硅基E波段差分低噪声放大器研究与设计

发布时间：2018-07-04 14:34

本文选题：CMOS + E波段　；参考：《华东师范大学》2015年硕士论文

【摘要】：毫米波频段(30-300GHz)的无线通信由于其超大的带宽,较高的安全性和抗干扰能力得到了广泛的发展,其中E-band频段相对于传统微波频段有着更加丰富的频谱资源、更高的传输速率以及更远的传输距离,近些年来成为学术界和工业领域的热点研究对象。硅基CMOS工艺的高频特性随着其特征尺寸的逐渐变小而变得越来越好,加上工艺成本低,集成度高等优点,从而越来越多的被使用到毫米波集成电路的设计中。本文基于65nm RFCMOS工艺,设计了两款用于E-band (71-76GHz81-86GHz)无线通信接收机前端的全集成低噪声放大器,主要的研究内容如下：1、通过对近些年来国内外毫米波低噪声放大器电路设计的调研总结,可以得出：毫米波低噪声放大器设计中常用的电路结构一般为单端或差分的源端电感负反馈的共源共栅结构；基本性能指标参数为电源电压1.5V左右、增益约为15dB、噪声系数6dB左右、输入输出反射系数小于-10dB。2、采用ADS Momentum对电路所使用无源器件,如片上螺旋电感、变压器以及变压器巴伦进行了仿真建模,其中所设计的片上螺旋电感的自谐振频率大于350GHz,品质因子Q在所工作的频段内大于16；进行级间耦合的变压器采用的是1：1的垂直耦合结构,在工作频段内S21小于-2dB。3、改变MOS管的偏置条件,对MOS管噪声系数和截止频率进行研究,从仿真结果中可以得出MOS管的电流密度偏置在0.15 mA/μm左右时,可以获得最优的噪声系数,电流密度偏置在0.3 mA/μm左右时,MOS管的截止频率最大；放大器电路采用伪差分结构以提高电路对共模干扰的抑制,其中传统共源共栅电路结构引入了级间电感以提升电路的频率响应特性,共栅管的栅端也添加了串联电阻以增强电路的稳定性；在有源器件MOSFET的版图的设计中尽量减小栅端、源端以及漏端金属连线之间的重叠部分以减小寄生电容,MOSFET版图设计后仿真值的fT为160GHz、fmax为406GHz,与器件模型前仿值相近。4、电路设计采用“场”、“路”协同仿真方法：因为高频时,电路的寄生效应变得更加严重,所以开始设计电路时就需要考虑到各种寄生效应的影响；电路的整体版图由有源模块的版图和无源模块的版图无缝拼接而成,将有源器件MOSFET版图抽出寄生,然后和电磁场仿真出的无源器件S参数一同带入到电路仿真中去,这样就可以考虑到有源模块和无源模块中的所有寄生,以便获得更加精确的仿真结果。5、总结出适用于CMOS毫米波频段的共源共栅源级电感负反馈结构低噪声放大器电路的设计方法：首先设置好每一级电路的最佳静态工作点;然后确定好晶体管的尺寸、完成版图绘制和寄生抽取；最后实现输入输出阻抗匹配。6、基于65nm RFCMOS工艺,设计了两款工作频段分别为71～76GHz和81～86GHz的三级伪差分结构低噪声放大器；在1.5V电源电压下,71～76GHz频段的低噪声放大器的后仿结果为：增益S21大于16dB、噪声系数NF小于7dB、输入输出反射损耗均小于-10dB、输入1dB压缩点为-17.5dBm; 81～86GHz频段的放大器的后仿结果为：增益S21大于15dB、噪声系数NF小于8dB、输入输出反射损耗均小于-10dB、输入1dB压缩点为-16.79dBm。本论文研究受上海市科委“科技创新行动计划”《E-band超高速无线通信毫米波信号源芯片技术研究》(13511500702)资助。
[Abstract]:The wireless communication of millimeter wave band (30-300GHz) has been widely developed because of its high bandwidth, high security and anti-interference ability, in which the E-band band has more abundant spectrum resources, higher transmission rate and farther transmission distance compared with the traditional microwave band, which has become an academic and industrial field in recent years. The high frequency characteristics of the silicon based CMOS process become better and better with its smaller feature size, with the advantages of low process cost and high integration, so more and more are used in the design of millimeter wave integrated circuits. Based on the 65nm RFCMOS technology, two types are designed for E-band (71-76GHz81-86GHz). The main research contents are as follows: 1. By summarizing the research on the design of the millimeter wave low noise amplifier in recent years, it can be concluded that the common circuit structure commonly used in the design of the millimeter wave low noise amplifier is the negative feedback of the single or differential source inductors. The common gate structure, the basic performance parameter is about 1.5V of power supply voltage, the gain is about 15dB, the noise coefficient is about 6dB, the input and output reflection coefficient is less than -10dB.2, using ADS Momentum to simulate the passive devices used in the circuit, such as on chip spiral inductor, transformer and transformer balun. The self resonant frequency of the rotating inductor is greater than 350GHz, and the quality factor Q is more than 16 in the working frequency band. The transformer with interstage coupling is the vertical coupling structure of 1:1. The S21 is less than -2dB.3 in the working band, and the bias condition of the MOS tube is changed. The noise coefficient and cut-off frequency of the MOS tube are studied, and M can be obtained from the simulation results. When the current density of the OS tube is biased at about 0.15 mA/ m, the optimal noise coefficient can be obtained. The cut-off frequency of the MOS tube is the largest when the current density is biased at about 0.3 mA/ mu m. The amplifier circuit uses the Pseudo differential structure to improve the suppression of the common mode interference. The traditional common source common gate circuit introduces the interstage inductance to improve the current density. The frequency response characteristic of the circuit, the gate end of the common gate also adds series resistance to enhance the stability of the circuit. In the design of the layout of the active device MOSFET, the overlap between the gate end, the source end and the leakage end metal connection is minimized to reduce the parasitic capacitance. The fT of the simulation value of the MOSFET layout is 160GHz, Fmax is 406GHz, and the device is used. The pre imitation value of the model is similar to.4. The circuit design adopts the "field" and "road" cooperative simulation method. Because the parasitic effect of the circuit becomes more serious because of the high frequency, the influence of various parasitic effects should be taken into account when the circuit is designed. The overall layout of the circuit is seamlessly spliced by the layout of the active module and the layout of the passive module. In addition, the active component MOSFET layout is extracted and parasitized, and then the S parameters of the passive device are brought into the circuit simulation with the EMF emulation. In this way, all the parasites in the active module and the passive module can be taken into account in order to obtain more accurate simulation results,.5, and the common source common grid source suitable for the CMOS millimeter wave band is obtained. The design method of the low noise amplifier circuit of the stage inductance negative feedback structure: first set the best static working point of each level circuit, then determine the size of the transistor, complete the layout and parasitic extraction; finally, realize the input and output impedance matching.6, based on the 65nm RFCMOS process, the two work bands are designed to be 71 to 76GH, respectively. Z and 81 to 86GHz three stage Pseudo differential structure low noise amplifier; under 1.5V power supply voltage, the result of low noise amplifier of 71 ~ 76GHz frequency band is: gain S21 is greater than 16dB, noise coefficient NF is less than 7dB, input and output reflection loss is less than -10dB, input 1dB compression point is -17.5dBm; 81 ~ 86GHz frequency amplifier's post imitation result In addition, the gain S21 is greater than 15dB, the noise coefficient NF is less than 8dB, the input and output reflection loss is less than -10dB, and the input 1dB compression point is -16.79dBm. this paper is funded by the research of the technology innovation action plan of Shanghai science and Technology Commission of 【学位授予单位】：华东师范大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN722.3

【相似文献】