Академический Документы
Профессиональный Документы
Культура Документы
net/publication/7520302
CITATIONS READS
13 219
4 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Robert X. Gao on 12 December 2013.
I. Introduction
In comparison, wireless sensors can be installed in
or process monitoring of injection molding, pressure
F measured within the mold cavity has been found to
be a more accurate indicator of product quality than pres-
tapped holes beneath the surface of the mold cavity
(Fig. 2), requiring much less modifications to the mold
structure. Therefore, wireless sensing is physically less in-
sures measured elsewhere, e.g., along the hydraulic lines vasive and economically more advantageous. Because of
or at the injection nozzle [1], [2]. Common practice in the their small sizes (e.g., 16 mm × 28 mm), more sensors can
industry uses a single cavity pressure sensor installed near be installed per mold as compared to the wired sensors,
the injection gate next to the mold cavity, and the per- leading to a more comprehensive spatial coverage of the
formance of the optimal molding parameters is tracked by pressure profile within the mold, thus improving product
the pressure measured at this position. However, a sensor quality control.
matrix arrangement often is desirable to measure pressures
Compared to the radio frequency (RF) wave that com-
at other strategically important locations, e.g., in the mid-
monly has been used as the information carrier for wireless
dle and at the end of the melt flow along the mold cavity,
communication, ultrasound wave is acoustic in nature and
to obtain a comprehensive pressure profile and gain in-
does not suffer from the electromagnetic shielding effect
sight into the flow pattern, especially for parts with long
prevalent in an injection mold environment, due to the
flow distances and complicated geometries. However, for
surrounding steels. Thus, it provides a better-suited alter-
conventional pressure sensors to be installed in an injec-
native for transmitting the pressure data out of the mold
tion mold cavity, the mold structure has to be modified to
cavity, as is experimentally verified [3]. Research has re-
provide a pathway for wire connections. Due to the com-
sulted in a wireless, pressure-sensor prototype that consists
plexity of the mold structure (Fig. 1) involving ejection
of three major components: an energy converter, a thresh-
holes, cooling lines, clamping plates, etc., the number of
old modulator, and an ultrasonic transmitter [4]. Through
wired sensors that can be installed in a mold is severely
the threshold modulator, electrical charges generated by
limited.
the energy converter due to the polymer melt pressure ex-
Manuscript received May 14, 2004; accepted January 21, 2005. The ceeding the preset threshold values are applied to the ultra-
authors gratefully acknowledge funding provided to this research by sonic transmitter. The transmitter in turn generates a se-
the National Science Foundation under award DMI-9988757. ries of ultrasound pulses, with each pulse corresponding to
L. Zhang and R. X. Gao are with the Department of Mechanical
and Industrial Engineering, University of Massachusetts, Amherst, a specific pressure threshold value. Reconstruction of the
MA 01003 (e-mail: gao@ecs.umass.edu). pressure is achieved by multiplying the number of pulses
C. B. Theurer is with the Pervasive Decisioning Systems Labora- received at an ultrasound receiver outside of the mold with
tory, GE Global Research, Niskayuna, NY 12309.
D. O. Kazmer is with the Department of Plastics Engineering, the respective pressure threshold values. This sensing prin-
University of Massachusetts, Lowell, MA 01854. ciple is illustrated in Fig. 3, where curve 1 represents the
0885–3010/$20.00
c 2005 IEEE
Authorized licensed use limited to: UNIV OF MASS-LOWELL. Downloaded on June 08,2010 at 15:24:15 UTC from IEEE Xplore. Restrictions apply.
zhang et al.: new mechanical wireless data transmission technique 1361
(a)
Authorized licensed use limited to: UNIV OF MASS-LOWELL. Downloaded on June 08,2010 at 15:24:15 UTC from IEEE Xplore. Restrictions apply.
1362 ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 52, no. 8, august 2005
Authorized licensed use limited to: UNIV OF MASS-LOWELL. Downloaded on June 08,2010 at 15:24:15 UTC from IEEE Xplore. Restrictions apply.
zhang et al.: new mechanical wireless data transmission technique 1363
TABLE I
List of Symbols
A0 B0 1 1 jφ2 /(2πf C0 ) cos γ0 + jzb sin γ0 Z0 (zb cos γ0 + j sin γ0 )
= , (7)
C0 D0 φQ j2πf C0 0 (j sin γ0 )/Z0 2(cos γ0 − 1) + jzb sin γ0
Authorized licensed use limited to: UNIV OF MASS-LOWELL. Downloaded on June 08,2010 at 15:24:15 UTC from IEEE Xplore. Restrictions apply.
1364 ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 52, no. 8, august 2005
of the input voltage (Fn−1 ) and current (un−1 ), and their shown in (13) (see next page), where Zl = Rl + jXsl is the
relationship was derived in the matrix form as: electrical impedance of the voltage measuring instrument
(e.g., an oscilloscope) connected to the ultrasound receiver
Fn−1 An Bn Fn and Zs = Rs + jXs represents the electrical impedance of
= . (8)
un−1 Cn Dn un the signal generation source (i.e., the threshold modulator
within the pressure sensor, as shown in Fig. 2) that excites
The transformation matrix is given as:
the transmitter. By taking the inverse Fourier transform of
An Bn cos γn jZn sin γn (13), the waveform of the ultrasonic signals arriving at the
= , n = 1, 2, and 3, ultrasound receiver could be obtained. The voltage trans-
Cn Dn (j sin γn )/Zn cos γn
(9) fer function expressed in decibel, i.e., the insertion gain
(IG), was determined as:
where the symbols Zn and γn represent the mechanical
impedance and normalized frequency of the corresponding Vo Rs
layer, respectively. They are obtained by replacing the sub- IG = 20 log + 20 log +1 . (14)
Vi Rl
script 0 with the corresponding layer subscript 1, 2, or 3, as
defined in (1)–(3). Combining the models for the piezoelec- The insertion gain is a function of the frequency and
tric layer and all nonpiezoelectric layers, the input-output characterizes the frequency response of the ultrasonic sig-
function for the entire transmitter system was obtained, nal transmitter. It was used in the present study as a per-
as described in the next section. formance measure for the design of individual ultrasonic
transmitters in a multiple-sensor, matrix configuration.
3. Incorporation of Layers: To model the overall fre-
quency behavior of the physically interconnected multiple
layers of the transmitter, their respective circuit models III. Frequency Behavior of the Layers
were modeled as a serial connection with each other. Mul-
tiplying the matrices for the various layers, the overall ma- For a transmitter with an n-layer structure, n resonant
trix that relates the excitation voltage Vi and current Ii modes exist, resulting in n resonant frequencies [12]. Ac-
(input to the piezoelectric layer) to the sound force F3 and cordingly, for the presented transmitter, consisting mainly
velocity u3 (output) to the mold steel was obtained as: of a piezoelectric layer, a bonding layer, and a front layer,
three major resonant frequencies can be found, denoted as
3 f1 , f2 , and f3 , respectively.
At Bt An Bn
= , (10) The existence of multiple resonant frequencies in a sin-
Ct Dt Cn Dn
n=0 gle transmitter is undesirable, when individual sensors in
and: a sensor matrix need to be distinguished. To enable re-
liable identification, it is required that the transmitter
Vi At Bt F3 within each sensor possess a unique and dominant resonant
= . (11)
Ii Ct Dt u3 frequency with nonoverlapping bandwidth (−6 dB band-
width) such that ultrasonic pulses received by the receiver
The input impedance of the entire transmitter under clearly can be associated with each transmitter. For a typ-
installed conditions, which determines the resonant fre- ical injection mold, four to six sensors may need to be in-
quencies of the transmitter, was derived as: stalled within the mold cavity to provide a comprehensive
At Z0t + Bt profile of the pressure variations during an injection mold-
Zi = , (12) ing process. Considering that most commercially available
Ct Z0t + Dt
broadband ultrasonic receivers cover a limited frequency
where Z0t is the mechanical impedance of the mold steel. bandwidth of about 2 MHz, the bandwidth of each trans-
To evaluate the modeling results, an ultrasonic receiver mitter needs to be designed as narrow as possible (e.g.,
that receives the ultrasonic pulses from the transmitter ≤ 300 kHz) such that all the transmitters can be accom-
and converts them into an electrical voltage signal needs modated by a single receiver, thus simplifying the hard-
to be modeled as well. Similar to the transmitter modeling, ware requirement of the measurement system. Given this
the governing matrices for the receiver (Ar , Br , Cr , and requirement, a distinctive dominant resonant frequency
Dr ) were derived using (6)–(11), in which the input and and a narrow bandwidth are the main performance mea-
output terms in (11) were interchanged such that the input sures for the transmitter design. In addition, high energy
for the receiver was the sound force F3 , and the output was efficiency, characterized by high magnitude for the inser-
the electrical voltage Vo . For the present study, a reciprocal tion gain, is needed to allow for maximum electrical-to-
approach was taken, by which the receiver and transmitter mechanical energy conversion and, consequently, transmis-
were assumed to have the same configuration and, there- sion power. To achieve such a design goal, the effect of the
fore, the same matrices. Using the transformation matrices front, bonding, and coupling layers on the overall transmit-
for the transmitter and receiver, the transfer function re- ter frequency response was investigated, using the equiv-
lating the input voltage Vi to the transmitter and the out- alent circuit model. In Table II, material properties used
put voltage Vo converted by the receiver was determined as for the study are illustrated.
Authorized licensed use limited to: UNIV OF MASS-LOWELL. Downloaded on June 08,2010 at 15:24:15 UTC from IEEE Xplore. Restrictions apply.
zhang et al.: new mechanical wireless data transmission technique 1365
Vo 2Rl Z0t
= , (13)
Vi [At Zl + Bt + Zs (Ct Z0t + Dt )] [Ar Zl + Br + Zl (Cr Z0t + Dr )]
TABLE II
Properties of Materials Used in the Design.
Longitudinal
Layer materials sound velocity Density
Piezoelectric (PKI502) 4,559 m/s 7.7 g/cm3
Front (Stainless Steel) 5,660 m/s 8.0 g/cm3
Bonding (E-Solder 3022) 2,110 m/s 2.25 g/cm3
Coupling (Grease) 1,200 m/s 0.95 g/cm3
Coupling
Permittivity factor
Piezoelectric (PKI502) 7.35 × 10−9 F/m 0.47
where the symbols v1L and v2L represent the longitudinal (b)
velocities of the ultrasound in the bonding and front layers, Fig. 8. Effect of front layer thickness on transmitter insertion gain
respectively. and bandwidth. (a) Magnitude of insertion gains as a function of the
front layer thickness, (b) bandwidth of insertion gains as a function
A. Selection of Front Layer Thickness of the front layer thickness.
Authorized licensed use limited to: UNIV OF MASS-LOWELL. Downloaded on June 08,2010 at 15:24:15 UTC from IEEE Xplore. Restrictions apply.
1366 ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 52, no. 8, august 2005
(a)
Fig. 9. Insertion gains for transmitters with 0.1 λf /4 and 2.0 λf /4
front layer thickness.
Authorized licensed use limited to: UNIV OF MASS-LOWELL. Downloaded on June 08,2010 at 15:24:15 UTC from IEEE Xplore. Restrictions apply.
zhang et al.: new mechanical wireless data transmission technique 1367
Fig. 11. Insertion gains for transmitters with various bonding layer
Fig. 12. Investigation of the effect of coupling layer thickness.
thickness.
was increased from 0.1 to 0.5 λb /4, the bandwidth grew ation in the range of 4–10 µm. To investigate the effect of
from 144 kHz to 246 kHz. It then remained essentially con- the coupling layer on the frequency response of a trans-
stant until the bonding layer thickness reached 1.0 λb /4, mitter that incorporates the optimal design (i.e., 2.0 λf /4
after which it showed a monotonic increase. As a bonding front layer and 1.0 λb /4 bonding layer thickness), numer-
layer thickness in the range of 0.1 to 0.5 λb /4 does not ical simulations were conducted, for which the coupling
satisfy the ≥ 20 dB requirement, and the bandwidth for layer thickness was varied from 4 µm to 10 µm, at a step
a layer thickness from 0.6 to 1.0 λb /4 remains essentially of 2 µm.
unchanged, the bonding layer thickness that enables the As shown in Fig. 12, increasing the coupling layer thick-
largest magnitude difference between f1 and f2 was cho- ness has led to a decrease of the f2 mode magnitude (e.g.,
sen as the optimal thickness. As shown in Fig. 10(b), this 5 dB reduction as the coupling layer thickness increases
thickness is 1.0 λb /4, at which the difference between the from 4 µm to 10 µm), thus reducing the energy efficiency
f1 and f2 modes is 24.6 dB, the bandwidth of the f2 mode of the transmitter. To minimize energy losses, the rough-
is 246 kHz, and the amplitude is −3.5 dB. In Fig. 11, ness of both the transmitter and mold surfaces needs to be
the effect of the bonding layer thickness of 0.1, 0.5, and minimized, e.g., by means of polishing, which could reduce
1.0 λb /4 on the insertion gains and the bandwidth is fur- the surface roughness down to below 0.5 µm, as has been
ther illustrated. The largest magnitude difference between practiced in the industry. Furthermore, as the coupling
the f2 and f1 modes is seen at the thickness of 1.0 λb /4. layer thickness varied from 4 µm to 10 µm, the location of
Based on the above analysis, the combination of the f2 center frequency has shifted toward lower frequency
2.0 λf /4 and 1.0 λb /4 presents the optimal thickness for range by about 20 kHz, while the bandwidth also has nar-
the front layer and bonding layer, respectively. When ex- rowed by about 30%. This variation needs to be considered
cited, a transmitter with these layer thickness values will when assigning the frequency band to the transmitters in
generate ultrasonic pulses centered at its f2 mode with a a sensor matrix configuration, to minimize frequency band
relatively large amplitude and narrow bandwidth. Such overlaps.
a transmitter, as a constituent component of a mold-
embedded pressure sensor, will facilitate signal distinction
in a multiple sensor matrix, which enables in-mold cavity IV. Frequency Band Assignment
pressure measurement.
To investigate the performance of the proposed multiple
C. Sensitivity to Coupling Layer Thickness sensor arrangement, a simulation was conducted on the
frequency band assignment, using transmitters with the
When installed into the steel mold, an air gap exists optimal layer thickness (2.0 λf /4 front layer and 1.0 λb /4
between the transmitter and the mold steel. To minimize bonding layer). The goal was to determine the maximum
the energy loss due to impedance mismatch between the number of transmitters that can be accommodated within
air (411 kg/m2 ·s) and steel (45.4×106 kg/m2 ·s), a coupling a fixed frequency range. Realistically, a frequency range
layer consisting of grease generally is needed. The thick- higher than 500 kHz was needed to minimize the influence
ness of the coupling layer can be difficult to control ac- of background noise in an injection mold machine [20]. For
curately in practical applications, due to the paste nature the present work, a frequency band of 2–4 MHz was chosen,
of the grease. Realistically, the thickness is determined by based on the frequency characteristics of the piezoelectric
the roughness of the mold and transmitter surfaces, which elements used for the transmitters and the fact that such
ranges from 2 µm (for a ground surface) to about 5 µm a bandwidth is readily covered by commercially available
(for a milled surface). This leads to a layer thickness vari- receivers.
Authorized licensed use limited to: UNIV OF MASS-LOWELL. Downloaded on June 08,2010 at 15:24:15 UTC from IEEE Xplore. Restrictions apply.
1368 ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 52, no. 8, august 2005
Fig. 13. Insertion gains of the selected six transmitters with nonover- (a)
lapping bandwidth.
TABLE III
Design Parameters for Six Transmitters with
Nonoverlapping Bandwidths.
Authorized licensed use limited to: UNIV OF MASS-LOWELL. Downloaded on June 08,2010 at 15:24:15 UTC from IEEE Xplore. Restrictions apply.
zhang et al.: new mechanical wireless data transmission technique 1369
(a)
(b)
TABLE IV Fig. 16. Measured and simulated ultrasonic pulses in the time and
frequency domains for transmitters with different front layer thick-
Comparison Between Simulated and Measured Transmitter ness: (a) 1.0 λf /4, (b) 1.6 λf /4, and (c) 2.0 λf /4.
Resonant Frequencies.
Authorized licensed use limited to: UNIV OF MASS-LOWELL. Downloaded on June 08,2010 at 15:24:15 UTC from IEEE Xplore. Restrictions apply.
1370 ieee transactions on ultrasonics, ferroelectrics, and frequency control, vol. 52, no. 8, august 2005
Authorized licensed use limited to: UNIV OF MASS-LOWELL. Downloaded on June 08,2010 at 15:24:15 UTC from IEEE Xplore. Restrictions apply.
zhang et al.: new mechanical wireless data transmission technique 1371
[10] C. S. Desilets, J. D. Fraser, and G. S. Kino, “The design of effi- Robert X. Gao (M’91–SM’00) received his
cient broad-band piezoelectric transducer,” IEEE Trans. Sonics M.S. and Ph.D. degrees from the Technical
Ultrason., vol. SU-25, no. 3, pp. 115–125, 1978. University of Berlin, Germany, in 1985 and
[11] J. H. Goll, “The design of broad-band fluid-loaded ultrasonic 1991, respectively. He is currently an associate
transducers,” IEEE Trans. Sonics Ultrason., vol. SU-26, no. 6, professor with the Department of Mechanical
pp. 385–393, 1979. and Industrial Engineering at the University
[12] T. Inoue, M. Ohta, and S. Takahashi, “Design of ultrasonic of Massachusetts, Amherst, MA.
transducers with multiple acoustic matching layers for medical His research focuses on sensing methodolo-
application,” IEEE Trans. Ultrason., Ferroelect., Freq. Contr., gies for machine health monitoring, diagno-
vol. 34, no. 1, pp. 8–16, 1987. sis and prognosis, sensor networks, multido-
[13] G. Kossoff, “The effects of backing and matching on the perfor- main signal processing, and wireless commu-
mance of piezoelectric ceramic transducers,” IEEE Trans. Son- nication.
ics Ultrason., vol. SU-13, no. 1, pp. 20–31, 1966. Dr. Gao received the 1996 National Science Foundation CAREER
[14] E. K. Sittig, “Effects of bonding and electrode layers on the Award and the 2000 University of Massachusetts Outstanding Engi-
transmission parameters of piezoelectric transducers used in ul- neering Junior Faculty Award. He was the Guest Editor for a special
trasonic digital delay lines,” IEEE Trans. Sonics Ultrason., vol. section on sensors for the Journal of Dynamic Systems, Measure-
SU-16, no. 1, pp. 2–10, 1969. ment, and Control, published by the American Society of Mechanical
[15] B. Carlin, Ultrasonics. New York: McGraw-Hill, 1960. Engineers in June 2004. Currently he serves as an Associate Editor
[16] L. L. Beranek, Acoustics. New York: McGraw-Hill, 1954. for the IEEE Transactions on Instrumentation and Measurement,
[17] W. P. Mason, Electro-Mechanical Transducers and Wave Fil- and chairs the Technical Committee on Built-in-Test and Self-Test
ters. New York: D. Van Nostrand, 1948. of the IEEE Instrumentation and Measurement Society.
[18] S. Sherrit and B. K. Mukherjee, “The use of complex coeffi-
cients to model piezoelectric materials,” in Proc. IEEE Ultrason.
Symp., 1998, pp. 633–640.
[19] M. Redwood, “Transient performance of a piezoelectric trans-
ducer,” J. Acoust. Soc. Amer., vol. 33, pp. 527–536, 1961. David O. Kazmer, P.E., is a graduate
[20] C. Theurer, L. Zhang, D. Kazmer, and R. Gao, “Acoustic teleme- of Cornell University, Ithaca, NY (B.S.,
try in injection molding,” in Proc. Soc. Plastics Eng. Annu. 1989) and Stanford University, Stanford, CA,
Tech. Conf. Process Monitoring Contr. Division, 2001, pp. 208– (Ph.D., 1995). He is currently an associate
213. professor with the Department of Plastics En-
gineering at the University of Massachusetts,
Lowell, MA, where he conducts research per-
taining to plastics product and process devel-
opment specializing in design of machinery,
Li Zhang received his M.S. degree from the instrumentation, and control systems.
University of Electronic Science and Technol- Dr. Kazmer received the 1998 National
Science Foundation CAREER Award, 1998
ogy of China, Sichuan, China, in 2000 and
the Ph.D. degree from the University of Mas- Young Investigator Award from the Office of Naval Research, and
sachusetts, Amherst, MA, (UMass) in 2004. the 2000 University of Massachusetts Outstanding Engineering Ju-
He is currently a Post-Doctoral research as- nior Faculty Award. He currently serves as an associate editor for
sociate at the Department of Mechanical and the American Society of Mechanical Engineers Journal of Mechan-
Industrial Engineering of UMass, working on ical Design, and is Chair of the American Society of Mechanical
a collaborative research project with the Na- Engineers (ASME) Technical Committee for Design for Manufac-
turing.
tional Institute of Standards and Technology
(NIST).
His research interests include electrome-
chanical systems design, smart sensing systems, machine condition
monitoring and health diagnosis, and mechanical signal processing.
Authorized licensed use limited to: UNIV OF MASS-LOWELL. Downloaded on June 08,2010 at 15:24:15 UTC from IEEE Xplore. Restrictions apply.
View publication stats