Search

Waveguide avalanche photodiodes (WG-APDs) have been developed for a 25-Gb/s operation. The waveguide structure was adopted for the APDs in order to achieve both high-speed performance and high responsivity. First, the dependence of the multiplication and absorption layer thickness on the 3-dB bandwidths of the WG-APDs was theoretically clarified. It was found that introducing an undepleted absorption layer was effective in improving the 3-dB bandwidth without reducing efficiency. The 3-dB bandwidth of the fabricated WG-APD based on the calculation results was over 20 GHz up to a multiplication factor of 7. The responsivities were 0.6 A/W for 1.31 m and 0.75 A/W for 1.55 m wavelength. High reliability of the WG-APDs was also demonstrated.

This paper reports the planar-structure InAl-GaAs-InAlAs superlattice avalanche photodiodes (SL-APDs) with a Ti-implanted guard-ring, which were developed for a compact and high-sensitivity 10-Gb/s optical receiver application. The design and fabrication of the novel concept planar-structure including the Ti-implanted guard-ring are described. The characteristics of the planar APDs are 0.36 /spl mu/A dark current at a multiplication factor of 10, 67% quantum efficiency, 110-GHz gain-bandwidth (GB) product, 15-GHz top-bandwidth, and - 27.2-dBm sensitivity at 10 Gb/s. The reliability was preliminary tested and the lifetime of longer than 10/sup 7/ h at 50/spl deg/C was estimated. The dark current characteristics including its temperature dependence and the excess noise characteristics are also analyzed. All the obtained characteristics exhibit the practical availability of the planar SL-APDs in the 10-Gb/s trunk line optical receiver uses.

The excess noise and ionization coefficient ratio of InAlAs avalanche photodiodes (APDs) with heterojunctions has been measured. To obtain these values accurately, we used a differential amplifier and determined the optical current at which the avalanche multiplication equals 1. These made it possible to conduct measurements on excess noise including that in the low multiplication region. Measured ionization coefficient ratios at M=10 were 0.18 and 0.17 with multiplication layer widths of 0.2 and 0.7 ?m , respectively. Excess noise was reduced with an increase in the optical input power. This tendency appeared more prominently in APDs with a thin multiplication layer.

According to the system self-organization theory, an agent-oriented system is studied and built, which can self-program object-oriented software for electromagnetic field computation.

At a factory which produces a wide variety of products such as ASIC's or microcomputers, it is necessary to detect any machine causing a bottleneck and return it to a high level of performance for fast and effective production. This paper introduces a workflow control system called "JUSTICE/MORAL" (just time process control system/method of optimum-buffer restriction and adjustment logic) which dynamically detects a machine causing a bottleneck and feeds work to that machine at an appropriate time. The advantages of this system are as follows: (1) manufacturing cycle time can be reduced by an average of 13% and a maximum of 50% and (2) throughput can be increased up to approximately 10%. Part of this system has been installed in our factory. We have seen factory cycle time improve by approximately 20% in the two month period from July to September 1998. Also, the machine causing the bottleneck was found quickly; as a result, the utilization of this machine increased by 3%,.

A wavelength conversion scheme based on a fiber Raman converter is proposed, in which an externally injected high power pump laser and the associated Stokes laser are used to assist the Raman conversion process of signal light coded with optical information. Because the Raman gain spectrum in fibers is extremely broad, a wavelength conversion device with wide-range tunability is feasible. We numerically demonstrate that wavelength conversion from 1.31 to 1.42 /spl mu/m can be realized using a fiber Raman converter at up to 10 Gb/s with an efficiency of 18%. It is also demonstrated that wide range conversion from 1.31 to 1.55 /spl mu/m for optical fiber communication is feasible at up to 5 Gb/s when the fiber Raman converters are cascaded twice.

A common problem in the modeling of anisotropic material behaviour is the description of saturation effects and the transition area between the measured data and the saturation model. A physical model for the saturation and a numerical method for the description of the material's behaviour are proposed. The nonlinear problem is treated using an iterative method. While normally the Newton-Raphson method is applied, modifications are needed to achieve convergence.

This paper describes an analysis process for a novel rare-earth permanent magnet lifting device by using finite element analysis. In the past ferrite and alnico magnets have been used in permanent magnet lifting devices which are extremely cheap compared to the neodymium-iron-boron magnets. Therefore optimized systems are not designed due to additional costs involved in purchasing and carrying out finite element analysis. Furthermore trial and error solutions have been associated with such designs for a long time, but now finite element analysis can not only enhance the design of the system but also predict whether the new design will function according to specifications.

Problems in analyzing 3-D stationary nonlinear magnetic field in non-oriented material taking into account hysteresis characteristics and eddy current, such as finite element formulation and convergence of nonlinear iteration, are discussed

The authors have developed a method of analysis which can examine the effects of the eddy currents and hysteresis in magnetic devices. In this paper, the method is applied to investigate the recording process of the metal-in-gap head and thin-film medium in a flexible disc drive. The effects of thickness and conductivity of thin sendust, and the recording current are clarified.

Factors affecting the convergence characteristics and results obtained by the optimal design method using the finite element method and simulated annealing are investigated systematically, and the optimal parameters for simulated annealing method are obtained. The optimal shape of the die mold for orientation of the magnetic powder (nonlinear magnetostatic problem) is obtained using finite elements and simulated annealing. The experimental verification is also carried out

A new type of 2nd order edge element is proposed. Various kinds of triangular edge elements for 2-D electromagnetic field analysis are discussed in terms of the definition of vectorial variables, shape functions and accuracy. It is illustrated that the new edge element is more favourable than other 2nd order edge elements from the standpoints of accuracy and selection of variables. The new element can be extended easily to higher order elements and to 3-D elements.

The effect of the scalar potential /spl phi/ in the A-/spl phi/ method on the convergence characteristic of the incomplete Cholesky conjugate gradient (ICCG) method using the edge element is investigated. Several 3-D eddy current models are analyzed both by taking into account /spl phi/ and neglecting /spl phi/ to compare the convergence characteristics. It is illustrated that the CPU time using /spl phi/ is less than 1/2 of that neglecting /spl phi/, and there are some models in which the use of /spl phi/ enables us to obtain convergent solutions.

Distribution of temperature and electric field at the joint of a dc power cable is analyzed taking into account the nonlinearity of electric resistance which is a function of temperature and electric field strength. >

A practical 3-D finite element method using edge elements for analyzing stationary nonlinear magnetic fields with eddy currents in electric apparatus, in which the flux interlinking the voltage winding is given, has been proposed. The method is applied to the analysis of magnetic fields in the Epstein frame

In order to reduce the CPU time for the modified Newton-Raphson method which introduces a relaxation factor, the effect of the relaxation factor on the residual of the Galerkin method is examined in detail. It is shown that a relaxation factor which always provides convergent solutions can be easily searched. Various methods of searching for the relaxation factor to be used are compared

The conventional method of analysis for nonconforming mesh with 3-D brick elements is improved in order to avoid the error due to the multiapproximation of potential. The method of adaptive mesh refinement using the nonconforming elements is described, and it is applied to a nonlinear magnetostatic model. It is shown that the error in potential at a nonconforming node due to multiapproximation can be avoided by improving the conventional method of analysis for the nonconforming mesh. The accuracy of the flux density using the novel adaptive mesh refinement scheme is improved compared with that using an ordinary conforming mesh with the same number of unknowns. Although the proposed method needs about twice as much CPU time as the conventional method, the mesh can be created automatically. >

The effectiveness of the edge element is investigated by comparing systematically the number of unknown variables with that of the nodal element. It is shown that the edge element is superior to the nodal element from the standpoints of computer storage and CPU time. The 3-D periodic boundary condition for the edge element is also derived in order to reduce the computer storage and the CPU time

The moving and fixed coordinate systems for eddy current analysis with moving conductors are examined and compared. It is shown that the moving coordinate system is superior for transient analysis from the standpoints of accuracy, computer storage, and CPU time. In DC steady state analysis of large Peclet number, spurious oscillations occur when using the fixed coordinate system with the ordinary Galerkin finite element method. The solution using the moving coordinate system, however, can be obtained stably. >

In order to evaluate the most suitable method of analysis (A- phi or T- Omega method) and finite element (nodal or edge element) for a given problem, the features of each method and element have been investigated. The accuracy, computer storage, and CPU time of each method and element are compared for a 3-D nonlinear magnetostatic model and a 3-D eddy current model. The flux and eddy current densities calculated are compared with those measured. It is shown that the accuracy and the CPU time of the edge element are better than those of the nodal element. The A- phi method is better than T- Omega method for nonlinear problems from the viewpoint of convergence characteristics of nonlinear iterations.

Over the past ten years, the ability of computers has progressed rapidly, and techniques of numerical analysis have improved vastly. The 3D analysis of electrostatic and magnetostatic fields can easily be carried out on a workstation, and results obtained contribute to the design and development of new products. The main factors of progress in numerical analysis using the finite-element method are reviewed.

It has been pointed out that the error in the power loss measured by a single sheet tester is mainly due to eddy currents in the specimen. The eddy current distribution is affected by the yoke construction, the overhang of the specimen, the distance between the yokes, and other factors. We previously reported on the effect of the yoke construction. In this paper, the effects on measurement error of the specimen overhang and the distance between yokes are examined both experimentally and through numerical analysis by the 3-D finite element method.

The error inherent in the T- Omega method due to cancellation between the current vector potential T and grad Omega terms is investigated by solving typical models. The cancellation error is examined using not only an axisymmetric simple model but also a three-dimensional one. The effect of permeability on the error of the T- Omega method is quantitatively examined using the simple model under DC and AC excitations. The errors in single precision and double precision are also investigated, and the allowable range of permeability is discussed. The magnetic field of the IEEJ model analyzed, and the calculated results are compared with measured ones. It is concluded that, although in general the error increases with the increase in permeability, the magnitude of the error is negligibly small if the calculation is carried out in double precision. >

A Si down-converter IC has been developed in order to satisfy the growing market for realizing lower cost in commercial microwave equipment, such as DBS (direct-broadcasting satellite) receivers and the GPS (Global Positioning System). Using a novel 0.6- mu m-emitter-width Si bipolar process, called DNP-III (direct nitride passivated base surface) for an f/sub T/ of 20 GHz, more than 10 dB of gain has been successfully obtained in the 1-2 GHz frequency range. Power dissipation is typically 125 mW for 5-V single supply voltage. Results confirmed the usefulness of the application of Si bipolar process to commercial microwave products. >

For compact and high-sensitivity 10 Gb/s optical receiver applications, we have developed low-dark-current planar-structure InAlGaAs-InAlAs superlattice avalanche photodiodes with a Ti-implanted guard-ring. The APDs exhibited dark current as low as 0.36 /spl mu/A at a gain of 10. The temperature dependence of the dark current was confirmed to be in a sufficient level for practical 10-Gb/s applications. The APDs also exhibited a quantum efficiency of 67%, a gain-bandwidth-product of 110 GHz, a top 3-dB bandwidth of 15.2 GHz, and a minimum gain for 10-GHz bandwidth of 1.6. Preliminary aging test also showed a stable dark current operation after aging of over 2200 h at 200/spl deg/C. These high-reliability, low-dark-current, high-speed, and wide-dynamic-range characteristics are promising for 10-Gb/s high-sensitivity optical receiver use.

The flux distribution and static form of a novel brushless DC linear motor were calculated using 3D finite element analysis. An analytical model to increase the force-to-volume ratio of the motor was also validated by numerical analysis. Experimental and numerical results show excellent correlation.

The effects of the impedances of the primary winding and the mutual inductor for airflux compensation of a single sheet tester (SST) on the distortion of flux waveform are investigated in order to design an optimal construction of SST. It is shown that the layers which compose the winding should be connected in series.

An effective acceleration technique using digital feedback to control the flux waveform of a single sheet tester has been developed. In this technique, the waveform of the applied voltage is pre-estimated using the equivalent circuit equation. The convergence characteristic of the proposed method using the new technique is compared experimentally with a conventional method.

The development of an electronically controlled fuel injection system is important to reduce the emissions and noise from the diesel engine. In order to design a high response magnetic solenoid actuator which is used for the fuel injection system, an optimal design of magnetic circuit is required. In this paper, a practical software using the 3-D finite element method with edge element is developed. The software is applied to the analysis of an actuator taking into account the eddy current and the motion of armature. >

Investigates the cause of quenching at the joint between a multifilamentary superconducting cable and a normal conducting cable. The 3D current distribution is analyzed. It is found that the cause of quenching is the extreme concentration of current in the joint due to the pronounced skin effect at liquid helium temperature. The mechanism of the concentration of current which is affected by the twist pitch and skin depth is also clarified quantitatively by numerical analysis

A type of transformer with reduced iron loss, which is excited by a semirotating magnetic field, has been investigated. Factors (method of excitation and slits in the core) affecting iron losses are examined. The iron loss of this type of transformer can be reduced below the material loss as measured for normal AC excitation by introducing a slit which cuts the eddy current path, and by improving the method of excitation

The optimal design of an injection mold for producing a stronger multipole magnet is carried out using the finite element method and the direct search method. It is shown that the maximum flux density in the cavity obtained by the optimal design is about 2.6 times higher than that of the initial shape determined empirically. 3-D analysis of the nonlinear magnetic field in the injection mold with complicated structure is also carried out. The calculated flux distribution on the cavity surface is in good agreement with the measured one

The 3-D finite element domain discretized by a nonconforming mesh using only first-order brick elements in investigated. The accuracy, memory requirements, and CPU time using the nonconforming mesh are compared with those using the ordinary conforming mesh by analyzing a 3-D linear magnetostatic model proposed by the IEE of Japan. It is shown that the memory requirements and CPU time can be decreased to about one-third of those using the ordinary mesh. >

The effects of the leakage flux, eddy current path, and end coil of electric machines on the flux and eddy current density analyzed by the 2-D approximation are investigated systematically by making a comparison with 3-D analysis. It is shown that, when the gap length is relatively large, the difference between the fluxes calculated by 2-D and 3-D analyses is pronounced due to the leakage flux. The flux in the pole obtained by 3-D analysis is larger than that obtained by 2-D analysis under constant current excitation, and the flux obtained by 3-D analysis is less than that obtained by 2-D analysis under constant voltage excitation. When the eddy current flows in the thin core, a larger error occurs in the 2-D analysis, because there is a great difference in eddy current paths between 2-D and 3-D models

A method for analyzing induced currents, including both eddy currents and circulating currents, in axisymmetric multiconductors connected in parallel to voltage sources has been developed using the finite-elements method. The method has the advantage that the nonlinear analysis of induced currents in conductors connected to voltage sources with distorted waveforms (for example, an inverted-fed transformer) is possible. Because the method can take into account the voltage sources, the external impedances, and the nonlinearity of the core it becomes possible to analyze flux and current distributions under actual operating conditions and to design windings which produce the minimum power loss. The effectiveness of the method is illustrated by applying it to a transformer model

The physical meaning of the grad phi in the A- phi method is investigated in order to determine when it can be safely neglected in order to reduce computation time. The relationship between eddy current distribution and grad phi and the effects of boundary conditions on grad phi are examined for several 2-D examples. It is shown that grad phi in 2-D analysis is a constant that modifies the interlinkage flux of the conductor, which is denoted by the magnetic vector potential A. >

An improved T- Omega method that can analyze magnetic fields produced not only by eddy currents but also by magnetizing currents is proposed. The method is applied to the analysis of 3-D eddy-current models with holes. The usefulness of the method is investigated by comparing calculated results with measured results and with results obtained by the A- Phi method. The T- Omega method has the advantage that the CPU time can be considerably reduced when most of the analyzed region is current-free and the eddy current flows two-dimensionally. The method is especially effective when the shape of the winding is simple and the distribution of magnetizing current is known beforehand. It is not effective when most of the analyzed region is current-carrying. >

This paper describes an Si monolithic microwave prescaler IC with a toggle frequency of 3.9 GHz. To optimize circuit parameters and to determine the dominant transistor parameters, a precise computer simulation has been used. An improved IC manufacturing process was developed to achieve the desired parameters and to obtain a transistor cutoff frequency of 10 GHz.

Six problems have been proposed for testing eddy-current computer codes. The problems represent two-dimensional and three-dimensional geometries, steady-state and transient time variation, and near and far boundaries. Five regional workshops were held to compare solutions of the problems, followed by an international workshop at Graz, Austria. The workshops and problems are described. >

To obtain the desired magnetic flux density distribution, the authors studied modifications of wiggler magnets for use in free electron lasers. The effects of magnet positions and the magnetization within magnets on the flux density distribution in two dimensions in the cavity between magnets were clarified. A three-dimensional finite element method was used to calculate distributions, and good qualitative agreement obtained with experiments using ferrite plastic magnets.

It is reported that the performance of a high-power free-electron laser amplifier using a conventional linear permanent-magnet wiggler could be improved by modifying the wiggler to provide a focusing force to the electron beam in both the vertical and horizontal planes simultaneously. The authors discuss several wiggler configurations found using the three-dimensional finite-element method. The effects of the distribution of magnetization, the structure of the rectangular block, and the magnet height on the flux distribution were investigated. >

When the flux distribution of a magnetic circuit is analyzed by using the conventional finite element method, the magnetizing currents must be given. Therefore, if the flux distribution is specified, it is difficult to obtain the distributions of magnetomotive forces or configuration of magnets producing the specified field distribution by the conventional finite element method. New methods which are called the "finite element method taking account of external power source" and the "finite element method with shape modification" have been developed. The processes of calculation in these methods are contrary to the conventional technique. These new methods have the following advantages: (a) If there are many unknown independent magnetizing currents, these currents are directly calculated by the new method. (b) When a flux distribution is specified, the optimum shapes of the magnets can be directly calculated. (c) As these new methods need no repetition, computing time can be considerably reduced. The principles and the finite element formulations of these new methods are described, and a few examples of application of these methods are shown. These new methods make it possible to design the optimum magnetic circuits by using the finite element method.

The flux distribution and torque characteristics of a stepping motor used in an analogue electronic watch are examined. In such an ultraminiature motor, it is difficult to investigate the flux distribution experimentally. The flux distributions at various rotor positions are numerically, analyzed by using the finite element method. Torque characteristics are obtained from these results. The effects of the exciting current and the shape of the stator and rotor on the flux distributions and the torques are discussed. These results give useful suggestions improving the design of stepping motors.

Computer codes developed for the A-φ, A-φ-Ω, A-Ω, T-Ω, and E-Ω methods are considered. Their accuracy, storage requirements, and CPU time are compared for linear eddy-current models. It is shown that the A-φ-Ω and T-Ω methods are preferable from the viewpoint of accuracy. The A-Ω and E-Ω methods are preferable with respect to storage and CPU time

The magnetic characteristis of cores with step-lap joints are analyzed by using the finite element method taking into account eddy current and magnetic saturation. The effects of the following factors on the magnetic characteristics such as flux and eddy current distributions and magnetizing current are clarified quantitatively. 1)step-lap length 2)length of air gap at the joint 3)number of laminations per one stagger layer 4)flux density (magnetic saturation) Obtained results give useful suggestions improving the design of the joints of transformer cores.

The iron loss characteristics of silicon steel were studied using the iron loss estimation method previously proposed by the authors. The rate of increase of the square wave iron loss with respect to the sine wave iron loss was only slightly greater than predicted. It was found that this iron loss estimation method provides adequate precision for widespread application to excitation due to square wave voltages, such as occur in inverter transformers.