Smart Cooling Equipment contributes to predictive maintenance

Releasing a quality-assurance bottleneck with the power of AI

Siemens Elektronikwerk Amberg has trained an algorithm to predict the probability of defects to streamline X-ray testing of printed circuit boards.

 

Ensuring solder joints on printed circuit boards are flawless is a vital part of quality assurance. The turnaround time of X-ray end-of-line PCB testing emerged as a bottleneck slowing overall output at Elektronikwerk Amberg (EWA). Instead of going the conventional route and adding X-ray machines, EWA turned to AI, edge and cloud computing to build an algorithm for predictive quality analytics. Embedded in an highly automated factory environment, this AI-driven model is one more building block supporting EWA on its path to zero defect manufacturing.

  

Siemens Elektronikwerk Amberg

Manufacturing tomorrow's electronics today

Elektronikwerk Amberg (EWA) produces 17 million Simatic products a year with 1,200+ product variants at a manufacturing quality of 99.999 percent. The award-winning pioneering smart factory went operational in 1990. Since then, EWA has automated 75% of its value chain and increased its output by factor 13. An IIoT-enabled central real-time quality management system supports EWA as it improves products, processes, and suppliers to reach Six Sigma quality.

Industrial Edge – Powering local intelligence for predictive quality analytics

For its AI-driven predictive-quality solution, EWA emphasizes local bulk processing power. Their platform of choice is Industrial Edge, an integrated edge-computing solution by Siemens. Industrial Edge supplies high availability and short latency times that enable fast responses. It processes bulk data close to the source without time-consuming edge-cloud interaction, making it more cost-effective and secure.

 

Learn about Industrial Edge

Solution in a Nutshell

In an AI-driven model it calls closed-loop analytics, EWA uses the analysis of process data at the shopfloor level to control and optimize the underlying process in real-time. The process at hand is the end-of-line X-ray testing of solder joints in PCB inspection. EWA decided to use artificial intelligence to predict whether a particular component even required time-consuming X-ray testing. 

EWA created transparency by installing sensors automated with Totally Integrated Automation (TIA). These sensors collect data from the solder paste printer, solder paste inspection, pick & place, and the automatic optical inspection (AOI). Their output of some 40 different data sets was structured and transferred to the MindSphere cloud platform. Here, MindSphere’s storage capacity, raw processing power, and artificial intelligence come together to provide what amounts to a machine-learning school in the cloud for algorithms. Once EWA’s algorithm had digested its historical data sets, the predictive-quality model moved to the edge device.

The AI-driven edge application harvests and pre-processes the data it needs to predict the quality of soldering joints right at the source. All production-critical data remains in the manufacturing environment. The edge application shares its prediction – whether or not a PCB requires X-ray – with the Simatic IT manufacturing execution system. This system, in turn, decides whether it opts for or skips the X-ray test. Each test skipped opens up the bottleneck.

Project Scope

EWA identified X-ray end-of-line testing as the critical bottleneck slowing the throughput of its surface-mounted device line. The project’s primary objective was to speed up the PCB inspection of solder joints for BUS-Connector PINs in the SIMATIC ET200SP Base Unit. The task was to design, build, and implement a smart digital solution for increasing the line’s throughput while providing an alternative to buying more X-ray machines. This alternative was to slot into and draw on EWA’s existing smart-factory infrastructure and contribute to EWA’s goal of zero-defect manufacturing. 

The business objective was to build a proof-of-concept model for predictive quality analytics that was cost-effective, secure, scalable, open to further optimization, and replicable in other production lines.

Connected Assets

  • SIMATIC S7-1500
  • SIMATIC IPC227E
  • Machine controllers and IPCs

Queried data

40000 production parameters from sources along the whole production line.
Such as:

  • PCB Order / Product Information
  • Soldering Temperature
  • X, Y, Z placement derivations on PCB
  • And many more

Our Achievements

MindSphere

30% fewer X-ray tests 

X-ray testing is down by up to 30%, unblocking a costly bottleneck in manufacturing. The target now is to drive X-ray testing down to zero

MindSphere

100% quality rate maintained

When the certainty of the algorithm‘s quality prediction is less than 100%, the PCB moves on to X-ray testing

MindSphere

500,000€ less capital investment 

Resolving the bottleneck no longer requires an extra X-ray machine. EWA saves the initial investment and costs, from machine integration to maintenance

Voices

In the future, artificial intelligence will help us better understand our processes. With AI, zero-defect production is no longer a vision for us but a tangible goal.

Gunter Beitinger, Plant Manager at Elektronikwerk Amberg

Offers to replicate this use case

  Cloud computing:

MindAccess:

  

MindAccess:

  

MindAccess:

  

MindAccess IoT
Value Plan

Visual Analyzer

MindSphere

Predictive Learning

  

  

  Edge computing:

MindAccess:

  

MindAccess:

  

MindAccess:

  

MindSphere

SIMATIC IPC227E Nanobox PC

MindAccess:

  

MindAccess:

  

MindAccess:

  

MindSphere

OPC UA Connector Edge App

MindAccess:

  

MindAccess:

  

MindAccess:

  

MindSphere

Flow Creator Edge App

MindAccess:

  

MindAccess:

  

MindAccess:

  

MindSphere

AI Runtime Edge App

Connect with an IIoT expert

Business case, scope, goals: There is no one-size-fits-all IIoT strategy. Explore how you can leverage the power of IIoT to drive the value of your operations and business model.