Title: Role of Op-Amps
1Role of Op-Amps
- widely used building blocks for analog circuits
- widely used in communication circuits
- Widely used in analog signal processing
- critical block for data-converters
- Widely Used in High Volume ICs
- memory read out
- Disk-driver reading-writing
2New challenges in Analog Design
- Decreased Supply Voltage
- Increased Digital/Analog Interference
- Reduced Testability
- Increased Parametric Variations
3Critical Specifications for Op-Amps
- High Speed/Large Gain-Bandwidth
- Sufficient Output Swing
- High power efficiency
Desirable digital built-in-self-test and
self-calibration
4Existing structures Limitations
Cascode Amplifier
DC Gain Modest
Speed Excellent
Power Efficiency Good
Frequency Response Good
Output Swing Small
5Existing structures Limitations
Cascaded Amplifier
DC Gain High
Output Swings Large
Frequency Response Poor
Speed Poor
6Existing structures Limitations
Positive Feedback Amplifier
DC Gain Large
Speed High
Output Swing Good
Frequency Response Good
Low Yield
7Our Objective
- Low Voltage Compatible
- High DC Gain
- High Speed
- Good Output Swing
- High Yield
- Standard Digital Process Compatible
- Good Power Efficiency
8Our Approach
- Find new amplifier architectures
- Rely on digital logic to enhance performance
- Use simple digital circuit sensing amplifier
performance - Integrate controllability
- Use adaptive feedback control
9Positive Feedback Gain-Boosting
- Can operate at very low voltages
- Can achieve very high DC gain
- No compromise in bandwidth
- No appreciable power increase
- Conventional wisdom has two concerns
- positive feedback leads to RHP poles
- gain boosting is limited if requiring robustness
w.r.t. process variations
10Our response to the concerns
- positive feedback causes RHP open-loop poles, but
do not necessarily cause RHP closed-loop poles! - RHP open-loop poles can actually improve
performance of closed-loop amplifiers! - F16/F18 fighter jets have open-loop RHP poles,
they play critical role in achieving their
superior closed-loop performance - no need for robustness across process variations
- use positive feedback pole control after
fabrication
11Positive Feedback Operational Amplifiers
Open loop transfer function
Closed-loop transfer function
Open-loop pole
Closed-loop pole
12Positive Feedback Amplifier Architecture
Exploration
- Some positive feedback amplifier architectures
have a dc gain that is much more sensitive to the
feedback control variable than others - Although architecture optimality may be difficult
to determine, synthesis techniques may yield
better structures than those already considered - Comparative study of several structures with
existing positive feedback structures
13Negative Output Conductance Op Amp
Vdd
As (A-1)gdsn ? gds1gds2,
ADC ? 8 !
14Implementation Block Diagram
First Stage
Negative Conductance Generator
Vi
Vi-
2nd stage
Vip
Vop
Basic Amp
CMFB
Vin
Von
2nd stage
Bias Generator
15Basic Amplifier and the Second Stage
Second stage
Basic amplifier
16Negative Conductance Generator
Vdd
M3
M2
M7
M6
Vi
Vi-
Mg1
Mg2
Mn1
Mn2
Vctrl
Vctrl
M1
M5
M4
M8
A
VXX
VXX
Vop
Von
17Low Gain Stage A
18Positive Feedback Amplifier Architecture
Exploration
M3
f1(-?1VA-?2VB-?3VOUT)
VB
M4
f2(-?4VA-?5VB-?6VOUT)
VOUT
M2
f3(-?7VA-?8VB-?9VOUT)
VA
M1
VIN
Overhead for realizing this structure can be very
small
19Example positive feedback amp1
Gain infinity
20Example positive feedback amp2
Gain infinity
21Example positive feedback amp3
Gain infinity
22Example positive feedback amp4
Gain infinity
23DC Sweep of amp1 in 0.1um process
24Preliminary Results DBISTSC PFAmp
A Digital Programmable Amplifier
Bifurcation in Positive Feedback Amplifier
Simulation and Measurement Results
25A digital controllable Amplifier
- Low-Voltage Compatible
- 3 transistors from VDD to VSS
- High Gain
- Attenuator provides positive feedback
- negative-conductance compensation
- High Speed
- Single Stage
26Small Signal Linear Equivalence
27Nonlinear Dynamic
Dynamic equation
Equilibrium manifold (DC Transfer)
28Amplifiers DC Characteristics
y
0
x
?
29Amplifiers DC Characteristics
y
?0.03
?0
?00.02
L
30Observations
31Phase Diagram in Y-? Plane
x0
32Phase Diagram in Y-? Plane
y
x-0.001
?
33Phase Diagram in Y-? Plane
y
x0.001
?
34Observations
35Pull up/Pull down Circuit
pullup
pulldown
36Pull up/Pull Down
y
OHOL10
x
37Pull up/Pull Down
Pull up, stay high
y
OHOL11
Pull down, return high
x
38Pull up/Pull Down
Pull up, return low
y
OHOL00
Pull down, stay low
x
39Sensing Process
40Linear MOS Attenuator
3. ? controlled by Aspect Ratio
4. k controlled by Aspect Ratio
5. Infinite input impedance at DC
41Adaptively Controllable Attenuator
42Programmable Attenuation
43Gain Enhancement Requirement
Gain is related to controllable Attenuator
Required Attenuator Control Code
Achievable Gain Lower Bound
44Design Specification for Attenuator DAC
1. Sufficient Coverage
2. Fine Resolution
3. DAC Size
45Effect of Offset on Sensing
offset
O
46Offset Compensated Sensing Circuit
47Design Specifications for input DAC
1. Sufficient Coverage
2. Fine Resolution
3. DAC Size
48Design Specifications
Aspec80dB
y
?0.0001
?0variation0.006
N1?6
Lspec20?V
offset variation 50mV
N2 ?12
49Combined Sensing and Control Logic
- Implement with inexpensive digital logic
- Time-efficient processing
- Accomplish offset compensation
- Realize optimal control code searching
- Adaptive feedback control
50Functional Block Diagram
51Nested attenuation/bifurcation control
initialize
? Attenuator DAC input
?(?L?H)/2
no
bifurcation detection
Bifurcation?
yes
no
?L?
?H?
52Bifurcation Detection Algorithm
initialize
? Input of offset compensation DAC
2N21 input codes
?(?L?H)/2
no
stepN2?
yes
Pull-up/Down
break without bifurcation
OHOL?
bifurcation detected
00
10
11
? H ?
break
? L ?
53Convergence Proof
Conditions on offset-DAC range guarantee initial
conditions
- at step 0
- 1.When ??L0, OHOL11
- 2.When ??H2N2, OHOL00
- 3. ?(?L)lt offset ??(?H)
- 4. ?H-?L2N2
54Convergence Proof (Continue)
- at step k1,
- ?(?L ?H)/2, pull up/down
- if OHOL10, bifurcation is detected, break
- if OHOL11, ?(?)ltoffset, let ?L ?
- if OHOL00, ?(?)?offset, let ?H ?
- if bifurcation is not detected, continue.
- then
- ?(?L)lt offset ??(?H)
- ?H - ?L2N2-(k1)
- At ?L, OHOL11 at ?H ,OHOL00
- This iteration ends either with bifurcation
detected or stepN2 without bifurcation.
55Convergence Proof (Continue)
- If stepN2, ?H-?L1
- ?(?L)lt offset ??(?H)
- when ??L, OHOL11, ?(?L)lt lower bound of
hysteresis - when ??H, OHOL00, ?(?H)gt up bound of hysteresis
- Lhysgt?(?H)- ?(?L)
?(?L)
?(?H)
offset
potential hysteresis width
Lhys
lower bound
up bound
56Convergence Proof (Attenuator)
Conditions on m-DAC range guarantee initial
conditions
- at step 0
- 1.When ? ?L0, no bifurcation
- 2.When ? ?H2N1, bifurcation
- 3. ?(?L)??0lt ?(?H)
- 4. ?H- ?L2N1
57Convergence Proof (Attenuator)
- step k1 ?(?L)??0lt ?(?H), ?H- ?L2N2-k
- let ?(?L ?H)/2, do bifurcation detection
- if bifurcation is detected, ?(?)gt ?0 ,let ?H ?
- else let ?L ?
- at both situations, ?(?L)??0lt ?(?H), ?H-
?L2N1-k-1
58Convergence Proof (Attenuator)
- This iteration ends with no bifurcation and finds
an optimal control code ?L
59Simulation Environment
- Ami 0.5um CMOS (most)
- Cadence
- Spectre (simulator)
- Hspice (simulator)
- SpectreVerilog (simulator)
- Analog Artist
- Schematic
- Virtuoso (layout tool)
- SEDSM (Floor-plan /Route )
- Synopsys
60Simulation Results
16-b R-1.8R DAC
??5?VltLspec
Offset Compensation DAC Characteristics
61M2
M4
M6
M1
M3
M5
62Digitally controlled PF Amplifier
VDD
?
?
-
-
cmfb
VSS
63Two Stage Amplifier
?
vo1
vi1
vop
PU
PD
A2
Amp
A1
vo2
vi2
von
64R-2R DAC
Matched 2R
-
R
Matched 2R
-
R
Designed 2R
-
xR
Designed 2R
-
xR
For K LSB
For K LSB
For N
-
K MSB
For N
-
K MSB
output
R
R
R
R
0.94R
0.94R
2R
1.9R
2R
1.9R
2R
2R
2R
2R
2R
2R
2R
2R
R-2R
0
1
N
-
2
N
-
1
0
1
N
-
2
N
-
1
DAC
ground
ground
reference
reference
Xlt1
Xlt1
LSB
LSB
MSB
MSB
65Two way switch
VH
Vo
Vo
0
0
VL
VH
VL
66Comparator
Vout
Comparator
67Self-Calibrated High Gain Amplifier
16-bit R-2R
DAC
code
RN
calibr
vcm
controller
Amp
Comparator
CLK
PU
PD
DS
68Algorithm for Controller
- Step 1 Initialize ?
- Step 2 Sweep input
- Step 3 Use pull-up/down method to detect
hysteresis - Step 4 if hysteresis exists, decrease ? else
increase ? - Step 5 if ? is very close to ?0, finish else
goto step 2
69Searching Process
Offset Compensation DAC
32768
16384
24576
16384
28672
25578
24577
No Bifurcation
Attenuator Control DAC
Max searching time 7x16 cycles
70Bifurcation Simulation
x0
71Pull-up/Pull Down
x-0.1mV
72AC Response
50dB gain enhancement
73Step Response
74Step Response
Significant Settling accuracy Improvement
75Robustness Over Temperature
76Simulated Performance in 90nm CMOS
Easy Migration
77One Fabricated Amplifiers Performance
Tab. 1 Simulated DC gain over wide temperature
range
Gain (dB) SS SF FS FF NN
0C 84 85 72 92 103
27C 82.2 89 84 85 110
80C 70.4 84.5 83.7 89 88
manual tuning
AMI 0.5um CMOS
Tab. 2 Measured DC gain of fabricated chips
Chip 1 2 3 4 5
Gain (dB) 90 87.6 89.3 84.5 86.5
78Two-Stage to enhance linearity
79Two stage amp DC Sweep
80Two stage Gain vs Vout
gt130dB over - 0.75V
81Amplifier performance summary
- Berkeley projected 90nm process
- power supply voltage Vdd1.1
- two stage total current Itot18ma
- Total power consumption Ptot20mw
- Capacitive load CL4p
- Unity gain frequency UGF 1.35GHz
- Phase margin at UGF 39 deg
- Frequency at gain 2 G2F 804MHz
- Phase margin at G2F 65 deg
82Switched-Capacitor Amplifier
Precision Multiply-by-Two Circuit
83Transient Simulation Results
100 MS/s 60 swing ?0.3Vdiff Settling accuracy
8.9e-4 10.1 bits
84Transient Simulation Results
100 MS/s full swing ?0.5Vdiff Settling accuracy
0.0013 9.6 bits
85Transient Simulation Results
80 MS/s 60 swing ?0.3Vdiff Settling accuracy 2
.1e-4 12.24 bits
86Transient Simulation Results
40 MS/s Settling accuracy 3.1e-5/0.56.2e-5 14.
0 bits
input0.25V output0.499969V
87Transient Simulation Results
10 MS/s Settling accuracy 6e-6/0.51.2e-5 16.3
bits
input0.25V output0.499994V