Trusted on the shop floor of
Zero-defect MLCCs at production speed. Intelgic's machine vision + AI platform automates inline and offline inspection from green tape to sintered chips, singulation, plating, and tape-and-reel — detecting sub-10 µm defects with full traceability.
MLCCs are tiny, brittle, and process-sensitive. Cosmetic flaws often correlate with latent reliability risks, while the smallest handling nick can become a catastrophic crack after reflow.
Chips as small as 0201/0402 (and smaller), millions per day.
Micro-chipping and micro-cracks are hard to see, easy to miss.
Matte vs. glossy end terminations, plating tone changes.
Green tape punching, stacking, lamination, dicing, tumble deburr, plating, marking, and packaging each introduce unique defects.
Inspection must match high-speed singulation and tape-and-reel rates without false rejects.
Defect categories covered across body/ceramic, end terminations, dimensional/handling and marking/packaging.
Note: Electrical parameters (capacitance/ESR/DF/IR) are typically verified by E-test/ATE. Intelgic can interface with those stations and correlate visual-to-electrical outcomes for richer SPC.
Dedicated stations tuned to each stage of the MLCC manufacturing flow.
Cameras, lenses, lighting geometry, and precision mechanics tuned for MLCC metrology.
20–65 MP industrial sensors, global shutter; optional line-scan for continuous web.
Telecentric for metrology; long-WD macro for TnR; motorized focus for thickness steps.
Vibration-isolated stages, fiducial-based alignment, flip modules for dual-side viewing.
Intelligent defect detection and classification powered by machine learning.
Learn "golden appearance" of ceramic and plating; flag novel defects without exhaustive labeling.
Classify chips (OK, minor chip, severe chip, crack, plating short, porosity, contamination, etc.).
Measure crack length/edge chip area in µm; compute wrap/coverage metrics.
Multi-angle reflectance or height proxies for subtle bevel defects.
One station handles multiple sizes/series — swap recipes in seconds; automatic scale re-calibration with a grid target.
Example acceptance rules used by Certainty AI — fully tunable to your QA standards.
L/W/T tolerance in µm; bevel min/max; wrap length %.
Max chip area (µm²) per edge; crack length/angle thresholds.
Minimum coverage; maximum porosity count/area; blister detection confidence.
Particle size/contrast thresholds; stain/mark acceptance windows.
OK / Rework / Reject with IPC-like severity bands (customizable).
All measurements are logged per serial/lot and bound to DQC (Digital Quality Certificate).
Designed to slot into existing MLCC lines with minimal disruption.
Tuned to match singulation and TnR lines (multi-camera parallelization supported).
Real-time PASS/FAIL with reject chute, marker, or diverter outputs.
OPC-UA / MQTT / REST / SQL; barcode/QR lot binding; ATE & MES/ERP hooks.
Operator, QE, Admin; recipe version control; audit trail.
From per-chip records to shippable digital certificates — full lineage on every reel.
Images, masks, metrics, decision, recipe version, operator/line ID, timestamps.
Pareto of top defects, FP/FR trend, heatmaps, SPC charts, OEE.
Link visual defects to upstream steps (lamination, sinter, plate, pack).
Digitally signed certificate for every shipped reel/tray.
Why MLCC manufacturers choose Intelgic's AOI platform.
Detect cracks & chips that drive field failures with unprecedented precision.
Learned normal modeling and robust lighting reduce false positive rates.
Reduce warranty exposure and improve overall cost of quality.
Handle multiple sizes/series with quick recipe switching.
Digital Quality Certificate (DQC) for compliance and customer audits.
For internal electrode faults, X-ray is typical. Intelgic focuses on external visual and metrology; we can integrate X-ray results into the same dashboard/DQC if available.
Yes — telecentric optics + recipe scaling and auto-calibration support a wide range of sizes.
Stage & scale calibration with traceable targets; periodic checks; recipe-bound scale factors; GR&R reporting.
Cross-polarized coaxial + darkfield blends and exposure bracketing; AI trained across illumination variants.
©2026 Intelgic Inc. All Rights Reserved.
