A High-Gain, Low-Noise GHz Ultra-Wideband LNA in a 0.18μm CMOS

Transcription

1 Majlesi Journal of Electrical Enineerin Vol., No., June 07 A Hih-Gain, Low-Noise GHz Ultra-Wideband LNA in a Behnam Babazadeh Daryan, Hamid Nooralizadeh * - Department of Electrical Enineerin, Islamshahr Branch, Islamic Azad University, Islamshahr, Iran - Department of Electrical Enineerin, Islamshahr Branch, Islamic Azad University, Islamshahr, Iran nooralizadeh@iiau.ac.ir; h_n_alizadeh@yahoo.com (Correspondin author) Received: Auust 06 Revised: January 07 Accepted: March 07 ABSTRACT An ultra-wideband (UWB) common ate-common source (CG-CS) low-noise amplifier (LNA) in a technoloy is presented in this paper. A capacitive cross-couplin fully differential amplifier with the current-reuse technique is described that in the entire GHz UWB band, achieves a hih and flat power ain with low noise and ood input impedance matchin amon low power consumption. The current reuse technique is used to achieve a wideband and reduce the power consumption. The capacitor cross couplin technique is used to m -boostin and hence to improve the NF of the amplifier. Therefore, the dependency between noise fiure (NF) and input impedance matchin is reduced. The proposed CG-CS amplifier has a fairly low NF compared with the other previous works in similar technoloy. In addition, a ood power ain over all bandwidth and a hih isolation with ood input/output impedance matchin are achieved. The minimum NF is.8 db, the maximum power ain is 4. db, the inverse ain is <-50 db, and in the entire GHz UWB band, the input and output matchin S and S are < -0.3 db and <-.3 db, respectively. The input third-order intercept point (IIP3) is -5 dbm. Moreover, the power consumption of the core is 0. mw with the supply voltae of.8 V. KEYWORDS: LNA; UWB, CMOS Technoloy, Current-reuse Technique, Capacitive Cross-couplin Technique, CG-CS Amplifier.. INTRODUCTION As the first stae of a receiver, low-noise amplifiers (LNAs) play a critical role in the overall performance of the system. The use of ultra-wideband (UWB) frequency rane ( GHz) for commercial applications was approved in 00 by the Federal Communications Commission []. The UWB technique provides hih data-rate transmission and reduces the power consumption. An UWB LNA must exhibit a flat power ain over the entire UWB bandwidth and have a broadband input matchin, a low noise fiure (NF), and a hih linearity while consumin a little power []. In modern wireless products, CMOS technoloy benefits are low cost, hih level of interation, and low power consumption. There are many research works related to UWB LNA desin with new structures and desin techniques [3], [4]. The distributed amplifier of [4] provides ood properties, but the cost is consumin lare power. In multistae LNAs, there are multiple DC bias paths, which result in a lare total power consumption [5]. A CMOS UWB LNA employin the noise-cancelin technique is reported in [6]. However, the ain performance of such a confiuration is often less superior [7]. The resistive shunt feedback amplifiers provide wideband input matchin. Nevertheless, resistive feedback amplifiers cannot provide required performance with low power consumption [8]. To realize low power, common ate-common source (CG-CS) and CS CS current-reuse structures are enerally adopted which hih ain also can be achieved [9-]. Compared with the narrowband LNA desins, a severe tradeoff between NF and source impedance matchin exists in a wideband LNA. To desin a fully differential capacitive cross-couplin CG LNA in [3], a m boosted technique is used and a low NF and hih linearity alon with a ood input matchin is achieved. The capacitor cross-coupled CG-LNA reduces the NF and the power consumption by a factor of two [4]. The cross-couplin techniques trade the power consumption and linearity to improve the amplifier noise and ain values. Nevertheless, the circuit needs to use a balun because of its differential input. In order to do the noise-cancellation, the wideband Balun CS- LNA in [5] converts the sinle-ended input to differential sinal at the output and can also be used as an active balun in the receiver. This amplifier with

2 Majlesi Journal of Electrical Enineerin Vol., No., June 07 active feedback and differential output cancels the nonlinearity of the feedback transistor and hence achieves better linearity in compare to the sinle-ended amplifier. The proposed method of [6] relaxes the trade-off between linearity and noise performance in LNAs with an active balun. In this paper, a fully differential CG CS LNA is proposed for usin in UWB receivers. We have combined current reuse and m -boostin techniques. In order to m -boostin, the capacitor cross couplin technique is used in this work. As a result, the dependency between NF and input impedance matchin is reduced. Moreover, a ood power ain over all bandwidth and a hih isolation with ood input/output impedance matchin are achieved. The proposed LNA has a relatively low NF compared with the other LNAs. In Section, the proposed CG-CS LNA is described. First, the CS-CG amplifier with the currentreuse technique is briefly studied. Then, the desin concepts such as bandwidth extension, input/output impedance matchin, and noise performance are analyzed. In Section 3, the simulation results obtained from the proposed fully differential CG-CS LNA are presented. This paper is concluded in Section 4.. PROPOSED CG-CS LNA.. CG CS with the current reuse technique Compared with CS LNA, the CG amplifier offers an appropriate impedance matchin in its input. The CG stae is well-known for a wideband input impedance matchin of R in =/ m where m is the trans conductance of the transistor. However, the amplifier suffers from poorer noise performance and the minimum available NF in the basic CG LNA of Fi. a is + (γ/α) [7]. For CG LNA, NF is approximately independent of frequency and bandwidth [8]. On the other hand, NF in CG is inversely proportional to m, however, we have restricted to increase m because of obtainin a proper input impedance matchin. The main problem of the CS architecture is the narrowband input impedance matchin, which makes it difficult to achieve a broadband input impedance matchin in the presence of parasitic capacitors. In [, 9], some useful techniques are studied applyin to the cascode CS topoloy. In the LNA desin for UWB receivers, the two stae CG-CS topoloies are used widely in the literatures [9, 0]. Fi. b shows a two-stae CG CS LNA. The CG, cascaded to CS topoloy, provides a current-reuse technique. The current-reuse technique reduces the power consumption [0, ]. Broadband impedance matchin is obtained by usin CG at the first stae. Furthermore, the current-reuse technique has staer tunin characteristics and is used to achieve the broadband operation as shown in Fi. c. The first stae resonates at low-frequency f L, and the second stae resonates at hih-frequency f H. Hence, a flat ain in the wide frequency rane can be achieved [], [].... Low-band resonant frequency of the first interstae f L The first stae with parasitic capacitances of the output is shown in Fi. a. The low-band resonant frequency, f L, is determined by the RLC elements, C, C, L, and R. C in is the total capacitance at the input of M, and C d is the parasitic capacitance at the drain of M. Nelectin the parasitic capacitances, the output impedance of the first stae can be calculated as: TL ( s ) Y ( Y R ) Y 3 sc () Where sl sl Y, Y, Y 3 () m m And sl sc (3) m m is the transconductance of M. The output impedance of the first stae is reduced as follows: N L TL (4) DL DL K A (5) Where K s s C L sc (6) () m m And 3 A( s ) s R C C L s C [ R C ( C ) m m ( L L)] sc ( R C) C m (7) The third-order frequency-dependent function of (7) in the denominator of TL (s) will not affect the lowband resonant frequency (f L ) because the poles of TL (s), determined by A(s), are located at the infinite of the frequency spectrum. The voltae ain of the first stae is calculated as v d m T L ( s ) TL ( s ) (8) v i mrs Where R s is the source impedance and m is the transconductance of M. Thus, f L is derived as follows: f L (9) LC It is seen that the low-band resonant frequency can be determined by adjustin C and L.

3 Majlesi Journal of Electrical Enineerin Vol., No., June 07 Fi.. (a) The basic CG LNA stae (b) The final schematic of the CG-CS amplifier with current-reuse technique and output buffer (c) Illustration of full-band frequency response of the staer tunin technique... Hih-band resonant frequency of the second interstae f H Hih resonance frequency of the band, f H, is determined by L and the parasitic capacitance at the drain of M. The schematic of the second stae and its equivalent small sinal circuit is shown in Fi. b. The hih-band resonant frequency, f H, is derived as follows [9]: f H (0) LC T Accordin to (0), f H can be tuned by L... Noise analysis and m -boostin Althouh the CG topoloy provides broadband matchin, the NF is not favorable and the noise performance of the CG LNA is stronly dependent on its input matchin network. Accordin to the noise analysis, the NF can be iven by [9], [3]: d 0 F m () Where γ is the coefficient of the channel thermal noise and d0 is the zero-bias drain conductance. Impedance matchin with m.r s = imposes a condition that prevents the increase in m to reduce NF. In order to reduce dependency between NF and impedance matchin as shown in Fi. c, a neative feedback path with an invertin amplification can be inserted between the source and ate terminals to effectively increase the transistor transconductance [3], [4]. The NF now becomes d 0 F m boostin () G A m, eff Where α = m / d0. It indicates that this technique reduces the channel noise by a factor (+A ne ) under input impedance matchin conditions. Of course, this ne analysis is based on the assumption that amplification stae does not contribute sinificant noise. Invertin amplification (A ne ) in the m -boostin CG LNA can be implemented in various ways. Usin passive elements occupies less space and does not add noise to the circuit. Hence, to boost m in this work, the capacitive cross-couplin technique is used in the differential CG LNA (Fi. d). By choosin a proper value for C c, A ne maxes out to [4]. This technique sinificantly reduces the noise of input transistors and improves the overall NF. Comparin the NF before and after m -boostin technique results in: F m boostin F (3).3. Input impedance matchin The proposed capacitive cross-couplin differential CG-CS LNA with the current-reuse technique is shown in Fi. 3. The parasitic capacitance (C par ) of the source terminal in M (M 5 ) and L S (L S ) provide 50 Ώ impedance matchin. Accordin to Fi. a, C par includes the source-body and source-ate capacitance of M (or M 5 ). Usin this fiure, the input impedance can be calculated as follows: TL in ( s ) SLS SC par Gm, eff r ds (4) Where r ds is the resistance between drain and source of M and TL (s) is the load impedance at the drain of M. In frequencies far from the input resonant frequency, the impedance of (/ S C par ) S Ls is relatively lare and can be inored. Then, the input impedance of (4) reduces to: 3

4 Majlesi Journal of Electrical Enineerin Vol., No., June 07 in TL Gm, eff r ds (5) Over a wide frequency rane in the CG stae, the impedance TL (s) is smaller than r ds. Thus, the input impedance matchin of 50 Ώ can be achieved under G m,eff = m and to ensure that / m =50Ώ the bias current and the size of the transistors M and M 5 can be adjusted by usin less power consumption. The source followers are added as buffers by insertin transistors M 4 and M 8 that are cascaded to the second stae. Fi.. (a) Schematic of the first stae and its equivalent small sinal (b) Schematic of the second stae and its equivalent small sinal (c) CG LNA stae with m -boostin feedback amplifier (d) Capacitive cross-couplin technique in a differential CG LNA. Fi. 3. Schematic of the proposed LNA 3. SIMULATION RESULTS The proposed fully differential CG CS amplifier with the current reuse and m -boostin techniques (Fi. 3) is desined and simulated usin ADS in TSMC 0.8 μm CMOS RF technoloy. The amplifier and the output buffers consume 5.6 ma and 5 ma from a.8 V DC supply, respectively. Then, the core power consumption is 0. mw and the buffer power consumption is 9 mw. As amplifier is fully differential and symmetric, the instance parameters are shown only for one side of the circuit in Table. Accordin to Fi. 4, a minimum NF of.8 db is achieved at frequency 5 GHz where the NF is db in the whole GHz UWB band. The input and output return losses are <-0.3 db and <-.3 db, respectively. Fi. 5 shows a flat small sinal ain in the entire GHz UWB band with a maximum ain of 4. db. Over the entire 7.5 GHz bandwidth, the isolation (S ) is <-50 db. Fi. 6 indicates an input third-order intercept point (IIP3) of -5.0 dbm applyin two tones with a 00 MHz spacin at 5.0 GHz RF sinals. In Table, we summarized the simulation results of this work. The results are also compared with those of some other reported LNAs usin similar technoloy. It 4

5 Majlesi Journal of Electrical Enineerin Vol., No., June 07 is seen that the proposed LNA has ood performance comparin with the other previous works and it outperforms most of them, especially, in NF and power ain. 4. CONCLUSION In this paper, a fully differential LNA for UWB application is desined with CG CS topoloy. The current reuse technique is used to achieve a wideband and to reduce the power consumption of the core to 0. mw. Moreover, the m -boostin technique is employed to improve the NF of the amplifier, db in the whole GHz UWB band, and this technique is implemented by usin the capacitive crosscouplin scheme. Compared with the other previous works in similar technoloy, the proposed LNA has a lower NF and consumes reasonable power. Furthermore, a flat and hih power ain is obtained in this work. Across the UWB frequency band, the power ain of.4-4. db is achieved. In addition, over the entire 7.5 GHz bandwidth, the input and output matchin, S and S, and the isolation, S, are <-0.3 db and <-.3 db, and <-50 db, respectively. M M M 3 M 4 Table. Summary of the instance parameters 8/0.8 (μm/ μm) C.6 pf L 9 nh R 5 kω 8/0.8 C 7 L. R 3 5 (μm/ μm) pf nh kω 44/0.8 C c 4 Pf L S 0.9 R 5 5 (μm/ μm) nh kω 4/0.8 C 6 7 V dd.8 (μm/ μm) pf V Fi. 4. (a) Simulated results of NF versus frequency (m9, m8 and m0 show NFs of 3. GHz, 0.6 GHz and minimum NF frequencies, respectively) (b) Simulated results of output return loss versus frequency (m6 and m7 for 3. GHz and 0.6 GHz) (c) Simulated results of input return loss versus frequency (m4 and m5 for 3. GHz and 0.6 GHz) 5

6 Majlesi Journal of Electrical Enineerin Vol., No., June 07 Fi. 5. Simulated results of (a) small-sinal power ain (b) inverse ain. Fi. 6. IIP3 with two input RF sinals of 5.0 GHz with 00 MHz frequency spacin References Technoloy Table. Performance summary and comparison BW Gain NF S [GHz] [db] [db] [db] IIP 3 [dbm] VDD [V] P dc [mw] S [db] This Work [5] < 0.3 < (Core) 30 < 50 < 65 [9] < (Core) < 43 [8] < [3] < < 60 [] < < 30 [] < < 40 [7] 0.3μm CMOS < [0] 0.3μm CMOS < < 8 REFERENCES [] FCC, Ultra-wideband transmission systems, Federal Reister, Vol. 67, No. 95, Feb.4, 00. [] K.-H. Chen, J.-H. Lu, B.-J. Chen, and S.-I. Liu, An ultra-wide band GHz LNA in 0.8-m CMOS, IEEE Trans. Circuits Syst. II, Exp. Briefs, Vol. 54, No. 3, pp. 7, Mar [3] I. Mohammadi, A. Sahafi, J. Sobhi,. Daei Koozehkanani, A linear, low power,.5-db NF LNA for UWB application in a 0.8 μm CMOS, Microelectronic Journal, Vol. 46, Issue, 05, pp [4] J. del Pino, R. Diaz, S.L. Khemchandani, Area reduction techniques for full interated distributed amplifier, Int. J. Electron. Commun., Vol. 64, Issue, 00, pp [5] S. M. Rezaul Hasan, Analysis and desin of a multi-stae CMOS band-pass low-noise preamplifier for ultra-wide-band RF receiver, IEEE Trans. Very Lare Scale Inter. Syst., Vol. 8, Issue 4, 00, pp [6] C. F. Liao, S. I. Liu, A broadband noisecancelin CMOS LNA for GHz UWB receivers, IEEE J. Solid-State Circuits, Vol. 4, Issue, 007, pp [7] Q. Li, Y.P. han, A.5-V 9.6-GHz inductorless low-noise amplifier in 0.3- m CMOS, IEEE Trans. Microw. Theory Tech., Vol. 55, Issue 0, 007, pp [8] A. Ismail, A. A. Abidi, A 3 0-GHz low-noise amplifier with wideband LC-ladder matchin network, IEEE J. Solid-State Circuits, Vol. 39, Issue, 004, pp [9] R. M. Wen, C. Y. Liu, P. C. Lin, A low-power full-band low-noise amplifier for ultra-wideband receivers, IEEE Trans. Microw. Theory Tech., Vol. 58, Issue 8, 00, pp [0] S. Arshad, R. Ramzan, K. Muhammad, Q. Wahab, A sub-0mw, noise cancellin, wideband LNA for UWB applications, Int J. Electron. Commun., Vol. 69, Issue, 05, pp [] M.T. Hsu, Y. C. Chan, Y.. Huan, Desin of low power UWB LNA based on common source topoloy with current-reused technique, Microelectronic Journal, Vol. 44. Issue, 03, pp

7 Majlesi Journal of Electrical Enineerin Vol., No., June 07 [] J. Shim, T. Yan, J. Jeon, Desin of low power CMOS ultra wide band low noise amplifier usin noise cancelin technique, Microelectronic Journal, Vol. 44, Issue 9, 03, pp [3] S. iabakhsh, H. Alavi-Rad, M. CE. Yaoub, A hih-ain low-power -4 GHz ultra-wide-band CMOS LNA for wireless receivers, Int. J. Electron. Commun., Vol. 66, 0; pp [4] W. huo, X. Li, S. Shekhar, S. H. K. Embabi, J. P. de Gyvez, D. J. Allstot, E.S. Sinencio, A capacitor cross-coupled common-ate low-noise amplifier, IEEE Trans. Circuits Syst. II: Express Briefs, Vol. 5, Issue, 005, pp [5] G.. Fatin, H.. Fatin, A wideband balun- LNA, Int. J. Electron. Commun., Vol. 68, Issue 7, 04, pp [6] F. Akbar, M. Atarodi, S. Saeedi, Desin method for a reconfiurable CMOS LNA with input tunin and active balun, Int. J. Electron. Commun., Vol. 69, Issue, 05, pp [7] Y. Shim, C. W. Kim, J. Lee, S. G. Lee, Desin of full band UWB common-ate LNA, IEEE Microw. Wirel. Compon. Lett., Vol. 7, Issue 0, 007, PP [8] B. Park, S. Choi, S. Hon, A low-noise amplifier with tunable interference rejection for 3.-to 0.6-GHz UWB systems, IEEE Microw. Wirel. Compon. Lett., Vol. 0, Issue, 00, pp [9] T. K. Nuyen, C. H. Kim, G. J. Ihm, M. S. Yan, S. G. Lee, CMOS low-noise amplifier desin optimization techniques, IEEE Trans. Microw. Theory Tech., Vol. 5, Issue 5, 004, pp