Material
Other, Global universal model
Condition
Other, Global universal model
Task
Other, Global universal model
Mathematical Model
Other, Global universal model
Signal
Other, Global universal model
Customized
Non-Customized
Structure
Other, Global universal model
SHINKAWA VM-5H3I. Overview
The SHINKAWA VM-5H3 is a high-precision semiconductor packaging inspection equipment mainly used for appearance inspection and defect analysis after chip packaging, suitable for quality control of semiconductor products such as integrated circuits (IC), power devices, and optoelectronic components. The equipment adopts a high-resolution optical imaging system and intelligent vision algorithms to automatically detect defects such as surface cracks, missing solder balls, wire bond misalignment, and contaminant residues on packaged chips. It features high speed, high sensitivity, and high repeatability, widely applied in mid-to-high-end production lines of semiconductor packaging test factories and wafer foundries. Its modular design supports switching between multiple inspection scenarios, adapts to different chip sizes and packaging types (e.g., QFP, BGA, flip chips), and meets the high-precision quality control requirements of semiconductor packaging.
II. Functional Features
High-Resolution Inspection Capability: Equipped with micron-level optical lenses (e.g., 10-500x variable magnification) and high-sensitivity CCD/CMOS cameras, the minimum inspection accuracy reaches 1μm, capable of identifying submicron-level surface defects.
High-Speed Automated Inspection: Maximum inspection speed up to 5,000 pieces/hour (depending on chip size and inspection accuracy), supporting continuous inspection of batch wafers, substrates, or individual chips to shorten quality control cycles.
Multi-Dimensional Defect Recognition: Supports 2D planar inspection (surface defects) and 3D topography analysis (e.g., solder ball height, wire bond arc), integrating algorithms such as grayscale analysis, edge detection, and template matching to cover over 99% of common packaging defects.
Flexible Inspection Modes: Compatible with multiple illumination methods including brightfield, darkfield, and polarized light, suitable for defect recognition on surfaces of different materials (e.g., metal, ceramic, molding compound), allowing customization of inspection areas and priorities.
Intelligent Data Management: Integrates defect classification statistics (e.g., defect types, location distribution), yield analysis, and historical data traceability functions, supports export in CSV/Excel formats, and can interface with factory MES systems for quality data sharing.
Human-Machine Interaction and Calibration: Equipped with a 17-inch touchscreen operation interface, supporting real-time image preview, parameter adjustment (e.g., light source intensity, inspection threshold), and built-in automatic calibration modules (e.g., focus, white balance calibration) to reduce operation complexity.
III. Technical Parameters
| Parameter Category | Specific Indicators |
|---|
| Inspection Range | X/Y axis stroke: 300mm×300mm, Z axis stroke: 100mm (adjustable focus range), θ axis rotation range: ±360° |
| Optical System | - Camera: 12MP color/monochrome CCD, frame rate ≥30fps - Lens: Motorized zoom lens (10-500X), adjustable depth of field - Light source: Multi-channel LED light source (red/green/blue/white), supporting multi-angle illumination |
| Inspection Accuracy | 2D inspection accuracy: ±1μm (3σ), 3D height inspection accuracy: ±0.5μm, repeat accuracy: ±0.3μm |
| Chip Size | Minimum inspection size: 0.1mm×0.1mm, maximum inspection size: 50mm×50mm, supported thickness: 0.1-5mm |
| Packaging Types | Supports QFP, BGA, CSP, flip chips, lead frame packaging, ceramic packaging, etc. |
| Motion System | Driven by servo motors + linear motors, positioning accuracy ±2μm, maximum acceleration: 3G, motion speed ≤500mm/s |
| Data Processing | Image processing speed: ≤20ms/frame, defect classification algorithm: supports deep learning (AI module optional) |
| Air Source and Power | Compressed air: 0.5-0.7MPa (for pneumatic components), power supply: single-phase AC 220V±10%, 50/60Hz, power consumption ≤3kW |
| Dimensions | 1600mm (L) × 1400mm (W) × 1800mm (H), weight approx. 800kg |
| Working Environment | Temperature: 23±3℃, humidity: 45%-65% (non-condensing), cleanliness: Class 1000 (ISO 7) |

IV. Working Principle
The working principle of SHINKAWA VM-5H3 is based on a collaborative architecture of "optical imaging + intelligent algorithms + motion control", with core processes as follows:
Workpiece Positioning and Pretreatment
Chips or substrates are fixed on the stage by vacuum adsorption or mechanical fixtures. The vision system first captures a global image to determine the coordinates of the inspection area.
Optical Imaging and Data Acquisition
Selects illumination modes (e.g., brightfield, darkfield) according to inspection requirements. High-resolution cameras scan the inspection area point-by-point or line-by-line to obtain 2D grayscale images or 3D point cloud data (via structured light or laser interferometry).
Defect Recognition and Analysis
The image processing system performs noise reduction, enhancement, and segmentation on images, identifies defects (e.g., cracks, foreign objects, dimensional anomalies) through preset algorithms (e.g., edge detection, template matching, grayscale threshold), and calculates deviations by comparing with standard templates.
For 3D inspection, surface topography data is obtained via laser triangulation or interferometry to analyze whether parameters such as solder ball height and wire bond arc meet specifications.
Data Processing and Judgment
The system automatically classifies defects (e.g., critical defects, acceptable defects), records information such as location, size, and type, determines whether the workpiece is qualified according to preset standards, and generates inspection reports.
Automation Cycle and Traceability
After inspection, the stage automatically moves to the next workpiece position, cooperating with robotic arms to achieve automated loading/unloading. Inspection data is stored in real-time, supporting subsequent quality traceability and process optimization.
V. Common Faults and Solutions
| Fault Phenomenon | Possible Causes | Solutions |
|---|
| Excessive inspection accuracy deviation | 1. Camera focus offset or lens contamination 2. Stage motion accuracy degradation 3. Unstable light source brightness | 1. Recalibrate focus and clean the lens 2. Check guide rail/screw wear and calibrate motion accuracy 3. Replace the light source or adjust brightness compensation |
| Defect missed detection or false detection | 1. Improper inspection algorithm parameter settings 2. Mismatched light source angle/wavelength 3. Insufficient image contrast | 1. Optimize threshold and template matching parameters 2. Adjust light source angle or replace wavelength (e.g., use blue light for metal defect inspection) 3. Enhance workpiece surface contrast (e.g., clean stains) |
| Abnormal camera imaging | 1. Poor camera connection contact 2. Camera sensor failure 3. Abnormal image processing card driver | 1. Check cable interfaces and restart the equipment 2. Replace the camera or sensor 3. Reinstall driver programs and upgrade firmware |
| Stage motion jamming | 1. Insufficient guide rail/screw lubrication 2. Motor driver failure 3. Mechanical structure interference | 1. Clean guide rails and add lubricant 2. Detect driver voltage/current and replace faulty modules 3. Check if transmission components are loose or blocked by foreign objects |
| Abnormal 3D inspection data | 1. Laser emitter power attenuation 2. Loss of 3D sensor calibration parameters 3. Workpiece surface reflection interference | 1. Replace the laser module and calibrate power 2. Re-perform 3D sensor calibration (e.g., Z-axis height calibration) 3. Add polarizers to reduce reflection or adjust inspection angle |
| Equipment alarm shutdown | 1. Safety sensor triggered (e.g., emergency stop button, safety door) 2. Overload protection or voltage abnormality 3. Software error | 1. Confirm safety device status and reset the alarm 2. Detect power voltage and load, repair abnormalities 3. Restart software, clear cache, and reinstall the system if necessary |