If you have an application requiring very high-speed control, signal processing or both, then an FPGA may be a better alternative than trying to push a conventional control system architecture too far and too fast. In the past, you'd be faced with arcane FPGA programming and complex integration, but now there's an easier way, as some suppliers offer FPGA functionality built in to the basic system architecture of their control systems.
FPGA stands for field-programmable gate array, and, as the name implies, an FPGA is a chip programmable by the end user. Programming an FPGA actually changes the chip's internal gate connections, allowing users to hard-code automation functionality at the chip level. This provides much more speed than any software-based solution, particularly when coupled with high-speed I/O.
Stuart McFarlane is the vice president of Viewpoint Systems, a systems integrator specializing in machine control, and his company is using FPGAs in a growing number of applications. "Using FPGA functionality is often the difference between a robust and a problematic solution," he says. "FPGAs are useful for both digital and analog elements in machine design, as they can make complex triggering, filtering and control not only possible, but relatively straightforward. Integration of specialty sensors and algorithms can be treated as special co-processors in the overall control system architecture."
"Using an FPGA to process a pulse train from a sensor directly instead of using a frequency-to-voltage converter allows for removal of spurious pulses, greater turndown ratio, specific frequency response and the ability to generate a precise clock or trigger signal as a function of the pulse train, explains McFarlane. "At the same time, one could generate a closed-loop pulse-width modulation signal to another device, eliminating several separate devices," he says. "Another example might be to process motion quality while implementing adaptive torque control using a brushless dc motor controller, all on the same device. This may be the enabling factor to solving the particular problem at all."
While FPGAs provide a solution to many tough automation problems, integration into the overall control system architecture can be problematic. One solution is to build a board-level controller with a CPU, an FPGA, I/O and other chips to support additional functionality as required. While this may be practical for machine-builder OEMs with annual product runs in the thousands, it doesn't make sense for lower-volume or custom applications or for those OEMs without extensive in-house board design capabilities.
For the vast majority of OEMs, what's needed is a way to easily integrate FPGA functionality into control system architecture, and National Instruments (NI) offers just such a solution. "Our off-the-shelf RIO platform integrates CPU, FPGA and I/O technology in a variety of form factors and performance levels," points out Eric Myers, a product manager at NI. "Our LabView system design software offers a consistent graphical programming approach across all elements of the platform; it abstracts low-level tasks and provides a wide range of IP libraries for automation tasks like motion control, machine vision, condition monitoring, prognostics and diagnostics."
An application example shows how this works in practice. Gleason creates the machines, tooling, processes, services and technologies needed to produce high-precision bevel and cylindrical gears. These gears are found in automobiles, airplanes, trucks, tractors, wind turbines, lawn mowers, power tools and other power transmission applications.
Standard control system technology simply wasn't fast enough to meet the demands of the gear production process, so an FPGA-based solution was employed. "We used an NI cRIO-9076 controller with an NI 9411 digital input module and an NI 9263 analog output module," explains McFarlane. "The digital input module monitors two digital rotational encoders, one on each spindle carrying the bevel gearset members."
Innovative analysis of these angular signals can tease out subtle variations in the average rotational speed, says McFarlane. "Coupled with sophisticated order analysis, these variations are used to modify the torque applied to the gearset at the proper angular positions and with appropriate amplitude," he explains. "Thus, the high-frequency dynamic torque components experienced by the gearset during lapping are no longer dominated by passive physics, but are actively controlled to achieve the desired results."