8/12/2019 L18_f07
1/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 1
Midterm Exam Statistics
0
1
2
3
4
5
6
10 15 16 17 17.5 18
Grade
Max.=20
Number ofOccurrence
Mean=16
Standard Deviation: 2
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 2
EE247Lecture 18
ADC Converters Sampling (continued)
Sampling switch considerations
Switch induced distortion
Sampling switch conductance dependence on inputvoltage
Clock voltage boosters
Sampling switch charge injection & clock feedthrough
Complementary switch
Use of dummy device Bottom-plate switching
Track & hold circuits
T/H circuit incorporating gain & offset cancellation
8/12/2019 L18_f07
2/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 3
Summary of Last Lecture
DAC Converters (continued) Dynamic element matching
DAC reconstruction filter
ADC Converters Sampling (continued)
Sampling switch considerations
Thermal noise due to switch resistance
Sampling switch bandwidth limitations
Switch induced distortion Sampling switch conductance dependence on input
voltage
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 4
Practical SamplingSummary So Far!
2
2
212
B
B
FS
C k TV
( )1 forinON o o ox DD thDD th
WVg g g C V V
V V L
= =
0.72
s
RB f C
8/12/2019 L18_f07
3/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 5
Switch On-ResistanceSwitch MOS operating in triode mode:
Vin
C
M1
1VDD( ) ( )
( )( )
0
1,
2
1 1
is a function of results in distortion
What if instead of connecting G to a fixed voltage,
DS
D triodeDSD triode ox GS TH DS
ON DS V
ON
ox GS th ox DD th in
ON in
dIW VI C V V V
L R dV
RW W
C V V C V V V L L
R V
=
= =
a floating and fixed voltage source is connected to G & S?
Desirable to maximize on voltage of GS Minimize ONR
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 6
Boosted & Constant VGS Sampling
VGS=const.
OFF ON
Increase gate overdrive voltage as
much as possible + keep VGSconstant
Switch overdrive voltage
independent of signal level Error due to finite RON linear (to
1st order)
Lower Ron lower time constant
Higher frequency of operation
Gate voltage VGS =low
Device off
Beware of signal
feedthrough due to parasiticcapacitors
8/12/2019 L18_f07
4/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 7
Constant VGS Sampling
(= voltage @ the switch input terminal)
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 8
Constant VGS Sampling Circuit
VP1
100ns
M12
M8
M9
M6
M11Vin
1.5V1MHz
Chold
P
C1 C2
M1 M2
VDD=3V
M3
C3
M5
M4
P
This Example: All device sizes:10/0.35All capacitor size: 1pF (except for Chold)
Note: Each critical switch requires a separate clock booster
Vg
Va Vb
Sampling switch & C
PB
Ref: A. Abo et al, A 1.5-V, 10-bit, 14.3-MS/s CMOS Pipeline Analog-to-Digital Converter, JSSC
May 1999, pp. 599.
PB
8/12/2019 L18_f07
5/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 9
Clock Voltage Doubler Operation
C1 C2
M10ff
M2Saturation
mode
VP1
=clock
PB
VDD=03V
P
a) Startup
03V
03V 00
03V 0(3V-VthM2)Acquire
charge C1 C2
M1Triode
M2off
VP1
PB
VDD=3V
P
3V0
3V0
3V03V (3V-VthM2) (6V-VthM2)
b) Next clock transition
03V
VP1
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 10
Clock Voltage Doubler Operation
C1 C2
M10ff
M2
VP1
PB
VDD=3V
P
03V
03V
3V ~6V
3V0
c) Next clock phase
(6V-VthM2) (3V-Vth
M2) ~ 3V
M2Triode
Acquires
charge
Both C1 & C2
charged to
VDD after one
clock cycle
Note that
bottom plate
of C1 & C2 is
either 0 or
VDD while top
plates are atVDD or 2VDD
8/12/2019 L18_f07
6/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 11
Clock Voltage DoublerSimulation
C1 C2
M1 M2
VP1Clock period: 100ns
PB
P_Boost
VDD
2VDD
0
VDD=3V
R1 R2
*R1 & R2=1GOhmdummy resistors added for simulation only
P
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 12
Constant VGS Sampler: Low
Sampling switch M11
is OFF
C3 charged to ~VDDInput voltage
source
M3Triode
C3
M12
Triode
M4
OFF
VS11.5V1MHz
Chold
1pF
~ 2 VDD
(boosted clock)
VDD
VDD
OFF M11OFF
DeviceOFF
VDD=3V
8/12/2019 L18_f07
7/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 13
Constant VGS Sampler: High
C3 previously
charged to VDD
M8 & M9 are on:
C3 across G-S of M11
M11 on with constant
VGS = VDD
C31pF
M8
M9 M11
VS11.5V1MHz
Chold
1pF
VDD
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 14
Constant VGS SamplingSimulation
Input Switch VGate
Input Signal
Chold Signal
8/12/2019 L18_f07
8/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 15
Boosted Clock SamplingComplete Circuit
Ref: A. Abo et al, A 1.5-V, 10-bit, 14.3-MS/s CMOS Pipeline Analog-to-Digital Converter, JSSC
May 1999, pp. 599.
Clock Multiplier
Switch
M7 & M13 for
reliability
Remaining issues:
-VGS constant only
for Vin
8/12/2019 L18_f07
9/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 17
Advanced Clock Boosting Technique
clk low Capacitors C1a & C1b charged to VDD
MS off Hold mode
Sampling
Switch
clk low
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 18
Advanced Clock Boosting Technique
Sampling
Switch
clk
high Top plate of C1a & C1b connected to gate of sampling switch Bottom plate of C1a connected to VIN Bottom plate of C1b connected to VOUT VGS & VGD of MS both @ VDD & ac signal on G of MS average of VIN &
VOUT
clk high
8/12/2019 L18_f07
10/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 19
Advanced Clock Boosting Technique
Gate tracks average of input and output, reduces effect of IR drop athigh frequencies
Bulk also tracks signal reduced body effect (technology used allowsconnecting bulk to S) Reported measured SFDR = 76.5dB at fin=200MHz
Ref: M. Waltari et al., "Aself-calibrated pipeline
ADC with 200MHz IF-sampling frontend,"ISSCC 2002, Dig.Tech. Papers, pp. 314
Sampling
Switch
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 20
Constant Conductance Switch
Ref: H. Pan et al., "A 3.3-V 12-b 50-MS/s A/D converter in 0.6um CMOS with over 80-dBSFDR," IEEE J. Solid-State Circuits, pp. 1769-1780, Dec. 2000
8/12/2019 L18_f07
11/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 21
Constant Conductance Switch
Ref: H. Pan et al., "A 3.3-V 12-b 50-MS/s A/D converter in 0.6um CMOS with over 80-dBSFDR," IEEE J. Solid-State Circuits, pp. 1769-1780, Dec. 2000
OFF
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 22
Constant Conductance Switch
Ref: H. Pan et al., "A 3.3-V 12-b 50-MS/s A/D converter in 0.6um CMOS with over 80-dBSFDR," IEEE J. Solid-State Circuits, pp. 1769-1780, Dec. 2000
ON
M2 Constant current
constant gds
M1 replica of M2
& same VGS
as M2
M1 also
constant gds
Note: Authors report requirement
of 280MHz GBW for the opamp for
12bit 50Ms/s ADC
Also, opamp common-modecompliance for full input range
required
8/12/2019 L18_f07
12/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 23
Switch Off-Mode Feedthrough Cancellation
Ref: M. Waltari et al., "A self-calibrated pipeline ADC with 200MHz IF-sampling frontend,"ISSCC 2002, Dig. Techn. Papers, pp. 314
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 24
Practical Sampling
Vo
C
M1
1
Rsw = f(Vi) distortion
Switch charge injection & clock feedthrough
Vi
8/12/2019 L18_f07
13/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 25
Sampling Switch Charge InjectionSwitching from Track to Hold
Vi VO
Cs
M1
VG
First assume Vi is a DC voltage When switch turns off offset voltage (V) induced on Cs Why?
VG
t
VH
Vi
VL
Vi +Vth
VO
Vi
toff
V
t
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 26
SamplingSwitch Charge Injection
Channel distributed RC network formed between G,S, and D
Channel to substrate junction capacitance distributed & voltage dependant
Drain/Source junction capacitors to substrate or well voltage dependant
Over-lap capacitance Cov =LDxWx Cox associated with G-S & G-D overlap
MOS xtor operating in triode region
Cross section view
Distributed channel resistance &
gate & junction capacitances
S
G
D
B
LD
L
Cov Cov
Cjdb
Cjsb
CHOLD
8/12/2019 L18_f07
14/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 27
Switch Charge InjectionSlow Clock
Slow clock clock fall time >> device speed During the period (t- to toff) current in channel discharges channel
charge acquired during the previous clock cycle into low
impedance signal source
Only source of error Clock feedthrough from Cov toCs
VG
t
VH
Vi
VL
Vi +Vth
tofft-
Device still
conducting
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 28
Switch Clock FeedthroughSlow ClockVG
t
VH
Vi
VL
Vi +Vth
VO
Vi
toff
V
t
D
Cov
VG
( )
( )
( ) ( )
( )
( )
ovi th L
ov s
ovi th L
s
o i
ov ov ovo i i th L i th L
s s s
o i os
ov ovos th L
s s
CV V V V
C C
CV V V
CV V V
C C CV V V V V V 1 V V C C C
V V 1 V
C Cwhere ; V V V
C C
= + +
+
= +
= + =
= + +
= =
t-
Cs
VO
8/12/2019 L18_f07
15/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 29
Switch Charge Injection & Clock FeedthroughSlow Clock- Example
( )
' 2ov ox th L
ov
s
ovos th L
s
C 0.1 fF / C 9 fF / V 0.4V V 0
C 1 0 x0 .1 fF / .1 %
C 1pF Al lo wi ng 1/ 2L SB AD C re so lu ti on ~ 9bi t
CV V V 0.4mV C
= = = =
= = =
= Wn preferable)
11B
1
1B
VG
t
VH
Vi
VL
1 1B
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 38
Switch Charge InjectionComplementary Switch
Fast Clock
In fast clock case
To 1st order, offset due to overlap caps
cancelled for equal device width
Input voltage dependant error worse!
1
1B
VG
t
VH
Vi
VL
( )
( )
( )
ch n n ox n H i th n
th pch p p ox p i L
ch pch no
s s
o i os
n ox n p ox p
s
Q W C L V V V
VQ W C L V V
Q1 QV
2 C C
V V 1 V
1 W C L W C L
2 C
=
=
= + +
+
8/12/2019 L18_f07
20/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 39
Switch Charge InjectionDummy Switch
Vi
VO
Cs
t
VH
Vi
VL
VG VGB
WM2=1/2WM1
VG VGB
M1 M2
M1 M11 ch ov
M 2 M 22 c h ov
2 M1 2 1
1Q Q Q
2Q Q 2Q
1For W W Q Q
2
+
+
= =
Q1 Q2
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 40
Switch Charge InjectionDummy Switch
Vi
VO
Cs
t
VH
Vi
VL
VG VGB
Dummy switch same L as main switch but half W
Main device clock goes low, dummy device gate goes high dummy
switch acquires same amount of channel charge main switch needs tolose
Effective only if exactly half of the charge stored in M1 is transferred to M2
(depends on input/output node impedance) and requires good matching
between clock fall/rise
WM2=1/2WM1
VG VGB
M1 M2
Q1 Q2
8/12/2019 L18_f07
21/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 41
Switch Charge InjectionDummy Switch
Vi VOM1
VG
M2
VGB
To guarantee half of charge goes to each side create the same
environment on both sides
Add capacitor equal to sampling capacitor to the other side of the switch
+ add fixed resistor to emulate input resistance of following circuit
Issues: Degrades sampling bandwidth
CsCs
R
WM2=1/2WM1
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 42
Dummy Switch Effectiveness Test
Ref: L. A. Bienstman et al, An Eight-Channel 8 13it Microprocessor Compatible NMOS D/A
Converter with Programmable Scaling, IEEE JSSC, VOL. SC-15, NO. 6, DECEMBER 1980
Dummy switch
W=1/2Wmain
As Vin is
increased Vc1-Vin
is decreased
channel charge
decreased less
charge injection
Note large Ls
good devicearea matching
8/12/2019 L18_f07
22/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 43
Switch Charge InjectionDifferential Sampling
Vi+
VO+
Vi-
VO-
To 1st order, offset terms cancel
Note gain error still about the same Has the advantage of better immunity to
noise coupling and cancellation of evenorder harmonics
Cs
Cs
( )
( )
( )( )
o o od i i id
o o i ioc ic
o i 1 os1
o i 2 os2
1 2od i d id 1 2 ic os1 os2
V V V V V V
V V V V V V
2 2V V 1 V
V V 1 V
V V V V V V 2
+ +
+ +
+ +
= =
+ += =
= + +
= + +
+= + + +
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 44
Avoiding Switch Charge InjectionBottom Plate Sampling
Switches M2 opened slightly earlier compared to M1
Injected charge by the opening of M2 is constant since its DS voltageis zero & eliminated when used differentially
Since Cs bottom plate is already open when M1 is opened
No charge injected on Cs
1aVH
VL
t
1bViVO
M1
1b
1a M2Cs
8/12/2019 L18_f07
23/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 45
Flip-Around Track & Hold
vIN
vOUT
C
S1A
1D
S2
2
S2A
2
S3
1D
1 S1vCM
Concept based on bottom-
plate sampling
1
21D
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 46
Flip-Around T/H-Basic Operation1high
vIN vOUT
C
S1A
1D
S2
2
S2A
2
S3
1D
1 S1vCM
Charging C
11D2
Note: Opamp has to be
stable in unity-gain
configuration
Q1
=VIN
xC
8/12/2019 L18_f07
24/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 47
Flip-Around T/H-Basic Operation
2high
vINvOUT
C
S1A
1D
S2
2
S2A
2
S3
1D
1 S1vCM
Holding
1
21D
Q2=VOUTxCVOUT=VIN
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 48
Flip-Around T/H - Timing
S1 opens earlier than S1A
No resistive path from Cbottom plate to Gnd charge
can not change
"Bottom Plate Sampling"
vIN
vOUT
C
S1A
1D
S2
2
S2A
2
S3
1D
1S1
vCM
1
21D
8/12/2019 L18_f07
25/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 49
Charge Injection
At the instant of transitioning from track to hold
mode, some of the charge stored in sampling
switch S1 is dumped onto C
With "Bottom Plate Sampling", only charge
injection component due to opening of S1 and
is to first-order independent of vIN Only a dc offset is added. This dc offset can be
removed with a differential architecture
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 50
Flip-Around T/H
vIN vOUT
C
S1A
1D
S2
2
S2A
2
S3
1D
1 S1vCM
Constant switch VGSto minimize distortion
Note: Among all switches
only S1A & S2A
experience full
input voltage swing
1
21D
8/12/2019 L18_f07
26/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 51
Flip-Around T/H S1 is chosen to be an n-channel MOSFET
Since it always switches the same voltage, its on-resistance, RS1, is signal-independent (to first order)
Choosing RS1 >> RS1A minimizes the non-linearcomponent of R = RS1A+ RS1 Typically, S1A is a wide (much lower resistance than S1) &
constant VGS switch
In practice size of S1A is limited by the (nonlinear) S/Dcapacitance that also adds distortion
If S1As resistance is negligible delay depends only on S1resistance
S1 resistance is independent of VIN error due to finitetime-constant independent of VIN
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 52
Differential Flip-Around T/H
Ref: W. Yang, et al. A 3-V 340-mW 14-b 75-Msample/s CMOS ADC With 85-dB SFDR at Nyquist Input, IEEE
JOURNAL OF SOLID-STATE CIRCUITS, VOL. 36, NO. 12, DECEMBER 2001 1931
Offset voltage associated with charge injection of S11 & S12 cancelled by differential
nature of the circuit
During input sampling phase amp outputs shorted together
S11
S12
8/12/2019 L18_f07
27/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 53
Differential Flip-Around T/H
Gain=1 Feedback factor=1
112
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 54
Differential Flip-Around T/HIssues: Input Common-Mode Range
Vin-cm=Vout_com-Vsig_com Amplifier needs to have large input common-mode compliance
8/12/2019 L18_f07
28/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 55
Differential Flip-Around T/HIssues: Input Common-Mode Range
Vin-cm=Vout_com-Vsig_com
Amplifier needs to have large input common-mode compliance
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 56
Input Common-ModeCancellation
Ref: R. Yen, et al. A MOS Switched-Capacitor Instrumentation Amplifier, IEEE JOURNAL OF
SOLID-STATE CIRCUITS, VOL. SC-17, NO. 6,, DECEMBER 1982 1008
Note: Shorting switch M3 added
8/12/2019 L18_f07
29/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 57
Input Common-Mode Cancellation
Track mode ( high)VC1=VI1 , VC2=VI2Vo1=Vo2=0
Hold mode ( low)Vo1+Vo2 =0
Vo1-Vo2= -(VI1-VI2)(C1/(C1+C3))
Input common-mode level removed
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 58
T/H + Charge Redistribution Amplifier
Track mode: (S1, S3 on S2 off)
VC1=VosVIN , VC2=0
Vo=Vos
8/12/2019 L18_f07
30/30
EECS 247- Lecture 18 Nyquist Rate ADCs-Sampling 2007 H.K. Page 59
T/H + Charge Redistribution AmplifierHold Mode
Hold/amplify mode (S1, S3 off S2 on)
Offset NOT cancelled, but not amplified Input-referred offset =(C2/C1) x VOS, & often C2