Bild von Institut mit Unilogo
home uniuni suchesuche sitemapsitemap kontaktkontakt
unilogoUniversität Stuttgart
Home | Institut | Lehre | Forschung | Stellen | Aktuelles

    Detektion von Kontaminationen (KombiSens Project)

 Institut für Technische Optik


project logo
Measurement task
Measurement principle
System configuration
Consortium
References & Links

Manufacturing of metallic parts includes several mechanical processing such as cutting, bending, shaping etc. All these operations are performed in short time due to mass production requirements. One of the most important quality parameters is the cleanness of produced parts. The issue includes the presence of particles and oils on its surface, which could decrease or even make impossible proper usage of these elements.

The goal of project KombiSens  is design of modular sensor for contamination measurement. The general concept includes existing of two or three main modules which could be flexibly added to the body of measurement system. Each module correspond to specific measurement task, e.g. in-line measurement or direct examination with hand-operating measurement head. Institüt für Technische Optik is responsible for design of the system for high speed and resolution measurement of particles and oil films in real time (in-line module).

 

Measurement task

The measurement task is detection and measurement of contaminations on mechanical parts surfaces. Two main types of contaminations  should be concerned – particles and oil films (Fig.1). The size of particles is in the range of 10µm - 500µm, while the thickness of oil films varies from 1µm to 50µm.
The tested part is placed on the moving belt which travels with speed in the range of  0.05m/s  - 0.3m/s. Roughness of the surface is defined according to following parameters:  Ra=0.1 - 1µm,
Rz = 0.5 - 10 µm, the width of area under test is 20mm.The maximum price of the system is 10000 €. 
measurement task
Fig.1. Measurement conditions for contamination detection system.

Existing measurement systems which are most relevant to this application are systems for inspection of printed circuits boards (PCB) [1,2,3]. Their dual task is first quality control of printed paths (continuity, proper thickness) and second soldering and elements position monitoring. It includes region detection and segmentation and pattern recognition image processing. Often linear sensor as the detector when high speed measurement is needed [4,5]. Three dimensional (3D) measurement systems which give information about surface topology and microshape are probably the most robust for both precise particle and oil thickness measurement, however their speed is not sufficient when the upper range of moving belt velocity is considered [6].  The technique of region spectral segmentation is directly applicable to oil detection system  (section 4) [3].

The ParticleGuard system developed by the member of KombiSens project consortium Advanced Clean Production GmbH  enables detection of particles on the tested surface using hand - operating measurement head. The minimal detectable size of particle is 2 µm. The sensor includes oblique illumination system and CCD array camera. Due to its application (static testing of different surfaces) the processing is relatively slow, single measurement cycle takes 30ms.  

Measurement principle

Typical measurement method for detection of particles is observation of the tested surface illuminated obliquely (also called dark field technique). In this case only the light which is reflected by the particle can reach the detector (Fig 2). This technique is applied in ParticleGuard developed by acp GmbH (section 2). The amount of light is strictly connected with the dimensions of the particle, therefore the resolution limit is connected with the sensitivity of the detector. For example the image intensifier must be used to detect 38 nm particles on LSI wafers [7]. In case of dynamic measurement the amount of light is limited by the time of acquisition, which corresponds to speed and resolution required. Another consequence of movement of the sample is application of  line scan camera as the detector. 



Thin film detection and quantitative measurement, including thickness and refractive index determination,  can be performed within two techniques: interferometry or ellipsometry. Both of them require static measurement conditions and isolations from vibrations.  Dynamic, high speed interferometric measurement is possible only when pulse laser is applied, but the disadvantage is high cost of the system, considerably exceeding required amount of 10 000 €. Ellipsometry is also sensitive for roughness of tested surface (see paragraph 5 for details). For these reasons only qualitative classification (yes/no) can be achieved. The measurement principle in this case is based on the of difference of indices of reflection between metallic and oil surfaces, combined with their chromatic properties. The illumination and observation angles are equal. While the surface is illuminated sequentially within two wavelengths, the index of reflection is determined and therefore the regions contaminated by oil are detected.

principle

Fig.2. Measurement principle for particle and oil detection 

System configuration

The measurement system consist of two channels: particle and oil film detection (Fig.3). In particle measurement channel red light (l0=625nm), LED is used, since the CCD camera have its sensitivity maximum in the red region of spectrum. In film measurement channel, in basic configuration, blue (470nm) LED is used. However additional red LED (625nm) is also built as the alternative source. This solution provides more flexibility, for example if more complex spectral analysis are required due to presence of surface with different spectral characteristics than already tested specimens  [11,12]. The polarizer combined with analyzer are used to minimize reflections from the edges of metallic parts. The spectral filter F is necessary because of relatively strong disturbing illumination in industrial environment. It is designed for two wavelengths used in the measurement within two channels.

system configuration
Fig.3 Configuration of the measurement system, L1, L2, L3 – lenses, P1, P2 – linear polarizers, A – analyzer, F – filter, FG – frame grabber, PC – personal computer.

The line scan camera Dalsa Piranha 1024 2P2 2x is 1024 pixels, 8 bit, high sensitivity camera with maximum line scan frequency 67kHz. Within 30kHz scanning and the moving belt velocity 0.3 m/s it provides resolution 10µm in x direction. The number of  pixels (1024) determines resolution in y direction (20mm wide measurement field) at the level of 19.5µm. All resolution parameters given here correspond to single channel measurement. The data acquisition transfer for this speed and resolution is 245Mb/s, which is high value concerning the necessity of further processing and export of the results in real time. The advanced frame grabber MVTitan-CL (Matrix Vision GmbH) has 2 inputs camera link interfaces, maximum data transfer 800Mb/s, 8MB local memory, and provides several image processing operations in the real time. The PC computer is Pentium 4 class, 2.4 GHz processor with 1GB RAM.  

Consortium

Public funding

Bundesministerium für Wirtschaft und Arbeit


externer LinkVDI VDE-IT
www.vdivde-it.de/innonet

externer LinkCoordinator
Fraunhofer Institut für Produktionstechnik und Automatisierung
www.ipa.fhg.de

externer LinkIndustrial partners
acp  - advanced clean production
www.acprod.com

externer LinkImplementa
www.implementa.com

externer LinkMAFAC Ernst Schwartz GmbH & Co.KG
www.mafac.de

externer LinkOTB Oberflächentechnik in Berlin GmbH & Co.
www.otb-group.com

References & Links



[1] Moganti-M et al. “Automatic PCB inspection algorithms: a survey”, Computer Vision and Image Understanding, USA: Academic Press, vol.63, no.2, 287-313 (1996).
[2]Y.Hara et al, “A system for PCB automated inspection using fluorescent light”, IEEE Transactions on pattern analysis and machine intelligence, vol.10, No.1, 69 – 78 (1988)
[3] S.Tominga, “Region segmentation by multispectral imaging”, proc. of 5th IEEE Southwest Symposium on Image Analysis and Interpretation (SSIAI’ 2002),  (2002)
[4]H.Yoda et al, “An automatic wafer inspection system using pipelined image processing techniques”, IEEE Transactions on pattern analysis and machine intelligence, vol.10, no.1, 4-16 (1998)
[5]P.Kim, S.Rhee, “Three dimensional inspection of ball grid array using laser vision systems”, IEEE Transactions on Electronics Packaging Manufacturing, vol. 22, No 2, 151-155 (1999)
[6]Y. Matsuyama, et al. Automated solder joint inspection systemusing 3D image detction”, proc. of the 3rd IEEE Workshop on Applications of Computer Vision (WACV’ 96), 116 - 122 (1996)
[7]M. Noguchi, Y. Kembo, “Detecting fine particles on LSI wafers”  Jpn. J. Appl. Phys., vol. 32, 352-357 (1993)
[8]Y.Huang, U.Liang, “Interferometric oil spill detection”, Optical Engineering, vol.40, no. 5, 740-745 (2001)
[9]Born, Wolf, “Principles of optics”, Pergamon Press, Oxford 1980
[10]T.Germer “Application of bidirectional ellipsometry to the characterization of roughness and defects in dielectric layers”, Proc. SPIE,  vol.3275, 121-31 (1998)
[11]P.Camagni et al, “Fluorescence response of mineral oils: spectral yield vs absorption and decay time”, Applied Optics, vo. 30.no. 1, 26 – 35 (1991)
[12]A.Barbaro et al. “Oil UV extinction coefficient measurement using standard specrometer“, Applied Optics, vol. 30, no.7, 852 – 857 (1991)

       

Links

externer LinkWeb page of Kombisens Project at VDI VDE - Innonet
Unser Online-Angebot enthält Links zu anderen Websites. Wir haben keinen Einfluss darauf, dass deren Betreiber die Datenschutzbestimmungen einhalten.
© Institut für Technische Optik zuletzt geändert: Aug 21 18:31:55 CEST 2002
 		 
Letzte Änderung 10.02.2010 (Mail towm) | © Universität Stuttgart | Impressum