardware for online diagnostics. In the second operating mode, during the physical process of Z-axis downward pressure, the logic processing module switches the application of the signals: the real-time head vibration signal is considered as a near-field excitation source signal generated by the internal mechanical excitation of the Z-axis actuator, which may include motor cogging effect or guide rail friction; the real-time substrate vibration signal is considered as a far-field response signal generated after the excitation is transmitted through the mechanical structure. The logic processing module inputs these two signals into an adaptive filtering algorithm, such as a low-order least mean square (LMS) algorithm. Specifically, it operates by using the near-field excitation source signal as input and the far-field response signal as the desired output. This adaptive filtering algorithm adjusts its internal set of filter coefficients in real time to minimize the mean square error between the input signal and the desired output signal after filtering. At the end of a preset calculation window, for example, a 20ms window covering the Z-axis from startup to contact with the substrate, the set of filter coefficients after convergence or stabilization is defined as the current characteristic coefficients representing the dynamic transfer characteristics between the two signals. Subsequently, the logic processing module will... With the benchmark characteristic coefficient template representing the healthy performance status Comparison; here The data template is obtained through a calibration procedure, which is as follows: After equipment debugging or maintenance, the calibration mode is activated, and the Z-axis actuator performs several placement operations in a confirmed healthy state. During this period, dual-channel signals are continuously acquired and the aforementioned adaptive filtering algorithm is run until its coefficients converge. The converged set of stable state coefficient values ​​(which can be...) are then used to... After averaging or statistically analyzing the dimensional vector, it is stored as the template for the baseline feature coefficients. The final comparison step is to... and All are considered Calculate the Euclidean distance between two feature vectors. ,in The system is based on this Euclidean distance. The size generates execution quality criteria regarding the current running state, for example, by using the distance... The adaptive filtering algorithm is characterized by its ability to suppress common-mode environmental noise unrelated to the Z-axis operation. When the criterion indicates an abnormal operating state, the system can output a predictive maintenance alarm through the equipment maintenance interface. The logic processing module can be implemented by a field-programmable gate array (FPGA), whose parallel processing architecture and high responsiveness make it suitable for completing a series of timing and calculation tasks within the manufacturing cycle, from high-frequency acquisition of vibration signals, relative signal differential analysis, zero-point window prediction, synchronization command triggering, instant mode switching, to real-time iteration of adaptive filtering coefficients.

[0023] Example 1: In a specific manufacturing scenario, a high-speed precision placement system for electronic components is deployed on a 24-hour continuous production line for high-density printed circuit boards. This production line environment experiences continuous common-mode vibration interference from nearby heavy stamping equipment, transmitted via the ground. Furthermore, the system must operate to maximize UPH (Uptime Per Hour), requiring its gantry to operate with extremely high acceleration and deceleration curves. During a placement cycle, the gantry moves at high speed above the target pad and receives a stop command. This drastic deceleration causes significant ringing in the placement head's mechanical structure, while common-mode vibration from the ground is also superimposed on the system. If conventional settling-time waiting methods are used, a 150ms waiting period must be set to avoid both gantry ringing and ground noise, constituting a major bottleneck to UPH. In contrast, the logic processing module in this system immediately activates the first operating mode. The first vibration sensing unit deployed on the placement head and the second vibration sensing unit deployed on the substrate stage simultaneously begin collecting high-frequency vibration data, acquiring the head vibration signal superimposed with gantry ringing and ground common-mode noise. And the substrate vibration signal, which mainly contains ground common-mode noise. The digital differential within the logic processing module calculates the relative vibration signals of the two components in real time. : In this calculation step, the common-mode vibration of the ground stamping equipment contained in both signals is subtracted and canceled out; the output of this calculation... It only reflects the gantry ringing signal of the placement head relative to the substrate stage; this exempts the system from environmental noise interference from the outset; the logic processing module then uses this clear... The attenuated sine wave model is used to lock its main frequency and phase, and a series of future zero-point windows are predicted. The system selects the first reachable zero-point window, such as time , and sends a synchronous trigger command to the Z-axis actuator. The placement action is completed with high precision far from the traditional 150ms waiting time.

[0024] At the instant the synchronization trigger command is issued, the logic processing module immediately switches to the second operating mode to perform in-situ diagnosis of the Z-axis actuator, which has been running continuously for 50 million cycles and whose physical health status is unknown. During the 20ms physical process of Z-axis compression, the same set of first and second vibration sensing units continuously collect signals. At this time, the real-time head vibration signal... Defined as a near-field excitation source signal, the real-time substrate vibration signal Defined as far-field response signals, both contain the nonlinear jitter characteristics generated during movement by the Z-axis lead screw due to minor wear; the adaptive filtering algorithm within the logic processing module uses... For input, with Real-time calculations are performed to obtain the desired output, and a set of current characteristic coefficients characterizing the current dynamic transfer properties of the Z-axis are calculated. Within the logic processing module, this current feature coefficient is... The reference characteristic coefficient template stored during system initial calibration Perform Euclidean distance The calculation; due to minor wear on the Z-axis, its physical transmission characteristics have shifted, leading to the calculated... deviation This makes the final Euclidean distance The system generates execution quality criteria based on the data exceeding the preset maintenance threshold and outputs a predictive maintenance alarm for the Z-axis through the equipment maintenance interface, thus achieving early fault diagnosis of the core moving parts of the equipment without interrupting current production.

[0025] Example 2: This example is used to verify the effectiveness of the aforementioned technical solution in improving placement efficiency and maintaining placement accuracy in a simulated manufacturing environment compared to the fixed settling time waiting method. The purpose of the experiment is to quantitatively compare the placement cycle time and final placement position accuracy of the system using the relative resonance zero-point synchronous triggering mechanism of this invention (experimental group) and the system using the fixed settling time waiting method (control group) at different operating speeds. The test platform is based on a standard high-speed placement equipment test bench, whose gantry has maximum acceleration. Maximum speed The motion capability; the Z-axis actuator response time is The experimental group added a first vibration sensing unit and a second vibration sensing unit to the platform, both of which are MEMS accelerometers with a sampling frequency set to 10kHz. The data was output to a logic processing module based on an FPGA. The control group used only the original equipment controller. To simulate structural vibrations transmitted by the operation of adjacent equipment in an industrial plant, the base of the entire test bench was mounted on a programmable six-degree-of-freedom vibration table, which applied a root mean square acceleration of [missing value]. Random vibration signals with a frequency range of 5Hz to 500Hz were used as common-mode noise to simultaneously affect the mounting head and substrate stage. Key parameters for the experiment were set as follows: The gantry motion was selected as a point-to-point movement task with a distance of 100mm from the starting point to the target position. The gantry accelerated to different target speeds at maximum acceleration (three sets were set). , , The system operates and stops with maximum deceleration when approaching the target position to induce varying degrees of mechanical ringing; the zero-error threshold is set based on the accuracy requirements for mounting 0201 size components, and is set to [value missing]. The settling time of the control group is to ensure that the maximum speed is maintained. The vibration energy decays to after stopping. Within the range, its fixed settling time was set to 150ms; Test procedure: For each set gantry target speed ( , , The control and experimental groups were run for 1000 consecutive placement cycles respectively; in each cycle, the time when the motion stop command was received from the gantry was recorded. The moment when the Z-axis actuator completes the mounting action ( The time difference is used as the single placement cycle time. Simultaneously, after each mounting is completed, an external high-precision coordinate measuring microscope (measurement accuracy) is used. ) Measure the deviation of the actual mounted component center from the target pad center We statistically analyzed the data from 1000 cycles, calculated the average cycle time and its standard deviation, as well as the average mounting error and standard deviation in the X and Y directions.

[0026] Table 1: Performance Comparison Data Between the Control Group and the Experimental Group at Different Operating Speeds

[0027] Average cycle time of the test group at all test speeds All were lower than the control group; for example, at maximum speed At that time, the control group required a fixed 150ms settling time plus Z-axis execution time, with an average cycle time of 156.5ms. The experimental group, by predicting the zero-point window and issuing a synchronous trigger command in the early stages of vibration (average 35.1ms), reduced its average cycle time by approximately 77.6%. As the target speed of the gantry increased, the amplitude and duration of the gantry ringing also increased, and the time required for the experimental group to predict the first usable zero-point window also increased accordingly (from...). The time increased from 23.5ms to (35.1ms); Regarding placement accuracy, the average error in the X and Y directions of the experimental group was comparable to that of the control group, both close to zero; the standard deviation of the error was... , The standard deviation of the experimental group at all speeds was slightly less than or equal to that of the control group, indicating that the experimental group did not sacrifice accuracy by triggering at the moment when the relative position error crossed the zero point, and may have achieved higher placement consistency. The test results confirmed the feasibility and effectiveness of the technical solution of the present invention. By calculating the relative vibration signal in real time to resist common mode noise interference, and selecting the zero point window to synchronously trigger Z-axis placement based on the prediction of the signal, the system can shorten the waiting time after high-speed movement and improve the overall placement efficiency of electronic components while maintaining or improving placement accuracy.

[0028] Example 3: To further illustrate the technical effect of using relative vibration signals for prediction and triggering in the technical solution of this invention, the following comparative experiment was conducted. This comparative example is used to verify the placement performance in an environment with common-mode vibration interference when a conventional technical path based solely on the absolute vibration signal of the placement head for stability judgment is used instead of a synchronous triggering method for execution instructions based on differential prediction of dual-path vibration signals. The test platform, external vibration conditions, and gantry motion parameters (including the target speed setting) of this comparative example are as follows: , , The three groups), mounting accuracy tolerance ( The threshold and accuracy measurement methods are consistent with those in Example 2; the only difference lies in the control logic: the system in this comparative example (hereinafter referred to as the comparative example group) only uses the first vibration sensing unit to collect head vibration signals. The logic processing module monitors this The amplitude, and when its amplitude first enters After reaching the threshold range, a Z-axis placement trigger command is issued; the test process is the same as in Example 2: for each set gantry target speed, the comparative group is run for 1000 consecutive placement cycles, and the time of each placement cycle is recorded. And measure the single placement accuracy error. ; Statistical analysis was performed on the data from 1000 cycles to calculate the average cycle time and its standard deviation, as well as the average mounting error and standard deviation in the X and Y directions.

[0029] Table 2: Performance Comparison Data of Comparative Example Group and Experimental Group of Example 2 at Different Operating Speeds

[0030] Referring to Table 2, although the comparative group attempted to shorten the waiting time by monitoring vibration signals, its average cycle time was... Compared to the fixed waiting time control group in Example 2, the time was shortened, but the placement accuracy deteriorated; the standard deviation of the error in the X and Y directions was [not specified]. , Much larger than the experimental group, for example in At speed, for , for Exceeded The accuracy requirements are not met; at the same time, the standard deviation of its cycle time increases, indicating that the triggering timing is unstable; the reason for this is that the comparative group only monitors... It is impossible to distinguish between the gantry ringing and the common-mode vibration transmitted from the ground. When the amplitude of the vibration signal temporarily enters the threshold range due to the ringing attenuation of the gantry, the placement head and the substrate stage may be in a state of large relative displacement due to common-mode vibration. This triggers placement, leading to a large random deviation in the actual placement position. Experimental results show that the conventional approach of relying solely on the absolute vibration signal of the placement head, without synchronous sensing of the substrate vibration signal and the step of calculating the relative vibration signal based on the difference between the two, cannot solve the accuracy control problem caused by high-speed movement under conditions of common-mode vibration interference. This invention introduces dual-channel sensing and differential processing to calculate the relative position error. Based on this signal, zero-point prediction and synchronous triggering are performed.

[0031] Example 4: This example combines Figures 1 to 3A description of a high-speed precision mounting system for electronic components, such as... Figure 1 As shown, it includes a substrate stage as a physical reference unit and a mounting head and Z-axis actuator as physical execution units. A first vibration sensing unit, such as a MEMS accelerometer, is used to sense the vibration of the physical execution unit to obtain the head vibration signal. A second vibration sensing unit, such as a MEMS accelerometer, is used to sense the vibration of the physical reference unit to obtain the substrate vibration signal. Both signals are input to a logic processing module. In mode 1, the logic processing module calculates the relative vibration and predicts the zero-point window to issue a synchronous trigger command to the mounting head and Z-axis actuator. In mode 2, it runs adaptive filtering to generate a criterion. The generation of this criterion also depends on a preset health state reference characteristic coefficient template for comparison. Finally, the logic processing module can output a predictive maintenance alarm to the equipment maintenance interface.

[0032] like Figure 2 As shown in the figure, this graph represents the curves of filter coefficient values ​​changing with the number of iterations. The graph uses three different curves to illustrate the differences in coefficient values ​​calculated by the adaptive filtering algorithm under three conditions: healthy state, slight wear, and abnormal state. These differences can be used to generate execution quality criteria. Figure 3 As shown, the process begins with the gantry moving at high speed to the target position and issuing a motion stop command. This action causes the gantry to generate mechanical ringing. At this time, the logic processing module starts the first operating mode and receives the real-time head vibration signal and the real-time substrate vibration signal provided by the first vibration sensing unit and the second vibration sensing unit, respectively. The logic processing module sequentially executes the steps of calculating the relative vibration signal with a digital differential, locking the resonant main frequency and phase, establishing an attenuated sine wave model, and predicting the zero-point window. Finally, during the selected reachable zero-point window, a synchronous trigger command is issued to the Z-axis actuator. After the command is issued, the logic processing module immediately switches to the second operating mode.

[0033] Example 5: This example describes the calibration steps for determining the key operating parameters and establishing a reference template for the aforementioned high-speed precision placement system for electronic components. First, the zero-error threshold is calibrated to define the acceptable range of relative positional error when predicting the zero-point window in the first operating mode. The initial state during calibration is that the system's mechanical structure is assembled, the first and second vibration sensing units are installed and connected to the logic processing module, the Z-axis actuator is functioning normally, and an external coordinate measuring microscope with sub-micron level measurement accuracy is provided on-site. The calibration steps are as follows: First, under static conditions without vibration introduced by gantry movement, the placement head is controlled to execute... The process involves repeating the pick-and-place or simulated mounting actions to the same target position on the substrate stage; secondly, after each action, a coordinate measuring microscope is used to measure the deviation of the actual landing point relative to the target position. ,in The third step is to calculate the radial error at all measurement points. The fourth step is to... Statistical analysis was performed on the radial error values, and their mean was calculated. with standard deviation Fifth, based on the statistical results, set a zero-error threshold. The threshold can be set as follows: If the calculation yields Value Considering measurement noise and minor environmental disturbances, the zero error threshold can be set as follows: This value is then stored in the logic processing module as the basis for determining whether the relative vibration signal has entered the zero error range in the first operating mode; secondly, the key signal processing steps involved in the zero-point window prediction algorithm in the first operating mode are determined; after the gantry movement stop command is issued, the logic processing module receives the continuously acquired relative vibration signal. The data collection time is set to 50ms; the processing flow is as follows: First, the collected data... The input is a digital bandpass filter. The center frequency and bandwidth of this filter are set based on the main resonant frequency of the equipment gantry structure. This resonant frequency can be determined during the design phase through modal analysis or during the commissioning phase through spectral analysis. If the main resonant frequency is 80Hz, the filter passband can be set to 75Hz to 85Hz. The second step involves using a parameter estimation algorithm on the filtered signal. This can be achieved by fitting a decaying sine wave model using the least squares method. The initial amplitude of the current ringing is obtained. attenuation coefficient angular frequency and initial phase The third step is to numerically solve the equations based on the determined model parameters. ZeroErrorThreshold was already specified in the previous step. ,exist All solutions within the interval The fourth step is to convert these solutions... Paired in chronological order, forming a series of time windows. Each window satisfies the condition within that time period. These windows are the predicted zero-point windows. The logic processing module will select one of the windows to issue a synchronous trigger command based on the Z-axis response time and the system control strategy.

[0034] Finally, the benchmark characteristic coefficient template for executing the quality criterion is generated in the second operating mode. Calibration and maintenance thresholds The initial state during calibration is that the system has completed the above calibration and the Z-axis actuator has been checked and confirmed to be in good condition. The calibration steps, i.e., entering the calibration mode in the specific implementation, include: First, controlling the system to execute... The second step is a substandard Z-axis mounting cycle; in each cycle... ( After the Z-axis trigger command is issued, head vibration signals are synchronously acquired within a preset 20ms time window. and substrate vibration signal The third step is to input these two signals into the LMS adaptive filtering algorithm implemented in the logic processing module, setting the algorithm parameters to the filter order. Step size factor Filter coefficient vector Initialize to a zero vector; the algorithm follows... Perform iterations, where This serves as the error signal, and the coefficient vector for that iteration is obtained by continuing until the end of the time window. The fourth step is to calculate the average value of the coefficient vector obtained in this iteration, which will serve as the baseline feature coefficient template. and will Store in non-volatile memory; fifth step, calculate the value obtained in each calibration cycle. Compared to the final square Euclidean distance Step 6, regarding this Distance values Perform statistical analysis and calculate its mean. and standard deviation Step 7: Set maintenance thresholds based on statistical results. Its settings need to balance the risks of false negatives and false negatives, and can be set as follows: The range of values ​​for the coefficients is determined based on the equipment maintenance sensitivity requirements. If set... Calculated , ,but identified as This threshold is also stored in the system configuration.

[0035] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0036] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A high-speed precision mounting system for electronic components, characterized in that the system... include: Placement head; A substrate stage for supporting a substrate; Z-axis actuator is used to drive the placement head to perform placement actions on the substrate stage; The first vibration sensing unit is used to acquire the real-time head vibration signal of the mounting head; The second vibration sensing unit is used to acquire the real-time vibration signal of the substrate stage. The logic processing module is used to: in the first operating mode, calculate the relative vibration signal between the head vibration signal and the substrate vibration signal in real time; based on the relative vibration signal, predict one or more zero-point windows in the future where the relative vibration signal crosses the zero error threshold; and issue a synchronization trigger command to the Z-axis actuator only during the zero-point window; immediately switch to the second operating mode after the synchronization trigger command is issued; in the second operating mode: use the real-time head vibration signal as the near-field excitation source signal and the real-time substrate vibration signal as the far-field response signal; input the near-field excitation source signal and the far-field response signal into an adaptive filtering algorithm to calculate the current characteristic coefficients that characterize the dynamic transmission characteristics between the two in real time; compare the current characteristic coefficients with a preset benchmark characteristic coefficient template representing a healthy execution state, and generate an execution quality criterion for the current operating state of the Z-axis actuator based on the comparison result.

2. The high-speed precision mounting system for electronic components according to claim 1, characterized in that, The system also includes a gantry for driving the placement head to move horizontally above the substrate stage; the logic processing module is used to immediately start the first operating mode after the gantry stops moving once. The first operating mode includes: subtracting the head vibration signal from the substrate vibration signal through a digital differential to generate a relative vibration signal; and predicting the zero-point window and issuing a synchronous trigger command based on the relative vibration signal.

3. The high-speed precision mounting system for electronic components according to claim 2, characterized in that, In the first operating mode, the logic processing module is also used to: lock the resonant frequency and phase of the relative vibration signal through a digital filter; establish an attenuated sine wave model representing the relative vibration signal based on the resonant frequency and phase; and calculate the zero-point window based on the attenuated sine wave model.

4. The high-speed precision mounting system for electronic components according to claim 1, characterized in that, The reference characteristic coefficient template is a set of reference coefficient values; the logic processing module is also used to enable the Z-axis actuator to run in a known healthy state in calibration mode, and during this period continuously acquire near-field excitation source signals and far-field response signals, run an adaptive filtering algorithm until convergence, and store a set of stable state coefficient values ​​after convergence as reference coefficient values.

5. The high-speed precision mounting system for electronic components according to claim 4, characterized in that, The logic processing module is used to process the current feature coefficients. With reference characteristic coefficient template All are considered dimensional feature vectors, where The dimension of the feature vectors; the logic processing module calculates the Euclidean distance between two feature vectors. To generate execution quality criteria, where Euclidean distance Determined in the following ways: ,in, and These are the current feature coefficient vector and the reference feature coefficient template vector, respectively. Each component; the logic processing module is also used to process the Euclidean distance. Compare with the preset maintenance threshold.

6. The high-speed precision mounting system for electronic components according to claim 1, characterized in that, The system also includes an equipment maintenance interface, and the logic processing module is also used to output a predictive maintenance alarm through the equipment maintenance interface when the execution quality criterion indicates that the Z-axis actuator is in an abnormal operating state.

7. The high-speed precision mounting system for electronic components according to claim 1, characterized in that, Both the first vibration sensing unit and the second vibration sensing unit are MEMS accelerometers, and are used to acquire real-time head vibration signals and real-time substrate vibration signals at a sampling frequency of 1kHz to 10kHz.

8. The high-speed precision mounting system for electronic components according to claim 1, characterized in that, In the second operating mode, the logic processing module is used as the input of the near-field excitation source signal to the adaptive filtering algorithm and the far-field response signal as the expected output of the adaptive filtering algorithm. Adaptive filtering algorithms are used to adjust a set of internal filter coefficients in real time to minimize the mean square error between the input signal after filtering and the desired output signal. The current feature coefficients are the filter coefficients of the adaptive filtering algorithm at the end of the preset calculation window.

9. The high-speed precision mounting system for electronic components according to claim 1, characterized in that, The zero error threshold is preset based on the inherent placement accuracy tolerance of the high-speed precision placement system for electronic components.

Images