circuit level models of cmos technology transistors prof. john choma, jr. university of southern...
TRANSCRIPT
Circuit Level Models Of CMOS Circuit Level Models Of CMOS Technology TransistorsTechnology Transistors
Prof. John Choma, Jr.
University of Southern CaliforniaDepartment of Electrical Engineering-Electrophysics
University Park: MC: 0271Los Angeles, California 90089-0271
213-740-4692 (USC )626-915-0944 (FAX )
818-384-1552 (CELL )[email protected]
Fall 2001 Semester
Circuit Level Models Of CMOS Circuit Level Models Of CMOS Technology TransistorsTechnology Transistors
Prof. John Choma, Jr.
University of Southern CaliforniaDepartment of Electrical Engineering-Electrophysics
University Park: MC: 0271Los Angeles, California 90089-0271
213-740-4692 (USC )626-915-0944 (FAX )
818-384-1552 (CELL )[email protected]
Fall 2001 Semester
EE 348: Lecture #04EE 348: Lecture #04
Canonic Cells OfAnalog MOS/CMOS Technology
Prof. John Choma, Jr.
University of Southern CaliforniaDepartment of Electrical Engineering-Electrophysics
University Park: MC: 0271Los Angeles, California 90089-0271
213-740-4692 (USC )626-915-0944 (FAX )818-384-1552 (CELL )
[email protected] [email protected]
Spring 2003 Semester
Canonic Cells OfAnalog MOS/CMOS Technology
Prof. John Choma, Jr.
University of Southern CaliforniaDepartment of Electrical Engineering-Electrophysics
University Park: MC: 0271Los Angeles, California 90089-0271
213-740-4692 (USC )626-915-0944 (FAX )818-384-1552 (CELL )
[email protected] [email protected]
Spring 2003 Semester
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 2
Lecture OverviewLecture OverviewLecture OverviewLecture Overview
Common Source Amplifier Model Low Frequency Performance High Frequency Performance
Broadband RF Amplifier Transfer Function I/O Impedances Design Scenario
Source Follower Model High Frequency Compensation
Common Gate Model High Frequency Performance Cascode Configurations
Common Source Amplifier Model Low Frequency Performance High Frequency Performance
Broadband RF Amplifier Transfer Function I/O Impedances Design Scenario
Source Follower Model High Frequency Compensation
Common Gate Model High Frequency Performance Cascode Configurations
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 3
CMOS Common Source AmplifierCMOS Common Source AmplifierCMOS Common Source AmplifierCMOS Common Source Amplifier
Assumptions Transistors Are Saturated: Transistor Drain And Source Series Resistances Are Negligible
Objectives Voltage Gain Transfer Function Frequencies Of Poles And Zeros
Assumptions Transistors Are Saturated: Transistor Drain And Source Series Resistances Are Negligible
Objectives Voltage Gain Transfer Function Frequencies Of Poles And Zeros
gg s hn1 o bias hn2V V V V V V
V g g
R s
V s
V b ia s
C l
V o
V d d
Schem atic Diagram
M 1
M 2
R s
V s
C l
V o s
AC Schem atic Diagram
M 1
M 2
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 4
CMOS Small Signal ModelCMOS Small Signal ModelCMOS Small Signal ModelCMOS Small Signal Model
Capacitances For Deep Submicron Technology Input Capacitance (Tens Of fF): Feedback Capacitance (Few fF): Output Capacitance (Tens -To- Hundreds Of fF):
Resistance Parameter (A Few -To- Tens Of K):
Capacitances For Deep Submicron Technology Input Capacitance (Tens Of fF): Feedback Capacitance (Few fF): Output Capacitance (Tens -To- Hundreds Of fF):
Resistance Parameter (A Few -To- Tens Of K):
i gs1 ols1C C C
o o1 o2r r r
o db1 db2 gd2 old2 lC C C C C C
R s
V s
V s
C l
V o s
AC Schem atic Diagram
M 1
M 2V o s
g Vm 1 1 r o
C f
C i C oV 1
R s
Sm all Signal M odel
f gd1 old1C C C
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 5
CMOS Amplifier Voltage GainCMOS Amplifier Voltage GainCMOS Amplifier Voltage GainCMOS Amplifier Voltage Gain
PoleRelationships
Zero
Zero Frequency Gain Magnitude
PoleRelationships
Zero
Zero Frequency Gain Magnitude
V s
V o s
g Vm 1 1 r o
C f
C i C oV 1
R s
vfos
vs
1 2
sA (0 ) 1
zVA ( s )
V s s1 1
p p
vv 2
A (0 ) 1 csA ( s )
1 as bs
f
f m1
C1c
z g
s i o o o m1 o s f1 2
f fs i o o
1 2 i o
1 1a R C r C r 1 g r R C
p p
C C1b R C r C 1
p p C C
v m1 oA (0 ) g r
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 6
CMOS Amplifier Time ConstantsCMOS Amplifier Time ConstantsCMOS Amplifier Time ConstantsCMOS Amplifier Time Constants
Time Constant Due To Poles (“a”) Sum Of Open Circuit Time Constants With Source Nulled Time Constant Of Any One Capacitor Is Computed With All Other
Capacitances Open Circuited And With Source Signal Nulled Time Constant Due To Zeros (“c”)
Sum Of Open Circuit Time Constants With Response Nulled Time Constant Of Any One Capacitor Is Computed With All Other
Capacitances Open Circuited And With Response Nulled Foregoing Is General Time Moment Theory For Any nth Order
Circuit
Time Constant Due To Poles (“a”) Sum Of Open Circuit Time Constants With Source Nulled Time Constant Of Any One Capacitor Is Computed With All Other
Capacitances Open Circuited And With Source Signal Nulled Time Constant Due To Zeros (“c”)
Sum Of Open Circuit Time Constants With Response Nulled Time Constant Of Any One Capacitor Is Computed With All Other
Capacitances Open Circuited And With Response Nulled Foregoing Is General Time Moment Theory For Any nth Order
Circuit
V s
V o s
g Vm 1 1 r o
C f
C i C oV 1
R s
vv 2
A (0 ) 1 csA ( s )
1 as bs
v m1 oA (0 ) g r
s i o o o m1 o s f
f m1
a R C r C r 1 g r R C
c C g
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 7
0
V o s
g Vm 1 1 r oV 1
R s
Due To C i
0
V o s
g Vm 1 1 r oV 1
R s
Due To C o
0
g Vm 1 1 r oV 1
V xR s
Due To C f
I x
I x
I x
V o s
Time Constant Due To PolesTime Constant Due To PolesTime Constant Due To PolesTime Constant Due To Poles
1ti s
x
i ti i s i
VR R
I
a R C R C
osto o
x
o to o o o
VR r
I
a R C r C
xtf o m1 o s
x
f tf f o m1 o s f
VR r 1 g r R
I
a R C r 1 g r R C
a
ai +
ao +
af
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 8
Time Constant Due To ZerosTime Constant Due To ZerosTime Constant Due To ZerosTime Constant Due To Zeros
V s
0
g Vm 1 1 r oV 1
R s
Due To C i
V s
0
g Vm 1 1 r oV 1
R s
Due To C o
V s
g Vm 1 1 r oV 1
V xR s
Due To C f
I x
I x
I x
0
c
ci +
co +
cf
1ni
x
i ni i
VR 0
I
c R C 0
nox
o no o
0R 0
I
c R C 0
xnf m1
x
f nf f f m1
VR 1 g
I
c R C C g
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 9
Bandwidth EstimationBandwidth EstimationBandwidth EstimationBandwidth Estimation
Approximate Bandwidth (B) Assume |cB| << 1 (Frequency Of Zero Is Much Larger Than B) Assume bB2 << 1 (Frequency Of High Frequency Pole Is Large)
Comments Crude First Order Bandwidth Estimate
Requires Real Poles OnlyRequires Nominally Dominant Pole Frequency Response
Practical Design GuidelineMay Not Be Supremely Accurate Bandwidth Estimate But Small “a” Is A
Necessary (But Not Sufficient) Condition For Broadband ResponseAn Examination Of The Constituent Terms Of “a” Identifies Energy
Storage Elements/Time Constants That Dominantly Limit Theoretically Achievable 3-dB Bandwidth
Approximate Bandwidth (B) Assume |cB| << 1 (Frequency Of Zero Is Much Larger Than B) Assume bB2 << 1 (Frequency Of High Frequency Pole Is Large)
Comments Crude First Order Bandwidth Estimate
Requires Real Poles OnlyRequires Nominally Dominant Pole Frequency Response
Practical Design GuidelineMay Not Be Supremely Accurate Bandwidth Estimate But Small “a” Is A
Necessary (But Not Sufficient) Condition For Broadband ResponseAn Examination Of The Constituent Terms Of “a” Identifies Energy
Storage Elements/Time Constants That Dominantly Limit Theoretically Achievable 3-dB Bandwidth
os vv 2
s
V A (0 ) 1 csA ( s )
V 1 as bs
v
v 2
A (0 ) 1 j cA ( j )
1 b j a
v vA ( jB ) A (0 ) 2
i o f1 2
1 1 1B
a a a a1 p 1 p
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 10
Frequency Response MetricsFrequency Response MetricsFrequency Response MetricsFrequency Response Metrics
CriticalFrequencyRelationships
Unity GainFrequency - u
Pole Dominance Metric kp = p2/u
kp > 1 Implies Amplifier Is A Dominant Pole Circuit
CriticalFrequencyRelationships
Unity GainFrequency - u
Pole Dominance Metric kp = p2/u
kp > 1 Implies Amplifier Is A Dominant Pole Circuit
V s
V o s
g Vm 1 1 r o
C f
C i C oV 1
R s
vfos
vs
1 2
sA (0 ) 1
zVA ( s )
V s s1 1
p p
v m1 o u v 1 p 2 uA (0 ) g r A (0 ) p k p @ @
s i o o o m1 o s f1 2
f fs i o o f f m1
1 2 i o
1 1R C r C r 1 g r R C
p p
C C1R C r C 1 z C g
p p C C
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 11
Negligible Feedback CapacitanceNegligible Feedback CapacitanceNegligible Feedback CapacitanceNegligible Feedback Capacitance
Poles
Zero At Infinity
Unity Gain Frequency For p2 >> p1
Pole Dominance Metric
Comments Negligible Feedback
Presumption Valid When No Feedback Capacitance Is Appended Pole Dominance Requires gm1Rs < Co/Ci
Dominant Pole Response Requires kp > 1Typical For First Stage; Atypical For Interstages
Unity Gain Frequency Is Effectively The Ratio Of Driver Transconductance -To- Output Capacitance Ratio
Poles
Zero At Infinity
Unity Gain Frequency For p2 >> p1
Pole Dominance Metric
Comments Negligible Feedback
Presumption Valid When No Feedback Capacitance Is Appended Pole Dominance Requires gm1Rs < Co/Ci
Dominant Pole Response Requires kp > 1Typical For First Stage; Atypical For Interstages
Unity Gain Frequency Is Effectively The Ratio Of Driver Transconductance -To- Output Capacitance Ratio
s i o o o m1 o s f s i o o1 2
1 1R C r C r 1 g r R C R C r C
p p
1 o o
2 s i
p 1 r C
p 1 R C
f fs i o o s i o o
1 2 i o
C C1R C r C 1 R C r C
p p C C
u v 1 m1 oω A (0)p g C
o2p
u m1 s i
Cpk
ω g R C@
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 12
Significant Feedback CapacitanceSignificant Feedback CapacitanceSignificant Feedback CapacitanceSignificant Feedback Capacitance Poles/
Zero
Unity Gain Frequency
Pole Dominance Metric
Comments Significant Feedback
Presumption Valid When Feedback Capacitance Is Appended, As In Pole Splitting Compensation
Pole Dominance Requires Large Source Resistancekp > 1Typical For Interstage; Atypical For Front End First Stage
Unity Gain Frequency Is Determined By Time Constant Formed Of Source Resistance And Feedback Capacitance
Poles/Zero
Unity Gain Frequency
Pole Dominance Metric
Comments Significant Feedback
Presumption Valid When Feedback Capacitance Is Appended, As In Pole Splitting Compensation
Pole Dominance Requires Large Source Resistancekp > 1Typical For Interstage; Atypical For Front End First Stage
Unity Gain Frequency Is Determined By Time Constant Formed Of Source Resistance And Feedback Capacitance
s i o o o m1 o s f m1 o s f 11 2
1 1R C r C r 1 g r R C g r R C 1 p
p p
f fs i o o
1 2 i o
C C1R C r C 1
p p C C
f fi o
i o
2 m1 f
C CC C 1
C C1
p g C
m1 sp
o i i
f o r
g Rk
C C C1
C C C
f m1 f m1 s uz g C g R ω
us f
1ω
R C
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 13
Frequency Response: UnstableFrequency Response: UnstableFrequency Response: UnstableFrequency Response: Unstable
A (0)v
u v/A (0 ) k p u
k o u
u
0
| |A (j ) v (in dB )
6 dB /O c tave
12 d B /O c ta ve
6 dB /O c tave
Potentially Unstable BecauseSecond Pole Lies At Frequency That Is Smaller Than Unity GainFrequency; That Is, kp < 1
v vf o uos
vs v
1 2 u p u
s sA (0 ) 1 A (0 ) 1z k ωV
A ( s )V A (0 )ss s s1 1 1 1
p p ω k ω
u v 1
2 p u
f o u
ω A (0 )p
p k ω
z k ω
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 14
Frequency Response: StableFrequency Response: StableFrequency Response: StableFrequency Response: Stable
A (0)v
u v/A (0 )
k p u k o u
u
0
| |A (j ) v (in dB )
6 dB /O c tave
12 d B /O c ta ve
6 dB /O c tave
Likely To Be Stable BecauseSecond Pole Lies At Frequency That Is Larger Than Unity GainFrequency; That Is, kp > 1
v vf o uos
vs v
1 2 u p u
s sA (0 ) 1 A (0 ) 1z k ωV
A ( s )V A (0 )ss s s1 1 1 1
p p ω k ω
u v 1
2 p u
f o u
ω A (0 )p
p k ω
z k ω
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 15
R s s
R o
R i
R s
R f
V s R l
+ V g g
V d d
V o
R in
R o u t
R f
R in
V s
V o s
g Vm 1 R o
R s sV 2
R s
R i
b m 2g V R l
R o u t
I
V 1
Match-Terminated Common SourceMatch-Terminated Common SourceMatch-Terminated Common SourceMatch-Terminated Common Source
Assumptions Very Large Transistor Channel Resistance Low Frequency Considerations Only For Initial Study Biasing Network Has Been Simplified To Expedite Analyses
Match Termination Circuit Can Be Designed For Rin = Rout When Rs = Rl
Match Termination Implies Rs = Rin = Rout = Rl R Match Termination Useful For Cascaded Stages
Assumptions Very Large Transistor Channel Resistance Low Frequency Considerations Only For Initial Study Biasing Network Has Been Simplified To Expedite Analyses
Match Termination Circuit Can Be Designed For Rin = Rout When Rs = Rl
Match Termination Implies Rs = Rin = Rout = Rl R Match Termination Useful For Cascaded Stages
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 16
Match-Terminated ConceptMatch-Terminated ConceptMatch-Terminated ConceptMatch-Terminated Concept
Terminations Single Stage: Rs = Rin = Rout = Rl R Cascade: For All Three Stages, Rs = Rin = Rout = Rl R
Comments Loading Effects On Each Stage Effectively Eliminated Reminiscent Of Transmission Line Loaded In Its Characteristic Impedance Results In Maximum Power Transfer At All I/O Ports
Terminations Single Stage: Rs = Rin = Rout = Rl R Cascade: For All Three Stages, Rs = Rin = Rout = Rl R
Comments Loading Effects On Each Stage Effectively Eliminated Reminiscent Of Transmission Line Loaded In Its Characteristic Impedance Results In Maximum Power Transfer At All I/O Ports
R
V sR
V o
R
M a tchTerm in a tedA m p lifie r
R
R
V sR
V o
R
M a tchTerm in a tedA m p lifie r
R R
M a tchTerm in a tedA m p lifie r
R R
M a tchTerm in a tedA m p lifie r
R
Gain: Av = Vos/Vs
Gain: Vos/Vs = Av3
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 17
Match-Terminated AnalysisMatch-Terminated AnalysisMatch-Terminated AnalysisMatch-Terminated Analysis
Critical Parameter IsGf = 1/Rf
Performance Indices
Return Ratios RT: Thévenin Resistance Seen
By Rf With Vs = 0 RTO: Thévenin Resistance Seen
By Rf With Vos = 0
Critical Parameter IsGf = 1/Rf
Performance Indices
Return Ratios RT: Thévenin Resistance Seen
By Rf With Vs = 0 RTO: Thévenin Resistance Seen
By Rf With Vos = 0
TO s l fov f s l os
s T s l f
T l fin f l ino
T l f
T s fout f s outo
T s f
1 R R ,R RVA G ,R ,R A
V 1 R R ,R R
1 R 0,R RR G , R R
1 R ,R R
1 R R ,0 RR G , R R
1 R R , R
R f
R in
V s
V o s
g Vm 1 R o
R s sV 2
R s
R i
b m 2g V R l
R o u t
I
V 1
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 18
I/O ImpedancesI/O ImpedancesI/O ImpedancesI/O Impedances
Straightforward Substitution Of Results Into I/O Equations
Comments If Ri = Ro And Rs = Rl , Rin Rout
More Commonly, Ri >> Rs And Ro >> Rl
Then Ri = Rin = Rout = Rl R Requires Rf = gmeR2
Note That Match Terminated Assumption Mandates R << Rs, Rl
Straightforward Substitution Of Results Into I/O Equations
Comments If Ri = Ro And Rs = Rl , Rin Rout
More Commonly, Ri >> Rs And Ro >> Rl
Then Ri = Rin = Rout = Rl R Requires Rf = gmeR2
Note That Match Terminated Assumption Mandates R << Rs, Rl
R s s
R o
R i
R s
R f
V s R l
+ V g g
V d d
V o
R in
R o u t
f o lin i
me o l
f i sout o
me i s
R R RR R
1 g R R
R R RR R
1 g R R
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 19
Match-Terminated GainMatch-Terminated GainMatch-Terminated GainMatch-Terminated Gain
Straightforward Substitution Of Results Into Gain Equation
Design Calculate gme For Desired Gain And Required Match Resistance Calculate Rss Based On Requisite gme
Calculate Required Rf Based On Required R And Computed gme
Straightforward Substitution Of Results Into Gain Equation
Design Calculate gme For Desired Gain And Required Match Resistance Calculate Rss Based On Requisite gme
Calculate Required Rf Based On Required R And Computed gme
R s s
R o
(b ig )
R i
(b ig )
R
g Rm e
2
V s R
+ V g g
V d d
V o
R
R os mev
s
V g R 1A
V 2
m
meb m ss
gg
1 1 λ g R @
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 20
Z o u t
(b - a )R(b - a )R
a
R
Inductive Output ImpedanceInductive Output ImpedanceInductive Output ImpedanceInductive Output ImpedanceMatch Terminated Output Impedance
Inductive Because Of Gate-Source Capacitance, Cgs
Potentially Troublesome With Capacitive Loads
Computation R Is Zero
FrequencyOutput Impedance
a: Time Constant Of Cgs With CurrentIx Constrained To Zero
b: Time Constant Of Cgs With VoltageVx Constrained To Zero
InductiveIf b > a
Match Terminated Output Impedance Inductive Because Of Gate-Source Capacitance, Cgs
Potentially Troublesome With Capacitive Loads
Computation R Is Zero
FrequencyOutput Impedance
a: Time Constant Of Cgs With CurrentIx Constrained To Zero
b: Time Constant Of Cgs With VoltageVx Constrained To Zero
InductiveIf b > a
R s s
R o
(b ig )
R i
(b ig )
R
g Rm e
2
+ V g g
V d d
V x
Z o u t
I x
C g s
effL b a R
xout
x
V 1 bsZ R
I 1 as
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 21
Inductance ComputationInductance ComputationInductance ComputationInductance Computation
ComputationsComputations
R f
g Vm y
R s sV 2
R
b m 2g V
I x
C g s
I y
V y
= g Rm e
2
V x
Effective Inductance Effective Inductance
2me ss
ss gsme me
effm ss
3me gs 2v
eff gsme v
g R RR R C
1 g R gL
1 g R
g R C 2 A 1L R C
1 g R 2 A 1
x
x
y gsssgs
y me b m ssI 0
y me ss gsb m ssgs gs
y m ss me m ssV 0
V CR Ra C
I 1 g R 1 1 λ g R
V 1 g R R C1 λ g Rb C RC
I 1 g R g R 1 g R
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 22
Comments On Inductive OutputComments On Inductive OutputComments On Inductive OutputComments On Inductive Output
Presumptions Very Large Channel Resistance (Reasonable For 50 Load) Very Large Input Port Resistance, Ri (Reasonable For 50 Load) Very Large Output Port Resistance, Ro (Reasonable For 50 Load) Negligible Gate-Drain Capacitance (Reasonable For Self-Aligning
Gate Technology And Possibly For Common Gate Cascode) Drain-Bulk Capacitance Absorbed Into Load Termination
Output Port Inductance Effective Inductance Estimate Is Low By About 20% -To- 30% Due
To Foregoing Presumptions Magnitude Is Generally Below A Nanohenrie Or So Can Cause Ringing/Settling Time Problems For Inappropriate
Capacitive Loads Incurs Non-Match-Terminated Conditions At Very High Signal
Frequencies
Presumptions Very Large Channel Resistance (Reasonable For 50 Load) Very Large Input Port Resistance, Ri (Reasonable For 50 Load) Very Large Output Port Resistance, Ro (Reasonable For 50 Load) Negligible Gate-Drain Capacitance (Reasonable For Self-Aligning
Gate Technology And Possibly For Common Gate Cascode) Drain-Bulk Capacitance Absorbed Into Load Termination
Output Port Inductance Effective Inductance Estimate Is Low By About 20% -To- 30% Due
To Foregoing Presumptions Magnitude Is Generally Below A Nanohenrie Or So Can Cause Ringing/Settling Time Problems For Inappropriate
Capacitive Loads Incurs Non-Match-Terminated Conditions At Very High Signal
Frequencies
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 23
R s R s
V s V s
R l
g o
C l
C oC g d
V o
V o s
V d dZ in Z in
Z o u t g Vm a b m bg V R l
C g s
V aV i
Z o u t
V b
Source FollowerSource FollowerSource FollowerSource Follower
Model Modifications
Simplified Model
Model Modifications
Simplified Model
o l sbC C C
a i os
b os
V V V
V V
b m bg VV b
b mg g Vm a
V o s g Vm i
V o s g Vm o s g Vm i
V o s g m
R s
V sC oC g d
V o s
Z in
g Vm i R ll
C g s
V i
outo b m
ll out l
osv m ll
s b
1R
g 1 λ g
R R R
V (0) 1A (0) g R
V (0) 1 λ
@
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 24
Source Follower ModelSource Follower ModelSource Follower ModelSource Follower Model
Source Follower
Common Source
Observation Topologically Identical Models Gain And Pole Frequency Expressions Derive From Respective
Common Source Results With Mere Changes In Circuit Branch Variable Notations
Source Follower
Common Source
Observation Topologically Identical Models Gain And Pole Frequency Expressions Derive From Respective
Common Source Results With Mere Changes In Circuit Branch Variable Notations
R s
V sC oC g d
V o s
Z in
g Vm i R ll
C g s
V i
V s
V o s
g Vm 1 1 r o
C f
C i C oV 1
R s
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 25
Source Follower GainSource Follower GainSource Follower GainSource Follower Gain Transfer Function
Poles And Zeros
Approximation
Transfer Function
Poles And Zeros
Approximation
v fosv
s
1 2
A (0) 1+s ZV (s)A (s)
V (s) s s1+ 1+
p p
ll o s gd ll m ll s gs1 2
gs gsll o s gd
1 2 gd o
gs
f m
1 1R C R C R 1 g R R C
p p
C C1R C R C 1
p p C C
C1
z g
Note Left Half Plane Zero
v m llb
ll l out lo b m
1A (0) g R
1 λ
1R R R R
g 1 λ g
bll o s gd ll s gs
1 2 b
λ1 1R C R C R R C
p p 1 λ
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 26
o gsbll o s gd ll s gs ll o gs
1 2 b 1 b m
gs gsll o s gd s ll o gs gd
1 2 gd o
s o gs gd o m s
2 gs o gs o T
C Cλ1 1 1R C R C R R C R C C
p p 1 λ p 1 λ g
C C1R C R C 1 R R C C C
p p C C
R C C C C g R1
p C C C C ω
Dominant Output Port CapacitanceDominant Output Port CapacitanceDominant Output Port CapacitanceDominant Output Port Capacitance
Approximate Pole And Zero Frequencies
Pole Dominance (p2 >> p1) Requires Implications
Small Source ResistanceLarge Device Unity Gain Frequency
Approximate Pole And Zero Frequencies
Pole Dominance (p2 >> p1) Requires Implications
Small Source ResistanceLarge Device Unity Gain Frequency
gs
f m
C1
z g
mT
gs gd
gω
C C
2
2m m s
T b s ogs o o
g g Rω 1 λ R C
C C C
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 27
Response CompensationResponse CompensationResponse CompensationResponse Compensation
TransferFunction
Feedforward Capacitive Compensation Incorporate Capacitance Cc Across Gate-Source Terminals Choose Cc Such That Cc + Cgs >> Co
May Be Difficult For Large Source-Bulk Or Load Capacitances Requires Negligible Threshold Modulation Factor
Gain And Bandwidth (B) Results
TransferFunction
Feedforward Capacitive Compensation Incorporate Capacitance Cc Across Gate-Source Terminals Choose Cc Such That Cc + Cgs >> Co
May Be Difficult For Large Source-Bulk Or Load Capacitances Requires Negligible Threshold Modulation Factor
Gain And Bandwidth (B) Results
gs
v f b mv
gs o m s o
1 2b m T gs o
sC11
A (0) 1+s Z 1 λ gA (s)
s s s C C sg R C1+ 1+ 1 1p p 1 λ g ω C C
b
vm s o
T c gs o
1 1 λA (s)
sg R C1
ω C C C
c gs oT
m s o
C C CωB
g R C
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 28
Compensated FollowerCompensated FollowerCompensated FollowerCompensated Follower
Transfer Function
Net Output Port Capacitance
Effect Of Feedforward Capacitance Is Effectively To Bypass Transistor At High Signal Frequencies
Transfer Function
Net Output Port Capacitance
Effect Of Feedforward Capacitance Is Effectively To Bypass Transistor At High Signal Frequencies
o l sb1 db2 gd 2C C C C C
R s
V s
C l
V o
V d d
Z o u t
V b ia s
C c
V s s
M 1
M 2
b
vm s o
T c gs o
1 1 λA (s)
sg R C1
ω C C C
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 29
Follower Output ImpedanceFollower Output ImpedanceFollower Output ImpedanceFollower Output Impedance
Dominant Pole/Zero Approximation
Compare Pole/Zero Time Constants Inductive Impedance Requires b > a Capacitive Impedance Requires a > b Calculate “a” As Sum Of Open Circuit
Time Constants With Ix = 0 Calculate “b” As Sum Of Open Circuit
Time Constants With Vx = 0
Dominant Pole/Zero Approximation
Compare Pole/Zero Time Constants Inductive Impedance Requires b > a Capacitive Impedance Requires a > b Calculate “a” As Sum Of Open Circuit
Time Constants With Ix = 0 Calculate “b” As Sum Of Open Circuit
Time Constants With Vx = 0
R s
V d d
Z o u t
C c
V s s
M 1
R s
C s bC g d
V x
g Vm i
I x
R o u t
C + Cc g s
V i
V x
I x
xout out
x
V 1 bsZ R
I 1 as
Z o u t
(b - a )R o u t
(b - a )R o u t
a
R o u t
eff outL b a R
outb m o b m
1 1R
1 λ g g 1 λ g
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 30
Time Constants — ZTime Constants — ZoutoutTime Constants — ZTime Constants — Zoutout
Due To Zero
Due To Pole
EffectiveInductance
Inductive Condition Satisfaction Likely
For Large Source Resistance Satisfaction Likely For Gate-Source Compensation Capacitance
Due To Zero
Due To Pole
EffectiveInductance
Inductive Condition Satisfaction Likely
For Large Source Resistance Satisfaction Likely For Gate-Source Compensation Capacitance
sbm s
c gs
Cg R 1
C C
Z o u t
(b - a )R o u t
(b - a )R o u t
a
R o u tR s
V d d
Z o u t
C c
V s s
M 1
I x
V x
s gd c gsb R C C C
s gd c gs out sb m s out c gsa R C C C R C 1 g R R C C
2eff out out m s c gs sb
L b a R R g R 1 C C C
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 31
C oC g d g Vm i R ll
C + Cc g s
V i
V i
I x
I x
C l
V d dZ in
C c
V s s
M 1
M 2V b ia s
V o s
V o
Source Follower Input AdmittanceSource Follower Input AdmittanceSource Follower Input AdmittanceSource Follower Input Admittance
Input Admittance
Voltage Ratio
Effective Input Capacitance
Input Admittance
Voltage Ratio
Effective Input Capacitance
x gd i gs c gs i os
x osin gd c gs ieff
i i
I sC V s C C C V V
I VY s C C C 1 sC
V V
bos
m s os
T c gs o
1 1 λV (s)
sg R CV (s)1
ω C C C
s
os os
i i bR 0
V V 1
V V 1 λ
bieff gd c gs
b
λC C C C
1 λ
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 32
I o
I o s
I o s
Z in
R l R l
R l
R sI + IQ s
V s s
V b ia s
V d d
Z o u t
C l
C o
C o
C iV i
V i
g Vm a
(1 + )g V b m i
b m ag V g o
g o
I s R s
V
a
V
b
V o s
V o s
C i
V i
I s R s
V
a
V
b
I 1 I 2
V o s
g Vo o sg i
I 1 I 2V i
g Vi i g o
Common Gate AmplifierCommon Gate AmplifierCommon Gate AmplifierCommon Gate Amplifier
o db gd l
i sb gs
i a b
os os l
C C C C
C C C
V V V
V I R
1 b m i o i os
2 b m i o os i
i o b m
I 1 λ g V g V V
I 1 λ g V g V V
g g 1 λ g
@
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 33
Common Gate Model/PerformanceCommon Gate Model/PerformanceCommon Gate Model/PerformanceCommon Gate Model/PerformanceI o
Z in
Z in
Z o u t
R l
R sI + IQ s
V s s
V b ia s
V d d
Z o u t
C l
V i
I o s
R lC o
V o s
V i
I s R s C i g R Io l o s
g Vi i g o
g i
o io ioi
s o l io s l
lin
i i l
sout
o o s
I A (s) A (s)A (s)
I 1 g R A (s) 1 T(s, R , R )
1 T(s, 0, R )1Z (s) =
g sC 1 T(s, , R )
1 T(s, R , 0)1Z (s) =
g sC 1 T(s, R , )
sii
i s
i o b m
loo
o l
s l o l io
RR
1 g R
g g 1 λ g
RR
1 g R
T(s, R , R ) = g R A (s)
@
@
@
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 34
Common Gate Gain ParametersCommon Gate Gain ParametersCommon Gate Gain ParametersCommon Gate Gain Parameters
Open Loop Gain Magnitude: |Aio(jω)| < 1 Dominant Pole Most
Likely Is Established By Co
Loop Gain Magnitude < 1 Magnitude Is Actually
Much Smaller ThanOne And Smaller YetAt High Frequencies
Closed Loop Gain
Open Loop Gain Magnitude: |Aio(jω)| < 1 Dominant Pole Most
Likely Is Established By Co
Loop Gain Magnitude < 1 Magnitude Is Actually
Much Smaller ThanOne And Smaller YetAt High Frequencies
Closed Loop Gain
Z in Z o u t
I o s
R lC o
V o s
V i
I s R s C i g R Io l o s g Vi i g og i
o l
i s
i s o loio
s ii i oo og R 0
g R 1
1 g R 1 g RIA (s)
I 1 sR C 1 sR C
i s o l
i s o ls l o l io
ii i oo o
g R g R
1 g R 1 g RT(s, R , R ) = g R A (s) =
1 sR C 1 sR C
i s
cl ioi s o l oo o
g RA (s) A (s)
1 g R 1 g R 1 sR C
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 35
Common Gate I/O ImpedancesCommon Gate I/O ImpedancesCommon Gate I/O ImpedancesCommon Gate I/O Impedances Input Impedance
OutputImpedance
I/O Impedances Are Both Capacitive And Approximately Dominant Pole Frequency Functions
Input Impedance
OutputImpedance
I/O Impedances Are Both Capacitive And Approximately Dominant Pole Frequency Functions
sii
i s
loo
o l
RR
1 g R
RR
1 g R
@
@
o l i inin
in il o o li
oo o i
o l o lin
i o b m b m
1 g R g Z (0)Z (s)
1 sZ (0)C1 sR C 1 g RsC
1 sR C g
1 g R 1 g R 1Z (0)
g g 1 λ g 1 λ g
i s o outout
out os i i so
ii i o
o b m s b m si sout
o o o
1 g R g Z (0)Z (s)
1 sZ (0)C1 sR C 1 g RsC
1 sR C g
1 g 1 λ g R 1 λ g R1 g RZ (0)
g g g
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 36
Common Gate SummaryCommon Gate SummaryCommon Gate SummaryCommon Gate Summary
Current Gain Frequency Response Less Than Unity Current Gain Magnitude At All Frequencies Dominant Pole Is Established By Net Output Port Capacitance
Input Resistance Is SmallOutput Resistance Is Very Large
I/O Impedances Low Frequency Input Impedance Is Small Low Frequency Output Impedance Is Very Large Both I/O Impedances Are Strongly Capacitive Input Impedance Is Capacitive Only To The Extent That The
Transistor Gate Is Biased By A Low Impedance Supply
Principle Applications Current Buffer With Capacitive I/O Ports Common Source-Common Gate (CS-CG) Cascode
Useful In Transconductor CircuitsNot As Effective As Bipolar Cascode Because Of Very Small Gate-
Drain Capacitance Of Common Source Driver
Current Gain Frequency Response Less Than Unity Current Gain Magnitude At All Frequencies Dominant Pole Is Established By Net Output Port Capacitance
Input Resistance Is SmallOutput Resistance Is Very Large
I/O Impedances Low Frequency Input Impedance Is Small Low Frequency Output Impedance Is Very Large Both I/O Impedances Are Strongly Capacitive Input Impedance Is Capacitive Only To The Extent That The
Transistor Gate Is Biased By A Low Impedance Supply
Principle Applications Current Buffer With Capacitive I/O Ports Common Source-Common Gate (CS-CG) Cascode
Useful In Transconductor CircuitsNot As Effective As Bipolar Cascode Because Of Very Small Gate-
Drain Capacitance Of Common Source Driver
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 37
CS—CG Cascode ArchitectureCS—CG Cascode ArchitectureCS—CG Cascode ArchitectureCS—CG Cascode Architecture
Performance Results Derive From General Common Gate Disclosures; Note Substitutions Below
Dominant Pole Likely Established At High Impedance Output Port
Performance Results Derive From General Common Gate Disclosures; Note Substitutions Below
Dominant Pole Likely Established At High Impedance Output Port
Z o u tR l
R s
V s s
V b ia s
V d d
Z o u t
C l R l C o
V o s
g o 1
g Vm 2 a g o 2
V s
V o
M 1
M 2
b 2 m 2 ag V
g Vm 1 s
I o s
V a
s m1 s m m2 o o2 s o1
os os l b b2 o l db2 gd 2
I = g V g = g g = g R = 1 g
V = I R λ = λ C = C C C
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 38
Cascode Circuit PerformanceCascode Circuit PerformanceCascode Circuit PerformanceCascode Circuit PerformanceVoltage Gain
About The SameAs For CommonSource Alone
Gate-DrainCapacitance Of M1 IsSmall, TherebyMitigating Miller EffectOn Cgd1
OutputResistance Very Large M2 Channel Resistance Multiplied By Approximately gm2/go1
Pole
If Rl Is Resistance Of Current Source, p Is Very Small Result Is Integrator Having Approximate Unity Gain Frequency Equal
To Gain-Bandwidth Product
Voltage Gain About The Same
As For CommonSource Alone
Gate-DrainCapacitance Of M1 IsSmall, TherebyMitigating Miller EffectOn Cgd1
OutputResistance Very Large M2 Channel Resistance Multiplied By Approximately gm2/go1
Pole
If Rl Is Resistance Of Current Source, p Is Very Small Result Is Integrator Having Approximate Unity Gain Frequency Equal
To Gain-Bandwidth Product
l out o l o1
= R R C R Cp
o2 b2 m2 b2 m2out
o2 o1 o2 o1
g 1 g 1 g1 1R = 1 1
g g g g
os l m1os os vv m1 l
s m1 s s
m1 l o2 lv m1 l
o1
o2 b2 m2
I R gV I A (0)A (s) = = = g R =
sV g V I 1 +p
g R 1 g RA (0) = g R
g1
g 1 g
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 39
Wilson Current AmplifierWilson Current AmplifierWilson Current AmplifierWilson Current Amplifier
Assumptions All Transistors Biased In Saturation Regime Aspect Ratio
Assumptions All Transistors Biased In Saturation Regime Aspect Ratio
Evaluate Performance Current Gain Input Resistance Output Resistance
Evaluate Performance Current Gain Input Resistance Output Resistance
M 1
M 2
M 3
R l
R s
I +
Is
Q
I 2
I 1
I o
V d d
V 1
V 2
R in
R o u t
1 1 2 2
2Q oQ2 2
3 3 1Q 1Q
W L = W L
I IW Lη = =
W L I I@
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 40
Model ManipulationModel ManipulationModel ManipulationModel Manipulation
Circuit Equations
Diode M2 Modeled ByTwo Terminal Net Conductance
Circuit Equations
Diode M2 Modeled ByTwo Terminal Net Conductance
M 2
M 3
I 1 s
I 1 s
V 1 s
V 1 s
I 2 s
V 2 s
g Vm 3 2 s g o 3 g m 2 g o 2
V 2 sI 2 s
I 1 sV 1 s
g o 3 g + gm 2 o 2
V 2 sI 2 s
g Im 3 2 s
g + gm 2 o 2
m31s o3 1s 2s
m2 o2
2s2s
m2 o2
gI = g V I
g g
IV =
g g
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 41
I 1 s
V 1 s
g o 3
g o 3
g + gm 2 o 2
g + gm 2 o 2
g Im 3 2 s
g Im 3 o s
g + gm 2 o 2
g + gm 2 o 2
M 1
R l
R s
R s
I s
I s
I o
V d d
R in
R o u t
I 2 s
R o u t
R lg Vm 1 a g o 1b 1 m 1 bg V
I o s
I = I2 s o s
V b
R in
V a
Wilson Small Signal ModelWilson Small Signal ModelWilson Small Signal ModelWilson Small Signal Model
Feedback
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 42
Wilson Small Signal PerformanceWilson Small Signal PerformanceWilson Small Signal PerformanceWilson Small Signal Performance
Current Gain Determined By
Relative DeviceGeometries
Best Suited ForCurrent SignalProcessing
I/O Resistances Small Input
Resistance Very Large Output
Resistance Excellent Current
Sink
Current Gain Determined By
Relative DeviceGeometries
Best Suited ForCurrent SignalProcessing
I/O Resistances Small Input
Resistance Very Large Output
Resistance Excellent Current
Sink
os io m2i
s m3m3io
m2 o2
m1 s o3 s m1 sio
o1 b1 m1o1 l
m2 o2
I A gA = = η
I gg1 A
g g
g R 1 g R g RA =
2g 1 λ g1 g R
g g
@
b1in
m1 m2 m1
b1 m1m1 s
m2 m1 sout
o1 o1
1 λ1 2ηR
g g g
1 λ g1 g R
g 2 g RR
g g
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 43
CS—Wilson CascodeCS—Wilson CascodeCS—Wilson CascodeCS—Wilson Cascode
Dominant Pole Established At High Resistance Output Port
Functions As A Cascode With Current Gain Allows Load Resistance To Be Reduced By Factor Of Wilson Current Gain
To Get Voltage Gain Equal To Conventional Cascode Lower Load Resistance Spells Lower Output Port Time Constant And
Hence, Increased Bandwidth, To The Extent That Output Port Capacitance Establishes Dominant Pole
Dominant Pole Established At High Resistance Output Port
Functions As A Cascode With Current Gain Allows Load Resistance To Be Reduced By Factor Of Wilson Current Gain
To Get Voltage Gain Equal To Conventional Cascode Lower Load Resistance Spells Lower Output Port Time Constant And
Hence, Increased Bandwidth, To The Extent That Output Port Capacitance Establishes Dominant Pole
M 1
M 2
M 3
M 4
R l
R s
I 4
I o
V d d
V s
V s s
I 3
I 2
Small Signal Equations
Low Frequency Voltage Gain
Small Signal Equations
Low Frequency Voltage Gain
4s 3s m4 s
os 2s 3s
I I g V
I I ηI
os m2m4 l m4 l
s m3
V gηg R g R
V g
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 44
Bandpass Low Noise Common SourceBandpass Low Noise Common SourceBandpass Low Noise Common SourceBandpass Low Noise Common Source
Inductors Assuming Infinite Q (Only In Academe), They Contribute No Noise Real Inductors Have Finite Q And Finite Self Resonant Frequency
Input Topology Allows Input Impedance Match To Source Resistance Maximum Power Transfer Critical Because Of Low RF Signal Power Match Is Only Bandpass → OK For Low Noise RF Applications
Inductors Assuming Infinite Q (Only In Academe), They Contribute No Noise Real Inductors Have Finite Q And Finite Self Resonant Frequency
Input Topology Allows Input Impedance Match To Source Resistance Maximum Power Transfer Critical Because Of Low RF Signal Power Match Is Only Bandpass → OK For Low Noise RF Applications
L s s
L s s
L g g
L g g
R s
R sV s
V s V s s
I o u t
C s b
C g s
ig a
C g d
C d bg vm b b a
v g a
r o
v
b a
T gks( ) ig a
Z in
Z in
I o u t
gdg
gs
Ck 1
C@
BiasingNot Shown
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 45
Bandpass Model ModificationsBandpass Model ModificationsBandpass Model ModificationsBandpass Model Modifications Ignore Gate-Drain Capacitance
Self-Aligning Gate Technology Possible Use Of Common Gate Cascode Small Device Thickness (W) Note kg = 1
Inductors Account For Winding Resistance (Finite Q) Ignore Self Resonance In Lgg (Analytical Tractability) Include Self Resonance In Lss
Capacitance Across Terminals Of Inductor WindingAbsorb Capacitance Into Csb; Replace Csb By Css > Csb
Miscellany Ignore Channel Resistance, ro (Reasonable For Typical Loads) Ignore Cdb By Ultimately Absorbing It Into Load Termination
Objective Demonstrate Bandpass Nature Of Amplifier Demonstrate Match Terminated Input (Signal) Port
Ignore Gate-Drain Capacitance Self-Aligning Gate Technology Possible Use Of Common Gate Cascode Small Device Thickness (W) Note kg = 1
Inductors Account For Winding Resistance (Finite Q) Ignore Self Resonance In Lgg (Analytical Tractability) Include Self Resonance In Lss
Capacitance Across Terminals Of Inductor WindingAbsorb Capacitance Into Csb; Replace Csb By Css > Csb
Miscellany Ignore Channel Resistance, ro (Reasonable For Typical Loads) Ignore Cdb By Ultimately Absorbing It Into Load Termination
Objective Demonstrate Bandpass Nature Of Amplifier Demonstrate Match Terminated Input (Signal) Port
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 46
Source Terminal ImpedanceSource Terminal ImpedanceSource Terminal ImpedanceSource Terminal Impedance
Source Lead Impedance Source Lead Impedance
Parameters Undamped Resonance
Quality Factor
Comments Desire Infinitely Large
Undamped Resonant Frequency
For ωs → ∞
Parameters Undamped Resonance
Quality Factor
Comments Desire Infinitely Large
Undamped Resonant Frequency
For ωs → ∞
sss ss
1ω
L C
ss ss
ss 2
s s s
sR 1 Q
ωZ (s)
s s1 + +
Q ω ω
ss ss s sss
sZ (s) R Q sL
ω
ss sss sss
ss ss
L Cω LQ
R R L s s
L g g R g g
R s
V s
C s s
C g s
i g a
g vm b b a
v g a
R s s
v
b a
T s( )i g aZ in
I o u t
Z s s
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 47
Modified RF Bandpass ModelModified RF Bandpass ModelModified RF Bandpass ModelModified RF Bandpass Model
Source Lead Current
Max Zss
Approximation
Small InductanceHigh Quality Factor (Low Rss)Connect Bulk To Source (λb = 0)
Input Port Impedance
Source Lead Current
Max Zss
Approximation
Small InductanceHigh Quality Factor (Low Rss)Connect Bulk To Source (λb = 0)
Input Port Impedance
m gs Tg g
mb ss
1 g sC ωI I I 1
1 g Z s
mb b m ss ss ssg λ g R C L
2 2ss s s ss s s ss ss ss ss
Z (jω ) Q R Q 1 Q R L R C
Tin gg gg ss
gs
ω1Z (s) sL R 1 Z
sC s
L g g R g g
R s
V s
C g s I g
g Z Im b s s T s( )I gZ in
I o u t
Z(s
)ss
I
V g
ss ss
ss 2
s s s
sR 1 Q
ωZ (s)
s s1 + +
Q ω ω
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 48
T ssss T ss ss
in gg gg 2gs
s s s
ω RR ω L sL
1 sZ (s) sL RsC
s s1
Q ω ω
RF Input PortRF Input PortRF Input PortRF Input Port
Input Impedance
Assume Large ωs
Small Inductance Connect Bulk To Source Terminal
Input Impedance
Assume Large ωs
Small Inductance Connect Bulk To Source Terminal
in gg ss T ss gg ss T ssgs
in inin
1 1Z (s) R R ω L s L L ω R
s C
1R sL
sC
@T s( )I gL in R in
R s
V s
C in I g
Z in
V g
I o u t
in gg ss T ss T ss
in gg ss
gsin
T ss gs
R R R ω L ω L
L L L
CC
1 ω R C
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 49
Input Port MatchingInput Port MatchingInput Port MatchingInput Port Matching
Resonate Lin And Cin With CarrierFrequency, ωc, Of Signal
Match Input Resistance WithSource Resistance Matching Accomplished Via Lss
Low Noise, Since No Resistance Is Explicitly Used For Matching Since Lss Is Used For Resistance Match, Lgg Affords Additional
Design Degree Of Freedom For Setting Center Frequency
Resultant I/O Relationships
Resonate Lin And Cin With CarrierFrequency, ωc, Of Signal
Match Input Resistance WithSource Resistance Matching Accomplished Via Lss
Low Noise, Since No Resistance Is Explicitly Used For Matching Since Lss Is Used For Resistance Match, Lgg Affords Additional
Design Degree Of Freedom For Setting Center Frequency
Resultant I/O Relationships
T ss gs
cgg ss gs
1 ω R Cω
L L C
s gg ss T ss T ssR R R ω L ω L
s sg
cc
c
V 2RI
ωω1 jQ
ω ω
Tout g
ωI j I
ω
T s( )I gL in R in
R s
V s
C in I g
Z in
V g
I o u t
2c in in
c c in s
ω L C 1
Q ω L 2R
out T cc
s s
I ω ω(jω ) j
V 2R
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 50
L s s
L d d
L g g L g g
R s R s
V s V s
V s s
I o u t
I o u t
V o u t
V o u t
C g s
C o
I g
T s( )I g
Z in Z in
V d d
Z(s
)ss Z
(s)
dd
Basic RF AmplifierBasic RF AmplifierBasic RF AmplifierBasic RF Amplifier
Approximate Model Large ro
Small Cgd
Small gmb
Zdd Is Same Form As Zss
Q Of Inductor (Resistance Rdd) Drain-Bulk And Output Port Capacitances (Cdd = Cdb + Co)
Approximate Model Large ro
Small Cgd
Small gmb
Zdd Is Same Form As Zss
Q Of Inductor (Resistance Rdd) Drain-Bulk And Output Port Capacitances (Cdd = Cdb + Co)
Biasing IsIncomplete
dd dd
dd 2
d d d
sR 1 Q
ωZ (s)
s s1 + +
Q ω ω
ddd dd
d ddd
dd
1ω
L C
ω LQ
R
Zin = Rs @ ω = ωc
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 51
RF Amplifier GainRF Amplifier GainRF Amplifier GainRF Amplifier Gain
Gain Relationships
Voltage Gain
Design Options Double Tuning Single Tuning Butterworth, Tchebyschev, Bessel Or Other Optimization
Gain Relationships
Voltage Gain
Design Options Double Tuning Single Tuning Butterworth, Tchebyschev, Bessel Or Other Optimization
L g g
R s
V sI o u t
V o u t
C g s I g
T s( ) I g
Z in
Z(s
)ss Z
(s)
dd
gout out out
s out g s
IV V I
V I I V
s sg
cc
c
V 2RI
ωω1 jQ
ω ω
Tout g
ωI j I
ω
out ddT
s s cc
c
V (jω) Z (jω) 1j
V (jω) 2R ωω1 jQ
ω ω
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 52
2out d ddT
s s c dc d
c d
V (jω) Q Rω 1 1j
V (jω) ω 2R ω ωω ω1 jQ 1 jQ
ω ω ω ω
Comments On RF Design OptionsComments On RF Design OptionsComments On RF Design OptionsComments On RF Design Options
I/O Gain
Single f Tuning ωd = ωc
Difficult To ImplementParametric UncertaintiesGate-Drain Feedback Capacitance Renders Port Tuning Non-Independent
Gain At Signal Carrier Frequency
Double Tuning Optimize Pole Locations For Butterworth Or Other Response Form ωc ≠ ωd; Qc ≠ Qd
Difficult To Implement Reliably On Chip For Foregoing Reasons
Single (Input Port) Tuning Ensure ωd Large In Comparison With Highest Frequency Of Interest Requires Very Small Geometry Device Because Of CdB
I/O Gain
Single f Tuning ωd = ωc
Difficult To ImplementParametric UncertaintiesGate-Drain Feedback Capacitance Renders Port Tuning Non-Independent
Gain At Signal Carrier Frequency
Double Tuning Optimize Pole Locations For Butterworth Or Other Response Form ωc ≠ ωd; Qc ≠ Qd
Difficult To Implement Reliably On Chip For Foregoing Reasons
Single (Input Port) Tuning Ensure ωd Large In Comparison With Highest Frequency Of Interest Requires Very Small Geometry Device Because Of CdB
2out c d ddT
s c c s
V (jω ) Q Rωj
V (jω ) ω 2R
University of Southern California/Choma
EE 348, Spring 2003: Lecture #04 53
L s s
L d d
L g g
R s
V s
V s s
V o u t
C o
C b
Z in
V d d
M 1
M 2M 3
R p
R b
Circuit Implementation Of RF CellCircuit Implementation Of RF CellCircuit Implementation Of RF CellCircuit Implementation Of RF Cell Transistor M1
Basic Transconductor Bulk Transconductance Is
Mitigated Transistor M2
Common Gate Cascode Mitigates Miller Multiplication
Of Cgd In M1 Must Have Large W For
Low Transconductance Transistor M3
Sets Biasing For M1 Behaves As Current Mirror Since, Assuming Ideal Inductors,
Gate-Source Biasing Of M1 And M3 Are Identical Note No Bias Current Through Resistor Rb
Stage Engineering Operates At Low Biasing Voltages Can Be Realized Differentially
Transistor M1 Basic Transconductor Bulk Transconductance Is
Mitigated Transistor M2
Common Gate Cascode Mitigates Miller Multiplication
Of Cgd In M1 Must Have Large W For
Low Transconductance Transistor M3
Sets Biasing For M1 Behaves As Current Mirror Since, Assuming Ideal Inductors,
Gate-Source Biasing Of M1 And M3 Are Identical Note No Bias Current Through Resistor Rb
Stage Engineering Operates At Low Biasing Voltages Can Be Realized Differentially