A servomechanism is a system in which motion is automatically controlled by means of comparing real physical motion to a variable representative of that motion to achieve a satisfactory unity between the two. This definition is not restricted to motion, but can be applied to any quantifiable variable.
Rudimentary servomechanisms have been employed in a variety of operations since the early days of the Industrial Revolution. Today, especially when incorporated with numerical controls, servomechanisms are the foundation of all modern industrial automation and a large number of other industries and applications.
The term ‘servo’ is often referred to as one of a number of powered actuators which manipulates an object or a component in a system, such as a valve or a motor. This common assumption is incorrect as a servomechanism is a complex system that is defined by three principles.
The Three Defining Principles of a Servomechanism
Human operators provided the means of control in the earliest industrial applications. The first steam engines, for example, were manually actuated by hand to operate valves to let either steam or cold water into and out of the cylinder. But when the valves are connected to the output of the piston, the piston will actuate the cold-water ports when fully extended and operate its steam ports when collapsed. The engine will now oscillate automatically although its operation still uncontrolled.
The first defining principle of a servomechanism is that it provides automatic control. James Watt accomplished automatic control of early steam engines by means of a centrifugal fly-ball governor driven proportionally with the output of the engine. As the velocity of the governor increases, it begins to close a throttle valve which regulates steam for the engine. Watt's mechanical governor is one of the earliest examples of a feedback device operating as a closed loop control, which is the second defining principle of a servomechanism.
The difference between desired operation of the system and the value or position received from the feedback device is known as 'error'(e). It is the purpose of all servomechanisms, and the final defining principle, to achieve an ideal error of zero. A servomechanism operating at zero error is operating at a continuous state of perfect stability. In real applications, a state of absolute stability is rarely achieved and an acceptable minimum of error must be achieved to make the application practical.
In Watt's engine, a manual steam valve serves to set a desired velocity. A second valve attached to the centrifugal governor provides a correcting feedback. The resulting regulation reduces the error between the desired velocity and the actual velocity.
Oscillation and Damping
Before the advent of modern electronics, achieving usably small levels of error in servomechanisms was challenging. In is important to understand, in the example of an engine, that the engine's purpose is to apply force in the form of torque necessary to move a load. The servomechanism's purpose is to regulate this work to be performed at a controlled and stable rate. In an engine's case this rate would be angular velocity or speed. When the θi desired velocity, or setpoint, is greater than the θo actual velocity, the result θi - θo is positive error and the controls will actuate to achieve an increase in torque of the engine.
As the available torque in an engine overcomes the resting inertia necessary to move a load, the result is acceleration. Once acceleration results in the actual velocity surpassing the desired velocity, the result is negative error and the controls will actuate to reduce the torque of the engine to achieve a deceleration the output. But this deceleration of the output will result in positive error again, once the actual velocity is once again less than the setpoint. The engine will now accelerate and decelerate indefinitely, which is known as continuous oscillation. Excessive continuous oscillation is generally unacceptable in most applications. Correcting continuous oscillation to achieve minimal error is known as damping.
Fortunately, natural damping is achieved by the load on the output of an engine in the form of an external oppositional force. Just as the mass of a steam engines load will bring the engine to a stop if the steam is shut off, that mass also dampens oscillation. The weight of the fly-balls on Watt's governor also acts as a damper against the governor’s velocity. Without the counter-effects of damping, servomechanisms would be useless but for the crudest of applications.
The decade preceding World War Two through its end was a period of rapid technological advancement . One of the more notable wartime innovations in applied servomechanics was Remote Position Control. These servomechanisms were configured to compute and predict flight paths of enemy aircraft and to position gunnery. These systems were some of the earliest applications of adaptive control in which the initial reference point of the servomechanism is unknown.
Amongst the most innovative examples of Remote Position Control were the Central Station Fire Control systems that controlled machine guns in the B-29 Superfortress. The principles remained the same: emerging technologies such as RADAR provided a variable reference and a rotary synchro transducer provided the feedback of the actual position of each of the guns axes. A second receiving synchro actuated the positioning of an axis. This pair of synchros was known as a selsyn.
In order to control any physical quantity it must be able to be measured. Today, nearly all of our controls are electronic based. There exists, for nearly any imaginable application, devices to transduce these measurable quantities into electronic signals. Once measured and fed back into a servomechanism, the output of that servo mechanism can manipulate powered controls to perform work proportional to the reference input of the servomechanism. This reference can be a constant setpoint or a variable from an external device.
A primary reason that servomechanisms are so widespread is that they control the work performed by a powerful system that can be many times greater than the input signal on the controlling servomechanism. For example, Remote Positioning Controls incorporating gyroscopes and selsyns controled valves in the hydraulic systems that positioned and stabilized cannon turrets in naval battleships as early as the 1930s.
Another example, similar to a synchro and one of the earliest methods of controlling the speed of an electric or hydraulic motor, is the tachogenerator or tachometer. This small electronic generator is coupled to a rotating axis. When the axis turns, the ‘tach’ generates a signal directly proportional to the angular velocity of the axis. In electronic controls, this signal is commonly referred to as a negative feedback signal. This negative feedback provides the error for the control.
Initially, all of these post-war electronic servomechanisms were analog and employed vacuum tubes as the building blocks of amplifiers and comparators to employ damping regulation. One of the most important aspects of an amplifier is that it can provide gain Κ.
With amplifier gain, the amount of error can be multiplied by a constant factor to create proportional control. This factor is known as proportional gain Κp.
With proportional gain, as the difference between the feedback signal and the setpoint reduces, the output of the amplifier reduces in direct proportion to the amount of error. Pout = Κp(e)(t). This results in non-linear acceleration and deceleration of the output of the servomechanism.
Another related innovation to emerge in the post-war years was numerical control. A Numerical control employs a resolver or an optical encoder to provide feedback. An encoder is a digital device and creates a pulse train that can be counted, compared and modulated with a high level of accuracy. The application of numerical control in servomechanisms resulted in systems which operated with unprecedented levels of accuracy and repeatability and revolutionized manufacturing by the 1960's.
These two features, repeatability and accuracy, are the defining characteristics of automation. The first successful demonstration of numerically controlled servomechanisms was in 1952 in the Servo Mechanisms Laboratory of the Massachusetts Institute of Technology on a machine tool. The machine demonstrated was a three axis milling machine on which each axis was driven by a leadscrew, powered by hydraulic motors and controlled by hydraulic servo valve.
A rotary resolver package attached to the leadscrew provided feedback for the numerical control system. In turn, the numerical controller then determined the error of the axis and controlled the feed rates, acceleration and deceleration for accurate and repeatable positioning control of the axis.
The hardware required to made numerical calculations for these early systems was extensive and prohibitive, requiring hundreds of vacuum tubes. But by the end of the 1950s, the rapid development of semiconductors enabled these circuits to be designed with transistors. By 1962 over a thousand machine tool applications with numerical controlled servomechanical axes were being put into U.S. production each year. The same technologies that were transforming machining also created a whole new industry: robotics.
From a servomechanical point of view, the means of controlling the axes of a machine tool and a robot are the same. In both applications throughout the 1960s and 1970s, any axis requiring power greater than 5 HP were powered hydraulically. The velocity, acc/decel rates and positioning of each axis was manipulated by the positioning of a servo valve. Servo valves operate at a constant pressure and control the rate of flow proportional to the desired velocity of the axis.
Fractional horsepower electric servo motors were developed during the war, such as those in the CSFC. But higher horsepower motors could not be controlled with the stability necessary for positioning control until the late 1970s. By then, continuous advances in field of solid-state electronics had completely transformed the electronics industry. Innovations in integrated circuitry that could perform pulse width modulation, for example, and dedicated microcontroller circuits were invented. With these new circuits, stable control of alternating current motors was now possible. The resulting motor drives or, inverters, could now control dedicated servo motors powerful enough to replace the hydraulic systems in all but the largest factory applications.
A list of obsolete, current and emerging technologies employing servomechanisms would be difficult to quantify. Servomechanisms are certainly not limited to manufacturing but found in a wide range of applications. The electromagnetic drum memory of early analog computers have long been obsoleted, yet modern hard disk drives also employ adaptive control servomechanisms. Sylsyn systems and punched tape readers have been obsoleted in favor of computerized control and servo motors. Servomechanisms are employed in flight controls, nautical controls, antenna positioning, internal combustion engines and even in Radio Controled applications for the hobbyist. New instruments, manipulators and applications are invented all the time. For every new innovation, the need to satisfactorily reduce error between desired and actual variables remains the basic function of all servomechanisms.
- West, John C. Servomechanisms English Universities Press, Warwick Square, London, 1953
- Childs, James J. Principles of Numeric Control Industrial Press, NY, NY, 1969
- Westinghouse Electric Corporation, et al. Industrial Electronics Reference Book John Wiley & Sons, NY, NY, 1948
- Harter, James Electromechanics: Principals, Concepts and Designs Prentice Hall, Engle Wood Cliffs, NJ, 1995
- Tanner, William R. Industrial Robots V.ol.1:Fundamentals, Society of Manufacturing Engineers, Dearborn, MI, 1979