Professional Documents
Culture Documents
6, DECEMBER
A Computer-Aided
BRUCE
A. WOOLEY,
357
1971
SUMMARYOF CHARACTERISTICS
wide-band monolithic amplifier, IEEE J. Solid-State Circuits,. vol. SC-1, Sept. 1966, pp. 1%28.
[21 L#l#rnan and H. Taub, Puke, Digital and Switching WaueNew York: McGraw-Hill, 1965, ch. 6.
[31 J. E. Solomon,. W. R. Davis, and P. L. Lee, (A self compensated monolithic operational amplifier with low input current
and high slew rate, ISSCC Digest Tech. Papers, Feb. 1969,
pp. 14-15.
[41 A. S. Grove, Physics and Technology oj Semrioaducto?Devices. New York: Wiley, 1967, ch. 11,
I.
INTRODUCTION
+
I
358
DEVICEMODELINGANDCHARACTERIZATION
In order to obtain an adequate simulation of the 741type amplifier, suitable device models and parameter
values must be established. In the ICL 8741 realization
of this amplifier there are six different types of transistors: the small n-p-n and lateral p-n-p devices that are
used throughout most of the amplifier, the large n-p-n
and substrate p-n-p output transistors Qll and QzO,the
double-collector lateral p-n-p Ql:;, and the double-emitter
substrate p-n-p Q2$. Plan views illustrating the planar
geometry of each of these devices are given in Fig. 2.
ivonlinear
For
DC
and
nonlinear
Time-Response
+15V
Q,2
Q8 14P
+ Ov
Q, Q2 _,N
Q9
14,U
)V +
0.732m
-IN
39k R5
Q3
Q4
+Vcc+14p
14p+124p
~
+27p -13.8
Q,,
Q,O
3k R4
BIASING
Q5
Q6
REFERENCE
NETWORK
Fig.
Q7 -13.7
STAGE
1.
INTERMEDIATE
OUTPUT
STAGE
STAGE
1,
B~
(b)
6.5
n
(c)
CollectorA
ot~
0.75
T
la,;,;,
?.25 9/16
~
1P!l
~3.25+
j/
(d)
f
(e)
Models
dc and time-response
analyses
of the
I_
(f)
359
Cj~
r-+---l
O.-O
Ic/Ipe.k
1.0
:
-a
<
-k.
0.05
(b)
(b)
b
c
~,k,
~.0316
0.1
0.316
1.0
3.16
Ic (mA)
,
EL
(c)
Ww
4
Fig, 4. Normalized plots of (a) (3, and (b) r, with collector current. (a) Curve a for small n-p-n, large n-p-n, and substrate
p-n-p; curve b for lateral p-n-p. (b) Curve a for large n-p-n;
curve b for small n-p-n and lateral p-n-p; curve c for specml
devices Q,, and Q=.
(d)
Fig. 3. Transistor circuit models. (a) Nonlinear model for an
n-p-n transistor. (b) Small-signal hybrid-r model for n-p-n
device. (c) Small-signal hybrid-~ model for lateral p-n-p demce. (d) Small-signal hybrid-m model for substrate p-n-p device.
(1)
c
=(+%+
c
(2)
360
IEEEJOURN.4L
OFSOLID-ST.lTE
CIRCUITS,
DECEMBER
1971
TABLE I
TRANSISTOR
Small
n-p-n
1.26
1.,(10-A)
290
03
072
2,5
670
300
1.45
1,15
405
0.65
().36
32
Do:..
I&~a,(mA)
B/(3,
B1
rblo 05mV(~)
rc(Q)
rl(lo+)
Tt(nS)
Tfr(ns)
C,.o(pF)
Cjc,(pF)
C,,(pF)
PARAMETERS
Large
n-p-n
Latera]
p-n-p
Substrate
p-n-p
0395
520
03
() 77
61
185
15
().97
O 76
243
28
1 55
7.8
3.15
95
(),11
().79
38
500
150
4,7
27 4
2540
0,1
1,05
17,6
130
0 11
0.90
4.8
80
156
4.47
26.5
2430
,-..
*. w)
~~
2rf, qI.
Linear
Svtall-Signal
(c,. + c,.)
+ T.c, c + !-,.
0.9
0.4
0.03
0.95
1.4
1000
(3)
Models
Double-Emitter
Substrate p-n;p
.4
0,79
94
0.015
0.85
1.5
1100
170
25
1.9
007
() 78
1,5
1600
120
3.1
27.4
220
0.1
(){)
80
2!,;
55
0.1
0.3
4,8
28
5.1
Q,,
Double%ollector
Lateral p-n-p
A
B
3.26
26.5
9550
1.1
2.4
4,8
00063
21
0.015
0.9
1.0
650
100
1,55
26,5
2120
;:
described by
(4)
r== Po
9.
(5)
1
7(J=
V9m
(6)
c. = C,e+ g.,,
(7)
Devices
I C13A
= 0.286
I C12
(8a)
I C13B
= 0,714.
I C12
(8b)
361
150100-
Quiescent
Conditions
50-
Frequency
Response
of the Complete
150,
:=
---
Simphfied
gom rmnver$[~n
i~Hz
100
Amplijier
---_..&
50
To obtain the gain-magnitude and gain-phase responses, it is necessary to simulate the complete openloop amplifier of Fig. 1 including a source resistance of
1 k~. For this simulation, the ac circuit model including
adequate device models is very large. To obtain the
desired type of results, three different CACA programs
were used. As a first step, the results of the dc state
simulation using BIAS-3 were used and the corresponding
transistor small-signal parameters were established using
the eauations of the last section. Next, the ac model of
the complete amplifier was analyzed using ROHRER,a
predecessor of the ac portion (for the sinusoidal steadystate response) of program CANCER[8].
This computer simulation verified that the smallsignal gain-frequency performance of the 741-type amplifier is determined almost entirely by the response of the
input and intermediate gain stages. This has been
checked by a simulation of only the first two stages. The
loading on the second stage by the output stage with the
short-circuit protection circuitry was modeled with a
simple emitterfollower. Further simulation runs also
showed that the p-n-p current sources could be replaced by ideal current sources in shunt with large resistors to represent the output resistance. Therefore,
further consideration of the gain-frequency response of
the amplifier could be carried out using a simplified
version of the first two stages as well as of the output
stage, This simplification permits additional analysis of
the response in terms of poles and zeros of the gain
function using program SLIC [7], which is limited at
present to circuits with less than 51 nodes. This program automatically establishes the small-signal device
model parameters after it has determined the dc state
of the circuit. The poles and zeros of the gain function
as well as the frequency response are then determined.
The results of the simulations using SLICare presented
in Fig. 5 and Table II. The amplifier provides a lowfrequency voltage gain of 111 dB. The unity gain frequency for the amplifier as compensated by the manu-
o%
\
,b,
,b,
Frequency
,b,
!
105 Hz
-D
(c)
Unbalance
Conversion
A double-to-single-ended
gain conversion network is
formed by transistors QE, Qe, and Q7 and provides approximately balanced transmission from either of the
two inputs to the single output. Consequently, excellent
common-mode rejection, which is the difference between
362
IB>!3E JOURNAL
TABLE
DOMINANT
OF SOLID-STATECIRCUITS,DECEMBER 1971
11
OPEN-LOOP
GAIN FUNCTION
Zeros
(10 rad/s)
Poles
(10 rad/s)
Imaginary
Real
No compensation
Compensation
5.4!2278E
6.82301E
1.69079E
1 69079E
2,40549E
1 14400E
+
+
+
+
02
01
01
01
01
02
2 75826E 05
2.40085E
+ 01
8.24558E
+ 01
3,9S042E + 01
1.44745E
+ 01
1,44745E
+ 01
0
0
0
0
(1
j 2248$)E
3.22489E
p-n-p
Performance
From the computer simulation runs, using various transistor parameter values, it has been established that the
second most dominant pole of the open-loop amplifier
gain function (compare Table II), can be associated to a
large extent with the charge storage in the lateral p-n-ps
Q, and Q,. The model element for these p-n-ps that procluces this effect is CT representing base charge.
The effect of replacing Q,+and Q~ in the 741 amplifier
z It must be noted that the common-mode rejection ratios of
Fig. 5(c) are for an idealized case where perfectly matched devices are assumed. For this aase, the rejection characteristics are
determined solely by the ci~cuit configuration. In an actual amplifier, common-mode rejection ratios somewhat lower than those
indicated are to be expected as a result of mismatched components in the input stage.
01
+ 01
Real
4,284 63E
6.556 20E
6 425 21E
8.386 15E
9.347 04E
1 368 57E
4,284 63E
6 844 47E
6.55620E
6 425 21E
8,386 15E
1.36857E
Imaginary
+
+
+
+
+
+
+
+
+
+
+
+
01
01
01
01
01
02
01
01
01
01
01
02
0
:
()
o
0
0
0
0
0
TABLE III
GAIN AND FREQUENCY
PERFORMANCE
OF 741 AMPLIFIERS
WITH LATERAL P-N+S AND IMPROVED P-N-IJS
Lateral
p-n-ps
Low-frequency gain
Uncom~ensated ~hase crossover
Compe&ation c~pacitance for
identical margin
Unity gain frequency when
compensated for identical
margin
111
dB
0.50 MHz
30
pF
1,54 MHz
HighPerformance
p-n-ps
112
dB
0.63 MHz
20
pF
2,87 MHz
Stage Characteristics
WOOLEY et Ill.:
EVALUATION OF THE 741 AMPLIFIER
v,,, (volts)
363
15-
and
QN. BY changing the transistor parameters of
QZOsuch that they are identical to those of QIA, the distortion in the output becomes less noticeable. A mismatch
in the ~. of Q14 and Q20 has little effect in distorting the
output .
(h.
10
5-
-70
. .
10
15
15
20
IV. SUMMARY
(a)
Computer simulated
Expenmentol
(b)
364
(1 + 12/i30)
(11)
= 1/(1
+ 2//30(30)
(12)
impedance of a diode-connected
transistor in series
with RI. This load imbalance for the composite input
pair results in a significant degradation in commonmode rejection, particularly at high frequencies. The
input voltage offset and offset drift of the amplifier may
also be adversely affected.
Intermediate
Gain Stage
of Fig. 1 consists of a
The intermediate gain stage
modified Darlington pair QIGQITwith an active current
source load Q13. Transistor QZZ is part of the output
short-circuit protection and is normally in an off condition.
The emitterfollower input QIGestablishes a high input
impedance for the stage, while the active load QR provides for large voltage gain. Because of large gain as well
as the fact that this stage is driven from the very high
output impedance of the first stage, unity-gain compensation of the complete amplifier can be achieved
through the Miller-effect multiplication of a relatively
small capacitance Cf.
The current level of the emitterfollower QIG is determined by the voltage drop across the 50-k~ resistance
Rq. The current-source transistor Ql~ is a dual-collector
lateral p-n-p biased from the biasing reference network.
Output
Stage
The output of the amplifier is a complementary classAB configuration with unity voltage gain. Transistors
QM and Q,,, establish a two-diode voltage drop between
the bases of the output emitterfollower QIA and QW
minimizing crossover distortion. The quiescent current
in Qls and Qlg is derived from collector A of the lateral
p-n-p current source Qls. The emitterfollower Q23 Provides a high load impedance for the intermediate gain
stage and isolates this stage from the amplifier output.
Q,, is a substrate p-n-p transistor with a second emitter
B added to prevent overdriving, and possible burnout,
of the intermediate gain stage. If QIG is driven hard
enough to saturate QIT, the emitter B of Qx, which is
normally inactive, begins to turn on and reduce the
input to Q16.
Short-circuit protection of the output stage during
positive output swings is provided by resistor R6 and
transistor Q15. If the output current through R6 becomes
excessive, Q15, which is normally off turns on and reduces
the drive to the base of the n-p-n output transistor Q14.
For negative output swings, short-circuit protection is
provided by R~, Qzl, QM, and Q2Z. When the negative
output current becomes excessive, the voltage drop across
R7 turns on the lateral p-n-p transistor Q~l. The resulting current in the diode-connected transistor QZ4 turns
on transistor QZZ and thereby reduces the input to QIG.
The voltage at the collector of Q17 rises correspondingly
and the drive to output substrate p-n-p Qz,o is reduced.
There is no internal provision in this amplifier to
ensure proper dc setup of the output stage. The external
wOOLEY
f?t
U/.
:EVALU.iTIONOF THE 741 AMPLIFIER
365
Reference
Network
_
.
~CC
VEE
(@n
412)
E5
(13)
R5
~~ln~
R, q
(14)
Clo
11 + 1C2 = lC [ 1 + BD +
1
2(1 +
l/B,,)
II
The experimental procedures used for device characterization are summarized briefly in this Appendix.
Current
Gain
Lollector
Saturation
Modulation
(16)
Resistance
(17)
Basewidth
l/qqm,
which is the reciprocal of the small-signal commonemitter output conductance. It is measured with a bridge,
with a low ac impedance at the transistor input. Measurements were made at several values of Ir to verify the
adequacy of (161.
(15)
where Icl and 102 are the collector currents of QI and Q2,
and 13p is the dc common-emitter current gain of the
lateral p-n-p devices, Q,3,Q~, QS, and QS. In establishing
(15) the base currents of Q, and Q2 are ignored with
regard to their contribution to the emitter currents of
these devices.
APPENDIX
r,, =
Capacitance
The junction capacitances are measured with an admittance bridge, such as the Boonton model 75C direct
capacitance bridge, at a frequency on the order of 100
.600
..
(18)
while for the basecollector and collector substrate junctions the measured capacitance is approximately
C,c(measured)
= C,. + 2Cpi.
(17)
and
CCS(measured) = C. + 3CP,.
Transit
(18)
Time
The authors are pleased to acknowledge the contribution of their present and former colleagues in the Integrated Circuit Group of the Electronics Research Laboratory, University of California, Berkeley. In particular,
A Five-Terminal
is to be noted that the initial phase of this investigation was the M.S. degree project of G. Pate] and
M. Schwartz. For the device characterization and for the
use and development of the simulators, we thank
I. Getreu, W. McCalla, and B. Rosario. The services of
the Computer Center of the Berkeley Campus of the
University are gratefully
acknowledged.
We thank
D. McDaniel for her assistance in the preparation of the
IC chip samples for device and stage characterizations.
Finally, we express our gratitude to D. Fullager and
Intersil, Inc., for making the 8741 IC available to us.
it
REFERENCES
[11 D. Fullager~ A new high performance monolithic operational amphtier, Fairchild Semiconductor, Tech. Paper.
[21 R. J. Widlar, A new monolithic operational amplifier design, National Semiconductorj Tech. Paper TP-2, June 1967.
[31 J. E. Solomon, W. R. Davis, and P. L. Lee, (A self-compensated monolithic operational amplifier with low input current and high slew rate, 1.969188CC Digest Tech. Papersl
vol. 12, pp. 1415.
[41 W. E. Hearn, Fast slewing monolithic operational amplifier,
IEEE J. Solid-State Circuits, vol. SC-6, Feb. 1971, pp. 20-24.
[51 W. R. Harden and M. J. Hellstrom, A triple-channel micronower operational amplifier. IEEE J. Solid-State Circuits.
iol. SC-~, Aug. 1970, Pp. 236240.
[61 R. J. Widlar. Desigm techniques for monolithic operational
amplifiers, IEEE J. Solid-State Circuits, vol. SC-4, Aug.
1969, pp. 184-191.
[71 T. E. Idleman, F. S. Jenkins, W. J. McCalla, and D. O.
Pederson, %LIC-A
simulator for linear integrated circuits, IEEE J. Solid-State Circuits, vol. SC-6, Aug. 1971,
PP. 188-203.
F. DAVIS