On the right track

A multi-product dairy company sorts various cartons using a cutting-edge vision sensor GERMAN dairy producer Hochwald Nahrungsmittel-Werke GmbH mapped out the route for worldwide success years ago with its high-quality B鋜enmarke and Glücksklee brands, and various export products. Approximately 180 people are employed at the Weiding factory in Polling, southern Germany, which turns about 200 million liters of milk per year into condensed milk, milk-powder specialties, and fresh products. Most of the condensed milk is intended for export and is filled into 80-, 170-, 340-, and 410-gram cans on five production lines. The cans are then packed in boxes in one, two, or four layers and are transported to the central box station on a common conveyor line. There the boxes must be sorted again on five delay lines before palleting. This requires clear identification of the individual containers. Initially, this identification was performed by two light barriers arranged one after the other, based on the different heights or lengths of the boxes. However, with the growing variety ?approximately 100 different variants in 18 sizes at the latest count ?it became increasingly difficult to sort the boxes by their dimensions alone. 揑n order to still be able to achieve reliable distribution of the boxes to the correct delay line without the need for manual intervention, we were looking for a system that would be easy to set up and reasonably priced and that could also detect the boxes 憃n the fly?by their color,?explains Markus Werkstetter, head of electrical engineering at Hochwald in Polling. Simple, compact, and low-cost After comparing several alternatives, Hochwald chose the Simatic MV220 vision sensor from Siemens. The innovative color-area sensor is the synthesis of highly efficient image processing technology and simple, compact sensor technology. It is designed with IP65 protection and is therefore suitable for use in the harsh industrial environment. All the elements necessary to easily test distinguishing color characteristics are accommodated in a compact housing: an image sensor with a CMOS chip for color evaluation of images with a resolution of 640 by 480 pixels and a continuously adjustable lens with a variable image field at object distances of 50 to 250 millimeters, plus white LEDs, a laser-based alignment tool, and a simple operating and indicating unit with keys, display, and LED indicators. The integrated digital inputs and outputs are fed via M12 connections, as is the power supply. The vision sensor can 搇earn?and save up to 16 different test models and reliably detect them on both stationary and moving objects automatically. Color detection without extensive programming The necessary manual alignment of the system is supported by two laser points projected into the image window. Enclosed templates assist the user with the menu-guided adjustment of the sensor to the appropriate ambient conditions. The learning of testing tasks is then reduced to the presentation of one or more objects to be detected. Test operation can begin as soon as the sensor has learned the models. Hochwald uses the 搈atching?test type at an average test density of 64 colors to detect a box color or a logo printed on the box ?16 and 2,048 colors are also possible. The particular color pattern to be detected is specified to the device by the master control via the digital inputs. 揂t the moment, detecting a single color pattern and linking the result to the following geometry scan is totally sufficient for our purposes,?says Werkstetter. From this information, the Simatic S7-300 master controller with a 315-2 DP CPU can determine the form and color clearly and accurately, and by this means determine the destination of the box. The link to the Simatic controller is achieved locally by Profibus, also from Siemens, via the digital inputs and outputs of theNIKE AIR HUARACHE