喜歡這套資料就充值下載吧。。。資源目錄里展示的都可在線預覽哦。。。下載后都有，，請放心下載，，文件全都包含在內，，【有疑問咨詢QQ:414951605 或 1304139763】
========================================喜歡這套資料就充值下載吧。。。資源目錄里展示的都可在線預覽哦。。。下載后都有，，請放心下載，，文件全都包含在內，，【有疑問咨詢QQ:414951605 或 1304139763】
========================================
technique Keywords: for angular the primary field. Using the finite element results, appropriate shape and size of ferromag- applicati sensor performance. And one important trend in sensor re- search has been into the area of sensor systems 2. ing distance between ferromagnetic component and GMR sensor chip surface, which is usually 1.5 mm or more 6 8. Note that if the working distance is smaller, the mag- netic intensity generated by the ferromagnetic component is usually so strong that it can upset its pinned layer, thus permanently damaging the sensor element. Furthermore, the GMR sensor bridge in fact converts any field direction to two component signals, independent of their source 0263-2241/$ - see front matter C211 2009 Elsevier Ltd. All rights reserved. * Corresponding author. Address: State Key Laboratory of Robotics and System (HIT), Harbin, PR China. Tel.: +86 451 864 120 42; fax: +86 451 864 183 06. E-mail addresses: (T. Lan), Hong.liudlr.de (H. Liu). Measurement 42 (2009) 10111016 Contents lists available at ScienceDirect Measurem journal homepage: www.elsevier doi:10.1016/j.measurement.2009.03.002 When developing sensors for measuring angular signal in miniature systems, it becomes obvious that the basic sen- sor has many limitations and imperfections (e.g. size, working distance, accuracy and offset) which prevent the designer from achieving the required specifications. To im- prove the basic sensor performance, one option is to try and refine the sensor itself using better materials, produc- tion methods, etc. This choice is generally an expensive one. Another option is to incorporate the sensor into a sys- tem which has the unique objective of improving the basic key advantages (e.g. high sensitivity, miniature size, con- tactless install and wear-free work) over other kinds of sensors 3,4. However, magnetic field sensor systems for demanding applications inevitably contain ferromagnetic parts 5. These ferromagnetic components are currently used as a part of the sensor itself, like in GMR sensor. The structure of such sensor systems is rather complex, rendering difficult their integration: in the case of GMR sensors, the issues include the mounting of the ferromag- netic component in limited assembly space, and the work- Magnetic sensor Finite element techniques 3D static magnetic analysis GMR sensor MEMS 1. Introduction In general, further system miniaturizati create demands for a continuous down-scaling functions in a variety of different netic component is compared and optimized. A simplified signal processing circuit is also given for the ultra-miniature GMR sensor system. Finally experimental results show an angular accuracy of less than C61 C14 with only the residual offset compensation of ultra-min- iature GMR sensor system is obtained. Integration experiences reveal the challenges asso- ciated with obtaining the required capabilities within the desired size. C211 2009 Elsevier Ltd. All rights reserved. on will certainly of sensor on fields 1. Accordingly, further scaling of sensor systems, is manda- tory for all applications where ultra-miniature size enables the integration of MEMS and other highly integrated systems. As a magnetic field sensor, GMR sensor offers several terous robot hand. A 3D static magnetic analysis modeling is used to avoid sensor damage by the excessive magnetic intensity and disorientation when other magnetic fields disturb Study of ultra-miniature giant magneto based on 3D static magnetic analysis T. Lan a , Y.W. Liu a , M.H. Jin a , S.W. Fan a , Z.P. Chen a State Key Laboratory of Robotics and System (HIT), Harbin, PR China b Institute of Robotics and System Dynamics, German Aerospace Center, Germany article info Article history: Received 27 November 2008 Accepted 3 March 2009 Available online 12 March 2009 abstract In this study, a method giant magneto resistance design appropriate resistance sensor system a , H. Liu a,b, * the modeling of ferromagnetic component of an ultra-miniature (GMR) sensor system is presented. The goal of the modeling is to sensor system for the highly integrated DLR/HIT II 5-finger dex- ent .com/locate/measurement 9. When a second magnetic field is add to the primary field, the resulting field is a superposed sum of both fields and it may lead to severe disorientation of the angular measurement. These two constraints obstruct the applica- tion of GMR sensor in highly integrated field. In this study, the ultra-miniature GMR sensor system with only 0.5 mm working distance is developed, and the disorientation by other magnetic field is avoided and com- pensated also based on 3D static magnetic analysis tech- nique. This paper firstly introduces the concept of GMR sensor and finite element modeling. Secondly, the shape and size of ferromagnetic component are compared and optimized by using the 3D static magnetic analysis results. Then a simplified signal processing circuit is also given. Fi- nally the experimental results obtained with the developed method are compared with theoretical results. 2. GMR sensor concept 1012 T. Lan et al./Measurement 42 (2009) 10111016 Giant magneto resistance means the very large change in resistance in ultra-thin magnetic multilayer films. The basic GMR material construction includes a pinned layer and a free layer; the free layer can be influenced by an external magnetic field. An applied magnetic field of ade- quate magnitude, in the range greater than the saturation field for the free layer and smaller than the standoff field for the pinned layer, will force the free layer magnetization to follow the field when it rotates. With a fixed reference layer magnetization, and an in-step following of the free layer magnetization, magneto resistance is a simple cosine function of the angle of the rotor relative to the stationary sensor. And the resistance R of a spin valve is related to the angle h between the free and the pinned layer magnetiza- tions in the following equation R=Rp 11=2GMR1C0cosh1 where the Rp is the lowest resistance when the two mag- netizations are parallel, and the GMR is the maximum per- centage magneto resistance. The angle sensor is used in combination with a planar permanent magnet attached to a live shaft (rotor), as shown in Fig. 1. The permanent magnet is magnetized in- Fig. 1. Schematic of an angle sensor-permanent magnet rotor 7. plan thus creating a field that is in the plane of the sensor chip and rotating with the shaft. This field forces the free layer magnetization to rotate in phase with it and the ro- tor. Therefore the output signal is a sinusoidal function of the angle. With the permanent magnet, sensor design, and their axial configuration the same, the distance be- tween the magnet and the sensor, or working distance, determines the magnitude and distribution of the mag- netic field acting on the free layer of the SV resistors. The basic requirement for the working distance is that the low- est field acting on the free layer is large enough to saturate it, and the highest field does not distort the reference layer 7. 3. Finite element modeling (FEM) The method of finite element modeling is based on a discretization of the solution domain into small regions. The program uses Maxwells equations as the basis for electromagnetic field analysis. In magneto static problems, the unknown quantity (degree of freedom) is usually the magnetic vector potential, and is approximated by means of polynomial shape functions. Other magnetic field quan- tities such as magnetic field flux density, magnetic inten- sity, current density, energy, forces, loss, inductance, and capacitance are derived from the degree of freedom 10. The size of elements must be small enough to provide suf- ficient accuracy 11. In this way, the differential equations of the continuous problem can be transformed into a sys- tem of algebraic equations for the discrete problem. The practical problems necessitate usually several thousands of unknowns. However, appropriate numerical techniques have been developed, enabling to obtain the solution of such systems within reasonable time, even when personal computers are used. As a highly integrated electromechanical system, robot hand needs to complete complex tasks, like fine manipula- tion, it is often necessary to get enough accurate angular signals to implement some control strategy 12. Therefore angular sensor plays as a very important role in the sensor system and its signal accuracy affects control effect di- rectly. With high sensitivity and miniature size, GMR sen- sor is very appropriate for highly integrated system, like DLR/HIT II 5-finger dexterous robot hand. As above stated, in spite of the benefits which GMR sen- sor has, there are still some imperfections in highly inte- grated application. Because in highly integrated system, it does not have enough space to install the sensor and per- manent magnet, then the sensor system should be de- signed to make the best use of the limited space. For example, in the DLR/HIT II dexterous robot hand, the dis- tance between shaft end and GMR sensor chip surface is only 0.5 mm (as shown in Fig. 2). By the limitation of ma- chine structure, there are two types of permanent magnet can be embedded into the shaft end. One is cylinder, the other is cube. In order to ensure enough strength and avoid interference, the diameter of the cylinder permanent mag- net should be less than or equal to 2.5 mm, the length of the cube permanent magnet should be less than or equal to 6 mm, and both the thickness should be less than or equal to 1 mm. Face the problems of the limited space and the demand of precise angular signals, and consider cost and time of sensor development, it is unaccepted that first manufacture permanent magnet and then measure it by measuring apparatus. Therefore, we present a valid method that uses 3D static magnetic analysis technique for designing the type of the permanent magnet to guaran- tee appropriate direction and magnitude of the magnetic field. The 3D finite element analysis model of a cylinder perma- nent magnet and a cube permanent magnet are built respectively, and their spatial magnetization vector distri- butions are also shown. Fig. 3(a) shows a spatial magneti- zation vector distributions generated by a cylinder permanent magnet U2:5C21mm, and Fig. 3(b) repre- sents a spatial magnetization vector distributions gener- ated by a cube permanent magnet 6C22C21mm. From the figure, it can be seen that the magnetization vectors generated by the cube permanent magnet are more planar than those of the cylinder permanent magnet. It is because the length of the cube permanent magnet is bigger than the cylinder permanent magnet. Whereas GMR sensor chip is only sensitive in the parallel plane of the chip, rather than orthogonally to the chip, so the cube permanent mag- net is better than the cylinder permanent magnet for appli- cation of GMR sensor in this paper. However, GMR materials not only require the direction of magnetization vector, but also need the magnitude of magnetic intensity in its working range to avoid the sensor element damage and signal distortion. Therefore the distri- bution of the magnetic intensity in the limited space is must be achieved. Fortunately, 3D static magnetic analysis technology can make these problems easy to be solved. With the help of 3D FEM software, we acquire 3D magnetic field distribution of different permanent magnet type (Figs. 4 and 5). From the date sheet of GMR sensor chip, we know Fig. 2. Application of the GMR sensor in 5-finger dexterous robot hand. T. Lan et al./Measurement 42 (2009) 10111016 1013 In order to reject disturbance of magnetic field, the live shaft is made of non-magnetic stainless steel material. Then the model can be simplified and only the permanent magnet is analyzed, so the burden of analysis calculation is lightened greatly. The permanent magnets analyzed in this paper are made of NdFeB35 material and have remanence equal to 1.2340 T and coercive force equal to 11339 A/m. Fig. 3. A spatial magnetization vector distributions generated by (a) cylinder 6 C22C21mm. the working range absolute values are from 2388 A/m to 15,920 A/m, namely B and C in Figs. 4 and 5. It means that the GMR sensor chip must be in domain between B and C. Fig. 4 shows a spatial distribution of magnetic intensity magnitude (capped plane along X = 0 mm and Z = 1 mm) generated by a cube permanent magnet which size is permanent magnet U2:5C21mm, (b) cube permanent magnet Fig. 4. A spatial distribution of magnetic intensity magnitude generated by a cube permanent magnet which size is 6C22C21 mm. (a) Capped plane along X = 0 mm, (b) capped plane along Z = 1 mm. Fig. 5. A spatial distribution of magnetic intensity magnitude generated by a cube permanent magnet which size is 6C21C20:5 mm. (a) Capped plane along X = 0 mm, (b) capped plane along Z = 1 mm. 1014 T. Lan et al./Measurement 42 (2009) 10111016 6C22C21 mm. From the figure, it can be seen that most of the GMR sensor chip are in domain between A and B that exceed the working range. Therefore the GMR sensor chip is unable to work normally, and we must find a way to make the GMR sensor chip in its working range, i.e. in the domain between B and C. It is well known that the magnitude of magnetic inten- sity can be diminished by dwindling the width and thick- ness of permanent magnet. Based on the 3D static magnetic analysis technique, the size can be dwindled step by step to achieve the adequate magnitude of magnetic intensity rather than first manufacture different sizes of permanent magnets and then measure them. Finally, the right size of the permanent magnet is acquired (as shown in Fig. 5). From the figure, it can be seen that the whole body of the GMR sensor chip is in domain between B and C. Therefore the size (6C21C20:5mm is the right one that can generate adequate magnitude of magnetic intensity. Additionally, the simulation results can be used to put other permanent magnet outside the working domain to avoid disorientation. And it also can be used to compensate deviation in the succeeding digital processing. 4. Signal detection and processing The best way to get a high signal-gain for sensing the GMR resistance change is to build up a Wheatstone bridge and sensing the differential voltage. In this case two oppo- site reference layer magnetizations are necessary to get the highest resistance change (Fig. 6). One bridge can measure a 180C176 angle-range, so it is necessary to build up two orthogonal bridges to detect angles of 0C176360C176. In this case, we need four different reference magnetization directions in total, which define the measurement angle-orientation and turning direction regarding to the chip package. The percent change of resistance available with this GMR material is about 5%. It means that the amplitude of output signal is too small to content with the demands of high accuracy angle detection. So a pair of analog multipli- ers and four low pass filters of the signal detection circuit are used to extract the amplitude of the signal and its phase. The advantages of this configuration are simplicity, low component count, low cost, and ultra-miniature size. Take sine signal for example, the first low pass filter has a capacitor in parallel with the feedback resistor, so the cir- cuit has a 6 dB per octave roll-off after a closed-loop 3 dB point defined by Fc 1=2C1pC1R2C1C1. And output voltage below this corner frequency is defined by Eq. (2). The cir- cuit may be considered as an AC integrator at frequencies well above Fc. However, the time domain response is that of a single RC rather than an integral. Parallel combination of R3 and R4 should be chosen equal to parallel combina- tion of the R1 and R2 to minimize errors due to bias cur- rent. The amplifier should be compensated for unity-gain or an internally compensated amplifier can be used. V sin out R1R2 R3R4 C1 R4 R1 C13:3C0 R2 R1 C1 V sin in 2 The second filter is a low pass filter formed by R5 and C2, which pretends minimizing the bandwidth of noise and limiting the valid spectrum in the system. Therefore the gent function of these two values. The angle is obtained without discontinuity or dead angle over full 360C176 (as shown in Fig. 10). sensors Fig. 8. Output signals of the GMR sensor compared with theoretical sine and cosine curves. T. Lan et al./Measurement 42 (2009) 10111016 1015 cut-off frequency for this filter is determined by the pole Frc 1=2C1pC1R5C1C2. 5. Experimental results The ultra-miniature GMR sensor system has been developed using the above explained method, which ful- fills the requirements of angular measurement and inte- grated level for DLR/HIT II 5-finger dexterous robot hand (as shown in Fig. 7). Experiments have been carried out to verify the validity of the model. When a constant voltage is applied to the bridge (as shown in Fig. 6), the output voltage V sin_in is a sine function, and V cos_in is a cosine function of the angle between GMR sensor chip and the live shaft, with a com- mon direct current offset half of the supply voltage +3.3 V. The output signals are captured by a microcontrol- ler and plotted with a 0.5 mm working distance, along with the theoretical sine and theoretical cosine curves (as shown in Fig. 8). The errors between measured values and theoretical values are also shown in Fig. 9. The curve with circles represents sine error and the curve with Fig. 6. Schematic diagram of the GMR squares is cosine error. The angle is extracted from the measured sine and mea- sured cosine function, with the four quadrant inverse tan- Fig. 7. GMR sensor system of DLR/HIT II 5-finger dexterous robot hand. signal detection and processing. Absolute non-linearity is defined as the deviation from the best linear fit with a unitary slope. ANL is often chosen because of the 360C176 periodicity. The measurement result for a 360C176 rotation magnetic field showed that the ANL er- ror is C66 C14 without any residual offset compensation. As shown in Fig. 10, an ANL of second harmonic is observed (dashed curve), and the sources of this type of ANL are gain mismatch and non-orthogonality of the sensor axes. It can be seen that the ANL repeat periodically. Therefore one cy- cle of the offset errors can be use to compensate the mea- sured signals. After offset compensation, the angular error (solid curve) is less than C61 C14 , which fulfills the require- ments of angular measurement in DLR/HIT II 5-finger dex- terous robot hand and many other applications. 6. Conclusions Based on 3D static magnetic analysis technique, the ultra-miniature GMR sensor system with only 0.5 mm all of which makes up for the shortcomings of traditional measurement sensors, such as bulk, complexity, high cost, strict demand to working environment and assembly diffi- culty. This work also has established a good base for fur- ther study of magnetic sensor for micro sensor systems of MEMS and highly integrated systems. Acknowledgements This project is supported by the National High Technol- ogy Research and Development Program of China (863 Pro- gram) (No. 2006AA04Z255) and Self-Planned Task (NO. SKLR200801A01) of State Key Laboratory of Robotics and System (HIT). 1016 T. Lan et al./Measurement 42 (2009) 10111016 Fig. 9. The errors between measured values and theoretical values. working distance is newly developed for highly integrated DLR/HIT II 5-finger dexterous robot hand. Through the experiment results mentioned above, the output charac- teristics of the GMR sensor system have been obtained such as the measurement range, accuracy, working dis- tance and the repeatability. It also has the qualities of ul- tra-miniature size, simple structure and high reliability, Fig. 10. The angular errors of ANL and compensation obtained using the four quadrant inverse tangent function. References 1 C. Hierold, C. Stampfer, T. Helbling, et al., CNT based nano elec