Vision systems capture quality at DaimlerChrysler
In the cylinder-head production section at DaimlerChrysler, German automobile manufacturer, robust and reliable vision sensors are being used to identify the fitted parts using a Data Matrix code. The technology ensures part traceability.
In the cylinder-head production section at DaimlerChrysler, German automobile manufacturer, robust and reliable vision sensors are being used to identify the fitted parts using a Data Matrix code. The technology ensures part traceability, an indispensable aspect of today's automotive-quality-management processes.
Every day, the employees at DaimlerChrysler's Bad Canstatt engine production plant in Stuttgart-Untertürkheim manufacture about 1600 V6 and V8 gasoline-powered engines in a two-shift operation. The production process for the 3200 cylinder heads is documented in the product history file for each vehicle. Before machining, the cylinders are marked with a Data Matrix code (DMC), making them uniquely identifiable. The code contains a part number describing the component, its year of production, the day of production, the sequential number in the relevant production series, as well as space for possible additional evaluations.
Error rate reduced
For several years, DaimlerChrysler had been using cameras to detect the code. However, their scanning error rate was sometimes more than 10%, an unacceptable level. In addition, to read the code it was necessary to blow off the residual oil from the machined surface of the cylinder heads. As a result, lubricating oil was deposited on the cameras themselves, and they had to be cleaned daily.
Due to these issues, the company's production planning department decided to test the Simatic VS130-2 vision sensor from Siemens (Elk Grove Village, IL, USA; www.siemenscnc.com) as a possible alternative to the existing camera system. "The scanning error rate with the VS130-2 was significantly lower right from the outset, even without blowing the oil off the Data Matrix code," reports Ralf Blesch, who is responsible for cylinder-head production at Bad Canstatt.
"Through close cooperation with Siemens, we succeeded in permanently cutting the scanning error rate to between 0.1% and 0.5%," production planner Bernd Hofacker adds. "And we no longer need to blow off the oil, saving us additional energy, handling, and investment costs."
A total of ten Simatic VS130-2 vision sensors have been in operation at the Bad Canstatt cylinder head production plant since mid-2005. The cylinder head is identified anew by every machining center. This is achieved by installing the camera diagonally to the scanning surface. An LED ring flash ensures adequate illumination.
"To prevent problems with oil shading and reflections, we installed the illumination unit separately from the camera and aligned it at a suitable angle," Blesch explains. At the same time, Siemens noted DaimlerChrysler's findings from the pilot phase and further developed the sensor so that the exposure control is now based on a freely selectable segment of the image rather than an integral measurement, as previously designed. This new feature, now a standard component of the VS130-2, has further improved the scanning error rate.
Data capture and analysis
The built-in CCD chip on the VS130-2 enables full-frame stationary images with an automatic exposure setting. For positive identification, the system only requires 5 pixels/dot of the DMC. This makes the system versatile. The dot size and scanning distance are defined by the selected optical system and can be varied within a broad range.
After scanning, the VS130-2 transfers the captured DMC image to the separate analysis electronics, installed close to the sensor yet slightly higher to protect the system against oil and metal chips. The image is analyzed and the evaluated data string is transferred via DP to the Sinumerik 840D machine controller (CNC) in the machining centers or to a Simatic S7-300 (PLC). The new system can be integrated into existing communication structures. The evaluated images are displayed "live" on the production-line control panels. In addition, it is possible to train new codes in situ. With the VS130-2, there is no need for complex planning or parameterization work. All the operator needs to do is configure the illumination and train the algorithms, without the need for user specifications using a code pattern.
The OP012 operator panel in the Sinumerik controller also links the vision sensors to the higher-level control system, where machine and operating data are captured and entered into the product history file. The systems communicate via Ethernet using the Transline 2000 HMI system jointly developed by DaimlerChrysler and Siemens, which provides personnel with password-protected access to stored fault images and live camera images from any connected terminal in the line.
DaimlerChrysler sees yet another advantage in the ability to configure the cameras as a standard open interface throughout the company via a Web browser at some time in the future. Direct access to the images not only enables all functions to be monitored from the operator's desk, but also allows faster commissioning of the DMC scanner, given that the positioning of the illumination unit can be monitored online and optimized as needed. In the "assess code quality" mode, the system can also detect whether the labeling is working correctly or whether the code itself contains a misalignment.