It is no secret that electronic motor controllers have become increasingly sophisticated. These advances have blurred the differences between two widely used kinds of motor controls: drives and servos. But there is a simple way to tell the difference between a motor drive and a servocontrol. A servo can power a motor to a stop, but generally a drive can’t. Put another way, the shaft of a motor powered by a servocontrol can stop at a precisely defined position. Moreover, it can accelerate and decelerate with equal precision.
Compare this behavior with that of a motor powered by a drive: The motor coasts to a stop based on the load inertia and time constants of the braking resistor. The technical explanation for this difference in behavior is that variable frequency motor drives have no bidirectional current flow. It is this bidirectional current capability which gives the servo its ability to power a motor to a precise stop.
Another key factor that has muddied the difference between servos and drives is closed-loop feedback. Servos were once the only controllers able to adjust motor torque and speed based on position or velocity of the load. But sophisticated new variable-frequency drives increasingly incorporate this feature. Consequently, it is now sometimes unclear whether the appropriate approach to solve a specific problem is a servo or a drive. The judgments needed to make this call involve fundamental issues in positioning, speed, and accuracy.
Feedback first
Most engineers assume that a positioning application must be a servo. The choice is usually obvious, but it can entail subtle aspects. Cost and performance considerations dictate that some applications use a drive instead. Operations involving large loads, as in pallet storage and retrieval systems, for example, are areas where variable-frequency drives are the best choice.
At first glance, grinding and polishing operations appear to be typical position-control applications. They really aren’t. It is true that the electronic controller must position the grinding tool to touch the part. The question is how to sense when the two touch. The only reliable indicator is the current through the grinding wheel motor, assuming it is electric, which rises as the wheel comes into closer contact with the part.
In this particular case, it turns out that the control requirement isn’t really position. It is the torque on the grinding wheel, which is proportional to the force applied to the part. When the torque is too low, the grinding tool doesn’t touch the workpiece. Torque too high means the grinding tool is digging into the part and will likely ruin it.
Cycle time can be another factor that bears on the servo-or-drive question. Consider a garage-door opener, for example. There is nothing to be gained by reducing its cycle time. It must position the door correctly, but it can move slowly with a lot of position error allowed. Limit switches serve as on/off controls and a simple constant-speed ac gearmotor does the job. Thus there is no justification for adding a costly servocontroller to this type of product.
Similarly, it is possible to design a conveyor with a simple gearmotor control. But the problem becomes more difficult when throughput rises. Accuracy and repeatability begin to suffer with increasing speed. Higher speeds start to make a more urgent argument for a servo. It is higher speed that forces a definition of the parameters dictating where a servo could best meet the need.
Four-quadrant operation refers to the concept that speed and torque can be independent. During forward velocity, it is possible to apply torque either in the direction of rotation, or against it. The latter case might come in handy for overhauling loads and in situations demanding fast deceleration.
The reverse is also true. Torque can be applied to the reverse direction of rotation, or in the forward direction to brake the load.
Unfortunately, many variable-frequency drives can’t provide four-quadrant operation. The problem is that when torque and speed are in the same direction the motor is “motoring.” When the load is “pulling” the motor (overhauling), as when an elevator lowers, the motor is generating current. In the power circuits of most drives, this generated current cannot go back into the line.
Dynamic braking in the drive takes place by shorting the stator winding and shunting the generated voltage to the motor. This indeed brakes the motor, but load deceleration is not controllable. The degree of braking is a function of the discharge circuit time constant (specifically, the value of the shunt resistor) and the load inertia.
Conventional variable-frequency drives cannot operate below 10% speed levels with any appreciable torque. Vector drives and other closed-loop drives can operate at stall with full torque for brief periods, but with the emphasis on “brief.” When operating web lines, for example, dc drives tend to work with relative ease in tension and over a wide range of speeds. High-performance ac drives are just beginning to perform with similar capability.
Recently, there has been a trend toward multiaxis power architectures that dramatically cut system costs by using a single common power supply. This arrangement reduces the per-axis cost once the first axis and power supply are installed. In such an application, with a mix of servos and variable-frequency drives, the incremental cost of a drive axis may exceed that of a servo.
There are usually substantial benefits to an all-servo system that may not be apparent until after commissioning the machinery. Because a variable-frequency drive accelerates and decelerates at a different rate than a servo, system starts and stops may produce some out-of-tolerance or scrap parts. If all axes are servocontrolled, parts produced during acceleration and deceleration will be as precise as those at running speeds. Servomotors are significantly smaller and weigh less than equivalent ac motors. Sometimes the reduced size and lower inertial mass will significantly improve application speed and throughput.
In cases where vector drives are candidates, the cost differential can be as low as 10 to 20%. Here the benefits of the servo approach can be substantial.
Dynamic response
Dynamic response is one of the toughest dimensions of performance to define. But once clearly understood, it helps clarify many ambiguities of an application.
Dynamic response defines how quickly the speed of the motor and drive (amplifier) can stabilize after a load disturbance. Each manufacturer specifies dynamic response somewhat similarly, but common wording might read, “velocity regulation of 0.5% with a 90% load change and 2 Hz to stabilize the motor speed.” Translated, this should mean that the motor speed will remain constant to 0.5% (of rated speed or set speed depending on the manufacturer) when there is a 90% load change, and the system should recover set speed within a half second. (Such a specification, by the way, indicates pretty good performance.)
Dynamic response might best be explained through a simple example. Suppose a 50-lb bag of dog food drops on an empty conveyor with a 50-lb. capacity. This changes the load instantly from 0 to 100%. The conveyor should slow momentarily and then return to the set speed. The time the conveyor takes to reestablish its set speed is a function of the drive dynamic response.
At a conveyor speed of 20 ft/min. (considered fairly slow by modern standards), the task of starting and stopping with an accuracy of 0.25 in. turns out to be more difficult than it might seem. At 20 ft/min., stopping in 0.25 in. incurs a 62-msec deceleration (corresponding to about 16 Hz). If the drive powering the application carries a 2 to 10-Hz rating, find a different solution.
Most simple conveying applications don’t need performance this critical. Requirements become more severe for coordination issues and part production. For example, consider what happens when the simple conveyor moves boxes into position for labeling. Bringing the box into position may require a full start and stop, and the accuracy of the final position imposes severe limits on system performance.
Issues relating to time and the motion come up constantly in real applications. Imagine the timing precision when cutting wallboard at 1,200 ft/min. while trying to custom-cut boards accurate to lengths within a fraction of an inch; or labeling two-liter soda bottles with the label material traveling at 800 ft/min., registering the label and cutting it to a few thousandths of an inch.
If torque output is proportional to current input in the input-output model of the motor and drive, then dynamic response resembles the current-rate-over-time (dI/dt) used to specify the operating limits of power transistors. This region of operation, also called Safe Operating Area, is a limit which cannot be exceeded safely. Transistors that do exceed it tend to fail, sometimes catastrophically.
It is not uncommon for servodrives to exhibit current rate-of-rise that is 50 to 100 times that of ordinary motor drives. It is this current in-rush that primarily sets the speed with which the servodrive can accelerate or decelerate a load.
Dynamic response is also analogous to the behavior observed when a simple ac motor switches on. How many times can the motor start and stop under starting load conditions before the system might fail? In starting a standard ac motor, the torque variation is 100% over typically no more than 10 starts per minute. In fact, if the rotor is not stationary when the restart takes place, the electrical load could be a dead short.
Many variable-frequency drives have the same problem. Suppliers offer a special feature to synchronize the variable frequency drive to the rotor at the zero crossing point of the ac sine wave.
Feedback loops
In most variable-frequency drives the only control is a speed reference which takes the form either of a speed pot or an external voltage input (velocity command).
However, a typical control block diagram for a servo contains both inner and outer feedback loops. The so-called inner loop is the means of controlling the motor velocity. The outer, or position loop, can sometimes be mathematically derived from the velocity loop. But many applications need a discrete external feedback device to get the desired position accuracy.
What complicates the position loop is that both time and position constraints must be considered simultaneously. The demand for speed also impacts the design of the controller. Products that implement high-speed motion often need a fast processor capable of real-time execution. For example, say a process line uses an 1,800-rpm motor and a 4,000-line encoder, whose quadrature frequency to the control is 126 kHz. A rule of thumb is that the processor may have to calculate the position at 10 times the speed of the feedback, or at 1.2 MHz in this case, to accurately control the load.
As the load speed increases, time available to control the motion drops proportionally. If the controller is a PLC, then its execution time enters into calculations of the position loop update time and accuracy. Suppose, for example, the PLC takes 20 msec to update the position control portion of the application. The question becomes whether 20 msec is acceptable given the speed of the moving parts.
Feedback rates also increase as speed rises. This has an impact on the selection of control hardware. Feedback rates in the megahertz range are common in most high-performance motion control equipment. But some low-cost controllers and PLC-based modules limit input frequencies to 250 kHz. High-resolution feedback devices can easily exceed these rates.
Dedicated motion controllers and PC-based motion controls frequently execute motion control algorithms in real time or near real time.
| Use a servo instead of a drive when..... •High-speed positioning is a must. • You need high dynamic response as with fast acceleration and deceleration. • The load is not large. Servos above 40 hp are rare and expensive because of the permanent-magnet material required. Also, large loads tend to require only slow dynamic response and coarse positioning. • There are stall conditions. Even drives operating closed-loop have severe limits at stall. Check the manufacturer’s ratings carefully. |