Technical Papers

Design of the all-quartz package (AQP) for a resonant SAW strain sensing element is discussed and strain transfer to the surface of the SAW substrate is studied. Analytical results are compared with the FEA results. Based on the analytical model, creep and relaxation of strain on top of the SAW substrate is simulated using a viscoelastic model for the bond and the AQP frame adhesives. An experimentally observed anomalous creep of the resonant frequencies under a constant load is explained theoretically.

Design of a torque sensor for aftermarket installation on marine propeller shafts is presented. The sensor is based on UHF SAW resonator sensing elements bonded to plate transducers that are clamped on the shaft and interrogated in a non-contact way through an easily installable RF coupler. A method of calibration of the transducers is described and the torque data logged during a one year period of testing on a ship are presented.

The paper overviews principles of operation of passive wireless SAW resonant sensing systems for temperature, strain, torque, and pressure measurements. Advantages and disadvantages of SAW resonators as wireless sensing elements are discussed in conjunction with the current EMC regulations. Design of differential resonant sensing elements and their sensitivity to strain and temperature are presented. Influence of antennas on sensor characteristics is discussed. Methods of interrogation of passive wireless resonant sensors are reviewed and design of some readers developed at Transense Technologies plc is presented. A trade-off between the sensor bandwidth, the read range and the sensor resolution is discussed. Applications of wireless passive SAW resonant sensors are presented for monitoring pressure and temperature in automotive tyres, for measuring torque in electrical power assisted steering, at the output of engines and at the input of automotive gearboxes, as well as for monitoring torque in wind turbine and marine applications.

Resonant SAW strain sensing elements used in non-contact torque, force and vibration sensors are investigated from the point of view of their stability and durability. Results of fatigue testing of the SAW sensing elements bonded to metal shafts with a stiff adhesive are presented. They demonstrate the sensor durability and stability of the strain sensitivity up to 13 million strain cycles. Stability of the frequency of the SAW differential resonant sensing elements is also investigated demonstrating an estimated sensor zero drift of 0.8-0.15 microstrain per year.

Statistical simulation is used to evaluate performance of four different frequency estimators for interrogation of resonant wireless SAW sensors. The first one is based on DFT and quadratic interpolation, the second one employs a weighted least-squares estimate of the phase difference between signal samples. The third and the fourth methods use singular value decomposition and apply a weighted linear predictor in the case of constant sine wave amplitude and an iterative least squares method in the case of decaying sine wave. Numerical receiver model includes additive and phase noises, SAW response limiting, non- linear phase distortions and parasitic SAW responses. Experimental results are also obtained for all the four frequency estimators.

Shear strain transfer from the part surface to the SAW substrate is analysed for the SAW strain sensor attached to the part by an elastic adhesive. An analytic expression for the strain transfer function is obtained. A model that takes into account viscoelasticity of the adhesive is proposed that allows prediction of the shear strain sensor output as a function of time for any temporal variation of the strain on the part surface. The model gives good match of the theoretical values of creep/relaxation and hysteresis to those observed in the SAW torque sensors experimentally.

The gearbox of a wind turbine experiences complex loading patterns due to the high variability of the wind and other transient events, resulting in high wear and tear leading to costly failure. In the past, condition monitoring (CM) has relied upon vibration, oil analysis and temperature monitoring. These techniques have been extensively used with some success; however, they can only monitor faults at a very late stage. This paper presents a detailed design and study of Surface Acoustic Wave (SAW) Torque sensors and the use of the Acoustic emission (AE) technique to monitor faults in the wind turbine drivetrain. A novel design to overcome inherent problems of SAW torque sensor use with large wind turbines is presented. Slow rotational speeds, make CM difficult due to the reduced energy loss rates, whereby the signal to noise (S/R) ratio is low. This makes defect detection using AE difficult or even impossible. However, we show here that with the use of the ‘Envelope’ technique, the use of AE is viable. AE technique was used to monitor the main bearing and gearbox bearings. Several features were extracted from the time domain AE signal and envelope analysis was applied to the AE signal to find periodical patterns produced by faults such as bearing defects or misalignment. In this work, AE and Torque techniques were evaluated in the laboratory and with real-life testing on the NEG MICON 750 wind turbine located at CRES (Greece).

A novel plate torque transducer based on UHF SAW resonant sensing elements has been developed allowing an easy aftermarket installation of passive wireless torque sensors on large diameter shafts by means of clamping. A design of a large diameter RF rotary coupler for wireless interrogation of the transducer was proposed that had a gap of 20 mm between the stator and the rotor and was insensitive to axial and radial misalignment as well as the rotation angle. The two plate transducers and the RF coupler were installed on a high-speed shaft of a wind turbine gearbox. The torque sensor accuracy achieved after calibration on a test shaft was better than 1% full range within the temperature range from +20°C to +80°C.

A new reader is presented that is based on a narrowband pulsed excitation of two SAW resonators contained in a differential wireless resonant sensor such as a torque senor. Simultaneous excitation of the resonators working at 433-437 MHz allows achieving a flat system frequency response up to Nyquist frequency. An interrogation algorithm and a structure of the reader ensure a torque sampling rate of at least 16 kHz and the system bandwidth of at least 8 kHz achievable at the processor clock of 150 MHz. Systematic errors caused by mutual interference of the two SAW responses, intermodulation products and aliasing are discussed. Random errors caused by the phase noise of the local oscillators are analyzed theoretically. Experimental results are also presented.

A new reader is presented that is based on a narrowband pulsed excitation of two SAW resonators contained in a differential wireless resonant sensor such as a torque senor. Simultaneous excitation of the resonators working at 433-437 MHz allows achieving a flat system frequency response up to Nyquist frequency. An interrogation algorithm and a structure of the reader ensure a torque sampling rate of at least 16 kHz and the system bandwidth of at least 8 kHz achievable at the processor clock of 150 MHz. Systematic errors caused by mutual interference of the two SAW responses, intermodulation products and aliasing are discussed. Random errors caused by the phase noise of the local oscillators are analyzed theoretically. Experimental results are also presented.

A new calibration method has been proposed for passive wireless torque sensors based on SAW resonators. It requires individual calibration of the sensors at the production stage only at room temperature and individual measurement of the difference frequency drift at zero torque over the operating temperature range. This approach greatly simplifies calibration and reduces its cost in the case of high volume manufacturing. Analysis of calibration statistics for the torque transducers has shown that the total measurement errors achieved in this case are very close to those obtained in the case of the individual sensor calibration in several temperature points over the entire operating temperature range.

The torque sensor capable of working in the extreme environment of race cars has been developed for the F1 kinetic energy recovery system. The sensor was based on two SAW sensing elements containing five one-port resonators that were attached to the KERS shaft. The achieved dynamic range of the sensor was 69 dB, the update period was 310 ms and it could work at the rotational speed up to 18000 rpm and temperatures up to 160°C. The mechanical design of the sensor ensured its small size and provided sufficient mechanical robustness allowing to survive very intense vibrations. Despite its small diameter, the non-contact rotary RF coupler connecting the sensor to the interrogation unit ensured strong enough SAW response at its input, very small rotational errors and small sensitivity to the axial movement. A number of changes in the interrogation algorithm, in particular, in the frequency tracking procedure, significantly improved stability of the sensor performance in dynamic regime. At the end of the development stage, the sensor successfully worked on race cars.

The paper is devoted to reduction of random errors in the wireless measurement of the resonant frequency of a SAW resonant sensor. It is achieved by using weighted averaging of multiple measurement results where the weights depend on the strength of the SAW responses picked up by the interrogation receiver. Optimum weights are found either by analytical method based on statistical information collected for a wireless tire pressure monitoring system (TPMS) or by means of genetic optimization algorithm using numerical model of the sensor. The optimized weights allow reduction of the time needed for averaging by at least a factor of two in comparison with simple averaging required to achieve the same standard deviation of the measured frequency.

Possible ways of reducing calibration complexity are considered for a non-contact passive torque and temperature sensor based on three one-port SAW resonators working at 433-437 MHz. A spread of sensor characteristics is investigated, and it is shown that dynamic calibration curves differ considerably from the static ones not allowing accelerated temperature variation during calibration. A simplified procedure is suggested using correction of generic calibration parameters based on individual sensor calibration only in one or two temperature points.

The main requirements for RF rotary couplers employed in contactless resonant SAW torque sensors have been formulated in the paper. They are: (a) flat frequency response, (b) small angular variation of the measured resonant frequency, (c) small angular variation of the insertion loss, and (d) insensitivity to manufacturing tolerances and contaminations. A theoretical analysis has been performed to establish influence of the coupler on the measured SAW natural oscillations. As a result, the frequency of the natural oscillations, their length and amplitude have been found as functions of the coupler’s output impedance and insertion loss. The results are applicable not only to RF rotary couplers but also to any combination of sensor and interrogation antennas including those positioned in the near field of each other. Computer simulations of several types of couplers have been done with the aim to calculate the angular variation of their output impedance and insertion loss. Two designs described in the paper, the small diameter coupler and the large diameter coupler, have demonstrated a good performance in terms of a small angular variation of the measured difference frequency and small contribution to random frequency measurement errors. Theoretical results have been confirmed by experiments with the two types of the couplers. Some ways to improve a tolerance of a large diameter coupler to contaminations within the rotor-stator gap have been proposed and tested.

An output engine torque measurement system employing 433 MHz SAW resonators as sensing elements is presented in the paper. Either one or two SAW sensing elements incorporating three or five one-port resonators are attached to a flexplate that couples the crankshaft to the torque converter of the car. An optimum design of the flexplate is described that minimizes the influence of bending and parasitic mechanical vibrations. Design of the large diameter RF rotary coupler as well as torque and temperature calibration characteristics of the sensor are presented. An interrogation algorithm and its influence on a system bandwidth are discussed. Experimental results of dynamic tests characterizing accuracy of the torque measurement are presented.

A wireless passive SAW sensor based on three one-port resonators and capable of simultaneous torque and temperature measurement is presented. The sensing element is made on Y+34° cut quartz substrate and the resonant frequencies are in the range from 433 to 437 MHz. Interrogation of the sensor is performed through an RF rotary coupler by short RF pulses. A microprocessor in the interrogation unit analyses the spectrum of SAW responses and calculates a temperature compensated torque value. Experimental results show that the temperature measurement error is within ±1° at any angle of the shaft and the torque measurement error does not exceed ±1% of full scale within the range from –40° to +125°C.

A complete pre-production in-car tire pressure monitoring system, employing 433 MHz SAW resonators as sensing elements is presented in this paper. The SAW sensing element incorporating three one-port resonators fabricated on a single substrate and connected to an antenna is described, along with calibration models for the sensor. An issue of sensor antenna tuning and sensor matching is investigated. A method of SAW sensor wireless interrogation is presented and the electronic interrogation unit is described. Experimental data characterizing accuracy of the remote resonant frequency measurement is presented.

Resolution of passive wireless resonant SAW sensing system is studied theoretically by means of stochastic simulations. It is shown that phase noise of a local oscillator of an interrogator’s receiver plays the main role in limiting the resolution at short distances between the interrogator and the sensor antenna. Additive noise of the receiver typically limits the resolution only at a distance larger than 1 to 3 m in free space at 433 MHz. Relationship between the standard deviation of the measured resonant frequency and the phase noise of the receiver is established for different number of coherently accumulated SAW sensor responses.

A contactless torque sensor based on four one-port SAW resonators on quartz substrates working in the 430-437 MHz frequency range is presented in the paper. Design of the sensor allows full compensation of the bending effect. A single stationary interrogation unit is used to measure resonant frequencies of all four resonators attached to the rotating shaft. The interrogation method is based on a pulsed excitation of the natural oscillations and their subsequent spectral analysis. The achieved standard deviation of the measured frequency difference is sufficient for EPAS applications in automotive industry.

A one-port SAW resonator with a non-uniformly strained substrate is studied both experimentally and theoretically. Apart from the expected variation of the resonant frequency due to static strain and a noticeable variation of the resonant value of S11 the emergence of a parasitic mode is also observed. Modelling explains this phenomenon and helps to predict a reduction of the sensitivity to strain due to its non-uniformity.

A detailed model of an interrogation system performing a continuous tracking of the resonant frequencies of the two SAW resonators is developed and the results of computer simulations are presented. A relationship is established between the amount of noise in the system, the Q-factor of the resonators, parameters of the frequency tracking system and the system bandwidth and the resolution. A variation of the SAW torque transducer sensitivity with temperature is discussed. A theoretical model describing this variation is presented and SAW sensors with a reduced sensitivity variation are suggested and experimentally tested.

Analytical model is built to study the influence of antenna impedance on the amplitude, length and frequency of a SAW sensor response received by an interrogation unit. Recommendations on the choice of optimum sensor antenna impedance are worked out in the case of the SAW resonators connected either in series or in parallel to the antenna. The accuracy of the resonant frequency measurement is analysed by means of statistical simulations.