Ceramic Industry

A <i>CI</i> ONLINE EXCLUSIVE: Advanced Inspection

May 1, 2006
Recent advances in automatic inspection technology are helping ceramic manufacturers reduce defects in finished components and improve profitability.

Above: The CatPro system, supplied by Image Labs International, can perform multiple inspection steps on ceramic substrates as they are conveyed through production.


Defects in finished parts have long been a challenge for ceramic manufacturers. Fortunately, recent advances in the world of automatic inspection are helping companies solve this problem. New imaging sensors incorporate powerful features, and software for image analysis has become more robust and user friendly. Robotics and other material handling systems have also become smaller and easier to integrate. Together, these advances allow automatic inspection machines to tackle tasks that were once thought impossible-and are helping manufacturers improve their bottom lines.

Camera Technology Advances

Camera technology has advanced considerably in the last few years. Sensors are being made with more pixels (photo-sensitive sites) at a much lower cost. Sensors with 1.3 million pixels (1300 x 1000) are common, and new cameras are available with up to 11 million pixels. Much of this capability is being driven by the new complementary metal oxide semiconductor (CMOS) sensors, which are used in consumer digital cameras.

High-speed digital camera interface standards have virtually replaced older analog signaling. Many of these sensors use the Firewire (IEEE-1394) or USB 2.0 serial interfaces made popular by consumer multimedia devices. Higher-speed image capturing is available in Camera Link and Gigabit Ethernet interfaces. These digital technologies allow multiple cameras on one cable set or full camera control from remote locations.

Another camera technology that can be appropriate for continuously moving products is the line scan camera. This device is a one-dimensional sensor that monitors a strip across a product path and can capture images as the product moves in front of the camera. Imagine looking at a train through a slit in the fence. At any one time, you only see a vertical stripe of the train. But as the train passes by, the entire train can be seen stripe by stripe. In this manner, an image of a large moving object can be obtained.

In addition to sensor and interface improvements, some manufacturers are packing intelligence into cameras. These "smart cameras" are capable of quickly performing surprisingly complex tasks. Vision system software provides the intelligence for harnessing camera technology for modern applications, such as optical character recognition (OCR) or verification (OCV). An extension of this capability is bar code or data matrix reading.

Additional development tools have made it easier to create solutions and user interfaces for vision systems. Vision system designers understand the tradeoffs of these products and can develop solutions tailored to factory applications.

One example of how vision systems can be applied is in catalytic converter inspection. These products are often extruded ceramic substrates consisting of many small, parallel channels called cells. Catalytic converters have a complex set of manufacturing requirements with regard to cell density, blocked or open cells, exterior contour (perimeter) shape, uniform cell structure (no cracks or broken cells), exterior surface consistency and labeling. These criteria can be monitored to help manufacturers verify the quality of their production and perform 100% inspection. Using modern technology, parts can be inspected either in a stationary inspection cell or as presented on a continuously moving conveyor process.

Figure 1. A cell structure with a face crack.

Cell Density

Cell density inspection can be measured in a system with either stationary or moving parts. In a stationary inspection, the part is loaded into an inspection station, and an area camera views the cell structure. Figure 1 shows how this image might look. Knowing the area of the optical field of view, the system can accurately measure cell dimensions, wall thickness, cell density and even cell formation (squareness) problems.

If the system handles parts in a continuously moving line, a line scan camera can be used to obtain high-resolution images of the end faces, and similar analyses can be performed on these images.

Figure 2. A defective diesel particulate filter edge with some of the cells improperly open.

Blocked or Open Cells

Typical flow-through converters for gas automobiles have an open cell structure, with the catalyst on the cell walls. In these devices, it is important for the cells to be completely open. If a product is released with too many cells plugged, it will affect the exhaust flow rates and ultimately engine performance. Camera systems can be used to check parts for partially plugged areas.

In diesel particulate filters (DPFs), the cells are extruded like gas ceramics, and the ends of the channels are plugged in an alternating checkerboard pattern. By carefully plugging alternate ends of each long cell, the exhaust is forced through the porous cell walls to filter out the particulates. Figure 2 shows a part edge with some of the cells improperly open. This type of defect can also be readily identified by a vision system.

Figure 3. An exterior contour graph of one part, with the variation from the desired shape amplified by a factor of five. Small contour variations show up clearly to the vision machine.

Exterior Contour Shape

The shape of catalytic converter parts is challenging to maintain during production, because the substrates shrink as they harden. Since the substrate will eventually go into a metal enclosure before it is installed in the vehicle, the dimensions of the substrate must be carefully monitored throughout the production process. One effective approach is to determine the outer boundary of the faces, as measured from high-resolution images. Figure 3 shows an exterior contour graph of one part, with the variation from the desired shape amplified by a factor of five. The central black line is the ideal contour; the red line is the observed part contour at the faces, with the radius scaled by a factor of five to exaggerate small variations that would not be visible to the human inspector but show up clearly to a vision machine.

Cell Structure

Many abnormalities in small cell structures can be seen by using an imaging approach similar to the cell density measurement method. Algorithms, or complex calculations, can verify the shape of each cell so irregularities can be seen. In the example shown in Figure 1, a crack is evident in the cell face. Other types of defects that can be seen include irregular cell shape, broken cells and cell wall variations. A typical image size is 8000 x 8000 pixels or larger, and almost any type of analysis can be performed on the image detail.

For defects to be seen, the vision system must have the following capabilities:

  • Sufficient lighting to make the defects stand out or have high contrast with the surrounding product.
  • Sufficient resolution to see the defects. This means the system must have enough pixels to uniquely describe the defect, but also enough optical sharpness to allow the pixels to capture transitions.
  • Sufficient processing power to analyze the image. In the case of the 8 x 8K images, some algorithms take up to one second or more to reach their decisions.


Figure 4. A face code printed on an open cell structure.

OCR/OCV

Another type of inspection involves reading or checking printed codes on the product. These may take the form of human readable text characters, bar codes or two-dimensional (2-D) data codes. Figure 4 shows a typical face code printed on the open cell structure.

The inspection machine can read the code to determine what processing is needed. Alternatively, the code can be applied after inspection to indicate a complete part. In this case, the code is usually checked by the system for quality to verify it was printed correctly. Most 2-D and bar codes include additional error correcting information, so if part of the code is illegible, the entire code information can still be read. This quality check should be done so the initial printing quality is high; if the code is subsequently degraded in handling, it will still be readable.

Figure 5. An edge chip and longitudinal crack on the surface of a ceramic part.

Three-Dimensional Defects

Imaging systems are not limited to flat 2-D information. By incorporating special lighting, 3-D defect information such as gouges, cracks or height variations can be seen. Figure 5 shows a small section of a surface map of the outside of a part and reveals an edge chip and longitudinal crack in the surface. The crack in this case is difficult to see without special low-angle lighting. The system has created this image by rotating the part in front of the line scan camera to map out the entire surface of the cylinder.

An Integrated Solution

While each of these inspections is believable in a laboratory setup, implementing them in a factory environment can be challenging. Fortunately, vision systems that combine line scan cameras for high-resolution imaging of face surfaces, specialized optics and lighting to look through a part, area scan images for verification of face printing, and line scan cameras with special lighting for 3-D measurements are commercially available and can be integrated into a robust inspection system for on-line, 100% inspection of parts in production. One such inspection system that provides multiple inspection steps performed on conveyorized ceramic substrates is shown on p. xx. While the instrument incorporates many independent inspections, the results are presented to the operator in a single user interface, which displays the inspection results and archives the images and measurement data for future analysis.

With today's advanced technologies and integrated systems, ceramic manufacturers have new tools for minimizing defects and improving profitability.

For more information about vision systems, contact Image Labs International, 151 Evergreen Dr., Bozeman, MT 59715; (406) 585-7225; e-mail brian@imagelabs.com ; or visit http://www.catalyticinspection.com or http://www.imagelabs.com .

Links