When casting ceramic tapes, various defects such as air or foreign particle inclusions as well as density and thickness variations can be encountered in the tape. In order to detect defects directly in-line in the film casting process, Fraunhofer IKTS has evaluated suitable measurement methods and implemented them in a demonstration tape casting plant. This allows material defects to be determined already in the manufacturing process and the casting process to be adjusted live in perspective. This would avoid high costs due to rejects for operators of tape casting plants.
At the explosion-proof start of the drying section, a camera-based inspection system was integrated to record the wet film thickness. Laser triangulation can be used to measure a tape width of up to 300 mm. These data are to be used later and correlated with dry film thickness. At the end of the dry line, another laser triangulation system, an NIR line scan camera, and a multi-channel eddy current system were used. This allows measurement of dry film thickness as well as defects with a minimum size of 80 μm and dielectric properties. The eddy current testing system can operate at measuring speeds of up to 6 m/min. The test systems were evaluated and optimized with regard to hardware and installation location as well as data generation and evaluation.
The data obtained in this way were combined on a measuring PC and visually displayed. In addition, algorithms for error detection were created. The "classic" image processing approach delivered the best result and is also more resource-efficient for a perspective-based in-line implementation of defect detection. A printer should then mark the defects directly on the tape.
With the various components, it was possible to make a digital description of the tape in order to take a first step towards the digitization of tape casting processes.
- Consulting services for plant constructors and plant operators for in-line inspection
- Installation and commissioning of measurement technology
- Creation of algorithms for pattern recognition