Motion Control Resources
- This technical feature is filed under:
Multi-Axis Drives Simplify Machine Design
by Kristin Lewotsy, MCA Contributing Editor
Motion Control & Motor Association Posted 05/15/2014
For a long time, motion control math was simple: one motor per axis, one drive per motor, connected by cables and housed in an enclosure. More axes meant more drives and bigger cabinets, along with increased cooling costs to port away all of the heat they generated. Multi-axis drives change that paradigm by combining the drive task, either at the level of the power bus or the processor. For the right application, multi-axes drives can cut costs, increase performance, and shrink machine footprints while simplifying network topology.
For decades, the classic machine design featured PLCs to manage overall I/O and timing, coupled with a dedicated motion controller for path planning and drive commands. Increasing performance demands prompted designers to push some of the computational burden from the controller down to the drive by adding memory and processing power to the drives themselves. These "somewhat smart" drives had enough processing muscle to close the local control loops (position, velocity, torque), which saved the compute cycles of the controller for ever more sophisticated path planning enabling. Industry standard commands were +-10V analog or step/direction signals that sent a position or a velocity command to the drive. The architecture as a whole remained centralized but machine performance improved.
The advent of digital buses like CANopen and EtherCAT, coupled with more sophisticated drive firmware, enabled functionality like storing simple motion programs onboard the drive, which enabled electronic gearing from drive to drive. These truly smart drives could be installed out on the machine next to the motion axis and daisy chained together in a master-slave configuration, eliminating the need for the centralized motion controller, the enclosure, and much of the cabling. Those capabilities lowered cost and reduced points of failure. For simple applications with limited numbers of axes and no need for synchronization, distributed architectures enabled by smart drives provided designers with important new options. Still, whether configured in a distributed or centralized architecture, each drive still remained discrete.
In recent years, with the advent of multi-axes drives, that has changed. Multi-axis drives combine capabilities in a single package. They include discrete drives packaged in the same housing, discrete drives at the package or board level that share a back plane, and drives that share a processor. The levels of integration vary, but the devices offer a variety of benefits ranging from increased efficiency to improved performance. One way to understand the distinctions is to consider how they would show up in a network. Loosely integrated multi-axis drives consisting of an enclosure with several single-axis drives will appear as individual nodes, one for each box. A version in which multiple drives are integrated onto a single board might show up as a single node to the higher level network, but may rely on a subnetwork – one primary drive and multiple sub drives, which typically reduces achievable performance or synchronization. In the case of a truly integrated multi-axis drive, the drive shows up as a single network node that powers multiple motors.
One integration approach involves working with individual drives but connecting them to a common DC bus across a shared backplane. While this, strictly speaking, still follows the one motor, one drive paradigm, the integration does offer benefits. A DC drive requires a converter to change the AC input to DC power. The use of a common bus allows a single converter to produce and distribute DC power to all the drives on the bus. The result is fewer components, lower cost, and smaller form factor. It's highly scalable approach so that adding an additional axis to an existing design can be straightforward.
The shared DC bus brings more benefits than simply eliminating converters. The approach reduces energy consumption for an overall lower cost of operations. One converter means that the device only requires branch protection (fuses, breakers, etc.) for one input rather than for each drive. This reduces component count, cabling, and the cost of labor for wiring and maintenance.
With proper design, the shared-bus approach can increase machine efficiency. Consider a wind or unwind module. During braking, an axis regenerates power that would normally be dumped into the shunt resistors, where it would be dissipated as heat, wasting energy and introducing a thermal management problem. With the DC shared bus, that power can be put back on the bus to power other axes (see figure 1). "Instead of pulling more energy from the main AC coming in, you can take advantage of that regenerative energy that was put back onto the DC bus to power other motors," says Mike Schweiner, product manager at Rockwell Automation (Milwaukee, Wisconsin). "You're using power that otherwise would be wasted."
Consider an eight-axis system. At any one time, axes are regenerating power, consuming power, or sitting idle. "With shared bus architecture, you can truly optimize your system," Schweiner says. "If you really understand what your machine is doing and the cycle times of each of these drives, you can calculate the RMS current requirements and size your converter appropriately." Consider a packaging machine sealing a box. That involves quick, brief movements with sharp peak power demands but low RMS requirements. If you're not using all the power of those eight axes at the same time, you can reduce the size of the converter and save money.
When you put drives on a common DC bus, you effectively share the capacitance of all the drives. "You develop a much stiffer bus," says Schweiner. "For aggressive acceleration profiles, a stiff bus can provide the energy that the axes need, which is another benefit of this type of multi-axis design."
It's important to choose an application that can take advantage of the benefits of the shared DC bus. Although conventional drives each need one converter, the overall multi-axis design carries a certain cost. For two or three axes, approach may not be cost-effective. Conversely, it can work quite well for tens or even more than 100 axes, such as you might find on a packaging line or printing line. Be open to a hybrid approach, as well. It may make sense to run certain high-powered axes with a single drive and then use a modular system for others. Although it's not unheard of to put this type of drive out on the machine in a distributed architecture, it's more the exception than the rule and driven, as always, by the application.
Some applications require performance that challenges even the best discrete-drive solutions. Here, true multi-axes drives provide an alternative. "A drive that can control multiple axes can allow you to achieve performance levels that are almost impossible with independent single-axis distributed drives," says Jason Goerges, General Manager at ACS Motion Control (Bloomington, Minnesota). In these designs, a single processor commands a separate pulse-width modulated (PWM) power circuit for each motor (see figure 2). The controller sends a multiplexed signal to the box’s microprocessor, where it is de-multiplexed into separate signals for each individual drive. Unlike distributed architectures or conventional single-drive centralized architectures, the level of synchronization is no longer driven by the communications bus or network. Instead, the drives are synchronized to the processor clock. This enables motion control to nano-scale accuracies.
The EtherCAT protocol allows devices in the network to be synchronized to 0.1 µs, worst case. "In most industrial automation applications, a delay of 0.1 µs between one box and another is not even noticeable," says Goerges, "but in extremely demanding applications such as positioning a semiconductor wafer to within nanometers, having a multiaxis node allows you to further synchronize the axes from a single processor, where jitter is typically less than 1 ns."
Currently, the axis counts for these types of drives are low, but they can be a good fit for many Cartesian applications such as semiconductor wafer inspection, where time is money. Yield for IC fabrication is directly tied to carefully mapping and avoiding flaws on the silicon substrate, which requires accurate, and fast, inspection. Systems for current 300-mm wafers may require a standstill following error (jitter) of 1 to 10 nm, and a +/-10- to 20-nm following error at less than 1 mm/s. Requirements for the next-generation 450-mm wafers are even more challenging: jitter of less than 1 nm and under +/-10 nm following error at over 1 mm/s. Of course, everything gets harder when the system gets bigger – longer runs, more inertia. This is where a multi-axis drive really shines.
As an example, consider an XXY or H-bar gantry, commonly used in wafer inspection systems. Each side of the gantry – X and X’ -- is powered by a separate motor/encoder pair. At the center of the gantry lies the Y-axis. The gantry can be best modeled as a system with the two sides of the gantry coupled, which requires a multiple-input, multiple-output (MIMO) algorithm. In order to achieve nanoscale accuracy, the system must close the feedback loop on all three axes and the motion must be highly synchronized. Because the multi-axes drive is synchronized to the clock cycle of the microprocessor, it can perform as required. Other uses for the technology include high-speed printing lines or high-accuracy applications such as beam collimation in particle accelerators.
Depending on machine complexity, deciding on the right multi-axis drive can be a challenge. Devices are available as anything from two to even eight axes. The problem arises when your axis count doesn't fit. If you have a seven-axis application, do you use two four-axis multi drives and just leave one without I/O or do you use a four-axis multi drive and three individual drives? Not surprisingly, it comes down to the application. What are the synchronization requirements? What are the cost and space constraints? Even an identical machine platform used in two different applications may require two different implementations. From a performance standpoint, at least, it's not a difficult choice, says Nate Holmes, Product Marketing Manager for motion control at National Instruments (Austin, Texas). "Other than the cost of having unused I/O, I don't see a big downside of using a multi-axis drive compared to the other drive solutions out there."
So far, this type of multi-axis drive technology is still a niche solution and requires very powerful controllers. Still, as technology becomes ever more exacting, that's likely to change. "I wouldn't say it's prevalent yet but I suspect it will become more so," says Goerges.
What do these trends mean for smart drives, "dumb" drives, and distributed control? Certainly, it's still a reliable and popular technology. Still, just as bandwidth limitations coupled with the demands for increased performance caused intelligence to be pushed down to the drives, so faster communication may push designs back toward centralized architectures, which may be a trend all on its own. “’Dumb’ drive motion is growing at a faster rate than the distributed technology where you have your intelligence inside the drive," says Craig Dahlquist, automation group supervisor at Lenze (Uxbridge, Massachusetts), citing recent market research. Dahlquist doesn't have a vested interest in the results – the company produces both types of technology..
Smart drives, drum drives, multiaxis, discrete – today's market gives designers a wide range of choices to develop the optimal system for their application. As always, though, it comes back to the application. Understand your requirements and limitations. Once you choose an option, remember Schweiner’s advice and optimize. The ideal design is out there.