Industrial Sensors and Control

7. Industrial Sensors and Control

7.1. Introduction
Current manufacturing strategy denes manufacturing systems in terms of sensors, actuators, eectors, controllers, and control loops. Sensors provide a means for gathering information on manufacturing operations and processes being performed. In many instances, sensors are used to transform a physical stimulus into an electrical signal that may be analyzed by the manufacturing system and used for making decisions about the operations being conducted. Actuators convert an electrical signal into a mechanical motion. An actuator acts on the product and equipment through an eector. Eectors serve as the hand that achieves the desired mechanical action. Controllers are computers of some type that receive information from sensors and from internal programming, and use this information to operate the manufacturing equipment (to the extent available, depending on the degree of automation and control). Controllers provide electronic commands that convert an electrical signal into a mechanical action. Sensors, actuators, eectors, and controllers are linked to a control loop. In limited-capability control loops, little information is gathered, little decision making can take place, and limited action results. In other settings, smart manufacturing equipment with a wide range of sensor types can apply numerous actuators and eectors to achieve a wide range of automated actions. The purpose of sensors is to inspect work in progress, to monitor the work-

in-progress interface with the manufacturing equipment, and to allow selfmonitoring of manufacturing by the manufacturing systems own computer. The purpose of the actuator and eector is to transform the work in progress according to the dened processes of the manufacturing system. The function of the controller is to allow for varying degrees of manual, semiautomated, or fully automated control over the processes. In a fully automated case, such as in computer-integrated manufacturing, the controller is completely adaptive and functions in a closed-loop manner to produce automatic system operation. In other cases, human activity is involved in the control loop. In order to understand the ways in which the physical properties of a manufacturing system aect the functional parameters associated with the manufacturing system, and in order to determine the types of physical manufacturing system properties that are necessary to implement the various desired functional parameters, it is important to understand the technologies available for manufacturing systems that use automation and integration to varying degrees. The least automated equipment makes use of detailed operator control over all equipment functions. Further, each action performed by the equipment is individually directed by the operator. Manual equipment thus makes the maximum use of human capability and adaptability. Visual observations can be enhanced by the use of microscopes and cameras, and the actions undertaken can be enhanced by the use of simple eectors. The linkages between the sensory information (from microscopes or through cameras) and the resulting actions are obtained by placing the operator in the loop. This type of system is clearly limited by the types of sensors used and their relationship to the human operator, the types of the eectors that can be used in conjunction with the human operator, and the capabilities of the operator. The manufacturing equipment that is designed for a manual strategy must be matched to human capabilities. The humanmanufacturing equipment interface is extremely important in many manufacturing applications. Unfortunately, equipment design is often not optimized as a sensor-operator-actuator/ eector control loop. A manufacturing system may be semiautomated, with some portion of the control loop replaced by a computer. This approach will serve the new

demands on manufacturing system design requirements. Specically, sensors now must provide continuous input data for both the operator and computer. The appropriate types of data must be provided in a timely manner to each of these control loops. Semiautomated manufacturing systems must have the capability for a limited degree of self-monitoring and control associated with the computer portion of the decision making loop. An obvious diculty in designing such equipment is to manage the computer- and operatorcontrolled activities in an optimum manner. The computer must be able to recognize when it needs operator support, and the operator must be able to recognize which functions may appropriately be left to computer control. A continuing machine-operator interaction is part of normal operations. Another manufacturing concept involves fully automated manufacturing systems. The processing within the manufacturing system itself is fully computer-controlled. Closed-loop operations must exist between sensors and actuators/eectors in the manufacturing system. The manufacturing system must be able to monitor its own performance and decision making for all required operations. For eective automated operation, the mean time between operator interventions must be large when compared with the times between manufacturing setups.

(7.1)
where = setup time

i = initial setup n = number of setups

The processes in use must rarely fail; the operator will intervene only when such failures occur. In such a setting, the operators function is to ensure the adequate ow of work in progress and respond to system failure. Several types of work cells are designed according to the concept of total manufacturing integration. The most sophisticated cell design involves fully automated processing and materials handling. Computers control the feeding of work in progress, the performance of the manufacturing process, and the removal of the work in progress. Manufacturing systems of this type provide the opportunity for the most advanced automated and integrated operations. The manufacturing system must be modied to achieve closed-

loop operations for all of these functions. Most manufacturing systems in use today are not very resourceful. They do not make use of external sensors that enable them to monitor their own performance. Rather, they depend on internal conditioning sensors to feed back (to the control system) information regarding manipulator positions and actions. To be eective, this type of manufacturing system must have a rigid structure and be able to determine its own position based on internal data (largely independent of the load applied). This leads to large, heavy, and rigid structures. The more intelligent manufacturing systems use sensors that enable them to observe work in progress and a control loop that allows corrective action to be taken. Thus, such manufacturing systems do not have to be as rigid because they can adapt. The evolution toward more intelligent and adaptive manufacturing systems has been slow, partly because the required technologies have evolved only in recent years and partly because it is dicult to design work cells that eectively use the adaptive capabilities. Enterprises are not sure whether such features are cost-eective and wonder how to integrate smart manufacturing systems into the overall strategy. The emphasis must be on the building-block elements necessary for many types of processing. If the most advanced sensors are combined with the most advanced manufacturing systems, concepts, and state-of-the-art controllers and control loops, very sophisticated manufacturing systems can result. On the other hand, much more rudimentary sensors, eectors, and controllers can produce simple types of actions. In many instances today, sensors are analog (they involve a continuously changing output property), and control loops make use of digital computers. Therefore, an analog-to-digital converter between the pre-processor and the digital control loop is often required. The sensor may operate either passively or actively. In the passive case, the physical stimulus is available in the environment and does not have to be provided. For an active case, the particular physical stimulus must be provided. Machine vision and color identication sensors are an active means of sensing, because visible light must be used to illuminate the object before

a physical stimulus can be received by the sensor. Laser sensors are also active-type sensors. Passive sensors include infrared devices (the physical stimulus being generated from infrared radiation associated with the temperature of a body) and sensors to measure pressure, ow, temperature, displacement, proximity, humidity, and other physical parameters.

7.2. Sensors in Manufacturing

Many types of sensors have been developed during the past several years, especially those for industrial process control, military uses, medicine, automotive applications, and avionics. Several types of sensors are already being manufactured by commercial companies. Process control sensors in manufacturing will play a signicant role in improving productivity, qualitatively and quantitatively, throughout the coming decades. The main parameters to be measured and controlled in industrial plants are temperature, displacement, force, pressure, uid level, and ow. In addition, detectors for leakage of explosives or combustible gases and oils are important for accident prevention. Optical-ber sensors may be conveniently divided into two groups: (1) intrinsic sensors and (2) extrinsic sensors. Although intrinsic sensors have, in many cases, an advantage of higher sensitivity, almost all sensors used in process control at present belong to the extrinsic type. Extrinsic-type sensors employ light sources such as LEDs, which have higher reliability, longer life, and lower cost than semiconductor lasers. They also are compatible with multimode bers, which provide higher eciency when coupled to light sources and are less sensitive to external mechanical and thermal disturbances. As described in Chapter 3, objects can be detected by interrupting the sensor beam. Optical-ber interrupters are sensors for which the principal function is the detection of moving objects. They may be classied into two types: reection and transmission. In the reection-type sensor, the light beam emitted from the ber is reected back into the same ber if the object is situated in front of the sensor.

In the transmission-type sensor, the emitted light from the input ber is interrupted by the object, resulting in no received light in the output ber located at the opposite side. Typical obstacle interrupters employ low-cost large-core plastic bers because of the short transmission distance. The minimum detectable size of the object is typically limited to 1 mm by the ber core diameter and the optical beam. The operating temperature range of commercially available sensors is typically 40 to +70C. Optical-ber sensors have been utilized in industry in many ways, such as: Detection of lot number and expiration datesfor example, in the pharmaceutical and food industries Color dierence recognitionfor instance, colored objects on a conveyer Defect detection, such as with missing wire leads in electronic components Counting discrete componentsfor example, bottles or cans Detecting absence or presence of labelsfor instance, packaging in the pharmaceutical and food industries

Fiber-optic sensors for monitoring process variables such as temperature, pressure, ow, and liquid level are also classied into two types: (1) the normally OFF type in which the shutter is inserted between the bers in the unactivated state. Thus, this type of sensor provides high and low levels as the light output corresponds to ON and OFF states, respectively, and (2) the normally ON type where the shutter is retracted from the gap in the inactivated state. In both types, the shutter is adjusted so it does not intercept the light beam completely but allows a small amount of light to be transmitted, even when fully closed. This transmitted light is used to monitor the cable (ber) continuity for faults and provides an intermediate state. Commercially available sensors employ bers of 200-m core diameter. The typical dierential attenuation, which determines the ON-OFF contrast ratio, is about 20 dB. According to manufacturers specications, these sensors operate well over the temperature range 40 to +80C with 2-dB variation in light output.

7.3. Temperature Sensors in Process Control

Temperature is one of the most important parameters to be controlled in almost all industrial plants, since it directly aects material properties and thus product quality. During the past few years, several temperature sensors have been developed for use in electrically or chemically hostile environments. Among these, the practical temperature sensors, which are now commercially available, are classied into two groups: (1) lowtemperature sensors with a range of 100 to +400C using specic sensing materials such as phosphors, semiconductors, and liquid crystals, and (2) high-temperature sensors with a range of 500 to 2000C based on blackbody radiation. 7.3.1. Semiconductor Absorption Sensors Many of these sensors can be located up to 1500 m away from the optoelectronic instruments. The operation of semiconductor temperature sensors is based on the temperature-dependent absorption of semiconductor materials. Because the energy and gap of most semiconductors decrease almost linearly with increasing temperature T , the band-edge wavelength
g (T )

corresponding to the fundamental optical absorption shifts toward

longer wavelengths at a rate of about 3 /C [for gallium arsenide (GaAs)] with T . As illustrated in Fig. 7.1, when a light-emitting diode with a radiation spectrum covering the wavelength g (T ) is used as a light source, the light intensity transmitted through a semiconductor decreases with T . Figure 7.2 shows the reection-type sensing element. A polished thin GaAs chip is attached to the ber end and mounted in a stainless-steel capillary tube of 2-mm diameter. The front face of the GaAs is antireection-coated, while the back face is gold-coated to return the light into the ber. The system conguration of the thermometer is illustrated in Fig. 7.3. In order to reduce the measuring errors caused by variations in parasitic losses, such as optical ber loss and connector loss, this thermosensor employs two LED sources [one aluminum gallium arsenide (AlGaAs), the other indium gallium arsenide (InGaAs)] with dierent wavelengths. A pair of optical pulses with dierent wavelengths along the ber. The light of
s

= 0.88 m and

= 1.3 m are

guided from the AlGaAs LED and the InGaAs LED to the sensing element
s

is intensity-modulated by temperature. On the

other hand, GaAs is transparent for the light of r , which is then utilized as a

reference light. After detection by a germanium avalanche photodiode (GeAPD), the temperature-dependent signal reference signal
r s

is normalized by the

in a microprocessor.

Figure 7.1. Operating principle of optical-ber thermometer based on temperature-dependent GaAs light absorption.

Figure 7.2. Sensing element of the optical-ber thermometer with GaAs light absorber.

Figure 7.3. System conguration of the optical-ber thermometer with GaAs light absorber.
The performance of the thermometer is summarized in Table 7.1. An accuracy of better than 2C is obtained within a range of 20 to +150C. The principle of operation for this temperature sensor is based on the temperature-dependent direct uorescent emission from phosphors. 7.3.2. Semiconductor Temperature Detector Using Photoluminescence The sensing element of this semiconductor photoluminescence sensor is a double-heterostructure GaAs epitaxial layer surrounded by two Al
x

Ga

1 x

As layers. When the GaAs absorbs the incoming exciting light, the electronhole pairs are generated in the GaAs layer. The electron-hole pairs combine

and reemit the photons with a wavelength determined by temperature. As illustrated in Fig. 7.4, the luminescent wavelength shifts monotonically toward longer wavelengths as the temperature T increases. This is a result of the decrease in the energy gap E
g

with T . Therefore, analysis of the

luminescent spectrum yields the required temperature information. The double heterostructure of the sensing element provides excellent quantum eciency for the luminescence because the generated electron-hole pairs are conned between the two potential barriers (Fig. 7.5). The system is congured as shown in Fig. 7.6. The sensing element is attached to the end of the silica ber (100-m core diameter). The excitation light from an LED, with a peak wavelength of about 750 nm, is coupled into the ber and guided to a special GRIN lens mounted to a block of glass. A rst optical inference lter IF 1, located between the GRIN lens and the glass block, reects the excitation light that is guided to the sensing element along the ber. However, this optical lter is transparent to the returned photoluminescent light. The reectivity of the second interference lter, IF 2, changes at about 900 nm. Because the peak wavelength of the luminescence shifts toward longer wavelengths with temperature, the ratio between the transmitted and reected light intensies if IF losses. The two lights separated by IF
2

changes. However, the ratio

is independent of any variation in the excitation light intensity and parasitic

2

are detected by photodiodes 1 and 2.

The detector module is kept at a constant temperature in order to eliminate any inuence of the thermal drift of IF 2.

Table 7.1. Characteristics of Semiconductor Sensors (Thermometer Performance)

Figure 7.4. Operating principle of optical-ber thermometer based on temperature-dependent photoluminescence from a GaAs epitaxial lm.

Figure 7.5. Sensing element of an optical-ber thermometer based on temperature-dependent photoluminescence.

Figure 7.6. Optical system of an optical-ber thermometer based on temperature-dependent photoluminescence.

The measuring temperature range is 0 to 200C, and the accuracy is 1C. According to the manufacturers report, good long-term stability, with a

temperature drift of less than 1C over a period of nine months, has been obtained. 7.3.3. Temperature Detectors Using Point-Contact Sensors in Process Manufacturing Plants Electrical sensors are sensitive to microwave radiation and corrosion. The need for contact-type temperature sensors have lead to the development of point-contact sensors that are immune to microwave radiation, for use in: (1) electric power plants using transformers, generators, surge arresters, cables, and bus bars; (2) industrial plants utilizing microwave processes; and (3) chemical plants utilizing electrolytic processes. The uses of microwaves include drying powder and wood; curing glues, resins, and plastics; heating processes for food, rubber, and oil; device fabrication in semiconductor manufacturing; and joint welding of plastic packages, for example. Semiconductor device fabrication is currently receiving strong attention. Most semiconductor device fabrication processes are now performed in vacuum chambers. They include plasma etching and stripping, ion implantation, plasma-assisted chemical vapor deposition, radio-frequency sputtering, and microwave-induced photoresist baking. These processes alter the temperature of the semiconductors being processed. However, the monitoring and controlling of temperature in such hostile environments is dicult with conventional electrical temperature sensors. These problems can be overcome by the contact-type optical-ber thermometer. 7.3.4. Noncontact SensorsPyrometers Because they are noncontact sensors, pyrometers do not aect the temperature of the object they are measuring. The operation of the pyrometer is based on the spectral distribution of blackbody radiation, which is illustrated in Fig. 7.7 for several dierent temperatures. According to the Stefan-Boltzmann law, the rate of the total radiated energy from a blackbody is proportional to the fourth power of absolute temperature and is expressed as:

Figure 7.7. Spectral distribution of blackbody radiation.

(7.2)
where is the Stefan-Boltzmann constant and has the value of 5.6697 10 W/m 2K
4. 8

The wavelength at which the radiated energy has its highest value is given by Wiens displacement law,

(7.3)
Thus, the absolute temperature can be measured by analyzing the intensity of the spectrum of the radiated energy from a blackbody. A source of measurement error is the emissivity of the object, which depends on the material and its surface condition. Other causes of error are deviation from the required measurement distance and the presence of any absorbing medium between the object and the detector. Use of optical bers as signal transmission lines in pyrometers allows remote sensing over long distances, easy installation, and accurate determination of the position to be measured by observation of a focused beam of visible light from the ber end to the object. The sensing head consists of a exible bundle with a large number of single bers and lens optics to pick up the radiated energy (Fig. 7.8). The use of a single silica ber instead of a bundle is advantageous for measuring small objects and longer distance transmission of the picked-up

radiated light. The lowest measurable temperature is 500C, because of the unavoidable optical loss in silica bers at wavelengths longer than 2 m. Air cooling of the sensing head is usually necessary when the temperature exceeds 1000C.

Figure 7.8. Schematic diagram of an optical-ber pyrometer.

Optical-ber pyrometers are one of the most successful optical-ber sensors in the eld of process control in manufacturing. Typical applications are: Casting and rolling lines in steel and other metal plants Electric welding and annealing Furnaces in chemical and metal plants Fusion, epitaxial growth, and sputtering processes in the semiconductor industry Food processing, paper manufacturing, and plastic processing

Figure 7.9 is a block diagram of the typical application of optical-ber pyrometers for casting lines in a steel plant, where the temperature distribution of the steel slab is measured. The sensing element consists of a linear array of fused-silica optical rods, thermally protected by air-purge cooling. Light radiated from the heated slabs is collected by the optical rods and coupled into a 15-m-long bundle of bers, which transmits light to the optical processing unit. In this system, each ber in the bundle carries the signal from a separate lens, which provides the temperature information at the designated spot of the slabs. An optical scanner in the processing unit scans the bundle, and the selected light signal is analyzed in two wavelength bands by using two optical interference lters.

Figure 7.9. Temperature distribution measurement of steel slabs by an optical-ber pyrometer using the two-wavelength method.

7.4. Pressure Sensors

If a pressure P acting on a diaphragm compresses a spring until an equilibrium is produced, the pressure can be represented as:

(7.4)
In this equation, F represents the force of the spring and A represents a surface area of the diaphragm. The movement of the spring is transferred via a system of levers to a pointer whose deection is a direct indication of the pressure (Fig. 7.10). If the measured value of the pressure must be transmitted across a long distance, the mechanical movement of the pointer can be connected to a variable electrical resistance (potentiometer). A change in the resistance results in a change in the measured voltage, which can then easily be evaluated by an electronic circuit or further processed. This example illustrates the fact that a physical quantity is often subject to many transformations before it is nally evaluated. 7.4.1. Piezoelectric Crystals Piezoelectric crystals may be utilized to measure pressure. Electrical charges are produced on the opposite surfaces of some crystals when they are mechanically loaded by deection, pressure, or tension. The electrical charge produced in the process is proportional to the eective force. This change in the charge is very small. Therefore, electrical ampliers are used to make it possible to process the signals (Fig. 7.11).

Pressure in this situation is measured by transforming it into a force. If the force produced by pressure on a diaphragm acts on a piezoelectric crystal, a signal which is proportional to the pressure measured can be produced by using suitable ampliers.

Figure 7.10. Deection as a direct indication of pressure.

Figure 7.11. Electrical ampliers are connected to a piezoelectric crystal.

Figure 7.12. Strain gauge for measurement of pressure.

7.4.2. Strain Gauges Strain gauges can also measure pressure. The electrical resistance of a wire-

type conductor is dependent, to a certain extent, on its cross-sectional area. The smaller the cross section (i.e., the thinner the wire), the greater the resistance of the wire. A strain gauge is a wire that conducts electricity and stretches as a result of the mechanical inuence (tension, pressure, or torsion) and thus changes its resistance in a manner that is detectable. The wire is attached to a carrier, which in turn is attached to the object to be measured. Conversely, for linear compression, which enlarges the crosssectional area of a strain gauge, resistance is reduced. If a strain gauge is attached to a diaphragm (Fig. 7.12), it will follow the movement of the diaphragm. It is either pulled or compressed, depending on the exure of the diaphragm.

7.5. Fiber-Optic Pressure Sensors

A Y-guide probe can be used as a pressure sensor in process control if a reective diaphragm, moving in response to pressure, is attached to the end of the ber (Fig. 7.13). This type of pressure sensor has a signicant advantage over piezoelectric transducers since it works as a noncontact sensor and has a high frequency response. The pressure signal is transferred from the sealed diaphragm to the sensing diaphragm, which is attached to the end of the ber. With a stainless-steel diaphragm about 100 m thick, hysteresis of less than 0.5 percent and linearity within 0.5 percent are obtained up to the pressure level of 3 10 the temperature range of 10 to +60C.
5

kg/m

(2.94 MPa) in

Figure 7.13. Schematic diagram of a ber-optic pressure sensor using a Y-guide probe with a diaphragm attached.

The material selection and structural design of the diaphragm are important to minimize drift. Optical-ber pressure sensors are expected to be used under severe environments in process control. For example, process slurries are frequently highly corrosive, and the temperature may be as high as 500C in coal plants. The conventional metal diaphragm exhibits creep at these high temperatures. In order to eliminate these problems, an all-fusedsilica pressure sensor based on the microbending eect in optical ber has been developed (Fig. 7.14). This sensor converts the pressure applied to the fused silica diaphragm into an optical intensity modulation in the ber. A pressure sensor based on the wavelength ltering method has been developed. The sensor employs a zone plate consisting of a reective surface, with a series of concentric grooves at predetermined spacing. This zone plate works as a spherical concave mirror whose eective radius of curvature is inversely proportional to the wavelength. At the focal point of the concave mirror, a second ber is placed which transmits the returned light to two photodiodes with dierent wavelength sensitivities. When broadband light is emitted from the rst ber to the zone plate, and the zone plate moves back and forth relative to the optical bers in response to the applied pressure, the wavelength of the light received by the second ber is varied, causing a change in the ratio of outputs from the two photodiodes. The ratio is then converted into an electrical signal, which is relatively unaected by any variations in parasitic losses.

Figure 7.14. Fiber-optic microbend sensor.

7.6. Nano-Positioning Capacitive Metrology Sensors

The nano-positioning capacitive sensor (Fig. 7.15) provides fast response, and precise trajectory control. It is capable of providing digital controllers

with a fast ber-link interface and an ID-chip for automatic calibration functions. It provides a resolution of 0.1 nm. 7.6.1. Nano-Capacitive Positioning Sensors Single-electrode capacitive sensors are direct metrology devices. They use an electric eld to measure the change in capacitance between the probe and a conductive target surface, without any physical contact. This makes them free of friction, and hysteresis, and provides high-phase delity and bandwidth. The selectable bandwidth setting allows the user to adapt the system to dierent applications. For the highest accuracy and sub-nanometer resolution, the bandwidth can be limited to 10 Hz.

Figure 7.15. A nano-positioning capacitive sensor.

For high-dynamics measurements, a bandwidth of up to 6.6 kHz is possible, with a resolution still in the 1-nm range. The user can choose a measurement range from 20 to 500 m, depending on the nominal measurement range of the selected sensor. The ten-channel system provides dierent extended measuring ranges for each selected sensor. The capacitive sensors measuring capacitance is the metrology system for most nano-positioning applications. They are absolute-measuring highdynamics devices. The capacitive sensors / control electronics use a high-

frequency AC excitation signal for enhanced bandwidth and drift-free measurement stability (Fig. 7.16). The electronics of the capacitive position sensor incorporate an innovative design providing superior linearity, low sensitivity to cable capacitance, low background noise, and low drift. The Integrated Linearization System (ILS) compensates for inuences caused by errors, such as non-parallelism of the plates. A comparison between a conventional capacitive position sensor system and the results obtained with the ILS is shown in Fig. 7.17. When nano-capacitance sensors are used with digital controllers, which add polynomial linearization techniques, a positioning linearity of up to 0.003 percent is achievable. The travel range is 15 m; the gain 1.5 m/V. Linearity is better than 0.02 percent; even higher linearity is achievable with digital controllers. Figure 7.18 shows the linearity of a piezo exure nano-positioning stage with an integrated capacitive position sensor operated in closed-loop mode with an analog controller. All errors contributed by the mechanics, PZT drive, sensors and electronics are included in the resulting linearity of better than 0.02 percent. Even higher linearity is achievable with digital controllers.

Figure 7.16. A 15-m piezo nano-positioning stage with an integrated capacitive position sensor.

Figure 7.17. The linearity of a conventional capacitive position sensor system versus ILS (Integrated Linearization System), shown before digital linearization.
The long-term stability of the nano-capacitive position sensor and electronics design is shown in Fig. 7.19. They enable direct measurement of the moving platform and are especially well-suited for multiaxis measurements in parallel metrology congurations. Parallel metrology means that the controller monitors all controlled degrees of freedom relative to ground, a prerequisite for active trajectory control. The capacitive sensors also provide high linear accuracy.

Figure 7.18. Linearity of a 15-m piezo nano-positioning stage operated with advanced control electronics.

Figure 7.19. Measurement stability of a capacitive position sensor control board with a 10-pF reference capacitor over 3.5 hours.

7.7. Electrode Geometry, Sensor Surface Flatness, and Finish

During sensor production, great care must be taken to maintain critical mechanical tolerances. Measuring surfaces are diamond-tool machined using sophisticated process control techniques. The result is the smooth, ultra-at, mirrored surfaces required to obtain the highest resolution commercially available. The target and probe plates are congured in such a way that the sensor capacitance in air is 10 pF at the nominal measuring distance. This means that one sensor electronics module can operate all sensor range models without modication.

7.8. Special Design Eliminates Cable Inuences

When measuring distance by detection of capacitance changes, uctuations in the cable capacitance can have an adverse eect on accuracy. This is why most capacitive measurement systems only provide satisfactory results with short well-dened cable lengths. Nano-capacitance systems use a special design that eliminates cable inuences, permitting use of cable lengths of up to 3 m without diculty. For optimum results, it is recommended that calibration of the sensor-actuator system be done by Metrology Laboratories. Longer distances between sensor and electronics can be spanned with special loss-free digital transmission protocols.

7.9. Materials Achieving Greater Accuracy

The best measurements are obtained when the coecient of thermal expansion of the sensor and the substrate to which it is axed are as nearly

equal as possible (Fig. 7.20a). If they dier, temperature changes lead to mechanical stress and deformation, which can compromise accuracy (Fig. 7.20.b). It is the material choice that aects performance, not the absolute value of the temperature coecient. A low-temperature coecient in a sensor mounted on a stainless steel stage will thus give poorer results than a steel sensor (Fig. 7.20b).

Figure 7.20. (a) The optimum match between sensor material and environment. (b) The CTE of sensor material and environment are dierent. Buckling can reduce sensor atness and accuracy.

7.10. Mounting, Calibration, and Measuring Ranges

The sensors should be mounted with a mid-range distance between the plates equal to the nominal measuring range (Fig. 7.21). The measuring range then comprises 50 to 150 percent of this distance. The corresponding range at the output of the sensor electronics is 10 volts in width. The probe/target capacitance at a distance equal to the nominal measuring range equals that of the 10-pF reference capacitor in the electronics. The nominal range itself can be increased or decreased by selecting a dierent reference capacitor in the electronics (Fig. 7.22).

Figure 7.21. The distance between the probe and target at the center position, with closest spacing and maximum separation.

Figure 7.22. Measuring ranges of dierent capacitive position sensors, standard ranges are 1- D-015, 3- D-050, and 4- D-100, while the extended ranges are 2- D016, 5- D050, and 6- D-100.

7.11. Parallelism of Measuring Surfaces

For optimum results, target and probe plates must remain parallel to each other during measurement (Figs. 7.23a and 7.23b). Positioning systems with multilink exure guidance reduce tip and tilt to negligible levels (Fig. 7.24) and achieve outstanding accuracy. This eect is negligible in nano-positioning systems with microradian-range guiding accuracy (Fig. 7.24). Flexure-guided nano-positioning systems oer sub-micro-radian guiding accuracy and are optimally suited for capacitive sensors (Fig. 7.24).

7.12. Electronics Support

Standard nano-capacitive position sensor electronics are supported by all digital controllers. Inputs are developed on I/O controller cards and on several OEM controllers for high-speed parallel port controllers (Fig. 7.25). In addition, nano-capacitive sensors may be developed to accommodate up to

seven channels to meet specic needs. Various sensor probes are available at 100 m, 50 m, and 20 m and are developed for nominal measurement range (Fig. 7.26).

Figure 7.23. (a) Relative sensor gain versus tilt. (b) Nonlinearity versus tilt.

Figure 7.24. Nano-positioning systems with a micro-radian-range.

Figure 7.25. A high-speed nano automation controller board.

Figure 7.26. High-precision capacitive sensor probes with signal conditioner electronics.
High-speed nano automation signal conditioner electronics are specially designed for single-electrode capacitive position sensor probes. They provide analog output with very high linearity, exceptional long-term stability, subnanometer position resolution, and have bandwidths of up to 6.6 kHz (Fig. 7.27).

Figure 7.27. The output linearity error of a high-speed nano controller signal conditioner.

7.13. Sensor Installation

The simple installation of the single-electrode probes is facilitated by the LED-bar indicating the optimum gap between probe and target. 7.13.1. Factory Calibration for Improved Linearity The highest possible linearity and accuracy are achieved with factory calibration of the sensor probe, together with the signal conditioner electronics. Two measurement ranges can be calibrated at the same time for one particular sensor probe. Factory calibration also optimizes parameters like ILS (linearization), gain, and oset and eliminates cable capacitance inuences.

7.14. Using the Integrated Linearization System (ILS) for Highest Accuracy
A linearization circuit compensates the inuences of parallelism errors between sensor and target to ensure excellent measuring linearity to 0.1 percent. 7.14.1. Multichannel Measurements The sensor electronics are equipped with I/O lines for the synchronization of multiple sensor systems.

7.15. Displacement Sensors for Robotic Applications

The operating principle of a displacement sensor using Y-guide probes is illustrated in Fig. 7.28. The most common Y-guide probe is a bifurcated ber bundle. The light emitted from one bundle is back-reected by the object to be measured and collected by another bundle (receiving bers). As a result, the returned light at the detector is intensity-modulated to a degree dependent on the distance between the end of the ber bundle and the object. The sensitivity and the dynamic range are determined by the geometrical arrangement of the array of ber bundles and by both the number and type of the bers. Figure 7.29 shows the relative intensity of the returned light as a function of distance for three typical arrangements: random, hemispherical, and concentric circle arrays. The intensities increase with distance and reach a peak at a certain discrete distance. After that, the intensities fall o very slowly. Most sensors use the high-sensitivity regions in these curves. Among the three arrangements, the random array has the highest sensitivity but the narrowest dynamic range. The displacement sensor using the Y-guide probe provides a resolution of 0.1 m,

Figure 7.28. The principle of operation of a ber-optic mechanical sensor using a Y-guide probe.

Figure 7.29. Relative intensity of returned light for three ber-optic arrangements
linearity within 5 percent, and a dynamic range of 100 m displacement. Yguide probe displacement sensors are well-suited for robotics applications as position sensors and for gauging and surface assessment since they have high sensitivity to small distances. One profound problem of this type of displacement sensor is the measuring error arising from the variation in parasitic losses along the optical transmission line. Recalibration is required if the optical path is interrupted, which limits the range of possible applications. In order to overcome this problem, a line-loss-independent displacement sensor with an electrical subcarrier phase encoder has been implemented. In this sensor, the light from an LED modulated at 160 MHz is coupled into the ber bundle and divided into two optical paths. One of the paths is provided with a xed retroreector at its end. The light through the other is reected by the object. The two beams are returned to the two photodiodes separately. Each signal, converted into an electric voltage, is electrically heterodyned into an intermediate frequency at 455 kHz. Then, the two signals are fed to a digital phase comparator, the output of which is proportional to the path distance. The resolution of the optical path dierence is about 0.3 mm, but improvement of the receiver electronics will provide a higher resolution.

7.16. Process Control Sensors Measuring and Monitoring Liquid

Flow
According to the laws of uid mechanics, an obstruction inserted in a ow stream creates a periodic turbulence behind it. The frequency of shedding the turbulent vortices is directly proportional to the ow velocity. The ow sensor in Fig. 7.30 has a sensing element consisting of a thin metallic obstruction and a downstream metallic bar attached to a multimode bermicrobend sensor. As illustrated in Fig. 7.31, the vortex pressure produced at the metallic bar is transferred, through a diaphragm at the pipe wall that serves as both a seal and a pivot for the bar, to the microbend sensor located outside the process line pipe. The microbend sensor converts the timevarying mechanical force caused by the vortex shedding into a corresponding intensity modulation of the light. Therefore, the frequency of the signal converted into the electric voltage at the detector provides the ow-velocity information. This ow sensor has the advantage that the measuring accuracy is essentially independent of any changes in the uid temperature, viscosity, or density, and in the light source intensity. According to the specications for typical optical vortex-shedding ow sensors, ow rate can be measured over a Reynolds number range from 5 10
3

to 6000 10

at temperatures

from 100 to +600C. This range is high compared to that of conventional ow meters. In addition, an accuracy of 0.4 and 0.7 percent, respectively, is obtained for liquids and gases with Reynolds numbers above 10,000.

Figure 7.30. Principle of operation of a vortex-shedding ow sensor.

Figure 7.31. Schematic diagram of a vortex-shedding ow sensor.

7.16.1. Flow Sensors Detecting Small Air Bubbles for Process Control in Manufacturing Another optical-ber ow sensor employed in manufacturing process control monitors a two-uid mixture (Fig. 7.32). The sensor can distinguish between moving bubbles and liquid in the ow stream and display the void fraction, namely, the ratio of gas volume to the total volume. The principle of operation is quite simple. The light from the LED is guided by the optical ber to the sensing element, in which the end portion of the ber is mounted in a stainless steel needle with a 2.8-mm outer diameter. When liquid is in contact with the end of the ber, light enters the uid eciently and very little light is returned. However, when a gas bubble is present, a signicant fraction of light is reected back. With this technique, bubbles as small as 50 m may be detected with an accuracy of better than 5 percent and a response time of only 10 s.

Figure 7.32. Flow sensor for two-phase mixtures.

Potential applications of this ow sensor for the control of processes in manufacturing systems are widespreadfor example, detection of gas plugs in production wells in the oil industry and detection of fermenters and distillers in the blood-processing and pharmaceutical industries. An optical-ber ow sensor for a two-phase mixture based on Y-guide probes is shown in Fig. 7.33. Two Y-guide probes are placed at dierent points along the ow stream to emit the input light and pick up the retroreected light from moving solid particles in the ow. The delay time between the signals of

the two probes is determined by the average velocity of the moving particles. Therefore, measurement of the delay time by a conventional correlation technique provides the ow velocity. An accuracy of better than 1 percent and a dynamic range of 20:1 are obtained for ow velocities up to 10 m/s. A potential problem of such ow sensors for two-phase mixtures is poor longterm stability, because the optical bers are inserted into the process uid pipes.

Figure 7.33. Flow sensor using two Y-guided probes based on a correlation technique.

7.16.2. Liquid Level Sensors in Manufacturing Process Control for Petroleum and Chemical Plants Several optical-ber liquid level sensors developed in recent years have been based on direct interaction between the light and liquid. The most common method in commercial products employs a prism attached to the ends of two single optical bers (Fig. 7.34). The input light from an LED is totally internally reected and returns to the output ber when the prism is in air. However, when the prism is immersed in liquid, the light refracts into the uid with low reection, resulting in negligible returned light. Thus, this device works as a liquid level switch. The sensitivity of the sensor is determined by the contrast ratio, which depends on the refractive index of the liquid. Typical examples of signal output change for liquids with dierent refractive indices are indicated in Table 7.2. The output loss stays at a constant value of 33 dB for refractive indices

higher than 1.40. The signal output of a well-designed sensor can be switched for a change in liquid level of only 0.1 mm.

Figure 7.34. Principle of operation of a liquid level sensor with a prism attached to two optical bers. Table 7.2. Refractive Index versus Output
Refractive Index, n 1.333 1.366 1.380 1.395 Loss Change, dB 2.1 4.6 6.0 31.0

Figure 7.35. (a) Principle of operation of a liquid level sensor for the measurement of boiler-drum water. (b) Liquid level sensor for the measurement of boiler-drum water with ve-port sensors.
Problems to be solved for this sensor are dirt contamination on the prism surface and bubbles in the liquid. Enclosing the sensing element with a ne lter helps keep it clean and simultaneously reduces level uctuations caused by bubbles. Since optical-ber liquid level sensors have the advantages of low cost and electrical isolation, their use is widespread in

petroleum and chemical plants, where the hazardous environment causes diculties with conventional sensors. They are used, for example, to monitor storage tanks in a petroleum plant. Another optical-ber liquid level sensor, developed for the measurement of boiler-drum water levels, employs a triangularly shaped gauge through which red and green light beams pass. The beams are deected as it lls with water so that the green light passes through an aperture. In the absence of water, only red light passes through. Optical bers transmit red or green light from individual gauges to a plant control room located up to 150 m from the boiler drum (Fig. 7.35). The water level in the drum is displayed digitally. This liquid level sensor operates at temperatures up to 170C and pressures up to 3200 lb/in
2

gauge. Many sensor units are installed in the boiler drum,

and most have been operating for seven years. This sensor is maintenancefree, fail-safe, and highly reliable.

7.17. Sensory MEMS Enable Certain Molecules to Signal Breast Cancers Spread
Metastasis is responsible for approximately 90 percent of all cancer-related deaths. Sensory MEMS technology enabled researchers to uncover how breast tumors use a particular type of molecule to promote metastasis. It also revealed that tumor cells learn to exploit cytokines to promote the spread of breast cancer. The work of several scientists examine how cells in the body communicate with each other through cytokines, signaling molecules that direct a wide range of activities such as cell growth and movement. One important cytokine transforming growth factor (TGF) normally suppresses tumor development. However, according to the ndings, cancer cells in humans are able to misuse these cytokines for their own gain by compelling TGF to enhance a tumors ability to spread instead of suppressing it. Using computer-based analysis to classify patient tumor samples based on their levels of TGF, it was observed that about half of all breast tumors contained active TGF. The aected tumors were found to be more aggressive and more likely to metastasize to the lung during the course of the patients disease.

Using mice for the next set of experiments, it was discovered that TGF prompts breast cancer cells to make a second cytokine, known as

angiopoietin-like 4 (ANGPTL4), which enhances the ability of the cancer to

spread to the lungs through the blood circulation. The results show that the breast cancer cells use ANGPTL4 to break down the thin capillaries of the lung, thus facilitating their escape into the lung tissue. TGF enhances human breast cancer metastasis and reveals how tumor cells learn to exploit cytokines by making them work as a relay system to promote the spread of breast cancer. The researchers are currently seeking to determine whether TGF and ANGPTL4 may also be active in other types of tumors, and are evaluating ways to interfere with the action of these cytokines to prevent metastasis in cancer patients. Deciphering how cancer cells take advantage of these cytokines is essential for developing therapies that can prevent this process since cytokines that act outside of cells can be more easily targeted by drugs that block their activity. The study provides support for developing agents that interfere with TGF in order to prevent and treat cancer metastasis. It points at ANGPTL4 as a possible target to interrupt the TGF stimulus of metastasis without interfering with the molecules benecial eects. Several pharmaceutical companies are currently testing TGF-blocking compounds in clinical trials as candidate drugs against breast cancer, melanoma, and other types of cancer.

7.18. On-Line Measuring and Monitoring of Gas by Spectroscopy

An optical spectrometer or optical ltering unit is often required for chemical sensors because the spectral characteristics of absorbed, uorescent, or reected light indicate the presence, absence, or precise concentration of a particular chemical species (Fig. 7.36). Sensing of chemical parameters via bers is usually done by monitoring changes in a suitably selected optical propertyabsorbance, reectance, scattering (turbidity), or luminescence (uorescence or phosphorescence), depending on the particular device. Changes in parameters such as the refractive index may also be employed for sensing purposes. The change in light intensity due to absorption is determined by the number of absorbing species in the optical path, and is related to the concentration C of the

absorbing species by the Beer-Lambert relationship. This law describes an exponential reduction of light intensity with distance (and also concentration) along the optical path. Expressed logarithmically,

Figure 7.36. (a) Light scattering from the waveguide surface increases as the antigen-antibody complexes are formed, and is detected by a side-mounted photodetector. (b) Fluorescence is excited in a uorescent marker, attached to an antigen molecule, by light passing through the waveguide. The uorescent light can be collected either sideways from the guide, or as light that is retrapped by the guide and directed to a photodetector. (c) Absorption of light by antibodyantigen complexes on the surface attenuates light traveling down the

waveguide.
where A is the optical absorbance, l is the path length of the light, is the molar absorptivity, and I
0

and I are the incident and transmitted light,

respectively. For absorption measurements via optical bers, the medium normally must be optically transparent. An accurate method for the detection of leakage of ammable gases such as methane (CH 4), propane (C 3H 8), and ethylene (C 2H 4) is vital in gas and petrochemical plants in order to avoid serous accidents. The recent introduction of low-loss ber into spectroscopic measurements of these gases oers many advantages for process control in manufacturing: Long-distance remote sensing On-line measurement and monitoring Low cost High reliability

The most commonly used method at present is to carry the sample back to the measuring laboratory for analysis; alternatively, numerous spectrometers may be used at various points around the factory. The new advances in spectroscopic measurements allow even CH 4 to be observed at a distance of 10 km with a detection sensitivity as low as 5 percent of the lower explosion

limit (LEL) concentration. The optical-ber gas measuring system employs an

absorption spectroscopy technique, with the light passing through a gasdetection cell for analysis. The overtone absorption bands of a number of ammable gases are located in the near-infrared range (Fig. 7.37). The optical gas sensing system can deal with a maximum of 30 detection cells (Fig. 7.38). The species to be measured are CH 4, C 3H 8, and C 2H 4 molecules. Light from a halogen lamp (an infrared light source) is distributed into a bundle of 30 single optical bers. Each of the distributed beams is transmitted through a 1-km length of ber to a corresponding gas detection cell. The receiving unit is constructed of three optical switches, a rotating sector with four optical interference lters, and three Ge photodiodes. Each optical switch can select any ten returned beams by specifying the number of the cell. The peak transmission wavelength of the optical lter incorporated

in the sensor is 1.666 m for CH 4, 1.690 m for C 3H 8, 1.625 m for C 2H 2, and 1.600 m for a reference beam. After conversion to electrical signals, the signal amplitudes for the three gases are normalized by the reference amplitude. Then, the concentration of each gas is obtained from a known absorption-concentration calibration curve stored in a computer.

Figure 7.37. Absorption lines of typical ammable gases in the nearinfrared, and the transmission loss of silica ber.

Figure 7.38. A gas detection system with 30 detection cells.

An intrinsic distributed optical-ber gas sensor for detecting the leakage of cryogenically stored gases such as CH 4, C 2H 4, and N 2 has also been developed. The sensors operation is based on the temperature-dependent transmission loss of optical berthat is, the optical ber is specially

designed so that the transmission loss increases with decreasing temperature by choosing the appropriate core and cladding materials. Below the critical temperature, in the region of 55C, most of the light has transferred to the cladding layer, and the light in the core is cut o. By connecting this temperature-sensitive ber between a light source and a detector and monitoring the output light level, the loss of light resulting from a cryogenic liquid in contact with the ber can be detected directly.

7.19. Using Avalanche Photodiodes to Improve System Performance

Avalanche photodiode detectors (APD) are used in many diverse
applications, such as laser range nders (Fig. 7.39), data communications, and photon correlation studies. This section of the chapter discusses APD structures, critical performance parameters, and excess noise factors. For low-light detection in the 200- to 1150-nm range, the designer has three basic detector choices: the silicon PIN detector, the silicon avalanche photodiode (APD), and the photomultiplier tube (PMT).

Figure 7.39. Avalanche photodiode detectors (APD)

APDs are widely used in instrumentation and aerospace applications, oering a combination of high speed and high sensitivity that is unmatched by PIN detectors, and quantum eciencies at >400 nm that is unmatched by PMTs. Providing a guideline to dening and selecting avalanche photodiodes should oer the user of spectroscopy sensing devices the fundamental tools needed to apply minimum signal-to-noise spectra and would include the following: Explanation of an avalanche photodiode How to select an APD

Discussion of excess noise factors Explanation of Geiger Mode Possible applications

7.20. Structures of Avalanche PhotodiodesAPD Structures

In order to understand why more than one APD structure exists, it is important to appreciate the design tradeos that must be accommodated by the APD designer. The ideal APD would have zero dark noise, no excess noise, a broad spectral and frequency response, a gain range from 1 to 106 or more, and a low cost. More simply, an ideal APD would be a good PIN photodiode with gain! In reality, however, this is dicult to achieve because of the need to trade o conicting design requirements. What some of these tradeos are, and how they are optimized in commercially available APDs, are listed next.

Figure 7.40. Schematic cross-section for a typical APD structure.

Consider the schematic cross-section for a typical APD structure (Fig. 7.40). The basic structural elements provided by the APD designer include an absorption region A, and a multiplication region M. Present across region A is an electric eld E that serves to separate the photogenerated holes and electrons, and sweeps one carrier towards the multiplication region. The multiplication region M is designed to exhibit a high electric eld so as to

provide internal photocurrent gain by impact ionization. This gain region must be broad enough to provide a useful gain, M, of at least 100 for silicon APDs, or 1040 for germanium or InGaAs APDs. In addition, the multiplying electric eld prole must enable eective gain to be achieved at eld strength below the breakdown eld of the diode. Figure 7.40 shows the reach-through structure, which provides a combination of high speed, low noise, capacitance, and extended IR response. The multiplication region M is designed to exhibit a high electric eld so as to provide internal photocurrent gain by impact ionization. A variation of the reach-through structure is epitaxial silicon APD (EPI-APD). The EPI-APD allows simpler and cheaper manufacturing methods, resulting in a lower-cost device relative to the standard reach-through device. The tradeo is a higher k factor, resulting in a higher excess noise factor for equivalent gain (Table 7.3).

Table 7.3. Typical Values of k, X, and F for Si, Ge, and InGaAs APDs
Excess Noise (Factor at Detector Type Ionization Ratio (k) Silicon (reachthrough structure) Silicon Epitaxial APDs Silicon (SLiK low-k structure) 0.002 0.17 500 3.0 0.06 0.45 100 7.9 0.02 X-Factor 0.3 Typical Gain (M) 150 Typical Gain) (F) 4.9

Germanium InGaAs

0.9 0.45

0.95 0.70.75

10 10

9.2 5.5

7.21. Critical Performance Parameters

An APD diers from a PIN photodiode by providing internal photoelectronic signal gain. Therefore, output signal current, I s, from an APD equals I M.Ro () P
s, s

where Ro() is the intrinsic responsivity of the APD at a gain

s

M = 1 and wavelength l, M is the gain of the APD, and P

is the incident
R

optical power. The gain is a function of the APDs reverse voltage, V shown in Fig. 7.41.

, and

will vary with applied bias. A typical gain-voltage curve for a silicon APD is

The gain as a function of the bias voltage varies with the structure of the APD. One of the key parameters to consider when selecting an APD is the detectors spectral noise. Like other detectors, an APD will normally be operating in one of two noise-limited detection regimes; either detector noise limited at low power levels, or photon shot noise limited at higher powers. Since an APD is designed to be operated under a reverse bias, sensitivity at low light levels will be limited by the shot noise and the APDs leakage current. Shot noise derives from the random statistical Poissonian uctuations of the dark current, I D (or signal current). Dark current shot noise (IN - SHOT) is normally given by I N(SHOT)= (2.q.B.I D )
1/2

, for a PIN
(4)

detector, where B is the system bandwidth. This diers for an APD, however, as bulk leakage current, I DB, is multiplied by the gain, M, of the APD leakage current I D is therefore equal to: . Total

(1)
where I DS is the surface leakage current. In addition, the avalanche process statistics generate current uctuations, and APD

performance is degraded by an excess noise factor (F) compared to a PIN, equation 4. The total spectral noise current for an APD in dark conditions is thus given by equation (1), ID = IDS + IDB M: where I DS is the surface leakage current.

(2)
where q is the electron charge. At higher signal light levels, the detector transitions to the photon shot noise limited regime where sensitivity is limited by photon shot noise on the current generated by the optical signal. Total noise from the APD in illuminated conditions will therefore equal the quadratic sum of the detector noise plus the signal shot noise. For a given optical signal power, P
s,

this is given by:

(3)
In the absence of other noise sources, an APD therefore provides a signal-tonoise ratio (SNR), which is worse than a PIN detector with the same quantum eciency. Noise equivalent power (NEP) cannot be used as the only measure of a detectors relative performance, but rather detector signal-to-noise (SNR) at a specic wavelength and bandwidth should be used to determine the optimum detector type for a given application. The optimum signal-to-noise occurs at a gain M, where total detector noise equals the input noise of the amplier or load resistor. The optimum gain depends in part on the excess noise factor, F, of the APD, and ranges from M = 50 to 1000 for silicon APDs, and is limited to M = 10 to 40 for germanium and InGaAs APDs.

7.22. Selecting an APD

APDs are generally recommended for high-bandwidth applications, or where internal gain is needed to overcome high preamplier noise. The following is a simple guide that can be used to decide if an APD is the most appropriate for ones light detection requirements.

Determine the wavelength range to be covered. (See the following section Types of APDs, to determine the specic APD type useful for the wavelength range to be covered.) Determine the minimum size of the detector that can be used in the optical system. Eective optics can often be more cost-eective than the use of an overly large PIN or avalanche photodetectors. Determine the required electrical frequency bandwidth of the system; again, over-specifying bandwidth will degrade the SNR of the system.

7.22.1. Types of APDs Avalanche photodiodes are commercially available that span the wavelength range from 300 nm to 1700 nm. Silicon APDs can be used between 300 nm to 1100 nm, germanium between 800 nm and 1600 nm, and InGaAs from 900 nm to 1700 nm. Although signicantly more expensive than germanium APDs, InGaAs APDs are typically available with much lower noise currents, exhibit an extended spectral response to 1700 nm, and provide higher frequency bandwidth for a given active area. A germanium APD is recommended for applications in high electromagnetic interference (EMI) environments, where amplier noise is signicantly higher than the noise from an InGaAs APD, or for applications where cost is a primary consideration. 7.22.2. Understanding the Specications Responsivity and Gain APD gain will vary as a function of applied reverse voltage, as shown in Fig. 7.41. In addition, for many APDs, it is not possible, or practical, to make an accurate measurement of the intrinsic responsivity: Ro(), at a gain M = 1. It is therefore inappropriate to state typical gain and diode sensitivity at M = 1 as a method for specifying diode responsivity at a given operating voltage.

Figure 7.41. A typical gain-voltage curve for a silicon APD.

In order to characterize APD response, one must specify APD responsivity (in amps/watt) at a given operating voltage. However, because of diode-to-diode variations in the exact gain voltage curve of each APD, the specic operating voltage for a given responsivity will vary from one APD to another. Manufacturers should therefore specify a voltage range within which a specic responsivity will be achieved. An example of a typically correct specication for diode responsivity, in this case for an InGaAs APD, is as follows:

7.23. Dark Current and Noise Current

As can be seen from the noise equation (Equation 2), the total APD dark current (and the corresponding spectral noise current) is only meaningful when specied at a given operating gain. Dark current at M = 1 is dominated by surface current, and may be signicantly less than

Since APD dark and spectral noise currents are a strong function of APD gain, these should be specied at a stated responsivity level. An example of a typically correct specication for diode dark current and noise currentin this case, for an InGaAs APDis as follows:

7.23.1. Excess Noise Factor All avalanche photodiodes generate excess noise due to the statistical nature of the avalanche process. This excess noise factor is generally denoted as F . As shown in the noise equation (Equation 2), F is the factor by which the statistical noise on the APD current (equal to the sum of the multiplied photocurrent plus the multiplied APD bulk dark current) exceeds that which would be expected from a noiseless multiplier on the basis of Poissonian statistics (shot noise) alone. The excess noise factor is a function of the carrier ionization ratio, k, where (k) is usually dened as the ratio of the hole to electron ionization probabilities. The excess noise factor may be calculated using the model developed by McIntyre (3), which considers the statistical nature of avalanche multiplication. The excess noise factor is given by:

(4)
Therefore, the lower the values of k and M, the lower the excess noise factor. The eective k factor (k
EFF)

for an APD can be measured experimentally by

tting the McIntyre formula to the measured dependence of the excess noise factor on gain. This is best done under illuminated conditions. It may also be theoretically calculated from the carrier ionization coecients

and the electric eld prole of the APD structure. The ionization ratio k is a strong function of the electric eld across the APD structure, and takes its lowest value at low electric elds (only in silicon). Since the electric eld prole depends upon the doping prole, the k factor is also a function of the doping prole. Depending on the APD structure, the electric eld prole traversed by a photogenerated carrier and subsequent avalanche-ionized carriers may therefore vary according to photon absorption depth. For indirect band gap semiconductors such as silicon, the absorption coecient varies slowly at the longer wavelengths, and the mean absorption depth is therefore a function of wavelength. The value of k
EFF,

and gain, M, for a silicon APD is thus a function of

wavelength for some doping proles. The McIntyre formula can be approximated for a k < 0.1 and M > 20 without signicant loss of accuracy as:

(5)
Also often quoted by APD manufacturers is an empirical formula used to calculate the excess noise factor, given as:

(6)
where the value of X is derived as a log-normal linear t of measured F-values for given values of gain M. This approximation is suciently appropriate for many applications, particularly when used with APDs with a high k factor, such as InGaAs and germanium APDs. Table 7.3 provides typical values of k, X, and F for silicon, germanium, and InGaAs APDs. The silicon APDs have three dierent values of ionization ration: The Super-Low k (SLiK) APDs used in photon counting modules The high-performance reach-through structures with k = 0.02 for applications requiring extremely low noise APD and high gain

The low-cost silicon epitaxial APD with k = 0.06 ideal for high SNR applications.

In germanium and InGaAs APDs, the k-value is generally quoted at M = 10, which somewhat overestimates F at M < 10 and underestimates F at M > 10.

7.24. The Geiger Mode

In the Geiger mode, an APD is biased above its breakdown voltage (V
BR )

>V

for operation at very high gain (typically 10

to 10 6).

When biased above breakdown, an APD will normally conduct a large current. However, if this current is limited to less than the APDs latching current, there is a strong statistical probability that the current will uctuate to zero in the multiplication region, and the APD will then remain in the o state until an avalanche pulse is triggered by either a bulk or photogenerated carrier. If the number of bulk carriergenerated pulses is low, the APD can therefore be used to count individual current pulses from incident photons. The value of the bulk dark current is therefore a signicant parameter in selecting an APD for photon-counting, and can be reduced exponentially by cooling.

7.25. Crack Detection Sensors for Commercial, Military, and Space Industry Use
Accurate and precise detection of crack propagation in aircraft components is of vital interest for commercial and military aviation and the space industry. A system has been recently developed to detect cracks and crack propagation in aircraft components. This system uses optical bers of small diameter (20 to 100 m), which can be etched to increase their sensitivity. The bers are placed on perforated adhesive foil to facilitate attachment to the desired component for testing. The ber is in direct contact with the component (Fig. 7.42). The foil is removed after curing of the adhesive. Alternatively, in glass-ber-reinforced plastic (GFRP) or carbon-berreinforced plastic (CFRP), materials that are used more and more in aircraft design, the ber can be easily inserted in the laminate without disturbing the normal fabrication process. For these applications, bare single ber or prefabricated tape with integrated bundles of bers is used. The system was

initially developed for fatigue testing of aircraft components such as frames, stringers, and rivets. In monitoring mode, the system is congured to automatically interrupt the fatigue test. The system has also been applied to the inspection of the steel rotor blades of a 2-MW wind turbine. A surveillance system has been developed for the centralized inspection of all critical components of the Airbus commercial jetliner during its lifetime. This ber nervous system is designed for in-ight monitoring and currently is accessible to ight and maintenance personnel.

Figure 7.42. An illustration of ber to be in direct contact with the component.

An optical-ber mesh has been tested for a damage assessment system for a GFRP submarine sonar dome. Two sets of orthogonally oriented bers are nested in the laminate during the fabrication process. When the bers of the mesh are properly connected to LEDs and the detectors, the system can be congured to visualize the location of a damaged area. As an alternative, a video camera and image processing are applied to determine the position of the damaged area. The ber end faces at the detection side of the mesh are bundled and imaged into the camera tube. Two images are subtracted: the initial image before the occurrence of

damage and the subsequent image. If bers are broken, their location is highlighted as a result of this image subtraction.

7.26. Control of the Input/Output Speed of Continuous Web Fabrication Using Laser Doppler Velocity Sensor
A laser Doppler velocimeter (LDV) can be congured to measure any desired component velocity, perpendicular or parallel to the direction of the optical axis. An LDV system has been constructed with a semiconductor laser and optical bers and couplers to conduct the optical power. Frequency modulation of the semiconductor laser (or, alternatively, an external beroptic frequency modulator) is used to introduce an oset frequency. Some commercial laser Doppler velocimeters are available with optical-ber leads and small sensing heads. However, these commercial systems still use bulk optical components such as acoustooptic modulators or rotating gratings to introduce the oset frequency. With an LDV system, the velocity can be measured with high precision in a short period of time. This means that the method can be applied for real-time measurements to monitor and control the velocity of objects as well as measure their vibration. Because the laser light can be focused to a very small spot, the velocity of very small objects can be measured, or if scanning techniques are applied, high spatial resolution can be achieved. This method is used for various applications in manufacturing, medicine, and research. The demands on system performance with respect to sensitivity, measuring range, and temporal resolution are dierent for each of these applications. In manufacturing processes, for example, LDV systems are used to control continuous roll milling of metal (Fig. 7.43), control the rolling speed of paper and lms, and monitor uid velocity and turbulence in mixing processes. Another industrial application is vibration analysis. With a noncontact vibrometer, vibration of machines, machine tools, and other structures can be analyzed without disturbing the vibrational behavior of the structure.

Figure 7.43. A ber-optic laser Doppler velocimeter at a rolling mill controls pressure by measuring input speeds.

Figure 7.44. Special probe for measurement of blood velocity.

Interestingly, the LDV system proved useful in the measurement of arterial blood velocity (Fig. 7.44), thereby providing valuable medical information. Another application in medical research is the study of motion of the tympanic membrane in the ear.

7.27. Ultrasonic/Laser Nondestructive Evaluation Sensor

Ultrasonic/laser optical inspection is a relatively new noncontact technique. A laser system for generating ultrasound pulses without distortion of the object surface is shown in Fig. 7.45. A laser pulse incident on a surface will

object surface is shown in Fig. 7.45. A laser pulse incident on a surface will be partly absorbed by the material and will thus generate a sudden rise in temperature in the surface layer of the material. This thermal shock causes expansion of a small volume at the surface, which generates thermoelastic strains. Bulk optical systems have been used previously to generate the laser pulse energy. However, the omnidirectionality of bulk sources is completely dierent from other well-known sources, and is regarded as a serious handicap to laser generation. To control the beamwidth and beam direction of the optically generated ultrasonic waves, a ber phased array has been developed.

Figure 7.45. Setup for beam steering of a laser-generated ultrasound by ber-optic phased array.
In this way, the generated ultrasonic beam can be focused and directed to a particular inspection point below the surface of an object (Fig. 7.45). This system has been optimized for the detection of fatigue cracks at rivet holes in aircraft structures. The combination of laser-generated ultrasound and an optical-ber interferometer for the detection of the resultant surface displacement has led to a technique that is useful for a wide variety of inspection tasks in manufacturing, including high-temperature objects and areas that are dicult to access, as well as more routine inspection and quality control in various industrial environments. Such a system can be applied to the measurement of thickness, velocity, aws, defects, and grain size in a production process.

7.28. Process Control Sensors for Acceleration

The principle of operation of the process control acceleration sensor is illustrated in Fig. 7.46. The sensor element, consisting of a small cantilever and a photoluminescent material, is attached to the end of a single multimode ber. The input light of wavelength
s

is transmitted along the

ber from a near-infrared LED source to the sensor element. The sensor element returns light at two dierent wavelengths, one of which serves as a signal light and the other as a reference light, into the same ber. The signal light at wavelength
s

is generated by reection from a small cantilever.

Since the relative angle of the reected light is changed by the acceleration, the returned light is intensity-modulated. The reference light of wavelength
r

is generated by photoluminescence of a neodymium-doped glass element

placed close to the sensor end of the ber.

Figure 7.46. Cantilever-type acceleration sensor.

The optoelectronic detector module has two optical lters to separate the signals
s

and

and two photodiodes to convert the signal and the

reference light into separate analog voltages. The signal processing for compensation is then merely a matter of electrical division. A measuring range of 0.1 to 700 m/s
2

and a resolution of 0.1 m/s

is obtained over the

frequency range of 5 to 800 Hz.

7.29. An Endoscope as Image Transmission Sensor

An imaging cable consists of numerous optical bers, typically 3000 to 100,000, each of which has a diameter of 10 m and constitutes a picture element (pixel). The principle of image transmission through the bers is shown in Fig. 7.47. The optical bers are aligned regularly and identically at both ends of the bers. When an image is projected on one end of the image

ber, it is split into multiple picture elements. The image is then transmitted as a group of light dots with dierent intensities and colors, and the original picture is reduced at the far end. The image bers developed for industrial use are made of silica glass with low transmission loss over a wide wavelength band from visible to near infrared, and can therefore transmit images over distances in excess of 100 m without signicant color changes. The basic structure of the practical optical-ber image sensing system (endoscope) is illustrated in Fig. 7.48. It consists of the image ber, an objective lens to project the image on one end, an eyepiece to magnify the received image on the other end, a ber protection tube, and additional bers for illumination of the object.

Figure 7.47. Image transmission through an image ber.

Figure 7.48. Basic structure of a ber scope.

Many examples have been reported of the application of image bers in process control. Image bers are widely employed to observe the interior of blast furnaces and the burner ames of boilers, thereby facilitating supervisory control. Image bers can operate at temperatures up to 1000C, when provided with a cooling attachment for the objective lens and its associated equipment. Another important application of the image ber bundles is observation, control, and inspection of nuclear power plants and their facilities. Conventional image bers cannot be used within an ionizing radiation environment because ordinary glass becomes colored when exposed to radiation, causing increasing light transmission loss. A highpurity silica core ber is well-known as a radiation-resistant ber for nuclear applications.

The endoscope has demonstrated its vital importance in medical and biochemical elds such as: Angioplasty Laser surgery Gastroscopy Cystoscopy Bronchoscopy Cardioscopy

7.30. Sensor Network Architectures in Manufacturing

In ber-optic sensor networks, the common technological base with communication is exploited by combining the signal generating ability of sensors and the signal transmitting capability of ber optics. This combination needs to be realized by a suitable network topology in various manufacturing implementations. The basic topologies for sensor networking are illustrated in Fig. 7.49. The basic network topologies are classied into six categories: Linear array network with access-coupled reective sensors (Fig. 7.49a ). Ring network with in-line transmissive sensors (Fig. 7.49b ).

Figure 7.49. Basic network topologies: (a) linear array, (b) ring, (c) reective star, (d) reective tree, (e) transmissive star, and (f) ladder network.
Star network with reective sensors (Fig. 7.49c ).

Star network with reective sensors; one or more sensors can be replaced by a separate star network, in order to obtain a tree network (Fig. 7.49d ). Star network that can also be operated with transmissive sensors (Fig. 7.49e ). Ladder network with two star couplers. A star coupler is replaced by several access couplers, the number required being equal to the number of sensors (Fig. 7.49f ).

Topological modications, especially of sensor arrays and ladder networks, may be desirable in order to incorporate reference paths of transmissive (dummy sensors) or reective sensors (splices, open ber end). The transmit and return bers, or ber highway, generally share a common single-ber path in networks using reective sensors. When a suitable ber-optic network topology is required, various criteria must be considered: The sensor type, encoding principle, and topology to be used The proposed multiplexing scheme, required number of sensors, and power budget The allowable cross-communication level The system cost and complexity constraints The reliability (i.e., the eect of component failure on system performance)

7.31. Power Line Fault-Detection Systems for Power Generation and Distribution Industries
In power distribution lines, faults such as short circuits, ground faults, and lightning strikes on the conductors must be detected in a very short time to prevent damage to equipment and power failure, and to enable quick repair. If the transmission line is divided in sections and a current or magnetic-eld sensor is mounted in each section, a faulty section can be determined by detection of a change of the level and phase of the current on the power line.

A system was developed as a hybrid optical approach to a fault-locating system that detects the phase and current dierence between two current transformers on a composite ber-optic ground wire (OPGW) wherein, due to induction, current is constantly passing (Fig. 7.50). The signal from a local electrical sensor, powered by solar cells and batteries, is transmitted over a conventional optical-ber communication link. By three-wavelength multiplexing, three sensor signals can be transmitted over a single ber. Seven sensors, three at each side of a substation and one at the substation itself, can monitor one substation on the power line, using one ber in the OPGW .

Figure 7.50. Fault locating system based on current and phase in ground wire.
Another system uses current transformers to pick up lightning current and thus detect lightning strikes. The signal is transmitted to a central detection point using the OPGW . Every sensor has its own OPGW ber. This system is on a 273-kV power line in Japan. The OPGW opens the possibility for using all kinds of sensors along the transmission line. These sensors may not only be used for locating faults, but also for monitoring structural integrity. The use of optical time-domain reectometry (OTDR) combined with passive intrinsic (distributed) sensors along the OPGW has future potential for providing a convenient and powerful monitoring method for power lines.

Further Reading
[p200199208990376001] Bailey Control Systems , Wicklie, Ohio.

[p200199208990376002] Bartman, R. K., B. R. Youmans, and N. M. Nerheim.

Integrated Optics Implementation of a Fiber Optic Rotation Sensor: Analysis and Development, Proc. SPIE , 719, 122134.
[p200199208990376003] Berthold, J. W ., Industrial Applications of Optical

Fiber Sensors, Fiber Optic and Laser Sensors III, Proc. SPIE , 566, 3744.
[p200199208990376004] Carrol, R., C. D. Coccoli, D. Cardelli, and G. T. Coate,

The Passive Resonator Fiber Optic Gyro and Comparison to the Interferometer Fiber Gyro, Proc. SPIE , 719, 169177 (1986).
[p200199208990376005] Chappel, A. (ed.), OptoelectronicsTheory and

Practice , McGraw-Hill, New York, 1978.

[p200199208990376006] Crane, R. M., A. B. Macander, D. W . Taylor, and J. Gagorik, Fiber Optics for a Damage Assessment System for Fiber

Reinforced Plastic Composite Structures, Rev. Progress in Quantitative NDE , 2B, Plenum Press, New York, 14191430, 1982.
[p200199208990376007] Doeblin, E. O., Measurement SystemsApplication

and Design , 4th ed., McGraw-Hill, New York, 1990.

[p200199208990376008] Fields, J. N., C. K. Asawa, O. G. Ramer, and M. K. Barnoski, Fiber Optic Pressure Sensor, J. Acoust. Soc. Am. , 67, 816 (1980). [p200199208990377001] Finkelstein, L., and R. D. Watts, Fundamental of

TransducersDescription by Mathematical Models, Handbook of Measurement Science ,vol. 2, P. H. Sydenham (ed.), Wiley, New York, 1983.
[p200199208990377002] Friebele, E. L. and M. E. Gingerich, Radiation-

Induced Optical Absorption Bands in Low Loss Optical Fiber Waveguides, J. Non-Crust. Solids , 38(39), 245250 (1980).
[p200199208990377003] Henze, M., Fiber Optics Temperature and

Vibration Measurements in Hostile Environments, Technical Material, ASEA Research and Innovation , CF23-1071E (1987).
[p200199208990377004] Hofer, B., Fiber Optic Damage Detection in Composite Structure, Proc. 15th Congress. Int. Council Aeronautical

Science , ICAS-86-4.1.2, 135143 (1986).

[p200199208990377005] Kapany, N. S., Fiber Optics, Principles and

Applications , Academic Press, London, 1976.

[p200199208990377006] Lagakos, N., et al., Multimode Optical Fiber Displacement Sensor, Appl. Opt. , 20, 167 (1981). [p200199208990377007] Liu, K., Optical Fiber Displacement Sensor Using a

Diode Transceiver, Fiber Optic Sensors II , A. M. Sheggi (ed.), Proc. SPIE ,

798, 337341 (1987). [p200199208990377008] Mizuno, Y., and T. Nagai, Lighting Observation

System on Aerial Power Transmission Lines by Long Wavelength Optical Transmission, Applications of Fiber Optics in Electrical Power Systems in Japan , C.E.R.L. Letterhead, paper 5.
[p200199208990377009] Mori, S., et al., Development of a Fault-Locating

System Using OPGW, Simitomo Electric Tech. Rev . (25), 3547.

[p200199208990377010] Norton, H. N., Sensors and Analyzer Handbook , Prentice-Hall, Englewood Clis, N.J., 1982. [p200199208990377011] Neubert, H.K.P., Instrument Transducers , 2d ed., Clarendon Press, Oxford, 1975. [p200199208990377012] Ogeta, K., Modern Control Engineering , 2d ed. Prentice-Hall, Englewood Clis, N.J., 1990. [p200199208990377013] Petrie, G. R., K. W . Jones, and R. Jones, Optical

Fiber Sensors in Process Control, 4th Int. Conf. Optical Fiber Sensors, OFS86 , Informal Workshop at Tsukuba Science City, VIII IVIII 19, 1986.
[p200199208990377014] Place, J. D., A Fiber Optic Pressure Transducer

Using A Wavelength Modulation Sensor, Proc. Conf. Fiber Optics 85 (Sira) ,

London, 1985. [p200199208990377015] Ramakrishnan, S., L. Unger, and R. Kist, Line Loss

Independent Fiberoptic Displacement Sensor with Electrical Subcarrier Phase Encoding, 5th Int. Conf. Optical Fiber Sensors, OFS 88 , New
Orleans, 133136, 1988. [p200199208990377016] Sandborn, V. A., Resistance Temperature

Transducers , Metrology Press , Fort Collins, Colo., 1972.

[p200199208990377017] Scruby, C. B., R. J. Dewhurst, D. A. Hutchins, and S.

B. Palmer, Laser Generation of Ultrasound in Metals, Press, London, 1982.

Research

Techniques in Nondestructive Testing , vol. 15, R. S. Sharpe (ed.), Academic

[p200199208990377018] Tsumanuma, T., et al., Picture Image Transmission-

System by Fiberscope, Fujikura Technical Review (15), 110 (1986).

[p200199208990377019] Vogel, J. A., and A. J. A. Bruinsma, Contactless

Ultrasonic Inspection with Fiber Optics, Conf. Proc. 4th European Conf. Non-Destructive Testing ,Pergamon Press, London, 1987.
[p200199208990377020] Yasahura, T., and W . J. Duncan, An Intelligent Field

Instrumentation System Employing Fiber Optic Transmission, Advances in Instrumentation , ISA, Wiley, London, 1985.
[p200199208990377021] Cova, S. et al., Avalanche Photodiodes for Near

Infrared Photon-Counting , SPIE Proc ., vol. 2388 (1995).

[p200199208990377022] Dautet, H., et al., Photon-Counting Techniques

with Silicon Avalanche Photodiodes, Applied Optics , 32 (21), 38943900

(1993). [p200199208990377023] Mclntyre, R. J., Multiplication Noise in Uniform

Avalanche Diodes, IEEE Trans. Electron Devices , ED13, 164168 (1966).

[p200199208990377024] Product Datasheet ED-0017/03/88: C30902 - 921 E/S, EG&G Canada Ltd.
Citation
EXPORT

Sabrie Soloman: Sensors and Control Systems in Manufacturing, Second Edition. Industrial Sensors and Control, Chapter (McGraw-Hill Professional, 2010 1994), AccessEngineering

2012 The McGraw-Hill Companies. All rights reserved. Customer Privacy Notice. Any use is subject to the Terms of Use, Privacy Notice and copyright information. For further information about this site, contact us.

Designed and built using SIPP2 by Semantico. This product incorporates part of the open source Protg system. Protg is available at http://protege.stanford.edu//