Turn every microscopic target into actionable quality data.
Vision inspection and metrology products for PCB micro-holes, post-dicing edge defects, and semiconductor test-pin arrays — replacing sampling with 100% inspection and subjective judgment with quantified data, from sample validation to line-side inspection to standard equipment delivery.
Numbers that come with a method.
Typical values — repeatability, speed, and other metrics vary with optical configuration, field size, and sample condition. Every metric is calibrated on your samples and written into the technical agreement.
Three product lines. One measurement platform.
Models, software platform, AI compute, and equipment — all in-house. Choose a product direction and start from a sample validation.
PCB Micro-Hole Measurement & Inspection
Every hole becomes traceable measurement data.
Large-field imaging plus massively parallel per-hole computation: diameter, roundness, position, concentricity, and burr/nick verdicts across 5K–10M holes — with a full-panel hole quality map that pinpoints local process drift.
Post-Dicing Inspection & Edge Quantification
Intercept bad dies before they consume packaging cost.
Full-wafer inspection at the post-dicing stage: low-k sidewall and hairline cracks, chipping, notches, burrs, and particles — quantified with position, size, and class, at under 0.5 s per stitch-and-inspect cycle.
Test-Pin Inspection & Lifecycle Control
From "visually checked" to "quantitatively confirmed".
Pin height, coplanarity, positional offset, and surface condition quantified pin by pin across sockets and fine-pitch arrays — with repair before/after comparisons and life-trend reports.
What sampling misses, you repay downstream.
High-density micro-holes, dicing edges, and test-pin arrays share the same traits: huge target counts, tiny defects, and gradual degradation. Low-coverage sampling and manual inspection are inherently poor at catching local drift. A quality rule of thumb puts the cost of escapes at 1–10–100: a defect fixed at its own step costs 1; at the next step, 10; at the customer, 100.
All in-house, one owner closes the loop
Models, software platform, AI compute unit, and standard equipment are all self-developed — accuracy, speed, and false-call issues are never bounced between vendors.
Large field × high res × high speed
High-speed micron-level stitching plus hardware-accelerated compute let 1000 mm fields coexist with micron-level metrology, with throughput scaling with compute.
Real-vs-false AI with verifiable verdicts
Dedicated models trained on your samples separate real defects from normal texture or saw marks; overkill and escapes are accepted against a jointly agreed sample set.
Phased adoption, controlled risk
Three delivery tiers — software → integrated unit → standard equipment — starting from sample validation, each step with clear deliverables and no one-shot bet.
The cost ratios are a general industry rule of thumb; actual magnitudes can be estimated jointly using customer process costs.
Every published metric is explainable, calibratable, and verifiable.
We provide the definition and the calibration method, not just the number. Repeatability, detection capability, inspection speed, and verdict consistency are each accepted on jointly confirmed sample sets.
Start with a sample validation.
Choose a product direction and prepare representative samples — good parts plus known anomalies. Insightek responds with a feasibility conclusion and a demo report. No need to decide on equipment up front.