摘要:A novel type of silicon condenser microphone with corrugated diaphragm has been proposed,designed,fabricated and tested.The e microphonecan be fabficated on a single wafer by use of silicon anisotropic etching and sacrificial layer etching techniques. The introduction of well performed corrugation has strongly increased the mechanicamesensitivity of the microphone diaphragm because of the reduction of the stthss in esin fnlms.Corrugations with different q values have been investigated,showing a close relationship between the ratio of the depth of the corrugation to the diaphragm thickness and mechanical sensitivity of the diaphragm. For the purpose of further decreasing thin film stress,composite siaphragm had been fabricated,reducing the initial stress to a much lower level of about 70MPa in tension. TTree etypes of corrugation placements in a diaphragm area of lmm by lmm have been fabricated. Microphones with higher than 10kHz bandwidth and sensitivity of 6. 8~16. 0mV/Pa under a relatively low bias voltage of about 5~15V have been fabricated. The experiments prove that the corrugation technique is promising for silicon condenser microphone.
摘要:This paper presents a kind of the microwave monolithic active filter circuit with GaAs MESFET negative resistance property, and gives a GaAs MESFET varactor construction considering the MMIC procedure-compatibility.Therefore, the active filter has the voltage-controlled function. The design method of monolithic circuit and the CAD results are also given. Finally,the MMIC active filter with chip size of 0.9 × 1.8mm2 is fabricated.The experiment results show that the design method presented is correct and effective.
摘要:The methodology of constructing scan chains in a design with multiple scan chains is proposed in this paper. Based on the circuit scale,the number of I/O pins available and the test time required,the number of scan chains is decided. Critical time is introduced,and dynamic programming method is used to locate the flip-flops of every scan chain. Using this algorithm,computing velocity is greatly improved and space is saved.
关键词:Multiple scan;design for testability;Dynamic programming;Scan chains
摘要:A new concept is introduced that characteristic damping coefficient kc is used to distinguish between the effects of mass loading and liquid damping of a quartz resonator vibrating in liquids. It is derived that the value of kco has nothing to do with the properties of liquid and solid deposited on the quartz crystal. It only depends on the parameters of the crystal resonator,such as fundamental frequency,the area of the electrode and the overtone number.The effect of mass loading occurs only when kc varies. By measuring the frequency and amplitude simultaneously,the influence on kco of fundamental frequency,electrode area and overtone number of a 10MHz AT cut quartz crystal resonator is studied.The experimental results fit the theory well.
摘要:The growing complexity of integrated circuits poses new challenges to the existing layout design tools. This paper presents a modified clustering approach for macrocell gate array placement. Both connectivities and sizes of cells are taken into account in the clustering process.When combined with Force-Directed-Relaxation method,the algorithm generates promising results and takes a short computation time. Examples are also given to demonstrate the effectiveness of the algorithm.
摘要:Two new current-mode Hamming neural networks are proposed in this paper.The first network consists of a current-mode template matching circuit and a fully interconnected current-mode Winner-Take-All(WTA)network.Simulation shows that it has high speed,high precision and low power dissipation. Based on the proposed network,another Hamming network which is composed of a current-mode template matching circuit and a voltage-mode WTA is also put forward.
摘要:A VLSI architecture for forward/inverse 2D discrete wavelet transform in realtime applications is presented in this paper. It is designed and verified by the hardware descirption language (VHDL)and is suitable for the hardware implementation. The VHDL of our architecture is synthesized by the Synopsys synthesizer. The synthesized circuits contain 7140 area. The results show that the speed of data processing can reach about 4M pixel/s.
摘要:MFNNs(Monolithic Fuzzy Neural Networks)proposed in this paper combine the advantages of Fuzzy Logic Control and Aritificial Neural Network Control,not only behave well in object control but also have abilities of model self-learning and on-line self-adjustment. A series of self-learning algorithms for MENNs can be obtained based on ANNs theory,Fuzzy Set theory and its special structure. Some simulation results based on MFNNs gradient algorithm show that the convergence performance of MFNNs is much better than ordinary ANNs,the self-learning and self-adjustment are practicable,simple and effective. The learning model can be used for intelligent control system including the complicated object control. The VLSI realization of the learning MFNNs can be implemented easily.
摘要:A novel idea is presented in this paper for the first time. The equivalent model of the interconnects in the VLSI circuits should be hierarchical. For the low level circuit designing,the interconnects should be treated as multiport networks with distributed parameters;for the high level module designing,they should be treated as logic elements. Based upon this idea,a table logic model of the interconnects in VLSI circuits is presented. Three major negative effects, namely,crosstalk,signal distortion and delay,’are all considered in this model. Numerical results show that this model is accurate and efficient. Besides,features of the algorithms for the fast calculation of the transient responses to multiple input patterns are also proposed.
摘要:This paper addresses the hardware overhead issue in the continuous checksumbased technique to design fault-tolerant linear analog circuits,and proposes the first optimization algorithm for hardware overhead in the error detection circuit of the checker. Without changing the original circuit,the proposed algorithm can not only reduce the number of passive elements,but also the number of analog operators so that the error detection circuitry in the checker has optimal hardware overhead. The algorithm is proved to be feasible and effective.
摘要:This paper presents equivalent conditions of correlation-immunity of p-valued logical functions with p≥3 being a prime or p= 4, 6. Based on this,necessary and sufficient conditions of correlation-immunity of p-valued logical functions are obtained by using Chrestenson transformation. Where Chrestenson transformation is concerned,Xiao-Massey Theorem is extended perfectly in the case of p≥3 being a prime.
摘要:The a-St: H TFT with thin active layer structure is studied in this paper. The experimental results show that the characteristics of a-Si: H TFT vary when the a-Si: H thickness is smaller than some critical value. The effects of the space charge in the active layer rear surface on the a-Si: H TFT characteristics are discussed in detail. The dependence of the threshold voltage on the active layer thickness is calculated with the effective surface space charge layer model,and the critical active layer thickness is 130nm,which is baSically consistent with experiment results.
摘要:Process is an important tool of describing dynamic characteristics of Petri net.Synchronous composition is an effective method for studing properties of complex Petri net. This paper researches the process characteristics in synchronous composition. The composition formulas set of basic processes and process language are obtained,respectively. The properties of out set and line are also obtained. These results are useful for studying dynamic characters of complex Petri net.
摘要:This paper introduces the basic concept,development process and application areas of fuzzy logic and fuzzy logic control.The function and character of fuzzy algorithms(fuzzification,selecting and trimming of fuzzy control rules, operators and methods of defuzzification)are discussed,some methods of combination fuzzy logic with artificial neural network and expert system are introduced. A survey of some kinds of realizing the fuzzy logic control is presented,the status of fuzzy control logic chips,in particular,current-mode multiple-valued logic CMOS VLSI technology are described.
摘要:The total charge Green’S function can be used in the MEI method taking the place of complete Green’s function to get the measuring function and then to determine the measured equation of invariance. The capacitance matrices of several kinds of multiconductor interconnects have been computed with this method.
关键词:Measured equation of invariance (MEI)method;Total charge Green’s function;Multiconductor interconnect
摘要:A fuzzy processor employing SUM & MAX operation is proposed,which can be used for pattern classification and pattern recognition.It receives multiple inputs that represent multiple features of an unknown pattern in time-shared way. Membership function generators generate memberships corresponding to each standard pattern according to the input features.Switched-capacitor accumulators sum multiple memberships to get synthetic memberships. Finally,WTA(Winner-Take-All)circuit finds the maximum synthetie membership and the recognition result is obtained.
摘要:Based on the defect size distributions on the silicon wafer, the fractal feature of the point defect distribution is analysed and the fractal model of describing the defect size distribu-nons is obtained by means of the system of the fractional dimension. This model has physical meaning clearly. A new dynamical model of expressing the change of defect size is given and discussed. The model can obtain defect characteristics which are necessary for accurate functional yield prediction during design for manufacturability of IC’s.
摘要:Based on thermodynamic analysis,the possible species containing arsenic and their concentrations at temperature from 500K to 1200K are calculated for Ga (CH3)3 - AsH3 - H2 system. A new method for arsenic abatement has been proposed through numerous simulation experiments. In order to improve the efficiency of arsenic abatement a three stage,counter flow showerbath has been inverstigated. The arsenic concentration in gases meets the State Standard for arsenic emission.
摘要:We put forward an external cavity laser using Faraday Anomalous Optical Filter (FADOF) as a frequency selecting device. Its linewidth is reduced and frequency locked to one of the four transmission pinks of FADOF. We also observed the laser frequency tunability (>0.2GHz) caused by the change of FADOF’s magnetic field. This external cavity laser can be used in free-space optical communication system.
摘要:Equivalent electricity quantity relation method (EEQR) is discussed. By means of this method,various switched capacitor DC-DC converters can be analysed without dealing with the complicated state equations.
关键词:Switched capacitor networks;DC-DC converters;Stable state analysis
摘要:A new active compensation scheme for the current integrator circuit using CCll is proposed. A minimum of passive RC components is needed. Both theoretical analysis and SPIGE simulation have verified a significant improvement in the gain and phase errors of an integrator. A second-order bandpass employing the new integrator is compared with the same structure employing uncompensated integrator.
摘要:This paper presents a very high speed ECL static frequency divider circuit,the structure of poly-St emitter NPN bipolar device is given and the key technologies such as salicide contact,deep trench isolation,shallow junction and thin base formation are described. By using this advanced St bipolar technology,the average gate delay less than 50ps is achieved for 19-stage ring oscillator with 3pm feature size. The circuit and layout design for a 1/2 divider as well as some strategies for enhancing performance are discussed in detail,and the 1/2 divider has got a maximum operating frequency of 3.2GHz.
摘要:Evaluating the noise performances of three configurations of transistor amplifier accurately is the crux of optimization design of circuit configurations for low--noise amplifier. New results about En-In noise analysis of three configurations of transistor are presented in this paper.
摘要:In this paper,a calculated program for high-level injection model of PET has been made based on the modified model of PET proposed by Yih-Feng Chyan et al. The input and output characteristic,current gain and frequency response (fT and fmax)of PET with exponential doping profile in emitter and base under high-level injection have been simulated by taking the measured data of studied device as the input parameters,and some characteristics have been analized theoretically. The program can be used to study the relation between the device characteristics and device structure parameters for PET.
关键词:Polysilicon emitter transistor(PET);High-level injection model
摘要:In this paper,a complete equivalent circuit of the photo negative resistance transistor with LBT structure(or photo lambda bipolar transistor PLBT)has been proposed for the first time.Based on this equivalent circuit,and by using the PSPICE circuit simulation program,the simulation on IPh-VcE characteristic of PLBT has been made. The simulation result is in good agreement with that measured from the PLBT experimental device. From initial investigation, it isshown that besides as a photodetector,this device has also the functions of photo-induced oscillation and photo-controlled frequency modulation.
摘要:An approach to the optimum design of CDT correcting lens and an example are presented in this paper on the basis of CPT correcting lens design. An accuracy analysis of the computer program for surface modificatin of the lens is made by using computer-modeling of Ebeam paths and ray-tracing.
摘要:Fractal cellular neural networks(FCNN)are first presented and applied to associative memory. From simulation experiments,it can be seen that FCNN is superior in performance to fractal neural networks (FNN) proposed by Baram for associative memory.
摘要:Timing-driven placement is the earliest approach to improve IC performance.However,the present algorithms run slowly and the designs are dissatisfactory. This paper,based on solving allocation problem,presents a placement algorithm that consists of initial placement,replacement and deleting overlapping. The experiment result shows that our algorithm runs fast and can reduce the IC delay.
摘要:A new competitive learning algorithm based on Partial Distortion Theorem is developed for vector quantization,which is called Partial-Distornon-Unform competitive learning a lgorithm. Compared with several popular competitive learning algorithms,the proposed algorithm shows. Compared with several populor competitive learning algorithms,the proposed algorithm the best performance for the codebook design with different sizes,especially for the design of large size codebook.
摘要:A method for simulation of nonlinear microwave, circuit under multitone excitation is discussed using arithmetic operator method. A new method to form Jacobian martix is given,it can analyse as many as multitone excitation.
摘要:The fuzzy knowledge acquirement is one of the key problems for fuzzy information processing. Based on fuzzy similarity relation and general knowledge,this paper introduces a kind of method for acquiring fuzzy knowledge including fuzzy concept and fuzzy rules.
摘要:A new analysis method based on time domain is presented for PLL frequency synthesizer with a charge pump and a digital phase-comparator. This method is used to design and improve this kind of synthesizer. CAA software has been applied to practice.
摘要:In this paper,we study the problem of stability for the cellular neural networks with delay. By making use of the Lyapunov function,the method of variation of parameters and combining with the method of inequality analysis,the present author establishes sufficient conditions of stability for cellular neural networks with delay.
关键词:cellular neural networks;delay;global stability;Lyapunov function