3.3 Inspection Method Validation (AOI/X-Ray/ICT/FCT Correlation)

Inspection is not a passive filter; it is a data acquisition system that defines the physical reality of the PCBA. Relying on unvalidated machines is arguably worse than no inspection at all, as it creates a false sense of security while satisfying neither physics nor the customer. This chapter defines the validation protocols and the critical correlation logic required to ensure your test gates—AOI, X-Ray, ICT, and FCT—actually detect the failures they are assigned to catch.

The Detection Physics (Coverage Mapping)

Do not assign a failure mode to a machine that cannot physically detect it. A camera cannot see through silicon; an electrical probe cannot measure solder fillet volume.

• AOI (Automated Optical Inspection): Line-of-Sight. Detects visible misalignment, polarity, and solder meniscus.
  • Blind Spot: BGA/LGA joints, hidden solder connections.
• AXI (Automated X-Ray Inspection): Density Delta. Detects voiding, shorts, and bridging under packages.
  • Blind Spot: Open circuits caused by "Head-in-Pillow" (HiP) defects where density exists but fusion does not.
• ICT (In-Circuit Test): Electrical Parametrics. Detects shorts, opens, and component values (R/L/C).
  • Blind Spot: Cold solder joints that make temporary contact under probe pressure.
• FCT (Functional Test): Behavioral Logic. Validates system performance.
  • Blind Spot: Latent defects (e.g., weak solder) that function now but fail under vibration.

Method Validation Protocol

Before mass production, validation must prove the machine can reject a known bad unit (Reliability) and accept a known good unit (Repeatability).

The "Golden" and "Red" Sample Test

Create a validation set comprising 1 verified "Golden Sample" (Perfect) and specific "Red Samples" (Defective) for each failure mode.

• Step 1: Run Golden Sample 30 times.
  • If any False Call (False Fail) -> Then retune thresholds. Capability is low.
• Step 2: Run Red Sample (e.g., Missing Component) 10 times.
  • If any Escape (False Pass) -> Then STOP. The program is invalid.
• Step 3: Introduce marginal defects (e.g., shifted component).
  • If machine passes marginal unit -> Then define the specific gray zone limit in the Quality Plan.

Pro-Tip: Never use a production unit as a Golden Sample without CT-Scan or Cross-Section verification. "It passed FCT" does not mean the solder joints are structurally sound.

Correlation & Feedback Loops

Individual machines are silos. Quality exists in the correlation between them.

Scenario A: AOI Pass -> ICT Fail (Short)

Analysis: The short is likely under a component body or caused by solder paste slump bridging traces.

Action:

• If systematic -> Then Update SPI (Solder Paste Inspection) volume limits.
• If random -> Then Check pick-and-place nozzle pressure (squishing paste).

Scenario B: X-Ray Pass -> FCT Fail (BGA Open)

Analysis: Classic "Head-in-Pillow" or micro-crack. X-Ray sees the ball aligned to the pad (density match), but no metallurgical bond exists.

Action:

• If frequent -> Then Switch to 2.5D or 3D X-Ray (Laminography) to inspect the specific interface layer.
• If unavoidable -> Then Mandate Boundary Scan (JTAG) at ICT/FCT to electrically validate the net.

Scenario C: High False Call Rate (AOI)

Analysis: Engineers often loosen AOI tolerances to keep the line moving, increasing the risk of Escapes.

Action:

• If False Call rate > 5000 PPM -> Then Halt line. Re-programming required. Operators ignore alarms when false calls are high ("The Boy Who Cried Wolf" effect).

Threshold Management

Define limits based on statistical reality, not default software settings.

• Voiding (X-Ray):
  • Standard: IPC-A-610 allows ≤ 25% voiding area.
  • High Reliability: Set warning limit at 15%.
  • If voiding > 20% in thermal pad -> Then Process Control failure (Reflow Profile).
• Component Shift (AOI):
  • Limit: ≤ 50% overhang.
  • Action: If shift is consistently 20% in one direction -> Then Offset pick-and-place coordinate data. Do not just widen the AOI window.

Final Checklist

Parameter	Rule / Threshold
Validation Frequency	Run Golden/Red samples at every shift start and changeover.
Escape Tolerance	0. An escaped "Red Sample" invalidates the entire shift's production.
False Call Limit	Max 3 false calls per panel. Exceeding this triggers engineering review.
BGA Inspection	2D X-Ray minimum; 3D X-Ray recommended for Class 3 products.
Program Change	Any change to thresholds requires re-validation with Red Samples.
Correlation Rule	If ICT fails, Then feed data back to AOI/SPI to close the detection gap.
Red Sample Storage	Keep Red Samples in a locked "Defect Library" box. Do not lose them.