Analog Circuits and Signal Processing
Series Editors:Mohammed Ismail. The Ohio State University
Mohamad Sawan. Ecole Polytechnique de Montreal
For further volumes:http://www.springer.com/series/7381
Pui-In Mak � Rui Paulo Martins
High-/Mixed-VoltageAnalog and RF CircuitTechniques for NanoscaleCMOS
Pui-In MakUniversity of MacauTaipa, Macao, China
Rui Paulo MartinsUniversity of MacauTaipa, Macao, China
ISBN 978-1-4419-9538-4 ISBN 978-1-4419-9539-1 (eBook)DOI 10.1007/978-1-4419-9539-1Springer New York Heidelberg Dordrecht London
Library of Congress Control Number: 2012933993
# Springer Science+Business Media New York 2012This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part ofthe material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission orinformation storage and retrieval, electronic adaptation, computer software, or by similar or dissimilarmethodology now known or hereafter developed. Exempted from this legal reservation are brief excerptsin connection with reviews or scholarly analysis or material supplied specifically for the purpose of beingentered and executed on a computer system, for exclusive use by the purchaser of the work. Duplicationof this publication or parts thereof is permitted only under the provisions of the Copyright Law of thePublisher’s location, in its current version, and permission for use must always be obtained fromSpringer. Permissions for use may be obtained through RightsLink at the Copyright Clearance Center.Violations are liable to prosecution under the respective Copyright Law.The use of general descriptive names, registered names, trademarks, service marks, etc. in thispublication does not imply, even in the absence of a specific statement, that such names are exemptfrom the relevant protective laws and regulations and therefore free for general use.While the advice and information in this book are believed to be true and accurate at the date ofpublication, neither the authors nor the editors nor the publisher can accept any legal responsibility forany errors or omissions that may be made. The publisher makes no warranty, express or implied, withrespect to the material contained herein.
Printed on acid-free paper
Springer is part of Springer Science+Business Media (www.springer.com)
Weng-Ieng & Long-ChengDita & Filipa, Diogo, Guilherme e Maria
Preface
The scope of more intelligent and higher data rate wireless connectivity in the
coming decades continuously push the performance envelope of wireless circuits
and systems. Higher integration level, more functionality, less cost and lower power
are the obvious goals. For the radio front-ends, inductorless broadband solutions
possess the highest potential to economically realize low-power multi-standard
solutions in nm-length CMOS technologies. The rapid downscaling of transistor
sizes and gate-oxide thickness, however, involves rapid reduction of supply voltage
for reliability. This fact, in addition to the changed device features such as lower
intrinsic gain and linearity, urges for more feasible techniques from different
dimensions, such that the performances can be aligned continuously with the
expectation from the global wireless chip industry.
In this book, high-/mixed-voltage analog and RF circuits are investigated as the
prospective solution for the next generation of wireless products in nm-length
CMOS technologies. The content starts by overviewing the design considerations,
pros and cons of high-/mixed-voltage circuits before describing three tailor-made
circuit designs targeting the mobile-TV applications. Mobile TV is recognized as
one of the key functions of handheld devices such as smart phones.
The first design is a 90-nmCMOS ultra-wideband low-noise amplifier withmixed-
voltageESDprotection for handling the full-bandofmobileTV.The second is a 90-nm
CMOShigh-voltage-enabledmobile-TVRF front-endwithTV-GSMinteroperability.
The third is a 65-nmCMOSmixed-voltage unified full-bandmobile-TV receiver front-
end averting any external balun,whilemeasuring favorably performanceswith respect
to the state-of-the-art.
Most techniques are generally extendable to different types of wireless systems
in ultra-scaled CMOS technologies.
Taipa, Macao, China Pui-In Mak
Rui Paulo Martins
vii
Acknowledgments
We are grateful to Macao Science and Technology Development Fund (FDCT)
and Research Committee of University of Macau (UM), for funding the project and
equipments of the State Key Laboratory of Analog and Mixed-Signal VLSI.We thank Mr. Ka-Fai Un andMiao Liu for their contribution to the design of the
multi-phase local oscillator generator and high-voltage amplifier, respectively.
We thank the staff at Springer, particularly Chuck Glaser, for guiding the
preparation of this book.
Finally, we send our heartfelt appreciation to our families, who endured our
dedication to this book.
ix
Contents
1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 The Global Development of Wireless Technologies . . . . . . . . . . . . . . . . . 1
1.2 Analog and RF Circuits in Nanoscale CMOS. . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Research Objectives. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Organization of the Book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2 General Considerations of High-/Mixed-VDD Analog
and RF Circuits and Systems. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 System Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Device Reliability in Ultra-scaled Processes . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3.1 AMR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3.2 HCI Lifetime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3.3 TDDB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3.4 NBTI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3.5 Punchthrough. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.4 Extend the Voltage Capability of Thin-
and Thick-Oxide Transistors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.5 High-/Mixed-Voltage Building Blocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.5.1 Power Amplifier and Wideband Balun-LNA
(High-VDD þ Mixed-Transistor) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.5.2 Passive Mixers (High-VDD þ Thin-Oxide Transistor) . . . . . . . 15
2.5.3 Differential Pair (High-VDD þ Thin-Oxide Transistor) . . . . . . 17
2.5.4 Recycling Folded Cascode Operational Amplifier
(High-VDD þ Thin-Oxide Transistor) . . . . . . . . . . . . . . . . . . . . . . . . 18
2.5.5 OpAmp-Based Analog-Baseband Circuits
(Mixed-VDD þ Mixed-Transistor) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.5.6 Low-Dropout-Regulator
(Mixed-VDD þ Mixed-Transistor) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
xi
2.5.7 Sample-and-Hold Amplifier
(High-VDD þ Mixed-Transistor) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.5.8 Line Driver (High-VDD þ Thin-Oxide Transistor) . . . . . . . . . . . 32
2.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3 A Full-Band Mobile-TV LNA with Mixed-Voltage
ESD Protection in 90-nm CMOS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.2 Circuit Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2.1 Full-Band Mobile-TV Tuner Architecture . . . . . . . . . . . . . . . . . . . 36
3.2.2 PMOS-Based Open-Source Input Structure
and Mixed-Voltage ESD Protection . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.2.3 Double Current Reuse for gm-Enhancement. . . . . . . . . . . . . . . . . . 39
3.2.4 Single-Stage Thermal-Noise Cancellation. . . . . . . . . . . . . . . . . . . . 41
3.3 Key Practical Design Issues. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.3.1 Package Effect on Input Impedance Match . . . . . . . . . . . . . . . . . . 43
3.3.2 Self-Startup Constant-gm Bias Circuit . . . . . . . . . . . . . . . . . . . . . . . . 44
3.3.3 Power-Up/Down Control with Reliability Concern . . . . . . . . . . 45
3.3.4 Mixed-Voltage ESD Protection Scheme . . . . . . . . . . . . . . . . . . . . . 45
3.4 Simulation Results, Discussions and Benchmarks. . . . . . . . . . . . . . . . . . . . 46
3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4 A High-Voltage-Enabled Mobile-TV RF Front-End
in 90-nm CMOS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2 Tuner Architecture for TV-GSM Interoperation. . . . . . . . . . . . . . . . . . . . . . 55
4.3 On/Off-Chip Circuit Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.3.1 Basic-Cell of the LNA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.3.2 ESD Protection Scheme. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.3.3 Programmable C-2C Attenuator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.3.4 I/Q Mixer Drivers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.3.5 Customized External Preselect Filters . . . . . . . . . . . . . . . . . . . . . . . . 70
4.3.6 Feedforward Gain Roll-Off Compensation . . . . . . . . . . . . . . . . . . . 72
4.4 Experimental Results, Discussions and Benchmarks. . . . . . . . . . . . . . . . . 73
4.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
5 A Mixed-Voltage Unified Receiver Front-End for Full-Band
Mobile TV in 65-nm CMOS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
5.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
5.2 RFE Architecture and Technology Features. . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.3 Wideband Balun-LNA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
xii Contents
5.3.1 CG-CS Noise-Canceling Balun-LNA . . . . . . . . . . . . . . . . . . . . . . 84
5.3.2 Proposed Gain-Boosting Current-Balancing
Balun-LNA with VGC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.4 Current-Reuse Mixer-LPF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.4.1 Circuit Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
5.4.2 Noise Figure. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
5.4.3 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
5.5 Multi-phase LOG. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.5.1 Brief Overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
5.5.2 Open Loop Multi-phase LO Generators
(Conventional and Proposed) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
5.5.3 Circuit Implementation and Simulation Results. . . . . . . . . . . . 105
5.6 Measurement Results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
5.6.1 The RFE. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
5.6.2 The LOG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5.6.3 Performance Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
5.6.4 Architecture Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
6.1 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
6.2 Recommendations for Future Research. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
Appendix: Open-Loop Multiphase LO Generators . . . . . . . . . . . . . . . . . . . . . . . 125
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Contents xiii
Abbreviations
1/f Flicker noise
AC Alternating current
ADC (A/D) Analog-to-digital converter
AMR Absolute maximum rating
ATT Attenuator
BB Baseband
BiCMOS Bipolar complementary metal oxide semiconductor
BPF Bandpass filter
BSF Band-selection filter
BTI Bias temperature instability
BUF Buffer
BW Bandwidth
CG Conversion gain
CG Common-gate
CLKGEN Clock generator
CM Common mode
CMFB Common mode feedback
CMOS Complementary metal oxide semiconductor
CMRR Common-mode rejection ratio
CQFP Ceramic quad flat-pack
CS Common-source
CSF Channel-selection filter
CT Continuous-time
DC Direct current
DMB-T Digital multimedia broadcasting-terrestrial
DRC Design rule check
DSB Double-side band
DSP Digital signal processor
DVB-T/H Digital video broadcasting-terrestrial/handheld
xv
ESD Electrostatic discharge
FET Field effect transistor
FF Fast-fast (corner)
FS Fast-slow (corner)
FS Frequency synthesizer
GBW Gain-bandwidth product
GSM Global system for mobile communications
HBM Human body model
HCI Hot carrier injection
HD2/HD3 Second-/third-order harmonic distortion
HPF Highpass filter
HR/HRR Harmonic rejection (ratio)
I In phase
I/P Input
IC Integrated circuit
IEEE Institute of electrical and electronics engineering
IIR Infinite-impulse response
IM3 Third-order intermodulation distortion
IP3 Third-order intercept point
IRR Image-rejection ratio
ISDB-T Integrated services digital broadcasting-terrestrial
ISM Industrial, scientific and medical
ISSCC IEEE international solid-state circuits conference
I-to-V Current-to-voltage
JSSC IEEE journal of solid-state circuits
LDO Low-dropout regulator
LIF Low IF
LNA Low-noise amplifier
LO Local oscillator
LOG Local oscillator generator
LPF Lowpass filter
LTCC Low-temperature co-fired ceramic
LV Low voltage
LVS Layout versus schematic
MOSFET Metal-oxide semiconductor field-effect transistor
NF Noise figure
O/P Output
OFDM Orthogonal frequency-division multiplexing
OpAmp Operational amplifier
OTA Operational transconductance amplifier
xvi Abbreviations
PCB Printed-circuit board
PLL Phase-locked loop
PM Phase margin
PVT Process, supply voltage and temperature
Q Quadrature phase/quality factor
QAM Quadrature amplitude modulation
RC Resistor-capacitor
RF Radio frequency
RFE Receiver front-end
RX Receiver
S2D Single-to-differential
SAW Surface acoustic wave
SF Slow-fast (corner)
SFDR Spurious-free dynamic range
SoC System-on-chip
SS Slow-slow (corner)
TDDB Time dependent dielectric breakdown
T-DMB Terrestrial-digital multimedia broadcasting
TX Transmitter
TXR Transceiver
UHF Ultra-high frequency
UWB Ultra wideband
VCO Voltage control oscillator
VGC Variable gain control
VHF Very-high frequency
V-to-I Voltage-to-current
WLAN Wireless local area network
ZIF Zero IF
Abbreviations xvii
Chapter 1
Introduction
1.1 The Global Development of Wireless Technologies
The convergence of wireless and semiconductor industries will turn into reality the
vision of fully autonomous and seamless wireless connectivity in the near future,
via combining advanced nanoscale CMOS technologies with innovative hybrid-
domain circuits and systems solutions [1, 2]. One aim inside this immense scope is
to develop a smart-mobile-companion device with high performance, adaptiveconnectivity and high power efficiency. High performance is the essential ingredient
to cope with the ever increasing add-on of functionalities in a small handheld
device. Adaptive connectivity is to automatically select the best wireless link,
maximizing the quality of service wherever and whenever possible. Power
efficiency is to extend the active-use days without entailing a big battery, or
worrying about the battery life after executing much of the device’s functionality.
To reach these three goals the system chips consisting of many analog and radio
frequency (RF) circuits play a key role. The aim of adaptive connectivity promotes
a full integration of many different radio technologies such as WiFi, Global
Positioning System (GPS), BluetoothTM, cellular and mmWave into one unit,
calling for an unprecedented high level of complexity among system planning,
architectures and circuits. The goal of maximizing hardware sharing without
compromising the performances challenges the designers throughout the front-to-
back-end development. Yet, advanced nanoscale CMOS technology constitutes,
still today, the most promising platform of wireless products for its tremendous
advance in speed, system-on-chip capability and maturity (to some extent).
In terms of power efficiency, 10-fold reduction of system power is the targeted
goal. This aggressive goal calls unavoidably for groundbreaking innovation in
silicon transceivers, sensors and mixed-signal interfaces. Investigation on new
architecture and circuit solutions benefiting the speed and area advantages of
nanoscale processes may eventually break the power wall. It is believed that,
the many new opportunities opened up by the wireless industry are bringing
P.-I. Mak and R.P. Martins, High-/Mixed-Voltage Analog and RF Circuit Techniquesfor Nanoscale CMOS, Analog Circuits and Signal Processing,
DOI 10.1007/978-1-4419-9539-1_1, # Springer Science+Business Media New York 2012
1
along another wave of synergy among the semiconductor industries including
the manufacturers, fabless IC design houses and CAD providers. As evidenced in
the last 20 years, the many challenges ahead continuously motivate large-scaleresearch and development in many aspects of integrated circuit design in the
coming decades.
1.2 Analog and RF Circuits in Nanoscale CMOS
Rapid downscaling of CMOS has led to faster analog and radio frequency (RF)
circuits, but with deteriorated linearity and accuracy due to lower device’s intrinsic
gain. Entered into the sub-1 V nanoscale regime, the downsizing of threshold
voltage (VT) is decelerated due to transistor variability, matching and leakage issues.
Along with the low supply voltage (VDD) constrained by the device reliability limits,
the trade-off between signal dynamic range and number of cascode transistors that
can be utilized become increasingly severe [3, 4]. While cascode structures are no
longer voltage efficient, cascade structuresmay not bewelcome also because of their
higher power consumption and lower operating speed. Thus, circuit techniques
that will continuously fit for sub-1 V processes must be investigated, in order to
keep driving up circuit performances in technology scaling.
1.3 Research Objectives
In this book, high-/mixed-voltage analog and RF circuits and systems are exten-
sively investigated. They have gained extensive attention in state-of-the-art
nanoscale-CMOS wireless systems such as mobile-TV, GPS and software-defined
radios [5–9]. As depicted in Fig. 1.1, for a wideband SAW-less receiver for multi-
standard compliance, maximizing the receiver dynamic range (DR) is essential to
detect weak signal in presence of high-power blockers.
0 dBm unwanted blocker (632 mVpp in 50W)
LO
RF BBLNA
High DR
-80 dBm desired signal(63.2 µVpp in 50W)
blockers
desiredFig. 1.1 Wideband multi-
band solution requires high
DR from all RF and baseband
circuits
2 1 Introduction
It is known that, for most analog and RF circuits, maintaining a constant
performance (i.e., signal to noise and distortion ratio) in a newer technology will
entail more power to compensate the loss of voltage headroom [10]. The basis can
be illustrated through the analysis of the relationship between VDD and the dynamic
range of a common-source (CS) amplifier with a resistive load, as shown in
Fig. 1.2a.
The channel-length modulation is neglected for simplicity. The input-referred
squared noise voltages imputable to M1 and RL are given by,
V2n;M1 ¼ 4kT
1
gmg; (1.1)
and
V2n;RL ¼ I2n;RL
1
g2m¼ 4kT
RL
1
g2m; (1.2)
respectively, where g is the noise factor and gm is the transconductance of M1, k isthe Boltzmann’s constant, T is the temperature (in Kelvin). The generic I-Vequation of a MOSFET considering the mobility degeneration parameter y is
given by,
ID ¼ 1
2
W
LmoCoxðVGS � VTÞ2 1
1þ yðVGS � VTÞ ; (1.3)
where W/L is the aspect ratio of the transistor and moCox is the transconductance
parameter. From (1.3), the input-referred third-order intercept point (IIP3) of a
MOSFET can be calculated [11],
V2IIP3 �
8
3
VGS � VT
y¼ 16
3
IDgmy
: (1.4)
RL
RL RL
VDD
VDD
VSS
Vout
Vin
ID
ID /2 ID /2
Vn,M1
In,RL
VCM.o+Vp
VCM,o-Vp
VCM,o
M1
M2
M1 M4
M3 M5 M6VLOp
VRFnVRFp
VLOp
Voutp Voutn
VLOn
2
2
a b
Fig. 1.2 (a) Typical CS amplifier and (b) typical active mixer with resistive load
1.3 Research Objectives 3
With (1.1), (1.2), and (1.4), the dynamic range DRCS of the CS amplifier can be
approximated,
DRCS � V2IIP3
V2n;M1 þ V2
n;RL
� 4
3
IDkTyg
; (1.5)
In practice, ID cannot be arbitrarily increased to boost DRCS when there is no
voltage headroom, i.e., the output common-mode voltage VCM,o must be withinVDD,
VCM;o ¼ xVDD ¼ VDD � IDRL; (1.6)
where x denotes a ratio value <1 and can be set as 0.5 for VCM,o ¼ VDD/2, which
roughly yields the highest 1-dB compression point (P1dB). Substituting (1.6) into
(1.5) lead,
DRCS � 2
3
VDD
RL
1
kTyg: (1.7)
When a high gain is desired, RL and VDD can be increased together by a factor
A such that the DR is maintained, while ID remains unchanged. The power is
increased by A times too. A similar situation holds for active mixer [12] as shown
in Fig. 1.2b. Its DR can be calculated following the same approach, i.e.,
DRMIXER � 2
3
VDD
RL
1
kTyg4
p2: (1.8)
Obviously, VDD cannot be arbitrarily increased to maximize the circuit’s DR due
to the reliability limitations. The major device reliability concerns of ultra-scaled
CMOS processes include the absolute maximum rating (AMR), hot carrier injection
(HCI), negative bias temperature instability (NBTI), time dependent dielectric
breakdown (TDDB) and punchthrough effect [13]. Boosting the device reliability
by transistor stacking can push upward the drain-source voltage limit. Differently,
the gate-drain/-source voltages should be controlled via proper bias in both active
and sleep modes. The device stacking can be simply based on thin-oxide devices,
or a hybrid of thin- and thick-oxide devices to balance the speed and voltage
withstand capability. Their implications to the circuit or system should be analyzed
according to the actual operation. For instance, when the maximum input and
output swings are defined, node-voltage trajectory checks can help to ensure there
is no device under overstressed at all time.
Inductorless wideband analog and RF techniques are considered as the key to
realize the next generation wireless multi-standard products in expensive nm-length
CMOS processes. As mentioned above, one of the keys to enhance the DR is
via elevating VDD, or in a wider extent, using mixed VDD as well as thin- and
thick-oxide transistors offered by advanced technologies [14]. Thin-oxide
transistors are normally for core design having a voltage limit which is the same
4 1 Introduction
as the nominal VDD. Thick-oxide transistors are for I/O design that can work at a
higher supply voltage. When design-for-reliability (i.e., prevents any device to be
under overstress) is included in the design flow, an elevated VDD outpacing the
technology roadmap can reliably enlarge the voltage headroom (VDD�VT) as
depicted in Fig. 1.3, where VT denotes the device threshold voltage.
As a way of demonstrating the techniques, the mobile-TV application is selected
as the common platform, which is recognized as one of the key functions for the next
generation smart-mobile-companion device [1]. The targeted mobile TV bands are
the VHF-III (174–248 MHz), UHF (470–862 MHz) and L (1.4–1.7 GHz)
bands as listed in Table 1.1. These three bands are where most mobile-TV standards,
2.5
3
3.5
4
Thr
esho
ld a
nd S
uppl
y V
olta
ges
(V)
0.13µm0.18µm
Technology Node
90nm 65nm0.25µm0
0.5
1
1.5
2
0.35µm
NominalVDD
An elevated VDD+
Design-for-Reliability
Elevated VDD
40nm
VT
Fig. 1.3 Benefiting thevoltageheadroomincrementviausinganelevatedVDDplusdesign-for-reliability
Table 1.1 Brief profiles of the most dominant mobile TV standards
Mobile TV standard
Frequency
(MHz) Data coding Air interface
T-DMB (Korea,
China, Europe)
174–245 Video: H.264 30 fps/QVGA OFDM (QPSK)
1,452–1,492 Audio: BSAC/AAC+
ISDB-T (Japan,
Brazil)
470–770 Video: H.264 15 fps/QVGA OFDM (QPSK/
16QAM)Audio: AAC+
DVB-H (Europe, US) 470–860 Video: H.264 30 fps/QVGA OFDM (QPSK/
16QAM/64QAM)1,670–1,675 Audio: AAC+
DMB-T (China) 470–860 Video: H.264 30 fps/QVGA OFDM (QPSK/
16QAM/64QAM)1,670–1,675 Audio: AAC+
MediaFLOTM (US) 712–722 Video: MPEG4 OFDM
1.3 Research Objectives 5
such as terrestrial–digital multimedia broadcasting (T-DMB), integrated services
digital broad-casting–terrestrial (ISDB-T), digital video broadcasting–handheld
(DVB-H) and digital multimedia broadcasting–terrestrial (DMB-T), reside.
Although the design examples in this book target mobile-TV applications, the
solutions are generally applicable to other wireless receivers in ultra-scaled
CMOS processes.
1.4 Organization of the Book
The materials are presented in six Chapters and one Appendix. Chapter 2 overviews
and studies, for the first time, state-of-the-art high-/mixed-voltage RF and analog
CMOS circuits and systems. Wherever appropriate their beneficial features
with respect to the conventional standard-voltage designs will be highlighted and
compared. From Chaps. 3 to 5, we present three circuit and system design examples
in nanoscale CMOS focusing on mobile-TV applications.
The first design is described in Chap. 3. It is a full-band mobile-TV low-noise
amplifier (LNA) with mixed-voltage ESD protection designed in 90-nm CMOS.
The details are outlined as follows:
• The LNA features a PMOS-based open-source input structure to optimize the
I/O swings under a mixed-voltage ESD protection while offering an inductorless
broadband input impedance match. The amplification core exploiting doublecurrent reuse and single-stage thermal-noise cancellation enhances the gain andnoise performances with high power efficiency. Optimized in a 90-nm 1.2/2.5-V
CMOS process with practical issues taken into account, the LNA using a
constant-gm bias circuit achieves competitive and robust performances over
process, voltage and temperature (PVT) variation. The simulated voltage gain
is 20.6 dB, noise figure is 2.4–2.7 dB and IIP3 is +10.8 dBm. The power
consumption is 9.6 mW at 1.2 V. |S11| < � 10 dB is achieved up to 1.9 GHz
without needing any external resonant network. Human Body Model ESD
zapping tests of �4 kV at the input pins cause no failure of any device.
Chapter 4 describes the second design, which is a high-voltage-enabled mobile-
TV RF front-end with TV-GSM interoperability fabricated in 90-nm CMOS.
The details are outlined as follows:
• It is an on/off-chip co-design employing externally three customized UHF/VHF
preselect filters, a RF switch and a balun. The integrated part includes:
(1) a cascode-cascade inverter-based low-noise amplifier that features a high
gain-to-power efficiency; (2) a linearized C-2C attenuator using reliable-
overdriven MOS switches; (3) an inductive-peaking feedforward path that
evens out the passband variation; and (4) two cascode I/Q mixer drivers
capable to drive passive mixers with small gain and bandwidth reduction.
Gate-drain-source engineering and self-biased structures are the keys enabling
6 1 Introduction
performance optimization with low power and no reliability risk. Fabricated in a
90-nm CMOS process with 1-V thin-oxide devices, the RF front-end measures
68-dB rejection at GSM-900 uplink, 0.7-dB passband roll-off, 3.9-dB noise figure
and �5.5-dBm IIP3 at a maximum voltage gain of 26.2 dB. The core occupies
0.28 mm2 and draws 15 mW. The achieved power-performance metrics
compares favorably with the prior arts.
The third design example is presented in Chap. 5. It is a mixed-voltage, unified
receiver front-end for full-band mobile TV fabricated in 65-nm CMOS. The details
are outlined as follows:
• The receiver front-end features three advances: (1) a gain-boosting current-balancing wideband balun-LNA shows low noise figure (NF), high IIP2 and
wideband output balancing concurrently; (2) a current-reuse mixer-lowpass filter(LPF) merges quadrature/harmonic-rejection mixing and third-order current-
mode post filtering in one block, enhancing linearity and noise just where both
are demanding, while saving power and area; (3) a direct injection-locked4-/8-phase local oscillator (LO) generator involving frequency division relaxes
themaster LO frequency. Fabricated in a 65-nmCMOS process, the RFE exhibits
4-dB NF, 35-dB maximum voltage gain and +32/�3.4-dBm out-of-channel
IIP2/IIP3 while dissipating 43–55 mW. The active die area is just 0.46 mm2.
The concluding remarks and recommendations for future work are summarized
in Chap. 6. The Appendix describes the theoretical analysis and design of open-loop
multi-phase local oscillator generators. It is a complementary part of the 65-nm
CMOS receiver front-end described in Chap. 5. Although this part is not the major
interest of this book, this self-contained Appendix allows the readers to digest such
a set of materials when concerning the design details in Chap. 5. Finally, it should
be noted that this book covers a wide range of topics. For in-depth investigation and
further study, some area requires additional reading of the references, such as [15]
about the mobile-TV standards and specifications. The key originalities of this
research have led to these publications [16–20].
References
1. G. Delagi, “Harnessing Technology to Advance to Next-Generation Mobile User-Experience,”
IEEE ISSCC, Digest of Technical Papers, pp. 18–14, Feb. 2010.2. L. Perre, J. Craninckx, A. Dejonghe, Green Software Defined Radios: Enabling Seamless
Connectivity While Saving on Hardware and Energy, Springer, 2009.3. P.-I. Mak, S.-P. U and R. P. Martins, Analog-Baseband Architectures and Circuits – for
Multistandard and Low-Voltage Wireless Transceivers, Springer, Sept. 2007.4. P.-I. Mak, S.-P. U and R. P. Martins, “On the Design of a Programmable-Gain Amplifier with
Built-in Compact DC-Offset Cancellers for Very Low-Voltage WLAN Systems,” IEEETransactions on Circuits and Systems – I: Regular Papers, vol. 55, no. 3, pp. 496–509,Mar., 2008.
References 7
5. R. Bagheri, A. Mirzaei, S. Chehrazi, M. E. Heidari, M. Lee, M. Mikhemar, W. Tang, and
A. A. Abidi, “An 800-MHz–6-GHz software-defined wireless receiver in 90-nm CMOS,”
IEEE J. Solid-State Circuits (JSSC), vol. 41, no. 12, pp. 2860–2876, Dec. 2006.6. Vassilios, K. Vavelidis and N. Haralabidis et al, “A 65-nm CMOS Multistandard, Multiband
TV Tuner for Mobile and Multi-Media Applications,” IEEE J. Solid-State Circuits, vol. 43,pp. 1522–1533, Jul. 2008.
7. C. Andrews, A. Molnar, “A Passive-Mixer-First Receiver with Baseband-Controlled RF
Impedance Matching, <6 dB NF, and >27dBm Wideband IIP3,” ISSCC Dig. Tech. Papers,pp. 46–47, Feb. 2010.
8. H. Moon, S. Lee, S-C. Heo, H. Yu, J. Yu, J-S. Chang, S-I. Choi, B-H. Park, “A 23 mW Fully
Integrated GPS Receiver with Robust Interferer Rejection in 65 nm CMOS,” ISSCC Dig. Tech.Papers, pp. 68–69, Feb. 2010.
9. J. Borremans, G. Mandal, V. Giannini, T. Sano, M Ingels, B. Verbruggenn and J. Craninckx
“A 40 nm CMOS Highly Linear 0.4-to-6 GHz Receiver Resilient to 0dBm Out-of-Band
Blockers,” ISSCC Dig. Tech. Papers, pp. 62–63, Feb. 2011.10. A. Annema, B. Nauta, R. Langevelde and H. Tuinhout, “Analog Circuits in Ultra-Deep-
Submicron CMOS,” IEEE Journal of Solid-State Circuits (JSSC), pp.132–143, Jan. 2005.11. W. Sheng, A. Emira and E. Sanchez-Sinencio, “CMOS RF Receiver System Design:
A Systematic Approach,” IEEE Trans. on Circuits and Systems – I: Regular Papers, vol. 53,no. 5, pp. 1023–1034, May. 2006.
12. M. El-Nozahi, E. Sanchez-Sinencio and K. Entesari, “Power-Aware Multiband–Multistandard
CMOS Receiver System-Level Budgeting,” IEEE Trans. on Circuits and Systems – II: ExpressBriefs, vol. 56, no. 7, pp. 570–574, Jul. 2009.
13. B. Serneels and M. Steyaert, Design of High voltage xDSL Line Drivers in Standard CMOS,Springer, 2008.
14. STMicroelectronics technology profile in Circuits Multi-Projet (R) [Online]: http://cmp.imag.
fr/products/ic/
15. A. Youssef and J. Haslett, Nanometer CMOS RFICs for Mobile TV Applications, Springer,2010.
16. P.-I. Mak and R. P. Martins, “Design of an ESD-Protected Ultra-Wideband LNA in Nanoscale
CMOS for Full-Band Mobile TV Tuners,” IEEE Transactions on Circuits and Systems – I:Regular Papers, vol. 56, no. 5, pp. 933–942, May 2009.
17. P.-I. Mak and R. P. Martins, “A 2 � VDD-Enabled Mobile-TV RF Front-End with TV-GSM
Interoperability in 1-V 90-nm CMOS,” IEEE Transactions on Microwave Theory andTechniques, vol. 58, pp.1664–1676, Jul. 2010.
18. P.-I. Mak and R. P. Martins, “High-/Mixed-Voltage RF and Analog CMOS Circuits Come
of Age,” IEEE Circuits and Systems Magazine, Issue 4, pp. 27–39, Dec. 2010.19. P.-I. Mak and R. P. Martins, “A 0.46 mm2 4-dB NF Unified Receiver Front-End for Full-Band
Mobile TV in 65 nm CMOS,” IEEE International Solid-State Circuits Conference (ISSCC),pp. 172–173, Feb. 2011.
20. P.-I. Mak and R. P. Martins, “A 0.46-mm2 4-dB NF Unified Receiver Front-End for Full-Band
Mobile TV in 65-nm CMOS,” IEEE Journal of Solid-State Circuits (JSSC), pp. 1970–1984,Sept. 2011.
8 1 Introduction
Chapter 2
General Considerations of High-/Mixed-VDD
Analog and RF Circuits and Systems
2.1 Introduction
Instead of just following the rapid downsizing of VDD in technology scaling,
high-/mixed-voltage RF and analog CMOS circuits and systems have emerged as
a prospective alternative [1], to deal with the wireless technology trends such
as software-defined radio and cognitive radio; both are hungry for bandwidth and
dynamic range. An elevated VDD, or a hybrid use of I/O and core VDD’s,
in conjunction with optimum selection of thin- and thick-oxide MOSFETS open
upmuch new design possibilities in re-defining circuit topologies, whilemaintaining
most speed and area benefits of advanced fine linewidth processes [2]. Voltage-
conscious bias techniques and overdrive protection circuits are simple and low
overhead techniques to ensure the reliability of all devices. This chapter studies
the basic design concept, system design considerations and some state-of-the-art
circuit examples. A wide variety of analog and RF CMOS circuits featuring high-/
mixed-VDD is discussed. Those circuits comprise power amplifier, low-noise ampli-
fier, mixer, operational-amplifier-based analog circuits, sample-and-hold amplifier
and line driver. Reliability metrics such as oxide breakdown voltage, hot carrier
injection (HCI), time dependent dielectric breakdown (TDDB), and bias tempera-
ture instability (BTI) will be briefly addressed. The involved concepts and
techniques are generally extendable to different wireless and non-wireless
applications.
2.2 System Considerations
A general system architecture of mixed-voltage wireless system-on-chip (SoC) for
portable applications is depicted in Fig. 2.1. Thin-oxide transistors powered byVDD,c
(core VDD) exhibit the simplest structure to maximize the speed-to-power efficiency
P.-I. Mak and R.P. Martins, High-/Mixed-Voltage Analog and RF Circuit Techniquesfor Nanoscale CMOS, Analog Circuits and Signal Processing,
DOI 10.1007/978-1-4419-9539-1_2, # Springer Science+Business Media New York 2012
9
of digital functions such as the digital signal processor. However, RF and analog
circuits such as theRFpower amplifier and the baseband operational amplifier are not
that efficient to work under the same VDD,c, which in the latest technologies, such as
the 65 and 40-nmCMOS, has values in the order of 1 to 0.9V [3], respectively. Such a
reduced VDD,c limits: (1) the overdrive voltages on transistors, and (2) the linear
output swing. Both of them can directly burden the performance optimization,
especially in multistandard wireless systems that demand high-linearity low-noise
wideband RF circuits [4]. Consequently, the exploration of a voltage islandingconcept in a power management unit would become essential for distribution of
the supply voltages to different RF and analog functions appropriately.
On the other hand, since many peripherals do not scale synchronously with the
silicon technologies, thick-oxide transistors are still kept available in advanced
processes to facilitate I/O communications. Thus, bringing thick-oxide transistors,
and their associated VDD,IO, into the RF and analog circuit design portfolio appears
to be a handy option to increase the design flexibility. Thick-oxide transistors can
be considered as devices from previous technology nodes: 0.25 and 0.18 mm.
Their reliable operating voltages are 2.5 and 1.8 V, respectively. Both are much
more comfortable values for RF and analog circuit design and can be easily
generated by a 3.6/3.7-V Li-ion battery. Obviously, circuits built with purely
thick-oxide transistors are not preferred as they cannot profit the speed and area
benefits of advanced processes. A hybrid use of thin- and thick-oxide transistors,
VDD,c and VDD,IO, emerges then as a new art in electronics that should be
adopted with a sensible balance. In the next sections, before the high-/mixed-
voltage-enabled circuits for wide types of RF and analog functions are described,
we will discuss the key device reliability concerns.
PowerManager
VDD,c
VDD,IO
Li-ion Battery2.7~ 4.2 V
1.2 V
2.5 V
RF & Analog
2.5 V I/O
Digital & 1.2 V I/O
Core Device(Thin-Oxide)
IO Device(Thick-Oxide)
LV Peripherals
HV PeripheralsPeripherals: SRAM, Flash, LCD
Display, LED, Key Pad, SIM Card….
SoC
Fig. 2.1 Increase the design flexibility of analog and RF circuits with more options on voltage
supplies and devices
10 2 General Considerations of High-/Mixed-VDD Analog. . .
2.3 Device Reliability in Ultra-scaled Processes
The technology Design Rule Manual provides the key device reliability concerns
including the absolute maximum rating (AMR), hot carrier injection (HCI),
electrostatic discharge (ESD), time dependent dielectric breakdown (TDDB), bias
temperature instability (BTI) and punchthrough effect. Complying with them in the
design indeed translates the term “design for reliability” into “voltage-conscious
design”, highly simplifying the design and verification methodologies [5]. Further-
more, in the topology formation phase, their implications to the circuits can be
easily identified. Other reliability issues related to interconnects and materials
like electromigration, stress-induced voiding and mechanical weakness are beyond
the scope of this work.
2.3.1 AMR
The AMR corresponds to the maximum voltage applied to a minimum-gate-length
device with no unrecoverable hard failure. AMR is concerned mainly with the gate-
oxide breakdown voltage as it is 3–4 times smaller than the junction breakdown
voltage [5]. A device biased close to the AMR limit may also lead to a deviation in
device parameters, degrading the long-term reliability. The tolerable AMR is
continuously reducing with the technologies (e.g., 1.6 V in 90-nm CMOS),
complicating the design of ESD protection in high-frequency pins.
2.3.2 HCI Lifetime
Degradation of MOS device characteristics occurs as a result of exposure to a high
VDS with a large drain current. Examples of degradation are a shift of VT and a
shorter gate-oxide breakdown lifetime. HCI normally happens in high-power
circuits such as the power amplifier, where the worst HCI bias conditions:
VDS≧VGS≧VT and VDS≧VDD/2 are concurrently satisfied. HCI degradation can be
reduced by lowering the drain current or increasing the device channel length (L).
2.3.3 TDDB
TDDB is the wear-out of insulating properties of silicon dioxide in the CMOS gate,
leading to the formation of a conducting path through the oxide to the substrate.
In order to protect the circuit against TDDB the catastrophic destruction of gate
oxides induced by the maximum DC gate oxide voltage at different temperatures
2.3 Device Reliability in Ultra-scaled Processes 11
must be considered. According to the maximum DC gate oxide voltages of 90 and
65-nm CMOS given in Table 2.1, NMOS has a higher voltage standing capability
than PMOS for all cases to prevent TDDB. Thus, NMOS is preferable when
considering TDDB in circuit design.
2.3.4 NBTI
BTI degradation happens under steady-state conditions. It is design dependent in
analog and RF circuits and primarily only PMOS devices are subjected to BTI
stress, namely negative BTI (NBTI). In a VDD-upscaled design, analyzing NBTI
involves detecting, in all modes of operation (DC and small signal), which PMOS
device is exposed to a peak or rms voltage value exceeding the standard VDD, which
is around 1 V in 90 and 65-nm CMOS. Thus, NMOS is also preferred when
implementing analog switches.
2.3.5 Punchthrough
Transistor gate length should be increased wherever possible to prevent the drain-
source depletion regions from punchthrough. In 90-nm CMOS, for a transistor
having an aspect ratio (W/L) of 10/0.1, a strong increment of drain current due to
punchthrough effect starts at a value of |VDS| around 2.3 V. Although the punch-
through effect is not intrinsically destructive it can accelerate, in the long term, the
gate oxide breakdown because of the induction of hot carriers. Punchthrough is a
critical concern in high-power circuits such as the power amplifier (PA), but it can be
avoided for low-power RF circuits such as the low-noise amplifier (LNA).
2.4 Extend the Voltage Capability of Thin-
and Thick-Oxide Transistors
With respect to the above-mentioned reliability concerns, individual thin (thick)-
oxide transistors can withstand maximally just one VDD,c (VDD,IO) voltage differ-
ence for any of the two terminals. In order to extend their voltage capability,
Table 2.1 Maximum DC gate oxide voltage to prevent TDDB
90 nm CMOS 65 nm CMOS
45�C 150�C 45�C 150�CGP NMOS 1.43 V 1.28 V GP NMOS 1.35 V 1.23 V
GP PMOS 1.29 V 1.17 V GP PMOS 1.23 V 1.11 V
12 2 General Considerations of High-/Mixed-VDD Analog. . .
stacking of devices can be applied. The concept is illustrated in Fig. 2.2. In steady
state, the possible structures can be a stack of two (or more) thin-oxide transistors,
thick-oxide transistors, or a hybrid use of both. Generally, the voltage capability
across the drain and source terminals can be multiplied by the number of stacked
transistors. One basic request of this technique is that the bulk should be tied to the
source terminal to avoid overstress between them, implying the need of a triple-well
process for NMOS to have an isolated bulk.
Among the three structures shown in Fig. 2.2 only pure stack of thin-oxide
transistors and a hybrid stack of thin- and thick-oxide transistors are relevant to
balance the speed and voltage capability. A pure stack of thick-oxide transistors
cannot take advantage of the area and speed features of advanced technologies.
In the hybrid case the thin-oxide transistor can serve as the amplification device
for minimization of the loading effect to the previous stage. The thick-oxide
transistor serves as the cascode device thus increasing the voltage capability.
High-/mixed-voltage RF and analog circuits are generally based on these two
stacking structures.
In addition to steady-state overstress, transient-state overstress should not be
allowed too. Depending on the nature of the signal processing, large-signal circuits
(e.g., line driver) requires checking the trajectory of all nodes. Alternative solutions
are to employ voltage-biased and self-biased circuit topologies; both of them
have the benefit that the internal node voltages can be easily controlled during
power up/-down transients. Examples of the techniques will be discussed in the
next section.
VDD,c
VDD,c
VDD,c
VDD,IO
VDD,IO
VDD,IO
VDD,IO
VDD,c
VDD,c
VDD,c
VDD,c
VDD,c
VDD,c
VDD,c
VDD,c
VDD,c
VDD,IO
VDD,IO+VDD,c
VDD,IO
VDD,IO
VDD,IO
VDD,IO
VDD,IO
2VDD,IO2VDD,c
VDD,IO
VDD,IO
Thin-Oxide MOS Thick-Oxide MOS
Fig. 2.2 Extending the voltage capability of thin-and thick-oxide transistors through stacking
2.4 Extend the Voltage Capability of Thin- and Thick-Oxide Transistors 13
2.5 High-/Mixed-Voltage Building Blocks
2.5.1 Power Amplifier and Wideband Balun-LNA(High-VDD þ Mixed-Transistor)
VDD-upscaling circuits have appeared in the literature for many years. The most
common application is on the PA. As shown in Fig. 2.3a a hybrid use of thin- (M1)
and thick-oxide device (M2) in cascode permits using of a higher VDD beyond the
standard value to maximize the possible output power in 0.13-mm CMOS [6].
The reliability test should be under the maximum output power level with the
entailed modulation (e.g., 64QAM OFDM), to ensure the steady-state overstress
conditions are met (i.e., │VGD,rms│, │VGS,rms│ and │VDS,rms│ have to be less than VDD
of each device type). M2 entails a triple well for independent bulk-source
connection.
On the other hand, in order to protect M1 from overstressing automatically
during the power-up/down transients, we propose to add a thick-oxide device
Mpt1 can be added to the Vx node. Its size is not critical as its aim is to ensure Vx
< VDD,c when VDD,elevated is activated first and can be turned off when VDD,c has
caught up automatically.
Mixed-transistor circuit topologies also find applications in recent small-signal
linearity-demanding wireless circuits and systems [7–9]. A 90-nm CMOS ultra-
wideband balun low-noise amplifier (LNA) [7] based on an elevated VDD (2.5 V)
VDD,elevatedVDD,elevated
VDD,elevated
VDD,elevated
Vb5VDD,c
VDD,c
VDD,elevated
VDD,C
Vb1
Vin
Vb
Vb4Vin
Mb1
Rb Vx
Vx
Vb2
LLLL
LL1
RLRL
M3
R2
R1
M2
M1
M1
M2
M4
C1
Voutn
Vout Voutp
Ib
C1
IMpt1
IMpt1
Automatic Overdrive Protection for M1
0t
V/I When VDD,elevated ramps upbefore VDD,c, M1 is cutoff
Mpt1,2 ensures Vx<VDD,c
Mpt1 is cutoff when innormal operation
Mpt1
a b
Fig. 2.3 RF circuits using an elevated VDD. (a) Power amplifier with automatic protection of M1
from being overstressed. (b) Wideband balun LNA
14 2 General Considerations of High-/Mixed-VDD Analog. . .
increases the output dynamic range while allowing more voltage drop at the
resistive load RL (Fig. 2.3b), achieving both high gain and high linearity but a
smaller output bandwidth due to an increased RL. A gain-peaking inductor LL can
be exploited to extend the output bandwidth.Mpt1 in Fig. 2.3a can also be applied to
this topology to protect M1 and M3.
2.5.2 Passive Mixers (High-VDD þ Thin-Oxide Transistor)
A passive current-mode downconversion mixer can be implemented with a resistor
Rff in-series with a MOS switch, and terminated with a virtual ground through the
use of an operational amplifier (OpAmp). The OpAmp provides linear I-V conver-
sion and first-order lowpass filtering at the output. In order to achieve a rail-to-rail
output swing the output dc-level should be at half of the supply. For a generic
1 � VDD design as shown in Fig. 2.4a, the clocked MOS switch can only have
1xVDD
1xVDD
2 xVDD
1xVDD
2xVDD
0
Vout
Vout
Vin
Vin
VSS
VSS
Rff
Rff
Rfb
Cfb
Rfb
Cfb
VCM(0.5xVDD)
VCM(1xVDD)
0.5xVDD-0.5xVDD
VGS =
1xVDDVGS =
(VDC = 0.5xVDD) (VDC = 0.5xVDD)
(VDC = 0.5xVDD)
(VDC = 1xVDD)
(VDC = 1xVDD)(VDC = 1xVDD)
0
a
b
Fig. 2.4 Downconversion passive mixers: (a) 1 � VDD design. (b) 2 � VDD design
2.5 High-/Mixed-Voltage Building Blocks 15
a maximum overdrive voltage of 0.5 VDD. However, by doubling the supply to
2 � VDD as shown in Fig. 2.4b, the overdrive voltage of the MOS switches is
maximized to the technology allowable limit, i.e., 1 � VDD. This act significantly
reduces the size of the MOS switch and its induced nonlinearity. The design of
2 � VDD OpAmp will be presented later in this chapter.
Another type of 2� VDD passive mixer with a mixer driver is shown in Fig. 2.5a.
A 2 � VDD 90-nm CMOS cascode amplifier serves as the mixer driver improving
the linearity and reverse isolation but the output dc-level is up-shifted to 1.5� VDD.
Under a 1.5 � VDD dc-level PMOS is preferred as the mixing MOS to maximize
the overdrive voltage. The unmatched input and output dc-levels of the OpAmp
require an extra bias current Ib to sink out the excess dc-current in the feedback
loop. Since Ib depends on the absolutely value of Rfb, a resistance-tracking
0
=
3R
R
Vout
lb
Vin
a
b
2xVDD
2xVDD
2xVDD
Rfb
Rfb
3Rfb
ID
Ib
R2
M2
M1
R1
Cfb
VSSVSS
VSS
VSSVSS
VSS
VCM(1.5xVDD)
(VDC ª 1xVDD)
(VDC ª 1.5xVDD)
(VDC ª 0.5xVDD)
(VDC ª 0.5xVDD)
(VDC ª 0.5xVDD)
1.5xVDD
1.5xVDD
0.5xVDD
1xVDDVGS =
2xVDD
0.5xVDD
Fig. 2.5 (a) 2 � VDD passive mixer with mixer driver and its (b) Ib generation circuit
16 2 General Considerations of High-/Mixed-VDD Analog. . .
bias circuit can be utilized for this purpose. As shown in Fig. 2.5b, an error
amplifier together with a 3Rfb and a current mirror generates the required value of
Ib ¼ 0.5 � VDD/Rfb. It should be noted that no device is under overstress.
2.5.3 Differential Pair (High-VDD þ Thin-Oxide Transistor)
A 3.3-V 0.18-mmCMOS two-stage operational amplifier (OpAmp) was demonstrated
in [10]. The concept of extending the voltage is related with the addition of
extra cascode transistors, boosting the voltage-withstand capability from 1.8 to
3.3 V. The input stage is of particular interest as it is based on a high-voltage-
enabled differential pair using a current mirror load. In order to understand the
performance difference in an advanced process, we re-designed and compared
only the input stage of the OpAmp (output stage is more customized) in 65-nm
CMOS, as shown in Fig. 2.6a and b. Table 2.2 summarizes the simulation
VinnVinp Vinp
a b
1xVDD
2xVDD
Vbp1
Vbp1
Vbp2
VinnM2
M2
M4
M4
M5 M5 M6
M7 M8
M3M3
M1
M1
CLCL
(VDC ª 0.3xVDD)
(VDC ª 1xVDD)
(VDC ª 0.55xVDD)
(VDC ª 0.7xVDD)
(VDC ª 1.5xVDD)
Fig. 2.6 Typical OpAmp’s input stage: (a) 1 V (typical) (b) 2 V-enabled
Table 2.2 Comparison between 65-nm1-V and 2-VOpAmp’s input stagewith a current-mirror load
Parameters 1 V OpAmp’s input stage 2 V OpAmp’s input stage
Technology 65 nm CMOS
Transistor type 1 V GP NMOS and PMOS
Power consumption 0.4 mW
Load CL 1 pF
DC gain 10 dB 18.4 dB
Unity-gain frequency 318.5 MHz 191 MHz
Phase margin 107� 95�
HD3 (@ 1 MHz input) 45.6 dB at 125 mVpp Output 44.3 dB at 330 mVpp Output
Output noise voltage
(@ 100 MHz)
14.6 nV/sqrtHz 14.7 nV/sqrtHz
2.5 High-/Mixed-Voltage Building Blocks 17
results showing that the dc gain and linear output swing of the 2-V design are
8.4 dB and 2.6 times better than its 1-V counterpart, respectively. On the other
hand, the unity-gain frequency is reduced by 40% due to the additional parasitic
poles in the 2-V design. Thus, when the speed is not that demanding, a 2-V-
enabled OpAmp is better since OpAmp-based circuits such as the active filter
are normally for baseband operation.
2.5.4 Recycling Folded Cascode Operational Amplifier(High-VDD þ Thin-Oxide Transistor)
Based on the above observation, a 2 � VDD-enabled recycling folded cascode
(RFC) OpAmp [10] is proposed and compared with its 1 � VDD RFC counterpart
[11] and its original 1 � VDD folded cascode (FC) counterpart. Without resorting
from a two-stage or multi-stage OpAmp, this 2 � VDD-enabled RFC OpAmp can
offer sufficient open-loop gain and linear output swing to realize high-precision
analog functions in a single stage. The reliability of the circuit is ensured via
voltage-conscious biasing and gate-drain-source engineering.
The single-stage FC OpAmp is shown in Fig. 2.7. Its performance can be
enhanced with the RFC technique [11] as shown in Fig. 2.8. It employs M1b, M2b,
M11, M12, M3b andM4b to improve concurrently the gain and speed. By controlling
the current mirror gain K, the small signal transconductance can be boosted,
M4M3
M5 M6
M8
M9M10
M7M1
Vbp1
Vbn2
Vbp2
Vcmfb
Vbn1
M0
M2
GND
Vinp Vinn
VOn
VDD
VOp
lblb2lb
2lb2lb
Fig. 2.7 Typical RFC OpAmp (1 � VDD design)
18 2 General Considerations of High-/Mixed-VDD Analog. . .
i.e., gmRFC ¼ gm1a (1 + K). Thus, under a fixed power budget, the RFC shows
a higher gain-bandwidth product (GBW) than the conventional FC structure.
Furthermore, the RFC OpAmp also exhibits larger output resistance than its FC
counterpart, leading to further gain enhancement.
The proposed 2 � VDD RFC OpAmp is depicted in Fig. 2.9. The aim of
doubling the supply is to enhance certain performance metrics that cannot be
simply obtained by doubling the bias current at 1 � VDD under the same power
budget. By appropriately doing transistor stacking and biasing, the output resis-
tance of the devices can be boosted while the voltage stress on them can also be
shared to meet the reliability limits. For instance, under a 2 � VDD, M00 can be
added to share the voltage stress on the current source M0 and improve its output
resistance, thereby the OpAmp’s common-mode rejection ratio (CMRR). On the
other hand, the current mirrors (M3a, M3b, M4a, M4b) are cascoded with (M3a0,
M3b0, M4a0, M4b0), whereas the current sources of the output stage (M11, M12) are
cascoded with (M9,M10). Thus, the overall DC gain should be enhanced due to the
boosted output resistance RO,RFC-2V by comparing it with the RO,RFC-1V as
summarized in Table 2.3.
The bias and common-mode feedback circuits (CMFB) are tailored to include
extra cascode devices to ensure all node voltages are within the reliability limits as
shown in Fig. 2.10.
M7
K:1
(K-1)Ib /2(K-1)Ib /2
1:K
GND
1xVDD
M4aM3a M3b M4b
M2b M2aM1a M1b
M0
M12
M5 M6
M11Vbn2
Vbn2
M8
M9 M10Vcmfb
VOpVOn
Vinn
2lbVbp1
Vbp2
Vinp
Fig. 2.8 Typical RFC OpAmp (1 � VDD design)
2.5 High-/Mixed-Voltage Building Blocks 19
In order to ensure the reliability and performance of the 2� VDD RFC OpAmp, it
should be tested in a closed-loop way based on the intended applications. Here, a
first-order active-RC circuit with unity gain is assumed, as shown in Fig. 2.11.
To check the reliability of all devices inside the OpAmp, a large square-wave
input at a common-mode voltage of 1 V is applied. Figure 2.12 shows the VGS–VGD
relationship at an input swing of 1.2 Vpp. Since the circuit is differential, only
half-circuit results are shown. It can be observed that both VGS and VGD vary within
the ±1-V boundary and the variation is indeed small. Next, the VDS trajectory is
checked. Figure 2.13a–c show that the VDS, in a period of square-wave input under
an input swing of 1.2, 1.6 and 2 Vpp, respectively. When the input swing is 1.2 V, all
VDS are within the ±1-V boundary. When the input swing is enlarged, some of the
VDS exceed 1 V, reached 1.13 and 1.25 V for 1.6 and 2-Vpp inputs, respectively.
According to the lifetime targets discussed in [12], VDS > 1 V may still be
acceptable for some applications.
In order to compare the performances between 1 � VDD FC, 1 � VDD RFC and
2� VDD RFC OpAmps fairly, they are designed under the same power budget. The
simulated open loop DC gain (Fig. 2.14) of the FC, 1-V RFC and 2-V RFC is
45.3 dB, 54.0 dB and 72.8 dB respectively, which indeed demonstrates the
enhanced output impedance of the 2-V RFC and the enhanced gain of the 1-V
RFC over the FC is apparent and within the expected range of 8–10 dB.
K:1 1:K
GND
M4aM4bM3a M3b
M3a’ M3b’ M4b’ M4a’Vbn0 Vbn0
Vinp
M2b M2aM1a M1b
Vinn
M0
M0’
lb
Vbp1′
Vbp1(K-1)Ib /4(K-1)Ib /4
M9 M10Vcmfb
M7’ M8’
Vbp3
M7 M8Vbp2
M5 M6Vbn2
VOpVOn
2xVDD
Fig. 2.9 Proposed 2 � VDD-enabled RFC OpAmp
20 2 General Considerations of High-/Mixed-VDD Analog. . .
Table
2.3
Designequationsof1�
VDDFC,RFCand2�
VDDRFCOpAmps
FC
1-V
RFC
2-V
RFC
Current
I D¼
1 2mC
oxW LðV
GS�VTÞ2
Transconductance
Gm;FC¼
gm1
Gm;1V�R
FC¼
gm1að1
þKÞ
Gm;2V�R
FC¼
gm1að1
þKÞ
Outputim
pedance
RO;FC�
½ðgm5þgmb5Þr o
5
ðr o1jjr
o3Þ�jjðg
m7r o
7r o
9Þ
RO;1V�R
FC�
½ðgm5þgmb5Þr o
5ðr o
1ajjr
o3a�
jjðg m
7r o
7r o
9Þ
RO;2V�R
FC�
fðg m
5þgmb5Þr o
5½ðr
o1ajjðg m
3a0
þgmb3
a0Þr o
3a0 ro3a�g
jjðg m
7r o
7g m
70 ro7
0 ro9Þ
Gain
Av;FC¼
Gm;FC�R
O;FC
Av;1V�R
FC¼
Gm;1V�R
FC�R
O;1V�R
FC
Av;2V�R
FC¼
Gm;2V�R
FC�R
O;2V�R
FC
GBW
GBW
FC�
Gm;FC
CL
GBW
1V�R
FC�
Gm;1V�R
FC
CL
GBW
2V�R
FC�
Gm;2V�R
FC
CL
Outputsw
ing
VO
Swing;FC¼
2�½1�ðV
ov3þ
Vov5þjV
ov7jþ
jVov9j�
VO
Swing;1V�R
FC¼
2�½1�ðV
ov3aþ
Vov5þjV
ov7jþ
jVov9j�
VO
Swing;2V�R
FC¼
2�½2�ðV
ov3a0þVov3aþVov5
þjV
ov7jþ
jVov70 jþ
jVov9j�
Slew
rate
SRFC¼
I b CL
SR1V�R
FC¼
KI b
CL
SR2V�R
FC¼
KI b
2CL
Thermal
noise
v2 iT;FC¼
8k BTg
g m1�½1
þgm
3
gm1þ
g m9
g m1��Df
v2 iT;1V�R
FC¼
8k BTg
g m1að1þK
Þ�½ð1
þK2Þ
1þK
þgm3a
gm1aþ
1ð1þK
Þg m
9
gm1a��Df
v2 iT;2V�R
FC¼
8k BTg
gm1að1þK
Þ�½ð1
þK2Þ
1þK
þg m
3a
g m1aþ
11þK
g m9
gm1a��Df
Flicker
noise
v2 if;FC¼
KFP
m PC2 oxW
1L1f½1þ2KFN
KFPðL 1 L
3Þ2
þðL 1 L
9Þ2 �
�Df
v2 if;1V�R
FC¼
KFP
m PC2 oxW
1aL1að1þK
Þf½ð1
þK2Þ
ð1þK
Þþ
KKFN
KFPðL 1
a
L3aÞ2
þðK
�1Þ
ð1þK
ÞðL1a
L9Þ2 �
�Df
v2 if;2V�R
FC¼
KFP
m PC2 oxW
1aL1að1þK
Þf½ð1
þK2Þ
ð1þK
ÞþK
KFN
KFPðL 1
a
L3aÞ2
þðK
�1Þ
ð1þK
ÞðL1a
L9Þ2 �
�Df
Power
P¼
I Total;FC�1
P¼
I Total;1V�R
FC�1
P¼
I Total;2V�R
FC�2
2.5 High-/Mixed-Voltage Building Blocks 21
As for the GBW, the FC, 1-V RFC and 2-V RFC is 68.2 MHz, 157.8 MHz and
97.5 MHz, The GBW of the 2 � VDD design, as expected, is less than to that in
the 1 � VDD design under the same power budget, but is fairly adequate for most
analog functions with signal bandwidth of less than 10 MHz. The phase margin
simulated for FC, 1-V RFC and 2-V RFC is 86.6�, 60.9� and 70.7� at their
respective GBWs. The phase margin of 1-V RFC is smallest since it has the largest
GBW, and has more poles compared to FC. The 2-V RFC shows less phase
margin compared to FC, since the larger GBW and the multiple poles added by
2xVDD
GND
Vref=1V
Ib
CMFBBias circuit
Vbn1
Vbn2
Vbp3
Vbp1
Vbp2
VBp1’
Vcmfb
VOutn
VOutp
Fig. 2.10 Bias and CMFB circuits for the 2 � VDD-enabled RFC OpAmp
R
R
R
R
Vinp
Vinn
C1 C1
C2
C2
C1 C1
Voutp
Voutn
Fig. 2.11 A unity-gain amplifier is used to assess the performance and reliability of a 2 � VDD-
enabled RFC OpAmp. R ¼ 600 kO, C1 ¼ 2 pF and C2 ¼ 4 pF. The input and output dc-levels are
1 � VDD to maximize the signal swing and allow ease of cascading
22 2 General Considerations of High-/Mixed-VDD Analog. . .
-0.75
0.0
0.75
1.5
-1.50.0 1.0u0.5u 1.5u 2.0u
Time [S]
0.95
: M0 : M0' : M2b : M2a : M4b’ : M4a’: M4b : M4a : M6 : M8 : M8’ : M10
: M0 : M0' : M2b : M2a : M4b’ : M4a’: M4b : M4a : M6 : M8 : M8’ : M10
: M0 : M0' : M2b : M2a : M4b’ : M4a’: M4b : M4a : M6 : M8 : M8’ : M10
1.2Vpp Input Swing 1.6Vpp Input Swing
2VppInput Swing
-1.50.0 2.0u1.0u0.5u 1.5u
Time [S]
0.0
1.5
1.13
-0.75
0.0
0.75
1.5
-1.50.0 1.0u0.5u 1.5u 2.0u
Time [S]
1.25
VD
S[V
]
VD
S[V
]
VD
S[V
]
a b
c
Fig. 2.13 VDS variation versus time at an input swing of (a) 1.2 V, (b) 1.6 V and (c) 2 V. The
notations correspond to Fig. 2.9. The maximum Vds is within 0.95/1.13/1.25 V, respectively
-0.5
0
0.5
1
M0 M0' M2b M2a M4b'
M8'
M4a'
M4b M4a M6 M8
-1 -0.5 0 0.5 11
M10
VGS[V]
VG
D[V
]
Fig. 2.12 2 � VDD-enabled RFC OpAmp’s VGS–VGD trajectories when a square-wave input is
applied with a signal swing of 1.2 Vpp
2.5 High-/Mixed-Voltage Building Blocks 23
cascode and current mirror. Nevertheless, neither amplifier shows any ringing in the
transient performance.
The linearity of the OpAmps is assessed as follows: a two-tone test centered
around 500 kHz (250 mVpp at 450 kHz and 250 mVpp at 550 kHz) was applied to
the three OpAmps, and their results are shown in Fig. 2.15a–c. The third intermod-
ulation distortion, IM3, is �49.7 dB (1-V FC), �57.2 dB (1-V RFC) and �76.5 dB
(2-V RFC), as shown in Fig. 2.16. For an analog-to-digital converter, the achieved
gain of the 2-V RFC OpAmp corresponds to>11-bit resolution for an output swing
as large as 0.8 Vpp. When they are in a unity-gain configuration, Fig. 2.17 confirms
the high gain accuracy of the 2-V RFC OpAmp over such a wide output swing.
The input-referred noises of the three OpAmps are shown in Fig. 2.18. When
integrated over a bandwidth of 1 Hz to 100 MHz, the noises are 74.4 mVrms (FC),
64.3 mVrms (1-V RFC) and 83.2 mVrms (2-V RFC). The latter is actually inferior
103 104 105 106 107 108 109
103 104 105 106 107 108 109
-40
-20
0
20
40
60
80
Frequency (Hz)
Mag
nitu
de (
dB)
-180
-160
-140
-120
-100
-80
-60
-40
-20
0
Frequency (Hz)
Pha
se (
deg)
FC1-V RFC2-V RFC
FC1-V RFC2-V RFC
a
b
Fig. 2.14 (a) Gain and (b) phase responses of 1-V and 2-V RFC OpAmps
24 2 General Considerations of High-/Mixed-VDD Analog. . .
2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5x 105
x 105
x 105
-100
-80
-60
-40
-20
0
Frequency (Hz)
FC
- A
mpl
itude
(dB
) -12.239
-61.9072
2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5-100
-80
-60
-40
-20
0
Frequency (Hz)
1-V
RF
C -
Am
plitu
de (
dB)
-12.1225
-69.3184
2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5-120
-100
-80
-60
-40
-20
0
Frequency (Hz)
2-V
RF
C -
Am
plitu
de (
dB)
-12.0629
-88.595
a
b
c
Fig. 2.15 Two tone FFT spectrums of (a) the FC, (b) the 1-V RFC and (c) the 2-V RFC for a
0.5 Vpp signal centered at 500 kHZ and separated by 100 kHz
2.5 High-/Mixed-Voltage Building Blocks 25
comparing with the 1-V ones. Table 2.4 summarizes their simulation results.
It concludes that the 2 � VDD RFC OpAmp is more efficient in improving the
gain precision and linearity of analog circuits. However, when GBW and noise are
the priorities, the 1-V RFC OpAmp becomes more superior. Voltage-oriented
OpAmp design, therefore, improves the performance metrics very different from
the current-oriented ones in nm-length CMOS processes.
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 210
20
30
40
50
60
70
80
Output Voltage (V)
Gai
n (d
B)
FC1V-RFC2V-RFC
Fig. 2.16 Open-loop gain versus output voltage. The common-voltage voltages are Vcm, FC ¼0.5 V, Vcm,1-V RFC ¼ 0.5 V and Vcm,2-V RFC ¼ 1 V
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
0.2
0.4
0.6
0.8
1
Output Voltage (V)
Gai
n (V
/V)
FC1V-RFC2-V RFC
Fig. 2.17 OpAmps in a unity gain configuration. The common-voltage voltages are Vcm,1-V FC ¼0.5 V, Vcm,1-V RFC ¼ 0.5 V and Vcm,2-V RFC ¼ 1 V
26 2 General Considerations of High-/Mixed-VDD Analog. . .
2.5.5 OpAmp-Based Analog-Baseband Circuits(Mixed-VDD þ Mixed-Transistor)
As shown above, a 2 � VDD recycling folded-cascode (RFC) OpAmp can achieve
better performances than its 1 � VDD counterpart such as DC gain and close-loop
linearity under a similar power budget. The applications of such an OpAmp are
extensive, as shown in Fig. 2.19. Depending on the selected impedances of Zfb and Zffseveral types of continuous-time and discrete-time circuits can be synthesized. Of
course, in some cases, it will need to interface between standard-VDD and high-VDD
building blocks.As shown inFig. 2.20a, in a transmitter the use of 1�VDD and2�VDD
OpAmp allows a progressively increase of linear output swing. The level shifter is a
current source Ib. Another case is shown in Fig. 2.20b for receiver, a 2� VDD OpAmp
(at the front) offerswider linear output swing to handle the out-of-channel interferer 1�VDD OpAmp (at the back) easily interfaces with 1 � VDD ADC. Again, Ib allows
seamless cascade of blocks having different common-mode voltages.
100 102 104 106 108
10-10
10-12
10-16
10-18
10-14
Frequency (Hz)
Spe
ctra
l Den
sity
(V
2/H
z)FC1-V RFC2-V RFC
Fig. 2.18 Input referred noise spectral power density
Table 2.4 Performance summary for 1-V and 2-V RFC OpAmps
Parameter FC 1-V RFC 2-V RFC
Power (bias current) [mA] 600 600 300
DC gain [dB] 45.3 54.0 72.8
GBW [MHz] 68.2 157.8 97.5
Open loop PM [deg] 86.6 60.9 70.7
Capacitive load [pF] 5.0 5.0 5.0
Slew rate (average) [V/ms] 53.3 96.6 65.4
1% settling time [ns] 24.8 9.8 18.0
Gain precision (closed loop, ideal case is 1) 98.6% 99.4% 99.8%
IM3, 0.5 Vpp at 0.5 MHz [dB] �49.7 �57.2 �76.5
Input referred noise (1 Hz–100 MHz) [mVrms] 74.4 64.3 83.2
2.5 High-/Mixed-Voltage Building Blocks 27
The Weighted Summer
Lowpass Filter
Current-Mode Mixer-+
-+
-+
Amplifier (or Buffer)
-+
Highpass Filter
-+
2xVDD
VSSVCM
Vin Vout(VDC = 1xVDD)
(VDC = 1xVDD)
Zfb
Zff
(1xVDD)
-
+
Fig. 2.19 Analog-baseband circuits based on a 2 � VDD OpAmp
Ib
Ib
Ri /2
Ri /2
Ri /2
Ri /2
Vx
Ri /2 Vx
0.5V1V
DAC
1xVDD
1xVDD
1xVDD-Vx
VSS
VSS VSS
VSS
VSS
VCM(0.5xVDD)
VCM(0.5xVDD)
VSSVCM(1xVDD)
VCM(1xVDD)
Signal
InterfererFullSwing
FullSwing
LNTAPassiveMixer
ADC
Signal
Interferer
(VDC ª 0.5xVDD)
(VDC ª 0.5xVDD) (VDC ª 0.5xVDD)
(VDC ª 0.5xVDD)
(VDC ª 0.5xVDD)
(VDC ª 1xVDD)
(VDC ª 1xVDD)
2xVDD
Rfb
Cfb
2xVDD
Rfb
Cfb
1xVDD
Rfb
Cfb
1xVDD
Rfb
Cfb
Vout
ID =
Ri /2
1xVDD-VxID =
a
b
Fig. 2.20 Level shifting: (a) 1 � VDD to 2 � VDD. (b) 2 � VDD to 1 � VDD
28 2 General Considerations of High-/Mixed-VDD Analog. . .
After lowpass filtering, the baseband signal can be driven off-chip or to the
analog-to-digital converter (ADC). In either case, a high-VDD source follower with
a thick-oxide MOS can be used as the buffer, as compared in Fig. 2.21a and b for
mixed-VDD and standard-VDD design. The mixed-VDD design inherently offers
level-shifting, avoiding any AC-coupling circuit that is area hungry at baseband,
i.e., the highpass cutoff frequency should be sufficiently low to prevent notching the
signal spectrum.
2.5.6 Low-Dropout-Regulator (Mixed-VDD þ Mixed-Transistor)
Low-dropout regulators (LDOs) are widely employed in SoC for improving the
power-supply rejection ratio (PSRR) of the internal circuit. PMOS-based LDO
(Fig. 2.22a) is more popular than its NMOS counterpart (Fig. 2.22b) for its lower
dropout voltage property. The main issue of PMOS-based LDO is the necessity of
an external big capacitor to ensure the stability. Such a requirement significantly
increases the manufacturing cost and pin counts because many LDOs are entailed
for a SoC. The NMOS-based LDO, on the other hand, is free from such a request
(i.e., cap-less) and features better stability and PSRR+. The key appeal is that VG
will need to be greater VDD12, which is not normally possible in a single-VDD
design. In a mixed-VDD design, VG > VDD12 can be solved as shown in
Fig. 2.22c. The abovementioned benefits of NMOS-based LDO are retained,
while an add-on benefit is that the maximum Vout,max can now be VDD12 (i.e., no
dropout voltage!). The drawback is that every 1-mA current to ZL yields 1.3-mW
power loss in the pass transistor. Thus, the technique befits better low-power high-
sensitivity circuits.
In addition to VDD-LDO, ground-LDO is becoming more crucial to desensitize
low-noise high-DR circuits, e.g., voltage-controlled oscillator (VCO), from sub-
strate noise coupling in a noisy SoC environment as shown in Fig. 2.23. Due to the
Vout,bufVout,buf
Thick-OxideSource Follower
Thick-OxideSource Follower
1xVDD 1xVDD1xVDD 2xVDD
R1 R1VDCª0.5xVDD VDCª0.5xVDD
VDCª0.5xVDDVDCª1xVDDM2
M2
M1 M1M3 M3
VoutVout
CDC
RDC
a b
Fig. 2.21 (a) Standard-VDD and (b) mixed-VDD source follower serves as a buffer
2.5 High-/Mixed-Voltage Building Blocks 29
presence of VDD25, VDD-LDO and ground-LDO can be jointly employed. The VCO
may employ a thick-oxide varactor or MOSFET capacitor as the frequency tuning
element, covering potentially a wider tuning range. Proper biases can ensure the
internal rail sufficiently large for the core circuit, surpassing the voltage-headroom
NMOSVG
Cap-less
Stability & PSRR +
PMOS
ZLCext
ZL
OTA OTAVref Vref
VG <VDD12
VG>VDD12 is solved
VG>VDD12
VG
VG
VG
External Cap
NMOS
Cap-lessStability & PSRR+
OutputStage
Vout,max=VDD12-VDSVout,max =VDD12-VDS
Vout,max =VDD12
Vout Vout
ZL
Vout
VDD25VDD25
VDD12 VDD12
VDD12OTA
a b
c
Fig. 2.22 LDO with (a) a PMOS pass transistor. (b) a NMOS pass transistor, and (c) mixed-
voltage design on a NMOS pass transistor
NMOS
Core Circuit
VDD,i
VDD,i
Mp1
Mn1 Mn2
Mp2OTA
PMOS
VGND,i
VGND,i
Vctrl
1.2V
0.65V
0.65V
~0.25V
~2.25V
0.65V
1.85VL
CC
Sufficient Internal Rail
Cap-lessStability & PSRR+/-
Thick-Oxide Varactor/MOS capacitor
VDD25VDD25
VDD25
Vref1
Vref2
VG1
VG2OTA
Fig. 2.23 Mixed-voltage mixed-transistor LDO for both positive and negative rails
30 2 General Considerations of High-/Mixed-VDD Analog. . .
tradeoff when employing LDO for PSRR improvements. Again, the drawback is
that every 1-mA current to ZL yields 1.3-mW power loss in the two pass transistors.
2.5.7 Sample-and-Hold Amplifier (High-VDD þ Mixed-Transistor)
Discrete-time analog-baseband circuits not only can benefit from the area and
power savings of a VDD-elevated OpAmp, but also the extra voltage headroom to
improve the linearity of sampling. Shown in Fig. 2.24a and b are two sample-and-
hold circuits with 1-V and 2.5-V supplies, respectively. At a 100-MHz sampling
rate, to sample-and-hold a 10-MHz 0.6-Vpp sinusoidal input at a dc level that is
midway to VDD/2, the former, based on thin-oxide MOS with a minimum channel
length of 60 nm can achieve 1.18-GHz tracking bandwidth (BW) but the HD3 is
limited to 31.5 dB. Alternatively, the latter based on thick-oxide MOS with a
minimum channel length of 280 nm can achieve 50-dB HD3, but the tracking
BW is almost halved. Then, this speed-linearity tradeoff is subject to applications
and can be flexibly selected in advanced processes, as both thin- and thick-oxide
devices are available. Since the clock is normally synthesized with the thin-oxide
circuit for power and area reduction, a clock level shifter, as shown in Fig. 2.25,
would be required for the 2.5-V design. Thus, power and speed overheads must be
considered.
1V
0V
0.8V0.5V
10/0.06
0.2V
1pF
10MHz
100MHz2.5V
0V
1.55V1.25V
10/0.28
0.95V
1pF
10MHz
100MHz
Output HD331.5 dB
Output HD350 dB
1.18 GHzTracking BW
0.57 GHzTracking BW
a b
Fig. 2.24 Sample-and-hold circuits: (a) 1-V design. (b) 2.5-V design
2.5V 2.5V
0V 0V
1V0
2.5V0
1V
0V
1V PLL
Fig. 2.25 Clock level shifter for 1–2.5 V
2.5 High-/Mixed-Voltage Building Blocks 31
2.5.8 Line Driver (High-VDD þ Thin-Oxide Transistor)
Reference [13] has demonstrated that a 5.5-V line driver realized in a standard 1.2-V
0.13-mmprocess is capable to attain state-of-the-art performances with no reliability
degradation. Figure 2.26 depicts the block schematic of such a central office (CO)
line driver topology (the output stage), where a 2 � VDD,c supply is adopted for
simplicity. The input signal is delayed (to synchronize with the upper path) and
buffered to drive the NMOS device M1, besides being also level-shifted up to drive
the PMOS device M4. The cascode transistors M2 and M3 serve to increase the
voltage-withstand capability. With a 2� VDD,c supply, the gate-bias voltages ofM2
andM3 are very simple, i.e., 1� VDD,c to ensure no overstress in both high- and low-
state outputs (Fig. 2.27). For a higher VDD multiplying design (i.e., greater than 2),
M1
M2
M3
M4
1xVDD,c1xVDD,c
1xVDD,c
2xVDD,c
2xVDD,c
2xVDD,c2xVDD,c2xVDD,c
1xVDD,c1xVDD,c
1xVDD,c
1xVDD,c
1xVDD,c
0
LevelShifter
0 0
Vin
Vout
0
0
Delay
Fig. 2.26 CO line driver with a 2 � VDD,c
0V
0V(Low)
0V
2xVDD,c 2xVDD,c
1xVDD,c
2xVDD,c
2xVDD,c
1xVDD,c
2xVDD,c(High)
1xVDD,c
1xVDD,c
1xVDD,c
1xVDD,c
1xVDD,c
1xVDD,c
M1
M2
M3
M4
M1
M2
M3
M4
Fig. 2.27 The internal node voltages when the output stage is delivering high and low-state
outputs showing all voltage differences are within 1 � VDD,c
32 2 General Considerations of High-/Mixed-VDD Analog. . .
a dedicated bias circuit for each cascode transistor is necessary to guarantee no
device is under overstress in both steady-state and transient operations. The circuit
needs two supplies 1 � VDD,c and 2 � VDD,c. The application-related performance
metrics are available elsewhere [13].
This 2 � VDD line driver has been recently modified to work on a 90-nm CMOS
switched-capacitor power amplifier [14] and 32-nm CMOS class-D power amplifier
[15]; both demonstrate state-of-the-art performances, further proving the impor-
tance of high-/mixed-VDD designs in nanoscale CMOS.
2.6 Summary
High-/mixed-voltage techniques feature high potential to boost up the
performances of RF and analog circuits without degrading the reliability. Bringing
the VDD,IO and thick-oxide transistors into the RF and analog circuit design portfo-
lio does not by itself require any add-on resource or technology option (at least up to
now), but it effectively increases the design flexibility. Circuit techniques play a
key role in this development, and are therefore long-term reusable when the
technology continues to advance.
This chapter only serves as a glimpse of this research trend and guiding direc-
tion, while highlighting the necessary gate-drain-source engineering skills to take a
broader advantage of available techniques. One of the critical points would be to
guarantee the circuit reliability compliance with the foundry guidelines when
considering device size and the potential adopted bias in transient and steady states.
Advantages of high-/mixed-voltage analog and RF CMOS circuits have been
demonstrated by several recent works, and are easily extendable to other
applications.
References
1. A.-J. Annema, B. Nauta, R. V. Langevelde and H. Tuinhout, “Analog Circuits in Ultra-
Deep-Submicron CMOS,” IEEE J. Solid-State Circuits (JSSC), vol. 40, pp. 132–143, Jan.,2005.
2. J. Borremans, G. Mandal, V. Giannini, T. Sano, M Ingels, B. Verbruggenn and J. Craninckx
“A 40nm CMOS Highly Linear 0.4-to-6GHz Receiver Resilient to 0dBm Out-of-Band
Blockers,” ISSCC Dig. Tech. Papers, pp. 62–63, Feb. 2011.3. STMicroelectronics technology profile in Circuits Multi-Projet (R) [Online]: http://cmp.imag.
fr/products/ic/
4. P.-I. Mak, S.-P. U and R. P. Martins, “Transceiver Architecture Selection – Review, State-
of-the-Art Survey and Case Study,” IEEE Circuits and Systems Magazine, Issue 2, pp. 6–25,Jun. 2007.
5. B. Serneels and M. Steyaert, Design of High voltage xDSL Line Drivers in Standard CMOS,Springer, 2008.
References 33
6. M. Zargari, L. Nathawad, H. Samavati, et al., “A Dual-Band CMOS MIMO Radio SoC for
IEEE 802.11n Wireless LAN,” IEEE J. Solid-State Circuits (JSSC), vol. 43, pp. 2882–2895,Dec., 2008.
7. R. Bagheri, A. Mirzaei, S. Chehrazi, M. E. Heidari, M. Lee, M. Mikhemar, W. Tang, and
A. A. Abidi, “An 800-MHz–6-GHz software-defined wireless receiver in 90-nm CMOS,”
IEEE J. Solid-State Circuits (JSSC), vol. 41, no. 12, pp. 2860–2876, Dec. 2006.8. C. Andrews, A. Molnar, “A Passive-Mixer-First Receiver with Baseband-Controlled RF
Impedance Matching, <6dB NF, and > 27dBm Wideband IIP3,” ISSCC Dig. Tech. Papers,pp. 46–47, Feb. 2010.
9. H. Moon, S. Lee, S-C. Heo, H. Yu, J. Yu, J-S. Chang, S-I. Choi, B-H. Park, “A 23mW Fully
Integrated GPS Receiver with Robust Interferer Rejection in 65nm CMOS,” ISSCC Dig. Tech.Papers, pp. 68–69, Feb. 2010.
10. L. Miao, P.-I. Mak, Z. Yan and R. P. Martins, “A High-Voltage-Enabled Recycling Folded
Cascode OpAmp for Nanoscale CMOS Technologies,” IEEE International Symposium onCircuits and Systems (ISCAS), pp. 33–36, May 2011.
11. R. S. Assaad, J. S. Martinez, “The Recycling Folded Cascode: A General Enhancement of the
Folded Cascode Amplifier,” IEEE J. Solid-State Circuits, vol. 44, no. 9, Sep. 2009.12. K. Ishida, A. Tamtrakarn and T. Sakurai, “An Outside-Rail Opamp Design Targeting for
Future Scaled Transistors,” IEEE Asian Solid-State Circuits Conference (A-SSCC), pp.73–76,Nov. 2005.
13. B. Serneels, M. Steyaert and W. Dehaene, “A 237 mW aDSL2+ CO Line Driver in a Standard
1.2V 130nm CMOS Technology,” ISSCC Dig. Tech. Papers, pp. 524–525, Feb. 2007.14. S.-M. Yoo. J.S. Walling, E. C. Woo and D.J. Allstot, “A Switched-Capacitor Power Amplifier
for EER/Polar Transmitters,” ISSCC Dig. Tech. Papers, pp.428–430, Feb. 2011.15. H. Xu, Y. Palaskas, A. Ravi, M. Sajadieh, M. A. El-Tanani and K. Soumyanath, “A Flip-
Chip-Packaged 25.3 dBm Class-D Outphasing Power Amplifier in 32 nm CMOS for WLAN
Application,” IEEE J. Solid-State Circuits (JSSC), vol. 46, no. 7, pp. 1596–1605, Jul. 2011.
34 2 General Considerations of High-/Mixed-VDD Analog. . .
Chapter 3
A Full-Band Mobile-TV LNA
with Mixed-Voltage ESD Protection
in 90-nm CMOS
3.1 Introduction
This chapter presents a number of circuit techniques enforced in the design of an
electrostatic discharge (ESD)-protected ultra-wideband (UWB) low-noise amplifier
(LNA) for mobile-TV applications. Unlike the design of narrowband LNAs, con-
current reception over a wide range of spectrum necessitates the LNA to feature
high linearity, preventing desensitization by the high-power blockers. This require-
ment, in conjunction with the obvious design goals of ESD protected input, low
noise figure (NF), low power, impedance match and high gain, constitute hard
tradeoffs to obtain a sensible balance and good compromise among all. The
proposed LNA is to cover the full band of mobile-TV services from 170 to
1,700 MHz such that only one LNA is necessary to support multiple standards.
It features a PMOS-based open-source input structure to optimize the I/O swings
under a mixed-voltage ESD protection while offering an inductorless broadband
input impedance match. The amplification core exploiting double current reuse andsingle-stage thermal-noise cancellation enhances the gain and noise performances
with high power efficiency. Optimized in a 90-nm 1.2/2.5-V CMOS process with
practical issues taken into account, the LNA using a constant-gm bias circuit
achieves competitive and robust performances over process, voltage and tempera-
ture (PVT) variation. The simulated voltage gain is 20.6 dB, noise figure is
2.4–2.7 dB and IIP3 is +10.8 dBm. The power consumption is 9.6 mW at 1.2 V.
│S11│ < –10 dB is achieved up to 1.9 GHz without needing any external resonant
network. Human Body Model ESD zapping tests of �4 kV at the input pins cause
no failure of any device.
P.-I. Mak and R.P. Martins, High-/Mixed-Voltage Analog and RF Circuit Techniquesfor Nanoscale CMOS, Analog Circuits and Signal Processing,
DOI 10.1007/978-1-4419-9539-1_3, # Springer Science+Business Media New York 2012
35
3.2 Circuit Description
3.2.1 Full-Band Mobile-TV Tuner Architecture
Figure 3.1 shows the block diagram of the full-band mobile-TV tuner where the
proposed LNA is inserted. The architecture is based on a direct-conversion receiver
that is differential to avoid common-mode pickups and even-order distortion [1]. As
addressed in many existing TV tuners (e.g., [2–5]), the baluns required for
generating the differential inputs in this frequency range are bulky in size, and
should be placed off chip. Wideband baluns having a center-tapped secondary can
be employed. The typical insertion losses of a 4.5–2,000-MHz 1:1 SMD balun are
0.32 and 2 dB in 4.5–1,000 MHz and 1–2 GHz, respectively [6].
Wideband reception suffers from the problem of harmonic mixings, i.e., in-band
blockers located at the harmonics of the local oscillator will become the co-channel
interferers. A 60-dB harmonic rejection ratio is expected. To achieve this, separated
RF filters in conjunction with a polyphase I/Q-mixer scheme are employed to doubly
reject the harmonic-mixing products. The polyphase I/Q-mixer scheme includes
three transconductance amplifiers having a gain ratio of 1:√2:1 to drive the passive
current-mode mixers [7]. A voltage-to-voltage LNA is therefore chosen to drive
those transconductance amplifiers that feature purely capacitive input impedance.
3.2.2 PMOS-Based Open-Source Input Structureand Mixed-Voltage ESD Protection
Figure 3.2 depicts the schematic of the proposed LNA. In order to save voltage
headroom and reduce the number of active devices, the cascode structure was not
applied. The main consequence of not using a cascode structure is a limited
470-860MHz
174-245MHz
1450-1700MHz
Balun
Iout+
Iout -
1
-45°
0°
45°
To Q-channel
LNA
This Work
1
Switches
Polyphase I/Q-Mixer SchemeOff-Chip
√2
Fig. 3.1 Full band mobile-TV tuner using a single wideband balun and LNA
36 3 A Full-Band Mobile-TV LNA with Mixed-Voltage ESD Protection. . .
reverse isolation. The results, however, show that with proper sizing the reverse
isolation is acceptable in the targeted frequency range. No matching network is
required as the source node of a nanoscale transistor offers a simple broadband
input impedance match.
The 1:1 balun equally splits the RF signal Vrf toM1 andM2 with opposite phases,
where reverse-biased P+-diffusion diode DP, and N+-diffusion diode DN, are
adopted for pin-to-rail ESD clamp. ESD-protection rail-clamp circuits incorporated
with DP and DN offer low-ohmic discharge paths among the I/O supply voltage –
VDD,I/O, the core supply voltage – VDD, and the common ground – GND. The aim of
choosing a PMOS-based input structure instead of NMOS is illustrated by
VDD
VDD,I/O
1:1Balun
Vrf
IbiasIbias
ESDProtectionRail-Clamp
Circuit
GND
On-Chip
Off-Chip
Package
RS
DP
DN
C2
C4
C1
C3
R1 R2
R5 R6
M1 M2
M3
R3M4
R4
Vout−Vout+
Vb1 Vb1
Vb2Vb2
0.5:
1
0.5:
1
Fig. 3.2 Schematic of the proposed LNA
3.2 Circuit Description 37
comparing the different bias conditions of the LNA and its pin-to-rail ESD clamp,
as follows:
In case NMOS is selected as the input device together with an input common-
mode voltage VCM,IN set to 0.6 V (Fig. 3.3a), the input swing is optimum as it is
midway the rail-clamp supply that is the VDD. Regrettably, VCM,IN ¼ 0.6 V will
offset the output common-mode level by the same amount, resulting in a limited
output swing. Though VCM,IN ¼ 0 V can resolve such an output swing limitation
(Fig. 3.3b), DN at such a VCM,IN is at a higher risk of forward bias, unavoidably
sacrificing part of the input swing.
Conventionally, the above tradeoff cannot be resolved by using PMOS
devices as shown in Fig. 3.4a. In this work, we propose to use VDD,I/O as the driving
voltage for the pin-to-rail ESD clamp. VDD,I/O is commonly available in modern
dual-oxide CMOS processes necessary for input–output (I/O) interfaces. Given that
VDD,I/O ¼ 2.5 V in the employed technology, the I/O swings can be concurrently
maximized, as shown in Fig. 3.4b. Besides that, considering the ESD robustness,
VCM,IN ¼ VDD balances the discharge capability of � zapping events.
The improvement of the proposed input structure in the employed 90-nm CMOS
technology is illustrated by plotting the diode-clamp I/O transfer curves and their
1.2 V
1.2 V
0.6 V+Vsat,n
Load
1.2 V
1.2 V
VCM,IN = 0VVCM,IN = 0.6 V
VinVin
DN
DP
DN
DP Vsat,n
VoutVout
Load
a b
Fig. 3.3 Diode clamps with a NMOS-input structure: (a) VCM,IN ¼ 0.6 V and (b) VCM,IN ¼ 0 V
1.2 V
Load
2.5 V (I/O Supply)
1.2 V- Vsat,p
0.6 V- Vsat,p
Load
VCM,IN = 1.2 VVCM,IN = 0.6 V
Vin
Vout
DN
DP
Vin
Vout
DN
DP
a b
Fig. 3.4 Diode clamps with a PMOS-input structure: (a) VCM,IN ¼ 0.6 V and (b) VCM,IN ¼ 1.2 V
but the clamping voltage is the VDD,I/O ¼ 2.5 V
38 3 A Full-Band Mobile-TV LNA with Mixed-Voltage ESD Protection. . .
first derivatives at typical and extreme temperatures (Fig. 3.5). With VCM,IN ¼ VDD
¼ 1.2 V, the linear input [vsig(t)], even at the highest temperature, has a swing of
roughly 3 Vpp in the PMOS case, while it is just roughly 0.8 Vpp in NMOS case with
VCM,IN ¼ GND (0 V).
3.2.3 Double Current Reuse for gm-Enhancement
Current reuse is power-efficient in gain enhancement. In this work, current is
doubly reused to boost the transconductance (gm). A simplified small-signal half-
circuit equivalent of Fig. 3.2 is shown in Fig. 3.6. Firstly, M1 gate-source terminals
are coupled with a gain of �1 (i.e., capacitive cross-coupling [8]) such that the
transconductance of M1, gm1, is enhanced to g0m1 as given by,
g 0m1 ¼ ð1þ AxÞgm1; (3.1)
where Ax stands for the voltage transfer of the highpass network: C3 and R1. The
capacitance division between M10s CGS and C3 gives Ax < 1.
The capacitive cross-coupling reuse in the NMOS device, M3, to further
enhances the overall transconductance Gm as given by,
Gm ¼ g 0m1 þ Aygm3; (3.2)
-2
-1
0
1
2
3
4
-2 -1 0 1 2 3 4 5Input Voltage, Vin (V)
Out
put V
olta
ge, V
out (
V)
1st derivativeof I/O curves
I/O curves
125°C
−55°C27°C
2.5 V
Vin = 0 V + Vsig
Vin = 1.2 V + Vsig
DN
DP
Vin Vout
VCM,IN = 1.2 V
VCM,IN = 0 V
Fig. 3.5 Diode-clamp I/O transfer curves and their first derivatives at different temperatures,
showing the linear input swing versus the selected VCM,IN
3.2 Circuit Description 39
where Ay is the voltage transfer of the highpass network:C1 and R3. R5 is a grounded
resistor. R5 and COUT in parallel with the output resistance of M3 (i.e., ro3), andthe resistance looking into the drain node of M1 [i.e., Rout,M1 ¼ ro1 + RS +
gm1ro1(Ax + 1)RS,] will lead to the total output impedance ZOUT,
ZOUT ¼ R5== ro1 þ RS þ gm1ro1ðAx þ 1ÞRS½ �==ro3== 1
joCOUT
: (3.3)
Grounded R5 implemented with a high-resistive polysilicon is to linearize ZOUTand optimize the output common-mode voltage. Since nanoscale transistors suffer
from strong channel-length modulation, the conventional 1/gm approximation of the
LNA’s source-node input impedance leads to poor matching between the calculated
and simulated results. An accurate expression of the resistive input impedance RIN
has to take into account all finite output impedances, i.e., ro1 of M1 and ro3 of M3,
yielding,
RIN ¼R5==ro3==
1joCOUT
þ ro1
1þ g 0m1ro1
: (3.4)
-1
-1
1:1Balun
Vrf
RS
0.5:
1
0.5:
1
RIN
ZIN
ZOUT
COUT
CIN
Ay
Ax
Vout+
VDD
R1Vb1
C3
C1
R3Vb2
M3 R5
M1
Fig. 3.6 Single-ended
small-signal equivalent
circuit of the proposed LNA
40 3 A Full-Band Mobile-TV LNA with Mixed-Voltage ESD Protection. . .
RIN in parallel with the input parasitic capacitance CIN leads to the input
impedance ZIN of the LNA given by,
ZIN ¼ RIN==1
joCIN
: (3.5)
After all, the differential voltage gain of the LNA, Av,diff, can be obtained as,
Av;diff ¼ Voutþ � Vout�Vrf
¼ 2ZINRS þ 2ZIN
GmZOUT: (3.6)
3.2.4 Single-Stage Thermal-Noise Cancellation
Noise-cancelling LNAs using a cascading configuration have been reported, e.g.,
[9, 10]. However, additional amplification stages lead to higher NF and poor
linearity. In this work, thermal-noise cancellation can be achieved in a single
stage by taking advantages from the bi-directional coupling behavior of the
balun, and the structure of the LNA after double-current reuse. Re-examining
Fig. 3.2 we can observe that the thermal noise of M1 (which can be modeled as a
noise current source connecting its drain and source) is partly injected to Vout+, and
partly negatively couples to the source of M2 through the balun, and to the gate of
M2 (M4) through the cross-coupling network R2-C4 (R4-C2). An identical noise
coupling operation occurs around M2. Differentially, certain noise transfer paths
will be out phased from the others, yielding a way to cancel out the noise ofM1 with
virtually no cost. Based on RIN (but not ZIN to simplify the noise analysis), it can be
shown that the noise factor F of the LNA is given by,
F ¼ 1þRSRIN
RS þ 2RINGm � 1
� �2gm1
g1a1
þ gm3g3a3
þ 1
R5
2RIN
RS þ 2RIN
� �24RIN
RS þ 2RIN
� �2Gm
2 RS
2;
(3.7)
where a1(a3) and g1(g3) are the process- and bias-dependent parameters ofM1 (M3),
respectively. Principally, the noise generated by M1 (M2) can be minimized by
designing,
RSRIN
RS þ 2RINGm ¼ RSRIN
RS þ 2RINg 0m1 þ Aygm3
� � ! 1: (3.8)
Regrettably, since (3.6) and (3.7) are inter-dependent by most parameters, there
is no straight way to minimize the overall system NF (NF ¼ 10·log F) since Av,diff
also affects the input- referred noise contribution of the succeeding circuits.
3.2 Circuit Description 41
In order to co-optimize the key performance metrics, a constraint-based
semi-computed design flow is applied. Given a power budget of 10 mW and a
�3-dB output bandwidth of >1.7 GHz when driving a 0.3-pF load (the input
capacitance of the polyphase I/Q-mixer scheme), the size of M1 and R5 are picked
to fulfill the required RIN, which must be close to RS/2 for an acceptable S11(<�10 dB). R5 is relatively low (250 O) to desensitize ro3 in Eq. 3.5 such that,
without affecting the input match, tuning the size ofM3 can trade the Av,diff, NF and
linearity. Simulation results are plotted in Fig. 3.7a and b, where the size multiplier
(N) of M3 is swept from 1 to 9. Selecting N ¼ 5 yields the highest input-referred
third-order intercept point (IIP3) since the third-order derivative of the DC charac-
teristic is close to zero at this N. The corresponding Av,diff is over 20 dB and NF is
2.5 dB. The �3-dB output bandwidth (1.84 GHz), │S11│ (�16.3 dB) and power
budget (9.6 mW) are satisfied.
-25
-20
-15
-10
-5
0
5
10
15
20
25
1 3 5 7 92
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
3
Av,diff
Av,
diff
(dB
)
Performance @ 1GHz
NF
(dB
)
NF
|S11|
|S11
| (dB
)
-17
-16.9
-16.8
-16.7
-16.6
-16.5
-16.4
-16.3
-16.2
-16.1
-16
1 3 5 7 97
7.4
7.8
8.2
8.6
9
9.4
9.8
10.2
10.6
11
IIP3
Power
N, size mutiplier of M3 (M4)
N, size mutiplier of M3 (M4)
Size of M3 (M4) = 10µm/0.16µm x N
Size of M3 (M4) = 10µm/0.16µm x N
Performance @ 1GHz
IIP3
(dB
m)
2
1.96
1.92
1.88
1.84
1.80
1.76
1.72
1.68
1.64
1.6O
utpu
t -3d
B B
andw
idth
(G
Hz)
Output -3dBBandwidth
8
8.5
9
9.5
10
10.5
11
11.5
12
Pow
er (
mW
)
a
b
Fig. 3.7 Size of M3(M4) versus RF performances: (a) │S11│, Av,diff and NF. (b) power consump-
tion, IIP3 and output �3-dB bandwidth
42 3 A Full-Band Mobile-TV LNA with Mixed-Voltage ESD Protection. . .
3.3 Key Practical Design Issues
3.3.1 Package Effect on Input Impedance Match
Taking into account the package effect (Fig. 3.8), the single-ended (left-half) input
impedance Z’IN of the LNA is given by,
Z 0IN ¼ 1
joCOC
== joLBW þ RBW þ RIN==1
joCIN
� �� �: (3.9)
where COC denotes the parasitic capacitance of the lead-frame and soldering pad on
the testing board. LBW and RBW stand for the inductance and resistance of the
bondwire, respectively. CIN is the total input capacitance resulting fromM1, DN and
DP, the parasitic capacitance of C1 and C3, and the bondpad (CPAD).
Similar to the analysis addressed in [11], targeting an in-band │S11│ < �10 dB
requires an input reflection coefficient magnitude │GIN│ to satisfy,
GINj j ¼ Z 0IN � RS=2
Z 0IN þ RS=2
��������� 0:32: (3.10)
With RS ¼ 50 O and a balun ratio of 1:1, RIN ¼ 25 O achieves a superior │S11│.However, the matched bandwidth is insufficient since CIN will lower ZIN when
frequency is increasing. Thus, as shown in Fig. 3.8 as well, the optimal RIN that can
result in a broaden bandwidth should be a higher value in practice. Here, with a
-35
-30
-25
-20
-15
-10
-5
0
0.1 101Frequency (GHz)
Equivalent Circuit
Vrf/2
RS/2 RBW
COC
COC = 2.7 pF
CIN = 2 pF
LBW
Z’IN
CINRINRIN = 25 Ω
RBW = 1 ΩLBW = 3 nH
RIN = 30 Ω
RIN = 35 Ω
RIN = 40 Ω
ZN
|S11
| (dB
)
Fig. 3.8 Input matching design including package effects
3.3 Key Practical Design Issues 43
value of RIN between 35 and 40 O it will lead to an adequate │S11│ over the desiredbandwidth, avoiding any external resonant network.
Since the exact value of LBW is uncertain in practice, the pickedRINmust guarantee
a │S11│ < �10 dB over an acceptable range of inductance variation [12]. As shown in
Fig. 3.9, the tolerable LBW variation range is from 1.5 to 3.5 nH for RIN ¼ 40 O.
3.3.2 Self-Startup Constant-gm Bias Circuit
To tackle the PVT variation, a self-startup constant-gm bias circuit is adopted. As
depicted in Fig. 3.10, an off-chip resistor RREF serves as the reference, such that gm,Mb2 can be fixed to the inverse of RREF by equalizing the dc bias current ofMb1 and
Mb2 (they exhibit a ratio of 4:1 and are both long-channel devices) [13]. The
generated Vb1 is the bias voltage of the LNA’s M1 and M2. RREF is scaled up by
10� to reduce the power consumption. Thus, the size ratio of Mb2 to M1 is set to
1:10 for a correct transconductance ratio.
The self-bias structure requires a start-up circuit. Diode-connected Mb7
guarantees that the bias circuit can be started up at power-on by satisfying,
VT;Mb5 þ VT;Mb3 þ VT;Mb7 þ jVT;Mb2j<VDD; (3.11)
and shut down afterwards by satisfying,
VGS;Mb5 þ VGS;Mb3 þ VT;Mb7 þ jVGS;Mb2j>VDD; (3.12)
where VT is the threshold voltage of the transistor.
-35
-30
-25
-20
-15
-10
-5
0
0.1 1 10Frequency (GHz)
1.5nH
2nH
2.5nH
3nH
3.5nH
In-band
|S11
| (dB
)
Fig. 3.9 Input matching versus the inductance (LBW) variation of the bondwire
44 3 A Full-Band Mobile-TV LNA with Mixed-Voltage ESD Protection. . .
3.3.3 Power-Up/Down Control with Reliability Concern
Nanoscale circuits for mobile TV require particular attention of reliability in power-
down condition since DVB-H uses time-slicing operation. In the employed 90-nm
CMOS process, the proposed LNA operating at 1.2 V is free from hot carrier
injection (HCI), time-dependent dielectric breakdown (TDDB) and absolute maxi-
mum rating (AMR). The primary concern is the negative bias temperature instabil-
ity (NBTI) of all PMOS devices. To prevent large VGS or VGD in power-down mode,
Mb1 and Mb2 in the self-startup constant-gm bias circuit (Fig. 3.10) are pulled up to
VDD. In this way, the NBTI stress is transferred to the pull-up switches, Mpu1 and
Mpu2, which show no impact to the LNA in active mode. This technique simulta-
neously pulls up Vb1 to VDD, guaranteeing M1 and M2 of the LNA are also shut
down in the same manner.
3.3.4 Mixed-Voltage ESD Protection Scheme
Protecting the nanoscale thin-oxide devices from ESD events requires well-
designed discharge paths to the supply rails. Figure 3.11 shows the designed
mixed-voltage ESD clamps. Power clamps based on P+/N-well diode chains
efficiently realize a sufficient high trigger voltage that is greater than the supply,
such that the leakage current and the chance of accidental latching due to normal
supply fluctuation are minimized. For instance, with a silicon diode threshold
Mb1
Vb1
VDD
Mb4
Mb2
Mb3
Mb7
Mb6Mb5
Mpu1 Mpu2
RREF(off-chip)
IbIb
4 : 1
Start Up PowerUp/Down
gm,Mb2 =1/RREF
Fig. 3.10 Self-startup constant-gm bias circuit (with power up/down control)
3.3 Key Practical Design Issues 45
voltage of ~0.65 V, the VDD,I/O-GND power clamp needs five diodes in series,
whereas for VDD-GND only three are necessary. The selected number of diodes in
the VDD,I/O-VDD power clamp have had the precaution of different supply start-up
sequences. For instance, VDD,I/O may start before VDD. The proposed scheme is
optimized to avoid the happening of any forward-bias current in all supply start-up
sequences.
The dimension of DP (DN) is 1 mm/50 mm � 10. Since the technology
determines that the parasitic capacitances resulting from DP and DN per unit area
are CDP ¼ 0.9 fF/mm2 and CDN ¼ 0.74 fF/mm2, respectively, the imposed total
parasitic capacitance at the input is ~870 fF, which occupies 44% of the total CIN
budget. The rest of the CIN budget can be reserved for the CPAD (~300 fF), the input
parasitic capacitance of M1 and the parasitic capacitances of C2, C4 and C5, which
are all metal-over-metal (MoM) capacitors.
3.4 Simulation Results, Discussions and Benchmarks
The LNA has been extensively characterized in a CadenceTM environment with
SPECTRE as the simulator. The devices dimensions are listed in Table 3.1. The
package effects and parasitic capacitances at the input and output nodes have been
modeled in all simulations. An extra load capacitor of 0.3 pF is added to account for
the input capacitance of the succeeding polyphase I/Q-mixer scheme. The ESD
robustness of the RF-input pins to rail is tested using the Human Body Model
(HBM) reference circuit [14]. A�HBM voltage pulse is applied to the LNA’s input
GND
I/P
VDD,I/O-GNDPower Clamp
VDD,I/O-VDDPower Clamp
VDD,I/O-GNDPower Clamp
LNA
+ zaps
− zaps
N+ diff. in P-wellDiode
P+ diff. in N-wellDiode
CDNxArea
CDPxArea
VDD,I/O (2.5 V) VDD (1.2 V)
VDC=1.2 V
DP
DN
Fig. 3.11 Mixed-voltage ESD-protection scheme
46 3 A Full-Band Mobile-TV LNA with Mixed-Voltage ESD Protection. . .
to induce a large +/– zapping discharge current that has a rising/falling time
of ~8 ns. Verified in all combinations, the LNA can withstand minimally �4 kV
of ESD zapping without causing internal or protection devices failure. This result
fulfills the standard of “safe level” (i.e., �4 kV) in the chip-level ESD
specifications.
It would be interesting to compare the effectiveness of the double-current reuse
technique to the simple capacitive cross-coupling. As shown in Fig. 3.12, with M3
serving as an amplification device, there are roughly 1-dB or 3-dB improvements in
terms of NF or Av,diff, respectively.
Multi-band reception of mobile-TV standards does not pose rigid group-delay
variation specification as the BW usage of each standard is relatively small when
comparing it with that of the LNA. The simulated inband gain delay is 233 � 21 ps.
The in-band linearity is verified by two-tone tests as shown in Fig. 3.13. Due to the
cascode-free structure of the LNA and optimal gate-source voltage biasing, high IIP3
of +10.8 dBm and �1-dB input-referred compression point (ICP) of �7.6 dBm are
Table 3.1 Devices dimensions
Device Size Device Size
M1, M2 (10 mm/0.08 mm) � 10 RREF 400 OM3, M4 (10 mm/0.08 mm) � 5 Mb1 (10 mm/0.08 mm) � 4
R1–R4 1 mm � 40 mm ¼ 40 kO Mb2 (10 mm/0.08 mm) � 1
C1–C4 1 mm � 40 mm ¼ 2.7 pF Mb3–Mb7 (1 mm/0.32 mm) � 8
R5, R6 10 mm � 3 mm ¼ 250 O DP, DN (1 mm/50 mm) � 10
-50
-40
-30
-20
-10
0
10
20
30
0
5
10
15
20
25
30
35
40Av,diff
0.01 0.1 1 10
Frequency (GHz)
NF
M3 as a current source
M3 as an amplifier
Av,
diff
and
|S11
| (dB
)
NF
(dB
)
|S11|
Fig. 3.12 RF performance comparison: M3 used as a current source (i.e., only capacitive cross-
coupling of M1) or M3 used as an amplifier (double-current reuse)
3.4 Simulation Results, Discussions and Benchmarks 47
concurrently achieved. Other IIP3’s among the desired signal band varies between
+10.5 and +13.7 dBm. Out-of-band two-tone test at 2.0 and 2.1 GHz is also conducted
(not shown), obtaining an out-of-band IIP3 of +9.9 dBm and a �1-dB ICP of
�3.5 dBm. Both are superior linearity performance metrics in this 1.2-V 9.6-mW
LNA design. The power-down leakage is 47 mW.
The effects of differential imbalance to the performances of the LNA are
simulated. Ten percent size mismatch between M1 and M2 still maintain the
in-band common-mode and supply rejection ratios >50 dB. Since the LNA is
intended to drive an integrated mixer, testing the reverse isolation S12, which
needs a 50-O test buffer, will not allow the determination of the stability infor-
mation of the LNA in its true operating condition. Thus, only the reverse voltage
gain (i.e., the differential outputs coupled back to the input node) and transient
simulation results can help to estimate the stability of the LNA. The simulated in-
band reverse voltage gain is < �28 dB. The tolerable CL is up to 1 pF without
affecting the targeted │S11│ of �10 dB. These tests inspire confidence about the
stability of the LNA while in practice.
As the impact of transistor variability in a nanoscale process is continuously
increasing, process corner and Monte-Carlo (MC) simulations are essential to
investigate the robustness of the RF performances over PVT. A set of simulations
counting the process (Fig. 3.14) supply voltage (Fig. 3.15) and temperature
(Fig. 3.16) variations have been conducted. In addition, a set of 100-time MC
simulation of the RF performances was also performed (Fig. 3.17).
-80
-60
-40
-20
0
-20 -10 0 10 20
Out
put S
igna
l Pow
er (
dBm
)
Input Signal Power (dBm)
20
40
60
IIP3 = +10.8 dBm
9
10
11
12
13
14
0.2 0.7 1.2 1.7Frequency (GHz)
IIIP
3(dB
m)
15
ICP = -7.6 dBm
Fig. 3.13 Two-tone tests at 695 and 700 MHz
48 3 A Full-Band Mobile-TV LNA with Mixed-Voltage ESD Protection. . .
16
16.5
17
17.5
18
18.5
19
19.5
20
20.5
21
0.1 0.5 0.9 1.3 1.7 2.1
Frequency (GHz)
SS
FFSNFP
FNSP
-40
-35
-30
-25
-20
-15
-10
-5
0
0.1 0.5 0.9 1.3 1.7 2.1
Frequency (GHz)
SS
FF
SNFP
FNSP
2.3
2.4
2.5
2.6
2.7
2.8
2.9
3
0.1 0.5 0.9 1.3 1.7 2.1
SS FFSNFPFNSP
Frequency (GHz)
|S11
| (dB
)N
F (
dB)
Av,
diff
(dB
)
a
b
c
Fig. 3.14 RF performance in process corners: FAST-FAST (FF), SLOW-SLOW (SS), FAST-
NMOS-SLOW-PMOS (FNSP) and SLOW-NMOS-FAST-PMOS (SNFP). (a) Av,diff (b) |S11|
and (c) NF
3.4 Simulation Results, Discussions and Benchmarks 49
These pre-silicon results rigidly verify the effectiveness of the proposed circuit
techniques and the completeness of the design consideration. The LNA will be
experimentally verified with the entire TV tuner.
Table 3.2 gives a comparison of the proposed LNA with respect to the recently
published wideband CMOS LNAs [15–32] (design with both simulation and
0
5
10
15
20
25
0.01 0.1 1 100
2
4
6
8
10
-40
-32
-24
-16
-8
0 1.4V1.2V1.0V
Power Consumption6.3mW @1.0V9.6mW @1.2V
Frequency ( GHz)
NF
1.4V1.2V1.0V
1.4V1.2V1.0V
Av,diff
Av,
diff
(dB
)
|S11|
|S11
| (dB
)
NF
(dB
)
Fig. 3.15 RF performances versus supply voltage variation
NF
Power Consumption8.5mW @-55°C9.6mW @ 27°C
11.5mW @125°C
0
5
10
15
20
25
0.01 0.1 1 100
2
4
6
8
10
-40
-32
-24
-16
-8
0
Frequency ( GHz)
Av,
diff
(dB
)
|S11
| (dB
)
NF
(dB
)
-55°C27°C
125°C
-55°C27°C
125°C
-55°C27°C
125°C
|S11|
Av,diff
Fig. 3.16 RF performances versus temperature variation
50 3 A Full-Band Mobile-TV LNA with Mixed-Voltage ESD Protection. . .
measurement results are considered). It can be observed that most performance
metrics of the current design are superior, in particular the voltage gain, power and
linearity, even under high ESD protection.
3.5 Summary
This chapter has described the design of an ESD-protected UWB LNA covering
170–1,700 MHz for full-band mobile TV tuners in a 90-nm CMOS process. �4-kV
ESD robustness at the RF input pins and +10.8-dBm IIP3 have been concurrently
achieved by exploiting a 1.2/2.5-V-mixed ESD protection scheme and a PMOS-based
20.2 20.3 20.4 20.5 20.6
5101520253035
-16.5 -16 -13.5 -120
4
8
12
16
20
2.37 2.38 2.39 2.4 2.41 2.42 2.43 2.44 2.45 2.460
5
10
15
20
25
0
-14.5 -12.5-15.5 -15 -14 -13
20.25 20.35 20.45 20.55
Av,diff @1GHz (dB)
NF @ 1GHz (dB)
Hits
Hits
Hits
Mean = 20.38SD = 0.058
N = 100
Mean = -14.54SD = 0.928
N = 100
Mean = 2.416SD = 0.014
N = 100
|S11| @ 1GHz (dB)
Fig. 3.17 Hundred-time MC simulation results of the RF performances
3.5 Summary 51
Table
3.2
Brief
perform
ance
benchmarkofrecentlypublished
widebandLNAs
Ref
–Year
Technology
Bandwidth
(GHz)
Gain(dB)
NF(dB)
IIP3(dBm)
Supply
(V)
Power
(mW)
No.ofinductors
ESD
protection
[15]TCAS-I’09
S90nm
CMOS
0.1
~1.89
20.6
VG
<2.7
+10.8
@0.7
GHz
1.2
*9.6
0�
4kV
[16]JSSC’08
M65nm
CMOS
0.2
~5.2
13–15.6
VG
<3.5
>0
1.2
21
0No
[17]JSSC’08
M130nm
CMOS
0.8
~2.1
14.5
VG
2.6
+16
1.5
17.4
0No
[18]ASSCC’07
M180nm
CMOS
0.05–0.86
15
VG
2.5
+8.3
1.8
7.2
0No
[19]RFIC’07
m90nm
CMOS
0.4
~1
16
PG
<5.3
�17@
1GHz
1.2
16.8
0No
[20]TCAS-II’07
S130nm
CMOS
0.2
~3.8
11.2
VG
<2.85
�2.7
@3GHz
1.2
1.9
0No
[20]TCAS-II’07
S130nm
CMOS
0.2
~6.2
10.5
VG
<2.85
�2.7
@3GHz
1.2
1.9
2No
[21]TCAS-I’07
M180nm
CMOS
2.8
~7.2
19.1
PG
<3.8
�1@
6GHz
1.8
32
7No
[22]PRIM
E’06
S130nm
CMOS
0.9
~2.5
17
PG
2�5
1.2
15.6
4No
[23]RFIC’05
M130nm
CMOS
0.1
~0.93
13
PG
4�1
0.2
1.2
0.72
0No
[24]CICC’05
M130nm
CMOS
0.1
~6.5
19
PG
<4.2
+1
1.8
11.7
4No
[25]ESSCIRC’05
M130nm
CMOS
3~5
26
VG
4�1
31.5
45
4�
1.5
kV
[26]ISCAS’05
S130nm
CMOS
3~10.7
11
PG
3�8
.21.2
4.8
5No
[27]ASSCC’05
M180nm
CMOS
0.04~0.9
20.3
PG
4�1
0.8
~�1
2.7
1.8
43.2
0No
[28]ISCAS’05
S180nm
CMOS
1.5
~2.6
15.4
PG
0.9
�2.5
1.5
11
4No
[28]ISCAS’05
S180nm
CMOS
3.2
~4.8
17.9
PG
1.6
�4.5
1.5
13.2
4No
[29]JSSC’05
M180nm
CMOS
2~4.6
9.8
PG
<5.2
�71.8
12.6
3No
[30]TCAS-I’05
M250nm
CMOS
3.2
~4.8
7PG
<3.7
4@
4GHz
2.5
20
2No
[31]ISSCC’04
M180nm
CMOS
2.3
~9.2
9.3
PG
5.2
�6.7
1.8
95
No
[31]ISSCC’04
M180nm
CMOS
2.4
~9.5
10.4
PG
5.3
�8.8
1.8
95
No
[32]MWSCAS’03
S180nm
CMOS
3~7
15.3
PG
<1.9
N/A
1.8
15
4No
Ssimulationresults,M
measurementresults*:2.5
VforESDprotectiondiode,PGpower
gain,VGvoltagegain
52 3 A Full-Band Mobile-TV LNA with Mixed-Voltage ESD Protection. . .
open-source input structure. Reliability- and PVT-conscious bias techniques lead to
competitive and robust RF performances. The LNA core employed double current-
reuse and single-stage thermal-noise cancellation achieves 20.6-dB voltage gain and
<2.7-dB NF with 9.6-mW power consumption.
References
1. P.-I. Mak, S.-P. U and R. P. Martins, “Transceiver Architecture Selection – Review, State-of-
the-Art Survey and Case Study,” IEEE Circuits and Systems Magazine, Vol. 7, Issue 2,
pp. 6–25, June 2007.
2. Vassiliou et al., “A 65 nm CMOS Multistandard, Multiband TV Tuner for Mobile and
Multimedia Applications,” IEEE J. of Solid- State Circuits, vol. 43, no. 7, pp. 1522–1533,Jul. 2008.
3. P. Antoine, et al., “A Direct-Conversion Receiver for DVB-H,” in IEEE ISSCC, Digest ofTechnical Papers, pp. 426–427, Feb. 2005.
4. D. Saias, et al., “A 0.12mm CMOS DVB-T Tuner,” in IEEE ISSCC, Digest of TechnicalPapers, pp. 430–431, Feb. 2005.
5. I. Vassiliou et al., “A 0.18mm CMOS, Dual-Band, Direct-Conversion DVB-H Receiver,” in
IEEE ISSCC, Digest of Technical Papers, pp. 606–607, Feb. 2006.6. Falco Electronics SMD RF Balun, Available [online]: http://www.falcomex.com/products/3/
04-00.asp
7. I. V R. Bagheri, A. Mirzaei, S. Chehrazi, M. Heidari, M. Lee, M. Mikhemar, W. Tang and A.
Abidi, “An 800MHz to 5GHz Software-Defined Radio Receiver in 90nm CMOS,” IEEE Int.Solid-State Circuits Conf. (ISSCC) Dig. Tech. Papers, pp. 480–481, Feb. 2006.
8. W. Zhuo, S. Shekhar, S. Embabi, J. Gyvez, D. Allstot and E. Sanchez-Sinencio, “A Capacitor
Cross-Coupled Common-Gate Low-Noise Amplifier,” in IEEE Trans. On CAS-II: ExpressBriefs, vol. 52, no. 12, pp. 875–879, Dec. 2005.
9. C.-F. Liao and S.-I Liu, “A Broadband Noise-Canceling CMOS LNA for 3.1–10.6-GHz UWB
Receivers,” IEEE J. of Solid-State Circuits, vol. 42, no. 2, pp. 329–339, Feb. 2007.10. E. A. Klumperink, F. Bruccoleri, P. Stroet and B. Nauta, “Amplifiers Exploiting Thermal
Noise Canceling: A Review” in Proc. of Gallium Arsenide and other compound semiconductorApplication Symp. (GAAS), pp. 371–374, Oct. 2004.
11. K. Bhatia, S. Hyvonen and E. Rosenbaum, “A Compact, ESD- Protected, SiGe BiCMOS LNA
for Ultra-Wideband Applications,” IEEE J. Solid-State Circuits, vol. 42, no. 5, pp. 1121–1130,May 2007.
12. G. Banerjee, K. Soumyanath and D. J. Allstot, “Desensitized CMOS Low-Noise Amplifiers,”
IEEE Trans. On CAS-I: Regular Papers, vol. 55, no. 3, pp. 752–765, Apr. 2008.13. R. Zele and D. Allstot, “Low Power CMOS Continuous-Time Filters,” IEEE J. of Solid-State
Circuits, vol. 31, no. 2, pp. 157–168, Feb. 1996.14. ESD Sensitivity Testing: Human Body Model (HBM) – Component Level, ESD Association
Standards, 1993.
15. P.-I. Mak and R. P. Martins, “Design of an ESD-Protected Ultra-Wideband LNA in Nanoscale
CMOS for Full-Band Mobile TV Tuners,” IEEE Transactions on Circuits and Systems – I:Regular Papers, vol. 56, no. 5, pp. 933–942, May 2009.
16. S. Blaakmeer, E. Klumperink, D. Leenaerts and B. Nauta, “Wideband Balun-LNA With
Simultaneous Output Balancing, Noise-Canceling and Distortion-Canceling,” IEEE J. ofSolid-State Circuits, vol. 43, no. 6, pp. 1341–1350, Jun. 2008.
17. W.-H. Chen, G. Liu, B. Zdravko and A. M. Niknejad, “A Highly Linear Broadband CMOS
LNA Employing Noise and Distortion Cancellation,” IEEE J. of Solid-State Circuits, vol. 43,no. 5, pp. 1164–1176, May 2008.
References 53
18. Y. Liao, Z. Tang and H. Min, “A CMOS Wide-Band Low-Noise Amplifier with Balun-Based
Noise-Canceling Technique,” in Proc. IEEE A-SSCC, pp. 91–94, Nov. 2007.19. M. Vidojkovic, M. Sanduleanu, J. Tang, P.Baltus and A. Roermund, “A 1.2 V, Inductorless,
Broad-band LNA in 90 nm CMOS LP,” in Proc. IEEE RFIC symp., pp. 53–56, Jun. 2007.20. A. Amer, E. Hegazi and H. Ragai, “A Low-Power Wideband CMOS LNA for WiMax,” IEEE
Trans. On CAS-II: Express Briefs, vol. 54, no. 1, pp. 4–8, Jan. 2007.21. Y.-J. Emery Chen and Y.-I. Huang, “Development of Integrated Broad-Band CMOS
Low-Noise Amplifiers,” IEEE Trans. On CAS-I: Regular Papers, vol. 54, no. 10,
pp. 2120–2127, Oct. 2007.
22. B. Martineau, et al., “A Wideband LNA for Wireless Multi-standard Receiver in 130 nm SOI
Process”, in Proc. IEEE PRIME, pp. 449–452, Jun. 2006.23. S. Wang, A. Niknejad and R. Brodersen, “A Sub-mW 960-MHz Ultra-Wideband CMOS
LNA,” in Proc. IEEE RFIC symp., pp. 35–38, Jun. 2005.24. S. Chehrazi, A. Mirzaei, R. Bagheri, and A. Abidi, “A 6.5 GHz Wideband CMOS Low Noise
Amplifier for Multi-Band Use,” in Proc. IEEE CICC, pp. 801–804, Sept. 2005.25. R. Salerno, M. Tiebout, H. Paule, M. Streibl, C. Sandner and K. Kropf, “ESD-Protected CMOS
3–5GHz Wideband LNA+PGA Design for UWB,” in Proc. ESSCIRC, pp.219–222, Sept.2005.
26. R. Molavi, S. Mirabbasi, and M. Hashemi, “A Wideband CMOS LNA Design Approach,” in
Proc. IEEE ISCAS, pp. 5107–5110, May 2005.
27. D. R. Huang, et al., “A 40–900 MHz Broadband CMOS Differential LNA with Gain-Control
for DTV RF Tuner”, in Proc. IEEE A-SSCC, pp. 465–468, Nov. 2005.28. Y. Wang, J. S. Duster, and K. T. Kornegay, “Design of an Ultra- Wideband Low Noise
Amplifier in 0.13mm CMOS,” in Proc. IEEE ISCAS, pp. 5067–5070, May 2005.
29. C.-W. Kim, M. S. Kang, P. T. Anh, H. T. Kim, and S. G. Lee, “An Ultra-Wideband CMOS
Low Noise Amplifier for 3–5-GHz UWB System,” IEEE J. Solid-State Circuits, vol. 40, no. 2,pp. 544–547, Feb. 2005.
30. J. Lerdworatawee and W. Namgoong, “Wide-Band CMOS Cascode Low-Noise Amplifier
Design Based on Source Degeneration Topology,” IEEE Trans. On CAS-I: Regular Papers,vol. 52, no. 11, pp. 2327–2334, Nov. 2005.
31. A. Bevilacqua and A. Niknejad, “An Ultra-Wideband CMOS LNA for 3.1 to 10.6 GHz
Wireless Receivers,” in IEEE ISSCC Digest of Technical Papers, pp. 382–383, Feb. 2004.32. H. Doh, Y. Jeong, S. Jung, and Y. Joo, “Design of CMOS UWB Low Noise Amplifier with
Cascade Feedback,” in Proc. IEEE MWSCAS, pp. II-641–II-644, Jul. 2004.
54 3 A Full-Band Mobile-TV LNA with Mixed-Voltage ESD Protection. . .
Chapter 4
A High-Voltage-Enabled Mobile-TV RF
Front-End in 90-nm CMOS
4.1 Introduction
High-voltage (HV)-enabled circuits offer a feasible alternative to cope with the
sub-1 V technologies at low cost. This chapter describes the first 2 � VDD-enabled
mobile-TV RF front-end with TV-GSM interoperability. It is an on/off-chip co-
design employing externally three customized UHF/VHF preselect filters, a RF
switch and a balun. The integrated part includes: (1) a cascode-cascade inverter-
based low-noise amplifier that features a high gain-to-power efficiency; (2) a
linearized C-2C attenuator using reliable-overdriven MOS switches; (3) an
inductive-peaking feedforward path that evens out the passband variation; and (4)
two cascode I/Q mixer drivers capable to drive passive mixers with small gain and
bandwidth reduction. Gate-drain-source engineering and self-biased structures are
the keys enabling performance optimization with low power and no reliability risk.
Fabricated in a 90-nm CMOS process with 1-V thin-oxide devices, the RF front-end
measures 68-dB rejection at GSM-900 uplink, 0.7-dB passband roll-off, 3.9-dB
noise figure and �5.5-dBm IIP3 at a maximum voltage gain of 26.2 dB. The core
occupies 0.28 mm2 and draws 15 mW. The achieved power-performance metrics is
favorably comparable with the prior arts.
4.2 Tuner Architecture for TV-GSM Interoperation
Figure 4.1 shows the proposed 2 � VDD-enabled mobile-TV tuner RF front-end. It
supports VHF-III (170–240 MHz) and UHF (470–860 MHz) bands in typical recep-
tionmode, and narrowerUHF band (470–750MHz) in TV-GSM interoperationmode.
A direct-conversion architecture [1] facilitates the frequency plan and hardware reuse
for multiband multistandard mobile-TV applications [2]. The external passives
include an off-the-shelf balun, an off-the-shelf SP3T RF switch and three customized
P.-I. Mak and R.P. Martins, High-/Mixed-Voltage Analog and RF Circuit Techniquesfor Nanoscale CMOS, Analog Circuits and Signal Processing,
DOI 10.1007/978-1-4419-9539-1_4, # Springer Science+Business Media New York 2012
55
preselect filters built with surface-mount devices (SMDs). A wideband LNA covering
the VHF-III and UHF bands is employed as it can cover both VHF III to UHF bands
without demanding reconfigurability [3]. The LNA has to exhibit low noise, high gain
and gain control such that the noise contribution of the latter circuitry is minimized.
However, due to the high gain of the LNA, the linearity of the front-end is limited
by the C-2C attenuator and I/Q mixer drivers. Since a 2 � VDD helps improving
linearity with low overhead, techniques for improving the linearity of the C-2Cattenuator and I/Q mixer driver are developed. All circuit blocks are differential
to desensitize them from bondwire variation, while minimizing common-mode
pickups and even-order nonlinearity. An inductive-peaking feedforward path is
added to even out the passband variation. The reason for the LNA to have three
sets of differential outputs will be addressed later. The gain control is implemented
digitally in three steps: (1) the LNA provides a one-step high-/low-gain control,
(2) the C-2C attenuator offers a coarse gain control with a 6-dB step size, and (3)
the I/Q mixer driver renders a fine gain control with a 0.75-dB step size.
The rationale for employing three preselect filters is related with the fact that
in the frequency range of interest the VHF-III band suffers from harmonic mixing
if only one preselect filter is adopted, since the UHF band is located at the
third harmonic of the local oscillator (LO) when downconverting the VHF-III
band by hard-switching mixers. This drawback is overcome by separation of the
Balun
LNA
C-2CAttenuator
I/Q MixerDrivers
I
Q
LOQ
LOI
880-915 MHz
PolyphasePassiveMixers
170-240 MHz
470-860 MHz
470-750 MHz
On-Chip
Preselect Filters
(for TV-GSMinteroperation)
GSM-900Transceiver
HPF:H(f)
900f (MHz)[log]
|H(f)| (dB)
750
Gain Ctrl
~10 dBIsolation
SP3T RFSwitch
Feedforward Gain-Roll-Off compensation
Off-Chip
Fig. 4.1 Proposed on/off-chip co-designed mobile-TV tuner RF front-end supporting VHF III
(170–240 MHz) and UHF (470–860 MHz) bands in typical reception mode, and narrower UHF
(470–750 MHz) band in TV-GSM interoperation mode
56 4 A High-Voltage-Enabled Mobile-TV RF Front-End in 90-nm CMOS
VHF-III and the UHF prefilters with transference of the mixers to a polyphase
harmonic-rejection mixer scheme [4, 5]. The VHF-III prefilter serves to reject the
interferers located at the harmonics of the LO, whereas the mixer scheme
interpolates a pseudo sinewave to suppress the third and fifth harmonics of the LO.
The preselect filter for the narrower-UHFband supports theTV-GSMco-integrated
terminals. The presence of the GSM-900 service poses a strict challenge to the tuner
operating in the UHF band [6]. Figure 4.2 outlines the spectrum of TV-GSM
interoperation, where the upper cutoff frequency is reduced from 860 to 750 MHz,
as allowed by theDVB-H specifications [7]. Since the isolation between the tuner and
GSM transceiver is limited to roughly 10 dB, external filtering is required for
seamless TV-GSM interoperation. For the GSM transceiver, a forefront off-chip
highpass filter (HPF) minimizes its leakage power from increasing the tuner’s input
noise floor. For the tuner, a preselect filter having a notch at 900MHz is required. An
attenuation level of not less than 58 dB is necessary to guarantee theGSM-900 uplink
signal, with a maximum power level of +33 dBm, still aligned with the maximum
input power outlined by theDVB-H standard. Such a rejection requirement cannot be
simply fulfilled by an off-the-shelf TV-tuner filter [8]. Typically, with no intent of
TV-GSM interoperation, only 30-dB attenuation at GSM-900 uplink is provided.
Moreover, due to a limited Q factor, a sloped passband is induced.
Here a customized preselect filter and a feedforward gain roll-off compensation
path are applied concurrently in the TV-GSM interoperation mode to meet the
GSM-rejection profile, while maintaining the flatness of the passband prior to
down-conversion. Likewise, the input-referred noise around the transition fre-
quency will not be pronouncedly degraded.
Other design specifications are concisely summarized here as they have been
extensively reported in the literature [2, 9]. The representative linearity specifica-
tion is given by the L1 pattern test of DVB-H, i.e., the TV tuner has to demodulate a
f (MHz)[log]
Power (dBm)
0
-35
WantedSignal
In-band blocker
≥58dB is required to suppress it as an in-band blocker
GSMSignal
23
-75
Slopedpassband
3rd OrderInter-modulation
distortion
2f1-f2 2f2-f1(900)
f1 f2
Fig. 4.2 170-to-750-MHz preselect filter’s profile for receiving the TV band in the presence of
GSM-900 uplink
4.2 Tuner Architecture for TV-GSM Interoperation 57
16-QAM signal when there are two interferers, one digital (40-dB stronger) and one
analog (45-dB stronger), two and four channels away from the desired, respec-
tively. This test sets the IIP3 to �5 dBm together with a noise figure (NF) of 8 dB.
The sensitivity specification is also given by the DVB-H standard. At the maximum
gain a NF of 5 dB is required. This value includes the insertion loss (IL) of the
external components.
4.3 On/Off-Chip Circuit Design
All the integrated circuits are of differential architecture though several schematics
are shown in their single-ended equivalents for simplicity.
4.3.1 Basic-Cell of the LNA
This sub-section describes the basic cell of the LNA that can be easily upscaled to
2 � VDD operation. As shown in Fig. 4.3, a 1 � VDD inverter-type amplifier with
the NMOS’s source terminal as the input enables a wideband input impedance
match. The capacitive cross-coupling technique [10] reuses the gate-source
transconductance (gm) of both NMOS and PMOS devices, resulting in an improved
gain-to-power efficiency. In this sub-circuit level, the input DC voltage is set at
Vinp
Rfb Rfb
VDD
Vp,bulk
Voutp Voutn
Vinn
RIN,diff
VRF
Vn,bulk Vn,bulk
Vp,bulk
VDD
C1C1
M1
M2M2
M1
(VDC = 0.5xVDD)
(VDC = VSS)Balun
Fig. 4.3 1 � VDD inverter LNA. Input DC level is assumed to be at Vss (0 V)
58 4 A High-Voltage-Enabled Mobile-TV RF Front-End in 90-nm CMOS
0 V (VDC ¼ VSS) through the off-chip balun. The gate voltage is self-biased by the
feedback resistor Rfb, resulting in an output DC voltage halfway of the supply (i.e.,
0.5 � VDD) that can maximize the linear output swing. The differential voltage gain
AV,diff is given by,
AV;diff ¼ Voutp � Voutn
Vinp � Vinn
¼ gm1 þ 2gm2 þ 1
ro2� 1
Rfb
� �ro1==ro2==Rfbð Þ: (4.1)
where gm1,2 and ro1,2 are the transconductance and output resistance of M1,2,
respectively. The overall transconductance (Gm) corresponds to the term gm1 +2gm2 + 1/ro2�1/Rfb which can be approximated to gm1 + 2gm2 for sufficiently
large ro2 and Rfb, implying roughly a tripled increment of transconductance when
comparing to that achieved by a single transistor. With VDD ¼ 1 V, proper sizings
lead to │VGS│ ¼ │VGD│ ¼ │VDS│ ¼ 0.5 V for bothM1 andM2. Given that |VT| of the
thin-oxide NMOS and PMOS transistors is roughly 0.3 V, each transistor is biased in
strong inversion with │VGS�VT│ ¼ 0.2 V. The Gm-to-Ib ratio is ~30 V�1 and the
output dynamic range is 0.6 Vpp.
The use of M2’s source node as the input terminal realizes a wideband input
impedance match. It can be shown that the differential input resistance (RIN,diff) of
the LNA is given by,
RIN;diff ¼ 2
2gm2 þ 1Rfb
þ 1Rfb
� 1ro2
� �AV; diff
: (4.2)
which can be simplified to a handy and observable form,
R0IN;diff �
2
gm1 þ 4gm2; (4.3)
when ro1 and ro2 are assumed to be infinite. However, in nanoscale technologies
Eq. 4.2 is essential for an accurate calculation of RIN,diff. RIN,diff and the total input
parasitic capacitance CIN determine the value and BW of the input reflection
coefficient magnitude │GIN│, as given by,
GINj j ¼1
joCIN==RIN;diff � RS
1joCIN
==RIN;diff þ RS
����������; (4.4)
where RS denotes the output resistance of the test source. The finite output imped-
ance ofM1 andM2 complicates the optimization between voltage gain, BW and NF
under an impedance-match condition. For the sizing of each component, a
constraint-based semi-computed design flow [11] is applied to optimize those
parameters concurrently. It is noteworthy that the capacitive cross-coupling
4.3 On/Off-Chip Circuit Design 59
technique and the use of an off- chip balun for signal injection allows effective
noise and distortion cancellation of M2.
Self-biased inverter-based circuits are sensitive to process, voltage and temper-
ature (PVT) variation. The back-gate control scheme highlighted in [12] is an
effective solution for this problem. It keeps the supply current constant and reduces
the sensitivity to supply ripple by returning the correcting signals to Vp,bulk and
Vn,bulk. A triple-well process is required to isolate the bulk of NMOS from substrate.
Because this work is the very first proof-of-concept prototype, this back-gate
control scheme has not been embedded.
Since the linearity of the receiver is mainly limited by the I/Q mixer drivers, the
main design consideration of the LNA is to operate it under a 2 � VDD with
guaranteed reliability. The associated biasing is simplified by introducing a VDD
partitioning concept. Based on it, a 2 � VDD cascode-inverter LNA, and a 2 � VDD
cascode-cascade-inverter LNA that befits the proposed RF front-end, are developed.
1. VDD Partitioning Concept: In order to reliably bias the thin-oxide transistors
under an elevated supply, it is convenient to equally divide the supply into four
regions, as shown in Fig. 4.4. From left to right, under a 2 � VDD, a 1 � VDD
inverter-type amplifier (IA1 to IA3) can be connected in three different ways
without affecting the performance and reliability. Such an amplifier is self-
biased by a feedback resistor. It can be observed that the inter-rail voltage levels
(0.5 � VDD, 1 � VDD and 1.5 � VDD) call for additional circuitry for generating
the inter supply rails. This overhead, however, can be avoided by exploiting a
cascode of two identical inverter amplifiers (IA4 and IA5). The intermediate
point is still the RF input node, self-biased to a value close to 1 � VDD because
of voltage division. Due to a matched I/O DC level, the voltage gain can be
enhanced further by cascading the cascoded inverter amplifiers (IA6 to IA9),
without needing any ac-coupling.
2xVDD
1.5xVDD
1xVDD
0.5xVDD
VSS
Single Stage Cascode Cascode-Cascade
IA1
IA2
IA4
IA5 IA7
IA6
IA9
IA8IA3
InverterAmplifier
self-biasedto1xVDD
Fig. 4.4 Three 2 � VDD partitioning schemes for the LNA
60 4 A High-Voltage-Enabled Mobile-TV RF Front-End in 90-nm CMOS
2. 2 � VDD Cascode-Inverter LNA: Figure 4.5 describes the transformation of a
1 � VDD inverter LNA into a 2 � VDD cascode-inverter LNA. It is noteworthy
that the main objective is to keep the RF performance, power consumption and
output resistance remain unchanged, but the voltage capability is doubled. The
LNA is structured by, first, splitting the 1 � VDD inverter LNA into two, relying
on the principle of device parallelism. The second is stacking them together. The
reliability of all devices is maintained at static and power-up/down transient as
the dc-levels of all internal nodes follow linearly with the 2 � VDD when it
ramps up (see Fig. 4.5, right). Due to the separation of circuits from 1 to 2, the
output impedance of each output (Vout1 and Vout2) is doubled from its original
1 � VDD design. This issue can be solved by ac-shorting the dual outputs
passively by a capacitor, or actively by another gain stage as shown in
Fig. 4.6. The latter can further boost the RF gain, at the expense of the BW
(since Vout is of even higher output impedance). The reliability is guaranteed
simply by adding two cascode devicesMcp andMcn with resistor Rb for the drain-
gate bias.
Here the passive combination has been selected for its excellent linearity, and
also the fact that the required summing capacitor can be merged into the
subsequent programmable C-2C attenuator, and the gain of the supply and
Ib x 0.5
Ib x 0.5
Rfb
Rfbx 2
Rfbx 2
Rfbx 2
VDD
VSS
VSS
2xVDD
VDD
Vout
M2
M2
M2M1
M1
M1
M4
M3
C1
C1x 0.5
C1 x 0.5
C1 x 0.5
Splitting bases onparallelism of each device
Cascoding (upper)
Cascoding (lower)
1xVDD
1xVDD
1xVDD
Vinn
Vinn Vout
Vout1
Vout2
Vinp
Vinp
Vinn
Vp,bulk
Vn,bulk
Vinn
Vinp
(VDC = Vss) (VDC = 0.5xVDD)
(VDC = 0.5xVDD)
(VDC = 1.5xVDD)
(VDC = 1xVDD)
(VDC = 1xVDD)
(VDC = 1xVDD)
(VDC = Vss)
0
0.5
1
1.5
2
0 0.5 1 1.5 2Supply Voltage (V)
DC
Lev
el (
V)
LNA
2xVDD2xVDD
0
ramp upA
B
C
D
A
B
C
D
W1/L1
W1/L1 x 0.5
W1/L1 x 0.5
W2/L2 x 0.5
W2/L2 x 0.5
W2/L2 x 0.5 x k
W1/L1 x 0.5 x k
W2/L2
Ib
Fig. 4.5 Constant-power constant-performance transformation of a 1 � VDD inverter LNA into a
2 � VDD cascode-inverter LNA (simplified half circuit)
4.3 On/Off-Chip Circuit Design 61
ground noises to the single-ended output can be remarkably reduced. The gain
enhancement is modified as a double capacitive cross-coupling technique. Thus,
the bias point, and the net voltage stressed on each transistor, are unaltered when
compared with a generic 1 � VDD design. This feature contrasts with the
outside-rail HV circuits [13] which operates with large signals, the trajectories
of the device terminal voltages in transients must be controlled to be within the
reliability limits such as HCI, TDDB and punchthrough. The factor k is
introduced as a correction constant since the roles of NMOS and PMOS in the
lower inverter are switched comparing with the upper one. It is possible to
design without k, but it helps to ensure that the upper and lower inverters can
achieve the same RF performances. The bulk of M2 (M3) should be tied to its
source to avoid voltage overdrive between the bulk and source terminals. The
bulks of M1 and M4 can be used to desensitize the circuit from PVT variation as
mentioned before. The DC level of the input node is self-biased to a value close
to 1 � VDD (halfway of the elevated supply).
3. 2 � VDD Cascode-Cascade-Inverter LNA: Its schematic is shown in Fig. 4.7.
Adding one more gain stage boosts gain and permits gain switching without
affecting the input impedance match. The second stage is realized by A3 and A4
that are inverter-like common-source amplifiers offering a simple gain-bypass
mode by using a switch (SW) in parallel with Rfb2. Considering the upper path
involving A1 and A3, the voltage gain AV,A3 of A3 in the high-gain mode can be
expressed by,
AV;A3 ¼ Vout;2nd
Vin;2nd¼ gm5 þ gm6 � 1
Rfb2
1ro5==ro6
þ 1Rfb2
: (4.5)
From
From
(VDC = 1.5xVDD)
(VDC = 1.5xVDD)
(VDC = 0.5xVDD)
(VDC = 0.5xVDD)
(VDC = 1xVDD) (VDC ª 1xVDD)
2xVDD
Mcp
VSS
Mcn
Mn
Mp
Rb
Vout1
Vout,sum
Vout2
Fig. 4.6 A 2 � VDD cascode-inverter LNA with its dual outputs combined actively by a dual-
input cascode amplifier
62 4 A High-Voltage-Enabled Mobile-TV RF Front-End in 90-nm CMOS
where gm5,6 and ro5,6 are the transconductance and output resistance of M5,6,
respectively. In the low-gain mode, Rfb2 is reduced (in parallel with the ON-
resistance of the SW). The input resistance of A3 RIN,2nd is given by,
RIN;2nd ¼ Rfb2 þ ro5==ro61þ gm5 þ gm6ð Þ ro5==ro6ð Þ : (4.6)
Since RIN,2nd loads to A1, the gains of A1 and A3 are simultaneously reduced in
the low-gain mode when Rfb2 is decreased, effectively increasing the linearity.
Besides, no reliability issue is induced as the gain control involves no change of
bias point. The consideration applies for the lower path involving A2 and A4. One
particular feature of this LNA is that in addition to the two high-impedance
output terminals (Vout1p and Vout2p), a low-impedance output terminal (Vout3p) is
available, which will be re-used for gain roll-off compensation.
The simulated performances of the standalone LNA are shown in Fig. 4.8. At noload condition the high-gain mode shows 31.4-dB voltage gain, 2.8-dB NF and
S11 < �10-dB BW of 0.1–1.5 GHz. At the low-gain mode, the performances are
21.1-dB voltage gain, 3.6-dB NF and S11 < �10-dB BW of 0.1–1.56 GHz. An
in-band 2-tone test at 400 and 500MHz shows an IIP3 of -3.3/5.4 dBm at high-/low-
gain mode. The LNA draws 9.8 mW at 2 V. It is noteworthy that those performance
metrics exclude the loading effect of the C-2C attenuator and I/Q mixer drivers.
Co-simulations between blocks are necessary to justify the overall RF performances.
Ib1 Ib2
A4
A1
A2
High/LowGain Mode
A3
NX
A3 & A4
SW
ESD Protection
2xVDD
2xVDD 2xVDD
VSS
VSS VSS
C1
C1
Vinp
Vinn(VDC = 1xVDD)
(VDC = 1xVDD)
Vout,2nd
Vout1p
Vout3p
Vout2p
Vin,2nd
RIN,2ndVp,bulk
M5
M6
Rfb2
(VDC = 1.5xVDD)
(VDC = 1xVDD)
(VDC = 0.5xVDD)
Fig. 4.7 A 2 � VDD cascode-cascade-inverter LNA with triple output terminals and high/low-
gain mode (simplified half circuit)
4.3 On/Off-Chip Circuit Design 63
4.3.2 ESD Protection Scheme
Since NX (see Fig. 4.7) is self-biased internally by the LNA at a DC level halfway of
the supply, it is convenient to apply forward-connected diode chains to boost the
ESD protection level. As shown in Fig. 4.9, D3’s and D4’s are three substrate PNP
diodes, which, together with the reverse-biased diodes (D1 and D2) and power
clamp construct the ESD protection scheme. In the ESD-robustness simulation, a
Human Body Model (HBM) voltage pulse is applied to the LNA’s input to induce a
large +/� zapping discharge current that has a rising/falling time of ~8 ns [14]. As
verified in all combinations the RF input pins can withstand minimally �4 kV of
ESD zapping without causing internal or protection devices failure. This result
fulfills the standard of “safe level” in the chip-level ESD specifications. The main
concern is the induced nonlinear parasitic capacitance, which is roughly 0.2 pF in
this design. With a rail-to-rail sinewave applied at the input, the total harmonic
distortion is 0.1%.
4.3.3 Programmable C-2C Attenuator
The linearity of the I/Q mixer driver limits that of the entire TV tuner because of the
voltage gain of the LNA. Instead of utilizing the current-steering gain-control
method [15] that can affect the operating points, a passive C-2C attenuator
(Fig. 4.10) is inserted between the LNA and the I/Q mixer drivers to control
coarsely the dynamicity of the RF signal. Conveniently, the input capacitor can
be divided into two equally-sized capacitors (Cx’s) for passively combining the
LNA’s upper (Vout1p) and lower (Vout2p) outputs. The five-stage attenuator offers an
0.1 1 10Frequency (GHz)
In-Band
-15-10-505
101520253035
S11
High-Gain ModeLow-Gain Mode
NF
Voltage GainV
olta
ge G
ain,
NF
, S11
(dB
)
no load
Fig. 4.8 Simulated performances of the standalone LNA with no load
64 4 A High-Voltage-Enabled Mobile-TV RF Front-End in 90-nm CMOS
attenuation range Av ¼ 0 to �30-dB with a �6-dB step size, i.e., Av ¼ Gn(2)�n for
n ¼ 0, 1, . . ., 5, where Gn ¼ (0, 1) is the gate control logic for switching the
attenuation levels. Sizing Cx has a tradeoff between the accuracy and BW. An
increase of Cx value can desensitize the gain step from parasitic capacitances,
whereas a decrease of Cx value can maximize the BW. Here the optimized Cx is
0.5 pF for the targeted BW. The simulated magnitude response of the C-2Cattenuator is shown in Fig. 4.11. The ON-resistance of the switches and capacitor
size determine the BW of the attenuator, which is well over 1 GHz among all
D1
D2
D3's
D4's
Powerclamp
RF Pin ESDProtection
LNA
(VDC ª 1xVDD)Vinp
2xVDD
VSS
Fig. 4.9 ESD protection scheme of the RF input pin
1.5xVDD
0.5xVDD
ON
OFF
Driving VoltageG0 ... G5
(to NY, biasedat 0.5xVDD)
0 to - 30 dB (-6dB/step)
Vin1 Vout
Vin2
CX
CX
CX = 0.5 pFAll NMOS: (5/0.1) x 4
CX CX 2CX
2CX2CX
VSS
G0
G0
G1 G2 G5
VSS VSS
Fig. 4.10 Programmable C-2C attenuator for coarse-gain control. The input capacitor 2Cx is
divided into 2 Cx’s to interface with the LNA
4.3 On/Off-Chip Circuit Design 65
attenuation levels. The output is not buffered before driving the mixer driver,
inducing 0.1-dB passband gain loss under a 0.1-pF load, which models the input
capacitance of the mixer driver.
An elevated VDD helps minimizing the size and nonlinearity of MOS switches.
In the proposed VDD partitioning scheme there are four gain switching methods
using digital inverters, as shown in Fig. 4.12. NMOS and PMOS switches using
the corresponding inverter ensure VGS ¼ 1 V in ON-state, VGS ¼ 0 V in OFF-
state. The main reliability concern of the C-2C attenuator is the BTI because
all switches conduct no static current, and to overcome it, triple-well NMOS
switches allowing bulk-source connection are employed as they are far less
affected by BTI than the PMOS devices. The associated tradeoff is that NMOS
-50
-45
-40
-35
-30
-25
-20
-15
-10
-5
0
0.1 1 10Frequency (GHz)
Mag
nitu
de R
espo
nse
(dB
)
In-Band
-1dB ~ - 31dB
@ 0.1pF load
Fig. 4.11 Simulated magnitude response of the C-2C attenuator
Vo2
Vo3Vo3
Vo1
Vo1
Vo2
Vo2
NMOS
NMOS
PMOS
PMOS
DigitalInverter
VSS
0.5xVDD
1xVDD
1.5xVDD
2xVDD
1(2xVDD)0(1xVDD)
1(1xVDD)0(GND)
1(1.5xVDD)0(0.5xVDD)
Fig. 4.12 Appropriate digital inverters for switches operating at different supply rails. They drive
NMOS and PMOS switches correspondingly, such that VGS of 1 � VDD in ON-state can be
achieved in the four cases shown
66 4 A High-Voltage-Enabled Mobile-TV RF Front-End in 90-nm CMOS
requires a deep n-well to permit bulk-source connection, which is a cheap option
widely available in advanced processes. The parasitic capacitances associated
with the deep-n-well diodes have to be accounted to minimize gain-step error.
The terminal voltage NY is purposely set to 0.5 � VDD (0.5 V) to interface with
the I/Q mixer drivers. As such, the gain-control logics [G0 to G8] can be reliably
up-shifted to 1.5 � VDD (1.5 V) in ON-state. Considering AMR and TDDB, the
RF input signals (Vin1 and Vin2) can be as large as 0.4 Vpp. In the linear region, a
0.5-V increment of overdrive (VGS�VT) leads to a better linearity and a smaller
transistor size for a given ON-resistance (RON). The RON of a CMOS transistor
based on a first-order model is given by,
RON ¼ 1
moCoxWL VGS � VTð Þ : (4.7)
The simulated IIP3 of the C-2C attenuator under a 50-Ω source impedance with
1.5 and 1-V gate-control voltages are shown in Fig. 4.13. The former achieves +9.8
to +16.1 dBm IIP3 among an attenuation range of 0 to �30 dB, showing minimally
7.3-dB improvement of IIP3 when comparing it with the latter.
The 0.5 � VDD-up level shifter is shown in Fig. 4.14. It features a simple
structure to realize low-to-high transition from 0/1-V input to 0.5/1.5-V output.
When the input is logic 1 (1 V), Vx level is close to ground, MOS switch Ma is
turned ON (with a resistance much smaller than R1), yielding Vy � 0.5 V [i.e., 2 V ·
(2R1//R1)/(2R1//R1 + 2R1)] and making [G0 to G5] equal to logic 1 (1.5 V). On the
other hand, when the input is logic 0 (0 V), Ma is OFF, yielding Vy � 1.5 V
0
2
4
6
8
10
12
14
16
18
-30 -24 -18 -12 -6 0
Gain (dB)
IIP3(
dBm
)
[G0 to G5]: 1V
[G0 to G5]: 1.5V
VDC=0.5V VDC=0.5V
VDC=0.5V VDC=0.5V
7.3 dB
Fig. 4.13 IIP3’s of the C-2C attenuator with 1-V and 1.5-V gate voltages. The triple-well NMOS
switches are biased at a DC level of 0.5 V
4.3 On/Off-Chip Circuit Design 67
[i.e., (2 V�1 V) · (2R1)/(2R1 + 2R1) + 1 V] and making [G0 to G5] equal to logic
0 (0.5 V). It can be verified that all terminal voltages of MOS devices satisfy the
reliable limits. The required reference voltages are generated by an on-chip high-
ohmic polysilicon resistor ladder.
4.3.4 I/Q Mixer Drivers
The proposed I/Q mixer drivers (Fig. 4.15) are based on a cascode structure with a
resistive load to drive a passive mixer. It benefits the most from an elevated VDD in
terms of linear output swing and reverse isolation. The fine-gain control [G6 to G8]
is set at the AC-switching part ofM3 (i.e.,M03), covering a 0–6-dB gain range with a
0.75-dB step size. The gain-control logics [G6 to G8] are up-boosted to 0.5 � VDD
in OFF-state and 1.5 � VDD in ON-state to improve the linearity, similar to the gain
control in the C-2C attenuator. The entire coarse-fine gain control involves no
change of bias points, ensuring the reliability in all operating modes. M4 and M5
are NMOS transistors with VGS ¼ VDS ¼ 1 � VDD and are of long channel length
(i.e., 1.2 mm) to avoid BTI and punchthrough.
M3 and M03 are linearized and self-biased by R3 at a VGS of 0.5 � VDD. They
deliver the signal current to the two cascode devices M4 and M5 while maintaining
adequate reverse and I/Q isolations of 76.3 and 38.4 dB in simulations, respectively.
The linear output swing is boosted to 0.6 Vpp without jeopardizing the reliability
limits in terms of RF stress [16], which is confirmed by checking the trajectories of
the terminal voltages in power-up/-down transients.
The mixer driver is supposed to deliver high swing output to a resistive load while
ensuring a sufficient voltage gain. When driving a passive mixer realized with
MOSFET, a small device size is preferred as the switching power of the local oscillator
(LO) path can be minimized. The mixer switchMMIX shown in Fig. 4.15 is of PMOS
type to take advantage of the 1.5 � VDD output dc-level, increasing the gate-source
2 V
1 V
2R1
2R1 R1
0 V
1 V
1.5 V
0.5 V
MaVGS, VGD, VDS, ≤ 1v
for all transistors
Gain Ctrl Logic
(1V, 0V)
G0 ...G5(1.5V, 0.5 V)
Vx
Vy
Fig. 4.14 0.5 � VDD-up level shifter
68 4 A High-Voltage-Enabled Mobile-TV RF Front-End in 90-nm CMOS
overdrive in ON-state as 1 � VDD. Since the I/Q mixer driver has resistive input
impedance, the loading effect between it and the LNA and C-2C attenuator must
be taken into account to check the overall gain and output BW under different loads.
As shown in Fig. 4.16, with RL1 ¼ 700O the voltage gain can be maintained>20 dB
0 to -30 dB(-6dB/step)
ProgrammableC-2C Attenuator
NZ (VDC ª 0.5xVDD)NY
(VDC ª 0.5xVDD)
ONOFF
GateVoltageG6 ...G8
Feedforward Gain roll-offCompensation Path
C
A A
B B
0 to 6 dB(0.75dB/step)
(VDC ª1.5xVDD) (VDC ª1.5xVDD)RL1
MMIX
loutp Qoutp
MMIX
M5
M3
M3'
R3
M4
LO90LO0
RL1
1.5xVDD1.5xVDD
1.5xVDD
0.5xVDD0.5xVDD
2xVDD 2xVDD
VSS
VSS
G6
G6G6 -G8
Vout1p
Vout2p
Vout3p
Fig. 4.15 The proposed cascode I/Q mixer driver. M3 is partially switchable for fine-step gain
control. The low impedance nodeNz interfaces the feedforward path to compensate the gain roll-off
0.1 1 10Frequency (GHz)
-20
-10
0
10
20
30
Mag
nitu
de R
espo
nse
(dB
) RL1 = 700Ω
In-Band
MMIX
3.6/0.1 x 4
3.6/0.1 x 33.6/0.1 x 2
3.6/0.1
Fig. 4.16 Magnitude response of the RF front-end at different sizes of the MMIX
4.3 On/Off-Chip Circuit Design 69
with around 1-GHz overall output BW with sizes ofMMIX ranging from 3.6/0.1 � n,where n ¼ 1, 2, 3, 4. With a 50-O RF test source and MMIX sized as 3.6/0.1, the
simulated differential voltage gain of the mixer driver is 5.4 dB and IIP3 is +8.6 dBm.
The I/Q mixer driver is extensively voltage biased to freeze the operating points
necessary for reliable operation. Since the DVB-H employs time-slicing operation [7]
a resistor ladder generating the reference and bias voltages avoids wrong voltage
buildup (remove) sequences in internal nodes during power-up (down) transients.
Figure 4.17 shows the simulated internal node voltages of the I/Q mixer drivers
(markers correspond to Fig. 4.15) when the 2 � VDD ramps up from 0 to 2 V. It can
be observed that the potential differences of all internal nodes are within the reliable
guides of AMR and TDDB.
4.3.5 Customized External Preselect Filters
The schematics, S21 andS11of the three preselect filters optimized for the 170-to-240,
470-to-860, and 470-to-750 MHz bands, are shown in Fig. 4.18a–c, respectively. The
source and input impedances of theRF front-end arematchedwith 75O. TheQ-factorsof surface- mount-device (SMD) inductor and capacitor at 1 GHz are 60 (QL) and 80
(QC), respectively. A 1-pF input parasitic capacitance is assumed to be at the input of
the RF front-end. The in-band │S11│ is less than�10 dB in all cases. The 470-to-750-
MHz preselect filter achieves 60 dB rejection at 900 MHz at the expense of 1.66-dB
passband variation near the cutoff. This effect is addressed by a gain roll-off
0
0.2
0.4
0.6
0.8
1
1.2
0 0.5 1 1.5 2
Supply Voltage (V)
Inte
rnal
nod
e V
DC (
V)
A C
C
B C
I/Q MixerDriver
0
ramp up2xVDD
2xVDD
M4 ,M5: VDS =
VGS =
M3: VGS = VDS =
VSS
Fig. 4.17 Simulated DC-node voltage change of Fig. 4.15 with the 2 � VDD ramps up from 0 to 2 V
70 4 A High-Voltage-Enabled Mobile-TV RF Front-End in 90-nm CMOS
-100
-80
-60
-40
-20
0
-35
-30
-25
-20
-15
-10
-5
0
56nH
8pF 17pF
56nH
5pF
S21
(dB
)
S11
(dB
)
0 0.5 1 1.5 2Frequency (GHz)
-3.5
-3
-2.5
-2
-1.5
-1
-0.5
0
-70
-60
-50
-40
-30
-20
-10
0
0 0.5 1 1.5 2Frequency (GHz)
9nH
0.5pF
-70
-60
-50
-40
-30
-20
-10
0
0 0.5 1 1.5 2-40
-35
-30
-25
-20
-15
-10
-5
0
Frequency (GHz)
1.66 dB Variation3.8nH
8.3pF
18nH
5pF 5pF
3.8nH
8.3pF
18nH
5pF 5pF
3.8nH
8.3pF
S21
(dB
)S
21 (
dB)
S11
(dB
)S
11 (
dB)
a
b
c
Fig. 4.18 Simulated S21 and S11 of the three preselect filters for: (a) 170-to-240-MHz band,
(b) 470-to-860-MHz band and (c) 470-to-750-MHz band
4.3 On/Off-Chip Circuit Design 71
compensation technique to be described next. The 170-to-240-MHzpreselect filter can
achieve more than 30 dB rejection at 510 MHz to prevent the harmonic mixing. In
order to measure the frequency response with minimum IL, the 470-to-860-MHz
preselect filter for typical reception is a simplified LC structure. The selected balun
and RF SP3T switch are wideband components. Up to 1 GHz, the 1:2 balun has a 1.2-
dB IL [17] and the SP3T RF switch has a 0.4-dB IL [18].
4.3.6 Feedforward Gain Roll-Off Compensation
The gain roll-off due to the 470-to-750-MHz preselect filter, and the finite BW
limitation of the LNA and C-2C attenuator in conjunction induce significant gain
roll-off near the cutoff. Vout3p (see Fig. 4.7) is a low-impedance output node of
the LNA reusable for gain roll-off compensation. Confirmed by on/off-chip
co-simulation a feedforward path (Fig. 4.19) from Vout3p with inductive peaking
and amplification realizes a low-Q highpass characteristic. The gain block of �1
implies cross-connection of the differential terminals. It compensates the passband
roll-off due to the external preselect filter, the LNA and the C-2C attenuator. The
inductor LFF is differential for area savings and the amplification is based on another
inverter amplifier A5. Since it is a low-Q peaking, the technique is insensitive to the
absolute value of LFF. An error amplifier loop around Mr generates a regulated
1 � VDD supply for reliable operation of A5. The error amplifier is based on a
differential pair with a current mirror load. Its current tail NMOS is diode-connected
to allow self-biasing, avoiding any overstress when the 2 � VDD ramps up. This
current tail is sized to consume a voltage stress of ~0.5 � VDD to reduce the voltage
stressed on the differential pair and current mirror. The simulated passband flatness
with and without compensation are depicted in Fig. 4.20. The passband flatness is
improved by 2.55 dB.
(Reference)1xVDD
-1
-+
A5
Gain Roll-off Compensation
(VDC ª1xVDD)
Mr
Mr
VoutTo NZ
VSSVSS
VSSVSS
CFF
LFF
2xVDD
2xVDD
1xVDD(Buffered)
Vout3p
~1.5xVDD
~0.5xVDD
Fig. 4.19 Simplified schematic of the feedforward gain roll-off compensation path
72 4 A High-Voltage-Enabled Mobile-TV RF Front-End in 90-nm CMOS
4.4 Experimental Results, Discussions and Benchmarks
Prototypes of the RF front-end have been fabricated in a 1-V 90-nm CMOS process.
The die micrograph and test fixture for resisting the ambient GSM signals from
affecting the test results are shown in Fig. 4.21a and b, respectively. The RF front-
end employs a single 2-V supply. It occupies 0.28-mm2 active area and is filled
symmetrically by obligatory dummy tiles to avoid mechanical strain differences.
The metal lines are set to 1 mm width per 1 mA DC current to prevent electro-
migration. An inductive-peaking 50-O test buffer was designed to drive the
equipments since the loading effect of the buffer cannot be simply de-embedded.
Figure 4.22a–c show theRF performancesmeasuredwith different preselect filters.
The measured peak voltage gain ranges from 24.9 to 26.5 dB and the minimum NF of
the RF front-end ranges from 3.5 to 3.9 dB, after de-embedding the loss of the RF
switch, balun and preselect filters. The in-band S11 is below �10 dB in all modes. In
TV-GSM interoperation mode, the rejection at GSM-900 uplink measures 68 dB, and
less than 0.7-dB gain roll-off within 470–750 MHz. Since the GSM-rejection filter is
realized in a discrete form, post-tuning for alignment is necessary to ensure a stable
and accurate notching at GSM-900 uplink against PVT. The manufacturable imple-
mentation of the filter prototypes should be with LTCC technology [19, 20] in
the next phase, which can have an accurately controlled frequency response.
Figure 4.23a and b show the voltage gain and gain error, NF and IIP3, measured
against the gain control words, respectively. The gain step error is within �1 dB
throughout a gain control range of 46 dB. It is believed that this error can be mainly
attributed to the C-2C attenuator, which should be re-optimized with the parasitic
capacitance caused by the deep-n-well diodes and the density tilings.
24
25
26
27
28
29
450 750650550
Frequency (MHz)
w/ compensation
w/o compensation
Vol
tage
Gai
n (d
B)
2.55 dB
Fig. 4.20 Simulated magnitude responses of the RF front-end with and without gain roll-off
compensation
4.4 Experimental Results, Discussions and Benchmarks 73
Throughout a 46-dB gain-control range, the IIP3 ranges from �5.5 to +4 dBm
under a 2-tone test at 400 and 500 MHz. Nevertheless, when the RF front-end is
switched to the low-gain mode (10-dB gain back-off), an IIP3 of �1.6 dBm
together with a minimum NF of 6.4 dB is adequate to pass the DVB-H L1 pattern
test with low sensitivity degradation. The RF front-end excluding the test circuitry
consumes 15 mW, out of which 10 mW is due to the LNA and the C-2C attenuator
(plus its associated bias circuit).
The desensitization of the RF front-end to the GSM-900 interference is
characterized by using two 2-tone tests as shown in Fig. 4.24. A 2-tone test at 0.5
and 0.7 GHz measures an IIP3 of �5.5 dBm. With a preselect filter notching at the
0.9 GHz, the 2-tone test at 0.7 and 0.9 GHz shows that the generated third-order
intermodulation distortion (IM3) becomes insignificant.
With no access to industrial long-term reliability testers, the reliability of the
chip was justified by operating it continuously at room temperature for 3 days. No
detrimental effect on the performance was noted. Due to a similar reason, the ESD
robustness could not be strictly characterized experimentally. The achieved ESD-
protection level is based on simulations. Comprehensive system-level
measurements including EVM degradation in a TV-GSM co-integrated scenario
should be chased in the latter stages of the research, with the presence of the digital
demodulator and other analog-baseband circuitry.
There is no similar on/off-chip co-design RF front-end reported in the literature
for direct performance comparison. Nevertheless, it is relevant to compare the
combined performance of the LNA and C-2C attenuator with the state-of- the-art
CMOS variable-gain low-noise amplifiers (VGLNAs) [12, 21, 22] in Table 4.1.
Fig. 4.21 (a) Test fixture for resisting the ambient GSM signals and (b) chip micrograph of the RF
front-end
74 4 A High-Voltage-Enabled Mobile-TV RF Front-End in 90-nm CMOS
-60
-50
-40
-30
-20
-10
0
10
20
30
50 100 150 200 250 300 350 400 450 500
5
10
15
20
25
30
35
40
45
0
NF
S11
Voltage gain
Frequency (MHz)
Vol
tage
Gai
n an
d S
11 (
dB)
Vol
tage
Gai
n an
d S
11 (
dB)
Vol
tage
Gai
n an
d S
11 (
dB)
NF
(dB
)N
F (
dB)
NF
(dB
)
In-band
-20
-15
-10
-5
0
5
10
15
20
25
30
50 350 650 950 1250 1550 18500
5
10
15
20
25
30
35
40
45
50
Frequency (MHz)
NF
S11
Voltage gainIn-band
-50
-40
-30
-20
-10
0
10
20
30
0
5
10
15
20
25
30
35
40
50 350 650 950 1250 1550 1850
Frequency (MHz)
NF
S11
Voltagegain In-band
a
b
c
Fig. 4.22 Measured RF performances of the RF front-end with the corresponding preselect filters
for: (a) 170-to-240-MHz band, (b) 470-to-860-MHz band and (c) 470-to-750-MHz band
4.4 Experimental Results, Discussions and Benchmarks 75
-18
-13
-8
-3
2
7
12
17
22
27
-3
-2
-1
0
1
2
3
Gai
n E
rror
(dB
)
Vol
tage
Gai
n (d
B)
Gain Control Word
0 10 20 30 40 50 60
Low Gain Mode
High Gain Mode
Voltage Gain
Gain Error
-23
32
0
4
8
12
16
20
24
28
32
36
-25
-20
-15
-10
-5
0
5
10
15
20
Low Gain Mode
High Gain Mode
Low Gain ModeHigh Gain Mode
Gain Control Word
0 10 20 30 40 50 60
NF
IIP3
IIP3
(dB
m)
NF
(dB
)
a
b
Fig. 4.23 Measured RF performances against the gain control word at 600 MHz: (a) voltage gain
and gain error, and (b) NF and IIP3
-90
-70
-50
-30
-10
10
30
-45 -35 -25 -15 -5 5
0.5 0.70.3 0.9f(GHz)
IM3Fundemental
IM3
0.7 0.90.5 1.1f(GHz)
IM3Fundemental
IM3
IM3Fundemental
-5.5 dBm
Input Power (dBm)
Out
put P
ower
(dB
m)
Fig. 4.24 Linearity measurements with 2-tone tests at: 0.5 and 0.7 GHz, 0.7 and 0.9 GHz
76 4 A High-Voltage-Enabled Mobile-TV RF Front-End in 90-nm CMOS
Table
4.1
Perform
ance
summaryandbenchmarkwiththestate-of-the-artCMOSVGLNAs
Thiswork
[12]
[21]
[22]
Frequency
(MHz)
170–240
470–750a
470–862
470–862
470–870
174–245
450–862
NF(dB)
3.8
3.9
3.5
2.6
4.3
3.0
4.5
Max.voltagegain(dB)
24.9
26.2
26.5
23
16
28
25
Rejectionat
GSM-900uplink(dB)
67b
68b
2b
n/a
n/a
n/a
n/a
ESD
protection/level
(V)
Yes/4
kc
no
no
Yes/n/a
IIP3(dBm)@
gain(dB)
�5.5
@26.2
�5.4
@23
�1.5
@16
�5to
�4@
25
�1.6
@16(low-gainmode)
+0.6
@9
Variable
gainrange(dB)
46
14
33
50
Power
(mW)@
VDD(V
)10@
2(LNA
+C-2
Cattenuator)
40@
1.2
22@
1.8
16@
1.8
15@
2(+
also
I/Qmixer
drivers)
Activearea
(mm
2)
0.283
0.067
0.32
0.52
Circuittopology
Differential
Single-endin
Single
end
Differential
Differential
out
Technology
90nm
CMOS
90nm
CMOS
0.18mm
CMOS
0.18mm
CMOS
aTV-G
SM
inter-operationmode
bWithexternal
filters
cTheESDprotectionlevel
isbased
onhuman
bodymodel
(HBM)simulation
4.4 Experimental Results, Discussions and Benchmarks 77
With similar NF and linearity performances, this work is advantageous for its lowest
power consumption, multi-standard conformity and TV-GSM interoperability.
The chip area is much larger than [12] due to the reason that here a wider gain
range is realized, and additionally two I/Q mixer drivers and one gain-roll-off
compensation path are implemented. The mixer drivers are capable to drive both
resistive and capacitive loads with low gain and BW reduction.
4.5 Summary
This design example has demonstrated thatRF circuits reliably powered by an elevated
VDD are capable of achieving high performances with low power consumption and no
reliability risk. The presented proof-of-concept prototype is a 2 � VDD-enabled
mobile-TV RF front-end with TV-GSM interoperability. Verified in a 1-V 90-nm
CMOSprocesswith standard1-V thin-oxidedevices, the circuit coredraws 15mWat a
custom-elevated 2-V supply. In TV-GSM interoperation mode, an inductive-peaking
feedforward path evens out the passband to 0.7-dB variation, while showing 68-dB
rejection at the GSM-900 uplink. The presented stress-conscious circuit architectures
and self-bias techniques are generally applicable for different designs. It is believed
thatHV-enabledcircuitswithdesign-for-reliability possess a highpotential inboosting
RF circuit performances in sub-1 V technologies at low cost.
References
1. P.-I. Mak, S.-P. U and R. P. Martins, “Transceiver Architecture Selection – Review, State-of-
the-Art Survey and Case Study,” IEEE Circuits and Systems Magazine, Issue 2, pp. 6–25,
Jun. 2007.
2. Vassilios, K. Vavelidis and N. Haralabidis, et al., “A 65 nm CMOS Multistandard, Multiband
TV Tuner for Mobile and Multimedia Applications,” IEEE J. Solid-State Circuits, vol. 43,pp. 1522–1533, Jul., 2008.
3. D. Im, H.-T. Kim, and K. Lee, “A CMOS Resistive Feedback Differential Low-Noise
Amplifier with Enhanced Loop Gain for Digital TV Tuner Applications,” IEEE Trans. onMicrowave Theory and Techniques, vol. 57, no. 11, pp.2633–2642, Nov. 2009.
4. J.A. Weldon, R. S. Narayanaswami and J. C. Rudell, et al., “A 1.75-GHz Highly Integrated
Narrow-Band CMOS Transmitter with Harmonic-Rejection Mixer,” IEEE J. Solid-StateCircuits, vol. 36, no. 12, pp. 2003–2015, Dec. 2001.
5. C.-Y. Cha, H.-B. Lee, and K. K. O, “A TV-Band Harmonic Rejection Mixer Adopting a gm
Linearization Technique,” IEEE Microw. and Wireless Compon. Lett., vol. 19, no. 9,
pp. 563–565, Sept. 2009.
6. V. Rambeau, H. Brekelmans, M. Notten, K. Boyle and J. V. Sinderen, “Antenna and input
stages of a 470–710 MHz silicon TV tuner for portable applications,” in Proc. of Europ. Solid-State Circuits Conf. (ESSCIRC), pp. 239–242, Sept. 2005.
7. Mobile and Portable DVB-T/H Radio Access-Parts 1 and 2: Interface Specification andInterface Conformance Testing, International Standard IEC 62002-1, IEC, Oct. 2005.
8. Datasheet of TDK Low Pass Filters for DVB-H/ISDB-T DEA Series DEA200710LT-1238A1.
78 4 A High-Voltage-Enabled Mobile-TV RF Front-End in 90-nm CMOS
9. M.-C. Kuo, S.-W. Kao, C.-H. Chen, T.-S. Hung, Y.-S. Shih, T.-Y. Yang and C.-N. Kuo, “A
1.2 V 114 mW Dual-Band Direct- Conversion DVB-H Tuner in 0.13 mm CMOS,” IEEE J.Solid-State Circuits, vol. 44, no. 3, pp. 740–750, Mar. 2009.
10. D. J. Allstot, L. Xiaoyong, and S. Shekhar, “Design Considerations for CMOS Low-Noise
Amplifiers,” IEEE Radio Frequency Integrated Circuits (RFIC) Symp., pp. 97–100., Jun.2004.
11. P.-I. Mak and R. P. Martins, “Design of an ESD-Protected Ultra-Wideband LNA in Nanoscale
CMOS for Full-Band Mobile TV Tuners,” IEEE Trans. on Circuits and Systems – I: RegularPapers, vol. 56, no. 5, pp. 933–942, May 2009.
12. L. Tripodi and H. Brekelmans, “Low-Noise Variable-Gain Amplifier in 90-nm CMOS for TV
onMobile,” in Proc. of Europ. Solid-State Circuits Conf. (ESSCIRC), pp. 368–371, Sept. 2007.13. K. Ishida, A. Tamtrakarn, T. Sakurai and H. Ishikuro, “An Outside-Rail Opamp Design
Targeting for Future Scaled Transistors,” in Proc. of Asian Solid-State Circuits Conf.(A-SSCC), pp. 73–76, Nov. 2005.
14. Y.-W. Hsiao and M.-D. Ker, “A 5-GHz Differential Low-Noise Amplifier with High Pin-to-
Pin ESD Robustness in a 130-nm CMOS Process,” IEEE Trans. on Microwave Theory andTechniques, vol. 57, no. 5, pp.1044–1053, May 2009.
15. J. Hu, M. Felder and L. Ragan, “A Fully Integrated Variable-Gain Multi-tanh Low-Noise
Amplifier for Tunable FM Radio Receiver Front-End,” IEEE Trans. on Circuits and Systems –I: Regular Papers, vol. 55, no. 7, pp. 1805–1814, Aug. 2008.
16. L. Larcher, D. Sanzogni and R. Brama, et al., “Oxide Breakdown After RF Stress: Experimen-
tal Analysis and Effects on Power Amplifier Operation,” in Proc. of Int. Reliability PhysicsSymp., pp. 283–287, Mar. 2006.
17. Datasheet of Mini-Circuits, JTX-2-10 T, RF Transformer.
18. Datasheet of Anadigics, AWS5523, SP3T Switch.
19. G. Brzezina, L. Roy, and L. MacEachern, “Design Enhancement of Miniature Lumped-Element
LTCC Bandpass Filters,” IEEE Trans. on Microwave Theory and Techniques, vol. 57, no. 4,pp. 815–823, Apr. 2009.
20. K. Huang and T. Chiu, “LTCC Wideband Filter Design with Selectivity Enhancement,” IEEEMicrow. and Wireless Compon. Lett., vol. 19, no. 7, pp. 452–454, Jul. 2009.
21. J. Xiao, I. Mehr and J. Silva-Martinez, “A High Dynamic Range CMOS Variable Gain
Amplifier for Mobile DTV Tuner,” IEEE J. Solid-State Circuits, vol. 42, p. 292–301, Feb.,2007.
22. T. Kim and B. Kim, “A 13dB IIP3 Improved Low-Power CMOS RF Programmable Gain
Amplifier using Differential Circuit Trans-conductance Linearization for Various Terrestrial
Mobile D-TV Applications,” IEEE J. Solid-State Circuits, vol. 41, pp. 945–953, Apr., 2006.
References 79
Chapter 5
A Mixed-Voltage Unified Receiver Front-End
for Full-Band Mobile TV in 65-nm CMOS
5.1 Introduction
With improved device parasitics in nm-length CMOS processes, wideband RF
circuits offer the desired compactness and power efficiency for realizing multi-
band multi-standard radios. This chapter describes a receiver front-end (RFE)
targeting the mobile-TV applications using mixed-voltage techniques. It covers
the VHF-III (174–248 MHz), UHF (470–862 MHz) and L (1.4–1.7 GHz) bands,
where standards like T-DMB, ISDB-T, DVB-H and DMB-T are resided. In order to
meet the noise and linearity specifications [1], while avoiding external baluns or
repeated RFEs that were still common in existing solutions [1, 2], a number of
circuit techniques are proposed to enhance the performance, power and area
efficiencies [3]. Together they lead to state-of-the-art performance, while saving
58% area compared to the 1.1 mm2 reported in [2]. The key design considerations
are outlined as follows.
A wideband balun low-noise amplifier (LNA) can nullify the cost and insertion
loss of external balun, but might suffer from low IIP2 and weak output balancing
compared to its fully-differential counterpart. The output balancing of a balun-LNA
significantly affects the IIP2 of the mixers following it. The common-gate (CG)
common-source (CS) balun-LNA [4] serves as an example. Although noise cancel-
lation and admittance scaling techniques can lower the NF to <3 dB, the imbalance
load and bias current between the CG and CS branches weakens the IIP2 and output
balancing. Output buffers were employed to alleviate those issues at the expense of
extra power. Alternatively, in [5], IM2 cancellation feedback was introduced to
enhance IIP2. This technique, however, penalizes the output bandwidth (BW) and
noise figure (NF), while the problem of weak output balancing remains unsolved. In
this work, a gain-boosting current-balancing balun-LNA is proposed. It not only
addresses the aforementioned issues, but also offers adequate input matching over
variable gain control (VGC). The VGC assists the balun-LNA in handling high
input power levels.
P.-I. Mak and R.P. Martins, High-/Mixed-Voltage Analog and RF Circuit Techniquesfor Nanoscale CMOS, Analog Circuits and Signal Processing,
DOI 10.1007/978-1-4419-9539-1_5, # Springer Science+Business Media New York 2012
81
Harmonic mixing and intermodulation distortion are critical concerns of wideband
RFEs. The former is due to the hard switching nature of MOS mixers, i.e., blockers
located at the harmonics of the LO become co-channel interferers after
downconversion. The latter is due to the potential co-existence of large amount of
blockers throughout the passband. Regarding this, especially for interoperableterminals as shown in Fig. 5.1, external SAW filtering is still the most reliable
solution [6]. The cellular transmitter can deliver power as high as 30 dBm whereas
the isolation between the internal terminals is just around 10–15 dB. If uninterrupted
operation of cellular and tuner is desired, external SAW filtering is highly necessary
to avoid saturation. As shown in Fig. 5.2, the SAW filters (e.g., [7]) can specifically
notch to the very nearby cellular bands.
VHF III
UHF
L
SAW
SAW
SAW
SAW
LimitedIsolation
WCDMA
GSM WiFi
TV
BT
FM
Co-Integration Terminal
Fig. 5.1 Limited isolation in co-integration terminal
UHF L
170 245 470 860 1400 1700
3 x VHF III 5 x VHF III
VHF III
3 x UHF
2580
8-PhaseHR Mixing
8-PhaseHR Mixing
4-PhaseQuad. Mixing
[MHz]
SAW
GSM 900
GSM 1800
WiFiWCDMA
SAW SAW
Wideband Balun-LNA
Fig. 5.2 Overview of SAW filtering, balun-LNA’s bandwidth and mixing strategies
82 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
With adequate pre-filtering the linearity of the RFE will be mainly limited by
the out-of-channel IIP2 and IIP3 [7]. Here, the mixer is merged with a current-mode
third-order lowpass filter (LPF) for improving the out-of-channel linearity,
while saving power and area for its simplicity. A third-order stopband rejection
profile suffices for shifting most baseband (BB) functions to the digital domain via
low-power analog-to-digital converters [2, 7].
When receiving the VHF-III and UHF bands, harmonic-rejection (HR) mixers
[8] can relax the rejection profile of the external SAW filters, and minimize noise
and distortion folding (i.e., the third and fifth harmonic bands are partially within
the BW of the balun-LNA as marked in Fig. 5.2). The entailed 8-phase LO, if
generated via frequency dividers, will necessitate a high master LO frequency (fLO).In fact, the frequency range (�0.9 GHz) of the RFE in [9] is limited by the required
high fLO (7.2 GHz) for generating the 8-phase LO, which significantly increases the
power and complicates the design of the phase locked-loop (PLL) and voltage-
controlled oscillator (VCO). In this work, a direct injection-locking 4-/8-phase LO
generator (LOG) lowers the master fLO. This dividerless solution particularly suits
this receiver-only application, as the problem of “VCO pulling” is irrelevant here in
the absence of power amplifier.
This chapter is organized as follows. Section 5.2 introduces the proposed RFE
architecture, and discusses the technology features which play a key role in
device sizing and topology selection. The conventional and proposed balun-LNAs
are described in Sect. 5.3. Sections 5.4 and 5.5 present the proposed current-reuse
mixer-LPF and the direct injection-locked 4-/8-phase LOG, respectively. In Sect. 5.6,
the experimental results will be presented.
5.2 RFE Architecture and Technology Features
Figure 5.3 depicts the proposed RFE architecture headed by three external SAW
filters. It integrates a wideband balun-LNAwith VGC, dual I/Q mixers, a 4-/8-phase
LOG and dual third-order BB LPFs merged with the mixers for current-reuse and
WidebandBalun-LNAwith VGC
BBout,I
BBout,Q
LOin
Current Reuse
4-/8-Phase LOG
1: 2 :1
IC
3rd-Order LPF
Lext
VHF III
UHF
LSAW
SAW
SAW
Fig. 5.3 Proposed full-band mobile-TV RFE
5.2 RFE Architecture and Technology Features 83
current-modefiltering.One external inductor is entailed at theRF input.Thedifferential
signals from the balun-LNA are ac-coupled to the mixers to prevent the balun-
LNA’s even-order distortion products (i.e., the concern of narrowband IIP2 [10])
and 1/f noise from leaking to the BB. The mixer has two operating modes: (1)
HR mode supports the VHF III and UHF bands, using a gain ratio of 1:√2:1among the three mixer cells and an 8-phase LO [8]; (2) quadrature modesupports the L band where HR is a surplus as the harmonic bands are far out
from the band of interest. In this case, the three mixer cells are parallelized and
the LO is switched to 4 phases. A direct injection-locking LOG synthesizes the
desired 4-phase and 8-phase LOs.
In nm-length CMOS processes the back-end features and options are highly
related to the performances because of the parasitic effects. Here, the available dual
supplies (1.2 and 2.5 V) and dual-oxide transistors (thin and thick) are employed to
optimize the performances of the RFE. Dual-supplies free more design headroom in
65-nm CMOS SoC [1] and beyond [11].
The employed I/O pads are offered by the foundry. They have passed the 2-kV
Human body Mode (HBM) ESD tests, but add roughly 1-pF input capacitance that
has to be taken into account in the design phase.
5.3 Wideband Balun-LNA
5.3.1 CG-CS Noise-Canceling Balun-LNA
Figure 5.4depicts its schematic. The single-to-differential (S2D) utilizes aCGamplifier
(M1) to generate the in-phase output. The source ofM1 is dc-grounded via an external
big inductor Lext to attain a wideband impedance match. The corresponding
Vb1
C1Lext
Vin
(gm1) M1
M2(gm2=ngm1)
VDD
von
vonvIN
vop
vop
RCG
RCG
Rin ≈ 1/gm
IDC nIDCCL CLisig+
isig+
isig+ = gm1Vinisig- = -ngm1Vin
isig-
isig-
RCS=RCG/nOutput balancing
at DC only
S2DS2D
RCG/n
Fig. 5.4 CG-CS noise-canceling balun-LNA [4]
84 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
transconductanceofM1 (gm1) is 20mS, given thatRin � 1/gm1 ¼ 50O. ACS amplifier
(M2) generates the anti-phase output, and assists noise canceling of M1. The noise
contribution ofM2 can be reduced by admittance scaling, e.g., gm2 ¼ 4gm1 ¼ 80mS.
To keep a balanced output in [4] a counterbalance load was set (RCS ¼ RCG/4). This
technique regrettably cannot ensure a wideband balanced output when driving the
mixers (with the input impedance modeled as CL). This fact, in addition to the bias
current mismatch between the CG and CS branches, leads to low IIP2, weak output
balancing and limited power supply rejection ratio (PSRR). Extra output buffers only
can reduce the output imbalance at the expense of power, but the problem of low IIP2
remains unsolved.
Another concern of such a balun-LNA is that the S2D cannot be simply bypassed
at high input power level. Tuning the load (RCG and RCS) or current steering-based
VGC does not help much when the linearity is limited by the nonlinearity ofM1 and
M2. This issue will be addressed by the proposed balun-LNA.
5.3.2 Proposed Gain-Boosting Current-BalancingBalun-LNA with VGC
The Single-to-Differential Stage (S2D) – The proposed balun-LNA (Fig. 5.5a) uses
a new S2D. The first idea is to use an ac-coupled gain stage (gmx) to enhance
the gain of the CS branch (Fig. 5.5b), avoiding scaling-up of gm2 with respect to gm1
(i.e., M1 and M2 are of equal size and bias current). The second idea is to reuse the
gain generated by M2 to enhance the gain of the CG amplifier by creating a loop
gain (1 + A) aroundM1, where A is given by │vo1n/vin│ (Fig. 5.5c). Due to the single-stage structure of gmx + gm2, and the low impedance of vo1n, A is ensured to
be stable. This loop gain A not only boosts the effective gain of M1 and lowers its
noise contribution, but also reduces the minimum IDC for M1 to offer input
impedance matching. An external big inductor Lext simplifies the input impedance
match and offers a DC-current path for the CG amplifier M1.
In order to simplify the bias and save component count, gmx is realized as an
inverter amplifier (M3 and M4) with resistive feedback Rfb for self-biasing. This
self-bias structure also handily biases M1-M2 as shown in Fig. 5.5c. Comparing
with the S2D topology in [4], just one capacitor C2 and one resistor Rfb are added as
M3-M4 can be counted as parts of M2.
The Differential Current Balancer (DCB) – Balanced output impedance is critical
to achieve wideband output balancing. Here, the differential balancing is realized in
the current domain by inserting a DCB between the S2D and the load as shown in
Fig. 5.6. The DCB serves as a differential current-control current source with unity
gain, forcing isig+ ¼ �isig� ¼ isig. In principle, a cascode amplifier (M5-M6) with
cross-coupled capacitors (C3-C4) might realize the DCB [12], as what it amplifies is
the difference of the input. However, accounting for the finite output resistance of
M5-M6, a double-cascode amplifier (M5-M8, C3-C6) is adopted to achieve wideband
output balancing.
5.3 Wideband Balun-LNA 85
TheDCB also benefits the S2D in two ways: (1) the condition isig+ ¼ �isig� ¼ isighelps mapping the loop gain of the CG branch to the gain given by the CS branch,
yielding A ¼ gmx/gm1, which is a definable ratio insensitive to process if the channel-
length modulation of M1-M4 is negligible. (2) The DCB improves the balun-LNA’s
reverse isolation and linearity by lowering the swing at vo1p and vo1n, where distortionarising from the nonlinear output resistance ofM1-M4 can be minimized.
The overhead of the DCB is the extra voltage headroom when compared with a
typical cascode stage. The use of dual supplies (1.2 and 2.5 V) overcomes this
limitation and allows more voltage gain without compromising the dynamic range
(DR). A high-voltage design also offers opportunities to realize more functions in
the current domain by cascoding them under one supply rail, potentially leading to a
wider RF BW (i.e., less high impedance nodes), better linearity (i.e., small internal
signal swing) and area savings (i.e., less ac-coupling capacitor between blocks).
Nevertheless, a high-voltage design with thin-oxide MOS requires node-voltage
trajectory checks, to ensure no gate oxide and p-n junction are overstressed at
transient and steady states. The bias scheme of the balun-LNA (Fig. 5.7) is designed
isig+ = (gm1(1+ )vin
Rin ≈1/[gm1(1+ )]
S2D DCB
IDC
VDD
VDD
VDD
IDCIDC
IDCisig+ isig-
-gmxVin
-gmxVin
-gm1vin
-gm1Vin
Vb1
C2
C2
M2 (gm1)
M2 (gm1)
C1
C1
RL
RL (gm1) M1
(gm1) M1
Lext
Lext
Vin
Vin
isig+ = gm1Vinisig- = -(gmx + gm1)Vin
isig- = -(gmx + gm1)Vin
Rin ≈ 1/gm
gmx ≈ gm3 + gm4
vo3n
vo1n
vo1n
vin
vo1n
vin
vINvo3p
vo1p
isig+
isig+
isig-
isig-
gmx
gmx
gmx
M3
M4Rfb
a b
c
Fig. 5.5 Generation of the proposed S2D: (a) the overall balun-LNA. (b) Ac-coupled gain
boosting avoids mismatch of IDC. M1 and M2 are of equal size and bias current. (c) The final
topology: gain-boosting the CG via the CS, where the gain-boost stage (gmx) is a self-biased
inverter amplifier (M3 and M4)
86 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
as a single-ended replica of it, generating the bias voltages Vb1 and Vb2 for M5-M8.
Such a scheme can detect the presence of VDD25 and VDD12, ensuring no device is
overstressed in power-up/down conditions. The reliability levels are guided by the
foundry design rule manual. The scheme ensures all terminal voltages meet the
reliability guidelines provided by the foundry.
Combining the S2D and DCB, the single-to-differential voltage gain Av,diff and
input impedance Zin of the balun-LNA can be obtained
Av;diff ¼ vo3p � vo3nvin
¼ 2gm1 1þ Að Þ RL==1
sCL
� �: (5.1)
Zin ¼ 1
gm1 1þ Að Þ ==sLext==1
sCin: (5.2)
respectively. The demand of both high Av,diff and low NF promotes setting a smaller
Zin via increasing gm1, which is, however, constrained by the desire of input
impedance matching. The minimum acceptable input reflection coefficient magni-
tude Gin can be determined as
Ginj j ¼ Zin � Rs
Zin þ Rs
��������� 0:32: (5.3)
1/gm5
S2D
DCBVDD25
VDD12
Vss12
RL RLIDC IDC
vo3nvo3n
vo2n
vo3pvo3p
vo1p
vo2p
CL CL
vo1n
vo1n
vo1p
isig
isig-isig+
(gm1) M1C2
C3 C4
C5 C6
C1
gmx
M2 (gm1)Rin
Vin
ZinLext Cin
Vb1 Vb1
Vb2 Vb2
M5 M6
M7 M8
(gm5)
(gm5)
Fig. 5.6 The S2D combined with the DCB. M1 and M2 are of equal size and bias current
5.3 Wideband Balun-LNA 87
It corresponds to a S11 value of < �10 dB. Accordingly, the tolerable ZIN for
RS ¼ 50O ranges from 26 to 97O. In this work, Zin is set to 33O (gm1¼7.5 mS and
gmx ¼ 3gm1) to enhance Av,diff and lower NF, while yielding S11 < �13 dB (3 dB
margin). Though acceptable here for the targeted frequency (0.17–1.7 GHz) in the
employed 65-nm CMOS process, it should be noted that the impedance match BW
is reduced when comparing it with a strict 50-O match.
TheGain-ControlAttenuator (ATT)–Thebalun-LNA is preceded by annMOS-onlyR-
2R network for coarseVGCas depicted in Fig. 5.8. It exploits the properties of partial
signal reflection (i.e., Zin deviates fromRS) and voltage division to realize the desired
attenuation levels. The control range was set to 18 dB, which offers a reasonable
tradeoff between IIP3 andNF at high input power levels.A 6-dB step size is related to
the gain-tuning characteristic of R-2R network when constant input impedance is
sought. The gain switching is executed via Sbypass, S1, S2 or S3 as shown in Fig. 5.9.Operating at the low-impedance input node of the balun-LNA with Lext provides thedc-ground, each nMOS enjoys a large VGS of 1.2 V for high linearity, while offering
VGC with a constant output BW. At the maximum gain, the ATT is bypassed with
Sbypass; it features a 1.8-O on-resistance (rON) under a practical device size of 200/0.06. The combined R0
in varies between 34.8 O (ATT bypassed) and 55 O (with the
ATT). Both values yield S11 < �13 dB, giving room for process variation. The
initial �2.2-dB attenuation at the highest gain is induced by the reflection and
division between ron and Rin (i.e.,�2.2 dB ¼ 20 log {[1 � (RS � R0in)/(Rs + R0
in)] ·
Rin/(ron + Rin)}.
Overdrive Protection
RLB
lb
Vb2
M4b
M3b
M1b
Mpt1
M2b
Vb1
VDD25
VDD12
VDD25 VDD12
Fig. 5.7 The bias scheme
of the balun-LNA
88 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
All MOS switches are operated in triode region with an ON-resistance value that
is either RIN or 2RIN. The size of the former is obtained according to
1
mCOX
WSW
LSWðVGS;SW � VTHÞ
zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{NMOS Switch
¼ 1
1
2mCOX
WM1
LM1
ðVGS;M1 � VTHÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}gm1
1þ Að Þ:
zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{Rin
(5.4)
ATT
gmx
S1 S2 S3
Sbypass
53
Rin ≈1/[gm1(1+A)]
Rin’ = rON+Rin or Rin
ATT(33W)
(34.8 or 55 W)
Rs
RFin
Rin
Rin
RinRinRin
rON=1.8W
2Rin
Rin
LextVin
C1M2
C2
M1 VDD12
VSS12
vo1p vo1n
Fig. 5.8 The balun-LNA is preceded by a passive ATT for coarse gain control
~2.2 dB attenuation
1.8 W1.8 Ω
33 W
1.8 W1.8 W
33 Ω
33 Ω
66 Ω
66 Ω66 Ω
33 Ω
33 Ω33 Ω
33 Ω
33 Ω
33 Ω
33 W33 W33 W
33 W33 W
33 W
33 W 66 W
33 W
33 Ω33 Ω
33 Ω
33 Ω 33 W33 W 33 W
8.4 dB attenuation
14.4 dB attenuation 20.4 dB attenuation
dc shortedto ground
dc shortedto ground
dc shortedto ground
dc shortedto ground
Ry=61.28 W
Rin=33 W
Ry=49.5 W
Rin=33 W
Ry=1.8 W
Rin=33 W
Ry=22 W
Rin=33 W
Rin’ = 55 WRin’ = 55 W
Rin’ = 34.8 W Rin’ = 55 W
vin
vinvin
vin
IDC,x
IDC,x
IDC,xIDC,x
Fig. 5.9 Equivalent circuits of the nMOS-only ATT in each gain step
5.3 Wideband Balun-LNA 89
On the left of (5.4) VGS,sw � VDD12, since all switches are switched [OFF, ON]
by [0, VDD12]. On the right VGS,SW � VDD12/2, which is set by the self-biased
topology of gmx (see Fig. 5.5c). In the employed process, VDD12 is around 4� of
the device’s threshold voltage VTH. Accordingly, an aspect ratio of Wm1/WSW
¼ 3/2 can be obtained by assuming A ¼ 3 and LSW ¼ LM1. A similar concept
applies for the MOS switch with 2Rin ON-resistance. The scheme is acceptable for
coarse VGC as process variation will affect the accuracy of the step size.
It is noteworthy that the bias current of M1 has to be passed through the ATT.
Considering the impedance back to the ATT (labeled as Ry), the equivalent resis-
tance at each attenuation step varies between 1.8 and 61.28 O. The source of M2,
thus, employs an ATT replica to track this variation (Fig. 5.8), matching the bias
currents of the CG and CS branches against VGC.
Performance Optimization via Using Dual VDD’s – The NF of the balun-LNA is
analyzed based on the simplified noise model shown in Fig. 5.10. For simplicity,
channel-length modulation is neglected and Ms is used to represent the equivalent
of M2-M4 in parallel. The transconductance gms of Ms correspondingly represents
gm2 + gm3 + gm4. Regarding a and g showed in Fig. 5.10 they are the process- and
bias-dependent parameters of the devices, respectively. The differential output
noise voltages due to each source are obtained as,
v2n;Rs;diffout ¼ Vn;RsgmsRL
�� ��2 (5.5)
v2n;M1;diffout ¼ 0 (5.6)
v2n;Ms;diffout ¼ In;Ms
2RL
1þ gmsRS
��������2 (5.7)
v2n;M5;diffout ¼ In;M5
gm1 1þ A0� �
gm1 1þ A0� �þ 1
RS
� 1
2
" #RL
����������2
(5.8)
v2n;M7;diffout ¼ 0 (5.9)
v2n;RL;diffout ¼ In;RLRL
�� ��2 (5.10)
where the loop gain around the CG branch is re-defined as 1 + A0 to match the set
assumption in this Section.With gm1 ¼ gm2, 1 + A0 is relatedwith a gm ratio given by,
1þ A0 ¼ 1þ gm3 þ gm4
gm1
(5.11)
90 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
The NF of the balun-LNA can be calculated by dividing the total output noise
power due to all devices by the noise power of RS leading to,
NF ¼ 1þ
v2n;M1;diffout þ v2n;Ms;diffout þ 2v2n;M5;diffout
þ2v2n;M7;diffout þ 2v2n;RL;diffoutv2n;Rs;diffout
(5.12)
If the input impedance match is exact (Rin ¼ RS), we can set gm1(1 + A0) ¼ gms
¼ 1/RS to simplify (5.12) as,
NF ¼ 1þ gaþ 2RS
RL: (5.13)
Equation 5.13 is the same as that of CG-CS noise-canceling balun-LNA withoutadmittance scaling: gmCG ¼ gmCS ¼ 1/RS [12]. The main difference here is the
reduction of the minimum Idc for 50-O matching by a factor of 1 + A. A smaller Idcmight compromise the linearity, but the presence of the DCB helps isolating the
In,RL
In,M7
2
2
In,M52
In,M72
In,M12
In,Ms2
Vn,Rs2
VDD
RL RL
RL
M5
M1
MS
MS = M2//M3//M4gms = gm2+gm3+gm4
M6
C3 C4
RS
C5 C6
M7 M8
= 4KTgm7 ag
In,M52
= 4KTgm5 ag
In,Ms2
= 4KTgms ag
In,M12
= 4KTgm1
Vn,Rs2
= 4KTRS
ag
In,RL2
=4KTvo3nvo3p
Fig. 5.10 Simplified noise model of the proposed balun-LNA. g is the excess channel thermal
noise coefficient. a is gm/gd0, with gm the device transconductance and gd0 the zero-bias channel
conductance
5.3 Wideband Balun-LNA 91
high-swing output node from the S2D, reducing the effect of S2D’s nonlinear
output resistance. For a conservative value of g/a ¼ 4/3, the NF is >3 dB.
For that reason, as mentioned before, the targeted Rin is 33 O to achieve better
NF and gain. The factual value can be computed via (5.12) under Rin ¼ 33 O.The 2.5-V supply allows a bigger RL without degrading DR. In the size optimi-
zation the optimum range that can balance IIP2, IIP3 and NF was found to be within
340–570 O under a 150-fF load as shown in Fig 5.11. In this region, the NF is
<2.8 dB and the IIP2/IIP3 is > + 45/+3 dBm. RL was chosen to be 410 O for a
maximum IIP2. The in-band NF justified at boundaries of the desired BW (0.17 and
1.7 GHz) has a difference of<0.1 dB. The corresponding mid-band gain is 25.6 dB
and the output BW is around 2.6 GHz. The size of the components of the balun-
LNA is summarized in Table 5.1.
Startup Considerations – Mixed-voltage circuits raise the concern of improper
power-up/-down sequences. As depicted in Fig. 5.12, in case VDD,12 has a delay in
start-up with respect to VDD25,M1-M4 are in the cutoff region overstressed by VDD25.
The proposed solution is to add two small-size thick-oxide PMOS transistors Mpt1
and Mpt2 to vo1p and vo1n, respectively. With them, the dc voltages at vo1p and vo1ncan be controlled to be within VDD12 when M1-M4 are still in cutoff region. Once
VDD12 ramps successfully, Mpt1 and Mpt2 are automatically turned off. A similar
operation holds in power-down condition when VDD12 is removed before VDD25.
2
2.3
2.6
2.9
3.2
3.5
200 700 1200 1700
NF at 170 MHz
RL (Ω)
NF
(dB
)
IIP2,
IIP
3 (d
Bm
)
@ CL = 150 fF
NF at 1.7 GHz
OptimumWindow
IIP3
IIP2
-15
0
15
30
45
60
Fig. 5.11 Simulated IIP2, IIP3 and NF with respect to RL. The optimum window of RL is
highlighted
Table 5.1 Components size
of the balun-LNAValue Value
M1, M2 2.4/0.07 � 6 C1 4 pF
M3 (for gmx) 2.4/0.06 � 20 C2 6 pF
M4 (for gmx) 2.4/0.06 � 40 C3–C6 2 pF
Rfb (for gmx) 15 kO CLa 150 fF
M5–M8 2.4/0.18 � 6 RL 410 OLext 100 nHaFor simulation only
92 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
Differential Balancing – The progression of gain and phase balancing within the
balun-LNA are assessed as shown in Fig. 5.13a and b, respectively. The gain-phase
imbalances are corrected progressively from [vo1p, vo1n] to [vo2p, vo2n] after the firstbalancing, and eventually to [vo3p, vo3n] after the second balancing. The in-band
gain and phase mismatches are <0.037 dB and <1.87�, respectively. From 0.1 to
10 GHz, the former is still acceptably small (<0.2 dB), whereas the latter is limited
at the low frequency side (<3.5�) due to the highpass characteristic of C3-C6 (2-pF
MiM capacitors) that were not oversized due to the add-on parasitics. The output
balancing can be affected by device mismatches. A matching-conscious layout is
entailed between M1 and M2, the differential RL and DCB.
PSRR – The voltage gain from each supply and ground rail to the differential output
(Vo3p�Vo3n) has been simulated as shown in Fig. 5.14. Due to a symmetrical RL and
the equal (and high) output impedance of the DCB, the PSRR with respect to VDD25
is > 48.1 dB. This is a significant result as VDD25 is generally just utilized for I/Os
featuring weaker quality when comparing it with the core supply. Differently, the
PSRR with respect to VDD12 and VSS12 are just 13.5 and 12.75 dB, respectively.
Multiple pads were therefore assigned between VDD12 and VSS12 to minimize noise
coupling and the effect of bondwire inductance on performance degradation.
VGC on Performances – Accounting a bondwire inductance of roughly 3.5 nH, anda 1-pF input parasitic capacitance due to the bondpad and ESD protection circuitry,
the simulated AC responses, S11 and NF against gain control are shown in
Fig. 5.15a–c, respectively. S11 < �10 dB is achieved up to 4 GHz. For the linearity
against VGC (Fig. 5.16), from the highest to lowest gain levels, the IIP2 ranges
Automatic Overdrive Protection for M1– M4
0t
V and I When VDD25 ramps up beforeVDD12 M1 - M4 are cutoff
vo1p
vo1p,n
vo1n
VDD25
VDD25
Mpt1
Mpt1,2 ensures vo1p,n< VDD12
Mpt1,2 is cutoff whenin normal operation
IMpt1
IMpt1,2
IMpt2
Mpt2
VDD25VDD12
VDD12
Fig. 5.12 A simple scheme for overdrive protection (the notations are referred to Fig. 5.6)
5.3 Wideband Balun-LNA 93
-1
0
1
2
3
4
5
6
0.1 1 10
Frequency (GHz)
Gai
n Im
bala
nce
(dB
)P
hase
Imba
lanc
e (D
egre
e)
-140
-120
-100
-80
-60
-40
-20
0
20
0.1 1 10
Frequency (GHz)
[vo2p, vo2n]
[vo2p, vo2n]
[vo3p, vo3n]
[vo3p, vo3n]
[vo1p, vo1n]
[vo1p, vo1n]
a
b
Fig. 5.13 Simulated gain (a) and phase (b) balancing progress inside the balun-LNA
-60
-50
-40
-30
-20
-10
0
10
20
30
0.1 1 10
Frequency (GHz)
20. lo
g(v o
3p-v
o3n)
(dB
)
From VDD25
From VSS12
From VDD12
From RFin
Fig. 5.14 PSRRs with respect to VDD25, VDD12 and VSS12
-30
-20
-10
0
10
20
30
0.1 1 10
Frequency (GHz)
Diff
. Vol
tage
Gai
n (d
B)
-40
-30
-20
-10
0
0.1 1 10Frequency (GHz)
S11
(dB
)N
F (
dB)
0
10
20
30
40
50
0.1 1 10
Frequency (GHz)
In Band NF@Highest Gain = 2.2 to 2.7 dB
Rin= 55Ω Rin= 34.8Ω
a
b
c
Fig. 5.15 Simulated balun-LNA’s (a) AC responses and (b) S11. (c) NF, against gain control
5.3 Wideband Balun-LNA 95
from +45 to +59 dBm, whereas the IIP3 ranges from +5.6 to +17 dBm. The former
is loosely related with the VGC. The latter is due to the ATT, which attenuates the
input signal before it reaches the S2D.
5.4 Current-Reuse Mixer-LPF
5.4.1 Circuit Description
Typically inwireless receivers, themixer andBBLPF are separately designed, and the
signal transfer is processed with in voltage (Fig. 5.17a). It implies that the out-of-
channel interferers are amplified at the output of the mixer prior to adequate filtering.
Here, with the use of a 2.5-V supply, more voltage headroom promotes the use of a
current-mode LPF to interface (cascode) with the mixer (Fig. 5.17b), offering ade-
quate filtering prior to BB I-V conversion. Figure 5.18 shows the block diagram of the
proposed I/Q mixer-LPF with switchable modes between HR mixing and simple
quadrature mixing. The circuit implementation is shown in Fig. 5.19, a polyphase
double-balanced structure is employed to lower the LO-to-BB leakage and even-order
distortion while rejecting the third and fifth harmonics of the LO. the HRmixer core is
based on a switched-gm cell [13], which is favored against the Gilbert cell in terms of
operating voltage and LO amplitude requirement.Moreover, the LOpath induces only
common-mode noise at the outputs that can be rejected differentially. Also due to the
2.5-V supply,Mm1-Mm4 can feature a large VGS, and use source degeneration (Rdeg) to
enhance their linearity and output resistance.
The LO buffers realized as CMOS inverters use cross-connected weak latches to
minimize the duty-cycle distortion. Small LO buffer’s ON-resistance reduces
the RF-to-LO coupling since Mm1-Mm4 can be well-grounded when they are in
0
5
10
15
20
25
30
5 15
Voltage Gain (dB)
40
45
50
55
60
0.9 & 1 GHz
2-Tone Test @
25
IIP3
(dB
m)
IIP2
(dB
m)
Fig. 5.16 Simulated IIP2 and IIP3 against gain control
96 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
BB
LO
RF
VoltageDomain
VoltageDomain
LO
RF
CurrentDomain
VoltageDomain
I/V Conversion
RBB
BB
a b
Fig. 5.17 (a) Voltage-mode mixer cascades with voltage-mode LPF. (b) Current-mode mixer
cascodes with current-mode LPF
0° 180°
-45° 135°
45° 225°
90°
2gm
gm
gm
gm
gm
270°
45° 225°
135° 315°(90° 270°)
(90° 270°)
(0° 180°)
(0° 180°)
Switchable LOI Waveform
Switchable LOQ Waveform
I-to-V
I-to-V
Current-Mode 3rd
Order LPF
Current-Mode 3rd
Order LPF
Voutp,Q
Voutp,I
Voutn,Q
Voutn,I
VRFn
VRFp
2gm
Fig. 5.18 Block diagram of the proposed I/Q mixer-LPF
5.4 Current-Reuse Mixer-LPF 97
Current-Mode Biquad LPF: HBiq(s)
2gm,eff
(8-Phase LO)(4-Phase LO)
LO Buffers
SwitchablePhase
from LOG
2
EquivalentCircuit
VDD25
RBBCBB /2
Cf2 /2
Cf1 /2
Cf1 /2
L=Cf2 /gmf2L=Cf2 /gmf
RBBML1 ML2
VO3p,BB
VO1p,BB
Rdeg
Vb
Vo2n,BBVo2p,BB
Vb
iout(s)
iin(s)
iout(s)
iin(s)
Mf3 (gmf)
Mf1 (gmf)
(gmf) Mf4
(gmf) Mf2
Mm1Mm2 Mm3Mm4
VO3n,BB
VO1n,BB
Vinn
VinpVinp
gm,eff
gm,eff
3rd-OrderLPF
135°180°135°180°180°180°
-45°0° 45°0° 0° 0°
a
b
Fig. 5.19 (a) Circuit implementation of the proposed current-reuse I/Q mixer-LPF. (b) the
schematic and equivalent circuit of the HBiq(s)
98 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
the ON-state. Assuming a square-wave-like LO, the conversion gain (CG) in
4-phase and 8-phase operations are given by
CG4PðsÞ ¼ cgm;eff 2þffiffiffi2
p RBB==
1
sCBB
� �HBiqðsÞ (5.14)
CG8PðsÞ ¼ cgm;eff 2ffiffiffi2
p RBB==
1
sCBB
� �HBiqðsÞ (5.15)
where c ¼ 2/p, accounting that only the first harmonic of the LO contributes to the
gain. On the other hand, gm,eff denotes the effective transconductance of Mm1-Mm4
after resistive degeneration, i.e., gm,eff ¼ gm,Mm1/(1 + gm,Mm1Rdeg). CG4p(s) intrinsi-
cally shows 1.6-dB higher gain than CG8p(s) since the polyphase paths in the 4-phase
mode are directly parallelized. This gain difference is indeed helpful as the L band
that uses a 4-phase LO is located much close to the �3-dB cutoff frequency of the
balun-LNA, where more gain droop exists. HBiq(s) is the frequency response of a
current-mode Biquad [14] with Q and cutoff frequency o0 as given by,
HBiqðsÞ ¼ ioutðsÞiinðsÞ ¼
g2mf
Cf1Cf2
s2 þ sgmf
Cf1
þ g2mf
Cf1Cf2
Q ¼ffiffiffiffiffiffiffiCf1
Cf2
ro0 ¼ gmfffiffiffiffiffiffiffiffiffiffiffiffiffi
Cf1Cf2
p (5.16)
This Biquad exhibits several distinct advantages [14]: (1) the Q of the complex
pole pair is insensitive to process; (2) there is no passband loss as the current gain at
dc is unity; (3) a constant-Q BW control can be achieved with C1 and C2 tuned
together; (4) the Biquad exhibits a zero in the noise transfer function of Mf1-Mf4
(Fig. 5.20a), resulting in in-band noise reduction. This property breaks the funda-
mental kT/C limit of the classic LPFs where the noise behavior should follow the
LPF’s profile (Fig. 5.20b); (5) high linearity is achieved since filtering is performed
in the current mode, while Cf1 filters out the high-frequency signals prior to
reaching the active elements (Mf1-Mf4); (6) the noise and linearity are tradable
with the cutoff frequency. A higher cutoff rejects more in-band noise because of the
zero exhibited at DC (Fig. 5.20c).
These good noise and linearity properties render it as a wise “first filter” in
direct-conversion receivers. Here, by integrating it with the mixer having a real pole
load [RBB//(1/sCBB)], a third-order LPF can be constructed without extra bias
currents, V-I and I-V converters that dominate the NF [14]. The BB output
common-mode voltage is controlled via a common-mode feedback circuit around
ML1-ML2. The targeted cutoff frequency is 12 MHz to balance the NF and selectiv-
ity. An automatic cutoff tuning is beyond the scope of this work, but should be
considered when accounting for process and temperature variations. A constant-QBW tuning can be achieved via co-tuning Cf1, Cf2 and CBB. The transistorsMf1-Mf4
are realized as thick-oxide MOS with a big device size to minimize NF and reduce
channel-length modulation. The latter is essential for an accurate filtering profile.
5.4 Current-Reuse Mixer-LPF 99
The protection circuit depicted in Fig. 5.12 is also applied here at vo1p,BB and
vo1n,BB to prevent Mm1-Mm4 (thin-oxide devices) from being overstressed during
the start-up/power-down condition. The bias circuit of the mixer-LPF is shown in
Fig. 5.21. The component sizes are given in Table 5.2.
5.4.2 Noise Figure
The simplified noise model of the mixer-LPF is shown in Fig. 5.22. Similar to the
noise analysis in Sect. 5.3, it can be derived that the single-side band (SSB) NF is
given by
NFSSB;MixLPFðsÞ ¼ 1
c2þ 2ggm;effRSc2
þ2
RBB
Rsc2g2m;effH2BiqðsÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
switched�gm mixer contribution
þ
2ggmf
sgmf
Cf2
s2 þ sgmf
Cf1
þ g2mf
Cf1Cf2
2664
37752
RSc2g2m;effH2BiqðsÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
LPF contribution
þ2ggmf
sCf2
sCf2 þ gmf
� �2
RSc2g2m;effH2BiqðsÞ|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
LPF contribution
(5.17)
“kT/C limit”
Flicker
DesiredSignal
Noise ωout-off1
Noise ωout-off2
Blockers
High ωout-off2reduces noise
Low ωout-off2rejects more Blockers
ωcut-off
Filter Profile
Noise
DesiredSignal
ωωpole ω
ωout-off2ωout-off1 ω
a b
c
Fig. 5.20 (a) In-band noise characteristic of Gyrator-C Biquad. (b) In-band noise characteristic of
generic filter. (c) Cutoff frequency poses a tradeoff between in-band noise and out-of-channel
linearity
100 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
where CBB is omitted and the noise due to the Rdeg is ignored to simplify the
expression. The first three terms of (5.17) are originated from the switched-gmmixer and are the same as those derived in [13]. The last two terms are due to the
Biquad, both are noise-shaped by an in-band zero as expected. Further design
details of the mixer and LPF are omitted as their individual designs have been
extensively addressed in [13] and [14], respectively.
5.4.3 Simulation Results
The filtering characteristic of the mixer-LPF has been assessed. The passband gain
and stopband rejection are increased progressively as shown in Fig. 5.23. The bias
scheme of themixer-LPF is a singled-ended replica of it generatingVb,BB1 andVb, BB2,
reducing the effects of process variation and mismatch. In Monte-Carlo simulations
Table 5.2 Components
sizes of the mixer-LPFValue Value
Mm1,–Mm4 (gm) 2.4/0.06 � 4 Cf1 160 pF
Mm1,–Mm4 (√2gm) 3.4/0.06 � 4 Cf2 30 pF
Mf1–Mf4 8.2/1 � 20 CBB 18 pF
Ml1, Ml2 8.2/0.5 � 160 RBB 1.5 kORdeg 90 O
VDD25
Vb,BB1(Mm1-Mm4)
Vb,BB2(Mf3-Mf4)
lb
VDD25
M5b M6b
M3b M4b
M1b M2b
R1b R2b
VDD12 VDD12
Fig. 5.21 Bias circuit of the
current-reuse mixer-LPF
5.4 Current-Reuse Mixer-LPF 101
(Fig. 5.24), the�3-dBcutoff frequency shows ameanof12.9MHzwith a reasonablesof 0.636MHz. The passband gain variation is from 8.6 to 12.2 dB. A stable 60-dB per
decade stopband attenuation is achieved at the output. Other simulated performances
metrics are summarized in Table 5.3.
5.5 Multi-phase LOG
5.5.1 Brief Overview
Multiphase LOGs have been extensively studied in wireless/wireline applications.
A solution based on an inverter-based ring-VCO [15] is compact and has wide-
range tunability, but the phase noise is still high for most high-tier wireless systems.
-1
VDD25
Vbias
HBiq(s)
RBB In,RBB2
In,M f32
In,Mf12
Vn,Mm12
Vn,Rs/22
vo3p,BB
(gmf) Mf3
(gmf) Mf1
Cf2
Cf1
Mm1 (gm,Mm1)
Rs/2
Fig. 5.22 Noise model of the current-reuse mixer-LPF
102 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
Lowering the phase noise promotes the use of coupled LC-VCOs [16]. The overhead
is the excessive number of inductors which will occupy a considerable amount of
chip area. Delay-locked loop (DLL) offers another way to realize a multiphase LOG
[17]. Regrettably, since it is based on delay units, the phase-noise performance is
heavily dependent on the number of output phases required. Currently, the most
common solution is LC-VCO plus frequency dividers [7]. For instance, a quadri-
phase LO can be generated via a divide-by-2 circuit. This division implies that the
associated PLL and VCO have to operate at a frequency that will be double of the
output. Thus, the design complexity dramatically rises with the number of phases
required, without mentioning the power, phase noise and phase error overheads.
Moreover, the PLL and VCO will be more sensitive to parasitic capacitances,
implying narrower locking and tuning ranges, respectively.
5.5.2 Open Loop Multi-phase LO Generators(Conventional and Proposed)
Recently, an open-loop quadri-phase clock generator for wireline applications was
proposed to surmount these constraints [18]. Unlike the active polyphase filter
entailing both inverters and capacitors [19], each phase corrector involves only
inverters in a ring oscillator configuration with interpolation for multi-phase
-120
-100
-80
-60
-40
-20
0
20
0.1 1 10 100 1000
Frequency (MHz)
Mix
er-F
ilter
BB
Gai
n R
espo
nse
(dB
) StopbandRejection
PassbandGain
60dB/dec
vo1p,BB
vo3p,BB
vo2p,BB
Fig. 5.23 Simulated mixer-LPF’s gain-filtering characteristics
5.5 Multi-phase LOG 103
Fig. 5.24 Monte-Carlo Simulation of mixer-LPF baseband AC responses
Table 5.3 Performance summary of the current-reuse mixer-LPF (simulation)
Parameters Value
�3 dB cutoff (MHz) (Monte Carlo simulation) Mean 12.9
s 0.636
LPF stopband profile (dB) 60/decade
Conversion gain (dB) 10.5
DSB NF (dB) 15.2
Out-of-channel IIP3 (dBm) +15.7a
Input capacitance (fF) 43
Power (mW) 9.8aTwo tones at [fLO + 20 MHz, fLO + 31 MHz]
IMD3 @ 9 MHz, IMD2 @ 11 MHz after mixing
104 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
outputs, resulting in a much more compact area and with a wider tuning range. The
phase precision can be optimized by increasing the number of phase correctors in
cascade. A master LO direct injection-locks the chain and defines the steady-state
frequency. The tuning range depends on the reference LO and the lock range of the
phase correctors; both can be optimized via proper sizing of the associated
inverters. The prime advantages of this method are its simplicity (i.e., open loop
and inverter only), wide bandwidth and the lack of power-hungry buffers. More-
over, the frequency of the driving source can be the same frequency as the multi-
phase outputs. The achieved frequency range in [18] for a quadri-phase output is
0.37–2.5 GHz, but the phase error is limited to 5�.Here, the proposed 4-/8-phase LOG extends such a concept, appropriately
narrowing the coverage of the LOG by using three chains of 4-/8-phase correctors
individually optimized for a specific band. As such, and through circuit-level
optimization, the phase error is controlled within 1� (simulation), being more
suitable for wireless applications.
The block schematic of the proposed 4-/8-phase LOG is depicted in Fig. 5.25. It is
based on 2 chains of 8-phase corrector (8PC) and 1 chain of 4-phase corrector (4PC);
each is dedicated to a specific mobile-TV band for minimum phase error. Selectors
with logic arrangement feature assign the correct LO phase to eachmixer in different
modes. The injection signal LOin might cover the desired bands by using a 1.27-to-
1.92-GHz PLL + VCO [20] (not integrated in this work) with selectable output
division ratios (1, 2, 3, 6 and 8). This LO plan corresponds to 3.8� relaxation of the
PLL + VCO’s operation frequency when comparing it with [9]. However, here, the
required division ratios can be of any number as they are unrelated to the phases of
LOout. In the L band, the PLL + VCO works at the same RF frequency.
5.5.3 Circuit Implementation and Simulation Results
The circuit implementations of the 4- and 8-phase paths are shown in Fig. 5.26a and b,
respectively. LOin drives a chain of all-digital 4PC/8PC to progressively improve the
phase precision of the 4-/8-phase LOout. A high-level algorithm has been developed
LO Buffers8-Phase LOout @fLO
4-Phase LOout @fLO
VHF-III BandUHF Band
L Band
8PC
LOin
8
8
4@fLO
Se l
ecto
r
Mode and PhaseArrangement
8PC8PC8PC8PC8PC
8PC8PC8PC8PC8PC8PC
4PC4PC4PC4PCS
elec
tor
Mode 4PC: 4-Phase Corrector8PC: 8-Phase Corrector
(From VCO)
Fig. 5.25 A direct Injection-locked 4-/8-phase LOG
5.5 Multi-phase LOG 105
in MATLAB to optimize the device size, number of stages and frequency range
using a linearizedmodel [21]. At the transistor level, each 4PC/8PC comprises 16/32
inverters classified into two types and three sets as depicted in Fig. 5.27a and b.
L-type features a larger device size than S-type for optimizing the phase
correctability of each 8PC. Set A is to interpolate the intermediate phases. Set B is
to suppress the self-oscillation frequency and thereby enlarging the locking window.
Set C is for injection lock and direct stage-to-stage cascading. The entire LOG is
implemented with thin-oxide MOSFETs operating at a 1.2-V supply.
4PC#1
in3in2in1in0
out3out2out1out0
4PC#4
in3in2in1in0
out3out2out1out0
Phase (degree)
0
90
180
270
Phase (degree)
0
90
180
270
Phase (degree)
0
90
180
270. . .
Phase (degree)
0
90
180
270
. . .
LO180
LOinLOout
LOin LOout
LO0
LO180
LO0
LO45LO90LO135LO180LO225LO270LO315
LO0
LO90LO180LO270
LO0
8PC#1
in3in2in1in0
out3out2out1out0
in7in6in5in4
out7out6out5out4
8PC#6
in3in2in1in0
out3out2out1out0
in7in6in5in4
out7out6out5out4
Phase (degree)
0
90
180
270
45
135225
315
Phase (degree)
0
90
180
270
45
135225
315
Phase (degree)
0
90
180
270
45
135225
315
. . .
Phase (degree)
0
90
180
270
45
135225
315
. . .
a
b
Fig. 5.26 Block schematic of the LOG path (a) 4-phase. (b) 8-phase
106 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
The phase precision of the LOG is mainly limited by the routing in the layout,
where the transistor intrinsic RC value is sensitive to the parasitics. The LOG was
laid out and extracted to tune out this effect in several iterations. In the post layout
simulation (PLS), the phase error for 8-phase output is optimized to be <1� for theVHF-III and UHF bands. This precision fairly meets our target of HR ratio of
A
B
B
A
BAB
A
A
A
A
A
in3
out3C
in0
out0C
in2
out2C
in1
out1C
in5
in4
out5
out4
in2
out2
in7
out7
in0
in1
out0
out1
in3out3
in6
out6
A
A
A
A
A
A
A
A
A
A
A
AA
A
A
A
BB
B
B
B
B
B
B
C
C
CC
C
C C
C
L-type S-type
a
b
Fig. 5.27 Detailed schematic of the (a) 4PC. (b) 8PC
5.5 Multi-phase LOG 107
33 dB, even under a possible gain mismatch up to 5%. For the 4-phase output, the
phase error is controlled to be less than 1.5� for the L band, which corresponds to an
image rejection ratio of around 38 dB for a possible gain mismatch up to 2.5%.
Similar to ring oscillators, the robustness of the phase corrector can be improved
by adopting a supply regulator and a Bandgap reference to cope with the voltage and
temperature variations, respectively [22, 23]. Simulations show that the phase noise
of the multi-phased LOwill not be degraded bymore than 1 dB (at 1-MHz frequency
offset) if the random noise superimposed on the supply is less than 80 mV.
5.6 Measurement Results
Prototypes of the RFE were fabricated in 65-nm CMOS. Deep n-well was employed
for essential bulk-source connection of certain nMOS devices (e.g., the ones in
cascode). The chip micrograph is shown in Fig. 5.28. The active die area is
0.46 mm2. A production socket was employed to facilitate tests of multiple samples
under the same fixture and environment (Fig. 5.29). Particular attention has been
paid to the layouts of the LOG and 8PC to avoid parasitic mismatches that can lead
to systematic phase error in the LOG.
Fig. 5.28 Chip micrograph of the implemented RFE
108 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
5.6.1 The RFE
The RFE measures 35 � 1-dB maximum voltage gain, 4 � 0.2-dB NF, and
consumes 43-to-55 mW of power over the desired bands (Fig. 5.30). The BB gain
responses show 60-dB/dec rejection and a constant cutoff of around 12 MHz
against gain (Fig. 5.31). The cutoff was not made tunable in this prototype. The
key distortion specification is justified by the out-of-channel linearity [1]; two tones
are applied at [fLO + 20 MHz, fLO + 31 MHz] with a power level sweeping from
�40 to �20 dBm. The measured IIP3 is around �3.4/+11 dBm at the highest/
lowest gain by tuning the ATT gain steps (Fig. 5.32). The IIP2 yields less variation
(32–35 dBm) against gain change. The concerned HR ratios (HRRs) are HRR3
¼ 35 dB and HRR5 ¼ 39 dB, aligning with that achieved in a typical HR mixer [8].
Note that to increase the data accuracy; the IIP2/HRR3/HRR5 is averaged from 12
samples with s of 3.9/1.8/2.4 dB. The worst S11 is close to –10 dB (Fig. 5.33),
which is degraded when comparing it with the simulations. The bondwire, test
socket and PCB parasitics should account for the discrepancy. The S11 curves are
consistent to the designed values of 34.8 O at high gain mode (without ATT) and
55 O at low gain modes (with ATT).
Fig. 5.29 PCB of the implemented RFE
5.6 Measurement Results 109
0
10
20
30
40
50
60
0.1 1
NF
(d
B),
Gai
n (
dB
), P
ow
er (
mW
)
RF Frequency, fRF (GHz)
NF
Power
LUHFVHF III
@ Highest Gain
Gain
Fig. 5.30 Measured RFE’s power, gain and NF
60dB/dec
-10
0
10
20
30
40
0.1 1 10 100
BB Frequency (MHz)
BB
Gai
n R
esp
on
se (
dB
)
UHF band8-Phase LO @ 665MHz
~18dB Gain Range
Fig. 5.31 Measured RFE’s BB gain responses against gain control
24
28
32
36
40
44
48
0.1 1RF Frequency, fRF (GHz)
IIP3
(dB
m)
IIP2(
dB
m)
-8
-4
0
4
8
16
12
IIP2 @ Highest Gain
IIP3 @ Lowest Gain
IIP2 @ Lowest Gain
IIP3 @ Highest Gain
LUHFVHF III
Fig. 5.32 Measured RFE’s IIP2 and IIP3 at the highest and lowest gain levels
-35
-30
-25
-20
-15
-10
-5
0
0 0.5 1 1.5 2
S11
(d
B)
RF Frequency, fRF (GHz)
S11 versus the 4gain steps
1700MHz170MHz
Rin’=34.8W
Rin’=55W
Fig. 5.33 Measured RFE’s S11 against gain control
Based on transient measurements, we can justify that the BB differential outputs
(Fig. 5.34a) and I/Q outputs (Fig. 5.34b) are of high-precision balancing, less than
0.5 dB gain and 1.2� phase errors in downconverting a single tone from 205 to 1MHz.
Fig. 5.34 Measured RFE’s (a) differential and (b) I/Q outputs under single-tone downconversion
from 205 to 1 MHz
112 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
5.6.2 The LOG
The performance of the LOG in each band was characterized using a signal
generator as LOin like it is shown in Fig. 5.35. The added phase noises (Fig. 5.36a–c)
are well below to that achieved in [1] which uses a practical PLL + VCO as LOin.
Thus, the phase noise induced by the LOG should be tolerable when the PLL +
VCO is present.
The injection-locking characteristic of the LOG is of interest. When LOin is
deactivated (Fig. 5.37a), LOout is free-running at its natural oscillation frequency of
around 240 MHz and noticeable spurious tones appear throughout the spectrum.
With LOin activated with a frequency of 205 MHz, LOout tracks LOin at the same
frequency and the dynamic range is >60 dB (Fig. 5.37b). On the other hand, the
second to sixth harmonics will be suppressed by the differential I/Q mixer with HR.
These results demonstrate that a high-purity multi-phase LO can be achieved
without frequency division.
5.6.3 Performance Comparison
Benchmarking with the state-of-the-art wideband RFEs that attain 4-dB NF [9, 24]
in Table 5.4, this work succeeds in extending the operating BW and BB selectivity
with comparable power, while reducing the external parts, chip area and fLO that can
ease the design of the PLL and VCO. We note that the on-chip HRR can be further
enhanced by incorporating the two-stage HR technique [9]. Though the pre-gain and
pre-filtering technique [24] can also enhance the HRR and lower the power, it
involves hard tradeoffs in impedance match, IIP3 and consistency of in-band
performances.
SignalGenerator
LOout(1phase)
SpectrumAnalyzer
DUT
8-Phase LOG
Phase-Noise Measurement
655MHz0.35mm
0.18
mm
LOin
Fig. 5.35 Test setup for phase noise measurement
5.6 Measurement Results 113
100 1000
-130
-120
-110
-100
Offset Frequency (kHz)
-90
-150
-140Pha
se N
oise
(dB
c/H
z)P
hase
Noi
se (
dBc/
Hz)
Pha
se N
oise
(dB
c/H
z)10
10dB/Dec
PLL-locked 4-Phase Output [5.1]
LOin
@205 MHz
100 1000Offset Frequency (kHz)
10
-130
-120
-110
-100
-90
-150
-140
100 1000Offset Frequency (kHz)
10
-130
-120
-110
-100
-90
-150
-140
LOout
LOout
LOout
PLL-locked 4-Phase Output [5.1]
PLL-locked 4-Phase Output [5.1]
10dB/Dec
10dB/Dec
LOin
LOin
@655 MHz
@1.55 GHz
a
b
c
Fig. 5.36 Measured phase
noise of the LOG in each
band comparing with
the data given in [1].
(a) 8-phase at 205 MHz.
(b) 8-phase at 655 MHz
and (c) 4-phase at 1.55 GHz
114 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
5.6.4 Architecture Comparison
Cascode of building blocks [25, 26] is an effective solution for current reuse and
maximizing the BW at RF, as the node between the low-noise amplifier (LNA)
and mixer can be of low impedance (Z) as shown in Fig. 5.38. The main
shortcomings are the limited reverse isolation (LO $ RF $ BB leakages) and
DR, due to insufficient voltage headroom in nanoscale CMOS technologies.
-80
01st
2nd 3rd
1st
2nd3rd
100 80070/
0
-90
(Frequency MHz)
(dB
m)
(dB
m)
100 80070/
0
70 MHz/Start 100 MHz Stop 800 MHz
70 MHz/Start 100 MHz Stop 800 MHz
-10
-20
-30
-40
-50
0
-90
(Frequency MHz)
-60
-70
-10
-90
-90
-20
-30
-40
-50
-60
-70
a
b
Fig. 5.37 Measured injection-locking characteristic of the LOG: (a) free-running at ~240 MHz
(b) Injection-locked at the desired 205 MHz
5.6 Measurement Results 115
Table 5.4 Measurement summary and benchmark with the state-of-the-art wideband RFEs
Parameters This work ISSCC 2009 [9] JSSC 2010 [24]
Operation frequency
fRF (GHz)
0.17–1.7a 0.4–0.9 0.3–0.8
Required master LO
frequency fLO(GHz)
fLO ¼ fRF (4 and
8 phases)
fLO ¼ 8 fRF (8 phases) fLO ¼ 4 fRF (8 phases)
Maximum gain (dB) 35 34 22–28b
RF gain control (dB) 17–35 No No
External components 1 inductor 2 inductors and
1 balun
2 inductors
Area (mm2) 0.46 1 0.5
BB filter order Third-order LPF
(1 biquad
+ 1 real)
Second-order LPF
(2 real poles)
First-order IIR LPF
(minor channel
selectivity)
Power (mW) @
fRF (GHz)
55 @ 1.7c 60 @ 0.9 18 @ 0.8
Input impedance
matching
Matched Matched Unmatched
DSB NF (dB) 4 [Spec: 4]d 4 0.8–4.3b
IIP3 (dBm) �3.4 [Spec: -5]d 3.5 �14 to �9b
IIP2 (dBm) 32 [Spec: 27]d
(balun LNA)
46 (differential LNA) 38–49c (balun LNA)
HRR3 (dB) 35 60 60
HRR5 (dB) 39 64 60
Supply voltage (V) 1.2 and 2.5 1.2 1.2
Technology 65 nm CMOS 65 nm CMOS 65 nm CMOSaVHF III, UHF and L bandsbIn-band variationcBalun-LNA (11.6 mW), I/Q mixer-LPFs (19.6 mW), LOG + LO buffers (24 mW at 1.7 GHz,
12 mW @ 170 MHz)dFrom [1]
Mixer
Balun-LNA
Load
Mixer
LNA
VCO+Load
Low Z @ RF Low Z @ RF
VDDVDD
VBBVBB
a b
Fig. 5.38 Cascode of building blocks: (a) LMV cell [25] and (b) Blixer [26]
116 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
Recently, a mixed-VDD design approach (Fig. 5.39) has been proposed to
alleviate these drawbacks. Both VDD25 (I/O VDD) and VDD12 (core VDD), thick and
thin-oxide devices are employed. The VDD25 enlarges the voltage headroom while
allowing more BW-/linearity-demanding nodes to be operated in the current
domain. Specifically, the balun low-noise amplifier (LNA) employs a differential
current balancer (DCB) to improve the output gain-phase balancing of the single-
to-differential stage (S2D), yielding better IIP2 and reverse isolation. The key
expense is there will be a high impedance node at RF (VRF) that may limit the RF
BW (depends on the required gain). For the analog baseband, the lowpass filter
(LPF) can be cascoded with the mixer for current-mode filtering, leading to lower
noise and better out-of-channel linearity (i.e., filtering happens before I-V conver-
sion). For the local oscillator generator (LOG), since it is operated as digital
circuits, structuring it with only thin-oxide devices and VDD12 (core VDD) are
sufficed to fully benefit the speed, power and area advantages of advanced
technologies.
5.7 Summary
A 65-nm CMOS RFE composed by a wideband balun-LNA with VGC, dual
current-reuse mixer-LPFs and a direct injection-locked 4-/8-phase LOG has been
described. Comparing with the existing wideband RFEs, the external parts and
repeated RF circuitry are reduced and the required master LO frequency is relaxed.
The RFE exploits a dual 1.2/2.5-V supply, allowing more functions to be
implemented in the current domain. The output balancing of the balun-LNA is
DCB
S2D
Load
LPF
Mixer
Load
Dual-VDDThin-Oxide
MOS Balun-LNA
I/O-VDDThick/Thin-OxideMOS Mixer-LPF
Core-VDDThin-OxideMOS LOG
LOG
VDD25
VDD12
VDD12
VBBVRF
VDD25
Fig. 5.39 Device and voltage plan of this work
5.7 Summary 117
enhanced by a capacitively cross-coupled cascode stage, and baseband selectivity is
efficiently implemented via a current-domain Biquad merged with the mixer.
Device reliability is ensured via a hybrid use of thin- and thick-oxide MOSFETs
and voltage-conscious biasing. The fabricated prototypes meet the key
specifications of mobile TV with a small die size of 0.46 mm2.
References
1. Vassilios, K. Vavelidis and N. Haralabidis et al, “A 65-nm CMOS Multistandard, Multiband
TV Tuner for Mobile and Multi-Media Applications,” IEEE J. Solid-State Circuits, vol. 43,pp. 1522–1533, Jul. 2008.
2. M. Jeong, B. Kim and Y. Cho et al, “A 65 nm CMOS Low-Power Small-Size Multistandard,
Multiband Mobile Broadcasting Receiver SoC,” ISSCC Dig. Tech. Papers, pp. 460–461, Feb.2010.
3. P.-I. Mak and R. P. Martins, “A 0.46 mm2 4-dB NF Unified Receiver Front-End for Full-Band
Mobile TV in 65 nm CMOS,” ISSCC Dig. Tech. Papers, pp.172–173, Feb. 2011.4. S. Blaakmeer, E. Klumperink, D. Leenaerts and B. Nauta, “Wideband Balun-LNA with
Simultaneous Output Balancing, Noise-Canceling and Distortion-Canceling,” IEEE J. Solid-State Circuits, vol. 43, pp. 1341–1350, Jun. 2008.
5. D. Mastantuono and D. Manstretta, “A Low-Noise Active Balun with IM2 Cancellation for
Multiband Portable DVB-H Receivers,” ISSCC Dig. Tech. Papers, pp. 216–217, Feb. 2009.6. Datasheet of TDK Low Pass Filters for DVB-H/ISDB-T DEA Series DEA200710LT-1238A1.
7. V. Giannini, P. Nuzzo, C. Soens et al, “A 2-mm2 0.1–5 GHz Software-Defined Radio Receiver
in 45-nm Digital CMOS,” IEEE J. Solid-State Circuits, vol. 44, pp. 3486–3498, Dec. 2009.8. J. Weldon, J. Rudell, L. Lin et al, “A 1.75-GHz Highly Integrated Narrowband CMOS
Transmitter with Harmonic-Rejection Mixers,” ISSCC Dig. Tech. Papers, pp. 160–161, Feb.2001.
9. Z. Ru, E. Klumperink, G. Wienk, and B. Nauta, “A Software-Defined Radio Receiver
Architecture Robust to Out-of-Band Interference,” ISSCC Dig. Tech. Papers, pp. 230–231,Feb. 2009.
10. B. Razavi, “Cognitive Radio Design Challenges and Techniques,” IEEE J. of Solid-StateCircuits, vol. 45, pp. 1542–1553, Aug. 2010.
11. J. Borremans, G. Mandal, V. Giannini, T. Sano, M Ingels, B. Verbruggenn and J. Craninckx
“A 40 nm CMOS Highly Linear 0.4-to-6 GHz Receiver Resilient to 0dBm Out-of-Band
Blockers,” ISSCC Dig. Tech. Papers, pp. 62–63, Feb. 2011.12. W. Zhuo, X. Li, S. Shekhar, S.H.K. Embabi, J. Pineda de Gyvez, D.J. Allstot and E. Sanchez-
Sinencio, “A Capacitor Cross-Coupled Common-Gate Low Noise Amplifier,” IEEE Trans. onCircuits and Systems II: Express Briefs, vol. 52, pp. 875–879, Dec. 2005.
13. E. Klumperink, S. Louwsma, G. Wienk and B. Nauta, “A CMOS Switched Transconductor
Mixer,” IEEE J. Solid-State Circuits, Vol. 39, pp. 1231–1240, Aug. 2004.14. A. Pirola, A. Liscidini and R. Castello, “Current-Mode, WCDMA Channel Filter with In-Band
Noise Shaping,” IEEE J. Solid-State Circuits, vol.45, no.9, pp.1770–1780, Sept. 2010.15. D. Y. Jeong, S. H. Chai, W. C. Song and G. H. Cho, “CMOS Current-Controlled Oscillators
Using Multiple-Feedback-Loop Ring Architectures,” ISSCC Dig. Tech. Papers, pp. 386–387,Feb. 1997.
16. L. C. Cho, C. Lee, and S. I. Liu, “A 1.2-V 37–38.5-GHz Eight-Phase Clock Generator in 0.13-
mm CMOS Technology,” IEEE J. Solid-State Circuits, vol. 42, no. 6, pp. 1261–1270, Jun.2007.
118 5 A Mixed-Voltage Unified Receiver Front-End for Full-Band Mobile TV. . .
17. X. Gao, E. A. M. Klumperink and B. Nauta, “Advantages of Shift Registers Over DLLs for
Flexible Low Jitter Multiphase Clock Generation,” IEEE Trans. on CAS–II: Express Briefs,vol. 55, no. 2, pp. 244–248, Mar. 2009.
18. K. H. Kim, P. W. Coteus, D. Dreps et al, “A 2.6 mW 370 MHz-to-2.5 GHz Open-Loop
Quadrature Clock Generator,” ISSCC Dig. Tech. Papers, pp. 458–459, Feb. 2008.19. F. Tillman, H. Sjoland, “A Polyphase Filter Based on CMOS Inverters,” NORCHIP Conf.,
pp. 12–15, Nov. 2005.
20. Lei Lu, Zhichao Gong, Youchun Liao, et al, “A 975-to-1960 MHz Fast-Locking Fractional-N
Synthesizer with Adaptive Bandwidth Control and 4/4.5 Prescaler for Digital TV Tuners”
ISSCC Dig. Tech. Papers, pp. 396–397, Feb. 2009.21. K.-F. Un, P.-I. Mak and R. P. Martins, “Analysis and Design of Open-Loop Multi-Phase
Local-Oscillator Generator for Wireless Applications,” IEEE Trans. on CAS-I: RegularPapers, vol. 57, no. 5, pp.970–981, May 2010.
22. T. Wu, K. Mayaram, and U. Moon, “An On-Chip Calibration Technique for Reducing Supply
Voltage Sensitivity in Ring Oscillators,” IEEE J. Solid-State Circuits, vol. 42, no. 4,
pp. 775–783, Apr. 2007.
23. Y.-T. Huang, C. M. Yang, S. C. Huang, H. L. Pan, T. C. Hung, “A 1.2 V 67 mW 4 mm2Mobile
ISDB-T Tuner in 0.13 mm CMOS,” ISSCC Dig. Tech. Papers, pp. 124–125, Feb. 2009.24. Z. Ru, E. Klumperink, C. Saavedra and B. Nauta, “A 300–800 MHz Tunable Filter and
Linearized LNA Applied in a Low-Noise Harmonic-Rejection RF Sampling Receiver,”
IEEE J. Solid-State Circuits, vol. 45, pp. 967–978, May 2010.
25. A. Liscidini et al., “Single-stage low-power quadrature RF receiver front-end: The LMV cell,”
IEEE J. Solid-State Circuits (JSSC), vol. 41, no. 12, pp. 2832–2841, Dec. 2006.26. S.C. Blaakmeer, E. Klumperink, D. Leenaerts, B. Nauta, “The Blixer, a Wideband Balun-
LNA-I/Q-Mixer Topology,” IEEE J. of Solid-State Circuits (JSSC), vol.43, no.12,
pp.2706–2715, Dec. 2008.
References 119
Chapter 6
Conclusions
6.1 Concluding Remarks
Technology downscaling has stressed the use of sub-1V VDD to maintain device
reliability. In this book, high-/mixed-voltage RF and analog CMOS circuits have
been presented as the technology comes of age, being apposite for realizing high-performance wireless systems in ultra-scaled CMOS technologies. Dual voltage
supplies have become very common since the deployment of nm-length CMOS
processes such as the 90-nm node. We also believe that multiple supplies offering a
VDD-on-demand approach will be more efficient in overall system power savings.
This book explored a number of analog and RF circuit techniques. From block level
to sub-system level, the techniques led to enhanced performances, lower power and
cost. When design-for-reliability (i.e., prevent overstress on any device) is included
in the design flow, an elevated VDD outpacing the technology roadmap directly
opens up more flexibility in defining circuit topologies while preserving sufficient
voltage headroom for signal swing. Table 6.1 summarizes the design considerations
of analog and RF circuits with different degrees of freedom, namely area, current
and the topic of this research, supply.Three high-/mixed-voltage design examples that are application-specified to
mobile TV have been presented as proofs of concept. The first design is a 90-nm
CMOS full-band mobile-TV low-noise amplifier (LNA) with mixed-voltage ESD
protection. The second is a 90-nm CMOS high-voltage-enabled mobile-TV RF
front-end with TV-GSM interoperability. The third is a 65-nm CMOS mixed-
voltage unified receiver front-end for full-band mobile TV. Together with a number
of system- and circuit-level techniques and optimization, the performances of the
describedworks have shown significant advances with respect to the state-of-the-art.
In conclusions, differing from digital circuits, analog and RF circuits will benefit
much more from technology scaling if the VDD (high, low or adaptive) is wisely
chosen for each block in a SoC environment. The transistor plan for each block can
be further customized among the available types such as thin-oxide, thick-oxide,
P.-I. Mak and R.P. Martins, High-/Mixed-Voltage Analog and RF Circuit Techniquesfor Nanoscale CMOS, Analog Circuits and Signal Processing,
DOI 10.1007/978-1-4419-9539-1_6, # Springer Science+Business Media New York 2012
121
high VT, standard VT and low VT. Mixed-voltage mixed-transistor design
methodologies are becoming a new art of circuit design in ultra-scaled CMOS
technologies.
6.2 Recommendations for Future Research
A mixed-VDD wideband transceiver architecture potentially suitable for software-
defined radios (SDRs) and cognitive radios (CRs) is depicted in Fig. 6.1. There are
numerous issues meriting further exploration in future research:
• Inductorless design is already an obvious approach for area savings in RF
receivers. When a high-VDD is adopted, a wideband low-noise transconductance
amplifier (LNTA) can deliver the desired gain with high DR when driving a
current-mode passive mixer. The first BB filter (e.g., OpAmp-based circuits)
worked as an I-V converter can directly benefit from the gain and DR offered by
a high-VDD. With the advance of data converters (both ADC and DAC), a
standard-VDD already can lead to excellent power efficiency for the requirements
of most wireless applications (i.e., speed of 20–60 MS/s and resolution of
8–11 bits). A mixed-VDD, however, will be relevant when more resolution is
entailed. For instance, when speed is not that demanding, the nonlinearity of the
input sample-and-hold circuit can be reduced effectively by using thick-oxide
MOS and a high-VDD clock (CLK). Moreover, the involved OpAmps in the ADC
and DAC can also benefit from a high-VDD to achieve a higher DC gain and a
wider signal swing.
• A high-VDD PA has already been a common practice in wireless transmitters to
boost output power and enhance power efficiency, especially for multi-band
communications where LC resonators should be avoided. For the mixer and
baseband LPF, passive implementation is becoming more relevant to minimize
nonlinearity and output noise floor; both are critical parameters for wideband
SDRs and CRs where external SAW filters may not be possible.
Table 6.1 Design strategies of analog and RF circuits in nanoscale CMOS
RF Analog
Increase area ✓ Gain-bandwidth (e.g., via inductors) ✓ Matching (DC-Offset)
✓ kT/C noise
Increase current ✓ Unity-gain frequency (e.g., higher device fT) ✓ Slew rate
✓ Bandwidth
Increase supplya ✓ Gain (e.g., allows a larger RL load) ✓ Gain precision
✓ Bandwidth (e.g., cascode of blocks to avoidhigh impedance nodes)
✓ PSRR (e.g., via LDO)
✓ Dynamic range
✓ Dynamic range ✓ Buffering and level
shifting in I/Os✓ Power efficiency (e.g., in PA)aDeviceoverdrive voltagesmustbewithin the technologyguidelines in both transient and steady states
122 6 Conclusions
• A low-power, area-efficient, power-management unit with multiple voltage
outputs based on single-inductor DC-DC converters and/or low-dropout
regulators (LDOs) are worth to explore in the nanoscale CMOS such that the
benefits of mixed-voltage design will become more obvious at the system level.
In particular, inductor-less and capacitor-less solutions will be highly beneficial
in silicon area savings, which is getting more and more expensive in ultra-scale
process nodes. High power supply rejection ratio (PSRR) over a wide bandwidth
will be a very challenging design goal for power-management circuits.
• The local oscillator (LO)/CLK generator can be based on a high-VDD VCO
which uses LDOs to enhance PSRR. Maximizing the VCO tuning range can be
resorted from thick- oxide varactor or MOS capacitor; both can withstand a
wider tuning voltage. For the involved frequency dividers that operate as digital
circuits, thin-oxide MOS and core-VDD are adequate.
• Last but not least, a time-adaptive VDD for the different digital blocks should be
an important direction for global power reduction (e.g., a very low-VDD in
standby mode to minimize leakage power).
High-VDD
High-VDDHigh-VDD
LNTA
ADC
Mixed-VDD
Mixed-VDD
Mixed-VDD
DAC
Mixer
PA
Mixer
LPF
LPFI I V
LO /CLK Generator
VVV
Power Mgr.Battery
... DifferentVDD’s
DSP
Adaptive-VDD
Device Oxide Thickness: Thick and ThinDevice VT: High, Standard, LowDC/DC
LDO
Passive
Passive Passive
Fig. 6.1 A prospective mixed-VDD wideband transceiver architecture for SDRs and CRs
6.2 Recommendations for Future Research 123
Appendix: Open-Loop MultiphaseLO Generators
A.1 Introduction
Many different types of multi-phase LO/clock generators have been proposed for
wireless and wireline applications. A ring voltage-controlled oscillator (VCO)
using inverters (Fig. A.1a) is a compact solution to realize a multi-phase LO with
a large frequency range [1, 2]. However, its phase noise performance is normally
unacceptable for high-tier wireless systems. Although the phase noise can be
substantially reduced by replacing all inverters with LC VCOs (Fig. A.1b) the
associated inductors occupy a significant part of chip area [3, 4]. Delay-locked
loop (DLL) using numerous delay units can also be a multi-phase clock generator
(Fig. A.1c) [5, 6]. The key drawback is that the phase-noise performance is heavily
dependent on the number of output phases required.
Alternatively, a multi-phase LO signal can be generated in an open-loop way by
using a low-noise LC VCO followed by a frequency divider (Fig. A.1d). Elementa-
rily, a quadri-phase LO can be generated via a div-by-2 circuit. Such a division factor
implies that the associated phase-locked loop (PLL) and VCO have to operate at a
doubled frequency. The design complexity, however, rises dramatically with the
number of phases required. For instance, to generate an octave-phase LO a div-by-4
circuit is necessary while the PLL and VCO have to operate at 4� of the output
frequency. A higher operating frequency unavoidably calls for more power to
lower the phase noise and phase error. Moreover, the PLL and VCO will be more
sensitive to parasitic capacitances, implying narrower locking and tuning ranges,
respectively. To surmount these constraints the frequency divider can be replaced
by a passive RC-CR polyphase filter (Fig. A.1e) [7, 8], but the performance can be
strongly affected by the temperature and process variations, while power-hungry
buffers are required for proper interface. Recently an open-loop quadri-phase clock
generator was proposed for wireline applications [9, 10]. Multiple phase correctors
are cascaded in an open-loop way to improve the phase precision (Fig. A.1f).
P.-I. Mak and R.P. Martins, High-/Mixed-Voltage Analog and RF Circuit Techniquesfor Nanoscale CMOS, Analog Circuits and Signal Processing,
DOI 10.1007/978-1-4419-9539-1, # Springer Science+Business Media New York 2012
125
The prime advantages of this method are its simplicity (i.e., open loop and inverter
only), no power-hungry buffer, and the number of output phases is independent of
the operation frequency of the circuit itself and its driving source. The achieved
frequency range in [9] is 0.37–2.5 GHz and the phase precision is �5� for a quadri-phase output.
In this content, we extend the concept of such open-loop architecture for wireless
applications with different requirements on the phase precision and the number of
output phases [11]. The targeted phase error is �1� and both quadri- and octave-
phase LO generators will be designed and analyzed.
...PLL
Inverter
...
PLL
LCVCO
LCVCO
LCVCO
LCVCO
LCVCO
LCVCO
LCVCO
LCVCO
DelayUnit 1
DelayUnit 2
DelayUnit N
Multi-Phase LO/Clock
PLL
Div-by-NPLLMulti-Phase
LO/Clock
Multi-PhaseLO/Clock
Multi-PhaseLO/Clock
PLLRC-CR
PolyphaseFilter
Multi-PhaseLO/Clock
Multi-PhaseLO/Clock
buffersbuffers
PLLPhase
CorrectorN
PhaseCorrector
1
…
……
…
………
…
…
a
b
c
d
e
f
Fig. A.1 Multi-phase LO/clock generation methods: (a) ring with inverters, (b) ring with LCVCOs,
(c) DLL (d) frequency divider, (e) RC-CR polyphase filter and (f) phase correctors in cascade
126 Appendix: Open-Loop Multiphase LO Generators
A.2 Mathematical Model of Open-Loop Multi-phaseLO Generator
A.2.1 Quadri-phase LO Generator
Architecture – The block diagram of a quadri-phase LO generator is depicted in
Fig. A.2. It is structured by putting numerous phase correctors in cascade to
interpolate a multi-phase LO from a two-phase differential input (vclk and vclkp).From left to right the phase correctors improve the phase precision progressively
until reaching the desired accuracy. The schematic of each inverter-based quadri-
phase corrector is shown in Fig. A.3. Every corrector is composed by 16 inverters
(CMOS) classified according to two different device sizes as L-type or S-type.L-type inverters feature a larger geommetrical size than the S-type to optimize the
phase precision in a specific frequency range. The inverters can be divided into
three groups according to their functionality: (1) Set A is for phase correction. With
three inverters in a loop it is able to oscillate and interpolate the intermediate
phases; (2) Set B is for natural-frequency suppression. It leads to a larger operating
frequency range [16]; (3) Set C is for signal injection. It allows multiple phase
correctors to be directly cascaded to improve the output phase precision.
Mathematical Model – In order to determine the optimum conditions in terms of
frequency range and phase precision, the quadri-phase corrector is modeled by a
signal flow graph (SFG) as shown in Fig. A.4. For simplicity, a linear model is
assumed [12, 13]. Each inverter is modeled as a single-pole amplifier with a transfer
function of h( f ) as given by,
hð f Þ ¼ � G
1þ jf
fC
; (A.1)
. . .
Cascade of Repeated Cell
Phase (degree)
090
180270
Open LoopQuadri-Phase
CorrectorN
...
vclk_270
vclk_180
vclk_90
vclk_0pq0(N)
pq1(N)pq2(N)pq3(N)vclkp
vclkvo0vo1vo2vo3
vi0vi1vi2vi3
Open LoopQuadri-Phase
CorrectorN
vo0vo1vo2vo3
vi0vi1vi2vi3
Open LoopQuadri-Phase
CorrectorN
vo0vo1vo2vo3
vi0vi1vi2vi3
pq0(N-1)pq1(N-1)pq2(N-1)pq3(N-1)
pq0(2)pq1(2)pq2(2)pq3(2)
pq0(1)pq1(1)pq2(1)pq3(1)
pq0(0)pq1(0)
pq2(0)pq3(0)
Fig. A.2 Block diagram of a quadri-phase LO generator. Phase error is reduced down progressively
from the correctors 1 to N
A.2 Mathematical Model of Open-Loop Multi-phase LO Generator 127
where G is the normalized DC gain and fC is the �3-dB cutoff frequency.
The constants a and b in Fig. A.5 represent the driving capability of L-type and
S-type inverters, respectively.Phasor-domain analysis is applied to obtain the phase correction transformation
of the nth phase corrector as expressed by,
ah
bh
ah
bh
ah bh ahbh
ah
ah
ah
ah
bh
bh
bh
bh
pq2(n-1)
pq2(n)pq1(n)
pq0(n) pq3(n)
pq3(n-1)pq4(n-1)
pq1(n-1)
Fig. A.4 SFG of a quadri-phase corrector
A
B
B
A
BAB
A
A
A
A
A
C
C
C
C
L-type S-type
vo2vo1
vo0 vo3
vi3
vi2vi1
vi0
Fig. A.3 Architecture schematic of an inverter-based quadri-phase corrector
128 Appendix: Open-Loop Multiphase LO Generators
AqpqðnÞ ¼ pqðn�1Þ
Aq ¼
1 �ah �ah �bh
�bh 1 �ah �ah
�ah �bh 1 �ah
�ah �ah �bh 1
26664
37775
pqðnÞ ¼ pq0ðnÞ pq1ðnÞ pq2ðnÞ pq3ðnÞ� �T
8>>>>>>>>><>>>>>>>>>:
; (A.2)
where Aq is the phase transformation matrix and pq(n) is the output phase vector.
Re-arranging (A.2) it yields,
pqðnÞ ¼ A�1q pqðn�1Þ: (A.3)
For N quadri-phase cascaded correctors pq(N) becomes,
pqðNÞ ¼ A�Nq pqð0Þ; (A.4)
where pq(0) is the input phase vector represented by,
pqð0Þ ¼ 1 1 �1 �1½ �Tbh; (A.5)
1.5 2 2.5 3 3.5-40
-30
-20
-10
0
10
20
30
40
n = 1n = 2n = 3n = 4
Reference Line
Normalized Frequency (f/fC)
Arg
(jpq1
(n)/p
q0(n
)) (
Deg
ree)
Fig. A.5 Static phase error of quadri-phase corrector with a/b ¼ 3.4 and G ¼ 10 (linear-model
simulation)
A.2 Mathematical Model of Open-Loop Multi-phase LO Generator 129
since the input phase can be either 0� or 180�. In the phasor domain only 1 or�1 are
available for the input vector. Figure A.5 shows the steady-state phase error
function defined by (A.6) with different number of stages in cascade,
FeqðnÞ ¼ Argð j pq1ðnÞpq0ðnÞ
Þ: (A.6)
wherea/b is chosen tobe3.4 toprovide anacceptable frequency range andG is selected
as 10 (practical DC gain value of a CMOS inverter in nanometer technologies).
The steady phase error depends on the ratio of the output frequency ( f ) to the cornerfrequency of an inverter (fc), as well as the number (n) of phase correctors in cascade.With n ¼ 4, the phase error is minimized over a wide range of f/fc (between 2.1
and 2.7). Figure A.6 also shows that the steady-state phase error can be minimized
by cascading additional stages of quadri-phase correctors for certain frequency ranges.
Based on the above definitions and following a particular context the transfer
function of the linearized quadri-phase LO generator will be derived next.
According to it, the optimal conditions for minimizing the phase error are obtained
to build-up the device-sizing strategy. From (A.2), we can simply prove that,
pq0ðnÞ ¼ �pq2ðnÞpq1ðnÞ ¼ �pq3ðnÞ
(: (A.7)
Thus, (A.2) and (A.3) can be simplified as,
1þ ah �ða� bÞhða� bÞh 1þ ah
� �pq0ðnÞpq1ðnÞ
� �¼ pq0ðn�1Þ
pq1ðn�1Þ
� �: (A.8)
OpenLoop
OctavePhase
Corrector1
. . .
Cascade of Repeated Cell
Phase (degree)
0
90
180
270
OpenLoop
OctavePhase
Corrector2
OpenLoop
OctavePhase
CorrectorN
vclk_315
vclk_270
vclk_225
vclk_180
vclk_135
vclk_90
vclk_45
vclk_0
vclkp
vclk
45
135
225
315
...
po0(N)vo0vo1vo2vo3vo4vo5vo6vo7
vi0vi1vi2vi3vi4vi5vi6vi7
vo0vo1vo2vo3vo4vo5vo6vo7
vi0vi1vi2vi3vi4vi5vi6vi7
vo0vo1vo2vo3vo4vo5vo6vo7
vi0vi1vi2vi3vi4vi5vi6vi7
po1(N)po2(N)po3(N)po4(N)po5(N)
po6(N)po7(N)
po0(2)po1(2)po2(2)po3(2)po4(2)po5(2)
po6(2)po7(2)
po0(1)po1(1)po2(1)po3(1)po4(1)po5(1)
po6(1)po7(1)
po0(0)po1(0)po2(0)po3(0)po4(0)po5(0)
po6(0)po7(0)
po0(N-1)po1(N-1)po2(N-1)po3(N-1)po4(N-1)po5(N-1)
po6(N-1)po7(N-1)
Fig. A.6 Block diagram of an octave-phase LO generator
130 Appendix: Open-Loop Multiphase LO Generators
And also for a simplification of the notations, A, B and C are introduced as
follows:
A ¼ 1þ ah
B ¼ ða� bÞh
C ¼ A �B
B A
� �: (A.9)
Similar to (A.4), we can obtain,
pq0ðNÞpq1ðNÞ
� �¼ C�N pq0ð0Þ
pq1ð0Þ
� �¼ C�N 1
1
� �: (A.10)
Since the matrix C can be diagonalized as,
C ¼ 1ffiffiffi2
p j �j1 1
� �� �Aþ jB 0
0 A� jB
� �1ffiffiffi2
p �j 1
j 1
� �� �; (A.11)
Substituting it in (A.10) will finally lead to,
pq0ðNÞpq1ðNÞ
� �¼ 2N
ð1þ jÞðAþ jBÞ�N þ ð1� jÞðA� jBÞ�N
ð1� jÞðAþ jBÞ�N þ ð1þ jÞðA� jBÞ�N
� �: (A.12)
Finally, the transfer function of pq1(N) divided by pq0(N) can be obtained as,
pq1ðNÞpq0ðNÞ
¼jþ 1þ ah� jða� bÞh
1þ ahþ jða� bÞh� �N
1þ j1þ ah� jða� bÞh1þ ahþ jða� bÞh
� �N : (A.13)
The criteria for phase error minimization is equivalent to set,
pq1ð1Þpq0ð1Þ
¼ �j; (A.14)
which implies a 90� phase shift. Substituting (A.14) into (A.13) leads to,
1� 2Gaþ Gbþ f
fc¼ 0
1� Gb� f
fc¼ 0
8>><>>: : (A.15)
A.2 Mathematical Model of Open-Loop Multi-phase LO Generator 131
By solving (A.15) the optimal conditions for phase error minimization can be
obtained as,
a ¼ G�1
f ¼ Gða� bÞfc
: (A.16)
It implies that an ideal 90� phase shift happens at the natural frequency of the
circuit: fn ¼ Gða� bÞfc. Also, a larger a/b ratio can provide a stronger phase
correcting ability.
A.2.2 Octave-Phase LO Generator
Architecture – Octave-phase LO generation can be obtained by further extending
the quadri-phase LO concept and architecture previously outlined. As such, the
block diagram of an octave-phase LO generator can be drawn as illustrated by
Fig. A.6. From left to right an octave-phase LO can be composed by multiple
octave-phase correctors. The number of stages in cascade will directly depend on
the final phase-precision requirement. The architecture schematic of an octave-
phase corrector is shown in Fig. A.7 which is composed by 32 inverters (CMOS).
Similar to the quadri-phase design the inverters are also classified as L-type or
S-type and the three sets (A, B and C) previously mentioned are also maintained.
C
A
B
BB
B
B
B
B
B
A
A
A
A
A
A
A
A
A
A
A A
A
A
A
C
C C
C
C
C Cvo4vo1
vo6
vi1
vi6
vi3
vo3
vi0
vo0 vo5
vi5
vo2
vi2
vo7
vi4
L-type S-type
Fig. A.7 Architecture schematic of an octave-phase corrector
132 Appendix: Open-Loop Multiphase LO Generators
Mathematical Model – The SFG of the octave-phase corrector is presented in
Fig. A.8 where a linear model is also assumed and each inverter is modeled,
similarly, as a single-pole amplifier. Again, phasor-domain analysis is applied to
obtain the phase transformation of the nth octave-phase corrector as given by,
AopoðnÞ ¼ poðn�1Þ
Ao ¼
1 �ah 0 0 �ah 0 0 �bh
�bh 1 �ah 0 0 �ah 0 0
0 �bh 1 �ah 0 0 �ah 0
0 0 �bh 1 �ah 0 0 �ah
�ah 0 0 �bh 1 �ah 0 0
0 �ah 0 0 �bh 1 �ah 0
0 0 �ah 0 0 �bh 1 �ah
�ah 0 0 �ah 0 0 �bh 1
266666666666664
377777777777775
poðnÞ ¼ po0ðnÞ po1ðnÞ ::: po7ðnÞ� �T
8>>>>>>>>>>>>>>>>>>><>>>>>>>>>>>>>>>>>>>:
; (A.17)
bh
bh
bh
bh
bh
bh
bh
bh
bh
bh
bh
bh
bhbh
bh
bh
ah
ah
ah a h
ah
ah
ahah
ah
ah
ah
ah
ah
ah
ah
ah
po4(n-1)
po4(n)po3(n)
po2(n)
po1(n)
po0(n) po7(n)
po6(n)
po5(n)
po3(n-1)
po2(n-1)
po1(n-1)
po0(n-1) po7(n-1)
po6(n-1)
po5(n-1)
Fig. A.8 SFG of an octave-phase corrector
A.2 Mathematical Model of Open-Loop Multi-phase LO Generator 133
where Ao is the octave-phase transformation matrix and po(n) is the output phase
vector of the nth octave-phase corrector. Re-arranging (A.17) it will yield,
poðnÞ ¼ A�1o poðn�1Þ: (A.18)
For N octave-phase correctors in cascade, po(N) is given by,
poðNÞ ¼ A�No poð0Þ: (A.19)
where po(0) is the input phase vector,
poð0Þ ¼ 1 1 1 1 �1 �1 �1 �1½ �Tbh; (A.20)
since the input phase can be either 0� or 180�. In the phasor domain, only 1 or�1 is
available for the input phase vector. The steady-state phase error function can be
defined by,
FeoðnÞ ¼ Arg ejp4po1ðnÞpo0ðnÞ
� �: (A.21)
As illustrated by Fig. A.9 the steady-state phase error can be minimized by
cascading additional stages of octave-phase correctors for a certain frequency range
from 0.7 to 1.3 (normalized frequency: f/fc). The a/b ratio is set to be 3.5 and G is
10. Although the optimum conditions in terms of frequency range and phase
Arg
(ejπ
/4p o
1(n)
/po0
(n))
(D
egre
e)
Normalized Frequency (f/fC)
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8-50
-40
-30
-20
-10
0
10
20
30
40
n = 2n = 3n = 4n = 5n = 6
ReferenceLine
Fig. A.9 Static phase error of octave-phase corrector with a/b ¼ 3.5 and G ¼ 10 (linear-model
simulation)
134 Appendix: Open-Loop Multiphase LO Generators
precision can be derived from the linear model the final phase permutation can only
be determined from transistor-level simulation because not all the transistors can be
operated, simultaneously, in the saturation region.
A.2.3 Design Considerations
Sizing – Both quadri- and octave-phase correctors have a limited operating frequency
range. The channel length of the inverter’s transistors is correlated to the upper
frequency limit (i.e., a smaller channel length allows a higher operating frequency).
When the channel length is fixed the a/b ratio is correlated to the range and the phasecorrecting ability of the phase corrector. A smaller a/b ratio can increase its operatingrange at the expense of a weaker phase-correcting ability. Thus, subject to different
applications, the optimum a/b ratio should be chosen such that the desired frequencyrange can be covered. On the other hand, the optimum number of phase correctors
needed in cascade can be determined according to the required phase precision.
PVT Variations – Similar to the ring oscillator the frequency range covered by the
phase corrector can be sensitive to process, voltage and temperature (PVT)
variations. For fast-fast (FF)/slow-slow (SS) process corner with temperature and
voltage variations, the covered frequency range is shifted up/down significantly, as
shown in Fig. A.10. For a reliable design the channel length of the inverters is
determined at “SS corner + low supply voltage + high temperature” for the highest
operating frequency to be larger than the desired frequency. Then, a suitable a/b at
“FF corner + high supply voltage + low temperature” is chosen for the lowest
operating frequency which is also lower than the desired. Increasing the width of
the transistors can only lead to a better variability control at the expense of power.
The operating principle is not dependent of the transistors’ width.
Again, similar to the ring oscillator, the robustness of the phase corrector can be
improved by adopting a supply regulator and a bandgap reference to cope with
the voltage and temperature variations, respectively [14, 15]. Simulations show that
Frequency
UsableFrequency Range
Frequency RangeCovered by Typical
FF + Supply+ Temperature
Typical
SS + Supply+ Temperature
Fig. A.10 Usable frequency range accounting for PVT variations
A.2 Mathematical Model of Open-Loop Multi-phase LO Generator 135
the regulator should stabilize the power supply with less than 80-mV fluctuation
such that the phase noise of the multi-phased LO will not be degraded by more than
1-dB at 1-MHz frequency offset. Those schemes are under development and have
not been included in this work.
Design and Verification Flow – The design and verification flow is graphically
summarized in Fig. A.11a and b. For simplicity only the quadri-phase LO generator
is considered. Based on the developed linear model and equations the phase error
Linear Model ofthe Quadri-Phase LO
Generator
Find phaseerror vector
Eq. (A1), (A2),(A4) & ( A5)
Evaluate thephase error
Eq. (A6)
Meets thespecifications?
Yes
To Transistor-LevelDesign
No
Adjusta/b or n
Modeling andParameter Determination
Transistor-Level Design,simulation and layout
Transistor Sizing[according to G, a/b & n obtain from theModeling and Parameter Determination]
TransientSimulation
Cover thedesired frequency
range?
Adjusttransistor’s
channellength
Adjusta/b No
No
Yes
Yes
Layout, Parasitics Extraction& Post-Layout Verification
Yes
DesignComplete
No
If ne
cess
ary
Eq. (A16)
InitialG, a/b & n
Confirm finalphase
permutation
Meet thespecifications ?
Phase precisionin the desired
frequency range?Phase noise?
a b
Fig. A.11 Design and verification flow: (a) modeling and parameter determination and
(b) transistor-level design, simulation and layout
136 Appendix: Open-Loop Multiphase LO Generators
vector can be determined with initial values of G, a/b ratio and n. These values canbe adjusted to minimize the phase error over the desired frequency range with fast
simulation speed. The obtained circuit parameters are then transferred to the
transistor-level design. Since the circuit is dynamic the optimization involves
mostly transient simulations. Except for the particular case of the phase noise
which was checked through periodic noise (pnoise) simulations. Although the
linear model can provide a set of parameters that are close to the optimum values,
transistor-level fine tuning is still necessary to account for PVT variations. Circuit
nonlinearity may also affect the final phase permutation and it must be confirmed at
transistor level. Finally, the optimized circuit can be transferred to the layout design
phase. Post-layout verification with parasitic effects is needed to re-confirm all
performance metrics and it will be repeated until all specifications are met.
References
1. D. Y. Jeong, S. H. Chai, W. C. Song and G. H. Cho, “CMOS Current-Controlled Oscillators
Using Multiple-Feedback-Loop Ring Architectures,” IEEE Int. Solid-State Circ. Conf.(ISSCC) Digest, pp. 386–387, 491, Feb. 1997.
2. J. Y. Chang, C. W. Fan, C. F. Liang, and S. I. Liu, “A Single-PLL UWB Frequency
Synthesizer Using Multiphase Coupled Ring Oscillator and Current-Reused Multiplier,”
IEEE Trans. on Circuits and Systems – II: Express Briefs, pp. 107–111, vol. 56, no. 2, Feb.2009.
3. L. C. Cho, C. Lee, and S. I. Liu, “A 1.2-V 37–38.5-GHz Eight-Phase Clock Generator in
0.13-mm CMOS Technology,” IEEE J. of Solid-State Circuits (JSSC), pp. 1261–1270, vol.42, no. 6, Jun. 2007.
4. A. Rofougaran, J. Rael, M. Rofougaran and A. Abidi, “A 900 MHz CMOS LC-Oscillator
with Quadrature Outputs,” IEEE Int. Solid-State Circ. Conf. (ISSCC) Digest, pp. 392–393,Feb. 1996.
5. X. Gao, E. A. M. Klumperink and B. Nauta, “Advantages of Shift Registers Over DLLs for
Flexible Low Jitter Multiphase Clock Generation,” IEEE Trans. on Circuits and Systems – II:Express Briefs, pp. 244–248, vol. 55, no. 2, Mar. 2009.
6. J. M. Chou, Y. T. Hsieh and J. T. Wu, “Phase Averaging and Interpolation Using Resistor
Strings or Resistor Rings for Multi-Phase Clock Generation,” IEEE Trans. on Circuits andSystems – I: Regular Papers, vol. 53, no. 5, pp. 984–991, May 2006.
7. M. J. Gingell, “Single-Sideband Modulation Using Asymmetric Polyphase Networks,” Elect.Commun. Vol. 48, pp. 21–25, 1977.
8. F. Behbahani, Y. Kishigami, J. Leete and A. A. Abidi, “CMOS Mixers and Polyphase Filters
for Large Image Rejection,” IEEE J. of Solid-State Circuits (JSSC), pp. 873–887, vol. 36, no.6, June 2001.
9. K. H. Kim, P. W. Coteus, D. Dreps, S. Kim, S. V. Rylov and D. J. Friedman, “A 2.6 mW
370 MHz-to-2.5 GHz Open-Loop Quadrature Clock Generator,” IEEE Int. Solid-State Circ.Conf. (ISSCC) Digest, pp. 458–627, Feb. 2008.
10. K. H Kim, D. M. Dreps, F. D. Ferraiolo, P. W. Coteus, S. Kim, S. V. Rylov, D. J. Friedman,
“A 5.4 mW 0.0035 mm2 0.48psrms-Jitter 0.8-to-5 GHz Non-PLL/DLL All-Digital Phase
Generator/Rotator in 45 nm SOI CMOS,” IEEE Int. Solid-State Circ. Conf. (ISSCC) Digest,pp. 98–99,99a, Feb. 2009.
A.2 Mathematical Model of Open-Loop Multi-phase LO Generator 137
11. K.-F. Un, P.-I. Mak and R. P. Martins, “Analysis and Design of Open-Loop Multi-Phase
Local-Oscillator Generator for Wireless Applications,” IEEE Trans. on Circuits and Systems –I: Regular Papers, vol. 57, no. 5, pp. 970–981, May 2010.
12. A. Rezayee and K. Martin, “A Three-Stage Coupled Ring Oscillator with Quadrature
Outputs,” in Proc. IEEE Int. Symp. on Circuits and Systems (ISCAS), pp.484–487, May 2001.
13. P. R. Gray and R. G. Meyer, Analysis and Design of Analog Integrated Circuits, Third.Ed.,New York: Wiley, 1993.
14. T. Wu, K. Mayaram, and U. Moon, “ An on-chip calibration technique for reducing supply
voltage sensitivity in ring oscillators,” IEEE J. Solid-State Circuits (JSSC), pp. 775–783,vol. 42, no. 4, Apr. 2007.
15. Y.-T. Huang, C. M. Yang, S. C. Huang, H. L. Pan, T. C. Hung, “A 1.2 V 67 mW 4 mm2Mobile
ISDB-T Tuner in 0.13 mm CMOS,” IEEE Int. Solid-State Circ. Conf. (ISSCC) Digest,pp. 124–125, Feb. 2009.
138 Appendix: Open-Loop Multiphase LO Generators
Biography of Authors
Pui-In Mak (S’00-M’08-SM’11) received the Ph.D. degree from University of
Macau (UM), Macao, China, in 2006. He is currently Associate Professor at UM.
He has been with the UM State-Key Laboratory of Analog and Mixed-Signal VLSI
as Research Assistant (2003–2006), Invited Research Fellow (2006–2008) and
Co-Faculty-in-Charge (2008–Present) coordinating the wireless and biomedical
research lines. He has a short-term work in Chipidea Microelectronics (’03), and
was a Visiting Fellow/Scholar at University of Cambridge, UK (2009), INESC-ID,
Portugal (2009) and University of Pavia, Italy (2010). His current research interests
are on analog, mixed-signal and RF circuits and systems for wireless, biomedical
and physical chemistry, and engineering education.
ProfessorMak authored two books: Analog-Baseband Architectures and Circuitsfor Multistandard and Low-Voltage Wireless Transceivers (Springer 2007), and
High-/Mixed-Voltage Analog and RF Circuit Techniques for Nanoscale CMOS(Springer 2012) and 90+ papers in journals and conferences. He holds three US
patents. He co-holds numerous research grants supported by Macau Government,
National Natural Science Foundation of China and UM. He is a frequent presenter at
leading fora such as DAC, ISSCC, ISCAS, VLSI and CICC, and has published
papers in JSSC, TCAS, TMTT and CASM, etc.
139
His service includes: Appointed/Elected Member of IEEE Circuits and Systems
Society (CASS) Board-of-Governors (2007–2011); Associate Editor of IEEE
POTENTIALS (2012–2014); Associate Editor of IEEE TRANSACTIONS ON CIRCUITS AND
SYSTEMS II – EXPRESS BRIEFS (2010–2013); Associate Editor of IEEE TRANSACTIONS ON
CIRCUITS AND SYSTEMS I – REGULAR PAPERS (2010–2011); Associate Editor of IEEE
CIRCUITS AND SYSTEMS SOCIETY NEWSLETTER (2010–Present); Member of CASS Publi-
cation Activities (2009–2011); Member of IEEE CASS CASCOM (2008–Present)
and CASEO (2009–Present) Technical Committees; Organization or Technical
Committee Member of AVLSIWS’04, APCCAS’08, PrimeAsia’09–’11,
ISCAS’10, VLSI-SoC’11, RFIT’11, SENSORS’11, APCCAS’12 and ISCAS’15.
He co-initiated the three special GOLD sessions in ISCAS’09–’11.
His paper awards include: ASICON Student Paper Award’03; MWSCAS Stu-
dent Paper Award’04; IEEJ VLSI Workshop Best Paper Award’04, DAC/ISSCC
Student Paper Award’05 and CASS Outstanding Young Author Award’10. He was
the (co)-recipient of University of Cambridge Visiting Fellowship’09; IEEE MGA
GOLD Achievement Award’09; CASS Chapter-of-the-Year Award’09; UM
Research Award’10; UM Academic Staff Award’11 and the National Scientific
and Technological Progress Award’11.
Professor Mak was decorated with the Honorary Title of Value for scientific
merits by the Macau Government in 2005.
140 Biography of Authors
Rui Paulo da Silva Martins (IEEE Member’88 – Senior Member’99 –
Fellow’08), born in April 30, 1957, received the Bachelor (5-years), the Masters,
and the Ph.D. degrees as well as the Habilitation for Full-Professor in electrical
engineering and computers from the Department of Electrical and Computer
Engineering, Instituto Superior Tecnico (IST), TU of Lisbon, Portugal, in 1980,
1985, 1992 and 2001, respectively. He has been with the Department of Electrical
and Computer Engineering/IST, TU of Lisbon, since October 1980.
Since 1992, he has been on leave from IST, TU of Lisbon, and is also with the
Department of Electrical and Computer Engineering, Faculty of Science and
Technology (FST), University of Macau (UM), Macao, China, where he is a Full-
Professor since 1998. In FST he was the Dean of the Faculty from 1994 to 1997 and
he has been Vice-Rector of the University of Macau since 1997. From September
2008, after the reform of the UMCharter, he was nominated after open international
recruitment as Vice-Rector (Research) until August 31, 2013. Within the scope of
his teaching and research activities he has taught 20 bachelor and master courses
and has supervised 24 theses, Ph.D. (11) and Masters (13). He has published: 16
books, co-authoring (5) and co-editing (11), plus 5 book chapters; 204 refereed
papers, in scientific journals (38) and in conference proceedings (166); as well as
other 70 academic works, in a total of 295 publications. He has co-authored three
US Patents (1 issued in 2009 and 2 in 2011) and has also submitted other 4. He has
created the Analog and Mixed-Signal VLSI Research Laboratory of UM: http://
www.fst.umac.mo/en/lab/ans_vlsi/website/index.html, recently elevated to State
Key Lab of China (the first in Engineering in Macao), being its Founding Director.
Professor Rui Martins is an IEEE Fellow, was the Founding Chairman of IEEE
Macau Section from 2003 to 2005, and of IEEE Macau Joint-Chapter on Circuits
And Systems (CAS)/Communications (COM) from 2005 to 2008 [2009World Chap-ter of the Year of the IEEE Circuits And Systems Society (CASS)]. He was the GeneralChair of the 2008 IEEE Asia-Pacific Conference on Circuits And Systems –
APCCAS’2008, and was elected Vice-President for the Region 10 (Asia, Australia,
the Pacific) of IEEE CASS, for the period of 2009–2012. He is Associate Editor of theIEEE Transactions on Circuits and Systems II: Express Briefs, for the period of
Biography of Authors 141
2010–2011. He was the recipient of two government decorations: the Medal of
Professional Merit from Macao Government (Portuguese Administration) in 1999,
and the Honorary Title of Value fromMacao SARGovernment (Chinese Administra-
tion) in 2001. In July 2010 was elected, unanimously, as Corresponding Member of
the Portuguese Academy of Sciences, Lisbon, Portugal.
142 Biography of Authors
Index
AAnalog, 1–4, 6, 9–33, 58, 74, 83, 116, 121, 122
Analog-to-digital converter (ADC), 24, 27, 29,
83, 122
Area, 1, 7, 9, 10, 13, 29, 31, 46, 72, 73, 77, 78, 81,
83, 86, 103, 105, 108, 113, 116, 121–123
Attenuator, 64, 65
BBalun, 6, 7, 14–15, 36, 37, 41, 43, 55, 59, 60,
72, 73, 81–96, 99, 116, 117
Bandwidth (BW), 6, 9, 15, 22, 24, 31, 42–44,
47, 52, 55, 59, 61, 63, 65, 69, 70, 72, 78,
81–83, 86, 88, 92, 99, 105, 113, 115,
116, 122, 123
Baseband (BB), 10, 18, 27–29, 31, 74, 83, 84,
96, 99, 101, 104, 110, 112, 113,
115–117, 122
Battery, 1, 10
BB. See Baseband (BB)
BW. See Bandwidth (BW)
CCircuit, 1, 2, 4, 6, 9–14, 16, 18, 20, 29–31,
33, 36–46, 50, 56, 58–74, 77, 78, 81,
96–101, 103, 105–108, 121, 122
CMOS, 1, 2, 4, 6, 7, 9–12, 14, 16, 17, 33,
35–53, 55–78, 81–118, 121–123
Cognitive radios (CRs), 9, 122, 123
Common-gate amplifier, 81
Common-source (CS) amplifier, 3, 4, 62, 81, 85
Constant transconductance bias, 6, 35, 44–45
Corner effect, 48
Cost, 29, 41, 55, 78, 81, 121
CRs. See Cognitive radios (CRs)CS amplifier. See Common-source (CS)
amplifier
Current, 6, 7, 11, 12, 15–19, 21, 22, 26, 27, 29,
31, 35, 36, 39–41, 44–47, 51, 53, 60, 64,
66, 68, 72, 73, 81, 83–102, 104,
115–118, 121, 122
balancer, 85–88, 116
domain, 85, 86, 117, 118
reuse, 6, 7, 35, 39–41, 83, 96–102, 104,
115, 117
DDC-DC converter, 123
Delay, 32, 47, 92, 103
Digital multimedia broadcasting–terrestrial
(DMB-T), 5, 81
Digital video broadcasting–handheld
(DVB-H), 5, 6, 45, 47, 57, 58, 70, 74, 81
Diode, 37–39, 45, 46, 52, 64, 67, 73
Direct conversion, 36, 55, 99
DMB-T. See Digital multimedia
broadcasting–terrestrial (DMB-T)
DVB-H. See Digital videobroadcasting–handheld (DVB-H)
DVB-H, 5, 6, 45, 57, 58, 70, 74, 81
Dynamic range, 2, 4, 9, 15, 59, 86, 92, 113,
115, 122
EElectrostatic discharge protection (ESD), 6, 11,
35–53, 64, 65, 74, 77, 84, 93, 121
143
FFolded-cascode, 18–27
Frequency divider, 83, 103, 123
GGain-bandwidth product, 19
Gain compensation, 57, 63, 70, 72–73, 78
Gain precision, 26, 27, 122
Global system for mobile (GSM), 57, 73, 74
Gyrator-C Biquad, 100
HHarmonic rejection, 7, 36, 57, 83
IInductor, 15, 70, 72, 84, 85, 116, 123
Integrated services digital broadcasting–
terrestrial (ISDB-T), 5, 6, 81
Inverter, 6, 55, 58, 60–63, 66, 72, 85, 86, 96,
102, 103, 105, 106
Inverter amplifier, 60, 72, 85, 86
ISDB-T. See Integrated services digital
broadcasting-terrestrial (ISDB-T)
KKT/C noise, 122
LLevel shifting, 28, 29, 122
Linearity, 2, 7, 10, 15, 16, 24, 26, 27, 31, 35,
41, 42, 47, 48, 51, 56, 57, 60, 61, 63, 64,
67, 68, 76, 78, 81, 83, 85, 86, 88, 91, 93,
96, 99, 100, 109, 116
Line driver, 9, 13, 32–33
LNA. See Low-noise amplifier (LNA)
Low-dropout (LOD) regulator, 29–31,
122, 123
Low-noise amplifier (LNA), 6, 7, 9, 12, 14–15,
35–53, 55, 56, 58–65, 69, 72, 74, 81–96,
99, 115, 116, 121
Lowpass filter, 7, 15, 21, 83, 116
MMatching, 2, 37, 40, 43, 44, 81, 85, 87, 89, 91,
93, 116, 122
Mixer, 14, 16, 48, 57, 66, 68, 70, 78, 83, 84, 96,
97, 99, 101, 105, 109, 115, 116, 118, 122
Mobile TV, 2, 5–7, 35–53, 55–78, 81–118, 121
Modeling, 41, 46, 66, 67, 83, 89, 91, 100,
102, 106
Multi-phase local oscillator generator, 7
NNoise, 3, 6, 7, 17, 21, 26, 27, 29, 35, 41, 55–57,
60, 81, 83–85, 89, 91, 93, 96, 99–102,
108, 116, 122
Noise cancellation, 6, 35, 41–42, 53, 60, 81,
84–85
OOperational amplifier, 9, 10, 15, 17–27
Overdrive protection, 9, 93
PPassive attenuator, 64
Phase corrector, 103, 105, 108
Phase-locked loop (PLL), 83, 103, 105, 113
Phase noise, 102, 103, 108, 113, 114
PLL. See Phase-locked loop (PLL)
Power, 1–4, 6, 7, 9–15, 17, 19–22, 27, 29, 31,
33, 35, 39, 42, 44–46, 48, 51–53, 55, 57,
61, 64, 68, 70, 77, 78, 81–83, 85, 87, 88,
91, 92, 100, 103–105, 109, 110, 113,
116, 121–123
amplifier, 9–12, 14–15, 33, 83
efficiency, 1, 6, 9, 35, 55, 58, 81, 122
Preselect filter, 6, 55–57, 70–75
RRadio frequency (RF), 1–4, 6, 7, 9–33, 36, 37,
42, 46–51, 53, 55–78, 81, 84, 86, 105,
115–117, 121, 122
Receiver front-end, 7, 81–118, 121
Recycling folded-cascode (RFC), 18–27
Reliability, 2, 4, 7, 9–12, 14, 18–20, 22, 32, 33,
45, 53, 55, 60–63, 66, 68, 74, 78, 87,
118, 121
RF. See Radio frequency (RF)
RFC. See Recycling folded-cascode (RFC)
Robustness, 48, 108
SSample-and-hold, 9, 31, 122
Simulation, 17, 26, 42, 46–52, 64, 68, 74, 77,
101–102, 104–109
144 Index
Slew rate, 21, 27, 122
Software defined radio, 2, 9
Supply, 2, 5, 10, 15, 16, 19, 32, 37, 38, 45, 46,
48, 50, 52, 59–62, 64, 66, 72, 73, 78, 86,
92, 93, 96, 106, 108, 116, 117, 121, 122
Supply voltage, 2, 5, 9, 10, 37, 48, 50, 87,
116, 121
System, 1, 2, 4, 6, 9–33, 41, 74, 102, 121, 123
TT-DMB. See Terrestrial–digital multimedia
broadcasting (T-DMB)
Temperature effect, 4, 11, 74
Terrestrial–digital multimedia broadcasting
(T-DMB), 5, 6, 81
Time-adaptive design, 123
TV-GSM co-operation, 6, 55–58, 73, 74, 77,
78, 121
UUltra high frequency (UHF), 5, 6, 55–57, 81,
83, 84, 107, 116
Ultra-wideband (UWB), 14, 35, 51
VVCO. See Voltage-controlled oscillator (VCO)
Very high frequency (VHF), 6, 55
Voltage-controlled oscillator (VCO), 29, 30,
83, 102, 103, 105, 113, 123
Voltage domain, 97
WWideband, 2, 4, 7, 10, 14–15, 36, 50, 52,
56, 58, 59, 72, 81–96, 113, 116, 117,
122, 123
Wireless technology, 1–2, 9
Index 145