Content uploaded by P.J. Fonseca
Author content
All content in this area was uploaded by P.J. Fonseca on Sep 16, 2016
Content may be subject to copyright.
REVIEW OF SCIENTIFIC INSTRUMENTS 83, 055007 (2012)
A new concept in underwater high fidelity low frequency sound generation
Paulo J. Fonseca and J. Maia Alvesa)
Faculdade de Ciências Universidade de Lisboa, Centro de Biologia Ambiental and SESUL,
Lisboa 1749-016, Portugal
(Received 19 November 2011; accepted 28 April 2012; published online 18 May 2012)
This article reports on a new type of system for high fidelity underwater sound generation (patent
pending PT105474). The system includes an underwater sound actuator and the corresponding elec-
tronic driver. The sound is generated by a rigid plate that is actuated (both for positioning/dumping
and excitation) using purely electromagnetic forces, thus, avoiding the use of any elastic membrane.
Since there is no compressible air inside the device, which is flooded by water, the operation of
this device is independent from depth, broadening its applications to any water pressure. Charac-
terization of the frequency response, the radiation characteristics, and the dynamic range of this
new device for underwater sound generation is presented. © 2012 American Institute of Physics.
[http://dx.doi.org/10.1063/1.4717680]
I. INTRODUCTION
Many fish communicate with acoustic signals mostly
in mating and agonistic contexts.1–5These signals are often
pulsed low frequency sounds with fast transients.6,7Playback
experiments, a widespread tool to study the function of an-
imals’ acoustic signals,8have been hampered in fish due to
limitations of commercially available underwater loudspeak-
ers, which do not reproduce fish sounds appropriately.
Most devices for underwater sound playback use either
(i) a moving coil associated to a diaphragm, as in traditional
loudspeakers; (ii) a piezoelectric effect, which is appropriate
to produce medium to high frequency sounds above 1 kHz,9
or (iii) a system that incorporates both a diaphragm and an
axial deformation mechanism.10
Commercial acoustic devices allowing low frequency
sound playbacks in water are relatively scarce. Underwa-
ter loudspeakers developed for swimming pools, such as the
UW30 from Lubell Labs (one of the most used in under-
water playback experiments) or the Clark Synthesis AQ339
Aquasonic Underwater Speaker, although suitable to play-
back music, do not represent appropriately low frequency
sounds with fast transients, such as the sounds of many fish
species. Although these devices can generate low frequency
sounds, usually above a few tens of Hertz, the equilibrium
position of the sound emitter is dependent on elastic compo-
nents, and so prone to resonation at certain frequencies as can
be clearly seen in Figure 1.
In addition, some of these devices are sealed and incor-
porate air, creating a further constraint caused by increased
pressures due to water depth. This is the main reason why
the UW30, for instance, cannot be used at depths in excess of
3 m. Moreover, the power output of these commercial loud-
speakers is often affected by depth, a further complication for
experiments requiring sound playback in places were the wa-
ter level changes significantly, as in the cases of sea shores and
estuaries. To the best of our knowledge even the higher quality
underwater loudspeakers available today have poor responses
a)jma@fc.ul.pt.
and/or are non-linear at low frequencies (below 100–200 Hz)
and cannot be used at depths beyond 20 m.
Several devices falling in categories (ii) and (iii) were
created to produce underwater sounds, some of them de-
scribed as appropriate to overcome some of the above men-
tioned problems. These include moving coil devices,11,12 or
hydraulic devices13 that use a piston and plate to drive oil that
is circumscribed by an elastic membrane responsible for ra-
diating the sound into water. Since all these devices not only
incorporate several elastic components but also keep some air
inside, it is expected that they are prone to resonances and
may be affected by water pressure, i.e., depth. Other devices
incorporating speakers kept in watertight containers14 are ex-
pected to be not only limited in water depth but also its fre-
quency response is probably affected by the air volume and
by the container itself. Thus, there was a need to develop a
device that could overcome these limitations.
II. DEVICE DESCRIPTION AND CHARACTERIZATION
An expanded schematic representation of a cross section
of the new device for underwater sound generation is pre-
sented in Figure 2.
The sound is generated by a rigid plate attached to a
cylinder that contains a permanent magnet inside. The vibra-
tion of both the cylinder and the plate is accomplished by ap-
plying a variable electrical current to a pair of coils surround-
ing the cylinder, wound in opposite directions (one clockwise
and the other anti-clockwise). A second pair of such coils,
both wound in the same direction, is used for setting the equi-
librium position of the cylinder. This is accomplished simply
because the magnetic field created by each coil will force the
permanent magnet located inside the cylinder in opposite di-
rections, the equilibrium position being the point along the
axis of vibration where these forces balance out to zero. The
intensity of the direct current applied to this pair of coils is
used to control the dumping effect on the cylinder and plate.
It should be noted that this prototype was developed to meet
the needs for playback experiments with small fishes. For this
0034-6748/2012/83(5)/055007/4/$30.00 © 2012 American Institute of Physics83, 055007-1
Downloaded 18 May 2012 to 194.117.2.51. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
055007-2 P. J. Fonseca and J. M. Alves Rev. Sci. Instrum. 83, 055007 (2012)
Acoustic signal produced by the device
time
100 ms
Acoustic signal produced by the fish
time
100 ms
Acoustic signal produced by UW30
time
100 ms
FIG. 1. Comparison of a fish signal with the playbacks of the same sound obtained using both the new device and a commercial underwater loudspeaker UW30.
reason the dimensions of the device were kept considerably
small. However larger devices should, in principle, perform
even better, because larger discs will be more efficient in low
frequency sound generation.
This unique combination of both electromagnetic exci-
tation and positioning/dumping makes it possible to exclude
any elastic component(s), such as membrane(s), from the ac-
tuator, and thus avoids unwanted resonant modes that char-
acterize all the existing equipment for underwater sound gen-
eration, as can be clearly seen in Figure 1. It should also be
noted that, since there is no compressible air inside the de-
vice, which is completely flooded by water, its operation is
independent from depth, broadening its applications to any
water pressure/depth.
The electronic driver developed to operate the device is
mainly composed by (i) a voltage controlled direct current
source to be applied to the positioning/dumping pair of coils;
and (ii) a transconductance amplifier whose output current is
applied to the excitation pair of coils. In normal use the po-
sitioning/dumping current is of the order of a few tenths of
ampere, and the peek excitation current is of the order of a
few ampere.
A suitable characterization of a sound playback device
(i.e., the actuator and its control electronics) includes its
frequency response, the radiation characteristics, and the
dynamic range. In order to measure these parameters an FFT
based transfer function was computed as the ratio of the cross
spectrum between the input (i.e., the electrical signal input
f
rigid plate
excitation coils
permanent magnet
positioning / damping coils
FIG. 2. Schematic diagram of a cross section of the device for underwater sound generation.
Downloaded 18 May 2012 to 194.117.2.51. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
055007-3 P. J. Fonseca and J. M. Alves Rev. Sci. Instrum. 83, 055007 (2012)
-30
-20
-10
0
10
20
0 500 1000 1500 2000 2500 3000
Frequency (Hz)
Magnitude (dB)
-180
-90
0
90
180
0 500 1000 1500 2000 2500 3000
Frequency (Hz)
Phase angle (degree)
1.7 cm
3 cm
6 cm
12 cm
0
0.2
0.4
0.6
0.8
1
0 500 1000 1500 2000 2500 3000
Frequency (Hz)
Coherence
Magnitude (dB)
Phase angle (deg)
Coherence
Frequency (Hz)
-180
-90
0
90
180
0 500 1000 1500 2000 2500 3000
Frequency (Hz)
Phase angle (degree)
Centered
disc edge
at disc plane
0
0.2
0.4
0.6
0.8
1
0 500 1000 1500 2000 2500 3000
Frequency (Hz)
Coherence
Magnitude (dB)
Phase angle (deg)
Coherence
Frequency (Hz)
(b)(a)
-30
-20
-10
0
10
20
30
FIG. 3. Device characterization: (a) Frequency response as a function of position; (b) frequency response as a function of distance.
to the playback device driver) and the system’s output (i.e.,
the sound produced by the device), to the power spectrum of
the input. For this purpose, the sound produced by the device
was recorded by a reference hydrophone Bruel&Kjaer 8103
(B&K 8103, frequency response 0.1 Hz–180 kHz, sensitivity
−211 dB re 1 V/μPa) connected to the conditioning electron-
ics of a Bruel and Kjaer Sound Level Meter 2238 Mediator
(B&K 2238 Mediator). The input signal was a sine sweep
(0–3000 Hz, 20 ms) allowing the evaluation of the frequency
response within this range. The output of the transfer function
gives both the gain and the phase responses as a function
of frequency, flat functions meaning that the response of
the system (output) is similar to the input signal in gain and
phase across frequency. A zero value of the transfer function
gain (dB) and phase (degree) means no amplitude difference
and no phase shift of the output relative to input, respectively.
Moreover, a coherence function, allowing an estimation of
the portion of the output power spectrum that is related to the
input spectrum was simultaneously computed. The coherence
function is normalized and a value of 1.0 means perfect coher-
ence, which in turn is a powerful statistical indication that the
transfer function result is valid, since the output is correlated
with the input. A value close to zero indicates that the result
of the transfer function at that frequency is not related to the
input and is therefore meaningless. Both functions were sub-
jected to a 25 stimulus presentations averaging. All stimuli
were delivered and recorded simultaneously with a through-
put rate of 100 kHz (National Instruments NI USB-6251
Multifunction I/O board). The recording (output) was delayed
(precision of ±0.005 ms at 100 kHz) relative to the stimulus
(input) to account both for the electronics and the propagation
time interval from the moment the stimulus was generated
to the instant when it was recorded, thus, preventing phase
change artifacts on the measurement. All functions were
computed using a 2048 points FFT based program made in
LabVIEW by the authors. Additionally, the responses at
discrete frequencies were measured by comparing the am-
plitudes of pure tone stimuli recorded both at the input of
the instrument chain (electric signal) and at the output of
the sound emitting device (sound recorded in water with the
hydrophone B&K 8103 connected to B&K 2238 Mediator).
This procedure confirmed the frequency response at very low
frequencies.
As observed in Figure 3(a), the frequency response mea-
sured from the sound recorded in front of the playback device
is not only very smooth and relatively flat (10 Hz–3 KHz,
±3 dB) but its characteristic does not change significantly
with distance, both in magnitude (gain) and phase angles
(Figure 3(b)), that are astonishingly even and close to zero.
The quality of the measurement is attested by the values
of the coherence function that remain close to unity within
the frequency range considered. Notice also the good agree-
ment with single frequency calibration data of two differ-
ent devices, represented by the circles and the triangles in
Figure 3(a).
Since the actuator disc is expected to act as a dipole, mod-
ifications of the sound field are predictable when measure-
ments are obtained progressively away from the axis. Addi-
tionally, the body of the device may interfere itself with the
sound waves. In fact, when recordings are obtained towards
the plane of the disc (Figure 3(a), blue line), the frequency re-
sponse changes not only in magnitude, but especially in phase
angles due to interactions of opposite phase waves generated
in both sides of the disc. The field is however very stable in
the region in front of the disc both in its gain and phase com-
ponents. Again here coherence close to 1.0 attests for the sig-
nificance of the measurements. Therefore, the device presents
a radiation diagram with an even sound field in the volume
in front of the disc, becoming considerably disturbed when
approaching the plane of the actuator’s disc.
In order to test the behavior of the device when reproduc-
ing fast transient sounds, a need for many researchers willing
Downloaded 18 May 2012 to 194.117.2.51. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions
055007-4 P. J. Fonseca and J. M. Alves Rev. Sci. Instrum. 83, 055007 (2012)
-90
-80
-70
-60
-50
-40
-30
-20
0 500 1000 1500 2000 2500
Frequency (Hz)
Amplitude (dB rel)
FishTalk playback
Fish Sound (electrical signal)
20 25 30 35 40 45 50
10 ms
(a)
-90
-80
-70
-60
-50
-40
-30
-20
0 500 1000 1500 2000 2500
Frequency (Hz)
Amplitude (dB rel)
noise
10 ms
(b)
FIG. 4. Device characterization: (a) Comparison of spectra of painted goby
sound (Pomasthoschistus pictus) and its playback with the device (both
recorded with B&K 8103); (b) representation of power spectra of a fish
sound (painted goby, Pomasthoschistus pictus) generated using the device,
and attenuated in 6 dB steps (recorded using a B&K 8103 hydrophone).
These results show that the dynamic range is greater than 36 dB. Notice that
the playback amplitude was not increased further because, at the represented
maximum, it was already above the amplitude generated by the fish.
to make playback experiments with fish, we used a pulse of
an agonistic sound of the fish painted goby (Pomasthoschistus
pictus). The device reproduced with very high fidelity the
fish sound wave as shown in Figure 4(a). The superimposed
oscillograms of the fish sound and the device playback (inset
in Figure 4(a)) are very similar, as are the power spectra
computed from a sequence of these signals.
Similar measurements were made to assess the dynamic
range of the device. As depicted in Figure 4(b) the power
spectra of recordings of the same sound reproduced with am-
plitudes decreasing by 6 dB steps shows a dynamic range in
excess of 36 dB and in the whole range down to the noise
level. In fact, the power spectra closely follow the sound am-
plitude decreases by the same 6 dB steps (cf. Figure 4(b)).
Notice that in these measurements the playback amplitude
was not increased further because, at the represented maxi-
mum, it was already largely above the amplitude of the sound
generated by the fish. It should be emphasized that a high dy-
namic range is desirable since it guarantees a proper represen-
tation of elements with different amplitudes within a sound
sequence.
III. CONCLUSIONS
A new concept in underwater sound generation was de-
scribed. The device presented does not keep any air inside,
and thus its operation is water depth independent. The equi-
librium positioning, excitation and dumping of the rigid plate
used to generate the sound is purely magnetically actuated.
Consequently, the resonances are almost absent, allowing low
frequency sounds to be reproduced with very high accuracy
from about 10 Hz up to 3 kHz, even when they present fast
transients as is typical of many fish communication sounds.
We believe that this device overcomes most of the limita-
tions of available emitters of underwater sound, and for this
reason may become an important tool in studies of fish be-
havior or other applications in deep water where low fre-
quency signal playback, namely of complex sounds, may be
requested.
ACKNOWLEDGMENTS
This research was funded by the Science and Technology
Foundation, Portugal (Project PDCT/MAR/68868/2006) and
pluriannual program UI&D 329.
1F. Ladich, S. P. Collin, P. Moller, and B. G. Kapoor, Communication in
Fish es (Science, Enfield, New Hampshire, 2006).
2R. O. Vasconcelos, J. M. Simões, V. C. Almada, P. J. Fonseca, and M. C. P.
Amorim, Ethology 116, 155–165 (2010).
3M. C. P. Amorim, J. M. Simões, V. Almada, and P. J. Fonseca, Behav. Ecol.
Sociobiol. 65(4), 707–716 (2011).
4M. C. P. Amorim and A. D. Hawkins, J. Fish Biol. 57, 895–907 (2000).
5A. D. Hawkins and M. C. P. Amorim, Environ. Biol. Fish. 59, 29–41
(2000).
6M. C. P. Amorim and A. S. M. Neves, Behaviour 145, 1065–1083 (2008).
7J. M. Simões, I. Duarte, P. J. Fonseca, G. F. Turner, and M. C. P. Amorim,
J. Acoust. Soc. Am. 124(2), 1332–1338 (2008).
8K. Gerhardt and F. Huber, Acoustic Communication in Insects and Anurans
(Chicago University Press, 2002).
9J. N. Kim and M. E. Stewart, “Sound and light emitting fish lure and
method of attracting fish,” U.S. patent 6,098,331, 8 August 2000.
10S. Hiroshi, “Bent type wave transmitter-receiver,” patent JP2001333487,
30 November 2001.
11S. Fumio, K. Kiichiro, and S. Mitsuharu, “Sealless mechanism for under-
water sound source,” patent JP8256394, 1 October 1996.
12S. Masaaki, “Sound-reproducing apparatus,” patent JP2005167565,
23 June 2005.
13S. Fumio, K. Hitoshi, and K. Kiichiro, “Hydraulically driven underwater
sound source device,” patent JP8009487, 12 January 1996.
14S. A. Stepanovich, L. O. Viktorovich, and P. S. Ivanovich, “Deepwater
low-frequency hydroacoustic emitter for acoustic tomography systems,”
patent RU2008112652, 20 December 2009.
Downloaded 18 May 2012 to 194.117.2.51. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions