Lecture 10

22
M.H. Perrott Analysis and Design of Analog Integrated Circuits Lecture 10 Frequency Response of Amplifiers Michael H. Perrott February 29, 2012 Copyright © 2012 by Michael H. Perrott All rights reserved.

description

general

Transcript of Lecture 10

Page 1: Lecture 10

M.H. Perrott

Analysis and Design of Analog Integrated CircuitsLecture 10

Frequency Response of Amplifiers

Michael H. PerrottFebruary 29, 2012

Copyright © 2012 by Michael H. PerrottAll rights reserved.

Page 2: Lecture 10

M.H. Perrott

Open Loop Versus Closed Loop Amplifier Topologies

Open loop – want all bandwidth limiting poles to be as high in frequency as possible

Closed loop – want one pole to be dominant and all other parasitic poles to be as high in frequency as possible

2

VoutZsrc

Vsrc

SourceAmp VoutZsrc

Vsrc

SourceAmp

Zf

Open Loop Closed Loop

Vout/Vin

(dB)

w (rad/s)w0 w1 w2

Vout/Vin

(dB)

w (rad/s)w0 w1 w2

VinVin

Page 3: Lecture 10

M.H. Perrott

OCT Method of Estimating Amplifier Bandwidth

OCT method calculates by the following steps:- Compute the effective resistance Rthj seen by each

capacitor, Cj, with other caps as open circuits AC coupling caps are not included – considered as shorts

- Form the “open circuit” time constant Tj = RthjCj for each capacitor Cj- Sum all of the “open circuit” time constants

3

⇒ BW ≈ 1Pmj=1RthjCj

rad/s

20log(Vout/Vin)(dB)

w (rad/s)w0 w1 w2

MidBand Gain f (Hz) =w (rad/s)

Note:

Page 4: Lecture 10

M.H. Perrott

Another Useful Analysis Tool: Miller Effect

Derive input impedance (assume gain of amplifier = A):

Consider the case where Zf is a capacitor

- For negative A, input impedance sees increased cap value- For A = 1, input impedance sees no influence from cap- For A > 1, input impedance sees negative capacitance!

Can be used to create active inductor for a specific frequency4

VoutA

Zf

Vin

Zin

iin

Zin =Viniin

=Vin

(Vin − Vout)/Zf=

VinZfVin −AVin

=Zf1−A

Zf =1

sC⇒ Zin =

1

s(1− A)C

Page 5: Lecture 10

M.H. Perrott

Key Capacitances for CMOS Devices

S D

GVGS

VD>ΔV

ID

LDLD

overlap cap: Cov = WLDCox + WCfringe

B

CgcCcb

Cov

CjdbCjsb

Cov

Side View

gate to channel cap: Cgc = CoxW(L-2LD)

channel to bulk cap: Ccb - ignore in this class

S D

Top View

W

E

L

E

E

source to bulk cap: Cjsb = 1 + VSB ΦB

Cj(0)

1 + VSB ΦB

Cjsw(0)WE + (W + 2E)

junction bottom wall cap (per area)

junction sidewall cap (per length)

drain to bulk cap: Cjsd = 1 + VDB ΦB

Cj(0)

1 + VDB ΦB

Cjsw(0)WE + (W + 2E)

23

(make 2W for "4 sided" perimeter in some cases)

L

5

Page 6: Lecture 10

M.H. Perrott

CMOS Hybrid- Model with Caps (Device in Saturation)

RS

RG

RD

RD

RS

RG

-gmbvsvgs

vs

rogmvgs

ID

Csb

Cgs

CgdCdb

Cgs = Cgc + Cov = CoxW(L-2LD) + Cov23

Cgd = Cov

Csb = Cjsb (area + perimeter junction capacitance)

Cdb = Cjdb (area + perimeter junction capacitance)

6

Page 7: Lecture 10

M.H. Perrott

OCT Thevenin Resistance Calculations

Cgs: Thevenin resistance between gate and source

Cgd: Thevenin resistance between gate and drain

7

Rthgs

Rthgd

Cgs

Cgd

Rthgs

Rthgd

RS

RG

RD RD

RS

RG

-gmbvsvgs

vs

rogmvgs

ID

Rthgs =RS(1 +RD/ro) +RG(1 + (gmb + 1/ro)RS +RD/ro)

1 + (gm + gmb)RS + (RS +RD)/ro

Rthgd = (RD +RG)(1− rods/ro) + rodsgmRGwhere rods = ro||

RD1 + (gm + gmb)RS

Page 8: Lecture 10

M.H. Perrott

OCT Example: Design Wide Bandwidth Amplifier

Step 1: identify AC coupling versus OCT capacitors- AC coupling caps will be regarded as shorts

Step 2: calculate individual OCT time constants Step 3: identify long OCT time constants and modify

circuit to improve its bandwidth

8

Rin

RLID

Vin

Vout

CL

Assumptions:gm = 1/(100), = 0, = 0Cgs = 10fF, Cgd = 3fFCsb = 5fF, Cdb = 4fFRin = 4kRL = 1kCL = 100fF

Page 9: Lecture 10

M.H. Perrott

Step 1: Identify OCT Capacitors

Which time constants are easy to calculate? How do we efficiently calculate the more difficult

cases?

9

Rin

RL

Vin

Vout

Rth1 (CL+Cdb)Rth2 (Cgs)

Rth3 (Cgd)

Assumptions:gm = 1/(100), = 0, = 0Cgs = 10fF, Cgd = 3fFCsb = 5fF, Cdb = 4fFRin = 4kRL = 1kCL = 100fF

Page 10: Lecture 10

M.H. Perrott

Step 2: OCT Time Constant Calculations

10

Rin

RL

Vin

Vout

Rth1 (CL+Cdb)Rth2 (Cgs)

Rth3 (Cgd)

Assumptions:gm = 1/(100), = 0, = 0Cgs = 10fF, Cgd = 3fFCsb = 5fF, Cdb = 4fFRin = 4kRL = 1kCL = 100fF

Easy ones:

Use formula for 3:

Rth1 = RL||Rthd = RL||∞ = RL = 1kΩ ⇒ τ1 = 1kΩ · 104fF = 104ps

Rth2 = Rin ||Rthg = Rin ||∞ = Rin = 4kΩ ⇒ τ2 = 4kΩ · 10fF = 40ps

Rthgd = (RD +RG)(1− rods/ro) + rodsgmRGwhere rods = ro||

RD1 + (gm + gmb)RS

= RD = RL

⇒ Rth3 = (RL +Rin)(1− 0) +RLgmRin = 5.5kΩ + 40kΩ = 45.5kΩ⇒ τ3 = 45.5kΩ · 3fF = 136.5ps

Page 11: Lecture 10

M.H. Perrott

Step 3: Identify Largest OCT Time Constant

11

Rin

RL

Vin

Vout

Rth1 (CL+Cdb)Rth2 (Cgs)

Rth3 (Cgd)

Assumptions:gm = 1/(100), = 0, = 0Cgs = 10fF, Cgd = 3fFCsb = 5fF, Cdb = 4fFRin = 4kRL = 1kCL = 100fF

Time constant associated with Cgd is the longest:

Why is this time constant so large given that it is associated with the lowest value capacitor?

How do we change the amplifier topology to reduce this time constant value?

τ3 = 45.5kΩ · 3fF = 136.5ps

Page 12: Lecture 10

M.H. Perrott

The Miller Effect Analysis Provides Helpful Intuition

Notice that Cgd is in the feedback path of the common source amplifier- Recall Miller effect calculation: - For this amplifier:

This analysis agrees well with OCT calculation of 136.5ps

12

Cgd

Rin

RL

Vin

Vout

A

Cin

Cgd

Cin

A = −gmRL ⇒ Cin = (1 + gmRL)Cgd = 11 · Cgd = 33fF

Cin = (1−A)Cgd

⇒ τ3 = RinCin = 4kΩ · 33fF = 132ps

Can we change the amplifier topology to lower this time constant?

Page 13: Lecture 10

M.H. Perrott

Consider Adding a Cascode Device

Examine the impact of this topological change using the Miller Effect analysis

13

Rin

RL

Vin

Vout

Cin

Cgd1

Vbias

M1

M2

A = −gm11

gm2≈ −1 ⇒ Cin = (1 + 1)Cgd1 = 2 · Cgd1 = 6fF

⇒ τ3 = RinCin = 4kΩ · 6fF = 24ps

Cascode device dramatically reduces the Cgd1 time constant!

Page 14: Lecture 10

M.H. Perrott

Does the Miller Effect Impact the Cascode Device?

Observe that the capacitance seen by Vbias is not of concern since this voltage is not part of the signal path

The signal path sees the time constant:

- This time constant is much smaller than the other time constants of the amplifier

14

Rin

RL

Vin

VoutCgd2

Vbias

M1

M2

τ4 = RL||Rthd2 · Cgd2 ≈ RL · Cgd2 = 1kΩ · 3fF = 3ps

Page 15: Lecture 10

M.H. Perrott

Perform OCT Calculations for Updated Amplifier

15

Rth4 (Cgd2)

Rin

RL

Vin

Vout

Vbias

M1

M2

Rth1 (CL+Cdb2)

Rth2 (Cgs1)

Rth3 (Cgd1)

Rth5 (Cgs2)

Rth6 (Cds1+Csb2)

Rth1 = RL||Rthd2 = RL = 1kΩ ⇒ τ1 = 1kΩ · 104fF = 104ps

Rth2 = Rin ||Rthg1 = Rin = 4kΩ ⇒ τ2 = 4kΩ · 10fF = 40ps

Rth5 = Rths2 ||Rthd1 ≈ 1/gm2||∞ = 100Ω ⇒ τ5 = 100Ω · 10fF = 1psRth4 = RL||Rthd2 ≈ RL = 1kΩ ⇒ τ3 = 1kΩ · 3fF = 3ps

Assumptions for all devices:gm = 1/(100), = 0, = 0Cgs = 10fF, Cgd = 3fFCsb = 5fF, Cdb = 4fFRin = 4kRL = 1kCL = 100fF

Rth6 = Rthd1 ||Rths2 =∞||1/gm2 = 100Ω ⇒ τ6 = 100Ω · 9fF = 0.9ps

Page 16: Lecture 10

M.H. Perrott

Perform OCT Calculations for Updated Amplifier

Use Thevenin formula for Cgd calculation:

16

Rth4 (Cgd2)

Rin

RL

Vin

Vout

Vbias

M1

M2

Rth1 (CL+Cdb2)

Rth2 (Cgs1)

Rth3 (Cgd1)

Rth5 (Cgs2)

Rth6 (Cds1+Csb2)

Assumptions for all devices:gm = 1/(100), = 0, = 0Cgs = 10fF, Cgd = 3fFCsb = 5fF, Cdb = 4fFRin = 4kRL = 1kCL = 100fF

Rth3 = (RD1 +RG1)(1− rods/ro1) + rodsgm1RG1

⇒ Rth3 = (1

gm2+Rin)(1− 0) +

1

gm2gm1Rin = 4.1kΩ+ 4kΩ = 8.1kΩ

⇒ τ3 = 8.1kΩ · 3fF = 24.3ps

where rods = ro1||RD1

1 + (gm1 + gmb1 )RS1

Page 17: Lecture 10

M.H. Perrott

Identify Longest OCT Time Constant

The load capacitance now presents the largest time constant:

17

Rth4 (Cgd2)

Rin

RL

Vin

Vout

Vbias

M1

M2

Rth1 (CL+Cdb2)

Rth2 (Cgs1)

Rth3 (Cgd1)

Rth5 (Cgs2)

Rth6 (Cds1+Csb2)

Assumptions for all devices:gm = 1/(100), = 0, = 0Cgs = 10fF, Cgd = 3fFCsb = 5fF, Cdb = 4fFRin = 4kRL = 1kCL = 100fF

Rth1 = RL||Rthd2 = RL = 1kΩ ⇒ τ1 = 1kΩ · 104fF = 104ps

Can we change the amplifier topology to lower this time constant?

Page 18: Lecture 10

M.H. Perrott

Add a Source Follower to the Output

Key idea: reduce the time constant associated with CL by decreasing the Thevenin resistance that it sees- Previous design presented RL = 1K to CL- Source follower presents Rths3 = 1/gm3 = 100 to CL

18

Rin

RL

Vin

VoutVbias

M1

M2

M3

Ibias CL

For all devices:gm = 1/(100), = 0, = 0Cgs = 10fF, Cgd = 3fFCsb = 5fF, Cdb = 4fFRin = 4kRL = 1kCL = 100fF

Source follower should reduce CL time constant by a factor of ten!

Page 19: Lecture 10

M.H. Perrott

Calculation of New CL Time Constant

Formal calculation:

19

Rth4 (Cgd2)

Rin

RL

Vin

VoutVbias

M1

M2

Rth1 (CL+Csb2)

Rth2 (Cgs1)

Rth3 (Cgd1)

Rth5 (Cgs2)

M3

Ibias

Rth8 (Cgd3)

Rth7 (Cgs3)

Rth6(Cds1+Csb2)

For all devices:gm = 1/(100), = 0, = 0Cgs = 10fF, Cgd = 3fFCsb = 5fF, Cdb = 4fFRin = 4kRL = 1kCL = 100fF

How large are the additional time constants created by M3?

Rth1 = Rths3 = 1/gm3 = 100Ω ⇒ τ1 = 100Ω · 104fF = 10.4ps

Page 20: Lecture 10

M.H. Perrott

Calculation of Additional Time Constants from M3

20

Rth4 (Cgd2)

Rin

RL

Vin

VoutVbias

M1

M2

Rth1 (CL+Csb2)

Rth2 (Cgs1)

Rth3 (Cgd1)

Rth5 (Cgs2)

M3

Ibias

Rth8 (Cgd3)

Rth7 (Cgs3)

Rth6(Cds1+Csb2)

For all devices:gm = 1/(100), = 0, = 0Cgs = 10fF, Cgd = 3fFCsb = 5fF, Cdb = 4fFRin = 4kRL = 1kCL = 100fF

Rth8 = RL||Rthd2 ≈ RL = 1kΩ ⇒ τ8 = 1kΩ · 3fF = 3ps

⇒ τ7 = 100Ω · 10fF = 1ps

Rth7 =RS3(1 +RD3/ro3) +RG3(1 + (gmb3 + 1/ro3)RS3 +RD3/ro3)

1 + (gm3 + gmb3 )RS3 + (RS3 +RD3)/ro3

⇒ Rth7 =1 + RD3/ro3 +RG3(gmb3 + 1/ro3)

gm3 + gmb3 + 1/ro3=

1 + 0 + 0

gm3 + 0 + 0= 100Ω

Page 21: Lecture 10

M.H. Perrott

Estimate Bandwidth Based on OCT Calculations

21

Rth4 (Cgd2)

Rin

RL

Vin

VoutVbias

M1

M2

Rth1 (CL+Csb2)

Rth2 (Cgs1)

Rth3 (Cgd1)

Rth5 (Cgs2)

M3

Ibias

Rth8 (Cgd3)

Rth7 (Cgs3)

Rth6(Cds1+Csb2)

τ1 = 10.4ps

τ2 = 40ps

τ3 = 24.3ps

τ4 = 3ps

τ5 = 1ps

τ6 = 0.9ps

τ7 = 1ps

τ8 = 3ps

BW ≈ 1Pmj=1RthjCj

=1

83.6ps= 11.96 Grad/s

⇒ BW ≈ 11.96

2π= 1.9GHz

Page 22: Lecture 10

M.H. Perrott

Summary

Two techniques prove very useful when designing amplifiers for desired frequency response behavior- Open Circuit Time Constant method- Miller Effect analysis

Thevenin resistance analysis in combination with the above offers tremendous insight for designing amplifier topologies- OCT method allows quick discovery of large time

constants- Miller effect provides intuition of the impact of placing

capacitors within feedback- Awareness of impedances presented by various

amplifier stages allows intuitive approach to achieve reduction of large time constants

22