IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 60, NO. 11, NOVEMBER 2011 3669 Thick-Film Ceramic Strain Sensors for Structural Health Monitoring Saad A. A. Jabir and Naren K. Gupta, Senior Member, IEEE Abstract—Structural Health Monitoring represents one of the primary field applications for new sensor technologies. Different kinds of sensors to monitor stability anomalies of civil structures are deployed on different kinds of building materials. Observation of stability must be durable and reliable for the lifetime of a structure. This paper is part of a research to investigate the possibility of application of thick-film (TF) piezoresistive sensors on building materials for the scope of monitoring structure’s stability and conditions of load. As a first approach toward this objective, TF piezoresistive sensors will be used like metal-foil strain gauges on two different building materials, namely brick (clay) and concrete. A boosted Wheatstone bridge interface circuitry will be proposed and simulated with a Saber simulator. Thick film ceramic sensors (TFCS) will be applied on red brick and concrete to investigate TF response proportionality and linearity when applied on these two kinds of building materials. Index Terms—Hybrids, piezoresistive, Saber, sensors, simulation, strain, stress, Structural Health Monitoring (SHM), thick film (TF). I. I NTRODUCTION I N ORDER to monitor the health of large structures such as highways, bridges, buildings, and dams, the outputs from various kinds of sensors, which are positioned at different places of structures, should be collected simultaneously and/or sequentially . Sensors could be of different nature and for measuring different physical entities. The most relevant entity to observe when monitoring the health of a structure is the strain due to material deformation. Strain indicates the load or load variation at the points where these sensors are located. The common way of measuring strains on structures is by applying foil film strain gauges , widely known as strain gauges. A myriad of shapes and configurations of such foil strain sensors suitable for force and torsion measurements on various structure geometries are commercially available [Fig. 1(a)]. On the other hand, the application of thick-film (TF) sensors (TFSs) is a well-consolidated technology that has been used for many years to build pressure transducers, as in Fig. 1(b). Certain TF pastes placed on a ceramic substrate with Manuscript received November 22, 2010; revised March 15, 2011; accepted March 16, 2011. Date of publication June 2, 2011; date of current version November 9, 2011. The Associate Editor coordinating the review process for this paper was Dr. Jesús Ureña. S. A. A. Jabir is with Synopsys GmbH, D-85609 Munich, Germany (e-mail: [email protected]). N. K. Gupta is with the School of Engineering and the Built Environment, Edinburgh Napier University, EH10 5DT Edinburgh, U.K. Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TIM.2011.2138310 a conventional silkscreen technique have a piezoresistive effect, i.e., will change in resistor value with strain developed on the ceramic substrate. The TF technology is the same technology used to build hybrid circuits. It was developed during the fifties and was considered as a replacement of printed circuit technology to provide mechanical assembly support for electronic circuits. A TF circuit comprises of layers of special inks or pastes deposited and fired onto an insulating substrate. The ink is composed of various metals, oxides, and ceramic and glass powders suspended in an organic vehicle. TF components make conductive, resistive, capacitive, and dielectric film patterns on a substrate surface. Deposition of the ink on the substrate surface is by screen printing through stainless-steel masks. The deposited inks are then fired in a kiln with predefined temperature profile. The steps to produce TF piezoresistive sensors are the same as those to produce hybrid circuits . In the 1960s and with the advent of resistors built with this technology using certain pastes, the piezoresistive effect was observed and used to build TF stress sensors. Table I highlights the main characteristics of TF strain sensors with respect to metal-foil and silicon strain gauge sensors. The combination of relatively high gauge factor (GF) and good temperature stability in addition to mass production capability and low cost are the major factors that justify the implementation of this technology to build reliable force and pressure sensors in general. TFS technology for measuring force related to deformation is applicable in automotive, industrial, medical, and sport apparatus fields. Long-term stable TF pressure transducers are commercially available. TF materials are known for their survival, longevity, and reliability in harsh environments, extreme temperatures, and high mechanical stresses, such as those present in automotive applications . This fact makes TF technology a good candidate for civil engineering applications. TF hybrid technology responds to the needs of civil and industrial structures of different kinds of sensors to measure various physical and environmental entities relevant to building security monitoring like temperature measurement – and chemical biosensors . An important topic in Structural Health Monitoring is the power supply for the intelligent sensors. The supply is necessary to drive electronics and the networking to monitor a structure during its life, which might last for tens of years. It is well known that cadmium sulfide can be screen printed and sintered to form films that display photosensitivity. Photoconductive sensors based on screen-printed cadmium sulfide and cadmium selenide thick films printed over standard silver–palladium conductors are feasible. Simple 0018-9456/$26.00 © 2011 IEEE 3670 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 60, NO. 11, NOVEMBER 2011 Fig. 1. Examples of (a) metal film strain gage and (b) Ceramic pressure membranes. The linearity and proportionality of response with material modulus of elasticity, for the first deployment method, will be discussed and experimented in this paper. IV. BASICS OF S TRAIN G AUGE C ONCEPTS AND R ELATIONSHIPS For the first deployment method, like foil-metal strain gauges, TFCS obeys R = ρ (L/A). (1) Fig. 2. Single piezoresistive TFS. photoconductive arrays and a potentiometric position sensor have been fabricated with TF technology –. Substrates made of ceramic material or steel have outstanding mechanical and thermal properties, which make them adequate to have stable electronic circuits on the one hand and to withstand harsh environmental conditions on the other. II. S AMPLE P REPARATION For the purpose of the investigation for the deployment of TFSs on building materials, a single TF resistor was developed on 0.3-mm-thick 96% alumina substrate (AL2 O3 ), as shown in Fig. 2. The piezoresistor aspect ratio, W : L, was 1.3. The paste used was DP2041 with a GF of 10. This paste is widely used for TFSs. Sheet resistance was 10 kΩ/. The resistance value obtained was in the range of 7.2–8 kΩ with a medium value of 7.5 kΩ. In general, if the effect of all variables on the resistor is considered ΔR Δρ ΔA ΔL = − + . (2) R ρ A L In this context, ΔL/L is the strain and ΔA/A = −2νε in an isotropic material stressed within its elastic behavior and characterized by a Poisson’s ratio ν . Accordingly, the following relationship is obtained: ΔR Δρ GF = = + 1 + 2ν. (3) RC ρC The GF is dependent on two terms, one is due to change in resistivity with strain and the other is for geometrical alteration. Solids may have a Poisson’s ratio ν of 0.2–0.45. Hence, even if the term dependent on resistivity of the material is nil, the geometrical contribution will give rise to a GF of 1.5–1.9. The GF for TF piezoresistive sensors ranges from 2 to 35, as indicated in Table I. V. C ONSIDERATION OF THE E LECTRONIC I NTERFACE III. D EPLOYMENT M ETHODS The sample sensor was deployed on structures in three ways. 1) Glued on the material surface like a common foil strain gauge. The force in this case acts parallel to the sensor surface, as in Fig. 3(a). 2) Similar to 1) with force perpendicular on sensor surface [Fig. 3(b)]. 3) Embedded in the material, as in Fig. 3(c), where the sensor is mounted on a printed circuit board or ceramic substrate in surface mount technology (SMT). For each of these methods of deployment, the TFS exhibits a different GF. The quarter Wheatstone bridge of Fig. 4 has one TF resistor, R + ΔR, changing with strain. After balancing the bridge i.e. Vo = VAB = 0, the applied stress on the TF sensor will change VAB as in (4) below: Vo is ΔR V × . (4) Vo = 4 R From (3), the change in resistance to axial strain is ΔR/R = GF C; hence V · C · GF (5) Vo = 4 1 4Vo C= × . (6) V GF JABIR AND GUPTA: THICK-FILM CERAMIC STRAIN SENSORS FOR STRUCTURAL HEALTH MONITORING 3671 TABLE I COMPARATIVE CHARACTERISTICS OF STRAIN GAGE TYPES  Fig. 5. Operational amplifier as current source. strain. The TF (R_TF) sensor is placed as a feedback resistor of the operational amplifier. The current through R_current is I= Fig. 3. TF force sensor deployment methods. (a) On material surface. (b) Direct load application on sensor surface. (c) Embedded in building material. V ref . R_current (7) The change in R_TF will be a change in the voltage across it, and consequently, the op-amp output voltage Vo will change following ΔVo = I × ΔR_TF. (8) Equation (9) results from dividing both sides of (8) by R_TF ΔR_TF ΔVo = . I × R_TF R_TF (9) To be able to compare (5) with (9), I ∗R_TF is replaced by the excitation voltage V , i.e., the maximum output voltage swing Vo ΔR = . V R (10) ΔR V. R (11) Hence Vo = Fig. 4. Wheatstone bridge circuit. This equation relates mechanical strain measurement to the bridge unbalance voltage (Vo ). Excitation voltage (V ) and gauge factor (GF ) are considered constants. Alternatively, the simple inverting operational amplifier in Fig. 5 can be used to increase output voltage variation with The higher is the current, the more is the voltage variation with strain. However, the current must be as high as that allowed by the amplifier dynamic range; thus, I ∗R_TF is almost the excitation voltage V . Output voltage swing is a function of the op-amp technology. Certain CMOS op amps have rail-to-rail output swing capability. 3672 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 60, NO. 11, NOVEMBER 2011 Fig. 6. Wheatstone bridge voltage amplified by an instrumentation amp. Fig. 7. Vout versus TF Rt1 change. ΔY should be divided by ten. From (3) Vo = V · GF · C. (12) This equation relates mechanical strain measurement to the op-amp output voltage (Vo ). Excitation voltage (V ) and gauge factor (GF ) are considered constants. Comparing (5) with (12), it is obvious that for the piezoresistive change with strain of one single sensor resistor, the output voltage swing in the op-amp configuration is four times bigger than the common quarter Wheatstone bridge setup at the same excitation voltage. A. Verification by Simulation To verify the derived mathematical conclusion before deploying the op-amp solution in real measurements, the two solutions were simulated with Saber simulator. The convention used in the circuitry is by having TFS resistances in compression denoted by Rc while those in tension are denoted by Rt . The full Wheatstone bridge of Fig. 6 shows four mechanically excited TF resistors. Resistors in the bridge are marked Rt for resistors in tension and Rc for resistors in compression. Those resistors must be positioned on the structure under test in a way to be excited mechanically in an opposite sense to enhance the signal output product. To verify the quarter bridge as in (4), only resistor Rt1 will be varied. The TF typical initial value resistance will be assumed 7500 Ω similar to the manufactured sample’s typical value. A GF of ten for a maximum strain of 1000 μstrain, which is approximately the maximum allowed strain for alumina , will produce maximum amount of piezoresistance change of 1%, i.e., 75 Ω. Counting for overload and tolerances, the circuit was simulated with Saber simulator, assuming a change of dr from 0 to 100 Ω in steps of ten. The instrumentation amplifier has a fixed gain of ten. During simulation cycles, the instrumentation amplifier (LT1101) gives an immediate reading of the voltage difference across the bridge and it is actually of the type that was used in the experiments. Fig. 8. TFS resistor in op-amp LMC6482 configuration. Transient analysis simulation of the circuit was setup for 100 μs. Fig. 7 shows the output voltage on the waveform viewer of Saber named cosmoscope. In the plot, there is one waveform for each value of Rt1 . The maximum obtainable voltage, ΔY can be seen from the measurement on the waveform to be = 0.397/10 = 0.0397 V. The division by ten is for the LT1101 amplifier fixed gain of ten. The same single TF resistor is made to change in a rail-torail operational amplifier configuration, as shown in Fig. 8. The amplifier is of type LMC6482 (CMOS rail-to-rail op amp). A current of (11 V/7500 Ω) A flows through the TF feedback resistor. With the change of R_TF by ΔR, the output remains in the allowed linear amplifier dynamic range. Simulation for 100-μs result with Rt1 changing in steps of 10 Ω is shown in Fig. 9. The measured voltage change ΔY = 0.14666 V. The ratio of the op-amp configuration to the quarter bridge configuration is thus 0.14666 V (op − amp) = = 3.69. V (bridge) 0.039 (13) JABIR AND GUPTA: THICK-FILM CERAMIC STRAIN SENSORS FOR STRUCTURAL HEALTH MONITORING Fig. 9. Output of op-amp with Rt1 stepping by 10 Ω. Fig. 10. Op-amp-driven Wheatstone bridge. This is almost four times the Wheatstone bridge interface circuit. The result is similar to the mathematical derivation. However, a higher value can be achieved by increasing the current as long as the maximum allowed output voltage swing of the chosen op-amp type is not exceeded. In Fig. 9, the small output voltage change is offset by 11 V from zero, while the waveform in Fig. 7 swings to almost 0.4 V from zero. One of the advantages of Wheatstone bridge is its elimination of the offset by having two branches or two voltage dividers that when subtracted, the difference is almost at zero level. This differential mode of functionality eliminates all nondesired common signals like noise and dc voltage. In addition, by using similar elements on opposite sides of the bridge, temperature drift of the sensors can be significantly reduced. To collect the advantages of both solutions, the circuit in Fig. 10 was simulated with a Saber simulator to examine the output voltage maximum swing with 1000 μstrain applied on 3673 Fig. 11. Op-amp-driven Wheatstone bridge max output voltage swing. all four bridge resistors Rt1 , Rt2 , Rc1 , and Rc2 . Rt1 and Rt2 are supposed to be under tension, hence increasing, and Rc1 and Rc2 are supposed to be contemporarily under compression, hence decreasing. ΔR changes in steps of 10 Ω. A maximum voltage change of 5.3 V can be observed at the instrumentation amplifier output, as shown in the waveform of Fig. 11. The differential amplifier removes the dc offset voltages at op-amp outputs; thus, its output swing starts from 0 to 5.3 V. The noise and temperature drift of the sensors will be also canceled by the differential final stage. This way, besides signal mitigation by using op amps, the advantage of having a differential output is maintained. The transfer function of the circuit in Fig. 10 after letting the (ΔR/R)2 term to zero is ΔR (14) Vo = 4Vr G × R where Vr input reference voltage (10 V in the figure); G instrumentation amplifier gain (set to ten); R nominal TF resistance value; ΔR change in resistance due to strain. It should be noticed here that the final output is a multiplication of a dc reference voltage and instrumentation amplifier gain (i.e., Vr × G). VI. A PPLICATION OF TFS ON B UILDING M ATERIALS In the circuit in Fig. 10, even if sensors are under the same mechanical excitation, they could be placed appropriately in the right place in the circuit to sum the effect at the output. In other words, it is possible to use the two resistor sensors in compression only while the sensors in tension are replaced with fixed resistances. The circuit in Fig. 12 is the same previous circuit with minor practical modifications. One important addition is the offset balance input amplifier. TFSs are usually of different resistance values so the voltage 3674 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 60, NO. 11, NOVEMBER 2011 Fig. 12. Differential amplifier with offset balance. output would be always out of zero. By varying V _DAC, the instrumentation amplifier output could be reset to zero before starting the test. The balance amplifier will alter the current in Rc2 to make V2 equal to V1 . This circuit was connected to LABVIEW on a personal computer through an interface POD named DAQPad-1200. This later has eight A/D converter inputs and two DAC outputs, one of which was used to zero the instrumentation amplifier output. The instrumentation amplifier LT1101 offers the possibility to set the gain externally, either to 10 or to 100. The analog switch DG201 was used to eventually increase the gain to 100. The analog switch control is from LABVIEW through DAQPod1200 the Gain_control digital output pin. To verify applicability of TFSs on different building materials and to evaluate their response as a function of the different materials, a test was done on the sample brick of dimensions 12.4 × 16.7 × 34.7 mm [Fig. 13(a)] and sample concrete of dimensions 23.3 × 22.3 × 50 mm [Fig. 13(b)]. The aim of the test was to verify the following: 1) the applicability and functionality of TFSs on concrete and bricks like foil-metal strain gauges; 2) to test the circuit and examine the possibility of adding the effect of two adjacent sensors glued on the same side of a structure and that the linearity is maintained, i.e., if Y1 = f (x1 ) and Y2 = f (x2 ), then Y1 + Y2 = f (x1 + x2 ); 3) the output entity and the proportionality of response of TFSs as a function of the material modulus of elasticity when applied to different materials like, in the specific case, red brick and concrete. The two TFSs were glued to the two samples with a twocomponent epoxy adhesive and were put in the locations indicated by Rc1 and Rc2 on the schematic. The other resistors Rt1 and Rt2 were replaced with a fixed resistor, as shown in Fig. 12. The transfer function equation of the circuit is Rc1 Rt2 − ×G (15) V out = Vr Rt1 Rc2 where Vr = 10 V and Rt1 = Rt2 = 7500 Ω. Fig. 13. block. (a) Sample brick of 12.4 × 16.7 × 34.7 mm. (b) Sample concrete Fig. 14. Concrete block sample stressed on a linear scale. VII. E XPERIMENTAL R ESULTS A. Experiment Overview The brick and concrete samples were put one at a time under pressure against an electronic linear scale in a bench vice, as shown in Fig. 14. The load was increased step by step up to a 100 kg load. Here, are the experimental steps. 1) The response of one single sensor at a time was considered first on the red brick sample. 2) The response of both sensors in compression at the same time on the red brick was then registered. 3) The mathematical sum of the two sensors compressed singularly was compared with the experimental result of both sensors compressed contemporarily. JABIR AND GUPTA: THICK-FILM CERAMIC STRAIN SENSORS FOR STRUCTURAL HEALTH MONITORING 3675 Fig. 18. S3 and S4 responses versus compressive force on concrete. Fig. 15. S1 response versus compressive force. Fig. 19. TFSs response on red brick and concrete. Fig. 16. S2 response versus compressive force. Fig. 17. S1 and S2 responses versus compressive force on brick. 4) The response of both sensors in compression at the same time on the concrete sample was then registered. 5) A comparison of the TF response of the two sensors on both samples was then considered based on the difference in the two materials elasticity modulus. B. Linearity of Response Step 1: Fig. 15 shows S1 response with S2 replaced by a fixed resistor of 7500 Ω. Fig. 16 shows S2 response with S1 replaced by a fixed resistor of 7500 Ω. Step 2: Fig. 17 shows the response of both S1 and S2 under mechanical compressive stress at the same time. Step 3: From the linear equations of the responses of S1 and S2 , it is clear that the sum of the two equations will produce the equation of the experimental sum in Fig. 17, i.e., 0.4005x + 0.5538x = 0.9543 x. C. Proportionality of Response Step 4: Fig. 18 shows the responses of other TFSs S3 and S4 on concrete block. Step 5: Fig. 19 shows the response of TFSs on the brick and concrete blocks put on the same graph for comparison. The sensor used to measure strain on material should have a response consistent with different materials under test. Following is the proof from the measurement curves of Fig. 19. If we consider Hook’s law E = σ/C and apply this on the two samples, red brick and concrete, where c = concrete and b = brick σ Ec = (16) ∈c σ Eb = . (17) ∈b By dividing (16) by (17) and considering the application of the same force change on both samples Ec = Eb ΔF Ac ×∈c ΔF Ab ×∈b = Ab × ∈b . Ac × ∈c (18) Since the output voltage is linearly proportional to strain and the applied ΔF for the two samples is the same Ec Ab × ΔVb = . Eb Ac × ΔVc (19) Considering the coefficients in the generated equations from the curves in Fig. 19 for the same 100 kg force change and substituting the values for the different sample areas, i.e., Ab = 205 mm2 and Ac = 545 mm2 Ec 205 × 0.9476 = 1.4524. = Eb 545 × 0.2454 (20) To evaluate the left side of (20), the elastic moduli of brick and concrete need to be substituted. From , the concrete elastic modulus depends on the ratio of sand to aggregate (S/A) and it varies from 22 to 25 k · N/mm2 . The sample under test has almost no aggregate, and hence, the modulus is around 3676 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 60, NO. 11, NOVEMBER 2011 22–23 k · N/mm2 ([7, p. 11, Fig. 5]). The experimental modulus of elasticity of red brick is around 15 k · N/mm2 . Hence Ec 22 000 = 1.466. (21) = Eb 15 000 The reason for the small difference is due to a) the different sensors which have been chosen randomly from a production lot. b) Exact modulus of elasticity of the samples, which could be different from the values considered in the calculation. However, result shows that response is consistent and proportional with material elastic modulus. VIII. S UMMARY AND D ISCUSSION This paper has examined the applicability of TF piezoresistive sensors on building material. For this purpose, a generic TFCS has been manufactured and used to investigate the possibility of its application on building materials. The designed sensor suits different deployment methods: 1) glued on the material with force parallel to the sensor surface; 2) glued on the material with force perpendicular on the sensor surface; and 3) mounted in surface mounting technology (SMT) on PCB to be embedded in the material. The paper has examined the linearity and proportionality of the sensor response for the first deployment method. As interface circuitry, an op-amp-boosted Wheatstone bridge has been verified to give better output response. Simulations proved the mathematical conclusion. This interface circuit has then been used to verify linearity and proportionality of response of TF piezoresistors on two kinds of building materials, namely, red brick and concrete. The measured response of the TF piezoresistive sensor has been shown to be, first, linear, i.e., if Y1 = f (x1 ) and Y2 = f (x2 ), then Y1 +Y2 = f (x1 +x2 ), and second, proportional to the modulus of elasticity of these two materials. To reach a practical deployment, besides linearity and proportionality, many other aspects need to be considered, studied, and experimented. Reliability of readings in time as a function of glue and structure deterioration, humidity and temperature effects, networking, and supply are part of the many considerations to qualify this technology. In addition, other kinds of substrates like LTCC and fiberglass have to be considered in light of exceeding the 1000-μstrain limitation of alumina and to have major integration. Different kinds of glue were experimented, and the most reliable was a hard two-component epoxy. However, there was no systematic scientific approach for which comparative results can be given at this stage. R EFERENCES  D.-H. Wang and W.-H. Liao, “Wireless transmission for health monitoring of large structures,” IEEE Trans. Instrum. Meas., vol. 55, no. 3, pp. 972– 981, Jun. 2006.  F. Tegtmeier and M. Peters, “Determination of structural damage to buildings using microsensors based on strain gage technology,” in Proc. XVI IMEKO World Congr., Vienna, Austria, Sep. 25–28, 2000, vol. VII, pp. 363–369.  N. M. White and J. E Brignell, “A planar thick-film load cell,” Sens. Actuators A, Phys., vol. 26, no. 1–3, pp. 313–319, Mar. 1991.  T. Wada, S. J. Stein, M. A. Stein, and S. M. Chitale, “The state-of-theart of thick film technology for automotive sensors,” in Proc. IEMT/IMC, 1997, pp. 41–46.  F. Vittorio, D. Marioli, and A. Taroni, “Displacement sensor based on pyroelectric thick-films and contactless light-spot cursor,” IEEE Trans. Instrum. Meas., vol. 51, no. 4, pp. 819–823, Aug. 2002.  M. K. Jain and C. A. Grimes, “A wireless magnetoelastic micro-sensor array for simultaneous measurement of temperature and pressure,” IEEE Trans. Magn., vol. 37, no. 4, pp. 2022–2024, Jul. 2001.  J. K. Su, S. W. Cho, C. C. Yang, and R. Huang, “Effect of sand ratio on the elastic modulus of self-compacting concrete,” J. Marine Sci. Technol., vol. 10, no. 1, pp. 8–13, 2002.  V. Gabriela, C.-C. Liu, and J. M. Anderson, “Electrochemical characterization and in vivo biocompatibility of a thick-film printed sensor for continuous in vivo monitoring,” IEEE Sensors J., vol. 5, no. 6, pp. 1147– 1158, Dec. 2005.  F. H. Nicoll and B. Kazan, “Large area high-current photoconductive cell using cadmium sulphide powder,” J. Opt. Soc. Am., vol. 45, no. 8, pp. 647– 650, Aug. 1955.  J. N. Ross, Thick-film Photo-sensors1994. [Online]. Available: http:// www.ecs.soton.ac.uk/publications/rj/1994/transduc/ross/ross.html  S. Vojdani, A. Sharifanai, and M. Doroudian, “Photovoltaic junctions formed on silk- screened cadmium sulphide layers,” Electron. Lett., vol. 9, no. 6, pp. 128–129, Mar. 1973.  B. Morten and M. Prudenziati, “Piezoresistive thick-film sensors,” in Handbook of Sensors and Actuators. Amsterdam, The Netherlands: Elsevier, 1994, p. 189.  M. Prudenziati, Handbook of Sensors and Actuators, Thick Film Sensors. Amsterdam, The Netherlands: Elsevier, 1994.  J. M. Nichols and Y. Z. Totoev, “Experimental determination of the dynamic Modulus of Elasticity of masonry units,” in Proc. 15th ACMSM, Melbourn, Vic, Australia, 1997.  D.-S. Lee, D.-D. Lee, S.-W. Ban, M. Lee, and Y. T. Kim, “SnO2 gas sensing array for combustible and explosive gas leakage recognition,” IEEE Sensors J., vol. 2, no. 3, pp. 140–149, Jun. 2002.  B. Morten and M. Prudenziati, “Piezoresistive thick-film sensors,” in Handbook of Sensors and Actuators. Amsterdam, The Netherlands: Elsevier, 1994, p. 193.  M. Prudenziati and B. Morten, “The state of the art in thick-film sensors,” Microelectron. J., vol. 23, no. 2, pp. 133–141, Apr. 1992. Saad Abdul Ameer Jabir received the B.Sc. degree in electrical engineering from Baghdad University, Baghdad, Iraq, and Ph.D. degree from Edinburgh Napier University, Edinburgh, UK. He is an Application Engineer Consultant for system and mixed-signal analysis software with Synopsys GmbH, Munich, Germany. He has worked as an Electronic System Designer in several industrial fields including nuclear medicine, orthopaedics, automatic test equipment, consumer electronics, and sensor design before joining Synopsys GmbH in 2002. He has over 30 successful industrial electronic projects. He is also the holder of patents in sports and orthopaedics apparatus and sensor elements. Naren K. Gupta (SM’87) received the B.Sc. degree in electrical engineering from Ranchi University, Ranchi, India, in 1969, the Postgraduate Diploma in technical science from the Institute of Science and Technology, The University of Manchester, Manchester, U.K., in 1972, the M.Sc. degree from Brunel University, Uxbridge, U.K., in 1976, the Ph.D. degree from the Institute of Science and Technology, The University of Manchester, in 1986, and the MBA degree from Edinburgh Napier University, Edinburgh, U.K., in 1996. He is a Professor in Electrical Engineering, a Teaching Fellow, and the Director of Quality in the School of Engineering and the Built Environment, Edinburgh Napier University, Edinburgh, U.K. He is an active Researcher and has published over 110 papers in international journals and conference proceedings. He has refereed papers for several journals, 545 in all, including those of IEEE and the Institution of Civil Engineers, U.K. He is currently a Consulting Editor for the journals EngineerIT and Energize. His current research involvement is in neural networks, measurements, and tests, including nondestructive testing, railway technology, and sensors and materials. He is also interested in pedagogical research. Dr. Gupta is a Chartered Engineer. He has been a fellow of the Institution of Engineering and Technology since 2001, a member of the Institution of Railway Signal Engineers, U.K., since 1982, and a fellow of the Higher Education Academy since 2000.