Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
SECTION 7 - COMPARATORS
Page VII.0-1
Page VII.0-1
VII. COMPARATORS
Contents
VI.1
VI.2
VI.3
VI.4
VI.5
Chapter 11
Analog Systems
SYSTEMS
Chapter 7
CMOS
Comparators
Chapter 8
Simple CMOS
OTA's
Chapter 9
High Performance
OTA's
COMPLEX
CIRCUITS
Chapter 5
CMOS
Subcircuits
Chapter 6
CMOS Amplifiers
SIMPLE
Chapter 2
CMOS
Technology
DEVICES
Chapter 3
CMOS Device
Modeling
Chapter 4 Device
Characterization
Page VII.1-1
VII.1 - CHARACTERIZATION OF
COMPARATORS
What is a Comparator?
A comparator is a circuit which compares two analog signals and
outputs a binary signal based on the comparsion. (It can be an op amp
without frequency compensation.)
Characterization of Comparators
We shall characterize the comparator by the following aspects:
Resolving capability
Speed or propagation time delay
Maximum signal swing limits
Input offset voltage
Other Considerations
Noise
Power
Etc.
Page VII.1-2
VOLTAGE COMPARATORS
Definition of a Comparator
VA
VB
VOUT
-
Noninverting
VOUT
VOH
VOUT =
V O H
V OL
when VA VB
VA - VB
when V A < V B
VOL
Inverting
VOUT
VOUT =
V O L
V OH
when VA VB
VOH
when V A V B
VA - VB
VOL
Page VII.1-3
COMPARATOR PERFORMANCE
Page VII.1-4
Type
Offset Voltage
(Power supply)
Resolution
Speed (8 bit)
Open-loop
1-10 mV
300V (5V)
10 MHz
Regenerative
Charge
Balancing
0.1 mV
50V (5V)
50 MHz
0.1 mV
5mV (5V)
30 MHz
Page VII.1-5
VOUT
VOH
VP - VN
VOL
Model
VP
+
+
fo VP - VN
VN
fo( V P
V O H
- VN ) =
V OL
VO
for ( V P - V N ) 0
for ( V P - V N ) 0
Page VII.1-6
Transfer Curve
VOUT
VOH
VIL
VP - VN
VIH
VOL
Model
VP
+
+
f1 VP - VN
VN
VO
V O H for ( V P - V N ) VIH
f1( V P - V N ) = AV( V P - V N ) f o r V I L ( V P - V N ) V I H
V OL for ( V P - V N ) VIL
Page VII.1-7
VOS
VIL
VP - VN
VIH
VOL
First Order Model with Offset
+-VOS
VP
V'P
+
+
f1 V'P - V'N
-
V'N
VN
VOUT
VOL
VIH
VP - VN
VO
v = VIH + VIL
2
tP
VIL
Time
Page VII.2-1
M2
I2
IB
M1
VBIAS
vO
VSS
Fig. 7.2-1 Simple inverting comparator
VIN
VDD
vO
vN
VTRP
Page VII.2-2
t.
VDD
VDD
vO = V IN + VT2
c
2a
t.
a
2s
VIN
M2
vO
VBIAS
M1
M1 sat.
M1 act.
VBIAS - VT1
VSS
Operating RegionsvDS1 v GS1 - VT
VSS
VSS
VIN
vN
VTRP
VDD
Trip PointAssume both M1 and M2 are saturated, solve and equate drain
currents for VTRP. Assume 0.
K N W1
2
iD1 = 2 L V BIAS - V SS - V T 1
1
K P W2
2
iD2 = 2 L V D D - v I N - VT2
iD1=iD2
KN( W1/L1)
KP( W2/L2) ( V BIAS - V SS - V T 1)
W1
W2
I.e. V DD = -VSS = 5V, VBIAS = -2V and KN L = KP L
1
2
VTRP = 5-1-(-2+5-1) = 4-2 = 2V
Page VII.2-3
M3
M4
vO
M1
vP
M2
VBIAS
vN
M5
vO
VOH = VDD
VOH'
M1 & M2 in
saturation
VOL'
VOL
VSS
-1
+1
vP - vN
Av
Page VII.2-4
vP > vN
M3
I1
vP
M4
I2
M1
M2
vO
vN
ISS
VBIAS
I4
4
I5
Kp'( W4/L4)
M5
VSS
vP < vN
V O H ' = VD D -
I5
Kp'( W3/L3)
V OL ' = v N - V T 2
3. For further decrease in vO, M2 is nonsat
and therefore the VGS2 can increase
allowing the sources of M1 and M2 to
fall(as v P falls).
4. Eventually M5 becomes non-sat and I5
starts to decrease to zero. M2 becomes a
switch and v O tracks V S2(VDS5) all the
way to VSS.
V OL = V SS .
Page VII.2-5
TWO-STAGE COMPARATOR
Combine the differential amplifier stage with the inverter stage.
Sufficient gain.
Good signal swing.
VDD
M3
M4
M6
vN
M1
M2
I8
M8
M5
VSS
vP
vO
M7
Page VI.3-1
Try to keep all devices in saturation - more gain and wider signal
swings.
I5
I4 = 2, then DC balance is achieved under the following:
S6
S
. 7 VDG4 = 0 M4 is saturated.
=
2
S4
S5
Page VI.3-2
KN = 24.75 A/V2
KP = 10.125 A/V2
VTN = -VTP = 1V
N = 0.015V -1
P = 0.020V -1
+
2V
20 2V
10
M3
- M6 40
M4
20
10
I8
20
10
vN
M1
vP
M5
VSS =0V
= 89.9
I5
2 + 4
vGS6
14.1 mV
= 0.157 mV
VOS = A (diff) = 89.9
v
A v(diff) =
vO =5V
10
10 3V
M7
vGS6 =
i6
i7
M2
20A
M8
10
20
10
10
10
Page VI.3-3
VG1
VDG1
-
+ M1
VGS1
-
+
VDS1
-
I5
VBIAS
v G1 (max) = V D D -
VDS5
VSS
I5
21
M5
I5
23 - VT3(max) + V T1 (min)
Example
Design M1 through M4 for a CM input range 1.5 to 9 Volts when VDD =
10 V, ISS = 40A, and VSS = 0V. Table 3.1-2 parameters with |VTN,P| =
0.4 to 1.0 Volts,
I5
vG1(min) = VSS + VDS5 +
1 + VT1(max)
40A
+ 1 (assumed VDS5 0.1V- it probably more
1.5 = 0 + 0.1 +
1
reasonable to assume 1 is already defined and find 5)
1 =
KNW 1
2
L1 = 250 A/V
vG1(max) = VDD 3 =
W1 W 2
L1 = L2 = 14.70
I5
3 - |VT3(max)| + VT1(min)
K PW 3
2
L3 = 250 A/V
W3 W 4
L3 = L4 = 31.25
Page VI.3-4
+
gm1vid
r ds2
r ds4
v1 gm6v1
-
r ds6
r ds7
vid = vP - vN
gm1
ds2 + g ds4
Av = g
2
Av =
gm6
g ds6 + g ds7
W1 W6
KNKP L L
1 6
( 2 + 4) ( 6 + 7)
vout
I1I6
W6
W1
Using L = 5, L = 5, N = 0.015V-1 , P = 0.02V-1
1
6
and Table 3.1-2 values;
2 (17)(8)(5)(5) . -6 95199.10-6
Av =
10 =
(0.015+0.02)2 I1I6
I1I6
Assume I1 = 10 A and I6 = 100 A
Av = 3010
V OH - VOL
= Resolution = 5 mV (assume)
Av
5 .
then VOH - V OL =
1000 3000 = 15 Volts
Page VI.3-5
M3
vN
VGS6 +
-
M1
M2
vP
CL1
VBIAS
M6
i6 key node
vO
i7
CL2
M5
M7
i5
VSS
V GS6 = VDD - v P + V D G 2
dv
iC = C dt , t =
v
CI
t2+ = C L2
K P W6 V
2 L6 ( D D
t+2
VTRP3
V
VDD
SS
VTRP3
V TRP3 - V S S
- v P - V D G 2 - |V T6 | ) 2 -
V DD - V TRP3
t2- = CL2 W L
7 5 i
L7 W5 5
t-2
Slew rate =
isource/sink
CLi
I7
Page VI.3-6
+5V
10
10
M3
40
10
C L1=0.3pF
M4
vDO
vN
M1
M2
20
10
vP
CL2=
10pF
I7 =40A
I5=20A
t = t1 + t2
( v DO (t 0 ) - V TRP2)
t1 =
CL1 ,
I5
vDO(t0) = 5 because vP = -1V
M6
I6
-5V
2I7
KP'( W6/L6)
2.40
= 2.58 V VTRP2 = 5 - 2.58 = 2.42 V
8.4
0.3pF
= 38.7ns
t1 = (5 - 2.42)
20A
CL2
CL2
t2 = v O (t 0 ) - 0
=5
I 6 - I 7
I 6 - I 7
KP6' W6
I6 = 2 L V D D - V D O (min) - VT6 2
6
[VDO(min) is an optimistic assumption based on vDS2 0]
VGS6 = 1 +
I5
= -1.77
KN.2
8.10-6
2
2 (4)(5 - (-1.77) -1) = 533 A
10 pF
t2 = 5 (533 - 40) A = 101 ns
I6 =
t = t1 + t2 139 ns
Second order consideration: Charging of Csb of M1 and M2
vO
Page VI.3-7
M3
M4
(6)
vN
M1
3v
2.42v
M8
20
10
M2
M6
CL1
40
10
(9)
vP
vO
CL2
10
10
M7
-5V
1v
vP
0v
tprop=167 ns
20
10
V(9)
Actual
-1 v
V(6)
-1.54v
Approx.
-3 v
-5 v
0ns
50ns
100ns
150ns
Time
200ns
250ns
300ns
Page VI.3-8
vin
-
+
gm1
-
gm2
R1
C1
R2
C2
vout
-
vout(s)
A o p1 p2
=
vin(s)
( s + p 1) ( s + p 2)
1
p1 = R C
1 1
1
p2 =
R2C2
Ao = gm1gm2R1R2
I5 = 20A R1 = g
1
1
=
40A(.03) = 833K
ds6 + g ds7
1
p2 =
(10pF)(833K) = 120Krps
I7 = 40A R2 = g
Page VI.3-9
M4
M6
M1
vN
M2
I8
M5
M8
vP
vO
M7
VSS
Key Relationships for Design:
i D = (v G S - V T ) 2
2
or
v DS (sat) =
Also,
gm =
where
KW
= L
2I D
2iD(sat)
iD (sat) = 2 [vDS(sat)]2
Page VI.3-10
2ID
3. Knowing the second stage current and minimum device size for M6,
calculate the second stage gain.
A2 =
-g m6
g ds6 + g ds7
4. Calculate the required first stage gain from A2 and gain specifications.
5. Determine the current in the first stage based upon proper mirroring
and minimum values for M6 and M7. Verify that Pdiss is met.
6. Calculate the device size of M1 from A1 and I DS1.
A1 = g
-g m1
ds1 + g ds3
and
gm1 =
2K'W/L
IDS1
7. Design minimum device size for M5 based on negative CMR requirement using the following (IDS1 = 0.5IDS5):
vG1(min) = VSS + VDS5 +
where VDS5 =
IDS5
1 + VT1(max)
2IDS5
5 = VDS5(sat)
IDS5
3 - VTO3 (max) + VT1
Page VI.3-11
Avo > 66 dB
Pdiss < 10 mW
VDD = 10 V
CL = 2 pF
VSS = 0 V
tprop < 1 s
K'W
Recall that = L
CMR = 4-6 V
Output swing is VDD - 2V and VSS + 2V
1). For t prop << 1 s choose slew rate at 100 V/s
dvOUT
. -12
. -6
dt = ( 2 10 ) ( 100 10 ) = 200 A
I7 = CL
2I7
7 =
W7
2(200A)
L7 > 5.88
17.0A/V 2( W7/L7)
M6:
2V > vDS6(sat) =
2( IOUT+I7)
=
6
-g m6
-1
=
3). A 2 = g
ds6 + g ds7
N + P
W6
2(400A)
L6 > 12.5
8.0A/V 2( W6/L6)
2KP'W6
I6L6 -10
Page VI.3-12
2KN'W1
W1
I
2 4 = 200
=
(
+
)A
[
1
4
1
]
I4L 1
L1
2KN'
(Good for noise)
8). S5 =
2I5
5 =
I5
1 - VT1(max)
(34)
-1 = 2.90 V
2(17.0)(200)
W5
2(34)
L > 0.48
(17)S5
5
I5
W5
34
S
=
(5.88)
=
1.0
7
I7
200
L5 = 1 . 0
Page VI.3-13
I5
3 - VTO3 (max) + VT1(min)
I5
34 A
= 2.76.10 -6
( 1 0 - 6 - 1 + 0 . 5) 2
W3 (2.76)(2)
W3 W 4
L =
=
0.69
8
L3 = L4 > 0.69
3
W4
(Previously showed L > 1.06 so no modification is necessary)
4
=
10). Summary
W
Wdrawn = (L - 1.6)
L
Design Ratios
W1
L1
W3
L3
W5
L5
W6
L6
W7
L7
W2
= L = 200
2
W4
= L = 1.0
4
= 1.0
= 12.5
= 5.88
W1
L1
W3
L3
W5
L5
W6
L6
W7
L7
Proper Mirroring
minimum geometry
and LD = 0.8m
W 2 1000
= L = 5
2
W4 5
= L =5
4
= 1.0
62.5
= 5
30
= 5
680
5
3.4 5
5
5
3.4 5
5
5
60
5
30
5
S6
S7
=
2
S4
S5
Page VII.4-1
MP3
MP4
MP12
MP13
MP8
MP6
MN25
MN1
MN2
vOUT
MN10
MN11
MN9
MN5
v1
v2
MN24
MN7
V SS
i1
1
gm13
+
i2
i2
gm2 v1
i1
rout
vout
-
where
R out (rds5gm11rds11)||((rds4||rds2)gm13rds13) =
=g
1
ds5gds11 (gds2+gds4)gds13
gm11 +
gm13
gm1 +gm2
vin
ds5gds11 (gds2+gds4)gds13
gm11 +
gm13
Page VII.4-2
i1
gm1 v2
gm12 gm2 v1
i2
C2
1
gm13
C3
i2
i1
rout
vout
-
where
C1 = C GS12 + C BS12 + C DG3 + C BD3
C2 = C GS13 + C BS13 + C DG4 + C BD4
and
where
1
3 =
routC3
Typical performanceW 1 W 2 W 11 W 13
ID1 = ID2 = 50A and ID3= I D4 = 100A,
L1 = L2 = L11 = L13
=1, assume C 3 0.5pF, and using the values of Table 3.1-2 gives:
gm1 = gm2 = gm11 =41.2S
gds5 = gds11 = 0.5S
gm13 = 28.3S
Page VII.4-3
BIAS
M1
M6
M8
M10
vO
M2
M3
M9
M11
M7
M4
M5
Performance (ISET = 50 A)
Rise time
= 100 ns into 50 pF
Fall time
Propagation delay = 1 s
Slew rate = 2.7 Volts/s
Loop Gain = 32,000
Comments
The inverter pair of M8-M9 and M10-M11 are for the purpose of
providing an output drive capability and minimizing the propagation delay.
Page VII.4-4
VDD
VDD
M6
M8
BIAS
VPB
M1
vO
-
+
M3
M2
M9
M4
VNB
M5
M7
VSS
Page VII.6-1
Comparator
threshold
Time
Comparator
output
VTRP+ comparator
output
Page VII.6-2
vB
vA
vOUT
VREFR2
R1 +R2
VOHR2
R1 +R2
R2
R1
+
V
- REF
VOL
vB
VOLR2
R1 +R2
Noninverting
vOUT
R2
vIN
VOH
R1
vA +
vB
VREF R1 +R2
R2
vOUT
vIN
+
- VREF
R1 V
R2 OL
VOL
R1 V
R2 OH
Page VII.6-3
M10 M11
M4
M8
M6
M1
M2
vO
BIAS
M5
M9
M7
VSS
1.0V
-600m
2.0V
3.0V
4.0V
5.0V
6.0V
-400m
-200m
0m
200m
400m
600m
Page VII.6-5
+
-
IDEAL
VOS
+
-
IDEAL
CAZ
VOS
VIN
+
VOS
+
VOS
IDEAL
+
0V
-
+
V
- OS
Page VII.6-6
vIN+
+
1
vIN-
VOS
+ CAZ
IDEAL
+
VOS
vOUT
Page VII.6-7
vOUT
vIN
+
CAZ
2
2
CAZ
-
vIN
vOUT
1
Page VII.6-1
C1
vIN
-
C2
C3
Cn
Latch
Q
Q
= input voltage
change
Latch
tn-1 )e-t/ ]
v out = An[1 - (1 + (n-1)!
v out = A(1-et/ )
A5
A4
A3
5
4
3
v out = A2[1 - (1 + t)e -t/ )]
2
v out = A(1-e-t/ )
n=1
A2
A
t3
tL
Time
Page VII.6-2
+
vIN
-
Latch
Q
Q
VDD
Q
FB
Reset
Q
FB
VB1
vIN
VB2
VSS
vIN-VOS
+ -
+
VOS
LATCH
Page VII.7-1
KW
2
2L (vGS-VT) ,
vDS(sat) =
2iD
K(W/L) , and gm =
2KWID
L