Chapter 3: Single-stage Amplifiers

3.1 Applications
3.2 General Considerations
3.3 Common-Source Stage
3.4 Source Follower
3.5 Common-Gate Stage
3.6 Cascode Stage

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education.
 Ideal vs Non-ideal Amplifier

• Ideal amplifier (Fig. a)

– Large-signal characteristic is a straight line

– α1 is the “gain”, α0 is the “dc bias”
• Nonlinear amplifier (Fig. b)

– Large signal excursions around bias point

– Varying “gain”, approximated by polynomial
– Causes distortion of signal of interest
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 2
 Analog Design Tradeof

• Along with gain and speed, other parameters

also important for amplifiers
• Input and output impedances decide interaction
with preceding and subsequent stages
• Performance parameters trade with each other
– Multi-dimensional optimization problem

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 3
 Common-Source stage with Resistive
load
• Very high input impedance at
low frequencies
• For Vin < VTH, M1 is off and Vout
= VDD

• When Vin > VTH, M1 turns on in

saturation region, Vout falls
• When Vin > Vin1, M1 enters triode
region
• At point A, Vout = Vin1-VTH

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 4
 Common-Source stage with Resistive
load
• For Vin > Vin1,

• If Vin is high enough to drive M1

into deep triode region so that
Vout << 2(Vin - VTH),

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 5
 Common-Source stage with Resistive
load

• Taking derivative of ID equation

in saturation region, small-signal
gain is obtained

• Same result is obtained from

small-signal equivalent circuit

• gm and Av vary for large input

signal swings according to

• This causes non-linearity

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 6
 Common-Source stage with Resistive
load
• For large values of RD, channel-length modulation of M1
becomes significant, Vout equation becomes

• Voltage gain is

• Above result is also obtained from small-signal

equivalent circuit

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 7
 Diode-Connected MOSFET
• A MOSFET can operate as a small-signal resistor if its
gate and drain are shorted, called a “diode-connected”
device
• Transistor always operates in saturation

• Impedance of the device can be found from small-signal

equivalent model

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 8
 Diode-Connected MOSFET

• Including body-effect, impedance “looking into” the

source terminal of diode-connected device is found as

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 9
 Diode-Connected MOSFET: Example
• Find RX if λ = 0

• Result is same compared to

when drain of M1 is at ac ground,
but
• Set independent sources to onlyapply
zero, whenVλ and
= 0 find
X
resulting IX
• Loosely said that looking into
source of MOSFET, we see 1/gm
when λ = γ = 0
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 10
 CS Stage with Diode-Connected Load

• Neglecting channel-length
modulation, using impedance
result for diode-connected
device,

where,
• Expressing gm1 and gm2 in terms of device dimensions,

• This shows that gain is a weak function of bias currents

and voltages, i.e., relatively linear input-output
characteristic
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 11
 CS Stage with Diode-Connected Load
• From large-signal analysis,

• If VTH2 does not vary much with Vout, input-output

characteristic is relatively linear.
• Squaring function of M1 (from its input voltage to its drain
current) and square root function of M2 (from its drain
current to its overdrive) act as inverse functions

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 12
 CS Stage with Diode-Connected Load

• As I1 falls, so does overdrive of M2 so that

• Subthreshold conduction of M2 eventually brings Vout to

VDD, but at very low current levels, finite capacitance at
output node CP slows down the change in Vout from VDD-
VTH2 to VDD.
• In high-frequency circuits, Vout remains around VDD-VTH2
when I1 falls to small values.
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 13
 CS Stage with Diode-Connected Load

• For Vin < VTH1, Vout = VDD – VTH2

• When Vin > VTH1, previous large-signal analysis predicts

that Vout approximately follows a single line

• As Vin exceeds Vout + VTH1 (to the right of point A), M1

enters the triode region and the characteristic becomes
nonlinear.
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 14
 CS Stage with Diode-Connected PMOS
device
• Diode-connected load can be implemented
as a PMOS device, free of body-effect
• Small-signal voltage gain neglecting
channel-length modulation

• Gain is a relatively weak function of device dimensions

• Since µn ≈ 2µp, high gain requires “strong” input device
(narrow) and “weak” load device (wide)
• This limits voltage swings since for λ = 0, we get

• For diode-connected loads, swing is constrained by both

required overdrive voltage and threshold voltage, i.e., for
small overdrive, output cannot exceed VDD - |VTH|.
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 15
 CS Stage with Current-Source Load

• Current-source load allows a high load resistance without

limiting output swing
• Voltage gain is
• Overdrive of M2 can be reduced by increasing its width,
ro2 can be increased by increasing its length
• Output bias voltage is not well-defined
• Intrinsic gain of M1 increases with L and decreases with ID

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 16
 CS Stage with Active Load

• Input signal is also applied to gate of load device, making

it an “active” load
• M1 and M2 operate in parallel and enhance the voltage
gain
• From small-signal equivalent circuit,

• Same output resistance as CS stage with current-source

load, but higher transconductance
• Bias current of M1 and M2 is a strong function of PVT
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 17
 CS Stage with Active Load: Supply
sensitivity

• Variations in VDD or the threshold voltages directly

translate to changes in the drain currents
• Supply voltage variations “supply noise” are amplified
too
• Voltage gain from VDD to Vout can be found to be

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 18
 CS Stage with Triode Load

• A MOS device biased in the

deep triode region acts as a
resistive load in a CS stage
• Vb is sufficiently low to ensure
M2 is in deep triode region for
all output voltage swings
• Voltage gain is Av = -gm1Ron2, where Ron2 is the MOS ON
resistance given by

• Ron2 depends on µpCox, Vb and VTHP which vary with PVT

• Generating a precise value of Vb is complex, which makes
circuit hard to use
• Triode loads consume lesser voltage headroom than
diode-connected
Copyright devices
© 2017 McGraw-Hill Education. All rights reserved. since
No reproduction Vwithout
or distribution the prior=
out,max Vconsent
written DD for the Education.
of McGraw-Hill former19
 CS Stage with Source Degeneration

• Degeneration resistor RS in series with source terminal

makes input device more linear
– As Vin increases, so do ID and the voltage drop across
RS
– Part of the change in Vin appears across RS rather
than gate-source overdrive, making variation in ID
smoother
• Gain is now a weaker function of gm
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 20
 CS Stage with Source Degeneration

• Nonlinearity of circuit is due to nonlinear dependence of

ID upon Vin
• Equivalent transconductance Gm of the circuit can be
defied as

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 21
 CS Stage with Source Degeneration

• gm is the transconductance of M1
• Small-signal voltage gain Av is then given by

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 22
 CS Stage with Source Degeneration

• Same result for Gm is obtained from small-signal

equivalent circuit, by noting that

• As RS increases, Gm becomes a weaker function of gm and

hence ID
• For , , i.e.,
• Most of the change in Vin across RS and drain current
becomes a “linearized” function of input voltage
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 23
 CS Stage with Source Degeneration

• Including body-effect and channel-length modulation, Gm

is found from modified small-signal equivalent circuit

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 24
 CS Stage with Source Degeneration

Large-signal behavior

RS=0 RS≠0

• ID and gm vary with Vin as • At low current levels,

derived in calculations in turn-on behavior is similar
Chapter 2 to when RS=0 since
and hence
• As overdrive and gm
increase, effect of RS
becomes more significant

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 25
 CS Stage with Source Degeneration

• Small-signal derived previously can be written as

• Denominator = Series combination of inverse

transconductance + explicit resistance seen from source
to ground
• Called “resistance seen in the source path”
• Magnitude of gain = Resistance seen at the drain/ Total
resistance seen in the source path

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 26
 CS Stage with Source Degeneration

• Degeneration causes increase in output resistance

• Ignoring RD and including body effect in small-signal

equivalent model,

• ro is boosted by a factor of {1 + (gm+gmb)RS} and then

added to RS
• Alternatively, RS is boosted by a factor of {1 + (gm+gmb)ro}
and then added to ro
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 27
 CS Stage with Source Degeneration

• Compare RS = 0 with RS > 0

• If RS = 0, and

• If RS > 0, and , obtaining negative gmV1

and gmbVbs
• Thus, current supplied by VX is less than VX/ro and hence
output impedance is greater than ro

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 28
 CS Stage with Source Degeneration
Intuitive understanding of increased output impedance
•Apply voltage change ΔV at output and measure resulting
change ΔI in output current, which is also the change in
current through RS

•Resistance seen looking into the source of M1 is 1/

(gm + gmb)
•Voltage change across RS is

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 29
 CS Stage with Source Degeneration
Intuitive understanding of increased output impedance

•Change in current across RS is

•Output resistance is thus

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 30
 CS Stage with Source Degeneration

• To compute gain in the general case including body effect

and channel-length modulation, consider above small-
signal model
• From KVL at input,

• KCL at output gives

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 31
 CS Stage with Source Degeneration

• Since voltage drops across rO and RS must add up to Vout,

• Voltage gain is therefore

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 32
 Lemma

• In a linear circuit, the voltage gain is equal to –GmRout

– Gm denotes the transconductance of the circuit when
output is shorted to ground
– Rout represents the output resistance of the circuit
when the input voltage is set to zero
• Norton equivalent of a linear circuit

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 33
 Source Follower

• Source follower (also called “common-drain” stage)

senses the input at the gate and drives load at the source
• It presents a high input impedance, allowing source
potential to “follow” the gate voltage
• Acts as a voltage buffer

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 34
 Source Follower

• For Vin < VTH, M1 is off and Vout = 0

• As Vin exceeds VTH, M1 turns on in saturation since VDS =
VDD and VGS –VTH ≈ 0 and ID1 flows through RS
• As Vin increases further, Vout follows the input with a
difference (level shift) equal to VGS
• Input-output characteristic neglecting channel-length
modulation can be expressed as

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 35
 Source Follower

• For Vin < VTH, M1 is off and Vout = 0

• Differentiating both sides of large-signal equation for Vout,

• Since ,

• Therefore,

• Note that
• Therefore,

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 36
 Source Follower

• Small-signal gain can be obtained more easily using

small-signal equivalent model

• We have,
• KVL:

• KCL:

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 37
 Source Follower

• Voltage gain begins from zero for Vin ≈ VTH (gm ≈ 0), and
monotonically increases
• As drain current and gm increase, Av approaches

• Since η itself slowly decreases with Vout, Av would

eventually become equal to unity, but for typical
allowable source-bulk voltages, η remains greater than
roughly than 0.2
• Even if RS = ∞, voltage gain of a source follower is not
equal to one
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 38
 Source Follower

• Drain current of M1 depends heavily of input dc level

• Even if VTH is relatively constant, the increase in VGS
means that Vout (=Vin-VGS) does not follow Vin faithfully,
incurring nonlinearity
• To alleviate this issue, the resistor can be replaced by a
constant current source
• Current source is itself is implemented as an NMOS
transistor operating in the saturation region

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 39
 Source Follower
Calculation of output impedance

• From small-signal equivalent circuit,

• It follows that and

• Body effect decreases output resistance of source

followers
• If VX decreases by ΔV so the drain current increases
– w/o body effect, VGS increases by ΔV
– with body effect, VTH decreases as well, thus (VGS-VTH)2
and ID1 increase by a greater amount, hence lower
output
Copyright © 2017 McGraw-Hill impedance
Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 40
 Source Follower

• Magnitude of the current source gmbVbs = gmbVX is linearly

proportional to the voltage across it, can be modelled by
a resistor equal to 1/gmb (valid only for source followers)
• This appears in parallel with the output, decreasing the
overall output resistance
• Since without 1/gmb, the output resistance is 1/gm, we
conclude that

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 41
 Source Follower

• Modelling effect of gmb by a resistor helps explain lower

than unity gain for RS = ∞
• From the Thevenin equivalent circuit

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 42
 Source Follower

• Small-signal equivalent circuit with a finite load

resistance and channel-length modulation is shown

• 1/gmb, rO1, rO2 and RL are in parallel, therefore,

• It follows that

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 43
 Issues with Source Follower

• Source followers exhibit high input impedance and

moderate output impedance, but at the cost of
– Nonlinearity
– Voltage headroom limitation
• Even when biased by ideal current source, there is input-
output nonlinearity due to nonlinear dependence of VTH
on the source potential
• In submicron technologies, rO changes substantially with
VDS and introduces additional variation in small-signal
gain
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 44
 Issues with Source Follower

• Nonlinearity can be eliminated if the bulk is tied to the

source
– Possible only for PFETs since all NFETs usually
share the same substrate
• PMOS source follower employing two separate n-wells
can eliminate the body effect of M1
• Lower mobility of PFETs yields a higher output
impedance than that available in the NMOS counterpart

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 45
 Issues with Source Follower

• Source followers also shift the dc level of the signal by

VGS, thereby consuming voltage headroom

• In the cascade of CS stage and source follower shown

above,
– w/o source follower, minimum allowable value of VX
would be VGS1-VTH1 (for M1 to remain in saturation)
– with source follower, VX must be greater than VGS2+
(VGS3-VTH3) so that M3 is saturated
• For comparable overdrive voltages in M1 and M3,
allowable
Copyright swing
© 2017 McGraw-Hill Education. at
All rights X is
reserved. reduced
No reproduction by
or distribution VtheGS2
without prior written consent of McGraw-Hill Education. 46
 Comparison of CS stage and Source
Follower

• Comparing the gain of source followers and CS stage

with a low load impedance
– E.g., driving an external 50-Ω termination in a high-
frequency environment
• When load is driven by a source follower, overall voltage
gain is

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 47
 Comparison of CS stage and Source
Follower

• Load can be included as part of a common-source stage,

providing a gain of

• Key difference between the two topologies is the

achievable voltage gain for a given bias current
• For example, if , source follower exhibits a
gain of at most 0.5 whereas the common-source stage
provides a gain close to unity
• Thus, source followers are not efficient drivers
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 48
 Common-Gate Stage

• A common-gate (CG) stage senses the input at the source

and produces the output at the drain
• Gate is biased to establish proper operating conditions

• Bias current of M1 flows through the input signal source

• Alternatively, M1 can be biased by a constant current
source, with the signal capacitively coupled to the circuit

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 49
 Common-Gate Stage: Large-signal
behavior
• Assume Vin decreases from a large
positive value and that λ=0
• For Vin ≥ Vb-VTH, M1 is off and Vout = VDD
• For lower values of Vin, if M1 is in
saturation,

• As Vin decreases further, so does Vout driving M1 into the

triode region if

• In the region where M1 is saturated, we can express the

output voltage as

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 50
 Common-Gate Stage
Input-output characteristic

• For M1 in saturation,

• Small-signal gain can thus be obtained

• Since , we have

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 51
 Common-Gate Stage

• Gain of the common-gate (CG) stage is positive

• Body effect increases the effective transconductance of

the stage
• For a given bias current and supply voltage (i.e., a given
power budget), voltage gain of the CG stage can be
maximized by
– Increasing gm by widening the input device,
eventually reaching subthreshold operation
[gm=ID/ζVT]
– Increasing RD and inevitably, the dc drop across it
• The minimum allowable value of Vout is VGS-VTH+VI1, where
VI1 denotes the minimum voltage required by I1
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 52
 Common-Gate Stage
• Consider output impedance of transistor and impedance
of the signal source

• In small-signal equivalent circuit, since current flowing RS

is –Vout/RD,
(1)

• Moreover, since current through rO is

(2)
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 53
 Common-Gate Stage

• Substituting V1 from (1) in (2),

• Therefore,

• The voltage gain expression is similar to that of a

degenerated CS stage
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 54
 Common-Gate Stage: Input Impedance

• From the small-signal equivalent circuit for finding input

impedance, we have

• The current through rO is equal to IX + gmV1 + gmbV1 =

IX - (gm+gmb)VX

• Voltages across rO and RD can be added and equated to

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 55
 Common-Gate Stage: Input Impedance

• Thus,

If (gm + gmb)rO >> 1

• The drain impedance is divided by (gm + gmb)rO when seen

at the source
– Important in short-channel devices because of their
low intrinsic
Copyright © 2017 McGraw-Hill gainNo reproduction or distribution without the prior written consent of McGraw-Hill Education. 56
Education. All rights reserved.
 Common-Gate Stage: Input Impedance

• Suppose RD = 0, then

• This is the impedance seen at the source of a source

follower, a predictable result since with RD = 0 the circuit
configuration is the same as a source follower

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 57
 Common-Gate Stage: Input Impedance

• If RD is replaced with an ideal current source, earlier

result predicts that input impedance approaches infinity
• Total current through the transistor is fixed and is equal
to I1
• Therefore, a change in the source potential cannot
change the device current, and hence IX = 0
• The input impedance of a CG stage is relatively low only
if the load impedance connected to the drain is small
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 58
 Common-Gate Stage

• In a CG stage with a current source load, substituting

RD=∞ in the voltage gain equation, we get

• Gain does not depend on RS

• From the foregoing discussion, if RD→∞, so does the
impedance seen at the source of M1, and the small-signal
voltage at node X becomes equal to Vin

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 59
 Common-Gate Stage

• In a degenerated CS stage, we loosely say that a

transistor transforms its source resistance up
• In a CG stage, the transistor transforms its drain
resistance down
• The MOS transistor can thus be viewed as an resistance
transformer

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 60
 Common-Gate Stage: Output
Impedance

• From the above small-signal equivalent circuit, we can

find output impedance as

• Result is similar to that obtained for a degenerated CS

stage

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 61
 Common-Gate Stage
• Input signal of a common-gate stage may be a current
rather than a voltage as shown below

• Input current source exhibits output impedance of RP

• To find the “gain” Vout/Iin, replace Iin and RP with a
Thevenin equivalent and use derived result to write

• Output impedance is simply given by

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 62
 Cascode Stage
• The cascade of a CS stage and a CG stage is called a
cascode topology

• M1 generates a small-signal drain current proportional to

the small-signal input Vin and M2 simply routes the current
to RD
• M1 is called the input device and M2 the cascode device
• M1 and M2 in this example carry equal bias and signal
currents
– Topology also called as “telescopic cascode”
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 63
 Cascode Stage: Qualitative Analysis

• Assume both transistors are in saturation and λ = γ = 0

• If Vin rises by ΔV, then ID1 increases by gm1ΔV
• This change in current flows through the impedance seen
at X, i.e., the impedance seen at the source of M2, which is
equal to 1/gm2
• Thus, VX falls by an amount given by gm1ΔV∙(1/gm2)
• This change in ID1 also flows through RD, producing a
drop of gm1ΔVRD in Vout, just as in a simple CS stage
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 64
 Cascode Stage: Qualitative Analysis

• Consider the case when Vin is fixed and Vb increases by

ΔV
• Since VGS1 is constant and rO1=∞, M1 can be replaced by an
ideal current source
• For node X, M2 operates as a source follower, it senses an
input ΔV at its gate and generates an output at X
• With λ = γ = 0, the small-signal voltage of the follower is
unity regardless of RD
• VX rises by ΔV, but Vout does not change since
ID2=ID1=constant, thus voltage gain from Vb to Vout is zero
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 65
 Cascode Stage: Bias Conditions
• For M1 to operate in saturation, we must
have
• If M1 and M2 are both in saturation, M2
operates as a source follower and VX is
primarily determined by Vb:
• Thus, and hence

• For M2 to be saturated,
• Thus,

if Vb is chosen to place M1 at the edge of saturation

• Minimum output level for which both transistors are in
saturation is equal to the sum of overdrives of M1 and M2
• Addition of M2 to the circuit reduces the output voltage
swing by at least its overdrive voltage
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 66
 Cascode Stage: Large-Signal Behavior

• For Vin≤VTH1, M1 and M2 are off, Vout=VDD, and VX≈Vb-VTH2

• As Vin exceeds VTH1, M1 draws current, and Vout drops
• Since ID2 increases, VGS2 must increase as well, causing VX
to fall
• As Vin becomes sufficiently large, two effects can occur:
– VX falls below Vin by VTH1, forcing M1 into the triode
region
– Vout drops below Vb by VTH2, driving M2 into triode
region
• Depending on device dimensions and RD and Vb, one
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 67
 Cascode Stage: Small-signal
characteristics

• Assume both transistors operate in saturation and λ=0

• Voltage gain is equal to that of a common-source stage
because the drain current produced by the input device
must flow through the cascode device
• This result is independent of the transconductance and
body effect of M2, the cascode device
• Can be verified using Av = -GmRout

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 68
 Cascode Stage: Output Impedance
• Important property of the cascode structure is its high
output impedance

• For calculation of Rout, the circuit can be viewed as a

common-source stage with a degeneration resistor equal
to rO1
• Thus,
• Assuming gmrO>>1, we have
• M2 boosts the output impedance of M1 by a factor of
(gm2+gmb2)rO2

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 69
 Triple Cascode
• Cascoding can be extended to three or more stacked
devices to achieve higher output impedance
• But required additional voltage headroom makes it less
attractive
• For a triple cascode, the minimum output voltage is equal
to the sum of three overdrive voltages

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 70
 Cascode stage with current source load
• Voltage gain can be maximized by maximizing Gm and/or
Rout
• Since Gm is typically determined by the transconductance
of a transistor and has trade-offs with the bias current
and device capacitances, it is desirable to increase
voltage gain by maximizing Rout

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 71
 Cascode Stage vs Increasing Length

• Increasing length of the input transistor for a given bias

current increases the output impedance
• Suppose the length of the input transistor is quadrupled
while the width remains constant
• Since , the overdrive
voltage is doubled and the transistor consumes the same
amount of voltage headroom as does a cascode stage,
i.e., circuits in (b) and (c) impose equal voltage swing
constraints

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 72
 Cascode Stage vs Increasing Length

• Since

• And , quadrupling L only doubles the value of

gmrO while cascoding results in an output impedance of
roughly gmrO2
• Transconductance of M1 in (b) is only half of that in (c),
degrading the performance
• For a given voltage headroom, the cascode structure
provides a higher output impedance
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 73
 Cascode Structure as Current Source
• High output impedance of cascode
structure yields a current source closer to
the ideal, but at the cost of voltage
headroom
• The current source load in a cascode stage
can be implemented as a PMOS cascode,
exhibiting an impedance equal to

• To find the voltage gain, Gm≈gm1

• Rout is the parallel combination of the NMOS and PMOS
cascode output impedances

• The gain is given by

• For typical values, this is approximated as

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 74
 Poor Man’s Cascode
• A “minimalist” cascode current source omits the bias
voltage necessary for the cascode device

• Called “poor man’s cascode”, M2 is placed in the triode

region because VGS1>VTH1 and VDS2=VGS2-VGS1 < VGS2-VTH2
• If M1 and M2 have the same dimensions, the structure is
equivalent to a single transistor having twice the length-
not really a cascode
• In modern CMOS technologies, transistors with different
threshold voltages are allowable, allowing M2 to operate
in saturation if M1 has a lower threshold than M2
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 75
 Poor Man’s Cascode: Shielding Property
• High output impedance arises from the fact
that if the output node voltage is changed
by ΔV, the resulting change at the source of
the cascode device is much less
– Cascode transistor “shields” the input
device from voltage variations at the
output
• Shielding property diminishes if cascode device enters
triode region
• In above circuit, as VX falls below Vb2-VTH2, M2 enters triode
region and requires a greater gate-source overdrive to
sustain the current drawn by M1, therefore

• As VX decreases, VP also drops to keep ID2 constant so

variation
Copyright of
© 2017 McGraw-Hill VX Allisrightsless
Education. reserved. attenuated as
No reproduction or distribution ittheappears
without prior written consent ofat P Education.
McGraw-Hill 76
 Folded Cascode
• The input device and the cascode device in a cascode
structure need not be of the same type

• In the figure above, (a) shows a PMOS-NMOS cascode

combination that performs the same function as a
telescopic cascode
• In order to bias M1 and M2, a current source must be
added as shown in (b)
• |ID1| + ID2 is equal to I1 and hence constant
• (c) shows an NMOS-PMOS folded cascode
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 77
 Folded Cascode: Small-signal operation

• If Vin becomes more positive, |ID1| decreases, forcing ID2 to

increase and hence Vout to drop
• The voltage gain and output impedance can be obtained
as calculated for the NMOS-NMOS cascode shown earlier
• (b) and (c) are called “folded cascode” stages because
the small-signal current is “folded” up [in (b)] or down [in
(c)]
• In the telescopic cascode, the bias current is reused
whereas those of M1 and M2 add up to I1 in (b) and (c),
leading to a higher bias current
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 78
 Folded Cascode: Large-signal operation

• Suppose Vin decreases from VDD to zero

• For Vin > VDD-|VTH1|, M1 is off and M2 carries all of I1,
yielding Vout = VDD - I1RD
• For Vin < VDD - |VTH1|, M1 turns on in saturation, giving

• As Vin drops, ID2 decreases further, falling to zero if ID1=I1

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 79
 Folded Cascode: Large-signal operation

• This occurs at Vin = Vin1 if

• Thus,

• If Vin falls below this level, ID1 tends to be greater than I1

and M1 enters the triode region to ensure ID1=I1
• As ID2 drops, VX rises, reaching Vb-VTH2 for ID2=0
• As M1 enters the triode region, VX approaches VDD
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 80
 Folded Cascode: Output Impedance

• M3 operates as the bias current source

• Using earlier results,

• The circuit exhibits a lower output impedance than a

nonfolded (telescopic) cascode

Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 81
 Folded Cascode with cascode load
• To achieve a high voltage gain, the load of a folded
cascode can be implemented as a cascode itself

• Increasing the output resistance of voltage amplifiers to

obtain a high gain may make the speed of the circuit
susceptible to the load capacitance
• A high output impedance itself does not pose a serious
issue if the amplifier is placed in a proper feedback loop
Copyright © 2017 McGraw-Hill Education. All rights reserved. No reproduction or distribution without the prior written consent of McGraw-Hill Education. 82