您的当前位置:首页正文

低噪声放大器输入阻抗匹配网络设计优化方法

2023-11-26 来源:年旅网
Dec.2011 Transactions of Nanjing University of Aeronautics&Astronautics V01.28No.4 OPTIMIZATIoN DESIGN METHOD FOR INPUT IMPEDANCE MATCHING NETWORK OF LOW NOISE AMPLIFIER Sun Ling,Wu Xianzhi,Ai Xuesong (Jiangsu Key Lab of ASIC Design,Nantong University,Nantong,226019,P.R.China) Abstract:According to the theories of optimal noise match and optimal power match,a method for calculating the optimal source impedance of low noise amplifier(LNA)is proposed based on the input reflection coefficient 11. Moreover,with the help of Smith chart,the calculation process is detailed,and the trade-off between the lowest noise figure and the maximum power gain is obtained during the design of LNA input impedance matching net— work.Based on the Chart 0.35-gm CMOS process,a traditional cascode LNA circuit is designed and manufac— tured.Simulation and experimental results have a good agreement with the theoretical analysis,thus proving the correctness of theoretical analysis and the feasibility of the method. Key words:low noise amplifier;power match;noise match;Smith chart CLC number:TN722.3 Document code:A Article ID:1005—1120(2011)04—0379.O6 INTRoDUCTIoN As illustrated in the ”analog design octagon”,the design of amplifiers is full of corn— promises[ ,especially for a low noise amplifier ing mathematics optimization technology was pro— posed to obtain optimal transistor sizes and com— ponent values in input matching network and load network.For a certain LNA,a power—con— strained optimization of simultaneous noise match (LNA),which is the first active stage in a receiv— er.LNA must operate at the same high frequency as carriers and must achieve impedance matching, and input match was given in Ref.[9].Different from those previously reported works,a conve— nient method based on Smith chart is advanced in low noise figure(NF),high power gain,good linearity and so on.Therefore。it is necessary to trade these parameters each other during the LNA design,which also becomes its design diffi— culties. According to the reported papers,many this paper to obtain the optimal source impedance for lower NF and the higher gain. 1 OPTIMAL NoISE MATCH AND PoWER MATCH As we all know,in the design of LNA,there exist an optimal source impedance to provide the lowest noise figure and an optimal source impedance to provide a maximum power gain. LNA circuits have been designed for different sys— tems[ ・引,and some optimization techniques have been considered.In Ref.[6],for example,a T— type input matching network was used to achieve a wideband input impedance matching.A novel However,these two optimal source impedances are not equal,in other words,the optimal noise match and the power match cannot be achieved si— noise optimization technique for inductively de— generated CMOS LNA was presented in Ref.[7].And in Ref.[83,a design method us— multaneously.Therefore,a trade—off should be Foundation items:Supported by the Nature Science Foundation for Key Program of Jiangsu Higher Education Institu— tions of China(O9KJA51OOO1);the Creative Talents Foundation of Nantong University;the Scientific Research Foundation of Nantong University(08B24,09ZWO05). Received date:2011—01—19;revision received date:2011-05—02 E—mail:sun.1@ntu.edu.cn 380 Transactions of Nanjing University of Aeronautics&Astronautics V01.28 made between them.After a brief review of opti- mel noise match and optimal power match。a method for calculating the optimal source impedance is proposed to achieve the 1owest noise figure and the maximum power gain under the constraint of input reflection coefficient S11. 1.1 Optimal noise match A typical equivalent representation of a noisy two—port is shown in Fig.1,where Z。is the impedance of input source, ZL the load impedance,Z|D the input impedance of the two- port network,镌and辞are the equivalent input noise voltage and current respectively,口。is the voltage source and 7j2 the noise voltage source. .The noise factor(F)in terms of Fmin and the source admittance is given by[ 。] F:F mi + U [(G—Go )z+(B。一Bopt)z] (1) where FmIn is the minimum noise factor,Rn the equivalent resistance,G。and B are the conduc— tance and the susceptance of the input source, while G叩t and B卵t the conductance and the suscep— tance of the input source at the minimum noise factor,respectively.The noise figure is the sim— ple noise factor expressed in dB(NF=1 01ogF). . 、 C, —r-弋 )vI 吒 ( ) Notwios-eporlest s 广, l B、 D、 Fig.1 Equivalent representation of noisy two—port The expression in Eq.(1)means that the contours of a constant noise figure are circles cen— tered about(Gopt'B。p )on a Smith chart.Fig.2 shows an example of constant noise figure circles plotted on a Smith chart.Some important conclu— sions can be drawn as follows:The centers of all constant noise circles locate along a line drawn from the origin of the Smith chart to the point Y。pt (Yop =Gopt+jB opt).For the smaller noise figure, the center of noise figure circle is farther to the o— rigin and the radius of the circle is smaller.When F=FlⅡin,the constant noise figure circle becomes a point and Y。一Gop +jB0pt.It means that if a matching network transfers the source impedance y目t0 Yo。 ,the circuit is in the state of best noise match.It is called the optimal noise match,so Yopt is the optimal noise source admittance and Z。pt the optimal noise source impedance. Fig.2 Bunch of constant noise figure circles 1.2 Optimal power match The optimal power match for LNA means that the input impedance and the source impedance should be conjugate matched.In other words。the input matching network of LNA ci卜 cult should transform the input impedance to 50 Q or transform 5 0 Q to the conjugate impedance of input.For there always exits a certain mismatch in Yeal systems,the return lOSS,or Sl1,is used to specify this deviation.As shown in Fig.2,S11can be expressed in terms of impedanceE ]as ZiZi 一 +Z。Z  (2) When Z|一Zi ,S1lhas the minimum value and Zi 。is called the optimal power source impedance. Substituting Zi 一Ri +jXi and Z 一R +jX into Eq.(2),we have Sl1 l = (Ri 一R。) +(Xi +X。)。 (Ri +R )。+(Xi +X。) (3) so that Ri 一2Ri R。l 1+lS l 1~IS I。 R。 +( i +X。) =0 (4) The expression in Eq.(4)denotes a set of constant Sl】circles on the Smith chart,as shown in Fig.3,where Z =R +jX。is set as the vari— able.Similar to a constant noise figure circle and No.4 Sun Ling,et a1.Optimization Design Method for Input Impedance… 381 its conclusions,the centers of all constant Sl】cir— cles locate along fl line drawn from the origin of the Smith chart to the point Zi 。(Zi 一Ri 一 jXi ).For the smaller S11,the center of the circle is farther to the origin and the radius of the circle is smaller.In other words,when the source impedance is closer to the optimal power source impedance Zi ,the Sl1 is smaller. Fig.3 Bunch of constant S11 circles 1.3 Trade-off between optimal noise match and power match As mentioned in previous section,the noise match or the power match can be achieved by moving source impedance to the optima1 noise source impedance or the optimal power source impedance.Then,there must be a suitable source impedance to trade power match with noise match[ .At specified Sl1,fl method for calculat— ing this source impedance is introduced as fol— lows. According to the location of optimal noise source impedance on the Smith chart,there exist two cases. (1)Optimal noise source impedance is in the constant Sl1 circle. As shown in Fig.4,when the optimal noise source impedance Z。pt is in the constant S11 circle, which is the ideal condition.the LNA circuit can meet the S】1 specification while achieving mini— mum noise figure.In this case,there is no need for compromise, just matching the source impedance to the optimal noise source impedance directly. (2)Optimal noise source impedance is out of constant 11 circle. As an example,assume that the required S1】 Fig.4 Z。pt in constant S“circle Fig.5 Z。pt out of constant S11 circle is一4.4 dB and the optimal power source impedance is Zi ,while the optima1 noise source impedance Z0pt is out of一4.4 dB circle,as shown in Fig.5.By drawing lf straight line between Zi 。 and Z。pt,there exist two points,denoted by A and B,where the line intersects the contour of S】】 circle with the constant value——4.4 dB.As we can see,point A is closer to ZODt,so it is the point of lowest noise figure at this S11 circle.By moving source impedance to point A,the optimal noise match is achieved at this specified S11.A detailed calculation process is shown as follows. The input impedance Zi :Ri +jXi (5) The optimal power source impedance Zi 。=Ri 一j i (6) The optimal noise source impedance Z0pt=Ropt+jX。 (7) The source impedance Z。=R。+jX。 (8) The linear equation connected Zi and Z0pt 静 ×(尺。一R叩 )+Zo (9) The circle equation of constant S1 S1I l。= (R 一R。) +( i + X。) 2 _(Ri +R )。+(X + X s) 2  (1O) 382 Transactions of Naniing University of Aeronautics&Astronautics Vo1.28 The distance from the cross point to the opti— mal noise source impedance is d。:(R。一R。 )。+( 。一X ) (11) From Eqs.(9,10),the impedances at point A(ZA=R^+jX^)and point B(ZB=RB+jXB) can be solved respectively.Then,using Eq.(1 1),the distance can be calculated.The point which has the closer distance is the desired solution.With the help of mathematical tools, the calculate process is easier and faster.Math— cad,one of engineering calculation software,is used in the design. 2 DESIGN EXAMPLE Based on the Chart 0.35一 m CMOS process, a cascode LNA circuit without input/output matching network is designed to verify the method.The total die area is 368/zm×554 m. Fig.6 shows its schematic and micrograph with test probes.And its on-chip measured S—parame~ ters are shown in Fig.7. (a)Schematic (b)Die micrograph with test probes Fig.6 Cascode LNA without input/output matching network -0 兽: 兽 : -1 Frequence/GHz Frequence/GHz (a)Measured 1 Measured 1 一5 一 。鲁 , :: 一6 0.0 1.0 2.0 3.0 4.0 Frequence/GHz Frequence/GI-Iz (c)Measured (d)Measured Fig.7 Measured S—parameters of LNA without match— ing network 2・1 Calculation of required optimal source impedance In this work,the designed LNA circuit is aimed at UHF RFID systems.and the centre op— eration frequency is 922.5 MHz.After an SP sier— ulation in Cadence Spectre at 922.5 MHz,we have: The input impedance Zi =74.533--j253.182 Q The equivalent resistance R =34.281 Q The optimal conductance of input source G。。 一0.664 ms The optimal susceptance of input source B =~1.514 ms The minimum noise figure NF一0.345 dB The noise factor F=2.374 Therefore,the optimal power and noise source impedances of this designed LNA at 922.5 MHz are Zi ’=Ri 一jXi 一74.533+j253.182 (12) Z。P =1/Yop 一1/(GopI+Bo )一 242.777+j553.754 (13) Assuming that the specified S11<一1O dB, from Eqs.(9,10),we obtain Z^=53.975+ j216.232 and ZB=102.921+j304.206.Further, from Eq.(11),we have 一(R^一R。 )。+(X^一X。p ) =1.49×10。 (14) 一(RB—Rop ) +(XB—X。 )。=::8.18×10 (15) NO.4 Sun Ling,et a1.Optimization Design Method for Input Impedance… 383 As we can see,以< ,the cross point ZBis the required optimal source impedance.The noise Based on the on—・chip measured S・‘parameters of LNA without matching network(Fig.7)and the model parameters of passive components(in— ductors and capacitors)from Murata Manufactur— ing Co.Ltd,the input and output impedance figures and the corresponding S11 at different source impedances are simulated in Cadence Spec— tre and the results are listed in Table 1. matching networks are accomplished in ADS and Table 1 NF and Sll at different source impedances the experimental results are shown in Figs.9,1 0. We can see that at 922.5 MHz,NF iS 0.656 dB 2.2 Design of input impedance matching net- work According to the calculated results,the input impedance matching process is to pull ZB to 50 Q or to pull 50 Q to ZB.There are many different topologies of matching network,the T—type is chosen in this design and a matching sketch is shown in Fig.8. (a)Parameters located on Smith chart 1.0 O O 1.0 (b)Locus ofmatching process Fig.8 Sketch of input impedance matching network (point m2)and S11 is一13.094 dB(point m1). Here,the value of NF iS 0.01 dB higher than the one listed in Table 1 while S】】iS 3 dB lower. O.85 0.8O ∞0.75 \ 0.70 \ \ ,m2 Z 0.65 \/ I \ \ 0.6O \ 、 ~ 0.55 800 850 900 950l ooO1 050l1o01 150l 2o0 Frequency/MHz Fig.9 Curve of noise figure —6 —8 / 一一 兽一10  ●一.7 一12 —14 j 朋l、 / —16 800 850 900 950 l 0001 050l 1oo1 150l 20o Frequency/MHz Fig.10 Curve of S11 3 CoNCLUSION A clearer and more straightforward under— standing of optimal noise match and optimal pow— er match iS presented.With the help of Smith chart,the detailed calculation of desired source impedance iS advanced for trade—off match at specified Sl1.Based on Chart 0.3 5一 m CMOS process,LNA for UHF RFID systems is designed and its simulation and experimental results are analyzed.Finally the design example proves the feasibility of the calculation method. References: [1]Razavi B.Design of analog CMOS integrated circuits 384 Transactions of Nanjing University of Aeronautics&Astronautics VoI.28 EM].New York:McGraw.Hill,2001. E23 Cai Li,Fu Zhongqian,Huang Lu.A low power high gain UWB LNA in 0.18一 m CMOS[J].Journal of Semiconductors,2009,3O(11):115004. [7]Geng Zhiqing,Wang Haiyong,Wu Nanjian.A novel noise optimization technique for inductively degener— ated CMOS LNA[J].Journal of Semiconductors, 2009,30(1O):105015. [3]Chang Chieh—Pin,Chen Ja—Hao,Wang Yeong—Her. A fully integrated 5 GHz low—voltage LNA using for— [8]Liu Mengmeng,Zhang Sheng,Wang Shuo,et a1. An optimization design methodology for CMOS ward body bias technology[J].IEEE Microwave and Wireless Components Letters,2009,19(3): 176—178. UWB LNA[J].Acta Electronical Sinica,2009,37 (5):1082—1086.(in Chinese) [9]Mao Xiaofeng,Lan Jialong,Huang Chaogang. CMOS narrow—band LNA design optimization with [4]Xu Leiiun,Wang Zhigong,Li Qin,et a1.Design of a 40一GHz LNA in 0.1 3一 m SiGe BiCMOS[J].Jour— na1 of Semiconductors,2009,3O(5):055005. low—power dissipation口].Semiconductor Technolo— gY,2007,32(5):417—421. [5]Lee J H,Chen C C,Lin Y S.3.7 mW 24 GHz LNA with 10.1 dB gain and 4.5 dB NF in 0.18一Ⅲn CMOS [1 o]Ludwig R,Bretchko P.RF circuit design:Theory and applications[M].Beijing:Science Press,2002. [11]Kurokawa K.Power waves and the scattering matrix technology[J].Electronics Letters,2010,46(19): 131O—l312. [J].IEEE Transactions on Microwave Theory and Techniques,1965,13(2):194—202. [6]Wang C H,Chiu Y T,Lin Y S.3.1 dB NF 2O一29 GHz CMOS UWB LNA using a T・-match input net-- [12]Circuh Sage.Optimal source impedance tO trade—off power and noise and to achieve a desired¥1 1[EB/ OL].http://www.circuitsage.com/lnapa/power— noise.pdf,2011-01. work[J].Electronics Letters,2010,46(19):1312- 】3】3. 低噪声放大器输入阻抗匹配网络设计优化方法 孙 玲吴先智 艾学松 (南通大学江苏省专用集成电路设计重点实验室,南通,226019,中国) 实验结果与理论分析相吻,表明了理论分析的正确性和方 法的可行性。 摘要:通过分析电路的最佳噪声匹配和最佳功率匹配,研究 了根据给定的输入反射系数( )计算低噪声放大器 (LNA)最佳源阻抗的方法。借助史密斯圆图给出了详细的 计算过程,实现了LNA输入匹配网络设计时的最佳噪声和 最佳功率之间的折衷匹配。基于Chart 0.35一Fm CMOS工 关键词:低噪声放大器;功率匹配;噪声匹配;史密斯圆图 中图分类号:TN722.3 艺完成了一种cascode结构LNA的核心电路设计,仿真和 (Executive editor:Zhang Huangqun) 基金项目:江苏省高校自然科学研究基金重大项目(0gKJA5100o1)资助项目;南通大学创新人才基金资助项目;南通大 学科学研究 ̄(08B24,09zw005) ̄目。 

因篇幅问题不能全部显示,请点此查看更多更全内容