sensors Article

Design and Implementation of an Electronic Front-End Based on Square Wave Excitation for Ultrasonic Torsional Guided Wave Viscosity Sensor Amir Rabani Advanced Manufacturing Technology Research Group, Faculty of Engineering, University of Nottingham, University Park, Nottingham NG7 2RD, UK; [email protected]; Tel.: +44-115-951-4079 Academic Editor: Dipen N. Sinha Received: 1 September 2016; Accepted: 7 October 2016; Published: 12 October 2016

Abstract: The market for process instruments generally requires low cost devices that are robust, small in size, portable, and usable in-plant. Ultrasonic torsional guided wave sensors have received much attention by researchers for measurement of viscosity and/or density of fluids in recent years. The supporting electronic systems for these sensors providing many different settings of sine-wave signals are bulky and expensive. In contrast, a system based on bursts of square waves instead of sine waves would have a considerable advantage in that respect and could be built using simple integrated circuits at a cost that is orders of magnitude lower than for a windowed sine wave device. This paper explores the possibility of using square wave bursts as the driving signal source for the ultrasonic torsional guided wave viscosity sensor. A simple design of a compact and fully automatic analogue square wave front-end for the sensor is also proposed. The successful operation of the system is demonstrated by using the sensor for measuring the viscosity in a representative fluid. This work provides the basis for design and manufacture of low cost compact standalone ultrasonic guided wave sensors and enlightens the possibility of using coded excitation techniques utilising square wave sequences in such applications. Keywords: torsional guided waves; viscosity sensor; ultrasonic front-end; square wave excitation

1. Introduction There has been an increasing demand for real-time monitoring and control of fluids both in process and in storage over the past 20 years. Sensor systems for real-time and in situ applications, which are capable of withstanding the process environment coupled with meeting health and safety requirements, are increasingly required by many industries. Many fluids found in industry are complex in nature. They can contain suspended solid or liquid particles, and more generally can exhibit complex rheological behaviour including time-, shear-, temperature-, and pressure dependence. These complex properties highlight the need for robust and innovative sensor system design for process industries. Exposure to hostile and hazardous process conditions, such as plant vibrations, fouling, cleaning agents, and toxic compounds pose additional challenges in instrument design. Among the rheological properties of fluids, viscosity is an integral and necessary component of many quality control procedures in the processing of complex liquids [1]. Monitoring viscosity on-line and in situ provides real-time data which can be used to optimise the process and support product quality. Finding and developing a method and instrument for in situ measurements of viscosity for different environments has been of continuing concern over many years in the area of industrial process control. Several technologies are commercially available for viscosity measurements on-line and in situ but have their own constraints [2–4]. They are normally expensive and need frequent and costly maintenance. To contain costs and provide disposability, the choice of sensor system and the design of the supporting electronic platform are important. Sensors 2016, 16, 1681; doi:10.3390/s16101681

www.mdpi.com/journal/sensors

Sensors 2016, 16, 1681

2 of 13

Recently, much attention has been paid to ultrasonic methods, particularly for measuring properties of different materials. Many instruments and techniques have been developed on measuring rheological properties of fluids using ultrasound during past decades to find the density and viscosity of fluids [5,6]. In particular, ultrasonic guided wave sensors have been used to measure viscosity and/or density in liquids by many researchers [7–21]. They can be formed into robust instruments which are inexpensive enough to be disposable when applied to hazardous materials, or can be used on-line in a process. They have advantages over conventional systems which are delicate, expensive, and not suitable for operation in a process line. The most popular ultrasonic guided wave technique for measuring fluids’ viscosity and density is based on the application of ultrasonic torsional wave propagation along a cylindrical rod or a thin wire waveguide, which is immersed in a viscous fluid. In operation, a torsional stress wave propagates along a solid or hollow waveguide while it is submerged in a fluid. The boundary layer of the fluid is alternately accelerated and decelerated as the torsional wave travels along the waveguide and reflects back from its distal end. The echoes thus received are amplified and processed to extract the propagation velocity and/or the attenuation of the waves in the waveguide. Both velocity and attenuation are affected by coupling between the motions of the waveguide surface associated with the propagating wave in the rod and motions in the liquid in contact with the waveguide. Under the common condition that phase speed within the rod is greater than the speed of viscous waves in the surrounding fluid, such coupling will cause radiation of energy into the fluid, reducing both speed and amplitude of the guided wave. McSkimin [7] made the first application based on this property of travelling torsional waves to measure fluid dynamic shear viscosity and stiffness of viscous liquids. Later, Lynnworth [22] designed a sensor based on the velocity of the propagating torsional stress waves to measure fluid density. A quantitative theory was developed by Bau [23] and optimised by Kim and Bau [24] and Fan et al. [25]. Kim and Bau [9] and Shepard et al. [14] applied this theory further using rectangular and circular cross-sections to measure density and viscosity simultaneously. Their work led to successful viscosity measurements of simple fluids. Costley et al. [12] and Vogt et al. [15] developed methods for measuring viscosity based on the attenuation of the fundamental torsional mode for measuring the viscosity of fluids. Their measurements showed good agreement with viscosity measurements using conventional rheometers when testing general purpose viscosity standards. Kim et al. [10] derived an asymptotic formula for attenuation of the fundamental torsional mode in a rod loaded by viscous fluid. Based on their model Rabani et al. [18] proved that the attenuation of the fundamental torsional wave is more sensitive to the viscosity of a surrounding fluid compared to the velocity of the guided fundamental torsional wave and derived an explicit expression to calculate viscosity in a fluid of known density from the measured attenuation. They showed successful viscosity measurements of Newtonian fluids; however, anomalous behaviour was observed in the measurement of the viscosity of engine, gear, and silicone oils. This anomalous behavior was explained by molecular relaxation phenomena, which has a significant effect on the viscosity of polymer-based fluids with long chain molecules. They extended their work further by developing an analytical model to estimate the torsional guided wave viscosity sensor shear rate [26]. It would thus appear that the ultrasonic torsional guided wave technique can be successfully employed to measure the viscosity of wide range of fluids. The electronic systems used in the works outlined above are mainly designed to support a wide range of guided wave experiments in the laboratories and the experiments were mainly done using a windowed sine wave burst as the excitation signal. These systems are bulky and expensive, and they can provide for many different signal settings which are not required for monitoring viscosity online and in situ in process plants. However, a self-contained sensor system based on bursts of square waves instead of sine waves would have a considerable advantage in terms of manufacturing simplicity and cost. This paper explores the possibility of using square wave bursts as the driving signal source and proposes a simple design of

Sensors 2016, 16, 1681

3 of 13

Sensors 2016, 16, 1681 3 of 13 a compact and fully automatic analogue square wave front-end for the ultrasonic torsional guided wave sensor.

simple design of a compact and fully automatic analogue square wave front-end for the ultrasonic torsional guided wave sensor.

2. Sensor Excitation Method

2.The Sensor Excitation Methodguided wave viscosity sensor consists of a waveguide and a torsional ultrasonic torsional wave transduction system [26].guided The waveguide is an elastic cylindrical rod which is aintorsional a test fluid. The ultrasonic torsional wave viscosity sensor consists of a waveguide and Thewave torsional waves are excited ultrasonic transducers. wave transduction system [26]. by Theutilizing waveguide is an elastic cylindrical An rod ultrasonic which is in torsional a test fluid. transducer converts electrical energybyinto acoustic energy in transmission whentorsional its activewave element The torsional waves are excited utilizing ultrasonic transducers. An mode ultrasonic transducer electrical acoustic in reception transmission mode its active is excited by a converts voltage or currentenergy signal;into on the otherenergy hand, in mode thewhen acoustic energy of element is excited by a voltage or current signal; on the other hand, in reception mode the acoustic the returning torsional ultrasonic wave is converted into an electrical signal. energy of the returning torsional ultrasonic wave is converted intoare anextensively electrical signal. Piezoelectric materials, particularly piezoelectric ceramics, in use for generating Piezoelectric materials, particularly piezoelectric ceramics, are extensively in use generating and receiving ultrasonic waves in medical imaging, non-destructive testing, andforprocess control and receiving ultrasonic waves in medical imaging, non-destructive testing, and process control applications [27]. These materials are also highly durable; they have high tensile strength and applications [27]. These materials are also highly durable; they have high tensile strength and their their polarization geometry can be designed-in. These features make them more suitable for sensor polarization geometry can be designed-in. These features make them more suitable for sensor design. design. PIC 255 piezoelectric ceramic plates (PI Ceramic GmbH, Lederhose, Germany) which operate PIC 255 piezoelectric ceramic plates (PI Ceramic GmbH, Lederhose, Germany) which operate in in thickness mode are are well wellsuited suitedfor foruse useininthe theultrasonic ultrasonic guided wave torsional sensor thickness shear shear mode guided wave torsional sensor ◦ transducer to generate torsional modifiedPZT PZT with C Curie temperature, transducer to generate torsionalmodes. modes. PIC255 PIC255 isisaamodified with 350350 °C Curie temperature, 24002400 permittivity in in thethe polarization 0.66 coupling couplingfactor factorforfor thickness shear oscillation permittivity polarizationdirection, direction, 0.66 thethe thickness shear oscillation −12−12N/C shear charge coefficient. It is suitable for low-power ultrasonic transducers due andand 550 550 × 10 × 10 N/C shear charge coefficient. It is suitable for low-power ultrasonic transducers due its low mechanical quality factor (i.e., which means largerpower powerfor foroff-resonant off-resonantexcitations excitations [28], to itstolow mechanical quality factor (i.e., 80),80), which means larger low temperature coefficient (i.e., 0.004/K). and[28], lowand temperature coefficient (i.e., 0.004/K). ordertotoexcite excite torsional waves in the rod, therod, design on the shearon to the In In order torsionalguided guided waves insensor the sensor thebased design based torsional mode conversion was adapted [29]. Hence, two rectangular shear plates (3 mm × 2 mm × shear to torsional mode conversion was adapted [29]. Hence, two rectangular shear plates 0.2 mm) are coupled on different sides at one end of the axially-elongated rod with direction of (3 mm × 2 mm × 0.2 mm) are coupled on different sides at one end of the axially-elongated rod polarization opposed each to other (Figure 1); to provide the required robustness, the whole assembly with direction of polarization opposed each to other (Figure 1); to provide the required robustness, is encapsulated within a PVC cap using general purpose epoxy resin. the whole assembly is encapsulated within a PVC cap using general purpose epoxy resin.

Figure guidedwave waveviscosity viscosity sensor. Figure1.1.Ultrasonic Ultrasonic torsional torsional guided sensor.

Once voltagesapplied appliedtoto the the plates plates in to to thethe direction of the Once thethe voltages in aa direction directionperpendicular perpendicular direction of the polarization, shear stressescouple couplein inaaplane plane perpendicular perpendicular totothe axis of of thethe rodrod which generates a polarization, shear stresses the axis which generates a torsional wave in the rod. For large diameter rods, it is possible to use more than two shear plates all torsional wave in the rod. For large diameter rods, it is possible to use more than two shear plates placed perpendicular to the axis and around the circumference of the rod such that the adjacent plates all placed perpendicular to the axis and around the circumference of the rod such that the adjacent have the same polarization and vibrate in the circumferential direction. In principle, shearing plates have the same polarization and vibrate in the circumferential direction. In principle, shearing piezoelectric transducers, if themselves perfect, and if perfectly attached to the sensor rod, should piezoelectric transducers, if themselves perfect, and if perfectly attached to the sensor rod, should excite only a torsional mode. However, such perfection is difficult to achieve in practice and some excite onlywill a torsional However, such perfection difficult to Figure achieve some energy exist in mode. non-torsional modes such as L(0,1) is and F(1,1) in 2, in forpractice example.and It is energy will exist in non-torsional modes such as L(0,1) and F(1,1) inthat Figure for example. It ismode therefore therefore necessary that the frequency-diameter product be such the2,required torsional necessary that the frequency-diameter product be such that the required torsional mode travels travels at a speed that is significantly different from that of the unwanted modes—in order to allow at a speed that is significantly different from that processing. of the unwanted modes—in to allow time an gating time gating of the torsional signal for further Referring to Figure order 2, this would require of the torsional signal for further processing. Referring to Figure 2, this would require an operating

Sensors 2016, 16, 1681

4 of 13

Sensors 2016, 16, 1681

4 of 13

frequency well below 1 MHz for the sensor with 1mm diameter steel rod where T(0,1) mode has higher operating frequency well below 1 MHz for the sensor with 1mm diameter steel rod where T(0,1) speed compare to F(1,1) and lower speed compare to L(0,1) modes. The maximum centre frequency mode has higher speed compare to F(1,1) and lower speed compare to L(0,1) modes. The maximum of thecentre induced torsional wave in the torsional viscosity sensor with 1mm diameter steel rod and two frequency of the induced torsional wave in the torsional viscosity sensor with 1mm diameter rectangular piezoelectric shear transducers with 3 mm length, 2with mm3width, and 0.2 mmwidth, thickness steel rod and two rectangular piezoelectric shear transducers mm length, 2 mm and is lying approximately in the 500 to 700 kHz range which perfectly satisfies operability requirement 0.2 mm thickness is lying approximately in the 500 to 700 kHz range which perfectly satisfiesfor the torsional viscosity sensor.for the torsional viscosity sensor. operability requirement

Dispersion curves carbon steel rods: frequencyFigureFigure 2. 2.Dispersion curvesforforcylindrical cylindrical carbon steel Group rods: velocity Groupversus velocity versus diameter product. frequency-diameter product.

The frequency response of the tone-burst excitation signal is narrow band with suppressed side

The frequency response of thetotone-burst excitation signal isguided narrowwave band with suppressed side lobes, so it is perfectly matched the requirements for torsional applications where the lobes,excitation so it is perfectly matched to the requirements for torsional guided wave applications where of fundamental torsional mode is of interest. Square wave burst signals have several the excitation ofand fundamental torsional mode isonly of interest. Square wave burst have several harmonics stronger side lobes. However, the first harmonic is within the signals frequency response of theand torsional sensor transducers and thusonly higher are not tothe have any effectresponse on the harmonics stronger side lobes. However, theharmonics first harmonic is likely within frequency modes in the sensor. Therefore, is possible to use the harmonic spectrum in theon of theexcited torsional sensor transducers and thusithigher harmonics are first not likely to have any effect attenuation measurements, since this has a similar frequency spectrum to the tone-burst signal. the excited modes in the sensor. Therefore, it is possible to use the first harmonic spectrum in the Figure 3 shows the since time and plotsfrequency of the received signalsto forthe tone-burst, unipolar, attenuation measurements, thisfrequency has a similar spectrum tone-burst signal.and bipolar square wave 10-cycle burst excitation signals in 1 mm carbon steel rod at 625 kHz immersed Figure 3 shows the time and frequency plots of the received signals for tone-burst, unipolar, and in a viscous liquid. The received signal is attenuated and deformed due to losing energy by travelling bipolar square wave 10-cycle burst excitation signals in 1 mm carbon steel rod at 625 kHz immersed in through the immersed sensor rod and capacitive nature of the transducers, respectively. In the a viscous liquid. The received signal is attenuated and deformed due to losing energy by travelling frequency domain, the higher harmonics that are expected from the square wave bursts are greatly through the immersed sensor rodare and capacitive nature of the respectively. suppressed. The lobes which clearly observable around thetransducers, first harmonic for unipolar In andthe frequency domain, the higher harmonics that are expected from the square wave bursts are greatly bipolar square wave bursts result from the fact that the window length for the transmitted burst was suppressed. lobes which are implies clearly observable harmonic for unipolar and bipolar 17.5 μs;The the reciprocal of this a frequency around domain the lobefirst width of 57 kHz. square wave bursts result from operation the fact that the window length for the transmitted burst was 17.5 To evaluate the sensor with different excitation signals, the probe was immersed in aµs; Newtonian liquidawith consistent behavior range of shear rates and frequencies (i.e. the reciprocal of sample this implies frequency domain lobeacross widtha of 57 kHz. olive oil), and threeoperation waveformswith (toneburst, unipolar, andsignals, bipolar the square wave bursts) each 10 To evaluate thethe sensor different excitation probe was immersed in a cycles long were excited using an arbitrary function generator (DUI, NDT Solutions Ltd, Chesterfield, Newtonian sample liquid with consistent behavior across a range of shear rates and frequencies (i.e., UK).and Fivethe attenuation measurements at each of threeand different frequencies (575bursts) kHz, 600 kHz, and olive oil), three waveforms (toneburst, unipolar, bipolar square wave each 10 cycles 625 kHz) were made using each waveform in random order. In order to measure the guided wave long were excited using an arbitrary function generator (DUI, NDT Solutions Ltd, Chesterfield, UK). attenuation in the sensor rod at the given frequencies, the sensor was immersed in the test liquid and Five attenuation measurements at each of three different frequencies (575 kHz, 600 kHz, and 625 kHz) the echo signals from the distal end of the sensor rod were recorded. The received signals from the were made using each waveform in random order. In order to measure the guided wave attenuation sensor were amplified up to 57 dB, digitised at 50 MHz, and averaged 100 times by the same arbitrary in thefunction sensor rod at the The given frequencies, the sensor was immersed in the anddomain the echo generator. received time domain signals were transformed intotest the liquid frequency signalsbyfrom the distal end of the sensor rod were recorded. The received signals from the sensor were FFT. The attenuation coefficient is calculated from the frequency domain relationship: amplified up to 57 dB, digitised at 50 MHz, and averaged 100 times by the same arbitrary function 1 XCAL (ω) α(ω) = transformed ln generator. The received time domain signals were into the frequency domain by(1) FFT. 2l X(ω) The attenuation coefficient is calculated from the frequency domain relationship: α (ω ) =

1 XCAL (ω ) ln 2l X (ω )

(1)

Sensors 2016, 16, 1681

5 of 13

where, l is the sensor length, and XCAL(ω) and X(ω) are the frequency domain amplitudes of 5the Sensors 2016, 16, 1681 of 13 Sensors 2016, 16, 1681from a sensor in air/vacuo and the received signal from an immersed sensor in 5 of 13 received signals the test liquid, The received signals from immersed sensordomain in a liquid diminish where, l isrespectively. the sensor length, and XCAL (ω) and X(ω)anare the frequency amplitudes of in the amplitude compared to the signals received from an un-immersed sensor. The sensor length was set received signals from a sensor in air/vacuo and the received signal from an immersed sensor in the where, l is the sensor length, and XCAL (ω) and X(ω) are the frequency domain amplitudes of the so as to bring the measured loss as close as possible to 1 Np to achieve the optimum condition for test liquid, respectively. Theinreceived an immersed in a liquid diminish in received signals from a sensor air/vacuosignals and thefrom received signal fromsensor an immersed sensor in the test minimum error in attenuation [18]. The average value of the measured attenuation on each waveform amplitude compared to the signals received from an un-immersed sensor. The sensor length was set liquid, respectively. The received signals from an immersed sensor in a liquid diminish in amplitude at each frequency ismeasured calculated and shown in 4a.sensor. At1each frequency, the optimum standard deviation of so as to bring loss as close asFigure possible to NpThe to achieve the condition for compared to thethe signals received from an un-immersed sensor length was set so as to bring the attenuation for each group of five readings was calculated for each signal type. These were found minimum error [18]. The value ofthe theoptimum measured attenuation on each waveform the measured lossinasattenuation close as possible to 1average Np to achieve condition for minimum error in to be less or equal to 6.7%, 4.4%, and 8.5% for tone-burst, unipolar square wave burst and bipolar at each frequency is calculated and shown in Figure 4a. At each frequency, the standard deviation of attenuation [18]. The average value of the measured attenuation on each waveform at each frequency square wave burst signals, respectively. This is shown on Figure 4a for the tone-burst as solid error the attenuation for each group of five readings was calculated for each signal type. These were found is calculated and shown in Figure 4a. At each frequency, the standard deviation of the attenuation bars. to each be less or equal to readings 6.7%, 4.4%, 8.5% forfor tone-burst, unipolar squarewere wave bursttoand bipolar for group of five wasand calculated each signal type. These found be less or square respectively. This unipolar is shownsquare on Figure 4a burst for the tone-burst as solidwave error equal towave 6.7%,burst 4.4%,signals, and 8.5% for tone-burst, wave and bipolar square 1 0.4 bars. signals, respectively. This is shown on Figure 4a for the tone-burst as solid error bars. burst 0 1

0.2 0.4

0

0 0.2

Amplitude (arb.) Amplitude (arb.)

Amplitude (mV) Amplitude (mV)

-1 1 -1 0 1

0

0.5

-1 1

1

0

-1 0 1

0

1 0.5

0.5

-1 600 0

620

640

620

0 1

0 0 0.5

660

Time (s)

-1 600

1

0.5

1

1.5

Frequency (MHz)

640

660

0

0

0.5

1

1.5

(s) spectrum (right) Frequency (MHz) T(0,1) mode, from 10 Figure 3. Time domain (left) andTime amplitude of the received cycles of tone-burst (top), unipolar square wave burst (middle), and bipolar square wave burst Figure 3. Time domain (left) and amplitude spectrum (right) of the received T(0,1) mode, from 10 cycles (bottom) excitation signals(left) in 1 mm carbon steel (right) rod at frequency of 625 T(0,1) kHz after travelling 3. Time domain and diameter amplitude spectrum of the received mode, from 10 ofFigure tone-burst (top), unipolar square wave burst (middle), and bipolar square wave burst (bottom) down the immersed sensor rod in a viscous liquid and reflection from the distal end. cycles of signals tone-burst (top), unipolar square wave burst (middle), and bipolar square wave excitation in 1 mm diameter carbon steel rod at frequency of 625 kHz after travelling downburst the (bottom) excitation in 1 mm diameter carbon steel frequency immersed sensor rodsignals in a viscous liquid and reflection fromrod theatdistal end. of 625 kHz after travelling down the immersed sensor rod in a viscous liquid and reflection from the distal end. 3.5

2.1 T one-burst Unipolar square-wave burst Bipolar square-wave burst

2

3.5 2.5

1.92.1

-1

-1

(Np m ) -1  (Np m )

(Np m ) -1  (Np m )

T one-burst Unipolar square-wave burst Bipolar square-wave burst

1.8 2 1.71.9

1.51.7

2

2 1

1.5 0.5

1.41.6

0.45

0.5

0.55

0.6

0.65

0.7

0.75

0.8

(a) 0.45

0.5

0.55

0.6

0.65

Frequency (MHz)

1 0 0.3

0.4

0.7

0.75

0.8

0 0.3

0.5

0.6

0.7

0.8

0.9

Frequency (MHz)

0.5

Frequency (MHz)

1.4 1.3 0.4

Unipolar square-wave burst Bipolar square-wave burst Predicted

3

2.5 1.5

1.61.8

1.31.5 0.4

Unipolar square-wave burst Bipolar square-wave burst Predicted

3

(b) 0.4

0.5

0.6

0.7

0.8

0.9

Frequency (MHz)

Figure steel rod immersed Figure4.4.Attenuation Attenuationmeasurements measurementsfor forthe theT(0,1) T(0,1)mode modefor for1mm 1mmdiameter diametercarbon carbon (b)steel rod immersed (a) in olive oil oil(a) (a)using using tone-burst, unipolar square and bipolar square wave burst as in olive tone-burst, unipolar square wavewave burst,burst, and bipolar square wave burst as excitation excitation signals. Average values of five recordings are shown as symbols. Solid error bars represent Figure Average 4. Attenuation for are the shown T(0,1) mode for 1mm diameter carbon steel rod signals. valuesmeasurements of five recordings as symbols. Solid error bars represent theimmersed standard the deviation of the measurements using(b) the tone-burst; (b) using unipolar and bipolar in standard olive of oilthe (a)measurements using tone-burst, square wave burst, and bipolar square wave burst deviation usingunipolar the tone-burst; using unipolar and bipolar square wave burstsas square wave bursts as excitation along with predicted values. as excitation along with predicted excitation signals. Average valuesvalues. of five recordings are shown as symbols. Solid error bars represent the standard deviation of the measurements using the tone-burst; (b) using unipolar and bipolar

The tone-burst signal gained general as the appropriate driving signal for guided square wave bursts as has excitation along withacceptance predicted values. The tone-burst signal has gained general acceptance as the appropriate driving signal for guided wave applications. As mentioned before, it is a narrowband signal without side lobes. In attenuation wave applications. As mentioned before, it is a narrowband signal without side lobes. In attenuation The tone-burst signal has gained acceptance as the appropriate signal measurement, a single frequency fromgeneral the frequency domain signal is used, driving in which case for the guided shape wave applications. As mentioned before, it is a narrowband signal without side lobes. In attenuation

Sensors 2016, 16, 1681 Sensors 2016, 16, 1681

6 of 13 6 of 13

measurement, a single frequency from the frequency domain signal is used, in which case the shape of the signal does not have an effect on the measurement. Hence, as can be seen in Figure 4a, the of the signal does not have an effect on the measurement. Hence, as can be seen in Figure 4a, differences between attenuation determined using different driving signals are statistically the differences between attenuation determined using different driving signals are statistically insignificant. The only difference between unipolar and bipolar square wave signals is the DC insignificant. The only difference between unipolar and bipolar square wave signals is the DC component associated with the unipolar square wave signal. Since the piezoelectric shear plates used component associated with the unipolar square wave signal. Since the piezoelectric shear plates used in the torsional probe are capacitive and act as bandpass filters, the unipolar square wave shows the in the torsional probe are capacitive and act as bandpass filters, the unipolar square wave shows the same behaviour as the bipolar square wave. Comparing the attenuation measured by unipolar and same behaviour as the bipolar square wave. Comparing the attenuation measured by unipolar and bipolar square wave burst signals shows that either of the signals can replace the tone-burst signal. bipolar square wave burst signals shows that either of the signals can replace the tone-burst signal. To further validate the applicability of the unipolar and bipolar square wave burst excitation To further validate the applicability of the unipolar and bipolar square wave burst excitation signals in the operation of the ultrasonic torsional guided wave viscosity sensor, attenuation signals in the operation of the ultrasonic torsional guided wave viscosity sensor, attenuation measurements for the T(0,1) mode in 1mm diameter carbon steel rod immersed in olive oil at measurements for the T(0,1) mode in 1mm diameter carbon steel rod immersed in olive oil at frequencies ranging from 525 kHz to 675 kHz with intervals of 25 kHz were carried out. As illustrated frequencies ranging from 525 kHz to 675 kHz with intervals of 25 kHz were carried out. As illustrated in Figure 4b, there is a good agreement between the measured and the predicted attenuation values in Figure 4b, there is a good agreement between the measured and the predicted attenuation values calculated from the attenuation expression for T(0,1) derived by Kim and Bau [9]. Having established calculated from the attenuation expression for T(0,1) derived by Kim and Bau [9]. Having established that no significant difference exists in the attenuation when using different driving signals, and that that no significant difference exists in the attenuation when using different driving signals, and that the measured attenuation from different signals agrees with expectation, the unipolar square wave the measured attenuation from different signals agrees with expectation, the unipolar square wave is is the best candidate on the grounds of cost and simplicity of design. the best candidate on the grounds of cost and simplicity of design. 3. Electronic Front-End System Design 3. Electronic Front-End System Design There are trade-offs associated with designing ultrasound front-end circuits which relate to There are trade-offs associated with designing ultrasound front-end circuits which relate to performance requirements, underlying physics, and cost. It is essential to understand the performance requirements, underlying physics, and cost. It is essential to understand the specifications specifications that are of particular importance and their potential effects on system performance that that are of particular importance and their potential effects on system performance that may limit may limit design choices. In the following, a simple implementation of a square wave ultrasonic design choices. In the following, a simple implementation of a square wave ultrasonic front-end for front-end for use with the ultrasonic torsional guided wave viscosity sensor is described. The use with the ultrasonic torsional guided wave viscosity sensor is described. The proposed circuit proposed circuit design contains four major components: a pulser circuit, a protection circuit, a design contains four major components: a pulser circuit, a protection circuit, a receiver amplifier, and a receiver amplifier, and a power supply. These analogue electronic components are key to the overall power supply. These analogue electronic components are key to the overall system performance. Their system performance. Their characteristics define the limits to that performance in terms of unwanted characteristics define the limits to that performance in terms of unwanted noise and signal distortions. noise and signal distortions. Figure 5 shows a schematic of the proposed front-end system. Figure 5 shows a schematic of the proposed front-end system.

Power Supply Circuit

(12V)

Square-Wave Burst Generator

Adaptor/Battery

Pulser Circuit

Protection Circuit

VGA

DLR

Digitiser

Transducers

Driver

Receiver Amplifier

Figure 5. Schematic of the ultrasonic unipolar square wave front-end. Figure 5. Schematic of the ultrasonic unipolar square wave front-end.

The pulser circuit acts as a transmitter which generates the basic square wave burst signal in two circuitsquare acts as wave a transmitter whichand generates the basic squareby wave burst signaldriver in two steps:The firstpulser a unipolar is generated this is then amplified a high voltage steps: first a unipolar square wave is generated and this is then amplified by a high voltage driver that connects to the transducers. On the receiver side, there is a protection circuit which blocks the that connects to the transducers. On the receiver side, there is a protection circuit which blocks the high voltage signals from the transmitter. It is followed by a low-noise variable gain amplifier and high voltage signals from the transmitter. It is followed by a low-noise variable gain amplifier and a differential line receiver. The amplified signal can then be sent to a digitiser for further processing.a differential line receiver. amplified signal can then be sent to for a digitiser for further processing. The power supply circuit The provides the different voltages required all the modules in the system. The power supply circuit provides the different voltages required for all the modules in the system. In the proposed system there is a capability to control the number of cycles, frequency, and repetition In the proposed system there is a capability to control the number of cycles, frequency, and repetition rate of the excitation signal, and also the gain of the receiver amplifier. These provide the ability rate of the excitation signal, and also the gain of the receiver amplifier. These provide the ability to

Sensors 2016, 16, 1681 Sensors 2016, 16, 1681

7 of 13 7 of 13

generate unipolar square wave bursts with different characteristics for experimental purposes. However, inunipolar real-timesquare and in wave situ applications ultrasonic torsional guided wave viscosity sensor to generate bursts withthe different characteristics for experimental purposes. will only require excitation with fixed frequency andtorsional number of cycles. A front-end However, in real-time and insignals situ applications the ultrasonic guided wave viscositysystem sensor with fixed frequency and number of cycles provides possibility for a compact design with no prewill only require excitation signals with fixed frequency and number of cycles. A front-end system with settings that canand be number easily integrated into the possibility sensor package for standalone permanent/remote fixed frequency of cycles provides for a compact design with no pre-settings installations. that can be easily integrated into the sensor package for standalone permanent/remote installations. 3.1. The The Pulser Pulser Circuit Circuit 3.1. The primary primary objective objective of of the the pulser pulser circuit circuit is is to to generate generate and and amplify amplify aa burst burst of of square wave wave The signals for for transmission. transmission. This This isis achieved achieved by by aa two-stage two-stage design, design, namely namely aa square square wave wave generator generator signals module and the driver module module module (Figure (Figure6). 6). In In the the first firststage, stage,the thesquare squarewave waveburst burstgenerator generatorinitiates initiatesa V. In the the next nextstage, stage, asequence sequenceofofunipolar unipolarpulses pulses(square (squarewave wavebursts) bursts)with withmaximum maximumamplitude amplitude of of 55V. the driver amplifies the sequence of pulses to 72 V amplitude. the Square-wave burst generator

Timer 2

Driver

Expander

Timer 1 Buffers

Ultrasonic pulser

Counter

Figure 6. Simplified diagram of the pulser circuit. Figure 6. Simplified diagram of the pulser circuit.

The square square wave wave burst burst generator generator is is formed formed of of two National The two CMOS CMOS timers timers (e.g., (e.g., LMC555, LMC555, National Semiconductor Corp., Clara, CA, USA), a binaryacounter HFE 4520 NXP Semiconductors Semiconductor Corp.,Santa Santa Clara, CA, USA), binary(e.g., counter (e.g.,B, HFE 4520 B, NXP Netherlands B.V., Eindhoven, The Netherlands), and three logic gates (Figure 6). Timer 1 and Timer6).2 Semiconductors Netherlands B.V., Eindhoven, The Netherlands), and three logic gates (Figure are configured as astable and monostable multivibrators, respectively. Timer 1respectively. generates theTimer square Timer 1 and Timer 2 are configured as astable and monostable multivibrators, 1 wave burst at frequencies which can be set up to 3 MHz. Timer 2 controls the repetition rate of the the generates the square wave burst at frequencies which can be set up to 3 MHz. Timer 2 controls burst. Therate counter the The logiccounter gates set of cycles fornumber each burst. The for frequency and repetition of theand burst. andthe thenumber logic gates set the of cycles each burst. repetition rate and are repetition settable byrate changing the timing resistors Timer 1 andfor Timer 2, 1respectively. The frequency are settable by changing the for timing resistors Timer and Timer This has been done using jumper switches on the prototype system in this study for experimental 2, respectively. This has been done using jumper switches on the prototype system in this study for purposes. As mentioned earlier, in industrial applications there is no need for variable frequency and experimental purposes. As mentioned earlier, in industrial applications there is no need for variable repetition rate a single timing for eachresistor timer will be sufficient. frequency and and repetition rate and aresistor single timing for each timer will be sufficient. An external 5 V trigger signal starts the driving process; Timer starts working working as as aa free free running running An external 5 V trigger signal starts the driving process; Timer 11 starts multivibrator, generating square wave signals. The counter counts counts the number pulses of generated multivibrator, generatingthe the square wave signals. The counter the of number pulses by Timer 1 and it triggers Timer 2 as soon as it reaches the maximum number of pulses set by the logic generated by Timer 1 and it triggers Timer 2 as soon as it reaches the maximum number of pulses set gates. 2 generates output pulseoutput which pulse resets which Timer resets 1. The Timer width 1. ofThe the output pulse by the Timer logic gates. Timer a2 single generates a single width of the from Timer defines the repetition raterepetition of the square wave circuit generates a continuous output pulse2 from Timer 2 defines the rate of the burst. squareThe wave burst. The circuit generates sequence of square wave burst packets automatically. a continuous sequence of square wave burst packets automatically. The driver driver consists consists of of two two input input buffers, buffers, an an ultrasonic ultrasonic pulser pulser and and an an expander expander circuit. circuit. The The signals signals The are fed fed to to the The ultrasonic ultrasonic pulser pulser are the ultrasonic ultrasonic pulser pulser through through inverting inverting and and non-inverting non-inverting buffers. buffers. The amplifies the square wave signals to an appropriate level to drive the transducers through the expander amplifies the square wave signals to an appropriate level to drive the transducers through the circuit. In circuit. common manywith ultrasonic circuits, thecircuits, expander consists of two parallel expander Inwith common many driver ultrasonic driver thecircuit expander circuit consists of diodes in opposite directions. It conducts high-voltage amplified signals generated by the pulser two parallel diodes in opposite directions. It conducts high-voltage amplified signals generatedand by connects pulser to thethe transducers during the transmission phase. Conversely, it stops conduction the pulserthe and connects pulser to the transducers during the transmission phase. Conversely, it at the time of termination of the burst sequence and isolates the pulser from the transducer during stops conduction at the time of termination of the burst sequence and isolates the pulser from the the reception phase. this work, the Supertex HV748 Sunnyvale, USA) is used transducer during theInreception phase. In this work, the(Supertex Supertex Inc., HV748 (SupertexCA, Inc., Sunnyvale, as the ultrasonic pulser. It is a monolithic high speed, high voltage, ultrasonic transmitter pulser. CA, USA) is used as the ultrasonic pulser. It is a monolithic high speed, high voltage, ultrasonic It is designed as a high driver piezoelectric transducers and other transducers applications.and Its circuit transmitter pulser. It isvoltage designed as afor high voltage driver for piezoelectric other

Sensors 2016, 16, 1681 Sensors 2016, 16, 1681

8 of 13 8 of 13

applications. Its circuitlogic consists of controller logic circuits, a level translator, gate driving buffers, consists of controller circuits, a level translator, gate driving buffers, and a high current and and high avoltage high current andoutput high voltage MOSFET outputofstage. The output stages this integrated circuit MOSFET stage. The output stages this integrated circuit canofprovide output voltages can voltages up to up ±75to V± and currents upattofrequencies ±1.8 A. It can at frequencies up provide to ±75 Voutput and output currents 1.8output A. It can operate upoperate to 20 MHz depending up MHz depending on the load capacitance. onto the20load capacitance. In Inthe theultrasonic ultrasonictorsional torsionalguided guidedwave waveviscosity viscositysensor sensorapplications, applications,ititisisrequired requiredto togenerate generate high amplitude ultrasonic signals to have a higher signal to noise ratio, particularly for viscosity high amplitude ultrasonic signals to have a higher signal to noise ratio, particularly for viscosity measurements forfor which thethe attenuation of the signal is large. Hence, the measurementsofofhighly highlyviscous viscousliquids liquids which attenuation of the signal is large. Hence, ultrasonic pulser is provided with a maximum 72 V voltage at its high voltage power supply pin. the ultrasonic pulser is provided with a maximum 72 V voltage at its high voltage power supply pin. This Thisenables enablesthe thegeneration generationof ofunipolar unipolar sequences sequences with with aa maximum maximum 72 72 V V amplitude amplitude at at the the output. output. Under pulses, thethe HV748 internal MOSFET-gate-drives switch several timestimes and Underthe thesequence sequenceofof pulses, HV748 internal MOSFET-gate-drives switch several this draws large instantaneous currents and results in a voltage drop due to the impedance of the and this draws large instantaneous currents and results in a voltage drop due to the impedance of voltage source and the between the voltage source and the pulser. Thus,Thus, the the voltage source andconnection the connection between the voltage source andultrasonic the ultrasonic pulser. decoupling capacitor shouldshould be chosen carefully to maintain the supply at thelevel ultrasonic the decoupling capacitor be chosen carefully to maintain thevoltage supplylevel voltage at the pulser. This capacitor introduces a minimum delay to the system for generating pulses with the ultrasonic pulser. This capacitor introduces a minimum delay to the system for generating pulses maximum amplitudeamplitude due to initial the maximum voltage. Avoltage. 168 nF decoupling capacitor with the maximum duecharging to initial to charging to the maximum A 168 nF decoupling was used in the proposed design. Figure 7 shows the minimum charging time required to reach to capacitor was used in the proposed design. Figure 7 shows the minimum charging time required to 99% maximum voltage of 72 V.ofThe reduction in the in pulse amplitude after after 10 cycles of the reachoftothe 99% of the maximum voltage 72 V. The reduction the pulse amplitude 10 cycles of square wavewave pulses is less 1.5%.1.5%. the square pulses is than less than 70 60

Voltage (V)

50 40 30 20 10 0

0

200

400

600

800

1000

1200

1400

Time (s)

Figure7.7.Decoupling Decouplingcapacitor capacitorcharging chargingtime timeand andchange changeininthe thevoltage voltageatatpulser pulserhigh highvoltage voltagepower power Figure supply pin (dashed-dot line). 10 cycle output (solid line). supply pin (dashed-dot line). 10 cycle output (solid line).

In In the the measurement measurement of of torsional torsional wave wave attenuation, attenuation, all all the the measurements measurementsare are made made using using aa reference signal and so the reduction in the pulse amplitude is present in both reference reference signal and so the reduction in the pulse amplitude is present in both reference and and measurement signals and does not affect the measurement result. measurement signals and does not affect the measurement result. 3.2. 3.2.The TheProtection ProtectionCircuit Circuit The Thedriver drivercircuit circuitgenerates generateshigh-voltage high-voltagesquare squarewave waveburst burstpackets packets(maximum (maximum72 72V) V)on onthe the node where the transducers are connected. The receiver amplifier circuit is designed for relatively node where the transducers are connected. The receiver amplifier circuit is designed for relatively low-voltage low-voltagesignals. signals.Hence, Hence,totoprevent preventhigh-voltage high-voltagesquare squarewave wavebursts burstsfrom fromdamaging damagingthe thereceiver receiver amplifier, amplifier,the theprotection protectioncircuit circuitisisplaced placedbetween betweenthe thetransducers transducersand andthe thereceiver receiveramplifier amplifiercircuit. circuit. The while the the system system is is in inthe thetransmission transmissionstate stateand andas asa Theprotection protection circuit circuit behaves as an open circuit while ashort shortcircuit circuitwhile whilethe thesystem systemisisininthe thereception reception state.The The protection circuit this design is based state. protection circuit in in this design is based on on design Camacho Fritsch design is fully automatic has many advantages thethe design by by Camacho andand Fritsch [30].[30]. TheThe design is fully automatic and and has many advantages such such as operating without any control signals orvoltages, bias voltages, low harmonic distortion, low insertion as operating without any control signals or bias low harmonic distortion, low insertion losses, losses, low low noise, low dissipation, power dissipation, high bandwidth, and small size.8 shows Figure the 8 shows the low noise, power high bandwidth, and small size. Figure protection protection circuit arrangement. It is based on a pair of N-channel depletion MOSFETs. circuit arrangement. It is based on a pair of N-channel depletion MOSFETs.

Sensors 2016, 16, 1681 Sensors 2016, 16, 1681

9 of 13 9 of 13

Figure 8. Schematic of the automatic protection circuit.

In operation operation when gate-source voltage zero and and the the drain-source drain-source voltage voltage is is In when the the gate-source voltage is is very very close close to to zero less than the difference between the gate-source voltage and the threshold voltage, both MOSFETs less than the difference between the gate-source voltage and the threshold voltage, both MOSFETs conduct with with low low drain-source drain-source impedance. impedance. In voltage, the the diode diode conduct In the the presence presence of of aa positive positive input input voltage, associated with Q1 conducts, Q2 operates in the saturation region, and the limiter diodes conduct associated with Q1 conducts, Q2 operates in the saturation region, and the limiter diodes conduct the the drain saturation saturationcurrent currentsafely safelytotoground. ground. This circuit operates negative input, in which drain This circuit alsoalso operates withwith negative input, in which case, case, Q1Q2 and Q2 swap their roles. Q1 and swap their roles. 3.3. The Receiver Amplifier A simple two-stage receiver amplifier is designed for low-noise operation, high gain and single-ended output (Figure 5). It It is is compact compact and inexpensive, comprising comprising of just two integrated circuits. ItIt consists consistsofofa avariable variable gain amplifier (e.g., AD8331, Analog Devices Inc., Norwood, MI, circuits. gain amplifier (e.g., AD8331, Analog Devices Inc., Norwood, MI, USA) USA)a and a differential line receiver (e.g., MAX4444, Integrated Products, Sunnyvale, CA, and differential line receiver (e.g., MAX4444, MaximMaxim Integrated Products, Sunnyvale, CA, USA). USA). The variable gain amplifier has a gain which is up to 55.5andB. It has anpreamplifier embedded The variable gain amplifier has a gain which is settable upsettable to 55.5 dB. It has embedded preamplifier and symmetrical differential usable as an ultralow-noise to and symmetrical differential output. It isoutput. usable Itasisan ultralow-noise element upelement to 120 up MHz. 120 MHz. The differential line receiver has symmetrical differential a single-ended output The differential line receiver has symmetrical differential inputs inputs and a and single-ended output and and closed-loop dB. When the system is reception in the reception mode, thewhich signalsare which are closed-loop gain gain of +6 of dB.+6When the system is in the mode, the signals received received by theamplifier receiver from amplifier from the transducers thecircuit protection will up be by the receiver the transducers through thethrough protection will becircuit amplified amplified up to 55.5 dB by the variable gain amplifier in the first stage. In the second stage, the to 55.5 dB by the variable gain amplifier in the first stage. In the second stage, the differential line differential line receiver convertsoutput the differential from variabletogain amplifier tosignal a singlereceiver converts the differential from the output variable gainthe amplifier a single-ended for ended signal for onward viaa coaxial to a distant onward transmission via transmission coaxial cable to distant cable processing unit. processing unit. 3.4. The Power Power Supply Supply Circuit Circuit 3.4. The The power supply circuit supplies electrical energy to each electronic module module in the system. It is designed to be robust, compact, and inexpensive. It consists of fixed fixed linear linear regulators, regulators, DC-DC voltage convertors, It works with 12 V DCDC external voltage source which can convertors, and andaahigh-voltage high-voltagepower powersupply. supply. It works with 12V external voltage source which be either an adaptor or a battery. The block diagram of the power supply circuit is given in Figure 9. can be either an adaptor or a battery. The block diagram of the power supply circuit is given in Figure A 5 V fixed linear regulator is employed to supply the square wave burst generator module. 9. Two A separate fixed linear regulators are provided to supply pulser with 3.3 Vgenerator and 9 V voltages. 5 V fixed linear regulator is employed to supply the the square wave burst module. The driver is also supplied with a high-voltage power supply which is formed by a triple output Two separate fixed linear regulators are provided to supply the pulser with 3.3 V and 9 V voltages. DC-DC voltage (e.g., aNMT0572SC, Muratasupply Powerwhich Solutions Inc., Mansfield, MA, USA) The driver is alsoconvertor supplied with high-voltage power is formed by a triple output DCwith voltages of 24 V, 48 V, and 72 V, selectable using jumpers, followed a high voltage adjustable DC voltage convertor (e.g., NMT0572SC, Murata Power Solutions Inc.,by Mansfield, MA, USA) with regulator to 24 regulate output. Also a DC-DCusing voltage convertor (e.g., IL0509S, XPvoltage Power, Sunnyvale, voltages of V, 48 the V, and 72 V, selectable jumpers, followed by a high adjustable CA, USA) is to supply floating Also 9 V, a reference the high-voltage power output, is regulator toused regulate the aoutput. a DC-DCto voltage convertor (e.g.,supply IL0509S, XP which Power, also required by the driver. The receiver amplifier circuit is supplied with ± 5 V fixed linear regulators. Sunnyvale, CA, USA) is used to supply a floating 9 V, a reference to the high-voltage power supply The introduction of noise by use a common power supply for thecircuit receiver amplifierwith and ±5 theVrest of output, which is also required byof the driver. The receiver amplifier is supplied fixed the system has been avoided by usingofa noise separate V linear regulator power for the receiver amplifier. linear regulators. The introduction by 5use of a common supply for the receiver

amplifier and the rest of the system has been avoided by using a separate 5 V linear regulator for the receiver amplifier.

Sensors 2016, 16, 1681

10 of 13

Sensors 2016, 16, 1681

10 of 13

Adaptor/ Battery

Power Supply Circuit

12V

5V DC-DC (Ref) 72V/48V/24V

3.3V

9V

Burst Square-Wave Generator

72V/ 48V/ 24V

Driver

DC-DC -9V

DC-DC Ref-9V

-5V

5V

Receiver Amplifier

Pulser Circuit

Figure 9. Block diagram of the power supply circuit. Figure 9. Block diagram of the power supply circuit.

SystemFunctional FunctionalVerification Verificationand andAnalysis Analysis 4.4.System To experimentally experimentally verify verify the the performance performance of of the the proposed proposed unipolar unipolar square square wave wave ultrasonic ultrasonic To front-end,the thetorsional torsional guided wave viscosity sensor immersed in aliquid test liquid (i.e.,virgin extra front-end, guided wave viscosity sensor was was immersed in a test (i.e., extra virgin olive oil) and connected to the designed analogue front-end through a 50 Ω coaxial cable olive oil) and connected to the designed analogue front-end through a 50 Ω coaxial cable in a in a temperature controlled room. amplified analogue received signals fromthe thetransducers transducersthat that temperature controlled room. The The amplified analogue received signals from weretransferred transferredto toaaportable portableoscilloscope oscilloscope(WaveJet (WaveJet324A, 324A,Teledyne TeledyneLeCroy LeCroyInc., Inc.,Chestnut ChestnutRidge, Ridge, were NY, USA) for visualization and then to a BNC connector block (NI BNC-2110, National Instruments NY, USA) for visualization and then to a BNC connector block (NI BNC-2110, National Instruments Corporation,Austin, Austin,TX, TX,USA) USA)using usingaa50Ω 50 Ωcoaxial coaxialcables. cables.The TheBNC BNCconnector connectorblock blockwas wasconnected connected Corporation, to a multifunction data acquisition board (NI PCI-6110, National Instruments Corporation, Austin, TX, to a multifunction data acquisition board (NI PCI-6110, National Instruments Corporation, Austin, USA) for digitization through a shielded cable (NI SH68-68-EPM, National Instruments Corporation, TX, USA) for digitization through a shielded cable (NI SH68-68-EPM, National Instruments Austin, TX, USA). A digital multimeter (FLUKE 289, Fluke Corporation, WA, USA) wasWA, used Corporation, Austin, TX, USA). A digital multimeter (FLUKE 289, FlukeEverett, Corporation, Everett, to monitor the temperature of the olive oil at the time of the measurements. The received analogue USA) was used to monitor the temperature of the olive oil at the time of the measurements. The signals were digitised at 5were MHzdigitised and averaged 50 times sent to50a times PC byand a GPIB (IEEE 488) received analogue signals at 5 MHz and and averaged sentinterface to a PC by a GPIB for further processing. The experimental is illustrated in Figure 10. interface (IEEE 488) for further processing.setup The experimental setup is illustrated in Figure 10. Attenuation measurements were performed at eight frequencies range to 708 Attenuation measurements were performed at eight frequencies in in thethe range 524524 to 708 kHzkHz at ◦ at 22.3 C. attenuation The attenuation coefficients are calculated using Equation (1). Theinvariation in the 22.3 °C. The coefficients are calculated using Equation (1). The variation the attenuation attenuation measurements, as a standard deviation over was five less recordings, was The less measurements, calculated as acalculated standard deviation over five recordings, than ±1.5%. than ± 1.5%. The measured attenuation of 1mm diameter carbon steel rod immersed in olive oil measured attenuation of 1mm diameter carbon steel rod immersed in olive oil is shown in Figure 11.is shownisin Figureagreement 11. There is a good the agreement between the measured There a good between measured attenuation and the attenuation predictions.and the predictions. Viscosities were calculated from the attenuation readings (eight readingsin intotal) total) using using the the Viscosities were calculated from the attenuation readings (eight readings explicit viscosity expression given by Rabani et al. [18]. The standard deviations of each group of explicit viscosity expression given by Rabani et al. [18]. The standard deviations of each group of eightcalculated calculatedviscosities viscositieswere wereless lessthan thanor orequal equalto to0.6%. 0.6%.Reference Referenceviscosity viscositymeasurements measurementswere were eight made for comparison using a Physica MCR 301 (Anton Paar GmbH, Graz, Austria) conventional made for comparison using a Physica MCR 301 (Anton Paar GmbH, Graz, Austria) conventional rotatingelement elementrheometer rheometeroperating operatingininsteady steadyshear shearmode modeusing usingcone coneand andplate plategeometry geometry(spindle: (spindle: rotating CP-50). A total number of 31 measurements were made on the test liquid at shear rates ranging from CP-50). A total number of 31 measurements were made on the test liquid at shear rates ranging from −1 . The average viscosity of the torsional guided wave sensor had an error less than 2% 0 to 1000 s 0 to 1000 s-1. The average viscosity of the torsional guided wave sensor had an error less than 2% comparedto tothe theaverage averageviscosity viscositymeasured measuredusing usingthe theconventional conventional rheometer rheometer (Table (Table 1). 1). compared

Sensors 2016,16, 16, 1681 Sensors Sensors 2016, 2016, 16, 1681 1681

11 of13 13 11 11 of of 13

Figure Figure10. 10.Unipolar Unipolarultrasonic ultrasonicsquare squarewave wavefront-end front-endexperimental experimentalsetup setupfor fortorsional torsionalguided guidedwave wave Figure 10. Unipolar ultrasonic square wave front-end experimental setup for torsional guided wave viscosity sensor. viscositysensor. sensor. viscosity 3 3 Experiment Experiment Prediction Prediction

-1-1

(Npmm ) ) (Np

2.5 2.5

2 2

1.5 1.5

1 1

0.5 0.50.3 0.3

0.4 0.4

0.5 0.5

0.6 0.6

0.7 0.7

Frequency Frequency (MHz) (MHz)

0.8 0.8

0.9 0.9

Figure 11.Measured Measured attenuationin in 1mmdiameter diameter carbonsteel steel rodembedded embedded inan an oliveoil oil versus Figure Figure 11. 11. Measured attenuation attenuation in 1mm 1mm diameter carbon carbon steel rod rod embedded in in an olive olive oil versus versus frequency, superimposed on the predicted attenuation curve. frequency, frequency, superimposed superimposed on on the the predicted predicted attenuation attenuation curve. curve. Table 1.Average Average viscosityof of thetest test liquid (torsional (torsional sensor vs. vs. conventionalrheometer). rheometer). Table Table 1. 1. Average viscosity viscosity of the the test liquid liquid (torsional sensor sensor vs. conventional conventional rheometer).

Testing Liquid Testing TestingLiquid Liquid 1 Extra virgin olive oil Extravirgin virginolive oliveoil oil11 Extra

Density Average Viscosity s) Viscosity (Pa· s) Average(Pa· Viscosity (Pa·Average s) Average Viscosity Density Average Viscosity (Pa· s) Average Viscosity (Pa· s)(Pa·s) 3) Density (kg·m−Physica −3) (kg· m MCR 301 Torsional Sensor −3 Physica Torsional Sensor (kg·m ) Physica MCR 301 MCR 301 Torsional Sensor 913.0 0.0756 0.0771 913.0 913.0 0.0756 0.0771 0.0756 0.0771 1 Sainsbury’s Supermarkets Ltd, London, UK. 1 Sainsbury’s 1 Sainsbury’s Supermarkets Ltd,London, London,UK. UK. Supermarkets Ltd,

5. 5. Conclusions Conclusions 5. Conclusions It It has has been been shown shown that that aa square square wave, wave, instead instead of of aa sine sine wave, wave, burst burst can can be be employed employed in in the the It has been shown that a square wave, instead of a sine wave, burst can be employed in the torsional guided wave viscosity sensor. The electronics for square wave burst signals are torsional guided wave viscosity sensor. The electronics for square wave burst signals are less less torsional guided wave viscosity sensor. The those electronicssine for squaresignals. wave burst signals the are less expensive expensive expensive and and simpler simpler to to implement implement than than those for for sine wave wave signals. However, However, the square square wave wave burst method is not in itself any better that the tone-burst case. Two types of square waves, burst method is not in itself any better that the tone-burst case. Two types of square waves, namely namely

Sensors 2016, 16, 1681

12 of 13

and simpler to implement than those for sine wave signals. However, the square wave burst method is not in itself any better that the tone-burst case. Two types of square waves, namely unipolar and bipolar, were considered for the design. The unipolar square wave was chosen as it requires a simpler electronic platform with fewer components thus reducing the size and the cost of the system. The design and implementation of an analogue front-end for the torsional guided wave viscosity sensor was presented. The design is simple, fully automatic, and robust. It works without the need of any external controlling signal. The cost of the device is considerably less than that of a conventional ultrasonic wave transmitter and receiver commonly used in many laboratories. It is compact and could be designed to fit into the same package as the transducer head for the torsional guided wave viscosity sensor. The proposed design is flexible for further design improvements; the output gain and the characteristics of the excited wave—such as the burst frequency, number of cycles, and repetition rate—are settable. It can be expanded to work with standalone processing chips such as microcontrollers and FPGAs. The characteristics and the performance of the designed ultrasonic front-end were verified experimentally. The operation of the system was stable and reproducible within an acceptable error range. The attenuation and viscosity measured by the torsional guided wave viscosity sensor using the designed device was in good agreement with the predicted values and with those measured using a commercial rheometer. Acknowledgments: The author would like to acknowledge Applied Ultrasonic Laboratory at the Department of Electrical and Electronic Engineering, The University of Nottingham for providing DUI ultrasonic wave generator for this research study. Conflicts of Interest: The author declares no conflict of interest.

References 1. 2. 3. 4.

5. 6. 7. 8. 9. 10. 11.

12.

13.

Darby, R. Chemical Engineering Fluid Mechanics; Marcel Dekker: New York, NY, USA, 1996. Barnes, H.A. On-line or process viscometry—A review. Appl. Rheol. 1999, 9, 102–107. Roberts, I.; McKenna, B. In-line and on-line rheology measurement of food. Texture Food 2003, 1, 161–182. Padua, A.A.H.; Tomida, D.; Yokoyama, C.; Abramson, E.H.; Berg, R.F.; May, E.F.; Moldover, M.R.; Laesecke, A. Viscometers. In Experimental Thermodynamics Volume IX: Advances in Transport Properties of Fluids; The Royal Society of Chemistry: Cambridge, UK, 2014; pp. 96–131. Papadakis, E.P. Ultrasonic Instruments and Devices: Reference for Modern Instrumentation, Techniques, and Technology; Academic Press: San Diego, CA, USA, 1999. Lynnworth, L.C. Ultrasonic Measurements for Process Control: Theory, Techniques, Applications; Academic Press: San Diego, CA, USA, 2013. McSkimin, H. Measurement of dynamic shear viscosity and stiffness of viscous liquids by means of traveling torsional waves. J. Acoust. Soc. Am. 1952, 24, 355–365. [CrossRef] Roth, W.; Rich, S.R. A new method for continuous viscosity measurement. General theory of the ultra-viscoson. J. Appl. Phys. 1953, 24, 940–950. [CrossRef] Kim, J.O.; Bau, H.H. Instrument for simultaneous measurement of density and viscosity. Rev. Sci. Instrum. 1989, 60, 1111–1115. [CrossRef] Kim, J.O.; Wang, Y.; Bau, H.H. The effect of an adjacent viscous fluid on the transmission of torsional stress waves in a submerged waveguide. J. Acoust. Soc. Am. 1991, 89, 1414–1422. [CrossRef] Kim, J.O.; Bau, H.H.; Liu, Y.; Lynnworth, L.C.; Lynnworth, S.A.; Hall, K.A.; Jacobson, S.A.; Korba, J.A. Torsional sensor applications in two-phase fluids. IEEE Trans. Ultra. Ferroelectr. Freq. Control 1993, 40, 563–576. [CrossRef] [PubMed] Costley, R.D.; Balasubramaniam, K.; Ingham, W.; Simpson, J.A.; Shah, V. Torsional waveguide sensor for molten materials. In Review of Progress in Quantitative Nondestructive Evaluation; Springer: New York, NY, USA, 1998; pp. 859–866. Costley, R.D.; Boudreaux, G.; Balasubramaniam, K.; Simpson, J.A. Waveguide sensor for liquid and slurry properties. In Review of Progress in Quantitative Nondestructive Evaluation; Springer: New York, NY, USA, 1999; pp. 1171–1176.

Sensors 2016, 16, 1681

14.

15. 16.

17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30.

13 of 13

Shepard, C.L.; Burghard, B.J.; Friesel, M.A.; Hildebrand, B.P.; Moua, X.; Diaz, A.A.; Enderlin, C.W. Measurements of density and viscosity of one-and two-phase fluids with torsional waveguides. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 1999, 46, 536–548. [CrossRef] [PubMed] Vogt, T.K.; Lowe, J.; Cawley, P. Measurement of the material properties of viscous liquids using ultrasonic guided waves. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 2004, 51, 737–747. [CrossRef] Ai, Y.; Lange, R.A. Theoretical analyses and numerical simulations of the torsional mode for two acoustic viscometers with preliminary experimental tests. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 2008, 55, 648–658. Prasad, V.; Balasubramaniam, K.; Kannan, E.; Geisinger, K. Viscosity measurements of melts at high temperatures using ultrasonic guided waves. J. Mater. Process. Technol. 2008, 207, 315–320. [CrossRef] Rabani, A.; Challis, R.E.; Pinfield, V.J. The torsional waveguide viscosity probe: Design and anomalous behavior. IEEE Trans. Ultra. Ferroelectr. Freq. Control 2011, 58, 1628–1640. Kazys, R.; Sliteris, R.; Raisutis, R.; Zukauskas, E.; Vladisauskas, A.; Mazeika, L. Waveguide sensor for measurement of viscosity of highly viscous fluids. Appl. Phys. Lett. 2013, 103, 204102. [CrossRef] Kazys, R.; Mazeika, L.; Sliteris, R.; Raisutis, R. Measurement of viscosity of highly viscous non-newtonian fluids by means of ultrasonic guided waves. Ultrasonics 2014, 54, 1104–1112. [CrossRef] [PubMed] Haser, A.N.; McGovern, M.E.; Behnia, B.; Buttlar, W.; Reis, H. Monitoring viscosity in asphalt binders using an ultrasonic guided wave approach. Insight-Non-Destr. Test. Cond. Monit. 2015, 57, 212–220. [CrossRef] Lynnworth, L.C. Slow Torsional Wave Sensors. In Proceeings of the Ultrasonics Symposium, Phoenix, AZ, USA, 26–28 October 1977; pp. 29–34. Bau, H.H. Torsional wave sensor—A theory. J. Appl. Mech. 1986, 53, 846–848. [CrossRef] Kim, J.O.; Bau, H.H. On line, real-time densimeter—Theory and optimization. J. Acoust. Soc. Am. 1989, 85, 432–439. [CrossRef] Fan, Z.; Lowe, M.; Castaings, M.; Bacon, C. Torsional waves propagation along a waveguide of arbitrary cross section immersed in a perfect fluid. J. Acoust. Soc. Am. 2008, 124, 2002–2010. [CrossRef] [PubMed] Rabani, A.; Pinfield, V.J.; Challis, R.E. Rate of shear of an ultrasonic oscillating rod viscosity probe. Ultrasonics 2016, 65, 18–22. [CrossRef] [PubMed] Krautkrämer, J.; Krautkrämer, H. Ultrasonic Testing of Materials; Springer Science & Business Media: New York, NY, USA, 2013. Erturk, A.; Inman, D.J. Piezoelectric Energy Harvesting; John Wiley & Sons: Chichester, UK, 2011. Vogt, T.K. Determination of Material Properties Using Guided Waves. Ph.D. Thesis, Imperial College London, University of London, London, UK, October 2002. Camacho, J.; Fritsch, C. Protection circuits for ultrasound applications. IEEE Trans. Ultrason. Ferroelectr. Freq. Control 2008, 55, 1160–1164. [CrossRef] [PubMed] © 2016 by the author; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).