2016 3 rd International Conference on Engineering Technology and Application (ICETA 2016) ISBN: 978-1-60595-383-0 Application of Golay Coded Pulse Compression in Air-coupled Ultrasonic Testing of Flexible Package Seal Defect Yisen Liu & Songbin Zhou* Guangdong Institute of Automation, Guangzhou, Guangdong, China Wei Han & Kejia Huang Guangdong Key Laboratory of Modern Control Technology, Guangzhou, Guangdong, China Chang Li & Weixin Liu Guangdong Public Laboratory of Modern Control and Manufacturing Technology, Guangzhou, Guangdong, China ABSTRACT: Air-coupled ultrasonic testing is a possible implementation method for online seal defect detection of flexible package. However, large acoustic impedance difference between air and solid material leads to difficulties in coupling ultrasonic energy into solid materials and results in a poor signal to noise ratio (SNR). To solve this problem, pulse compression technique based on Golay codes has been applied in air-coupled ultrasonic testing of flexible package seal defect. The effects of different Golay codes (4, 8 and 16 bits) and Golay codes with different bit length (1, 2, and 3 circles) are investigated to obtain a reasonable parameter selection. The results demonstrate that the Golay coded pulse compression technique can improve the SNR ratio of the received signal remarkably, and suppressed the side lobe of the compressed pulse effectively. An improvement in SNR of the seal region image up to 9.4 db is achieved, compared with the non-modulated excitation method. Keywords: Golay code; pulse compression; air-coupled ultrasonic; seal defect 1 INTRODUCTION Flexible Package has been widely applied in the industry of food, medical products and daily chemical. Various types of seal defects, such as bubble, contamination and channel leak, will have serious impact on product quality, especially the channel leak defects. Currently, detection of defects in the seal region of flexible packages is accomplished through visual inspection or through destructive testing, which is time consuming and cannot achieve online detection in the manufacturing process. Ultrasonic testing is one of the nondestructive testing methods of flexible package seal quality developed in recent years. A number of works have been reported on experiments and imaging algorithms. However, in traditional ultrasonic testing, immersion ultrasound transducers were used and the seal region *Corresponding author: sb.zhou@gia.ac.cn of flexible package needs to be immersed into fluid couplant. Therefore, it is still an off-line test method. Air-coupled ultrasound allows testing and evaluating materials in non-contact manner, which has shown bright prospects in testing of composite material, food and package quality. Unfortunately, ultrasonic attenuation in air and large acoustic impedance difference between air and solid material lead to difficulties in coupling ultrasonic energy into solid materials. Poor signal to noise ratio (SNR) is the critical problem that limits the applications of air-coupled ultrasonic testing. Approaches including transducer design, high voltage power amplifier and signal processing technique have been investigated to solve the SNR problem. Pulse compression is one of the effective signal processing techniques that can be utilized to improve the SNR without increasing the emitting power. Pulse compression technique was first used in radar signal processing, and has made great progress in sound applications in recent years. In this paper, pulse com- 667
pression based on Golay coded excitation was used for improving the SNR of air-coupled ultrasound. The effects of different Golay codes and the bit length of Golay codes were studied. Then the Golay code with reasonable parameter was used to imaging channel leak defect of flexible package in non-contact manner. 2 GOLAY CODED PULSE COMPRESSION METHOD Pulse compression increases SNR by emitting a long coded pulse and correlates the received signal with the input pulse. Chirp and binary are two major encoding types in pulse compression technique. Frequently used binary codes include Barker codes, pseudo-random PN sequences, and Golay complementary pairs. Golay complementary sequences fill with one circle sine-wave per bit. The auto-correlation functions of the A code and the B code and their sum are also shown in this figure. Being excited by a pair of Golay complementary sequences, the received signals can be defined as convolution between the emitted Golay sequences and the transducer response during emission and reception (see Eq. (2)). (2) and are the received signals of a pair of Golay pulses, e and are the A code and the B code of the Golay sequence, and and are the transducer response during emission and reception, respectively. Use the Golay sequence as matched filter in pulse compression process, then the sum of auto-correlation functions of received Golay pulses can be represented as: R R R R (3) where R and are the auto-correlation of the Golay sequence. It can be seen in Eq. (3) that the side-lobes can be effectively eliminated by involving Golay complementary sequences as excitation and matched filter in pulse compression process. 3 APPARATUS AND EXPERIMENT Figure 1. (a) 16-bit Golay sequences fill with one circle sine wave per bit (b) Auto-correlation function of the 16-bit Golay sequences (c) Sum of auto-correlation function. The work principle of Golay complementary sequences is displayed as follows. Let A[n] and B[n] be a pair of the binary sequences of length N. The sum of auto-correlation functions of A[n] and B[n] has a main peak and zero side-lobes. The characteristic of Golay sequences can be represented as: 2, 0 0, 0 (1) where R and are the auto-correlation functions of A[n] and B[n] respectively. For example, a pair of 16-bit Golay complementary sequences can be written as [-1, -1, 1, -1, -1, -1, -1, 1, -1, 1, 1, 1, 1, 1, -1, -1] and [-1, -1, 1, -1, -1, -1, -1, 1, 1, -1,- 1, -1, -1, -1, 1, 1]. Fig. 1 illustrates the waveform of a 16-bit The schematic diagram of experimental setup is shown in Figure 2. The air-coupled inspection system is composed of a pair of spot focused air-coupled ultrasonic transducers with a beam diameter (BD) of 1 mm, a personal computer, a signal generator (Agilent 33210A), a power amplifier (Trek 2100HF), a small signal amplifier, a DAQ card (Spectrum M2i.4021 -exp), and an electric scanning stage. Figure 2. Schematic of the air-coupled testing system. The resonant frequency of the transducers is 1 MHz, and the bandwidth is about 150 khz. In the study of 668
pulse compression scheme selection, there was no sample placed between the transducers. Golay pulses were generated by the signal generator and amplified to 25 Vp-p by the power amplifier, then emitted by the transducer. The received signals from the receiving transducer were amplified with a 40 db gain and collected by the DAQ card. The personal computer controlled the whole inspection process and was responsible for the pulse compression of received signals. In the study of seal defect inspection, the excitation voltage was increased to 50 Vp-p, and the amplify gain of received signal was increased to 60 db to obtain enough signal strength. The measured sample was a seal region of polyethylene package with a 1 mm (L) channel leak defect. The sample was placed between the transducers and the transmitted signals were collected. Transducers scanned on the X-Y surface driven by the electric scanning stage, to form an image of the seal region. The size of scanning area is 30 mm 5 mm, with a 0.25 mm scanning step for x direction (Sx) and a 1 mm scanning step for y direction (Sy) (see Figure 3). Figure 4 shows the received signals corresponding to Golay pulses excitations in 4-bit, 8-bit and 16-bit respectively. The excitation is different Golay pulses fill with one circle sine-wave per bit, which can be seen in the top right corner of the figures. The received signals are broadened in time domain and have increased amplitudes with the increasing of excitation bits. The peak values of received signals are 0.57 V, 0.72 V and 0.74 V corresponding to excitation in 4-bit, 8-bit and 16-bit respectively. The received signals of A codes are not shown in the figures because they are very similar to those of B codes. Figure 5. Compressed pulses corresponding to excitation in 16-bit: (a) correlation of A code (b) correlation of B code (c) correlation sum of A code and B code. Figure 3. Scanning strategy of the flexible package seal region. 4 RESULT AND DISCUSSION 4.1 Selection of pulse compression scheme Figure 5 shows the pulse compression process of the received signals excited by the 16-bit Golay codes. The received signals corresponding to A code and B code were correlated with the corresponding excitation signals. As shown in Figure 5(a) and (b), the correlation signals have much higher amplitudes than the original signals. However, the side-lobes are obvious in these two figures. The correlations of A code and B code were summed and shown in Figure 5(c). The amplitude of main peak is further increased while the side-lobe is effectively suppressed due to the complementary property of Golay code. Figure 4. Received signals corresponding to different Golay coded excitation: (a) 4-bit (b) 8-bit (c) 16-bit. Figure 6. Envelopes of correlation sum corresponding to different Golay coded excitation: (a) 4-bit (b) 8-bit (c) 16-bit. 669
Figure 6 shows the envelopes of compressed pulses with different Golay coded excitations. It can be clearly seen that the signal amplitudes significantly increase and the signal widths become much contracted, compare with the original received signals. With the increase of excitation bits, the amplitudes of the main peak become larger, which means a better SNR. On the other hand, the main peaks of compressed pulses are slightly narrowing with the increase of excitation bits. The amplitude gains in pulse compression process are 34.6 db, 36.5 db and 42.2 db, corresponding to excitation in 4-bit, 8-bit and 16-bit respectively. The increase in amplitude with excitation bits can be attributed to two reasons: the larger auto-correlation index and the stronger received energy from broadening in time duration of excitation. To obtain a better SNR, 16-bit Golay complementary sequences were chosen for the following investigation. Figure 7. Received signals corresponding to excitation with different bit lengths: (a) 1 circle (b) 2 circles (c) 3 circles. The impact of bit length was also studied. The received signals corresponding to excitation in 16-bit Golay pulses with 1, 2 and 3 circles sine wave per bit are shown in Figure 7. The received signals broaden in time domain and have increases in amplitude with the increase of bit length. The peak values of received signals are 0.74 V, 0.86 V and 0.95 V for bit length of 1, 2 and 3 circles, respectively. The envelopes of compressed pulses with excitations with different bit lengths are shown in Figure 8. The main peak amplitudes of compressed pulses increase as the bit length increase from 1 circle to 3 circles. The voltage amplitude gains in pulse compression process are 42.1 db, 50.8 db and 53.0 db, corresponding to bit length of 1, 2 and 3 circles respectively. It should also be noticed that the peak widths are slightly broadened with the increase of circles, especially from 2 circles to 3 circles. The side-lobe becomes obvious and separates from the main peak while bit length increases to 3 circles. Figure 8. Envelopes of compressed pulses corresponding to excitations with different bit lengths. The reason for the increasing amplitude of compressed pulses along with the increase of bit length is the bandwidth matching between excitation and transducers. The bandwidth of transducers used in this work is about 150 khz. Figure 9 shows the frequency spectra of the excitation signals with different bit lengths. While the bit length is one circle, the bandwidth is much wider than the bandwidth of transducers. As a result, the excitation energy cannot be effectively used, and leading to a lower peak value of the compressed pulse. With the increasing of bit length, the bandwidths of excitation signals become narrower, as a result, have better matching situations with the transducers. However, increasing bit length makes the excitation more time-consuming and would reduce the temporal resolution and the scanning speed in air-coupled ultrasonic testing. Hence, we chose the 16-bit Golay complementary sequences with 2 circles per bit as the excitation in the following seal defect inspection experiment. Figure 9. Frequency spectra of the excitation signals with different bit lengths: (a) 1 circle (b) 2 circles (c) 3 circle. 670
4.2 Imaging of the flexible package seal defect A polyethylene package with a 1 mm channel leak defect was placed between the emitting and receiving transducers. As a comparison, the transducers are also excited by non-modulated sine pulses and the received transmitted signals are shown in Figure 10(a). The excitation voltage was increased to 50 Vp-p and the gain of received signal was increased to 60 db to obtain enough signal strength. However, it can be seen in Figure 10(a) that the transmitted signal is severely attenuated due to the interfacial reflection of package sample, and the received signal is almost submerged by noises. The received signal was band-pass filtered within 0.9 MHz to 1.1 MHz and shown in Figure 10(b). The transmitted signal still cannot be easily distinguished from noises. The SNR of the band-pass filtered signal is 8.9 db. Fig. 10(c) shows the envelope of compressed pulse which excited by 16-bit Golay sequences with a bit length of 2 circles. A main peak can be clearly seen in the compressed pulse. The SNR of this signal is 16.6 db, which has been dramatically improved compared with Figure 10(b). Figure 10. (a) Transmitted signal excited by non-modulated sine pulses, (b) band-pass filtered result of (a), (c) the envelope of correlation sum which excited by the 16-bit Golay sequences with a bit length of 2 circles. The scanning images of flexible package seal region with different excitation methods are shown in Figure 11. As we can see in Figure 11(a), the image has a severe noise while the received signal amplitude with non-modulated excitation was regarded as the imaging characteristics. It s hard to recognize the channel leak defect in this image. The received signals were band-pass filtered and formed Figure 11(b). The image SNR is improved and the channel leak defect can be recognized, although the profile of the channel defect is not clear. The scanning image forming by pulse compression method is shown in Figure 11(c). It can be seen that the image SNR is improved compared with Figure 11(a) and 11(b). The noise in background is effectively suppressed and the profile of the channel defect is clear. The image SNR of Figure 11(a) to Figure 11(c) is 3.7 db, 8.5 db and 13.1 db, respectively. Figure 11. Seal region scanning images forming by different methods: (a) signal with non-modulated excitation (b) band-pass filtered signal with non-modulated excitation (c) compressed pulses with Golay codes excitation. 5 CONCLUSION In this paper, it has been demonstrated that the pulse compression technique based on Golay codes improves the SNR of air-coupled testing and helps to realize non-contact seal defect detection. The effects of different Golay codes (4, 8 and 16 bits) and different bit lengths (1 circle to 3 circles) of Golay codes are analyzed to obtain an optimal parameter selection. 16-bit Golay sequences with a bit length of 2 circles were chosen as excitation in the defect detection experiment for the balance of SNR and temporal efficiency. The SNR of scanning image increases 9.4 db, compared with the non-modulated excitation method. Golay coded pulse compression technique contributes a better SNR, an improved accuracy of time measurement, and effective suppression of side lobe. This allows a wide range of applications in air-coupled ultrasound testing that not limited to detection of seal defect, such as composite material examination, elastic properties testing and surface metrology. ACKNOWLEDGMENT The authors would like to acknowledge funding for this work from Science and Technology Planning Project of Guangdong Province (GN: 2013B0911 00013, 2014A010104010, 2015A010103008). REFERENCES [1] Frazier C.H. et al. 2000. High contrast ultrasound images of defects in food package seals. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control. 47(3): 530-538. 671
[2] Gan T.H. et al. 2001. The use of broadband acoustic transducers and pulse-compression techniques for air-coupled ultrasonic imaging. Ultrasonics. 39: 181-194. [3] Gan T.H. et al. 2003. High-resolution, air-coupled ultrasonic imaging of thin materials. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control. 50(11): 1516-1524 [4] Gomez T.E. 2004. Acoustic impedance matching of piezoelectric transducers to the air. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control. 51(5): 624-33. [5] He C. et al. 2009. Defect detection in seal region of flexible packages using ultrasonic non-destructive testing technique. Chinese Journal of Scientific Instrument. 30(4): 750-755. [6] Han X.M. et al. 2005. High frame rate ultrasonic imaging system based-on linear frequency-modulated signal. Chinese Journal of Biomedical Engineering. 24(6): 700-704. [7] Kazys R. et al. 2006. Air-coupled ultrasonic investigation of multi-layered composite materials. Ultrasonics. 44: 819-822. [8] Morris S.A. et al. 1998. New sensors help improve heat-seal microleak detection. Packaging Technology & Engineering. 7(7): 42-49. [9] Morris S.A. et al. 1999. Evaluation of defects in seal region of food packages using the ultrasonic contrast descriptor, ΔBAI. Packaging Technology and Science. 12(4): 161-171. [10] Meyer1 S. et al. 2006. Non-contact evaluation of milk-based products using air-coupled ultrasound. Measurement Science and Technology. 17: 1838-1846. [11] Misaridis T. & Jensen J.A. 2005. Use of modulated excitation signals in medical ultrasound. Part I: basic concepts and expected benefits. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control. 52: 177-191. [12] Nowicki A. et al. 2004. Ultrasonic examination of hard tissues using Golay complementary series excitation. Acoustical Imaging. 27: 619-626. [13] Nowicki A. et al. 2004. Ultrasonic examination of hard tissues using Golay complementary series excitation. Acoustical Imaging. pp: 619-626. [14] Ozguler A. et al. 1998. Ultrasonic imaging of micro-leaks and seal contamination in flexible food packages by pulse-echo technique. Journal of Food Science. 63(4): 673-678. [15] Ozguler A. et al. 2001. Ultrasonic monitoring of the seal quality in flexible food packages. Polymer Engineering and Science. 41(5): 830-839. [16] Ricci M. 2012. Exploiting pseudorandom sequences to enhance noise immunity for air-coupled ultrasonic nondestructive testing. IEEE Transactions on Instrumentation & Measurement. 61: 2905-2915. [17] Seco F. et al. 2006. Air coupled ultrasonic detection of surface defects in food cans. Measurement Science and Technology. 17: 1409-1416. [18] Song J. et al. 2006. Spherically focused capacitive-film, air-coupled ultrasonic transducer. Journal of the Acoustical Society of America. 119(2): EL1-EL6. [19] Tian Q. et al. 2000. Parametric modeling in food package defect imaging. IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control. 47(3): 635-643. [20] Thiele S. et al. 2014. Air-coupled detection of nonlinear Rayleigh surface waves to assess material nonlinearity. Ultrasonics. 54: 1470-1475. [21] Yoo Y.M. et al. 2001. A low voltage portable system using modified Golay sequences. Proceedings of IEEE Ultrasonics Symposium. 2: 1469-1472. [22] Zhou Z. et al. 2014. Application of wavelet filtering and Barker-coded pulse compression hybrid method to air-coupled ultrasonic testing. Nondestructive testing & evaluation. 29(4): 297-314. 672