From Hours to Minutes: AI-Powered PCB X-Ray Inspection for Vinyas
An AI inspection pipeline is replacing manual BGA void detection with target full-board inspection in under 10 minutes.
For Vinyas Innovative Technologies, ITChamps is building an AI pipeline on top of Nordson Quada 3 infrastructure to automate BGA localization, solder-ball detection, void measurement, and traceable report generation.
The Challenge
Manual BGA X-ray inspection was too slow and structurally incomplete
The existing workflow depended on operators to navigate BGAs, mark shadow-zone balls, estimate voids, and manually log outcomes, creating throughput, consistency, and traceability problems.
- A single PCB inspection consumed hours of skilled operator time.
- The Quada 3 semi-auto workflow missed shadow-zone solder balls entirely.
- Void estimation was subjective and varied between operators.
- Inspection data lacked automated traceability and searchable reporting.
- Throughput scaled with headcount rather than capability.
The Solution
An AI pipeline layered onto existing Nordson Quada 3 hardware
The system detects BGA locations, identifies all solder balls including occluded ones, segments voids, classifies pass/fail against IPC-7095D thresholds, and automatically generates traceable reports.
YOLOv8 and Grounding DINO detect and map all BGA locations from X-ray images.
SAM 2 and a Vision Transformer identify all balls, including shadow-zone cases the native system misses.
U-Net and PatchCore segment void regions and calculate precise void percentages against IPC-7095D thresholds.
The pipeline generates PDF and Excel reports linked to serial numbers, dates, and operators in a searchable quality database.
Results
Measured business impact
Full PCB inspection target from a current baseline of several hours.
Coverage target includes shadow-zone balls missed by the current workflow.
Planned accuracy target comparable to dedicated AXI systems.
The design aims to eliminate manual marking, detection, and report generation.
Highlights
- Implementation is currently in progress for April to September 2026.
- No new X-ray hardware replacement is required.
- The design supports IPC-7095D-compliant traceable reporting.
- The system is intended to create a searchable quality intelligence database.
Services and technology
Want to know more about this story?
More Stories
Related case studies
How AI-Powered Inventory Optimization Exposed the 800 SKUs Driving 70% of Cost
For a global server enclosure manufacturer, ITChamps deployed an AI-powered inventory optimizer that exposed SKU concentration risk, improved forecasting, and strengthened service levels without capital investment.
Read case study
Grinding Optimization Where 40% of Operating Cost Is Electricity
ITChamps used historical plant data to identify and validate optimal grinding parameters offline, then converted those findings into an operational knowledge base for sustained use in production.
Read case study
How AI Unlocked 30%+ Efficiency Gains in a Thermal Power Plant
A thermal power operator used ITChamps' Process Optimizer to identify the highest-leverage control variable, map the productivity zone, and sustain materially better efficiency without capital investment.
Read case study