You are on page 1of 17

A Review of Advanced Phased

Array Approaches & Techniques


Educational Note

A Review of Advanced Phased


Array Approaches & Techniques

Page 1 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com
Contents
Introduction............................................................................................................................................ 3

Classic phased array - beamforming ................................................................................................ 4

Beamforming in transmission: ...................................................................................................... 4


Beamforming in reception ............................................................................................................. 5
Scanning patterns for beamforming phased array .................................................................. 6
A variation on beamforming: Dynamic Depth Focusing (DDF) .................................................. 7

Total Focusing Method ......................................................................................................................... 9

TFM visualization of various discontinuities ........................................................................... 10


Variations on FMC/TFM ...................................................................................................................... 12

Half Matrix Capture ....................................................................................................................... 12


Sparse Matrix Capture (or Selective Matrix Capture) ........................................................... 13
Adaptive TFM ................................................................................................................................... 13
Modifying the transmission parameters....................................................................................... 14

Increasing the transmission aperture ....................................................................................... 14


Virtual Source aperture ................................................................................................................ 14
Plane wave imaging ....................................................................................................................... 15
Implications of phased array modes .............................................................................................. 15

References............................................................................................................................................. 17

Page 2 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com
Introduction

This is the fourth in a series of notes related to ultrasonic non-destructive inspection,


they are intended to give internal and external staff involved in the sale and support of
Sonatest products a brief introduction to the technology and concepts involved. They
are not intended as a tuition course in how to use the equipment, although a few
relevant aspects may be highlighted.

These education notes should be considered as a ‘work in progress’ please contact


the author with any comments or suggestions.

At the time of writing (January 2021), phased array ultrasonic testing has been part of
the NDT toolkit for nearly 20 years, the first portable instrument (from Harfang, now
Sonatest Canada) being launched in 2002 at the 8th European Conference for NDT in
Barcelona.

Since that time, it has become affordable and widely used. Many internationally agreed
standards have been developed, enabling the routine use of phased array techniques
without requiring complex ‘performance demonstration’ activity. In turn, this has meant
that the technique can be specified in procedures for a wide range of manufacturing
and in-service inspections.

As the technique has evolved, several different approaches to creating a sound field and
evaluating the reflections that come back have been developed. While these expand the
capabilities of the technique, they also add to the complexity of choices and potential
confusion. This paper is an attempt to explain the different varieties of phased array,
and where they have advantages and disadvantages. In some applications, the new
approaches allow reliable and accurate inspections that were not previously possible. In
other applications, they are completely unnecessary. To understand which is
appropriate, an understanding of the capabilities is required.

Page 3 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com
Classic phased array - beamforming

The beamforming approach uses phased array


principles to simulate the operation of a conventional
ultrasonic probe, with the key difference that its
properties can be controlled electronically and
changed many times (potentially thousands) per
second to generate a scan. Because the individual
probe elements are very narrow, they will send energy
over a wide range of angles (see Figure 1 ). Within this Figure 1 Field at -10dB from a typical
range we can ‘phase’ the elements so that the phased array probe element (5MHz,
transmitted beam strength and receiving sensitivity is 0.6mm)
directed at a specific angle or focus.

When using the beamforming approach – in common with all phase array techniques –
we look at the transmission and reception properties separately.

Beamforming in transmission:

The idea of beamforming is to fire multiple elements of an array, so that the sound field
produced by them reinforces at a selected point. To do this, we fire each element with a
delay, so that the sum of the applied delay and the time for the sound to reach that
point is the same for all elements. Therefore, the sound pulses from each element
reach that point simultaneously, and the sound intensity is maximised.

On the Veo series instruments, or on the UT studio software, we can show this delay
graphically:

Figure 2 With all elements firing Figure 3 With the outer elements of the
simultaneously, the array acts as a single array firing earlier, the beam focus will
probe of similar size. The beam will shorten
propagate at the nominal refraction angle
of the wedge, focusing at the equivalent
near field length

Page 4 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com
Figure 4 Firing the upper elements first Figure 5 Firing the lower elements first
will steer the beam down will steer the beam upwards

This will produce a sound beam that is essentially equivalent in beam spread and
energy profile to that produced by a mono-element probe of the same size and
frequency. This allows experience of what can be detected gained with mono-element
transducers to be directly transferred to phased array use: for example, a Sonatest X4
series probe used in natural focus mode will have nearly identical beam characteristics
to a standard 8mm x9mm probe, such as a Sonatest MMA or Waygate MWB.

Beamforming in reception

Figure 6 Block diagram of beamformer receive section.

The reception side, shown schematically in Figure 6, works similarly. A digital delay is
applied to the received signal on each element, so that the effective delay (physical
transit time plus added digital value) from the target point is identical. Ultrasound from
that point will then be enhanced as the channels are added together in synchronisation.
For most applications of beamforming phased array the delays for transmission and
reception are identical – the probe focuses at the same point in transmission and
reception. This combination of transmit and receive delays, potentially alongside other
parameters such as sensitivity and variable gain amplifier characteristics, is called a
focal law.

The beamformer circuitry is designed so that we can very quickly (a few microseconds)
load a new focal law, generating a beam with different characteristics.

Page 5 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com
Scanning patterns for beamforming phased array

A beamformer design allows the software to select direction, focus and which elements
are fired for each beam or focal law. In turn, these focal laws are sequenced to create a
frame or scan pattern.

Being able to select a new focus and direction for each beam means that the scanning
be optimised to cover a critical region.

Using a larger number of elements to generate the beam gives a tighter focus – more
accurate resolution within a selected region, but a worse resolution outside it. This is
exactly analogous to using a larger aperture on a manual camera.

The images below show a near identical configuration with the region of optimised
focus highlighted. The ‘natural focus’ is mostly used when evaluation is carried out by an
amplitude method; this means that the amplitude response of the beam is very similar
to a mono-element probe of similar aperture, avoiding distortion due to focusing
effects.

a) Natural Focus b) Constant Path Focus

c) Constant Depth Focus d) Constant Offset Focus

Figure 7 Beamforming phased array, different focus patterns

Most beamformer circuitry includes a multiplexor. This allows a circuit with only a
limited number of channels (typically 16 or 32) to be connected to one or more arrays
with a much greater number of elements. Each focal law can only address elements
corresponding to the number of channels, but successive focal laws can address

Page 6 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com
different elements. This allows linear scans to be produced, or more than one probe to
be connected; for example, allowing a weld to be scanned from both sides at once.

Figure 9 Dual-sided weld scan


Figure 8 Linear scan, 64 elements total,
eight in any focal law

A variation on beamforming: Dynamic Depth Focusing (DDF)

Standard beamforming works well, but has the disadvantage that it can be difficult to
get an image with acceptable focus covering all of the region of interest. One approach
to solving this is Dynamic Depth Focusing (DDF): a single focal law is applied at
transmission, but multiple subsequent delay laws are applied on reception, forming a
long pulse-echo focal spot for each beam. Combining these by selecting the data from
the optimal scan allows an acceptable focus to be used at each depth. Because we are
focusing only on reception, this will usually not be as sharply focused on an individual
point as a scan optimized for that point.

Figure 10 Focus for transmission (Natural Figure 11 Zone 1 Focus


‘unfocused’)

Figure 12 Zone 2 Focus Figure 13 Zone 3 Focus

Page 7 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com
Figure 14 Zone 4 Focus Figure 15 Combined region of acceptable
focus

DDF is used in some industries, particularly for inspection of thick material welds.

Page 8 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com
Full Matrix Capture

Full Matrix Capture uses a quite different approach. Rather than combining multiple
elements of the array to form a transmitted beam, each element is fired individually.
Since the individual elements are extremely small, they approximate to an
omnidirectional source. After each element is fired, the reflected sound is separately
collected from each element on receive. If there are more elements than channels, it is
possible to fire the element more than once – e.g. fire element 1, and collect sound on
elements 1-32, then fire element 1, collect sound on 33-64, then fire element 2 etc.
Once the data has been collected, it can be processed and displayed immediately (Live
TFM) or recorded for offline analysis.

This approach generates huge amounts of data. For example, a frame of standard
phased array data, containing maybe one hundred beams with 2kB of data each, would
be 200kB; collecting a meter of data at one frame per millimetre would be 200MB.
Large, but manageable.

If we collect the same scan as full matrix data, again with 2kB per A-scan and a 64-
element probe, we have 64 (transmit elements) x 64 (receive elements) x 1000
(number of points) x 2kB– around 8GB of data. Storing and processing this much data
can rapidly use up disk space, and Live TFM, where it must be analysed at the capture
rate, requires a lot of processor power.

Total Focusing Method

The TFM approach is the most popular method of analysing FMC data (Holmes et al,
2004i). This can be done in real time, or the data can be stored for later analysis. Storing
the data has the advantage that different processing approaches can be used as
required on the same data – important if you are not certain what types of defects you
expect to find.

For each point in the region of interest, we take the data from each A-scan in turn and
sum the amplitudes at the time corresponding to the round-trip distance to that point.

Figure 16 TFM combination showing the first two firings in the simple case of an 8-element
probe. We sum 64 points of amplitude data measured at the round-trip time to obtain the
intensity of the reflection from a particular point.

Page 9 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com
Again, it is apparent that this requires a lot of processing time; for a typical image we
must select and sum many millions of data points.

Probably the biggest advantage with TFM is that, because we are reflecting sound from
a discontinuity at multiple angles, the detectability is less dependent on its orientation.
With some reconstruction algorithms we can find the orientation at which the reflection
is greatest and, at least partly, characterise the indications. By calculating the beam
path to correspond to various reflection patterns and wave modes, we can enhance
them and reduce or eliminate others. We can often size indications far more accurately.
Typically, TFM imaging systems can display various combinations of longitudinal(L) and
transverse(T) waves, propagating directly or reflected via a material surface in
transmission and/or reflection.

Figure 17 LL or TT Figure 18 LTT, TLT, Figure 19 LLLL, TTTT Figure 20 TTTTT


modes TTL, LLL, or TTT modes mode
modes

The TT and TTTT modes correspond closely to ‘half’ and ‘full’ skip in conventional mono-
element inspections. The addition of compound modes can be especially useful in
imaging unfavourable oriented indications, in a similar way to the use of tandem probes
for a mono-element system.

TFM visualization of various discontinuities:

Figure 21 TFM Visualization of plate edge. Note that this shows just corner reflectors on
the TT and TTTT modes; the 'tandem' modes show it more clearly

Page 10 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com
Figure 22 TFM visualization of centre line crack

Figure 23 TFM visualization of root defect

Page 11 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com
Figure 24 TFM visualization of inclusion

As in Figures 21-24, the TFM image often gives


lower sensitivity than a comparable beamforming
phased array scan. However, for certain types of
defects, especially vertically oriented ones, it
works very well.

Full Matrix Capture/TFM can be useful not only


for sector scan-type applications, but also where
a linear scan is used; notably for imaging of
corrosion, where it can be very helpful in finding Figure 25 TFM imaging of rear wall
small irregular pits and accurately assessing corrosion

surfaces that are not parallel. (Figure 25)

Variations on FMC/TFM

Full matrix capture, particularly when we are both capturing data and displaying it in
real time, requires a lot of processing power, many pulser cycles, and generates huge
amounts of data. There are many strategies to reduce some or all of these.

Half Matrix Capture

Half matrix capture relies on the reciprocity principle: that the received signal
transmitting from A to B is the same as that from B to A, and therefore only half the
data collected needs to be stored for processing as the other half is essentially identical.
Storing the sum of the two may provide a slight improvement in signal to noise by
averaging, but half matrix capture generally has no significant downside.

Page 12 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com
Sparse Matrix Capture (or Selective Matrix Capture)

This covers a range of options where only received data on a fraction of the elements is
recorded. This selection may be either random or optimised to those where the
strongest signal is expected (e.g., those adjacent to the transmission elements).
Reducing the number of elements from which data is analysed can greatly improve the
processing speed, but at the cost of some image distortion or an increase noise. Figure
26 and Figure 27 show the effect of reducing the number of elements used to
reconstruct the received signals.

Figure 26 50% of elements selected - Figure 27 6% of elements selected - Gives


Gives an SNR of 25dB and a speed of 30 an SNR of 15dB and a speed of 45 frames
frames per second per second

Sparse matrix capture should not be confused with the use of a ‘sparse matrix array’ for
example a phased array probe with a 16x16 matrix would have 256 elements. By
connecting only half of its elements in a quasi-random selection, it can be used on a
128 channel phased array system. There will be a degree of signal degradation (Diarra
et al, 2012ii ) but for some applications this provides an effective solution without
requiring prohibitively expensive instrumentation. The focal law calculation software will
need a detailed definition of the probe.

Adaptive TFM

Because we do not have to make any special modification to the transmission pulses,
The FMC/TFM approach is very suitable for inspection of irregular surfaces or using a
water column (Le Jeune et al, 2014iii) The approach taken is to use the first reflected
signals to map out the surface of the part, then calculate the timings to each point,
knowing the transit time in water plus the transit time in the material.

Appropriate applications include complex shaped parts, inspection through a weld


bead, or inspection of flat parts using a variable length water column such as a wheel
probe.

Again, the amount of calculation to implement this approach makes it difficult to


achieve without substantial processing power and energy consumption.

Page 13 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com
Modifying the transmission parameters

We can change the reception parameters to optimise the signal to noise / processing
requirement compromise, but the time needed to fire each element separately is still a
fundamental speed limitation, as is the vast amount of data produced.

In addition, the sound field from a single element is very weak; and although we are
averaging many data points, the stronger the incident sound, the stronger the reflection
– and the better the signal to noise is likely to be.

Several approaches exist to potentially improve these issues. They are most effective in
improving the speed when we have a system with plenty of processing power at the
receive side, where the transmission speed will become more of a bottleneck.

Increasing the transmission aperture

A simple method is to fire two or more adjacent elements together then skip to the next
set. This is quite effective, and will give a stronger sound field; however, the larger
aperture will give a more directional sound beam which, if carried too far, may mean
that we do not get enough sound to some parts of the region of interest. The resolution
may also be reduced because the sound is originating form a larger source. This
method can be effective if used carefully; for example, the situation shown in Figure 28
gives a reasonable beam divergence with two elements, which is likely to be acceptable,
but four elements together give a much narrower beam, which may be insufficient for
many applications.

Figure 28 Sound field from1, 2, and 4 elements of a 5MHz, 0.6mm pitch array

Virtual Source aperture

Transmission using a synthetic virtual source creates a special focal law to fire a group
of elements so that they create a divergent beam which correspond to emission from a
single point source behind the array. (Sutcliffe et al, 2012iv)

Useful imaging can be done with only a few transmission pulse sets from virtual sources
distributed along the array.

Page 14 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com
Using a VSA approach has advantages:
i) A strong, highly divergent ultrasound source is
created.
ii) A relatively large number of elements can be
used to generate the sound source.
iii) The position for calculation is well defined.
iv) A reduced number of firings is required to
Figure 29 Using a virtual sound
generate a high-quality image. source.

Plane wave imaging

Originally developed for medical application, this approach sends a succession of sound
beams out at different angles. A reduced number of angles with a reasonably divergent
beam can provide acceptable results, potentially faster (given sufficient processing
power) than a beamforming phased array inspection (Le Jeune et al, 2015v, see also
Hoyle, 2018 vi )

Implications of phased array modes

We have briefly discussed a range of detailed techniques within the general class of
phased array inspection.

All of these have different advantages and disadvantages, and (as far as the author is
aware) there is not yet a single piece of equipment that combines all of them in a
portable, affordable package.

The new modes do have advantages in specific applications, but it is important to


understand that conventional beamforming phased array has many advantages of its
own:
1) Very few of the advanced phased array methods have yet been incorporated into
accepted standards. These are only just starting to appear for FMC/TFM, and it is
likely to be some years before they are in wide use; therefore, the use of novel
methods in routine inspection work is likely to be some years away.
• These methods are still at the ‘performance demonstration’ stage, where
beamforming phased array was 10-15 years ago. This means that, in the
absence of agreed approaches, each application must be carefully
evaluated to be confident that the solution employed will reliably find all
required defects.
2) Evaluation of the relative performance of the different approaches, and the
parameters that can be chosen, is very much in the research stage.

Page 15 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com
3) It is much easier to transfer existing quality standards for conventional ultrasonics
to a beamforming phased array approach than to TFM or complex modes, where
the response of discontinuities is likely to be significantly different. Notably,
amplitude-based inspection criteria are not easily adapted to TFM.
4) Beamforming phased array, especially when combined with TOFD, still has a better
probability of detection for many types of defects.
5) TFM evaluation will normally require multiple transmission modes to be checked.
If the wrong mode is used, it is easy to miss ‘unexpected’ defects.
6) Live TFM is useful but has some limitations: where any quality record must be
kept, it is still likely that full waveform data must be archived. This takes a lot of
space – a typical inspection job can easily run to hundreds of gigabytes.
7) While TFM is unquestionably better for accurate sizing of defects – which is
sometimes needed, especially when assessing growth rates – beamforming
phased array is more than adequate for most go / no acceptance standards.
8) Portable (and therefore relatively low-powered) systems are still likely to be
significantly slower in TFM modes than in beamforming phased array. State-of-the-
art systems for research purposes typically employ desktop or rack-mounted PCs
with multiple graphics processors.

The author’s current recommendation is that for most applications, while FMC/TFM
capability (or the ability to add it in future) is ‘nice to have’, the performance and ease of
use in conventional modes should still be the main equipment selection criteria.

Page 16 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com
References

i
C Holmes, B Drinkwater and P Wilcox, The post-processing of ultrasonic array data using the
total focusing method. Insight - Non-Destructive Testing and Condition Monitoring, BINDT,
Volume 46, Number 11, November 2004, pp. 677-680

Bakary Diarra, Hervé Liebgott, Piero Tortoli, Christian Cachard. Sparse array techniques for
ii

2D array ultrasound imaging. Acoustics 2012, Apr 2012, Nantes, France. ffhal-00810649f

iii
Léonard Le Jeune, Sébastien Robert, Phillippe Dumas, A. Membre, Claire Prada, Adaptive
Ultrasonic Imaging with the Total Focusing Method for Inspection of Complex Components
Immersed in Water QNDE2014 proceedings.

M Sutcliffe, M Weston, P Charlton, B Dutton and K Donne, Virtual source aperture imaging for
iv

non-destructive testing. Insight - Non-Destructive Testing and Condition Monitoring, BINDT,


Volume 54, Number 7, July 2012, pp. 371-379

v
Léonard Le Jeune, Sébastien Robert, Eduardo Lopez Villaverde, Claire Prada, Plane Wave
Imaging for ultrasonic non-destructive testing: Generalization to multimodal imaging.
Ultrasonics, Elsevier, 2016, 64, pp.128-138. 10.1016/j.ultras.2015.08.008. hal-01322017

vi
Ewan Hoyle, Comparison of FMC, VSA and PWI and the Array Limitations when Beam Forming
- Presentation from BINDT 2018

Page 17 © Sonatest, 2020. All rights reserved. All the information here is subject to change without prior notification.

www.sonatest.com

You might also like