DON THOMAS
From the time Henry Ford established the first continuously-moving assembly line, engineers have faced the challenge of making precision manufacturing processes operable at ever-higher rates. Eighty-eight years later, high-speed automated manufacturing equipment, packaging equipment, and robots have greatly improved product quality and manufacturing capacity. Along with the resulting high speed and precision comes the challenge of maintaining and troubleshooting machines that perform assembly tasks with durations of only milliseconds.
As machine speeds have increased, so have the high-speed imaging motion-analysis tools used to help engineers see, measure, and understand motions that are a blur to the eye in real time. Standard video cameras operating at 30 frames per second are designed to capture and play at real-time rates, and thus are of little help in capturing fast events. In high-speed imaging motion analysis, images are recorded very quickly—usually at 500 to 2000 images per second for most industrial applications—and then played back slowly to visually expand time. Engineers can then see and measure timing errors, bounce, vibration, displacements, and many other anomalies that cause manufacturing problems.
Once an engineer is able to observe and measure the problematic motion, an opinion can be formed according to actual data; changes can then be made to correct the problem. What may have once taken hours or days to correct can often be completed in minutes. In a high-volume manufacturing environment, the results of such rapid corrections can show in greater productivity, reduced down time, reduced waste and higher quality goods. Time-saving benefits of motion analysis also are found in research, development, and test environments, resulting in greater accuracy of test data, faster confirmation of mechanical prototype operation, and faster time to market.
Camera types
High-speed cameras fall into three different categories: stand-alone camera systems, handheld cameras, and personal-computer (PC) peripheral camera systems. All are based on common digital recording architecture. Using solid-state memory as a circular recording medium, the cameras can record until told to stop by the operator or by a trigger signal from a machine fault indicator or other related source. The digital recording architecture provides continuous recording by overwriting the oldest image in memory with the newest. This recording architecture also allows the trigger signal to be set anywhere in the acquired image memory bank. By preselecting where the trigger signal (time zero) is established in memory, the user is able to window the event to capture pre- and post-event (trigger) images. The images will show what led up to the failure, the failure itself that was detected by the sensor (trigger), and post-failure status. This technology is valuable for trouble-shooting intermittent machine problems and failure testing since the camera system will record until triggered, saving the event of interest.
When deciding what type of camera to select, the first thing to consider is the application itself, starting with the level of image analysis. Specifics of the application will help determine which of two camera formats will better meet the needs of the engineer, with the two choices being quick-look or in-depth analysis. Quick-look motion analysis permits the engineer to observe motion without collecting measurement data that is required for more-critical analysis. This level of motion analysis is very common in manufacturing where engineers observe timing relationships to adjust machine performance. For example, packaging engineers use high-speed cameras to optimize machine processes by fine-tuning each machine function, minimizing dwell time between steps. As throughput demands on machines continue to increase, performance optimization, faster line changeovers, and minimal down times become essential in hitting production quality and productivity targets. In-depth analysis often is used in troubleshooting more-intricate applications where data extracted from the images is used to measure and study complex motions and related data.
Any portable high-speed camera that is self-contained and capable of operating without a computer is termed a stand-alone camera system. Such a system usually consists of a camera head that is attached to a processor box by a cable. Difficulties of working with large manufacturing equipment are eased by the ability to locate the camera up to 100 ft from the processor. A stand-alone camera system has record rates of up to 8,000 images per second and is widely used for manufacturing floor operations.
A second type of stand-alone high-speed camera—the handheld—was recently introduced by Redlake Imaging. Applications suitable for such a camera may require a simple observation of a fast mechanical event to determine the cause of a failure or to visualize motion dynamics. Resembling a larger consumer digital camera, the instrument (called the MotionMeter) is a point-and-shoot high-speed camera capable of recording up to 1000 images per second. The advantages of a handheld camera—size, portability, and lower cost—appeal to maintenance engineers and field-service technicians, who can now afford to carry high-speed cameras on service calls.
A third class of high-speed camera system—the PC peripheral—offers the same attributes as stand-alone systems but resides in the PC environment. Such a system usually differs from a stand-alone system in its intended use. Because the camera sends data to a PC, it becomes easy to collect and perform in-depth analysis of motion and data using software. Once the images are captured on a peripheral connection interface (PCI) board, they are instantly available for viewing and analysis prior to archiving on the PC hard drive or on removable media.
A tool consisting of a standard reticle allows easy calibration and measurements to be made from images, with the results displayed in a spreadsheet format. This level of analysis is typically found in research and development, test, and manufacturing applications that require precision information. For example, measuring the bounce, displacements, and timing of a complex high-speed folding operation is necessary in order to make mechanical design changes that will improve performance.
Engineers often use instrumentation data as well as high-speed images to understand machine dynamics. Electronic signals from accelerometers, strain gauges, load cells, switch positions, and voltages are a few of the data points that characterize the state of the mechanical process. The challenge of capturing high-speed images and electronic test data related to an event was greatly simplified with the introduction of a specialized software package by Xcitex (Cambridge, MA). The package simplifies image and data capture by synchronizing the high-speed image acquisition from a PCI board with high-speed data acquisition from input-output hardware. This process results in images and data that are correlated at the time of acquisition. For example, engineers doing drop-testing of products can now see the instrumentation data and images of the product deformation during impact. Images of the impact with related graphs of the impact data are instantly displayed.
Camera monitors position
In one application of high-speed imaging, a single camera recording at 2000 images per second is used to continuously monitor tightly toleranced positioning of paper in a paper-sheet feeder that is part of a printing press. Converted paper stock traveling at 800 feet per second is moved into printing position. The Xcitex MiDAS software controls capture of the images of the process and collects data from three machine sensors to monitor spindle speed, machine vibration, and air gap between sheets. In the process-monitoring system, an automatic triggering system captures, archives, and logs the video and data from each line stoppage and then exports the information across the company network to the production engineer for analysis. The software system re-arms itself without operator action.
Although primarily used to troubleshoot and test technology, the high-speed image and data technology has been adopted by manufacturing engineers for use as a production monitoring technique. By capturing high-speed images and data of a line failure, engineers can see images of the actual fault as well as the state of the machine at that time. This data can be collected and analyzed to determine the state of the machine only moments before it begins to fail. With the performance profile known, the next step is to apply a real-time monitoring and control system that can adjust the manufacturing line when it begins to enter the failure profile. Adjustments can include throttling back the line speed, increasing tension guides, or other such corrective action.
Over the next few years, high-speed cameras for motion analysis will monitor fast manufacturing processes and provide intelligent output similar to machine-vision technology. Images will be acquired and analyzed within the camera and motion performance parameters fed to the machine control system every few milliseconds. Imaging technology will thus become common on manufacturing lines.
Don Thomas is vice president of marketing at Redlake Imaging, 18450 Technology Drive, Suite A, Morgan Hill, CA 95037; e-mail: [email protected].