Motion Control Resources
- This technical feature is filed under:
Tomorrow's Integration Today
by Kristin Lewotsky, MCA Contributing Editor
Motion Control & Motor Association Posted 11/17/2014
A decade ago, integration in motion control meant smart components that could do double duty: motors that included their own drives, drives that could act as controllers, encoders that could operate in master/slave configurations, etc. The next level was to apply integration at the design level via mechatronics. Instead of using a serial approach that started with first tackling the mechanical portion of the design, then the electrical, then the controls aspects, mechatronics involved all the stakeholders upfront. Now, integration has become even more comprehensive, combining input from components across the machine to allow motion systems to adapt to changing conditions and optimize performance over the lifetime of the equipment.
There was a time machines consisted of dumb axes managed by centralized controllers. The motion controller and PLC handled path planning and I/O, while the motors and drives did what they were told. As onboard processors and memory began to appear in devices like motors, drives, and even actuators, intelligent controls capabilities began to propagate down from the plant and machine level to the component level.
Early integrated motor drives offered key advantages such as eliminating the need for centralized control cabinets and long cabling runs. Devices could be hung out on the machine itself and daisy chained together to save floor space and cost while reducing points of failure. The problem was that for the most part distributed control architectures could not handle sophisticated, highly synchronized motion. They could execute two or three axes of perpendicular travel, for example, but typically could not draw a circle. Today, that has changed. Integrated smart components operating over a shared communications bus can achieve a high degree of synchronization.
The amount of information available from components has also grown significantly. Drives, for example, used to deliver simple diagnostic information such as whether they were powered up or whether they were operating within normal ranges. Today, advanced servo drives can track heat capacity, number of cycles, loading curves, etc. That wealth of information supports diagnostics and troubleshooting from the axis level right on up to the system level.
Every machine has a characteristic spectrum of vibration frequencies generated when the moving load excites resonances in the frame. During the commissioning of a motion system, the integrator or OEM will tune the servo motors and add notch filters to suppress vibration. The problem is that machines change over the course of their lifetime as mechanical components age, loads differ, and tooling gets adjusted. Maybe a nut loses its preload or bearing begins to wear, which shifts the vibration frequencies on that axis. That, in turn, affects every other part of the machine as well as the controls. Now, the machine displays new characteristic frequencies and all of those carefully set notch filters no longer perform their intended tasks. This compromises performance and reduces throughput.
Traditionally, system diagnostics were based on historical data, often gathered manually by maintenance staff wandering around with data collectors and then analyzed off-line. They’d capture data points at intervals, analyze them, then wait for an opportunity to take the machine down so that they could rebalance it, damp vibrations, tune the servos, etc. To avoid impacting operations, these activities would be scheduled weeks in the future. Eventually, the tuning or repairs would take place, but in the meantime, the problems continued, components continued to wear prematurely, product quality began to drop, and machine speeds were reduced to avoid jams or bad product.
Vibration analysis can reveal damage early in the process, making it possible to address the root causes. The frequencies and locations of the vibration allow maintenance to not only pinpoint the source but to correct the cause, for example by lubricating a bearing or tightening a belt.
Resonant frequencies can be characterized and monitored by using the components themselves as condition monitoring devices. One approach is to add accelerometers to the load. Another method is to use the motor as a sensor in which the drive performs a frequency analysis on the encoder signal to extract vibration data. A high-resolution encoder provides feedback on the order of 16 million counts. If there’s a jam, a loose coupling, or some other mechanical problem, frequency analysis at the drive should reveal it. This not only enables early maintenance to extend the lifetime of the axis with the problem, it also prevents damage to other elements in the system.
Vibration is just one aspect of condition monitoring. Components can track changes in machine efficiency. Drives can monitor current draw to catch changes in torque demand introduced by a failing bearing, for example. All of it gets delivered to the HMI in real-time. “With real-time condition monitoring, that imbalance is seen by the operator, not by some maintenance department that comes through monthly,” says Mike Burrows, director of market development for integrated architecture systems at Rockwell Automation. “The operator can adjust the system live to remove the imbalance, or dynamic rebalancing can be calculated by the control system. Not only have you made the equipment more efficient, you’ve actually eliminated failures.”
There was a time that installing a servo was a tedious, time-consuming job that involved using an oscilloscope to iteratively tune the PID loops to optimize performance. More recently, vendors have developed software routines that could tune servo motor performance to a first approximation. They saved an enormous amount of time, although installers still found themselves doing the final iterations manually. Today, auto tuning has improved to the point that integrators and maintenance staff alike have little, if any, tuning to perform after motor installation.
In a fully integrated system, once the sensors detect changes in vibration caused by component wear, auto tuning can be used to modify system performance (see figure 1). “Auto tuning capabilities allow the machine to adapt the control loop,” says Steve Yutkowitz, principal software engineer at Siemens. “Yes, eventually maintenance needs to be done on that machine, but for a situation in which the machine can’t be taken down in the immediate term, the auto servo tuning capability allows the machine to continue to produce parts just long enough to get it to its next maintenance cycle.” Typically, the operator or maintenance staffer simply invokes auto tuning in the HMI. After repair, the machine can be characterized again and auto tuning repeated to bring it back to baseline.
The latest generation of smart motor drives takes auto tuning one step further to continuous or adaptive auto tuning. For these real time devices, some or all of the parameters of the loop control and filters adapt continuously while the machine is running in production. That means that if a problem arises, the machine can adapt to load vibration, resonance and emerging mechanical issues in real time without human intervention, preserving performance.
Adaptive auto tuning provides other benefits. Imagine a servo gantry that is picking packages of cheese off a conveyor and loading them into a case. Ideally, the system should be able to adapt to a change from 1-lb blocks of cheese to 3-lb blocks of cheese without maintenance staff having to establish a different set of tuning parameters for every new product. “If not, that leads to the case where you have a set of tuning, filter and vibration-suppression parameters that only works for one product on one machine,” says Bryan Knight, automation solutions team leader at Mitsubishi Electric. “If you have a room full of machines doing the same process and they all have unique tuning settings that are manually configured, it becomes a maintenance problem.”
Auto tuning can provide enormous benefits but it’s important to note that it can also mask larger problems. Ideally, the system should make the update while simultaneously informing the operator and maintenance so that they are able to check for underlying causes.
Throughput plays an essential role in profitability. The higher the throughput of the machine, the more quality product it can manufacture in a given timeframe. That said, there’s a point of diminishing returns past which the moving load begins to worsen vibration or excite new frequencies. Cranes and gantries and pick-and-place arms can sway during acceleration / deceleration. Liquid in containers can slosh while the vessels are being moved. The system may require extra settling time to properly execute very high-resolution movements. In the worst-case scenario, excess speed can damage to the product and/or the machine. These types of issues impose an upper limit on axis speed.
Specialized motion profiles such as S-curve profiles help mitigate the effects of acceleration and deceleration by slowing during those portions. The problem with this sort of jerk-limiting approach is that it still reduces throughput. The ideal approach is to avoid exciting resonance modes altogether.
One way to do this is to close the position loop at the drive level using the encoder. The idea is to plan the path to avoid exciting resonances over each point-to-point move of the motion (see figure 2). “We’re not doing it over the entire motion control profile, but from every point to every other point, every 250 µs,” says Bipin Sen, senior business development engineer at Bosch Rexroth. Applied to the slosh problem, this can improve throughput by as much as 50%. In the case of gantry sway, it can deliver up to a 40% improvement, which translates to 30 to 45 picks per minute.
The approach is another example of integrated components enabling distributed control, but in this case, that control enables far more sophisticated motion, orchestrated across the machine. "Because every drive has its own intelligence and path planning, each axis can look different from every other axis,” says Sen. “You can eliminate frequencies that are unique to each particular axis and load.” With this level of flexibility, the system can accommodate load-dependent frequencies, for example a top loader that carries a full pallet during one part of the cycle and returns empty.
With the emerging industrial Internet of Things, integration is poised to move beyond a machine-by-machine basis to occur on a whole line. An integrated response allows the whole line to be interrogated and controlled in real-time. If a bottling machine jams, for example, intelligent line control takes over. The robot arm loading flats slows down. The upstream modules reduce demand. Once the jam is cleared, the line returns to full speed. The line as a whole maintains productivity even though one module is off-line. It comes down to leveraging intelligence.
“We need to make not just one machine smarter but make the entire line smarter,” says Derrick Stacey, Solutions Engineer at B&R Industrial Automation. “Then we can use safe motion and integrated control to increase overall productivity. Even though you may see peaks and valleys in your output, you aren’t having pure stoppages. That can save time on rehoming and all of the associated processes that go with stopping and starting machines.”
Taking full advantage, however, requires another level of integration – integrated information. In the traditional model, a factory had a machine-based system that managed the equipment and a separate information system to evaluate productivity, overall equipment effectiveness, etc. Data harvested manually from the equipment was input into the information system and analyzed off line. This further delayed any response to issues that may have cropped up. As intelligent components increase, however, that situation has begun to change. Leveraging industrial Ethernet protocols, machines, lines, and even plant floors are increasingly integrated, sharing information in a way that brings significant benefits.
The key characteristic is that these devices and systems don’t just use data, they provide contextualized information. “It’s not just a bunch of data points that you’re trying to interpret in a database and software,” says Burrows. “That allows you to get great value out of the system quickly as opposed to having to engineer and work very hard in order to program it to do what you want.”
Intelligent drives have the ability to recognize when an axis is an idle state and make simple decisions on whether to power down. That is useful in and of itself, but the real value emerges when the axes work together. “That system-level intelligence is where you have much greater opportunity for efficiency gains and predictive maintenance and things like that,” he adds. “It’s usually not an individual device on its own that can make a decision or really understand what’s going on.
The big challenge will be supporting increased functionality without making the system too difficult to integrate or use. “We see a large trend toward adopting distributed control systems,” says Christian Fritz, principal product manager for Smart Machine Control at National Instruments. “Customers are trying to either consolidate multiple automation tasks in more powerful control systems or build interconnected control systems with smart subsystems. The questions to consider are how do we effectively connect these intelligent nodes and how can we simplify the system complexity for development and maintenance. Software is a critical component to address these challenges.”
The availability of highly granular real-time data, coupled with increasing processing power, is enabling ever greater levels of integration across the machine, the line, and the production floor. Components are both better able to collect information and better able to leverage it. “The result is improved condition monitoring, condition-based maintenance, OEE, real-time historical analysis, and security management delivered by open-source networking protocols,” says Burrows. “These are all values you get by having an integrated system.”
Thanks go to Reid Hunt, product manager at Kollmorgen, for providing useful input.