READERS PROJECTS

Ultrasonic Directive Speaker

50+ piezo transducers generate
audible sound beam
By Kazunori Miura (Japan)

The acronym LRAD sadly

and worryingly made it
to newspaper stories in
connection with Somali
pirate attacks on vessels in
the Gulf of Aden, and that is how the general public
first got word of sound being used as a defence
weapon. This article explores the design of a DIY
50-metre range directive sound beam using an array of
piezo ultrasonic transducers (up to 200!) driven with a
PWM signal, all strictly experimental.

Spurred by the success of their Long Range Parametric speaker arrays typically employ practice, 14 kHz is commonly found to be
Acoustic Device® (LRAD) systems, American ultrasonic waves, the same as used in car the real limit at least for adults.
Technology Corporation changed its name parking ‘radars’, distance meters, metal So how can humans perceive a supersonic
to LRAD Corporation on March 25, 2010 [1]. analyzers, etc. However it was not until sound wave? Several methods have been
For non military applications, Audio Spot- recently that approaching a real parametric devised to convert a supersonic wave into
light® is a product of Holosonic Research speaker is possible using commonly avail- a sound wave you can hear. One method
Labs, Inc. [2]. Audio Spotlight produces a able components. is to passively obtain an audible frequency
very sharp sound beam and has found appli- from two supersonic wave sources with
cations in museums, exhibits and galleries. Principle of the parametric a slightly different frequency. For exam-
Those who hear sound from a parametric speaker ple, an undulating 1 kHz tone is obtained
speaker for the first time are typically sur- A parametric speaker achieves high directiv- from two supersonic waves of 40 kHz and
prised and sometimes frightened by the ity thanks to the almost line-of-sight prop- 41 kHz. As illustrated in Figure 1, where two
effect. Sounds appear to be heard from agation of sound waves in the supersonic supersonic waves intersect, a sound within
extremely nearby, although the person range. Supersonic is often loosely defined the audible domain is perceived. The disad-
standing right beside you does not hear as ‘above 20 kHz’ because it exceeds the vantage of this method is that only weak
anything. upper frequency limit of human hearing. In audible sounds are produced, by no means

Note. Readers’ Projects are reproduced based on information supplied by the author(s) only.
The use of Elektor style schematics and other illustrations in this article does not imply the project having passed Elektor Labs for replication to verify claimed operation.

56 03-2011 elektor
 READERS PROJECTS

U/S #1
F1
U/S #2
F2
Compression

(fast)

enough to stun or incapacitate people like

the LRAD.
Other ways of producing an audible sound Rearranging
from supersonic waves include amplitude
100442 - 11 (slow) 100442 - 12
modulation (AM), double sideband mod-
ulation (DSB), single sideband modula-
tion (SSB), frequency modulation (FM) all Figure 1. Where ultrasonic waves from Figure 2. Shock waves come about by
employ the recently developed parametric sources with frequencies F1 and F2 air molecules on their way back to their
speaker system. intersect, audible sounds amounting to F3 original position colliding with other
Inevitably, a 110 dB+ supersonic wave will = |F1–F2| may be heard. molecules being compressed at the same
be irregular in terms of sound pressure dis- time by a sound wave.
tribution as it propagates through a long
air mass, and an audible sound seems to
appear by itself owing to these non-linear of distortion, which is undesirable for ‘nar- processing using DSPs to reduce distortion
characteristics. As a result, the audible rowcasting’ applications like in a museum. to a minimum, often in combination with
sound perceived is marked by a fair amount Manufacturers typically resort to signal a highly sophisticated parametric speaker

+24V

V+
+12V J1
EXT
D1

1N4007 Q1

+12V 1 8
R4
47R
VCC VB 7 C2
HO
IRF540
R8 IC2
12 2 6 100n
47R IN VS A
VCC
IR2111 Q2
16 11
2IN+ C2 4
15 10 LO R5
2IN- IC1 E2 COM GND1
1 8 47R
1IN+ C1 3
K2 2 9
C3 1IN- E1 IRF540
4 6
DTC RT
100n
14
VREF
TL494 CT
5 R2

3 13
CIN OC 20k
D2
GND
7
R1 R3 1N4007 Q3
C8 C1
R7
1k5

1 8
47R
3n3 VCC VB 7 C5 GND
100n 20k HO
IRF540
R9 IC3
+24V 2 6 100n
47R IN VS B
IR2111 Q4
C4
4
LO R6
1000u COM
47R
IC4 3
D3 7812 +12V IRF540

K1
1N4007
C14 C6

100n 47u
20...24VDC
100442 - 13

Figure 3. Circuit diagram of the PWM power driver for the ultrasonic parametric speaker unit.
The audio input signal is connected to jack socket K2. Channel B is optional.

elektor 03-2011 57
READERS PROJECTS

lator is given in Figure 3. There are no spe-

cial components. The TL494 PWM control
circuit and the IR2111 half bridge MOSFET
driver are used in their standard application
circuits. The TL494 has an internal oscilla-
tor whose frequency is deter-
mined by trimpot R2 and
capacitor C1. The basic
pulsewidth is adjusted
with R1. You need to set
up optimum modula-
tion with trimpots
R1 and R2.
The audio input sig-
J J
nal is connected to K2
TR
(loudspeaker level required,
not microphone or line). The board
has two outputs, A and B, each driv-
100442 - 14
ing an array of piezo transducers, option-
ally through an inductor (see below). Each
channel is suitable for up to 200 transduc-
Figure 4. The ‘mini’ version of the directive Figure 5. A radio ferrite rod makes a good ers. The normal supply voltage is 20-24 DVC
loudspeaker consists of 50 ultrasonic piezo core for winding a 60-160 μH adjustable
to K1. The FET stages may be powered by an
transducers connected in parallel and inductor that enables the largely capacitive
external supply via the EXT terminal after
in phase. The step-up transformer and transducers to be peaked at their resonant
removing wire link J1. Heatsinks may be
associated resonance caps are optional. frequency (usually, 40 kHz).
The associated PWM driver in its basic form required on the IRF540 FETs depending on
is suitable for up to 200 transducers. the supply voltage and the transducers’ rat-
ings (up to 60 VDC may be possible).
The U/S speaker schematic is large but
unsurprising, see Figure 4. It represents
setup. in numbers, meaning the high directivity is one channel and a ‘mini’ version with just
The ‘non linear characteristic’ is due to the due to many small transducers arranged in a 50 transducers.
fact that it takes more time for air molecules plane-like shape. This is essential for making
to be restored to their original density than a truly directional speaker unit. Speaker unit and optional coil
to be compressed (Figure 2). When the There are several type of ultrasonic trans-
sound pressure is high, and frequency too, A parametric 2-channel speaker ducer around. The author used 16 mm
a shock wave may be produced by returning modulator diameter devices specified for 40 kHz and
air molecules colliding with the ones being Double sideband modulation (DSB) is eas- 28 kHz. A minimum of 50 transducers
compressed. In fact, an audible sound is ily implemented using analogue switches. is required to make an effective speaker
produced by any molecule not completely Frequency Modulation (FM) has the same unit. You need more than 100 transducers
‘returning’. When the frequency of the effects basically if you look at the way if you want to the unit to have any sort of
vibration rises, the ‘non-linear characteris- supersonic sound waves compress air and range outdoors. All transducers should be
tic” tends to become noticeable by an effect interact. carefully distributed to maintain phase.
best described as ‘air viscosity’. The author first attempted a DSB modu- Remember, the wavelength is about 8 mm
There is another reason for the high direc- lator. The result: big sound, lots of distor- so a positioning error of 1 mm causes phase
tivity (i.e. small beamwidth) exhibited by a tion and the method might be suitable for a errors and loss of SPL.
parametric speaker array. The supersonic sound beam weapon. Next, a PWM system Ultrasonic transducers are made from pie-
wave is actually generated by a large num- was built. Looking at the net result, PWM zoelectric ceramic materials. When a volt-
ber of small loudspeakers called transduc- is very similar to FM. The audible sound age is applied to the device, a special type of
ers. The piezo-electric transducer is widely obtained from PWM is weaker than from foil is deformed inside, generating a super-
used both as a sensor and a transmitting DSB, but of a better quality. A PWM modu- sonic sound wave of a specific frequency.
device in car and home automatic systems. lator may be compared to a class-D ampli- Typically, the transducer’s sound output
The directivity of the piezo transducer by fier without its low pass filter! reaches 105–120 dB (at 30 cm distance)
itself is not too high. However, strength is The schematic of a 2-channel PWM modu- when a voltage of 10–20 Vrms is applied,

58 03-2011 elektor
 READERS PROJECTS

Caution — Health Hazard.

Appropriate measures must
be taken to prevent long term
exposure to high ultrasonic
sound levels.

[1] www.lradx.com
[2] www.holosonics.com

CH1 CH2

1 1 1
correspond- ducer version of the U/S speaker
ing to 28–56 V pp about 55 turns of wire gave best results 40 kHz
and that raises the ques- (60-80 μH). TX RX
U/S U/S
tion if a similarly high supply The ultrasonic transducers need to be
voltage is needed. checked individually to determine their
Electrically, an ultrasonic transducer has polarity (phase). This may be done using
the properties of a capacitor, which can be an oscillator and a 2-channel oscilloscope
100442 - 15
made part of a series resonant circuit by as illustrated in Figure 6. One U/S device
putting an inductor in series. Tuning the is connected to an oscillator (or generator)
inductor to about 40 kHz enables the trans- supplying a 40 kHz source signal that’s also Figure 6. Test arrangement to establish the
ducer to be driven from a low supply volt- fed to one channel of the oscilloscope. The polarity of each and every piezo transducer
age. A step-up transformer as shown in the ‘receiver’ device gets connected to scope used for the parametric array.
speaker schematic is another way to get the channel 2. Now you can view the signal and
transducers to operate at resonance. the timing at a glance (Figure 7).
The resonance frequency fr may be calcu-
lated from Fun with the parametric speaker
It should be reiterated here that the project
fr = 1 / (2 p × L C) is experimental and intended to promote
your own experiments. Connect the audio
Each ultrasonic transducer equals about sound source through 3.5 mm jack socket
2,000–3000 pF worth of capacitance. Con- K2, and connect the power supply to K1.
necting 50 of them you get roughly 0.1– You can probably hear a weak sound from
0.15 μF. To obtain resonance an inductance the transducer array. Carefully adjust R1
of about 60– 160 μH is called for, to con- and R2 for optimum sound quality. Check
nected between the driver’s A and B out- if the sound beam is anything like directive
puts and the respective transducer arrays. — it should be, even when using one chan-
Fine tuning is required to peak for reso- nel (A or B). The author has tentatively indi-
nance and the author produced an adjust- cated a range of about 50 meters (150 ft.) Figure 7. Scope image obtained with
able inductor from enamelled copper wire for a 200-transducer (!) system. correct polarisation of the receiving U/S
and a ferrite rod (Figure 5). For a 200-trans- (100442) transducer.

elektor 03-2011 59