Motion Control Resources
Linear Actuators Combine Performance and Economy
by Kristin Lewotsky, Contributing Editor
Motion Control & Motor Association Posted 10/11/2012
Solutions like ballscrews, belts, and rack drives can provide surprisingly high accuracy and speed over tens and even hundreds of meters.
There was a time linear motors were the go-to choice for applications that required linear motion with any degree of precision. Analogous to an unrolled servo motor, a linear motor can travel over lengths limited only by the availability of sufficient magnetic material for the magnet track. The properties of rare-earth magnets initially positioned linear motors as reasonably priced high-speed solutions. Combining multiple forcers together on the same magnet track even allowed the units to handle reasonably high loads, making them an increasingly popular choice for designers. All that changed a few years ago when the price of rare-earth magnets went through the roof, however. Suddenly, engineering teams were looking for more cost-effective options. Although demanding applications like semiconductor manufacturing and metrology still need linear motors, in many cases, linear actuators like ballscrews, belt drives, and rack drives provide surprisingly effective—and economical—alternatives.
There is good reason that linear motors are considered the gold standard when it comes to positioning. They deliver high rigidity, high acceleration, high velocity, in addition to sub-micron positioning accuracy and repeatability of several dozen nanometers. “You also reduce the amount of mechanical drive train components to control the load so you have in essence no backlash,” says George Rabuzin, Director of Sales at the Drive Systems Group. As with most things in engineering, there are trade-offs. Linear motors are far more complex than linear actuators. Depending on the design, the systems can suffer from a periodic error known as cogging. Above all though, there’s the question of cost. “Everything about it is more expensive,” says Rabuzin. “Some young engineer who has linear actuator experience might say that technology makes sense but it’s extremely expensive so the reality is if you don’t actually need it, you can’t justify its cost.”
Linear actuators convert the motion of a rotary servo motor or stepper motor into linear displacement. In the case of a ball-screw actuator, the load is attached to a nut that travels up and down the screw as it turns. The balls inside the nut decrease friction as well as helping distribute the load, allowing them to handle loads as high as several hundred pounds (see figure 1). Ballscrew actuators do have limitations, however. Turning the screw at too rapidly can excite resonant modes, leading to mechanical distortion known as whip. This problem worsens as distance increases, so for practical purposes, ballscrew actuators should be limited to no more than about 3 m in length.
Within those bounds, however, they can deliver impressive performance, with positioning accuracies over 12-in of travel on the order of ±5 µm. “Obviously there’s a limit to what you can do with a ball screw, but if you can get a high-speed rotary motor and a precision-ground ball screw, you can actually do a lot of precise, pretty fast moves and it’s going to be a lot cheaper than a linear motor,” says Aaron Dietrich, marketing manager at Tolomatic Inc. (Hamel, Minnesota). Adding a linear encoder to the actuator or positioning stage for closed loop feedback, in addition to the motor feedback loop, can improve accuracy even further. “Now you can get almost like poor man’s linear motor performance,” he adds.
The planetary roller screw provides a ballscrew alternative that delivers high linear speed and hundreds to thousands of pounds of thrust over short distances of travel (typically 24 in. or less). The roller screws are analogous to the planetary gears in that they are robust enough to withstand high duty cycles (see figure 2). Preloaded to eliminate backlash and integrated with closed-loop feedback, they can also position very precisely. “Depending on the resolution of the feedback device, we could be talking about tens of microns of positioning accuracy,” says Travis Lake, Regional Sales Manager at Exlar Corp.
The classical ballscrew actuator consists of a moving screw and a stationary nut attached to the load. Taking the technology to the next level requires turning that arrangement on its ear, as several vendors have done. “I wanted a design with the main advantages of the linear motor which are high-speed, high fidelity, high stiffness, and the ability to have multiple carriages on the same driveline,” says Mike Everman, Chief Technology Officer at Bell-Everman Inc. “That leads you to a design where you have to spin the ball nut, not the ballscrew.”
In the new device, the ball nut actuator, the nut is part of a precision traveling spindle that is placed inside a hollow-core motor. When the motor rotates, it turns the nut, and the whole assembly moves up or down the stationary screw. Because only the nut turns, it exposes the motor to low rotary inertia. That allows users to make much more aggressive moves with smaller motors. Screw whip is also much less of a problem. That said, moving the nut up and down the screw can still excite resonant frequencies so there are some speed limitations, but Everman points to a system that operates at 2.5 m/s.
Rack and belt drives
For applications requiring longer distances and higher speeds, rack drives provide a practical alternative. In a rack drive, a motor-driven pinion—essentially a gear—travels along a toothed track. By adding in closed-loop feedback with an encoder and using control algorithms to compensate for wear, rack and pinion systems can be coaxed into delivering very high accuracies. Modifying the pinion and the rack also brings big benefits. To eliminate backlash, for example, designers use a pair of pinions linked together with a 0.5º to 1º axial phase mismatch. When the spring-loaded pinions are pushed into the rack so their teeth mesh, it forces them into tension so that they maintain their position during operation, eliminating hysteresis. Another improvement is to use a helical ground rack with angled teeth on the pinion rather than a simple cut soft steel rack.
Rack and pinion drives are not distance limited, but the high-performance versions do carry associated costs. “They’re not as expensive as a linear motor but they can be expensive,” says Rabuzin. “A really good-quality rack can cost more than a belt system. You have to get a matched helical ground rack as opposed to cut soft steel rack with a split pinion. You can buy a standard pinion for $80, whereas a split pinion might cost $1000 because it’s an engineered device. Sometimes what they’ll do is integrate the rack with a linear bearing. On one side you have a linear rail and on the other side is the rack, so it’s almost your guidance system as well as your driving system. But that’s going to cost you some money because it’s an engineered product.”
Rack drives provide good performance over long distances. Because they involve metal on metal, however, they are speed limited. For applications that require repetitive high-speed motion, particularly over long ranges, belt drives offer a better solution.
There was a time belt drives were considered the bottom of the heap when it came to positioning. Although they could handle distances from inches to well over 100 feet, the belts stretched and required perpetual monitoring and adjustment. They could break without warning, making the technology inappropriate for vertical applications. Most important, their accuracy and repeatability were orders of magnitude higher than those of competing technologies.
In the past five years, much of that has changed. Today’s belts are highly structured products incorporating metal and Kevlar to ensure robustness and minimize stretching. CNC machining has tightened the tolerances on both the belt teeth and the pulleys, allowing belt drives to reliably deliver accuracies of around ±75 µm. With the aid of closed loop control and compensation algorithms in the control loop, the error can be reduced even further. They can operate at speeds of around 10 m/s and move loads equivalent to thousands of pounds.
Despite the level of structure, a certain amount of give in a belt is inescapable. Achieving the kind of performance mentioned above requires essentially preloading the belt by determining the maximum force to be applied by the load and adding a safety factor of typically 10%. “You’re pre-stretching the belt past any load it will ever see,” says Dan Lutz, Lead Engineer at item North America (Akron, Ohio). “As long as you’re not exceeding that load, you’re not going to stretch the belt.” It’s important not to overdo, however. Doing so could deform the teeth can ultimately introduce backlash. “We don’t even get close to the maximum material tolerance—the most you actually stretch a normal timing belt is basically about 4 mm/m.”
When it comes to machine design, there’s a tendency to over specify components. Rather than calculate motor size and gear reduction ratios, which can admittedly be complicated, design teams often start with the motor size from an existing platform and scale it by some safety factor. The problem is that that same technique may well have been used already multiple times in a machine line over the years, resulting in a gear motor that provides far more torque than necessary. It’s essential that during setup, limit switches are in place so that the belts don’t become overstressed. “We’ve had people who have basically [started up the machine] and stretched the belt because they had 15 to 20 times the torque that was required for the gearbox,” says Lutz. “We have to install another one because it won’t maintain accuracy.”
A new style of belt drives is more like a hybrid of a traditional belt drive and a rack and pinion drive. It’s based on two belts: a static belt that is bonded teeth up to an aluminum rail and an active belt turned by a motor-driven pinion on the carriage (see figure 3).
Like a linear motor, the design is compatible with using multiple carriages on the same driveline. “It has a consistent and very high stiffness from end to end,” says Everman. “The upper dynamic belt meshes with the lower belt, which is bonded down, so it’s kind of like an über rack drive with its teeth completely covered.” The only part of the belt that can stretch is the segment in the carriage, which is only about an inch and a half long; the rest of it is under minimal tension (see video). “Instead of having a belt that is 20 feet long and springy like a rubber band, we now have the ability to belt drive a payload with a very high and constant stiffness driveline.” The system delivers backlash-free operation and has inherent mechanical damping, which allows higher tuning gains. The design minimizes belt stretch, allowing the system to achieve speeds as high as 3 m/s with unloaded unidirectional repeatability of ±5 µm and bidirectional repeatability of ±50 µm.
Video of ServoBelt in action; inside look starting at 1:00
For precision applications, linear motors provide a level of performance that’s hard to beat. For all but the most demanding uses, however, engineering teams today have a variety of alternatives from which to choose. By determining the key performance parameters in their application, whether that is load, length of travel, speed, absolute accuracy, or cost, they can find a solution that will satisfy their needs.