Servomotors

The actuator, the "muscle" in a motion control system is the servomotor. Any motor (or other actuator for that matter, including hydraulic and pneumatic cylinders) can be made a servo by providing appropriate feedback and control, but motors specifically designed to be used in servo applications have a small number of distinguishing characteristics. Broadly speaking, they are of high quality and cost, and not generally encountered in the consumer marketplace. They almost always have ball bearings for durability and stability, and they will usually run well at low speeds without "cogging". The family of motors used in these discussions is the very popular permanent magnet configuration. Servomotors are available with and without gearheads to simultaneously reduce output speed and increase torque. The field housing on permanent magnet servomotors is nearly always a smooth uninterrupted cylinder. Servomotors are also nearly always "face mount", rather than having a mounting foot or bracket on their back, side, or bottom. Servomotors may be sold with or without an encoder. If the encoder is not built in to the motor, it will have to be added to the motor or some other part of the driven mechanism by the end user. Servo systems depend on feedback to tell the controller where they are and how fast they are going. A servomotor without an encoder or some other type of feedback device is just another motor. It is important to note that in the motors illustrated here, the motor, encoder, and geartrain are easily distinguished. Together they comprise the servomotor, but the motor, per se, is the part inside the field housing.

There is enormous variety in families and types of servo devices. To distinguish the types of motors discussed here from other families and types of servo motors, some adjectives can be applied which are used in catalogs and reference sources.

Permanent Magnet (sometimes written PM)
DC (Direct Current, as opposed to AC)
gearhead (or gearmotor)
brush type (as opposed to brushless)
encoder (or incremental encoder)

Most combinations of the above terms would make sense when used together to describe our motors, but one would probably not use them all together. The result would be an unwieldy mouthful. Brush-type gearhead servomotor would probably be adequate.

These three images are of the type of motor used to turn the base in the Barrel project. The motor has been partially disassembled to show its encoder and gearhead. Taken together, the photointerrupter and encoder disc comprise the shaft encoder. Notice that the lines in the encoder disc are coarse enough to be visible. This encoder disc has 100 lines, which is on the coarse end of the resolution scale for this type of motor. Since the gearhead has a 12.5 :1 ratio (every 12.5 turns of the motor shaft produces 1 turn of the output shaft), the encoder shows 12.5 x 100 = 1250 lines per revolution of the output shaft. Useable resolution is actually four times that. See the section below on quadrature encoders for why.

Gear trains tend to be modular so the manufacturer can make one motor frame and supply different gear ratios within a model family by installing the required gear train when a customer places an order. The entry point to the gear train is the motor shaft, which is cut to form the first gear in the train.

This motor is of the type used to move the camera arm lead screw in the Barrel project. It produces about 3000 RPM and has no gearhead.

Here is the shaft encoder of the same motor. Notice that the encoder lines are not separately visible in the photograph. Each line is finer than a hair and difficult to see without a lens. The orange spot is corrosion on the disc, probably from a thumbprint. Encoder discs are very thin, delicate metal made by a process called chemical machining.

The same motor as above showing the armature and field magnets. These magnets are from which the permanent magnet (PM) motor derives its name. PM motors are variable speed over a wide range, and reversible, making them appropriate for servoing. Brushless DC motors will probably replace the PM variety for reasons of reliability, but for now they cost significantly more.

Shown here is the back end of the same motor above. The function of the commutator is twofold. First, it carries power from the brushes to the armature windings. (See the BRUSH page for more discussion on carbon brushes). This armature has eight separate windings (coils of copper wire) which become strongly magnetized when current flows through them. As the armature rotates, the commutator performs its second function. As each commutator segment, in turn, comes in contact with the pair of brushes, that winding turns on, and the one previous turns off, In an 8-pole motor like this one, turning 3000 RPM, each winding is turned on and off 100 times per second. The goal of this is to place the magnetic field of the armature in the same place in space all the time, even though the armature is rotating. The pushing of the armature's magnetic field against the field of the two permanent magnets causes the armature to rotate.

Another object visible in the photo above, is a small capacitor. As the windings in the armature are being rapidly turned on an off by the brushes, they are also being turned on and off even more rapidly by the controller. When any coil of wire, also referred to as an inductor, is switched, especially by mechanical switches (which the commutator and brushes really are) electrical noise is produced. This is not a trivial matter. At all. Noise propagates as radio energy and as "hash" which gets into signal and control lines and corrupts data. Effects of noise can vary from unnoticeable to complete system failure. And it can be the very devil to find. The best thing to do to avoid noise problems is to try to stop it at the source where possible, and use proper grounding, shielding, and suppressing techniques whenever you can. One thing commonly done when switching inductors with mechanical switches is to put a small value capacitor, between 0.01 and 0.1 microfarads across the switch points, or to ground (a good, solid ground), or some combination. In this motor, the manufacturer used two capacitors, one from each brush lead and connected to a ground tab which was secured under the head of an assembly screw, thus connecting the capacitors to the case. This is a perfectly normal method of noise suppression, and usually works fine. In the case of this specific case of this motor, however, so much noise is coupled to the case that it gets into everything and makes the controller erratic. An oscilloscope probe showed noise was getting into the encoder leads and confusing the controller about what the motor was doing. Removing the capacitors completely is not an option, but bending the capacitor ground tab (see photo above) out from under the assembly screw effectively left one capacitor across the brush leads, which is another common way of suppressing motor brush noise. And it worked.

The camera rotator motor, cabled and ready for installation.

Digital shaft encoders come in two broad categories: absolute and incremental and are not necessarily sold as an internal component of a motor. Absolute encoders are rare. They tell the controller absolutely, as a multi-digit number where the shaft is in its rotation. Another way of locating the shaft in its rotation, is to count off distance relative to some known position in small increments. (see LIMITS AND HOMING for more on known positions.) The incremental encoder used in most servomotors has a disc and photointerrupter module. The disk is evenly divided along its edge with lines and spaces that are the same size. Two photodetectors in the photointerrupter module look through the disk at an infrared light source, also in the module. The detectors are staggered exactly the space of one-half the width of a line (actually a multiple of this width, but it works out to have the same effect). So only one detector at a time changes from light to dark or vice versa. The two detectors are customarily called "channel A" and "channel B".

"A leads B"

"B leads A"

Outputs from the two channels are read by the controller. The controller can see two things about each channel: levels and edges. Level means whether a channel is high or low, that is, whether that detector is light or dark. Levels are the horizontal lines in the diagrams above. Edges represent the change from low to high, or high to low. Those changes are the vertical lines in the diagrams. When A leads B, the rising edge of A happens when B is low. When B leads A, the rising edge of A happens when B is high. The edges of the lines and slots are both counted all the time. A complete cycle of a wave is all the area from one rising edge up to the next rising edge of the same wave. Looking at "A leads B", and starting with the first rising edge of A, that's one count. Then B rises; that's another count, then A falls, making a third count, and finally, in the space of one line and slot pair, A falls, for a total of four counts. So out of a 500 line encoder, each rotation of the motor shaft measures 2000 counts. Pretty efficient, no? The other thing a quadrature encoder can tell the controller is which direction the motor is turning. Let's say that "A leads B" represents the motor turning forward. A is rising while B is low (and B is falling while A is low, and so on). If the motor changes direction then the channels will look like "B leads A". Now A is rising while B is high. That's how the controller determines direction. The controller doesn't assume anything about what the motor is doing, including the direction it's turning. It is always measuring, and correcting. One caveat about wiring. Since the controller is directionally sensitive, if the channel A and B wires were to be inadvertently reversed, the motor will "run away", because it is driving the load (it thinks) toward the goal position. But the error is increasing. The controller drives the motor even harder, the error grows, and so on. This isn't too rare the first time a servo system is powered up, and if it happens, power down quickly and flip the channel A and B leads on the encoder.

Nomenclature in the shaft encoder business can be problematic. There are lines, counts, ticks, pulses, and perhaps one or two other terms to describe the physical encoder and the data that is extracted from it. When looking at an encoder, be sure to discover, either by context, datasheets, or a phone call to the vendor's sales support office whether the counts per revolution means the number of holes in the disc, or whether the term in question means the resolution when read in quadrature (counting edges, instead of holes). Definitions do seem to vary from place to place.