6

Operational amplifiers

6.1 Introduction
Integrated circuit technology made possible the operational amplifier (op amp) on a chip,
which is a high-gain, multitransistor voltage amplifier, encapsulated in a small chip and
costing less than a dollar. The first popular chip was the 741, a 24-transistor op amp. It
came out in the late 1960s and continues to be popular to this day. Op amps were originally
used in analog computers where, by changing the external circuitry which is connected to
the op amp, they could be used as adders, integrators, and differentiators. Today these ver-
satile amplifiers find applications in all fields, including signal processing, filtering,
switching circuitry, instrumentation, and so forth. Their use is only limited by the ingenu-
ity of the designer.
An op amp is designed to handle two inputs at the same time. The output signal will be
an amplified version of the difference between the two inputs. That is, if υ+ and υ
are the
two inputs, υout ¼ A(υ+
υ
), where A is the voltage gain of the op amp. If we set one of the
inputs to zero (grounded), υout will then just be an amplified version of the nonzero input
signal, that is, υout ¼ Aυ+ or υout ¼
Aυ
. The ability to amplify the difference between two
signals makes the op amp valuable in instrumentation, where unwanted interference sig-
nals common to both inputs are automatically canceled out.

6.2 OP AMP: An almost ideal amplifier

In Section 5.2, we learned that an ideal amplifier is characterized by infinite gain (A ¼ ∞),
infinite input resistance (Ri ¼ ∞), and zero output resistance (Ro ¼ 0). To these character-
istics we could also add that the bandwidth should be infinite–in other words, the ideal
amplifier amplifies all frequencies from DC to the highest frequencies with the same
gain—and furthermore that all of the preceding characteristics should remain stable with
temperature changes. The equivalent circuit of such an amplifier is given by Fig. 5.3.
The op amp matches these characteristics to a high degree. A popular model is the μA741,
a cheap but high-performance operational amplifier for which A ¼ 105, Ri ¼ 2 MΩ, and
Ro ¼ 75 Ω. Fig. 6.1a gives the equivalent circuit of an op amp enclosed by a triangle, which
is the traditional symbol for the op amp. It is basically a differential amplifier as it responds
to the difference in the voltages applied to the inverting and noninverting inputs, i.e.,
υo ¼ Aðυ +
υ
Þ ¼ Aυd (6.1)

Electronics and Communications for Scientists and Engineers. https://doi.org/10.1016/B978-0-12-817008-3.00006-1 231

© 2020 Elsevier Inc. All rights reserved.
232 Electronics and communications for scientists and engineers

Inverting input iF

– i1 RF
Ro P i≈0
–
ϑd + –
+ Ri –
A
Aϑd R1 ϑd ≈ 0
+ +
+
+ +
ϑ– + +
ϑo ϑi ϑo
ϑ–
Non inverting input

(a) (b)

iF

i1 RF
P
–
R1
ϑd +
–
+ Aϑd
+ +
ϑi
ϑo

(c)
FIG. 6.1 (a) The equivalent circuit (open-loop) of an op amp showing two input terminals, (b) The inverting amplifier
consists of an op amp and external circuitry (closed-loop). For simplicity, power supply connections to the op amp are
not shown, (c) The equivalent circuit of the inverting amplifier, using the ideal representation for the op amp.

where A is the open-loop (no feedback) gain of the op amp (typically 105 or larger). Note
that the polarity of υd is given by the + and
signs of the input terminals. For most circuit
calculations the ideal representation of an op amp (shown within the dashed triangle in
Fig. 6.1c) can be used; that is, in Fig. 6.1a the input resistance Ri is replaced by an open
circuit (Ri ¼ ∞) and the output resistance Ro is replaced by a short circuit (Ro ¼ 0).
An op amp, even though almost ideal in the first three characteristics (A, Ri, and Ro), has
a gain that varies substantially as the frequency of the input signal changes, and further-
more its characteristics are not very stable with temperature changes. Therefore, typical
use of an op amp is in a closed-loop configuration in which negative feedback is applied
which “kills” much of the open-loop gain of the op amp, resulting in a lesser but very stable
gain. The loop is closed by using external resistors to feed back some of the output voltage
to the input. An example is provided in the following section.

6.2.1 The inverting amplifier

The simplest example of a closed-loop configuration is the inverting amplifier shown in
Fig. 6.1b, in which an input signal υi is applied to the inverting (negative) terminal through
the R1 resistor and the noninverting input is grounded, while RF serves to “feedback” from
 Chapter 6 • Operational amplifiers 233

output to input. The circuit equivalent of this popular amplifier is shown in Fig. 6.1c. Using
Kirchoff’s voltage law we can readily write for the input loop υi ¼ i1 R1
υd, and for the
output loop υo ¼ Aυd ¼
iF RF
υd. Note that in the equivalent circuit we show a connec-
tion from the controlled voltage source Aυd to ground that is not present in Fig. 6.1b this
can be confusing. However, in Fig. 6.1b the op amp is represented simplistically by a tri-
angle symbol and such a connection would add little meaning. Besides, when analyzing a
circuit which has op amps, the equivalent circuit for each op amp should always be used.
The voltage gain Ar with feedback is obtained as follows: using the ideal representation
(Ri ¼ ∞, Ro ¼ 0) of the op amp shown in Fig. 6.1c, we conclude that since Ri, ¼ ∞,
i ¼ 0. Therefore all current through R1 passes through RF

i1 ¼ iF (6.2)

Furthermore, we note that υd ¼ υo/A  0 as A ! ∞, so the op amp input is in effect

shorted and the inverting terminal is effectively grounded.1 Therefore, the gain can be
stated as

υo
iF RF RF
Ar ¼ ¼ ¼
(6.3)
υi i1 R1 R1

This is a surprising result, as it says that the gain of the op amp with the external cir-
cuitry is equal to a ratio of the resistances of a feedback resistor and an input resistor. The
gain is thus determined by external resistors only and is independent of A as long as A is
high. Obviously, resistors are much less frequency and temperature sensitive than op
amps. Hence, we have obtained an amplifier whose gain is not as high as A of an op
amp alone, but it is very stable and has a constant value. The minus sign in Eq. (6.3) means
that there is a phase reversal in the output voltage.
Let us elaborate on the result obtained in Eq. (6.3). A consequence of the output voltage
fed back by resistor RF to the inverting terminal is to make that terminal (point P) a virtual
ground (suggested by the dashed ground symbol at P), which means that the voltage at
that terminal, υ
, is zero. What makes this result so remarkable is that a real short circuit
between ground and point P does not exist as then a substantial current would flow to
ground. Point P is at ground potential but no current flows as i ¼ 0 for the op amp, and
hence the label virtual ground. Point P, for the inverting configuration, remains a virtual
ground for any variations in the input signal υi.
In addition to being a virtual ground, point P, for the inverting configuration, is also
referred to as summing point. To sum several signals, for example, we can connect several
resistors to point P as shown in Fig. 6.4a. The sum of the signal currents in the input resis-
tors equals the current in RF, since no current to ground exists at point P. Because the addi-
tion appears to take place at point P, it is called a summing point.

1
Note that the output voltage υo is limited by the power supply voltage, which is typically between 5 and
15 V. Hence, the input voltage υd would be on the order of microvolts, given that A > 105.
234 Electronics and communications for scientists and engineers

n n n
Example 6.1
For the inverting amplifier of Fig. 6.1b, find the gain, input resistance, υd, υi, i, and iF. Assume
A ¼ 5  105, Ri ¼ 106 Ω, Ro ¼ 0, RF ¼ 100 kΩ, R1 ¼ 1 kΩ, and the power supply voltage is 5 V. In
order to obtain a better understanding, derive the op amp gain, Eq. (6.3), without initially set-
ting υd ¼ 0.
Since the amplifier input impedance Ri is large, the current i flowing into the op amp is neg-
ligible. Therefore, as in Eq. (6.2), current in R1 equals the current in RF, i.e.,
υi + υd
υd
υo
¼
R1 RF
Using Eq. (6.1), υd ¼ υo/A in the above expression, we obtain


1 RF RF
υo 1 + + ¼
υi
A AR1 R1
Given that the op amp gain is very large, we can take the limit as A ! ∞ and obtain

RF
υo ¼
υi
R1
which is the desired result of Eq. (6.3).
The gain of the op amp with external circuitry depends only on the external resistors that are
connected to the op amp. Hence, from Eq. (6.3), Ar ¼
RF/R1 ¼
100/1 ¼
100. The minus
sign in the gain expression implies that the amplified output signal is 180° out of phase with
the input signal.
The input resistance (it is the resistance that a source would see when connected to the υi
input terminals) is simply R ¼ υi/i1 ¼ R1 ¼ 1 kΩ. This is a rather low input impedance (in prac-
tice the term impedance is used when addressing any kind of input resistance), not suitable
when a high-impedance source is to be connected to the input terminals, as then only a small
portion of the source voltage would be driving the amplifier. In addition, a feeble source might
not be able to provide the large currents which a low input impedance requires. Ideally, a high-
impedance voltage source should work into an infinite-impedance voltage amplifier.
The maximum magnitude of the output voltage υo is limited by the supply or battery voltage
(it usually is less by about 2 V). Any figure displaying a load line such as Fig. 4.13b or Fig. 4.17c
shows that υo,max  υpowersupply. Assuming υo ¼
5 V, we obtain for the differential voltage
υd ¼
5 V/
A ¼ 5/5  105 ¼ 10 μV, which is a very small voltage and is usually neglected.
From Eq. (6.3), we have that υi ¼ (
R1/RF) υo ¼ (
1/100)(
5) ¼ 50 mV. The current i flowing
into the op amp input is given by i ¼ υd/Ri ¼ 10 μV/1 MΩ ¼ 0.00001 μA ¼ 10 pA. This current is
so small, that the approximation by zero is valid.
The feedback current iF ¼ i1. Hence, iF ¼ υo/RF ¼ 5 V/100 kΩ ¼ 50 μA.
In summary we state two basic rules for the analysis of op amp circuits. Rule 1 is assume the
two input terminals are at the same voltage, i.e., the differential voltage υd ¼ 0 (or equivalently
υ+ ¼ υ
). Rule 2 is assume no current flows into either input terminal (i ¼ 0). Practically any new
op amp configuration with feedback can be analyzed this way.
n n n
 Chapter 6 • Operational amplifiers 235

6.2.2 The noninverting amplifier

If we apply the input signal to the noninverting terminal and the feedback voltage to the
inverting terminal, as shown in Fig. 6.2a, the result is an amplifier with very high input
impedance, low output impedance, and no phase reversal. Such an amplifier is ideal as
it can be driven by a high-impedance source (it does not load the source) and can drive
a low-impedance load (the load does not affect the amplifier).
To realize the noninverting amplifier, we ground R1of Fig. 6.1b or C and apply the input
signal υi, at the noninverting terminal. The voltage i1R1 is then applied to the inverting ter-
minal as negative feedback voltage υ1. Again, we assume that the input current i is zero and
υd ¼ 0, and hence υi ¼ υ1. With op amp input current i equal to zero, the currents through
RF and R1 are the same, therefore
υo
υ1 υ1
¼
RF R1
The real gain of the op amp with external circuitry is then
υo υo i1 ðRF + R1 Þ RF
Ar ¼ ¼ ¼ ¼1+ (6.4)
υi υ1 i1 R1 R1
As in the case of the inverting amplifier, gain depends on the ratio of two external
resistances.
Whereas the input impedance of the inverting amplifier was low, equal to R1 in
Fig. 6.1b, the input impedance of the noninverting amplifier for practical purposes can
be approximated by infinity. Using Fig. 6.2a, the input impedance can be stated as
R0 i ¼ υi/i. Kirchoff’s voltage law for the input loop gives us

υi + υd + i1 R1 ¼ 0 (6.5)

where υd ¼ i Ri. Solving for i in Eq. (6.5), we have for the input impedance
υi υi Ri Ri
R0i ¼ ¼ ¼  ¼∞
i ðυi
i1 R1 Þ=Ri 1
i1 R1 =υi 1
1

i=0 i
+ +
υd A ϑd A
– –
RF
i = 0 i1
+ + + +
ϑi ϑo ϑi ϑo
+
ϑ1 i1 R1

(a) (b)
FIG. 6.2 (a) The noninverting op amp configuration, (b) A voltage follower for which υo ¼ υi.
236 Electronics and communications for scientists and engineers

where, using Eq. (6.5), we have approximated i1 R1  υi. A more careful analysis would
show that the input impedance of the noninverting amplifier is R0 i ≫ Ri, and since Ri, is
on the order of several megaohms for op amps, the approximation of R0 i by infinity is very
useful. The noninverting amplifier is thus ideally suited to amplify signals from feeble
(high-impedance) sources.2

6.3 Voltage followers and the unit gain buffer

A special case of the noninverting amplifier is the useful configuration known as the volt-
age follower, shown in Fig. 6.2b. It is obtained by letting RF ¼ 0 (short circuit) and R1 ¼ ∞
(open circuit) in the circuit of Fig. 6.2a. The gain of this configuration is then

υo RF
Ar ¼ ¼1+ ¼1 (6.6)
υi R1

which means that the output voltage follows the input voltage. The input impedance R0 i of
the voltage follower is found by applying Kirchoff’s voltage law to the circuit of Fig. 6.2b. This
results in υi ¼ υd + υo ¼ i Ri + A υd ¼ i Ri(1 + A)  i Ri A, where we have used that υd ¼ i Ri
and υo ¼ A υd. As the input impedance is the ratio of input voltage and input current, we
obtain

υi
R0i ¼ ¼ ARi (6.7)
i

The approximation of R0 i by infinity is again valid since typically for op amps Ri ¼ 1 MΩ

and A ¼ 106, giving R0 i ¼ 1012 Ω ¼ 1 TΩ. A million megaohms, for all practical purposes, is
an open circuit.
Such a device, when placed between source and load, protects the source from having
to deliver high load currents. It is then called a unit gain buffer. Buffering, or isolating the
source from its load, is frequently necessary since sources, such as transducers, sensors,
microphones, and tapeheads, do not produce any significant power. Connecting a low-
impedance load directly to a high-impedance source would drop the available voltage
to the load to negligible levels. On the other hand, a buffer with a practically infinite input
impedance and an almost zero output impedance would deliver the entire source voltage
to the load.3

2
Furthermore, the output impedance of a noninverting amplifier is much less than Ro, and Ro for most op
amps is about 100 Ω.
3
Of course, a noninverting amplifier provides buffering and amplification. However, there are many instances
when only buffering is required.
 Chapter 6 • Operational amplifiers 237

n n n
Example 6.2
A pickup device has an internal impedance of Rs ¼ 10 kΩ and produces a voltage of υs ¼ 2 V. If it
is to drive a chart recorder which can be represented by a load impedance of 500 Ω, find the
voltage and power available to the chart recorder. Repeat when a unity gain buffer is introduced
between source and load.
The load voltage is υL
υo ¼ υs0.5/(10 + 0.5) ¼ 0.095 V and power delivered to the load is
PL ¼ υ2L/RL ¼ (0.095)2/500 ¼ 18.14 μW.
Placing a buffer between source and load, as shown in Fig. 6.3a, and approximating the
buffer by an ideal one (Ri ¼ ∞, Ro ¼ 0), which means that the buffer voltage equals the source
voltage (no input current flows because Ri ¼ ∞) and the load voltage equals the buffer
voltage (because Ro ¼ 0), the load voltage is then υL ¼ 2 V and the power delivered is

+
Rs – Rs
+ ϑ +ϑ
+ RL s o RL
ϑo
ϑs ϑs

(a) (b)
FIG. 6.3 (a) A voltage follower is used as a buffer between a weak source and a load, isolating the source from
excessive current demands, (b) The buffer is shown as an ideal buffer which draws no current from the source and has
zero internal voltage drop.

PL ¼ υ2L/RL ¼ 22/500 ¼ 8 mW. The introduction of a buffer results in a voltage gain of 2 V/

0.095 V ¼ 21 and a power gain of 8 mW/18.14 μW ¼ 441, which demonstrates the effective-
ness of a buffer. It goes without saying that most pickup and transducer sources could not
deliver the necessary current or power to drive low-impedance loads without the use of
a buffer.
n n n

6.4 Summers, subtracters, and digital-to-analog converters

By applying several inputs to the inverting amplifier, as shown in Fig. 6.4a, we obtain the
summer amplifier, which gives an output voltage that is the sum of input voltages.
Such a configuration can be used, for example, to mix audio signals. As pointed out in
the paragraph preceding Example 6.1, point P is a current summing point as no current
can flow to ground. Therefore,
i1 + i2 + i3 ¼ iF (6.8)
υ1 υ2 υ3 υo
+ + ¼
(6.9)
R1 R2 R3 RF
238 Electronics and communications for scientists and engineers

R3
ϑ3
i3
R2 RF
ϑ2
i2 iF
R1
ϑ1 P
– ϑo
i1 A
+

(a)

ϑ3 ϑ2 ϑ1 ϑo

RF 0 0 0 0
ϑ2 0 0 5 1
R1 R 0 5 0 2
0 5 5 3
5 0 0 4
ϑ1 R1 A ϑo 5 0 5 5
R A 5 5 0 6
5 5 5 7
(b) (c)
FIG. 6.4 (a) A summing amplifier, (b) A subtracting amplifier, (c) A table giving 3-bit binary numbers (logic 0 and 1 are
represented by voltages 0 V and 5 V, respectively) and their decimal equivalents.

which equals
RF
υo ¼
ðυ1 + υ2 + υ3 Þ (6.10)
R1
when R1 ¼ R2 ¼ R3. We refer to this as an inverting adder. A noninverting adder, that is, one
giving Eq. (6.10) but without the minus sign, can be designed by connecting an inverter to
output υo. An inverter is an inverting amplifier like that of Fig. 6.1b with RF ¼ R1, resulting
in υo ¼
υi.
A subtracter is shown in Fig. 6.4b. Input υ1 passes through an inverter, which has a gain
of
1 (that is, the output signal is phase-shifted by 180° with respect to the input signal).
The resulting output voltage is thus the difference of the two input voltages, or
RF
υo ¼
ðυ2
υ1 Þ (6.11)
R1
By choosing RF ¼ R1, we would obtain a simple subtracter for which υo ¼ υ1
υ2.
A digital-to-analog converter (DAC) translates a binary number to an analog signal. For
example, the table in Fig. 6.4c gives the 3-bit binary numbers 000–111 and their equivalent
decimal numbers 0–7. The digital input signals are υ1
υ3 and the decimal-equivalent
voltage is υo. The binary digits 0 and 1 are represented by the input voltages 0 V and
5 V, respectively.
We can select the inverting summer, Fig. 6.4a, to perform the conversion. The output
voltage for such a summer, using Eq. (6.9), is


υ1 υ2 υ3
υo ¼
RF + + (6.12)
R1 R2 R3
 Chapter 6 • Operational amplifiers 239

If we choose (by trial and error) RF ¼ 8 kΩ, R3 ¼ 10 kΩ, R2 ¼ 20 kΩ, and R1 ¼ 40 kΩ, a
binary input signal 001 (υ3 ¼ 0, υ2 ¼ 0, υ1 ¼ 5 V) applied to the summer would give us
the following output voltage: using Eq. (6.12) we obtain υo ¼
8(5/40 + 0 + 0) ¼
l.4 Sim-
ilarly, for input 111 (υ3 ¼ 5, υ2 ¼ 5, υ1 ¼ 5), we would obtain υo ¼
8(5/40 + 5/20
+ 5/10) ¼
7. Hence, the inverting summer performs a digital to analog conversion. By
adding more inputs υ4, υ5, …, to the summer, we can handle larger binary numbers.

n n n
Example 6.3
A noninverting summer is shown in Fig. 6.5. Analyze the circuit and show that it performs the
mathematical summing operation. As the current into the noninverting terminal of the op amp
is vanishingly small, the node at υp acts as a current summing point for currents flowing in the
two R resistors. Therefore,

R
ϑ2
ϑP
R + + ϑo
ϑd A
ϑ1 –
RF

+ R1
ϑ′1

FIG. 6.5 A noninverting summer.

υ2
υp υ1
υp
+ ¼0 (6.13)
R R
which results in

υ2 + υ1 ¼ 2υp (6.14)

At the output terminal we have that the current through RF and R1 is the same, that is,
(υo
υ0 1/RF ¼ υ0 1/R1, because current into the inverting terminal of the op amp is vanishingly
small. Noting that voltage υ0 1 ¼ υp, because υd  0, we have
υo
υp υp
¼ (6.15)
RF R1
Substituting into this equation υp from Eq. (6.14) gives us the desired result

R1 + RF
υo ¼ ðυ1 + υ2 Þ (6.16)
2R1
which is that the circuit of Fig. 6.5 performs the mathematical summing operation. A note of
caution though: since the node at υp is not at zero volts, there can be interference (cross talk)

4
For simplicity, we are ignoring the minus sign on υo. This can be rectified by adding an inverter to the summer,
or by using a noninverting summing amplifier.
240 Electronics and communications for scientists and engineers

between the input signals. In that sense, the inverting summer is superior and hence more use-
ful since point P is a virtual ground in the inverting amplifier.
n n n

6.5 The differential amplifier

In many practical environments, such as instrumentation, biomedical applications, con-
trol systems, etc., we need to amplify the difference between two signals. In these situations,
the differential amplifier, typically an op amp configuration as shown in Fig. 6.6, is used. An
advantage of a differential amplifier is that any signal common to both inputs (common
mode signal) is canceled and does not appear in the output voltage υo. Undesirable signals
such as noise, AC hum, DC level, drift, etc., are canceled as they are picked up equally at both
inputs. On the other hand, even the smallest difference in the inputs (differential mode sig-
nal) is amplified. In situations where an interference signal is much stronger than a desired
signal in need of amplification, a difference amplifier is frequently the only solution. One
could conclude that an op amp by itself should serve that purpose, since after all its output is
given by Eq. (6.1) as υo ¼ A(υ+
υ
) ¼ A υd. Practically, this would result in a very poor,
unstable differential amplifier and, as discussed previously, operation in a closed-loop con-
figurations which introduces negative feedback is needed to give stable amplification.
Op amps, in the configuration of Fig. 6.6, are widely used differential amplifiers. To
analyze this circuit, recall that current into an ideal op amp is zero, which gives for the
loop equation at the inverting terminal
υ1
υo
υ1 ¼ ii R + υ
¼ R + υ
(6.17)
R + RF
At the noninverting terminal we can write either a loop equation or by voltage division
obtain
RF
υ + ¼ υ2 (6.18)
R + RF
Recall that for an ideal op amp, υd ¼ 0, or equivalently υ+ ¼ υ
; combining this with the
above two equations gives us the desired result for the differential amplifier
RF
υo ¼ ðυ2
υ1 Þ (6.19)
R

R RF

+ i1 i1
ϑ1
ϑ– –
A
ϑ+ + +
ϑ2 ϑo
+ i2 i2

R RF
FIG. 6.6 A differential amplifier, showing the external feedback circuitry.
 Chapter 6 • Operational amplifiers 241

Hence, the output voltage is proportional to the difference of the input signals ampli-
fied by the closed-loop gain RF/R. The gain is stable as it is independent of any op amp
parameters such as A, which can vary significantly with temperature, frequency, etc.,
and furthermore, A is specified by the manufacturer. On the other hand, the gain RF/R
depends only on the values of two resistors, and is therefore stable and can easily be chan-
ged by the circuit designer.

6.5.1 Practical versus ideal

The above output voltage, Eq. (6.19), is due to differential-mode signals, υ2
υ1, and
would be the only output voltage if the op amp were perfectly balanced, that is, ideal.
Unfortunately, practical op amps never are; they produce a small output voltage even
when the differential input signal is zero. For example, let us put identical signals υcm
on both inputs,5 where υcm is called the common-mode voltage as υcm contributes
equally to υ1 and υ2. We now have that υ1 ¼ υ2 and that υcm ¼ υ1 and υcm ¼ υ2. (By adding
υ1 and υ2, we can also express the common-mode signal as υcm ¼ (υ1 + υ2)/2, which is
just the average of the two input signals.) Any output voltage that is now obtained must
be an amplified version of the common-mode signal only and therefore can be stated
as υo ¼ Acmυcm, where Acm is the common-mode gain. Acm for an ideal differential
amplifier is obviously zero as an ideal amplifier responds only to the difference voltage
υ2
υ1 and not to the common-mode voltage υcm. The fact that a practical op amp
responds to the average signal level can create unexpected errors. For example, if
one signal is 10 μV and the second
10 μV, the output voltage will be different than
if the inputs had been 510 and 490 μV, even though the difference signal of 20 μV is
the same in both cases.
To answer the question of how good a practical differential amplifier is in comparison
with an ideal one, we define a common-mode rejection ratio, CMRR, as

Adm
CMRR ¼ (6.20)
Acm

where Adm is the differential-mode gain and from Eq. (6.19) is equal to RF/R. Thus, CMRR
for a practical op amp is a measure of quality—the larger the number, the better, as CMRR
for an ideal op amp is infinite.6 Because CMRR is normally a large number, we express it in
decibels as CMRR(dB) ¼ 20 logCMRR (typical values are 50–100 dB).

5
Equivalently, we could short the inputs together and apply a voltage υcm to the shorted input. This would
guarantee that the difference voltage υ2
υ1 equals zero at the input.
6
It should be noted in passing that CMRR for an op amp alone, Fig. 6.1a, is the same as that for the op amp
circuit with feedback, Fig. 6.6. The CMRR for the 741 op amp is given in the manufacturers spec sheet as 90 dB
and thus is also the CMRR of the differential amplifier of Fig. 6.6.
242 Electronics and communications for scientists and engineers

Since for a practical amplifier, the output voltage Eq. (6.19) has an additional compo-
nent due to the common input voltage υcm, we can write the total voltage by linear super-
position as
υ1 + υ2
υo ¼ ðυ2
υ1 ÞAdm + Acm

2  (6.21)
1 ðυ1 + υ2 Þ=2
¼ ðυ2
υ1 ÞAdm 1 +
CMRR υ2
υ1

where the voltage gain7 for the difference-mode signal is Adm and that for the common-
mode signal is Acm. The desired output is (υ2
υ1) Adm but the presence of a common-
mode input adds an error term, which is small if CMRR is large or if the common-mode
input υcm is small. Ideally, the two input signals should be of equal strength and 180° out of
phase. Then, the common-mode signal would be zero, the differential-mode signal a max-
imum, and the output voltage a linear magnification of the input signal, given by the first
term of Eq. (6.21), or by Eq. (6.19). Unfortunately, in most practical situations, it is just the
opposite. Typically in practice, a signal of only a few millivolts rides on a common-mode
signal of several volts as shown in the following example.

n n n
Example 6.4
Find the output voltage υo for a differential amplifier (Fig. 6.6) if the input voltages are υ1 ¼ 9 V
and υ2 ¼ 9.02 V. Use R ¼ 1 kΩ, RF ¼ 120 kΩ, and a 741 op amp for which specifications are given
as a CMRR of 90 dB and an open-loop gain of A ¼ 2  105.
Before proceeding with the solution, we should note that only a small 20 mV signal is riding
on a large common-mode voltage of 9 V. The closed-loop gain is obtained from Eq. (6.19) as
Adm ¼ RF/R ¼ 120. Expressing the 90 dB common-mode rejection ratio as a numerical value
gives CMRR ¼ 1090/20 ¼ 31,623. To obtain the common-mode gain, we use Eq. (6.20), which
gives Acm ¼ Adm/CMRR ¼ 120/31,623 ¼ 0.00379. The output voltage can now be given, using
Eq. (6.21), as
υ1 + υ2
υo ¼ ðυ2
υ1 ÞAdm + Acm
2
9 + 9:02
¼ ð9:02
9Þ120 + 0:00379
2
¼ 2:4 + 0:034 ¼ 2:434V
The deviation from the expected value of 2.4 V caused by the common-mode signal is small
in this case, only 0.034 V or 1.4%. This is due to the high CMRR value of the fine 741 op amp. A
lesser value than 90 dB for CMRR would have substantially increased the percentage deviation.
In some situations, the closed-loop common-mode gain Acm for a differential amplifier is
not known, but the open-loop (no feedback) gain A for the op amp and its CMRR is specified

7
For example, Adm can be measured by setting υ1 ¼
υ2 ¼ 0.5 V, which results in (υ2
υ1) ¼ 1 V and
υcm ¼ (υ1 + υ2)/2 ¼ 0. The measured output voltage would then be υo ¼ Adm, which is the gain for the difference
signal. Likewise, setting υ1 ¼ υi ¼ 1 V results in (υ2
υ1) ¼ 0 and υcm ¼ 1 V; consequently υo ¼ Acm and the output
voltage gives the gain for the common-mode signal.
 Chapter 6 • Operational amplifiers 243

by the manufacturer. In these situations it is important to know that the CMRR for an op amp
alone and an op amp with feedback is the same, as can be seen from an examination of
Eq. (6.21). Therefore, CMRR ¼ Adm/Acm ¼ A/Acm,openloop. This should allow calculation of
Acm as Adm can be obtained directly from the relationship Adm ¼ RF/R.
n n n

6.5.2 Interference signals

Amplifiers with high-impedance inputs are well suited to the amplification of feeble sig-
nals since the sources for these signals in turn have high impedance. For example, a mod-
erately high-impedance input might be 1 MΩ. If the source also has a 1 MΩ internal
impedance and generates a 10 mV signal, 5 mV will then be available across the amplifier
input terminals for amplification.
The difficulty with high-impedance inputs is that any stray electric or magnetic fields
can induce interfering voltages, at times very large, across the input terminals. For exam-
ple, electric noise fields due to power lines, lightning, sparking, switching circuits, motors,
arc welders, etc., can set up I Ro voltage drops across the input impedance. Equivalently,
the electric noise fields can be viewed as capacitive coupling between the noise source and
the amplifier input as suggested in Fig. 6.7a. By voltage-divider action, the interfering sig-
nal at the input will be VnoiseRin/(Rin + l /jωC).8 Even though the stray coupling capaci-
tances are very small (less than picofarad for C), if the input impedance Rin or if the
frequency ω of the interfering signal is high, much of the noise voltage Vn is developed
across the input terminals and will be amplified with the desired signal Vs. Using shielded
wires, with the outside conducting sheath connected to ground, can substantially reduce
capacitive pickup as then the interfering signal will be conducted harmlessly to ground.
A second source of interference are magnetic fields which are produced wherever
electric current flows. Thus, power lines and power machinery are a source of stray mag-
netic fields. Faraday’s law tells us that a time-varying magnetic field will induce a
voltage, also known as induced emf, in wires which surround the magnetic field. That
is, υind ¼
AdB/dt, where A is an area formed by wires that close around the magnetic field
B.9 Fig. 6.7b shows such an area A, in this case a long rectangular area formed by the parallel,
connecting wires between a source and the amplifier input. This area is threaded by a
changing magnetic field B, most likely caused by 60 Hz power lines. Such induced voltages
are unwelcome as they are amplified along with the desired signal, causing errors and dis-
tortion in the amplifier output. A remedy for this type of interference is to minimize the
enclosing area by running the wires parallel and close to each other. If this does not reduce

8
To be strictly correct, Rin should be replaced by Rin k Zs, since Rin is in parallel with Zs in Fig. 6.7a. Also, for
brevity, the phasor expression for the interfering signal is given. However, since a signal must be a real number, we
can obtain its magnitude by multiplying the phasor expression by its complex conjugate and taking the square
root as outlined in Chapter 2 (see Eq. 2.5 or Eq. 2.11).
9
If the magnetic field is sinusoidal with angular frequency ω, then the magnitude of υind ¼ AωB. The induced
voltage is thus proportional to the area looped by the wires, the frequency, and the strength of the magnetic field.
244 Electronics and communications for scientists and engineers

C
ϑnoise magn. field B

+ ϑ Zs Rin + ϑ Zs Rin
– s – s

(a) (b)
FIG. 6.7 (a) Small, stray capacitances exist between noise source and amplifier that couple an interfering field to the
input, (b) Time-varying magnetic fields due to noise sources (mainly power lines) induce a voltage in wires that close
around the magnetic field.

the magnetic pickup sufficiently, the wires should be twisted, which further reduces the loop
area and causes cancelation of induced voltage in successive twists.
In the above two paragraphs, we considered sources of interfering signals and
advanced some procedures to minimize the pickup of such signals on leads that connect
a transducer, probe, or any other signal source to the amplifier. If after all these proce-
dures, we still experience a strong noise at the source, a further decrease of such noise
may be accomplished by a differential amplifier (see the following example).

n n n
Example 6.5
With an electrocardiogram (EKG) amplifier, skin contacts are used to measure brain activity
and heart activity. These high-impedance sources produce faint signals, sometimes in the
microvolt range. It is not unusual to find that even after twisting and shielding the input leads,
a large 60 Hz voltage in addition to the desired, small signal still exists. Fig. 6.8a shows pickup
wires, which at one end are connected by skin contacts (electrodes) to the chest of a patient,
and at the other end to an EKG amplifier (an EKG amplifier is just a differential amplifier as
shown in Fig. 6.6). Since the leads are of equal length and otherwise symmetric, the noise

ϑ2
ϑ1 or ϑ2 ϑ1

+ + AC signal
ϑ2 ϑ1
DC level

0
t ϑcm
EKG signal
ϑ2 – ϑ1
Skin contact
electrodes t
(a) (b)
FIG. 6.8 (a) Leads for EGK signal pickup. The ground is usually a contact point at the lower extremities of a patient,
such as the right leg. (b) Signals υ1 and υ2 generated by the electrodes make up the difference signal. The common-
mode signal is primarily the interfering AC voltage and the interfering DC level.
 Chapter 6 • Operational amplifiers 245

pickup, which here is shown as a DC level and an AC voltage, is the same for both leads. Each
lead also picks up the desired EKG signal, which is shown as riding on the noise signal. The EKG
signal has a complicated, almost periodic waveform (it contains frequencies in the range from
0.05 to 100 Hz). For simplicity, let us model the EKG signal by a single spectral component, say,
sin 500t. The total signals, picked up by leads 1 and 2, could then have the form

υ1 ¼ 2 + 3 sin377t + 0:001 sin500t ffi υcm + 0:001 sin500t

and

υ2 ¼ 2 + 3 sin377t + 0:002 sin500t ffi υcm + 0:002 sin500t

and are shown in Fig. 6.8b. The EKG signal, even though much smaller then the 2 V DC level or
the 3 V AC signal (which is due to the 60 Hz power lines: ω ¼ 2π  60 Hz ¼ 377), is still 1 mV and
2 mV at leads 1 and 2, respectively. For an ideal differential amplifier, the output voltage would
simply be

υo ¼ Adm ðυ2
υ1 Þ ¼ Adm  0:001sin 500t

where Adm is the differential-mode gain, given by Eq. (6.19) as RF/R. We could try to optimize the
difference signal by moving the skin contact electrodes around and possibly find positions
where υ1 and υ2 are almost 180° out of phase. Then, the total signal at the differential amplifier
would be υ2
υ1 ¼ 0.003 sin 500t, a threefold increase in signal strength.
If the op amp in the differential amplifier circuit of Fig. 6.6 were a 741 and its CMRR had
deteriorated to 70 dB, the total output voltage would then be given by Eq. (6.21) as
υ1 + υ2
υo ¼ ðυ2
υ1 ÞAdm + Acm
 υ1 + υ2 
2
υo ¼ Adm ðυ2
υ1 Þ +
2CMRR
This is the result for a practical differential amplifier. It shows that the common-mode volt-
age, which includes the undesirable interference signal, is reduced by the numerical value
of CMRR—which, after converting from the logarithmic scale, is equal to CMRR ¼ 1070/20
¼3200. Thus at the input, the 2 V DC level is equivalent to a 2 V/3200 ¼ 0.6 mV and the
3 V AC signal is equivalent to only 3 V/3200 ¼ 0.9 mV, which is smaller than the 1 and 2 mV
EKG signals, but not by much. For example, using a 741 op amp with a CMRR of 80 dB would
reduce the interference signals by an additional factor of 10, and a new 741 which is rated at
90 dB would reduce it by 100, which for practical purposes leaves only the desirable EKG signal
to be amplified. Thus, at least for the data given in this example, an op amp with a 90 dB CMRR
compares favorably with an ideal op amp for which CMRR is infinite.
There are situations where the signals are so weak that an amplifier with essentially infinite
input impedance is needed. For such cases an improved version of the difference amplifier,
called an instrumentation amplifier, is used. Basically, such an amplifier preamplifies each
of the two signals before they reach the difference amplifier. The preamplifiers are noninverting
op amps which are characterized by very high input impedances. As signal loading is avoided
and CMRR is improved, instrumentation amplifiers also make superior EKG amplifiers.
n n n
246 Electronics and communications for scientists and engineers

6.6 Differentiating, integrating, and logarithmic amplifiers

Op amps can be used as accurate integrators and differentiators. Before proceeding, we
should note that a capacitor or inductor10 can perform integration and differentiation.
Ð
For example, for a capacitor we have that υ ¼ C1 i dt. The difficulty here is that we have
two variables, the current i and voltage υ. What is needed is a one-variable operation: if
the input is current i, the output should be the integral or derivative of i. Nevertheless,
by adding one more circuit element to a capacitor or inductor, we obtain a limited inte-
grator or differentiator in the form of a low-pass filter (Figs. 2.6 or 1.25a) or high-pass filter
(Fig. 2.7). Thus, if we consider a square wave of period Tas the input to a low-pass filter, the
output will be the integral of the input as long as the time constant RC of the filter is much
larger than the period T. Under these conditions (RC ≫ T), we have a useful integrator for
signals whose variations are much faster than the time constant of the filter. Similarly, the
output will be the derivative of the input, if the time constant of a high-pass filter is RC ≪ T.
A square-wave input would then result in an output consisting of sharp up and down
spikes corresponding to the leading and trailing edges of the square input pulses.
Fig. 6.9a shows an op amp integrator which is an inverting amplifier with the feedback
resistor replaced by a capacitor C. As for the inverting amplifier, υd  0, indicating that
point P is at ground potential. Since P is a virtual ground, the voltage across R1 is simply
Ð
υ1 and the voltage across C is υo ¼
C1 iC dt. Also, no current flows into the op amp, i  0,
which makes point P a current summing point. Therefore, iR1 ¼ iC or
ð
1
υo ¼
v1 dt (6.22)
R1 C

which states that the output voltage is the integral of the input voltage. Note that for the op
amp integrator, there is no restriction on the input signal as there was for the low-pass
filter integrator. Only the magnitudes of the output signal of the op amp integrator cannot
exceed the power supply voltage for the op amp.

+ +
C
+ P i≈0 + P R
– –
R1 ϑd ≈ 0 A C1 A
+ + + + + +
ϑ1 ϑo ϑ1 ϑo

(a) (b)
FIG. 6.9 (a) An op amp integrator, (b) An op amp differentiator.

10
Inductors are avoided in integrated circuits as they are bulky, heavy, and expensive in comparison to
capacitors.
 Chapter 6 • Operational amplifiers 247

n n n
Example 6.6
In situations where several voltages have to be added and their sum integrated we would ordi-
narily use a summing amplifier followed by an integrating amplifier. Show that in such situa-
tions both operations can be performed by a single op amp. The circuit that can combine both
operations in one is shown in Fig. 6.10. Current i is the sum of three currents flowing through
resistors R1, R2, and R3, that is, i ¼ i1 + i2 + i3. Since point P is a summing point, current i con-
tinues as current through the capacitor, or ic. Point P is also a virtual ground which gives for the
output voltage
R3 C
ϑ3
i3 iC
R2 i
ϑ2 P
–
i2 ϑo
+
R1
ϑ1
i1

FIG. 6.10 A summer-integrator which adds voltages υ1, υ2, υ3 and integrates the sum.

ð ð ð

1 1 1 υ1 υ2 υ3
υo ¼
iC dt ¼
i dt ¼
+ + dt
C ð
C C R1 R2 R3
(6.22a)
1 R1 R1
¼
υ1 + υ2 + υ3 dt
R1 C R2 R3
which is the desired result. For R1 ¼ R2 ¼ R3 we have a simple addition of the voltages in the
integrand. For different resistances we have a weighted addition and the output is the integral
of a weighted sum.
A differentiator is obtained by interchanging the capacitor with the input resistor, as shown
in Fig. 6.9b. Again equating the currents that flow through the resistor and capacitor, we obtain

dυ1 υo
C ¼

dt R
Solving for the output voltage, we obtain the desired result

dυ1
υo ¼
RC 1 (6.23)
dt
which states that output voltage is proportional to the derivative of the input voltage. The op
amp differentiator is not as stable as the integrator.11 It is seldom used in practice because it has
problems with noise and instabilities at high frequencies. The tendency to oscillate with some
signals can be decreased by placing a low-value resistor in series with C1, and by using the fast-
est op amp.

11
The differentiator (unlike the integrator which smoothes noise) magnifies noise spikes, because of the large
slopes that are present in noise voltage systems.
248 Electronics and communications for scientists and engineers

iC

P
–
R A
+ +
ϑi +
ϑo

FIG. 6.11 A logarithmic amplifier.

Nonlinear input–output relationships can be produced by placing a nonlinear element in

the feedback path of an op amp. A logarithmic amplifier gives an output voltage which is pro-
portional to the logarithm of the input voltage. Such an amplifier can handle a wide range of
input voltages (a logarithm compresses the scale of a variable) which would normally saturate a
linear amplifier. Essential to the operation of the logarithmic amplifier is to have a device in the
feedback loop with exponential characteristics. A diode or transistor has such characteristics.
Fig. 6.11 shows a grounded-base transistor for which we have, using Eq. (5.11),

ic ¼ αIo exp ð
eυo =kT Þ (6.24)

where υo is the emitter-to-base voltage. The collector current can be expressed in terms of the
input voltage as ic ¼ υi/R, recalling that P is a summing point and virtual ground. Taking the
logarithm of Eq. (6.24), we obtain

kT υi
υo ¼
ln (6.25)
e αIo R
which shows that the output voltage is the logarithm of the input voltage.
If we were to place a diode or transistor in the input loop and a resistor R in the feedback loop,
we would obtain an exponential or antilog amplifier. Now that we have a log and an antilog ampli-
fier we can construct a multiplier, that is, a circuit that can multiply two signals. If we sum the
output of two logarithmic amplifiers and then pass the output of the combination through an
antilog amplifier, the resulting signal will be proportional to the product of the two input signals.
n n n

6.7 Active RC filters

Filters can be used to remove, emphasize, or deemphasize certain frequencies of a signal.
This property can be used to reshape a signal or to block noise which is concentrated in a
specific band of frequencies. For example, the passive, low-pass filter of Fig. 2.6 allows
passage of DC and low frequencies, but attenuates the high frequencies. It is called pas-
sive, because the circuit includes only resistors, inductors, and capacitors. An active (con-
tains sources or active elements in addition to R, L, and C), low-pass filter, on the other
hand, can have the same frequency characteristics, but can also provide gain in the
pass-band. Of course, we could tag on an amplifier to a passive filter to obtain gain,
but it would require more circuit elements.
 Chapter 6 • Operational amplifiers 249

CF

gain Ar
RF
RF = 100
100 R1
–
A 70.7
+ R1 +
+ 50
ϑi ϑo
0
ω (log)
0.1 ωo ωo 10 ωo
(a) (b)
FIG. 6.12 (a) An op amp low-pass filter, (b) Frequency response of a low-pass filter which has a gain of 100 in the
pass-band.

An active, low-pass filter is shown in Fig. 6.12a. The filter elements are placed directly in
the feedback loop of an inverting amplifier. We can express the closed-loop gain as a func-
tion of frequency by simply replacing in Eq. (6.3) the resistances by impedances, that is,
RF
ZF 1 + jωCF RF
Ar ¼
¼
(6.26)
Z1 R1
where ZF is the parallel combination of RF and CF, and Z1 ¼ R1. Normally we are interested
in plotting the frequency response of the filter, which is obtained by taking the absolute
value of Eq. (6.26),
RF 1
j Ar j ¼  qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (6.27)
R1
1 + ðωCF RF Þ2

where the cutoff, corner, or half-power frequency is ωo ¼ 1 /RFCF. Note that ωo is the same
as that for the previously considered, low-pass filters of Eqs. (2.14) and (5.22). The above
expression consists of two terms: one is simply the gain of the inverting amplifier RF/R1,
and the other is a low-pass filter function, the same as that for a passive filter as previously
considered in Fig. 2.6. The advantages of an active filter are the ease with which the gain
and the bandwidth can be varied by controlling RF/R and the corner frequency ωo ¼ 1 /
RFCF. The frequency response of such a device for which the DC and low-frequency gain
is RF/R ¼ 100 and for which the low-pass bandwidth is determined by the half-power
frequency 1 /RFCF is plotted in Fig. 6.12b.

n n n
Example 6.7
Design a low-pass filter of the type shown in Fig. 6.12a with a closed-loop gain of 100 and a half-
power frequency of 500 Hz. The input impedance of the device is to be 1 kΩ.
As the negative input terminal of an inverting amplifier is a virtual ground, the input
impedance Zi ¼ R1 ¼ 1 kΩ. The low-frequency gain, which is given by Ar ¼ RF/R1, is specified
to be 100. Therefore, RF ¼ ArR1 ¼ 100  1 kΩ ¼ 100 kΩ. The remaining circuit
element which must be determined is the capacitance. Since the cutoff (half-power,
corner) frequency is specified and is given by fo ¼ ωo/2π ¼ 1/2πRFCF, we can solve for
250 Electronics and communications for scientists and engineers

RF gain Ar

– RF/R1
R1 C1 A
+ +
ϑi +
ϑo
0 ω (log)
0.1 ωo ωo 10 ωo
(a) (b)
FIG. 6.13 (a) An op amp high-pass filter, (b) Frequency response of a high-pass filter.

CF ¼ 1/2πfoRF ¼ l/2π  500  100 ¼ 3.2 μF. The gain-frequency plot is given by Fig. 6.12b with
angular half-power frequency ωo ¼ 2πfo ¼ 6.28  500 ¼ 3.14 krad/s.
An active high-pass filter is shown in Fig. 6.13a. The real gain is given again by Eq. (6.3) as

ZF RF
Ar ¼
¼
(6.28)
Z1 R1 + 1=jωC1
the absolute value of which is

RF =R1
jAr j ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (6.29)
1 + ð1=ωR1 C1 Þ2

This expression is plotted in Fig. 6.13b and shows the typical frequency response of a high-
pass filter with a half-power frequency ωo ¼ 1/R1C1 and a high-frequency gain of RF/R1.
Combining the responses of a low-pass and high-pass filter, we can construct a band-pass
filter. An active band-pass filter is an op amp with a parallel combination of RF and CF in the
feedback loop (same as in Fig. 6.12a) and a series combination of R1 and C1 in the input loop
(same as in Fig. 6.13a). If we choose the cutoff frequency 1/RFCF of the low-pass filter to be
larger than the cutoff frequency 1/R1C1 of the high-pass filter, only a band of frequencies equal
to the difference of cutoff frequencies will be passed.
n n n

6.8 Comparators and analog-to-digital converters

6.8.1 Comparator
An op amp by itself (open loop, no feedback) makes a very effective comparator, because
the gain A of an op amp is so high ( 105). Since the output voltage υo of an op amp cannot
exceed the power supply voltage,12 which, let us say, is 10 VDC, the output voltage will
swing from +10 V to
10 V for an input voltage swing of only +0.1 mV to
0.1 mV. Hence,
a change of only a fraction of a millivolt at the input will give a large output voltage.
Fig. 6.14a shows how an op amp is used to compare the magnitude of two signals.
Applying a reference signal to the inverting input and an unknown signal to the

12
Typically an op amp needs two power supplies, a positive voltage supply VBB and a negative voltage supply

VBB. If the op amp is to be operated from batteries, this can be a disadvantage as two separate batteries are
needed as shown in Fig. 6.14a.
 Chapter 6 • Operational amplifiers 251

ϑo
Ideal op amp
– 10V
VBB
+ Practical op amp
ϑd + A ϑo
– ϑ? (volts)
+ ϑ + – 0 4.999 5 5.001
– ? Vref + VBB RL
–10V VBB = 10V

(a) (b)
FIG. 6.14 (a) A comparator is an op amp in the open-loop mode with υo ¼ Aυd ¼ A(υ?
Vref). Two power supplies are
needed, which allows the output voltage υ0 to swing between + VBB and
VBB. (b) The unknown input voltage–
output voltage characteristics for a reference voltage of 5 V.

noninverting input, the output of the op amp is zero only when the two signals are equal.
When they are not equal, the output υo is either saturated at the supply voltage of +10 V
when the unknown signal υ? is υ? > Vref, or υo ¼
10 V when υ? < Vref. Fig. 6.14b shows the
transfer (input–output voltage) characteristics of an op amp when the reference signal is a
5 V battery and the supply voltage is 10 V. It shows that even a fraction of a millivolt in the
unknown signal above or below the 5 V reference signal will give a large indication of +10 V
or
10 V in the output.
When the reference voltage is chosen as Vref ¼ 0 (equivalent to grounding the inverting
input), we have a zero-crossing comparator. A very small input voltage (fraction of a mil-
livolt) will swing the output to 10 V and similarly a small negative voltage will swing the
output to
10 V. It is interesting to observe that the output of such a device is a square
wave for practically any periodic input signal. For example, a sinusoidal input, starting
at 0 V and increasing, will immediately saturate the output at 10 V. The output will stay
saturated at 10 V until the input sinusoid goes negative, at which time the output saturates
at
10 V and stays saturated until the sinusoid goes again positive. This repeats, generat-
ing a square-wave output voltage.

6.8.2 A/D converter

There are many ways to convert an analog signal to a finite number of ones and zeros.
Basically, the analog signal is sampled at evenly spaced, short-time intervals and the sam-
pled values converted to binary numbers. A simple (nonsampling) analog-to-digital con-
verter is shown in Fig. 6.15a which can convert a continuous signal to one of four values,
that is, to a 2-bit number. The four resistors, R, constitute a voltage divider across the ref-
erence voltage Vref and thus provide a fixed reference voltage for each comparator in the
stack of three comparators. For example, the top comparator has a reference voltage of
(3/4) Vref applied to the inverting input, the second (1/2)Vref, and the bottom one (l/4)Vref.
If the analog input signal υs < (l/4)Vref, the output of each comparator is 0 V. As υs
increases, first the bottom comparator saturates to the op amp power supply voltage
VBB, then the middle one, and finally the top one. Depending on the input signal, we
can have four different voltages at υ1, υ2, and υ3, which are shown in Fig. 6.15b as 0 V
252 Electronics and communications for scientists and engineers

Vref

R
– ϑ3
+ ϑ3 ϑ2 ϑ1 2-bit No.
R 0 0 0 0 0
– ϑ2 0 0 5 0 1
ϑS + 0 5 5 1 0
R 5 5 5 1 1
– ϑ1 (b)
+

(a)
FIG. 6.15 (a) A stack of three comparators can convert a continuous range of values of υs to one of four values.
(b) A table showing the coding of the outputs of υ1
υ3 and a 2-bit number.

and 5 V (computers use 5 V power supplies, i.e., VBB ¼ 5 V). These four values can be inter-
preted by a code converter or decoder (not shown in Fig. 6.15a) as 00, 01, 10, and 11. For an
8-bit digital output, we would need a stack of 255 (28–1) comparators.

6.9 The analog computer

Now that we have op amp integrators, differentiators, and summers, we can use these to
construct an analog computer, which can solve for us the differential equations that
describe the mechanical and electrical systems which we are designing for manufacturing
and production. For example, a system that all of us are familiar with is the suspension
system for an automobile, which in its elementary form is a weight mounted on top of
a spring and shock absorber, which in turn are connected to a wheel that moves along
a usually not-too-smooth of a road. Fig. 6.16 shows such a model in which a mass m
(the car) moves up and down along ym and a wheel follows a sinusoidally rippled road

ym
m

k D

yr

Rippled road surface

FIG. 6.16 A simple model for the suspension of an automobile. Mass m represents the car, k represents the springs,
and dashpot D the shock absorbers.
 Chapter 6 • Operational amplifiers 253

yr (t) ¼ Y sin ωt which applies an up and down force to the spring k and shock absorber D.
Summing forces on mass m

 fm + fD + fk ¼ 0
d 2 ym dy m dy r (6.30)
m +D
+ k ðym
yr Þ ¼ 0
dt 2 dt dt

where the vertical motion of the mass is given by fm ¼ ma ¼ md2ym/dt2, the force on the
mass transmitted by the shock absorber is modeled by a dashpot fD ¼ Ddy/dt (D is a veloc-
ity damping constant), and the force on the mass transmitted by the spring is fk ¼ ky,
where k is a linear spring constant. Let us move all terms which represent forces exerted
by the road on mass m to the right side of the equation
d 2 ym dy m dy
m 2
+D + ky m ¼ D r + ky r (6.31)
dt dt dt
By modeling the road by yr(t) ¼ Y sin ωt, where Y is the height of the ripples in the road
(road roughness), we can finally express the equation which determines the vertical
motion of a car traveling along a rippled road, that is,
d 2 ym dy m
m 2
+D + ky m ¼ DωY cos ωt + kY sinωt (6.32)
dt dt
A solution for ym could be used to design an optimum suspension system which would
give the smoothest ride over a rippled or a washboard road.
Analog computers were widely used to solve differential equations throughout the
1960s until the digital computer largely replaced their use. However, for specialized appli-
cations such as suspension design or vibrational analysis, analog computers accurately
mimic physical systems, can be easily changed for different parameters, and quickly give
answers that are readily displayed on an oscilloscope. In this example, the vertical dis-
placement ym in the suspension system will be represented by a voltage in the electrical
analog computer.
The first step is to solve for the highest-order derivative of the above differential equa-
tion, or
d 2 ym D dy m k DωY kY
2
¼

ym + cos ωt + sinωt (6.33)
dt m dt m m m
which is now in appropriate form for solution by double integration. We observe that the
sum of the four terms on the right represents d2ym/dt2. The next step is to arrange an inte-
grating amplifier which is preceded by a summing amplifier with four inputs to add the
right-side terms, as shown in Fig. 6.17. In Example 6.6 it was shown that the summing and
integrating operation can be combined such that only one op amp is needed (op amp 1 in
Fig. 6.17). The output of the first integrator13 is
dym/dt, which must be multiplied by D/m

13
In order to give the indicated integrator outputs, the RC constants must be chosen as unity according to the
integrator Equation (6.22).
254 Electronics and communications for scientists and engineers

kY sinωt R
m 4
DωY R
cosωt –dym
m 3
C dt
R C
2 ym
R –
(1) –
+ R (4) output ym
1 +

D
R
m Assume RC=1
R –
– (2)
(3) +
+ k
dym D dym
–D m dt m
m dt
–
(5)
k y +
–
m m
FIG. 6.17 Analog computer solution to Eq. (6.30), which describes a suspension system for an automobile.

(op amp 2) before it can be routed to the summer (after multiplication we need to change
sign by
1, which op amp 3 does). Another integration by op amp 4 gives us ym, which after
multiplying by
k/m is routed as the second input to the summer. The remaining two
terms in Eq. (6.33) are road condition inputs to the summer as the third input (DωY/m
cos ωt) and the fourth input (kY/m sin ωt). These are external inputs and can be changed
as necessary to simulate a variety of roads. The solution to the problem is ym; it is the
desired output which can be directed to a display device such as an oscilloscope. In an
analog computer such as this, the various parameters relating to the suspension system
can be varied and the new performance of the suspension readily evaluated.

n n n
Example 6.8
Design an efficient analog computer program for the solution of the familiar mass–spring–
damper problem in mechanics as shown in Fig. 6.18a. The vibrating mass mo is acted upon
by an external, sinusoidal force f ¼ F sin ωt N (newtons). The solution is obtained when the dis-
placement and velocity of mass mo are known. Assume mo ¼ 500 kg, the damping coefficient
D ¼ 2.5  104 N  s/m, and the spring constant k ¼ 106 N/m. Specify all resistor and capacitor
values in the op amp circuitry of the analog computer.
The equation describing the motion of the vibrating mass mo is the same as that for the auto-
mobile suspension given by Eq. (6.31) except for the right side of Eq. (6.31), which is the external
driving force and which in this example is given by f. Therefore, summing forces on mass mo, we
obtain in this case

d2 y dy
mo 2
+D + ky ¼
F sin ωt (6.34)
dt dt
 Chapter 6 • Operational amplifiers 255

f
y
mo

k D

(a)

1 μF dy
ϑ3
dt 1 μF
R3 = 20kΩ 1 MΩ
–y
ϑ2 1 MΩ
– 1 MΩ
R2 = 500Ω –
+ –
f + y
mo ϑ1 +
R = 1 MΩ
(b)
FIG. 6.18 (a) Vibrating mass supported by spring k and damper D and acted upon by an external force f. (b) Analog
computer simulation of vibrating mass. The external force input is f.

Rearranging in a form suitable for analog simulation, we have

d2 y dy F
2
¼
50
2000y
sin ωt
dt dt mo
where D/mo ¼ 50 and k/mo ¼ 2000. Integrating once gives
ð

dy F dy
¼
sin ωt + 2000y + 50 dt (6.35)
dt 500 dt
In order to find the resistance values in the summer-integrator op amp, let us refer to
Fig. 6.10 and compare Eqs. (6.35)–(6.22a). If we identify R1C with unity (R1 ¼ 106 Ω ¼ 1 MΩ,
C ¼ 10
6 F ¼ 1 μF) and identify υ1 with F sin ωt/500, υ2 with y, and υ3 with dy/dt, then
R1/R2 ¼ 2000 and R1/R3 ¼ 50. The values of the remaining resistors are therefore R2 ¼ 500 Ω
and R3 ¼ 20 kΩ. A circuit diagram for the analog computer simulation is given in Fig. 6.18b. Note
that this is a more efficient implementation than that outlined in Fig. 6.17, where for pedagogical14
reasons we have shown the summer-integrator with four identical input resistors R. Using the
same R0 s required two additional op amps (op amp 2 to change the scale and op amp 3 to invert).
As Eq. (6.22a) shows, we can use the input resistors of a summer-integrator to change the scale of
the inputs, which we have done in this example with the result that two op amps were eliminated.
n n n

14
It might be even better to first show the summer-integrator as two separate op amps, one for summing and
one for integrating, and only then proceed to the combined summer-integrator of Fig. 6.17 and finally to the most
efficient implementation shown in Fig. 6.18b.
256 Electronics and communications for scientists and engineers

6.10 Summary
An op amp is a high-gain, direct-coupled integrated circuit amplifier of about 20 transis-
tors on a tiny silicon chip. By itself, it is not particularly stable with a gain that varies with
frequency, dropping off to a negligible level at about a megahertz. However, applying neg-
ative feedback15 to the op amp, we obtain a moderate-gain amplifier that is very stable and
responds to frequencies much higher than a megahertz. Combining that with the charac-
teristics of high input impedance and low output impedance, we almost have an ideal
amplifier that is used by itself or as a building block in numerous applications in many
diverse fields. In analog design, they have surpassed transistors as the basic building
block. The versatility of this amazing chip allows it to be cascaded without loading prob-
lems, to be used as an oscillator, to be used as a differential amplifier that amplifies a dif-
ference signal but attenuates a common-mode (interfering) signal such as hum, and to be
used widely in “operation” applications which we started to explore in this chapter (com-
parators, D/A converters, integrators, etc.). The op amp fundamentals learned in this
chapter should be readily applicable to new situations encountered by designers in many
engineering fields. Because designing with them is so simple, even casual users should be
able to construct simple amplifiers and filters.
● There are two rules that govern the design of op amp circuits. The first is that the two
input terminals of an op amp in any circuit are at the same voltage, and the second is that
no current flows into either of the two input terminals. These were elaborated on in
Section 6.2, where the application of the two rules to the inverting amplifier showed that
point P, for practical purposes, is at ground level—a virtual ground—and also that point P
is a summing point—current flowing into the input resistor continues into the feedback
resistor. A thorough understanding of these rules can expedite the design of op amp
circuitry.
● There are some physical limitations of op amps, some of which are obvious and others
which are more subtle. An obvious one is the limit on the output voltage of an op amp
which cannot be larger than the power supply voltage.
The finite bandwidth is another limitation that can be overcome to some extent by
applying negative feedback—the more feedback, the larger the bandwidth, but the gain
is proportionately reduced. To quantify it, we introduce the gain-bandwidth product,
which is constant for any given op amp circuit. Hence reducing gain increases bandwidth
and vice versa. Initially op amps were limited to a gain-bandwidth product of 1 MHz,
which restricted their use to low-frequency applications, primarily audio. At present, since
the gain-bandwidth product has been pushed to 500 MHz, op amps are suited for video
applications as well.

15
Routing a portion of the output signal back to the input, such that it is 180° out of phase with the input signal,
is called negative feedback. It results in decreased gain but improved frequency characteristics, and reduces
waveform distortion. In positive feedback, a portion of the output signal that is fed back is in phase with the input
signal, usually leading to an unstable situation in which the circuit oscillates.
 Chapter 6 • Operational amplifiers 257

Another limitation of op amps is the presence of a small voltage, called the input offset
voltage, which due to imbalances in the internal op amp circuitry is always present. Max-
imum values are quoted in manufacturers’ data sheets. For the same reason, a small input
bias current is present at the inverting and noninverting terminals of an op amp that is
connected in a circuit.
A further limitation is the slowed response of an op amp to a quickly changing
input. If the input changes too quickly, there will be a delay in the output, given by
the slew rate limit. Typically, the vertical part in an ideal step input will be reproduced
slanted.
A final limitation is the common-mode rejection ratio (CMRR) in differential ampli-
fiers, which are amplifiers that provide a high degree of discrimination against common-
mode (interfering) signals while amplifying a differential signal. CMRR is a ratio of
differential gain to common-mode gain and was fully explored in this chapter. A popular,
general-purpose and cheap op amp is the 741. Its specifications are Ri ¼ 2 MΩ, Ro ¼ 75 Ω,
A ¼ 2  105, CMRR ¼ 90 dB, and supply voltage ¼  15 V.

Problems
1. Explain why a high input resistance and a low output resistance are desirable
characteristics of an amplifier.
2. Calculate the gain of the inverting op amp given in Example 6.1 without initially
assuming that υd ¼ 0. Use the resistance values specified in the example and compare
the gain to the value of
100 obtained by using the gain expression
RF/R1.
Ans: Error ¼
0.02%.
3. In the inverting amplifier of Fig. 6.1b, find the input impedance if R1 ¼ 10 kΩ and
RF ¼ 200 kΩ. Assume the op amp is ideal.
4. A particular microphone which produces an open-circuit voltage of 50 mV can be
modeled by 50 mV source voltage (υs) in series with a 10 kΩ source resistance (Rs). If
the microphone voltage needs to be amplified to a level of 5 V, design an inverting
amplifier to accomplish it.
5. For the microphone case described in Problem 4, design a noninverting amplifier.
Ans: If R1 ¼ 1 kΩ, then RF ¼ 99 kΩ in Fig. 6.2a.
6. A mediocre op amp (see Fig. 6.1a) with A ¼ 104, Ri ¼ 100 kΩ, and Ro ¼ 0.5 kΩ is to be
used as a unit gain buffer (Fig. 6.2a). By writing the circuit equations for Fig. 6.2a, show
that υo  υi and show that the input impedance Zi ¼ (1 + A)Ri. Give numerical values.
7. Repeat the calculations in Example 6.2 when the input device is a microphone with
υs ¼ 10 mV (rms) and source impedance Rs ¼ 50 kΩ. Draw a conclusion regarding the
effectiveness of a buffer.
8. An op amp such as the 741 is used in an inverting amplifier (Fig. 6.1b). If the input
impedance to the inverting amplifier is to be 2 kΩ, design the amplifier for a gain of

50, that is, find R1and RF.
Ans: RF ¼ 100 kΩ, R1 ¼ 2 kΩ.
258 Electronics and communications for scientists and engineers

9. An inverting amplifier of the type shown in Fig. 6.1b uses a 741 op amp that is powered
by a 15 V supply. If the input current i1 is not to exceed 20 μA, design the circuit for
a gain of
100, that is, find R1 and RF. Assume that υo,max is limited by the 15 V power
supply voltage.
10. Two voltages υ1 and υ2 are to be added by a summing amplifier to give an output that is
υo ¼
υ1
5υ2. Design a summer of the type shown in Fig. 6.4a. Use RF ¼ 10 kΩ.
Ans: R1 ¼ 10 kΩ, R2 ¼ 2 kΩ.
11. Find the gain Ar ¼ υo/υ1 of the op amp amplifier circuit shown in Fig. 6.19. What
operation does it perform?

RF
–
R1 A
+ ϑo
ϑ1

FIG. 6.19

12. Repeat Problem 10 but change the desired output voltage to υo ¼
υ1 + υ2.
13. A voltage-to-current converter converts a voltage signal to a proportional output
current. Show that for the circuit in Fig. 6.20 we have iL ¼ υs/R1. As the circuit is
basically a noninverting amplifier (Fig. 6.2a), the load current is independent of the
source impedance Rs and load impedance RL. Hence, the amplifier requires very
little current from the signal source due to the very large input resistance of an
op amp.
14. A current-to-voltage converter converts an input current to a proportional output
voltage. Show that for the circuit in Fig. 6.21 we have υo ¼
isRF. Note that due to the
virtual ground at the inverting amplifier input, the current in Rs is zero and it flows
through the feedback resistor RF.

+
A
–
RS iL RL

+
–
ϑS R1

FIG. 6.20
 Chapter 6 • Operational amplifiers 259

RF

–
A
+ ϑO
RS
iS

FIG. 6.21

100kΩ

–
A ϑO
+ id +

FIG. 6.22

15. A reverse-biased photodiode in the circuit of Fig. 6.22 generates 0.4 μ A of current (id)
per 1 μW of light power falling on it. Find the voltage υo that a voltmeter would read
when the diode is illuminated with radiant power of 50 μW.
16. Two differential amplifiers, differing only in their CMRRs, are available. If the inputs
are υ1 ¼ 1 mV and υ2 ¼ 0.9 mV, calculate the output voltages for the two amplifiers and
show the effect that a higher CMRR has.

(a) Differential-mode gain 2000, common-mode gain 100

(b) Differential-mode gain 2000, common-mode gain 104
Ans: (a) 219 mV with 9.5% error; (b) 200.2 mV with 0.1% error.
17. A 1 μV signal is to be amplified in a differential amplifier for which the CMRR is
specified as 100 dB. Find how large the magnitude of an interfering signal can be
before its output magnitude is equal to that of the desired signal.
18. Find the magnitude of the common-mode output signal of an op amp for which the
CMRR ¼ 80 dB, Ad ¼ 104, and an interfering common-mode signal of 2 V exists.
Ans: υo,cm ¼ 2 V.
19. In the op amp circuit of Fig. 6.9, if capacitor C is replaced by an inductor L, what
operation would the circuit perform?
20. Show that the output voltage υo of a circuit in which R and the transistor in Fig. 6.11
are interchanged is that of an exponential or antilog (inverse log) amplifier.
Ans: υo ¼ α RIoexp(
e υi/kT); the orientation of the npn transistor is such that the
input is at the emitter of the transistor.
260 Electronics and communications for scientists and engineers

21. An op amp low-pass filter is to have a cutoff frequency of 100 Hz and a gain of
magnitude 50. Determine the remaining parameters if the capacitor is specified as
C ¼ 1 μ F.
22. An op amp low-pass filter (Fig. 6.12a) has R1 ¼ 2 kΩ, RF ¼ 22 kΩ, and CF ¼ 0.1 μ F.
Determine the corner frequency and the DC gain.
Ans: 72.3 Hz,
11.
23. Find the unity-gain bandwidth for the low-pass filter of Problem 22; that is, find the
frequency at which the gain has dropped to unity.
24. Design a high-pass op amp filter with a high-frequency gain of
100 and a cutoff
frequency of 1 kHz. Resistors with values of 1 kΩ, 10 kΩ, and 100 kΩ are available.
Determine the capacitance and resistance values of the filter.
25. A 741 op amp for which A ¼ 2105 has a supply voltage of 15 V. If it is used as a
comparator (Fig. 6.14a with Vref ¼ 5 V), determine the change in υ? which will drive
the output voltage from negative saturation (
15 V) to positive saturation.
Ans: 4.999925 V to 5.000075 V.
26. Repeat Problem 25, except for a zero reference voltage.
27. Design a comparator circuit to set off an alarm when the temperature in a boiler
reaches 160 °C. You have available a temperature-to-voltage transducer which
generates a voltage of 5 V at 160 °C and an alarm which activates at
15 V and is off for
larger voltages. Sketch the transfer characteristics.
28. A zero-crossing op amp comparator is powered by two batteries, one with VBB ¼ 15 V
and the other with VBB ¼
15 V, as shown in Fig. 6.14a. Such an arrangement has an
output voltage υo which is +15 V when υ? > 0 and
15 V when υ? < 0. If the input
voltage is given by υ? ¼ 0.01 sin ωt V, sketch the output voltage.
29. Design an analog computer to determine the current i(t) for t > 0 in the series RL
circuit shown in Fig. 1.26a. The switch is turned on at t ¼ 0 and connects a battery V to
RL. The resulting current is determined by di/dt ¼ V/L
Ri/L.
30. A typical solution for Problem 6–29 involves more than one op amp. It is possible, by
combining addition and integration in one op amp (see Fig. 6.10), to design an
efficient, single-op-amp solution to Problem 29. Show the circuit.
31. Determine the differential equation whose analog is shown in Fig. 6.23.

R3
C1
t=0 R3
–
A –
+ R1 A R1C1=1
VB +
R2

FIG. 6.23
 Chapter 6 • Operational amplifiers 261

d2 x
32. Design an analog computer to solve dt 2
+ 3x ¼ cos ωt.
Ans:

d2x –dx
C
dt2 dt C
cosωt
R1 R1
–
–
+ x
R1 +
R1C=1
3R

–3x R
–
+

FIG. 6.24

33. Design an analog computer to solve for the current i(t) in the series RLC circuit of
Fig. 2.2a. The current is the solution to Eq. (2.3), which can be restated here as
dυðt Þ diðt Þ
+ L ddtið2t Þ + iðCt Þ
2

dt ¼ R dt