Published in: Proceedings of the 26th European Solid-State Circuits Conference, 2000, ESSCIRC '00, September 2000, Stockholm, Sweden

Transcription

1 Uing capacitive cro-coupling technique in RF low noie amplifier and down-converion mixer deign Zhuo, Wei; Embabi, S.; Pineda de Gyvez, J.; Sanchez-Sinencio, E. Publihed in: Proceeding of the 6th European Solid-State Circuit Conference, 000, ESSCIRC '00, 19-1 September 000, Stockholm, Sweden DOI: /ESSCIR Publihed: 01/01/000 Document Verion Publiher PDF, alo known a Verion of Record (include final page, iue and volume number) Pleae check the document verion of thi publication: A ubmitted manucript i the author' verion of the article upon ubmiion and before peer-review. There can be important difference between the ubmitted verion and the official publihed verion of record. People intereted in the reearch are advied to contact the author for the final verion of the publication, or viit the DOI to the publiher' webite. The final author verion and the galley proof are verion of the publication after peer review. The final publihed verion feature the final layout of the paper including the volume, iue and page number. Link to publication General right Copyright and moral right for the publication made acceible in the public portal are retained by the author and/or other copyright owner and it i a condition of acceing publication that uer recognie and abide by the legal requirement aociated with thee right. Uer may download and print one copy of any publication from the public portal for the purpoe of private tudy or reearch. You may not further ditribute the material or ue it for any profit-making activity or commercial gain You may freely ditribute the URL identifying the publication in the public portal? Take down policy If you believe that thi document breache copyright pleae contact u providing detail, and we will remove acce to the work immediately and invetigate your claim. Download date: 15. Jul. 018

2 Uing Capacitive Cro-Coupling Technique in RF Low Noie Amplifier and Down-Converion ixer Deign Wei Zhuo, Sherif Embabi, *Joé Pineda de Gyvez, Edgar Sánchez-Sinencio Analog ixed-signal Center, Texa A& Univerity, College Station, Texa, U. S. A. * Philip Reearch Laboratorie, Eindhoven, The Netherland. Abtract We report an approach to improve the noie performance of RF low noie amplifier (LNA) and down-converion mixer. The technique we decribed here i baed on capacitive cro-coupling acro the two ide of a differential input tage. A LNA and mixer have been implemented in 0.5µ COS proce to demontrate the viability of thi technique. The meaurement how that the LNA achieve 3.0dB noie figure and 1. db voltage gain (optimized for the maximum power tranfer), and the mixer ha 5.dB DSB noie figure and 13. db voltage converion gain. Both LNA and mixer operate at.7v voltage upply and conume 7mW and 8.1mW power, repectively. 1. Introduction Growth of the wirele communication market put increaing demand on low power, low cot and high performance receiver. RF LNA and mixer are among the mot critical building block in the receiver chain. The COS technology ha hown it feaibility for high frequency application [1-]. Thi paper decribe how a capacitive cro-coupling (CCC) technique can be ued for RF amplifier and mixer to improve the performance. Section II dicue a capacitive cro-coupling (CCC) technique. Section III preent the LNA implementation uing the CCC technique. Detail of mixer implementation uing CCC are decribed in Section IV. Experimental reult are hown in Section V, followed by concluion in Section VI.. Capacitive cro-coupling technique for noie reduction Common gate and common ource input tage are tructure widely een in the LNA deign. The main drawback of common-gate amplifier i their relatively high noie figure. Ignoring the noie contribution due to the load, a common-gate LNA ha a minimum NF [1]: F = 1+ γ (1) Where, γ repreent the channel thermal noie coefficient. It i uually difficult for common-gate LNA to provide a NF well below 3dB. A Common ource input tage offer the poibility to achieve the bet noie performance by increaing their input quality factor Q. However, it degrade the linearity and increae the enitivity of the input matching. Common-gate amplifier can be more eaily matched and uually exhibit better linearity than common-ource amplifier. Fig. 1 Capacitive cro-coupling tech niq ue The capacitive cro-coupling (CCC) ha been ued for gain enhancement and matching [3]. In thi ection, we will demontrate how the CCC can be ued to improve the NF of a common-gate input tage. Fig.1 how the chematic of a differential common-gate input tage with CCC. It can be hown that the cro-coupling n caue the noie of 1 and, namely v and v, to produce common-mode noie voltage at the output node d1 and d. Intuitively, Thi can be explained a follow. Let aume the following intantaneou polarity for noie ource v ; poitive polarity at the gate terminal of input tranitor 1 and negative at the other ide (polarity can be exchanged a well). One can eaily ee that the drain voltage of 1 will decreae due to the poitive diturbance at the gate terminal. On the other hand, becaue of the negative diturbance at S, the drain voltage of will alo decreae. Hence, the input noie ource v caue a common-mode voltage at d1 and d. Similar analyi can be applied to the input noie ource d1 d V V n Inp S1 v. n 1 C 1 A more rigorou mall ignal noie analyi wa ued for a mathematical derivation. Here, only channel C S Inn

3 thermal noie ource i conidered, and other noie mechanim are neglected for implicity. The crocoupling capacitor C 1 and C are aumed to be much greater than gate-ource capacitance C g. The derived noie factor expreion of the input pair 1 and i given by : iolation of the LNA. Tranitor 5, 6, 7, and 8 are ued to realize large reitor to iolate the ignal path from the biaing circuitry. The LNA load i a 300Ω reitance (auming an on-chip mixer). F = 1+ γ () It can be een that due to the capacitive cro-coupling the noie contribution from the input tranitor i reduced though not completely eliminated. 9 R 1 R I bia outn outp VC C Worthy to mention i that the derivation of the noie factor aume that the input impedance i matched to the ource impedance for maximum power tranfer. atching for the optimal noie performance can reduce the NF value. Thi can be demontrated by changing the effective input impedance (1/ g m ). Fig. illutrate the change of the NF v. the effective tranconductance ( g m ). The noie performance will be improved a the tranconductance i increaed. Another advantage of the common gate input tage with CCC i that the input effective tranconductance i doubled due to the capacitive cro-coupling. Thu, the current conumption can be reduced. Finally, it i alo intereting to note that the capacitive cro-coupling doe not add much cot and complexity. C 3 Inp Fig. 3 Sch ematic ofthe LNA with CCC A pointed out in the previou ection, the LNA with CCC provide noie advantage over the conventional differential common-gate LNA. However, the dynamic range of a circuit i not only determined by the noie but alo by the linearity. Auming that the tranconductance g m i the only ource which caue nonlinearity, the input referred IP3 of the conventional differential commongate LNA and the LNA with CCC are derived uing volterra erie. The derived expreion are given in (3) and (4), repectively. L 1 C L Inn IP3 = 8 3 ( K G 3gm + K gm / G ) (3) IP3 G = 4 (4) 3 K 3gm Fig. NFofthe inputpair w ith CCC v. gm 3. LNA Implementation uing the CCC Technique The full chematic of a LNA employing the CCC technique i depicted in Fig.3. The input impedance matching i achieved by izing and adjuting the bia current of the input tranitor 1 and uch that 1/( g m )=50Ω. The cro-coupling capacitor C 1 and C are 10pF poly-to-poly capacitor in our implementation. Off-chip inductor L 1 and L are ued to reonate with the gate-ource capacitor C g and the input paraitic capacitance at the frequency of interet. Cacode tranitor 3 and 4 are added to improve the revere Equation (3) and (4) are implified expreion evaluated under input matching condition. The parameter K gm and K 3gm are the econd-order and thirdorder tranconductance nonlinear coefficient. G i the ource conductance. It i oberved that the IP3 of the conventional differential common-gate LNA i not only determined by the third-order nonlinear coefficient but alo by the econd-order nonlinear coefficient. The IP3 of the LNA with CCC i only influenced by the thirdorder nonlinear coefficient. For a firt-order analyi, we may conider the OSFET tranitor to be a quare-law device. Thi mean that the third-order nonlinear coefficient i uually much maller than the econd-order one. Thu, the IP3 of the LNA with CCC i higher than that of the conventional differential common-gate LNA. Although the econd-order intercept point (IP) i not important in a heterodyne receiver, it i a limiting factor in a direct converion receiver. The IP of a differential circuit i uually limited by mimatch between differential path. Auming that only tranconductance

4 (g m ) ha mimatch ( g m ) and applying the ame procedure for the IP3 modeling, the input-referred IP of both LNA are given by (5) and (6), repectively. gm 16 G IP (5) 3 K g m IP (6) Due to mimatch, the conventional differential commongate LNA how a finite IP, while the mimatch effect i minimal on the IP of the LNA with CCC. 4. ixer Implementation uing the CCC Technique R C 3 C 4 IF+ 1 R I bia LO LO C 9 1 C 3 RF+ RF- C L 1 L IF- LO+ 5. Experiment Reult Both LNA and mixer have been implemented in an AI 0.5µ COS proce. eaurement of the LNA how a 1.dB voltage gain and a NF of 3.0dB at 900Hz. The 3.0dB NF, which eem higher than recently reported value, can bet be explained by the fact that the LNA wa deigned for maximum power tranfer and not optimized for the noie figure. One can achieve better noie performance by increaing the tranconductance a illutrated in Fig.. The gain and NF of the LNA are plotted in Fig. 5 with frequency wept from 800 to 950. The input-referred IP3 i extracted by applying two tone tet. The power of both tone wa wept while the reulting third-order intermodulation product wa meaured. Fig. 6 how the fundamental tone and the third-order intermodulation product at 899.5Hz and 900Hz. Since output impedance of LNA i not 50Ω, VdBm i defined a voltage correponding to a power level of 0dBm in a 50Ω ytem. Becaue ource reitance i 50Ω, the input VdBm i ame a original dbm. 11 Fig. 4 Sch ematic ofthe mixe r w ith CCC The meaured mixer DSB noie and converion gain are 5.dB and 13.dB, repectively. The meaured NF i lower than mot publihed mixer. The reult of a twotone meaurement on the mixer are hown in Fig. 7. For the mixer meaurement, the LO drive amplitude i equivalent to 300mV. Both LNA and mixer prototype are meaured in TQFP package with ESD protection. The meaured LNA and mixer reult are ummarized in Table.1 and Table. repectively. icrophotograph of both circuit are hown in Fig. 8. In the Gilbert cell mixer, the linearity i uually determined by RF tranconductance. Thu, mot of the deign effort i pent attempting to find better way of providing V-I converion. One imple, but powerful technique conit of uing degeneration reitor in the ource of V-I converter. But, additional reitor will penalize the mixer noie performance, which may, in turn, increae deign difficultie for the circuitry preceding it. The capacitive cro-coupling technique depicted in previou ection can be ued in RF V-I converter to minimize the noie contribution and improve the linearity imultaneouly. The propoed double-balanced Gilbert cell mixer with CCC i hown in Fig. 4. Tranitor 1 and, cro-coupled by capacitor C 1 and C, form a differential tranconductance tage. The drain current of the 1 and i teered through 3 and 4 or 5 and 6 to perform the mixing function. Output capacitor C 3 and C 4 combined with output reitance R 1 and R to form a low pa filter. The purpoe of thi filter i to attenuate high frequency component at the IF output. Biaing of the mixer i imilar to that of the LNA. 6. Concluion In thi paper, we have hown how the capacitive cro-coupling technique can be ued to improve the noie performance of a common-gate input tage. A LNA and mixer with CCC have been implemented in 0.5µ COS proce to verify our theoretical finding. The LNA achieve around 3.0dB NF up to 900Hz, In addition, the LNA ha ignificant +6.7dBm inputreferred IP3 and only conume 0mW. The mixer ha very low DSB NF at 5.dB and +13.8dBm output IP3. Yet, both circuit do not rely on uing on-chip piral inductor. Reference [1] A. Rofougaran, J.Y-C.Chang,. Rofougaran, A. A. Abidi, A 1GHz COS RF front-end IC for a direct converion wirele receiver, IEEE J. Solid-State Circuit, July 1996, vol. 31, pp [] A. R. Shahani, D. K. Shaffer, and T. H. Lee, A 1-mW Wide Dynamic Range COS Front-End for Portable GPS Receiver, IEEE J. Solid-State Circuit, Dec 1997, vol. 3, pp

5 [3] T. Cho et al, A Single-Chip COS Direct-Converion Tranceiver for 900Hz Spread-Spectrum Digital Cordle Phone, ISSCC Diget of Technical Paper, Feb 1999, pp Table.1 Sum mary oflna meaurement Frequency 900Hz NF Gain (voltage) IP3 (input) IP (input) 1dB (input) 3.0dB 1.dB +6.7dBm +34dBm -7dBm Power Supply.7V Power Conumption 0mw Fig. 5 eaured LNA gain and NFv. frequency Silicon area mm Table. Sum mary ofmixe r meaurement RF Frequency 900Hz LO Frequency LO Amplitude NF (DSB) Gain (voltage) IP3 (output) IP (input) 910Hz 300mV 5.dB 13.dB +13.8dBm +5.6dBm Fig. 6 eaured LNA IP3 Power Supply.7V Power Conumption 8.1mw Silicon area mm Fig. 7 eaured m ixe r two tone tet Fig. 8 icroph otograph ofthe LNA and m ixe r