-
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise
Amplifier
by Heng Jin
A thesis submitted in conformity with the requirements for the
degree of Master of Applied Science
Department of Electrical and Computer Engineering UniversiSr of
Toronto
2000
O Copyright by Heng Jin 2000
-
National Library 161 of Canada Bibliothèque nationale du Canada
Acquisitions and Acquisitions et Bibliographie Services services
bibliographiques 395 Wellington Street 395, nie Wellington OttawaON
K1AON4 Ottawa ON K1A ON4 Canada Canada
The author has granted a non- L'auteur a accordé une licence non
exclusive licence ailowing the exclusive permettant a la National
Library of Canada to Bibliothèque nationale du Canada de reproduce,
loan, distnbute or seil reproduire, prêter, distribuer ou copies of
this thesis in rnicroform, vendre des copies de cette thèse sous
paper or electronic formats. la forme de rnicrofiche/film, de
reproduction sur papier ou sur format électronique.
The author retains ownenhip of the L'auteur conserve la
propriété du copyright in this thesis. Neither the droit d'auteur
qui protège cette thèse. thesis nor substantial extracts &om it
Ni la thèse ni des extraits substantiels may be printed or
otherwise de celle-ci ne doivent être imprimés reproduced without
the author's ou autrement reproduits sans son permission.
autorisation.
-
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier
Master of Applied Science. 2000
Heng Jin
Department of Electrical and Computer Engineering
University of Toronto
Abstract
This thesis deals with the design and implernentation of a IV,
1.9GHz low noise
amplifier (LNA) using a 0.5pm CMOS on SOI technology with 3
levels of metal. The
arnplifier is optimized for CDMA applic2tions operating in the
1.93- 1.99GHz band. The
inductive degeneration topology used in the LNA implementation
provides low noise and
low power dissipation. The use of CMOS on SOI rechnology may
lead to an optimum
single chip implementation of both the analog and digital
building blocks of a 1.9GHz
transceiver operating from a IV supply. Such an implementation
offers reduced cost and
improved reliability.
The LNA consists of two amplifying stages with on-chip inductors
and capacitors. At
1.96GHz, the arnplifier has a 1.9d.B noise figure, a 14dB gain
and a 3dBm IIP3. It also
exhibits 17.4dB input and 28.3d.B output renirn losses
respectively in a 50R system without
extemd matching networks. The circuit draws 10.6mW from a IV
supply and the chip area
is 1.1~24 mm2.
-
Acknowledgments
I would like to express my sincere gratitude to Professor C.A.T.
Salama for his
insightful guidance and invaluable assistance throughout the
course of this work.
My sincere thanks to Jaro Pristupa for his technical assistance.
My appreciation
extends to al1 the staff and students in the Microelectronic
Research Laboratory including
Dana Reem, Anthoula Kampouris, Milena Khazak, Richard Barber,
Dod Chettiar, Eck
Kubowicz, Mehrdad Rarnezani, Dusan Suvakovic, Homoz Djahanshahi,
John Ren,
Farhang Vessai. Sotoudeh Hamedi-Hagh. Naoto Fushijima. Sameh G.
Nassif-Khalil and
Koji Yano for a11 their help.
1 would like to thank my friend. Song Ye, who kept me Company
with his constructive
discussions and cheerful chats, and the test of my friends:
Louis Zhang, S e m Hsien-en
Peng, Jianghong Hu, Yucai Zhang, Shuo Chen, Franklin Zhao,
Zhixian Jiao, Wei Yang,
Derek Hing Sang Tarn and Hongfei Lu who were always there for
me.
A special word of thanks to rny parents and my sister who have
been a constant source
of support and encouragement.
And to my wife Wei An, thank you for your patience, support and
love. To my daughter
Dian Jin, for being the best baby a father could have.
This work was supported by NSERC, Micronet, Gennum. Mitel,
Norte1 Networks and
PMC-S ierra.
-
Table of Contents
Page
7 1.1 The Role of the LNA in Wireless Receivers
............................................................ 1.2
Fundamental Characteristics of Low Noise Amplifiers
........................................... 3
............................................................................................................
S-parameters 3 Noise Figure
.............................................................................................................
5 Third-order Intermodulation Intercept Point Referred to the Input
(IIP3) .............. 5
1.3 Why CMOS on SOI
.................................................................................................
7
......................................................................................
1.4 CMOS LNA Architectures 7
Resistive termination
...............................................................................................
8
.............................................................................................
S hun t-series feedbac k 9 Inductive degeneration
............................................................................................
-9
1.5 Previous Work on CMOS LNA ..................... ......
.................................................... 9 1.6 Thesis
Objective and Outline
.................................................................................
IO
CHAPTER 2 Design of a Low Noise Amplifier for Wireless
Transceivers ..mm..a....a.... 13
Introduction
............................................................................................................
13 Classical Two-Port Noise Theory
..........................................................................
13 2.2.1 Noise figure and two-port noise parameters
................................................. 14 2.2.2 Optimum
source admittance
......................................................................
16
................................................................................................
Noise in MOSFETs 17 Derivation of the MOSFET Two-Port Noise
Parameters ...................................... 19
Power-Constrained Noise Optirnization
................................................................ 22
LNA Design
...........................................................................................................
25 LNA Layout
...........................................................................................................
27 2.7.1 Transistor layout
...........................................................................................
28 2.7.2 Capacitor layout
............................................................................................
28 2.7.3 Inductor layout
.......................................................................*......................
29 2.7.4 Final Layout .....................................
........................................................... 30
iii
-
2.8 Post-Layout Simulation Results
.............................................................................
3 1
......................................................................................................
S-parameters 32
.....................................................................................................
Intermodulation -33
.................................................................................................................
S tability 34
......................................................................................................................
Y ield -34
..............................................................................................
Temperature effects -34
2.9 Summary
................................................................................................................
35
............................................... CHAPTER 3
Experimental Results .oooo.ooooooo.ooooooooooomoooooo37
......................................................................................
3.1 The LNA Implementation 37
..........................................................................................................
3.2 LNA Testing 38
3.2.1 Gain and noise figure
....................................................................................
38 .................................................. 3.2.2
Reflection coefficient and reverse isolation 39
........................................................................
3.2.3 Intermodulation distortion 1
..........................................................................
3.2.4 Performance characteristics 42
................................................................................................................
3.3 Summary 42
CHAPTER 4 Conclusions
...............................................................D............................
c.45
-
List of Figures
Page
7 . Fig 1 . 1 . Portable PCS receiver architectures
......................................................................
Fig . 1 2: LNA connected as a two-port network for S-parameten
characterization .......... 4 Fig . 1.3. Amplifier input versus
output characteristics
....................................................... 6
....................................................................................
Fig . 1.4. CMOS LNA architectures 8 Fig . 2.1 : Cascaded noisy
stages
........................................................................................
13 Fig . 2.2. Equivalent noise mode1 for noisy two-port dnven by
noisy source ................... 14 Fig . 2.3: Noise circuit model
for the MOSFET . (a) Conventional mode1 .
(b) Mode1 with al1 noise sources refiected to the input
....................................... 18 Fig . 2.4. Dependence of
NF on Qs as a function of power dissipation
............................. 24 Fig . 2.5. Inductive 1 y
degenerated common-sourced amplifier
......................................... 25 Fig . 2.6. Low noise
amplifier schematic
...........................................................................
27 Fig . 2.7. Transistor layout
.................................................................................................
28 Fig . 2.8. Capacitor layout
..................................................................................................
29 Fig . 2.9. Typical inductor layout
.......................................................................................
30 Fig . 2.10. Final full circuit layout
.....................................................................................
31 Fig . 2.1 1 : Final circuit schematic for the LNA
..................... .. ................................ 31 Fig .
2.12. S-parameters venus frequency for the LNA
.................................................... 32 Fig . 2.13.
LNA output spectrum using a two-tone RF input
............................................ 33 Fig . 3.1 :
Micrograph of the LNA
......................................................................................
37 Fig . 3 . 2 Test setup for gain measurements
......................................................................
38 Fig . 3.3. Test setup for noise figure measurements
.......................................................... 38 Fig .
3.4. Gain and noise figure versus frequency for the LNA
......................................... 39 Fig . 3.5. Test setup
for S-parameter measurements
.......................................................... 39 Fig .
3.6. (a) Input and (b) output reflection coefficients for the LNA
.............................. 40 Fig . 3.7. Reverse isolation (S
12) for the LNA
...................................................................
40 Fig . 3.8. Test senip for two-tone third-order intermodulation
rneasurements .................. 41 Fig . 3.9. Measured LNA output
spectrum
........................................................................
41
-
List of Tables
Page
Table 1.1. Surnmary of recentiy reported CMOS LNA
characteristics ............................ 10 Table 2.1 :
Simulation results of LNA at different temperatures
....................................... 35 Table 3.1 : Summary of
LNA characteristics (25OC)
......................................................... 42 Table
3.2. Measured performance and cornparison with previous design
........................ 43
-
Chapter 1 : Introduction 1
HAPTER 1
Introduction
The analog front-end in a wireless transceiver acts as the
interface between the antenna
and the digital signal processor. In the digital signal
processor, low power is essential,
making submicron CMOS technology the best implementation choice.
The analog front-
end and specifically the low noise amplifier (LNA) require a
high speed technology, such as
GaAs or silicon bipolar. However, the use of low-cost submicron
CMOS or SOI CMOS
technology in the analog front-end may lead to an optimum single
chip implementation of
both the analog and digitai building blocks in wireless
transceivers used in modem high
capacity mobile communication systems. Such an implementation
ofters reduced cost and
improved reliability.
SOI CMOS provides a variety of advantages over bulk CMOS, such
as reduced
parasitic capacitance and elimination of backgating. Moreover,
SOI devices exhibit high
speed performance even at low supply voltage (-IV). Furthemore,
due to the low-loss
dielectric substrate, SOI inductors have higher self-resonant
frequencies and higher quality
factors than those fabricated on bulk silicon.
This thesis descnbes a IV, 0.5pm SOI CMOS LNA optimized for CDMA
applications
and operating in the 1.93-1 -99GH.z band. Compared to previously
reported designs [9, 10,
12, 131, the present design offers Iower noise, high gain, low
intermodulation distortion and
onthip 500 input output impedance rnatching.
A l-V, CMOS on SOI, 1 .!&GHz CDMA Low Noise Amplifier
University of Toronto
-
Chapter 1 : Introduction 2
1.1 The Role of the LNA in Wireless Receivers
In a wireless receiver, the information being relayed is either
voice or data which is
used to modulate a radio frequency (RF) carrier. Narrowband
circuitry in the radio
transceiver is used to access the portion of the frequency
spectrum that has been allocated
for the communications service. For example, the 1930-1990 MHz
band is used by the
North Amcricm CDMA standard for the PCS (Persona! Communication
Semices)
Receiving (Rx) band.
PCS offers the following advantages over conventional cellular
service: (1) PCS is an
dl-digital system from its inception, (2) PCS allows for
wireless voice. data, and video
transmissions, and (3) PCS has searnless roaming in North
America.
Antenna
RF mixer IF amplifier IF mixer Data out
(a) Heterodyne receiver
'XI +%+
rXI
Bandpass filter Image-reject filter
Antenna
Bandpass filter
Bandpass filter 1 Low-pass filter
rL 'XI
kz l (b) Homodyne receiver
Fig. 1.1 : Portable PCS receiver architectures
+ LNA 'XI
A 1 -V, CMOS on SOI, t .%GHz CDMA Low Noise Amplifier University
of Toronto
rL 'XI De-rnodulator
-
Chapter 1 : Introduction 3
The two receiver architectures that are considered viable for
portable PCS applications
are shown in Fig. 1.1. The conventional superheterodyne
(heterodyne or dual-conversion)
receiver illustrated in Fig. 1.1 (a), uses a high quaiity
bandpass filter centered at an
intermediate frequency (IF) to select the desired communications
channel. The radio
frequency signal received at the RF input to the mixer is
multiplied by the local oscillator
(LO) signal and it is thereby translated to an intermediate
frequency (IF) band. This
facilitates signal processing at a lower frequency where higher
quality filters and arnplifiers
can be economicaily constructed.
The direct-conversion or homodyne architecture, illustrated in
Fig. 1.1 (b), is an
alternative to the heterodyne scheme. The signal from the
antenna is directly convened to
the baseband where al1 of the filtering must be done to select
the desired channel.
It should be noted that in the RF section of the receivers, the
low-noise amplifier is
cornmon to both the heterodyne and homodyne schemes. In the
conversion process from the
RF to the IF bands, the received signal is degraded by the noise
from the mixers, and
consequently low-noise prearnplifiers are placed ahead of the
mixers in the receive chahs to
improve the signal to noise ratio.
1.2 Fundamental Characteristics of Low Noise Amplifiers
The main characteristics of low noise arnplifiers are discussed
in the following
paragraphs.
S-parameters
S-parameters cm be used to characterize the performance of a
LNA. Szl is the gain of
the LNA when its input and output impedances are rnatched to 50Q
. S and S22 are used to
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 1: Introduction 4
c harac tenze the input and output impedance matching
respectively. S 2 represents the
reverse isolation of the LNA.
- * * -
Er1 - * -En : 11 '2 : R1=50!2 -?y- LNA
4" 8 * A
I Y I
* 1
~eferen& Plane 1 Reference Plane 2
Fig. 1.2: LNA connected as a two-port network for S-parameters
characterization
A block diagram of the LNA connected as a two-port network for
S-parameters
characterization is shown in Fig. 1.2. The forward transmission
coefficient S2, (also cailed
power gain or simply gain), is related to the terminal voltages
V? and El by the fonvard
voltage gain (A,,), and is given by the following equation
l m ICI Power delivered to the Ioad
V? A, = - ( 1 2) El
and resistances RI and R2 are the source and load impedances,
which are typically 5042
S 7 , can be interpreted as twice the voltage gain (A,,)
developed between the load (V2) and
the opencircuited source voltage (El) , when R1=R2.
Similady, the reverse voltage gain is related to SIZ (IS 121 is
also called reverse isolation)
by the equation:
Power delivered to the source - - Power available from the
load
A 1-V, CMOS on SOI, 1 .%GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 1 : Introduction 5
The remaining S-parameters: SI and S2*, define the input and
output reflection
coefficients of the two port network, and are related to input
(2,) and output impedances
(ZouJ of the LNA respectively and the characteristic impedance
of the test system. Zo, by
the following equations
- - Power delivered to the input - Zin - Z~ ' i n - I - - -
Power available from the source Zin + Zo Zin + Ri Power delivered
to the output - Z ~ ~ t - Z ~ - Z ~ ~ t - Rz
SZ2 = - - Power available from the load Z,,, + Zo Z,,, + R2
Noise Figure
The noise figure (NF) is used to specify degradation in the
signal-to-noise ratio caused
by the LNA and it is always expressed in dB. The noise factor F
is the linear equivalent of
the noise tigure, and it is given by the following
expression
Input signal-to-noise ratio - F= - Total output noise Output
signal-to-noise ratio Total output noise due to the source (
1.7)
Noise and the way to reduce the NF in the LNA are discussed in
Chapter 2.
Third-order Intermodulation Intercept Point Referred to the
Input (m3)
While noise puts a limit on the lower Ievel of the signal which
c m be received, an upper
limit is imposed by the finite power supply voltage used for
active circuits and the non-
linear transfer characteristic inherent in the semiconductor
devices. As the input signal
increases in amplitude, these limitations cause distortion,
compression and saturation of the
signal being processed by the amplifier. Intermodulation
distortion caused by non-
linearities of the active components cm result in spurious
distortion being generated within
the receive channel bandwidth in a narrowband application. In
such cases, the interference
caused by the distortion products can be mistaken for the
desired signai.
A t -V, CMOS on SOI, 1 -9-GHz CDMA Low Noise Amplifier
University of Toronto
-
Chapter 1 : Introduction 6
Even-order intermodulation products are normally found at
frequencies well above or
below the signals which generated them and are usually of Iittle
concern. The odd-order
distortion produced by an LNA c m give rise to distortion
products which c m rnask or
interfere with the desired signal. The third-order products are
the most significant and c m
lie very close to the signals which generated them in frequency.
Third-order non-Iinearîiy is
customa-ily characterized by a specification called the
third-order intercept point.
Input Power
Fig. 1.3: Amplifier power output venus power input
characteristics
Fig. 1.3 illustrates the relationship between the fundamental
and intermodulation
distortion products generated by an amplifier for two equal
amplitude input signals at
different frequencies. The upper line shows the relationship
between the fundamental
output and the input signal (i.e., the gain), which is linear
(1:l dope) at low input power
A I -V, CMOS on SOI, 1 .%GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 1 : Introduction 7
levels, but rolls-off or compresses at high input power levels.
The third order
intermodulation lies well below the fundamental or linear curve
at lower input powers, but
increases with a 3: 1 slope for increasing input power. The
fundamental and intermodulation
distortion curves c m be extrapolated to a point of
intersection, which is called the third-
order intercept point. The intercept point cm be referred to
either the input or output power
axes; then respective points are known as the third-order
intercept point referred to the input
(IIP3) or to the output (OIP3).
1.3 Why CMOS on SOI
In this work. CMOS on SOI is used to implement the LNA. SOI CMOS
provide a
variety of advantages over bulk CMOS such as reduced parasitic
capacitance and
elimination of backgating. Moreover, SOI devices exhibit high
speed performance even at
low supply voltage (-IV). Furthemore, due to the low-loss
dielectric substrate, SOI
inducton have higher self-resonant frequencies and higher
quality factors than those
fabricated on bulk silicon. The inductors fabricated on silicon
suffer substrate losses, which
are caused by both inductive and capacitive coupling between the
inductor metal Iayer(s)
and the conductive substrate.
1.4 CMOS LNA Architectures pl
Much work on CMOS LNA's has been done [2-131. The architectures
reported cm be
classified into four distinct types, illustrated in simplified
form in Fig. 1.4. Each of these
architectures may be used in a single-ended forrn (as shown), or
in a differential form.
Differential architectures require the use of a balun to
transform the single-ended signal
from the antenna into a differentid signal. Practical baluns
introduce extra loss which adds
directly to the noise figure of the system, so single-ended
architectures are preferred.
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 1: Introduction 8
Zin Zin I
5 L-+ r- T (a) Resistive termination (b) l/g, temination
(c) Shunt-series feedback (d) Inductive degeneration
Fig. 1.4: CMOS LNA architectures
Resistive termination 131
The resistive termination configuration, shown in Fig. 1.4 (a),
is also called the
common source configuration. The use of real resistors provides
SOC2 input impedance.
However, it results in a noise figure of 3dB or higher,
independent of the minimum noise
performance of the transistor. Two effects are responsible for
this degradation in noise
figure. First, the added resistor contributes its own noise to
the output which equals the
contribution of the source resistmce. Second, the input is
attenuated by the resistor.
The common-gate topology, shown in Fig. 1.4 (b), provides good
isolation between the
input and output in cornparison to the common source topology.
In order to match the input
to SOR, the device transconductance g, have to be set to 1/50Q.
The minimum noise figure
in this configuration tends to be around 3dB theoretically and
greater in practice.
pp
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 1 : Introduction 9
Shunt-series feedback [5]
LNA design for CDMA transceivers typically requires a noise
figure around 2dB. So,
the previously described architectures are not candidates for
this application. Fig. i.J(c)
illustrates a topology, which uses resistive shunt and senes
feedback, to set the input and
output impedances of the LNA. It is reported [5] that amplifiers
using shunt-series feedback
often have high power dissipation compared to others with
similar noise performance. The
higher power is partially due to the fact that such amplifiers
are naturally broadband. and
hence techniques which reduce the power consumption through LC
tuning are not
applicable.
Inductive degeneration [2.4,7-9. 1 1- 131
The fourth topology employs inductive source degeneration, as
shown in Fig. 1.4 (d), to
generate a real term in the input impedance. Tuning of the
amplifier input becomes
necessary, making this a narrow-band approach. However this
requirement is not a
limitation for a CDMA transceiver. This architecture meets the
low noise and low power
performance specifications of the application and will be the
one used in this thesis.
1.5 Previous Work on CMOS LNA
Table 1.1 summarizes recently reported CMOS LNA results. Most of
the designs have
difficulty in simultaneously meeting low noise figure (-2dB),
low supply voltage (IV), low
power consumption, low input and output retum loss
specifications of CDMA receivers.
A l-V, CMOS on SOT. 1 .g-GHz CDMA Low Noise AmpIifier University
of Toronto
-
Chapter 1 : Introduction 10
Table 1.1 : Summary of recently reported CMOS LNA
characteristics
1.6 Thesis Objective and Outline
The main objective of this thesis is to develop a IV, 1.9GHz low
noise amplifier with
minimum power dissipation and minimum chip area using a silicon
on-insulator (SOI)
CMOS process. The LNA is specifically geared for CDMA receiver
application in the 1.93-
I .99GHz band,
The SOI CMOS process used for this project is provided by
Peregrine Serniconductor
Corporation. The process is a OSym, double poly CMOS process
with maximum unity gain
cutoff frequency fT of SOGHz. This process provides 6 transistor
types, difised and
polysilicon resistors, and triple layer metal interconnection
[14].
The target specifications of the LNA are listed in Table 1.2. An
inductive source
degeneration architecture was chosen to achieve the
specifications.
Sheng et al.. 1996 [5]
--
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
COMA 0 . 8 ~ CMOS
FioIougdran el d i . f G J o iôi 1 A ;pm C:dCS
0.6pm CMOS
0.35pmChIOS
0.25pm CMOS
0.5pm SOI
0.25pm SOI
0 . 8 ~ CMOS
O.6pm SOI
0.2pm SOI
Shaeffer et al.. 1997 [2]
Shahani e t d . 1997 r] Huang et al., 1998 f81
Johnson et aL. 1998 [91
Komurasaki et al.. 1998 [IO]
Floyd et ah. 1999 [Il]
Jin et al.. 1999 (121
Harada et al.. 2000 (131
Shunt-Set. FB
GPS
GPS
GPS
NIA
M A
N/A
PCS
N/A
Ggm-Tùrm.
L-Oegen.
L-Degen.
L-Degen.
L-Degen.
NIA
L-Degen.
L-Degen.
LaOegen.
3.3
3.Û
1.5
1.5
2.5
1.5
1.0
2.7
t .5
1.0
7.5
3.5
3.5
3.8
1.85
2.8
3.6
1.5
3.5
3.5
11 .O
22
22
17
16.2
10
7.0
12.4
23.5
IO
N A
A
-9.3
-6
-7.25
4
6,O
-1.9
NIA
O
36
27
30
12
t6
14
5.0
13.5
M A
19
0.9
%'A
NIA
NIA
-123
-12
NIA
-21
N A
NIA
N/A
û.3
1.5
1.5
0.9
2.4
1.9
0.9
t.8
2
N A
:&*A
-5-15
WA
-9.5
-7.5
N A
-5
NIA
NIA
-
Chapter 1 : Introduction I l
Table 1.2: Target specifications of the LNA
i Noise figure i e.5dB i
Process
Architecture
Power supply
Gain I 14dB
Target specification
0.5pm CMOS on SOI
Inductive degeneration
I V
1 Power consumption 1 1 Frequency 1 1 -93-1.99GHz 1 1 Input
output return loss 1 rl O ~ B 1 Terminal impedances 1
The thesis is organized as follows. Chapter 2 describes die
architecture, design and
layout of the LNA. Simulation results are presented. Chapter 3
presents the experirnental
results obtained.
Chapter 4 gives a bief summary of what has been achieved and
outlines areas for
future work.
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 1: Introduction 12
References J. Crols and M. S teyaert, "CMOS Wireless Tronsceiver
Design", Kluwer Academic Publishers, 1997.
D. Shaeffer and T. H. Lee, "A lSV, 1 .SGHz CMOS Low Noise
Amplifier," IEEE J. Solid-State Circuits, vol. 32, pp.745-759,
1997.
J. Y.-C. Chang, A. A. Abidi and M. Gaitan, "Large suspended
inductors on silicon and their use in a 2-pm CMOS RF amplifier,
"IEEE Electron Device Lett., vol. 14, pp. 246- 248, 1993.
A.N. Karanicolas, "A 2.7V 900MHz CMOS LNA and Mixer," IEEE J.
Solid-State Cir- cuits, vo1.3 1, No. 12, pp. 1939- 1944. 1996.
S. Sheng, L. Lynn, J. Peroulsa, K. Stone and 1. O'Donnell, "A
Low-Power CMOS Chipset for Spread Spectrum Communications", lEEE
ISSCC, San Francisco, Techni- cal Digest, pp.346-347, 1996.
A. Rofougaran, J. Y.-C. Chang, M. Rofougaran and A. A. Abidi, "A
lGHz CMOS RF front-end IC for a direct-conversion wireless
receiver," IEEE J. Solid-state Circtiits, ~01.3 1, pp. 880-889,
July 1996.
A.R. Shahhani, D.K. Shaeffer and T. H. Lee, " A 12mW Wide
Dynamic Range CMOS Front-End for a Portable GPS Receiver," IEEE J.
Soiid-State Circuits, vol. 32, pp. 306 1 -3070, 1 997.
Q. Huang, F. Piazza, P. Onatti and T. Ohguro, "The Impact of
Scaling Down to Deep Subrnicron on CMOS RF Circuits," iEEE J.
Solid-State Circuits. vol. 33, pp. 1023- 1036, 1998.
R.A. Johnson, P.R. Houssaye, C.E. Chang, P. Chen, M.E. Wood,
G.A. Garcia, 1. Lagnado and P.M. Asbeck, "Advanced Thin-Film
Silicon-on-Sapphire Technology: Microwave Circuit Applications,"
IEEE Transactions on Eiectron Devices. vol. 45, pp. 1047- 1054,
1998.
[IO] H. Kornurasaki, H. Sato, N. Sasaki, K. Ueda, S. Maeda, Y.
Yamaguchi and T. Miki, "A Sub 1-V SOI CMOS Low Noise Amplifier for
L-Band Applications,'' IEEE R F Inte- grated Circuits Symposium,
Digest of Technical Papers, pp. 153- 156, 1998.
[ l I] B. A. Floyd, J. Mehta, C. Gamero and Kenneth K. O, "A
900-MHz, 0.8-pm CMOS Low Noise Amplifier with 1.2-dB Noise Figure,"
IEEE CICC, Digest of Technical Pupers, pp. 66 1-664, 1999.
[12] W. Jin, P.C.H. Chan and C. Hai, "1.5-V 1.8-GHz SOI Low
Noise Amplifiers for PCS Receivers," IEEE Int. SOI Conference
Pmceedings, pp. 16-17, 1999.
[13]M. Harada, T. Tsukahara and J. Yamada, "0.5-1V 2GHz RF
Front-end Circuits in CMOS/SIMOX," IEEE ISSCC, San Francisco,
Technical Digest, pp.378-379,2000.
[14] Ultra Thin Silicon (UTSI) Luyuut Rules and Design Manual,
Peregrine Serniconductor Corporation, 1998.
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for Wireless
Transceivers 13
Design of a Low Noise Amplifier for Wireless Transceivers
2.1 Introduction
For a low noise amplifier with a cascade of stages, as shown in
Fig. 2.1, the overall
noise figure (NFmmr) cm be obtained in terms of the NF of each
stages given by Friis'
equation [ I I
NF, - i NF, - I ......... NFtotal = 1 + ( N F , - 1 ) + - +
+
A , A I A2......A (2.1)
( m - 1 )
where A!, A2,--, A, are the unloaded voltage gains of the
respective stages and the NF of
each stage is calculated with respect to the source impedance
driving the stage. This
equation indicates that the noise contributed by each stage
decreases as the gain preceding
the stage increases, implying that the first stage in a cascade
is the most critical.
Fig. 2.1 : Cascaded noisy stages
A 1 -V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise AmpIifier for WireIess
Transceivecs 14
In this chapter, a design strategy for a low noise amplifier
(LNA) that balances gain,
input impedance, noise figure, and power consumption is
developed by first introducing
power-constrained noise optimization of the input MOSFET,
because this transistor is the
most important source of noise in the LNA. The theory,
architecture, design and layout of
the LNA as well as simulation results are presented in the
following sections.
2.2 Classical Two-Port Noise Theory [2]
2.2.1 Noise figure and two-port noise parameters
Any noisy 2-port network driven by a source that has an
admittance Y, and an
equivalent shunt noise current 7s can be represented by a
noiseless 2-port network plus a
noise voltage source Zn and a noise current source Tn as
illustrated in Fig. 2.2.
r - - - - - - - - - - -
'i- Noisy Two-Port Network I %
Fig. 2.2: Equivalent noise modal for noisy two-port driven by
noisy source
#
I - 1
Noiseless 1
The noise figure is defined as
YS
NF = Total output noise power Output noise power due to input
source (2.2)
Two-Port Network 1
A calculation based directly on (2.2) requires computing of the
total power due to d l
1 i
_1
the noise sources and dividing that result by the noise power
due to the input source. An
equivalent and simplet method is to compute the total
shortcircuit mean-square noise
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for Wireless
Transceivers 15
current due to the input source. The noise figure of the noisy
two-port can then be expressed
In order to accommodate the possibility of correlations between
e, and in, in cm be
expressed as the sum of two components
where i , is correlatecl with en, and i,, is not and
where Y, is the correlation admittance.
Cornbining (2.3), (2.4). and (2.5). the noise
i: + li, + (Y, + ~, )e , l ' NF = l
figure becomes
Equation (2.6) contains three independent noise sources, each of
which may be treated
as thermal noise produced by an equivalent resistance or
conductance (whether or not such
a resistance or conductance actually exits) defined as
follows
Using these equivalences, the expression for the noise figure c
m be written purely in
terrns of impedances and admittances as
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for WireIess
Transceivers 16
where each of the admittances Y, and Y, have been decomposed
into a sum of a
conductance G and a susceptance B, defined as
The parameten Ge. Be. Rn, and G,, define the noise performance
of the two-port
network.
2.2.2 Optimum source admittance
Once a given two-port noise has been characterized by its four
noise parameters, the
general conditions for minimizing the noise figure c m be
identified from (2.10). Taking the
first derivative with respect to the source admittance (B, and
G, respectively) and setting it
equal to zero yields
Hence, to minimize the noise figure, the source susceptance
should be made equal to the
inverse of the correlation susceptance, while the source
conductance should be set equal to
the value defined in (2.14).
The noise figure corresponding to these choices is found by
direct substitution of (2.13)
and (2.14) into (2.10) yielding
The noise figure in ternis of NFmin and the source adminance can
be expressed as
A 1 -V, CMOS on S 01, 1 -9-GHz CDMA Low Noise Amplifier
University of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for Wireless
Transceivers 17
2.3 Noise in MOSFETs [2,3]
Since MOSFETs are essentially voltage-controlled resistors, they
exhibit noise.
Detailed theoretical considerations [4] lead to the following
expression for the drain current
noise ind in the MOSFET
where k is Boltzmann's constant, T is the absolute temperature
in Kelvin, Af is the noise
bandwidth in hem, and gdo is the drain-source conductance at V@.
y is typically 1-3 for
s hort-c hannei MOSFETs [SI.
In addition to drain current noise. the thermal agitation of
channel charge results in gate
noise. The Buctuating channel potential couples capacitively
into the gate terminal, leading
to a noisy gate current. The gate noise current ing can be
expressed as [6]
where 6 is the coefficient of gate noise (equal to 1.33 for
longchamel devices and 4-6 for
short-channel devices) and the parameter gg is given by [6]
where o is the angular frequency in radians per second and Cg,
is gate to source
capacitance.
The gate noise in a MOSFET is partially correlated with the
drain noise through a
correlation coefficient c defined as
with the value of c theoretically estimated as j0.395 [6].
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for Wireless
Transceivers 18
An additional source of noise in MOS devices is the noise
generated by the distributed
gate resistance 171. This noise source can be modeled by a
series resistance in the gate
circuit and an accompanying white noise generator. The
distributed gate resistance is given
by V I
where Ro is the sheet resistance of the polysilicon gate, W is
the total gate width of the
device, L is the gate length and n is the number of gate fingers
used to lay out the device.
The factor of 1112 arises from a distributed analysis of the
gate resistance, assuming that
each gate finger is contacted at both ends. By contacting at one
end only, this term increases
to 113. In addition, this expression neglects the interconnect
resistance used to connect the
multiple gate fingers together. This interconnect is normally
routed in a metal layer that
exhibits very low sheet resistance and hence is insignificant.
The noise circuit model for the
MOSFET is illustrated in Fig. 2.3 (a). Fig. 2.3 (b) shows the
same circuit with the noise
source Tnd reflected to the input and labelled inl and en
respectively.
O O drain gate
+ - ' ng g9 Cgs
source O O source
O gate l I I I
source I I I 1
drain
source
Fig. 2.3: Noise circuit model for the MOSFET (a) Conventional
model. (b) Mode1 with al1 noise sources reflected to the input
A 1-V, CMOS on SOI, 1 -9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for WireIess
Transceivers 19
The equivalent input noise voltage generator Zn and current
generator Tnl account for the
output noise or dnin current noise Tnd (2.17) are given by
and
2.4 Derivation of the MOSFET Two-Port Noise Parameters
The objective of this section is to correlate the noise circuit
model of the MOSFET
presented in Fig. 2.3 to the model of Fig. 2.2.
The total equivalent input noise voltage generator in is given
by (2.21). Using (2.7). R,l
is given by - 9
The total equivdent input current noise ln is the sum of the
reflected drain noise
contribution Zn/ and the induced gate current noise Ing. The
latter consists of two terms, one - ing, is hllly
uncorrelated
w here
correlated with the drain current noise, while the other Tngu is
completely
with the drain cunent noise. That is
Hence, 7; in (2.4) is given by
and Tu is
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for WireIess
Transceivers 20
The correlation admittance Y, in (2.5) may be expressed using
(2.28) and &, as follows
From equation (2.22). Y, can be expressed as
To express Y, in a more useful form, the gate noise correlation
coefficient c (2.20) must
be incorporated. The last term in (2.3 1 ) cm be expressed in
tems of cross-correlations by
multiplying both numerator and denorninator by
then avenging each. resulting in
the conjugate of the drain noise current and
Using (2.20), the correlation admittance Y,
where
may be wrinen as
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for Wireless
Transceivers 21
G, and Bc may be obtained by comparing (2.12) and (2.33)
resulting in
Using the definition of the correlation coefficient, the induced
gate noise rnay be
expressed as follows
The 1 s t term in (2.37) is the uncorrelated portion of the gate
noise curent. so that, Gu is
Al1 the four MOSFET two-port noise parameters Rn, G,, Bc, and
G,, have been derived
in (2.25), (2.35), (2.36) and (2.39). With these parameters,
rninirnizes the noise figure as well ûs the minimum noise
from (2.13), (2.14) and (2.15) resulting in
B,, = -B, = -o~,,(i + a ICI &) The reai part of the optimum
source admittance is
both the source impedance that
figure itself c m be detennined
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for WireIess
Tmsceivers 22
The width of the input MOSFET must be selected using (2.42),
which expresses GoPt
as a Function of gare capacitance. Setting GoPr equal to the
source conductance yields the
"optimum" value of Cg,, which, in tum, allows computation of the
necessary device width.
The required value of Cg, for a 50R system at 1.96GHz was
calculated to be around
3.5pF using y=2,6=4, lcl=0.395, and a4.85. A device large enough
to produce the required
value of Cs, wouM bc raughly Imm wide. Furthemorc, ;hc bias
currcnt for süch a largc
device would typically be over 100mA. Hence the power consumed
would be unacceptably
high for virtually any application, even though the noise figure
would correspond very
closely to NFmin. Therefore power-constrained noise optimization
must be introduced as
discussed in the following section.
2.5 Power-Constrained Noise Optimization (21
The goal here is io reformulate equation (7.16) in tems of power
consumption. Once
an equation is denved, it will be minirnized subject to the
conscraint of tixed power and then
solved for the width of the transistor corresponding to this
optimum condition.
Because the source susceptance B, is zero and Bopt (2.40) is
found to be very cloes to
zero [2], the difference between the two cm be neglected. The
expression for the noise
figure reduces to:
Rearranging (2.42) for GoPt to define a parameter QOp, with the
dimensions of a quality
factor results in
To accommodate the possibility of operation with source
conductances other than Gap, a
sirnilar Q in which Gop, is replaced by Gs (the actual source
conductance) can be defined as
A 1-V, CMOS on SOI, 1 -9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for Wireless
Tmnsceivers 23
Expressing (2.45) using (2.46) and (2.47), and the noise
parameters of (2.25) and (2.34)
results in the following noise figure expression
The pararneters a, g,, Qop, and Q, in (2.48) are linked to power
dissipation. To rewrite
these terms directly in terms of power, first recall that an
expression for the drain current is
191 7
P- b = w L C ~ ~ v ~ a t E s a t r + P (2.49)
where
and
where Cm is the gate oxide capacitance per unit area, v,,, is
the saturation velocity, and E,,
is the velocity saturation field strength. p, is the electron
rnobility.
The power dissipation can be written as follows
Furthemore, the transconductance g, c m be found by
differentiating (2.49). After
rearranging, this may be expressed as:
Cg, c m be expressed as:
Substituting for Cm from (2.52) results in
A 1-V, CMOS on SOI, 1 .%GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of ri Low Noise Amplifier for Wueless
Transceivers 24
and
where
Wth the aid of ihese expressions, the noise figure (2.48) cm be
written in ternis of p
and PD. Minimizing the resulting equation is complex enough but
it is best solved
graphically as shown in Fig. 2.4. From this figure. it can be
concluded that the value of QSp
that leads to the power-constrained minimum noise figure is
between 3.5 and 5.5.
Fig. 2.4 : Dependence of NF on Qs as a function of power
dissipation, for L=0.5pm, Rb5011, (u=2d=2m1.96GHz, Vdd=l V, ~2.5,
6=5.0, ld=0.395, v S a ~ l XI 05m/s, and ~ ~ ~ ~ 4 . 7 ~ 1
06v/rn
Once Qsp has been determined, the width of the optimum device c
m be expressed
using (2.54) and (2.56) resulting in
- - - - - - - -- - - - - - - - -
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for Wireless
Transceivers 25
With a device of width WVtp the noise figure obtained using
(2.48) is
This noise optimization method is used in the design because it
accounts for al1 the
parameters of interest. An exceilent input match can be
achieved, as discussed in next
section by the inductive source degeneration technique, while
providing nearly the best
noise figure possible in a given technology with a specified
power dissipation. For
W=2nfi2m 1.96GHz, L=0.5pm, C,,=2.79x 1 o - ~ F / ~ ' , RS=50R
and QSp=4.5, WqrP is found
from (2.58) to be 3961m.
2.6 INA Design
As described in Chapter 1, inductive degeneration topology as
shown in Fig. 2.5 (a)
meets the low noise and low power performance specifications of
the LNA and will be used
in this thesis. The LNA was designed in a 0.5pm CMOS on SOI
technology with 3 levels of
Fig. 2.5: lnductively degenerated common-sourced amplifier
Assuming the device mode1 with only a transconductance and a
gate-source
capacitance as shown in Fig. 2.5 (b) [IO], the input impedance
Zin has the following form
A 1 -V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for Wireless
Tmnsceivers 26
At resonance, Zin becomes Zin = a&s (3.6 1 )
At resonance, the impedance is purely real and proportional to
L,. By choosing L,
appropriately, the real term can be made equd to 5OR The gate
inductance Lg is used to set
the resonance frequency once L, is chosen to satisw the
criterion of a SOC2 input impedance.
It is important to emphasize that L, does not generate thermal
noise.
In an amplifier specification where the necessary gain cannot be
obtained in one stage.
the logical step is to try a multistage design, allowing the
desired gain to be spread over
multiple stages, reducing the gain for each stage to a value
that c m be achieved in one stage.
As (2.1) suggests, only the NF of the first stage plays a
criticai role. So, the design
procedure developed in the previous section can be used for that
stage. Fint, (2.58) is used
to determine the necessary device width. Then, the device is
biased using the arnount of
current allowed by the power constraint. Next. the value of the
source inductance L, is
chosen to provide the desired input match, using the value of
CO, that corresponds to the bias
conditions. The expected noise figure can then be computed from
(2.59). Findly, sufficient
inductance Lg is added in series with the gate to bring the
input loop into resonance at the
desired openting frequency.
In this design, a two-stage amplifier topology with inductive
degeneration at the source
was used as illustrated in Fig. 2.6. Inductive source
degeneration offers simultaneously
optimal noise and impendance matching. In addition. good
lineaity is achieved. On-chip
spiral inductors were used to provide input and output matching
and to act as loads in the
two stages of this design. Lbl and LM provide both matching and
gate bias simultaneously.
Capacitor Ci, Cc and Co are used for both matching and DC
decoupling. MI and M2 are
intrinsic N channel MOSFETs with OV threshold voltage. The
topology allows the amplifier
to operate at a IV supply voltage V' A gate width of 396pm for
Ml was chosen to
optimize the noise figure and minirnize DC power dissipation.
The gaie width M2 was
chosen the same as Ml, because the two stages can be biased by a
simple bandgap circuit
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for WieIess
Transceivers 27
without increasing the power consumption. Circuit simulations
were carried out using
Agilent's Advanced Design System (ADS) to determine the optimal
values of the remaining
components in the circuit. These values are given in Fig. 2.6.
The bias voltage Vbias is
generated off-chip for flexibility in testing.
Fig. 2.6: Low noise amplifier schematic
Careful layout of the LNA was necessary to meet the low noise
and high frequency
performance specifications. An appropriate layout not only
maximizes the yield but also is
the only way to achieve the required specifications predicted by
simulations. Care must be
taken to minimize resistive Iosses associated with metai lines
and to ensure that other losses
in the circuit do not diminish the overall performance of the
circuit.
A 1-V. CMOS on SOI, 1 -9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for WireIess
Transceivers 28
2.7.1 Transistor layout
The layout of the transistor must ensure that the distributed
gate resistance does not
adversely impact the noise figure. By interdigitating the device
with each gate finger
contacted at both ends, the contribution of this source of noise
can be reduced to
insignificant levels. The interconnect was routed in a metai
layer that possesses significantly
lower sheet resistance. The layout of transistor M! and MT is
illustrated inFig. 2.7. They
were split into two separate banks of devices, with poly gate
contacted at both ends.
Fig. 2.7: Transistor layout
2.7.2 Capacitor layout
The capacitoa are implemented as poly-poly structures with a
dielectric oxide in
between. The capacitor structure has parasitic resistance
associated with the resistance of
the polysilicon plates. This series resistance must be accounted
for. If the capacitor is just a
bypass capacitor, the senes resistance may not rnatter, but if
the capacitor is in the signal
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for Wireless
Transceivers 29
path, the series resistance is critical, and cm significantly
lirnit the performance of the
circuit.
In the circuit described in this work, three sets of signal
coupling capacitors were used.
These capacitors rnust be laid out to reduce their senes
resistance. This reduction can be
sirnply achieved by placing a large number of smdl capacitors in
parallel, as shown in Fig.
2.8. with the overal l capacitance heine the desired coupling
capacitor. This layout procedure
reduces the series resistance of the poly-poly capaciton by a
factor equal to the number of
paralle1 capacitors used, thus reducing the resulting resistance
[ I 11.
Via
One of small capacitors in parallel
Fig. 2.8: Capacitor layout
2.7.3 lnductor layout
Contact
' Metal and laye rs
The spiral inducton were assembled using ail three metal layers
to minimize series
resistance. The typical layout of an inductor can be seen in
Fig. 2.9. Sufficient space was
provided between inductors and their surroundings to keep
unwanted parasitic effects from
affecting the inductors' electricai characteristics.
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for Wireless
Transceivers 30
2.7.4 Final layout
The final layout of the LNA is illustrated in Fig. 2-10. The
signal Row is from left to
right. The wiring for the signal paths in the layout must be
wide enough to keep interconnect
resistance low so as to ensure the minimum signal loss. Also,
metai Iayers are used as much
as possible for the same reason. Crossing between metal lines is
kept to a minimum. The
physical layout in Fig. 2.10 corresponds to the electricai
circuit topology shown in Fig. 2.6.
The area of the chip is 1.1x2.4 mm2 including bonding pads. Less
than 5% of the chip area is
used for the active devices.
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for Wireless
Transceivers 31
--
Fig. 2.1 O: Final full circuit layout
2.8 Post-Layout Simulation Results
The simulations of the design were done using Agilent's Advanced
Design System
(ADS) md Spice models provided by Peregrine Semiconductor
Corporation. The mode1 of
the inductors is a series LR from ports 1 to 2 with a parallel
shunt capacitor C. These
inductor models c m be obtained using extracted equations
provided by Peregrine [12].
Fig. 2.1 1 : Final circuit schematic for the LNA
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for Wireless
Transceivers 32
The final circuit, dong with al1 bondwire inductors, is shown in
Fig. 2.11. Al1 the
inductor values as well as parasitic resistance and capacitance
are shown on the diagram.
S-parameters versus frequency are shown in Fig. 2.12. The power
gain (SZ1) is very Rat
over the full band ( 1.93- 1.99GHz). The input and output
reflection coefficient (S I I and S2*)
curves indicate good input and output matching and the reverse
isolation (SI,) is better than
4SdB ovet the fuII band,
Fig. 2.12: S-parameters versus frequency for the LNA
- - -
A 1-V, CMOS on SOI. 1 -9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for Wireless
Transceivers 33
Intermodulation
Fig. 2.13 illustrates the LNA output spectrum when a two-tone RF
input at
fl=1.960GHz and f2=1.965GHz is applied. The W output power level
is Pfl=-34.465dBm
for each tone. The third-order intercept point referred to the
output (OIP3) c m be calculated
using
where Pzfldfl is output power level at (2fl-f2)= i .955GHz. So,
the output referred third-order
intercept (OIP3) is 19.25dBm. IIP3 is given by
IIP3 = OIP3 - (Power gain of the LNA)
and in this case is 3.6dBm.
! r c q . GHz
Fig. 2.13: LNA output spectrum using a two-tone RF input
A 1 -V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for Wireless
Transceivers 34
The stability of the LNA was determined using Rollen's stability
factor k which is
defined in terms of the S-panmeters of the amplifier by the
equation [13]
w here
It wûs found that k is greater than one over the full band,
implying that the LNA is
unconditionally stable.
The yield simulation of the design were also done using the ADS
software within the
- 12db was 78% when al1 the passive full band. The yield for
S21min=14dB, SI Imax=S22maK-- component values varied by t 10%.
bondwire inductors varied by S0%. Process toleration
(including the variation of the threshold voltage) were taken
into account.
Temperature effects
The temperature of the environment will definitely affect the
performance of the LNA.
Table 2.1 gives the simulation results of the LNA at -50°C, 25°C
and 75°C respectively.
Temperature has little impact on the characteristics of the
LNA.
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for WireIess
Transceivers 35
Table 2.1 : Simulation results of LNA at different temperatures
- - - r Temperature 1 -500~ 1 25°C 1 7 5 0 ~
I ~ o w e r supply 1 IV 1 1 Frequency 1 1.96GHz 1 1.96GHz 1
1.96GHz 1 Noise figure 1 12.5dB* 1 9.5dB* 1 c2.5dB* y--.--- - - I 1
Gain 1 16.0dB / 15.8dB / 15.4dB
1 Power consumption 1 10.6mW 1 10.6rnW 1 10.9mW - - -
I n p u t return l o s ~ ISI1I 1 18.0dB 1 18.3dB 1 19.0dB 1
Output return loss IS221 1 18.6dB ( 18.6dB 1 l8.4dB
*First order analytical calculation
2.9 Summary
In this chapter, the power-constrained noise optimization of the
input MOSFET and
design principles of the LNA were descnbed. Sources of noise in
the LNA, the architecture,
design and layout of the amplifier as well as simulation results
were presented. These
results meet the specifications for application in a CDMA
transceiver operating in the 1.93-
1.99GHz band.
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 2: Design of a Low Noise Amplifier for Wireless
Transceivers 36
References H. T. Fnis, "Noise Figure of Radio Receivers", Proc.
IRE, vol. 32, pp. 419-422, 1944.
T. H. Lee, The Design of CMOS Radio-freqitency Integrated
Circuits, Cambridge Uni- versity Press, Cambridge, New York, NY,
USA. 1998.
D. K. Shaeffer, The Design and implementation of low-power CMOS
radio receivers, Kluwer Academic. Boston. USA, 1999.
A. van der Ziel, "Thermal Noise in Field EfFect Transistors,"
Proc. IEEE, pp. 1801- 1813, 1962.
A. A. Abidi, "High-frequency Noise Measurements on Fm's with
Srnail dimensions," IEEE Trans. Electron Devices, vol. ED-33, pp.
180 1 - 1805, 1986. A. van der Ziel. Noise in Solid State Devices
and Circuits, Wiley, New York, USA, 1986.
R. P. Jindal, "Noise Associated with Distributed Resistance of
MOSFET Gate Struc- tures in Integnted Circuits," IEEE Trans.
Electron Devices, vol. ED-3 1, pp. 1505- 1509, 1984.
B. Razavi, R.-H. Yan, and K. F. Lee, "Impact of Distributed Gate
Resistance on the Performance of MOS Devices," IEEE Tram Circuits
Syst. I , vol. 4 1, pp. 750-754, 1994.
N. G. Einspruch Ed., VLSI Electronics: Microstrrïcture Science,
vol. 18, chapter 1, pp. 1-37, Academic Press, New York, 1989.
[IO] D. Shaeffer and T. H. Lee, "A [SV, 1.5GHz CMOS Low Noise
Amplifier," IEEE J. Solid-State Circuits, vol. 32, pp.745-759,
1997.
[ I l ] R. S. Narayanaswami, The Design of A 1.9GHz Born W CMOS
Power Amplifier For DECT, M.S. Thesis, University of California at
Berkeley, 1996.
[12] Ultra Thin Silicon (UTSi) Lqoiit Rides and Design Manual,
Peregrine Semiconductor Corporation, 1998.
[ 1 31 B. Razavi, RF Microelectronics, Prentice-Hall, NJ,
1998.
- - - - - - - -
A 1-V. CMOS on SOI, 1 .%GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 3: Experimental Results 37
CHAPTER 3
Experimental Results
3.1 The LNA Implementation
The micrograph of the LNA implemented using the 0.5ym double
poly SOI CMOS
process provided by Peregrine Semiconductor Corporation is shown
in Fig. 3.1 and
corresponds to the electrical circuit topology shown in Fig.
2.6. The chip was packaged in a
24-led cerarnic Hat package (24-Pin CFP). This package is
suitable for frequencies up to
2GHz.
Fig. 3.2: Micrograph of the LNA
A 1-V, CMOS on SOI, f .%GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 3: Experimentai Results 38
The packaged LNAs were evaluated and tested using the Canadian
Microelectronics
Corporation's PCB-TF2 printed circuit board test fixture [ I l .
Typicai results obtained on a
sample of six chips are presented in the following sections.
3.2.1 Gain and noise figure
By ensuring appropriate DC bias and supply voltages, the gain
and nois figure of the
LNA were tested. The gain test senip is shown in Fig. 3.2. The
noise test setup is shown in
Fig. 3.3. The measured power gain and noise figure versus
frequency (from 1.92 to 2.0GHz)
are shown in Fig. 3.4.
Fig. 3.2: Test setup for gain measurements
Noise Figure Noise Source Meter
Drive
HP346A
Noise )
source
Fig. 3.3: Test setup for noise figure measurements
A 1 -V, CMOS on SOI, 1 %GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 3: Experimental Results 39
- _ Simulation Gain - 0 -
' - 0 - - - 0 - - . ~ . . - O - O - . -
' O . - Measured Gain
frequency (GHz)
Fig. 3.4: Gain and noise figure versus frequency for the LNA
3.2.2 Reflection coefficient and reverse isolation
The input and output reflection coefficients (SI and S77) -- and
the reverse isolation (S 12) were tested using a HP8720B network
anaiyzer. The setup is shown in Fig. 3.5. Fig. 3.6
shows the measured input and output reflection coefficients (SI
1 and S 7 7 ) _- indicating good input and output matching. Fig.
3.7 shows the reverse isolation (S 12) which is better than
38.44dB over the full band.
t =P net aramets en+ Port 2
O Analyzer 1 Port 1 ' I G J L U G L
HP87206
Fig. 3.5: Test setup for S-parameter measurements
A I-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 3: Experimental Results 40
-1 OdB -20dB -3068 -40dB
OdB -1OdB -20dB -30dB -40dB
Fig. 3.6: (a) Input and (b) output reflection coefficients for
the LNA
Fig. 3.7: Reverse isolation (SI2) for the LNA
A f -V. CMOS on SOI, 1 -9-GHz CDMA Low Noise Amplifier uni ver
si^ of Toronto
-
Chapter 3: Experirnental Results 41
3.2.3 Intermodulation distortion
The test setup for two-tone third-order intermodulation
distortion measurements is
shown in Fig. 3.8. Generator frequencies (two-tones) were set at
1.950MHz and 1.955MHz
respectively. The tone levels were set so as not to saturate the
LNA. The final output
spectrum of the LNA obtained from the spectrum andyzer is
illustrated in Fig. 3.9. The
input refened third-order intercep (I[P?) was calculated to he
3dBm using the method
described in Chapter 2.
RF Generator Colby SG8000A
f1 T-connecter Power > -0 HP8563E , Combiner + LNA Spectrum
Analyzer
RF Generator
Rohde & Schwarz Signal Generator ÇMT03
Fig. 3.8: Test setup for two-tone third-order intermodulation
measurements
Fig. 3.9: Measured LNA output spectrum, two-tones power level -1
OdBm, third- order intermodulation power level -60.83dBm
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 3: Experimental Results 42
3.2.4 Performance characteristics
Table 3.1 summarizes the characteristics of the LNA. Generaily,
there is good
agreement between experimental and simulation
characteristics.
Table 3.1 : Summary of LNA characteristics (25°C)
Power supply
1 Frequency 1 1.96GHz
Simulation
2.64mm2
I
1 Noise figure 1 1.9dB
Die area
1 Gain 1 14dB
Experimental
2.64mm2
1 Power consumption 1 1 0.6mW 1 f0.6mW 1 Input return Ioss IS1 1
1 17.4dB 1 18.3dB 1 1 Output return loss IS221 1 28.3dB 1 18.6dB 1
( Reverse isolation ISI21 1 44.7dB 1 45.8dB 1 1 Terminal impedances
1 50Q I 5022 I
* First order analytical calculation
This chapter discussed the implementation and the experimental
results of the LNA
implemented using 0Spm double poly SOI CMOS process provided by
Peregrine
Semiconductor Corporation. The test results of the LNA were
presented and are generally
in good agreement with simulations and expected
specifications.
As outlined in Table 3.2, compared to previously reported SOI
CMOS LNA designs,
the present design offers lower noise, high gain, low
intermodulation distortion and on-chip
- - - - - - - - - -
A 1-V, CMOS on SOI, 1 -9-GHz CDMA Law Noise Amplifier University
of Toronto
-
Chapter 3: Experimental Results 43
50R input output impedance matching. The LNA is viable for CDMA
receiver application
in the I .93- 1.99GHz band.
Table 3.2: Measured performance and comparison with previous
design
1 Main author, 1 This year [Ref.] , design
I
Johnson 1 Komumsaki. 1998 [2]
I 1998
Jin, 1 Harada, 1999 141 2000 [5!
I
1 Application 1 CDMA NIA 1 NIA PCS 1 N/A 0.5pm
Technology SOI 0.5pm 0.25pm SOI
0.61m 0.2prn SOI
1 Architecture ( L-Degen
1 Gain (dB) 1 14
-7.5 1 NIA 1 NIA -12 1 NIA NIA
A 1-V, CMOS on SOI, 1 -9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 3: Experimentd Results 44
References Przhted Circuit Board Test F i ~ r e for 24-Pin CFP
(PCF-TF2): User Guide, Canadian Microelectronics Corporation,
Version 2.0, 1999.
R.A. Johnson, PR. Houssaye, C.E. Chang, P. Chen, M.E. Wood, G.A.
Garcia, 1. Lagnado and P.M. Asbeck, "Advanced Thin-Film
Silicon-on-Sapphire Technology: Microwave Circuit Applications ,"
IEEE Transactions on Electron Devices. vol. 45, pp. 1047- 1054,
1998.
H. Komurasaki, H. Sato, N. Sasaki, K. Ueda, S. Maeda, Y.
Yamaguchi and T. Miki, "A Sub 1-V SOI CMOS Low Noise Amplifier for
L-Band Applications," IEEE RF Inte- grated Circii ifs Symposium,
Digest of Technical Papers, pp. 153- 1 56, 1998.
W. Jin, P.C.H. Chan and C. Hai, "1.5-V 1.8-GHz SOI Low Noise
Amplifien for PCS Receivers," IEEE Int. SOI Conference,
Proceedings, pp. 16- 17, 1999.
M. Harada, T. Tsukahara and J. Yamada, "0.5- 1 V 2 G b RF
Front-end Circuits in CMOSISIMOX," ISSCC, Technical Digest,
pp.378-379.2000.
- - - - - - - - - - - -
A 1-V, CMOS on SOI, 1.9-GHz CDMA Low Noise Amplifier University
of Toronto
-
Chapter 4: Conclusions 45
Conclusions
This *ork dedt with the design of 3 !ow noise amplifier (LN.4)
for the mdog front-end
of a wireless CDMA transceiver using CMOS on SOI technology. The
use of this
technology may lead to an optimum single chip implementation of
both the analog and
digital building blocks of a 1.9GHz transceiver operating from a
IV supply. Such an
implementation offers reduced cost and improved reliability.
A Iow noise amplifier optimized for CDMA applications and
operating in the 1.93-
1.99GHz band was designed and implemented using the 0.5pm double
poly SOI CMOS
process provided by Peregrine Semiconductor Corporation. The
inductive degeneration
topolopy used in the LNA implementation provides low noise and
low power dissipation.
The LNA consists of two stages with onchip inductors and
capacitors.
The experimental results at 1.96GHz show that the LNA has a
1.9dB noise figure, a
i4dB gain and a 3dBm iIP3. The LNA dso exhibits 17.4dB input and
28.3dB output return
losses respectively in a 50R system without extemal matching
networks. The circuit draws
1 O.6mW from a 1 V supply and the chip area is 1.1~2.4 mm2.
Future work should focus on the use of more advanced
technologies such as 0.25pm or
0.18pm CMOS on SOI processes which would provide MOSFETs with a
higher unity gain
cutoff frequency. A LNA realized using such technologies is
expected to operate at
frequencies in the 2-5GHz range with low noise and low power
dissipation. Negative
feedback could be added to the circuit, if the LNA is used in
wideband applications. The
performance of the amplifier could be further improved if
inducton with a higher quaiity
factor are available.
- - ---- -
A 1-V, CMOS on SOI, 1 -9-GHz CDMA Low Noise Amplifier University
of Toronto