【作者】 郭本青；

【导师】 文光俊；

【作者基本信息】 电子科技大学 ， 电路与系统， 2011， 博士

【摘要】 伴随无线通信市场的迅猛增长和集成电路技术的快速发展，射频前端集成电路技术得到了广泛深入的研究。设计者对于射频前端电路优越电学性能的无止境追求，也不断驱使着相关设计技术的推陈出新，与时俱进。其中，直接变频结构以结构简洁，高度集成而受到研究者的大为推崇，被视为最有希望成为未来射频集成收发前端的主导架构。不过，与性能上的优越相伴随的是技术上的挑战。直到今天，业界围绕着其直流失调、闪烁噪声、本振泄漏等关键技术问题所展开的研究与探索从没有间断过。另一方面，混频器是射频接收前端的核心模块之一，其噪声的周期时变特性使得相关分析变得十分复杂，也正是这个原因，其一直是射频集成电路学科颇具挑战性的研究课题。既有的相关噪声分析模型已经不能有效提供在亚微米、高频等当前实际应用场景下的混频器电路设计指导。基于此，如何开发出具有更广泛适用性的混频器噪声解析模型显得尤为迫切。本文针对CMOS混频器的噪声特性及射频接收前端的若干关键技术问题进行了研究，取得了一些有益的结论和成果，主要研究工作和创新点如下：1.提出亚微米CMOS有源混频器噪声模型：借助最新的基于亚微米物理机理的MOSFET器件I-V模型和器件噪声模型，通过数值迭代来求解混频器大信号I-V方程，进而结合小信号推导得到包含沟道调制效应的混频器各级噪声转换函数，最后得到包含亚域区导电效应，沟道调制等二级效应的噪声解析模型。2.提出包含记忆效应的CMOS有源混频器噪声分析方法：基于线性周期时变理论，推导得到包含记忆效应的混频器噪声转移方程。使用电路小信号分析方法，推导得到混频器各级的周期时变转移函数。包含尾电容记忆效应的混频器电路各级噪声源至输出的噪声变换系数得以数值求解，进而最终得到适用于高频下的混频器噪声解析模型。以上两个噪声模型均为包含热噪声和闪烁噪声的一元化解析式，可以应用于具有不同中频特点的该型混频器的噪声设计优化。虽然理论上两者可以合二为一，但是出于解析式的简洁直观目的，文中分开进行论证。3.提出本振失调下CMOS有源混频器的转换增益模型：通过对电路的共源共栅小信号等效，推导出亚微米工艺条件下的包含器件输出电阻效应的混频器有效驱动级跨导解析式。在此基础上，提出了本振信号存在各种幅度、相位失调情况下的混频器转换增益解析模型。4.提出一种低噪放和混频器融合结构的增益可调直接变频射频接收前端：采用折叠结构降低开关对的偏置电流取得良好的闪烁噪声性能，同时兼有优越的增益性能；通过共模与差模反馈来改善直流失调，稳定中频输出；利用差分电路结构以及版图优化布局获得高的二阶交调指标，两个管子堆叠的电路结构，利于获得低压低功耗。更多还原

【Abstract】 With the rapid growth of wireless communications market and the rapiddevelopment of integrated circuits technologies, integrated circuits technologies for RFfront-end were extensively researched. Designers of RF front-end circuits for theendless pursuit of excellent electrical properties, has also been driving relatedtechnological innovation from time to time. Among them, the direct conversionarchitecture with simple structure, highly integrated level is greatly welcomed by theresearchers, and is considered the most promising to become the leading RF integratedtransceiver architecture in the future. However, the superior performance isaccompanied by technical challenges. Even today, the studies undertaken by theindustry around the DC offset, flicker noise, LO leakage and other key technicalproblems never stop. The mixer is one of the core modules of RF receiving front-end,the time-varying characteristics of whose noise complicates related analysis much. It isfor this reason that the noise analysis of mixers has been a challenging problem in RFintegrated circuits subject. Meanwhile, the existing noise analysis model can noteffectively provide design guidelines for the mixer circuits in current sub-micron andhigh-frequency application scenarios. Based on this, how to develop a noise analysismodel of mixers with broader applicability is particularly urgent.The noise characteristic of CMOS mixers and a number of key technical issues forRF receiving front-end have been studied in the dissertation, and some usefulconclusions and results are obtained, where primarily research and innovation are asfollows:1. A sub-micron CMOS active mixer noise model is proposed: Based on the latestsub-micron physical noise model and I-V model of MOSFETs, the large-signal I-Vequation of the mixer is solved by numerical iteration. And then, the noise transferfunctions of the mixer with channel modulation are derived by small-signal analysis.Finally, a noise analytical model considering secondary effects such as subthresholdconductivity and channel length modulation is obtained. 2. A noise analysis method for CMOS active mixer with the memory effect isproposed: Based on linear periodic time-varying theory, the noise transfer equationsof the mixer including the memory effect have been rigorously derived. Bysmall-signal analysis, the periodic time-varying transfer functions of the mixer areobtained. The noise transform coefficients including tail capacitor memory effectcan be solved numerically, and then a high frequency noise analysis model of themixer is finally captured. These two noise models both are unified analytic modelsincluding thermal noise and flicker noise, and can be applied to the noise designand optimization of the mixer with different intermediate frequency characteristics.Although theoretically the two can be combined, for the simple and intuitivepurposes, they are separately demonstrated.3. A conversion gain model with LO signal offsets for CMOS active mixer isproposed: With the approximation of the small-signal cascode circuit, an analyticdriver stage transconductance including the channel length modulation effect isderived suitable for submicron technology. On this basis, an analytical conversiongain model for the mixer is proposed with the existence of various amplitude andphase imbalance of local oscillator signal.4. A gain-adjustable direct conversion RF receiving front-end with merged low-noiseamplifier and mixer is presented: Low flicker noise and high gain are achieved bydecreasing the bias current of switching pairs; DC offset is improved by bothcommon mode feedback and differential mode feedback, making the IF output levelstable. High IIP2is obtained by the differential circuits’ architecture and thesymmetrical layout. The circuit with only two stacked MOSFETs, achieves thetarget of low voltage and low power consumption.更多还原