Review of techniques for On-line Monitoring and Inspection of Laser Welding
J Shao and Y Yan
Department of Electronics, University of Kent at Canterbury, Kent CT2 7NT, UK
Emails: [email protected], [email protected]
Abstract: Laser welding has been applied to various industries, in particular, automotive, aerospace and microelectronics. However, traditional off-line testing of the welds is costly and inefficient. Therefore, on-line inspection systems with low cost have being developed to increase productivity and maintain high welding quality. This paper presents the applications of acoustic, optical, visual, thermal and ultrasonic techniques and latest development of laser welding monitoring. The advantages and limitations of these techniques are also discussed.
1. Introduction In resent years, laser welding has been widely used in manufacturing from vehicle assembly in automotive production to the joining of microelectric components in electronics industry due to its high speed, non-contact and precision with low heat effects. A primary concern over the industry spectrum is to detect weld defects fast, reliably and cost-effectively. Therefore, a number of on-line inspection systems have being developed to improve weld quality and reduce overall costs.
This paper is to review on-line sensing systems developed over past a few years for real time quality monitoring and inspection of laser welding processes. The paper begins with a brief introduction of process signals during laser welding. Then, sensing techniques such as acoustic and optical emission sensors are presented.
2. Process signals of laser welding The laser welding is a welding process that obtains fusion by directing a highly concentrated beam of coherent light on a very small spot. The laser-material interactions that occur during laser welding emit energy in a variety of forms.
Optical and acoustical process signals can be measured from the emissions of the welding using suitable sensors. Since the signals contain information about the beam-material interaction, welding defects can be detected during the process and recorded for each single work piece. Figure 1 shows a selection of detectable emissions, which can be used as the process signals. The reflected laser is the amount of the radiation of the laser source which is not absorbed by the material. Acoustic emissions, which can be divided into air-borne and structure-borne emissions, are originated from the stress waves induced by changes in the internal structure of a work piece, the surface the metal vapour and laser beam back reflection. The metal vapour and the molten pool emit continuous radiation which spectrum varies with different laser applications. For instance, during Nd:YAG-laser spot welding, the process radiation is in the visible and infrared range [1]. For CO2 laser keyhole welding, the plasma generated is known to emit light with a wavelength between 190nm and beyond 400nm, and the spatter emits light with a wavelength between 1000nm and 1600nm [2]. In addition, the geometrical
Institute of Physics Publishing Journal of Physics: Conference Series 15 (2005) 101–107doi:10.1088/1742-6596/15/1/017 Sensors & their Applications XIII
101© 2005 IOP Publishing Ltd
parameters of the keyhole and melt pool also contain useful information which can be used to inspect the welding quality.
It is high desirable to diagnose laser process quality using these emissions, in particular, to understand the relationship between emission characteristics and weld quality characteristics.
Laser beam
Mirror
Metal vapour & plasma
Geometrical parameters of keyhole and melt pool
Process radiation emitted by:- Work piece- Melt pool- Metal vapour & plasma
structure-borne emission
Reflected laser
Air-borne acoustic emission
Lens
Nozzle
Laser beam
Mirror
Metal vapour & plasma
Geometrical parameters of keyhole and melt pool
Process radiation emitted by:- Work piece- Melt pool- Metal vapour & plasma
structure-borne emission
Reflected laser
Air-borne acoustic emission
Lens
Nozzle
Figure 1. Process signals during laser welding
3. Review of the sensing techniques 3.1 Acoustic emission techniques Acoustic emission (AE) involves a sensor which converts process sounds into electrical output to a measurable variable. Air-borne emission has the human audible of ranges between 20Hz and 20kHz. The typical sensor for this emission is a microphone placed nearby the weld zone (Figure 2). The frequency of structure-borne emission is usual from 50kHz to 200kHz. Piezoelectric transducers are typically mounted at acoustic mirror, acoustic nozzle and work piece to measure AE (Figure 2).
AE transducer
Laser beamMirror
AE transducer foracoustic nozzle
AE transducer foracoustic mirror
Lens
MicrophoneMetal vapour &
plasma AE transducer
Laser beamMirror
AE transducer foracoustic nozzle
AE transducer foracoustic mirror
Lens
MicrophoneMetal vapour &
plasma
Figure 2. Typical experimental setups for acoustic emissions
The application of AE to the monitoring and characterisation of laser material processing has been reported frequently in the literature [3-6]. Farson and Kim [4] monitored laser welding using a condenser microphone with frequency response bandwidth ranging from 20Hz to 20kHz. The microphone was located 165mm from the plume. The signals were recorded with a computer data acquisition system at a sampling rate of 50kHz and were used to evaluate the analytical model of acoustic emission of the laser welding developed by the authors. Steen and Weerasinghe [5] investigated the ultrasonic acoustic emissions based on the study of signals mounted on the back of a beam guide mirror, i.e. acoustic mirror. Their work showed that there was a slight rise in the acoustic
102
signal as the keyhole began to fail. Gu and Duley [6] used an omni-directional microphone with a flat frequency response between 20kHz and 500kHz to observe the resonant acoustic emission during laser welding of metals. Li [3] mounted broad band (100kHz to 1MHz) ultrasonic piezoelectric sensors on the back of beam guide mirrors and on a metallic plate mounted on the laser processing nozzle by silicon vacuum grease. The work piece was placed at various locations relative to the laser beam focal plane and laser conditions were varied to examine the heating, melting, vaporisation, plasma generation and keyhole generation on the response of various acoustic sensors. He found that the acoustic mirror signal is generated mainly by the melt pool modulated laser beam back reflection (between 100kHz and 600kHz). The signal is strongest on the beam guide mirror closest to the work piece. When a keyhole is generated, the acoustic mirror signal becomes weaker, while the acoustic nozzle (principally around 100kHz) signal becomes stronger.
AE techniques are still investigated in academia, however, applications in this field have declined due to their limitations. For structure-borne emission monitoring, contact installation of transducers is needed, therefore, it is not suitable for mass production. For air-borne emission inspection, it is often difficult to acquire useful signal from hostile and noisy environments in industry. 3.2 Optical detector Optical detector, particularly ultraviolet (UV), visible or infrared (IR) detectors, has been widely used to converted the flux density of the radiation emitted by the welding process into an electrical signal. Optical filter is often placed in front of the detector to confine the spectral ranges of the whole sensor system. Typical setups using co-axial and off-axial arrangements are illustrated in Figure 3.
Laser beamBeam divider
Lens
Optical detectors (Off-axial arrangement)
Metal vapour & plasma
Optical detectors (Co-axial arrangement)
Laser beamBeam divider
Lens
Optical detectors (Off-axial arrangement)
Metal vapour & plasma
Optical detectors (Co-axial arrangement)
Figure 3. Typical setups for optical detectors using co-axial and off-axial arrangements
Tonshoff et al [8] developed an online process monitoring system. The sensor, which was installed using the co-axial arrangement, consisted of a silicon photodiode and a preamplifier with an adjustable gain. Two strategies were used to detect the defects of the welding. They claimed their system was able to be integrated into the automated production.
Park et al [9] used two UV photodiodes and one IR photodiode with off-axial arrangements to measure the plasma and spatter of the CO2 laser welding for automotive industry. They conducted the experiments under different laser power, welding speed and nozzle position conditions. The relationship between the signals from the photodiodes and the welding conditions were studied. Based on the signal variation, they classified the laser weld quality into the categories of optimal heat input, slightly low heat input, low heat input, partial joining due to gap mismatch and nozzle deviation. Furthermore, they defined the fuzzy rule base and fuzzy membership functions and developed a laser welding quality evaluation system using a fuzzy pattern recognition algorithm. Park and Rhee [10, 11] also used the photodiodes to explain the relationship between the plasma and spatter and bead shape according to the welding variables. Through a correlation between these signals and weld quality, they
103
developed a multiple regression analysis and neural network to estimate the penetration depth and width of the weld bead.
Sforza and Blasiis [12] used four optical detectors, one IR, one UV and two visible detectors, to measure the arc welding process. During the tests, three different signals, i.e. IR, UV and electronic temperature which derived from two visible detectors, were simultaneously recorded. They developed a discrimination algorithm to evaluate the characteristics of the weld and distinguish bad welds from good ones.
Zhang et al [13] developed a sensing system containing two optical sensors to detect the IR and UV waveband of the optical emissions induced in the underwater laser welding. The relationship between the optical signals and the weld quality with various shielding conditions were investigated. Their results showed that the detected signal well reflects the shielding condition variations of the local dry cavity. The optimal shielding condition could be determined by signal stability.
Ostendorf et al [1, 7, 14] investigated the laser spot welding in the micro joining applications. The sensor system was based on a silicon photodiode and a dedicated software package. The optical bandwidth of the photodiode covered from 300nm to 1100nm. Their experimental results showed that the system was capable of detecting and analysing the process emission for laser micro spot welding.
The work, which was untaken by Chen et al [15], Farson et al [16] and Hugel et al [17], also showed that detecting the plasma/plume signal or the reflected laser using optical detectors is a kind of simple and effective way to real time monitor the welding process.
3.3 Other sensing techniques Besides the sensors presented above, there are also other sensors being used for welding monitoring such as visual, thermal and ultrasonic and etc. A brief introduction of their applications will be given in this section.
Beersiek [18, 19] presented a system for process monitoring of laser beam welding based on a CMOS-camera. The system observed the welding process online and coaxial to the laser beam. It was used to investigate the geometrical parameters of the keyhole. Zhang et al [20] conducted measurement of the geometrical appearance of the weld pool using a camera system. The neural network was used to identify the parameters in real time.
Jeon et al [21] studied the surface temperature variation in the laser brazing of a pin-to-hole joint using an infrared radiation sensor. The transient heat flow was analyzed using the finite element method. The effect of process parameters was investigated to enable the prediction of the appropriate process parameters from the measured and calculated results. Bertrand et al [22] applied the bi-dimensional monochromatic and the 1-spot multi-wavelengths pyrometers to monitor the surface temperature in Nd:YAG continuous laser welding. They identified the variation of brightness temperature with operational parameters and detected certain typical welding defects. Bad’yanov et al [23] also presented a system, which consisted of a pyrometer and three Si photodiodes, to monitor and control laser welding. Lim et al [24-26] applied a point infrared sensor for the study of the pulsed laser spot welding.
Li et al [27] used a plasma charge sensor (PCS), as shown in Figure 4, to measure weld penetration and detect weld defects. The plasma behaviour was observed during welding through measurement of the space charge voltage induced on an electrically insulated welding nozzle. They showed that the induced voltage is a measure of plasma temperature and thus of the welding performance. An expert system was used in their system to analyse the PCS signal automatically and classify weld defects. It was reported that a greater than 90% success rate was achieved in detecting and classifying defects in high-speed industrial laser can welding.
Laser ultrasonics was investigated by Miller et al [28], Klein and Bodenhamer [29] and Kercel et al [30]. Their research showed that this technique held a great promise in weld monitoring applications. Wang and Chen [31] applied a voltage Hall effect transducer and a current Hall effect transducer to acquire weld voltage and current. Their experimental results showed the signals varied with the status
104
of the weld pool. Therefore, this low cost and high speed system could be used for the penetration quality of the welding.
Laser beamMirror
Lens
Metal vapour & plasma
Insulating collar
PCS nozzle
RL C
Conductor
Vo
PCS Output
Laser beamMirror
Lens
Metal vapour & plasma
Insulating collar
PCS nozzle
RL C
Conductor
Vo
PCS Output
Figure 4. Schematic diagram of PCS sensor
4. Discussions The study of the previous work reviews that a wide range of sensors has been used to monitor the laser welding processes. Every sensor has its advantages and limitations.
Structure-borne emission sensor has been extensively investigated and appears to be a good choice for detecting material phase transformation and crack formation. However, the sensor needs contact installation with the work piece. Air-borne emission sensor can detect weld surface defects. But it is difficult to be used in noisy and hostile environments. There has few applications of AE reported in microelectronics industry so far. Optical detectors have been widely used over a wide range of industries since they are relatively simple, cheap and effective. Although some commercial systems are available, signal processing and classification would be crucial for the further development of this technique. CCD, CMOS camera or array sensor has been often used to monitor the continuous process. However, it is tricky to apply this technique to the pulsed laser spot welding monitoring as the welding process is dynamic and transient.
To fully exploit the advantages of every sensor, more and more multiple sensor systems have been investigated. It is believed that sensor fusion system in conjunction with latest advance in statistical and artificial intelligence would play an important role in laser welding monitoring and inspection.
References [1] Ostendorf, A, Specker, W, Stallmach, M and Zeadan, J, 3D-MID and process monitoring for
micro joining applications, Proceedings of the society of photo-optical instrumentation engineers (SPIE) 4977, 508-517, 2003.
[2] Ono, M, Nakada, K and Kosuge, S, An investigation on CO2 laser-induced plasma, J Jpn Weld Soc, 10(1992): 239-245.
[3] Li, L, A comparative study of ultrasound emission characteristics in laser processing, Applied surface science, 186 (2002): 604-610.
[4] Farson, D, F and Kim, K, R, Generation of optical and acoustic emissions in laser weld plues, Jurnal of applied physics, 85(1999), 1329-1336.
[5] Steen, W, M and Weerasinghe, W, M, Monitoring of laser material processes, SPIE Proc. 650(1986), 160-166.
[6] Gu, H, P and Duley, W, W, Resonant acoustic emission during laser welding of metals, J Phys. D: Appl. Phys. 29 (1996) 550-555.
[7] Ostendorf, A, Kulik, C, Stallmach, M and Zeadan, J, Basic investigations for controlling the laser spot welding process when packaging 3-dimensional molded interconnect devices, Proceedings of the society of photo-optical instrumentation engineers (SPIE) 5339, 441-447, 2004.
105
[8] Tonshoff H, K, Ostendorf A, Guttler R, Specker W, Online monitoring and closed-loop control of laser welding processes, 12th international symposium for electromachining(ISEM), Aachen, Germany, May 11-13, 1998, 1405 603-612, 1998.
[9] Park, Y, W, Park, H, Rhee, S and Kang, M, Real time estimation of CO2 laser weld quality for automotive industry, Optics & laser Technology 34 (2002) 135-142.
[10] Park, H and Rhee, S, Analysis of mechanism of plasma and spatter in CO2 laser welding of galvanized steel, 31 (1999) 119-126.
[11] Park, H and Rhee, S, Estimation of weld bead size in CO2 laser welding by using multiple regression and neural network, J of Laser Appl, 11 (1999), 143-150.
[12] Sforza, P and Blasiis, D, On-line optical monitoring system for arc welding, NDT&E international 35 (2002) 37-43.
[13] Zhang, X, Chen, W, Ashida, E and Matsuda, F, Relationship between weld quality and optical emissions in underwater Nd:YAG laser welding, Opticas and lasers in engineering 41(2004), 717-730.
[14] Ostendorf, A, Temme, T and Zeadan, Laser spot welding of electronic micro parts, Proceedings of the society of photo-optical instrumentation engineers (SPIE) 5662, 306-312, 2004.
[15] Chen, W, Zhang, X, Bao, J, Wang Y, Li T, Jiang P and Zhang H, Sensing and control system of process quality in CO2 laser deep penetration welding, Proc SPIE 3550, 287-297, 1998.
[16] Farson, D, Ali, A and Sang, Y, Relationship of optical and acoustic emissions to laser weld penetration, Weld Res Suppl, 77 (1998), 142-148.
[17] Hugel, H, Muller, M, Hohenberger, B and Dausinger F,Laser beam welding: resent developments on process conduction and quality aaurance, Proc SPIE 3571, 52-60, 1999.
[18] Beersiek, J, A CMOS camera as a tool for process analysis not only for laser beam welding, 20th ICALEO Vols 92&93, 1185-1193, 2001.
[19] Beersiek, J, On-line monitoring of keyhole instabilities during laser beam welding, ICALEO (R) 99: Section D, 49-58.
[20] Zhang, Y, Kovacevic, R and Li, L, Characterization and real-time measurement of geometrical appearance of the weld pool, Int. J. Mach. Tools Manufact. Vol 36 No 7, 799-816, 1996.
[21] Jeon, M, Kim, W, Han, G and Na, S, A study on heat flow and temperature monitoring in the laser brazing of a pin-to-plate joint, Journal of materials processing technology, 82 (1998) 53-60.
[22] Bertrand, P, Smurov, I and Grevey, D, Application of near infrared pyrometry for continuous Nd:YAG laser welding of stainless steel, Applied surface science 168 (2000) 182-185.
[23] Bad’yanov, B, Elizarov, A and Kolupaev, Y, Application of the step-by-step approximation method for the computer control of laser welding processes, Measruement Techniques, Vol 46, No. 2, 2003.
[24] Lim, D, Cho, Y, Gweon, D, A robust in-process monitoring of pulsed laser spot welding using a point infrared sensor, Proceedings of the Institution of Mechanical Engineers, Part B, Engineering Manufacture, Vol. 212, Issue 3, 241-250, 1998.
[25] Lim, D, Gweon, D, A new criterion for quality monitoring of pulsed laser spot welding using an infrared sensor. Part 1: The radiation feature as a criterion for quality monitoring, Proceedings of the Institution of Mechanical Engineers, Part B, Engineering Manufacture, Vol. 213 Issue 1, 41-49, 1999.
[26] Lim, D, Gweon, D, A new criterion for quality monitoring of pulsed laser spot welding using an infrared sensor. Part 2: Quality estimation using an artificial neural network, Proceedings of the Institution of Mechanical Engineers Part B, Engineering Manufacture, Vol. 213 Issue 1, 41-49, 1999.
[27] Li, L, Brookfield, D, J, Steen, W, M, Plasma charge sensor for in-process, non-contact monitoring of the laser welding process, Meas, Sci, Technol. 7 (1996), 615-626.
[28] Miller, M, Mi, B, Kita, A and Ume, C, Development of automated real-time data acquisition system for robotic weld quality monitoring, Mechatronics, 12 (2002), 1259-1269.
[29] Klein, M and Bodenhamer, T, Laser ultrasonics technology report, www.industrial-laser.com.
106
[30] Kercel, S, Klein, M and Pouet, B, Bayesian separation of lamb wave signatures in laser ultrasonics, Proc. SPIE 4055, 350-361, 2000.
[31] Wang, Y and Chen, Q, On-line quality monitoring in plasma-arc welding, Journal of materials processing technology, 120 (2002), 270-274.
107
Recommended
