The modern automobile is witnessing rapid growth in sensor systems spurred by the growing need for higher efficiency and reduced emissions. The progression of sensor applications can be seen from the early days of measuring oil and water pressure/temperature, which were followed by the addition of crank and cam position, air mass flow/temperature/pressure and exhaust gas analysis. On the chassis, anti-lock braking system (ABS) wheel speed sensors compete with direct pressure sensors to determine the status of tire inflation. Accelerometers for airbag deployment have been supplemented by gyro-based inertial module units. The question is: Which sensors will be next to support the advances of vehicle performance and safety?
The drive for reduced exhaust emissions means it is important to control engine performance over the life of the vehicle. Modern engine control is typically based on look-up tables of nominal output torque versus a range of input variables, all of which are derived from dynamometer testing of a few engines. Because engines vary with production tolerances and by wear throughout their service life, torque estimates for control of the engine and automatic transmission gear changes are generally non-optimal.
First-generation tire pressure monitoring systems (TPMS) used ABS wheel sensing (tires with low pressure rotate faster) or battery-powered tire pressure sensors and transmitters. ABS sensors have relatively poor accuracy and the algorithms need minutes to stabilize. Battery-powered sensors provide accuracy but are relatively heavy, have limited service life, and present battery disposal issues. With more than a billion tires purchased annually, discarding batteries is not ecologically sound.
Surface acoustic wave (SAW) technology can be applied as a strain transducer that is lightweight (<2 grams), small, rugged, recyclable and both battery-less and wireless in nature. SAW sensors can readily sense torque and temperature for steering EPS and powertrain applications as well as pressure and temperature for TPMS. SAW is particularly suitable for monitoring rotating components or those difficult or dangerous to access. This article reviews the design and interrogation of SAW sensor systems produced by Honeywell and discusses several key automotive applications.
Surface acoustic waves on solids were predicted and characterized in the 19th century but not used in electronic systems until the latter part of the 20th century. A SAW device has the ability to convert an electrical signal into an acoustic signal with the same frequency, but a much smaller wavelength because of a reduction in propagation velocity of about five orders of magnitude; e.g., a 100 MHz electrical signal with a wavelength of about three meters is converted to a wavelength of only 30 microns on the SAW device. This allows manipulation of an RF signal in a very small package. Low propagation velocity enabled the use of SAW devices for time delays and filtering in radar systems and television but their application burgeoned within the mobile phone market. Honeywell SAW sensors use small piezo-electric quartz die upon which two or three single-port resonators, with natural resonant frequencies around 434 MHz, are fabricated in aluminum using standard photo-lithographic techniques (Figure 1). Because of the established high-volume market for SAW filters, no new manufacturing tooling is required to produce SAW resonators, only a new mask, providing benefits of low unit cost and easy second sourcing.
A SAW resonator is excited by a short radio-frequency (RF) burst. The centrally placed interdigital transducer (IDT) converts the electrical input signal to a mechanical wave through the piezo-electric effect. The waves propagate from IDT to reflectors and back until a forced resonance exists as a standing wave. After the transmit signal is switched off, the resonator continues to oscillate but at a frequency modified by any applied mechanical and/or thermal strain. The decaying oscillation is converted back to an electrical signal via the piezo-electric effect and retransmitted to the SAW interrogation board where the frequencies are analyzed and converted to engineering parameters.
The electronics system performs a variety of tasks to convert changes in the state of the SAW sensors into meaningful pressure, torque or temperature signals for the vehicle controller. It must wirelessly stimulate the SAW elements, read back their resonant signals, determine their frequencies, and then calculate the torque, pressure or temperature value using stored calibration information.
Dependent on the sensing application, up to five individual SAW resonators are interrogated to complete a single measurement of the output parameter. The sensor is a narrowband device typically containing two or three resonators designed with nominal frequency peaks occupying the 433.05 to 434.70 MHz ISM band. Spacing between peaks must be sufficiently large to prevent any crossover of individual frequency peaks as they shift in response to the external condition under measurement. Otherwise the system could lose track of the relationship between each resonator and its corresponding fre-quency resulting in ambiguity in the calculation of the final parameter. Regardless of mechanical packaging, all resonators in a given application are connected in parallel so the interrogation system sees the equivalent of a single one-port resonator with a number of resonant peaks.
A DSP controller that is connected to an RF ASIC directs wireless interrogation of a SAW resonator. The process begins with transmission of a narrowband RF burst signal with a frequency close to the resonant peak of one of the SAW resonators. The wireless interface between interrogation electronics and passive sensor varies dependent on the application. Planar micro-strip couplers are typically used for torque applications to allow uninterrupted linkage between the stationary and rotating components. Dipole antennas have been used for TPMS applications. Transmission power levels during this phase of interrogation typically range from about 0.2 to 3 mW. Following wireless excitation of the SAW resonator, the RF ASIC switches from transmit to receive mode so that it can capture the returning RF signal. The receive path in the ASIC consists of a low-noise amplifier (LNA) followed by a single sideband (SSB) mixer that downconverts the ∼433 MHz signal to a first intermediate frequency (IF) of 11 MHz. The signal is then filtered and amplified before entering an I-Q mixer, which downconverts the signal further to its second IF of 1 MHz.
At this point the SAW signal has been split into separate low-frequency channels in quadrature as I + jQ. These I and Q signals are sent from the ASIC to the DSP where they are simultaneously sampled by an internal analog-to-digital converter (ADC). This interrogation process is repeated a number of times so that multiple I-Q response signals from a SAW element can be combined in a time-synchronized fashion known as coherent accumulation, which improves the signal-to-noise ratio (SNR) of the signal by reducing the effect of random errors and phase noise in the RF ASIC. The next processing step involves the calculation of a discrete Fourier transform (DFT) on the complex signal formed by sampled I-Q data in order to determine the exact frequency of the resonant peak. This approach is more accurate than a DFT computed from a single-channel input. A full spectrum fast Fourier transform (FFT) is not performed because the sensor is narrowband and the peak frequency in the sampled signal will lie within the second IF ± the maximum frequency shift of the resonator when subjected to its full range of measurement. A coarsely spaced set of spectral lines is first calculated to determine the rough location of the frequency peak. A finer spacing of spectral lines is then computed as the basis of a final inter-polation to determine the exact resonant frequency.
The process of interrogation and determination of resonant frequency is sequentially performed for each SAW resonator in the transducer resulting in a set of frequencies that are the basis for calculating differential shifts proportional to the sensed parameter. These frequency differences provide input to a set of model-based equations that the DSP uses to determine final measurement values of torque, temperature or pressure for transmission to the vehicle controller or for other higher-level processing.
As shown in Figure 2, the quartz die, carrying SAW resonators, is typically packaged within a stainless steel pellet circa 11 mm diameter by 2 mm thick, mass <2 gm.
For tire pressure monitoring, the die rests on two ledges and is deformed by the upper surface diaphragm pressing against it via a central pip. Interrogation and backscatter signals are broadcast from a simple whip antenna.
For torque sensing (Figure 3), the die is bonded to the base of the pellet, which is in turn attached to the component under torque. The interrogation and return signal are transmitted via non-contacting planar couplers.
In TPMS and torque sensing dies, three SAW resonators generate two frequency differences, one proportional to pressure or torque, the other to temperature. This provides temperature-compensated pressure and torque and independent temperature monitoring.
For TPMS, SAW sensors can be mounted in various ways including on the tire using a bonded rubber patch, on the valve (both rubber and metallic) or on a runflat band clamped in the wheel well. Sensitivities can be adjusted to suit trucks at 10 bar down to racecars at 2 bar. Typical accuracy is 1% of full scale.
For torque measurement (Figure 4), sensors can be attached to shafts or disks with resonators in classic ±45° orientation to sense the tension/compression components of shear strain.
Measurement accuracy in the 1% class can be obtained with full-scale strains between 50 and 500 microstrain. Applications to date include steering shafts for EPAS, engine flexplates, auto-transmission output shafts and drive shafts.
SAW-sensing technology provides new opportunities for sensing pressure, torque and temperature, especially on rotating components, within the automotive environment. In many cases, there is the opportunity to switch from indirect parameter measurement, in order to estimate a wanted variable, to real-time output sensing for both monitoring and closed loop control.
Gary O'Brien is an Engineering Fellow for the Sensing and Control Division of Honeywell International in Freeport, IL. Ray Lohr is technical director for Transense Technologies plc in Oxon, UK.