Evaluation of air coupling ultrasonic transducers for surface roughness measurement.

Transcription

1 Evaluation of air coupling ultrasonic transducers for surface roughness measurement. Nicolás Pérez* 1, Carlos Negreira* * Laboratorio de Acústica Ultrasonora, Universidad de la República, URUGUAY Francisco F. Montero de Espinosa **3 ** nstituto de Acústica, CSC, Madrid, ESPAÑA Serrano 144, 86 Madrid 1 nico@fisica.edu.uy carlosn@fisica.edu.uy 3 pmontero@ia.cetef.csic.es Abstract Ultrasonic techniques are widely used for non-destructive testing NDT. When dealing with standard applications, a fluid is used to couple the transducer and the sample to be measured. The use of transducers coupled by air opens a wide range of possibilities for these tests, allowing the test of pieces that do not support a coupling liquid nor direct contact bonding, because the constituent materials or due to constraints related to on line operation or spatial geometries. The main difficulty to use transducers coupled by air is the huge mismatch of the acoustic impedance between a solid and the air. Due to the great difference of elastic properties and density in the interface transducer-air, the energy transferred from the transducer is very low. n this work we present a special application that consist of the surface roughness measurement in solid interfaces using air coupled transducers. Specially designed piezoelectric transducers were made for this application. These transducers are made of pzt-polymer piezocomposites, coupled to the air using two quarter length matching layers. The main objective of this work is the evaluation of these transducers to measure surface roughness. The set of samples selected are in the range of those found in industrial applications (irons submissive the corrosion and grinder s of commercial use).. Keywords: Piezocomposites, air transducers, industrial applications, roughness. ntroduction During the last decade ultrasonic aircoupled transducers are displaced from laboratory curiosity to a reality for the nondestructive testing [NDT ] at industrial scale. [1] [] The design of transducers and electronic systems specially adapted to air-coupled ultrasound, made possible a big number of tests that could not be made using coupling fluids because the materials involved or the geometries. Although originally this technique was conceived for aerospace compound materials, the applications have extended to a great range of materials where the traditional NDT methods are difficult to apply. [3] [4] The use of air as coupling media between the transducer and the sample has two main problems: the huge difference of acoustic impedances between a solid and the air and the high absorption of the ultrasound when propagating in the air. To evaluate the energy transfer between two materials, the specific acoustic impedance Z is the relevant parameter. Table shows the specific acoustic impedance of the air in comparison with usual materials in NDT applications. Table Materials Acoustic mpedance Z [Mrayl] Air.4 Water 15 Aluminium 17 Steel 45 The transmission amplitude coefficient between two media of specific impedances Z1 and Z in the case of normal incidence, T is [5]: 63

2 T 4 Z ( Z + Z ) 1 1 Z Eq.1 For example, when an ultrasonic pulse coming from water impinges normally an aluminium flat surface, 3 % of the signal amplitude is transmitted to the aluminium. When the same experiment made in made in air, only a,1 % of the signal amplitude is transmitted to the aluminium. This fact is one of the great difficulties to use air coupled transducers because the dynamic range decreases dramatically. So, specially designed transducers to improve the delivering of energy to the air must be done. The second problem is the great ultrasound energy absorption in air at the frequency range normally used in NDT applications. This effect increases with the frequency and limits the application range of the technique. Frequencies upper than MHz are forbidden for standard through transmission techniques. n this work we use transducers with a central frequency of 8 khz [6]. The transducer is made using 1-3 piezoceramic-polymer piezocomposites ( pz7, Ferroperm & Araldit H, Cyba&Geigy) without backing section. Two quarter wavelength layers (polymer + microporous cellulose) are used to match the piezocomposite polymer impedance to the air. nsertion losses less than db at 1 MΩ coupling are obtained- see Figure 1-.. Since insertion losses are measured using two transducers faced 8 mm in air, it can be observed a ripple associated with the resonances of this air column. The Gain-Phase module of an impedance analyzer HP 4194A were used. Figure.- Picture of the transducers used in this work, the terminals are BNC to be used with commercial pulse-echo equipment. The high efficiency of these transducers allows their use for surface roughness characterization of solids analysing of the reflected signal at the air-solid interface. The roughness is a very important parameter in online quality control processes as, for instance, polishing of pieces and predictive maintenance. Frequently, mechanical faults of metallic pieces suffering stress begin in their surface. The origin can be an isolated manufacturing defect or the damage produced by superficial corrosion. The periodic surface roughness test is then recommended. On the other hand, the traditional method to measure the surface roughness is based on a pick up profiling instrument. This procedure is not easily implemented on-line and is forbidden to be used for delicate materials. Ultrasonic techniques appear like a good alternative for the determination of the roughness in industrial scale [7] [8]. The main title should be printed in Arial typefont, 14 pt., boldface. n the main title, please use initial caps; do not capitalize articles, coordinate conjunctions, and prepositions under four letters in length. For subheadings use 1 pt., boldface type and initial caps. Do not capitalize articles, coordinate conjunctions, and prepositions under four letters in length. Surface Roughness and the Kirchoff model Figure 1.- nsertion losses of the 85 khz used transducers. The ripple superposed is due to the resonances of the air column. The surface roughness can be described using a one-dimensional model [9]. Be the surface of the material represented by a cross cut as represented in Figure 3 where h(x) is the height measured with respect to their average. 64

3 This function verifies: L ( x) h dx Eq. The parameter that characterizes this behaviour is the spatial correlation function CR defined by h( x) h( x + X ) C R ( X ) σ Eq.5 Figure 3: Definition of surface height. From the function h(x), the roughness Rrms (Root Medium Square) is defined as : 1 Rrms h ( x) dx L Eq.3 This parameter gives average information. Another description parameter of the roughness is its space surface distribution. Periodic surface distribution is usual in industrial machine processes. Random surface distribution is common in natural processes when a great quantity of random events occurs. Very frequent random patterns are described by the Gaussian distribution, which will be taken as work hypothesis for the samples used in this paper. The probability distribution p(h) will then be given by p( h) 1 h exp σ π σ Eq.4 where σ is the standard deviation When an incident sound wave reaches to a surface with acoustic impedance different from the media in which the wave propagates, one part of the energy is transmitted through the surface and another is reflected. Each point of the reflecting surface can be considered like a source of small waves, forming the reflected total acoustic field (Huygens Principle). The specification of the rms roughness does not give information about the spatial periods or the scale in which the portions of the surface have a relation to neighbouring regions. For a smooth surface, it exists one direction in which all these wavelets are in phase. This is the specular direction and the reflected angle equals the incident angle with respect to the normal of the surface. For the other directions a destructive interference between the waves appears and energy does not propagate. f the zone illuminated by the wave have finite extension then it exists diffraction with a main lobe around the specular direction. The lobe size depends on the wavelength and the size of the illuminated zone. The field produced by these waves in phase is called the coherent field. For a rough surface the phase of the reflected waves cannot be predicted without the knowledge of the surface profile. Reflections in different directions take place without fix phase relation with respect to the incident wave. This component of the reflected field is called the diffuse or incoherent field. Figure 4: Reflected energy. A smooth surface, energy is reflected in specular direction and the field is coherent. B less rough surface, there s two components one coherent and the other diffuse. C very rough surface, the field is diffuse. n this way when a reflection at a rough surface takes place, the total reflected intensity is formed by two terms, one coherent and another incoherent. + total coherent incoherent Eq.6 Kirchhoff's theory or physical optics approximation is one of the most used to study rough surfaces [8]. This is because it has a simple physical base and, in some practical 65

4 situations, it gives analytical expressions to calculate the intensity of the reflected waves. The fundamental physical hypothesis is to consider the surface locally flat, which is called the approach of the tangent plane. Other additional hypothesis for the field reflected at a rough surface are: Far field approximation: the relation between the transducer radius r and the emitted wavelength λ satisfies r > 1 λ Eq.7 Plane monochromatic wave The surface is gently undulating, the radius of curvature is large compared with the wavelength. The heights follow the normal distribution, also called Gaussian. When it has not privileged directions about a point is called isotropic. The illuminated area is greater than the correlation length of the surface. Acusto-electrical response. The acusto-electrical response measurement is an habitual technique to evaluate ultrasonic transducers performance. This technique consists of applying a short pulse (like a Dirac delta) to the transducer receiving the echo from a smooth parallel plane reflector placed in front of it. The received signal y can be expressed by a convolution product of several factors y He D Hr v Eq.9 Here, He is the emission transfer function of the transducer, D includes the diffraction term and the reflection coefficient at the surface, Hr is the reception transfer function and v is the excitation electrical signal. From here we can obtain the frequency spectrum of the transducer. Figure 5 shows the experimental acoustoelectric response obtained from the above described transducers. Following this model, the expressions for the coherent and incoherent intensity are coherent incoherent ω R exp c CR c 4 ω R rms S rms Eq.8 Where is the intensity measured at a smooth reflecting plane (where there is single coherent reflection) ω is the angular frequency of the emitted pulse, S the illuminated region of the surface and c the speed of sound, 15 m/s in water and 34 m/s in air. Experimental results n this work we present three experimental results. First the air-coupled transducer characterization by the acusto-electrical response, second the measure of a complete set of rough surface samples using the aircoupled transducer and third the measure of the same set of samples in water using a commercial immersion piezoelectric transducer. The acoustic signals are emitted and received by a Sonic FTS Mark V, commercial pulse-echo NDT equipment. A HP546 1 MHz oscilloscope were used to digitize the signals Figure 5: Acousto-electrical response The sample frequency in this experience is 1 MHz, allowing the reconstruct ion of the spectrum to a maximum frequency of 5 MHz - Nyquist theorem. Figure 6 shows the frequency spectrum of the time signal shown in Figure 5. As it can be observed, the transducers present a 45% frequency bandwidth. This bandwidth allows to generate relatively short pulses as shown in figure 5. The centre of the band is located at 8 khz. Roughness measurement. A selected set of samples were used to measure the surface roughness. Next table describes the samples and shows the estimated roughness measured by a stylus profilometer. The measured roughness varies from cero to 1 μm. 66

5 Figure 6: frequency spectrum of the acousto-electrical response. TABLE Description of the materials used to test the method and the roughness measured by contact profilometry. ndex Description Roughness [μm] 1 Glass plate less one Less 17 corrosion steel 3 Medium 34 corrosion steel 4 Fine 51 sharpening 5 Medium 68 sharpening 6 Gross 86 sharpening 7 Fine grinder 13 8 Gross grinder 1 Figure 7: Normalized amplitude, air-coupled transducer. A gradual amplitude diminution is observed when the sample roughness increases. These experimental data were then used to estimate the surface roughness via the Kirchoff model. ncoherent or diffuse term is small compared with the coherent one in the experimental cases considered in this work. This term is then not considered here. nverting the Kirchoff`s formula, the correspondent roughness values are obtained. The inversion formula is: R rms c ln ω Eq.1 The ratio / is plotted in figure 8, ω is the angular frequency near 4.6 Mrad/s and the sound speed c 34 m/s To decrease the uncertainty of the acoustic measurement, a region of 5x5 mm was selected for each sample, scanning the square area with 1mm steps. So, 5 echoes were recorded for each sample. The mean value of the echo signal amplitude over the matrix is assigned as the surface echo value and de standard deviation is used as an error limit. Figure 7 shows the results obtained using the air-coupled transducer. The amplitude is normalized by the plane echo amplitude from the glass plate sample. Figure 8: Roughness data fit using the Kirchoff model n the next the same samples were measured using a commercial immersion transducer (Panametric V5MHz). The fluid used was water. The selected transducer has a 5 MHz central frequency. We use this frequency because in the usual commercial range of frequencies, this the one with the closest wavelength 67

6 transducers are useful for industrial applications when coupling fluids cannot be used. The Kirchoff model is a good tool to predict the roughness from ultrasonic pulse-echo techniques in the range of low and medium roughness. Figure 9: Normalized amplitude, watercoupled transducer. Figure 11.- Air coupled roughness measurements *.Water coupled o. Gray line shows the measured roughness using a stylus. References [1].-J. Buckley. Air-Coupled Ultrasound - A Millennial Review. WCNDT Rome. Figure 1: Roughness data fit using Kirchoff model. Finally, Figure 11 shows the comparison between the predictions made with the aircoupled transducers and the corresponding with the commercial immersion transducer. n the same figure, the value of the roughness measured with the contact profilometer are depicted grey line-. t can be observed that the ultrasonic method is useful but also that the measurements from air-coupled ultrasound are still better than those made with the immersion technique. Conclusions n this work the viability of using air-coupled ultrasonic transducers based on PZT-polymer piezocomposites for the measurement of surface roughness is shown. Specially designed 8 khz air-coupled transducers were made. The technique was tested with samples having corrosion standard roughness up to 1μm. The comparison with the same measurement made with commercial immersion transducers shows that air coupled ultrasonic [].-H. Loertscher, B. Grandia, J. Strycek, and W. Grandia. Airscan Transducers, Technique and Applications. Ndt.net Sept [3].-W. Grandia, and C. Fortunko. NDE Applications of Aircoupled Ultrasonic Transducers EEE Ultrasonic Symposium, Proceedings, Vol 1, pp , SSN [4].-G. Blessing, P. Bagley and J. James. The Effect of Surface Roughness on Ultrasonic Echo Amplitude in Steel. Materials Evaluation, Vol 4, October 1984, pp [5].-B. Auld.. Acoustic fields and waves in solids. Krieger Publishing Company, second edition 199 [6].- Y. Gómez-Ullate and F. Montero de Espinosa Ultrasonic Air-Coupled Metrology of Material Surfaces.Proc. 4 EEE Ultras. Symp., 98-31, 5 [7].-G. Blessing, J. Slotwinski, D. Eitzen and H. Ryan. Ultrasonic measurements of surface roughness. Applied Optics, Vol. 3, No 19, July 1993, pp [8].-G. Blessing and D. Eitzen. Ultrasonic Sensor for measuring Surface Roughness. Surface Measurement and Characterization, Proc. SPE, Vol. 19, Sep1988, pp [9.]-J. Ogilvy. Theory of Wave Scattering From Random Rough Surfaces. Hilger, Philadelphia, Pa