An FAU (fiber array unit) is a key packaged component in optical modules and silicon-photonics coupling, and its appearance and assembly condition directly affect downstream assembly reliability. Before delivery, the finished part needs inspection across overall dimensions, V-groove offset, cover-plate defects, adhesive condition, shell scratches and component misalignment. Conventional single-side or manual visual inspection can hardly cover all six orientations in one operation. The FAU six-side inspection machine from Shenzhen Xingte Technology Co., Ltd. uses multi-station vision cameras to synchronously capture images from six directions, completing a full-area appearance check in a single loading cycle.
1. Inspection Objects and Defect Types
The machine targets finished FA fiber arrays and fiber components, serving as a full-appearance quality gate at the end of the line. Defects to be screened out include overall dimension deviation, V-groove offset relative to fibers, cover-plate surface defects, adhesive overflow or shortage, shell scratches, and misaligned components. Some defects come from assembly alignment, others from dispensing and handling; if they flow into downstream steps, coupling and packaging quality will be affected. Six-side inspection exists to screen out these defects before finished parts leave the line.
2. Principle of Six-side Synchronous Imaging
Multi-station vision cameras are arranged around the fixture. Once a part is loaded, cameras in each direction capture appearance images synchronously, covering all six faces of the component for full-area imaging with no blind spots. Compared with flipping part by part or rotating by hand, synchronous acquisition shortens per-part cycle time and avoids bumping caused by repeated handling. With a fast inspection cycle, the machine supports the quality-inspection needs of high-volume FA component production.
3. Inspection Items
In one imaging pass the machine screens multiple dimensions: overall dimensions judge whether the outline is within control; V-groove offset checks the relative position between the fiber array and the grooves; cover-plate defects inspect surface appearance; overflow and shortage correspond to adhesive application status; shell scratches trace handling and transport damage; component misalignment verifies whether each assembled part is in place. Thresholds for each item can be set according to process requirements, keeping defect judgment consistent.
4. AI Recognition of Multiple Defect Classes
The machine uses an AI algorithm to recognize multiple defect classes in parallel on the synchronous images, including dimension deviation, adhesive defects, appearance bumps and assembly misalignment. The algorithm compares image features with a preset defect library and outputs a judgment, reducing missed and false calls caused by eye fatigue and subjective differences in manual inspection. Parallel recognition of multi-class, multi-face defects is what distinguishes six-side inspection from single-point inspection.
5. Automatic Marking and Data Reports
Parts judged defective are automatically marked for downstream sorting, while results are compiled into data reports recording defect types and distribution. Reports can be used to trace batch quality and support iterative improvement of line processes. Modular quick-change fixtures let the machine adapt quickly to different fiber-array specifications. Shenzhen Xingte Technology Co., Ltd. can provide vision inspection solutions for FAU finished-part quality control.
