Motion Control Resources
- This technical feature is filed under:
Understanding Sensors for Motion Control and Manufacturing
by Kristin Lewotsky, Contributing Editor
Motion Control & Motor Association Posted 09/29/2020
Modern machines are festooned with sensors. The role of sensors begins with motive power – after all, a DC motor only becomes a servo motor with the addition of closed-loop feedback. Sensors perform many other tasks, however. Sensors detect parts, check fill levels, inspect labels, perform quality control, govern the interaction between different motion axes, between the motion control assemblies and other subsystems, and between the machine and the load. They monitor equipment health to minimize downtime. Without this input, the factory as we know it couldn’t function.
The primary functions of sensors on a motion system are:
- Monitoring machine operation
- Monitoring machine condition
- Monitoring interaction with the load or with other systems
Each of these tasks requires different measurement modalities and sensor technologies. Let’s consider a few:
Monitoring Machine Operation
Probably the most essential sensor for a motion control system is the motor feedback sensor. Typically attached to the motor shaft or the load, these devices supply the control system with input that can be converted into speed or position in order to drive down the error between actual and commanded position/speed.
The most common devices used for this task are encoders and resolvers. In an encoder, the motion of the motor or load causes a patterned disc or strip to modulate a constant input signal. Encoders can be divided into incremental and absolute types. Incremental encoders generate a pulse stream that can be converted into speed or to angular displacement from some home position established at start up. For absolute encoders, output is an n-bit digital word corresponding to a specific angular position.
Encoders can be implemented as optical, magnetic, or inductive designs. Optical encoders use an LED on one side of the code disc and a photodiode or optical phased array detector on the other side (see Figure 1). As the motor shaft turns, the patterned disc modulates the beam. Linear encoders operate analogously. Optical encoders offer the highest resolution output, making them good fits for demanding applications. Because they are optical, they tend to be sensitive to thermal variation and shock and vibration, as well as contamination.
Magnetic encoders operate similarly to optical encoders, using a magnetically patterned wheel spinning over an array of magnetoresistive detectors. The sensing technology makes them very rugged and effective for harsh environments involving vibration and contamination. Because they involve magnets, there are some limitation to shock and temperature swings. For highest resolution and performance, Hall-effect encoders use a chip-based array to increase sensitivity and reduce error.
Encoder selection involves a number of considerations; click here for a detailed tutorial on how to choose the most effective encoder for your application.
A resolver is a rotary position sensor that contains no electronics. Resolvers are differential transformers in which a coil attached to the rotor induces a voltage in two readout coils that are 90° out of phase (see Figure 2). The angular position is determined by taking the arctangent of the ratio of the sine coil voltage and the cosine coil voltage. As analog devices, resolvers have theoretically infinite resolution, although they do require analog-to-digital conversion. Because they do not incorporate onboard electronics, resolvers are very robust and good candidates for harsh environments such as those with extreme temperatures, high shock and vibration, or high amounts of radiation.
Inductive encoders are chip-level versions of a resolver. The coils are lithographically patterned on a PCB that can be mounted to the motor stator. The rotor/shaft excites the coils via a conductive disc.
Many motion-control applications require torque monitoring capabilities, for example, twisting the cap onto a bottle or tube to a prescribed level of tightness. There are several technologies used for torque sensing.
Monitoring Machine Condition
The high cost of downtime, especially unscheduled downtime has led to increasing interest in condition monitoring for purposes of predictive maintenance. Three key sensor technologies for condition monitoring are current sensors, temperature sensors, and vibration sensors. Drive current is directly related to torque output for motor. If the current draw of a motor suddenly increases, it means that the axis requires more torque to move the load. Common causes include bearing defects, lubrication breakdown, improperly tensioned belts, misalignments, etc.
Modern drives and some smart motors include current sensors and data loggers that make it possible to track current over time, with built-in analytics and the ability to send alerts when thresholds are exceeded.
Used in conjunction with model-based algorithms, current sensors can be used as virtual torque sensors. They even can act as virtual quality control devices – if the current draw meets spec, then the manufacturing step has been properly executed.
Vibration monitoring is an effective tool for rotating assets, particularly continuously rotating assets such as fans, pumps, and blowers. Every mechanical system has a natural frequency of oscillation. Defects in the system alter the vibration spectrum. By monitoring that spectrum over time for changes in magnitude and the emergence of new frequencies, vibration analysis can provide early warning of defects with a high degree of specificity (see Figure 3). Here, too, common issues include bearing cage defects, lubrication problems, damaged impeller blades, and more.
The most common vibration monitoring types are based on piezoelectric and micro electromechanical systems (MEMS) technology.
In a MEMS-based accelerometer, a small proof mass is suspended between parallel plates to form dual capacitors. When the device vibrates, the proof mass oscillates so that the capacitance on one side increases while the capacitance on the other side increases (see Figure 4). This data can be used to calculate vibration frequency. The structure is typically repeated multiple times in a MEMS sensor to average out performance. The active region is suspended in the sensor housing by flexures. By modifying the mass of the test mass and the stiffness of the flexures, it is possible to make MEMS sensors effective over a variety of vibration frequency ranges.
Piezoelectric sensors use piezoelectric ceramic as the sensing modality. When a piezoelectric ceramic is subjected to pressure, it generates a voltage.
Monitoring Interaction with the Load
The industrial environment can be a challenging one marked by contamination, temperature swings, shock and vibration, and EMI. The needs of each application differ. As a result, OEMs and integrators need to consider a variety of sensing modalities in order to develop the best solution for each project. The primary sensing modalities used in automation are:
- Optical sensing
- Capacitive sensing
- Inductive sensing
- Magnetic sensing
- Ultrasonic sensing
These technologies can be used to monitor how products are moving through the line, how they are interacting with the equipment, and the quality and properties of the products themselves.
One of the most common sensing tasks in manufacturing is present or absence detection. The objective may be as simple as determining whether a box is on a conveyor. It may be as complex as determining whether the lid of a new soda bottle has been screwed on properly or is skewed. The best modality to use depends upon the properties of the objects being moved, the distance from the sensor head, the speed of motion, and the accuracy needed.
Because photoelectric sensors use light as the sensing modality, they are compatible with a variety of materials, from metal to plastic to paper. A photoelectric sensor consists of an emitter that generates the signal and a receiver that detects the response. Photoelectric sensors use either LED or laser diode sources. The receivers are photodiodes that convert light into photocurrent. Wavelengths very depending on the conditions of the application. IR and red sources are most common but other wavelengths may be used depending on the environment, the ambient lighting, and the materials involved.
Photoelectric sensors can be classed by sensing mode into through-beam, retroreflective, and diffuse-reflective sensors. As with most things in engineering, each has its pros and cons and sweet spot applications
In a through-beam sensor, the emitter is placed on one side of the active region and the receiver is placed on the other side (see Figure 5). The source generates a beam that can be received by the photodetector. A sensing object passing between the two attenuates the beam, reducing the photo current interrupts the photocurrent and registers as a detection; a continuous beam counts as an absence.
Note: Some photodetectors are designed so that the absence of light generates a signal in the presence of light does not. These so-called dark-eyed photodetectors can be used to detect absence rather than presence. For the sake of simplicity, in this article we will focus on presence detection.
Applications can be as simple as counting objects on a conveyor belt or checking perforated material using a narrow-beam laser source.
Through-beam sensors can measure across a long distance with high accuracy. The solid-state design makes it extremely accurate. Because the system consists of two components, it’s more complex. In particular, accurate alignment is essential. Applications include object detection, determination of characteristics like stack heights and fill levels, quality checks like skewed-lid detection, parts counting, and triggering vision systems.
In a retroreflective sensor, both emitter and receiver are housed in the same sensor head. The sensor head is mounted opposite a retroreflector. The source sends out a signal, which reflects from the retroreflector and returns to the receiver (see Figure 6). When a object passes between them, attenuating the beam, the system registers a presence.
Retroreflective sensors tend to be insensitive to the color and inclination of the objects under test. They are not, however, completely insensitive to finish. A very shiny object mimics the signal returned by the retroreflector and can result in a false-negative. The solution is to modify the system slightly by using a corner cube retroreflector and adding a polarization filter before the detector. The retroreflector will return polarized light, which will pass through the matched polarization filter. Any reflection from the shiny surface will be unpolarized, and thus attenuated by the filter.
Retroreflective sensors can also be used to detect transparent objects. This function requires customized low-hysteresis circuitry that makes the system sensitive to even very small attenuations. Adding polarization filters on the emitter and receiver provisions false negatives caused by the reflection.
Because the only require a single sensor head, retroreflector sensors are both simpler and more economical than through-beam versions. The trade-off is a reduction in range from centimeters to multiple meters. The polarization filters also attenuate the signal, further reducing range. Because the emitter and the receiver are spatially separated, the light from the retroreflector actually returns at a slight angle. This reduces the accuracy of the sensor somewhat compared to the through-beam versions.
Applications include object detection, particularly of transparent parts; parts counting, even at high speeds, and more.
The diffuse-reflective photoelectric sensors are the simplest of the bunch. As with the retroreflective design, the sensor heads include both emitter and receiver. The design doesn’t make use of a retroreflector, however. Instead, the light from the emitter travels out to the sensing object and is reflected back to the receiver (see Figure 7).
Light bouncing off a reflective surface tends to return in a narrow (specular) cone. The light returning from a nonreflective surface reflects in many directions, the rougher the surface, the more random (diffuse) the reflection. Some of the light returns to the receiver in the sensor head; above a certain threshold, the system detects the sensing object.
Diffuse reflective sensors are simple to install and the least expensive of the three modes. They can operate at distances of centimeters to a few meters. Because these devices use diffuse reflectance, however, the signal-to-noise ratio can be very high. As a result, sensing distance and other performance of basic diffuse-reflective sensors is highly dependent on the color and finish of the sensing object, as well as ambient light conditions. All the details of the application, particularly of the objects being monitored, need to be taken into account. Detecting a dark object in front of a light background, for example, would require special design.
To address some of these concerns, diffuse-reflective sensors can be designed to reject ambient light (see Figure 8). A sensor can be set to detect only objects at a specific distance using a two-part photodiode (see Figure 8). Light reflected from a surface in the foreground would propagate to the near (N) photodetector, while light propagating from a surface in the background would wind up on the far (F) photodetector. Only light falling equally on both photodetectors would register as a presence. This principle can be used for rejecting foreground and background optical noise. It could also be used to convert a diffuse-reflective sensor into a distance-measuring device.
Proximity sensors are non-contact sensors that can detect the presence of an object based on the interaction with its materials. The three sensor modalities used in proximity sensors are capacitive, inductive, and magnetic.
Capacitance varies as a function of proximity and area, which can be used to determine the size and distance of the sensing object. Capacitive sensors have ranges of up to several tens of millimeters. As they are solid-state and thus very robust and economical. They work with metallic objects as well as dielectric materials such as resins, liquids, and powders. Measurement performance can be affected by nearby objects, and EMI from power and signal cables. Performance also varies as a function of temperature.
Inductive proximity sensors are based on the interaction between a winding and circuit connected to form a tuned frequency oscillator. The oscillator generates a constantly changing magnetic field that can induce eddy currents in a nearby conductor. The magnitude of the eddy currents is proportional to the proximity of the field. The eddy currents, in turn, affect the magnitude of the oscillator circuit. The oscillator thus functions as interrogator and readout.
The effectiveness of an inductive sensor varies depending upon the properties of the metal object it’s trying to detect. The technology works best with ferrous materials. The eddy current effect is reduced in nonferrous metals such as aluminum and brass. As a result, the operating distance is scaled by the material – the sensing range of a brass or bronze object, for example, might be one half that of an iron object.
Magnetic proximity sensors metallic read switch that can be controlled by magnetic field. When a magnetic object or a small sensor magnet placed on the object of interest nears sensor, the switch closes, generating a signal. Magnetic proximity sensors are simple and robust. They are EMI immune and unaffected by moisture and contamination. On the downside, they do require a magnetic device on the object of interest.
For helpful conversations, thanks go to Michael Hamoy, product manager, sensors at Omron Automation Americas