An MD&DI January 1999 Column
Designers of motion control systems today must choose from among an ever-expanding array of components and technologies.
With the continuing advances in microprocessors, programmable logic technology, power semiconductors, and software, designers of motion control systems can achieve more reliable, accurate, and sophisticated results than ever before. But to realize these improvements, the designers must be able to choose the most suitable motion control components and technologies from among a wide variety of products and suppliers.
Making these choices involves not just picking the highest-quality components, such as controllers, motors, motor drives, and feedback sensors, but also planning how all of these components and others will work together in the finished system.
A typical multiaxis x-y-z motion platform applied to a sample analyzer.
The first step for choosing motion control components is to clearly define the system requirements. Does the system need to control force, speed, position, or a combination of these? Is accuracy the most important goal or is repeatability more critical? How many motors or axes of control are required? Do multiple axes need to be coordinated, as they are in an x-y-z arrangement, or can they be treated as independent axes? A common mistake is not taking into account the unique needs of an application when choosing components.
Once the designer has taken the time to gain a complete understanding of the requirements, he or she is ready to determine which products will best meet the needs of the system.
Mechanical factors have much more effect on the electronic design of motion control systems than the electronic design has on mechanics. Product flow and throughput, human operator requirements, and maintenance issues help determine mechanics, which in turn help decide electronic and software requirements. Therefore, electronics engineers must understand the mechanics of motion control systems to achieve successful electronics designs.
Mechanical Actuators. Actuators provide a method of converting a motor's rotary motion into linear motion. Options include leadscrews, ball screws, rack and pinion, or belt/cable/chain drives. Some actuators have backlash and all have finite levels of torsional and axial stiffness, which directly affect the system's frequency response characteristics.
Linear Guides and Linear Bearings. These bearings keep a translating load constrained to a single degree of freedom. Technologies include recirculating and nonrecirculating rolling elements; sliding (friction) type; round and profile- shaped guides; and air, hydrostatic, and magnetic bearings. Important attributes are dynamic and static friction, rigidity, straightness, flatness, smoothness, load capacity, and amount of mounting surface preparation for installation.
Machine Structure. Machine structure directly affects a motion system's performance in ways that aren't always obvious. A properly designed structure will minimize the injection of external disturbances by incorporating a highly damped and compliant barrier to isolate the motion system from its environment, yet have a stiff enough and sufficiently damped structure to avoid resonance problems. A high static mass–to–reciprocating mass ratio can also prevent the motion system from exciting the structure it is mounted on.
Other Components. Other moving mechanical components include cable carriers that retain moving cables, end stops that restrict travel, shock absorbers to dissipate energy during a crash, and way covers that protect against dirt. Any components that move will affect a system's response by changing the amount of inertia, damping, friction, stiffness, or resonance.
Most electronic motion control systems consist of several key elements: a controller, motor drive, motor, and feedback sensor.
The brain of the motion control system, the motion controller is responsible for all the computational requirements of motion path planning, servo loop closure, and sequence execution. Essentially a computer dedicated to motion control, the controller is programmed by the end-user to perform desired tasks. The controller outputs a low-power motor command signal in digital or analog format to the motor drive.
Brushless servomotors with integral feedback are shown here in several frame sizes, stack lengths, and windings.
The motor drive takes this low-power signal and amplifies it to deliver the appropriate current to the motor windings. The motor produces torque proportional to its winding current and sets the load in motion.
The feedback sensor provides position or velocity information back to the controller, which determines whether to change the current requested from the drive. In a typical closed-loop system, the sequence of reading the feedback and updating the motor current is done at 1 kHz or faster. This fast sampling frequency (or servo update rate) allows the motor to smoothly follow varying velocity profiles as well as react quickly to external disturbances.
The wide variety of motion controllers available gives a designer considerable flexibility; however, of all the component choices made in designing a motion control system, the choice of the motion controller can have the most serious ramifications. The selection of most motors, drives, feedback devices, and mechanical components can often be tweaked or changed midway through the equipment design process or even much later as part of a field upgrade (albeit with great inconvenience). However, the motion controller usually involves a software component, and changing it represents much more than just a parts replacement--it also includes a learning curve and a test-and-debug process. Also, controllers can differ in their feature sets, communication protocols, and hardware interfaces.
Motion controllers have evolved considerably in the last few years, following the trend of improving price/performance ratios for microprocessors, digital signal processors, and programmable logic devices. Equipment designers faced with a build-or-buy decision usually realize quickly that the level of development time and technical expertise contained in the hardware, firmware, and software of these specialized products often rules out a competitive in-house design. Selecting a vendor focused on motion control is typically the best choice.
Software configuration utilities allow checking of status and variables for troubleshooting and diagnostics.
Servo frequency response and stability may be evaluated by today's advanced software tools without using traditional test equipment such as oscilloscopes.
For those rare applications that require unique embedded solutions, a designer may consider using specialized motion control chip sets incorporating application-specific firmware and hardware logic. Such chip sets require printed circuit board designs that include the component-level hardware and software interfaces to host microprocessors and input/output (I/O). Considering the amount of work required and the need to obtain the latest features and performance to remain competitive, most users decide to purchase an appropriate controller.
Motion controllers are typically available as bus-based cards or in stand-alone configurations. Designed to be incorporated within a host computer, bus-based cards are available in most popular formats, including ISA, PCI, compact PCI, STD, PC-104, and VME. By residing on the computer's internal expansion bus, the cards can provide communication speed and flexibility.
Unlike bus-based cards, stand-alone controllers operate without requiring installation in a computer. They have their own power supplies and enclosures. Communications take place via RS-232 serial links.
Space requirements and cost are usually about the same whether a designer chooses a stand-alone or card controller. A bus-based card typically requires an external breakout board to allow the many signals from its single high-density edge connector to be transmitted to the outside environment. In many cases, a powerful stand-alone motion controller with significant analog and digital I/O can function as the entire machine controller and eliminate the need for a computer.
Whether choosing a bus-based or stand-alone motion controller, a manufacturer needs to ensure that the controller can not only control the number of motors required, but also the types of motors. For example, some multiaxis controllers control both step and servomotors, allowing a designer to easily use both motor technologies in one system.
Additional considerations for choosing motion controllers include the ease of use and power of the programming language and setup software tools; multitasking capabilities; number of I/O points; coordinated motion requirements, such as linear and circular interpolation, electronic gearing, or camming; synchronization to internal and external events; and error-handling capabilities.
The most popular of the many types of motors available are step, permanent magnet (PM) brush, and PM brushless (Table I).
|Smoothness||Low to moderate||Good to excellent||Good to excellent|
|Speed range||0–1500 rpm typical (higher speeds possible with special drive schemes)||0–6000 rpm||0–100,000 rpm|
|Torque||High, but rapid falloff with speed||Moderate||High|
|Required feedback||None||Position or velocity||Commutation and position or velocity|
Table I. A comparison of step and permanent-magnet motor types.
Step motors are often selected simply because they can be run open loop; that is, without any feedback sensor. A step motor is designed with a number of discrete positions where the shaft will rest while producing a holding torque. The lack of feedback plus a relatively simple motor drive design makes step motors a cost-effective and reliable choice for many applications.
Digital servo drives are software configurable for torque, velocity, position, and follower modes of operation.
However, performance demands often require that designers consider either brush or brushless servomotors. Both can typically achieve higher top speeds, smoother low-speed operation, and finer resolution than step motors, but the required feedback sensor and relative complexity imparts a cost disadvantage when compared to step motors.
Brush servomotors are constructed with a wire-wound armature rotating within a magnetic field produced by a stationary PM assembly. As the motor rotates, current is sequentially applied to the appropriate armature windings by a mechanical commutator consisting of multiple brushes sliding on a ring of copper segments. Brush motor technology is quite mature and can provide very high performance. However, the mechanical commutator often becomes a limiting factor.
Brushless motors incorporate electronics to eliminate the need for mechanical commutation. Essentially an inside-out brush motor, a brushless motor consists of a PM rotor with windings distributed about the outer stationary housing. The mechanical commutator is replaced by either extra transistors and logic in the motor drive and noncontact Hall effect sensors in the motor; a commutating encoder; or, in some cases, extra software in the controller or drive.
Compared to brush motors, brushless motors typically exhibit lower rotor inertia, lower winding thermal resistance, and no sliding mechanical contacts. These motors can offer higher top speeds, higher continuous torque, and faster acceleration without the brush dust, commutator wear, contamination, arcing, and electromagnetic interference (EMI) or radio-frequency interference (RFI) of a brush assembly. Recent advances in semiconductors, specifically in power devices and microprocessors, have significantly reduced the cost disadvantage of brushless motor systems. The majority of new designs employ brushless technology.
Lately there has also been increasing market interest in linear motors. Although these motors have been around for many years, recent trends in digital drives, software commutation, performance requirements, and competition in the linear motor industry have increased the popularity of this motor type.
A linear motor can be envisioned as a rotary motor that has been sliced open and unrolled. Linear motors are available in many of the same designs as rotary motors: step, brush, brushless, and induction. Linear motors can replace conventional mechanical actuators, such as leadscrews and belt drives, and eliminate most moving and wearing of mechanical components. Their noncontact design offers advantages in speed, acceleration, cleanliness, and maintenance. Additionally, linear motors do not experience performance degradation commensurate with system length, as do most other methods of mechanical actuation.
However, linear motors typically require a more expensive feedback device than rotary motors do. They also require moving cables and often use exposed high-energy magnets, which may present safety hazards by attracting loose ferrous objects. Also, the only way to obtain greater force is to choose a larger motor. There are no other options, such as reducing gears or changing screw pitch or pulley size.
MOTOR DRIVES (AMPLIFIERS)
Motor drives, also referred to as motor amplifiers, must match the type of motor that is used, such as step, brush, or brushless.
Basic step motor drives are fairly simple devices consisting of several power transistors that sequentially energize the motor phases according to the number of digital step pulses received from the controller. More advanced microstepping versions, which are common, control the phase currents in a staircase sinusoidal fashion, permitting a motor to operate with a much greater resolution and reducing a step motor's inherent resonance problem.
Brush motor drives typically accept a ±10-V analog signal from the controller. This signal corresponds to a current or voltage command. Four power transistors arranged in an H-bridge control both the direction and magnitude of current in the motor windings.
Like a brush drive, a brushless motor drive typically accepts a ±10-V analog signal from the controller. However, just as the typical brushless motor has a three-phase winding (delta or wye), brushless drives typically contain six power transistors arranged in three half–H-bridges as well as additional logic to control their timing as a function of rotor-position feedback. This logic is often expanded upon to employ advanced commutation schemes, such as sinusoidal current control and phase advance, to extract even more performance.
Drives are either linear, which have power transistors that operate in the active region, or pulse width modulated (PWM), which have power transistors that switch between full on and full off. PWM drives are by far the most common and offer advantages in efficiency, size, and cooling requirements. Linear drives, however, produce much less EMI/RFI and may offer performance advantages when used with low-inductance motors.
For optimum performance, drives must suit the motor's characteristics. For example, a motor has a peak current limit before demagnetization occurs and a continuous current limit before it exceeds its thermal limit. The drive must be set to avoid exceeding these limits. Many drives use closed-loop feedback of winding current, and this loop must be tuned. Also, the scaling of the drive must be set so that it operates over its full dynamic range.
Considering all of these factors, it is often in the designer's best interest to procure both motor and drive from the same vendor to ensure easy setup. Many of today's drives are digitally controlled by microprocessors, which can tremendously simplify configuration through the use of Windows-based utilities.
While step motors are often run open loop, not requiring feedback sensors, all servomotors must use these components.
The most common feedback sensor is the incremental encoder. An incremental encoder produces a number of pulses proportional to the distance moved (a linear encoder) or the rotation of a shaft (a rotary encoder). These pulses are counted by the controller to determine the motor's position and direction of travel at any time. Motor speed is determined by the controller mathematically differentiating pulses in real time. Systems that require very smooth, low-speed operation are usually fitted with a high-resolution encoder that provides accurate velocity data. However, some systems incorporate an analog tachometer that produces an analog signal directly proportional to speed.
In addition to incremental encoders, other types of feedback sensors include resolvers, linear variable-differential transformers (LVDTs), potentiometers, absolute encoders, sinusoidal encoders, capacitive sensors, inductive sensors, and laser interferometers. Selection criteria include accuracy, repeatability, environmental concerns, operating temperature, cost, and physical envelope.
Because a sensor can rarely be mounted directly to the end effector, feedback is usually taken from a mechanically coupled component with the assumption of a robust connection. In actuality, how a feedback sensor is mounted can have a far greater effect on accuracy than the type of sensor that is chosen. Care must be taken, because mechanical linkages between the feedback sensor and the moving load can have backlash and compliance, introducing both static and dynamic errors into the system.
OTHER SYSTEM ELEMENTS
Motion controllers, feedback sensors, motor drives, and motors are the most significant components of motion control systems that designers must choose. But there are many other elements that contribute to the success of the design. For example, cabling must be chosen for signal integrity, flex life, bend radius, and exposure to chemicals.
Travel-limit switches should be used to avoid end-of-stroke collisions and home switches employed to establish a zero-reference position. To avoid unacceptable levels of EMI/RFI emissions and to minimize susceptibility, attention needs to be given to choosing line filters, connectors, cable shielding, and grounding methods.
Any relevant regulatory compliance issues must be addressed, and safety issues must be dealt with. For instance, designers must address what might happen in the event of component failure, power failure, or programming errors. Maintenance and diagnostic features should also be incorporated to ease troubleshooting and service. Finally, the method by which a human operator or other system controllers and computers will interface with the motion control system must be addressed.
PUTTING IT ALL TOGETHER
Although the number and variety of choices in motion control components and their vendors give designers considerable flexibility, they also increase the likelihood of compatibility problems.
A component may seem to be the ideal choice when considered alone, but is it really best when its interdependency on other components in the system is taken into account? For example, a low-inductance motor may seem suitable, but designers must consider whether the low PWM switching frequency of a digital drive could lead to excessive ripple currents, effectively derating the motor. Perhaps a linear drive or external inductors would better suit such an application.
One common problem when choosing components is that designers often assume that the products will operate as advertised even under less than ideal conditions. For example, the speed-torque envelope listed in a motor manufacturer's data sheet is generated under optimal conditions: using a drive that has a particular peak current, continuous current, switching frequency, and minimum bus voltage; cables of minimum resistance; particular motor heat sinks, and ideal ambient temperature conditions. If the motor will not operate under these conditions in an application, its published performance cannot be expected for that application.
Typically, the performance requirements for a motion system cannot be defined in terms of one component's specifications. For example, if a positioning system requires an accuracy of 0.0002 in. and an encoder with 0.0002-in. resolution is being used, several problems may still occur. All digital servo systems can hunt plus and minus one count, microprocessor controllers may experience round-off error, there are no perfect couplings between the feedback sensor and the end effector, and both dynamic and static accuracy is partly governed by system tuning.
Considering the system as a whole and keeping all the requirements of the system in mind can help designers to choose from among the increasing variety of motion control components.
Michael Backman is manager of the Motion Control Division of Thomson Industries Inc. (Port Washington, NY).