A wireless demodulation system for passive surface acoustic wave torque sensor Xiaojun Ji, Yanping Fan, Hongli Qi, Jing Chen, Tao Han, and Ping Cai Citation: Review of Scientific Instruments 85, 125001 (2014); doi: 10.1063/1.4902180 View online: http://dx.doi.org/10.1063/1.4902180 View Table of Contents: http://scitation.aip.org/content/aip/journal/rsi/85/12?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Local oscillator phase noise limitation on the resolution of acoustic delay line wireless passive sensor measurement Rev. Sci. Instrum. 85, 065001 (2014); 10.1063/1.4880455 Remote vibration measurement: A wireless passive surface acoustic wave resonator fast probing strategy Rev. Sci. Instrum. 83, 055001 (2012); 10.1063/1.4705728 Surface acoustic wave devices as passive buried sensors J. Appl. Phys. 109, 034905 (2011); 10.1063/1.3504650 High-frequency, high-sensitivity acoustic sensor implemented on ALN/Si substrate Appl. Phys. Lett. 83, 1641 (2003); 10.1063/1.1604482 Surface transverse waves in polymer-coated grating configurations J. Appl. Phys. 91, 5700 (2002); 10.1063/1.1465502
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP: 202.28.191.34 On: Sat, 28 Feb 2015 07:28:46
REVIEW OF SCIENTIFIC INSTRUMENTS 85, 125001 (2014)
A wireless demodulation system for passive surface acoustic wave torque sensor Xiaojun Ji,1,a) Yanping Fan,1,b) Hongli Qi,2 Jing Chen,1 Tao Han,1 and Ping Cai1 1 School of Electronic Information and Electrical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China 2 No. 704 Research Institute, China Shipbuilding Industry Corporation, Shanghai 200031, China
(Received 20 July 2014; accepted 7 November 2014; published online 1 December 2014) Surface acoustic wave (SAW) resonators are utilized as torque sensors for their passive and wireless features. However, the response of a SAW torque sensor is difficult to detect because of the transient response duration and interruption of channel noise, which limit the application of SAW torque sensors. The sensitive mechanism and response function of a passive wireless SAW torque sensor are analyzed in this study. A novel demodulation system involving both hardware and software is developed for the SAW torque sensor. A clipping amplifier is utilized in the hardware to widen the dynamic response range and increase the length of the valid signal. Correlation extension and centroid algorithms are designed to lengthen the received signal and improve the estimation accuracy of the center frequency of the response signal, respectively. Meanwhile, a fast binary search algorithm is proposed to accelerate the scanning cycle according to the developed response function. Finally, the SAW torque sensor demodulation system is set up and SAW resonators with high sensitivity are fabricated on a quartz substrate. The presented demodulation system is tested, and a standard deviation of 0.28 kHz is achieved. This value is much smaller than that of classic and modern spectrum estimation methods. The sensitivity of resonance frequency shift versus torque on the shaft of the assembled senor is 2.03 kHz/Nm; the coefficient of determination is 0.999, and the linearity is 0.87%. Experimental results verify the validity and feasibility of the proposed SAW torque sensor demodulation system. © 2014 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4902180] I. INTRODUCTION
Surface acoustic wave (SAW) resonators consist of an interdigital transducer (IDT) and shorted electrode gratings. These resonators are manufactured on the surface of a piezoelectric crystal substrate. The shorted electrode gratings located on both sides of IDT form a resonant cavity. The surface acoustic waves are excited by the IDT placed in the cavity and are reflected by the electrode gratings to form a standing wave pattern.1 SAW devices can operate in passive and wireless modes without additional electronic circuitry for signal transmission,2 and thus offer exciting insights into wireless sensing and control of moving parts in harsh environmental or high-temperature conditions. The variations in environmental condition can change the material constants of the SAW substrate and result in a shift in the resonance frequency of the SAW device. The tested physical parameter can be determined by measuring this frequency shift. Thus, SAW devices are widely utilized in sensing applications for temperature,3 pressure,3 and vibration,4 because of their high reliability. For the passive wireless SAW sensor, no energy amplification mechanism exists at the sensing site. The energy of the retransmitted response from a passive SAW sensor is very weak. The response of a SAW sensor can be received only after the request is switched off. The response duration is transient and only lasts for a very short time. In addition, the noise in the wireless channel and receiving unit increases the diffia) Electronic mail: [email protected] b) Electronic mail: [email protected]
0034-6748/2014/85(12)/125001/8/$30.00
culty of frequency estimation. With the increasing demand for reliable SAW sensors, estimating the frequency of the SAW response rapidly and accurately has become increasingly important. Currently, many researchers focus on the estimation of the response of wireless SAW sensors. Zhang et al.5 utilized a genetic algorithm to estimate the frequency of a passive SAW torque sensing signal. However, a genetic algorithm is time consuming and cannot effectively solve problems in which the only fitness measure is a single correct or incorrect measure. Ostermayer6 presented correlative signal processing and wavelet transform methods to obtain the information of interest in the output response of SAW sensors. However, the algorithms can only estimate the response time. Shrena7 proposed a signal processing algorithm based on fast Fourier transform (FFT) to accurately estimate the properties of SAW as a function of temperature. However, achieving the required resolution through FFT is impossible because the amounts of sampled data and channel noise have a significant impact on the result of FFT. In the mathematical field of numerical analysis, interpolation is a method of constructing new data points within the range of a discrete set of known data points; it has been applied to demodulate SAW parameters.8, 9 However, the interpolation algorithm enhances the frequency resolution in the calculation rather than lengthen the duration time of the effective signal. Therefore, none of the above mentioned algorithms meet the requirements of high accuracy and high speed. In addition, all the aforementioned methods involve only software to estimate the frequency.
85, 125001-1
© 2014 AIP Publishing LLC
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP: 202.28.191.34 On: Sat, 28 Feb 2015 07:28:46
125001-2
Ji et al.
Rev. Sci. Instrum. 85, 125001 (2014)
We propose a novel demodulation system that involves both hardware and software for the passive wireless SAW torque sensor. A clipping amplifier is utilized in the hardware to widen the dynamic response range and increase the length of the valid signal. The clipping amplifier also helps alleviate the problem that the magnitude variation of the received signal increases the dynamic range loading in receiver circuits.10 A correlation extension algorithm is developed to help to suppress noise and improve the resolution of frequency estimation. A centroid algorithm is designed in the software to estimate the SAW response with high accuracy. A fast scanning method is proposed to accurate the scanning cycle. A demodulation system is then built and tested. The experimental results show that the presented demodulation system is correct in theory and acceptable.
II. ANALYSIS OF THE MECHANISM AND RESPONSE OF A SAW TORQUE SENSOR
Owing to the high quality of the SAW resonator, more energy can be stored in the same excitation duration, and a longer process of energy discharge can be achieved after the excitation signal disappears. In addition, the narrow operational frequency band makes it possible to achieve multiple accesses to a passive resonator sensor array through frequency divisions. Hence, the SAW resonator is often utilized as a sensing unit. Figure 1 shows the structure of the SAW torque sensor. The sensor consists of the shaft, a SAW resonator, and an antenna. The antenna and SAW resonator are connected by Al/Si lines. To hold the SAW torque sensor and preserve the symmetry of strain distribution, four rectangular planes are grinded to form a square cross section with a length of b. When torque is applied to the shaft, normal strain oriented at a direction of ±45◦ to the shaft axis is at its maximum E±45◦ .11 According to the mechanics of elasticity,9 the relation between maximum shear strain and applied torque is deduced as E45◦ = −E−45◦ =
M , 0.2305Gb3
(1)
where G is the modulus of elasticity in shear, b is a square side length of the cross section, and M is the applied torque. According to first-order perturbation theory proposed by Tiersten,12 the SAW resonance frequency shift induced by ap-
Antenna Al/Si line
Reflector IDT Test shaft FIG. 1. Schematic of SAW torque sensor.
plied torque can be described as ∗ ui,j Cˆ ij kl uk,l dV (f0 − f0 ) = . f0 2ρω02 u∗i ui dV
(2)
The Cartesian tensor notion is employed in the expression above. In Eq. (2), f0 is the resonance frequency, f0 is the resonance frequency under a torque bias, ui is the mechanical displacement vector, ρ is the density of the SAW substrate, * ˆ ijkl is the perturbation bias denotes a complex conjugate, and C that depends on the applied torque, which is defined as12 Cˆ ij kl = Tik δj l + Cij klmn Emn + Cij nl Wl,n + Cinkl Wj,n , (3) where i, j, k, l = 1, 2, 3, δ jl is the Kronecker delta, Tij is the stress tensor, Cijklmn is the third-order elastic constants, Cijnl is the second-order elastic constants, and Emn and Wl,n are the biasing strain and displacement gradient component, respectively. Ejn = 1/2 Wl,n .12 Equation (3) can then be rewritten as Cˆ ij kl = Gij kl M Gij kl = δj l Cikmn emn + Cij klmn emn + Cij nl eln + Cinkl ej n
. (4)
In Eq. (4) eij is defined as
eij =
⎛
1
0
⎜ 1 ⎜ 0 −1 3 0.2305Gb ⎝ 0 0
0
⎞
⎟ 0⎟ ⎠.
(5)
0
Hence, the resonance frequency shifts versus torque can be obtained as ∗ ui,j Gij kl uk,l dV (f0 − f0 ) = . (6) M 8π 2 f0 ρ u∗i ui dV According to Eq. (6), to measure the torque accurately, one must ensure that the resonance frequency shifts are measured accurately. The impulse response of the IDTs is13
xm , (7) h(t) = C1 Wm δ t − Vs m where C1 is a constant, Wm is the weight factor, and xm and Vs are the mth position of the acoustic source and the velocity of SAW, respectively. In the charge interval, a short sinusoidal pulse train with angular frequency ω [shown in Eq. (8)] is received by the antenna connected to the SAW resonator and converted into SAW propagating on the substrate. In the following analysis, we suppose that no loss exists during the SAW propagation, 0, t
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP: 202.28.191.34 On: Sat, 28 Feb 2015 07:28:46
REVIEW OF SCIENTIFIC INSTRUMENTS 85, 125001 (2014)
A wireless demodulation system for passive surface acoustic wave torque sensor Xiaojun Ji,1,a) Yanping Fan,1,b) Hongli Qi,2 Jing Chen,1 Tao Han,1 and Ping Cai1 1 School of Electronic Information and Electrical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China 2 No. 704 Research Institute, China Shipbuilding Industry Corporation, Shanghai 200031, China
(Received 20 July 2014; accepted 7 November 2014; published online 1 December 2014) Surface acoustic wave (SAW) resonators are utilized as torque sensors for their passive and wireless features. However, the response of a SAW torque sensor is difficult to detect because of the transient response duration and interruption of channel noise, which limit the application of SAW torque sensors. The sensitive mechanism and response function of a passive wireless SAW torque sensor are analyzed in this study. A novel demodulation system involving both hardware and software is developed for the SAW torque sensor. A clipping amplifier is utilized in the hardware to widen the dynamic response range and increase the length of the valid signal. Correlation extension and centroid algorithms are designed to lengthen the received signal and improve the estimation accuracy of the center frequency of the response signal, respectively. Meanwhile, a fast binary search algorithm is proposed to accelerate the scanning cycle according to the developed response function. Finally, the SAW torque sensor demodulation system is set up and SAW resonators with high sensitivity are fabricated on a quartz substrate. The presented demodulation system is tested, and a standard deviation of 0.28 kHz is achieved. This value is much smaller than that of classic and modern spectrum estimation methods. The sensitivity of resonance frequency shift versus torque on the shaft of the assembled senor is 2.03 kHz/Nm; the coefficient of determination is 0.999, and the linearity is 0.87%. Experimental results verify the validity and feasibility of the proposed SAW torque sensor demodulation system. © 2014 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4902180] I. INTRODUCTION
Surface acoustic wave (SAW) resonators consist of an interdigital transducer (IDT) and shorted electrode gratings. These resonators are manufactured on the surface of a piezoelectric crystal substrate. The shorted electrode gratings located on both sides of IDT form a resonant cavity. The surface acoustic waves are excited by the IDT placed in the cavity and are reflected by the electrode gratings to form a standing wave pattern.1 SAW devices can operate in passive and wireless modes without additional electronic circuitry for signal transmission,2 and thus offer exciting insights into wireless sensing and control of moving parts in harsh environmental or high-temperature conditions. The variations in environmental condition can change the material constants of the SAW substrate and result in a shift in the resonance frequency of the SAW device. The tested physical parameter can be determined by measuring this frequency shift. Thus, SAW devices are widely utilized in sensing applications for temperature,3 pressure,3 and vibration,4 because of their high reliability. For the passive wireless SAW sensor, no energy amplification mechanism exists at the sensing site. The energy of the retransmitted response from a passive SAW sensor is very weak. The response of a SAW sensor can be received only after the request is switched off. The response duration is transient and only lasts for a very short time. In addition, the noise in the wireless channel and receiving unit increases the diffia) Electronic mail: [email protected] b) Electronic mail: [email protected]
0034-6748/2014/85(12)/125001/8/$30.00
culty of frequency estimation. With the increasing demand for reliable SAW sensors, estimating the frequency of the SAW response rapidly and accurately has become increasingly important. Currently, many researchers focus on the estimation of the response of wireless SAW sensors. Zhang et al.5 utilized a genetic algorithm to estimate the frequency of a passive SAW torque sensing signal. However, a genetic algorithm is time consuming and cannot effectively solve problems in which the only fitness measure is a single correct or incorrect measure. Ostermayer6 presented correlative signal processing and wavelet transform methods to obtain the information of interest in the output response of SAW sensors. However, the algorithms can only estimate the response time. Shrena7 proposed a signal processing algorithm based on fast Fourier transform (FFT) to accurately estimate the properties of SAW as a function of temperature. However, achieving the required resolution through FFT is impossible because the amounts of sampled data and channel noise have a significant impact on the result of FFT. In the mathematical field of numerical analysis, interpolation is a method of constructing new data points within the range of a discrete set of known data points; it has been applied to demodulate SAW parameters.8, 9 However, the interpolation algorithm enhances the frequency resolution in the calculation rather than lengthen the duration time of the effective signal. Therefore, none of the above mentioned algorithms meet the requirements of high accuracy and high speed. In addition, all the aforementioned methods involve only software to estimate the frequency.
85, 125001-1
© 2014 AIP Publishing LLC
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitationnew.aip.org/termsconditions. Downloaded to IP: 202.28.191.34 On: Sat, 28 Feb 2015 07:28:46
125001-2
Ji et al.
Rev. Sci. Instrum. 85, 125001 (2014)
We propose a novel demodulation system that involves both hardware and software for the passive wireless SAW torque sensor. A clipping amplifier is utilized in the hardware to widen the dynamic response range and increase the length of the valid signal. The clipping amplifier also helps alleviate the problem that the magnitude variation of the received signal increases the dynamic range loading in receiver circuits.10 A correlation extension algorithm is developed to help to suppress noise and improve the resolution of frequency estimation. A centroid algorithm is designed in the software to estimate the SAW response with high accuracy. A fast scanning method is proposed to accurate the scanning cycle. A demodulation system is then built and tested. The experimental results show that the presented demodulation system is correct in theory and acceptable.
II. ANALYSIS OF THE MECHANISM AND RESPONSE OF A SAW TORQUE SENSOR
Owing to the high quality of the SAW resonator, more energy can be stored in the same excitation duration, and a longer process of energy discharge can be achieved after the excitation signal disappears. In addition, the narrow operational frequency band makes it possible to achieve multiple accesses to a passive resonator sensor array through frequency divisions. Hence, the SAW resonator is often utilized as a sensing unit. Figure 1 shows the structure of the SAW torque sensor. The sensor consists of the shaft, a SAW resonator, and an antenna. The antenna and SAW resonator are connected by Al/Si lines. To hold the SAW torque sensor and preserve the symmetry of strain distribution, four rectangular planes are grinded to form a square cross section with a length of b. When torque is applied to the shaft, normal strain oriented at a direction of ±45◦ to the shaft axis is at its maximum E±45◦ .11 According to the mechanics of elasticity,9 the relation between maximum shear strain and applied torque is deduced as E45◦ = −E−45◦ =
M , 0.2305Gb3
(1)
where G is the modulus of elasticity in shear, b is a square side length of the cross section, and M is the applied torque. According to first-order perturbation theory proposed by Tiersten,12 the SAW resonance frequency shift induced by ap-
Antenna Al/Si line
Reflector IDT Test shaft FIG. 1. Schematic of SAW torque sensor.
plied torque can be described as ∗ ui,j Cˆ ij kl uk,l dV (f0 − f0 ) = . f0 2ρω02 u∗i ui dV
(2)
The Cartesian tensor notion is employed in the expression above. In Eq. (2), f0 is the resonance frequency, f0 is the resonance frequency under a torque bias, ui is the mechanical displacement vector, ρ is the density of the SAW substrate, * ˆ ijkl is the perturbation bias denotes a complex conjugate, and C that depends on the applied torque, which is defined as12 Cˆ ij kl = Tik δj l + Cij klmn Emn + Cij nl Wl,n + Cinkl Wj,n , (3) where i, j, k, l = 1, 2, 3, δ jl is the Kronecker delta, Tij is the stress tensor, Cijklmn is the third-order elastic constants, Cijnl is the second-order elastic constants, and Emn and Wl,n are the biasing strain and displacement gradient component, respectively. Ejn = 1/2 Wl,n .12 Equation (3) can then be rewritten as Cˆ ij kl = Gij kl M Gij kl = δj l Cikmn emn + Cij klmn emn + Cij nl eln + Cinkl ej n
. (4)
In Eq. (4) eij is defined as
eij =
⎛
1
0
⎜ 1 ⎜ 0 −1 3 0.2305Gb ⎝ 0 0
0
⎞
⎟ 0⎟ ⎠.
(5)
0
Hence, the resonance frequency shifts versus torque can be obtained as ∗ ui,j Gij kl uk,l dV (f0 − f0 ) = . (6) M 8π 2 f0 ρ u∗i ui dV According to Eq. (6), to measure the torque accurately, one must ensure that the resonance frequency shifts are measured accurately. The impulse response of the IDTs is13
xm , (7) h(t) = C1 Wm δ t − Vs m where C1 is a constant, Wm is the weight factor, and xm and Vs are the mth position of the acoustic source and the velocity of SAW, respectively. In the charge interval, a short sinusoidal pulse train with angular frequency ω [shown in Eq. (8)] is received by the antenna connected to the SAW resonator and converted into SAW propagating on the substrate. In the following analysis, we suppose that no loss exists during the SAW propagation, 0, t
