Achieving Less Than 2% 3- σ Mismatch With Minimum Channel-Length CMOS Devices

Offset and speed are critical yet conflicting design parameters in high-speed amplifiers and comparators, especially those used to process the characteristically high-frequency low-amplitude signals of today's wireless transceiver systems. As device area is decreased to reduce parasitic capacitances and hence achieve higher bandwidth, random mismatches inherently increase. The proposed Survivor strategy circumvents this tradeoff by fabricating a number of small-geometry device pairs on-chip (each of which have high bandwidth) and having the IC self-select the best-matched set of devices during start-up and/or power-on-reset events and use them in critical portions of the circuit. In the experiments conducted on a prototype fabricated using a 0.6-mum CMOS technology, a mirror using the best-matched minimum channel-length pair chosen from a bank of 32 pairs (6 mum/0.6 mum) had a 3-sigma offset performance (1.94%) similar to that of a mirror using 48 mum/4.8 mum devices (1.91%) and hence a bandwidth that was 64 times higher (BW_6/0.6ap64BW_48/4.8)     

A Short Polarization Converter Using a Triangular Waveguide

A novel polarization converter using a triangular waveguide is proposed and analyzed by the imaginary-distance beam-propagation method based on Yee's mesh and the finite-difference time-domain method. The polarization conversion length is investigated as a function of relative refractive index difference. It is found, for a silicon core embedded in a silica cladding, that the conversion length is 2 mum, while the insertion loss is 0.5 dB at a wavelength of 1.55 mum. The extinction ratio is more than 20 dB over a wide wavelength range of 1.25 to 1.65 mum. Using a geometrically expanded model, the polarization conversion behavior is verified in the experiment at a microwave frequency of 15 GHz. Finally, reasonable polarization conversion is obtained with a modified structure, in which the two corners of the triangular waveguide are cut and the cut plane is aligned with a square input (output) waveguide.     

Micromechanical resonators with submicron capacitive gaps in 2 μm process

An innovative gap reduction technique is reported to achieve sub-micron capacitive gaps for micromechanical resonators to boost the output signal using the standard low-cost 2 mum commercially available foundry process from MEMSCAP. Electrostatic actuation was used to reduce the gap size below the fabrication limitation. To demonstrate the proposed idea, a 6.35 MHz Lame-mode square resonator was designed, fabricated and tested. The resonator gap size was experimentally measured to be 0.64 mum, which boosted the resonance peak by 20 dB.     

干涉法测量光学材料光学非均匀性

为了实现光学材料光学非均匀性的高精度检测,采用将光学材料制成一小楔角(小于0.117°)光学元件并利用菲索干涉法检测其光学非均匀性的方法,对测量过程进行了理论仿真分析和熔石英材料光学非均匀性的实验验证,取得了熔石英材料的光学非均匀性数据,峰谷值为4.33×10-6,均方根值为0.862×10-6,测量准确度优于4.8×...     

On the RF Series Resistance Extraction of Nanoscale MOSFETs

A new extraction method of series resistance based on the radio frequency S-parameter measurement for sub -0.1 mum metal oxide semiconductor field-effect transistor is presented. The practical limit of conventional methods is analyzed from measurement and simulation. From this analysis, analytical expressions are derived, and linear regression techniques are used to extract the series resistances. The proposed method improves the accuracy and reduces the measurement frequency.     

雷达高度表最小均方误差拟合高度估计算法研究

针对采用波门分裂法和半功率点法的传统雷达高度表,受噪声和地面各散射单元散射起伏的影响较大,进而影响高度估计的精度问题。文中提出了采用最小均方误差拟合高度估计算法,该算法基于地面回波解析模型,充分利用地面回波能量进行最小均方误差拟合高度估计,从而提高高度估计精度。仿真结果表明,在不同高度上,最小均方误差的精度均优于波门分裂法和半功率点法,降低了噪声和地面散射起伏对高度估计精度的影响。     

一种基于LS-WTLS的稳健平面拟合方法

为了解决平面数据点位精度差异性及平面模型常数项解算精度较低的问题,提出了一种基于最小二乘-加权总体最小二乘(LS-WTLS)的稳健平面拟合方法。该方法采用加权最小二乘模型与稳健估计IGGⅢ方案相结合的方式对平面模型误差项参量进行解算,然后通过设置阈值剔除粗差数据,利用最小二乘法对平面模型常数项进行解算,以此进一步提高了平面模型各参量的解算精度。结果表明, 新方法相对于最小二乘(LS)法、总体最小二乘(TLS)法、LS-TLS法、IGGⅢ-LS-TLS法,其单位权中误差分别提高了53.6%, 195.0%, 47.5%和5.1%,平面拟合精度分别提高了49.4%, 179.3%, 48.7%和46.99%,表现出了良好的抗粗差干扰能力。该研究验证了新方法在平面拟合领域的优越性和可靠性。     

一种改进的基于Viterbi的语音切分算法

主要针对文本提示型说话人识别中语音切分高精确度要求的问题,在利用Viterbi算法的语音切分基础上,提出了向后平滑搜索多帧能量极小值的语音切分方法。该算法首先对0~9的每个数字建立模型,然后利用Viterbi算法对随机数字串进行切分得到初始切分点,最后利用搜索多帧能量极小值的方法更新原始切分点。实验表明,相比于传统的切分算法,在误差范围小于20 ms之内,改进算法的切分准确率由82.1%提高到88%。     

基于PSO和改进神经网络的图像滤波方法的研究

提出了一种基于改进BP神经网络和粒子群优化算法（PSO）的图像滤波方法。该方法利用对数最小均方误差函数（LNLS）代替BP神经网络传统的最小均方误差函数（LMS）,用来减小图像噪声对神经网络精度的影响;并将改进后的BP神经网络利用PSO算法优化,从而避免神经网络陷入局部极小值点,进一步提高神经网络滤波能力。实验结果表明,与传统滤波方法相比,该方法不仅能有效地滤除图像中的高斯噪声而且能很好地保护图像细节。     

空基多平台对空中机动目标的被动跟踪定位算法

针对空中机动目标的被动定位跟踪问题,提出了一种先用静态估计理论对空中目标进行最小二乘估计,再采用基于“当前”统计模型的自适应滤波算法进行滤波处理的方法,取得了比最小二乘估计与卡尔曼滤波相结合的算法更好的效果。仿真结果表明,在跟踪非机动目标时,该算法和最小二乘估计与卡尔曼滤波结合的办法相当;在跟踪机动目标时,该算法的误差明显小于原算法。     

基于总体最小二乘准则的OFDM系统时变信道估计

信道估计的准确程度直接影响正交频分复用系统的性能。为了提高时变信道估计算法的精度,基于总体最小二乘准则( TLS)提出了一种时变信道的估计方法。该方法用线性模型对时变信道进行建模,不仅考虑了信道噪声,同时也兼顾了模型误差。该方法能较好地跟踪信道的变化,显著消除模型误差。仿真结果表明所提算法的均方误差介于最小二乘算法与最小均方误差算法之间,在不同归一化多普勒频移下,该算法具有较好的稳健性。     

GPS作为间接基准鉴定精确导航系统的新方法

随着技术进步,精确导航系统与GPS处于同一精度,传统的直接比对法不再适用。为此,研究了一种新的间接法,以提高基准精度,满足鉴定要求。间接法是利用GPS天线阵测得多个基准值,然后采用最小二乘法,解出基准位置的精确值。编出了试验数据处理软件,客观准确地评估了精度性能。它已成功地用于某项精确导航系统的鉴定试飞中,对后续项目具有参考价值。     

条纹投影动态3维测量中相位高精度估计

为了提高条纹投影动态3-D形貌测量精度, 采用加窗傅里叶分析辅助相移的方法来减小运动导致的相移误差。首先采用加窗傅里叶分析法估计变形条纹间的实际相移量, 然后采用最小二乘法估计出变形条纹的高精度相位分布, 最后由估计的相位计算得到场景3维形貌。理论分析了物体运动对相移量的影响, 通过仿真研究了所提方法的相移量估计精度, 并搭建了实验系统进行验证。结果表明, 实验中采用所提方法的相位恢复精度达到0.1673rad, 比现有方法有明显提高。该方法用来提高条纹投影动态3-D形貌测量中相位精度是有效的。     

Servo Performance Enhancement of Motion System via a Quantization Error Estimation Method—Introduction to Nanoscale Servo Control

When compared to the accuracy of nanoscale control, the resolution of current positioning sensors is relatively low. Because of this, the output from low-precision sensors normally includes quantization errors that could degrade control performance. As a result, in this paper, a method of quantization error estimation based on the least square method is examined. In the proposed method, estimation accuracy is improved by taking into account the effect of input disturbances. Furthermore, a bias adjustment method is proposed that is expected to satisfy the constraints on quantization error. The effectiveness of the proposed method is demonstrated by simulations and experiments.     

基于贝叶斯模型和数字图像相关的视觉测量

为了能够快速准确地实现3维测量,提出了一种基于贝叶斯模型和数字图像相关的视觉测量方法。该方法采用数字图像相关法对校正后的图像进行立体匹配,克服了传统立体匹配方法精度不高的问题;采用贝叶斯模型估计图像视差,并将其作为数字图像相关法非线性迭代优化的视差初值,克服粗搜索方法寻找视差初值计算量大、精度低的缺点;基于校正后图像与原图像之间的投影关系,由最小二乘法计算出匹配点的3维坐标。结果表明,基于贝叶斯模型和数字图像相关的视觉测量算法,能够快速准确地实现3维视觉测量。     

Broadband Polarizers Using Dual-Layer Metallic Nanowire Grids

Broadband polarizers functioning in an extremely wide wavelength range from 0.3 to 5 mum were developed based on structural design and numerical simulation. By properly selecting the dual-layer nanowire grids including metal materials, grid periods, grid thicknesses, and substrate materials, broadband polarizers could be developed to realize high optical performance. We proposed a broadband polarizer using aluminum nanowire grids on a magnesium fluoride substrate with 80-nm square grids. The proposed polarizer can reach an extinction ratio from 47 to 70 dB and a transmission efficiency from 38% to 94% in the wide wavelength range.     

基于ARMA模型的电力系统负荷预测方法研究

采用加权最小二乘法参数估计方法，得到应用于电力系统日负荷预测和月负荷预测的ARMA模型，实验预测结果表明，用ARMA模型进行电力负荷预测是非常有铲的。尤其是采用加权最小二乘估计的ARMA模型，预测精度更高。     

直升机桨叶共锥度测量中圆形标记点高精度定位方法

针对直升机桨叶共锥度测量中背景复杂、圆形标记点较小,易出现定位不准的问题,提出一种新的圆形标记点定位方法.首先,利用OTSU图像分割方法计算阈值;然后,以OTSU阈值为边界阈值,对图像进行多重分割;利用圆度特征和质心变化规律,对每一次分割后的连通域轮廓进行干扰排除,实现圆形标记点的粗定位;最后,采用最小二乘拟合法完成标记点的精确定位.通过圆形标记点定位实验,表明该方法在复杂场景下对较小的圆形标记点具有较高的定位精度,可用于提高大场景复杂环境下直升机桨叶共锥度测量精度.     

基于LSSVM的威布尔分布形状参数估计

固态介质击穿寿命特性通常用威布尔分布来描述,形状参数卢反应了固态介质的失效特征,因而需要精确估计β值.提出了在小样本情况下基于最小二乘支持向量机(LSSVM)的参数评估方法,并给出了LSSVM在MOS电容与时间有关的击穿寿命分布评估中的应用实例,并与常规的最小二乘评估方法相比,得到的结果表明LSSVM的评估精度更高(均方误差更小)、鲁棒性更好,在小样本情况下能更精确地确定威布尔分布的形状参数.     

A High Linearity, Fast-Locking Pulsewidth Control Loop With Digitally Programmable Duty Cycle Correction for Wide Range Operation

A high linearity pulsewidth control loop (PWCL) is proposed in this paper. Using the linear control stage (CS) and digital-controlled charge pump (DCCP), the proposed PWCL can be operated within a wide-range of both input and output duty cycles over a wide frequency range. A simple detection circuit is utilized to control the DCCP in a complementary architecture such that the proposed PWCL can reduce the locking time ratio to 4.5. The test chip is fabricated using 0.18 mum CMOS process. The measurement results show that the frequency range of the input signal was 1 MHz to 1.3 GHz, the duty cycle range of the input signal is from 30% to 70% and the programmable duty cycle of the output signal is from 30% to 70% in steps of 5%. The measurement power dissipation and the peak-to-peak jitter are 4.8 mW and 13.2 ps, respectively, at an operating frequency of 1.3 GHz.