Flaw finders

The ever-increasing use of automated inspection technology aims to improve product consistency and ensure fewer quality failures. Colin Carter reports.


To ensure output remains consistent and the end-user benefits from good quality, manufacturers try to ensure that processes conform to tight specifications and products are checked for conformity before they leave the factory.

There are various methods of assessing whether ‘product’ is within specification or to be rejected, and many of these rely on visual methods where either people or a sensor of some sort automatically decides whether it passes or fails.

The outside of your car is important in terms of what is an acceptable level of surface finish. No-one wants a new vehicle with even fairly small scratches or dents. Manual inspection methods, where lamps are shone on the surface to detect faults, are still common but as in many other fields, technology is taking over.

Micro-Epsilon’s ‘reflectControl’ vision system, for example, uses deflectometry to compare a reflected image with the surface of an object, such as a car panel. The degree of difference of the reflected image to the real object is a measure of the surface finish on the part.

The technology mimics the human eye and brain to produce a measure of the surface quality. Robot arms scan all areas of a vehicle’s body and the system claims to be able to pick up surface defects in as little as a minute and a half.

The company has been working closely with BMW and from the initial work, detecting imperfections in small plastic and rubber components, it has developed a system that checks a vehicle’s sides, doors and mudguard areas. Some 36 per cent of the body is scanned for defects, such as dents, inclusions in coatings, porosity and grinding and polishing marks. Constant reference to a CAD version of the ideal car enables the system to accurately pass or reject the parts.

Another example is Industrial Vision Systems’ German sister company, Neurocheck, which has developed a solution to identify surface textures and defects in small metallic automotive filter parts.

Due to the speed of production, manual examination for the two main types of defects — minor cosmetic and more serious functional problems (defined as those which will lead to non-uniform wear patterns) — is a very expensive process. NeuroCheck’s automatic inspection takes place after a sharpening procedure and is claimed to be able to deal with around 40 pieces/min.

The inspection performs two main functions: scratch recognition and sharpening errors, and the examination of the outer surface for breaks and dents or inclusions.

A NeuroCheck digital camera with a macro lens and an LED ring lighting unit comprising two separate units at different incident angles gives definition to any surface errors.

A picture of the surface structure is built and a decision made on whether the unit passes or fails inspection. The system eliminates the need for much manual inspection and the associated human errors in diagnosing problems.

And it’s not just the exterior of a vehicle where quality control is important. Carpet sets that are patchy in colour or texture are as unacceptable as scratched bodywork, an important factor when Cambridgeshire carpet maker Collins and Aikman upgraded its manufacturing process from a manual to fully automatic colour matching process.

One problem that had to be overcome was that of reading true colour, independent of the nap, which in some systems can change the perceived colour.

The company used technology supplied by one of the world’s leading sensor manufacturers, Balluff. using its BFS 26K sensors the system was taught to recognise the colour, regardless of the orientation of the carpet or the nap.

With sensors fixed at an angle to, and a set distance from, the carpets, the technology was able to decide whether the colour variations were acceptable or not. It is also claimed to have raised productivity and lowered the scrap rate of carpets.

Another high-volume product where quality failures can mean expensive downtime as well as product rejection is vinyl flooring.

Here, the manufacturing process involves producing a continuous ‘web’ of flooring, drawn between rollers in a similar fashion to the way paper is produced. However, for the product to be uniform, production of the flooring requires exact temperature control.

As a way of inspecting the product to ensure its quality, US vinyl flooring manufacturer Mannington Mills has installed a Raytek ES100 automated temperature monitoring system on its production line.

This produces a 3ft-wide strip of vinyl flooring, and prior to the installation of Raytek’s technology the quality varied greatly depending upon the final temperature the web reached.

The ES100 system uses the company’s MP50 process imager to monitor the temperature distribution of the web by analysing the infrared radiation emitted by the material.

Temperature profile data is sent direct to a PC, which translates this into a 2D thermal image, with alarms set for when the flooring is likely to be out of spec.

The system has the advantage that adjustments can be made as the temperature begins to move — a factor that is claimed to have saved much wasted scrap product since installation.

So the message is clear. As technology gets better, and manual inspection gradually becomes a thing of the past, products can only improve in quality — which is good news for the end-user.