Frequency response of different couplant materials

for mounting transducers

Sabrina Colomboa, A. Giannopoulosb, M.C. Fordeb,*, Roy Hassonb, Jennifer Mulhollandb

aFaberMaunsell Ltd, Imperial House, 31 Temple Street, Birmingham B2 5DB, UKbUniversity of Edinburgh, School of Engineering and Electronics, Institute for Infrastructure and Environment, King’s Buildings, Edinburgh EH9 3JL, UK

Received 24 August 2003; accepted 31 March 2004

Available online 16 September 2004

Abstract

Sensors often are piezoelectric crystal transducers that convert movement (a variation of pressure) into an electrical voltage. Several non-

destructive techniques involve the use of transducers, such as sonic testing, tomography, acoustic emission and pulse-impact echo. There are

different types of transducers according to their different aims and applications, but in all cases the mounting of a sensor is an essential

requirement in order to record good quality data—a good acoustic coupling between the transducer and the surface of the structure has to be

ensured. It is common practice to use Cyanoacrylate adhesive glue (e.g. superglue) for most applications, but the authors found its use

problematic during temporary installations due to the difficulties encountered to remove the sensor at the end of the experiment. For this

reason, a study has been carried out to investigate possible alternative couplant materials. Eight different materials have been selected, and

their amplitude of response in terms of time-domain and the frequency-domain has been compared. A final evaluation based on several pre-

defined criteria has then been obtained, showing the feasibility of ‘plasticine’ as a valid alternative to superglue.

q 2004 Published by Elsevier Ltd.

Keywords: Non-destructive testing; Transducer; Sensor; Couplant material; Time-domain; Frequency-domain

1. Introduction

Several non-destructive methods (acoustic emission,

sonic and ultrasonic testing, tomography, pulse-impact

echo) involve the use of piezoelectric transducers [1].

Physically, the transducers detect a movement that leads to a

redistribution of the electrical charges inside the crystal,

resulting in a change in the voltage [4]. Although the

different techniques might employ the sensors in a different

way and there are different types of sensors depending on

their specific use, in all circumstances their reliability is

fundamental for the success of a test. The reliability depends

not only on the intrinsic characteristics of the sensor (i.e.:

type, manufacturer, and so on), but also on the way it is

mounted on the structure under investigation. The require-

ment of a ‘correct’ mounting of the transducer includes:

0963-8695/$ - see front matter q 2004 Published by Elsevier Ltd.

doi:10.1016/j.ndteint.2004.03.008

* Corresponding author. Tel.: C44-131-650-5721; fax: C44-131-452-

8596.

E-mail addresses: sabrina.colombo@fabermaunsell.com (S. Colombo),

m.forde@ed.ac.uk (M.C. Forde).

†

Firstly that a good acoustic coupling between the sensor

and the surface is ensured—in terms of frequency content

and amplitude

†

Secondly that the sensor is properly fixed to the surface

of the tested material.

This paper deals with the first of these issues.

The coupling affects both the quality and quantity of data

and a good acoustic couplant is essential in order to record

good data. To achieve this, the sensor surface needs to be

smooth and clean and the couplant material should be thin

so as to fill any eventual air gaps. The nature of the coupling

substance affects the quality of the bond and has

implications on the quality and reliability of the recorded

signal.

It is common practice to use Cyanoacrylate adhesive glue

(e.g. superglue), but the authors [2,3], found some major

drawbacks. In fact, when the sensor’s installation was

temporary (i.e. it had to be removed at the end of test) its

removal was difficult and the sensor was at risk of being

damaged.

NDT&E International 38 (2005) 187–193

www.elsevier.com/locate/ndteint

S. Colombo et al. / NDT&E International 38 (2005) 187–193188

For this reason an experimental study of different materials

was designed and carried out. The experiment consisted of

comparing the signal response obtained by transducers

coupled with the varied materials. On this basis, as well as

taking into account the practical characteristics of the

substances, it was evaluated which material could provide a

valid and/or better alternative to the use of superglue.

2. The experiment

2.1. Materials and criteria

As mentioned earlier, the experiment described in this

paper investigated the use of different sensors couplant

materials. The following eight substances were selected,

and the specific products were chosen to be representative of

the commonly available brands of each substance:

†

superglue—Cyanoacrylate adhesive, manufactured by

RS components

†

micro soft beads of wax

†

general purpose light brown grease by RS components

†

plasticine—indicating the trademark used for soft

modelling material (generally used by children); the

‘Humbrol’ brand was specifically used during this

experiment

†

general purpose adhesive—manufactured by RS

components

†

sealant—indicating a waterproof all purpose sealant; the

‘Uni-bond—Henkel’ brand was specifically used

†

adhesive pads—the ‘Pritt Sticky Pads’ brand was

specifically used

†

Hot gluegun glue—indicating the hot glue commonly

used in gluegun; the ‘Loctite’ glue was specifically used

As the thickness of some of the substances also affects

the response, three different thicknesses were considered in

those cases: 2, 5 and 8 mm. Since the best results were

always obtained using a 2 mm thickness, only this case will

be considered and discussed.

Based on the problems encountered during previous

tests, in order to evaluate the global performance of the

couplant, the following criteria were considered:

†

quality of the received signal—e.g. the frequency content

and the amplitude;

†

adaptability to surface—e.g. ability of the substance to

mould to the surface;

†

repeatability—e.g. ability to produce the same quality of

results during repeated tests;

†

durability—e.g. ability of the substance to hold the

transducer over time;

†

ease of installation;

†

ease of removal—e.g. time taken and effort required to

remove the accelerometer from the surface and the

couplant from the accelerometer, without the latter being

damaged;

†

location limitations—e.g. necessity of power supply that

could limit the ease of use in some fieldwork;

†

time for substance to set, if necessary.

2.2. Equipment and calibration

A general accelerometer was used to carry out the

experiment. The test was undertaken using an impact-echo

system, which tied into MATLAB. The impact-echo

technique consists of generating waves in a material through

an impact applied by an external source, such as a hammer

or a ball bearing [5]. The system used for this experiment

had three channels: one recorded the hammer impulse, the

remaining two recorded the signal detected by the accel-

erometers. The latter were connected to a pre-amplifier and

through it to a switch box and then on to a laptop. The

transducers were PCB Piezoelectric 308B15 acceler-

ometers, which have a broadband frequency range of

1–3000 Hz. They weigh about 55 g and use quartz as

piezoelectric material; they are connected through coaxial

cables and have a built in amplifier. A metal mounting base

was used for their protection. A PCB 086B03 impulse

hammer was used to generate the impacts at the centre of the

bottom of the slab. The hammer has a working frequency

range from 0 to 8000 Hz, a resonant frequency at 31 kHz

and a mass of 280 g. Ten impacts were generated for each

case in order to have a reliable sample of signals.

A calibration was undertaken by mounting the two

accelerometers with the same couplant and verifying that

the recorded signals were identical.

The test consisted of three stages, during which the

signals were compared in the time-domain and frequency

domain. The comparison was carried out using the signal

processing MATLAB toolbox that automatically performs

the conversion from time to power spectrum. The power

spectrum describes the distribution, over frequency, of the

power contained in the signal. Different methods can be

used to estimate the power spectral density and The Welch’s

method available in MATLAB was used for the data of this

experiment as it produced smoother plots. This method

divides the time-domain data into (possibly overlapping)

segments, producing a power spectrum for each of them and

then averages them to produce the final plot.

3. First stage: experiments and results

During the first stage, the propagating medium was

represented by an octagonal solid concrete plate, 25 mm

thick and with a side of 150 mm. The plate was supported on

eight brick columns, one positioned at each corner of the

plate. Photos of the equipment, hammer and the slab can be

seen in Fig. 1.

Fig. 1. Photos of the equipment, hammer and ball bearing and concrete slab.

S. Colombo et al. / NDT&E International 38 (2005) 187–193 189

Two accelerometers were used simultaneously, one

accelerometer was mounted as a ‘control’, using superglue,

whilst the other was mounted with different couplants that

could then be compared against the ‘control’ signal. The

substances that provided a good signal with no major

drawbacks were then advanced to the second stage of the

investigation, where they were directly compared to each

other.

The results of this first stage permitted the elimination of a

few materials. The sticky pad couplant was excluded, as the

quality of the signal (see Fig. 2) was exceptionally poor: both

amplitude and number of peaks in the time domain are much

Fig. 2. Time-domain signals of s

lower than those in the superglue signal. As a result it was

considered unnecessary to convert the signal to its frequency

domain. Wax and grease were disqualified due to installation

difficulties. In fact the use of wax was impractical. The beads

of wax must be placed in position and melted using a heating

source such as a cigarette lighter or blow lamp. This requires

a large amount of heat for a considerable length of time and it

can be extremely difficult to ensure that all the wax is

fully melted. Once all the wax was in a liquid state and the

heat source removed, the wax set very quickly. Therefore the

accelerometer had to be placed in position accurately and

almost instantly. In practice it was easy to fail to get

uperglue and sticky pads.

Fig. 3. Concrete slab results: time domain (left) and power spectral density (right) plots for the different couplant types, shown in comparison with each other.

S. Colombo et al. / NDT&E International 38 (2005) 187–193190

the accelerometer mounted at all; it was difficult to mount it

in the perfect position and impossible to adjust it once the

wax had set. Finally, this whole process would be impossible

to carry out on anything but a horizontal surface. The use of

grease was also problematic, as its bounding characteristics

were very poor, resulting in poor coupling of the accel-

erometer. Grease also stains the surface. Both the sealant and

the general adhesive were finally eliminated due to their slow

set up times. In both cases approximately twelve hours were

required in order for the couplant to set properly. Moreover,

significant time and effort had to be put into removing the

substances from the transducer and the concrete.

4. Second stage: experiments and results

Three materials reached the second stage: superglue,

2 mm thick plasticine and gluegun glue. By comparing

Fig. 4. Testing set up: concrete slab and cube.

S. Colombo et al. / NDT&E International 38 (2005) 187–193 191

the amplitude of the recorded waveforms and their

frequency components, the superglue always provided the

best quality signal. The signal recorded using the gluegun

glue (top graph of Fig. 3) showed some losses of amplitude

compared to the superglue signal. Some losses can be also

seen in the plasticine signal (middle graph of Fig. 3) when

compared to the superglue, although the plasticine

performed better than the Gluegun glue (bottom graph of

Fig. 3). The following ranking in terms of signal quality was

thus deduced:

Fig. 5. Concrete cube results: time-domain si

1.

gnal

superglue;

2.

plasticine;

3.

gluegun glue.

5. Third stage: experiments and results

In the final stage a test on a cast concrete cube (150 mm)

specimen was carried out using these couplants. A ball bearing

(13 mm in diameter) attached by an elastic band to a frame that

s for the three candidate couplants.

Fig. 6. Concrete cube results: power spectral density plots for the three candidate couplants.

S. Colombo et al. / NDT&E International 38 (2005) 187–193192

sat above the impact point, was used as an impact source. The

frame was located such that the ball would strike the surface

only once. This system set-up would ensure an adequate,

constant and reproducible impact. Only one accelerometer at a

time was used and this was mounted on the centre of one of the

vertical sides of the cube. This was due to the fact that an initial

calibration using two accelerometers showed significant

differences between the recorded signals. This was probably

caused by internal flaws or differences in the concrete, as well

as a result of modal resonance from the cube sides [5]. The test

was repeated using two different cube faces and the setting of

the experiment can be seen in the right photo of Fig. 4.

The results (Figs. 5 and 6), both in terms of time and

frequency domain, showed that superglue provided by far

the better signal, whilst the response of plasticine and

gluegun glue were very similar to each other.

Table 1

Transducers mounting substances: final assessment table

Criteria Superglue Plasticine Gluegun glue Grease

Quality of received

signals

5 3 2 2

Adaptability 3 5 5 4

Repeatability 5 5 5 2

Durability 5 3 5 2

Ease of installation 4 4 3 2

Ease of removal 1 5 4 4

Location limitations 5 5 3 5

Time to set 3 5 4 4

Total/40 31 35 32 25

6. Final discussion and comparison

A final comparison table was then produced (Table 1) to

globally evaluate the performance of the substances. In

order to do that, the previously mentioned criteria were

taken into account and a value between 1 and 5 (5Zexcellent; 1Zpoor) was assigned to each of them. The three

materials that reached the final stage of the experiments

were herein considered. The grease was also included, as an

indication of negative performance. Overall, plasticine

appeared to be the best choice of material.

It should be pointed out at this stage that the weighting

system used is arbitrary. It derives from the main aim of the

experiments, which was to find a material that would give

reasonably good results and at the same time would be easy

to use and remove during practical testing. Although the

quality of the data is undoubtedly higher using Cyanoacry-

late glue (the amplitudes of the recorded peaks for the

plasticine and the glue gun glue in Fig. 6 are very low and a

frequency shift can also be noted), the difficulty of its

removal heavily penalizes his final score in the assessment

table. Different aims and/or necessities might possibly lead

to a different weighting system and final score of the

materials.

7. Conclusions

This paper presented the results of an experiment aimed

to find couplant materials to be used as an alternative to

superglue. Eight substances were initially selected and their

feasibility was evaluated on the basis of the signals response

S. Colombo et al. / NDT&E International 38 (2005) 187–193 193

(in term of signal amplitude in the time-domain and the

frequency-domain) and some established practical criteria.

Although superglue always provided the best quality

signals, when the practical factors were considered,

plasticine was found to be the better performing material

overall.

Acknowledgements

The authors wish to acknowledge the technical and

support staff of the University of Edinburgh as well as the

University facilities that were made available.

References

[1] Bungey JH, Millard SG. Testing of concrete structures. London:

Chapman & Hall; 1996. p. 286.

[2] Colombo S. Feasibility study of the application of the acoustic emission

technique to concrete bridges. PhD Thesis, University of Edinburgh;

2003; p. 299.

[3] Colombo S, Forde MC. AE experiments on concrete beams: general

overview and research in progress on bridges. In: Forde M, editor.

Proceedings of the international conference on structural faults Crepair 2001, volume Cd-Rom, London. ISBN 0-947644-47-4.

[4] Ohtsu M. The history and development of acoustic emission in concrete

engineering. Concr Libr Jpn Soc Civil Eng 1995;(25):121–34.

[5] Sansalone MJ, Streett WB. Impact-echo: nondestructive evaluation of

concrete and masonry. Bullbrier Press; 1997 p. 339.