How AI-powered automated X-ray inspection (AXI) is transforming PCB assembly quality control — detecting defects in BGA solder joints, hidden vias, and multi-layer assemblies that human operators cannot reliably identify.
AI Meets X-Ray: The Next Frontier in PCB Inspection
Express Manufacturing, Inc. (EMI) has expanded its inspection capabilities with the TR7600FB SII automated X-ray inspection system from Test Research, Inc. (TRI). The system uses AI-driven inspection algorithms to detect solder defects in high-density assemblies that traditional visual inspection cannot reach.
As PCB assemblies grow denser and more complex — with BGA packages, QFN thermal pads, and package-on-package stacking — human-dependent inspection becomes insufficient. AI-powered AXI fills the gap by identifying defects inside solder joints without requiring human operator judgment calls.
Key Technical Capabilities
The TR7600FB SII brings several advances:
- 5 µm resolution: Detects micro-voids and hairline cracks invisible to standard X-ray
- AI-driven void detection: Machine learning algorithms automatically identify and classify solder voids without manual programming
- Automated programming: AI reduces setup time — the system learns acceptable solder joint profiles from reference samples
- 20 fields of view per second: Maintains production throughput while increasing inspection depth
- Board capacity: Handles assemblies up to 850 mm × 520 mm and 12 kg
Why This Matters: Hidden Defects in Modern Assemblies
Traditional AOI (Automated Optical Inspection) can only see surface-level defects. For high-reliability assemblies, the critical defects are hidden:
Defects only X-ray can find:
- BGA solder ball voids (>25% void = reliability risk)
- Head-in-pillow (HiP) defects in fine-pitch BGA
- Insufficient solder under QFN thermal pads
- Cold joints in bottom-terminated components
- Internal layer misregistration in assembled modules
Where AI adds value over conventional X-ray:
- Reduces false calls by 40–60%
- Consistent judgment with no operator fatigue
- Adaptive thresholds based on component criticality and IPC standards
- Fast enough for 100% inline inspection rather than statistical sampling
Industry Standards Integration
The system supports modern factory communication protocols:
- SMEMA: Standard mechanical-electrical interface
- SECS/GEM: Equipment communication standard
- IPC-CFX-2591: Connected Factory Exchange for IIoT
- IPC-HERMES-9852: Board transport and traceability
This connectivity means inspection data integrates directly with MES, enabling real-time yield tracking, automatic process adjustment triggers, and full traceability.
The Broader AI Inspection Trend
EMI's investment reflects acceleration in AI adoption across PCB inspection:
| Company | AI Capability | Year |
|---|---|---|
| Express Mfg / TRI | AI void detection + auto-programming | 2026 |
| Jabil / ViTrox | AI digital factory, multi-inspection fusion | 2026 |
| Koh Young | Deep learning solder classification | 2025 |
| Mirtec | AI-powered false-call reduction | 2025 |
The pattern is clear: inspection equipment is transitioning from rule-based (human-programmed criteria) to learning-based systems.
What This Means for Hardware Engineers
- Design for inspectability: Component placement should allow clear X-ray access
- Void budget planning: With consistent AI measurement, tighter void requirements are achievable
- First-pass yield data: AI inspection generates rich datasets for DFM optimization
For engineers designing assemblies with BGA, QFN, and hidden-joint components, the availability of AI-powered inspection means higher quality is achievable — but designs should still consider X-ray accessibility during layout.
Source: PCB Directory, June 1, 2026.
Top comments (0)