CMOS low-noise amplifier for 6-11 GHz MB-OFDM UWB wireless radio system

(1)

CMOS

Low-Noise

Amplifier for

6-1

lGHz

MB-OFDM

UWB

Wireless Radio

System

Zhe-Yang

Huang',

Che-Cheng Huang2

Dept.of Communication Engineering, National Chiao Tung University, Hsin-Chu, Taiwan

2Instrument

Technology Research Center, NARL, Hsin-Chu, Taiwan

Abstract - This paper presents a low-power low-noise M ixer

amplifier (LNA) with switching bands for MB-OFDM Group-C /

andGroup-Dultra-wideband wireless radiosystem.TheLNA is designed and implemented in TSMC 0.18um RF CMOS process.

Simulation results show that power gain of12.4dB,inputand PLL DSP

output matching lower then -8.5dB and -14.5dB, and a r

-minimum N1F of 4.0dB can be achieved, while the power I

consumptionis 11.2 mWthrough1.8V powersupply. A D

I. INTRODUCTION Fig.2UWBTransceiver Architecture of MB-OFDM

The Ultra Wide-Band (UWB) is a new technology for

wireless personal area networks (WPAN) which contains II. DESIGN OF ULTRA-WIDEBAND LNA manyadvantages ofhigh data transmission rate, low cost, and

low power consumption that is better than narrowband A. Wideband

Amplifier

Design

wireless system.InFebruary, 2002,UWB system is approved A sketch of the wideband

amplifier

which contains

input

by

the Federal Communications Commission

(FCC)

[1]*

matchmg network,

main amplifier

and output buffer and that Although the UWB standard (IEEE 802. 15.3a) has not been is shown as

Figure

3. The

specification

of ultra-wideband defined which there are two specifications, DS-UWB and system is definedas3.

1GHz-10.6GHz,

thereforeaverywide

MB-OFDM,

and this paper

mainly design

for MB-OFDM

bandwidth

input matching

network is necessary

in

the UWB

tJWB system in the band of 6.0GHz-IlGHz for future LNA In this wideband

amplifier,

the main

amplifier

is

applications. The band definition of MB-OFDM is illustrated _including _the _low-noise _{amplifier and switching}, _capacitors, in

Fig.

1 which extended from 3168MHzto

10296MHz,

and and the

switching capacitors

are

design

for

switching

groups

the bandwidth of MB-OFDM is

containing Group-A,

for MB-OFDM UWB

wireless radio

system. For Group-C andGroup-D.; and theproposedLNAisdesignfor measurement

consideration,

theoutput

impedance

is

always

Group-C and Group-D. The Group-B is not considered in

designed

for 50 ohms i the output buffer. The

design

current UWB system which causedbythe U-NII band and g

WLAN (IEEE 802.11a). considerations_return_{loss, power gain, and noise figure (NF), linearity}oflow-noise amplifier are mainly in input

_(P1dB,

IIP3) and power consumption, but thereare some trade-off

GROUPA GROUP B GROUP C GROUPD

between theseimportantcharacteristics.

iBand Band Band Band Ban.d Ban.d Ban.d Band Band Band Band Ban.d Ban.d #1

#2 #3 #4 #5 #6 #7 #8 #9 #10 #11 #12 #13

4 _

~~~~~~~~~~~~~~~~0

RF IN

I

RF OUT

3432 3960 4488 5016 5808 6336 6864 7392 7920 8448 8976 9504 10032 |Matchingf LN

MHzll xnu MHzll Ng-L MHz MHzll xnu MHzll MHzll xnu MHz MHz MHz NetworkLNATttXI

Fig. IBand Definition of MB-OFDM

2) T~~~~~CslCs2 Cs3 Buffer ()

In RF wireless receiver, LNA is one of themost critical

building blocks caused by the noise figure is dominated inI't w W3 stage of the receiver that is illustrated inFig. 2. For LNA

design, there are many trade-off between different - - -

-specifications. For example, the power gain affects noise Fig.3WidebandAmplifier

figure, the die areaaffect cost, and the power consumption

affects thebatterylife. B. Proposed Low-Noise

Amplifier

IntheFigure 4, the proposed low-noise amplifier comprises the input matching network which is implemented by the

588

1-4244-0797-4/07/$20.00

( 2007 IEEE

(2)

Chebyshev Filter

[3]

where are Li, Ci, L2, C2, Lg, Ls and Cgsi; Rs

Ll

Cl

Zins

Ls Cgsi the main amplifier is containing Mi, M2, LLI, Csl, Cs2, Cs3,

SWi, SW2and SW3; the output buffer is the M3; and Lsp is '1

the series peakinginductor,the Cdcl, Cdc2 are DC blocking 5

capacitors. The voltage gain of cascade amplifier is Av which j Vs L2 T RL is described in (1), where gmi is the transconductance of Mi, Zin

LLI is the loading of the cascode amplifier Ccaps is the total __

capacitor which including the switching capacitor and Fig.6 Input Matching Network in Small Signal Models

parasitic capacitors. And Zin is expressed in equation (3), B(s) is the equivalent impedance of Li, L2, Ci and C2; Zin is an impedance of a

6th

VDD orderband-pass filter which is too complicated to express in

F,g VDD theequation; and the equation (3) is the basic form.

VDD

_El

I

T; |LL1S

_F,

-Zns

=

_9~~~~~in

sL

_SCC+

S

+ ml

Ls

sg + +

TLs

(2)

gsl

gsl gsI T

1

M3 RF_OUT Csl s2

Cs3

Cd

Zin

=

(sLI

+

)

(sL2

)

(sLg +

+

w,

LS)

Lsp

cd~1

s RF_IN

LI

CI

TC2

_S _S_1W =B(s)II(sL + 1 +

Ls)

Lg

~~ ~ ~ ~

]g SgsiI OS ()

DI~.7~i

1

MI -L2

(3) D. Series

Peaking

VBias

Fig.4ProposedLow-NoiseAmplifier

vWhen

designing

a

wideband

amplifier,

the most

critical

problem is the bandwidth limited by large capacitance which

A = -s9 1(1) affects the

high frequency

response.

Thus,

the

high frequency

1+

g,1Ls

+

s(LLICcas

+ L L

Ccp

) response is difficult to reach higher frequency. Therefore,

series

peaking

is a good

solution

to solve the

important

C. Wideband InputMatching Network problem that is adding inductance Lsp alongthe RF path.

The wideband

input matching

networkare

containing

Li

Figure

7 illustrates the

frequency

response with series

CI, L2, C2, Lg,LsandCgsIwhichis shown in

Figure

5;and

peakig

inductor, the inductor raise the

high frequency

is

input impedance

fromthe source terminal. The modified response. In the other way, series

peaking

is also could be

equivalent

circuit with small

signal

models is shown in the viewed in timeconstantissue. And the timeconstantwithout Fig. 6. The equivalent input impedance Zins is expressed in Eq.

series peaking

is shown in (4); adding the inductor Lsp, and 2where

RL

=

cLs.

RLis

always design

for 50 ohms;

Lg

thenewtimeconstantisderivedas

(5).

The

Ceq

ismodifiedas

(2),Cgs

I designtocancel each other in image

impedance.

a newcapacitor, and the capacitance is alsosmaller; forthis

and

Cgsl1

are design to cancel each other in imageimpedance. reason, the time constant is becoming reduced.

Zins 20- I I I RFIN LI C1 TV ml v V

('V~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

_ _ ~~~~~~~~~~~~~~~-*-Q.QQnH L- --O--.59nH . -. -_ n .On

Fig.5 Input Matching Network -- 1.36nH

SG 6G 7G SG 9G lOG 11G 12G

Frequency

Fig.7Inductance of SeriesPeaking

(3)

1

eRC

rq

Cgs2

(4)

2

--Band#12,13

--- -- -- 11

Band#8, 9

where Ceq = (Cgs2 1lCeaps) ' -4- ._\ . -v Band#6, 7

req 4-1+( m m 2 )r 2 ] o o ---T---,----

I---n---I--n---I---2=RC

=req

.*Ceq

(5) -12- -- ---

--->~~~---64- ---~~

where

C = 2spcaps

-10-

2 .

z'2

=RC

(C00 r

Wcaps ° 2.OG 4.0G 6.QG 8.oGS 10.OG 12.OG

Frequency

Fig.8 Input Return LOSS(Si11)

req = [1 +

(g,12

+ gmb2

)'o2 ]iOl

+

rO2

III. SIMULATION RESULTS -.-Band#12,13

\ ~~~~~~~~BandX8, 9

The simulation results of the proposed

UJWB

LNA using 13 .^ ---Band# 6,7 Agilent ADS2005A simulator are given in Figure 8 to Figure

12. In Figure 8 that can be seen the input return loss (511) are _

lower than -12.7dB/-9.5dB/-1 1.3dB/-8.5dB (Band #6,

7/1#8,

9/ r~5 4 #10, 11/ #12, 13) between 6.0GHz to 10.6GHz. In Figure 9 -;

that can be seen that the output return loss (S22) are lower than 203 7 -14.5dB/-14.8dB/-15.8dB/-15.6dB (Band #6, 7/#8, 9/#10, 11/

#12, 13)between 6.GHzto 10.6GHz, respectively. The power gain whose peak value are 8.3dB, 10.8dB, 12.4dBandl11.7dB ; ; ; ; .

at 6.5GHz, 7.5GHz, 8.9GHz and 9.7GHz which covers the

0o20G

4.0G 60G 80G 100G 12.0G

Group-C and Group-D ofthe MB-OFDM, and that is shown in Frequency

Figure 10. In Fig. 11, it can be seen that the noise figure is Fig.9Output Return Loss (S22) below 4.9dB, 4.3dB, 4.1dB and 4.5dB between 6.0GHz to

10.6GHz and the minimum noise figure are 4.8dB, 4.1dB,

4.0dB and 4.0dB at 6.6GHz, 7.5GHz, 8.8GHz and 9.5GHz 20 -'---r--'---r

I--(Band #6, 7/ #8, 9/ #10, 11/ #12, 13) through 1.8V supply 0

voltage. In Fig. 12, the input-referred 1dB compression point (IPldB) is -15dBm, -17dBm, -18dBm and -17dBm at 6.5GHz, o

7.5GHz, 9.0GHz and 10.0GHz (Band #6, 7/ #8, 9/ #10, 11/ 10 ---g---16---#12, 13). Third-Order Input Intercept Point (IIP3)

[528MHz]-t

at6336MHz and6864MHzis-4dBm

(Band

7);[10M

#6, Hz] at

---20---

--I--- ----I---

I---6864MHz and 6874MHz is -3dBm (Band #6, 7); Third-Order 30 ;

Ceq~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~l

Input Intercept Point [528MHz] at 6864MHz and 7392MHz

is--2dBm (Band #8, 9);[10MHz] at 7392MHz and 7402MHz is ~ 3

-ldBm (Band #8, 9); Third-Order Input Intercept Point 5Q

-Band#1O,

11 [528MHz] at 8448MHz and 8976MHz is +ldBm (Band #10, 60

Band# 8,

9 11);[10MHz] at 8448MHz and 8458MHz is +ldBm (Band --ad

#10, 11); Third-Order Input Intercept Point

[528MH-Iz]

at 00 2OG 4.0G 60G 8.0G 10.OG 120G

6336MHz and 6864MHz is +ldBm (Band #12, 13),[10MHz] Frequency

at 6336MHz and 6346MHz is +ldBm (Band #12, 13); The

Fig.

1

PowerGain(S21)

power consumption is 11.2mW through 1.8V supply voltage

which neglects the power of output buffer. And the layout is illustrated inFigure 16.

(4)

8

-*-Band#12,13'

Band 1 6,7 #8,9 #10,11 #12,13 -E-Band#1O,11 FI 511I (dB) < -12.7 < -9.5 < -11.3 < -8.5 Band# 8, 9 Band# 6, 7 S22 (dB) <-14.5 <-14.8 <-15.8 <-15.6 S21(dB) 5.9-8.3 8.7-10.8 9.5-12.4 8.7-11.7 - S21 Max.(dB) 8.3 10.8 12.4 11.7 * X;5;m m . Working Bandwidth (GHz) 6.0-7.2 7.0-8.2 8.1-9.3 8.7-10.6 LLm \ -++ t. _ NF(dB) 4.8-4.9 4.1-4.3 4.0-4.1 4.0-4.5 IP1 dB(dBm) -15 -17 -18 -17 X i IIP3(dBm) [lOMHz] -3 -1 +1 +1 IIP3(dBm) [528MHz] -4 -2 +1 +1 PowerConsumption (mW) 11.2 11.2 11.2 11.2

0.0 2.OG 4.OG 6.OG 8.OG 1MG 12.OG Table.1Performance Conclusions

Frequency Fig.1I Noise Figure

14 ACKNOWLEDGMENT

12- *++++++ +_...++ X@ _,. The authors would like tothank the chip implementation center(CIC)for technical support.

10-._

i..

O , , A ~~~~~~~~~~~~~REFERENCES

[1] FCC,"FinalRule of the FederalCommunications Commission, 47 CFR

^ . t

~~~~~~~~~~~~~~~~~~Part

15,Sec. 503", Federal Register, vol. 67,no. 95,May 2002.

2 -*-Band4l21v,13 9 [2] _

-9BandS1O, 11 4Cellll)6

~4 Band# 8, 9. . _

e

TBand#

6, 7 . [3] Andrea Bevilacqua, and Ali M. Niknejad," An Ultrawideband CMOS -60 -50-40-30 -20 10 0

~~~Low-Noise

Amplifier for 3.1-10.6-GHz Wireless Receivers "IEEE

InputPower~dBm) JOURNAL OF SOLID-STATE CIRCUITS, Vol. 39, No.

Fig. 12 Input-Referred 1dB Compression Point (P1dB)12p25926De204

[4] Chang,C.-P.; Chuang, H.-R., '0.18 um 3-6 GHz CMOS broadband LNA for UWB radio", ElectronicsLetters, Volume 41, Issue 12, June 2005

IV. CONCLUSIONS Page(s):33-34.

A CMOSlow-nise aplifir is esignfor M-OFDM [5] Chang-Wan

Kim, Min-Suk

Kang, Phan Tuan Anh, Hoon-Tae

Kim,

and

Ultra-Wideband wireless radio system which using the Sang-Gug Lee, 'An Ultra-Wideband CMOS LowxNoise Amplifier for

switching capacitor technique to implement the multi-group 3-5-GHz UWB System", IEEE JOURNAL OF SOLID-STATE applications. The simulation results show that the proposed CIRCUITS, Vol. 40, No.2,pp.544-547,Feb. 2005.

UWB LNA gives 8.3dB/10.8dB/12.4dB/1 1.7dB (Band#6,7/ [6] Chih-Fan Liao; Shen-IuanLiu, "A broadband noise-canceling CMOS #8, 9/ #10, 11/ #12, 13) maximum power gain between LNA for 3.1-10.6-GHz UWB receiver, Custom Integrated Circuits 3.1GHz to 10.3GHz while consuming 11.2mW though 1.8V Conference, 2005. Proceedings of the IEEE 2005, 18-2t1 Sept. 2005

power supply. Page(s):161-164.

[4] 13-stagescomm.-sourceT0.l8umCMOSI <-12.2 <-10.1 T13.5-15.81 3.0-6.0 15.7 T4.7-6.7 T 4.7 59.4

[5] |2-stageshunt-peaked

t0.l8um

CMOS |<-9.5 |<-8.1 T15.3-18.0 |3.1-5.0 |18.03 T2.7-3.1 T 2.7 |25.9 |

[6] 3-stages shunt-peaked 0.18um CMOS <-7 < -12 6.7-9.7 1.2-11.9 9.7 4.5-5.1 4.5 20.0

This Work Switching Capacitors 0.1l8umCMOS < -8.5 <-14.5 5.9-12.4 6.0-10.6 12.4 4.0-4.9 4.0 11.2 Table.2 PerformanceComparisons