Numerical control relieving machine for relieving tooth type cooling fins

Abstract

The invention discloses a numerical control relieving machine for relieving tooth type cooling fins, and belongs to the technical field of numerical control machining. The equipment comprises an encoder acquisition module, an edge intelligent preprocessing unit, a dual-channel signal processing module, a self-adaptive predictive compensation system, a torque double-closed-loop control system and a dual-stage rigidity switching coupling. FPGA intelligent filtering is carried out at a signal source end, the response speed and precision are balanced by using a two-channel architecture, and lag is eliminated in combination with prediction compensation; at a mechanical transmission end, cutting vibration is restrained through a variable-rigidity coupler and torque feedback. According to the invention, the problems of tooth pitch drifting, processing line trembling and non-uniform feeding of an existing relieving machine are solved, high-precision and high-consistency batch production of relieving radiating fins is realized, and the tooth pitch error is controlled within + / -0.01 mm.

Description

Technical Field

[0001] This invention belongs to the technical field of precision CNC machining equipment, and relates to a CNC machine for high-precision scraping of metal (such as copper and aluminum) heat sinks, specifically a CNC tooth scraping machine for toothed heat sinks. Background Technology

[0002] The shaving process is a manufacturing technique that uses a specialized planer to continuously cut the surface of a metal material (aluminum or copper) to create high-density, high-efficiency heat dissipation fins in one piece. Compared to traditional shaving or welding processes, shaving heat sinks have extremely high heat dissipation efficiency due to the absence of thermal resistance between the substrate and the fins, and are widely used in high-power LEDs, server CPU heat sinks, and heat dissipation modules for electronic control systems in new energy vehicles.

[0003] However, with the increasing power density of electronic devices, extremely high requirements are placed on the pitch accuracy, thickness uniformity, and surface finish of heat sinks. Existing CNC gear-shaving machines face the following main technical bottlenecks when dealing with high-precision, high-volume production: Distortion and delay of position feedback signal: Traditional tooth-shaving machines mostly use linear encoders or rotary encoders to directly provide position feedback. During high-frequency reciprocating motion (shaving action), the encoder signal is highly susceptible to electromagnetic noise from the servo motor, mechanical vibration, and the impedance of the transmission cable, leading to a decrease in the signal-to-noise ratio. Existing solutions involve increasing the filtering depth of the back-end controller, but this inevitably introduces phase lag, preventing the servo system from responding to rapidly changing commands in real time, resulting in periodic drift in the tooth pitch.

[0004] The contradiction between rigidity and flexibility: Gear cutting involves two processes: "rapid retraction during idle stroke" and "loaded grinding." Rapid retraction requires extremely high dynamic response from the system, typically necessitating light load inertia and flexible buffering to reduce impact. The grinding process (especially with large depths of cut) requires extremely high transmission stiffness to overcome significant cutting resistance and prevent tool chatter. Existing equipment typically uses couplings with single stiffness, which cannot accommodate both conditions, resulting in either large start-stop impacts or noticeable chatter in the machined texture.

[0005] Unpredictability of load fluctuations: During the tooth-shaving process, the cutting resistance fluctuates dynamically due to the inhomogeneity of the internal crystal structure or changes in microhardness of the material. Existing single closed-loop (position loop only) control systems can only correct position errors after they occur, which is "post-compensation." They cannot adjust the current instantaneously during sudden torque changes, resulting in uneven thickness at the root of the shavingd heat sink, or even broken teeth or scrap.

[0006] The lag in processing data: Existing CNC controllers mostly adopt a centralized processing architecture, where all sensor signals are fed into the main CPU for processing. In multi-axis high-speed linkage, the data throughput is huge, resulting in longer interpolation calculation cycles, making it difficult to achieve microsecond-level real-time vibration suppression and micrometer-level error compensation. Summary of the Invention

[0007] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides a CNC tooth-shaving machine for tooth-shaped heat sinks. It has the advantages of eliminating noise from the signal source, having feedforward prediction capability, and being able to adaptively adjust mechanical stiffness and control strategy according to the processing state, thereby significantly improving the tooth pitch accuracy, surface consistency and production efficiency of tooth-shaving processing.

[0008] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: a CNC tooth-shaving machine for toothed heat sinks, comprising: The encoder signal acquisition module is used to detect the real-time position of the linkage spindle; The edge intelligent signal preprocessing unit, located at the encoder end, is used to perform high-speed analog-to-digital conversion, impedance buffering, and anti-aliasing filtering on the sampled signal, and to achieve adaptive noise suppression through the linear and nonlinear composite filter circuit in the FPGA. A dual-channel time-separation signal processing module is connected to the edge intelligent signal preprocessing unit, which is used to independently filter the fast dynamic response signal and the steady-state high-precision positioning signal respectively. The adaptive prediction and compensation system is electrically connected to the dual-channel module. It predicts position changes in advance and performs filtering and delay compensation based on the motion model of the servo motor. The torque sensing dual closed-loop control system is used to simultaneously detect actual torque and position feedback, forming a composite control of position loop and torque loop - a dual-stage stiffness switching coupling, installed at the output end, is used to automatically switch between flexible buffer mode and rigid precision mode according to the machining stage. The servo drive execution unit is connected to the torque sensing dual closed-loop control system and is used to control the feed motion based on the fused compensation signal.

[0009] Preferably, the edge intelligent signal preprocessing unit includes a high-speed ADC, an active impedance buffer, a fourth-order Bessel filter, and an FPGA composite filter array. The FPGA has a built-in first-order FIR filter and a second-order Volterra nonlinear filter, and the FPGA has a built-in adaptive LMS algorithm to automatically correct the filter parameters according to temperature drift changes.

[0010] Preferably, the dual-channel time-separated signal processing module includes: Fast channel with a sampling rate of 10~20MHz, using an adaptive notch filter to suppress resonance interference generated by the main shaft; A precision channel with a sampling rate of 5~10MHz, using a multi-phase pipelined FPGA filter unit for high-precision steady-state data processing; It also uses a fuzzy logic controller to smoothly switch between speed, acceleration, and load current channels.

[0011] Preferably, the adaptive predictive compensation system establishes a discretized predictive model including motor inertia, friction coefficient, and PID parameters. It performs multi-step prediction and feedforward compensation within each sampling period, predicting the position value for the next N sampling periods based on the current state variables, where N is determined by the group delay of the current filter. This reduces the response delay caused by filtering.

[0012] Preferably, the torque sensing dual closed-loop control system includes: The coupling integrates a strain gauge torque sensor with a measurement accuracy better than ±0.5%FS and uses non-contact wireless signal transmission. A dual-channel data acquisition card is used to acquire the position signal and torque signal of the grating ruler, respectively. The control logic unit is used to dynamically adjust the current feedforward based on the deviation between the position loop and the torque loop to achieve stable loading and feeding.

[0013] Preferably, the dual-stage stiffness switching coupling is composed of an elastic rubber transmission component and an electromagnetic locking mechanism connected in parallel; Low-rigidity transmission is achieved through rubber transmission components when the power is off, and high-rigidity transmission is achieved through an electromagnetic locking mechanism when the power is on. The coupling is in a flexible mode during the start-up phase, with a stiffness range of 3000~6000 Nm / rad; During the steady phase, it is in rigid mode with a stiffness range of 40,000 to 60,000 Nm / rad. Furthermore, the switching signal is determined by comprehensively considering three conditions: stable speed, low torque fluctuation, and small position error.

[0014] Preferably, the servo drive execution unit, the predictive compensation system, and the torque control system adopt a fusion parameter control algorithm, which includes three weighted components: position command, speed feedforward, and torque compensation. The weight ratio can be dynamically adjusted according to the processing conditions to simultaneously optimize response speed and positioning accuracy.

[0015] Preferably, the system adopts an LVDS differential signal transmission structure and is configured with a continuous-time linear equalizer and a pre-emphasis circuit to improve the integrity of the encoder signal during long-distance transmission.

[0016] Preferably, the edge intelligent signal preprocessing unit and the dual-channel signal processing module are connected via star grounding and independent voltage regulation to avoid multi-axis interference and improve stability during synchronous interpolation processing.

[0017] Preferably, the entire machine structure is used in the processing of shovel-shaped heat sinks, with the tooth pitch consistency error controlled within ±0.01mm, the difference in tool mark depth controlled within ±3%, and capable of continuous eight-axis synchronous interpolation precision shovel processing.

[0018] (III) Beneficial Effects Compared with the prior art, the present invention provides a CNC tooth-shaving machine for toothed heat sinks, which has the following advantages: Extremely high signal fidelity: Through edge intelligent preprocessing, the problem of long-distance transmission interference is solved at the signal source, and the signal-to-noise ratio is improved by more than 20dB, laying the foundation for high precision.

[0019] Balancing dynamic and static performance: The dual-channel architecture perfectly solves the contradiction in traditional filtering where "clean filtering results in slow response, while fast response results in high noise," enabling the gear hoist to both retract quickly and perform precise cutting.

[0020] Zero-delay control experience: The adaptive predictive compensation system effectively offsets the computational delay of the digital system, achieving near-analog system instantaneous response characteristics and significantly reducing the following error during high-speed machining.

[0021] Intelligent vibration suppression and shearing stability: The dual-stage stiffness coupling, combined with the torque closed loop, suppresses vibration through both mechanical and electronic control methods, resulting in mirror-like surfaces on the heat sink teeth and a stable tooth pitch error within ±0.01mm.

[0022] Highly adaptable: It can adapt to the processing requirements of different materials (copper, aluminum, alloy) and different tooth heights and tooth densities without the need for cumbersome manual mechanical adjustments. Attached Figure Description

[0023] Figure 1 This is a block diagram of the control system of the present invention. Detailed Implementation

[0024] To better understand the purpose, structure, and function of this invention, the following detailed description of a CNC tooth-shaving machine for tooth-type heat sinks is provided in conjunction with the accompanying drawings.

[0025] Please see Figure 1 The present invention: The overall concept of the present invention To address the common machining quality issues encountered by CNC gear hobbing machines during heat sink processing, such as tooth pitch drift, uneven feed, and inconsistent tool mark depth, the solution focuses on improvements in five key areas: 1. Enhanced preprocessing at the encoder signal source end A high-speed ADC+FPGA edge intelligence module is added to the position encoder to achieve source noise suppression, nonlinear correction, and transmission optimization.

[0026] 2. Dual-channel time-separated signal processing architecture Rapid dynamic response and high-precision steady-state positioning are handled separately to meet the needs of different processing stages.

[0027] 3. Adaptive Predictive Compensation Filtering System By using a servo motor motion model to predict position and compensate for filtering delay in advance, a balance between high precision and fast response is achieved.

[0028] 4. Torque Sensing Dual Closed-Loop Control System The position-torque dual closed loop compensates for coupling transmission errors in real time, improving feed stability.

[0029] 5. Dual-stage stiffness switching coupling The system features flexible buffering during startup and rigid transmission during stabilization, enabling adaptive switching between rigidity and flexibility while balancing vibration reduction and improved accuracy.

[0030] II. Core technology modules include 1. Edge intelligent signal preprocessing unit Objective: To improve the quality of encoder feedback signals and reduce the negative impact of sampling delay and transmission distortion on CNC systems.

[0031] Hardware design: High-speed ADC: resolution ≥ 14 bits, sampling rate 10~12MHz.

[0032] Front-end conditioning: Active impedance buffer (input ≥10MΩ, output ≤50Ω) + fourth-order Bessel anti-aliasing filter (cutoff frequency ≈ sampling rate 0.4).

[0033] FPGA: ≥50K logic cells, ≥200MHz clock frequency, integrated DSP module.

[0034] Parallel processing pipeline: a combination of first-order linear FIR and second / third-order Volterra nonlinear filtering, with a feedforward compensation path to avoid group delay accumulation.

[0035] LMS adaptive algorithm module: Updates filter coefficients using temperature drift and zero-position reference signals.

[0036] Transmission and power supply: LVDS differential interface, pre-emphasis and continuous-time linear equalizer optimize long-distance transmission.

[0037] Independent linear voltage regulation + star grounding suppresses multi-axis crosstalk.

[0038] 2. Dual-channel time-separated signal processing architecture Objective: To resolve the conflict between fast response and deep filtering in a single-channel architecture.

[0039] Architecture Design: Fast channel: High sampling rate ADC (10~20MHz), three-stage adaptive notch filter to suppress spindle-related vibration interference + small-order FIR filter to reduce noise, processing delay ≤30μs.

[0040] Precision channel: High-precision ADC (16-bit, 5~10MHz), FPGA multi-phase filter pipeline (8~12 stages), stopband attenuation ≥80dB, delay ≤80μs.

[0041] Channel switching: Fuzzy logic control combines speed, acceleration, and load current for smooth transition (spline interpolation fusion), and the switching transient error is controlled within ±3μm.

[0042] 3. Adaptive Predictive Compensation Filtering System Objective: To reduce the response delay introduced by filtering while maintaining the accuracy of deep filtering.

[0043] Design Methodology: Dual-channel fusion: Dynamic weighting of the measurement channel (actual encoder values) and the prediction channel (motion model generated values).

[0044] Motion model: A discretized prediction model is established based on the motor inertia J, friction coefficient B, and position loop PID parameters.

[0045] Multi-step prediction (Np = 2~4 sampling periods) calculates the position in advance and provides feedforward compensation instructions.

[0046] Weight adaptation: Increase the weight of predicted values ​​during the high-speed / acceleration phase, and increase the weight of actual values ​​during the steady-state phase.

[0047] The filter cutoff frequency is dynamically adjusted and optimized based on the high-frequency error energy ratio and phase lag.

[0048] 4. Torque Sensing Dual Closed-Loop Control System Objective: To eliminate fluctuation errors in the transmission process of the coupling and improve feed accuracy.

[0049] Hardware structure: The coupling integrates a high-precision strain gauge torque sensor (±0.5%FS accuracy, range of 1.5 times the rated torque) and wireless transmission (2.4GHz, latency <2ms).

[0050] Dual-channel acquisition card: the first channel provides position feedback from the grating ruler, and the second channel provides torque signal.

[0051] Non-contact power supply for couplings (moving core transformer, 3~5mm air gap).

[0052] Control logic: Position loop: Traditional PID, mainly based on grating ruler values.

[0053] Torque loop: Monitors the deviation between actual torque and expected torque, and writes it into the current feedforward through a proportional-integral compensation strategy.

[0054] Threshold adaptive: loosen during roughing (8~12% of rated torque), tighten during finishing (3~5%).

[0055] Calibration process: Static multi-point calibration + actual cutting verification to ensure that the error is controllable.

[0056] 5. Dual-stage stiffness switching coupling structure Objective: To provide damping and vibration absorption during the startup acceleration phase and high-rigidity precision transmission during the stabilization processing phase.

[0057] Structural design: Main body: Integrated metal-rubber vulcanization structure, with a stainless steel wire mesh skeleton embedded in the corrugated rubber ring.

[0058] Flexible mode: stiffness ≈ 5000 Nm / rad, absorbs torque impact and reduces vibration.

[0059] Rigid mode: The locking pawl engages with the toothed ring, with a stiffness of approximately 50,000 Nm / rad.

[0060] Actuator: Fast-response electromagnetic drive (≤20ms), guide chamfer design reduces switching shock.

[0061] Switching criteria: Rigid mode is triggered only when all three conditions are met simultaneously: stable speed, low torque fluctuation, and small position error. Detailed Implementation

[0062] A CNC tooth-shaving machine for toothed heat sinks, comprising: The encoder signal acquisition module is used to detect the real-time position of the linkage spindle; The edge intelligent signal preprocessing unit, located at the encoder end, is used to perform high-speed analog-to-digital conversion, impedance buffering, and anti-aliasing filtering on the sampled signal, and to achieve adaptive noise suppression through the linear and nonlinear composite filter circuit in the FPGA. A dual-channel time-separation signal processing module is connected to the edge intelligent signal preprocessing unit, which is used to independently filter the fast dynamic response signal and the steady-state high-precision positioning signal respectively. The adaptive prediction and compensation system is electrically connected to the dual-channel module. It predicts position changes in advance and performs filtering and delay compensation based on the motion model of the servo motor. The torque sensing dual closed-loop control system is used to simultaneously detect actual torque and position feedback, forming a composite control of position loop and torque loop - a dual-stage stiffness switching coupling, installed at the output end, is used to automatically switch between flexible buffer mode and rigid precision mode according to the machining stage. The servo drive execution unit is connected to the torque sensing dual closed-loop control system and is used to control the feed motion based on the fused compensation signal.

[0063] Preferably, the edge intelligent signal preprocessing unit includes a high-speed ADC, an active impedance buffer, a fourth-order Bessel filter, and an FPGA composite filter array. The FPGA has a built-in first-order FIR filter and a second-order Volterra nonlinear filter, and the FPGA has a built-in adaptive LMS algorithm to automatically correct the filter parameters according to temperature drift changes.

[0064] Preferably, the dual-channel time-separated signal processing module includes: Fast channel with a sampling rate of 10~20MHz, using an adaptive notch filter to suppress resonance interference generated by the main shaft; A precision channel with a sampling rate of 5~10MHz, using a multi-phase pipelined FPGA filter unit for high-precision steady-state data processing; It also uses a fuzzy logic controller to smoothly switch between speed, acceleration, and load current channels.

[0065] Preferably, the adaptive predictive compensation system establishes a discretized predictive model including motor inertia, friction coefficient, and PID parameters. It performs multi-step prediction and feedforward compensation within each sampling period, predicting the position value for the next N sampling periods based on the current state variables, where N is determined by the group delay of the current filter. This reduces the response delay caused by filtering.

[0066] Preferably, the torque sensing dual closed-loop control system includes: The coupling integrates a strain gauge torque sensor with a measurement accuracy better than ±0.5%FS and uses non-contact wireless signal transmission. A dual-channel data acquisition card is used to acquire the position signal and torque signal of the grating ruler, respectively. The control logic unit is used to dynamically adjust the current feedforward based on the deviation between the position loop and the torque loop to achieve stable loading and feeding.

[0067] Preferably, the dual-stage stiffness switching coupling is composed of an elastic rubber transmission component and an electromagnetic locking mechanism connected in parallel; Low-rigidity transmission is achieved through rubber transmission components when the power is off, and high-rigidity transmission is achieved through an electromagnetic locking mechanism when the power is on. The coupling is in a flexible mode during the start-up phase, with a stiffness range of 3000~6000 Nm / rad; During the steady phase, it is in rigid mode with a stiffness range of 40,000 to 60,000 Nm / rad. Furthermore, the switching signal is determined by comprehensively considering three conditions: stable speed, low torque fluctuation, and small position error.

[0068] Preferably, the servo drive execution unit, the predictive compensation system, and the torque control system adopt a fusion parameter control algorithm, which includes three weighted components: position command, speed feedforward, and torque compensation. The weight ratio can be dynamically adjusted according to the processing conditions to simultaneously optimize response speed and positioning accuracy.

[0069] Preferably, the system adopts an LVDS differential signal transmission structure and is configured with a continuous-time linear equalizer and a pre-emphasis circuit to improve the integrity of the encoder signal during long-distance transmission.

[0070] Preferably, the edge intelligent signal preprocessing unit and the dual-channel signal processing module are connected via star grounding and independent voltage regulation to avoid multi-axis interference and improve stability during synchronous interpolation processing.

[0071] Preferably, the entire machine structure is used in the processing of shovel-shaped heat sinks, with the tooth pitch consistency error controlled within ±0.01mm, the difference in tool mark depth controlled within ±3%, and capable of continuous eight-axis synchronous interpolation precision shovel processing.

[0072] Example 1: System Hardware Architecture Setup The CNC tooth-shaving machine of the present invention has a core mechanical structure including a marble base, a high-rigidity gantry frame, X / Y / Z axes driven by linear motors, and a special tooth-shaving tool holder.

[0073] For the encoder signal acquisition module, a Heidenhain or equivalent precision sine / cosine grating ruler is selected as the main position feedback. Unlike traditional designs, this embodiment directly installs an edge intelligent signal preprocessing unit 3cm behind the grating ruler reading head. This unit uses a multi-layer PCB design and integrates a 16-bit ADC chip with a sampling rate of up to 12MHz (such as Analog Devices' AD76xx series) and a low-power FPGA (such as Xilinx Spartan-7).

[0074] Signal transmission adopts the LVDS (Low Voltage Differential Signaling) standard, transmitting the processed digital signal to the main controller via shielded twisted-pair cable. This architecture completely eliminates the electromagnetic interference experienced by traditional analog signals when transmitted through drag chain cables.

[0075] Example 2: Edge Intelligent Signal Processing Flow The workflow of the edge intelligent signal preprocessing unit is as follows: Impedance matching: The original analog signal is first buffered by an operational amplifier with high input impedance (>10MΩ) to prevent the load effect from lowering the signal amplitude.

[0076] Anti-aliasing filtering: High-frequency noise above 6MHz is filtered out by a fourth-order Bessel analog filter to protect the ADC from aliasing.

[0077] FPGA digital cleaning: The quantized data from the ADC is fed into the FPGA.

[0078] Linear filtering: Use a first-order FIR filter to remove white noise.

[0079] Nonlinear correction: Utilizing the Volterra series model, subdivision errors caused by misalignment of the grating ruler or engraving errors are compensated for in real time. The FPGA internally stores the factory calibration table and, based on data from the onboard temperature sensor, uses the LMS adaptive algorithm to fine-tune the calibration coefficients to compensate for thermal drift.

[0080] Example 3: Dual-channel and Predictive Compensation Algorithm After receiving the preprocessed signal, the main controller enters the dual-channel time separation processing module: Channel A (Fast Channel): The design focus is on "speed". It employs a CIC (Integral Comb) filter, retaining only basic noise reduction functions, with the total delay time locked at 25μs. Data from this channel is primarily used to monitor sudden speed changes and emergency stop protection.

[0081] Channel B (Precision Channel): The design focus is on "accuracy". It adopts an 8-stage pipelined FIR filter with a low cutoff frequency setting and a stopband attenuation of 85dB, which can filter out micron-level jitter, but the delay is about 80μs.

[0082] Fusion mechanism: The fuzzy controller reads the motor's command acceleration in real time. When the acceleration is greater than 1G (such as during rapid tool return), the weights tend to favor channel A (e.g., 0.8A + 0.2B); when the acceleration is less than 0.1G (such as during stable cutting), the weights tend to favor channel B (0.1A + 0.9B).

[0083] The fused signal enters the adaptive predictive compensation system. This system, based on the Kalman filter concept, establishes a state-space model of the motor: X(k+1)=A·X(k)+BU(k) Where X represents the position state and U represents the control input. Based on the current state at time k, the system predicts the position at time k+N (where N is the number of cycles caused by the filtering delay), and sends the predicted value as a feedback signal into the position loop. This is equivalent to "catching up" on the time axis by the time consumed by the filter, achieving zero-phase lag control.

[0084] Example 4: Electromechanical Cooperative Control and Execution The actuator of this invention employs an innovative dual-stage stiffness switching coupling and torque sensing dual closed-loop control.

[0085] Two-stage coupling construction: The coupling body consists of an inner vulcanized rubber ring (flexible body) and an outer electromagnetic jaw clutch (rigid body).

[0086] Phase 1 (Reverse / Idle Run): The electromagnetic coil is de-energized, and the jaws disengage. Motor power is transmitted through a rubber ring with a stiffness of approximately 5000 Nm / rad. The damping properties of the rubber effectively absorb the high-frequency impacts generated by frequent motor starts and stops, protecting the lead screw and guide rails.

[0087] Phase Two (Shoveling): When the controller issues the "feed" command and the Z-axis position reaches the cutting point 10ms beforehand, the electromagnetic coil is energized, and the jaw clutch engages and locks. The power bypass rubber ring transmits power directly through the metal jaw clutch, instantly increasing the stiffness to 55000 Nm / rad. At this point, the system becomes a rigid connection, ensuring that the tool does not elastically retract when cutting into the metal.

[0088] Dual closed-loop torque control: The strain gauges on the coupling monitor the transmitted torque in real time, and the data is transmitted to the controller via a 2.4GHz wireless module at a frequency of 1kHz.

[0089] In addition to the standard position PID loop, a torque inner loop has been added to the controller.

[0090] When the blade encounters a hard spot in the material, the resistance increases instantly, and the torque sensor detects a torque peak. At this point, the position loop has not yet generated sufficient error, but the torque loop has already taken the initiative, directly increasing the current feedforward and boosting the motor's output torque.

[0091] Experiments show that after introducing the torque ring, the fluctuation of the shovel tooth depth is reduced from ±0.05mm to ±0.008mm.

[0092] Example 5: Overall Processing Effect A batch of 6063 aluminum alloy heat sinks (size 100x100mm, tooth height 30mm, tooth pitch 1.5mm) were processed using the tooth-shaving machine described in this invention.

[0093] Setting parameters: Spindle cutting speed 120 times / minute.

[0094] Processing results: Tooth pitch consistency: As measured by a coordinate measuring machine, the error between any two teeth is within ±0.006mm (current equipment is usually ±0.03mm).

[0095] Surface quality: The shoveled surface is as bright as a mirror, with no visible chatter marks, and the roughness Ra reaches 0.8μm.

[0096] Production efficiency: By switching between a flexible mode for rapid retraction and a rigid mode for precise cutting, the cycle time has been reduced by 15%, resulting in a significant increase in overall production capacity.

[0097] It is understood that the present invention has been described through some embodiments, and those skilled in the art will recognize that various changes or equivalent substitutions can be made to these features and embodiments without departing from the spirit and scope of the invention. Furthermore, under the teachings of the present invention, these features and embodiments can be modified to adapt to specific situations and materials without departing from the spirit and scope of the invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the protection scope of the present invention.

Claims

1. A CNC tooth-shaving machine for toothed heat sinks, characterized in that, include: The encoder signal acquisition module is used to detect the real-time position of the linkage spindle; The edge intelligent signal preprocessing unit, located at the encoder end, is used to perform high-speed analog-to-digital conversion, impedance buffering, and anti-aliasing filtering on the sampled signal, and to achieve adaptive noise suppression through the linear and nonlinear composite filter circuit in the FPGA. A dual-channel time-separation signal processing module is connected to the edge intelligent signal preprocessing unit, which is used to independently filter the fast dynamic response signal and the steady-state high-precision positioning signal respectively. The adaptive prediction and compensation system is electrically connected to the dual-channel module. It predicts position changes in advance and performs filtering and delay compensation based on the motion model of the servo motor. The torque sensing dual closed-loop control system is used to simultaneously detect actual torque and position feedback, forming a composite control of position loop and torque loop - a dual-stage stiffness switching coupling, installed at the output end, is used to automatically switch between flexible buffer mode and rigid precision mode according to the machining stage. The servo drive execution unit is connected to the torque sensing dual closed-loop control system and is used to control the feed motion based on the fused compensation signal.

2. A CNC tooth-shaving machine for toothed heat sinks according to claim 1, characterized in that: The edge intelligent signal preprocessing unit includes a high-speed ADC, an active impedance buffer, a fourth-order Bessel filter, and an FPGA composite filter array. The FPGA has a built-in first-order FIR filter and a second-order Volterra nonlinear filter, and the FPGA has a built-in adaptive LMS algorithm to automatically correct the filter parameters according to temperature drift changes.

3. A CNC tooth-shaving machine for toothed heat sinks according to claim 1, characterized in that: The dual-channel time-separation signal processing module includes: Fast channel with a sampling rate of 10~20MHz, using an adaptive notch filter to suppress resonance interference generated by the main shaft; A precision channel with a sampling rate of 5~10MHz, using a multi-phase pipelined FPGA filter unit for high-precision steady-state data processing; It also uses a fuzzy logic controller to smoothly switch between speed, acceleration, and load current channels.

4. A CNC tooth-shaving machine for toothed heat sinks according to claim 1, characterized in that: The adaptive prediction and compensation system establishes a discretized prediction model that includes motor inertia, friction coefficient, and PID parameters. It performs multi-step prediction and feedforward compensation in each sampling period, and predicts the position value for the next N sampling periods based on the current state variable, where N is determined by the group delay of the current filter. This reduces the response delay caused by filtering.

5. A CNC tooth-shaving machine for toothed heat sinks according to claim 1, characterized in that: The torque sensing dual closed-loop control system includes: The coupling integrates a strain gauge torque sensor with a measurement accuracy better than ±0.5%FS and uses non-contact wireless signal transmission. A dual-channel data acquisition card is used to acquire the position signal and torque signal of the grating ruler, respectively. The control logic unit is used to dynamically adjust the current feedforward based on the deviation between the position loop and the torque loop to achieve stable loading and feeding.

6. A CNC tooth-shaving machine for toothed heat sinks according to claim 1, characterized in that: The dual-stage stiffness switching coupling is composed of an elastic rubber transmission component and an electromagnetic locking mechanism connected in parallel. Low-rigidity transmission is achieved through rubber transmission components when the power is off, and high-rigidity transmission is achieved through an electromagnetic locking mechanism when the power is on. The coupling is in a flexible mode during the start-up phase, with a stiffness range of 3000~6000 Nm / rad; During the steady phase, it is in rigid mode with a stiffness range of 40,000 to 60,000 Nm / rad. Furthermore, the switching signal is determined by comprehensively considering three conditions: stable speed, low torque fluctuation, and small position error.

7. A CNC tooth-shaving machine for toothed heat sinks according to claim 1, characterized in that: The servo drive execution unit, the predictive compensation system, and the torque control system adopt a fusion parameter control algorithm, which includes three weighted components: position command, speed feedforward, and torque compensation. The weight ratio can be dynamically adjusted according to the processing conditions to simultaneously optimize response speed and positioning accuracy.

8. A CNC tooth-shaving machine for toothed heat sinks according to claim 1, characterized in that: The system adopts an LVDS differential signal transmission structure and is equipped with a continuous-time linear equalizer and a pre-emphasis circuit to improve the integrity of the encoder signal during long-distance transmission.

9. A CNC tooth-shaving machine for toothed heat sinks according to claim 1, characterized in that: The edge intelligent signal preprocessing unit and the dual-channel signal processing module are connected via star grounding and independent voltage regulation to avoid multi-axis interference and improve stability during synchronous interpolation processing.

10. A CNC tooth-shaving machine for toothed heat sinks according to any one of claims 1 to 9, characterized in that: The entire machine structure is used in the processing of shovel-shaped heat sinks, with the tooth pitch consistency error controlled within ±0.01mm, the difference in tool mark depth controlled within ±3%, and it can perform precision shovel machining with continuous eight-axis synchronous interpolation.

Images