4.4 Programming & Tuning

Inspection programming involves translating raw component and solder data into actionable pass or fail criteria. The primary operational goal of tuning an inspection system is balancing a minimized false alarm rate against the goal of zero escapes (false negatives). This requires a structured approach that relies on Golden Boards, organized library management, and continuous iterations driven by the Pareto analysis of actual false calls. Inspection programs are not a substitute for process control; they function as a sensitive process monitor that flags upstream drift in the solder paste printing or Pick & Place operations.

The Inspection System Programming Loop

Programming and tuning transform static design data into a robust, repeatable inspection judgment that stays consistent across different shifts and operators.

1. Data Import: CAD or Gerber data (centroid location, rotation, polarity, generic component shape) is imported and mapped to the machine’s internal package library.
2. Teaching and Calibration: The machine is taught by scanning the Golden Board. This establishes the optimal lighting profiles for Automated Optical Inspection (AOI) and the specific slice planes for Automated X-ray Inspection (AXI) Laminography.
3. Tuning: Thresholds and inspection criteria are iteratively adjusted to balance detection capability against the operational cost of manual verification.

Critical Factor

The success of an inspection program is measured not only by the defects it catches, but equally by the low number of false calls it generates. An effective program catches anomalies without causing excessive line stoppages.

Library and Data Standardization

A standardized central library is helpful for smooth NPI changeovers and steady throughput across multiple products.

• One Package, One Library: A single, central library item should be created for each unique component package (e.g., standard 0402, 0.5 mm QFN). This central item defines the Presence Window, Polarity Region, and Solder Pad ROI (Region of Interest) for inspection across all products.
• Reusability: Reusing library items globally across all product programs ensures consistency and minimizes programming time for new products.
• AXI Specifics: For Automated X-ray Inspection (AXI), the library item should define structural parameters such as ball pitch, component height, and slice plane locations, along with clear voiding rules (e.g., voiding per individual ball vs. total combined area).
• Golden Set: The baseline program must be built using data derived exclusively from the Golden Board (the known-good assembly verified during First Article Inspection (FAI)). When possible, this should be supplemented with Near-Limit Boards (assemblies with barely acceptable features) to help train the system on acceptable boundaries.

Risk Classification and Limit Setting

Not all defects carry the same risk. Inspection limits must be outlined according to the actual risk level to prevent cosmetic issues from unnecessarily blocking production flow.

Risk Class	Defect Type	Limit Strategy	Action upon Failure
Class A (Critical)	Missing Component, Polarity Reversal, BGA Bridge.	Defined Limits. Typically triggers a stop or rejection to the rework station.	Should be addressed and cleared before the next panel begins.
Class B (Major Quality)	Tombstoning, Insufficient Fillet, Minor component skew.	Moderate Limits. Requires human verification and potential rework; data is logged for Statistical Process Control (SPC) tracking (Chapter 4.5).	Serves as a leading indicator to monitor upstream process drift.
Class C (Informational)	Smudges, Silk Screen Nicks, minor Flux Residue.	Tracking Only. Generally does not block Work In Progress (WIP). Used to gather data for periodic housekeeping checks.	Data is captured in the background but may not trigger a line alarm.

Tuning and the False Call Reduction Loop

Tuning is the continuous engineering process of adjusting parameters to reduce false calls while maintaining strong defect detection. This process should be driven by data.

1. Pareto Analysis: A sample lot (50–200 boards) must be inspected, and a Pareto chart of all false calls must be generated, classified precisely by component reference designator (RefDes) and the specific failure reason (e.g., “U10 fillet too reflective”).
2. Iterative Fixes: The top 3 false-call sources must be addressed with the smallest applicable programmatic change—for instance, adjusting a single lighting angle or subtly widening one component’s acceptable X/Y window. Global threshold relaxation should be avoided unless necessary.
3. Verification: A small mini-lot (10 boards) must be re-run to confirm that the false call rate drops, while verifying that the detection of true defects remains intact.
4. Process vs. Program: If tuning cannot resolve a consistent false call, such as recurring insufficient solder calls on a specific component, the issue is likely process drift (e.g., low Solder Paste Inspection (SPI) volume, poor paste release). Focus must be shifted from camera tuning to the physical stencil or printer recipe.

Change Control and Program Maintenance

Because inspection programs guide what ships to the customer, they benefit from the same change control rigor as Pick & Place (PnP) placement programs.

• Version Control: The Automated Optical Inspection (AOI)/Automated X-ray Inspection (AXI) program revision must be ensured to update with any change to the libraries, limits, or algorithm settings. The program version must be tied to the product’s Golden Recipe.
• Documentation: A brief explanation for every program change must be included in a changelog (e.g., “v2.1: Reduced false bridge calls on C47-C49 by softening the side-lighting angle”).
• Machine Learning (ML): If the inspection system uses Machine Learning (ML)/Artificial Intelligence (AI) for defect classification, a diverse training set must be utilized. A holdout set of images must be reserved for regular, independent verification of the model’s accuracy to ensure it continues to identify defects correctly.

---

Recap: Defect Classification and Response Protocol

Defect Class	Defect Examples	Limit Strategy	Action Upon Failure
Class A (Critical)	Missing Component, Polarity Reversal, BGA Bridge	Defined Limits	Stop/Reject; Clear before next panel
Class B (Major Quality)	Tombstoning, Insufficient Fillet, Minor Skew	Moderate Limits	Human Verification & Rework; Log for SPC
Class C (Informational)	Smudges, Silk Screen Nicks, Minor Flux Residue	Tracking Only	Background Data Capture; No Line Block
False Call Reduction	Pareto of False Calls by RefDes/Reason	Iterative Parameter Tuning	Address top 3 sources; Verify on 10-board mini-lot
Library Standardization	One Package, One Library Item	Central Library for all products	Ensure consistency across all programs