You are on page 1of 10

90 4  Phased Array Beam Modeling (1-D Elements)

Fig. 4.14   A 32 element


array radiating into water at
5 MHz. Element size is two
wave lengths with a gap size
equal to one tenth of an ele-
ment length

Fig. 4.15   A 32 element array


radiating into water at 5 MHz
with a delay law to produce a
steering angle of 20° and no
focusing (Phi = 20, F  = inf).
Element size is one half a
wave length with a gap size
equal to one tenth of an ele-
ment length

generated by this array. It can be seen that there are no grating lobes visible and the
array generates a near field beam structure similar to that of a single element trans-
ducer. However, if one changes the element length to be two wavelengths, with all
other parameters staying the same, then we see definite grating lobes in addition to
the main beam (Fig. 4.14). If we apply a steering law only to the array considered in
Fig. 4.13 the wave field in Fig. 4.15 looks as if it was coming from a rotated trans-
ducer as discussed in Chap. 3. Figure 4.16 shows the 32 element array considered
in Fig. 4.13 when a focusing delay law is used to focus the beam at a distance of
20 mm. It can be seen that the beam does develop a beam “waist” near the speci-
fied focal point. As a final example, consider the array example of Fig. 4.13 again
4.6 Use of Gaussians for Modeling Phased Array Beam Fields 91

Fig. 4.16   A 32 element array


radiating into water at 5 MHz
with a delay law to produce
focusing and no steering
(Phi = 0, F = 20  mm). Element
size is one half a wave length
with a gap size equal to one
tenth of an element length

Fig. 4.17   A 32 element


array radiating into water at
5 MHz. Element size is one
half a wave length with a gap
size equal to one tenth of an
element length. A Blackman
apodization law is applied to
the elements

but where we apply Blackman apodization weights to the array. As can be seen in
Fig. 4.17, this apodization produces a beam with a much smoother profile and with-
out the visible side lobes of Fig. 4.13.

4.6 Use of Gaussians for Modeling Phased Array Beam


Fields

In Chap. 3 we showed that a multi-Gaussian beam model is an effective model for


large single element transducers. Because of the use of the paraxial approxima-
tion in the multi-Gaussian beam model, which assumed that the entire beam of the
92 4  Phased Array Beam Modeling (1-D Elements)

element was contained in a narrow region centered around the z-axis (normal to
the element), that model cannot accurately predict the fields at large angles from
the z-axis of small elements, as is required in phased arrays when significant beam
steering is present. However, it is still possible to use the Gaussian basis functions
developed by Wen and Breazeale in conjunction with a high frequency line source
model to model properly the wave field at large angles of small elements by using a
non-paraxial expansion [6]. To see this, we start from the high frequency line source
model with a Gaussian velocity profile on the surface at z = 0 (see Eq. (3.99)), which
we rewrite as
 +∞
k 2 exp(ikr )
p (x, ω ) = ρcv0 (ω ) A ∫ exp  − B( x ′) / b  r dx ′, (4.53)
2

2π i −∞

where x = ( x, z ) and r = ( x − x ′ ) 2 + z 2 . However, instead of expanding the ra-


dius, r, about the z-axis, as is done with the paraxial approximation, we let

r= x 2 + z 2 − 2 xx ′ + ( x ′ ) 2
(4.54)
= r02 − 2 xx ′ + ( x ′ ) 2
( x ′ ) 2 − 2 xx ′
= r0 1 + ,
r02

where r0 = x 2 + z 2 and assume that x ′ / r0 << 1 (which is valid for a small ele-
ment if we are not too close to the element). This same expansion was discussed in
Chap. 2 when looking at the validity of the far field approximation (see Eqs. (2.39),
(2.40), and (2.41)) and we found that to keep at least quadratic terms in the expan-
sion we had to use three terms in the binomial expansion to find
 xx ′ ( x ′ ) 2 cos 2 θ
r ≅ r0 − + , (4.55)
r0 2r0

where cos θ = z / r0 and θ is the angle that the radius r0 makes with respect to the
z-axis. Note that the original derivation [6] of this approach kept only two terms in
the expansion which had the effect of making the replacement cos 2 θ → 1. Placing
the correct second order approximation of Eq. (4.55) into Eq. (4.53) then gives

k exp(ikr0 )
+∞
 2  cos θ
2
iB    −ikxx′ 
p (x, ω ) = ρcv0 (ω )
2π i
A
r0 −∞ ∫ exp  ik ( x ′ ) 
 2R
+ 2   exp 
kb    R 
 dx′. 

(4.56)

But the integral in Eq. (4.56) can be done [Schmerr-Song], leading to the explicit
expression
4.6 Use of Gaussians for Modeling Phased Array Beam Fields 93

Fig. 4.18   A 32 element array radiating into water at 5 MHz with a delay law to produce a steering
angle of 20° and no focusing (Phi  = 20, F = inf). Element size is one half a wave length with a gap
size equal to one tenth of an element length. This is the same case considered in Fig. 4.15 where
the wave field was calculated with a multiple line source model. Here, the wave field is calculated
with (a) a non-paraxial Gaussian model and (b) with a Hankel function model.


ρcv0 (ω ) A exp(ikr0 )  −ikx 2 
p ( x, ω ) = exp  , (4.57)
 2r0 (cos θ + iBr0 / D) 
2
cos θ + iBr0 / D
2

where D = kb 2 / 2 is the Rayleigh distance [Schmerr-Song]. Here we will use the


10 rather than the 15 Wen and Breazeale coefficients for piston behavior [7] in order
to keep the calculation times to a minimum. We then have a complete beam model
for an element in the form

10
ρcv0 (ω ) An exp(ikr0 )  −ikx 2 
p (x, ω ) = ∑ exp  . (4.58)
n =1 cos θ + iBn r0 / D
2 2
 0 r (cos 2
θ + iB r
n 0 / D ) 

The MATLAB® function NPGauss_2D (Code Listing C.16) which has the calling
sequence
>>p=NPGauss_2D(b, f, c, e, x, z);
implements Eq. (4.58) for an element whose center is offset a distance, e, in the x-
direction, so that r0 = ( x − e) 2 + z 2 . It uses the ten Wen and Breazeale coefficients
which are contained in the MATLAB® function gauss_c10 (Code Listing C.17). One
can replace the multiple line source beam model ls_2Dv in the MATLAB® script
mls_array_model with NPGauss_2D to generate comparable results for phased ar-
rays. Figure 4.18a shows the same setup considered in Fig. 4.15 (a 32 element array
radiating into water at 5 MHz with a delay law to produce a steering angle of 20°
and no focusing. The element size was one half a wave length with a gap size equal
94 4  Phased Array Beam Modeling (1-D Elements)

to one tenth of an element length) but with the use of the non-paraxial Gaussian
model in the MATLAB® script mls_array_model in place of ls_2Dv. Figure 4.18b
shows again the same results but with the Hankel function beam model (contained
in the MATLAB® function rs_2Dv) used in place of ls_2Dv in the same script. It
can be seen from Figs. 4.15 and 4.18 that the Hankel function model and multiple
line source model results appear identical and the non-paraxial Gaussian model also
agrees quite well with the other two models but with some differences appearing in
the very near field region adjacent to the face of the array where the expansion of
Eq. (4.55) cannot be expected to be accurate. However, the non-paraxial Gaussian
model took 13 times longer to evaluate than the multiple line source model and the
Hankel function model took 8.3 times longer than the multiple line source model so
the multiple line source model appears to be the best overall choice for simulating
phased arrays in 2-D problems. This is not surprising since the multiple line source
model uses only a single line source term for the small element size considered
here (element length less than a wavelength) whereas the other models use multiple
element segments or basis functions in their calculations. Even for larger array ele-
ments the multiple line source model remains efficient, losing its advantage only
when the element length is tens of wavelengths, as found in large, single element
transducers. In contrast, for large, single element circular transducers a paraxial
multi-Gaussian beam model only needs ten or fifteen Gaussians to produce accurate
wave field calculations [Schmerr-Song] so it is typically the fastest beam model
available for those cases, particularly when calculating the fields through interfaces.

4.7 Beam Steering and Focusing through a Planar


Interface

In Chap. 2 we also described modeling the radiation of an element through a planar


interface. We can use that model and combine it with the beam steering and focus-
ing laws developed in Chap. 5 for this case and the discrete apodization laws of
Sect. 4.4 to simulate array wave fields with an interface present. All these elements
are combined in the MATLAB® script mls_array_int (Code Listing C.18) which is
very similar in structure to the mls_array_modeling script for a single medium used
in Sect. 4.5. The calling sequence for this script is simply

>>mls_array_int

The script uses the beam modeling MATLAB® function ls_2Dint discussed in
Chap. 2 as well as the MATLAB® discrete_windows function for the apodiza-
tion laws. The time delay laws are generated in the MATLAB® function delay_
laws_2D_int (Code Listing C.19) developed in Chap. 5. The calling sequence for
this delay law function is

td=delay_laws2D_int(M, s, angt, ang20, DT0, DF, c1, c2, n)


4.7 Beam Steering and Focusing through a Planar Interface 95

Fig. 4.19   Geometry and


material parameters for steer-
ing and focusing an array
through a planar interface

Fig. 4.20   Normalized pres-


sure wave field in steel for
a 32 element array located
in water at a distance of
Dt 0 = 25.4 mm from
a water-steel interface.
Other setup parameters
are: d = 2b = 0.25 mm,
∆ = 0.05 mm, θt = 0 ,
f = 5MHz, no apodization or
steering/focusing

where td holds the delay times (in microseconds), M is the number of elements in
the array, s is the array pitch (in mm), angt = θt is the angle the array makes with
the interface (in degrees), ang20 = θ 20 is the specified steering angle (in degrees) as
measured in the second medium, DTO =  Dt 0 is the distance of the center of the array
from the interface (in mm), DF = D f is the specified focal depth in the second me-
dium (in mm). If DF = inf is specified then steering without focusing is present. The
variables (c1, c2) are the wave speeds in the first and second media, respectively, in
m/s. The final input string parameter (’y’ or ’n’) specifies if a plot of the rays corre-
sponding to the delay law parameters is wanted. Figure 4.19 illustrates these various
parameters and Chap. 5 gives more details of this function.
To illustrate the use of this script, consider the case of a 32 element array in
water located 25.4 mm from a water-steel interface and oriented parallel to the in-
terface ( θt = 0° ). Figure 4.20 shows results for the case where the element length is
0.25 mm, the gap length is 0.05 mm, the frequency is 5 MHz, and no apodization or
96 4  Phased Array Beam Modeling (1-D Elements)

Fig. 4.21   Normalized pres-


sure wave field in steel for
a 32 element array located
in water at a distance of
Dt 0 = 25.4 mm from
a water-steel interface.
Other setup parameters
are: d = 2b = 0.25 mm,
∆ = 0.05 mm, θt = 0°,
f = 5MHz, θ 20 = 30°,
D f = inf, no apodization

Fig. 4.22   Normalized pres-


sure wave field in steel for
a 32 element array located
in water at a distance of
Dt 0 = 25.4 mm from
a water-steel interface.
Other setup parameters
are: d = 2b = 0.25 mm,
∆ = 0.05 mm, θt = 0°,
f = 5MHz, θ 20 = 30°,
D f = 8 mm, no apodization

delay law is specified for the array. The near field structure of this array wave field
in the steel can be clearly seen in Fig. 4.20. Figure 4.21 shows the same setup as in
Fig. 4.20 but where now a delay law is chosen to steer the array, without focusing,
at an angle θ 20 = 30° in the steel. Figure 4.22 is for the case where a steering angle
θ 20 = 30° is again specified, along with a focal depth D f = 8 mm. Steering and fo-
cusing effects consistent with these choices of the delay law are clearly evident.
Figure 4.23 shows the wave field of a 16 element array where the element length is
0.325 mm and the gap length is 0.05 mm but where Dt 0 = 50.8 mm the array is now
4.7 Beam Steering and Focusing through a Planar Interface 97

Fig. 4.23   Normalized pres-


sure wave field in steel for
a 16 element array located
in water at a distance of
Dt 0 = 50.8 mm from a water-
steel interface, θt = 10.217°.
Other setup parameters
are: d = 2b = 0.325 mm,
∆ = 0.05 mm, f = 5MHz,
θ 20 = 45° (no steering),
D f = inf, no apodization

Fig. 4.24   Normalized pres-


sure wave field in steel for
a 16 element array located
in water at a distance of
Dt 0 = 50.8 mm from a water-
steel interface, θt = 10.217°.
Other setup parameters are
d = 2b = 0.325 mm,
∆ = 0.05 mm, f = 5MHz,
θ 20 = 30°, D f = inf, no
apodization

at an angle θt = 10.217° from the interface. By Snell’s law this will generate a re-
fracted beam at 45° in the steel. In this case the time delay law parameters were cho-
sen to be θ 20 = 45°, D f = inf, so that there is no steering or focusing. The size of this
array is the same as the large single element transducer example shown in Fig. 2.17
so it is not surprising that without steering or focusing the wave field images are
very similar. Figure 4.24 shows the same array of Fig. 4.23 but with beam steering
specified as θ 20 = 30°. It can be seen that the wave field has indeed been shifted to
the new specified refracted angle. The 2-D modeling studies of this chapter are in
98 4  Phased Array Beam Modeling (1-D Elements)

the spirit of those done by Wooh et al [1–3]. See also [4–5] for some similar model-
ing simulations and discussions of efficiency.

References

1. S.-C. Wooh, Y. Shi, A simulation study of the beam steering characteristics for linear phased
arrays. J. Nondestr. Eval. 18, 39–57 (1999).
2. A. Clay, S.-C. Wooh, L. Azar, J.-Y. Wang, Experimental study of phased array beam steering
characteristics. J. Nondestr. Eval. 18, 59–71 (1999).
3. S.-C. Wooh, Y. Shi, Optimum beam steering of linear phased arrays. Wave. Motion. 29, 245–
265 (1999).
4. E. Kuhnicke, Plane arrays—fundamental investigations for correct steering by means of sound
field calculations. Wave. Motion. 44, 248–261 (2007).
5. P. Crombie, A.J. Bascom, R. Cobbold, Calculating the pulsed response of linear arrays: accu-
racy versus computational efficiency. IEEE T. Ultrason. Ferr. 44, 997–1009 (1997).
6. X. Zhao, T. Gang, Non-paraxial multi-Gaussian beam models and measurement models for
phased array transducers. Ultrasonics. 49, 126–130 (2009).
7. J.J. Wen, M.A. Breazeale, A diffraction beam field expressed as a superposition of Gaussian
beams. J. Acoust. Soc. Am. 83, 1752–1756 (1988).
Chapter 5
Time Delay Laws (2-D)

In Chap. 3 we introduced continuous time delay laws for beam steering and focus-
ing of large, single element transducers and we obtained discrete versions of those
laws in Chap. 4 for phased arrays. The explicit focusing delay laws discussed in
both previous chapters used the paraxial approximation. This approximation gave
us a simple delay law to implement and it was consistent with the paraxial ap-
proximation used in the beam models in Chap. 3. However, there is no requirement
to introduce such approximations in the delay laws designed for phased arrays. In
this chapter we will derive exact delay laws for combined steering and focusing in
a single medium and describe the case where the steered/focused beam must pass
through a planar interface. In all cases, we will limit our discussion in this chapter to
1-D arrays radiating waves in two-dimensions. The corresponding delays laws for
2-D arrays radiating into three dimensions will be discussed in Chap. 8.

5.1 Delay Laws for a Single Medium

In designing a delay law for steering and focusing a 1-D array in two dimensions,
one can parameterize the delay law in several ways. One way is to specify the steer-
ing angle, Φ , as measured along the central axis of the entire array, and the focal
distance, F, as measured along this axis (see Fig. 5.1a and b). For both cases shown
in Fig. 5.1 the distance from the center of the array to the center of the first element
is B − b = ( M − 1) s / 2 , where 2B is the total length of the array, 2b is the length of
an element, and s is the pitch of the array (see Eq. (4.5)). The quantity ( M −1) / 2
will appear frequently in our delay law expressions so we will abbreviate that quan-
tity as M :
(5.1) ( M − 1)
M = .
2

The distance to the centroid of the mth element, em , in terms of M is then


(5.2)
em = [(m − 1) − M ]s.
L. W. Schmerr Jr., Fundamentals of Ultrasonic Phased Arrays, 99
Solid Mechanics and Its Applications 215, DOI 10.1007/978-3-319-07272-2_5,
© Springer International Publishing Switzerland 2015

You might also like