Guide to Industrial Automation -- Components and Hardware: Sensors

Home | Articles | Forum | Glossary | Books

Sensors provide input data to control systems and can take many different forms. Discrete sensors may signal the absence or presence of an object or the position of an actuator, while analog sensors may be used to sense pressure, position, or many other physical qualities that can be described numerically.

3.1 Discrete Devices

Discrete sensors are digital in nature and provide an on or off signal.

They often come with an attached cable for termination into a control cabinet, but they also have a variety of "quick disconnect" (QD) cabling options. They are typically available in 24VDC, 120VAC, or contact closure (relay) output configurations.

DC sensors use solid-state transistors as a switching method.

There are two different types depending on the nature of input device they are being interfaced with: PNP, or "sourcing," and NPN, or "sinking." A sourcing sensor provides a positive reference signal to an input, or "sources" current. This means that it must be attached to a sinking, or NPN-type, input device. The opposite is true for a sinking sensor, which is connected to a sourcing input point that provides positive-to-negative current flow into the sensor.

QD cables are standardized for sensors; most are of the Micro or Pico QD variety. Cables are available in three, four, or five wire varieties, depending on the configuration of the sensor. For large devices such as light curtains, a QD cable with more conductors is specified.

Buttons, Switches, and Contact Closures

Buttons and switches are used by machine or system operators to signal a control system to perform a task or set a state, such as automatic or manual control mode. A push button typically has only two states, on or off, and may be maintained in each position (toggle)-momentary on or momentary off. Most push buttons are mechanical in nature and have a set of electrical contacts attached to the backside of the button. The contacts may be of the normally open (NO) or normally closed (NC) configuration. Some buttons may be touch sensitive or capacitive in nature with solid-state or mechanical contacts. FIG. 4 shows schematic symbols for some of these different kinds of discrete input devices.

Selector switches may have multiple positions, each with a separate contact or group of contacts associated with it. Switches may be maintained at each position or spring return, giving the switch a "home" position or momentary effect.

Contact closures may also be controlled by the coil of a relay or solid-state signal. These are often wired to the inputs of a controller to indicate a status or condition. One by-product of using physical contacts in an electrical circuit is transients. Whenever a switch is opened or closed on an electrical circuit, a spike of voltage is created.

Current is not interrupted immediately, and a small arc usually forms between the contact points. This can have an effect on the contacts themselves, causing pitting. It can also create a spark, which can cause problems in flammable or explosive atmospheres. If a controller is looking for a single change of state on an input, it can sometimes detect multiple "bounces" of the contacts due to transients. This can cause problems if an input is used as a counter.

Physical devices such as diodes are sometimes used on coils and contacts to help minimize these effects, while software "debouncing" can be used to ensure pulses are of a specified duration before accepting them as valid input. Solid-state devices are also commonly used to minimize the effects of transients.

FIG. 4 Contact closure schematic symbols.


Photoelectric sensors, or photo-eyes, transmit and receive a light signal. The sensor changes state when the light changes from being received to not being received or vice versa. There are two conditions for the output of a photo-eye-Light On, where the output of the sensor is energized when light is detected, and Dark On, where the sensor output is energized when no light is being received. This is usually a selectable parameter with a switch or wire selection.

Photo-eyes are generally available in AC or DC varieties, although DC is much more common. DC photo-eye outputs are configured for PNP (sourcing) or NPN (sinking) outputs. There are also usually indicator lights on the body of the photo-eye for indication of power, switching status, or margin (amount of light received).

Usually, photo-eyes use an LED to generate the light signal. A lens is typically placed in front of both emitter and receiver to help amplify the light signal. The LED may be of various colors in the visible light range or in the infrared spectrum, which gives the light a longer range. Lasers are also often used for precise detection or longer-range applications. Visible LEDs are usually red, but green or blue are also used in diffuse or color-sensing applications.

When photo-eyes are placed too close together there is the possibility that light from one photo-eye's transmitter will trigger the receiver of a different eye. To reduce this possibility, manufacturers often modulate the light at different frequencies for different eyes within the same product group. Not all photo-eyes have this feature, but for those that do, there are typically series numbers or other markings that allow the eyes to be differentiated.

In addition to classifying photo-eyes by their output type, there are several physical configurations of the sensor.

Through-beam photo-eyes have a separate emitter, which transmits the light, and receiver, which receives the signal and controls the output state. FIG. 5 shows this configuration. Because the sensor has to have two separate cables terminated for power and signal and it has two physical pieces, it is more expensive in both installation time and in hardware cost than other configurations. The through beam photo-eye has a longer range and is considered to be most reliable when detecting the absence and presence of objects.

Retroreflective photo-eyes use a reflective tape or a plastic reflector to bounce the light off back into its receiver, as shown in FIG. 6. The emitter and receiver are both built into the same housing and use a common power wire, which reduces cost over the through-beam type, but the range is shorter.


FIG. 5 Through-beam photo-eye. Sensed Object Emitter Receiver


To minimize interference from the light bouncing off other reflective surfaces, a polarized signal is often used. A corner cube type reflector shifts the light 90° before it is received, and only light out of phase with the transmitted light is accepted as a signal. This allows the gain of the input circuit to be set at a higher level since the sensor will ignore signals from highly reflective objects that are not out of phase.


FIG. 6 Retroreflective photo-eye. Sensed Object; Polarized Reflector

FIG. 7 Diffuse photo-eye. Sensed Object


Diffuse reflective photo-eyes use the object to be detected as a target, as shown in FIG. 7. Light is transmitted from the emitter and received similar to the other configurations; however, the light received indicates the presence of a target rather than its absence, as with the through-beam and retroreflective configurations. Diffuse reflective photo-eyes are not the ideal sensor to use for simple absence or presence of an object since the amount of light received is affected by the reflectivity and color of the target. This property can, however, be used to an advantage when using the sensor to differentiate between colors. A red LED signal will reflect much more strongly from a red object than from a green object and vice versa. Techniques using red, green, blue, yellow, and white LEDs as light sources are often used in color detection photo-eyes.

There are also various additional physical configurations for photo-eyes. The amplifier that powers the transmitting and receiving LEDs may be self-contained with lenses or it may have a head for attachment of fiber optics; see FIG. 8). Fibers may be made of plastic or glass. Plastic fibers usually have an opaque plastic cladding around the outside the clear inner fiber. This protects the fiber and acts as a waveguide for the light to reflect back into the core, keeping the light confined. Glass fibers typically can be used in longer-distance applications; however, they are more fragile than plastic fibers. Plastic fibers can be bent in a tighter radius than glass fibers.

Lenses and right-angle tips are often used for fiber optics. The mounting end of fiber-optic cables may also be threaded for use with nuts. Plastic fibers may be cut to length using a small cutter that is often available from the photo-eye manufacturer. Glass fibers usually have factory ends on them to prevent damage and must be purchased in the required length.

FIG. 8 Fiber-optic amplifiers.

Proximity Switches

Proximity switches are used to detect the position of an object. While photo-eyes are also sometimes referred to as proximity switches, here we are discussing inductive, capacitive, limit switch, and Hall effect types. Proximity switches are often called proxes for brevity.

Inductive proximity switches are used to detect metal objects.

A coil of thin wire is energized with a weak current that is connected to an oscillating circuit. When a large enough piece of metal enters the field created by the current flowing through the coil, the oscillator is stalled and a discrete signal is generated, signaling the presence of an object. The type of metal being detected strongly influences the range of an inductive proximity switch. Metals, such as steel containing iron, make the best target, while aluminum reduces the sensing range by about 60 percent.

In addition to the PNP or NPN output designation, inductive proximity switches are categorized as being shielded or unshielded.

Shielded proxes have a metal housing all the way up to the sensing face of the switch. This reduces the range but allows the sensor to be mounted flush with a metal surface without detecting the metal to the side. Unshielded proxes have a longer range since the field extends out from the sides of the prox, but are more susceptible to damage or interference.

Inductive proximity switches are available in a threaded barrel style, a flat surface mount, or various other configurations. Barrel proxes typically have a metal housing with a plastic-covered sensing surface, but they can be made entirely of stainless steel for ruggedness.

FIG. 9 shows a shielded barrel proximity switch sensing a metal target block. Note that it is threaded completely into its mounting block, indicating that it is shielded.

Capacitive proximity switches use a capacitive sensing surface that discharges when an object is placed close to it. As such, it can be used to detect nonmetal solid or liquid objects. A common use of capacitive proxes is to detect a liquid through the sides of a plastic or fiberglass vessel. As long as the vessel walls are fairly thin, the prox can be set to detect the difference in mass between an empty and full vessel. Capacitive proxes are also sometimes used as operator push buttons for ergonomic purposes since they take no pressure to activate, unlike mechanical push buttons.

Like the inductive proximity switch, capacitive proxes have a very short detection range. They are usually larger than their inductive cousins.

FIG. 9 Inductive barrel proximity switch.

Hall effect sensors create a voltage difference based on the amount of magnetic field they sense. Hence, they are used to sense magnetic objects, such as a magnet moving with a piston inside a cylinder body.

Any current-carrying conductor creates a slight magnetic field of its own, transverse to the direction of current flow. With a known magnetic field, its distance from the Hall plate can be determined. This makes Hall effect sensors a great choice for end-of-cylinder stroke sensors since they can sense the magnet through a metal (typically aluminum) body. This is illustrated in FIG. 10; inside the aluminum body of the Festo cylinder is a magnet attached to the cylinder's piston head. These electromagnetic transducers are used for proximity switching, positioning, speed detection, and current-sensing applications.

Hall sensors and inductive proximity switches are the most common sensors used in detecting cylinder or actuator position. What is the major difference between a Hall switch and an inductive prox? Essentially, a Hall effect sensor can sense a magnetic field, whereas an inductive sensor creates its own magnetic field.

FIG. 10 Hall effect sensor on pneumatic cylinder.

Limit switches are mechanically activated devices that open or close electrical contacts when an object contacts it. There are wide variety of configurations, sizes, and degrees of ruggedness for limit switches. Roller limit switches have a metal or plastic roller that allows an object, such as a cam, to slide along the contact point. Lever arm and "whisker"-style switches extend the reach of the switch.

FIG. 11 shows several roller and lever arm switches mounted on a rotating display at a trade show; as the arm rotates in the center of the display, the lever arms move actuating the internal switch contacts.

Precision limit switches are used to precisely control the actuation point of a switch for positioning or measurement. They are typically plunger-type switches with a very short stroke.

FIG. 11 Roller limit switches.

3.2 Analog

Analog sensors produce an output that is proportional to a measured property. There can often be offsets and linear errors associated with analog sensors that must be taken into account when using the resulting measurements, and calibration to a known standard is often required. Analog sensors are often known as transducers.

Analog sensors are often used in automated and manual gauging.

Special-purpose machines are often built around a specific type of gauge or group of gauging devices as a test station.

Pressure, Force, Flow, and Torque Sensing

Force can be measured using a variety of devices. One common element in measuring the amount of force exerted on an object is a strain gauge. Because strain gauge wires are fragile and difficult to handle, they are typically attached by an adhesive, such as superglue, to an insulating flexible backing, such as plastic. As stress is applied to the mounted strain gauge, the foil is deformed, causing its electrical resistance to change. The resulting resistance change-usually measured using a Wheatstone bridge circuit-is related to the strain by the "gauge factor," or the ratio of electrical resistance to mechanical strain, factoring in temperature, which also plays a small role. This circuit is illustrated in FIG. 12.

A strain gauge can be configured in a variety of physical packages to measure force or weight or to determine vibration and acceleration.

For small strain measurements, semiconductor strain gauges known as piezoresistors are preferred because they usually have larger gauge factors-or changes in resistance over a range of strain- than a foil gauge, thus allowing for more accuracy. Downfalls of semiconductor gauges include the higher cost, fragility, and greater sensitivity to temperature changes.

A load cell is a transducer that converts an input mechanical force into a measurable electrical output signal. When weight, or load, is applied, the strain gauge deforms, changing the electrical resistance of the gauges in proportion to the load. The strain gauge measures the deformation, or strain, as an electrical signal as current is passed through the gauge element.

FIG. 12 Wheatstone bridge strain gauge circuit.

In order to ensure maximum sensitivity and account for temperature changes, the typical load cell consists of four strain gauges in a Wheatstone bridge configuration. Load cells with one strain gauge, a quarter bridge, or two strain gauges, a half bridge, are also available.

Because of the small amount of electrical signal output produced, typically only in the range of a few millivolts, amplification by an instrumentation amplifier is required. The amplified output is then fed into an algorithm to calculate and scale the force applied to the transducer.

Pressure may be measured by using a piezoresistive strain gauge as described previously. The gauge is attached to a force collection element, such as a diaphragm, piston, or bellows, and deflection is measured proportional to the change in pressure. Absolute, differential, gauge, and vacuum pressures can be measured using this method. A diaphragm with a pressure cavity can be used to form a variable capacitor that is effective in detecting low pressure changes.

Displacement of the diaphragm can also be measured inductively by measuring deflection of a magnet, use of a linear variable differential transducer (LVDT), or detection of an induced eddy current. These methods are known as electromagnetic pressure sensing. Optical methods can also be used by detecting changes in light transmission through an optical fiber as it is deformed.

Flow of liquids or gases can be measured in a number of ways. A rotary potentiometer (resistive element) is often used when attached to a vane that turns in the fluid or gas. Other flow sensors are based on devices that measure the transfer of heat caused by the moving medium. This principle is common when using microsensors to measure flow. Flow meters are related to devices called velocimeters that measure velocity of fluids flowing through them. Laser-based interferometry is often used for airflow measurement, but for liquids, it is often easier to use a physical deformation of some kind to measure the flow. Another approach is Doppler-based methods for flow measurement. Hall effect sensors may be used on a flapper valve, or vane, to sense the position of the vane, as displaced by fluid flow.

Detection of flow and pressure along with the measurement of valve positions in the process industry is known as instrumentation.

Analyzers that detect properties such as acidity, viscosity, or density can also be included in this group. Outputs from instrumentation are often connected to transmitters, which convert signals into standard ranges such as 4 to 20 mA or 0 to 10 V signals.

Commonly, torque sensors or torque transducers use strain gauges applied to a rotating shaft or axle. Because of the relative movement of the shaft a noncontact means to power the strain gauge bridge is necessary, as well as a means to receive the signal from the rotating shaft. This can be accomplished using slip rings, wireless telemetry, or rotary transformers. Newer types of torque transducers add conditioning electronics and an ADC to the rotating shaft (rotor). Stator electronics then read the digital signals and convert those signals to a high-level analog output signal, such as +/-10VDC

Color and Reflectivity

As described in the digital sensor section 3.1, various colors of LED light reflect from different colored materials with varying intensity.

This property can be used to sample the amount of light returning to a receiver and determine color. Combinations of reflected red, green, and blue light can be analyzed to determine shades and hues to separate items of different color properties. Despite this being listed in this current section, color sensors are often "taught" a color and an output is then switched if the color is detected.

For more accurate determination of color, a CCD is used to capture a colored region. CCDs react to photons, and when a filter called a Bayer Mask is placed over the CCD, it becomes a color-sensitive device. Red, blue, and green again are the operative colors for color CCDs. CCDs are also used to create black-and-white images that can be converted to a scale for intensity measurement.


LVDTs are a type of electrical sensor used for measuring linear displacement. The transformer-like device has three solenoidal coils placed end to end around a tube. The center coil is the primary, and the two outer coils are the secondaries. A cylindrical ferromagnetic core, attached to the object whose position is to be measured, slides along the axis of the tube.

An alternating current is driven through the primary, causing a voltage to be induced in each secondary proportional to its mutual inductance with the primary. The frequency is usually in the range 1 to 10 kHz.

As the core moves, these mutual inductances change, causing the voltages induced in the secondaries to also change. The coils are connected in reverse series, so that the output voltage is the difference (hence "differential") between the two secondary voltages. When the core is in its central position, equidistant between the two secondaries, equal but opposite voltages are induced in these two coils, so the output voltage is 0.

When the core is displaced in one direction, the voltage in one coil increases as the other decreases. This causes the output voltage to increase from 0 to a maximum. The output voltage is in phase with the primary voltage. When the core moves in the other direction, the output voltage also increases from 0 to a maximum, but its phase is opposite to that of the primary. The magnitude of the output voltage is proportional to the distance moved by the core (up to its limit of travel), which is why the device is described as "linear." The phase of the voltage indicates the direction of the displacement. FIG. 13 illustrates the internal arrangement of an LVDT.

Because the sliding core does not touch the inside of the tube, it can move without friction, making the LVDT a highly reliable device.

The absence of any sliding or rotating contacts allows the LVDT to be completely sealed against the environment.

LVDTs are commonly used for position feedback in servomechanisms and for automated measurement in machine tools and many other industrial and scientific applications.

FIG. 13 LVDT.; Voltage Source; Output


Ultrasonic sensors transmit sound pulses at a high frequency and evaluate the echo received back from the sensor. Sensors calculate the time interval between sending the signal and receiving the echo to determine the distance to an object.

Ultrasonic sensors are often used for distance measurement but are common in liquid and tank level applications. The technology is limited by the shapes of surfaces and the density or consistency of the material; for example, foam on the surface of a fluid in a tank could distort a reading.

Because of the effect of the air medium on the speed of sound on the signal, ultrasonic sensors are not particularly repeatable or precise; however, they can be used over fairly long distances and tend to have a smoothing or averaging effect when measuring irregular or moving surfaces.

Distance and Dimensions

Photoelectric sensors, proximity switches, LVDTs, ultrasonics, and encoders can all be used to measure distance and dimensions.

With optical sensors such as photo-eyes, the property of reflectivity can be used to determine the relative distance of an object from the sensor. As an object moves farther away, the amount of light received by the sensor becomes less. The color of the target also has an effect on the received signal, however, so optical distance measurement is best used on a consistent target. Laser-based devices can be used similarly to LED photoelectrics with longer range and less dependence on color.

Rows of LEDs or lasers that can measure dimensions based on the number of beams broken or the amount of light received and CCD based devices that can measure distances accurately are other usable optical methods. These methods do not depend on reflectivity and can be used to measure nearly any object as long as it is not too large.

Techniques using precision tooling and physical contact with the target such as LVDTs are also commonly used where contact with the target is feasible.

For longer-distance strokes LVDTs may not offer enough accuracy for an application. An excellent option for measuring distance is time based magnetostrictive position sensing. Magnetostriction uses a ferromagnetic measuring element known as a waveguide, along with a movable position magnet. The magnet generates a direct-axis magnetic field within the waveguide. When a current or "interrogation pulse" is passed through the waveguide, a second magnetic field is created radially around the guide. The interaction between the two fields generates a strain pulse that travels at a constant speed from its point of generation at the magnet (the measuring point) to the end of the waveguide. A sensor detects the pulse and generates a highly accurate positional reading through the electronics of a high-speed counter.

Magnetostrictive sensors provide an absolute position reading that never needs recalibration or homing after a power loss. This can be a significant advantage over using LVDTs and encoders. The only limitation of this technology is that it cannot be used for short-distance dimensional measurements; the minimum range is about 25 mm.

A well-known manufacturer and the first to develop products using this technology is MTS Systems, developer of Temposonics sensors.

Thermocouples and Temperature Sensing

There are a variety of devices that can be used to measure temperature.

One of the most commonly used is the thermocouple. A thermocouple is a junction between two different metals that produces a voltage related to a temperature difference. They are inexpensive and interchangeable, have standard connectors, and can measure a wide range of temperatures. The main limitation when using a thermocouple is its accuracy; system errors of less than one kelvin (K) can be difficult to achieve.

Any circuit made of dissimilar metals will produce a temperature related difference of voltage. Thermocouples for measurement of temperature are made of specific alloys, which in combination have a predictable and repeatable relationship between temperature and voltage. This relationship is not linear, however, and the voltage curve must be linearized in the input instrument. Temperature loop controllers contain linearization algorithms for the most common types of thermocouples. Selection of the thermocouple type can be made by setting dipswitches or software parameters.

Different alloys are used for different temperature ranges and to resist corrosion. Where the measurement point is far from the measuring instrument, the intermediate connection can be made by extension wires, which are less costly than the materials used to make the sensor itself. Thermocouples are standardized against a reference temperature of 0°C; instruments then use electronic methods of cold junction compensation to adjust for varying temperature at the instrument terminals. Electronic instruments also compensate for the varying characteristics of the thermocouple within the linearization algorithm and help improve the precision and accuracy of measurements. An example of a thermocouple is shown in FIG. 14; the probe at the bottom is the sensing element inside a protective

"well," while the container at the top is the head, which contains the termination points for the thermocouple wire.

FIG. 14 Thermocouple

Thermocouples are widely used in science and industry; a few applications would include temperature measurement for kilns and injection molding of plastics, measurement of exhaust temperature of gas turbines or diesel engines, ovens, and many other industrial processes.

The most common type of thermocouple in use is the K thermocouple (chromel-alumel). This covers temperature ranges from -200°C to 1350°C. It is inexpensive and available in a variety of styles. J thermocouples (iron-constantan) are less popular than K because of their lower usable temperature range of -40°C to 750°C. Other types of thermocouple include E, N, B, R, S, T, C, M, and chromel-gold/iron. A table for the different types of thermocouples is located in App. F.

One note on thermocouple polarity: there is a polarity labeled + and - for connection to input terminals. Counter to the common thought that the red wire is positive in many DC circuits, red is always the negative lead for thermocouples. Not every thermocouple pair has a red wire, but when using the American National Standards Institute (ANSI) color code, the red lead will always be negative.

Thermistors are a type of resistor with resistance proportional to its temperature. Thermistors are usually made of a ceramic or polymer material. They have a high precision over a limited temperature range, typically -90°C to 130°C.

Resistance temperature detectors (RTDs) also change resistance proportionally with temperature, but are made of pure metals. They are useful over a wider temperature range than thermistors but are less accurate. RTDs and thermistors may both be used with standard analog inputs and an excitation voltage because of their linearity, unlike thermocouples, which must use a special input to linearize the signal.

Infrared thermocouples or infrared temperature sensors are used as noncontact methods of sensing temperature. They use the thermal emission from the target to scale temperature to a readable value.

They are usually manufactured to be used in place of a J- or K-type thermocouple for convenience.

3.3 Special-Purpose Sensors

There are various sensing devices that do not meet the criteria of being either digital or analog as they use elements of both.

Encoders and Resolvers

An encoder is a type of transducer that senses position or orientation, usually for use as a reference or active feedback to control position.

Encoders may be rotary or linear, optical or magnetic, analog or digital, depending on the type of application.

Rotary optical encoders use a rotating glass or metal disk with slots or perforations along the circumference. An LED emits light along the path of the slots creating a train of pulses that can be used to count or measure distance. FIG. 15 shows an open-style encoder.

Encoders may be open as in the illustration but are usually housed in a rugged metal housing with bearings and shaft for attaching to a motor with a coupling. The housing is generally watertight and may have either a potted cable or a multi-pin connector for termination.

By placing two sets of slots 90° out of phase with each other, the direction of rotation can also be determined. These two signals are known as the A and B pulses of the encoder. The inverse of the encoder A and B pulses are also often used, commonly known as A not and B not. A single slot is also placed along the circumference and is known as the Index or Z pulse; this is used for identifying the home or reference position of the encoder or device attached to it. This offset A and B pulse configuration is known as quadrature. This configuration is illustrated in FIG. 16.


FIG. 15 Encoder. LED/Optical Sensor Rotation Axis Optical Disk Optical/Encoder; Components


Encoders are often of the multiturn variety; that is, they will turn multiple times, providing a count much higher than the number of slots on the disk. This means that the high-speed counter or servo module that the encoder is connected to must keep track of the number of turns or total count of the pulses. If the power is removed from the counter or control system, it is necessary to "home" the axis or device attached to the encoder, typically to an external "home" sensor and the index pulse.

Absolute encoders use a parallel signal to provide a binary count of the position of the encoder. The signal from an absolute encoder gives an unambiguous position within its travel range without requiring knowledge of any previous position. This means that the encoder will have a fixed range or number of turns. Absolute encoders are often used when a system or axis must retain its position even when powered off and moved. Absolute and incremental encoders provide the same accuracy, but the absolute encoder is more robust to interruptions in transducer signal.

Resolvers are also used to detect rotary position and velocity. A resolver is best described as a rotating electrical transformer that provides a sinusoidal output, which is then converted to a digital value representing position. A common type is the brushless transmitter resolver. This type of resolver is similar to an electric motor in that it has a rotor and stator. The stator portion is made up of an exciter winding and two two-phase windings, labeled X and Y. These windings are located at 90° angles to each other. When an alternating current is induced into the exciter winding, the signal is transferred into the rotor windings and then back into the X and Y windings. This provides a sine and cosine feedback current, which is measured to determine the angle of the rotor. On one full revolution, the two feedback signals repeat their waveforms. Because resolvers are analog, they effectively have infinite resolution. FIG. 17 shows a cutaway resolver in an industrial housing.

FIG. 16 Incremental encoder disk quadrature track patterns. Clockwise (A leads B) Counterclockwise (B leads A)

Vision Systems

Also known as machine vision, vision systems apply microprocessor or computer-based vision processing to inspection, measurement, and guidance tasks. While computer vision is mostly focused on image processing, machine vision may also require digital I/O devices to control other manufacturing equipment. Machine vision is used in the inspection of manufactured goods such as semiconductor chips, automotive parts, food, and pharmaceuticals. It is also often used as a guidance method for robots.

Just as human inspectors working on assembly lines visually inspect parts to judge the quality of workmanship, so machine vision systems use smart cameras or digital cameras with computer-based image-processing software to perform similar inspections. Individual characteristics of parts can be assigned parameters to judge on a pass fail basis for absence/presence, measurement tolerances, color, surface defects, and a number of other visually determined aspects.

Machine vision systems are also programmed to perform simpler tasks, such as counting objects on a conveyor, reading serial numbers, and measuring parts. Manufacturers favor machine vision systems for cases that require high-speed, high-magnification, 24-hour operation, and/or repeatability of measurements. Vision systems are more consistent than human beings because of distraction, illness, and other physical or mental limitations; humans are better at making finer qualitative judgments and adapting to new undefined defects.

Computers do not "see" in the same way human beings do.

FIG. 17 Resolver.

Cameras are not equivalent to human optics. Computing devices see by examining the individual pixels of images, processing them, and attempting to develop conclusions with the assistance of knowledge bases and features, such as pattern recognition engines. Although some machine vision algorithms have been developed to mimic human visual perception, no machine vision system can yet match the capabilities of human vision in terms of image comprehension, tolerance to lighting variations, image degradation, and part variability.

A number of unique processing methods have been developed to process images and identify relevant image features in an effective and consistent manner. Among these are various line, circular, or area tools to detect edges or count pixels within a defined intensity or brightness range; "blob" tools to identify patterns or shapes of a certain size; and color and text recognition tools.

A typical machine vision system will consist of several of the following components:

    1. A digital or analog camera (black-and-white or color) with optics for acquiring images.
    2. Camera interface for digitizing images (widely known as a "frame grabber"). This converts the image to a digitized format, typically a two-dimensional array of intensity values.
    3. This is then placed in memory for analysis by the software algorithms. A processor (often a PC or embedded processor, such as a digital signal processor [DSP]).
    4. I/O hardware (digital I/O) or communication links (usually a network or RS-232 connection) to report results.
    5. A lens to focus the desired field of view onto the image sensor.
    6. Suitable, often very specialized, light sources (LED illuminators, fluorescent or halogen lamps, direct on-axis, and others). The lighting is designed to enhance or highlight certain features while obscuring or minimizing those that are not of interest. Generating or eliminating shadows is one of the principal purposes of adding lighting.
    7. A program to process images and detect relevant features.
    8. A synchronizing sensor for part detection (often a photo-eye . or proximity switch) to trigger image acquisition and processing. This sensor may also be used to trigger a synchronized lighting pulse to freeze a sharp image.

In some cases, some or all of the above are combined within a single device, called a smart camera. The use of an embedded processor eliminates the need for a frame grabber card and external computer, reducing cost and complexity of the system while providing dedicated processing power to each camera. Smart cameras are typically less expensive than systems made up of a camera and a board and/or external computer. FIG. 18 shows two different types of camera.

The one on the right is used for reading bar code images. Note the adjustable lens on the left camera; this is used to regulate the light input and focus the image.

The camera itself typically uses a CCD or CMOS image sensor.

Both of these devices perform the task of converting light into an electrical signal. The array of pixels creates an image by patterning the light and dark pattern focused onto the sensor by the optics of the camera. The intensity levels are then processed by the software into patterns that can be analyzed by the various tools.

Usually the software takes several steps to process an image. First the image is processed to reduce noise. It may also convert the many analog shades of gray into a simpler combination of black-and-white pixels, a process known as binarization. To do this an analog threshold is set in the software. After this simplification of the image, software can count or identify objects and measure or determine the size of patterns and features. The final step is to pass or fail the captured image based on the criteria entered by the user. The result is then communicated by digital signals or communications to a control system that can then act on the information to reject or process the part.

FIG. 18 Cognex In-Sight smart cameras.

Though most machine vision systems rely on black-and-white cameras, the use of color cameras is becoming more common. It is also increasingly common for machine vision systems to include digital camera equipment for direct connection rather than a camera and separate frame grabber, thus reducing signal degradation.

X-ray sensors are sometimes used to look inside materials for flaws such as cracks or bubbles. When combined with vision technology, these sensors can be used for automatic material sorting.

Gas Chromatography

Gas chromatography-mass spectrometry (GC-MS) is a method that is used in some chemical and process plants as a means of identifying and separating substances. This requires that a sample of the substance be captured, ionized, accelerated, deflected, and detected at the molecular level. The instruments that do this are quite expensive but are used in the food and beverage, perfume, and pharmaceutical industries.

Bar Codes, RFID, and Inductive ID

A bar code is a method of representing data by putting it into a visible, machine-readable format. Originally, bar codes were only represented by parallel lines that varied in width and spacing to encode alphanumeric data. This is called one-dimensional (1-D) or linear bar coding. Two-dimensional methods are also widely used today as the reader technology has evolved.

Linear or 1-D readers contain a light source that reflects off the black-and-white lines similar to a diffuse photo-eye. The light source is generally a red LED or laser. To cover a larger read area, the transmitted light will sometimes "raster" or move up and down.

FIG. 19 shows a commonly seen linear bar code.

The mapping of patterns into characters is known as a "symbology." This specification includes the coding for the alphanumeric characters along with the start and stop characters and computation of a checksum (a simple error-detection scheme).

There are more than 30 different 1-D codes in use. Most of these fall into two groups, discrete or continuous, depending on whether characters begin and end with a bar or not. There are also two-width or many-width classifications. Some of the more common symbologies are UPC, Code 39, and interleaved 2 of 5. Most 1-D readers can be set to read any of the common formats.

FIG. 19 1-D bar code.

Bar codes later evolved into other geometric patterns in two dimensions (2-D). These bar codes are usually made up of rectangles, dots, hexagons, or other geometric shapes arranged in a grid pattern.

Readers for 2-D bar codes generally use a CCD camera to capture the bar code image. Two-dimensional symbologies cannot be read by a laser as there is typically no sweep pattern that can encompass the entire symbol. FIG. 20 shows a 2-D bar code.

Some of the more common codes for 2-D symbologies are DataMatrix, Codablock, EZCode, and QR code. The automotive industry is a major user of 2-D DataMatrix codes since the pattern can be directly imprinted into a metal part using a pinstamp or "dot peen" marking system. Laser etching can be used for the same purpose.

Radio frequency identification (RFID) systems are another method of tagging parts and identifying them. Unlike bar codes, however, the tag does not have to be within line of sight range of the reader and may even be embedded inside an object. A common use of RFID systems in industrial automation is to track pallets or carriers through a process. An RFID system consists of a radio transmitter-receiver for two-way communication interfaced with a processor for the received information and the RFID tags that contain the information. Tags consist of an integrated circuit containing data and an antenna. The RFID reader sends a signal to the tag and reads its response. It may also work as a read-write system that transmits data to the tag for tracking purposes.

RFID tags can be passive, using the radio energy transmitted by the reader to power its circuit, or active with a tiny battery. Another option is a battery-backed passive tag, which is only activated when in the presence of a reader. Passive tags can be made much smaller and less expensively than active or battery-backed passive tags but must be very close to the reader for the field to be strong enough to activate the tag. FIG. 21 shows a reader and several RFID tags.

FIG. 20 2-D bar code.

Tags may contain a pre-coded unique serial number for lookup in a database or hold product-related information such as a part number, production date, or lot code. Read-write tags may be coded at various locations as a part travels through production, or they may be of the write-once read-multiple variety, also known as a field programmable tag. Information coded into a tag is stored electronically using nonvolatile memory.

RFID systems usually operate either in the high-frequency (HF) or ultrahigh-frequency (UHF) range of the radio spectrum. The distance at which an RFID tag can be read varies from less than a foot for some inexpensive, small passive tags to hundreds of feet for some larger active tags. More than one tag at a time may respond to the interrogation signal transmitted by a reader, so collision detection is often an important feature for an RFID controller.

Inductive ID systems serve a similar function to RFID systems but use a coil of wire similar to a proximity switch. The reader will excite an oscillator circuit in the tag that will transmit a serial code. Inductive ID systems can be lower cost and less susceptible to radio interference, but typically handle less information. They also have a much shorter range. Like RFID tags, inductive tags also come in active or passive, read-write, or read-only varieties.

FIG. 21 RFID reader and tags.

Keyboard Wedge

A keyboard wedge is an interface that allows a device such as a bar code scanner or magnetic strip reader to emulate a keyboard. The name "wedge" describes the physical position it occupies wedged between the keyboard and the computer port- FIG. 22 illustrates this arrangement. For example, a bar code reader converts the scanned code into a human readable alphanumeric format and then passes it through the wedge as if it were typed on the keyboard. The computer does not know whether the data came from the keyboard or another device, and the data is translated seamlessly.

A keyboard wedge may also be a software program that takes information into a USB or COM port and routes it through the keyboard buffer. Again, this is a transparent process from the perspective of the computer. This is a less expensive method of interfacing card scanners or bar code readers with a control system, but typically designers would choose a dedicated port for peripheral devices that are to be used often.


FIG. 22 Keyboard wedge diagram. Barcode Reader Hardware Wedge Keyboard; Computer


Top of Page

PREV   NEXT   Guide Index HOME