01_639091_Delmar_Blake 8/16/01 2:51 PM Page 1

Introduction to
CHAPTER
1 Communication
Systems

1.1 Introduction OBJECTIVES

1.2 Elements of a After studying this chapter, you should be able to:
Communication System • Describe the essential elements of a communication system,
1.3 Time and Frequency • Explain the need for modulation in communication systems,
Domains • Distinguish between baseband, carrier, and modulated signals and give
1.4 Noise and examples of each,
Communications • Write the equation for a modulated signal and use it to list and explain
1.5 Spectrum Analysis the various types of continuous-wave modulation,
• Describe time-division and frequency-division multiplexing,
• Explain the relationship between channel bandwidth, baseband band-
width, and transmission time,
• List the requirements for distortionless transmission and describe some
of the possible deviations from this ideal,
• Use frequency-domain representations of signals and convert simple sig-
nals between time and frequency domains,
• Use a table of Fourier series to find the frequency-domain representa-
tions of common waveforms,
• Describe several types of noise and calculate the noise power and volt-
age for thermal noise,
• Calculate signal-to-noise ratio, noise figure, and noise temperature for
single and cascaded stages,
• Use the spectrum analyzer for frequency, power, and signal-to-noise
ratio measurements.

1
1
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 2

ractical electrical communication began in 1837 with Samuel

E LECTRONICS P Morse’s telegraph system. It was not the first system to use elec-
tricity to send messages, but it was the first to be commercially suc-
…REWIND
cessful. Though not electronic, it had all the essential elements of the
communication systems we will study in this book. There was a trans-
Communications:
mitter, consisting of a telegraph key and a battery, to convert informa-
The Past
tion into an electrical signal that could be sent along wires. There was
a receiver, called a sounder, to convert the electrical signal into sound
that could be perceived by the operator (a variant of this system used
marks on a paper tape). There was also a transmission channel, which
consisted of wire on poles. Later, beginning in 1866, telegraph cables
were run under water. By 1898, there were twelve transatlantic cables
in operation.
Voice communication by electrical means began with the invention of the
telephone by Alexander Graham Bell in 1876. Since that time, telephone sys-
tems have steadily expanded and have been linked together in complex ways.
It is now possible to talk to someone almost anywhere in the world from nearly
anywhere else.
The electronic content of telephony has also increased. In its early days, the
telephone system did not involve any electronics at all. Eventually the vacuum
tube and later the transistor allowed the use of amplifiers to increase the dis-
tance over which signals could be sent. Radio links were installed to allow
communication where cables were impractical. Changes to digital electronics
in both signal transmission and switching systems are now under way.
Radio is an important medium of communication. The theoretical frame-
work was constructed by James Clerk Maxwell in 1865 and was verified ex-
perimentally by Heinrich Rudolph Hertz in 1887. Practical radiotelegraph sys-
tems were in use by the turn of the century, mainly for ship-to-shore and
ship-to-ship service. The first transatlantic communication by radio was ac-
complished by Guglielmo Marconi in 1901.
Early radio transmitters used spark gaps to generate radiation and were not
well-suited to voice transmission. By 1906, some transmitters were using spe-
cially designed high-frequency alternators, and one of these was used experi-
mentally to transmit voice. The one-kilowatt transmitter was modulated by
connecting a microphone in series with the antenna. The microphone was al-
legedly water-cooled. Regular radio broadcasting did not begin until 1920, by
which time transmitters and receivers used vacuum-tube technology. The
diode tube had been invented by Sir John Ambrose Fleming in 1904, and the
triode, which could work as an amplifier, by Lee De Forest in 1906.
By the late 1920s, radio broadcasting was commonplace and experiments
with television were under way. The United States and several European coun-
tries had experimental television services before World War II, and after the
war it became a practical worldwide reality.

2
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 3

Section 1.2 ● Elements of a Communication System 3

1.1 Introduction
Communication was one of the first applications of electrical technology. Today,
in the age of fiber optics and satellite television, facsimile machines and cellular
telephones, communication systems remain at the leading edge of electronics.
Probably no other branch of electronics has as profound an effect on people’s
everyday lives.
This book will introduce you to the study of electronic communication sys-
tems. After a brief survey of the history of communications, this chapter will
consider the basic elements that are common to any such system: a transmitter, a
receiver, and a communication channel. We will also begin discussions of sig-
nals and noise that will continue throughout the book.
Chapter 2 will review some of the fundamentals that you will need for later
chapters and introduce you to some of the circuits that are commonly found in
radio-frequency systems and with which you may not be familiar. We will then
proceed, in succeeding chapters, to investigate the various ways of implementing
each of the three basic elements.
We begin with analog systems. The reader who is impatient to get on to digital
systems should realize that many digital systems also require analog technology
to function. The modern technologist needs to have at least a basic understanding
of both analog and digital systems before specializing in one or the other.
It is often said that we are living in the information age. Communication
technology is absolutely vital to the generation, storage, and transmission of this
information.

1.2 Elements of a Communication System

Any communication system moves information from a source to a destination
through a channel. Figure 1.1 illustrates this very simple idea, which is at the
heart of everything that will be discussed in this book. The information from the
source will generally not be in a form that can travel through the channel, so a de-
vice called a transmitter will be employed at one end, and a receiver at the other.

The Source
The source or information signal can be analog or digital. Common examples are
analog audio and video signals and digital data. Sources are often described in
terms of the frequency range that they occupy. Telephone-quality analog voice
signals, for instance, contain frequencies from about 300 Hz to 3 kHz, while ana-
log high-fidelity music needs a frequency range of approximately 20 Hz to 20 kHz.
Video requires a much larger frequency range than audio. An analog video
signal of television-broadcast quality needs a frequency range from dc to about
4.2 MHz.
Digital sources can be derived from audio or video signals or can consist of
data (alphanumeric characters, for example). Digital signals can have almost any
bandwidth depending on the number of bits transmitted per second, and the
method used to convert binary ones and zeros into electrical signals.

Figure 1.1 Elements of a

Source Transmitter Receiver Destination communication system
Channel
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 4

4 Chapter 1 ● Introduction to Communication Systems

The Channel
A communication channel can be almost anything: a pair of conductors or an
optical fiber, for example. Much of this book deals with radio communication, in
which free space serves as the channel.
Sometimes a channel can carry the information signal directly. For example,
an audio signal can be carried directly by a twisted-pair telephone cable. On the
other hand, a radio link through free space cannot be used directly for voice sig-
nals. An antenna of enormous length would be required, and it would not be pos-
sible to transmit more than one signal without interference. Such situations
require the use of a carrier signal whose frequency is such that it will travel, or
propagate, through the channel. This carrier wave will be altered, or modulated,
by the information signal in such a way that the information can be recovered at
the destination. When a carrier is used, the information signal is also known as the
modulating signal. Since the carrier frequency is generally much higher than that
of the information signal, the frequency spectrum of the information signal is
often referred to as the baseband. Thus, the three terms information signal, mod-
ulating signal, and baseband are equivalent in communication schemes involving
modulated carriers.

Types of Modulation
All systems of modulation are variations on a small number of possibilities. A
carrier is generated at a frequency much higher than the highest baseband fre-
quency. Usually, the carrier is a sine wave. The instantaneous amplitude of the
baseband signal is used to vary some parameter of the carrier.
A general equation for a sine-wave carrier is:

e(t)  Ec sin (ct  ) (1.1)

where e(t)  instantaneous voltage as a function of time

Ec  peak voltage
c  frequency in radians per second
t  time in seconds
  phase angle in radians

Using radians per second in the mathematics dealing with modulation makes the
equations simpler. Of course, frequency is usually given in hertz, rather than in
radians per second, when practical devices are being discussed. It is easy to con-
vert between the two systems by recalling from basic ac theory that   2f.
In modulation, the parameters that can be changed are amplitude Ec , frequency
c , and phase . Combinations are also possible; for example, many schemes for
transmitting digital information use both amplitude and phase modulation.
The modulation is done at the transmitter. An inverse process, called demod-
ulation or detection, takes place at the receiver to restore the original baseband
signal.

Signal Bandwidth
An unmodulated sine-wave carrier would exist at only one frequency and so
would have zero bandwidth. However, a modulated signal is no longer a single
sine wave, and it therefore occupies a greater bandwidth. Exactly how much
bandwidth is needed depends on the baseband frequency range (or data rate, in
the case of digital communication) and the modulation scheme in use. Hartley’s
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 5

Section 1.2 ● Elements of a Communication System 5

Law is a general rule that relates bandwidth and information capacity. It states
that the amount of information that can be transmitted in a given time is propor-
tional to bandwidth for a given modulation scheme.
I  ktB (1.2)
where I  amount of information to be sent
k  a constant that depends on the type of modulation
t  time available
B  channel bandwidth
Some modulation schemes use bandwidth more efficiently than others. The
bandwidth of each type of modulated signal will be examined in detail in later
chapters.

Frequency-Division Multiplexing
One of the benefits of using modulated carriers, even with channels that are capa-
ble of carrying baseband signals, is that several carriers can be used at different
frequencies. Each can be separately modulated with a different information sig-
nal, and filters at the receiver can separate the signals and demodulate whichever
one is required.
Multiplexing is the term used in communications to refer to the combining
of two or more information signals. When the available frequency range is di-
vided among the signals, the process is known as frequency-division multiplexing
(FDM).
Radio and television broadcasting, in which the available spectrum is divided
among many signals, are everyday examples of FDM. There are limitations to the
number of signals that can be crowded into a given frequency range because each
requires a certain bandwidth. For example, a television channel occupies a band-
width of 6 MHz. Figure 1.2 shows how FDM applies to the VHF television band.

Time-Division Multiplexing
An alternative method for using a single communication channel to send many
signals is to use time-division multiplexing (TDM). Instead of dividing the avail-
able bandwidth of the channel among many signals, the entire bandwidth is used
for each signal, but only for a small part of the time. A glance at Equation (1.2)
will confirm that time and bandwidth are equivalent in terms of information ca-
pacity. A nonelectronic example is the division of the total available time on a tel-
evision channel among the various programs transmitted. Each program uses the
whole bandwidth of the channel, but only for part of the time. Examples of TDM
in electronic communication are not as common in everyday experience as FDM,
but TDM is used extensively, especially with digital communication. The digital
telephone system is a good example.

Other Services Figure 1.2 Frequency-division

FM Radio (Police, Aircraft, multiplexing in the VHF television
Broadcasting Amateur, etc.) Channel band
2 3 4 5 6 7 8 9 10 11 12 13

54 66 76 88 108 180 192 204 216

60 72 82 174 186 198 210

Frequency (MHz)
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 6

6 Chapter 1 ● Introduction to Communication Systems

It is certainly possible to combine FDM and TDM. For example, the available
bandwidth of a communication satellite is divided among a number of transmitter-
receiver combinations called transponders. This is an example of FDM. A single
transponder can be used to carry a large number of digital signals using TDM.

Frequency Bands
Hertz’s early experiments in the laboratory used frequencies in the range of ap-
proximately 50 to 500 MHz. When others, such as Marconi, tried to apply his
results to practical communication, they initially found that results were much
better at lower frequencies. At that time, little was known about radio propagation
(or about antenna design, for that matter). It is now known that frequencies rang-
ing from a few kilohertz to many gigahertz all have their uses in radio communi-
cation systems.
In the meantime, a system of labeling frequencies came into use. The fre-
quencies most commonly used in the early days, those from about 300 kHz to
3 MHz, were called medium frequencies (MF). Names were assigned to each order
of magnitude of frequency, both upward and downward, and these names persist to
this day. Thus we have low frequencies (LF) from 30 to 300 kHz and very low fre-
quencies (VLF) from 3 to 30 kHz. Going the other way, we have high frequencies
(HF) from 3 to 30 MHz, very high frequencies (VHF) from 30 to 300 MHz, and so
on. Figure 1.3 shows the entire usable spectrum, with the appropriate labels.

FREQUENCY WAVELENGTH SAMPLE USES

300 GHz 1 mm
Extremely High Frequencies Millimeter Waves
(EHF)
30 GHz 1 cm
Super High Frequencies Radar
(SHF)
3 GHz 10 cm Microwaves Communications Satellites,
Ultra High Frequencies Microwave Ovens, Personal Communication
(UHF) Systems, Cellular Telephones
300 MHz 1m
Very High Frequencies TV Broadcast
(VHF) FM Broadcast
30 MHz 10 m
High Frequencies Short Waves Shortwave Broadcast
(HF) Commercial
3 MHz 100 m
Medium Frequencies Medium Waves AM Broadcast
(MF)
300 kHz 1 km
Low Frequencies Long Waves Navigation
(LF)
30 kHz 10 km Submarine Communications
Very Low Frequencies
(VLF)
3 kHz 100 km
Voice Frequencies
(VF)
300 Hz 1000 km Audio
Extremely Low Frequencies
(ELF)
30 Hz 10,000 km Power Transmission

Figure 1.3 The radio-frequency spectrum

01_639091_Delmar_Blake 8/16/01 2:51 PM Page 7

Section 1.2 ● Elements of a Communication System 7

Radio waves can also be described according to their wavelength, which is

the distance a wave travels in one period. The general equation that relates fre-
quency to wavelength for any wave is:

  f (1.3)

where   velocity of the wave in meters per second

f  frequency of the wave in hertz
  wavelength in meters
For a radio wave in free space, the velocity is the same as that of light, which is
approximately 300  106 meters per second. The usual symbol for this quantity
is c. Equation (1.3) then becomes
c  f (1.4)

where c  speed of light, 300  106 m/s

f  frequency in hertz
  wavelength in meters

It can be seen from Equation (1.4) that a frequency of 300 MHz corresponds to a
wavelength of 1 m and that wavelength is inversely proportional to frequency.
Low-frequency signals are sometimes called long wave, high frequencies corre-
spond to short wave, and so on. The reader is no doubt familiar with the term
microwave to describe signals in the gigahertz range.

EXAMPLE 1.1 Calculate the wavelength in free space corresponding

to a frequency of:
(a) 1 MHz (AM radio broadcast band)
(b) 27 MHz (CB radio band)
(c) 4 GHz (used for satellite television)

Solution Rearranging Equation (1.4).

c  f
c

f
300  106 m/s
(a)  
1  106 Hz
 300 m
300  106 m/s
(b)  
27  106 Hz
 11.1 m
300  106 m/s
(c)  
4  109 Hz
 0.075 m
 7.5 cm
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 8

8 Chapter 1 ● Introduction to Communication Systems

Distortionless Transmission
The receiver should restore the baseband signal exactly. Of course, there will be a
time delay due to the distance over which communication takes place, and there
will likely be a change in amplitude as well. Neither effect is likely to cause prob-
lems, though there are exceptions. The time delay involved in communication via
geostationary satellite can be a nuisance in telephone communication. Even
though radio waves propagate at the speed of light, the great distance over which
the signal must travel (about 70,000 kilometers) causes a quarter-second delay.
Any other changes in the baseband signal reflect distortion, which has a cor-
rupting effect on the signal. There are many possible types of distortion. Some
types are listed below, but not all of them will be immediately clear. Much of the
rest of this book will be devoted to explaining them in more detail. Some possible
types of distortion are:
• Harmonic distortion: harmonics (multiples) of some of the baseband compo-
nents are added to the original signal.
• Intermodulation distortion: additional frequency components generated by
combining (mixing) the frequency components in the original signal. Mixing
will be described in Chapter 2.
• Nonlinear frequency response: some baseband components are amplified more
than others.
• Nonlinear phase response: phase shift between components of the signal.
• Noise: both the transmitter and the receiver add noise, and the channel is also
noisy. This noise adds to the signal and masks it. Noise will be discussed later
in this chapter.
• Interference: if more than one signal uses the same transmission medium, the
signals may interact with each other.
One of the advantages of digital communication is the ability to regenerate
a signal that has been corrupted by noise and distortion, provided that it is still
identifiable as representing a one or a zero. In analog systems, however, noise and
distortion tend to accumulate. In some cases distortion can be removed at a later
point. If the frequency response of a channel is not flat but is known, for instance,
equalization in the form of filters can be used to compensate. However, harmonic
distortion, intermodulation, and noise, once present, are impossible to remove
completely from an analog signal. A certain amount of immunity can be built into
digital schemes, but excessive noise and distortion levels will be reflected in in-
creased error rates.

SECTION 1.2 REVIEW Explain what modulation is, why it is necessary, and how it can be accomplished.
QUESTION ●

1.3 Time and Frequency Domains

The reader is probably familiar with the time-domain representation of signals.
An ordinary oscilloscope display, showing amplitude on one scale and time on
the other, is a good example.
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 9

Section 1.3 ● Time and Frequency Domains 9

Signals can also be described in the frequency domain. In a frequency- 1

domain representation, amplitude or power is shown on one axis and frequency is

v (V)
displayed on the other. A spectrum analyzer gives a frequency-domain represen- 0
tation of signals. 0.5 1
Any signal can be represented either way. For example, a 1 kHz sine wave is
–1
shown in both ways in Figure 1.4. The time-domain representation should need
no explanation. As for the frequency domain, a sine wave has energy only at its t (ms)
fundamental frequency, so it can be shown as a straight line at that frequency. (a) Time domain
Notice that our way of representing the signal in the frequency domain fails
to show one thing: the phase of the signal. The signal in Figure 1.4(b) could be a
cosine wave just as easily as a sine wave.

Vp ( V )
Frequency-domain representations are very useful in the study of communi- 1
cation systems. For instance, the bandwidth of a modulated signal generally has
some fairly simple relationship to that of the baseband signal. This bandwidth can
easily be found if the baseband signal can be represented in the frequency do- 1.0
main. As we proceed, we will see many other examples in which the ability to f (kHz)
work with signals in the frequency domain will be required. (b) Frequency domain

Figure 1.4 Sine wave in time and

Fourier Series frequency domains
Any well-behaved periodic waveform can be represented as a series of sine and/or
cosine waves at multiples of its fundamental frequency plus (sometimes) a dc off-
set. This is known as a Fourier series. This very useful (and perhaps rather sur-
prising) fact was discovered in 1822 by Joseph Fourier, a French mathematician,
in the course of research on heat conduction. Not all signals used in communica-
tions are strictly periodic, but they are often close enough for practical purposes.
Fourier’s discovery, applied to a time-varying signal, can be expressed math-
ematically as follows:

A0
f(t)   A1 cos t  B1 sin t  A2 cos 2t
2 (1.5)
 B2 sin 2t  A3 cos 3t  B3 sin 3t    

where f(t)  any well-behaved function of time as described above. For

our purposes, f(t) will generally be either a voltage v(t) or a
current i(t).
An and Bn  real-number coefficients; that is, they can be positive, nega-
tive, or zero.
  radian frequency of the fundamental.

The radian frequency can be found from the time-domain representation of

the signal by finding the period (that is, the time T after which the whole signal
repeats exactly) and using the equations:

1
f (1.6)
T

and

  2f (1.7)
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 10

10 Chapter 1 ● Introduction to Communication Systems

The simplest ac signal is a sinusoid. The frequency-domain representation of

a sine wave has already been described and is shown in Figure 1.4(b) for a voltage
sine wave with a period of 1 ms and a peak amplitude of 1 V. For this signal, all
the coefficients of Equation (1.5) are zero except for B1, which has a value of 1 V.
The equation becomes

(t)  sin (2000) V

which is certainly no surprise.

The examples that follow use the equations for the various waveforms shown
in Table 1.1.

EXAMPLE 1.2 Find and sketch the Fourier series corresponding to the square wave in
Figure 1.5(a)

1
v (V)

0.5 1 1

Vp (V)

1 2 3 4 5 6 7
t (ms)
f (kHz)

(a) Time domain (b) Frequency domain

Figure 1.5

Solution A square wave is another signal with a simple Fourier representa-

tion, although the representation is not quite as simple as that for a sine wave.
For the signal shown in Figure 1.5(a), the frequency is 1 kHz, as before, and
the peak voltage is 1 V.
According to Table 1.1, this signal has components at an infinite number
of frequencies: all odd multiples of the fundamental frequency of 1 kHz.
However, the amplitude decreases with frequency, so that the third harmonic
has an amplitude one-third that of the fundamental, the fifth harmonic an am-
plitude of one-fifth the fundamental, and so on. Mathematically, a square
wave of voltage with a rising edge at t  0 and no dc offset can be expressed
as follows:

asin t  sin 3t  sin 5t     b

4V 1 1
(t)  (1.8)
 3 5
where V  peak amplitude of the square wave
  radian frequency of the square wave
t  time in seconds
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 11

Section 1.3 ● Time and Frequency Domains 11

For this example, Equation (1.8) becomes:

c sin (2  103t)  sin (6  103t)

4 1
(t) 
 3
 sin (10  10 t)     d V
1 3
5

This equation shows that the signal has frequency components at odd mul-
tiples of 1 kHz, that is, at 1 kHz, 3 kHz, 5 kHz, and so on. The 1 kHz compo-
nent has a peak amplitude of
4
V1   1.27 V

The component at three times the fundamental frequency (3 kHz) has an
amplitude one-third that of the fundamental, that at 5 kHz has an amplitude
one-fifth that of the fundamental, and so on.
4
V3   0.424 V
3
4
V5   0.255 V
5
4
V7   0.182 V
7
The result for the first four components is sketched in Figure 1.5(b). Theo-
retically, an infinite number of components would be required to describe the
square wave completely, but, as the frequency increases, the amplitude of the
components decreases rapidly.

The representations in Figure 1.5(a) and 1.5(b) are not two different sig-
nals but merely two different ways of looking at the same signal. This can be
shown graphically by adding the instantaneous values of several of the sine
waves in the frequency-domain representation. If enough of these components
are included, the result begins to look like the square wave in the time-domain
representation. Figure 1.6 (page 14) shows the results for two, four, and ten
components. It was created by simply taking the instantaneous values of all the
components at the same time and adding them algebraically. This was done for
a number of time values, resulting in the graphs in Figure 1.6. Doing these cal-
culations by hand would be simple but rather tedious, so a computer was used
to perform the calculations and plot the graphs. A perfectly accurate represen-
tation of the square wave would require an infinite number of components, but
we can see from the figure that using ten components gives a very good repre-
sentation because the amplitudes of the higher-frequency components of the
signal are very small.
It is possible to go back and forth at will between time and frequency do-
mains, but it should be apparent that information about the relative phases of the
frequency components in the Fourier representation of the signal is required to
reconstruct the time-domain representation. The Fourier equations do have this
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 12

12 Chapter 1 ● Introduction to Communication Systems

TABLE 1.1 Fourier series for common repetitive waveforms

1. Half-wave rectified sine wave

V

v
T/2 T

a    b
V V 2V cos 2t cos 4t cos 6t
(t)   sin t   
 2  13 35 57
2. Full-wave rectified sine wave
(a) With time zero at voltage zero
V

v
T/2 T

a    b
2V 4V cos 2t cos 4t cos 6t
(t)    
  13 35 57
(b) With time zero at voltage peak
V

0
v T/2 T

a1     b
2V 2 cos 2t 2 cos 4t 2 cos 6t
(t)   
 13 35 57
3. Square wave
(a) Odd function
V

0
v

T/2 T
−V

asin t  sin 3t  sin 5t     b

4V 1 1
(t) 
 3 5
(b) Even function
V

0
v

−V

acos t  cos 3t  cos 5t     b

4V 1 1
(t) 
 3 5
(continues)
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 13

Section 1.3 ● Time and Frequency Domains 13

TABLE 1.1 Continued

4. Pulse train τ
V

v
−T T

t
2V sin /T
a cos 3t     b
V sin 2/T sin 3/T
(t)   cos t  cos 2t 
T T /T 2/T 3/T

5. Triangle wave
V

v
−T T
−V

acos t  2 cos 3t  2 cos 5t     b

8V 1 1
(t) 
2 3 5

6. Sawtooth wave V
(a) With no dc offset
0
v

−T T
−V

asin t  sin 2t  sin 3t     b

2V 1 1
(t) 
 2 3

(b) Positive-going
V

0
v

−T T
−V

t
(t)   asin t  sin 2t  sin 3t     b
V V 1 1
2  2 3
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 14

14 Chapter 1 ● Introduction to Communication Systems

v2(t) (Volts) −2
0 0.001
t (Seconds) (a) Two components

2
v4(t) (Volts)

−2
0 0.001
t (Seconds) (b) Four components

2
v10(t) (Volts)

−2
0 0.001
t (Seconds) (c) Ten

Figure 1.6 Construction of a square wave from Fourier components

3
v(t) (V)

−3
0 0.001
t (Seconds)

Figure 1.7 Addition of square-wave Fourier components with wrong phase angles
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 15

Section 1.3 ● Time and Frequency Domains 15

information, but the sketch in Figure 1.5(b) does not. If the phase relationships
between frequency components are changed in a communication system, the sig-
nal will be distorted in the time domain.
Figure 1.7 illustrates this point. The same ten coefficients were used as in
Figure 1.6(c), but this time the waveforms alternated between sine and cosine:
sine for the fundamental, cosine for the third harmonic, sine for the fifth, and so
on. The result is a waveform that looks the same on the frequency-domain sketch
of Figure 1.5(b) but very different in the time domain.

Find the Fourier series for the signal in Figure 1.8(a). EXAMPLE 1.3
Solution The positive-going sawtooth wave of Figure 1.8(a) has a Fourier
series with a dc term and components at all multiples of the fundamental fre-
quency. From Table 1.1, the general equation for such a wave is

 a b asin t  sin 2 t  sin 3 t     b (1.9)

A A 1 1
(t) 
2  2 3
The first (dc) term is simply the average value of the signal.
For the signal in Figure 1.8, which has a frequency of 1 kHz and a peak
amplitude of 5 V, Equation (1.9) becomes:
(t)  2.5  1.59 [sin (2  103t)  0.5 sin (4  103t)
 0.33 sin (6  103t)  . . .] V

4
3.5
5 3
4 2.5
Vp (V)
v (V)

2
3
1.5
2
1
1 0.5
0 0
1 2 3 1 2 3 4 5

t (ms) f (kHz)
(a) Time domain (b) Frequency domain

Figure 1.8

Effect of Filtering on Signals

As we have seen, many signals have a bandwidth that is theoretically infinite.
Limiting the frequency response of a channel removes some of the frequency
components of these signals and causes the time-domain representation to be dis-
torted. An uneven frequency response will emphasize some components at the ex-
pense of others, again causing distortion. Nonlinear phase shift will also affect the
time-domain representation. For instance, shifting the phase angles of some of the
frequency components in the square-wave representation of Figure 1.7 changed
the signal to something other than a square wave.
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 16

16 Chapter 1 ● Introduction to Communication Systems

However, Figure 1.6 shows that while an infinite bandwidth may theoreti-
cally be required, for practical purposes quite a good representation of a square
wave can be obtained with a band-limited signal. In general, the wider the band-
width, the better, but acceptable results can be obtained with a band-limited
signal. This is welcome news, because practical systems always have finite band-
width.
Although real communication systems have a solid theoretical base, they
also involve many practical considerations. There is always a trade-off between
fidelity to the original signal (that is, the absence of any distortion of its wave-
form) and such factors as bandwidth and cost. Increasing the bandwidth often in-
creases the cost of a communication system. Not only is the hardware likely to
be more expensive, but bandwidth itself may be in short supply. In radio com-
munication, for instance, spectral allocations are assigned by national regulatory
bodies in keeping with international agreements. This task is handled in the
United States by the Federal Communications Commission (FCC) and in Canada
by the communications branch of Industry and Science Canada. The allocations
are constantly being reviewed, and there never seems to be enough room in the
usable spectrum to accommodate all potential users.
In communication over cables, whether by electrical or optical means, the
total bandwidth of a given cable is fixed by the technology employed. The more
bandwidth used by each signal, the fewer signals can be carried by the cable.
Consequently, some services are constrained by cost and/or government
regulation to use less than optimal bandwidth. For example, telephone systems
use about 3 kHz of baseband bandwidth for voice. This is obviously not distor-
tionless transmission: the difference between a voice heard “live,” in the same
room, and the same voice heard over the telephone is obvious. High-fidelity
audio needs at least 15 kHz of baseband bandwidth. On the other hand, the main
goal of telephone communication is the understanding of speech, and the band-
width used is sufficient for this. Using a larger bandwidth would contribute little
or nothing to intelligibility and would greatly increase the cost of nearly every
part of the system.

SECTION 1.3 REVIEW What bandwidth is required to transmit the first seven components of a triangle wave
QUESTION with a fundamental frequency of 3 kHz?
●

1.4 Noise and Communications

Noise in communication systems originates both in the channel and in the com-
munication equipment. Noise consists of undesired, usually random, variations
that interfere with the desired signals and inhibit communication. It cannot be
avoided completely, but its effects can be reduced by various means, such as re-
ducing the signal bandwidth, increasing the transmitter power, and using low-
noise amplifiers for weak signals.
It is helpful to divide noise into two types: internal noise, which originates
within the communication equipment, and external noise, which is a property of
the channel. (In this book, the channel will usually be a radio link.)
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 17

Section 1.4 ● Noise and Communications 17

External Noise
If the channel is a radio link, there are many possible sources of noise. The reader
is no doubt familiar with the “static” that afflicts AM radio broadcasting during
thunderstorms. Interference from automobile ignition systems is another common
problem. Domestic and industrial electrical equipment, from light dimmers to vac-
uum cleaners to computers, can also cause objectionable noise. Not so obvious,
but of equal concern to the communications specialist, is noise produced by the
sun and other stars.

Equipment Noise Noise is generated by equipment that produces sparks. Ex-

amples include automobile engines and electric motors with brushes. Any fast-
risetime voltage or current can also generate interference, even without arcing.
Light dimmers and computers fall into this category.
Noise of this type has a broad frequency spectrum, but its energy is not
equally distributed over the frequency range. This type of interference is generally
more severe at lower frequencies, but the exact frequency distribution depends
upon the source itself and any conductors to which it is connected. Computers, for
instance, may produce strong signals at multiples and submultiples of their clock
frequency and little energy elsewhere.
Man-made noise can propagate through space or along power lines. It is usu-
ally easier to control it at the source than at the receiver. A typical solution for a
computer, for instance, involves shielding and grounding the case and all con-
necting cables and installing a low-pass filter on the power line where it enters the
enclosure.

Atmospheric Noise Atmospheric noise is often called static because lightning,

which is a static-electricity discharge, is a principal source of atmospheric noise.
This type of disturbance can propagate for long distances through space. Most of the
energy of lightning is found at relatively low frequencies, up to several megahertz.
Obviously nothing can be done to reduce atmospheric noise at the source.
However, circuits are available to reduce its effect by taking advantage of the fact
that this noise has a very high peak-to-average power ratio; that is, the noise oc-
curs in short intense bursts with relatively long periods of time between bursts. It
is often possible to improve communication by simply disabling the receiver for
the duration of the burst. This technique is called noise blanking.

Space Noise The sun is a powerful source of radiation over a wide range of fre-
quencies, including the radio-frequency spectrum. Other stars also radiate noise
called cosmic, stellar, or sky noise. Its intensity when received on the earth is nat-
urally much less than for solar noise because of the greater distance.
Solar noise can be a serious problem for satellite reception, which becomes
impossible when the satellite is in a line between the antenna and the sun. Space
noise is more important at higher frequencies (VHF and above) because atmos-
pheric noise dominates at lower frequencies and because most of the space noise
at lower frequencies is absorbed by the upper atmosphere.

Internal Noise
Noise is generated in all electronic equipment. Both passive components (such as
resistors and cables) and active devices (such as diodes, transistors, and tubes)
can be noise sources. Several types of noise will be examined in this section, be-
ginning with the most important, thermal noise.
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 18

18 Chapter 1 ● Introduction to Communication Systems

Thermal Noise Thermal noise is produced by the random motion of electrons

in a conductor due to heat. The term “noise” is often used alone to refer to this
type of noise, which is found everywhere in electronic circuitry.
The power density of thermal noise is constant with frequency, from zero to
frequencies well above those used in electronic circuits. That is, there is equal
power in every hertz of bandwidth. Thermal noise is thus an equal mixture of
noise of all frequencies. It is sometimes called white noise, in analogy with white
light, which is an equal mixture of all colors.
The noise power available from a conductor is a function of its temperature,
as shown by the equation:

PN  kTB (1.10)

where PN  noise power in watts

k  Boltzmann’s constant, 1.38  1023 joules/kelvin (J/K)
T  absolute temperature in kelvins (K); this can be found by adding
273 to the celcius temperature
B  noise power bandwidth in hertz
v

This equation is based on the assumption that power transfer is maximum.

That is, the source (the resistance generating the noise) and the load (the ampli-
fier or other device that receives the noise) are assumed to be matched in
t impedance.
(a) Voltage as a function of time Equation (1.10) shows that noise power is directly related to bandwidth over
which the noise is observed (the noise power bandwidth). Filters are often speci-
fied in terms of their bandwidth at half power, that is, at the points where the gain
Power Density W/Hz

of the filter is 3 dB down from its gain in the center of the passband. An ideal
bandpass filter would allow no noise to pass at frequencies outside the passband,
kT so its noise power bandwidth would be equal to its half-power bandwidth. For a
practical filter, the noise power bandwidth is generally greater than the half-power
bandwidth, since some contribution is made to the total noise power by frequen-
cies that are attenuated by more than 3 dB. An ordinary LC tuned circuit has a
f (Hz) noise power bandwidth about 50% greater than its half-power bandwidth.
The power referred to in the previous paragraph is average power. The in-
(b) Power density as a function of
stantaneous voltage and power are random, but over time the voltage will average
frequency
out to zero, as it does for an ordinary ac signal, and the power will average to the
Figure 1.9 Noise voltage and power value given by Equation (1.10). Figure 1.9 shows this effect.

EXAMPLE 1.4 A receiver has a noise power bandwidth of 10 kHz. A

resistor that matches the receiver input impedance is connected across its
antenna terminals. What is the noise power contributed by that resistor in
the receiver bandwidth, if the resistor has a temperature of 27°C?

Solution First, convert the temperature to kelvins:

T(K)  T(°C)  273

 27  273
 300 K
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 19

Section 1.4 ● Noise and Communications 19

According to Equation (1.10), a conductor at a temperature of 300 K would

contribute the following noise power in that 10 kHz bandwidth:
PN  kTB
 (1.38  1023 J/K) (300 K) (10  103 Hz)
 4.14  1017 W

This is obviously not a great deal of power, but it can be significant at the
signal levels dealt with in sensitive receiving equipment.

Thermal noise power exists in all conductors and resistors at any temperature
above absolute zero. The only way to reduce it is to decrease the temperature or
the bandwidth of a circuit (or both). Amplifiers used with very low level signals
are often cooled artificially to reduce noise. This technique is called cryogenics
and may involve, for example, cooling the first stage of a receiver for radio astron-
omy by immersing it in liquid nitrogen. The other method, bandwidth reduction,
will be referred to many times throughout this book. Using a bandwidth greater
than required for a given application is simply an invitation to noise problems.
Thermal noise, as such, does not depend on the type of material involved or
the amount of current passing through it. However, some materials and devices
also produce other types of noise that do depend on current. Carbon-composition
resistors are in this category, for example, and so are semiconductor junctions.

Noise Voltage Often we are more interested in the noise voltage than in the
power involved. The noise power depends only on bandwidth and temperature, as
stated above. Power in a resistive circuit is given by the equation:

V2
P (1.11)
R
From this, we see that:

V 2  PR (1.12)
V  1PR
Equation (1.12) shows that the noise voltage in a circuit must depend on the
resistances involved, as well as on the temperature and bandwidth.
Figure 1.10 shows a resistor that serves as a noise source connected across
another resistor, RL , that will be considered as a load. The noise voltage is repre-
sented as a voltage source VN, in series with a noiseless resistance RN. We know VN/2 RN
from Equation (1.10) that the noise power supplied to the load resistor, assuming
a matched load, is: RL VN /2
VN
PN  kTB

In this situation, one-half the noise voltage appears across the load, and the Noisy Resistor
rest is across the resistor that generates the noise.
The root-mean-square (RMS) noise voltage across the load is given by: Figure 1.10 Thermal noise voltage

VL  1PR
 1kTBR
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 20

20 Chapter 1 ● Introduction to Communication Systems

An equal noise voltage will appear across the source resistance, so the noise
source shown in Figure 1.10 must have double this voltage or

VN  21kTBR (1.13)
 14kTBR

EXAMPLE 1.5 A 300  resistor is connected across the 300 

antenna input of a television receiver. The bandwidth of the receiver is
6 MHz, and the resistor is at room temperature (293 K or 20°C or 68°F).
Find the noise power and noise voltage applied to the receiver input.

Solution The noise power is given by Equation (1.10):

PN  kTB
 (1.38  1023 J/K) (293 K) (6  106 Hz)
 24.2  1015 W
 24.2 fW
The noise voltage is given by Equation (1.13):

VN  14kTBR
 14 (1.38  1023 J/K) (293 K) (6  106 Hz) (300 )
 5.4  106 V
 5.4 V
Of course, only one-half this voltage appears across the antenna terminals;
the other half appears across the source resistance. Therefore the actual noise
voltage at the input is 2.7 V.

Shot Noise This type of noise has a power spectrum that resembles that of ther-
mal noise, having equal energy in every hertz of bandwidth, at frequencies from
dc into the gigahertz region. However, the mechanism that creates shot noise is
different. Shot noise is due to random variations in current flow in active devices
such as tubes, transistors, and semiconductor diodes. These variations are caused
by the fact that current is a flow of carriers (electrons or holes), each of which
carries a finite amount of charge. Current can thus be considered as a series of
pulses, each consisting of the charge carried by one electron.
The name “shot noise” describes the random arrival of electrons arriving at
the anode of a vacuum tube, like individual pellets of shot from a shotgun.
One would expect that the resulting noise power would be proportional to the
device current, and this is true both for vacuum tubes and for semiconductor junc-
tion devices.
Shot noise is usually represented by a current source. The noise current for
either a vacuum or a junction diode is given by the equation:

IN  12qI0B (1.14)
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 21

Section 1.4 ● Noise and Communications 21

where IN  RMS noise current, in amperes

q  magnitude of the charge on an electron,
equal to 1.6  10 19 coulomb
I0  dc bias current in the device, in amperes
B  bandwidth over which the noise is observed, in hertz

This equation is valid for frequencies much less than the reciprocal of the carrier
transit time for the device. The transit time is the time a charge carrier spends in
the device. Depending on the device, Equation (1.14) may be valid for frequen-
cies up to a few megahertz or several gigahertz.
Since shot noise closely resembles thermal noise in its random amplitude and
flat spectrum, it can be used as a substitute for thermal noise whenever a known
VN out
level of noise is required. Such a noise source can be very useful, particularly
when conducting measurements on receivers. A calibrated, variable noise genera-
tor can easily be constructed using a diode with adjustable bias current. Figure
1.11 shows a simplified circuit for such a generator. Figure 1.11 Diode noise generator

EXAMPLE 1.6 A diode noise generator is required to produce 10 V

of noise in a receiver with an input impedance of 75 , resistive, and a
noise power bandwidth of 200 kHz. (These values are typical of FM
broadcast receivers.) What must the current through the diode be?

Solution First, convert the noise voltage to current, using Ohm’s Law:

VN
IN 
R
10 V

75 
 0.133 A

Next, rearrange Equation (1.14) and solve for I0:

IN  12qI0B
I 2N  2qI0B
I 2N
I0 
2qB
(0.133  106 A)2

2 (1.6  1019 C) (200  103 Hz)
 0.276 A or 276 mA

Partition Noise Partition noise is similar to shot noise in its spectrum and
mechanism of generation, but it occurs only in devices where a single current sep-
arates into two or more paths. An example of such a device is a bipolar junction
transistor, where the emitter current is the sum of the collector and base currents.
As the charge carriers divide into one stream or the other, a random element in the
currents is produced. A similar effect can occur in vacuum tubes.
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 22

22 Chapter 1 ● Introduction to Communication Systems

The amount of partition noise depends greatly on the characteristics of the

particular device, so no equation for calculating it will be given here. The same is
true of shot noise in devices with three or more terminals. Device manufacturers
provide noise figure information on their data sheets when a device is intended for
use in circuits where signal levels are low and noise is important.
Partition noise is not a problem in field-effect transistors, where the gate cur-
rent is negligible.
Excess Noise Excess noise is also called flicker noise or 1/f noise (because the
noise power varies inversely with frequency). Sometimes it is called pink noise
because there is proportionately more energy at the low-frequency end of the
spectrum than with white noise, just as pink light has a higher proportion of red
(the low-frequency end of the visible spectrum) than does white light. Excess
noise is found in tubes but is a more serious problem in semiconductors and in
carbon resistors. It is not fully understood, but it is believed to be caused by vari-
ations in carrier density. Excess noise is rarely a problem in communication cir-
cuits, because it declines with increasing frequency and is usually insignificant
above approximately 1 kHz.
You may meet the term “pink noise” again. It refers to any noise that has
equal power per octave rather than per hertz. Pink noise is often used for testing
and setting up audio systems.
Transit-Time Noise Many junction devices produce more noise at frequencies
approaching their cutoff frequencies. This high-frequency noise occurs when the
time taken by charge carriers to cross a junction is comparable to the period of the
signal. Some of the carriers may diffuse back across the junction, causing a fluc-
tuating current that constitutes noise. Since most devices are used well below the
frequency at which transit-time effects are significant, this type of noise is not
usually important (except in some microwave devices).

Addition of Noise from Different Sources

All of the noise sources we have looked at have random waveforms. The ampli-
tude at a particular time cannot be predicted, though the average voltage is zero
and the RMS voltage or current and average power are predictable (and not zero).
Signals like this are said to be uncorrelated; that is, they have no definable phase
relationship. In such a case, voltages and currents do not add directly, but the total
voltage (of series circuits) or current (of parallel circuits) can be found by taking
the square root of the sum of the squares of the individual voltages or currents.
Mathematically, we can say, for voltage sources in series:

VNt  2V 2N1  V 2N2  V 2N3     (1.15)

and similarly, for current sources in parallel:

INt  2I 2N1  I 2N2  I 2N3     (1.16)

EXAMPLE 1.7 The circuit in Figure 1.12 shows two resistors in series
at two different temperatures. Find the total noise voltage and noise power
produced at the load, over a bandwidth of 100 kHz.
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 23

Section 1.4 ● Noise and Communications 23

Figure 1.12

R1
100 Ω
300 K
RL
300 Ω
R2
200 Ω
400 K

Noise Load
Source

Solution The open-circuit noise voltage can be found from Equation (1.15)

VNt  2V 2R1  V 2R2

 3(24kTBR1)2  14kTBR2)2
 24kT1BR1  4kT2BR2
 14kB(T1R1  T2R2)
 14 (1.38  1023 J/K) (100  103 Hz) [(300 K) (100 )  (400 K) (200 )]
 779 nV

This is, of course, the open-circuit noise voltage for the resistor combina-
tion. Since in this case the load is equal in value to the sum of the resistors,
one-half of this voltage, or 390 nV, will appear across the load.
The easiest way to find the load power is from Equation (1.11):

V2
P
R
(390 nV)2

300 
 0.506  1015 W
 0.506 fW

Signal-to-Noise Ratio
The main reason for studying and calculating noise power or voltage is the effect
that noise has on the desired signal. In an analog system, noise makes the signal un-
pleasant to watch or listen to and, in extreme cases, difficult to understand. In digi-
tal systems, noise increases the error rate. In either case, it is not really the amount
of noise that concerns us but rather the amount of noise compared to the level of the
desired signal. That is, it is the ratio of signal to noise power that is important,
rather than the noise power alone. This signal-to-noise ratio, abbreviated S/N
and usually expressed in decibels, is one of the most important specifications
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 24

24 Chapter 1 ● Introduction to Communication Systems

of any communication system. Typical values of S/N range from about 10 dB for
barely intelligible speech to 90 dB or more for compact-disc audio systems.
Note that the previous ratio is a power ratio, not a voltage ratio. Sometimes
calculations result in a voltage ratio that must be converted to a power ratio in
order to make meaningful comparisons. Of course, the value in decibels will be
the same either way, provided that both signal and noise are measured in the same
circuit, so that the impedances are the same. The two equations below are used for
power ratios and voltage ratios, respectively:

PS
S/N (dB)  10 log (1.17)
PN

VS
S/N (dB)  20 log (1.18)
VN

At this point, the reader who is not familiar with decibel notation should refer
to Appendix A. Decibels are used constantly in communication theory and prac-
tice, and it is important to understand their use.
Although the signal-to-noise ratio is a fundamental characteristic of any
communication system, it is often difficult to measure. For instance, it may be
possible to measure the noise power by turning off the signal, but it is not possi-
ble to turn off the noise in order to measure the signal power alone. Consequently,
a variant of S/N, called (S  N)/N, is often found in receiver specifications. It
stands for the ratio of signal-plus-noise power to noise power alone.
There are other variations of the signal-to-noise ratio. For instance, there is the
question of what to do about distortion. Though it is not a random effect like ther-
mal noise, the effect on the intelligibility of a signal is rather similar. Consequently,
it might be a good idea to include it with the noise when measuring. This leads to
the expression SINAD which stands for (S  N  D)/(N  D), or signal-plus-
noise and distortion, divided by noise and distortion. SINAD is usually used in-
stead of S/N in specifications for FM receivers.
Both (S  N)/N and SINAD are power ratios, and they are almost always ex-
pressed in decibels. Details on the measurement of (S  N)/N and SINAD will be
provided in Chapter 6.

EXAMPLE 1.8 A receiver produces a noise power of 200 mW with

no signal. The output level increases to 5 W when a signal is applied.
Calculate (S  N)/N as a power ratio and in decibels.

Solution The power ratio is

5W
(S  N)/N 
0.2 W
 25

In decibels, this is

(S  N)/N (dB)  10 log 25

 14 dB
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 25

Section 1.4 ● Noise and Communications 25

Noise Figure
Since thermal noise is produced by all conductors and since active devices add
their own noise as described previously, it follows that any stage in a communica-
tion system will add noise. An amplifier, for instance, will amplify equally both
the signal and the noise at its input, but it will also add some noise. The signal-to-
noise ratio at the output will therefore be lower than at the input.
Noise figure (abbreviated NF or just F) is a figure of merit, indicating how
much a component, stage, or series of stages degrades the signal-to-noise ratio of
a system. The noise figure is, by definition:

(S/N)i
NF  (1.19)
(S/N)o

where (S/N )i = input signal-to-noise power ratio (not in dB)

(S/N )o = output signal-to-noise power ratio (not in dB)
Very often both S/N and NF are expressed in decibels, in which case we
have:

NF (dB)  (S/N)i (dB)  (S/N)o (dB) (1.20)

Since we are dealing with power ratios, it should be obvious that:

NF (dB)  10 log NF (ratio) (1.21)

Noise figure is occasionally called noise factor. Sometimes the term noise
ratio (NR) is used for the simple ratio of Equation (1.19), reserving the term
“noise figure” for the decibel form given by Equation (1.20).

EXAMPLE 1.9 The signal power at the input to an amplifier is

100 W and the noise power is 1 W. At the output, the signal
power is 1 W and the noise power is 30 mW. What is the amplifier
noise figure, as a ratio?

Solution

100 W
(S/N)i   100
1 W
1W
(S/N)o   33.3
0.03 W
100
NF (ratio)  3
33.5

EXAMPLE 1.10 The signal at the input of an amplifier has an S/N of

42 dB. If the amplifier has a noise figure of 6 dB, what is the S/N at the
output (in decibels)?
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 26

26 Chapter 1 ● Introduction to Communication Systems

Solution The S/N at the output can be found by rearranging Equation (1.20):

NF (dB)  (S/N)i (dB)  (S/N)o (dB)

(S/N)o (dB)  (S/N)i (dB)  NF (dB)
 42 dB  6 dB
 36 dB

Equivalent Noise Temperature

Equivalent noise temperature is another way of specifying the noise perform-
ance of a device. Noise temperature has nothing to do with the actual operating
Power Gain A temperature of the circuit. Rather, as shown in Figure 1.13, it is the absolute tem-
(S/N)i (S/N)o perature of a resistor that, connected to the input of a noiseless amplifier of the
same gain, would produce the same noise at the output as the device under dis-
(a) Real (noisy) amplifier
cussion. Figure 1.13(a) shows a real amplifier which, of course, produces noise at
its output, even with no input signal. Figure 1.13(b) shows a hypothetical noise-
Power Gain A less amplifier with a noisy resistor at its input. The combination produces the
NF=0 dB same amount of a noise at its output as does the real amplifier of Figure 1.13(a).
(S/N)i (S/N)o From Equation (1.19), the noise figure is given by:
Noiseless
Amplifier (S/N)i
Equivalent NF 
(S/N)o
Noise
Resistance (Si/Ni)

So/No)
(b) Equivalent with noise (1.22)
resistance and ideal SiNo
(noiseless) amplifier 
NiSo
Figure 1.13 Equivalent noise Si No
resistance 
So Ni

where Si  signal power at input

Ni  noise power at input
So  signal power at output
No  noise power at output
If we are going to refer all the amplifier noise to the input, then this noise, as
well as the noise from the signal source, will be amplified by the gain of the am-
plifier. Let the amplifier power gain be A. Then:
So
A (1.23)
Si
Substituting this into Equation (1.22), we get
No
NF  (1.24)
Ni A
or
No  (NF )Ni A (1.25)

Thus the total noise at the input is (NF)Ni.

01_639091_Delmar_Blake 8/16/01 2:51 PM Page 27

Section 1.4 ● Noise and Communications 27

Assuming that the input source noise Ni is thermal noise, it has the expres-
sion, first shown in Equation (1.10):

Ni  kTB

The equivalent input noise generated by the amplifier must be

Neq  (NF)Ni  Ni
 (NF)kTB  kTB (1.26)
 (NF  1)kTB

If we suppose that this noise is generated in a (fictitious) resistor at temperature

Teq, and if we further suppose that the actual source is at a reference temperature
T0 of 290 K, then we get:

kTeqB  (NF  1)kT0B

Teq  (NF  1)T0 (1.27)
 290(NF  1)

This may seem a trivial result, because it turns out that Teq can be found
directly from the noise figure and thus contains no new information about the
amplifier. However, Teq is very useful when we look at microwave receivers con-
nected to antennas by transmission lines. The antenna has a noise temperature due
largely to space noise intercepted by the antenna. The transmission line and the
receiver will also have noise temperatures. It is possible to get an equivalent noise
temperature for the whole system by simply adding the noise temperatures of the
antenna, transmission line, and receiver. From this, we can calculate the system
noise figure merely by rearranging Equation (1.27):

Teq  290(NF  1)
Teq
NF  1  (1.28)
290
Teq
NF  1
290

Equivalent noise temperatures of low-noise amplifiers are quite low, often less
than 100 K. Note that this does not mean the amplifier operates at this tempera-
ture. It is quite common for an amplifier operating at an actual temperature of
300 K to have an equivalent noise temperature of 100 K.

EXAMPLE 1.11 An amplifier has a noise figure of 2 dB. What is its

equivalent noise temperature?

Solution First, we need the noise figure expressed as a ratio. This can be
found from Equation (1.21):
NF (dB)  10 log NF (ratio)
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 28

28 Chapter 1 ● Introduction to Communication Systems

From which we get:

NF (dB)
NF (ratio)  antilog
10
 antilog 0.2
 1.585
Now we can use Equation (1.27) to find the noise temperature:
Teq  290(NF  1)
 290(1.585  1)
 169.6 K

Cascaded Amplifiers
When two or more stages are connected in cascade, as in a receiver, the noise
figure of the first stage is the most important in determining the noise perform-
ance of the entire system because noise generated in the first stage is amplified in
all succeeding stages. Noise produced in later stages is amplified less, and noise
generated in the last stage is amplified least of all.
While the first stage is the most important noise contributor, the other
stages cannot be neglected. It is possible to derive an equation that relates the
total noise figure to the gain and noise figure of each stage. In the following
discussion, the power gains and noise figures are expressed as ratios, not in
decibels.
We will start with a two-stage amplifier like the one shown in Figure 1.14.
The first stage has a gain A1 and a noise figure NF1, and the second-stage gain and
noise figure are A2 and NF2, respectively.
Let the noise power input to the first stage be
Ni1  kTB
Then, according to Equation (1.25), the noise at the output of the first stage is:
No1  NF1Ni1A1
(1.29)
 NF1kTBA1
This noise appears at the input of the second stage. It is amplified by the sec-
ond stage and appears at the output multiplied by A2. However, the second stage
also contributes noise, which can be expressed, according to Equation (1.26), as:
Neq2  (NF2  1)(kTB) (1.30)

Stage 1 Stage 2

Gain A1 Gain A2
Noise Figure NF1 Noise Figure NF2

Figure 1.14 Noise figure for a two-stage amplifier

01_639091_Delmar_Blake 8/16/01 2:51 PM Page 29

Section 1.4 ● Noise and Communications 29

This noise also appears at the output, amplified by A2. Therefore the total
noise power at the output of the two-stage system is:
No2  No1A2  Neq2A2
 NF1A1kTBA2  (NF2  1)kTBA2 (1.31)
 kTBA2(NF1A1  NF1  1)
The noise figure for the whole system (NFT) is, according to Equation (1.24):
No2
NFT 
Ni1AT
Substituting the value for No2 found in Equation (1.31) and making use of the fact
that the total gain AT is the product of the individual stage gains, we get
kTBA2(NF1A1  NF2  1)
NFT 
kTBA1A2
NF1A1  NF2  1
 (1.32)
A1
NF2  1
 NF1 
A1
This equation, known as Friis’ formula, can be generalized to any number of
stages:

NF2  1 NF3  1 NF4  1

NFT  NF1        (1.33)
A1 A1A2 A1A2A3

This expression confirms our intuitive feeling by showing that the contribu-
tion of each stage is divided by the product of the gains of all the preceding
stages. Thus the effect of the first stage is usually dominant. Note once again that
the noise figures given in this section are ratios, not decibel values, and that the
gains are power gains.

A three-stage amplifier has stages with the following specifications: EXAMPLE 1.12
Calculate the power gain, noise figure, and noise temperature for the entire
amplifier, assuming matched conditions. Power Noise
Stage Gain Figure
Solution The power gain is simply the product of the individual gains.
1 10 2
AT  10  25  30  7500 2 25 4
3 30 5
The noise figure can be found from Equation (1.33):
NF2  1 NF3  1
NFT  NF1  
A1 A1A2
41 51
2 
10 10  25
 2.316
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 30

30 Chapter 1 ● Introduction to Communication Systems

If answers are required in decibels, they can easily be found:

gain (dB)  10 log 7500  38.8 dB
NF (dB)  10 log 2.316  3.65 dB

The noise temperature can be found from Equation (1.27):

Teq  290(NF  1)
 290(2.316  1)
 382 K

SECTION 1.4 REVIEW Why is it more important to have a low noise figure in the first stage of an amplifier
QUESTION than in other stages?
●

1.5 Spectrum Analysis

In this chapter, we have discovered that there are two general ways of looking
at signals: the time domain and the frequency domain. The ordinary oscillo-
scope is a convenient and versatile way of observing signals in the time domain,
providing as it does a graph of voltage with respect to time. It would be logical
to ask whether there is an equivalent instrument for the frequency domain: that
is, something that could display a graph of voltage (or power) with respect to
frequency.
The answer to the question is yes: observation of signals in the frequency
domain is the function of spectrum analyzers, which use one of two basic
techniques. The first technique uses a number of fixed-tuned bandpass filters,
spaced throughout the frequency range of interest. The display indicates the out-
put level from each filter, often with a bar graph consisting of light-emitting
diodes (LEDs). The more filters and the narrower the range, the better the resolu-
tion that can be obtained. This method is quite practical for audio frequencies, and
in fact a crude version, with five to twenty filters, is often found as part of a stereo
system. More elaborate systems are often used in professional acoustics and
sound-reinforcement work. Devices of this type are called real-time spectrum an-
Input
LED alyzers, as all of the filters are active all the time. Figure 1.15 shows a block dia-
Column gram of a real-time spectrum analyzer.
Displays When radio-frequency (RF) measurements are required, the limitations of
for
Each real-time spectrum analyzers become very apparent. It is often necessary to ob-
Channel serve signals over a frequency range with a width of megahertz or even gigahertz.
Obviously, the number of filters required would be too large to be practical. For
RF applications, it would be more logical to use only one, a tunable filter, and
sweep it gradually across the frequency range of interest. This sweeping of the
Bandpass filter could coincide with the sweep of an electron beam across the face of a
Filters
cathode-ray tube (CRT), and the vertical position of the beam could be made a
Figure 1.15 Real-time spectrum function of the amplitude of the filter output. This is the essense of a swept-
analyzer frequency spectrum analyzer like the one shown in Figure 1.16.
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 31

Section 1.5 ● Spectrum Analysis 31

Figure 1.16 Swept-frequency spectrum analyzer (Photograph courtesy of Agilent

Corp).

In practice, narrow-band filters that are rapidly tunable over a wide range
with constant bandwidth are difficult to build. It is easier to build a single, fixed-
frequency filter with adjustable bandwidth and sweep the signal through the filter.
Figure 1.17 shows how this can be done. The incoming signal is applied to a
mixer, along with a swept-frequency signal generated by a local oscillator in the
analyzer. Mixers will be described in the next chapter; for now, it is sufficient to
know that a mixer will produce outputs at the sum and the difference of the two
frequencies that are applied to its input.
The filter can be arranged to be at either the sum or the difference frequency
(usually the difference). As the oscillator frequency varies, the part of the spec-
trum that passes through the filter changes. The oscillator is a voltage-controlled
oscillator (VCO) whose frequency is controlled by a sawtooth wave generator
that also provides the horizontal sweep signal for the CRT.
The filter output is an ac signal that must be rectified and amplified before it is
applied to the vertical deflection plates of the CRT. If the amplification is linear, the

Attenuator IF IF Envelope Vertical

Mixer Filter Amp Detector Amp
Input

Bandwidth Reference
Control Level Control

Span Control
VCO Horizontal
Center Amp
Frequency Sawtooth CRT
Control Timebase
Generator

Figure 1.17 Simplified block diagram for a swept-frequency spectrum analyzer

01_639091_Delmar_Blake 8/16/01 2:51 PM Page 32

32 Chapter 1 ● Introduction to Communication Systems

vertical position of the trace will be proportional to the signal voltage amplitude at
a given frequency. It is more common, however, to use logarithmic amplification,
so that the display can be calibrated in decibels, with a reference level in decibels
referenced to one milliwatt (dBm).

Using a Spectrum Analyzer

A typical swept-frequency spectrum analyzer has four main controls, which are
shown on the block diagram of Figure 1.17. Their functions are described in the
following paragraphs.
The frequency control sets either the center frequency of the sweep or the fre-
quency at which it begins, called the start frequency (this is the frequency
corresponding to the left edge of the display). Some analyzers provide a switch so
that the display can show either the start or the center frequency.
The span control adjusts the range of frequencies on the display. There are two
methods of specifying the span. Sometimes it is calibrated in kilohertz or megahertz
per division. Other models give the span as the width of the whole display, which is
normally ten divisions. Dividing this number by ten will give the span per division.
The frequency of a displayed signal can be found by adjusting the frequency
control (also called tuning control) until the signal is in the center of the display
(assuming the start/center control, if present, is at center). The frequency of a sig-
nal not at the center can be found by counting divisions and multiplying by the
span per division.
The bandwidth control adjusts the filter bandwidth at the 3 dB points. Since
the smaller the bandwidth, the better the potential resolution of the analyzer, it
might seem logical to build the filter with as narrow a bandwidth as possible and
use it that way at all times. However, sweeping a signal rapidly through a high-Q
circuit (such as a narrow bandpass filter) gives rise to transients, which will reduce
the accuracy of amplitude readings. If the frequency span being swept is small, a
narrow bandwidth is necessary so that good resolution is obtained, but when a
wide span is in use, narrow bandwidth is unnecessary. This is fortunate, for the
sweep will naturally be faster, in terms of hertz per second, for the wider span. For
wide spans, then, a wide bandwidth is appropriate, with narrow bandwidth being
reserved for narrow span. Very often, the bandwidth and span controls are locked
together, so that the correct filter bandwidth for a given span will be selected au-
tomatically. Provision is made for unlocking the span and bandwidth controls on
the rare occasions when this is desirable.
3 dB It is important to realize that as viewed on a spectrum analyzer, a sine wave
Reference Bandwidth
Level of IF Filter
or a component of a more complex signal does not always appear as a single ver-
tical line. If the span is sufficiently narrow that the filter bandwidth is a significant
10 dB / portion of a division, the signal will appear to have a finite width. Such a display
division is shown in Figure 1.18. What is seen in the display is the width and shape of the
filter passband, not that of the signal.
The fourth major adjustment is the reference level control. Spectrum analyz-
Noise ers are set up so that amplitudes are measured with respect to a reference level at
Level the top of the display (not the bottom, as one might assume). This level is set by
the reference level control, which is normally calibrated in dBm.
Span Center Frequency There will also be an adjustable attenuator at the input to reduce strong sig-
nals to a level that can be handled by the input mixer. The attenuator control is
Figure 1.18 Spectrum analyzer normally coupled with the reference level setting, so that changing the attenuation
display automatically alters the displayed reference level.
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 33

Section 1.5 ● Spectrum Analysis 33

–20 40

–30 30

–40 20

–50 10

–60 0

–70 –10

–80 –20

–90 –30
10 mV 100 mV 1 V 10 V
10 µV 100 µV 1 mV 10 mV
Voltage (rms)

Figure 1.19 Conversion chart, dBm to voltage (for 50 ) (Courtesy Hewlett

Packard Co.)

A switch that sets the vertical scale factor is often closely associated with
the reference level control. This switch typically has three settings: 10 dB/division,
1 dB/division, and linear. In the first position, one division vertically represents a
10 dB power difference at the input. The second position gives an expanded scale,
with 1 dB per division. With either of these settings, the power of a signal can be
measured in either of two ways: by setting the signal to the top of the display with
the reference level controls and then reading the reference level from the controls,
or by counting divisions down from the top of the screen to the signal, multiply-
ing by 1 or 10 dB according to the switch setting, and subtracting the result from
the reference level setting.
The linear position is a little more complicated. The reference level is at the
top of the screen, as before. However, this time the screen is a linear voltage scale.
Zero voltage is at the bottom, and the voltage corresponding to the reference level
is at the top. This can easily be calculated from the power in dBm and the spec-
trum analyzer impedance, or you can use the graph in Figure 1.19. Please note,
however, that this graph is valid only if the analyzer has an input impedance of
50 , the most common value. Some analyzers allow the reference level to be set
directly in volts.
Once the voltage that corresponds to the reference level is known, the signal
voltage can be found. If the reference level has been adjusted so that the signal is
just at the top of the screen, then the signal voltage is that corresponding to the
reference level. Otherwise the signal voltage can be found by using proportion.
The signal voltage will be:

Vref Usig
Vsig  (1.34)
Ndiv

where Vsig = signal voltage in RMS volts

Vref = RMS voltage corresponding to the reference level
Usig = position of the signal in divisions up from the bottom of the screen
Ndiv = number of divisions vertically on the display (usually 8)
The linear scale is not used very often with spectrum analyzers, but it can be
handy on occasion.
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 34

34 Chapter 1 ● Introduction to Communication Systems

EXAMPLE 1.13 For each of the spectrum-analyzer displays in Figure 1.20, find the
frequency, power level (in dBm and watts), and voltage level of the
signals. The input impedance of the analyzer is 50 .

(a) Reference level 10dBm (b) Reference level 10 dBm (c) Reference level 20 dBm
Vertical: 10 dB/division Vertical: 1 dB/division Vertical: linear
Center frequency 110 MHz Center frequency 7.5 MHz Center frequency 543 MHz
Span: 10 kHz/division Span: 100 kHz/division Span: 1 MHz/division

Figure 1.20

Solution
(a) The signal is in the center of the display horizontally, so its frequency is
displayed on the analyzer’s frequency readout, assuming that it is set to
read the center frequency. In this example it is, so the signal frequency is
110 MHz. The signal peak is two divisions below the reference level of
10 dBm, with 10 dB/division, so its level is 30 dBm. The equivalent
power can be found from:
P
P (dBm)  10 log
1 mW
dBm
P (mW)  antilog
10
30
 antilog
10
 1  103 mW
 1 W
The voltage can be found from the graph in Figure 1.19, or it can be found
more accurately from:
V2
P
R
V  1PR
 11 V  50 
 7.07 mV

(b) The signal is one division to the left of center, with 100 kHz/division. The
frequency is 100 kHz less than the reference frequency of 7.5 MHz, so
f  7.5 MHz  0.1 MHz  7.4 MHz
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 35

Section 1.5 ● Spectrum Analysis 35

With regard to the amplitude, note that the scale is 1 dB/division and the
signal is one division below the reference level, so the signal has a power
level of
P (dBm)  10 dBm  1 dB  9 dBm
This can be converted to watts and volts in the same way as in part (a):
dBm
P (mW)  antilog
10
9
 antilog
10
 7.94 mW
V  1PR
 17.94 mW  50 
 630 mV
(c) The signal is three divisions to the right of the center reference frequency
of 543 MHz, with 1 MHz/division. Therefore, the frequency is:
f  543 MHz  3  1 MHz
 546 MHz
It is easier to find the signal voltage before finding the power, because the
amplitude scale here is linear. To arrive at the signal voltage, we need the volt-
age that corresponds to the power level at the top of the screen. Either from the
graph or analytically, we can easily find that the reference level of 20 dBm
corresponds to a power of 10 W, which represents 22.4 mV. The level on the
display is 68 of this, because the scale is linear with 8 divisions representing the
reference level voltage. Therefore the signal level is:
6
V  22.4 mV   16.8 mV
8
Now it is easy to find the power. In watts, it is:
V2
P
R
(16.8 mV)2

50 
 5.64 W
Now we find the power in dBm:
P
P (dBm)  10 log
1 mW
5.64 W
 10 log
1 mW
 22.5 dBm

Why are swept-frequency spectrum analyzers preferred over real-time spectrum SECTION 1.5 REVIEW
analyzers for radio-frequency use? QUESTION
●
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 36

USING Purpose:
To view a SIMULINK
• A sine block (A cosine is implemented by using a
phase of π/2 radians. Use a sample time of 0.1 ms =
MATLAB implementation of a full-wave 0.0001 sec).
rectified sine wave using a • Various gain blocks for the multipliers (e.g., –4/3π)
Fourier series approach. • A sum block.
In addition, the simulation results are displayed using
Background:
the scope block
MATLAB and its associated program SIMULINK can
be used to model and implement various system behav-
Implementation:
iors. In this section, SIMULINK will be used to imple-
Start MATLAB and then type simulink at the com-
ment the following equation, which represents the first
mand prompt to start the SIMULINK program.
five terms of the Fourier series of a full-wave rectified
sine wave. >> simulink
2 4cos(2t) 4cos(4t) Select the created model by using the File-> Open menu
v(t)    option to select and open the file "full_wave.mdl".
 (1)(3) (3)(5)
Run the simulation by selecting the Simulation-> Start
4cos(6t) 4cos(8t)
  menu option in the model window.
(5)(7) (7)(9)
• Double-click on the scope to maximize it and view
where, with a frequency of 60 Hz (in this implementa- all six waveforms. (Double-clicking on a block
tion): allows one to view the various parameters and
change them where permitted).
  2f  (2)(60)
Modify some of the block parameters to implement a
The full-wave rectified sine wave equation indicates the 120-Hz full-wave rectified signal using three Fourier se-
need for the following SIMULINK blocks: ries terms.
• Constant block with a value 2/π.

36
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 37

atellite communication is a relatively new form of radio. The first

S artificial satellite, Sputnik I, was launched by the Soviet Union in
1957, but practical, reliable communication via satellite really began
ELECTRONICS
… FAST FORWARD
with the launch of Intelsat I in 1965. This satellite was still partly ex-
perimental; the first modern commercial geostationary satellite, Anik Communications: The
A-1, was launched by Canada in 1972. Future
Modern communication systems have benefited from the invention of the
transistor in 1948 and the integrated circuit in 1958. Modern equipment makes
extensive use of both analog and digital solid-state technology, often in the
same piece of equipment. The current cellular telephone system, for instance,
uses analog FM radio coupled with an elaborate digital control system that
would be completely impractical without microprocessors in the phones them-
selves and in the fixed stations in the network.
The coming years are likely to bring more exciting developments. Optical
fiber is already a very important communication medium, and its use is bound
to increase. The trend toward digital communication, even for such “analog”
sources as voice and video, will continue.
It is impossible to predict accurately what will be developed in the future.
Both cellular telephone and facsimile machines are improvements on systems
that had been in use for many years and were regarded as not particularly ex-
citing, yet their use has grown very quickly in the last several years. About the
only thing that is certain about the future of communication technology is that
it will be interesting.

Here are the main points to remember from this chapter: SUMMARY
1. Any communication system has three essential elements: the transmitter, the
receiver, and the communication channel.
2. The design of the transmitter and receiver must take into account the charac-
teristics of the channel, particularly noise, distortion, and limited bandwidth.
3. Modulation is necessary with many types of communication channels. In
modulation some characteristic of a carrier waveform is changed in accor-
dance with the amplitude of a lower-frequency signal known as the baseband,
information signal, or modulating signal.
4. The carrier characteristics that can be modulated are amplitude, frequency, and
phase.
5. Time-division and frequency-division multiplexing are two systems for shar-
ing a channel among several information signals.
6. For a given modulation scheme, the amount of information that can be trans-
mitted is proportional to the time taken and the channel bandwidth employed.
7. Signal transmission is said to be distortionless if the information signal at the
receiver output is identical to that at the transmitter input, except for time
delay and change in amplitude. Any other change is known as distortion.

37
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 38

38 Chapter 1 ● Introduction to Communication Systems

8. Signals can be represented in either the time or frequency domain. Fourier

series can be used to find the frequency-domain representation of a periodic
signal. Signals can be observed in the frequency domain using a spectrum
analyzer.
9. Noise is present in all communication systems and has a degrading effect on
the signal. There are many types of noise, but the most common is thermal
noise, which is a characteristic of all electronic equipment.

IMPORTANT EQUATIONS

Note: For handy reference, each chapter will contain a IN  12qI0B (1.14)
brief list of the most commonly used equations. These
equations must not be used blindly: you must understand VNt  1V 2N1  V 2N2  V 2N3     (1.15)
how and where to use them. Refer to the text when nec-
essary. These are not the only equations you will need to S/N (dB)  10 log (PS/PN) (1.17)
solve the problems in each chapter.
S/N (dB)  20 log (VS/VN) (1.18)
e(t)  Ec sin (ct  ) (1.1)
(S/N)i
  f (1.3) NF  (1.19)
(S/N)o
f(t)  A0 /2  A1 cos t  B1 sin t  A2 cos
NF (dB)  (S/N)i (dB)  (S/N)o (dB) (1.20)
2t  B2 sin 2t  A3 cos 3t  B3 sin (1.5)
3t    
Teq  290(NF  1) (1.27)
f  1/T (1.6)
NF2  1 NF3  1
  2f (1.7) NFT  NF1  
A1 A1A2
(1.33)
NF4  1
PN  kTB (1.10)     
A1A2A3
VN  14kTBR (1.13)

GLOSSARY
baseband The band of frequencies occupied by an information modulation The process by which some characteristic of a car-
signal before it modulates the carrier. rier is varied by an information signal.
carrier A signal that can be modulated by an information multiplexing The transmission of more than one information
signal. signal over a single channel.
channel A path for the transmission of signals. noise Any undesired disturbance that is superimposed on a sig-
distortion Any undesirable change in an information signal. nal and obscures its information content.
Fourier series A way of representing periodic functions as a noise figure Ratio of the input and output signal-to-noise ratios
series of sinusoids. for a device.
frequency domain A representation of a signal’s power or am- noise temperature Equivalent temperature of a passive system
plitude as a function of frequency. having the same noise-power output as a given system.
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 39

Questions 39

receiver Device to extract the information signal from the sig- time domain Representation of a signal’s amplitude as a func-
nal propagating along a channel. tion of time.
signal-to-noise ratio (S/N) Ratio of signal to noise power at a transmission Transfer of an information signal from one loca-
given point in a system. tion to another.
spectrum analyzer Device for displaying signals in the fre- transmitter Device that converts an information signal into a
quency domain. form suitable for propagation along a channel.

QUESTIONS
1. Identify the band of each of the following frequencies. 8. Describe the way in which each of the following types of
(a) 10 MHz (used for standard time and frequency broad- noise is generated:
casts) (a) equipment (e) shot
(b) 2.45 GHz (used for microwave ovens) (b) atmospheric (f) partition
(c) 100 kHz (used for the LORAN navigation system for (c) space (g) excess
ships and aircraft) (d) thermal (h) transit-time
(d) 4 GHz (used for satellite television) 9. What is the difference between white and pink noise?
(e) 880 MHz (used for cellular telephones) 10. What is meant by the noise power bandwidth of a system?
2. Suppose that a voice frequency of 400 Hz is transmitted on 11. Why is the noise power bandwidth greater than the half-
an AM radio station operating at 1020 kHz. Which of these power bandwidth for a typical bandpass filter?
frequencies is
12. Why are amplifiers that must work with very weak signals
(a) the information frequency? sometimes cooled to extremely low temperatures? What is
(b) carrier frequency? this process called?
(c) the baseband frequency? 13. What is meant by the signal-to-noise ratio? Why is it impor-
(d) the modulating frequency? tant in communication systems?
3. Which parameters of a sine-wave carrier can be modulated? 14. Give two ratios that are similar to the signal-to-noise ratio
and are easier to measure.
4. Explain briefly the concept of frequency-division multiplex-
ing and give one practical example of its use. 15. Explain how noise figure is related to signal-to-noise ratio.
16. What is the relationship between noise figure and noise
5. Explain briefly the concept of time-division multiplexing
temperature?
and give one practical example of its use.
17. Explain why the first stage in an amplifier is the most impor-
6. Describe what is meant by each of the following types of
tant in determining the noise figure for the entire amplifier.
distortion:
18. Name the two basic types of spectrum analyzer, and briefly
(a) harmonic distortion
describe how each works.
(b) nonlinear frequency response
19. Name the four most important controls on a swept-
(c) nonlinear phase response frequency spectrum analyzer, and describe the function of
7. (a) What is the theoretical bandwidth of a 1 kHz square each.
wave? 20. What happens to the noise level displayed on a spectrum an-
(b) Suggest a bandwidth that would give reasonably good alyzer as its bandwidth is increased? Why?
results in a practical transmission system.
(c) Suppose that a 1 kHz square wave is transmitted
through a channel that cannot pass frequencies above
2 kHz. What would the shape of the wave be after
passing through the channel?
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 40

40 Chapter 1 ● Introduction to Communication Systems

PROBLEMS
Section 1.2 26. Sketch the spectrum for the pulse train shown in Figure 1.23.
21. Visible light consists of electromagnetic radiation with
wavelengths between 400 and 700 nanometers (nm).
Express this range in terms of frequency.
22. Equation (1.3) applies to any kind of wave. The velocity of
4
sound waves in air is about 344 m/s. Calculate the wave-

v (V)
length of a sound wave with a frequency of 1 kHz.
2
Section 1.3
23. Sketch the spectrum for the half-wave rectified signal in
Figure 1.21, showing harmonics up to the fifth. Show the
voltage and frequency scales, and indicate whether your 0 10 20 30 40 50 60
voltage scale shows peak or RMS voltage.
t (ns)

Figure 1.23
v (V)

17 V
27. Sketch the spectrum for the sawtooth waveform in Figure
1.24. Explain why this waveform has no dc component,
unlike the sawtooth waveform in Example 1.3.
8.335 16.67

t (ms)

Figure 1.21

24. Sketch the frequency spectrum for the triangle wave shown 3
in Figure 1.22, for harmonics up to the fifth. Show the volt-
v (V)

age and frequency scales.

0
1 2 3 4 5 6

−3

3
v (V)

t (ms)
−4 4 8 12
Figure 1.24

Section 1.4
t (µS) 28. A 50  resistor operates at room temperature (21°C). How
much noise power does it provide to a matched load over the
Figure 1.22 bandwidth of
(a) a CB radio channel (10 kHz)?
25. A 1 kHz square wave passes through each of three commu- (b) a TV channel (6 MHz)?
nication channels whose bandwidths are given below. Express your answers in both watts and dBm.
Sketch the output in the time domain for each case.
29. What would be the noise voltage generated for each of the
(a) 0 to 10 kHz conditions in Problem 28?
(b) 2 kHz to 4 kHz 30. Calculate the noise current for a diode with a bias current of
(c) 0 to 4 kHz 15 mA, observed over a 25 kHz bandwidth.
01_639091_Delmar_Blake 8/16/01 2:51 PM Page 41

Problems 41

31. The signal-to-noise ratio is 30 dB at the input to an amplifier Section 1.5

and 27.3 dB at the output. 36. The display in Figure 1.25 shows the fundamental of a
(a) What is the noise figure of the amplifier? sine-wave oscillator and its second harmonic, as seen on a
(b) What is its noise temperature? spectrum analyzer. The analyzer has an input impedance of
50 . Find:
32. (a) The signal voltage at the input of an amplifier is
100 V, and the noise voltage is 2 V. What is the (a) the oscillator frequency
signal-to-noise ratio in decibels? (b) the amplitude of the signal, in dBm and in volts
(b) The signal-to-noise ratio at the output of the same (c) the amount in decibels by which the second harmonic is
amplifier is 30 dB. What is the noise figure of the less than the fundamental (usually expressed as “x dB
amplifier? down”)
33. (a) A receiver has a noise temperature of 100 K. What is its
noise figure in decibels?
(b) A competing company has a receiver with a noise tem-
perature of 90 K. Assuming its other specifications are
equal, is this receiver better or worse than the one in part
(a)? Explain your answer.
34. Suppose the noise power at the input to a receiver is 1 nW in
the bandwidth of interest. What would be the required signal
power for a signal-to-noise ratio of 25 dB?
35. A three-stage amplifier has the following power gains
and noise figures (as ratios, not in decibels) for each
stage:

Power Noise Figure 1.25

Reference level 10 dBm
Stage Gain Figure
Vertical: 10 dB/division
1 10 3 Center frequency 100 MHz
Span: 20 MHz/division
2 20 4
3 30 5

Calculate the total gain and noise figure, and convert both to 37. Find the frequency, power (both in dBm and in watts), and
decibels. voltage for each of the signals shown in Figure 1.26.

(a) Reference level 30 dBm (b) Reference level 18 dBm (c) Reference level 12 dBm
Vertical: 10 dB/division Vertical: 1 dB/division Vertical: linear
Center frequency 872 MHz Center frequency 79 MHz Center frequency 270 MHz
Span: 10 MHz/division Span: 500 kHz/division Span: 5 kHz/division

Figure 1.26
01_639091_Delmar_Blake 8/16/01 2:52 PM Page 42

42 Chapter 1 ● Introduction to Communication Systems

Systems 41. The frequency response of an amplifier can be measured by

38. The four stages of an amplifier have gains and noise figures applying white noise to the input and observing the output
(in decibels) as follows: on a spectrum analyzer. When this was done for a certain
amplifier, the display shown in Figure 1.27 was obtained.
Find the upper and lower cutoff frequencies (at 3 dB down)
Power Noise
and the bandwidth of the amplifier.
Stage Gain Figure

1 12 dB 2 dB
2 15 dB 4 dB
3 20 dB 6 dB
4 17 dB 7 dB

Calculate the overall noise figure in decibels.

39. A three-stage amplifier is to have an overall noise tempera-
ture no greater than 70 K. The overall gain of the amplifier
is to be at least 45 dB. The amplifier is to be built by adding
a low-noise first stage to an existing two-stage amplifier that
has the gains and noise figures shown below. (The stage
Figure 1.27
numbers refer to locations in the new amplifier.)
Reference level 4 dBm
Vertical: 1 dB/division
Power Noise Center frequency 210 MHz
Stage Gain Figure Span: 10 MHz/division

2 20 dB 3 dB 42. A receiver has a noise bandwidth of 200 kHz. A resistor is

3 15 dB 6 dB connected across its input. What is the noise power at the
input if the resistor temperature is 20 degrees Celsius? Give
(a) What is the minimum gain (in decibels) that the first your answer in dBf.
stage can have?
43. An amplifier has a noise figure of 3 dB. If the signal to noise
(b) Using the gain you calculated in part (a), calculate the ratio at its input is 30 dB, what is the signal to noise ratio at
maximum noise figure (in decibels) that the first stage its output, in dB?
can have.
44. An amplifier consists of two stages. Stage one has a gain of
(c) Suppose the gain of the first stage could be increased by
12 dB and a noise figure of 2 dB. Stage two has a gain of
3 dB without affecting its noise figure. What would be
20 dB and a noise figure of 5 dB. Calculate the noise figure,
the effect on the noise temperature of the complete
in decibels, and the equivalent noise temperature, in kelvins,
amplifier?
for the amplifier.
40. The (S  N ) /N ratio at the output of an amplifier can be
45. What is the effect on the signal to noise ratio of a system, in
measured by measuring the output voltage with and without
dB, of doubling its bandwidth, other things being equal?
an input signal. When this is done for a certain amplifier, it
Consider thermal noise only.
is found that the output is 2 V with the input signal switched
on and 15 mV with it switched off. Calculate the (S  N ) /N 46. What is the effect on the signal to noise ratio of a system, in
ratio in decibels. dB, of doubling the transmitter power?

ANSWERS TO SECTION REVIEW QUESTIONS

Section 1.2 Section 1.4

Modulation is the use of the baseband signal amplitude to Noise introduced in the first stage will be amplified by all later
change some parameter of the transmitted signal. It is necessary stages.
to allow information to be transmitted at higher frequency and
for many information channels to be multiplexed. The basic Section 1.5
methods of modulation involve changing the amplitude, fre- Swept-frequency spectrum analyzers can cover a much wider
quency or phase of the transmitted signal. frequency range than can real-time spectrum analyzers.
Section 1.3
39 kHz