SENSORS USED IN COMPUTER CONTROL
Sensors are an important part of closed-loop systems. A sensor is a device that outputs a signal which is related to the measurement of (i.e. is a function of) a physical quantity such as temperature, speed, force, pressure, displacement, acceleration, torque, flow, light or sound. Sensors are used in closed-loop systems in the feedback loops, and they provide information about the actual output of a plant. For example, a speed sensor gives a signal proportional to the speed of a motor and this signal is subtracted from the desired speed reference input in order to obtain the error signal.
Sensors can be classified as analog or digital. Analog sensors are more widely available, and their outputs are analog voltages. For example, the output of an analog temperature sensor may be a voltage proportional to the measured temperature. Analog sensors can only be connected to a computer by using an A/D converter. Digital sensors are not very common and they have logic level outputs which can directly be connected to a computer input port.
The choice of a sensor for a particular application depends on many factors such as the cost, reliability, required accuracy, resolution, range and linearity of the sensor. Some important factors are described below.
Range. The range of a sensor specifies the upper and lower limits of the measured variable for which a measurement can be made. For example, if the range of a temperature sensor is specified as 10–60 ◦C then the sensor should only be used to measure temperatures within that range.
Resolution. The resolution of a sensor is specified as the largest change in measured value that will not result in a change in the sensor’s output, i.e. the measured value can change by the amount quoted by the resolution before this change can be detected by the sensor. In general, the smaller this amount the better the sensor is, and sensors with a wide range have less resolution. For example, a temperature sensor with a resolution of 0.001 K is better than a sensor with a resolution of 0.1 K.
Repeatability. The repeatability of a sensor is the variation of output values that can be expected when the sensor measures the same physical quantity several times. For example, if the voltage across a resistor is measured at the same time several times we may get slightly different results.
Linearity. An ideal sensor is expected to have a linear transfer function, i.e. the sensor output is expected to be exactly proportional to the measured value. However, in practice all sensors exhibit some amount of nonlinearity depending upon the manufacturing tolerances and the measurement conditions.
Dynamic response. The dynamic response of a sensor specifies the limits of the sensor characteristics when the sensor is subject to a sinusoidal frequency change. For example, the dynamic response of a microphone may be expressed in terms of the 3-dB bandwidth of its frequency response.
In the remainder of this chapter, the operation and the characteristics of some of the popular sensors are discussed.
1.7.1 Temperature Sensors
Temperature is one of the fundamental physical variables in most chemical and process control applications. Accurate and reliable measurement of the temperature is important in nearly all process control applications.
Temperature sensors can be analog or digital. Some of the most commonly used analog temperature sensors are: thermocouples, resistance temperature detectors (RTDs) and ther- mistors. Digital sensors are in the form of integrated circuits. The choice of a sensor depends on the accuracy, the temperature range, speed of response, thermal coupling, the environment (chemical, electrical, or physical) and the cost.
As shown in Table 1.1, thermocouples are best suited to very low and very high tempera- ture measurements. The typical measuring range is from −270 ◦C to +2600 ◦C. In addition, thermocouples are low-cost, very robust, and they can be used in chemical environments. The typical accuracy of a thermocouple is ±1 ◦C. Thermocouples do not require external power for operation.
RTDs are used in medium-range temperature measurements, ranging from −200 ◦C to +600 ◦C. They can be used in most chemical environments but they are not as robust as thermocouples. The typical accuracy of RTDs is ±0.2 ◦C. They require external power for operation.
Thermistors are used in low- to medium-temperature applications, ranging from −50 ◦C to about +200 ◦C. They are not as robust as thermocouples or RTDs and they cannot easily be used in chemical environments. Thermistors are also low-cost devices, they require external power for operation, and they have an accuracy of ±0.2 ◦C.
Integrated circuit temperature sensors are used in low-temperature applications, ranging from −40 ◦C to +125 ◦C. These devices can be either analog or digital, and their coupling with the environment is not very good. The accuracy of integrated circuit sensors is around ±1 ◦C. Integrated temperature sensors differ from other sensors in some important ways:
• They are relatively small.
• Their outputs are highly linear.
• Their temperature range is limited.
• Their cost is very low.
• Some models include advanced features, such as thermostat functions, built-in A/D convert- ers and so on.
• An external power supply is required to operate them.
Analog integrated circuit temperature sensors can be voltage output or current output devices. Voltage output sensors give a voltage which is directly proportional to the measured tempera- ture. Similarly, current output sensors act as high-impedance current sources, giving an output current which is proportional to the temperature.
A popular voltage output analog integrated circuit temperature sensor is the LM35DZ, manufactured by National Semiconductors Inc. (see Figure 1.9). This is a 3-pin analog output sensor which provides a linear output voltage of 10 mV/◦C. The temperature range is 0 ◦C to +100 ◦C, with an accuracy of ±1.5 ◦C.
The AD590 is an analog integrated circuit sensor with a current output. The device operates in the range −55 ◦C to +150 ◦C and produces an output current of 1 µA/◦C.
Digital integrated temperature sensors produce digital outputs which can be interfaced to a computer. The output data format is usually nonstandard and the measured temperature can be extracted by using suitable algorithms. The DS1620 is a popular digital temperature sensor which also incorporates digitally programmable thermostat outputs. The device provides a 9-bit serial data to indicate the measured temperature. Data is extracted from the device by sending clock pulses and then reading the data after each pulse. Table 1.2 shows the sensor’s measured temperature–output relationship.
There can be several sources of error during the measurement of temperature. Some impor- tant possible errors are described below.
Sensor self-heating. RTDs, thermistors and integrated circuit sensors require an external power supply for their operation. The power supply can cause the sensor to heat, leading to an error in the measurement. The effect of self-heating depends on the size of the sensor and the amount of power dissipated by the sensor. Self-heating can be avoided by using the lowest possible external power, or by considering the heating effect in the measurement.
Electrical noise. Electrical noise can introduce errors into the measurement. Thermocouples produce very low voltages (of the order of tens of microvolts) and, as a result, noise can easily enter the measurement. This noise can usually be minimized by using low-pass filters, and by keeping the sensor leads as short as possible and away from motors and other electrical machinery.
Mechanical stress. Some sensors such as RTDs are sensitive to mechanical stress and should be used carefully. Mechanical stress can be minimized by avoiding deformation of the sensor. Thermal coupling. It is important that for accurate and fast measurements the sensor should make a good contact with the measuring surface. If the surface has a thermal gradient then incorrect placement of the sensor can lead to errors. If the sensor is used in a liquid, the liquid should be stirred to cause a uniform heat distribution. Integrated circuit sensors usually suffer from thermal coupling since they are not easily mountable on surfaces.
Sensor time constant. The response time of the sensor can be another source of error. Every type of sensor takes a finite time to respond to a change in its environment. Errors due to the sensor time constant can be minimized by improving the coupling between the sensor and the measuring surface.
Position sensors
Position sensors are used to measure the position of moving objects. These sensors are basically of two types: sensors to measure linear movement, and sensors to measure angular movement. Potentiometers are available in linear and rotary forms. In a typical application, a fixed voltage is applied across the potentiometer and the voltage across the potentiometer arm is measured. This voltage is proportional to the position of the arm, and hence by measuring the voltage we know the position of the arm. Figure 1.10 shows a linear potentiometer. If the
applied voltage is Vi , the voltage across the arm is given by
Potentiometer type position sensors are low-cost, but they have the disadvantage that the range is limited and also that the sensor can be worn out by excessive movement of the arm.
Among other types of position sensors are capacitive sensors, inductive sensors, linear variable differential transformers (LVDTs) and optical encoders. Capacitive position sensors rely on the fact that the capacitance of a parallel plate capacitor changes as the distance between the plates is changed. The formula for the capitance, C, ofa parallel plate capacitor is
where ε is the dielectric constant, A the area of the plates and d the distance between the plates. Typically, the capacitor of the sensor is used in the feedback loop of an operational amplifier as shown in Figure 1.12, and a reference capacitor is used at the input. If a voltage Vi is applied, the output voltage Vo is given by
From (1.3), if we apply a constant amplitude sinusoidal signal as the input, the amplitude of the output voltage is proportional to the distance between the plates.
LVDT sensors (see Figure 1.13) consist of one primary and two secondary windings on a hollow cylinder. The primary winding is in the middle, and the secondary windings have equal number of turns, series coupled, and they are at the ends of the cylinder (see Figure 1.14). A sinusoidal signal with a voltage of 0.5–5 V and frequency 1–20 kHz is applied to the primary winding. A magnetic core which measures the position moves inside the cylinder, and the movement of this core varies the magnetic field linking the primary winding to the secondary windings. Because the secondary windings are in opposition, the movement of the core to one position increases the induced voltage in one secondary coil and decreases the induced voltage in the other secondary coil. The net voltage difference is proportional to the position of the core inside the cylinder. Thus, by measuring the induced voltage we know the position of the core. The strong relationship between the core position and the induced voltage yields a design that exhibits excellent resolution. Most commercially available LVDTs come with built-in signal-conditioning circuitry that provides an easy interface to a computer. The device operates from a d.c. supply and the signal conditioner provides the a.c. signal required for the operation of the circuit, as well as the demodulation of the output signal to give a useful d.c.
voltage output. The range of an LVDT is from ±125 µm to ±75 mm and the sensitivity ranges from 0.6 to 30 mV per 25 µm under normal excitation of 3–6 V.
The advantages of LVDT are:
• low cost;
• robust design;
• no hysteresis effect;
• fast response time;
• no friction resistance;
• long life.
The main disadvantage of the LVDT is that the core must have direct contact with the measured surface, which may not always be possible.
Velocity and acceleration sensors
Velocity is the differentiation of position, and in general position sensors can be used to measure velocity. The required differentiation can be done either in hardware (e.g. using operational amplifiers) or by the computer. For more accurate measurements velocity sensors should be used. There are two types of velocity sensors: linear sensors, and rotary sensors.
Linear velocity sensors can be constructed using a pair of coils and a moving magnet. When the coils are connected in series, the movement of the magnet produces additive voltage which is proportional to the movement of the magnet.
One of the most widely used rotary velocity sensors is the tachometer (or tachogenerator). A tachometer (see Figure 1.15) is connected to the shaft of a rotating device (e.g. a motor) and produces an analog d.c. voltage which is proportional to the speed of the shaft. If ω is the angular velocity of the shaft, the output voltage of the tachometer is given by
Vo = kω,
where k is the gain constant of the tachometer.
Another popular velocity sensor is the optical encoder. This basically consists of a light source and a disk with opaque and transparent sections where the disk is attached to the rotating shaft. A light sensor at the other side of the wheel detects light and a pulse is produced when the transparent section of the disk comes round. The encoder’s controller counts the pulses in a given time, and this is proportional to the speed of the shaft. Figure 1.16 shows a typical commercial encoder.
Acceleration is the differentiation of velocity, or the double differentiation of position. Thus, in general, position sensors can be used to measure acceleration. The differentiation can be done either by using operational amplifiers or by a computer program. For accurate measurement of the acceleration, semiconductor accelerometers can be used. For example, the ADXL202 is an accelerometer chip manufactured by Analog Devices Inc. This is a low-cost 8-pin chip with two outputs to measure the acceleration in two dimensions. The outputs are digital signals whose duty cycles are proportional to the acceleration in each of the two axes. These outputs can be connected directly to a microcontroller and the acceleration can be measured very easily, requiring no A/D converter. The measurement range of the ADXL202 is ±2 g, where g is acceleration due to gravity, and the device can measure both dynamic acceleration (e.g. vibration), and static acceleration (e.g. gravity).
Force sensors
Force sensors can be constructed using position sensors. Alternatively, a strain gauge can be used to measure force accurately. There are many different types of strain gauges. A strain gauge can be made from capacitors and inductors, but the most widely used types are made from resistors. A wire strain gauge is made from a resistor, in the form of a metal foil. The principle of operation is that the resistance of a wire increases with increasing strain and decreases with decreasing strain.
In order to measure strain with a strain gauge, it must be connected to an electrical circuit, and a Wheatstone bridge is commonly used to detect the small changes in the resistance of the strain gauge.
Strain gauges can be used to measure force, load, weight pressure, torque or displacement. Force can also be measured using the principle of piezoelectricity. A piezoelectric sensor produces voltage when a force is applied to its surface. The disadvantage of this method is that the voltage decays after the application of the force and thus piezoelectric sensors are only useful for measuring dynamic force.
Pressure sensors
Early pressure measurement was based on using a flexible device (e.g. a diaphragm) as a sensor; the pressure changed as the device moved and caused a dial connected to the device to move and indicate the pressure. Nowadays, the movement is converted into an electrical signal which is proportional to the applied pressure. Strain gauges, capacitance change, inductance change, piezoelectric effect, optical pressure sensors and similar techniques are used to measure the pressure.
Liquid sensors
There are many different types of liquid sensors. These sensors are used to:
• detect the presence of liquid;
• measure the level of liquid;
• measure the flow rate of liquid, for example through a pipe.
The presence of a liquid can be detected by using optical, ultrasonic, change of resistance, change of capacitance or similar techniques. For example, optical technique is based on using an LED and a photo-transistor, both housed within a plastic dome and at the head of the device. When no liquid is present, light from the LED is internally reflected from the dome to the photo-transistor and the output is designed to be off. When liquid is present the dome is covered with liquid and the refractive index at the dome–liquid boundary changes, allowing some light to escape from the LED. As a result of this, the amount of light received by the photo-transistor is reduced and the output is designed to switch on, indicating the presence of liquid.
The level of liquid in a tank can be measured using immersed sensor techniques, or non- touching ultrasonic techniques. The simplest technique is to immerse a rod in the liquid with a potentiometer placed inside the rod. The potentiometer arm is designed to move as the level of the liquid is changed. The change in the resistance can be measured and hence the level of the liquid is obtained.
Pressure sensors are also used to measure the level of liquid in a tank. Typically, the pressure sensor is mounted at the bottom of the tank where change of pressure is proportional to the height of the liquid. These sensors usually give an analog output voltage proportional to the height of the liquid inside the tank.
Nontouching ultrasonic level measurement is very accurate, but more expensive than the other techniques. Basically, an ultrasonic beam is sent to the surface of the water and the echo of the beam is detected. The time difference between sending the beam and the echo is proportional to the level of the liquid in the tank.
The liquid flow rate can be measured by several techniques:
• paddlewheel sensors;
• displacement flow meters;
• magnetic flow meters;
Paddlewheel sensors are cost-effective and very popular for the measurement of liquid flow rate. A wheel is mounted inside the sensor whose speed of rotation is proportional to the flow rate. As the wheel rotates a voltage is produced which indicates the flow rate.
Displacement flow meters measure the flow rate of a liquid by separating the flow into known volumes and counting them over time. These meters provide good accuracy. Displacement flow meters have several types such as sliding vane meters, rotary piston meters, helix flow meters and so on.
Magnetic flow meters are based on Faraday’s law of magnetic induction. Here, the liquid acts as a conductor as it flows through a pipe. This induces a voltage which is proportional to the flow rate. The faster the flow rate, the higher is the voltage. This voltage is picked up by the sensors mounted in the meter tube and electronic means are used to calculate the flow rate based on the cross-sectional area of the tube. Advantages of magnetic flow rates are as follows:
• Corrosive liquids can be used.
• The measurement does not change the flow stream.
Figure 1.17 shows a typical magnetic flow meter.
Air flow sensors
Air flow is usually measured using anemometers. A classical anemometer (see Figure 1.18) has a rotating vane, and the speed of rotation is proportional to the air flow. Hot wire anemometers
have no moving parts (Figure 1.19). The sensor consists of an electrically heated platinum wire which is placed in the air flow. As the flow velocity increases the rate of heat flow from the heated wire to the flow stream increases and a cooling occurs on the electrode, causing its resistance to change. The flow rate is then determined from the change in the resistance.