Review of reflective fiber optic sensors for surface topography measurement

YANG Rui-feng, HU Chen-hao, GUO Chen-xia, GAI Ting, LANG Guo-wei

(1. School of Instrument and Electronics, North University of China, Taiyuan 030051, China；2. Automatic Test Equipment and System Engineering Research Center of Shanxi, Taiyuan 030051, China)

0 Introduction

With the refinement and diversification of the tools we used in all works, people’s demand for manufacturing technology has gradually increased. In the industrial fields like aerospace, automotive, shipbuilding and mold industries, which are closely related to manufacturing, surface processing quality directly affects their products’ performance, reliability and life. At present, the measurement and inspection of the surface topography are problems frequently encountered by production departments and have not yet been well solved. This is mainly reflected in the fact such as the low efficiency or accuracy of measurement and the complexity of surface texture modeling and processing.

Traditional contact measuring system is slow in measurement, low in efficiency and un-compact in structure, which will cause scratches on the surface. It is not suitable for on-line measurement of fragile materials[1]. Due to the advent of low-loss optical fibers and the development of related sensors, optical fiber sensing technology has been gradually applied to surface measurement. Fiber optic sensors (FOS) have the advantages of electromagnetic interference resistance, low cost, good stability and so on. They possess a good development potential and have a wide range of applications in communication, urban construction, medical, transportation, energy and aerospace[2]. According to the research situation, FOS for surface topography measurement (including displacement, roughness and other parameters) can be classified into three categories: the interference type based on phase modulation, the polarize type based on state modulation of polarization and reflective intensity-modulated FOS (RIM-FOS) type based on intensity modulation[3]. The first two types are expensive, complex and technically difficult to implement or debug. On the contrary, the last type based on intensity modulation is simple, inexpensive, easy to design and also is the earliest optical fiber sensing technique. Therefore many scholars have tried to research its application in surface topography measurement.

1 Principle and performance

1.1 Principle of RIM-FOS

RIM-FOS belongs to the intensity modulation type. It is a device that converts measured parameters (such as distance, curvature, roughness, etc.) into the signal of light intensity. Its basic working principle is that the light from light source is transmitted by the emitting fiber to the probe, and projected onto the surface. The probe is made up of fiber bundle in a certain arrangement. The reflection beam carries the information of the surface to be measured, which is received by the receiving fiber. Then reflection beam intensity is converted to voltage by photosensor. The measured parameters such as displacement are obtained by analyzing numerical changes of the voltage signals under different conditions. Furthermore, surface morphology of the object can be obtained by processing these measured parameters. Figs.1 and 2 show the working principle in the ideal smooth surface and actual rough surface under different visual angles.

Fig.1 Working principle in ideal case

Fig.2 Working principle in actual case

The light modulated by reflector is considered as a mixture of specular reflection and diffuse reflection. The total intensity can be determined by using an empirical formula summarized by Yuan Libo[4]. Because of the length of this paper, here, we just take a single fiber pair with equal core diameter as an example, and some parameters can refer to Fig.1. Let the distance between two optical fibers isb, the cone radius of the reflected light at the fiber end isC, the diameters of the emitting and receiving fiber are both 2r, the numerical aperture isNA=sin(a), the reflectivity of reflector isR0, the distance from the reflector to the input (output) fiber end isd, the light intensity of the receiving fiber and emitting fiber areI(x) andI0,ξis modulation parameter related to light source, fiber coupling and so on. Then, we have

(1)

When the light source is in good coupling with the emitting fiber,ξ≈0, the formula can be expanded and deformed as Eq.(2) by neglecting higher order terms,

(2)

where tana=C/2d.

Obviously, whileCb-r, reflected light enters the receiving fiber and it is inside the reflected light cone, soI(d)≠0. Fig.3 shows the relationship betweenIandd.

Fig.3 Output characteristic curved line of sensor

TheY-axis in Fig.3 represents the luminous flux (LM) of reflected light, which can also be replaced by an electrical signal converted from the output optical power or the ratio of output to input. LM is used to characterize changes in light intensity.X-axis in Fig.3 is the distance between the fiber end and the reference line of the object to be measured, which can reflect the displacement and the roughness of measured surface. Instead of ideal linear variations, the front slope of characteristic curve is narrow and steep while the back slope is wide and slow due to the rough surface. The front slope has good sensitivity and high resolution, but the linear range is small. It is suitable for small changes in high-precision measurement. The back slope has large linear range and low sensitivity. It is suitable for low resolution and large range measurement. The mathematical relationship between surface morphology and the electrical signal converted from light intensity signal can be established, and the whole shape information can be obtained after the scanning measurement.

1.2 Performance comparison

At present, the main methods of non-contact surface topography measurement are laser trigonometry[5], focusing method[6], scanning phase-shifting method[7], interferometry[8]and two microscopy methods in atomic level[9]. The measurement modes, characteristics and accuracy of these methods are shown in Table 1.

Table 1 Comparison of common measurement methods

As seen from Table 1, the first two methods suitable for rapid on-line detection are simple in structure, but the accuracy is not ideal and performance may be unstable. Other methods have high accuracy or sensitivity, but always have the disadvantages of small measurement range and high cost. Whereas, RIM-FOS is very suitable for non-contact measurement when considering cost and sampling rate for its simple principle, reliable performance and fast frequency response. RIM-FOS has the advantages of small size and flexible design, so that it can be well adapted to various measuring surfaces. Due to the characteristics of the output curve, cross range measurement with different sizes can be realized. Its accuracy magnitude can reach about 100 nm under laboratory conditions. But the light intensity signal is very weak, and easy to be affected by light source, circuit noise and the vibration of the platform. There is still room for accuracy improvement. Hence researchers have been working on eliminating interference to improve accuracy and realize its practicability for surface topography measurement as soon as possible.

2 Research history and present status

RIM-FOS is one of the earliest sensing methods that is widely used and maturely researched. Using RIM-FOS for measuring topography and its parameters (such as roughness, curvature, etc.) also can be seen as a micro displacement measurement based on datum plane. And the research results are always based on the output characteristic curved line. Therefore, all the experimental schemes used are same as displacement measurement experiment. As early as 1966, the American Frank W E put forward this reflective fiber-optic sensor and applied for patents[10]. The patent describs a method of measuring blood pressure and heart beat by displacement with just one emitting fiber and one receiving fiber. Meanwhile, Kissinger C D improved the arrangement mode of the fibers in probe[11]. Cook R O et al. have made great achievements in the frequency response, dynamic range, linear range and working distance of RIM-FOS[12]. Previous researchers have analyzed the distribution of reflected light intensity by complex calculations based on the papers of Beck-man and J M Bennett[13-14]. These results laid theoretical foundation for its application in surface topography measurement.

With the development of all aspects, the design of reflective optical fiber sensor is gradually becoming mature. Around 1975, D Spurgeon, G C Lin, W P T North first used reflective fiber sensing technology to measure the quality parameter of surface topography[15-17]. They used simplest single fiber pair to investigate the correlation between reflected light intensity and roughness, because different characteristic curves can be obtained under different rough surfaces. In 1986, for the first time, reflective fiber sensor was applied to detect the surface topography parameters of semiconductor technology, and could work within the range of 20 nm

Over the same period, based on the primary foreign study results, domestic research about using RIM-FOS to measure the surface topography was carried out. Since 1990, the reflective optical fiber surface roughness detector has been studied systematically by Professor Lu H B[28], a famous photoelectric expert in China. The work of this team mainly focuses on the compensation of light source[29-30], the establishment of mathematical model[31-32]and the influence of various parameters on the modulation characteristics of light intensity[33-34]. In 1992, Professor Li Y F made a detailed study of the characteristics of probes with different structures. On the basis of cruciform probe structure, it was proposed that double concentric circle structure can significantly reduce the interference of light source fluctuation and has a wide measurement range[35].

At the end of 20th century, Professor Zhao Y of Harbin Institute of Technology, according to the detection need of internal thread in aerospace, put forward the internal measurement by using RIM-FOS. His team has done a great deal of research on improving the resolution and stability of the system, the stability of the laser power output of their test system is better than 0.07%, the whole system stability is better than 0.15%, the lateral resolution is 8 μm and the location resolution is 0.05 μm[36-37]. However, their study just aimed at internal structure inspection of a kind of thread used in aerospace, not in design for measuring universal surface characteristics since the project requirements.

Professor Chen Y P of Huazhong University of Science and Technology has studied the modeling of RIM-FOS[38]and completed the design of light source control system and signal processing circuit. In 2010, his students studied the path planning and intensity compensation of optical fiber sensing systems by using neural networks, and designed a system for surface profile of simple object with accuracy of 100 nm[39]. In recent years, domestic related papers are less than the past. Master student Liu D from Zhongyuan University of Technology has done the coin surface measurement by using RIM-FOS[40]and Master student Wang C L from University of Jinan has used the RIM-FOS for measuring surface defects of steel balls[41].

3 Research hotspots and current situation

It can be seen that the research time of RIM-FOS for surface topography measurement is still short, and the key problems are center on the probe structure, interference compensation and mathematical model building. These are also the research hotspots for scholars.

3.1 Structure design of probe

The research of the probe structure can be divided into three optimization directions: the distribution of fiber bundle, the arrangement of fiber and the configuration of fiber end face.

Through a large number of previous research, the distribution mode of fiber bundle has been standardized by four kinds of parallel (single pair), coaxial, random and semicircle, as shown in Fig.4.

In recent years, few novel fiber bundle distribution methods have been proposed. The mainstream adopts double circle coaxial distribution, and our team has designed a new type fiber sensor probe of concave ladder based on this kind of distribution, as shown in Fig.5.

This new probe not only has better optical compensation, but also expands the measurement range and improves the measurement sensitivity. The numerical aperture, core diameter, fiber number and other parameters of the fiber in the probe have also been systematically studied for the quantitative analysis of modulation functions. Based on these results, a lot of practical products have been sold, such as Keyence-FU series, MTI2000 and so on.

Fig.4 Fiber bundle distribution

Fig.5 New type fiber sensor probe of concave ladder

According to the present research situation, the arrangement of fiber can be simply summarized as parallel, dislocation and angles. In the initial research history, the probe structures studied by scholars are generally parallel, just like Fig.1. In recent years, some scholars have proposed a new type of probe that receiving fiber and emitting fiber at different level with a certain amount of dislocation, which reduces the interference of ambient light, obtains better sensitivity and successfully measures displacement and roughness[42]. In the case of dislocation, the emitting fiber should keep the distance from the receiving fiber to avoid blocking the light, so the probe is not compact enough. Therefore, some scholars have put forward another new arrangement with an angle, and deduced the formula of receiving light intensity[43-45]. These probe structures are shown in Fig.6.

Fig.6 Some new structures of sensor

In the aspect of fiber end face configuration, due to the difficulty of modeling, simulation and manufacturing of complex configuration, there is no systematic research. But some researchers designed optical fiber ramp with needle head to increase light receiving area or reduce ambient light interference[46], this structure is suitable for measuring the inner surface of hole. Here, a new kind of inclined plane structure is given in this paper, as shown in Fig.7, which combines the concept of double circle coaxial, dislocation and angle.

In addition, some scholars used the lens[47]and collimator[48]on the probe to shape the beam, so as to improve the working distance and measurement range. Under studies, the new probes emerge in an endless stream, and they also have advantages and disadvantages in different situations. It is worth to study whether it is possible to invent a switchable multifunctional probe with different structures to accommodate different measurements.

Fig.7 New inclined plane structure

3.2 Interference compensation

Interference compensation is the most important problem that RIM-FOS faced in surface topography measurement, the common compensation method are shown in Fig.8. In surface measurement using RIM-FOS, the optical signal carrying the surface topography information is very weak, so it is easy to be disturbed by external various factors. The common interference sources are light source fluctuation, ambient light, fiber loss, circuit noise, platform vibration and so on. The design of fiber probe structure and optical path in all kinds of new probes is considered to weaken and compensate the interference. Therefore there is also research on vibration reduction of test platform. Besides hardware compensation, in recent years, the use of cost-effective software compensation methods has been becoming increasingly popular[49]. This will be a major way in the future because the hardware method has high test cost and complex consideration in design. Among such software compensation technology, the neural network and information fusion are very effective to reduce the light source fluctuation and temperature influence.

However, most of the compensation methods can only work for single interference source, there is no research on integrated system combining hardware and software to make up for various interference sources.

Fig.8 Common compensation method

3.3 Mathematical model

As the study moving on, the mathematical model initially proposed by R.O.Cook is no longer applicable, RIM-FOS also does not simply receive and analyze the reflected light with fixed angle.

In view of this circumstance, some new mathematical models are proposed following Backman’s electromagnetic scattering theory. However, it is difficult to get universal conclusions since no systematic correlation between the mathematical models of different probe structures. Only researchers in Huazhong University of Science and Technology have obtained the general model of discrete general model and integral general model according to the number of fiber bundles[50].

At present, the research in this field is still lacking, and it is still necessary to explore a general mathematical model which can guide the design of optical fiber probe. The reflective light intensity model and light distribution model of fiber end are two important parts for establishing characteristic function.

The intensity distribution model plays an significant role in surface parameter measurement, but the latest model is still quite different from the actual situation, which is also one of the problems should be further explored. For this problem, we can try to use the self-learning ability of neural networks to skip the establishment of mathematical models.

Except the hot issues mentioned above, in the past two years, some scholars have put forward new ideas from light source and optical fiber materials, for instance, using multi wavelength light source[51]and plastic optical fiber[52-53]to improve the performance of RIM-FOS.

But when we want to get higher measurement accuracy (the area of measurement is smaller than the area of projected spot), the sensor will be deceived. This phenomenon is shown in Fig.9.

Fig.9 Phenomenon of surface deception

For the sake of analysis, assume that there are 5 beams (a,b,c,d,e) uniformly distributed in 2D space. According to reflection criterion, photodetector will receive the same light intensity sequence (0, 0, 2, 1). Then the measurement system will consider Figs.9(a) and (b) are same surface. But, obviously, the surface topography in the circles are different between Figs.9(a) and (b). This phenomenon is more serious in 3D space.

At present, there is no research on solving this problem. But we can try to divide fiber bundle into different regions to get more detailed information. And change sampling strategy to get overlapped information, then use recurrent neural networks (RNN) for surface restoration.

4 Conclusion

Under current studies, it can be seen that the main trend of probe structure is coaxial type; compensation methods prefer to use software algorithms to work in with optical layout; it tends to use the neural network to jump over the difficult problem of mathematical model. So far, the developed RIM-FOS surface measurement system is mostly used for measuring the parameters of morphology characterization. The use of this method to reconstruct the contour image or 3D view is less studied. This simulation result (contour image) combined with digital result (parameters’ measurement) should be the direction of future research.

RIM-FOS is a mature sensor technology, but in the surface topography measurement, the research is not enough. At present, RIM-FOS industrial system used to accurately measure surface has not been formed. However, due to the potential use value, experts and scholars are making great efforts to change its application status.

[1] Manojlovic L M, Zivanov M B, Marincic A S. White-light interferometric sensor for rough surface height distribution measurement. IEEE Sensors Journal, 2010, 10(6): 1125-1132.

[2] Gong Y. Research on key technology of three dimensional multi-elements measurement of parts and development of measure system. Beijing: China Academy of Machinery Science & Technology, 2013.

[3] Yang Y, Yamazaki K, Aoyama H. Fiber optic surface topography measurement sensor and its design study. Precision Engineering, 2000, 24(1): 32-40.

[4] Yuan L B. Modulation types and its theoretical analysis methods. Optical Communication Technology, 1994, 18(2): 143-165.

[5] Li X Q, Wang Z, Fu L H. A fast and in-situ measuring method using laser triangulation sensors for the parameters of the connecting rod. Sensors, 2016, 16(10): 1679.

[6] Chun B S, Kim K, Gweon D. Three-dimensional surface profile measurement using a beam scanning chromatic confocal microscope. Review of Scientific Instruments, 2009, 80(7): 073706.

[7] Xu J Q, Wang Y S, Si S C, et al. Study on scan-phase-shifting profilometry for measurement of 3-D object shapes. Acta Photonica Sinica, 2004, 33(10): 1210-1213.

[8] Laopornpichayanuwat M, Visessamit J, Tianprateep M. 3-D Surface roughness profile of 316-stainless steel using vertical scanning interferometry with a superluminescent diode. Measurement, 2012, 45(10): 2400-2406.

[9] Zhang W. The research on measurement of micro-structure with sharp curvature. Hangzhou: Zhejiang University, 2013.

[10] Frank W E. Detection and measurement device having a small flexible fiber transmission line. US: 3273447A, 1966.

[11] Kissinger C D. Fiber optic proximity probe. US: 332-7584A, 1967.

[12] Cook R O, Hamm C W. Fiber optic lever displacement transducer. Applied Optics, 1979, 18(19): 3230-3241.

[13] Bennett J M, Mattsson L. Introduction to surface roughness and scattering. Washington D.C. Optical Society of America, 1989, 13(1): 62.

[14] Beckmann P. The seattering of electromagnetic waves from rughness surfaces.New York: Norwood, MA, Artech House, Inc., 1983.

[15] Spurgeon D, Slater R A C. In-process indication of surface roughness using a fibre-optics transducer. In: Proceedings of the Fifteenth International Machine Tool Design and Research Conference, Macmillan Education UK, 1975: 339-347.

[16] Lin G C I, Shea T, Hoang K. Measurement of surface roughness with a laser beam. In: Australian Conference on Manufacturing Engineering: Preprints of Papers, Institution of Engineers, Australia, 1977: 132.

[17] North W P T, Agarwal A K. Surface roughness measurement with fiber-optics. Journal of Dynamic Systems Measurement & Control, 1983, 46(6): 1001-1002.

[18] Domanski A, Ejchart W, Jedrzejewski J, et al. The fiber-optic instrument for extremely small roughness measurement. In: Proceedings of Optical Fibres and Their Applications IV, International Society for Optics and Photonics, 1986, 670: 116-119.

[19] Fawcett S C, Keltie R F. Use of a fiber optic displacement probe as a surface finish sensor. Sensors & Actuators A Physical, 1990, 24(1): 5-14.

[20] Persson U. A fibre-optic surface-roughness sensor. Journal of Materials Processing Technology, 1999, 95(1/3): 107-111.

[21] Yang Y, Yamazaki K, Aoyama H. Fiber optic surface topography measurement sensor and its design study. Precision Engineering, 2000, 24(1): 32-40.

[22] Zhao Y, Li P, Wang C, et al. A novel fiber-optic sensor used for small internal curved surface measurement. Sensors & Actuators A Physical, 2000, 86(3): 211-215.

[23] Zhang K, Butler C, Yang Q, et al. A fiber optic sensor for the measurement of surface roughness and displacement using artificial neural networks. IEEE Transactions on Instrumentation & Measurement, 1997, 46(4): 899-902.

[24] Suparta G B, Nugroho W, Swakarma I K, et al. Metal roughness profile inspection using a micro-displacement fiber optic bundled sensor. Journal of Optoelectronics & Advanced Materials, 2009, 3(1): 65-68.

[25] Ma Y Z, Li G P, Liu H G, et al. Compensation of the optical fiber sensor based on the surface quality detection for steel balls. In: Proceedings of Chinese Control and Decision Conference, IEEE, 2011: 1265-1268.

[26] Rahman H A, Rahim H R A, Ahmad H, et al. Fiber optic displacement sensor for imaging of tooth surface roughness. Measurement, 2013, 46(1): 546-551.

[27] Lian S W, Yuan G. Real-time non-contact optical detection system for roller deformation based on artificial neural network. Optik-International Journal for Light and Electron Optics, 2015, 126(20): 2291-2294.

[28] Lu H B, Zhou D H, Qi X M. Study on fiber sensor for measuring surface roughness. Chinese Journal of Scientific Instrument, 1990, 11(4): 441-445.

[29] Zhou D H, Lu H B, Qi X M. The characteristic of laser intensity fluctuation and its compensation technology in the fiber optic sensor system. Opto-Electronic Engineering, 1991, 18(5): 31-36.

[30] Yang H Y, Lu H B, Xu T, et al. Research on intensity compensation of the RIM fiber optic sensor. Instrument Technique and Sensor, 2001, (5): 4-6.

[31] Yang H Y. Research on the mathematical model and key technologies of the reflective intensity modulated fiber-optic sensor. Changsha: National University of Defense Technology, 2002.

[32] Lu H B, Yu Z X, Yan S H, et al. The mathematical model of the output characteristics for optic fiber pair of reflection type optic fiber sensor. Opto-Electronic Engineering, 1998, 25(5): 17-24.

[33] Yang H Y, Lu H B, Xu T, et al. Effects of optical fiber parameters on the intensity modulation property of the reflective fiber optic sensors. Acta Photonica Sinica, 2002, 31(1): 74-78.

[34] Lu H B, Zhou D H, Qi X M, et al. Research on cause for effect of polarized propertles of laser upon intensity compensation in fiber optic sensing system. Optical Instruments, 1994, (1): 12-16.

[35] Li Y F, Ren J K. Modeling and simulation research on the optic fiber displacement sensor with multi-concentric distribution bundles. Journal of Changsha University of Science and Technology (Natural Science), 2009, 6(4): 59-63.

[36] Zhao Y, Li P S. Research on improving the resolution of laser optic fiber measuring system. Opto-Electronic Engineering, 1999, 26(5): 27-30.

[37] Zhao Y, Li P S, Pu Z B. MJ internal threads for aerospace purposes and their noncontact inspection methods. Measurement Technique, 1999, (11): 3-4.

[38] Chen Y P, Cao H M, Zhang G, et al. Modeling and simulation of reflective optical fiber bundle displacement sensors. Journal of Optoelectronics Laser, 2005, 16(6): 653-658.

[39] Wang Y, Yuan C M, Chen Y P. Application of BP neural network in non-contact measurement path planning. Machinery & Electronics, 2012, (9): 35-39.

[40] Liu D. The Technology and application of 3D profile surface measurement based on optical fiber sensor. Zhengzhou: Zhongyuan University of Technology, 2016.

[41] Wang C L. Study on the methods of multiple optical fibers for steel ball surface defects inspection. Jinan: University of Jinan, 2014.

[42] Wang X D. Intelligent signal processing of fiber optic sensor for simultaneous measurement of displacement and surface roughness. In: Proceedings of SPIE, the International Society for Optical Engineering, 2002, 4920: 427-432.

[43] Patil S S, Shaligram A D. Design, development and experimental study of novel configuration of retro reflective fiber optic sensor for measuring of surface roughness. In: Proceedings of International Symposium on Physics and Technology of Sensors, IEEE, 2015: 129-132.

[44] Puangmali P, Althoefer K, Seneviratne L D. Mathematical modeling of intensity-modulated bent-tip optical fiber displacement sensors. IEEE Transactions on Instrumentation & Measurement, 2010, 59(2): 283-291.

[45] Jia B H, He L, Yan G D, et al. A differential reflective intensity optical fiber angular displacement sensor. Sensors, 2016, 16(9): 1-15.

[46] Xu X M. Inner surface roughness measurement of holes based on reflective intensity modulation fiber optic sensor. Harbin: Harbin Institute of Technology, 2010.

[47] Shen W, Wu X W, Meng H Y, et al. Long distance fiber-optic displacement sensor based on fiber collimator. Review of Scientific Instruments, 2010, 81(12): 123104.

[48] Lee H S, Lee S S. Reflective-type photonic displacement sensor incorporating a micro-optic beam shaper. Optics Express, 2014, 22(1): 859.

[49] Guo Y, Wang Y T, Jin M. Improvement of measurement range of optical fiber displacement sensor based on neutral network. Optik-International Journal for Light and Electron Optics, 2014, 125(1): 126-129.

[50] Liao S H. Research and development of surface contour detection system based on optical fiber sensor. Wuhan: Huazhong University of Science and Technology, 2012.

[51] Zhu N N, Zhang J. Surface roughness measurement based on fiber optic sensor. Measurement, 2016, 86: 239-245.

[52] Hang Z Y, Xue G Q, Dong L, et al. A review of recent developed and applications of plastic fiber optic displacement sensors. Measurement, 2014, 48(1): 333-345.

[53] Shi X M. Research on reflected intensity modulated optical fiber distance sensor based on plastic optical fiber. Harbin: Harbin Engineering University, 2013.