CHAPTER 4

COAXIAL CABLE SYSTEMS

A coaxial cable is simply a transmission line consisting of an unbalanced pair made

up of an inner conductor surrounded by a grounded outer conductor, which is held
in a concentric configuration by a dielectric. The dielectric can be of many different
types, such as solid “Poly” (Polyethylene or polyvinyl chloride), foam, spirally, air,
or gas. In the case of air/gas dielectric, the center conductor is kept in place by
spacers or disks.
Systems have been designed to use coaxial cable as a transmission medium with a
capability of transmitting a frequency division multiplex configuration ranging from
120 voice channels to 10, 800. Community antenna television (CATV) systems used
single cables for transmitted bandwidths in the order of 300 M Hz.
Frequency division multiplex was developed originally as a means to increase the
voice channel capacity of wire systems. At a later date the same techniques were
applied to radio. Then for a time, the 20 years after World War II, radio systems
became the primary means for transmitting long haul, toll telephone traffic. Lately,
coaxial cable has been making a strong comeback in this area.
One advantage of coaxial cable systems is to reduce noise accumulation when
compared to radio links. For point-to-point multichannel telephony the FDM line

Communication Systems I, First Edition. 73

By Osama A. Alkishriwo Copyright c 2020 John Wiley & Sons, Inc.
74 COAXIAL CABLE SYSTEMS

frequency configuration can be applied directly to the cable without further modula-
tion steps as required in radio links. Thus substantially reducing system noise.
In most cases radio links will prove more economical than coaxial cable. Coaxial
cable should be considered instead of radio links using the following general guide-
lines.

In areas of heavy microwave (including radio links) RFI.

On high density routes where it may be more economical than radio links.
(Think here of a system that will require 5000 circuits at the end of 10 years).

On long national or international backbone routes where the system designer is

concerned with noise accumulation.

Coaxial cable systems may be attractive for the transmission of television or other
video applications. Some activity has been noted in the joint use of TV and FDM
telephone channels on the same conductors. Another advantage in some circum-
stances is that system maintenance costs may prove to be less than for equal capacity
radio links.
Nowadays, the long distance coaxial cable is replaced by fiber links. One deter-
rent to the implementation of coaxial cable systems or fiber communication systems
as with any cable installation, is the problem of getting right-of-way for installa-
tion, and its subsequent maintenance (gaining access), particularly in urban areas.
Another consideration is the possibility of damage to the cable once it is installed.
Construction crews may unintentionally dig up or cut the cable.

4.1 Basic Construction Design of Coaxial

Each coaxial line is called a “tube”. A pair of these tubes is required for full duplex,
long haul application. One exception is the CCITT small bore coaxial cable system
where 120 voice channels, both “go” and “return”, are accommodated in one tube.
For long-haul systems more than one tube is included in a sheath. In the same sheath
filler pairs of quads are included, sometimes placed in the interstices, depending on
the size and lay-up of the cable. The pairs and quads are used for order wire and
control purposes as well as for local communication. Some typical cable lay-ups
are shown in Fig. 4.1. Coaxial cable is usually placed at a depth of 90/120 cm,
depending on frost penetration, along the right-of-way. Tractor drawn trenchers or
plows normally are to open the ditch where the cable is placed, using fully automated
procedures.
Cable repeaters are spaced uniformly along the route. Secondary or “dependent”
repeaters are often buried. Primary power feeding of main repeaters are installed
in surface housing. Cable lengths are factory-cut so that the splice occurs right at
repeater locations.
 COAXIAL CABLE CHARACTERISTICS 75

Figure 4.1 Some basic coaxial lay-ups.

4.2 Coaxial Cable Characteristics

For long-haul transmission, standard cable sizes are as follows:

inch (in) millimeter (mm)

0.047/0.174 1.2/4.4 small diameter
0.104/0.375 2.6/9.5
The fractions express the outside diameter of the inner conductor over the inside
diameter of the outer conductor. For instance, for the large bore cable, the outside
diameter of the inner conductor is 0.104 inch and the inside diameter of the outer
conductor is 0.375 inch. This is shown in Fig. 4.2. As can be seen from Eq. (4.1),
the ratio of the diameters of the inner and outer conductors has an important bearing
on attenuation. If we can achieve a ratio of b/a = 3.6, a minimum attenuation per
unit length will result. For air dielectric cable pair  = 1.0, outside diameter of inner
conductor = 2a, inside diameter of the outer conductor = 2b. Attenuation constant
α dB/km is given as
√
2.12 × 10−4 f a1 + 1b


α= (4.1)
1.6 log(b/a)
where a is the radius of the inner conductor and b is the radius of the outer conductor
(a and b are in cm). The characteristic impedances (Zo ) can be written as follows
   
138 b b
Zo = √ log = 138 log in air (4.2)
 a a
The characteristics impedance of coaxial cable is Zo = 138 log(b/a) for an air
dielectric. If b/a = 3.6, then Zo = 77 Ω. Using dielectric other than air reduces the
characteristic impedance. If we use the disks mentioned above to support the center
conductor, the impedance lowers to 75 Ω.
76 COAXIAL CABLE SYSTEMS

Figure 4.2 Basic electrical characteristics of coaxial cable.

Figure 4.3 is a curve giving attenuation per unit length in decibels versus fre-
quency for the two most common types of coaxial cable discussed in this chapter. At-
tenuation increases rapidly as a function of frequency and is a function of the square
root of frequency as shown in Fig. 4.3. The transmission system engineer is basically
interested in how much bandwidth is available to transmit FDM line frequency con-
figuration. For instance, the 0.375 in cable has an attenuation of about 5.8 dB/mi
at 2.5 M Hz and the 0.174 in cable 12.8 dB/mi. At 5 M Hz the 0.174 in cable has
about 19 dB/mi and the 0.375 in cable 10 dB/mi. Attenuation is specified for the
highest frequency of interest.

Figure 4.3 Attenuation frequency response per kilometer of coaxial cable.

Coaxial cable can transmit signals down to DC, but in practice, frequencies below
60 KHz are not used because of difficulties of equalization and shielding. Some
engineers lift the lower limit to 312 KHz. The high frequency limit of the system
is a function of the type and spacing of repeaters as well as cable dimensions and
the dielectric constant of the insulating material. It will be appreciated from Fig. 4.3
that the gain frequency characteristics of the cable follows a root frequency law, and
equalization and “preemphasis” should be designed accordingly.
 SYSTEM DESIGN OF COAXIAL 77

4.3 System Design of Coaxial

Figure 4.4 is a simplified application diagram of a coaxial cable system in long-

haul, point-to-point multichannel telephone service. To summarize system operation,
an FDM line frequency is applied to the coaxial cable system via a line terminal
unit. Dependent repeaters are spaced uniformly along the length of the cable system.
These repeaters are fed power from the cable itself. In the ITT design, the dependent
repeater has a plug-in automatic level control unit. In temperate zones where cable
laying is sufficient and where diurnal and seasonal temperature variations are within
the “normal” (a seasonal swing of ±10o C) a plug in level control (regulating) unit
is incorporated in every fourth dependent amplifier (See in Fig. 4.5). We use the
word “dependent” for the dependent repeater (DA in Fig. 4.5) for two reasons. It
depends on a terminal or main repeater for power and it provides to the terminal or
main repeater fault information.

Figure 4.4 Simplified application diagram of a long-haul coaxial cable system for
multichannel telephony.

Let us examine Fig. 4.4 and 4.5 at length. Assume we are dealing with a nominal
12 M Hz system on a 0.375 in (9.5 mm) cable. Up to 2700 voice channels can be
transmitted. To accomplish this, two tubes are required, one in each direction. Most
lay-ups, as shown in Fig. 4.1, have more than two tubes. Consider Fig. 4.4 from
left to right. Voice channels in a four-wire configuration connect with the multiplex
equipment in both “go” and “return” directions. The output of the multiplex equip-
ment is the line frequency (baseband) to be fed to the cable. The line signal is fed to
the terminal repeater, which performs the following functions:

Combines the line control pilots with the multiplex line frequency.

Provides “preemphasis” to the transmitted signal, distorting the output signal

such that the higher frequencies get more gain than the lower frequencies, such
as shown in Fig. 4.3.

Equalizes the incoming wide band signal.

78 COAXIAL CABLE SYSTEMS

Feeds power to dependent repeaters.

Figure 4.5 Detail of application diagram. DA dependent amplifier (repeater), Reg.

regulation circuitry, d distance between repeaters.

The output of the terminal repeater is a preemphasized signal with required pilots
along with power feed. In the ITT design this is a dc voltage up to 650 V with
a stabilized current of 110 mA. A main (terminal) repeater feeds, in this design,
up to 50 dependent repeaters in each direction. Thus a maximum of 30 dependent
repeaters appear in a chain for every main or terminal repeater. Other functions of
a main repeater are to equalize the wide band signal and to provide access for drop
and insert of telephone channels by means of through group filters.
Figure 4.5 is a blow up of a section of Fig. 4.4 showing each fourth repeater with
its automatic level regulation circuitry. Distance d between DA repeaters (dependent
amplifier) is 4.5 km or 2.8 mi for a nominal 12 M Hz system (0.375 in cable).
Amplifiers have gain adjustments of ±16 dB, equivalent to varying repeater spacing
±570 m (1870 f t).
As can be seen from the above, the design of coaxial cable systems for both
long-haul multichannel telephone service as well as CATV systems has become, to
a degree, a “cookbook” design. Basically, system design involves the following:
Repeater spacing as a function of cable type and bandwidth.
Regulation of signal level.
Temperature effects on regulation.
 SYSTEM NOISE CALCULATIONS 79

Equalization.

Cable impedance irregularities.

Fault location or the so-called supervision.

Power feed.

Other factors are, of course, right-of-way for the cable route with access for main-
tenance and the laying of the cable. With these factors in mind, consult Table 4.1
which review the basic parameters of the CCITT approach. For the 0.375 in coaxial
cable systems practical noise accumulation is less than 1 pwp/km, whereas radio
links allocate 3 pwp/km. These are good guideline numbers to remember for gross
system considerations. Noise in coaxial cable systems derives from the active de-
vices in the line (e.g. the repeaters) as well as the terminal equipment, both line
conditioning and multiplex. Noise design of these devices is a trade-off between
thermal and intermodulation (IM) noise. IM noise is the principal limiting param-
eter forcing the designer to install more repeaters per unit length with less gain per
repeater.

Table 4.1 Characteristics of CCITT Specified coaxial cable systems (larger diameter cable).
Item Nominal Top Modulation Frequency
2.6 M Hz 4 M Hz 6 M Hz 12 M Hz 60 M Hz
CCITT rec. G.337A G.338 G.337B G.332 G.333
Repeater type Tube Tube Tube Transistor Transistor
Video capability No Yes Yes Yes Not stated
Video + FDM capability No No No Yes Not stated
Nominal repeater spacing 6 mi/9 km 6 mi/9 km 6 mi/9 km 3 mi/4.5 km 1 mi/1.55 km
Main line reg. pilot 2604 kHz 4092 kHz − 12435 kHz 12435/4287 kHz
Auxiliary reg. pilots − 308.6 kHz − 4287.3 kHz 61.16 kHz and 40.92 kHz

4.4 System Noise Calculations

The gain, G, of a device is often expressed in decibels, as the ratio of the output to
input voltages or powers, i.e., GdB = 20 log(Vo /Vi ) or GdB = 10 log 10(Po /Pi ).
Being a ratio gain is not related to any particular power level. There is a logarithmic
unit of power, however, which defines power in dB above a specified reference level.
If the reference power is 1 mW the units are denoted by dBm (dB with respect to a
1 mW reference) and if the reference level is 1 W the units are denoted by dBW .
80 COAXIAL CABLE SYSTEMS

4.4.1 Noise Temperature

A convenient way to specify noise power is via an equivalent thermal noise temper-
ature, Te , given by
N
Te = (K) (4.3)
kB
The noise temperature of a subsystem (say an amplifier) is not the temperature of the
room it is in nor even the temperature inside its case. It is the (hypothetical) temper-
ature which an ideal resistor matched to the input of the subsystem would need to
be at in order to account for the extra available noise observed at the device’s output
over and above that which is due to the actual input noise. This idea is illustrated
schematically in Fig. 4.6. The total available output noise of a device is therefore
given as

N = (kTs B + kTe B)G (4.4)

where the first term in the brackets is the contribution from the input (or source) noise
and the second term is the contribution from the subsystem itself. G is the (power)
gain (expressed as a ratio) of the subsystem which can be greater or less than 1.0.
Ideal Practical
noiseless amplifier noisy amplifier

Figure 4.6 Equivalent noise temperature (Te ) of an amplifier.

4.4.2 Noise Factor or Noise Figure

Another measure of the excess internal noise generated by two-port is the integrated
noise figure f that is defined as the actual output noise power divided by the output
noise power if the two-port were noiseless assuming that the noise source at the input
is at room temperature T0 or 17o C). The noise figure is a power ratio written as

k(T0 + Te )BG Te
f= =1+ (4.5)
kT0 BG T0
Also, the noise factor, f , of an amplifier is defined as the ratio of SNR at the
system input to SNR at the system output when the input noise corresponds to a
temperature of T0 , i.e.
 SYSTEM NOISE CALCULATIONS 81

(S/N )i
f= (4.6)
(S/N )o Ni =kT0 B

where
 
S signal power at input, Si
= (4.7)
N i kT0 B
and
 
S signal power at output, So
= (4.8)
N o k(T0 + Te )BG
The specification of the input noise temperature as T0 allows all devices and sys-
tems to be compared fairly, on the basis of their quoted noise factor. f is therefore a
figure of merit for comparing the noise performance of different devices and systems.
Thus,
Si /(kT0 B) Te
f= =1+ (4.9)
Si G/[Gk(T0 + Te )B] T0
The noise factor of any device can be converted to an equivalent noise temperature
using

Te = T0 (f − 1) (4.10)

4.4.3 Noise Temperature for a Transmission Line

It is important to remember that the “gain” of the individual subsystems can be
greater or less than 1.0. In the latter case the subsystem is lossy. (A transmission
line, for instance, with 10% power loss would have a gain of 0.9 or −0.46 dB.) The
equivalent (input) noise temperature, Te,l of a lossy device at a physical temperature
T0 can be found from its “gain” using
T0 (1 − Gl )
Te,l = (4.11)
Gl
(The subscript l here reminds us that a lossy device is being considered and that
the gain Gl is therefore less than 1.0.) Equation (4.11) is easily verified by consider-
ing a transmission line terminated in matched loads as shown in Fig. 4.7. Provided
the loads, Rs and RL , and the transmission line are in thermal equilibrium there can
be no net flow of noise power across the terminals at either end of the transmission
line. The power available to the transmission line from Rs is

Ns = kT0 B (4.12)

If the transmission line has (power) gain, Gl ( < 1.0), then the power available to
RL from Rs is
82 COAXIAL CABLE SYSTEMS

Figure 4.7 Source, transmission line, and load in thermal equilibrium.

0
Ns = kT0 BGl (4.13)

The transmission line itself must supply the balance of power to RL i.e.

Nline = kT0 B − kT0 BGl = kTe BGl (4.14)

Dividing Eq. (4.14) by kB gives

T0 (1 − Gl )
Te = (4.15)
Gl
The noise figure of lossy devices (such as transmission lines or passive mixers) is
related to their “gain”, Gl, by

1
f= (4.16)
Gl

where f and Gl are expressed as ratios, or alternatively by

F = −Gl (dB) (4.17)

4.4.4 Noise Temperature of Cascaded Subsystems

The total system noise temperature, at the output of a device or subsystem, Tsyst out ,
can be found by dividing Eq. (4.4) by kB, i.e.

Tsyst out = (Ts + Te )G (4.18)

If several subsystems are cascaded, as shown in Fig. 4.8, the noise temperatures
at the output of each subsystem are given by

Tout 1 = (Ts + Te1 )G1

Tout 2 = (Tout 1 + Te2 )G2
Tout 3 = (Tout 2 + Te3 )G3
 SYSTEM NOISE CALCULATIONS 83

Figure 4.8 Partitioning of system into cascaded amplifier subsystems.

The noise temperature at the output of subsystem 3 is therefore

Tout 3 = {[(Ts + Te1 )G1 + Te2 ]G2 + Te3 }G3

= Ts G1 G2 G3 + Te1 G1 G2 G3 + Te2 G2 G3 + Te3 G3 (4.19)

This temperature can be referred to the input of subsystem 1 by dividing Eq. (4.19)
by the total gain, G1 G2 G3 , i.e.
Te2 Te3
Tsyst in = Ts + Te1 + + (4.20)
G1 G1 G2
or

Tsyst in = Ts + Te (4.21)

where Te is the equivalent (input) noise temperature of the system excluding the
source and Tsyst in is the overall input noise temperature including the source. The
total noise temperature output of a set of cascaded subsystems is then simply given
by

Tsyst out = Tsyst in G1 G2 G3 (4.22)

Comparing Eqs. (4.20) and (4.21) the equivalent noise temperature of the cas-
caded subsystems is
Te2 Te3
Te = Te1 + + (K) (4.23)
G1 G1 G2
This can be rewritten in terms of noise factors as
(f2 − 1)T0 (f3 − 1)T0
(f − 1)T0 = (f1 − 1)T0 + + (4.24)
G1 G1 G2
Dividing by T0 and adding 1,
(f2 − 1) (f3 − 1)
f = f1 + + (4.25)
G1 G1 G2
This is called the Friis noise formula.
84 COAXIAL CABLE SYSTEMS

4.5 Coaxial Repeater Design

4.5.1 General

Consider a coaxial cable system 100 km long using 0.375 in cable of transmitting
up to 2700 V F channels in an FDM/SSB configuration (12 M Hz). At 12 M Hz
cable attenuation per kilometer is approximately 8.3 dB/km (from Fig. 4.3). The
total loss at 12 M Hz for the 100 km cable is 8.3 × 100 = 830 dB. Thus one
approach the system design engineer take would be to install a 830 dB amplifier at
the front end of the 100 km section. This approach is rejected out of hand. Another
approach would be to install a 415 dB amplifier at the front end and another at the
50 km point. Suppose the signal level was −15 dBm composite at the originating
end. This −15 dBm + 415 dB = +400 dBm or +370 dBW . Remember that
+60 dBW is equivalent to a megawatt; otherwise we would have an amplitude with
an output of 1037 W or 1031 M W . Still another approach is to have 10 amplifiers
with 83 dB gain, each spaced at 10/km intervals. Another would be 20 amplifiers
or 830/20 = 41.5 dB each; or 30 amplifiers at 830/30 = 27.67 dB, each spaced
at 3.33 km intervals. As we shall see later, the latter approach begins to reach an
optimum from a noise standpoint keeping in mind that the upper limit for noise
accumulation is 3 pW p/km. The gain most usually encountered in coaxial cable
amplifiers is 30 − 35 dB.
If we remain with the 3 pW p/km criterion, in nearly all cases radio links will be
installed because of their economic advantage. Assuming 10 full-duplex RF channels
per radio system at 1800 V F channels per RF channel, the radio link can transmit
18, 000 ful-duplex channels, and do it probably more cheaply on an installed cost
basis. On the other hand, if we can show noise accumulation less on coaxial cable
systems, these systems will prove in at some number of channels less than 18, 000
if the reduced cumulative noise is included as an economic factor. There are other
considerations such as maintenance and reliability, but let us discuss noise further.
Suppose we design our coaxial cable systems for no more than 1 pW p/km. Most
long-haul coaxial cable systems being installed today meet this figure. However, we
will use the CCITT figure of 3 pW p/km in some of the examples that follow.
Noise for this discussion consists of two major components, namely

Thermal noise (white noise)

Intermodulation (IM noise)

Coaxial cable amplifier design, to reach a goal of 1 pW p/km of noise accumula-

tion, must walk a “tightrope” between thermal and intermodulation noise. It is also
very sensitive to overload, with its consequent image on intermodulation noise.
 COAXIAL REPEATER DESIGN 85

4.5.2 Thermal Noise

Thermal noise threshold, Pn may be calculated for an active, two port, device as
follows:

Pn = 10 log(KT ) + N F + 10 log(Bw )
Pn = −174 dBm/Hz + N F + 10 log(Bw ) (4.26)

where, Bw : Bandwidth (Hz)

K: Boltzamn’s constant
N F dB: Noise figure of the amplifier
T : Room temperature = 290 K.

Restating the equation above for a voice channel with a nominal Bw of 3000 Hz,
we then have,

Pn = −139 dBm + N F (dBm/3kHz) (4.27)

Assume a coaxial cable system with identical repeaters, each with gain Gr spaced
at equal intervals along a uniform cable section. Here Gr exactly equals the loss of
the intervening cable between repeaters. The noise output of the first repeater is
Pn + Gr (in dBm ). For N repeaters in cascade, the total noise (thermal) output of
the N th repeater is

Pn + Gr + 10 log(N ) (dBm) (4.28)

An important assumption all along is that the input/output impedance of the re-
peaters just equals the cable impedance Z0 .

4.5.3 Overload and Margin

The exercise of this section is to develop an expression for system noise and discuss
methods of reducing it. In the previous section we developed a term for thermal
noise for a string of cascaded amplifiers (Pn + Gr + 10 log(N )). The next step
is to establish 0 dBm as a reference or more realistically −2.5 dBmp, because we
are dealing with a voice channel nominally 3 kHz wide and we want it weighed
psychometrically. Now we can establish a formula for a total thermal noise level as
measure at the end of a coaxial cable system with N amplifiers in cascade.

Pt = Pn + Gr + 10 log(N ) − 2.5 (dBmp) (4.29)

As before we assume that all the amplifiers are identical and spaced at equal
intervals and that the gain of each is Gr which is exactly equal to the loss of the
intervening cable between each amplifier.
86 COAXIAL CABLE SYSTEMS

Examining the equation above, we see that the operation level is high. The next
step is to establish an operating level which should never be exceeded and call it
L. A margin to that level must also be established to take into account instability
of the amplifiers caused by aging effects, poor maintenance, temperature variations,
misalignment, and so forth. The margin to the maximum operating level point is
Mg . All units are in decibels. A more realistic equation can now be written for total
thermal noise including a suitable margin.

Pt = Pn + Gr + 10 log(N ) − 2.5 (dBmp) + L + Mg (4.30)

These levels are shown graphically in Fig. 4.9. A number of interesting rela-
tionships can be developed if we consider a hypothetical example. CCIIT permits
3 pW p/km noise accumulation (CCITI Rec. G222). Allow 2 pW p/km of that fig-
ure to be attributed to thermal noise. If we were to build a system 100 km long, we
could then accumulate 200 pW p of thermal noise. Now set 200 pW p equal to Pt , in
the previous equation. Hence, convert 200 pW p to dBmp (−67 dBmp). Thus

Figure 4.9 Graphic representation of reference level, signa/levels, and noise levels in a
coaxial cable system. (Note Levels are not drawn to scale).

−67 dBmp = Pn + Gr + 10 log(N ) − 2.5 dBmp + L + Mg

Pn + Gr + 10 log(N ) + L + Mg = −64.5 dBmp

Let us assign some numbers to the equations that are somewhat reasonable. To
the 100 km system install 20 repeaters at equal intervals. Cable loss is 5 dB/km,
or 500 dB total loss at the highest operating frequency. Thus repeater gain, Gr is
25 dB, with N = 20. Let,
 COAXIAL REPEATER DESIGN 87

L + Mg = 15 dB

From Eq. (4.27), we have

Pn = −139 + N F

Thus,

−139 + N F + 25 + 10 log(20) + 15 = −64.5 dBmp

N F = 21 dB or less

This is a noise figure that is fairly easy to meet.

Let us examine this exercise a little more closely and see if we cannot derive some
important relationship that can offer the system and amplifier design engineer some
useful guidance.

1. By doubling the length of the system, system noise increases 3 dB, or by dou-
bling the number of amplifiers, Gr being held constant, system noise doubles
(i.e., 10 log(2N )).

2. By making the terms L and Mg smaller, or in other words, increasing the max-
imum operating level, reducing the margin, system thermal noise improves on
a decibel for decibel basis.

3. Of course by reducing N F , system noise may also be reduced. But suppose

N F turned out to be very small in the calculations, a figure that could not be
met or would imply excessive expense. Then we would have to turn to other
terms in the equation, such as reducing terms Gr , L, and Mg . However, there is
little room to maneuver with the latter two, 15 dB in the example. That leaves
us with Gr . Of course, reducing Gr is at the cost of increasing the number of
amplifiers (or increasing the size of the cable to reduce attenuation, etc.). As we
reduce Gr the term 10 log(N ) increases because we are increasing the number
of amplifiers, N . The trade-off between the term 10 log(N ) and Gr occurs
where Gr is between 8 and 9 dB.

Another interesting relationship is the attenuation of the cable. It will be noticed

that the loss in the cable is approximately inversely proportional to cable diameter.
As an example, let us assume that the loss of a cable section between repeaters is
40 dB. By increasing the cable diameter 25%, the loss of the cable section becomes
40/(1+0.25) = 32 dB. In our example above, by increasing cable diameter, repeater
gain may be decreased with the consequent improvement in system noise (thermal).
(note The examples given above are given as exercises and may not necessarily be
practicable owing to economic constraints).
88 COAXIAL CABLE SYSTEMS

4.5.4 Intermodulation Noise

The second type of noise to be considered in coaxial cable system and repeater design
is intermodulation (IM) noise. IM noise on a multichannel FDM system may be
approximated by a Gaussian distribution and consists of second, third, and higher
order intermodulation products. Included in these products, in the wide band systems
we cover here, are second and third harmonics. IM products (e.g. intermodulation
noise) are a function of the nonlinearity of active devices.
To follow our argument on intermodulation noise in coaxial cable repeater design,
the reader is asked to accept the following. If a simple sinusoid wave is introduced
at the input port of a cable amplifier, the output of the amplifier could be expressed
by an equation with three terms, the first of which is linear representing the desired
amplification. The second and third terms are quadratic and cubic representing the
non-linear behavior of the amplifier (i.e. second and third order products). On the
basis of this power series, for each 1 dB change of fundamental input to the amplifier,
the second harmonic changes 2 dB, and the third harmonic 3 dB. Furthermore, for
two waves A and B, a second order sum (A + B) or difference (A − B) is equivalent
to the second harmonic of A at the output plus 6 dB. Likewise, the sum of A+B +C
would be equivalent to the level of the third harmonic level at the output plus 15.6 dB
of one of the waves. We consider that all inputs are of equal level. The situation for
2A + B would be equivalent to 3A + 9.6 dB, and so forth. These last three power
series may be more clearly expressed when set down as follows, where PH is the
harmonic intermodulation power,
PH(A±B) = PH(2A) + 6 dB
PH(A±B±C) = PH(3A) + 15.6 dB (4.31)
PH(2A±B) = PH(3A) + 9.6 dB
Repeater Design:
For this discussion let IM2 and IM3 express the nonlinearity of a repeater; they are,
respectively, the power of the second and third harmonics corresponding to a 0 dBm
fundamental (−2.5 dBmp). Adjusted to the maximum operating level L (see Fig.
4.9), the second harmonic power, P2A , is

P2A = IM2 + 10 log(N ) + L (dBmp) (4.32)

L is assumed to be positive as in our argument in the preceding section. Now
decrease the applied signal level to a repeater by L dB, assuming the power of the
fundamental of a wave at the 0 TLP (test level point) was L dBmp. It follows
that, by decreasing the applied power L dB so that the fundamental of a wave is
now 0 dBmp at the 0 test level point (TLP), the magnitude of the fundamental is
decreased L dB at the input of the first amplifier. For every decrease of 1 dB in the
fundamental. Therefore the new power of the second harmonic amplitude will be
decreased by 2L dB, or
P2A = IM2 + 10 log(N ) + L − 2L dBmp
 COAXIAL REPEATER DESIGN 89

for the system or the second harmonic noise power level.

P2A = IM2 + 10 log(N ) − L (dBmp) (4.33)

Let us consider the A + B product. For fundamentals of equal magnitudes, such

a product in a single repeater will be 6 dB higher than a 2A product. It also varies
by 2 dB per 1 dB variation in both fundamentals and adds in power addition as a
function of the number of repeaters. Thus,

P(A+B) = IM2 + 10 log(N ) − L + 6 (dBmp) (4.34)

Similarly for the 3A condition (e.g. third harmonic) of a wave fundamental, A,

P(3A) = IM3 + 10 log(N ) − 2L (dBmp) (4.35)

For the A + B − C condition,

P(A+B−c) = IM3 + 20 log(N ) − 2L + 15.6 dBmp (4.36)

Again we find “2L” because third order products vary 3 dB for every 1 dB change
in the fundamental. We use 20 log rather than 10 log assuming that the products
add in phase (i.e. voltage-wise) versus the number of repeaters. The 10 log term
represented power addition. For the 2A − B condition,

P(2A−B) = IM3 − 2L + 20 log(N ) + 9.6 dBmp (4.37)

4.5.5 Total Noise and Its Allocation

In summary there are three noise components to be considered

Thermal noise

Second order IM noise

Third order IM noise

Let us consider them two at a time. If a system is thermal and second order IM
noise limited, minimum noise is achieved allowing an equal contribution. For the
3 pW p/km case, we would assign 1.5 pW to each component.
For thermal and third order IM noise limited systems, twice the contribution is
assigned to thermal noise as to third order IM noise. Again for the 3 pW p case,
2 pW p is assigned to thermal noise and 1 pW p to third order IM noise.
The parameter L is established such that these divisions can be met by adjusting
repeater spacing and repeater design. As an example in practice, Fig. 4.10 shows
noise allocation of the North American L4 system.
90 COAXIAL CABLE SYSTEMS

Figure 4.10 Allocation of noise in a practical system. Accumulated noise over 4000 mi of
North American L4 system.

4.6 Coaxial Equalization

4.6.1 Introduction

Consider the result of transmitting a signal down a 12 M Hz coaxial cable system

with the amplitude-frequency response shown in Fig. 4.3. The noise per voice chan-
nel would vary from an extremely low level for the channels assigned to the very
lowest frequency segments of the line frequency (baseband) to extremely high lev-
els for those channels that were assigned to the spectrum near 12 M Hz. For long
systems there would be every reason to believe that these higher frequencies would
be unusable if nothing was done to correct the cable to make the amplitude response
more uniform as a function of frequency. Ideally we would wish it to be linear.
Equalization of cable deals with the means used to assure that the signal-to-noise
ratio in each FDM telephone channel is essentially the same no matter what its as-
signment in the spectrum. In the following discussion we consider both fixed and
adjustable equalizers.

4.6.2 Fixed Equalizers

The coaxial cable transmission system design engineer has three types of cable
equalization available which fall into the category of fixed equalizers. These are
as follows

Basic equalizers

Line build-out (LBO) networks

Design deviation equalizers

 COAXIAL EQUALIZATION 91

The basic equalizer is incorporated in every cable repeater. It is designed to com-

pensate for the variation of loss frequency characteristic of uniform cable sections.
This is done by simply making the fixed gain proportional to the square root of fre-
quency, matching loss for the nominal length. For the North American L4 system
the gain characteristic is shown in Fig. 4.11. For the case of 12 M Hz cable, the
section nominal length would be 4.5 km (CCITI Rec. G.332).

Figure 4.11 Repeater gain characteristic for North American L4 system.

The key word in the preceding paragraph is uniform. Unfortunately some cables
sections are not uniform in length. It is not economically feasible to build tailor-
made repeaters for each nonuniform section. This is what line build-out networks
are used for. Such devices are another class of fixed equalizer for specific variations
of nominal repeater spacing. One way of handling such variations is to have available
LBO equalizers for 5%, 10%, 15% etc., of the nominal distance.
The third type of equalizer compensates for design deviation of the nominal char-
acteristics standard for dependent repeaters and actual loss characteristics of the ca-
ble system in which the repeaters are to be installed. Such variation is systematic
such that the third level to equalization, the design deviation equalizer, is installed
one for each 10, 15, or 20 repeaters to compensate for gross design deviation over
that group of repeaters.

4.6.3 Variable Equalizers

Figure 4.12(a) shows the change of loss of cable as a function of temperature vari-
ation and Fig. 4.12(b) shows approximately each temperature variation with time.
Adjustable equalizers are basically concerned with gain frequency variations with
time. Besides temperature, variations due to aging of components may also be a
problem; however, this is much less true with transistorized equipment. Cable loss
per kilometer shown in Fig. 4.3 is the loss at a mean temperature. The 3/8 in cable
92 COAXIAL CABLE SYSTEMS

used in L system applications has a variation of ±20o F/mi at 20 M Hz or about

±0.6% dB/mile at 60 M Hz.

(a)

(b)

Figure 4.12 (a) Change in loss of 1 mi of 3/8 in coaxial cable for ±20o F change in
temperature. (b) Approximate earth temperature variation with time at 4 f t depth.

The primary purpose of automatic regulation is to compensate for the gain varia-
tion due to temperature changes. Such automatic regulation usually is controlled by
a pilot tone at the highest cable frequency. For instance, in the ITT cable design for
12 M Hz, the pilot controlled system will always apply exact compensation at the
pilot frequency of 12, 435 kHz, an error may occur at other frequencies. On a single
amplifier this error is very small but will add systematically along the route. Such
error is usually corrected using manually adjustable equalizers.
 COAXIAL LEVEL AND PILOT TONES 93

4.7 Coaxial Level and Pilot Tones

Intrasystem levels are fixed by cable and repeater system design. These are the L
and Mg established in section 4.4.3. Modern 12 M Hz systems display an overload
point of +24 dBm or more. Remember that the
Overload point= equivalent peak power level + relative sending level + margin
The margin is Mg as in section 4.4.3 and L may be related to relative sending
level. Mg can be reduced, depending on how well system regulation is maintained.
Typically pilot levels are −10 dBm0. The level is a compromise, bearing in
mind system loading, to minimize the pilots contribution to intermodulation noise
in a system that is multichannel in the frequency domain (FDM). Another factor
tending to force the system designer to increase pilot level is the signal-to-noise
ratio of the pilot tone required to effectively actuate level regulating circuitry. Pilot
level adjustment is the injection point usually requires a stability better than 0.1 dB.
Internal pilot stability should display a stability improved over desired cable system
level stability. If system level stability is to be ±1 dB, then internal pilot stability
should be better than ±0.1 dB.
The number of system pilots assigned and their frequencies depend on bandwidth
and the specific system design. Commonly 12, 435 kHz is used for regulation and
13.5 M Hz for supervisory. In the same system and auxiliary pilot is offered at
308 kHz and, as an option, a frequency comparison pilot at 300 kHz.
The only continuous in-band pilot in the L4 system is located at 11, 648 kHz.
Supervisory pilot tones are transmitted in the band 18.50 − 18.56 M Hz. An L
multiplex synchronizing pilot is located at 512 kHz.

4.8 Coaxial Supervisory

The term supervisory in coaxial cable system terminology refers to a method of

remotely monitoring repeater condition at some manned location. As mentioned
above, the L4 system uses 16 pilot tones brought upon command giving status of 16
separate, buried repeaters.

4.9 Powering the Coaxial System

Power feeding of buried repeaters in the ITT system permits the operation of 15
dependent repeaters from each end of a feed point (12 M Hz cable). Thus up to
30 dependent repeaters can be supplied power between power feed points. A power
feed unit at the power feed point provides up to 650 V DC voltage between center
conductor and ground using 110 mA stabilized direct current. Power feed points
may be as far apart as 140 km (87 mi) on large diameter cable.