Why Visual Quality Inspection for PCBs is a Necessity During the Component Shortage—and Beyond

by DarwinAI

As the global pandemic enters its third year, a perfect storm of a sustained demand surge and supply chain disruption is presenting original equipment manufacturers (OEMs) with an array of challenges, including:

  • Component shortages
  • Long lead times
  • Higher prices

OEMs—particularly those who manufacture printed circuit boards (PCBs)—are sandwiched by these issues: on the input end, they depend on supplies to feed their production processes; on the output end, they’re doing their best to meet demand for the products and components they produce.

The short-term outlook: more of the same

Unfortunately, the underlying causes and contributing factors are not going to be resolved in the short term.  Global semiconductor sales increased by 26.8% from January 2021 to January 2022, indicating that demand for chip-hungry products—from automobiles, smart appliances, and other consumer devices to medical equipment and 5G infrastructure—is a long way from returning to pre-pandemic levels.

Similarly, it will take at least the rest of 2022—and perhaps much longer—for the global supply chain to return to normal operations (even without additional large-scale COVID disruptions). What’s more, significant production increases are much further off, as new facilities are years away from coming online.

Compounding matters, the well-documented labor shortage that’s slowing down production lines are more likely to be the new normal than merely a short-term blip.

To top it off, geopolitical uncertainties abound, threatening future supply disruptions and giving risk management professionals much to consider.

PCB inspection remains a production bottleneck and a contributor to high cost of poor quality (COPQ)

To satisfy demand, OEMs are under pressure to produce and assemble PCBs at record-setting rates. However, these initiatives are complicated by several factors, including:

  • Shortfalls in foil copper supplies
  • New suppliers, brought in to help alleviate shortages
  • New materials, as manufacturers look for alternatives to those in short supply

While new suppliers and materials build supply chain redundancy, they also complicate operations by introducing new variables.

But perhaps the largest barrier to ramping up production is the bottleneck created by highly manual PCB inspection processes. For high-mix, low-volume PCB inspection, some OEMs attempt to address the issue by employing a small army of quality inspectors. However, the aforementioned shortage of skilled workers renders this approach an expensive “band-aid” at best; moreover, while throughput may superficially increase, human error offsets these gains and lowers efficiency.

The result is a process largely dependent on human labor, one that suffers from high escape rates and high false positives. For the business, this means a high cost of poor quality (COPQ) due to unnecessary rework, less scrap, and more expensive failures uncovered only later in production.

To help mitigate the supply and demand crunch in the short term, OEMs need to make the most of the resources, materials, and personnel available today.

Over the longer term, OEMs can lower COPQ by investing in innovative solutions.

Fortunately, there’s a proven way for OEMs to meet their short-term needs while laying the groundwork for long-term efficiency gains.

Using visual quality inspection to improve product quality and production efficiency

COPQ captures the direct and indirect costs that wouldn’t exist if systems, processes, and products were perfect. To lower COPQ, manufacturers can invest in solutions that improve the detection and prevention of failures, so long as those solutions return more in savings than they cost to implement.

One such solution making waves in manufacturing is visual quality inspection (VQI), which leverages artificial intelligence (AI) to improve the speed, accuracy, and efficiency of quality control processes.

Here’s how a typical VQI inspection point works:

First

One or more cameras capture multiple images of the PCB under inspection.

Second

An AI engine examines the images and identifies production defects.

Third

The AI presents the findings—usually, a pass/fail plus the reasons for the decision—to an operator through a visual user interface.

Fourth

The operator either validates the AI’s decision or overrides it.

The result is a versatile, non-destructive, highly scalable, and highly automated inspection process that can be applied to a range of inspection use cases, such as:

  • Surface defects, including distinguishing between cracks, dents, etc.
  • Coatings, to ensure desired coverage uniformity
  • Metal alloys, by inspecting grain size
  • Solder joints, for worker and non-worker defects
  • Assemblies, to meet product specifications

Advanced systems can even distinguish between debris and damage—including specific types of damage—to inform rework decisions, creating further efficiencies.

Perhaps more importantly, the performance of such human-in-the-loop solutions continue to improve over time: while the system’s initial capabilities are largely determined by the training data with which it’s provided, it rapidly improves based upon the feedback provided by the human operator. The result is a fast and more accurate solution that accommodates different board types while reducing the operational burden of fine-tuning existing systems (and one that’s never impeded by fatigue).

In the short term, VQI can lessen the impact of many of the challenges OEMs are experiencing today, by:

  • Reducing scrap, ensuring the manufacturer makes the most of their constrained supply
  • Acting as a force multiplier, enabling a single skilled employee to perform at the level of an entire team—but without overburdening them—and allowing the organization to reassign team members where they’re most valuable
  • Making inspection viable—in terms of performance and cost—earlier in the production process, when failures are cheaper

In the long run, VQI can deliver extraordinary savings over the life of a production line.

For example, a leading global manufacturer of aerospace components used DarwinAI’s VQI solution to improve the speed and accuracy of conformal coating inspection of PCBs—the solution paid for itself in the first month, and the savings will continue for as long as the line is operating.

How can DarwinAI help?

Our Solution

While the VQI process outlined above sounds simple, the technology behind it is anything but—and that’s where DarwinAI comes in.

We provide an end-to-end deep learning-based VQI solution for manufacturers to improve product quality and reduce production costs by easily integrating AI into their existing production process. We specialize in building accurate, efficient, and robust AI with less data than conventional AI methods—delivering faster time to value.

Whether you want to automate a VQI inspection point so you can reassign your inspectors to more complicated tasks, or you want to bring in some AI assistance to scale the impact of your existing team, we can help.

Here are a few ways you can learn more about how DarwinAI can help your manufacturing process:

Of course, you can always contact us directly—we’d be happy to hear about your manufacturing process and to discuss ways in which VQI can help.

More Resources