Closing the quality loop at Krunon with Limen

Mastering a demanding material

In injection moulding, consistency is built on thousands of small decisions: the design of the mould, the choice and condition of the material, the machine settings and the experience to understand how they influence one another.

Krunon has spent more than 30 years developing that understanding. From its facility in Bunnik, the company produces high-quality plastic components in both small and large series. Its compact team combines mould design, material knowledge and production expertise with an approach that is deliberately practical: understand the customer’s application, build a stable process and keep looking for ways to make it better.

One of Krunon’s key production lines shows why that experience matters. The component is moulded from a specialised elastomer with demanding processing characteristics and a narrow process window. Producing it consistently requires close control of the material preparation, temperature and pressure. Relatively few injection moulders have the practical experience to work with it reliably. Krunon can because Jesper de Groot and his team understand the relationship between the material, the mould and the machine.

But a controlled process is not the same as an entirely predictable one. Because the material is highly sensitive to small variations in the process, roughly one in every 25 to 30 components can show a faint surface mark. The defect is subtle, but the part no longer meets the required visual standard. Until now, those components had to be found and removed by hand.

The inspection works, but it places a repetitive burden on a small team. It also happens at the end of the process: the rejected part is removed, while the information contained in that defect largely disappears with it.

From seeing a defect to acting on it

For Krunon, the question was not simply whether a camera could see the mark. The system also had to fit naturally into the existing line, be trained for this specific product and communicate with the production equipment without turning a focused quality problem into a large automation project.

That is where Limen comes in.

An iPhone positioned at the inspection point monitors each component as it moves through the line. Limen’s vision model assesses the part in real time and distinguishes an acceptable product from one showing the characteristic surface mark. When the system detects a non-conforming—or NOK—part, it sends a signal through Modbus to Krunon’s control setup. The machine or downstream handling equipment responds immediately and routes the part out of the accepted product flow.

The result is more than automated inspection. Vision becomes part of the production loop: the defect is detected, the line is informed and the rejected component is separated at the moment it appears.

Making vision the easy part

The part of the project that sounds hardest—bringing a vision system onto a running production line—turned out to be the shortest. Deciding where the camera should look took genuine deliberation: light, angle and the geometry of the part all matter. Mounting it, once that decision was made, took two minutes.

Connecting seeing to acting was nearly as quick. Krunon’s handling equipment already speaks Modbus—the same language Limen signals in—so wiring a detection to a response came down to mapping triggers to actions in the Limen app. A few clicks, configured by Krunon’s own technician. No integrator on site, no PLC code rewritten, no automation project.

The first model was trained together with Limen’s team, and it moved faster than anyone had planned for. The line had already snapshotted every part that passed; teaching came down to labelling them OK or NOK and letting the model learn the difference. Jesper’s conclusion at the end of the session: the next one, the team trains itself.

And that model is deliberately never finished. The line runs around the clock, and every cycle adds to its dataset—so week by week its sense of OK and NOK sharpens, and Krunon’s grip on the line tightens with it.

Turning quality into a process signal

The surface mark is the visible outcome of a much larger process. That makes every detection useful.

Limen gives Krunon a continuous record of OK and NOK results, including the image connected to each rejection and the way the NOK rate develops over time. Instead of relying only on periodic checks or discovering a pattern after a batch has been produced, the team can follow quality while production is running.

If isolated defects remain isolated, the process can continue with confidence. If the rejection rate begins to rise, that change becomes an early signal that something in the material, process or machine needs attention—even when the underlying issue is difficult to see from the machine alone.

Over time, this creates a new feedback loop. Krunon can review when defects occur, compare production runs and use the evidence to guide process improvements. A rejected component no longer represents only lost time or material; it becomes a data point that helps the team understand the line more deeply.

A natural next step

Krunon already combines technical craftsmanship with modern machinery, automation and careful quality control. The company’s own philosophy is built around staying flexible, communicating openly and continually improving the way products are made. For Jesper and the team, adding vision is a continuation of that approach.

Limen is designed to make industrial vision easier to adopt. It uses accessible hardware, can be deployed directly around an existing process and gives production teams clear software for monitoring quality, reviewing detections and defining what should happen when a defect appears. That makes it possible to begin with one well-defined challenge, learn in production and expand where the system creates value.

This first quality application is an important step. From there, Krunon and Limen plan to bring the same approach to other key lines: monitoring quality continuously, responding to NOK products immediately and using inspection data to improve the process behind them.

For Krunon, the aim is not to add technology for its own sake. It is to connect the knowledge already present in the team with what is happening on the line, at production speed. One component is inspected. One defect is caught. One more signal helps make the next cycle better.