ItInsightsWorld AIapplications: How Siemens and NVIDIA Move PCB Inspection Preparation Into a Digital Twin Workflow
At the application layer, the practical question is where AI removes waiting, sample handling, and repeated manual tuning from quality inspection. Siemens and NVIDIA have arranged industrial software, Siemens’ Digital Twin technology, and NVIDIA Cosmos around that question for PCB inspection preparation. The use case centers on automated optical inspection (AOI) and starts at Siemens Electronics Factory Erlangen, where product and process changes can occur three to five times per week.
At the Siemens Electronics Factory in Erlangen, Germany, a digital twin of the SMT AOI inspection area is produced with the Digital Twin Composer. The factory contributes the manufacturing setting, a high-mix use case, and expert knowledge. The objectives include up to an 85% reduction in manual engineering effort for AOI program generation, a 40% reduction in false calls, and more than 5% first-pass-yield improvement.
Application path
Before inspection was often prepared after a product variant was physically prototyped. The revised flow moves part of preparation into a digital setting.
- Siemens Teamcenter provides PCB design files, bill of materials data, and component details.
- Process Preparation X checks, normalizes, and merges inputs into a digital twin of the printed circuit board assembly (PCBA).
- The golden board, meaning the reference board’s digital twin, is paired with NVIDIA Cosmos to create a photorealistic virtual representation and synthetic samples for accepted and faulty states.
- Siemens uses NVIDIA Cosmos Data Factory to generate synthetic circuit board defect images, allowing calibration before physical boards exist.
- During production, secondary inspection assesses AOI flags and helps decide whether a possible flaw is real or false.
Comparison
| Dimension | Prototype path | Digital path |
|---|---|---|
| Timing | Wait for a physical board. | Start virtual setup early. |
| Reference | Use physical board. | Use golden board twin. |
| Defects | Use physical samples. | Use synthetic defect images. |
| Span | Pre-production setup. | Preparation, calibration, and inspection assistance. |
Readout
Images of the Golden Board and potential defects are automatically produced from the digital twin of the PCB using Process Preparation X software and NVIDIA Cosmos. The approach is meant to support earlier inspection setup, reduce manual engineering and verification demand, limit reliance on physical samples, and strengthen inspection calibration across virtual and physical phases.
Checklist
- Choose an AOI setup pain point.
- Map Teamcenter design data for the pilot.
- Confirm Process Preparation X normalization into the digital twin.
- Define golden board, accepted, and faulty states.
- Plan secondary inspection review for false calls.
FAQ
Is this only AOI program generation?
No. The workflow covers virtual preparation, calibration, and inspection assistance after production begins.
What is generated before boards exist?
A virtual representation and synthetic samples, including defect images, support calibration.
Why does Erlangen matter?
It is Customer Zero for adaptive AI-supported production in a high-mix environment.
Common misconceptions
- It is not a model-architecture showcase; it is inspection-workflow embedding.
- It does not replace production validation; it shifts part of setup earlier.
- Synthetic images do not promise zero false calls; they support calibration toward stated targets.
Action suggestions
- Begin with one production line, not factorywide transformation.
- Use Teamcenter as the data entry point.
- Validate twin output against inspection criteria.
- Track false-call review and yield movement.
The useful takeaway is that manufacturing AI often works first as a workflow bridge: it shifts physical setup into digital preparation, then supports decisions after production starts.