Especially in precision motion control applications, the question often comes up, “What are control loops and what do they do?” To answer that, let’s look at one of your activities. When you drive a car, you are part of at least one control loop. If you control the speed without using the speedometer, you are using open-loop control (assuming you neglect the inputs from the seat-of-the-pants). By contrast, if you observe the speedometer, you have then closed the control loop. That is, you specify an action by pressing on the accelerator and you receive a precise feedback on the final results. Then you mentally note any error between your intended speed and the actual speed, and take the necessary corrective action, Figure 1.
In reality, everything you do is — in some manner — a closed loop system. Any action taken produces a response, a feedback, and an attempt to correct the action by the controlling devices. Some loops take a long time to close, or complete the loop. Others are completed in a few thousandths of a second.
When controlling motion, control loops are usually associated with the physical control of a motor shaft that positions something in relation to something else measured in distance or time. Controlling an electric motor, requires controlling three basic functions of the motor shaft:
Position — the location of one part referenced to some other physical point within the application. For example, a point on an indexing table relative to a drill head, or a point on a linear slide in relation to an inserter.
Speed — usually defined in revolutions per minute (rpm), radians/sec (rad/sec), or some other measure of velocity with respect to time.
Torque — generally defined by the electrical current flowing through the motor, which is indicative of the torque delivered by the motor.
Position control loop
A position control loop is used when you want to control the location of one device relative to the location of another device. During normal driving on a crowded highway, you use a driver’s version of a closed position loop. You establish a gap between the front of your vehicle and the back bumper of a vehicle directly in front of you, Figure 2. You attempt to keep the gap distance constant by adjusting your speed.
If the vehicle in front slows down, the gap gets smaller. To maintain the desired gap, you slow down at the same rate plus a “bit” more. The bit more makes up for the gap change that occurs before you respond to the change.
The motion control term used to describe the change in the gap is following error. It represents a difference between the actual gap and the desired gap. This distance related to time includes the time it takes to notice the change plus the time needed to take action.
Response time is quantified with “bandwidth,” the reciprocal of response time (1/response time). It establishes the limit of performance for a control loop or for a drive and system being controlled. A larger bandwidth value means a device can execute more commands per unit of time.
Speed control loop
To control position, you must control speed, as we discussed previously with closed-loop control and Figure 1. Continuing the traveling example, assume one car after another passes you, so you decide to check the accuracy of your speedometer. You use the mile markers and the car clock as a speed feedback loop. Traveling at an indicated 60 mph, you know that each mile should take one minute. If the time between each mile marker is 66 instead of 60 seconds, you have a 10% error signal, so you increase the speed shown on the speedometer by 10% to reduce the time between mile markers.
With one marker per mile, the speedloop correction time is several seconds. If there were markers at ½-mile intervals, the correction time is less. With 60 markers per mile, correction time is less yet, and you can maintain good speed regulation.
Continue on Page 2
The importance of frequent markers is also important in motor speed control. This usually requires corrections every hundredth or sometimes every few thousandths of a second. Therefore, the intervals between markers (electrical pulses) is very short. Typical digital sensing devices (encoders, which are mounted on a motor and turned by the motor shaft) have 500 or more pulses per revolution. By having a large number of pulses, which is typical when controlling electric motors, the correction time is a fraction of one motor shaft revolution.
To enable speed loops to correct for speed errors quickly, they are generally designed to operate two to three times faster than the response of the position loop. Here’s why: in the highway example, you can try to correct the speed by waiting until the next mile marker is in sight. You can then rapidly accelerate or decelerate to pass the mile marker at 60 seconds. The average speed between markers would be 60 mph; but during the mile, the car would have been at different speeds.
If you wait too long before changing the speed, the car may lack the ability to respond and keep the 60 mph average for the mile. The car’s acceleration rate is determined by the engine, load weight, and road traction.
If you replace the highway example with an industrial application, road traction is similar to belt tension on a pulley system.
A typical diagram of a speed loop (or any loop) is shown in Figure 3 . The command signal can be any kind of signal. In the highway example, the command signal is your desired speed. The summing junction (Σ) is that point where the command signal (your desire) and feedback signal (road observation) come together to form the error signal (needed change in gas pedal position).
Current control loop
To control position, you must control speed; and to control speed, you must control motor torque. However, sensing shaft torque is difficult, so we control current to an electric motor, because motor torque is proportional to current.
If you need to control the position of a machine to better than 0.100” linear position or the shaft of the motor within 5 deg, you must specify precise speed regulation.
To achieve precise speed regulation over all operating conditions, you need a motor capable of producing the torque required to move the load at the commanded speeds. We then must select a motor controller with sufficient power capacity and with a high enough bandwidth (short enough response time) so the controller does not limit motor performance.
Most drives can deliver more responsive speed changes at low speeds than at higher speeds. To understand this, consider an internal combustion engine. Here, combustible fuel is converted into pressure to move the car’s engine, but there is a physical limitation on how fast this can happen. Electric motors have similar time constraints. Motors have a built-in time delay called the motor time constant. This constant defines how long it takes a motor’s magnetic field to reach full strength for rated torque output. If the motor is commanded to turn quickly before these magnetic fields have reached full strength, the motor may not turn as expected due to decreased torque output.
How control loops relate
Figure 4 shows the relationship between these three control loops. The position loop has the slowest response. Again you can use the driving example. Driving your car down the highway, you are positioning the vehicle by moving faster or slower than the car in front of us. To control position, you must control velocity at a faster rate than you control position. From a more distant perspective, it may take us hours to position the vehicle from Cleveland to Chicago, but you must continuously control velocity in terms of miles per hour or engine revolutions per minute.
Each section of the control system has its own characteristics. Therefore, each must be considered so it doesn’t limit the preceding section. Increasing the bandwidth, for example, of the current control loop will not improve the system performance if the motor can not develop the torque or force required to achieve expected performance. Similarly, increasing the bandwidth of N1 may force N2 and N3 to higher values that may not be commercially or technically available. Determining only what is required for each loop will prevent run away system costs and over sensitive controls.
As we discussed, the torque a motor can deliver is limited when its magnetic field is at maximum strength (when the iron is saturated with magnetic lines of flux), and additional current no longer produces a stronger magnetic field. However, replacing a motor with a larger motor permits more force to be developed, but it places more stress on the other parts of the machine and requires a larger electronic controller (amplifier). This is another example of “larger is not always better.”
Finally, to get the most economical solution for the application, it is important that the user specify only the performance that is required. To specify a “wish list” of performance instead of required performance will result in higher costs and unessential features that will make the system more complicated.
Raymond C. Blatt is product manager of servo drives, and Howard G. Murphy is manager of application engineering for standard drives, Allen-Bradley Co., Mequon, Wis.