Professional Documents
Culture Documents
10.1
The Two-Stage CMOS Op Amp
"
"
Figure 10.1
!
!
Figure 10.1
! Cc
Figure 10.1
(W )6
(W )7
L =2
L
W
W
(
)
(
)
L4
L5
Figure 10.1
VICM (min)
or
VICM VSS + Vtn + VOV 3 Vtp
!
!
VICM (max)
or equivalently
VICM VDD VOV 5 Vtp VOV 1
!
Output Swing
Analog Integrated Circuit Lab
vO (max)
vO (min)
Voltage Gain
Analog Integrated Circuit Lab
Figure 10.2
10
Voltage Gain
Analog Integrated Circuit Lab
Rin =
!
o2
o2
o4
11
o4
Voltage Gain
Analog Integrated Circuit Lab
1
1
=
+
VOV 1 VA2 VA 4
! The second-stage transconductance Gm 2 is
2
Gm 2 = g m 6 =
!
2I D6
VOV 6
V
V
VA 6
, ro 7 = A7 = A7
I D6
ID7
I D6
12
Voltage Gain
Analog Integrated Circuit Lab
1
1
+
Av = A1 A2
= Gm1 R1Gm 2 R2
= g m1 (ro 2 ! ro 4 ) g m 6 (ro 6 ! ro 7 )
!
13
The CMRR of the two stage op amp is determined by the first stage,
so the CMRR is
CMRR = [ g m1 (r02 ! r04 )][ 2 g m3 RSS ]
'
14
Frequency Response
Analog Integrated Circuit Lab
!
!
CC + C gd 6 CC
15
Frequency Response
Analog Integrated Circuit Lab
1
2 R1Gm 2 R2CC
fP2
Gm 2
2 C2
fZ
Gm 2
2 CC
(1 + Gm 2 R2 )CC Gm 2 R2CC
f t = Av f P1
Gm1
2 CC
16
Frequency Response
ft must lower than f P 2 and f Z ,so the design must satisfy the two conditions:
Gm1 Gm 2
<
, Gm1 < Gm 2
CC
C2
! The uniform -20-db/decade gain rolloff obtained at frequencies f f P1
Figure 10.3
17
Phase Margin
Analog Integrated Circuit Lab
!
!
uniform -20db/decade
18
Phase Margin
Analog Integrated Circuit Lab
ft
)
fZ
Frequency Response
Analog Integrated Circuit Lab
!
!
!
= Gm 2Vi 2
1
R+
sCC
1
R)
! Thus, the new zero is at sZ = 1 CC (
Gm 2
20
Slew Rate
Analog Integrated Circuit Lab
!
!
3
21
Slew Rate
Analog Integrated Circuit Lab
I
CC
I
t
CC
Gm1
and Gm = g m1 = I VOV 1 , so SR = 2 ftVOV ,or SR = tVOV
2 CC
!
!
A =
PSRR + =
!
!
vdd
v
A = o
vss
ro 7
ro 6 + ro 7
23
Thus,
A
!
vo
ro 7
=
vss ro 6 + ro 7
Ad
= g m1 (ro 2 ! ro 4 ) g m 6 ro 6
A
2
24
Design Trade-offs
Analog Integrated Circuit Lab
!
!
!
!
The performance of the two-stage CMOS op-amp are primarily determined by two design parameters:
1. The length L used for the channel of each MOSFET.
2. The overdrive voltage VOV at which each transistor is operated.
A large L and correspondingly larger VA increases the gain, CMRR and PSRR.
Operating at a lower VOV can also increase these three parameters and the range ofVICM
and output swing.
Also, the offset is minimized by operating at a lower VOV .
However, the transition frequency fT requires the selection of a larger VOV , which
determines the high-frequency performance of the MOSFET,
fT =
gm
2 (C gs + C gd )
25
Design Trade-offs
!
26
Example 10.1
Let it be required to design the circuit to obtain a dc gain of 4000 V V .
Assume that the available fabrication technology is of the 0.5 m type for
which Vtn = Vtp = 0.5V , kn ' = 200 A V 2 , k p ' = 80 A V 2 , VAn ' = VAp ' = 20 V m 2
, and VDD = VSS = 1.65V. Use L = 1 m for all devices. Operate all devices at
the same VOV , in the range of 0.2V to 0.4V. Use I = 200 A, I D 6 = 0.5mA.
Specify the W L for all transistors. Also give the values realized for the
range of VICM , the maximum possible output swing, Rin and Ro .
Also determine the CMRR and PSRR. If C1 = 0.2pF and C2 = 0.8pF, find
the required values of CC and the series resistance R to place the transmission zero at s = and to obtain the highest possible ft consistent with
a phase margin of 75$. Evaluate the values obtained for ft and SR.
27
Solution
(a) W L ratio:
Av = g m1 (ro 2 ! ro 4 ) g m 6 (ro 6 ! ro 7 )
=
2I
2( I 2) 1 VA
1 V
D6 A
VOV
2 ( I 2) VOV 2 I D 6
VA 2
)
VOV
To obtain AV = 4000, given VA = 20V
400
4000 =
, VOV = 0.316V
2
VOV
To obtain the required ratio of Q1 and Q2 ,
=(
I D1 =
1 ' W
k p VOV 2
2 L 1
1
W
100 = 80 0.3162
2
L 1
25 m
W 25 m W
Thus, =
, =
L 1 1 m L 2 1 m
28
Solution
Analog Integrated Circuit Lab
(a) W L ratio:
For Q3 and Q4 ,
1
W
100 = 200 0.3162
2
L 3
To obtain
W W 10 m
= =
L 3 L 4 1 m
For Q5 ,
1
W
200 = 80 0.3162
2
L 3
W 50 m
Thus, =
L 5 1 m
29
Solution
Analog Integrated Circuit Lab
(a) W L ratio:
Since Q7 is required to conduct 500 A,its
(W L )
W
W 125 m
=
2.5
=
1 m
L 7
L 5
For Q6 ,
1
W
500 = 200 0.3162
2
L 6
W 50 m
Thus, =
L 6 1 m
Let's select I REF = 20 A,
W
W 5 m
Thus, = 0.1 =
L 8
L 5 1 m
30
Solution
Analog Integrated Circuit Lab
Solution
Analog Integrated Circuit Lab
Thus,
2( I 2) 1 VA
2( I 2) VA
CMRR =
2
VOV
2 ( I 2)
VOV
I
= 2(
VA 2
20 2
) = 2(
) = 8000
VOV
0.316
32
Solution
Analog Integrated Circuit Lab
2I
2( I 2) 1 VA
V
D6 A
VOV
2 ( I 2) VOV I D 6
= 2(
VA 2
20 2
) = 2(
) = 8000
VOV
0.316
Expressed in decibels,
33
Solution
Analog Integrated Circuit Lab
Thus, f P 2
2 I D 6 2 0.5
=
= 3.2 mA V
VOV 6 0.316
3.2 103
=
= 637MHz
12
2 0.8 10
1
1
=
= 316
Gm 2 3.2 103
34
Solution
Analog Integrated Circuit Lab
(e) For a phase margin of 75$ , the phase shift due to the second pole at f = ft
must be 15$ ,
That is , tan 1
ft
Gm1 = g m1 =
Thus,
The value of SR is
Gm1
, where
2 f t
2 100A
= 0.63mA V
0.316V
0.63 103
CC1 =
= 0.6pF
2 171103
SR = 2 171106 0.316
= 340 V s
35
10.2
The Folded-Cascode CMOS Op Amp
36
!
!
Figure 10.8
37
CL denotes the total capacitance at the output node (including internal transi
-stor capacitances).
Figure 10.8
38
Figure 10.9 is a more complete circuit for the CMOS folded-cascode op amp.
Q11 provides the constant current I utilized of biasing the differntial pair.
Figure 10.9
39
VDD VOV 9
BIAS 1
40
VICM
VICM (min)
Output Swing
Analog Integrated Circuit Lab
41
Voltage Gain
Analog Integrated Circuit Lab
! The
Ro is the parallel equivalent of the output resistance of the cascode amplifier and the
output resistance of the cascode mirror, thus
Ro = Ro 4 ! Ro 6
Ro 4 ( g m 4 ro 4 )(ro 2 ! ro10 )
Ro 6 g m 6 ro 6 ro8
So, Ro = [ ( g m 4 ro 4 )(ro 2 ! ro10 )] ! ( g m 6 ro 6 ro8 )
42
Figure 10.10
Voltage Gain
Analog Integrated Circuit Lab
! The
1
Gm
Ro
R
o
1 + Av Av
Thus,
Rof = 1 g m1
43
Frequency Response
!
This circuit has poles at the input, at the connection between the CS and CG
transistors, and at the output terminal.
Normally, the first two poles are at very high frequencies. CL is usually large,
and the pole at the output becomes dominant
fp =
1
2 CL Ro
f t = Gm Ro f p =
44
Gm
2 CL
Frequency Response
!
In other words, a heavier capacitive load decreases the bandwidth of the foldedcascode amplifier but does not impair its response.
45
Slew Rate
Analog Integrated Circuit Lab
5
3
IB I
ch is I B I .
!
IB
IB I
IB I
46
Slew Rate
Analog Integrated Circuit Lab
I
CL
The reason for selecting I B > I is to avoid turning off the current mirror completely;
if the current mirror turns off, the output distortion increases.
! Finally,
47
Example 10.2
Analog Integrated Circuit Lab
Consider a design of the folded-cascode op amp of Fig. 10.9 for which I = 200 A,
I B = 250 A, and VOV for all transistors is 0.25V. Assume that the fabrication process provides kn ' = 100 A V 2 , k p ' = 40 A V 2 , VA' = 20 V m. VDD = VSS = 2.5V,
and Vt = 0.75V. Let all transistors have L = 1m, and assume that CL = 5pF.
Find I D , g m , ro , W L for all transistors, the allowable range of VICM and output swing.
Determine the values of Av , ft , f P and SR.
What is the power dissipation of the op amp?
48
Solution
Analog Integrated Circuit Lab
ro =
VA
ID
20
,
ID
2 I Di
W
=
'
2
L i k VOV
W L
Q2
Q3
Q4
Q5
Q6
Q7
Q8
Q9
Q10
Q11
100 100 150 150 150 150 150 150 250 250 200
0.8
2.0
2.0
1.6
80
100
32
0.8
32
1.2
1.2
1.2
120 120 48
1.2
49
48
1.2
48
1.2
48
200 200 64
Solution
!
Ro 6 = 21.28M
Ro = Ro 4 ! Ro 6 = 6.4M
!
0.8 103
! ft =
= 25.5MHz
12
2 5 10
!
fp =
f t 25.5MHz
=
= 5kHz
Av
5120
50
Solution
Analog Integrated Circuit Lab
I
200 106
SR =
=
= 40 V s
CL
5 1012
The total current is 500A=0.5mA, and the total supply voltage is 5V,thus
PD = 5 0.5 = 2.5mW
51
We have found that the upper limit on the input common-mode range exceeds the
supply voltage VDD , the magnitude of lower limit is significantly lower than VSS .
The opposite situation occurs if the input differential amplifier is made up of PMOS.
It follows that an NMOS and a PMOS differential pair placed in parallel would
provide an input stage with a common-mode range that exceeds the power supply
voltage in both directions.
52
! Each
Vo = 2Gm RoVid
Figure 10.11
53
The cascode current mirror limits the negative swing to 2 VOV + Vt above VSS .
The cascode mirror reduces the voltage swing by Vt volts.
Figure 10.12
54
! To
permit the output voltage at the drain of Q3 to swing as low as 2VOV , we must
10.12(b) is the modified mirror circuit, which is known as the wide-swing current mirror.
! The
Thus the output voltage can go down to 2VOV with Q3 still in saturation.
! Also,
the voltage at the drain of Q1 is now VOV and thus Q1 is operating at the edge
of saturation.
55
10.3
The741 Op-Amp Circuit
56
Stage 3
Stage 1
Stage 2
57
Bias Circuit
Analog Integrated Circuit Lab
2.
58
Device Parameters
Analog Integrated Circuit Lab
For the standard npn and pnp, the following parameters will be used:
Q13 will be assume to be equivalent to two transistors, Q13 A and Q13 B , with parallel
base-emitter junctions and haveing following saturation currents:
I SA = 0.25 1014 A, I SB = 0.75 1014 A
Q14 and Q20 will be assumed to each have an area three times that of a standard device.
62
10.4
DC Analysis of the741
63
I REF =
64
Input-Stage Bias
Analog Integrated Circuit Lab
Fig. 10.14 is the Widlar current source that biases the input stage.
Assume 10 to be large, we have
VBE11 VBE10 = I C10 R4
I REF
= I C10 R4
I C10
= I S 11 , we can get I C10 = 19A.
VT ln
Assumed that I S 10
Figure 10.14
65
Input-Stage Bias
Analog Integrated Circuit Lab
! From symmetry,
I C1 = I C 2
IE3 = IE4 I
2I
IC 9 =
1+ 2 P
2 I I C10
Input-Stage Bias
Analog Integrated Circuit Lab
IC 6 IC 4 I
!
2I
VBE 6 + IR2
R3
I
IS
Substituing I S = 1014 A and I = 9.5A results in VBE 6 = 517mV.
VBE 6 = VT ln
I OS = I B1 I B 2
68
Second-Stage Bias
Analog Integrated Circuit Lab
VBE17 = VT ln
So, I C16 I E16 = I B17 +
I C17
= 618mV
IS
I E17 R8 + VBE17
R9
69
assume that P 1
Output-Stage Bias
Analog Integrated Circuit Lab
I C19
= 530mV
IS
= VBE18 + VBE19 = 588 + 530 = 1.118V
VBE19 = VT ln
VBB
VBB = VT ln
I C14
I
+ VT ln C 20
I S 14
I S 20
10.5
Small-Signal Analysis of the 741
71
vi
4re
V
25mV
re = T =
= 2.63k
I
9.5A
Rid = 4 ( N + 1) re
ie =
= 4 ( 200 + 1) 2.63
= 2.1M
!
Figure 10.18
io = 2 ie
Gm1
io
=
vi 2re
Figure 10.20
Figure 10.21
is Ro1 = 6.7M.
Figure 10.21 shows the equivalent circuit of the input stage.
73
Example10.3
We wish to find the input offset voltage resulting from a 2% mismatch between the
resistances R1 and R2 in Fig. 10.13.
74
Solution
Analog Integrated Circuit Lab
VBE 5 + IR = VBE 6 + ( I I )( R + R )
VBE 5 VBE 6 = I R I ( R + R )
VBE 5 VBE 6 Ire
I
R
=
I
R + R + re
Substituing R = 1k and re = 2.63k
shows that I = 5.5 103 I .
So, VOS
I 5.5 103 I
=
=
Gm1
Gm1
Example10.4
It is required to find the CMRR of the 741 input stage. Assume tha the circuit is balancde
except for mismatches in the current-mirror load that result in an error m in the mirror's
current-transfer ratio; that is, the ratio becomes(1- m ).
76
Solution
Analog Integrated Circuit Lab
Ro = Ro 9 ! Ro10
2i +
2i
vicm
Ro
vicm
2 Ro
io = mi
Gmcm
assume P 1
io
i
= m
vicm vicm
=
CMRR=
m
2 Ro
Gm1
= 2 g m1 Ro m
Gmcm
Input Resistance
Figure 10.24
ic17
vi 2
Figure 10.25?
Output Resistance
Ro 2 = ( Ro13 B ! Ro17 )
Ro13 B = ro13 B
For the 741 component value we obtain Ro13 B = 90.9k.
Figure 10.27
80
Figure 10.28
81
Small-Signal Model
vo 2 = Gm 2 Ro 2 vi 2
A2
vi 3
Rin 3
= Gm 2 Ro 2
vi 2
Rin 3 + Ro 2
Figure 10.29
! Ro 23 =
Ro 2
+ re 23
23 + 1
Rout =
Ro 23
+r
20 + 1 e 20
Figure 10.30
84
10.6
Gain, Frequency Response, and Slew
Rate of the 741
85
Small-Signal Gain
Analog Integrated Circuit Lab
vo vi 2 vo 2 vo
=
vi vi vi 2 vo 2
= Gm1 ( Ro1 ! Ri 2 )( Gm 2 Ro 2 ) Gvo 3
Thus,
Ao
RL
RL + Rout
vo
= 476.1 (526.5) 0.97
vi
= 243147 V V
107.7dB
86
Figure 10.31
Frequency Response
!
Rt = Ro1 ! Ri 2
= 6.7M ! 4M = 2.5M
Thus, the dominant pole f P is
1
fP =
= 4.1Hz
2 Cin Rt
f t = A0 f3dB
Figure 10.32
assume that all the other poles are at very high frequency.
A Simplified Model
Analog Integrated Circuit Lab
A( s )
Vo ( s ) Gm1
=
Vi ( s ) sCC
A( j ) =
Gm1
jCC
t =
Gm1
CC
t
1MHz
2
88
Figure 10.33
Slew Rate
Analog Integrated Circuit Lab
Figure 10.34
2I
t
CC
2I
CC
Figure 10.35
89
SR =
10.7
Modern Techniques for the Design of
BJT Op Amps
91
2V to 3V.
This is done for the following reasons:
1. Modern small-feature-size IC fabrication technologies require low power-supply voltages.
2. Compatibility must be achieved with other parts of the system that use low-voltage supplies.
3. Power dissipation must be minimized, especially for battery-operated equipment.
Figure 10.36
92
Since the positive input terminal is connected to ground, ground voltage has to be within
the allowable input common-mode range.
The input common-mode range should extend below the negative-supply rail.
Consider the unity-gain voltage follower.
The input common-mode voltage is equal to the signal vI .
Thus, the input common-mode range shoulde include also the positive supply rail.
Therefore, the input common-mode range is more than rail-to-rail operation.
93
Figure 10.37
Bias Design
Analog Integrated Circuit Lab
I
VBE 2 = VT ln
I
S2
IS 2
VBE1 VBE 2 = VT ln
I
S1
But,
I
V
I = T ln S 2
R2 I S 1
Figure 10.38
94
Bias Design
Analog Integrated Circuit Lab
Figure 10.39
95
vo
= g m1,2 RC
vid
=
VR
I 2
RC = C
VT
VT
97
! Consider the
99
ICM
Figure 10.42
100
! Figure 10.42
shows an input stage that achieves more than rail-to-rail input common-
mode range by utilizing a pnp differential pair (Q1 and Q2 ) and an npn differential
pair (Q3 and Q4 ) connected in parallel.
The dependence of the differential gain on the input common-mode VICM is usually
undesirable and can be reduced considerably by arranging that on of the two differential
pairs is turned off when the other one is active.
101
Example 10.5
It is required to find the input resistance and the voltage gain of the input stage shown
in Fig. 10.42. Let VICM 0.8V so that the Q3 Q4 pair is off. Assume that Q5 supplies
10 A, that each of Q7 to Q10 is biased at 10A, and that all four cascode transistors
are operating in the active mode. The iuput resistance of the second stage of the op amp
(not shown) is RL = 2M. The emitter-degeneration resistances are R7 = R8 = 20K,
and R9 = R10 = 30K. Recall that the device parameters are N = 40, P = 10, VAn = 30V,
VAp = 20V.
102
Solution
Analog Integrated Circuit Lab
Rid = 2r 1 = 2 P g m1
I C1 5 106
g m1 =
=
= 0.2 mA V
VT 25 103
2 10
Rid =
= 100k
0.2
i
Gm1 = c 7
vid 2
ro1 =
VAp
I C1
R7 = 20k
20V
= 4M
5A
ro 7 =
VAn 30V
=
= 3M
I C 7 10A
re 7
1
V
25mV
= T =
= 2.5k
g m 7 I C 7 10A
103
Solution
Analog Integrated Circuit Lab
v R7
ie 7 g m1 id
2
R
+
r
7
e7
v 20
v
= g m1 id
= 0.89 g m1 id
2 20 + 2.5
2
v
io ie 7 = 0.89 g m1 id
2
i
Gm1 o = 0.89 g m1 = 0.89 0.2 = 0.18 mA V
vid 2
R = Ro 9 ! Ro 7 ! ( RL 2 )
Ro 9 = ro 9 + ( Ro 9 ! r 9 )(1 + g m9 ro 9 )
ro 9 =
g m9
VAp
IC 9
20V
= 2M
10A
I C 9 10 106
=
=
= 0.4 mA V
VT 25 103
104
Solution
Analog Integrated Circuit Lab
! r 9 =
P
g m9
10
= 25k
0.4 mA V
= 12.9M
Ro 7 = ro 7 + ( R7 ! r 7 )(1 + g m 7 ro 7 )
gm7
I C 7 10 106
=
=
= 0.4 mA V
3
VT
25 10
r 7 =
N
gm7
R = 12.9 ! 23 ! 1 = 0.89M
v 2
Ad = od = Gm1 R1
vid 2
40
= 100k
0.4 mA V
For the cascode circuit in Fig. 10.42, the cascode transistors Q7 through Q10 must
operate in the active mode at all time.
However, the currents supplied by Q9 and Q10 will not be exactly equal to the
currents supplied by Q7 and Q8.
These changes in turn can cause one set of the current sources to saturate.
Thus, we need a circuit that detects the change in the dc or VCM of vO1 and vO 2 , and
adjusts the bias voltage on the bases of Q7 and Q8. , VB to restore current equality.
106
107
Thus, VE VCM .
The voltage VB is simply equal to VE
plus the voltage drop of diode D1.
Figure 10.45
108
Example 10.6
Consider the operation of the circuit in Fig. 10.44. Assume that VICM 0.8V and thus
the npn input pair (Fig. 10.42) is off. Hence I 3 = I 4 = 0. Also assume that only dc voltages are present and thus I1 = I 2 = 5A. Each of Q7 to Q10 is biased at 10A, VCC = 3V,
VBIAS 3 = VCC 1, R7 = R8 = 20k, R9 = R10 = 30k. Neglect base currents and neglect
the loading edffect of the CMF circuit on the ouput nodes of the cascode circuit. The
CMF circuit provides VB = VCM + 0.4.
(a) Determine the nominal values of VB and VCM . Does the value of VCM ensure
operation in the active mode for Q7 through Q10 ?
(b) If the CMF circuit were not present, what would be the change in vO1 and vO 2
as a result of a current mismatch I =0.3A between Q7 Q8 and Q9 Q10?
Use the output resistance values found in Example 10.5.
(c) Now, if the CMF circuit is connected, what change will it cause in VB to eliminate
the current mismatch I ? What is the corresponding change in VCM from its
nominal value?
109
Solution
Analog Integrated Circuit Lab
for Q7 Q8 to be active
That is,
VCM > 0.4V
VCM < VBIAS 3 + 0.6
Solution
(b) For I C 9 I C 7 = I C10 I C 8 = I
VCM = IRo1
Ro1 = Ro 7 ! Ro 9 = 23 ! 12.9 = 8.3M
Thus,
If VCM is positive,
VCM = 0.6 + 2.5 = 3.1V
which exceeds the 2.6V maximum allowed value before Q9 Q10 saturate.
If VCM is negative,
VCM = 0.6 2.5 = 1.9V
111
Solution
Analog Integrated Circuit Lab
(c)
I C 7 = I C 8 =
I =
VB
re 7 + R7
VB
re 7 + R7
VB = I ( re 7 + R7 )
Correspondingly
25mV
= 0.3A
+ 20k
10A
Thus, to restore the current equality, the change required in VB and VCM is only 6.75mV.
112
! Its
! However,
! Rather a
to drive Q p .
! Because of
114
Figure 10.48
115
I SN and I S 7
iN iP
I
vE = VT ln
iN + iP I SN I S 7
are the saturation currents of QN and Q7 .
iN
vE VT ln
I SN
I
+ VT ln
S7
i
vE = VT ln N
I SN
116
+ VEB 7
iP
vE = VT ln
+ VEB 6
I
SN
We can obtain a relationship between iN and iP :
I
vE = VREF = VT ln REF
I S 10
I REF
+
V
ln
T
S 11
2
I REF
I SN I S 7
iN iP
=
iN + iP I I S 10 I S 11
I
I
I
S 10 S 11
117
! Thus,
So,
1
1
I Q , and that for iP iN , iN I Q .
2
2
Thus the circuit not only establishes the quiescent current I Q but also sets the
minimum current in the inactive output transistor at
118
1
IQ .
2
Chapter 11
Filters and Tuned Amplifiers
11.1
Filter Transmission, Types, and
Specification
Filter Transmission
Analog Integrated Circuit Lab
!
!
Figure 11.1
Vo ( s)
Vi ( s)
Filter Transmission
Analog Integrated Circuit Lab
T ( j ) = T ( j ) e j ( )
!
G ( ) 20 log T ( j ) , dB
or, alternatively, in terms of the attenuation function.
A( ) 20 log T ( j ) , dB
Vo ( j ) = T ( j ) Vi ( j )
Figure 11.2
!
!
!
Figure 11.3
6
Figure 11.4
7
11.2
The Filter Transfer Function
aM ( s z1 )( s z2 ) ( s zM )
( s p1 )( s p2 ) ( s pN )
Transfer function: T ( s) =
a4 ( s 2 + l21 )( s 2 + l22 )
s 5 + b4 s 4 + b3 s 3 + b2 s 2 + b1s + b0
Figure 11.4
10
Transfer function:
a5 s ( s 2 + l21 )( s 2 + l22 )
T (s) =
s 6 + b5 s 5 + b0
s=0
11
a0
sN + bN 1s N 1 + + b0
12
11.3
Butterworth and Chebyshev
Filters
13
14
!
!
T ( j ) =
1+ 2 (
!
At = p ,
T ( j p ) =
2N
)
p
1
1+ 2
15
max
/10
!
!
17
18
!
!
19
Example 11.1
Analog Integrated Circuit Lab
= 0.588
20
Solution
3.find 0 from 0 = p (1/ )
Analog Integrated Circuit Lab
1/ N
4.find poles
p1 = 0 ( cos80! + j sin 80! ) = 0 (0.1736 + j 0.9848)
09
T ( s) =
( s + 0 )( s 2 + s1.87940 + 02 )( s 2 + s1.53210 + 02 )( s 2 + s0 + 02 )( s 2 + s0.34720 + 02 )
21
Figure 11.12 Sketches of the transmission characteristics of representative (a) even-order and (b) odd-order Chebyshev
filters.
22
!
!
!
2. T ( j )
1
2
1 + cosh [ N cosh ( / p )]
23
for p
Amax = 10log(1 + )
= 10 A
max
24
/10
2k 1
1
1
2k 1
1
1
) sinh( sinh 1 ) + j p cos(
) cosh( sinh 1 )
N 2
N
N 2
N
k = 1, 2,3, N
!
2 N 1 ( s p1 )( s p2 ) ( s pN )
where K is a constant equal to the required DC gain of the filter
25
Example 11.2
Analog Integrated Circuit Lab
max
/10
= 0.588
pk = p sin(
2k 1
1
1
2k 1
1
1
) sinh( sinh 1 ) + j p cos(
) cosh( sinh 1 )
N 2
N
N 2
N
26
Solution
Analog Integrated Circuit Lab
p1 , p5 = p (0.0895 j 0.9901)
p2 , p4 = p (0.2342 j 0.6119)
p5 = p (0.2895)
5. T ( s) =
T (s) =
K p2
2 N 1 ( s p1 )( s p2 ) ( s pN )
5p
where
p = 2 104 (rad / s)
27
11.4
First-Order and Second-Order
Filter Function
28
!
!
!
!
First order
Second order
29
First-Order Filter
Analog Integrated Circuit Lab
a1s + a0
T ( s) =
s + 0
!
a1
The numerator coefficients, a0 and a1 , determine the
type of filter (e.g., LP, HP, etc.)
!
A transmission zero at s = a0 / a1
The high frequency gain that approaches
30
First-Order Filter
Analog Integrated Circuit Lab
31
32
All-Pass Filter
!
!
a2 s 2 + a1s + a0
0 ) s + 2
0
Q
s2 + (
!
p1 , p2 = 0 j0 1 1 2
4Q
2Q
pole frequency 0
pole quality factor
The higher the value of Q the closer the poles
are to the j axis
! Q = ,yield sustained oscillation
!
34
!
!
!
35
36
37
38
!
!
!
1 , 2 = 0 1 + (1/ 4Q 2 ) 0 Thus, BW 2 1 = 0
Q
2Q
39
!
!
Regular notch, = 0
Low-pass notch, > n
High-pass notch, < n
Two transmission zeros are in the right side half of the s plane
The magnitude response is constant over all frequencies , called
flat gain
40
11.5
The Second-Order LCR Resonator
41
Figure 16.17 (a) The second-order parallel LCR resonator. (b, c) Two ways of exciting the resonator of (a) without changing its natural structure:
Figure 11.17
42
1
) + sC + ( 1 )
sL
R
s
C
= 2
s + s( 1 ) + ( 1 )
CR
LC
Q = 0CR
43
T ( s) =
44
V0 ( s )
Z 2 (s)
=
Vi ( s ) Z1 ( s ) + Z 2 ( s )
T ( s)
V0 ( s )
Z 2 (s)
Y1
=
=
Vi ( s ) Z1 ( s ) + Z 2 ( s ) Y1 + Y2
1/ sL
1/ LC
= 2
(1/ sL) + sC + (1/ R ) s + s (1/ CR ) + (1/ LC )
T ( s)
Transfer function
T ( s)
V0 ( s)
Z 2 (s)
Y1
=
=
Vi ( s) Z1 ( s ) + Z 2 ( s) Y1 + Y2
1/ R
s(1 / RC )
= 2
(1 / sL) + sC + (1 / R ) s + s(1 / CR) + (1 / LC )
Transfer function
s 2 + 02
T ( s) a2 2
s + s (0 / Q) + 02
46
! L1 , C1 are
selected so that
L1C1 = 1/ n2
L1 || L2 = L
s 2 + n2
T ( s ) a2 2
s + s(0 / Q) + 02
! s
V0
C1
=
Vi C1 + C2
,thus
a2 =
C1
C1 + C2
T(s )
48
s 2 s(0 / Q) + 02
T ( s) = 2
s + s(0 / Q) + 02
s 2(0 / Q)
s 2 + s(0 / Q) + 02
s (0 / Q)
= 0.5 2
s + s (0 / Q) + 02
T ( s) = 1
49
11.6
The Second-Order Active Filters Based on
Inductor Replacement
50
Z in V1 / I1 = sC4 R1R3 R5 / R2
L = C4 R1R3 R5 / R2
51
52
0 = 1/ LC6 = 1/ C4C6 R1 R3 R5 / R2
Q = 0C6 R6 = R6
!
C6 R2
C4 R1 R3 R5
Select C4 = C6 = C,R1 = R2 = R3 = R5 = R
0 = 1/ CR Q = R6 / R
Figure 11.21 (a) An LCR resonator. (b) An op ampRC resonator obtained by replacing the inductor L in the LCR
resonator of (a) with a simulated inductance realized by the Antoniou circuit of Fig. 11.20(a). (c) Implementation of the
buffer amplifier K.
53
Ks 2
T ( s) =
s2 + s
K = DC gain
1
R2
+
C6 R6 C4C6 R1R3 R5
K = High-frequency gain
54
T (s) =
Ks / C6 R6
1
R2
s2 + s
+
C6 R6 C4C6 R1R3 R5
K [ s 2 + ( R2 / C4C6 R1R3 R5 )]
T ( s) =
1
R2
s2 + s
+
C6 R6 C4C6 R1R3 R5
K = Center-frequency gain
55
C61
T ( s) = K
C61 + C62 s 2 + s
s 2 + ( R2 / C4C61 R1 R3 R5 )
1
R2
+
(C61 + C62 ) R6 C4 (C61 + C62 ) R1 R3 R5
n = 1/ C4C61R1 R3 R5 / R2
C61 + C62 = C6 + C
C61 = C (0 / n )2
0 = 1/ C4 (C61 + C62 ) R1 R3 R5 / R2
Q = R6
K = DC gain
C62 = C C61
C61 + C62 R2
C4
R1 R3 R5
56
s 2 + ( R 2 / C 4 C 6 R1 R 3 R 5 1 )
T (s) = K
R2
1
1
1
s2 + s
+
(
+
)
C 6 R 6 C 4 C 6 R1 R 3 R 5 1 R 52
n = 1 /
0 =
Q = R6
C 4 C 6 R1 R 3 R 5 1 / R 2
R2
1
1
(
+
)
C 4 C 6 R1 R 3 R 5 1 R 5 2
C 6 R2
1
1
(
+
)
C 4 R1 R 3 R 5 1 R 5 2
K = High-frequency gain
1
1
1
+
=
= 0C
R51 R52 R5
R51 = R5 (0 / n ) 2
R52 = R5 / [1 (n / 0 ) 2 ]
57
1 r2
R2
+
C6 R6 r1 C4C6 R1R3 R5
T (s) =
1
R2
s2 + s
+
C6 R6 C4C6 R1R3 R5
s2 s
r1 = r2 = r (arbitary)
Adjust r2 to make Qz = Q
11.7
Second-Order Active Filters Based on the
Two-Integrator-Loop Topology
59
Vhp
02
1 0
Vhp + ( Vhp ) + ( 2 Vhp ) = KVi
Q s
s
02
1 0
Vhp = KVi ( Vhp ) ( 2 Vhp )
Q s
s
!
60
Thp Vhp / Vi
(0 / s)Vhp
Vi
(02 / s 2 )Vhp
Vi
K 0 s
= Tbp ( s)
2
2
s + s(0 / Q) + 0
K 02
= 2
= Tlp ( s)
2
s + s(0 / Q) + 0
61
Circuit Implementation
Analog Integrated Circuit Lab
Vhp = Vi
!
R
R
R
R3
R2
(1 + f ) + Vbp
(1 + f ) V1 p f
R2 + R3
R1
R2 + R3
R1
R1
Rf
Rf
R f 02
R3
0
R2
Vhp =
(1 + )Vi +
(1 + )( Vhp )
( Vhp )
R2 + R3
R1
R2 + R3
R1
s
R1 s 2
2
1
0
0
! Equating Vhp = KVi ( Vhp ) ( Vhp )
Q s
s2
! R / R =1
f
1
! R3 / R2 = 2Q 1
! K = 2 (1 / Q )
62
Circuit Implementation
Vo = (
!
RF
R
R
R
R
R
Vhp + F Vbp + F V1 p ) = Vi ( F Thp + F Tbp + F Tlp )
RH
RB
RL
RH
RB
RL
63
An Alternative Two-Integrator-Loop
Biquad Circuit
!
!
!
!
64
Vo = (
V
V
V
rV 1
1
1
+ sC1 )( i ) + (
)Vo + [( 0 i ) i ](
)
R
sC
sQCR
sCR sCR2
R3 sCR
= p
= Vi ( s 2
Vo
=
Vi
s2 (
C1 s 1
r
1
+ (
)+ 2
)
C C R1 RR3 C R2 R
C1
1 1
r
1
)+s (
)+ 2
C
C R1 RR3 C RR2
1
1
s2 + s
+ 2 2
QCR C R
Figure 11.26 The TowThomas biquad with feed forward. The transfer function of Eq. (11.68) is realized by feeding
the input signal through appropriate components to the inputs of the three op amps. This circuit can realize all special
second-order functions. The design equations are given in Table 11.2.
65
!
!
!
!
HP
! Notch
(all types)
! AP
!
R3 =
66
11.8
Single-Amplifier Biquadratic Active Filters
67
!
!
68
!
!
!
!
69
Bridged-T networks
70
Bridged T network
Analog Integrated Circuit Lab
Vo ( s ) a2 s 2 + a1s + ao
T ( s) =
=
Vi ( s ) s 2 + 0 s + 2
0
Q
1
1 1
1
s2 + s + +
V
C1 C2 R3 C1C2 R3 R4
T (s) = a =
Vb
1
1
1
1
s2 + s
+
+
+
C1 R3 C2 R3 C1 R4 C1C2 R3 R4
1
C1C2 R3 R4
0 =
s2 + s
Q =[
+ 02 = s 2 + s (
C1C2 R3 R4
R3
71
1
1 1
1
+ ) +
C1 C2 R3 C1C2 R3 R4
1
1
+ )]1
C1 C 2
t ( s) =
I3 = (Vb-Va)/R3
Ia = 0
Va
I2 = I 3
I1
(Vb Va )
R3
and
I2 = I3 =
(Vb Va )
R3
V12
I4
Va
Vb
V12 = Va I 2 Z C 2 = Va
(Vb Va )
R3
Vb
1
1
= Va 1 +
sC2
sR3C2 sR3C2
1
Vb
1 +
I4 =
Vb
1
1 1
1
+
s 2 + s +
V
C1 C2 R3 C1C2 R3 R4
t (s) = a =
Vb
1
1
1
1
+
s 2 + s
+
+
C1R3 C2 R3 C1R4 C1C2 R3 R4
72
V12 Va
=
R4 R4
1
1
1 1
1
+ Vb +
= Va 1 +
R3 sR3 R4C2
R4 sR3C2 R3
Vb = I1Z C1 + V12
=
Va 1
1 1 Vb 1
1
+
1
+
+
1
Vb
+Va 1 +
sR3C2 sR3C2
Rearranging we get the final result for t(s)
Vo
s ( / C1 R4 )
=
1
Vi s 2 + s ( 1 + 1 ) 1 +
C1 C2 R3 C1C1 R3 R4
73
!
!
74
A
(1 t ) = 0
A +1
75
Feedback Loops
Feedback loop obtained by applying
the complementary transformation
to the loop
R3 = R R4 = R / 4Q 2
CR = 2Q / 0 C1 = C2 = C
76
Figure 11.34 (a) Feedback loop obtained by placing the bridged-T network of Fig. 11.28(b) in the negative feedback
path of an op amp.
(b) Equivalent feedback loop generated by applying the complementary transformation to the loop in (a). (c) A low-pass
filter obtained by injecting Vi through R1 into the loop in (b).
1
C3C4 R1 R2
Q =[
C3C4 R1 R2
C4
1
1
+ )]1
R1 R2
2
! R1 = R2 = R, C4 = C , and C3 = C / m yield m = 4Q
77
CR = 2Q / 0
11.9
Sensitivity
78
Sensitivity
Analog Integrated Circuit Lab
79
Example 11.3
Analog Integrated Circuit Lab
For the feedback loop of Fig. 11.29, find the sensitivities of 0 and Q relative to all
the passive components and the op-amp gain.Evalate these sensitivities for the design
considered in the preceding section for which C1 = C2
Solution
1
1
0 =
,to obtain SC10 = SC20 = S R30 = S R40 =
2
C1C2 R3 R4
1
1 1
For Q we have Q = [ C1C2 R3 R4 ( + ) ]1
C1 C2 R3
C1 C2
C1
1 C2
Q
apply the sensitivity definition to obtain SC1 =
C + C
2 C1
C2
1
2
80
1 Q
1
SR 4 =
2
2
Solution
we assume the op amp to have a finite gain A, the characteristic equation
for the loop becomes 1 + At ( s) = 0
using the design values C1 = C2 = C , R3 = R, R4 = R / 4Q 2 , and CR=2Q /0
s 2 + s (0 / Q) + 02
we get t ( s) = 2
s + s (0 / Q )(2Q 2 + 1) + 02
where the 0 and Q have the nominal or design values of the pole frequency and Q factor
the actual values are obtained by substituting for t ( s ) in 1 + A ( s ) = 0
s2 + s
2Q
(
Q
+ 1 + 02 + A( s 2 + s
+ 02 ) = 0
Q
assumeing the gain A to be real and dividing both sides by A + 1, we get
0 2Q 2
2
2
s + s 1 +
+ 0 = 0
Q
A +1
we see the actual pole frequency,0 a , and the pole Q, Qa , are
0 a = 0 , Qa =
Q
1 + 2Q 2 / ( A + 1)
81
Solution
Analog Integrated Circuit Lab
Thus SwA0 a = 0
S
Qa
A
A
2Q 2 / ( A + 1)
=
A + 1 1 + 2Q 2 / ( A + 1)
82
11.10
Switched-Capacitor Filter
83
Switched-Capacitor Filter
Analog Integrated Circuit Lab
84
85
86
CiVi
Tc
Req Tc / C1
we obtain an equivalent time constant for the integrator: Time constant =C2 Req = Tc
!
87
C2
C1
Practical Circuit
Figure 11.36 A pair of complementary stray-insensitive, switched-capacitor integrators. (a) Noninverting switchedcapacitor integrator.
(b) Inverting switched-capacitor integrator.
88
Figure 11.37 (a) A two-integrator-loop, active-RC biquad (b) its switched-capacitor counterpart.
89
Fig.11.37(a) yields
0 =
1
, replace R2 and R4 with their switched-capacitor equivalent values,that is
C1C2 R3 R4
R3 = Tc / C3 and R4 = Tc / C4
givs 0 =
1
Tc
C3 C4
select the C1 = C2 = C then C3 = C4 = KC
C2 C1
Q=
90
11.11
Tuned Amplifiers
91
Tuned Amplifiers
Analog Integrated Circuit Lab
The response is characterized by the center frequency 0 ,the 3-dB bandwidth B, and the skirt
selectivity, which is usually measured as the ratio of 30-dB bandwidth to the 3-dB bandwidth
The 3-dB bandwidth is less than 5% of 0 , this narrow-band property makes possible centain
approximations that can simplify the design process
92
Figure 11.39 The basic principle of tuned amplifiers is illustrated using a MOSFET with a tuned-circuit load. Bias
details are not shown.
93
Figure 11.39
Analog Integrated Circuit Lab
V0 =
g mVi
g mVi
=
YL
sC + 1/ R + 1/ sL
V0
g
s
= m 2
Vi
C s + s (1/ CR) + 1/ LC
0 = 1/ LC
a 3-dB bandwidth of B=
1
CR
a Q factor of Q 0 / B = 0CR
and a center-frequency gain of
V0 ( j0 )
Vi ( j0 )
= gm R
94
Example 11.4
Analog Integrated Circuit Lab
Solution
Center-frequency gain = 10 = 5 R.Thus R = 2 k .Since R = RL ! r0 then R = 2.5k
B = 2 104 =
1
CR
Thus
1
= 7958 pF
2 104 2 103
Since 0 = 2 106 = 1/ LC , we obtain
C=
L=
1
= 3.18 H
4 2 1012 7958 1012
95
Inductor Losses
The power loss in the inductor is usually
represented by a series resistance rs
L
Q0 0
rs
Typically, Q0 is in the range of 50 to 200
1
1
1
1 1 + j (1/ Q0 )
=
=
rs + j0 L j0 L 1 j (1/ Q0 ) j0 L 1 + 1/ Q02
1
1
For Q0 1 Y ( j0 ) =
1
+
j
j0 L
Q0
Y ( j0 ) =
Q0 =
Rp
0 L
R p = 0 LQ0
96
Use of Transformers
Figure 11.41 A tapped inductor is used as an impedance transformer to allow using a higher inductance, L, and a
smaller capacitance, C.
Figure 11.42 (a) The output of a tuned amplifier is coupled to the input of another amplifier via a tapped coil. (b) An
equivalent circuit.
Note that the use of a tapped coil increases the effective input impedance of the second amplifier stage.
97
Figure 11.43 A BJT amplifier with tuned circuits at the input and the output.
To avoid the loading effect of the bias resistors RB1 and RB 2 on the
input tuned circuit, a radio frequency choke(RFC) is inserted in series
with each resistor
The analysis and design is complicated by the Miller effect
due to capacitance C
98
99
Synchronous Tuning
! In the design of tuned with multiple tuned circuit, the question of the
frequency to which each circuit should be tuned arise for the overall
response to exhibit high passband flatness and skirt selectivity
B=
0
Q
21/ N 1
100
Stagger-Tuning
Analog Integrated Circuit Lab
0
0
1
1
j0 1
s+
+ j0 1
s+
2
2
2
Q
4
Q
2
Q
4
Q
T (s) =
, we can know p = s j0 so s = p + j0
p + 0 / 2Q
The result implies that the response of second-oder bandpass filter in the neighborhood of
Transfer function is T ( p ) =
its center fequency s = j0 is identical to the response of a first-order low-pass filter with
a pole at ( -0 / 2Q ) in the neighborhood of p = 0
Stagger-Tuning
102
Stagger-Tuning
Figure 11.48 Obtaining the poles and the frequency response of a fourth-order stagger-tuned, narrow-band bandpass
amplifier
by transforming a second-order low-pass, maximally flat response.
Stagger-Tuning
Analog Integrated Circuit Lab
01 = 0 +
B
2 2
, B1 =
B
, Q1
2
20
B
20
B
B
, B2 =
, Q2 =
B
2 2
2
Note that for the overall response to have a normalized center-frequency gain of unity,
02 = 0
104
Chapter 12
Signal Generators and
Waveform-Shaping Circuits
Introduction
Linear
oscillators
Signal
Generators
RC frequency-selective network
(operation in 10~100K Hz)
- Wien-Bridge Oscillator
- Phase-Shift Oscillator
- Quadrature Oscillator
- Active-Filter Tuned Oscillator
LC frequency-selective network
(operation in 100K~hundreds MHz)
-LC-Tuned Oscillators
-Crystal Oscillators
Nonlinear
oscillators
Bistable multivibrators
Astable multivibrators
Monostable multivibrators
2
A( s )
1 A( s) ( s )
Loop gain : L(s) A(s) ( s)
Af (s) =
characteristic equation :
1-A(s) ( s) = 0
For the curcuit to produce sustained oscillations at a frequency 0
the characteristic equation has to have roots at s = j 0 .
Thus 1-A(s) (s) should have a factor of the form s 2 + o 2 .
3
Oscillation Criterion
Barkhausen criterion : at 0 the phase of the loop gain should be zero and
Oscillation Criterion
Analog Integrated Circuit Lab
d
is large,
d
the resulting change in
If
0 will be small.
vO = R f R1 vI
vA = V
R3
R2
+ vO
R2 + R3
R2 + R3
vB = V
R5
R4
+ vO
R4 + R5
R4 + R5
vO = L - = -V
R3
R
- VD 1 + 3
R2
R2
vO = L + = V
R4
R
+ VD 1 + 4
R5
R5
Wien-Bridge Oscillator
R2 Z p
L ( s ) = 1 +
R1 Z p + Z s
1 + R2 R1
L (s) =
3 + sCR + 1 sCR
s = j
L ( j ) =
1 + R2 R1
3 + j (CR 1 CR)
To sustained oscillations:
(1) the phase will be zero
w0 =
1
CR
(To ensure that oscillation will start, one chooses R2 R1 slightly greater than 2)
10
R 3 = R 6 , and R 4 = R 5 .
11
Phase-Shift Oscillator
(1) It consists of a negative-gain amplifier(-K) with a three-section
(thrid-order) RC ladder network in the feedback.
(2) The circuit will oscillate at the frequency for which the phase shift
of the RC network is 180 .
(3) For oscillations to be sustained, the value of K should be equal to
the inverse of the magnitude of the RC network transfer function at
the frequency of oscillation.
13
Quadrature Oscillator
The quadrature oscillator is based on the two-intergrator loop.
Amplifier 1 is connented as an inverting as an inverting Miller intergrator
with a limiter in the feedback for amplitude control.
Amplifier 2 is connected as a noninverting integrator.
Norton
equivalent
15
Quadrature Oscillator
Analog Integrated Circuit Lab
For R f = 2 R :
The circuit will be a perfect noninverting integrator.
vO 2 =
1 t
vO1dt
0
CR
loop gain:
L (s)
Vo 2
1
= 2 2 2
Vx
sC R
oscillation frequency:
0 =
1
CR
16
Bandpass
filter
Limiter
17
This circuit uses a variation on the bandpass circuit based on the Antoniou
inductance-simulation circuit.
Limiter
18
LC-Tuned Oscillators
Analog Integrated Circuit Lab
0 = 1
Hartley oscillator:
CC
L 1 2
C1 + C2
0 = 1
19
( L1 + L2 ) C
At node C:
since V 0 :
s 3 LC1C2 + s 2 ( LC2 R ) + s ( C1 + C2 ) + g m + = 0
R
20
1 2 LC2
3
g m + R R + j ( C1 + C2 ) LC1C2 = 0
s j :
For oscillations to start, both real and imaginary parts must be zero.
0 = 1
CC
L 1 2
C1 + C2
C2 C1 =g m R
for oscillations to start :
g mR > C2 C1
21
22
Crystal Oscillators
A piezoelectric crystal, such as quartz, exhibits electromechanical-resonance
characteristics that are very stable(with time and temperature) and highly
selective(having very high Q factors).
Since the Q factor is very high, we may neglect the resistance r.
s 2 + (1 LC s )
1
1
Z ( s ) = 1 sC p +
= sC 2
sL
+
1
sC
s
p s + ( C p + Cs ) LCs C p
a series resonance at s
Cp Cs
s = 1
s j ,
LCs
a parallel resonance at p
p = 1
1 2 s2
Z ( j ) = j
C p 2 p2
23
CsC p
L
Cs + C p
Crystal Oscillators
CsC p
p = 1 L
> s = 1
C
+
C
p
s
But C p Cs p s
LCs
25
Bistable Multivibrator
Analog Integrated Circuit Lab
The bistable multivibrator has two stable stables. The circuit can remain in
either stable state indefinitely and moves to the other stable state only
when appropriately triggered.
R1 ( R1 + R 2 )
v + = vo
26
R1 ( R1 + R2 )
Assume that vo is L+ , and thus v+ = L+
(1) vI is increase from 0
vI > v + (i.e L+ ), vo goes negative.
op amp saturating in the negative direction
v o = L- , v + = L - and VTH = L +
(2) vI is decreased
vI < v + (i.e L ), vo goes positive.
op amp go to its positive-saturation state
v o = L + , v + = L + and VTL = L
27
vO = L+ , v+ = 0 and vI = VTL
VTL = L+ ( R1 R2 )
Superposition:
R2
R1
v+ = vI
+ vO
R1 + R2
R1 + R2
vO = L , v+ = 0 and vI = VTH
VTH = L ( R1 R2 )
29
30
31
saturation voltages of the op amp are by cascading the op amp with a limiter circuit.
L + = VZ1 + VD
L + = VZ + VD1 + VD2
L - = - ( VZ2 + VD )
L - = - VZ + VD3 + VD4
32
Astable Multivibrator
Analog Integrated Circuit Lab
33
For T2 :
v = L+ ( L+ L ) e t
v = L ( L L+ ) e t
= CR and v = L+ at t = T1
v = L at t = T2
T1 = ln
1 ( L L+ )
1
T2 = ln
1 ( L+ L )
1
For T1 = T2 , L+ = L
T = 2ln
34
1+
1-
35
For T1 :
VTH VTL L+
=
T1
CR
T1 = CR
VTH VTL
L+
T2 = CR
VTH VTL
L
For T2 :
VTH VTL L
=
T2
CR
VTH VTL
L+
Monostable Multivibrator
In some applications the need arises for a pulse of known height and width
generated in response to a trigger signal. The monostable multivibrator has
one stable stage in whcih it can remain indefinitely, and a quasi-stable stage
to which it can be triggered and in which it stays for a predetermined interval.
vB ( t ) = L ( L VD1 ) e t C1R3
and vB (T ) = L :
L = L ( L VD1 ) e T C1R3
Trigger
circuit
V L
T = C1 R3 ln D1
L L
For VD1 << L
recovery
period
37
1
T C1 R 3 ln
1-
For comparator 1:
2
VTH = VCC
3
For comparator 2 :
1
VTL = VCC
3
38
2
vc = VTH = Vcc at t = T :
3
T = CRln3 1.1CR
vc = Vcc (1 et RC )
39
40
In interval TH :
vC = VTH e t CRB
TL = CRB ln 2 0.69CR B
Period T of the square wave:
T = TH + TL = 0.69C ( RA + 2 RB )
Duty cycle of the square wave:
Duty cycle
TH
R + RB
= A
TH + TL R A + 2R B
42
43
R5
R4 + R5
vI is small:
Transfer characteristic : almost linear
(as a sine waveform is near its zero crossings)
vI is large:
Transfer characteristic : bend
(as a sine waveform approaches its peak)
44
vO = vI
The straight-line transfer characteristic
vo -v I almost pass through the origin.
suitable for small input signal.
vI < 0 : Diode : Off
vO = 0
The op amp will be operating in an open-loop
fashion and its output will be at the
negative saturation level.
it requires some time to back into its
linear region (limit the operation frequency)
45
An Alternative Circuit
Catching
diode
vI > 0 : D2 : On D1 : Off
vO = 0
vI 0 : D2 : Off D1 : On
vO = vI ( R1 = R2 )
46
Measuring AC Voltages
Analog Integrated Circuit Lab
v2 : mostly DC ; V2 =
VP R2 R4
R1 R3
First-order
low-pass filter
47
49
vI > 0 :
vI < 0 :
D2 : On ; D1 : Off
D1 : On ; D2 : Off
vO = v I
when R1 = R2
(virtual short )
vO = v I
vI > vO :
Diode : Off
vO follow vI
51
53
Chapter 13
Output Stages and
Power Amplifiers
Figure 13.1 Collector current waveforms for transistors operating in (a) class A,
(b) class B, (c) class AB, and (d) class C amplifier stages.
2
vO =vI vBE1
vO min = -IR L
Signal waveforms
VCEsat
of Fig.13.2
Figure 13.4 Maximum signal waveforms in the class A output stage of Fig. 13.2 under the condition
I = VCC /RL or, equivalently, RL = VCC /I. Note that the transistor saturation voltages have been neglected.
Example13.1
Analog Integrated Circuit Lab
= 100.
Sol:
( b ) For a sinusoidal output voltage of maximum possible amplitude ( i.e.,10-V peak ) , the instantaneous
power dissipation in Q1 will be as shown in Fig.13.4 ( d ) .Thus the average power dissipation in Q1
will be
1
1
PD1 = VC C I = 10 0.1 = 0.5W
2
2
For Q2 , the current is constant at I = 0.1A and the voltage at the collector will have an average value of
0V. Thus the average voltage across Q2 will be VC C and the average dissipation will be
PD 2 = I vCE
= I VC C = 10 0.1 = 1W
average
PL =
2
orms
V
=
RL
10 / 2
100
= 0.5W
Power-conversion efficiency
Analog Integrated Circuit Lab
Load power(PL )
Supply power(PS )
Assuming that the output voltage of Fig13.2 is a sinusoid with the peak value V!o , the
average load power will be
(
P =
V"o / 2
RL
2
"
1 VO
=
2 RL
PS = 2VC C I
Thus
2
!
1 VO
1 V!o V!o
=
=
4 IRLVC C 4 IRL VC C
V! O = VC C = IRL
10
Figure 13.6 Transfer characteristic for the class B output stage in Fig. 13.5.
11
12
Power-Conversion Efficiency
Analog Integrated Circuit Lab
2
1 V!o
PL =
2 RL
The average power drawn from each of the two power supplies is
1 V!o
PS + = PS =
V
RL C C
2 V!o
PS =
VC C
RL
1 V! 2
Thus the efficiency will be given by = o
2 RL
2 V!o
V!o
VC C =
RL
4 VC C
= 78.5% This value is larger than that obtained in the class A stage (25%)
2
1 VCC
The maximum average power available from a class B output stage is PL =
2 RL
13
Power Dissipation
2 V!o
1 V!
The average power dissipated in the class B stage is given by PD = PS PL =
V
RL C C 2 RL
When V!o
P D max
= VCC , PD max
2Vcc2
Vcc2
= 2
Thus, PDN max = PDP max = 2
RL
RL
14
Example13.2
It is required to designe a class B output stage to deliver an average power of 20W to an 8- load.
The power supply is to be selected such that VCC is about 5V greater than the peak output voltage.
This avoids transistor saturation and the associated nonlinear distortion, and allows for including
short-circuit protection circuitry. ( The latter will be discussed in Section13.8 ) Determine the supply
voltage required, the peak current grawn from each supply, the total supply power, and the powerconversion efficiency. Also determine the maximum power that each transistor must be able to dissipate safely.
Sol:
2
1 V!o
Since PL =
then V!o = 2 PL RL = 17.9V Therefore we select VCC = 23V.
2 RL
V!
The peak current drawn from each supply is I!o = o = 2.24A
Rl
1
Thus the average power drawn from each supply is PS + = PS = 2.24 23 = 16.4W
P
20
The power-conversion efficiency is = L =
= 61%
PS 32.8
The maximum power dissipated in each transistor is PDN max = PDP max
15
Vcc2
=
= 6.7W
RL
The crossover distortion of a class B output stage can be reduced substantially by employing a
high-gain op amp and overall negative feedback.
The original dead band 0.7V is reduced to 0.7 Ao volt, where Ao is the dc gain of the op amp.
Single-Supply Operation
Figure 13.10 Class B output stage operated with a single power supply.
17
iN = iP = IQ = I S eVBB /2VT
The value of VBB is selected to yield the
required quiescent current IQ.
18
Circuit Operation
When vI goes positive, VB of QN increases by the same amount and vo = vI +
19
VBB
vBEN
2
20
Output Resistance
Rout
VT
VT
= reN ! reP and reN = , reP =
iN
iP
Thus,
Rout =
VT VT
V
= T
iN i p iP + iN
21
Example13.3
In this example we explore the details of the transfer characteristic, vo versus vI , of class AB
circuit in Fig.13.11. For this purpose let VCC =15V,I Q =2mA, and RL = 100. Assume that QN
and QP are matched and have I S = 1013 A. First, determine the required value of the bias voltage
VBB . Then, find the transfer characteristic for vO in the range-10V to +10V.
Sol:
To determine the required value of VBB we use iN = iP = I Q = I S eVBB /2VT
with I Q
= 2mA and
Example13.3
It is also useful to find iP and vEBP as follows: iP = iN iL , vEBP = VT ln ( iP iS )
A similar process can be employed for negative vO . However, symmetry can be utilized, obviating
the need to repeat the calculations. The resualts obtained are displayed in the following table:
The table also provides value for the dc gain vO vI as well as the increemental gain vO vi at the
various values of vO . The incremental gain is computed as follows
vo
RL
=
vi
RL + Rout
where Rout is the small-signal output resistance of the amplifier, given by Rout
23
VT V
VT
T =
iN i
iP + iN
P
Example13.3
Analog Integrated Circuit Lab
The incremental gain is the slope of the voltage transfer characteristic, and the magnitude of its
variation over the range of vO is an indication of the linearity of output stage. Observe that for
0 vo 10V, the incremental gain changes from 0.94 to 1.00, about 6%. Also observe as vO
becomes positive, QN supplies more and more of iL and QP is correspondingly reduced. The
opposite happens for negative vO .
24
Figure 13.14
25
Example13.4
Consider the class AB output stage under the conditions that VCC = 15V, RL = 100, and the
output is sinusoidal with a maximum amplitude of 10V. Let Q N and Q P be matched with
I S = 10 13 A and =50. Assume that the biasing diodes have one-third the junction area of the
output devices. Find the value of I BIAS that quarantees a minimum of 1mA through the diodes
at all times. Determine the quiescent current and the quiescent power dissipation in the output
transistors ( i.e.,at vO = 0 ) . Also find VBB for vO = 0, +10V, and-10V.
Sol:
The maximum current through QN is approximately equal to iL max = 10 V 0.1k =100mA. Thus
the maximum base current in QN is approximately 2mA. To maintain a minimum of 1mA through
the diodes, we select I BIAS =3mA. The area ratio of 3 yields a quiescent current of 9 mA through Q N
and QP . The quiscent power dissipation is PDQ = 2 15 9 = 270mW
For vo = 0, the base current of QN is 9/51 0.18mA, leaving a current of 3-0.18=2.82mA to flow
1
through the diodes. Since the diodes have I S = 10 13 A, the voltage VBB will be
3
VBB = 2V
2.82mA
ln
I
= 1.26V
26
At vO = +10V, the current through the diodes will decrease to 1mA, resulting in VBB 1.21V.
At the other extreme of vO = 10V, QN will be conducting a very small current; thus its base
current will be negligibly small and all of I BIAS ( 3mA ) flows through the diodes, resulting in
VBB 1.26V
27
BE 1
R1
R2
R1
VBB = I R ( R1 + R2 ) = VBE1 1 +
And VBE1 = VT ln
I C1
; I C1 = I BIAS I R
I S1
28
29
Example 13.5
It is required to redesign the output stage of Example 13.4 utilizing a VBE multiplier
for biasing. Use a small-geometry transistor for Q1 with I S = 1014 A and design for a quiescent current I Q = 2mA.
Sol:
Since the peak positive current is100mA,the base current of QN can be as high as 2mA. we
shall therefore select I BIAS = 3mA, thus providing the multiplier with a minimum current of
1mA. Under quiescent conditions ( vO = 0 and i L = 0 ) the base current of QN can be neglected and all of I BIAS flows through the multiplier. We now must decide on how this current
will be almost cut off at the positive peak of vO . Therefore, we shall select I R = 0.5mA, leaving2.5mA for I C1. To obtain a quiescent current of 2mA in the output transistors,
VBB
2 1013
2.5 103
VBB = 2VT ln 13 = 1.19V and R1 + R2 =
= 2.38k . At IC1 = 2.5mA VBE1 = VT ln
= 0.66V
IR
10
1014
Thus R1 =
VBE1
= 1.32k R 2 = 2.38 1.32 = 1.06k
IR
30
31
1
1
2
For Q1 : I D1 =I BIAS = kn ' (W L )1 (VGS 1 Vtn ) ; For Q2 : I D 2 =I BIAS = k p ' (W L ) 2 VSG 2 Vtp
2
2
1
k n ' (W L )1
= Vtn + Vtp + 2 I Q
1
k n ' (W L ) n
1
k P ' (W
L)
2
1
k P ' (W
L)p
L)
k n ' (W L )1 + 1
k P ' (W L )2
k n ' (W L ) n + 1
k P ' (W
32
(W L )n
(W L )1
BIAS
BIAS
Vtn vOVN
(W L ) N ,P
33
BIAS
Vtn vOVN
When vO is negative
vO min = VSS + vOVN < VSS + VOV I + Vtp + vOVP
The allowable range of vO has been increased.
Disadvantage :
The output resistance
(R
out
= ron rop
is very high.
34
Fig.13.19
35
Output Resistance
feedback factor : =1
v
open-loop gain : A o = g mp ( rop RL )
vi
open-loop output resistance : Ro =RL rop
output resistance with feedback :
Rof =
Ro
1 + A
(R
rop )
1 + g mp ( rop RL )
output resistance :
Routp
1
1
1
1
1
=1
rop
; Routn
=
g mp
g mn
Rof RL g mp
Rout 1 ( g mp + g mn )
36
Figure 13.21
37
1
2
k p ' (W L ) p VOV2
1 2
kVOV
2
v v
v v
1
1
2
iDP = k [VOV ( vO vI )] = kVOV 2 1 O I = I Q 1 O I ,
2
2
VOV
VOV
v v v
v v
iDN = I Q 1 + O I iL = iDP iDN = 4 IQ O I = O
VOV
VOV
RL
VOV
VOV
VOV
vO = vI 1 +
,
usually
v
=
v
1
O
I
4 I R
4 I R
4 I Q RL
Q
L
Q
L
38
Gain error vO vI =
VOV
4 I Q RL
2IQ
vOV
gain error =
1
2 g m RL
39
Example 13.6
In this example we explore the design and operation of a class AB common-source
output stage of the type shown in Fig.13.19, required to opertae from a 2.5V power
supply to feed a load resistance RL = 100. The transistors available have Vtn = Vtp = 0.5V
and kn '=2.5k p '=250 A V 2 . The gain error is required to be less than 2.5% and I Q = 1mA.
Sol:
Gain error:
VOV
4 I Q RL
We are given the required maximum gain error of -0.025, I Q = 1mA, and RL = 100. In order
to keep low and also obtain as high a g m as possible [g m =2I Q /VOV ], we select VOV to be as
low as possible. Practically speaking, VOV is usually 0.1V to 0.2V. Selecting VOV = 0.1V results in which yields = 10 which is within the typically recommened range.
Fig13.22(a)shows the circuit in the quiescent state with the various dc voltages and currents indicated. The required(W/L)ratios of QN and QP can be found as follows:
40
Example 13.6
Analog Integrated Circuit Lab
IQ =
(W L ) P
1
2
k p ' (W L ) P VOV
(W L ) P = 2000; (W L ) n =
= 800
2
kn ' k p '
Thus QN and QP are very large transisitors, not an unusual situation in a high-power output stage.
To obtain the out put resistance at the quiescent point, we use Rout = 1 ( g mp + g mn ) where
g mp = g mn =
2IQ
VOV
Next we wish to determine the maximum and minimum allowed value of vO . Since the circuit is symmetrical, we need to consider only either the positive-output or negative-output case. For vO positive, QP conducts more of the output current iL . Evemtually, QN turns off and QP conducts all of iL .
To find the value of vO at which this occurs, note that QN turns off when the voltage at its gate drops
from the quiescent value of -1.9V to-2V, at which point vGSN = Vtn . An equal change of -0.1V appears
at the output of the top amplifier, as showns in Fig13.22(b). Analysis of the circuit in Fig.13.22(b) show
that: iL = iDP = 4mA; vO = iL RL = 0.4V
For vO > 0.4V, QP must conduct all the current iL . The situation at vO = vO max is illustrated in Fig.13.22(c).
Analysis of this circuit resulat after some straightforward but tedious manipulations, in
vO max 2.05V ; iL max = 20.5mA
41
Junction Temperature:
for silicon devices, TJ max is in the range of 150C to 200C.
Thermal Resistance:
TJ TA = JA PD ;
42
JA
43
Example 13.7
Analog Integrated Circuit Lab
TJ max TA0
= 62.5 C W
PD 0
(b) PD max =
TJ max TA
JA
= 1.6w
(c) TJ = TA + JA PD = 87.5C
44
Figure 13.25 The popular TO3 package for power transistors. The case is metal with a
diameter of about 2.2 cm; the outside dimension of the seating plane is about 4 cm. The
seating plane has two holes for screws to bolt it to a heat sink. The collector is electrically
connected to the case. Therefore an electrically insulating but thermally conducting spacer
is used between the transistor case and the heat sink.
45
JA = JC +CA
JA : thermeal resistance between junction and ambience.
JC : thermeal resistance between junction and case(package).
CA : thermeal resistance between case and ambience.
If a heat sink is ultilized CA = CS + SA
CS : thermeal resistance between case and sink.
SA : thermeal resistance between ambience and sink.
TJ TA = ( JC +CS + SA ) PD
46
PD max =
TJ max TC
JC
47
Example 13.8
A BJT is specified to have TJ max = 150C and to be capable of dissipation maximum
Analog Integrated Circuit Lab
TJ max TA
JA
= 1.6W
48
TJ max TA
JC
= 32W
49
51
Figure 13.30
52
53
54
Figure 13.33 A class AB output stage utilizing a Darlington npn and a compound pnp.
Biasing is obtained using a VBE multiplier.
55
Short-Circuit Protection
Figure 13.34
56
Thermal Shutdown
57
Figure 13.36 The simplified internal circuit of the LM380 IC power amplifier. (Courtesy National Semiconductor
Corporation.)
58
VS 3VEB
R1
R2
R2
1
1
Let I 3 = I 4 , and R1 = 2 R2 VO = VS + VEB
2
2
The DC feedback from the output to the emitter of Q4 , through R2 acting to stabilize
the debias voltage at approximately half the power-supply.
59
Small-siganl Analysing
60
Figure 13.38 Power dissipation (PD) versus output power (PL) for the LM380 with RL = 8. (Courtesy
National Semiconductor Corporation.)
61
Power Op Amps
62
The voltage swing of the output of bridge amplifier is twice the input.
64
Figure 13.42
65
Temperature Effects
66
67
Figure 13.44 A class AB amplifier with MOS output transistors and BJT drivers. Resistor
R3 is adjusted to provide temperature compensation while R1 is adjusted to yield the
desired value of quiescent current in the output transistors. Resistors RG are used to
suppress parasitic oscillations at high frequencies. Typically, RG = 100.
68
Transistor Q6 is placed in direct thermal contact with the output transistors.By appropriate
choice of the VBE multiplication factor of Q6 , the bias voltage VGG (between the gates of the
output transistors) can be made to decrease with temperature at the same rate as that of the
sum of the threshold voltages(VtN + VtP ) of the output MOSFETs.
VGG = (1 +
R3
R
)VBE 6 + (1 + 1 )VBE 5 4VBE
R4
R2
VGG
R V
= (1 + 3 ) BE 6
T
R4 T
Choosing R3 R4
The other VBE
and