Sensor Selection for Inspection and Control

In early manufacturing lines the only devices that could detect a product on the line were switches that actually touched the product. In contrast, sensors deliver noncontact detection — eliminating the mechanical wear and potential product damage created by prior methods.

Because they don’t touch the target, sensors must detect it apart from the background based upon some distinguishable change in conditions. For example, the sensor may beam a light to a receiver positioned across a manufacturing line as a way of detecting the object. To ensure reliable detection, the sensor must repeat this operation with high accuracy, despite potential variances in the target object color, orientation, or environment.

It is for this reason that machine builders cannot rely upon a “one-sensor-fits-all” mentality. The traits of the object or feature of interest, as well as the application environment, will determine what sensor type will prove more advantageous. Some of the most common sensor types today include standard photoelectric sensors, lasers, fiber-optic sensors, ultrasonic sensors, and vision sensors.

Photoelectric sensing
The go-to sensor — the first many engineers consider when designing inspection systems — is the photoelectric sensor, also known as a photo or electric eye. A photoelectric sensor uses a photo diode and photo transistor pair to inspect for the presence or absence of a product or product feature, making it useful for discrete applications. The sensor itself is an electrical device that identifies and responds to changes in light intensity. The light source that originates with the emitter is designed to reach the receiver in one of several sensing modes:

In opposed-mode sensing, the emitter and receiver sit in separate housings with the emitter aligned directly opposite the receiver. Product flows along a conveyor between the two devices and interrupts (or breaks) the light beam to activate the sensor signifying an object is present.

Retroreflective-mode sensing typically has both emitter and receiver within the same sensor housing. A retroreflector takes the position of the receiver in opposed-mode sensing, reflecting the light beam from the sensor emitter back to the receiver in the same housing. The receiver detects the drop in light intensity when the target breaks the reflected beam. An advantage to this sensor is that it can work with conveyors placed along a wall, as the retroreflector doesn’t need much depth for mounting. This arrangement also eliminates the need for power on both sides of the conveyor.

Diffuse-mode sensing also locates emitter and receiver elements in the same housing, but the target object makes, rather than breaks, the light beam. The prior two sensing modes show product presence when the light beam either did not reach the receiver or was significantly attenuated from the presence of the product. Here, light energy reflected back from the target object indicates when the target is in position.

A background suppression-mode sensor also relies upon the target object to complete the light beam, but uses a defined cutoff point to prevent false readings caused by distracting or reflective backgrounds. Both fixed-field and adjustable-field sensors are available: fixed, for applications where parameters are used long term and to prevent tampering; and adjustable, for applications where parameters are prone to change.

Opposed-mode sensing offers the highest reliability to sense opaque objects. These sensors also feature high excess gain, which is the amount of light reaching the receiver of a sensing system that is over and above the minimum amount of light needed to activate the sensor output. High excess gain lets opposed-mode sensors deliver reliable results in contaminated environments by “burning through” dust and other airborne particles. This beam-block arrangement also suits objects made of challenging materials, such as those having reflective surfaces, as long as the object is opaque. Clear objects, though, are unsuitable targets as a photoelectric beam will pass through a clear object as if it wasn’t there. This technique also offers the quickest response of all the sensing modes, making opposed-mode sensing well suited for detecting and counting objects on high-speed production lines.

One primary reason opposed-mode sensing isn’t used is that this arrangement needs two electrical devices, the emitter and receiver, installed, aligned, and powered on opposite sides of the product flow. Retroreflective-mode sensing only needs power on one side, as the retroreflector is used on the opposing side making it also particularly useful for confined areas. It is also a more economical solution in many cases than opposed-mode sensing. However, in this arrangement the light beam is actually traveling twice the distance. Unlike an opposed-mode photoelectric that creates a direct, one-way path from emitter to receiver, this beam travels to the retroreflector and back. Not only does this double the beam’s travel distance, but it also reduces the excess gain of the sensor. That can make it more susceptible to environmental contamination of the type found in dusty areas.

Diffuse-mode sensing, another economical solution, better suits scenarios where opposed-mode and retroreflective-mode sensing cannot be used either because of location or beam-reflection length. Because diffuse sensors rely upon the light energy reflecting off the target object to activate, objects with dark or light-absorbing features, such as foam or black rubber, are not ideal candidates for this sensing mode. In addition, an inconsistent or reflective background can easily distract these sensors leading to errant output signals.

In those cases, background-suppression sensors make a better choice. Background-suppression sensors have a definite range limit, so backgrounds that can distract other diffuse-mode photoelectrics are ignored. They also offer high excess gain and can handle surface reflectivity well, though shiny surfaces beyond the sensing range can falsely trigger these sensors.

Two less commonly used sensing modes, divergent-diffuse and convergent-mode sensing, differ in how the light beam focuses on the object. In both instances the target object makes rather than breaks the beam. Divergent-mode sensing uses a wide light beam with almost no “blind spots” to detect small objects at a short distance, as well as clear materials and some shiny surfaces. Convergent-mode sensing employs a lens system to focus emitter and receiver elements on a small, precise point in front of the sensor. This arrangement is useful for counting radiused objects placed close together on a production line. It can detect the front of one can located a set distance from the sensor, while ignoring the rest. Standard photoelectrics may produce a constant signal in these cases, as the sensor can be activated by any section of the can, including sides that may be touching the next can in line.

In some photoelectric sensors, the traditional light beam is replaced with a Class 1 or Class 2 laser. The laser is particularly ideal for small-object detection, as the beam is smaller and easier to focus on precise targets. The lasers can be infrared or visible red, which aids in sensor alignment. Lasers are also capable of employing the triangulation principle either to determine presence or to handle analog applications such as distance measurement. This capability allows these sensors to determine not only if a bottle cap has been applied, but also whether it has been fastened properly. Lasers are also used in equipment such as measuring light screens for similar purposes.

Fiber-optic sensors
Standard photoelectric sensors work well in many plants, but their electronics can prove hazardous in potentially explosive environments. Instead, photoelectric sensors in these environments are made with fiber-optic assemblies. Small, flexible fibers transmit the light from the sensor emitter to the target and back to the receiver.

Fiber-optic assemblies are made with glass or plastic fibers. Each glass fiber strand usually has a diameter of 0.002 in. Hundreds of these fibers are typically bundled within a flexible armored cable or other protective sheathing material. Plastic fibers are packaged with a larger, monofilament core to become a single fiber-optic strand from 0.01 to 0.06 in. in diameter. These tiny optics are far more rugged than they appear: They can survive corrosive conditions, including extreme moisture, and some can tolerate elevated temperatures as high as 900°F.

Their compact design makes fiber-optic assemblies particularly well suited for mounting and positioning within confined areas. Because fiber-optic assemblies serve as passive mechanical components, they can be safely piped in and out of hazardous areas while delivering complete immunity to electrical noise. Their low mass also lets these assemblies withstand high levels of vibration and mechanical shock.

While fiber-optic assemblies fall under photoelectric sensing, this type of sensing costs more than standard photoelectric systems. Much of the emitted light energy is attenuated through the fibers giving these sensors shorter range and less excess gain than self-contained photoelectrics. Therefore, it’s best to avoid fiber-optic sensors in highly contaminated environments.

Ultrasonic sensors
Though light energy offers the fast response time often desired on manufacturing lines, the traits of the target object or its background may deem a photoelectric solution unsuitable. Photoelectrics may detect objects of one color more easily than others because different colors reflect different amounts of light. A black object, for instance, will appear quite different to a diffuse-mode sensor than a white object. And a photoelectric may be unable to distinguish a clear object from its background.

Ultrasonic sensors use sound waves instead of light beams to detect objects, making them color-indifferent and also good for detecting reflective or translucent objects. The sensors use sound energy at frequencies of 20 kHz or higher, slightly beyond the range of human hearing, created by a ceramic transducer that vibrates when electrical energy is applied to it. Vibrations compress and expand air molecules in waves from the sensor face to a target object. The ultrasonic sensor both transmits and receives sound, first emitting a sound wave and then “listening” over a designated time period for the return echo before retransmitting. The time it takes to receive the return echo relates to the object’s distance from the sensor, while a lack of return echo indicates no product present.

Sound waves are affected differently by different materials. For example, the returning echo is different between a full and a half-empty bottle of liquid. The sensor can be set to identify this difference to indicate when a bottle is or is not properly filled. Ultrasonics can additionally be used in applications where the target objects may be located at varying distances from the sensor. In this setup, the background rather than the target object is identified as the “good” condition that the sensor seeks. Whenever this condition is not identified, the sensor’s output switches to indicate that an object is present.

If an ultrasonic sensor is selected, it becomes necessary to ensure consistent temperature and airflow conditions in the application environment. The speed of an ultrasonic sensor sound wave is affected by temperature fluctuation. Sound travels faster to and from the target as temperature rises. This effect can make the target object appear closer to or further from the sensor than its true distance. Whether caused by fans or pneumatic equipment, air currents may also deflect or disturb the path of the ultrasonic wave, making it difficult for the sensor to identify the correct location of the target object.

Vision sensors
At times, engineers need to see the big picture of more than a single point or a single type of object. For instance, a full pattern or texture may need to be verified, or several features may need to be inspected at once. Multiple sensors could be used in some of these inspection scenarios with a single sensor used to verify the presence of each feature. But often these features are difficult to distinguish from their background using ordinary optical sensors. In addition, the process of purchasing and installing the multitude of sensors that may be needed for more complex applications may prove inconvenient or cost-prohibitive.

These are areas where vision sensors excel by obtaining and analyzing a full snapshot of the application in process. In addition to confirming the presence of a feature or region of interest on the target object, a vision sensor facilitates control: process control, machine-tool control, robot control, or quality control. It accomplishes this using the three major components that make up a vision inspection system: the lighting, the sensor, and the software.

Proper dedicated lighting creates the optimal contrast between the feature or region of interest and its background to let the camera identify the target efficiently and with higher accuracy. Numerous lighting styles and arrangements are available to highlight differences in color, texture, or shape. These arrangements can be likened to the different photoelectric-sensing modes. A backlight, for instance, located behind the target object produces a silhouette of the product, a setup similar to opposed-mode sensing.

The vision sensor contains a lens that focuses the reflected light onto an imager chip. The imager consists of an array of tiny light-sensitive cells called photosites, which convert the light into a 2D image. The light energy is stored in the photosite as an analog signal, proportional to the brightness of the light striking the cell. This information is then digitized by an a/d converter and typically displayed as an 8-bit gray-scale image.

The image gets processed by vision software consisting of controls, a graphical-user interface, and tool algorithms. Before the inspection begins, users define the region of interest. They then select vision tools within the software that will search the region of interest in each captured image to determine if the correct features are present. One region of interest within the full captured image is drawn for each vision tool, and many tools can be used to let users efficiently inspect several features at once. A test tool evaluates the results of all other tools and outputs a pass/fail signal accordingly.

If the proper feature of interest is identified in each instance, the target passes the inspection; if not, the target fails the inspection. These results are communicated to the manufacturing line through discrete outputs, a serial connection, or an Ethernet port. These outputs often work along with external devices such as diverters, so any products that fail inspection are diverted from the line.

Vision systems do carry a higher cost than traditional sensors, and they require significantly more programming. However, as graphical-user interfaces have become more user friendly, the programming factor is becoming less of a concern. Many first-time users can set up an inspection with ease on the latest systems.

Perhaps one of the most advantageous aspects of a vision sensor is its flexibility. With the right vision tools and lighting, the same vision sensor can be used to solve virtually all of the above applications. In automation environments where the products manufactured vary significantly, a system designed to provide an equal number of inspection options may be the most effective and cost-efficient option in the long run.

© 2010 Penton Media, Inc.