CHAPTER 9 FREQUENCY RESPONSE

Chapter Outline
9.1 Low-Frequency Response of the CS and CE Amplifiers
9.2 Internal Capacitive Effects and the High-Frequency Model
9.3 High-Frequency Response of the CS and CE Amplifiers
9.4 Tools for the Analysis of the High-Frequency Response of Amplifiers
9.5 A Closer Look at the High-Frequency Response
9.6 High-Frequency Response of the CG and Cascode Amplifiers

NTUEE Electronics – L.H. Lu 9-1

Frequency response of amplifiers
Midband:
 The frequency range of interest for amplifiers
 Large capacitors act as short circuits and small capacitors act as open circuits
 Gain is constant and can be obtained by small-signal analysis
Low-frequency band:
 Large capacitors no longer act as short circuits (small capacitors still act as open circuits)
 The gain roll-off is mainly due to coupling and by-pass capacitors
 Gain drops at frequencies lower than fL
High-frequency band:
 Small capacitors no longer act as open circuits (large capacitors still act as short circuits)
 The gain roll-off is mainly due to parasitic capacitances of the MOSFETs and BJTs
 Gain drops at frequencies higher than fH

NTUEE Electronics – L.H. Lu 9-2

 9.1 Low-Frequency Response of the CS and CE Amplifiers

The CS amplifier
Small-signal analysis for transfer function:
𝑅
𝑉 = 𝑉
1
𝑅 + +𝑅
𝑠𝐶
1
𝐼 = 𝑉
1 1
+ ||𝑅
𝑔 𝑠𝐶
𝑅
𝑉 =− 𝑅 𝐼
1
𝑅 + 𝑅 + 𝑠𝐶

𝑉 𝑉 𝐼 𝑉 𝑠 𝑠+𝜔 𝑠
𝑠 = =𝐴
𝑉 𝐼 𝑉 𝑉 𝑠+𝜔 𝑠+𝜔 𝑠+𝜔
𝑅
𝐴 =− 𝑔 𝑅 ||𝑅
𝑅 +𝑅
1
𝜔 =
𝐶 𝑅 +𝑅
𝑔 + 1/𝑅
𝜔 =
𝐶
1
𝜔 =
𝐶 𝑅 +𝑅
1
𝜔 =
𝐶𝑅

NTUEE Electronics – L.H. Lu 9-3

The complete frequency response (gain and phase) are determined by the transfer function analysis
Determining the lower 3-dB frequency (fL) by dominant pole
 Coupling and by-pass capacitors result in a high-pass frequency response with 3 poles and 1 zero
 If poles are sufficiently separated (dominant-pole case):
Bode plot can be used to evaluate the response for simplicity
The lower 3-dB frequency is the highest-frequency pole
fP2 is typically the highest-frequency pole due to small resistance of 1/gm

1
𝑅 +𝑅 ≫𝑅 +𝑅 ≫
𝑔
𝑓 ≪𝑓 ≪𝑓 ≈𝑓

NTUEE Electronics – L.H. Lu 9-4

Determining the lower 3-dB frequency (fL) by poles and zeros
 If poles are located closely, the lower 3-dB frequency (at a frequency higher than poles and
zeros) can be evaluated by the transfer function
𝑉 𝑠 𝑠+𝜔 𝑠
𝑠 =𝐴
𝑉 𝑠+𝜔 𝑠+𝜔 𝑠+𝜔
𝑉 1 1 + 𝜔 /𝑗𝜔 1
𝑗𝜔 = 𝐴
𝑉 1 + 𝜔 /𝑗𝜔 1 + 𝜔 /𝑗𝜔 1 + 𝜔 /𝑗𝜔
1 1 + 𝑓 /𝑗𝑓 1 𝐴
𝐴 =
1 + 𝑓 /𝑗𝑓 1 + 𝑓 /𝑗𝑓 1 + 𝑓 /𝑗𝑓 2
𝑓 𝑓 𝑓 𝑓 𝑓 𝑓 𝑓 𝑓
1+ 1+ 1+ =2 1+ →1+ + + ≈2 1+
𝑓 𝑓 𝑓 𝑓 𝑓 𝑓 𝑓 𝑓

𝑓 ≈ 𝑓 +𝑓 +𝑓 − 2𝑓

Determining the pole and zero frequencies by inspection (capacitors do not interact)
 A pole is determined by the time constant of the capacitor
1 1
𝜔 = =
𝜏 𝐶 𝑅 +𝑅
1 𝑔 + 1/𝑅
𝜔 = =
𝜏 𝐶
1 1
𝜔 = =
𝜏 𝐶 𝑅 +𝑅

NTUEE Electronics – L.H. Lu 9-5

  A transmission zero is the value of s at which the input does not reach the output

Coupling capacitor CC1: Vo = 0 as the capacitor is open (zero at s = 0)

By-pass capacitor CS: Vo = 0 as ZS is open (zero at s = −1/RSCS)
Coupling capacitor CC2: Vo = 0 as the capacitor is open (zero at s = 0)
Selecting values for coupling and by-pass capacitors
 These capacitors are typically required for discrete amplifier designs
 It is desirable to minimize the total capacitance for coupling and by-pass purposes
 CS is first determined to satisfy fP2  fL
 CC1 and CC2 are chosen such that poles fP1 and fP3 are 5 to 10 times lower than fL

NTUEE Electronics – L.H. Lu 9-6

The method of short-circuit time constant
Useful technique to estimate the lower 3-dB frequency for circuits where the capacitors interact
Time-constant method is used to evaluate fL without deriving the transfer function
The method is predicated on the assumption that one of the poles is dominant
The estimation is usually very good even a dominant pole does not exist
 Set independent source to zero (Vsig = 0)
 Consider each capacitor one at a time by set all other capacitors ideal (as short circuit)
 For each capacitor Ci, find the total equivalent resistance Ri seen by Ci
 Calculate the 3-dB frequency fL using:
1 1
𝑓 =
2𝜋 𝐶𝑅

For the CS amplifier example:

𝐶 =𝐶 →𝑅 =𝑅 +𝑅
𝐶 = 𝐶 → 𝑅 = 𝑅 ||(1/𝑔 )
𝐶 =𝐶 →𝑅 =𝑅 +𝑅
1 1
𝑓 =
2𝜋 𝐶𝑅

1 1 1 1
= + +
2𝜋 𝐶 𝑅 +𝑅 1 𝐶 𝑅 +𝑅
𝐶 𝑅 || 𝑔

NTUEE Electronics – L.H. Lu 9-7

The CE amplifier
Small-signal equivalent circuit:

 The capacitors interact (3 capacitors not necessarily for 3 poles)

 Transfer function can be derived but is rather complex
Method of short-circuit time constants to evaluate fL:
 Considering only C1 = CC1:
Set Vsig = 0 and treat CE and CC2 as short
The equivalent resistance:
R1 = RC1 = Rsig+RB||r

NTUEE Electronics – L.H. Lu 9-8

  Considering only C2 = CE:
Set Vsig = 0 and treat CC1 and CC2 as short
The equivalent resistance:
R2 = RCE = re+(Rsig||RB)/(1+)
 Considering only C3 = CC2:
Set Vsig = 0 and treat CC1 and CE as short
The equivalent resistance:
R3 = RC2 = RC+RL
 Estimate the lower 3-dB frequency:
fL  [1/1+1/2+1/3]/2
= [1/CC1RC1+1/CERCE+1/CC2RC2]/2

Selecting values for the capacitors

 Typically required for discrete amplifier designs
 CE is first determined to satisfy needed fL
(choose 1/CERCE = 80% of L)
 CC1 and CC2 are chosen for smaller time constant
(1/CC1RC2 and 1/CC2RC2  10% of L)

NTUEE Electronics – L.H. Lu 9-9

 9.2 Internal Capacitive Effects and the High-Frequency Model

The MOSFET device

There are basically two types of internal capacitance in the MOSFET
 Gate capacitance effect: the gate electrode forms a parallel-plate capacitor with gate oxide
 Junction capacitance effect: the source/body and drain/body are pn-junctions at reverse bias
The gate capacitive effect
 MOSFET in triode region:
1
𝐶 =𝐶 = 𝑊𝐿𝐶 +𝐶
2
 MOSFET in saturation region:
2
𝐶 = 𝑊𝐿𝐶 +𝐶 and 𝐶 =𝐶
3
 MOSFET in cutoff region:
𝐶 = 𝐶 = 𝐶 and 𝐶 = 𝑊𝐿𝐶
 Overlap capacitance:
𝐶 = 𝑊𝐿 𝐶
The junction capacitance
 Includes components from the bottom
and from side walls
 The simplified expression are given by
𝐶 𝐶
𝐶 = and 𝐶 =
1 + 𝑉 /𝑉 1 + 𝑉 /𝑉

NTUEE Electronics – L.H. Lu 9-10

The high-frequency MOSFET model
 Based on low-frequency small-signal models including gm and ro
 All parasitic capacitances (Cgs, Cgd, Csb and Cdb) for MOSET in saturation are included
Simplified high-frequency MOSFET model
 Source and body terminals are shorted
 Cdb is also neglected to simplify the analysis
 Cgs is typically the largest parasitic capacitance and plays
the most important role in amplifier frequency response

simplified model

NTUEE Electronics – L.H. Lu 9-11

The unity-gain frequency (fT)
 The frequency at which the current gain of the MOSFET becomes unity
 Is typically used as an indicator to evaluate the high-frequency capability
 Smaller parasitic capacitances Cgs and Cgd are desirable for higher unity-gain frequency
𝐼 = 𝑠𝐶 𝑉 + 𝑠𝐶 𝑉
𝑠𝐶 𝑉 + 𝐼 = 𝑔 𝑉
𝐼 𝑔 − 𝑠𝐶 𝑔
→ = ≈
𝐼 𝑠 𝐶 +𝐶 𝑠 𝐶 +𝐶
1 𝑔
→𝑓 =
2𝜋 𝐶 + 𝐶

 The unity-gain frequency can also be expressed as

𝑊
1 𝑔 1 𝜇 𝐶 𝐿 𝑉 3𝜇 𝑉
𝑓 = ≈ =
2𝜋 𝐶 + 𝐶 2𝜋 2 4𝜋𝐿
3 𝑊𝐿𝐶
The unity-gain frequency is strongly influenced by the channel length
Minimizing the channel length effectively increases the unity-gain frequency of MOSFET
Higher unity-gain frequency can be achieved for a given MOSFET by increasing the bias
current (ID) or the overdrive voltage (VOV)

NTUEE Electronics – L.H. Lu 9-12

The BJT Device
High-frequency small-signal model:
 Includes all parasitic capacitances in the hybrid- model or T-model
 Parasitic capacitances:
The base-charging or diffusion capacitance (Cde)
The base-emitter junction capacitance (Cje)
The total base-emitter capacitance (C = Cde+Cje)
The collector-base junction capacitance (C)
 C is typically in the range of a few pF to a few tens of pF
 C is typically in the range of a fraction of a pF to a few pF

NTUEE Electronics – L.H. Lu 9-13

The cutoff (unity-gain) frequency:
 The frequency at which the current gain of the MOSFET becomes unity
 Is typically used as an indicator to evaluate the high-frequency capability
𝐼
𝑉 =
1/𝑟 + 𝑠𝐶 + 𝑠𝐶
𝑔 𝑉 = 𝐼 + 𝑠𝐶 𝑉
𝐼 𝑔 − 𝑠𝐶
→ ℎ = =
𝐼 1/𝑟 + 𝑠 𝐶 + 𝐶
𝑔 𝑟 𝛽
≈ =
1+𝑠 𝐶 +𝐶 𝑟 1 + 𝑠/𝜔
1 𝑔 1 𝐼
→𝑓 = =
2𝜋 𝐶 + 𝐶 2𝜋 (𝐶 + 𝐶 )𝑉

NTUEE Electronics – L.H. Lu 9-14

 9.3 High-Frequency Response of the CS and CE Amplifiers

The common-source amplifier

The coupling and by-pass capacitors act as shirt circuits at high frequencies
The MOSFET is represented by its high-frequency small-signal model
high-frequency equivalent circuit

simplified small-signal model

𝑅 = 𝑅 ||𝑅
𝑅 = 𝑅 ||𝑅 𝑅 = 𝑟 ||𝑅 ||𝑅 𝑉 =𝑉 𝑅 / 𝑅 +𝑅

NTUEE Electronics – L.H. Lu 9-15

Analysis of the high-frequency transfer function:
𝑉 −𝑉 𝑉
=
+ 𝑠𝐶 𝑉 + 𝑠C 𝑉 −𝑉
𝑅 𝑅
𝑉
𝑠C 𝑉 − 𝑉 = 𝑔 𝑉 +
𝑅
𝑉 𝑔 𝑅 1 − 𝑠 𝐶 /𝑔
𝑠 =−
𝑉 1+𝑠 𝐶 +𝐶 1+𝑔 𝑅 𝑅 + 𝐶 +𝐶 𝑅 +𝑠 𝐶 𝐶 +𝐶 𝐶 +𝐶 𝐶 𝑅 𝑅
𝑉 𝑔 𝑅 𝑅 1 − 𝑠 𝐶 /𝑔
𝑠 =−
𝑉 𝑅 +𝑅 1+𝑠 𝐶 +𝐶 1+𝑔 𝑅 𝑅 + 𝐶 +𝐶 𝑅 +𝑠 𝐶 𝐶 +𝐶 𝐶 +𝐶 𝐶 𝑅 𝑅
1 − 𝑠/𝜔
=𝐴
1 + 𝑠/𝜔 1 + 𝑠/𝜔
 The transfer function is a second-order one with two poles and one zero
 The poles and zero result in gain roll-off at higher frequencies
 The high-frequency response can be evaluated by
𝑔 𝑅 𝑅
1 − 𝑗𝜔 𝐶 /𝑔
𝑉 𝑅 +𝑅
𝑗𝜔 = −
𝑉 1−𝜔 𝐶 𝐶 +𝐶 𝐶 +𝐶 𝐶 𝑅 𝑅 + 𝑗𝜔 𝐶 +𝐶 1+𝑔 𝑅 𝑅 + 𝐶 +𝐶 𝑅

 The bandwidth is typically defined by the 3-dB frequency (fH)

NTUEE Electronics – L.H. Lu 9-16

Determining the upper 3-dB frequency (fH):
 The 3-dB frequency is defined by the dominant pole (lowest-frequency) of the transfer function
 If dominant pole does not exist, the 3-dB frequency can be evaluated by the poles and zeros of
the high-frequency response:
1 + 𝑠/𝜔 1 + 𝑠/𝜔
𝐴 𝑠 = 𝐴 𝐹 (𝑠) = 𝐴
1 + 𝑠/𝜔 1 + 𝑠/𝜔
1 1
1+𝜔 +
1 1 + 𝜔 /𝜔 1 + 𝜔 /𝜔 𝜔 𝜔
𝐹 (𝑗𝜔 ) = = ≈
2 1 + 𝜔 /𝜔 1 + 𝜔 /𝜔 1 1
1+𝜔 +
𝜔 𝜔
1 1
𝑓 ≈
2𝜋
1 1 1 1
+ −2 +
𝜔 𝜔 𝜔 𝜔
𝜔
𝑓 ≈
2𝜋
(for dominant-pole case)

NTUEE Electronics – L.H. Lu 9-17

The common-emitter amplifier
The coupling and by-pass capacitors act as shirt circuits at high frequencies
The BJT is represented by its high-frequency small-signal model

high-frequency equivalent circuit

simplified small-signal model

𝑅 = 𝑅 ||𝑅 𝑅 = 𝑅 ||𝑅

𝑅 = 𝑟 ||𝑅 ||𝑅 𝑉 =𝑉 (𝑅 ||𝑟 )/ 𝑅 ||𝑟 + 𝑅

NTUEE Electronics – L.H. Lu 9-18

Analysis of the high-frequency transfer function:
𝑉 −𝑉
= 𝑠𝐶 𝑉 + 𝑠C 𝑉 − 𝑉
𝑅
𝑉
𝑠C 𝑉 − 𝑉 = 𝑔 𝑉 +
𝑅
𝑉 𝑔 𝑅 1 − 𝑠 C /𝑔
𝑠 =−
𝑉 1+𝑠 𝐶 +C 1+𝑔 𝑅 𝑅 +C 𝑅 +𝑠 𝐶 C 𝑅 𝑅
𝑉 𝑔 𝑅 (𝑅 ||𝑟 ) 1 − 𝑠 C /𝑔
𝑠 =−
𝑉 (𝑅 ||𝑟 ) + 𝑅 1+𝑠 𝐶 +C 1+𝑔 𝑅 𝑅 +C 𝑅 +𝑠 𝐶 C 𝑅 𝑅
1 − 𝑠/𝜔
=𝐴
1 + 𝑠/𝜔 1 + 𝑠/𝜔
 The analysis is very similar to common-source amplifier
 The transfer function is a second-order one with two poles and one zero
 Poles and zero result in gain roll-off at higher frequencies
 High-frequency response can be evaluated by
𝑔 𝑅 (𝑅 ||𝑟 )
1 − 𝑗𝜔 C /𝑔
𝑉 (𝑅 ||𝑟 ) + 𝑅
𝑗𝜔 = −
𝑉 1−𝜔 𝐶 C 𝑅 𝑅 + 𝑗𝜔 𝐶 + C 1 + 𝑔 𝑅 𝑅 +C 𝑅

 The bandwidth is typically defined by the 3-dB frequency (fH)

NTUEE Electronics – L.H. Lu 9-19

Miller’s Theorem:
A technique to replace the bridging capacitance
The bridging impedance Z can be equivalently divided into shunt Z1 at node 1 and shunt Z2 at node 2
The equivalent input and output impedances are:
𝑉 −𝑉 𝑉 − 𝐾𝑉 𝑉 𝑉
𝐼= = = =
𝑍 𝑍 𝑍/(1 − 𝐾) 𝑍
𝑉 −𝑉 𝑉 /𝐾 − 𝑉 −𝑉 −𝑉
𝐼= = = =
𝑍 𝑍 𝑍/(1 − 𝐾 ) 𝑍
1
→ 𝑍 = 𝑍/(1 − 𝐾) and 𝑍 = 𝑍/ 1 −
𝐾

Simplified analysis of CS amplifier by Miller Theorem:

Assume the gain from G to D is nearly constant for frequencies close to the midband
Bridging capacitance Cgd is equivalently divided into C1 = (1-K)Cgd and C2 = (1-K-1)Cgd

𝐾 ≡ 𝑉 /𝑉 = −𝑔 𝑅

𝐶 𝐶

NTUEE Electronics – L.H. Lu 9-20

Transfer function of the CS amplifier by using Miller’s theorem:
𝑉 𝑔 𝑅
𝑠 =−
𝑉 1 + 𝑠𝑅 𝐶 +𝐶 1 + 𝑠𝑅 𝐶 + 𝐶
𝑉 𝑔 𝑅 𝑅 1
𝑠 =−
𝑉 𝑅 +𝑅 1 + 𝑠𝑅 𝐶 +𝐶 1+𝑔 𝑅 1 + 𝑠𝑅 𝐶 + 𝐶 1 + 1/𝑔 𝑅

 The transfer function has two poles:

1
𝜔 =
𝑅 𝐶 +𝐶 1+𝑔 𝑅
1 1
𝜔 = ≈
1 𝑅 𝐶 +𝐶
𝑅 𝐶 +𝐶 1+
𝑔 𝑅

 The assumption that gain from G to D is constant no longer valid as frequency increases
 The midband gain remains the same
 The transfer function can not be used to predict high-frequency response very accurately
 It can be used to simplify the analysis for the 3-dB frequency estimation if a dominant pole exists
 The equivalent response of the CS amplifier is typically approximated by a STC circuit:
𝑔 𝑅 𝑅 1 𝐴
𝐴 𝑠 ≈− =
𝑅 +𝑅 1 + 𝑠𝑅 𝐶 +𝐶 1+𝑔 𝑅 1 + 𝑠/𝜔

NTUEE Electronics – L.H. Lu 9-21

Simplified analysis of CE amplifier by Miller Theorem:
Assume the gain from B to C is nearly constant for frequencies close to the midband
Bridging capacitance C is equivalently divided into C1 = Ceq and C2 (negligible)

The midband gain is defined as

𝑔 𝑅 𝑅 ||𝑟
𝐴 =−
𝑅 + 𝑅 ||𝑟
Ceq is considered Miller capacitance multiplying C by a factor of the midband gain
Cin is effectively the largest capacitance in the CE amplifier due to Miller Effect
It is not necessary to derive the poles and zeros as the high-frequency response deviates
The 3-dB frequency can be estimated by the dominant pole as
1 1
𝑓 =
2𝜋 𝑅 𝐶 +𝐶 1+𝑔 𝑅

NTUEE Electronics – L.H. Lu 9-22

 9.4 Useful Tools for the Analysis of the High-Frequency Response of
Amplifiers
Open-circuit time constant method to evaluate amplifier bandwidth
General transfer function of the amplifier
𝑠 𝑠 𝑠
1+𝜔 1+𝜔 .. 1 + 𝜔 1 + 𝑎 𝑠 + 𝑎 𝑠 + ⋯+ 𝑎 𝑠
𝐴 𝑠 =𝐴 =𝐴
𝑠 𝑠 𝑠 1 + +𝑎 𝑠 + 𝑎 𝑠 + ⋯ + 𝑎 𝑠
1+ 1+ .. 1+
𝜔 𝜔 𝜔
Open-circuit time constant:
 Remove all independent sources
 Consider the capacitors one at a time and the others are considered idea (open-circuit)
 Find the equivalent shunt resistance for each capacitor
 The coefficient can be obtained as 𝑏 = 𝐶 𝑅 + 𝐶 𝑅 + ⋯ + 𝐶 𝑅
Dominant pole approximation:
1 1 1
𝑏 = + + ⋯+
𝜔 𝜔 𝜔
1 1
𝜔 ≈𝜔 ≈ =
𝑏 𝐶 𝑅 +𝐶 𝑅 +⋯+𝐶 𝑅
For an amplifier with 2nd order transfer function and a dominant pole:
𝑠 𝑠
1+ 1+ 1+𝑎 𝑠+𝑎 𝑠
𝜔 𝜔
𝐴 𝑠 =𝐴 =𝐴
𝑠 𝑠 1+𝑎 𝑠+𝑎 𝑠
1+𝜔 1+𝜔
1 1 1 1 1 𝑏 𝑏
𝑏 = + ≈ →𝜔 ≈ and 𝑏 = ≈ →𝜔 ≈
𝜔 𝜔 𝜔 𝑏 𝜔 𝜔 𝜔 𝑏

NTUEE Electronics – L.H. Lu 9-23

Analysis of CS amplifier by using open-circuit time constants

Open-circuit time constants:

𝑅 =𝑅 𝑅 =𝑅 1 + 𝑔 𝑅 +𝑅 𝑅 =𝑅

𝑏 =𝐶 𝑅 +𝐶 𝑅 +𝐶 𝑅 =𝐶 𝑅 +𝐶 𝑔 𝑅 𝑅 +𝑅 +𝑅 +𝐶 𝑅
1 1
𝜔 ≈ =
𝑏 𝐶 𝑅 +𝐶 𝑔 𝑅 𝑅 +𝑅 +𝑅 +𝐶 𝑅

NTUEE Electronics – L.H. Lu 9-24

Analysis by using Miller’s Theorem and open-circuit time constants
Bridging capacitance Cgd is replaced by C1 and C2

𝐾≡ ≈-𝑔 𝑅

𝐶 =𝐶 1−𝐾 = 𝐶 1+𝑔 𝑅

𝐶 =𝐶 1−𝐾 =𝐶 1 + 1/𝑔 𝑅 ≈𝐶

Transfer function and H:

𝑉 𝑅 1
= −𝑔 𝑅
𝑉 𝑅 +𝑅 1 + 𝑠/𝜔 1 + 𝑠/𝜔
1 1
𝜔 = = ≈𝜔
𝐶 𝑅 𝐶 +𝐶 1+𝑔 𝑅 𝑅
1 1
𝜔 = =
𝐶 𝑅 𝐶 +𝐶 𝑅
Open-circuit time constant analysis based on Miller’s equivalent circuit:
𝑅 =𝑅
𝑅 =𝑅
𝑏 =𝐶 𝑅 +𝐶 𝑅 = 𝐶 +𝐶 1+𝑔 𝑅 𝑅 + 𝐶 +𝐶 𝑅
1 1
𝜔 ≈ =
𝑏 𝐶 +𝐶 1+𝑔 𝑅 𝑅 + 𝐶 +𝐶 𝑅

NTUEE Electronics – L.H. Lu 9-25

Analysis of CE amplifier by using open-circuit time constants

Open-circuit time constants:

𝑅 =𝑅 𝑅 =𝑅 1 + 𝑔 𝑅 +𝑅 𝑅 =𝑅

𝑏 =𝐶 𝑅 +𝐶 𝑅 +𝐶 𝑅 =𝐶 𝑅 +𝐶 𝑔 𝑅 𝑅 +𝑅 +𝑅 +𝐶 𝑅
1 1
𝜔 ≈ =
𝑏 𝐶 𝑅 +𝐶 𝑔 𝑅 𝑅 +𝑅 +𝑅 +𝐶 𝑅

NTUEE Electronics – L.H. Lu 9-26

 9.5 A Closer Look at the High-Frequency Response

Analysis by transfer function

Transfer function of the amplifier:
𝑉 𝑔 𝑅 1 − 𝑠 𝐶 /𝑔
𝑠 =−
𝑉 1+𝑠 𝐶 +𝐶 1+𝑔 𝑅 𝑅 + 𝐶 +𝐶 𝑅 +𝑠 𝐶 𝐶 +𝐶 𝐶 +𝐶 𝐶 𝑅 𝑅
𝑉 𝑔 𝑅 𝑅 1 − 𝑠 𝐶 /𝑔
𝑠 =−
𝑉 𝑅 +𝑅 1+𝑠 𝐶 +𝐶 1+𝑔 𝑅 𝑅 + 𝐶 +𝐶 𝑅 +𝑠 𝐶 𝐶 +𝐶 𝐶 +𝐶 𝐶 𝑅 𝑅
1 − 𝑠/𝜔
=𝐴
1 + 𝑠/𝜔 1 + 𝑠/𝜔
Poles and zero for dominant-pole approximation:
𝑔 𝑅 𝑅
𝐴 =−
𝑅 +𝑅
𝑔
𝜔 =
𝐶
1 1
𝜔 ≈ =
𝑏 𝐶 +𝐶 1+𝑔 𝑅 𝑅 + 𝐶 +𝐶 𝑅 𝑉 =𝑉 𝑅 / 𝑅 +𝑅

𝐶 +𝐶 1+𝑔 𝑅 𝑅 + 𝐶 +𝐶 𝑅 𝑅 = 𝑅 ||𝑅
𝑏
𝜔 ≈ = 𝑅 = 𝑟 ||𝑅 ||𝑅
𝑏 𝐶 𝐶 +𝐶 𝐶 +𝐶 𝐶 𝑅 𝑅

 The 3-dB frequency is given by H  P1 (is typically limited by Rsig and Miller effect)

NTUEE Electronics – L.H. Lu 9-27

CS amplifier with a small source resistance
For the case where source resistance is zero
The transfer function of the amplifier:
𝑉 1 − 𝑠 𝐶 /𝑔
𝑠 =− 𝑔 𝑅
𝑉 1+𝑠 𝐶 +𝐶 𝑅
 Midband gain:
𝐴 = −𝑔 𝑅 𝑉 =𝑉

 The transfer function has one pole and one zero 𝑅 =0

1 𝑅 = 𝑟 ||𝑅 ||𝑅
𝜔 =
𝐶 +𝐶 𝑅
𝑔
𝜔 =
𝐶
Midband gain and zero are virtually unchanged
The lowest-frequency pole no longer exists
Gain-bandwidth product:
 Gain rolls off beyond fH (20 dB/decade)
 The gain becomes 0 dB at ft:
𝑔
𝑓 = 𝐴 𝑓 =
2𝜋 𝐶 +𝐶

NTUEE Electronics – L.H. Lu 9-28

 9.6 High-Frequency Response of the CG and Cascode Amplifiers

High-frequency response of the CG amplifier

High-frequency circuit model for CS amplifier

The frequency response by neglecting ro:

𝑉 𝑔 𝑅 1
𝑠 =
𝑉 1+𝑔 𝑅 1 + 𝑠/𝜔 1 + 𝑠/𝜔
1
𝜔 =
1
𝐶 𝑅 ||
𝑔
1
𝜔 =
𝐶 +𝐶 𝑅
 The upper 3-dB frequency is typically defined by H  P2  P1
 The poles of CG amplifier is usually much higher than the dominant pole of the CS amplifier

NTUEE Electronics – L.H. Lu 9-29

Time constant method to evaluate the 3-dB frequency of CG amplifier by neglecting ro:
1
𝑏 =𝜏 +𝜏 =𝐶 𝑅 || + 𝐶 +𝐶 𝑅
𝑔
1 1
𝜔 ≈ =
𝑏 1
𝐶 𝑅 || 𝑔 + 𝐶 +𝐶 𝑅

The frequency response as ro included (ro has to be taken into account for IC amplifiers)
 The transfer function derivation is rather complex if ro is not negligible
 Time constant method can be used to evaluate the 3-dB frequency

𝑟 +𝑅
𝜏 =𝐶 𝑅 =𝐶 𝑅 || 𝜏 = 𝐶 +𝐶 𝑅 = 𝐶 +𝐶 𝑅 || 𝑔 𝑟 𝑅 +𝑟 +𝑅
1+𝑔 𝑟

NTUEE Electronics – L.H. Lu 9-30

High-frequency response of the cascode amplifier
Open-circuit time constant method:
 Capacitance Cgs1 sees a resistance Rsig
 Capacitance Cgd1 sees a resistance Rgd1
𝑅 =𝑅 1+𝑔 𝑅 +𝑅

 Capacitance (Cdb1+Cgs2) sees a resistance Rd1

𝑟 +𝑅
𝑅 = 𝑟 ||𝑅 = 𝑟 ||
1+𝑔 𝑟
 Capacitance (Cgd2+CL) sees a resistance Ro||RL
𝑅 ||𝑅 = 𝑔 𝑟 𝑟 +𝑟 +𝑟 ||𝑅

 Effective time constant

𝑏 =𝐶 𝑅 +𝐶 𝑅 + 𝐶 +𝐶 𝑅 + 𝐶 +𝐶 𝑅 ||𝑅

The upper 3-dB frequency:

1 1
𝜔 ≈ =
𝑏 𝐶 𝑅 +𝐶 𝑅 + 𝐶 +𝐶 𝑅 + 𝐶 +𝐶 𝑅 ||𝑅

NTUEE Electronics – L.H. Lu 9-31

The effective time constant can be expressed as:
𝜏 =𝑅 𝐶 +𝐶 1+𝑔 𝑅 +𝑅 𝐶 +𝐶 +𝐶 + 𝑅 ||𝑅 𝐶 +𝐶

 The 1st term arises at input node; the 2nd term at the middle node; the 3rd term at output node
 In the case of a large Rsig:
The first term dominates if the Miller multiplier is large (typically with large Rd1 and RL)
A small RL (to the order of ro) is needed for extended bandwidth
The midband gain drops as the value of RL decreases
A trade-off exists between gain and bandwidth
 In the case of a small Rsig:
The 1st term becomes negligible
A large RL (to the order of A0ro) can be used to boost the amplifier gain
The 3rd term dominates
 In the case of zero Rsig:
1
𝑓 ≈
2𝜋 𝑅 ||𝑅 𝐶 +𝐶

Choose Ro||RL larger than RL, which is defined in CS amplifier, by a factor of Ao

The fH of the cascade will be lower than that of the CS amplifier by the same factor Ao
The unity-gain frequency is unchanged at
𝑔
𝑓 ≈
2𝜋 𝐶 +𝐶

NTUEE Electronics – L.H. Lu 9-32

Summary of CS and cascode amplifiers

NTUEE Electronics – L.H. Lu 9-33

High-frequency response of the cascode amplifier
Open-circuit time constant method:
 Capacitance C1 sees a resistance R1
𝑅 =𝑅 ||𝑟

 Capacitance C1 sees a resistance R 1

𝑅 = 𝑅 ||𝑟 1+𝑔 𝑅 +𝑅

 Capacitance (Ccs1+C2) sees a resistance Rc1

𝑟 +𝑅
𝑅 = 𝑟 ||𝑅 = 𝑟 || 𝑟
𝑟 + 𝑅 /(𝛽 + 1)

 Capacitance (C 2+CL) sees a resistance Ro||RL

𝑅 ||𝑅 ≈ 𝛽 𝑟 ||𝑅

 Effective time constant

𝑏 =𝐶 𝑅 +𝐶 𝑅 + 𝐶 +𝐶 𝑅 + 𝐶 +𝐶 𝑅 ||𝑅

High-frequency response:
𝑟
𝐴 =− 𝑔 𝛽𝑟 ||𝑅
𝑟 +𝑅
1 1
𝜔 ≈ =
𝑏 𝐶 𝑅 +𝐶 𝑅 + 𝐶 +𝐶 𝑅 + 𝐶 +𝐶 𝑅 ||𝑅

NTUEE Electronics – L.H. Lu 9-34

 9.7 High-Frequency Response of the Source and Emitter Followers

The source follower:

High-frequency transfer function:

𝑉 1 + 𝑠/𝜔
(𝑠) = 𝐴
𝑉 1+𝑏 𝑠+𝑏 𝑠
𝑅 𝑔 𝑅
𝐴 = = ≈1
𝑅 + 1/𝑔 1+𝑔 𝑅
𝑔
𝜔 =
𝐶
𝐶 𝐶 +𝐶
𝑏 = 𝐶 + 𝑅 + 𝑅
1+𝑔 𝑅 1+𝑔 𝑅
𝐶 𝐶 +𝐶 𝐶 +𝐶 𝐶
𝑏 = 𝑅 𝑅
1+𝑔 𝑅

NTUEE Electronics – L.H. Lu 9-35

Frequency response with dominant pole approximation:
1 + 𝑏 𝑠 + 𝑏 𝑠 = 1 + 𝑠/𝜔 1 + 𝑠/𝜔
1 𝐶 𝐶 +𝐶
𝜔 ≈𝜔 ≈ = 𝐶 + 𝑅 + 𝑅
𝑏 1+𝑔 𝑅 1+𝑔 𝑅
𝑏 𝐶 𝐶 +𝐶 𝐶 𝐶 +𝐶 𝐶 +𝐶 𝐶
𝜔 ≈ = 𝐶 + 𝑅 + 𝑅 / 𝑅 𝑅
𝑏 1+𝑔 𝑅 1+𝑔 𝑅 1+𝑔 𝑅

Frequency response for non-dominant pole cases (real poles):

1 1
𝑓 ≈
2𝜋
1 1 1
+ −2
𝜔 𝜔 𝜔
Frequency response for complex pole cases (Q > 0.5):
1 𝑠 𝑠
1+𝑏 𝑠+𝑏 𝑠 =1+ +
𝑄𝜔 𝜔

1 𝑔 𝑅 +1
𝜔 = =
𝑏 𝑅 𝑅 𝐶 𝐶 +𝐶 𝐶 +𝐶 𝐶

𝑏 𝑔 𝑅 +1 𝑅 𝑅 𝐶 𝐶 +𝐶 𝐶 +𝐶 𝐶
𝑄= =
𝑏 𝐶 +𝐶 𝑔 𝑅 +1 𝑅 + 𝐶 +𝐶 𝑅

NTUEE Electronics – L.H. Lu 9-36

The emitter follower:

High-frequency transfer function:

𝑉 1 + 𝑠/𝜔
(𝑠) = 𝐴
𝑉 1+𝑏 𝑠+𝑏 𝑠
𝑅
𝐴 = ≈1
𝑅 +𝑟 +𝑅 / 𝛽+1
1
𝜔 =
𝐶 𝑟
𝐶 + 𝐶 1 + 𝑅 /𝑟 𝑅 + 𝐶 +𝐶 1+𝑅 /𝑟 𝑅
𝑏 =
1 + 𝑅 /𝑟 + 𝑅 /𝑟
𝐶 𝐶 +𝐶 𝐶 +𝐶 𝐶 𝑅 𝑅
𝑏 =
1 + 𝑅 /𝑟 + 𝑅 /𝑟

NTUEE Electronics – L.H. Lu 9-37

 9.8 High-Frequency Response of Differential Amplifiers

Resistively Loaded MOS Differential Amplifier:

Differential-mode operation:
 Use differential half-circuit for analysis
 Identical to the case for common-source amplifier
 Can be approximated by a dominant-pole system
Common-mode operation:
 Use common-mode half-circuit for analysis
 CSS is usually significant and can not be neglected
 CSS results in a zero at lower frequency
 Other capacitance  high-frequency poles and zeros
𝑅 ∆𝑅
𝐴 =−
2𝑅 𝑅
𝑅 ∆𝑅 𝑅 ∆𝑅 1
𝐴 𝑠 =− =− + 𝑠𝐶
2𝑍 𝑅 2 𝑅 𝑅

𝑅 ∆𝑅 𝑠
=− 1 + 𝑠𝐶 𝑅 =𝐴 1+
2𝑅 𝑅 𝜔

1
𝜔 =
𝐶 𝑅

NTUEE Electronics – L.H. Lu 9-38

Common-mode rejection ratio (CMRR):
 The frequency dependence of CMRR can be evaluated
 CMRR decreases at higher frequencies due to the pole of Ad and the zero of Acm

NTUEE Electronics – L.H. Lu 9-39

Active-Loaded MOS Differential Amplifier:
Differential-mode operation:
 Equivalent capacitances:
𝐶 =𝐶 +𝐶 +𝐶 +𝐶 +𝐶
𝐶 =𝐶 +𝐶 +𝐶 +𝐶 +𝐶
 Transconductance (Gm):
𝑔 𝑉 /2
𝑉 =−
𝑔 + 𝑠𝐶
𝑔 𝑔 𝑉 /2 𝑔 𝑉 /2
𝐼 = −𝑔 𝑉 = =
𝑔 + 𝑠𝐶 1 + 𝑠𝐶 /𝑔
𝑔 𝑉 /2
𝐼 =𝐼 +𝐼 = +𝑔 𝑉 /2
1 + 𝑠𝐶 /𝑔
𝐼 1 + 𝑠𝐶 /(2𝑔 )
𝐺 ≡ =𝑔
𝑉 1 + 𝑠𝐶 /𝑔
The pole and zero of Gm(s) are at very high frequencies
Minor pole and zero near unity-gain frequency
𝑔 𝑓
𝑓 = ≈
2𝜋𝐶 2
2𝑔
𝑓 = ≈𝑓
2𝜋𝐶

NTUEE Electronics – L.H. Lu 9-40

  The transfer function of the amplifier:
𝑅 𝑟 ||𝑟
𝑉 =𝐺 𝑉 =𝐺 𝑉
1 + 𝑠𝐶 𝑅 1 + 𝑠𝐶 𝑟 ||𝑟
𝑉 1 + 𝑠𝐶 /(2𝑔 ) 1 1 + 𝑠𝐶 /(2𝑔 ) 1
=𝑔 𝑅 =𝑔 𝑟 ||𝑟
𝑉 1 + 𝑠𝐶 /𝑔 1 + 𝑠𝐶 𝑅 1 + 𝑠𝐶 /𝑔 1 + 𝑠𝐶 𝑟 ||𝑟

The dominant pole is typically

1 1
𝑓 = =
2𝜋𝐶 𝑅 2𝜋𝐶 𝑟 ||𝑟
Common-mode operation:
 By taking CSS into account for the mid-band common-mode gain:
1 1 + 𝑠𝐶 𝑅
𝐴 ≈− =−
𝑅 2𝑔 𝑅
2𝑔 1 + 𝑠𝐶 𝑅
Contributes to a zero at lower frequency
Common-mode rejection ratio (CMRR):
 CMRR decreases at higher frequencies due to the dominant pole of Ad(s) and the zero of Acm(s)

NTUEE Electronics – L.H. Lu 9-41