Understanding Surface Acoustic Wave
(SAW) Devices for Mobile and Wireless
Applications and Design Techniques
by Colin K. Campbell, Ph.D., D.Sc.
Session 19: "An Overview of SAW
Devices For Mobile/Wireless Communications"
( 68 Questions and Answers for Year 2008 )
(Including Real-Time SAW Fourier Transformers)
(You may wish to print a copy
of this web page for future reference)
Question 1. What is the current world-wide production level of surface acoustic wave (SAW) devices?
Answer 1: Major SAW manufacturers/suppliers include Japan, USA, Germany, mainland China, and Taiwan. While I have not been able to obtain up-to-date world-wide levels, my ownunofficialestimate is that these have to be several million SAW devices a year. For example, one company alone in one of these countries is reportedly producing 3 million devices per day !
UNUSUAL PROPERTIES OF SAW DEVICES
Question 2: Before we go any further, tell me if surface acoustic wave (SAW) filters are analogor digital devices?
Answer 2: Tricky question! My own view is that some configurations (as in the basic bidirectional interdigital transducer (IDT) structure of Figure 1), can be considered to operate as passive HYBRIDanalog/digital devices! The basic SAW filter sketched in Figure 1 is indeed a passive analog device. It is just a thin metal film structure deposited on top of a piezoelectric crystal substrate, with no power supplies to worry about. However, this is not the complete answer! Now for the digital part. Look at the constituent input/output IDTs. The layout pattern of these input/output thin metal film patterns is designed to provide the desired bandpass filtering function H(f) = Voutput/Vinput as the SAW propagates along the piezoelectric crystal surface. But these bidirectional IDTs may be considered to act asspatially-sampledversions of the corresponding time-evolving Inverse Discrete Fourier Transform (IDFT) h(t). (Remember that there is a unique correspondence between the frequency response H(f) of a filter, and its impulse response h(t). (Simple concepts for digital signal-processing engineers. Not so simple for old analog circuit designers like me !). Because of this, many digital signal processing techniques can be employed in the design of the IDT patterns. Additionally, SAW filters find applications in many digital communications systems.
3: Give me three examples of the digital
signal-handling equivalence of a SAW filter.
Answer 3: (a) Digital signal-processing window function techniques can be applied to shape the IDT patterns, and thereby shape the filter bandpass frequency response. Examples of these include Hamming, Cosine weighting, Kaiser, Kaiser-Bessel, Taylor-weighting, and Dolph-Chebyshev. (See Chapter 3 of my 1998 SAW book).
(b) The well-known (??) Remez Exchange algorithm - originally applied to the design of optimum Finite Impulse Response (FIR) linear-phase digital filters - can also be applied to the design of SAW bandpass filters.(See Chapter 8 of my 1989 SAW book Surface Acoustic Wave Devices and Their Signal Processing Applications ( Academic Press:Boston,1998 ), which also includes a FORTRAN Remez program for SAW applications. Also see: J. H. McClellan, T. W. Parks and L. R. Rabiner, "A computer program for designing optimum FIR linear phase digital filters," IEEE Transactions on Audio and Electroacoustics, vol. AU-21, pp. 506-526, December 1973.)
(c) As a third hybrid-performance example, SAW Nyquist filters are employed in Quadrature-Amplitude-Modulation (QAM) digital radio modems.(See Chapter 19 of my 1998 SAW book).
4: Can SAW bandpass filters operate at harmonic frequencies?
Answer 4: Yes. They can operate at selected harmonic frequencies, depending on the metalization ratio h = a/b in Figure 2. Rayleigh-wave delay-line filters employing split-electrode IDTs on YZ-lithium niobate have been reported as operating efficiently up to the 11thharmonic. (See: W. R. Smith, "Basics of the SAW interdigital transducer," in J. H. Collins and L. Masotti (eds.)Computer-Aided Design of Surface Acoustic Wave Devices. Elsevier: New York, 1976. Also see: W. R. Smith and W. F. Pedler, "Fundamental- and harmonic-frequency circuit model analysis of interdigital transducers with arbitrary metalization ratios and polarity sequences," IEEE Transactions on Microwave Theory and Techniques, vol. MTT-23, pp. 853-864, November 1975). The IDTs in Figure 2(a) and Figure 2(b) can operate at selected odd-harmonic frequencies, while the IDT structure in Figure 2(c) can operate at selected even and odd harmonics, depending on the metalization ratio.
5: But why would I want to operate a SAW filter
in a harmonic mode?
Answer 5::a) Say I am using SAW filters fabricated on single-crystal piezoelectric substrates. One good reason why I might want to use a SAW filter operating in a harmonic mode relates to possible interference from acoustic bulk waves, which may be generated to various levels by an excited interdigital transducer (IDT), in addition to the desired SAW. Bulk waves can propagate in any direction within the propagating single-crystal piezoelectric substrate on which the IDTs are fabricated. These can have three components: namely those for 1) longitudinal bulk waves, 2) fast transverse shear waves, and 3) the slow transverse shear waves. (See Chapter 2 of my 1998 SAW book). Those components that arrive at the output IDT will generate interfering voltages there, in addition to the desirable SAW. These can cause undesirable passband as well as out-of-band degradation. If, however, I operate in a high-enough harmonic mode, it may be possible to "bypass" such bulk wave interference. (See References 37 and 38 in Chapter 6 of my 1998 SAW book).
b) Also, one good reason why I might need to operate at SAW filter (i.e., Rayleigh-wave or leaky-SAW (LSAW) type) in a harmonic-frequency mode relates to the operational frequency for my SAW filter. Remember that the SAW acoustic wavelength lois given by lo =v/fo, where v = SAW velocity and fo = fundamental operating frequency. This makes for very small SAW devices at frequencies above about 1.5 GHz. As an example, a packaged 1.880-GHz SAW Tx-filter for USA Personal Communications Services (PCS), (see Figure 1.4 in my SAW book), may only have an area in the order of 3 mm x 3 mm. (If you do not think this is a small filter, get out a millimeter scale and think about this!)
Again consider that I want to use a high-frequency SAW filter design on a piezoelectric crystal substrate. If the operating frequency is to be above about 1.5 GHz, then I must be concerned as to whether or not I can have the desired photolithographic resolution in the fabrication of my IDT patterns. Recall that the acoustic wavelength lo at filter center frequency fo is given by lo= v/fo , where v = SAW/LSAW velocity. Remember from our previous web-page discussions that an electrode finger width in a SAW IDT is typically lo/4. So, in order to maximize my photolithography, I would want to use a SAW substrate with the largest acoustic velocity v. For frequencies above about 1 GHz this would suggest the use of a LSAW substrate cut, with acoustic velocity in the order of 4000 meter/sec. If the filter fundamental frequency is to be fo = 2 GHz, this would give lo = 2.0 micron (1 micron = 10-4 cm). For lo/4 IDT fingers this would result in required finger widths of only 0.5 micron (1 micron = 10-4 cm). If I want to make my own 2-GHz SAW filter with this fundamental frequency, I would require use of a high-resolution photolithographic camera. As well, I could encounter additional deterioration of the IDT finger edges in the follow-up microelectronic lithographic etching processes. If, as a result of these degradations, the fundamental frequency bandpass response was not achievable, or acceptable, I could try to use a suitable Mth harmonic-frequency design , while employing IDT finger dimensions as if for frequency fo/M. I have often fabricated SAW intermediate frequency (IF) filter designs for operation at the 5th harmonic, because of lithographic resolution limitations.
DEVICE GENERAL CLASSIFICATIONS
Question 6: SAW devices my be classified into four (4) general groups, relating to their mobile/wireless signal processing applications. (a) List these four groups. (b) Give a few representative signal processing applications for each group.
Answer 6: (a) Group 1: Linear Resonator and Resonator-Filter Devices. Group 2: Linear Devices Using Unidirectional IDTs. Group 3: Linear Devices Using Bidirectional IDTs. Group 4: Nonlinear Devices.
(b) Group 1 : Antenna duplexers (2 to 4 W) for mobile/wireless transceivers, RF filters for front-end interstage coupling, Resonator-filters for one-way and two-way pagers, Resonators and resonator-filters for medical alert transmitters, Resonators and resonator-filters for automobile keyless locks, Resonators for garage door openers, Fixed frequency and tunable oscillator circuits.
Group 2 : Low-loss Intermediate Frequency (IF) filters for mobile and wireless circuits, Low-loss RF front-end filters for mobile/wireless circuitry, Multimode frequency-agile oscillators for spread-spectrum secure communications, Low-loss delay lines for low-power time-diversity wireless receivers.
Group 3 : Nyquist filters for microwave digital radio, Voltage-controlled oscillators (VCOs) for first or second-stage mixing in mobile transceivers, Fixed, or variable, delay lines for path-length equalizers, Pseudo-Noise (PN)-coded delay lines for combined Code-Division-Multiple-Access/ Time-Division-Multiple-Access (CDMA/ TDMA), Clock-recovery filters for fiber-optics communication repeater stages, Intermediate frequency (IF) filters for mobile/wireless receivers and pagers. (See page 223 of my 1998 SAW book for variable SAW delay lines).
Group 4: Synchronous and asynchronous convolvers for indoor/outdoor spread-spectrum communications.
Question 7: (a) By way of illustrating an analog-cellular type mobile communications system, sketch the basic circuit for a dual-heterodyne 800-MHz band Advanced Mobile Phone Service (AMPS) transceiver and illustrate where SAW devices can be employed in it. (b) Briefly describe the functions and merits of these components.
Answer 7: (a) Figure 3 shows the basics of such an AMPS transceiver, employing six (6) possible SAW components. This operates as a narrow-band frequency-modulation (FM) system, employing Frequency Division Multiple Access (FDMA).(See Chapter 10 and Table 10.1 in my 1998 SAW book). Mobile Tx and Rx bandwidths are 824-859 and 869-894 MHz, respectively, with 832 channels and a channel spacing of 30 kHz.
(b) The antenna duplexer filters can typically be leaky-SAW (LSAW) low-loss ladder-type filters. LSAW devices are normally preferred here over Rayleigh wave structures, as they have greater sub-surface penetration than Rayleigh waves, which allows for higher power handling capabilities (1-2 W) before the onset of device degradation. As well, the receiver preselect filter Rx#1 requires 1) low insertion loss (Less than about 3 dB), 2) a highly-selective bandwidth to prevent overloading of the follow-up Low Noise Amplifier (LNA), and 3) a dynamic range capability of about 120 dB. The follow-up RF filter RX#2, which can be a LSAW resonator-filter type, is required to suppress (i) harmonics, (ii) image-frequency noise, and (iii) noise generated by Class C (remember this ?) amplifier noise. The antenna-duplexer transmit filter Tx#1 must handle power levels of up to 30 dBm. The preceding RF filter Tx#2 , which can be a LSAW resonator-filter type, is required to suppress close-in noise. The SAW component in the Voltage-Controlled Oscillator (VCO) in the first mixer stage can typically incorporate a dual-mode SAW resonator-filter, or a wideband SAW delay line. Since the channel spacing is only 30 kHz here, the IF SAW filter must be very selective and also temperature stable. Typically this could be a two-pole waveguide-coupled resonator- filter on a stable-temperature cut (e.g. ST-X) of quartz piezoelectric-crystal substrate.
Question 8: So far so good! Now sketch an illustrative transceiver for a digital-cellular communications transceiver, such as for the Global System for Mobile Communications (GSM). Again indicate the possible location of constituent SAW components.
Answer 8: Figure 4 outlines a basic European GSM digital cellular transceiver, using In-phase/Quadrature-phase (I-Q) modulation/demodulation, and, showing up to seven (7) possible SAW components. As given in Table 10.3 of my 1998 SAW textbook, this system has a Tx band from 890-915 MHz, and an Rx band from 925-960 MHz. In contrast to the analog transceiver of Figure 3, this digital system only has 124 channels, with 8 users per channel, but with a carrier channel spacing of 1250 kHz. The access scheme here is TDMA/FDM with Gaussian Minimum Shift Keying (GMSK) modulation. The SAW RF components are similar to those discussed in Figure 3. The IF filter here is spectrally shaped, however, to cater for the power spectral distribution of MSK signals. (See page 418 and Figure 15.2 of my 1998 SAW textbook).
NYQUIST FILTERS FOR MICROWAVE DIGITAL RADIO
Question 9: (a) What microwave common carrier bands are used in North America for long-haul and data communications traffic? (b) What is the purpose of a Nyquist filter in a digital microwave radio system? (c) Sketch a block diagram outline for the circuitry of a basic digital microwave transmitter employing Quadrature Amplitude Modulation (QAM), showing the location of the SAW signal-processing Nyquist IF filter. (d) Is the Nyquist filtering only carried out in the transmitter section?
Answer 9: (a) North American microwave common carrier bands are 4, 6, 8, and 11 GHz.
(b) To attain freedom from Inter Symbol Interference (ISI).(See page 588 of my 1998 SAW textbook)
(c) Figure 5 outlines the basic form of a typical microwave digital radio transmitter employing Quadrature Amplitude Modulation (QAM). Note that the SAW Nyquist filter also incorporates an X/(sinX) filter to compensate for spectral distortion when Non-Return-To-Zero (NRZ) binary signaling is employed. (See page 581 of my 1998 SAW textbook). (d) Not necessarily. If matched filtering is required, the total required Nyquist filter response is split evenly between IF stages in both the transmitter and receiver. (See page 590 of my 1998 SAW textbook).
Question 10: In your response in Answer 7 you used the term "dBm". (a) What does this mean? (b) Give some illustrative dBm numbers related to SAW front-end components and oscillators for mobile/wireless systems.
Answer 10: (a) The term "dBm" is a base-10 logarithmic parameter and means "decibels referred to 1 milliwatt (mW)". Thus 1 mW = 0 dBm.
(b1) Consider an RF signal at the input to a wireless receiver with a voltage level of 0.8 microvolt (mV) across a 50 ohm input impedance. The input power is (0.8 x 10-6)2/50 = 1.28 x 10-14 watts. The corresponding dBm value is dBm = 10 x log(1.28 x 10-14/10-3) = ~ -109 dBm. I have typically used this signal level output from a frequency synthesizer when testing the required Signal-Noise-Distortion (SINAD) performance specifications for a mobile radio receiver. (For SINAD information see page 267 of my 1998 SAW textbook).
(b2) "Off the shelf" Rayleigh-wave oscillators are typically limited to an upper power level in the order of 15 dBm, while leaky-SAW oscillators perform up to about 30 dBm.(See page 542 of my 1998 SAW textbook).
(b3) Some wireless pagers are required to operate with input signal levels less than -100 dBm. Figure 6 outlines one front-end circuit for achieving this. It employs a low-loss leaky-SAW antenna duplexer, followed by a dual-mode leaky-SAW resonator-filter. Down conversion to the IF stage is achieved using a differential active mixer, a differential local oscillator, feeding a differential IF stage. The merits of the conversion circuit in Figure 6 can include 1) low front-end insertion loss, 2) good out-of-band rejection, 3) signal swings are doubled compared with single-ended circuits, 4) improved common-mode rejection, 5) small package size, 6) no balance-to-unbalance transformer (Balun) required, 7) input/output impedance matching capability, 8) reduced power consumption, and 9) frequency capability up to 2 GHz.
( See Reference 112 in my web publication listing. Also see G. Endoh, M. Ueda, O. Kawachi, and Y. Fujiwara, "High performance balanced type SAW filters in the range of 900 MHz and 1.9 GHz," Proceedings of 1997 IEEE Ultrasonics Symposium, vol. 1, pp. 41-44.)
ACOUSTIC WAVE OSCILLATOR CONFIGURATIONS
Question 11: Illustrate the types of SAW oscillators and configurations that can be employed in mobile/wireless communications.
Answer 11: The artistic representation of Figure 7 hopefully serves to illustrate the wide variety of available oscillator configurations. These include fixed-frequency oscillators, tunable oscillators, frequency-hopping oscillators and injection- locked oscillators. Moreover, oscillator types include those employing Rayleigh-wave propagation, leaky-SAW propagation and Surface-Skimming Bulk Wave (SSBW) propagation. (For details of these see Chapter 18 of my 1998 SAW book).
12: Why can STW or LSAW oscillators have a higher
power output capability than Rayleigh-wave ones?
Answer 12: Because Rayleigh-wave sub-surface penetration is only about 1 acoustic wavelength, excessive power densities can degrade the IDT metalization to the point of destructive failure. Typically, STW oscillators can operate up to about 2 W ( + 33 dBm), before the onset of piezoelectric nonlinearities. They can also have excellent far-out phase noise responses, and can be preferred for enhanced noise suppression above 10-kHz Fourier frequency offset.
NOISE IN SURFACE ACOUSTIC WAVE OSCILLATORS
Question 13: How good is the phase noise capability and long-term stability of the Rayleigh wave oscillators noted in Figure 7?
Answer 13: Typical noise floors of Rayleigh-wave resonator oscillators on ST-quartz are down to about -176 dBc/Hz at a Fourier frequency offset of 20 kHz. Long-term aging, attributed to random-walk processes, can be less than 1 ppm/year. Vibrational sensitivity capabilities are given as df/f = 1 x 10-9/g, (which are at least as good as bulk-wave AT-cut devices). Multiple-pole oscillators can have phase noises down to -80 dBc/Hz at 10-kHz Fourier frequency offset, with noise floors of -183 dBc/Hz. Typically, four-pole VCOs can have flat group delay over 400 ppm, to compensate for 1) five-year aging, 2) temperature changes from -40 to +70oC, and 3) frequency accuracy. The phase noise for hybrid Rayleigh-wave VCOs can be about -100 dBc/Hz at 1 kHz offset. Power levels of Rayleigh wave oscillators are typically limited to less than about +15 dBm. (See Chapter 18 of my 1998 SAW book).
14: In your Answer 13 you use the term "dBc/Hz".
(a) What does this mean ? (b) How do you measure this?
Answer 14: (a) This term means "Decibels below the carrier in a 1-Hz bandwidth." It relates to phase-noise measurements, and is measured at a desired frequency offset (called the Fourier Frequency Offset), usually anywhere from a 1-kHz offset to a 1-MHz offset from the nominal carrier frequency. Figure 8 illustrates the phase noise of a hybrid 422-MHz Rayleigh wave oscillator that I used for a wireless application, before and after locking with a Phase-Locked Loop (PLL) for frequency selection. The measured frequency stability of this particular locked oscillator was +/- 10 Hz over a 1 second measuring period.
(b) I used a commercial frequency stability analyzer which can be run to obtain the stability in the phase domain or in the time domain. Time domain measurements are quoted in terms of "Allan Deviation", or "Sigma y of tau" . (Note: If you want to learn more about noise and noise measurements, a VERY good reference book is the USA National Bureau of Standards Monograph 140, called "TIME AND FREQUENCY: Theory and Fundamentals," (B. E. Blair, Editor), U.S. Department of Commerce, Issued May 1974, Library of Congress Catalog Number: 73-600299. Also, for more information on definitions used in frequency and time measurements, see: E. Ferre-Pikal, J. R. Vig, J. C. Camparo, L. S. Cutler, L. Maleki, W. J. Riley, S. R. Stein, C. Thomas, F. L. Walls, J. D. White, ""Draft revision of IEEE STD 1139-1988 standard definitions of physical quantities for fundamental frequency and time metrology - random instabilities, " Proc. 1997 IEEE International Frequency Control Symposium, pp. 338-357, 1997) .
15: What kinds of phase noise do I
have to consider in SAW oscillators?
Answer 15: i) White phase noise, ii) Flicker phase noise, iii) White frequency noise, iv) Flicker frequency noise , v) Random walk. (See Table 18.1 on page 537 of my 1998 SAW text book).
16: (a) What is special about the phase noise characteristic
of an injection-locked SAW oscillator listed in Figure 7? (b) What are
some wireless applications of injection-locked SAW oscillators?
Answer 16: (a) As discussed in Chapter 18 of my 1998 SAW book, within the maximum injection-locking bandwidth, the oscillator tracks (and amplifies) the input signal. Most importantly, the oscillator adopts the phase noise of the input signal source. (See Ref #63 on my web publication page, as well as R. Adler, "A study of locking phenomena in oscillators", reprinted in Proc. IEEE, vol. 61, pp. 1380-1385, Oct. 1973. (I THINK that these injection-locking relationships should apply to ALL types of injection-locked electronic oscillators. )
(b) Applications of injection-locked SAW oscillators include 1) FM demodulation at UHF frequencies which offers good signal-to-noise performance, as well as fabrication simplicity over lumped Inductance-Capacitance (LC) tuning networks, (See Ref. #49 in my web publication page), and 2) Carrier recovery at gigahertz frequencies in Binary Phase Shift Keying (BPSK) modulation systems, with demonstrated Bit-Error Rates (BER) of about 10-7 at a Carrier-to-Noise (C/N) ratio of 14 dB. (See Ref. #87 in my web publication page).
LADDER FILTERS FOR ANTENNA DUPLEXERS
Question 17: Some of the circuits you sketched above related to front-end circuitry employing leaky-SAW (LSAW) low-loss ladder filters with antenna duplexers. (a) What are the merits of such LSAW ladder filters? (b) Sketch a LSAW "building block" component of such a ladder filter. (c) Sketch an illustrative LSAW antenna duplexer employing such building blocks, and illustrate a typical frequency response for a 2.45-GHz ladder filter in a Wireless Local Area Network (WLAN) circuit.
Answer 17: (a) Merits include capabilities for 1) low loss operation (e.g., less than about 3 dB for Tx and Rx stages), 2) high rejection at mutual frequency bands, 3) power handling of at least 1 W, 4) good sidelobe suppression, 5) high rejection at the image frequency and at second and third harmonic frequencies, and 6) very small and light package sizes.
18: What are the chief components of surface
acoustic wave front-end ladder filters and antenna duplexers?
Answer 18: One-port resonators are configured in series-shunt combinations to act as Inductance-Capacitance-Resistance (LCR) Impedance Elements (IE). For energy storage and resonator action the individual one-port resonators can either consist of a long IDT with significant finger reflections, or a short IDT in conjunction with end reflection gratings as shown in Figure 9. Because their relatively deeper sub-surface wave penetration results in a higher power-handling capability, leaky-SAW (LSAW) resonators (e.g. using 42o Y-X LiTaO3) are normally preferred over Rayleigh-wave ones (e.g. using 128o LiNbO3). (See Chapter 13 of my 1998 SAW textbook for more details).
19: Sketch a basic LSAW ladder-filter antenna duplexer.
Answer 19: Figure 10 sketches an illustrative ladder-filter example. (There are many possible variations, depending on the required filtering specifications). The impedance/frequency characteristics for resonator elements IE-1, IE-2, IE-3, IE-4, are selected to provide the desired Tx and Rx filtering responses.(See Chapter 13 of my 1998 SAW textbook for more details).
20: Sketch an illustrative frequency response for
a 2.45-GHz Wireless Local Area Network (WLAN) as designed using LSAW ladder-filter
Answer 20: Figure 11 illustrates such a response. Ladder filters can often be recognized by the characteristic sidelobe "wings" shown here.
SAW IF FILTERS FOR SATELLITE COMMUNICATIONS
Question 21: a) Give an example of a communications system employing wideband linear-phase SAW IF filters with fractional bandwidths of 50%. (b) What types of piezoelectric substrates are used for the wideband SAW filter ? c) What finger slant angles are typically employed ? d) What are some desirable response specifications?
Answer 21: (a) 70-MHz SAW IF filters with 50% fractional bandwidth have been employed in digital data terminals in Mobile Earth Stations (MES) for INMARSAT-C satellite communications. Such wideband filters have employed IDTs with slanted-finger geometry. Their characteristics include 1) extremely-flat passband response, 2) excellent linear phase response across the passband, and 3) large out-of-band suppression. Figure 12 illustrates one such response of a 70-MHz slanted-finger SAW IDT structure with 50% fractional bandwidth.
(b) Typically, Y-Z LiNbO3 or 128 Y-X LiNbO3.(For more on wideband slanted-finger IDTs see Chapter 8 of my 1998 SAW textbook).
(c) Finger slant angles of less than about 7 degrees are used to contain the SAW.
(c) Passband specifications could typically require a passband amplitude ripple of less than about 0.6 dB, a phase ripple of less than 5 degrees, and at least 50 dB out-of-band suppression.
(For detailed computer-aided design techniques for these slanted-finger structures see: H. Yatsuda, "Automatic computer-aided design of SAW filters using slanted finger interdigital transducers," IEEE Transactions Ultrasonics, Ferroelectrics, and Frequency Control, vol. 47, pp. 140-147, January 2000.)
(ASH) WIRELESS RECEIVER
Question 22: (a) In your Answer 6 above, you mention "Time Diversity Receivers." With reference to a basic block diagram circuit, outline the difference between a time diversity (ASH) wireless receiver and a single-conversion superheterodyne receiver. Briefly highlight the operation of the time-diversity (ASH) wireless receiver. (b) What are some of the merits of this time-diversity wireless receiver?
Answer 22: (a) Figure 13 shows the basics of a time-diversity receiver, as compared with those of single-conversion superheterodyne receiver. As shown, the time-diversity receiver has no local oscillator for down conversion. Instead the incoming RF data-modulated signal is time-gated into a low-loss SAW RF delay line. The time-gating is controlled by a pulse generator which alternately switches on/off the RF amplifiers at the input and output to the delay line. The low-loss (e.g. less than ~ 3 dB) SAW RF delay line structure can typically employ Single Phase Unidirectional Transducers (SPUDTs). It is designed to hold hundreds of samples per incoming data bit. Typical delays are in the order of 0.5 microsecond. Since the input/output RF amplifiers are not "on" at the same time, there is no undesirable feedback to cause instability. The gating pulse signals can subsequently be removed from the message data-bit signals in the output detector stage. Signal-processing gains obtained with the time-diversity are comparable with that of the single-conversion superheterodyne receiver.
(b) Some of the target specifications applied to the
time-diversity (Ash) wireless receiver are as follows: 1) Center
frequency 180 to 450 MHz, 2) -100 dBm sensitivity at a 1.0-kb/s data rate,
3) 500 kHz minimum RF bandwidth, and 4) very-low power consumption..
(For more on the time-diversity (Ash) receiver see 1) D. L. Ash, "New UHF receiver architecture achieves high sensitivity and very low power consumption," RF Design, pp. 32-44, December 1994, and 2) "1995 Product Data Book", RF Monolithics, Inc, Dallas, Texas, USA).
(Also for more on low-loss SAW SPUDTs, see Chapter 12 of my 1998 SAW textbook).
CIRCUITS FOR FIBER-OPTICS DATA-COMMUNICATIONS NETWORKS
Question 23: What are the merits of SAW-based clock-recovery circuits in digital regenerative repeater circuits in fiber-optic data communication networks?
Answer 23: SAW-based timing-recovery modules can have excellent jitter-free performance in many instances. One example of their application is for digital regenerative-repeater circuits for fiber-optic networks operating under an Asynschronous Transfer Mode/Synchronous Optical Network/Synchronous Digital Hierarchy (ATM/SONET/SDH) mode in data communications, as outlined in Figure 14. Bit-Error-Rate(BER) performance in each repeater is aimed at BER < 10-11, in conjunction with reliability and long life. Depending on the fiber-optic Synchronous Transfer Mode (STM) employed, these are clock-recovery SAW filters whose center frequencies fb correspond to bit rates of 155.52 Mb/s (STM-1), 622.08 Mb/s (STM-4) or 2488.32 Mb/s (STM-16). Effective Qs of these transversal SAW filters are normally in the approximate range 700 < Q< 1500. Insertion losses for these linear-phase clock-recovery SAW filters are typically in the range 15 to 20 dB, with very low phase-slope ripple across the passband.
24: How are these SAW clock-recovery filters employed
in regenerative repeaters?
Answer 24: Figure 15 shows the basics of an illustrative regenerative repeater for a fiber-optic data communications system employing Non-Return-To-Zero(NRZ) modulation. (This outlines the circuitry for the timing-recovery "block" in Figure 14). One portion of the down-converted electrical signal is applied to a clock-frequency extraction circuit. Since the power spectrum of an NRZ signal has nulls at the signaling ratefb and a maximum at fb/2, an indirect method is used to extract clock frequency fb. As shown, the down-converted signal is first pre-filtered at the power spectrum peak fb/2. This pre-filtered signal output is applied to a frequency-doubling (squaring) circuit, for extraction of signaling frequency fb, which is then applied to the SAW filter with center frequency fo = fb. Timing comparisons and "0" or "1" signaling decisions are obtained, following which the regenerated electrical signal is up-converted to the optical output and passed "down the line" to the next repeater stage. Note that in some applications the SAW filter and central components in Figure 15 can be combined into an Application Specific Integrated Circuit(ASIC) for circuit packaging.
25: What limits the usable Q =
= 3-dB bandwidth) of these SAW filters?
Answer 25: Static detuning over the entire repeater-circuit chain places a upper limit on the usable Q value for the SAW filter. Additionally, the ringing time of the SAW filters places a lower limit on usable Q.
26: What types of SAW filters are used in these
regenerative repeater modules, and how difficult is their design?
Answer 26: Since the SAW clock-recovery filters are required to have extremely-high phase linearity across the passband, transversal (i.e., delay-line) types of SAW clock filters appear to be favored over SAW resonator filters. The design of SAW filters operating at center frequency fo= 2488.32 MHz can be especially demanding. To date, design techniques for SAW clock-recovery filters over 2 GHz have included 1) delay-line structures operating at the fundamental center frequency on piezoelectric crystal substrates, 2) filters operating at the third harmonic on piezoelectric crystal substrates, and 3) thin-film filters fabricated with composite layers of Silicon Dioxide/Zinc Oxide/Diamond/Silicon. Figure 16 shows the response of an illustrative SAW clock filter employing Silicon Dioxide/Zinc Oxide/Diamond/Silicon, and operating at 2.488 GHz.
(See Chapter 19 of my 1998 SAW textbook).
SAW CONVOLVERS FOR INDOOR/OUTDOOR WIRELESS COMMUNICATIONS
Question 27: What are real-time SAW convolvers used for?
Answer 27: They find application in indoor/outdoor spread-spectrum wireless for packet-data and packet-voice communications. They also can be well suited to combat multipath interference due to spurious reflections in indoor environments.
28: What are some of the merits of SAW convolvers
for wireless communications?
Answer 28: They can have the merits of broad bandwidth, large processing gain, and small size. Also, as I just mentioned in Answer 27, they can provide improved performance against multipath interference. Indeed, SAW convolvers with RF bandwidths greater than the coherence bandwidth are well suited to indoor spread-spectrum communications in buildings with highly-reflecting structures. They can also give good jamming protection if pseudo-noise spreading codes are employed. ( NOTE: SAW convolvers are used in IF stages, not RF ones!) .
29: In their operation, are SAW convolvers
designed to actually implement the convolution of two signals ?
Answer 29: No! They are actually implemented to effect autocorrelationbetween an incoming signal message bit and a locally-provided time-reversed reference replica of the coding applied to the message signal.
30: I really do not understand the difference between
convolution and autocorrelation. Can you demonstrate this to me in a simple,
Answer 30: Convolution and autocorrelation relate to the way the interaction between two signals is processed as a function of time. Maybe Figure 17 will help to demonstrate this. Here, an autocorrelation peak occurs at a time when the animals are identically overlapping one another.
31: Now give me a block-diagram sketch of a very
basic real-time SAW convolver.
Answer 31: Figure 18 shows a basic real-time SAW convolver on a single-crystal piezoelectric substrate. The input message-coded IF signal at frequency fis applied at Port 1. A time-reversed replica of the message-coding sequence , also at frequency f, is applied to Port 2. Their related SAW signals propagate under the metal film, where autocorrelation takes place. The metal film must be long enough to contain an entire code bit. The autocorrelated output at frequency 2f is obtained at Port 3.
32: Why is the Port 3 output signal at frequency
2f, while the input (Port 1) and reference (Port 2)
signals are only at frequency f?
Answer 32: The situation here is exactly the same as for an ordinary three-terminal analog mixer component. Since the input and reference signals have to MIX, the convolver has to operate non linearly! To do this, at least one of the input signals (normally at Port 2 ) has to be large enough to drive the sub-surface SAW region into nonlinearity. Note that a Rayleigh-wave crystal cut is therefore preferred, instead of a Leaky-SAW (LSAW) one. The reason for this is that the sub-surface penetration of a Rayleigh wave is much less than a LSAW one. This means that for the same input powers, the power DENSITY of the Rayleigh wave will be higher, and make it easier to get in to nonlinear operation.
33: In Figure 18 you gave an outline of a very
basic real-time SAW convolver. Sketch an outline of a more sophisticated
one, and mention some of its relative advantages.
Answer 33: OK! Figure 19 shows the basics of a dual-track real-time IF SAW convolver. Some such convolvers have been reported with correlation interaction times of up to 22 microsecond. The design trick here is to arrange the polarities of the interdigital transducers (IDTs) at Port 1 input so that they excited in-phase SAWs in both tracks. However, the polarities of the IDTs at reference Port 2 are arranged to excite 180oout-of-phase SAWs between the two tracks. The autocorrelated signals in Track 1 and Track 2 can be summed by a differential summer. This is not the end of the story, however! Any spurious undesirable SAW reflections within Track 1 and Track 2 will be in-phase, and will therefore cancel out in the differential summer. Typical reported processing bandwidths B for this structure are B = 50 MHz at 350-MHz center frequency, with Tme-Bandwidth (TB) products in the order of TB = 150.
34: (a) Define the convolution efficiency hc
a real-time SAW convolver. (b) What are some typical values of convolution
efficiency for real-time SAW convolvers?
Answer 34: (a) This is normally defined as hc = 10.log10[(Pout)/(Ps.Pr)]. It is usually expressed in (dBm)-1. In this evaluation the output power Pout at Port 3 is normally measured with signal and reference powers Psand Pr both set at 0 dBm (i.e. 1 mW).
(b) Representative values of convolution efficiency vary from about -70 dBm for a dual-track basic convolver on a piezoelectric crystal substrate to about -46 dBm for a layered structure involving ZnO/SiO2/Si. (See Chapter 17 in my 1998 SAW textbook).
35: A SAW convolver has a rated convolution efficiency hc=
-46 dBm. If the signal input power Ps is 10 dBm
(10 mW) and the reference power Pr is 20 dBm (100 mW),
what is the correlated output power Pout?
Answer 35: Expressed in dBm units we have Pout = hc+ Ps + Pr . Before going any further, however, we must remember that the IDTs at Port 1 and Port 2 are bidirectional. This means that each IDT will lose 3 dB from the autocorrelation process. As a result the output power at Port 3 will be Pout = (-46) + (10 - 3) + (20 - 3) = -22 dBm = 6.3 microwatt.
36: If the output noise floor level in the previous
SAW convolver is -75 dBm, determine the output Signal- to-Noise (S/N)
Answer 36: This gives the output signal/thermal noise ratio (at output frequency 2f) as S/N = (-22) - (-75) = 53 dB, which also corresponds to the dynamic range in this convolver example.
37: Can a SAW convolver be used for synchronous
or asynchronous communications?
Answer 37: Yes, but a given design will only be for only one mode - not both. For example, SAW convolvers have been designed for synchronous packet-data communication using Binary Phase Shift Keying (BPSK)/Frequency Hopping (FH) modulation and 255-chip orthogonal Kasami code sequences. Asynchronous types have employed Direct Sequence (DS)/Frequency Shift Keying (FSK) or Direct Sequence (DS)/Code Shift Keying (CSK) spectral spreading, using Pseudo-Noise (PN) 127-chip maximal sequence generators. (See Chapter 17 in my 1998 SAW textbook).
38: What are some of the frequency bands that modems
with these IF SAW convolvers have operated in ?
Answer 38: These include 1) the 900-MHz spread spectrum band using the DS/CSK mode, 2) Full-duplex operation in the 2-GHz spread-spectrum band, and 3) the licence-free spread-spectrum band in Japan below 322 MHz. (See Chapter 17 in my 1998 SAW textbook).
WIRELESS BAGGAGE LABEL SECURITY IDENTIFICATION "TAGS"
Question 39: What are SAW wireless label identification "tags", and what are they used for?
Answer 39: SAW wireless label "tags" are used for identifying a wide range of luggage or commercial shipping-container items. Instead of scanning an item with an optical scanner, as at a supermarket checkout counter, the SAW inspection transmitter circuit sends a high frequency radio signal pulse (e.g., at 1000 MHz) from a transmitter to a SAW "tag" on the item to be inspected. The SAW baggage tag itself is a passive component. Basically, it is a coded SAW interdigital transducer (IDT) which has a small antenna attached to it . When excited by the interrogating radio signal pulse from the nearby RF transmitter, it can radiate a coded RF signal back to the source, for identification, as sketched in the basic circuit of Figure 20. These tags can be very small indeed ! (For artistic illustration the size of the SAW label sketched in Figure 20 is very greatly exaggerated here !)` A choice of different code-length sequences can be employed in each IDT fabrication , depending on its length (e.g., 128 bit-codes).
40: What are
some of the reported merits of commercial SAW wireless Radio Frequency
Identification (RFID) "tags" ?
Answer 40: (a) SAW RFID tags are entirely passive. (b) They can be read with only milliwatt levels of RF interrogation power. (c) They have a high level of radiation "hardness" under gamma-ray sterilization of medical and food products requiring sterilization with gamma radiation. (d) "Read" ranges of 3 to 20 meters depending on the system. (d) Good electromagnetic interference filtering. (d) Tag temperature range capabilities from -100oC to over +200oC. (e) EPC compatible with EPC-64 and EPC-96 RFID specifications. (f) SAW tag capabilities for 24-, 32-, 48-, 64- and 96-bit capacities. (g) Operational capabilities for operation in the 1.7 GHz and 2.5 GHz frequency bands.
(For more on SAW wireless tags and their potential see, for example, C. S. Hartmann, "Future high volume applications of SAW devices," Proceedings of 1985 IEEE Ultrasonics Symposium, vol. 1, pp. 64-73, 1985.
41: When I check
out my groceries at the supermarket, the optical scanner at the checkout
counter can only scan one item at a time. In the case of SAW RFIDs using
electromagnetic wave interrogation, can the SAW inspection/detection circuit
only handle one RFID at a time? What happens if there are several RFID
tags close together, with the scope of the wireless detector circuit?
Answer 41: Another tricky question! When interrogated by a single wireless transmitter/receiver, multiple reflection signals from RFID tags could of course occur when several of these are close together ( such as placed on a number of different jars of jam on a shelf), within the radiation pattern of the single interrogating antenna. This would result in a multiplicity of received codes at the interrogator in the same time interval! To overcome this, one reported technique uses a phase modulation of selected finger pairs on each SAW RFID device, which places a unique identifier on the signal returned to the wireless interrogator circuit. (See, for example, P. J. Edmonson and C. K. Campbell, United States Patent No; US 6,827,281 B2, Dec. 7, 2004, "Encoded SAW RFID tags and Sensors for Multi-User Detection Using IDT Finger Phase Modulation).
42: In your
answer to Question 40 you mention the terms "EPC-64" and "EPC-96". What
do you mean by these ?
Answer 42: (a) "EPC" stands for "Electronic Product Code" and represents a numbering scheme for the unique identification of objects. EPC may be considered as a Radio Frequency Identification (RFID) evolution of the Universal Product Codes (UPC) currently used as optical-scanning barcodes in supermarkets and elsewhere. There are several proposed standards of EPC, relating to the amount of data stored in the interrogating transponder. Current EPC standards include EPC-64 employing 64 bits of information data and EPC-96 employing 96 bits of information data.
43: Give an
example of the coding distribution for an EPC-96 system.
Answer 43: Consider that we can divide the 96-bit code into four Segments from left to right.
Segment 1 is the Header and is 8 bits in length (0 to 7 bits), This identifies the EPC version in use.
Segment 2 is the EPC Manager, which employs 28 bits of data (8 to 35 bits). This is used to identify the particular Manufacturer of the product in question. The binary number 228 gives us 228 = > 268 million identifiable Manufacturers !
Segment 3 is the Product Object Class and is 24 bits in length (36 to 59 bits) . This gives us 224= > 16 million products to identify.
Segment 4 is the Serial Numberfor a given product, and is 36 bits in length (60 to 95 bits). This gives us 236 = > 68 billion possible serial numbers!
44: But before
I figure out how the above EPC data be met by a SAW RFID tag design,
first of all sketch a simple binary-coded SAW wireless RFID label tag,
and explain its operation.
Answer 44:I have sketched a simple illustrative SAW wireless tag in Figure 21, employing IDT reflector pairs configured, for example, as a 110011011 binary code, as governed by the individual IDT relative "polarities". The antenna is shown as a simple one-turn loop antenna. Note that input/output IDTs have a common bus bar. The RF pulse transmitter in Figure 20 sends an interrogation pulse to this SAW tag. After a short time delay the SAW tag re-radiates an RF signal as a 110011011-coded RF waveform. This is subsequently detected by the time-gated receiver and phase-detector circuit of Figure 20. Note that an operational requirement for this particular circuit is that the free-propagation distance between transmitter and SAW tag must be greater than the IDT code length.
45: But am I
restricted to the use of IDT sections as reflectors in Figure 21 above
Answer 45:No. It is normally much easier if I use thin metal film reflectors strips - each with modest SAW reflectivity capability - as shown in Figure 21a.
46: How do these
reflector strips work in the one-port device of Figure 21a ?
Answer 46:The IDT to the left is directly connected to the tag's antenna which receives an interrogation RF signal. The RF signal is converted to a SAW which is reflected sequentially from the various reflector strips and returned to the antenna. These reflector strips can be placed on the piezoelectric crystal substrate (typically 128o LiNbO3) to encode the RFID tag using amplitude weighting, phase weighting or other variables.
47: (a) Give
me an example of the level of bit encoding I can attain with the
RFID tag configuration of Figure 21a. Assume that I only have a maximum
of 16 reflectors.
(b) Highlight, (without giving mathematical details), how you could improve the above simple 16-bit design to meet EPC tag specifications. Also give a reference to such a design
Answer 47:(a) First of all, consider the simplest design where the 16 reflectors are separated at fixed intervals. Further consider that the placement of each individual reflector strip corresponds to a binary "1", while the absence of a reflector strip corresponds top a binary "0". This will give us a capability of only 216= 65, 000 unique tags, which would not be of any use for the EPC-64 or EPC-96 tag numbers mentioned above.
(b) However, recent SAW design techniques involving a different type of data encoding - using a higher number of data bits for each signal pulse, together with phase encoding of reflector strip placements and a higher data density -- have shown that it is possible to attain 264 = 1.8 x 1019unique RFID tag numbers using the same size SAW device as for the simple 16-bit one considered above.
For reference to the SAW design of part (b) see C. S. Hartmann, "A global SAW ID tag with large data capacity, Proceedings of 2002 IEEE Ultrasonics Symposium, vol. 1, pp. 65-69, 2002.
(THIN) FILM BULK ACOUSTIC RESONATORS AND FILTERS FOR THE 2 TO 5 GHz
Question 48: a) What is an "FBAR" and b) where is it used in wireless/mobile systems ?.
Answer 48:a) "FBAR" stands for "(Thin) Film BulkAcousticResonator. By themselves FBAR resonators can be employed as feedback elements in high frequency VCOs. Bandpass FBAR ladder-filter modules constructed from FBAR resonators can also be employed as front-end duplexer filters in the 2-GHz to 5-GHz range. As well as a small package size (e.g., ~ 125 m3 in a PCS duplexer), FBAR duplexers have good power-handling capability (e.g. > ~32 dBm in a PCS duplexer).
49:What are the merits
of FBAR filters compared SAW filters at these frequencies - especially
in the 5-GHz range?
Answer 49: SAW filter dimensions decrease with increasing frequency. As I noted in Answer 5, a packaged 1.880-GHz SAW Tx-filter for USA Personal Communications Services (PCS), (see Figure 1.4 in my SAW book), may only have an area in the order of 3 mm x 3 mm. And as we get up into the 5-GHz range, (and unless we may choose to operate in a harmonic mode), SAW fabrication IDT line width dimensional limitations and tolerances become too severe for all but the most sophisticated fabrication systems. But while SAW device fabrication resolution is concerned with width parameters , the FBAR designs are dictated by depth parameters thereby offering the potential for less stringent fabrication constraints.
50: a) What piezoelectric thin-film
materials are currently employed or examined for FBAR filters? b)
Give three important FBAR filter design parameters? c) Why
are these important?
Answer 50: a) These currently include Zinc Oxide (ZnO), Aluminum Nitride (AlN), and PZT (PbZrxTi1-xO3). b) Three important parameters are i) Electromechanical coupling factor k2, ii) Temperature Coefficient of Delay (TCD), and iii) acoustic velocity v. c) i) Higherk2 values mean larger fractional bandwidth capability ZnO has a larger k2 (~ 8.5%) than AlN (~ 6.4% in an epitaxial film), while PZT has reportedly still-higher k2values. ii) However, the TCD of ZnO (~ 60 ppm/oC) is not as good as AlN (~ 25 ppm/oC). Low values of TCD are required for maintaining frequency accuracy over a wide temperature range. iii) AlN has a higher acoustic velocity (~ 10,400 m/s) than ZnO (~ 6330 m/s). A higher acoustic velocity means that the device can operate at a higher frequency using the same physical dimensions.
(For more on FBAR resonators and filters, see, for example, a) K. M. Lakin, "Thin film resonators and filters," Proceedings of 1999 IEEE Ultrasonics Symposium, vol. 2, pp. 895-906, 1999, b) H. P.. Lobl et al, "Piezoelectric materials for BAW resonators and filters," Proceedings of 2001 IEEE Ultrasonics Symposium, vol. 1, pp. 807-811, 2001).
51: Why are we now talking about bulk acoustic
wave (BAW) filters and resonators, when we have been so far discussing
SAW filters and resonators?
Answer 51: Their are many circuit equivalencies in the modelling of SAW and BAW resonator and filter circuits. For example, one equivalent circuit for SAW filter modeling employs the Mason Equivalent Circuit that was first applied to BAW filters and resonators.
(For more on the Mason Equivalent Circuit, see Chapter 4 of my 1998 SAW book as well as, for example, J. F. Rosenbaum,Bulk Acoustic Wave Theory and Devices, Artech House, Boston, 1988)
52: a) Sketch the basic configuration of one type
of FBAR resonator, and highlight its operating principles.
b) Give some typical response and size parameters for GHz frequency FBAR
ladder filter front-end duplexers employing series-shunt FBAR resonators.
Answer 52: Figure 22 shows the basics of one type of FBAR resonator. The resonator itself is composed of a piezoelectric layer contained between input/output connectors, which is excited to implement mechanical resonance. It is deposited on top of a highly resistive wafer substrate, such as silicon (Si). For optimum performance the design aim is to deposit an epitaxial (i.e., single crystal) piezoelectric layer, with a particular crystal axis orientation for a given piezoelectric. This can be tricky ! Analogous to a microwave resonator, the fundamental resonance frequency is that which results when the piezoelectric film thickness is 1/2 acoustic wavelength. In order to minimize mechanical damping, the resonator requires a large acoustic mismatch with outer boundaries. This is achieved in the design of Fig. 22 by cutting away the bottom support base, using micro machining or plasma etching.
b) A reported 5-GHz FBAR of this type on AlN had an unloaded series-resonanceQs= 913 at 5.173 GHz, with a k2 x Qs product of 58. Using such an FBAR in a 5-GHz front-end ladder filter, (in the same way as for the SAW ladder filter of Fig. 10 above), a fractional bandwidth of 5.0% was obtained, with a 2-dB bandwidth of 210 MHz and a 3-dB bandwidth of 260 MHz, suitable for 5-GHz WLAN applications. It was indicated that this particular FBAR response outperformed an equivalent SAW ladder filter in both the passband and out-of-band responses. The filter package size in this design was 2.5 x 2.0 x 0.9 mm.
(For further details of this particular 5-GHz FBAR resonator and filter see, T. Nishihara, T. Yokoyama, T. Miyashita, Y. Satoh, "High performance and miniature thin film bulk acoustic wave filters for 5 GHz," Proceedings of 2002 IEEE Ultrasonics Symposium, (to be published).
53:Is the FBAR filter structure of Fig. 22 the only design
under study at this time?
Answer 53: No. Instead of having a "free-space" piezoelectric membrane as in Fig. 22, another type of FBAR under development uses a Solidly Mounted Structure (SMR), where the bottom resonator section is not "free", but is deposited on layered films which are configured to act as a reflecting "mirror". This layered film structure is known as a Bragg reflector. (For more on SMR filters, see, for example, R. Lanz, M-A Dubois, P. Muralt, "Solidly mounted BAW filters for the 6 to 8 GHz range based on AlN thin films," Proceedings of 2001 IEEE Ultrasonics Symposium, vol. 1, pp. 843-846, 2001).
54:Sketch an LCR equivalent circuit, and illustrative
frequency response for an FBAR resonator.
Answer 54: As indicated in Fig. 23(a), the same LCR equivalent circuit representations can be used both for both FBAR and one-port SAW resonators.(For more on one-port and two-port SAW resonators see Chapter 11 of my 1998 SAW book). Resonator equivalent parametersCs,Ls, and Rs establish the series resonance with minimumimpedance at notch frequency fs in Fig. 23(b). But the resonator is also just a capacitor , with parallel capacitance Cp and tan(delta) dissipation loss resistanceRp. At frequencies above fs, therefore,Cp and Rpprovide a parallel resonance with Rs, Ls, resulting in an impedance maximum at frequency fp. Rleadrepresents contact and lead resistance here. Depending on the design some connection inductanceLlead may also be present.
SAW COMB FILTERS
55: Is a SAW filter
constrained to having just a single passband response, such as in the example
of Figure 11?
Answer 55: No. This is where the analog/digital hybrid capability of the SAW filter can be used, as mentioned in Answer 2 ! For example, we can apply digital-filter concepts to the design of a SAW filter. One such sample design illustrated here employed the Remez Exchange Algorithm used in linear phasedigital filter design. This was derived in the early 1970s as a tool for designing finite impulse response (FIR) linear phase digital filters. (See, for example, J. H. McClellan, T. W. Parks and L. R. Rabiner, "A computer program for designing optimum linear phase digital filters," IEEE Transactions Audio and Electroacoustics, vol. AU-21, No. 6, pp. 506-526, December 1973. Essentially, given a desired frequency response, it supplies a finite set of impulse response coefficients for the digital filter synthesis, thus yielding an optimum approximation to the desired linear phase response. Its application to SAW filters is covered in some detail in Chapter 8 of my earlier 1989 SAW book listed below. Figure 24 illustrates a prototype singleSAW filter, designed in this way to perform as a 10-band comb filter. Other Remez examples are given in my 1989 SAW book.
SAW WIRELESS BIOSENSORS FOR VAPOR DETECTION AND IDENTIFICATION
56: (a) Can SAW resonators
be used as biosensors? (b) If so, give two examples.
Answer 56: (a) Yes.
(b) 1. Uncoated SAW resonators have been used in fast gas chromatography for electronic nose simulation of olfactory responses. This is used to obtain a high resolution visual image of specific vapour fragrances containing a variety of chemicals. (See, for example, E. J. Staples, "Electronic nose simulation of olfactory response containing 500 orthogonal sensors in 10 seconds" Proc. 1990 IEEE Ultrasonics Symposium.)
2. Bio-coated SAW resonators have been used for on-the-spot vapour phase detection of plastic explosives containing nitro groups such as TNT, RDX and others, using a SAW resonator immunosensor array. Detection sensitivity is dependent on the biolayer deposited on the surface of the SAW resonator. ( See, for example, S-H Lee, D. D. Stubbs, W. D. Hunt, and P. J. Edmonson, "Vapor phase detection of plastic explosives using a SAW resonator immunosensor array" Proc. IEEE Sensors Conference, Irvine, California, 2005).
57: Sketch a basicuncoated
two-port SAW resonator, and highlight its important parameters for
the sensor used considered here.
Answer 57: Figure 25 depicts the basics of a two-port SAW resonator. Reflection gratings "bounce" back SAW that would otherwise "escape" from the IDTs. Reflection gratings can be fabricated using open or shorted metal strips. Shorted gratings, such as shown in Figure 25, can have better reflection qualities. SAW resonators are generally designed to have low insertion losses in the range 1 to 3 dB, and high-Q values (greater than 1000). Where Q = fo/Df at resonance frequency fo, and Df is usually measured at the 3-dB points in Figure 26. For temperature stability they are fabricated on temperature-stable substrates such as ST-cut quartz. The resonance is critically dependent on the spacing between the IDTs and the spacing between the reflections gratings and adjacent IDTs. The higher the Q, the higher will be the resolution of the oscillator spectral response. Figure 26 shows a typical frequency response, for a particular spacing between gratings and IDTs. (See Chapter 11 of my 1998 SAW book).
58: Sketch, and discuss,
the basics of biocoated
two-port SAW resonator oscillator circuit, such as reported for plastic
explosive detection as in your Answer 57, and highlight its important parameters.
Answer 58: Figure 27 depicts the basics of one biocoating configuration of a two-port SAW resonator oscillator for vapor detection and identification. The biolayer comprises an antibody coating to detect the presence of target molecules from the vapor of the small molecules from the gas phase . This causes an immobilization of the antibody coating of the target molecules structure, and results in a baseline shift of the oscillator frequency. The oscillator is transmitted to a test site for analysis. A special analyses can subsequently be applied to identify the vapor in question. A bank of such resonator oscillators with different identifying biolayer antibody coatings (e.g., anti-TNT or anti-RDX antibodies) can be employed for identification of more than one vapor. The normal linear relationship between frequency shift and mass loading of the resonator surface has been extended to cater for the more complex case of such antibody layer perturbations. (See, for example, W. D. Hunt, D. D. Stubbs and S-H Lee, "Time-dependent signature of acoustic wave biosensors," Proc. IEEE, vol. 91, pp. 890-901, 2003)
SAW SENSORS AND IDENTIFIERS USING SELECTABLE REFLECTOR ARRAYS
A "normal" SAW reflection grating, such as shown in Figure 25 can be designed
as an "shorted strip" one or as an "open-strip" one.
(a) How many strips are typically used in these?
(b) Are there any other ways we can effect the reflection of SAW waves, but in a controllable "on" or "off" manner?
(c) Explain your above response in some more detail.
(a) An ordinary mirror only has one reflecting surface, as it reflects all of the incident light. However a single SAW metal strip can only reflect about 1% of an incident SAW. That is why we need many strips - usually 100 or more - strategically separated - so that the combined SAW reflections reinforce each other to totally reflect an incident SAW. (But in practice, things are not always perfect!)
(b) Yes, using split-electrode IDTs of the type shown in Figure 2(b)
(c) The split-electrode IDT shown in Figure 2(b), has some unusual properties compared with the solid-electrode type of Figure 2(a). In Figure 2(a) each solid electrode is one-quarter of a wavelength wide, while each of the split electrodes in Figure 2(b) is one-eighth of a wavelength wide. That means that the electrical resonance frequency of the split-electrode IDT is one half of its mechanical resonance frequency. So that the mechanical reflections cancel at exactly the electrical resonance frequency. (See Figure 6.15 in my 1998 SAW book). However, there will still be significant SAW reflections just off the electrical resonance frequency. But if we put a short-circuit load across the split-electrode IDT, the SAW will pass right under it with no reflections! (See, for example, A. J. DeVries, "Surface wave bandpass filters", in text book, Surface Wave Filters, H. Matthews (Ed.), Chapter 6, Wiley, New York, 1977). This gives us the means for controlling SAW reflectivity by opening or shorting a load across the split-electrode IDT. With intermediate magnitude and phase reflectivities by using other than a short-circuit load.
(a) How can we in situ open or short load across a split-electrode IDT, in order to control its reflectivity?
(a) How many strips are typically used in these?
(b) Mention a wireless communication example of the above technique
(a) A fluidic channel can be built into the surface structure of a split electrode IDT, to inject a conductive fluid across a split-electrode pair, as sketched in Figure 28, and thereby short out the IDT in question.
(b) Individual split-electrode IDT in an array of these, as outlines in Figure 28, can then be switched on or off, so that the output data from such an array resembles a Pulse Position Modulation (PPM) type of data transfer. (See, P. J. Edmonson and C. K. Campbell, US Patent No: US 6,967,428 B2, Nov. 22, 2005, "Selectable reflector arrays for SAW sensors and identification devices')
Sketch a basic SAW linear FM chirp filter and briefly describe its construction and operating principles.
(a) Figure 29 illustrates the construction of a very basic SAW linear fM chirp filter. Here the finger widths and spacings of the IDT electrodes are fabricated so that, when impulsed, the detected signal at the (wideband) output IDT varies linearly with frequency. This will be in the form of a frequency up-chirp or frequency down-chirp, depending on placement of the output IDT. (Note that the phase of the output signal will have a linear term in time t, and a quadratic term in t2.) The signal processing gain corresponds to the time-bandwidth (TB) product, where T = chirp filter dispersion time (normally quoted in microseconds), and B = chirp bandwidth (normally expressed in MHz), The linear FM chirp slope m is given as m= B/T (in MHz/ msec). Typical TB products for linear SAW linear FM chirp filters are TB = 10,000, while TB = 1 for a SAW filter with uniform finger spacing).
(See Chapter 8 of my 1998 SAW book for more on various types of SAW chirp filters),
(a) What do we mean by the term Fourier Transform Pair as applied to signal processing, and especially to SAW applications? Express in general terms, without equations.
(b) What we mean by the term Convolution as applied to signal processing?
(a) The impulse response h(t) of any system is related uniquely to its frequency response H(f) - and vice versa - by a Fourier Transform Pair. As one application to basic SAW filter design, the IDT finger pattern is a sampled version of the impulse response h(t) of the desired frequency response H(f), where h(t) represents the Inverse Discrete Fourier Transform
(b) The time-domain convolution of signal functions f1(t) and f2(t) corresponds to the multiplication of their respective frequency response functions H1(f) and H2(f). Convolution corresponds to a reversal of one of the time responses, together with a relative time displacement of one of the responses, so that the two signals are mathematically manipulated as moving towards one anther, and overlapping, as in Figure 17.
(a) Name three types of SAW real-time processors for mobile/wireless applications utilizing SAW linear chirp filters and Fourier Transform techniques.
(b) Where can I find more information on these Fourier Transform Processors?
(1) Single-stage real-time Fourier-Transform Processor as a compressive receiver for spectrum analysis of signals.
(2) Two-stage real-time Fourier-Transform Processor for Cepstrum Analysis.
(3) Two-stage Fourier-Transform Processor for real-time on-line filtering.
(b) See for example, M. A. Jack, P. M. Grant, and J. H. Collins, "The theory, design and applications of surface acoustic wave Fourier-transform processors," Proc. IEEE, vol. 68, pp. 229-247, 1980. Also see Chapter 16 of my 1989 book: Surface Acoustic Wave Devices and Their Signal Processing Applications.
(a) Sketch the basic circuitry for the single-stage real-time SAW Fourier Processor mentioned in your Answer 63, and highlight its principles of operation. Exclude circuit components such as compensation of inherent delays etc.
(b) Give some typical operational parameters for such s single-stage Fourier Transform Processors.
(a) A very basic circuit for this Processor is shown in Figure 30, which employs two, or three, linear FM chirp filters with the same chirp slopem. This is based on the mathematical trick that the Fourier Transform of the product of signal s(t) and the impulse response time h(t) for the linear FM chirp filter can be expanded mathematically into three separate terms involving a pre-multiplication, convolution , and post-multiplication. The corresponding circuit is as shown in Figure 30. Note that for convolution to be achieved the convolver chirp slope must be the opposite of that for the pre-multiplier The optional output chirp filter serves to remove a residual quadratic phase term if both the magnitude and phase of the output are required for network analysis.(
(b) These can have 100% duty cycle, with spectral resolutions, with analytic bandwidths up to 1 GHz. Spectral resolutions can vary from the kHz to the MHz range. IF frequencies can be in the GHz range with processing times in the 25 to 60 microsecond range. This can be much less than for digital Fourier Transform Processors of the same price.
(a) What is Cepstrum signal processing used for ?
(b) State very briefly how Cepstrum signal processing can be achieved using a two-stage real-time Fourier Transform Processor mentioned in Answer 62.
(c) Give a classic reference paper dealing with Cepstrum analysis
(a) Cepstrum signal processing is a method for analyzing the power spectrum of a signal which contains a periodic echo. It is based around the observation that the logarithm of the power spectrum of a signal with a small echo component has an added periodic component due to that echo. Thus, the echo component should be separable from the signal if a second Fourier transform is applied to the logarithmof the power spectrum, (i.e., log(A.B) = log (A) + log(B) ).This gives the Cepstrum response output in a pseudo-time domain, with the dimensions of seconds.
(b) The Cepstrum processor utilizes two cascaded processors of Figure 30, with a logarithmic amplifier and detector located between the output of the first processor and the input of the second one. In this way high-speed real-time processing can determine pulse durations and repetition rates from about 50 nanoseconds to 50 microseconds, as well as the bit rates of binary codes.
(c) A classic Cepstrum paper - with a most unusual title - is: B. P. Bogert, M. J. R. Healey and J. W. Tukey, The quefrency analysis of time series for echoes: Cepstrum, pseudo-autocovariance, cross-cepstrum and saphe cracking, " in M, Rosenblatt (ed), Proc. Symposium on Time Series Analysis, Wiley: New York, pp. 209-243, 1963.
How do we achieve real-time on-line filtering, using a two-stage real time Fourier processor as mentioned in Answer 62?
Instead of using a logarithmic amplifier and detector between the first a nd second processors as in Answer 64, we use a third mixer, gated by a real-time filter function H(2pmt). This achieves amplitude-clipping or time-gating of the signal output from the first Fourier processor, and so allows for on-line adaptive-filtering or fixed-filtering of spread spectrum signals for suppression of narrow-band interference. (See Chapter 16 of my 1989 SAW book)
DEFINITIONS AND ABBREVIATIONS
Question 67: I do not understand some of the phrases used to describe mobile/wireless handset units. Tell me what the following abbreviations mean, namely (a) GPS-enabled, (b) Bluetooth-enabled, (c) Multi-band, (d) Multi-mode, (e) 3G?
Answer 67: (a) GPS stands for Global Positioning System. This enables accurate position determination by means of the triangulation of signals from satellites, and lets you locate where you are (e.g. while traveling in your car, or in a boat fishing in one of Canada's many lakes, etc.). Typically, an integrated GPS unit can use 1 or 2 front-end SAW RF filters for enhanced detection of the satellite signals. AGPS-enabledunit means that the GPS unit can be combined with other add-on services.
(b) Bluetooth involves short-range wireless systems designed for operation in the unlicensed 2.4-GHz Industrial, Scientific and Medical (ISM) band. Bluetooth-enabled systems are intended for portable linking to various units such as mobile handsets or notebooks. A Bluetooth receiver can typically employ 1 front-end RF SAW filter. (For more on license-free spread-spectrum bands see page 528 of my 1998 SAW book).
(c) Multi-band mobile/wireless transceivers can operate in more than one frequency band. One example is for GSM three-band Worldphones that can operate in GSM, DCS, or PCS modes. Recall that GSM stands for Global Systems for Mobile Communications, DCS stands for Digital Cellular System, whilePCSstands for Personal Communications Services . The latter operate in the 1800-MHz and 1900-MHz frequency bands. (See the Glossary definitions section on pages 613-618 of ny 1998 SAW book.)
***** Note that GSM is often referred to as the world's firstdigital wireless technology. However, I personally consider it to be the world's second digital wireless technology - the first digital one being Morse Code wireless transmission, that has been around for many, many years! ******
mobile/wireless transceivers are those that can operate in more than one
mode of operation. These modes include AMPS, GSM, TDMA. (Again,
see the Glossary definitions section on pages 613-618 of ny 1998 SAW book.)
(e) 3G refers to Third-Generation mobile/wireless systems operating in the 2100-MHz band.
68: What do we mean
by Direct-Conversion Zero-IF?
Answer 68: Direct Conversion Zero-IF (ZIF) receivers are those which directly down-convert an incoming RF signal to baseband, as opposed to traditional superheterodyne receivers that incorporate one or more intermediate-frequency (IF) filter stages between RF and baseband. The use of ZIF stages will, of course, depend on the mobile/wireless system involved, and on the sensitivity specifications placed on incoming RF signals.
My 1998 SAW text book is:
Colin K. Campbell, Surface Acoustic Wave Devices for Mobile and Wireless Communications. Academic Press:
Boston, 633 pages, 1998. (ISBN Number 0-12-157340-0).
My 1989 SAW textbook, which includes
chapters on the Remez Exchange Algorithm, as well as on real-time SAW Fourier
Transform Processors is:
Colin Campbell, Surface Acoustic Wave Devices and Their Signal Processing Applications. Academic Press: Boston, 470 pages, 1989. (ISBN Number 0-12-157345-1).
Check my 1998 book contents at
You can see my biographical sketch and photo at http://www3.sympatico.ca/colin.kydd.campbell/ckcbiog.htm
Or find my author and co-author list of publications athttp://www3.sympatico.ca/colin.kydd.campbell/ckcpub.htm
NOTE: Publication #76 in my list
of publications is an Invited Review Paper in the October 1989 Proceedings
of the IEEE, entitled "Applications of Surface Acoustic and Shallow
Bulk Acoustic Wave Devices." This Review paper includes 322 References.
(One of my SAW illustrations in that paper was also used as the front cover
design for that Proceedings issue.) This paper may now be downloaded from
the IEEE web site for Ultrasonics, Ferroelectrics, and Frequency
Control, at Internet address:
My email address is firstname.lastname@example.org
Some - but not all - of the above SAW topics were discussed in previous Sessions on this web-page site. These were:
Session 1: "How Many SAW Devices Can
Be Used In a Typical AMPS Mobile Transceiver ?"
Session 2: "Using a Leaky-SAW Differential Mode Resonator Filter in Conjunction with a Differential
Active Mixer in the Front End of a Low-Power Wireless Receiver."
Session 3: "On the Merits of Using SAW Convolvers For Wireless Communications."
Session 4: "Example of a Fast Frequency-Hopping SAW Oscillator Circuit."
Session 5: "Phase Noise in Surface Acoustic Wave Oscillators."
Session 6: "Leaky-SAW Front-End Ladder Filters and Antenna Duplexers."
Session 7: "SAW Nyquist Filters for Digital Microwave Radio."
Session 8: "SAW Clock-Recovery Filters for Fiber-Optic Data-Communications Networks."
Session 9: "Wideband SAW IF Filters With Slanted Finger IDTs for Satellite Communications."
Session 10: "So Far So Good, But How Do I Design a Basic SAW IF Filter ?"
Session 11: "Why Would I Want (Or Need) To Use a SAW Filter Operating In A Harmonic Mode?"
I have saved copies of each of the
above web-page Sessions as MS
97 Word documents. Let me know if you would like me to e-mail
you any of the above session files.
Copyright © Colin Campbell, 2008
In December 2004 this "hit" counter
was reset to zero after a count of 38,000