Escolar Documentos
Profissional Documentos
Cultura Documentos
Hoi Lee
Professor
Administrative
Class Information
Time: Fridays 1:00 p.m. 3:45 p.m.
Location: JSOM 2.902
TA: Qi Cheng (qxc150030@utdallas.edu)
1
Grading Policy and Arrangement
Email Use
2
Textbook & Reference Books
Textbook
Class notes and research papers (all provided in the course
webpage; Not a single book can cover all materials in the
course)
Reference Books
Analog IC Design with Low Dropout Regulators, by Gabriel
Rincon-Mora, McGraw-Hill Professional Publisher, 2009. (ISBN:
0-071-60893-1)
Pulse-Width Modulated DC-DC Power Converters, by Marian K.
Kazimierczuk, John Wiley & Sons Publisher, 2008 (ISBN: 0-470-
77301-7)
Fundamentals of Power Electronics, 2nd Edition, by R. W.
Erickson and D. Maksimovic, Kluwer Academic Publisher, 2001.
(ISBN: 0-792-37270-0)
Elements of Power Electronics, Philip T. Krein, Oxford University
Press, 1998. (ISBN: 0-135-25643-7)
EE6378 Lecture 1 pg. 5
Course Topics
Main Topics
Introduction to Magnetics
3
Course Contents
Introduction
4
Converter Classification
Problems:
Input supply voltage is not constant
Current consumption of electronic equipment varies under
different conditions
Performances of internal systems in the equipment are
not consistent under different supply voltages and current
consumptions
Power Management ICs:
Always provide an regulated power source to the
electronic equipment
High power and area efficiencies, and low cost
5
Needs and Features of Portable Devices
But, Limited
Battery Capacity
Energy Source
6
Types & Comparisons of DC-DC Converters
No step-up conversion
Simplest and lowest
LDO cost Poor efficiency if the
voltage mismatch between
Low output noise input and output is larger
Inductor
-less High switch count
SCPC (or Step up / down
conversion Lack of effective control
Charge Pump
scheme
without Low cost
Regulation) Poor power efficiency in
High energy density
regulator
EE6378 Lecture 1 pg. 13
SMPCs
DC-DC conversions from Li-Ion cell battery (nominal 3.6V) to low
supply voltage (1.2V) for internal digital baseband processors
Dynamically adjust supply voltages in RF power amplifier
Linear Regulators (LDOs)
Sensitive RF or audio blocks which have stringent noise and
PSRR requirements
Post-regulators following switching regulators
SC Regulators
LCD displays, backup-battery boost converters, etc.
Lower cost solution to substitute SMPCs as no inductor is
required
7
Combining Power Converters
8
Integrated DC-DC Converters (2)
9
Integrated DC-DC Converters (4)
Low-dropout regulator
Main feature:
Has only one on-chip power
transistor
Area is much smaller than SCPC
and SMPC
Chip area: 0.31mm2
Technology: AMS 0.6m CMOS
Publication: JSSC Oct. 03
10
Voltage References (2)
11
Review on
Poles & Zeros; Bode Plots; and Modeling of
Single and Two-Stage Amplifiers
Transfer Function
vi(s) System
vo(s)
H(s)
12
Transfer Function (2)
R
A transfer function is defined by
v (s ) vi(s) C vo(s)
H (s ) = o
v i (s )
13
Poles and Zeros (2)
For a physical system, all coefficients are real. Hence, the
poles and zeros can have only two forms: (1) the roots are
real, and (2) if one root is complex, its conjugate has to be
another root (e.g. If z1 = -p+jq is a root, then z1* = -p-jq is
another root).
14
Frequency Responses and Bode Plot (2)
20log|H(s)| p1 10 p1 2
20log|H(s)| p1 102p1
0.1p1 10p1 103p1 3
0.1p1 10p1 10 p1
0dB w w
0dB
-20dB (log scale) (log scale)
-20dB
-40dB -40dB
-60dB -60dB
2
H(s) Unstable!
H(s) p1 10 p1 3
0.1p1 10p1 10 p1
o
w
o
0 90
o (log scale) o
-45 45
-90
o
0
o
w
0.1p1 10p1 103p1 (log scale)
p1 102p1
1 1
LHP Pole : H (s ) = RHP Pole : H (s ) =
s s
1+ 1-
p1 p1
EE6378 Lecture 1 pg. 29
15
Modeling of a Single-Stage Amplifier (1)
vo
vi
vi Rin Ro CL vo
Gmvi
CL
vi, vo, Rin, Gm, Ro, and CL are the input signal, output signal,
input resistance, effective transconductance, output
resistance, and loading capacitance, respectively.
16
Modeling of a Single-Stage Amplifier (3)
v1 vo
vi
C1 CL
v1
vi Rin Ro1 C1 vo
Gm1vi Gm2v1 Ro2 CL
17
Modeling of a Two-Stage Amplifier (2)
The transfer function A(s) is given by, again, solving the nodal
equations:
v (s ) Gm1Gm2Ro1Ro 2 Ao
A(s ) = o = =
v i (s ) (1 + sRo 2CL ) (1 + sRo1C1) (1 + s ) (1 + s )
p1 p2
18
Modeling of a Two-Stage Amplifier (4)
20log|A(s)|
As will be discussed in the next -20dB/dec
Ao
chapter, when there are 2 poles
-40dB/ dec
before the unity-gain frequency, the
amplifier is generally unstable. Thus, a0dB w
(log scale)
p1 p2
circuit technique known as frequency
compensation is utilized to stabilize the A(s)
w
amplifier. Particularly, Miller 0
o
(log scale)
Compensation is commonly used in
two-stage amplifier design. o
-180
19
Extra Notes (-3dB Frequency)
20
Extra Notes (Phase Shift)
Phase Shift
The phase shifts of LHP/RHP pole/zero are shown below
Ho w
1. LHP pole : H (s ) = H ( jw ) = - tan-1( )
s p1
1+
p1
Ho w
2. RHP pole : H (s ) = H ( jw ) = + tan-1( )
s p1
1-
p1
s w
3. LHP zero : H (s ) = Ho (1 + ) H ( jw ) = + tan-1( )
z1 z1
s w
4. RHP zero : H (s ) = Ho (1 - ) H ( jw ) = - tan-1( )
z1 z1
Example 1
21
Solution to Example 1
1 + 0.22
20 log10 H (2p * 2 * 106 ) = 20 log10 = -32.85dB
1 + 202 1 + 22
H (2p * 2 * 106 ) = tan-1(0.2) - tan-1(20) - tan-1(2) = 11.3! - 87.1! - 63.4! = -139!
Review on
Stability & Frequency Compensation of
Two-Stage Amplifiers
22
Simplification for Two-Stage Amplifier
M3 M4 Cm
V1 ML
Vo gm1 gmL
Vin Vo
Cm
-A1 -AL
M1 M2
Vin CL r1 rL CL
C1
Mb2
Mb1 Mb3
Fig. 1 Fig. 2
Cm
Cm
gm1 gmL V1 Vo
Vin Vo
-A1 -AL gmLV 1
gm1Vin r1 C1 rL CL
r1 rL CL
C1
Fig. 1 Fig. 2
23
Example 1
M3 M4
V1 ML
Vo
Cm
M1 M2
Vin
Mb2
Mb1 Mb3
Answer
Cm
V1 Vo
gmLV 1
gm1Vin r1 C1 rL CL
24
Why Frequency Compensation?
Cm
V1 Vo
gmLV 1
gm1Vin r1 C1 rL CL
Cm
V1 Vo Cm
g m1gmL r1rL (1 - s)
gmLV 1 Vo g mL
gm1Vin r1 C1 rL CL =
Vin (1 + sC g r r )(1 + s CL )
m mL 1 L
gmL
Numerator:
DC gain
Zero: (1-as) RHP zero, (1+as) LHP zero
1 RHP zero exits phase margin degradation
Denominator:
Poles: (1+as+bs2+), all terms should be positive (LHP poles);
otherwise amplifier is unstable
2 LHP poles exist
25
Miller Compensation in Two-Stage Amplifier (2)
Cm
g m1gmL r1rL (1 - s)
DC Gain P-3dB Vo g mL
=
Vin (1 + sC g r r )(1 + s CL )
m mL 1 L
-20dB/decade gmL
0dB
s DC Gain = gm1gmLr1rL=A1AL
UGF
p2 RHP zero (zRHP) = gmL/Cm
-40dB/decade zRHP p-3dB = 1/CmgmLr1rL
-20dB/decade p2 = gmL/CL
UGF = DC Gainp-3dB =
gm1/Cm
26
Dimension Condition of Cm
Quiz
27
Answer
zRHP zRHP
No Free Lunch!!!
Probe Further
DC Gain P-3dB
P-3dB
DC Gain
Increasing gm1
p2
0dB UGF s
UGF zRHP
p2
zRHP
28
Correct Way to Increase BW
Rule of Thumb:
Larger current should be allocated to the
output stage for UGF enhancement!!
gmL >> gm1!
EE6378 Lecture 1 pg. 57
29
Miller Compensation with Null Resistor
Rm Cm
gm1 gmL
Vin Vo
-A1 -AL
r1 rL CL
C1
1
No change in pole locations
gm1gmL r1rL [1 - sCm ( - Rm )]
Vo gmL Rm is used to improve PM as
=
Vin CL zRHP is removed by Rm = 1/gmL
(1 + sCm gmL r1rL )(1 + s )
gmL PM = 63.4o
low-power design condition
Dimension Condition of Rm
LHP zero is generated if Rm > 1/gmL
If the LHP zero is used to cancel p2, then Rm is set as
CL 1 g 1
Rm = (1 + ) = (1 + mL )
Cm gmL 2gm1 gmL
Both zRHP and p2 are cancelled
Rm cannot be too large since very large Rm causes
open circuit and no pole-splitting effect due to Miller
compensation (Rm < r1/10)
Rule of Thumb:
1/gmL Rm < r1/10
30
Circuit Implementation of Miller Compensation
M3 M4
ML
Vo
Rm Cm
M1 M2
Mb2
Mb1 Mb3
31
Design Example
Given:
CL=10pF
UGF > 3MHz
PM > 60o
DC Gain > 80dB
SR > 2.5V/s
Power Consumption < 160W
Supply Voltage = 2V
Designers job: choose Rm, Cm, (W/L)i, Li, I to
meet specifications!!
What are the relationship between designers
job and the specifications?
EE6378 Lecture 1 pg. 64
32
Design Example (2)
33
Design Example (3)
M3 M4
ML
Vo
Rm Cm
M1 M2
Mb2
Mb1 Mb3
Performance Summary
Actual Circuit
Specifications Theoretical Hybrid-p
(TSMC 0.35m CMOS)
Supply
2V
Voltage
Power
< 160 W 144.7 W N. A. 150 W
Consumption
DC Gain > 80 dB 84 dB 84 dB 85 dB
CL 10pF
34
Simulation Tips
References
35