Professional Documents
Culture Documents
Long-Haul Communication
Long-Haul Communication
Martin P. Clark
Copyright 1991, 1997 John Wiley & Sons Ltd
ISBNs: 0-471-97346-7 (Hardback); 0-470-84158-3 (Electronic)
3
Long-haul
.
Communication
None of the circuits that we have so far discussed are suitable as they stand for long haul
communication. To get us anywhere with long haulwe need to address ourselves to the following
inescapably pertinent topics:
0
multiplexing (how to increase the number of circuits that may be obtained from
one physical cable)
In thischapter we discusspredominantlyhowtheselineissues
affect analoguetransmission
systems and how they may be countered. The effects on digital transmission are discussed in
later chapters.
30
LONG-HAUL
Signal attenuation occurs in simple wireline systems, in radio and in optical fibre
systems. The effect of attenuation on a line follows the function shown in Figure
3.1,
where the signal amplitude (the technical term for signal strength) can be seen to fade
with distance travelled according to a negative exponential function, the rate of decay
of the exponential function along the length of the line being governed by the attenuation constant, alpha ( a ) .
Complex mathematics, which we will not go into here, reveal that the value of the
attenuationconstantforanyparticular
signal frequency onan electrical wire line
transmission system is given by the following formula:
cli
The larger the value of (, the greater the attenuation, the exact value depending on the
following line characteristics:
R : the electrical resistance per kilometre of the line, in ohms
G: the electrical leakance per kilometre of the line, in mhos
L: the inductance per kilometre of the line, in henries
C: the capacitance per kilometre of the line, in farads
f : the frequency of the particular component of the signal.
1\
signol omplitude= Ta
Dosronced
31
LINE LOADING
CR/G henrieslkm
This, in effect, reduces the electrical properties of the line to a simple resistance, minimizing both the attenuation and the distortionsimultaneously. In practice, the attenuation and distortion of a line can be reduced artificially by increasing its inductance
L ideally in a continuous manner along the lines length. The technique is called line
loading. It can be achieved by winding iron tape or someothermagneticmaterial
directly around the conductor, but it is cheaper and easier to provide a lumped loading
coil at intervals (say 1-2 km) along theline. The attenuationcharacteristics of unloaded
and loaded lines are shown in Figure 3.2.
Unloaded line
-.
Continuously
loaded line
Speech
band
Signalfrequency
32
LONG-HAUL
The use of line loading has the disadvantage that it acts as a high frequency filter,
tending to suppress the high frequencies. It is therefore important when lump loading is
used, tomakesurethatthewanted
speech band frequencies suffer only minimal
attenuation. If the steep part of lump loaded line curve (marked by an asterisk in
Figure 3.2) were to occur in the middle of the speech band, then the higher frequencies
intheconversationwouldbedisproportionatelyattenuated,resultingin
heavy and
unacceptable distortion of the signal at the receiving end.
3.3 AMPLIFICATION
Although lineloading reduces speech-band attenuation, there is still a loss of signal
which accumulates with distance, and at some stage it becomes necessary to boost the
signal strength. This is done by the use of an electrical amplifier. The reader may well
ask what precise distance line loading is good for. There is, alas, no simple answer as it
depends on the gauge of the wire and on the transmission bandwidth required by the
user. In general, the higher the bandwidth, the shorter the length limit of loaded lines
(10-15 km is a practical limit).
Devices called repeaters are spaced equally along the length of a long transmission
line, radio system or other transmission medium. Repeaters consist of amplifiers and
other equipment, the purpose of which is to boost the basic signal strength.
Normallya repeater comprisestwo amplifiers, oneforeachdirection
of signal
transmission. A splitting device is also required to separate transmit and receive signals.
This is so that each signal can befed to a relevant transmit or receive amplifier, as
Figure 3.3 illustrates.The splitting device is called a hybrid or hybridtransformer.
Essentially it converts a two-way communication over two
wires into two one-way,
two-wire connections, and it is then usually referred to as a four-wire communication
(one direction of signal transmission on each pair of a two-pair set). So, while a single
two-wire line is adequate for two-way telephone communication over a short distance,
as soon as the distance is great enough to require amplification then a conversion
to
Repeater
r
--- - _ _ _ _ - - I
- - -l
Amplifier
33
AMPLIFICATION
four-wire communication is called for. Figure 3.3 shows a line between two telephones
with a simple telephonerepeater in the line. Eachrepeaterconsists
of twohybrid
transformers and two amplifiers (one for each signal direction).
The amplijication introduced at eachrepeaterhas
to be carefully controlledto
overcomethe effects of attenuation,without adversely affecting what is called the
stability of the circuit, and without interfering with other circuits in the same cable.
Repeaters too far apart orwith too little amplification would allow the signal current to
fade to such an extent as to be subsumed in the electrical noise present on the line.
Conversely, repeaters that are too close together or have too much amplification, can
lead to circuit instability, and to yet another problem known as crosstalk.
A circuit is said to be unstable when the signal that it is carrying is over-amplified,
causingfeedback and even moreamplification.Thisin
turn leads to even greater
feedback, and so on and on, until the signal is so strong that it reaches the maximum
power that the circuit can carry. The signal is now distorted beyond cure and all the
listener hears is a very loud singing noise. For the causes of this distressing situation let
us look at the simple circuit of Figure 3.4.
The diagram of Figure 3.4 shows a poorly engineered circuit
which is electrically
unstable. At first sight, the diagram is identical to Figure 3.3. The only difference is that
various signal attenuation values (indicated as negative) and amplification values
(indicated as positive) have been marked using the standard unit of measurement, the
decibel (dB). The problem is that the net gain around the loop is greater than the net
loss. Let us look more closely. The hybrid transformer H1 receives the incoming signal
from telephone Q and transmits it to telephone P, separating this signal from the one
that will be transmitted on the outgoing pair of wires towards Q. Both signals suffer a
3 dB attenuation during this line-splitting process. Adding the attenuation of 1 dB
which is suffered on the local access line by the outgoing signal coming from telephone
P, the total attenuation of the signal by the time it reaches the output of hybrid H1 is
therefore 4 dB.The signal is further attenuated by 5 dB as a result of line loss. Thus the
input to amplifier A1 is 9 dB below the strength of the original signal. Amplifier A1 is
set to more than make up forthis attenuation by boosting the signal by 13 dB, so that at
+ 13 dB
Line loss
Arnplifler
+ l 3 dB
34
LONG-HAUL
its output the signal is actually 4dB louder than it was at the outset. However, by the
time the second hybrid loss (in H2)and Qs access lineattenuation have been taken into
account, the signal is back to its original volume. This might suggest a happy ending,
but unfortunately the circuit is unstable. Its instability arises from the fact that neither
of thehybridtransformerscanactuallycarryouttheirline-splittingfunctionto
perfection; a certain amount of the signal received by hybrid H2 from the output of
amplifier A1 is finding its way back onto the return circuit (Q-to-P).
A well-designed and installed hybrid would give at least 30 dB separation of receive
and transmit channels. However, in our example, the hybrid has either been poorly
installed or become faulty, and the unwanted retransmittedsignal (originated by P but
returned by hybrid H2) is only 7 dB weaker at hybrid Hss point of output than it was at
the output from Al, and is then attenuated by 5 dB andamplified 13 dB before finding
its way back to hybrid H l , where it goes through another undesired retransmission,
albeit at a cost of 7 dB in signal strength. The strength of this signal, which has now
entirely lapped the four-wire sectionof the circuit, is 2 dBless than that of the original
signal emanating from telephone P. However, on its first lap it had a strength 4dB
lower than the original. In other words, the
re-circulated signal is actually louder than it
was on thefirst lap!What is more, if it goesaround again it will gain 2dB in strength for
each lap, quickly getting louder and louder and out
of control. This phenomenon is
called instability. The primary cause is the feedback path available across both hybrids,
which is allowing incoming signals to be retransmitted (or fed back) on their output.
The path results fromthenon-idealperformance
of thehybrid.Inpracticeit
is
impossible to exactly balance the hybrids resistance with
that of the end telephone
handset.
One way to correct circuit instability is to change the hybrids for more efficient (and
probablymore expensive)ones.Thissolutionrequirescarefulbalancing
of each
telephone handset and corresponding hybrid, and is not possible if the customer lines
and the hybrids are on opposite sides of the switch matrix (as would be the case for a
two-wire customer local line connected
via a local exchange to a four-wire trunk or
junction). A cheaper and quicker alternative to the
instabilityproblem is simply to
reduce the amplificationinthefeedbackloop.This
is done simply by adding an
attenuating device.Indeedmostvariableamplifiers
are in fact fixed gainamplifiers
(around 30 dB) followed immediately be variable attenuators or pads. For example, in
the case illustrated in Figure 3.4 a reduction in the gain
of both amplifiers A1 and A2 to
12 dB will mean that the feedback signal is exactly equal in amplitude to the original. In
this state the circuit may just be stable, but it is normal to design circuits with a much
greater margin of stability, typically
at least 10dB. For the Figure 3.4 example this
would restrict amplifier gain to no more than 7 dB. Under these conditions the volume
of the signalheardintelephone
Q will be 6dB quieterthanthattransmitted
by
telephone P, but this is unlikely to trouble the listener.
Crosstalk is the name given toanoverheard
signal onanadjacentcircuit.It
is
broughtabout by electromagneticinduction of an over-amplifiedsignalfrom
one
circuit onto its neighbour, and Figure3.5 gives a simplified diagram of how it happens.
Becausethe transit amplifier on the circuitfromtelephone
Ato telephone B in
Figure 3.5 is over-amplifying the signal, it is creating a strong electromagnetic field
around the circuit and the same signal is induced into the circuit from P to Q. As a
result the user of telephone Q annoyingly overhears the user of telephone A as well as
CIRCUITS
TWO- AND FOUR-WIRE
35
Repeaters
Figure 3.5
Crosstalk. Q hears A
Repeater 1
Repeater 3
2 -wire
2 -wire
I ine
I ine
LONG-HAUL
36
Repeater 1
Repeater 2
Repeater 3
L- wire
Line
L-wire
line
3.5
EQUALIZATION
We have mentioned the need for equalization on long haul circuits, to minimize the
signal distortion. Speech and data signals comprise a complex mixture of pure single
frequency components, each of which is affected differently by transmission lines. The
result of different attenuation of the various frequencies is tonal degradation of the
received signal; at worst, the entirehigh or low-frequency range couldbe lost. Figure 3.8
shows the relative amplitudes of individual
signal frequencies of a distorted and an
undistorted signal.
Amp1 i t u d e
(signal strength)
I-
-r-
listortedsignal
/
p
I
Speechband
D Signal frequency
Figure 3.8 Amplitude spectrum of distorted and undistorted signals (attenuation distortion)
Consists
of
37
Bandwidth
name
Channel
(1 telephone channel)
Group
Supergroup
Basic hypergroup
(also called a super
mastergroup)
Basic hypergroup
(alternative)
Mastergroup
Hypergroup
(12 MHz)
Hypergroup
(60 MHz)
of
24 telegraph
subchannels
120 Hz spacing
12 channels
5 groups
15 supergroups
(3 mastergroups)
4 kHz
48 kHz
240 kHz
3.7 MHz (3.6 MHz used)
(240 kHz per supergroup
with 8 kHz spacing
0-4 kHz
60-108 kHz
3 12-552 kHz
312-4082 kHz
12
60
900
(4 MHz line)
16 supergroups
60-4028 kHz
5 supergroups
9 mastergroups
1.2 MHz
12 MHz
2 700
36 mastergroups
60 MHz
10 800
960
300
38
LONG-HAUL
One. L wire
FDM line 1
( supergroup
12 individual
circuits
( & -wire 1
12 individual
circuits
El
+Transmit
CTE
ST E
12 individual
circuits
2 X 18 k H z
bandwith,
&-wirelines
Receive
DIVISION
FREQUENCY
39
(FDM)
channels. The same principle can be applied at other levels in the hierarchy. Thus, for
example, all the supergroups of a hypergroup could be broken down into their component groups using HTE and STE.Alternatively, some of the supergroups could be used
directly for bandwidth applications of 240 kHz.
Before multiplexing,the audio signals which areto bemultiplexed-up are first
converted to four-wire transmission, if they are not so already. The signals on each
transmit pair are then accurately
filtered so that stray signals outside the allocated
bandwidth are suppressed. In fact, a telephone channel is filtered to be only 3.1 kHz in
bandwidth (of the available 4 kHz). The remaining 0.9 kHz separation prevents speech
interference between adjacent channels. Groups are filtered to 48 kHz, supergroups to
240 kHz, etc. Each filtered signal is then modulated by a carrier frequency, which has
the effect offrequency shifting the original signal into another part of the frequency
spectrum. For example,atelephonechannelstarting
out in thebandwidthrange
300-3400 Hz might end up in the range 4600-7700 Hz. Another could be shifted to the
range 8300-11 400Hz, and so on.
Each component channel of an FDM group is frequency shifted by the CTE to a
different bandwidth slot within the 48 kHz available;so that in total 12 individual
channels may be carried. Likewise, in a supergroup, five already made-up groups of
48 kHz bandwidth are slotted in to the 240 kHz bandwidth by the STE.
The frequency shift is achieved by modulation of the component bandwidths with
different carrier signal frequencies. The frequency of the carrier signal which is used to
modulate the original signal (or baseband) will be equal to the value of the frequency
shift required. Each carrier signal must be produced by the translating equipment.
Themodulation of a signal inthefrequency
band 300-3400Hz using acarrier
frequency of 8000Hz produces a signal of bandwidth from 4600Hz (8000-3400) to
1 1400 Hz (8000 + 3400). The original frequency spectrum of300-3400 Hz is reproduced
in two mirror image forms,called sidebands. One sideband is in the range 4600-7700 Hz
and the other is in the range 8300-1 1 400 Hz. Both sidebands are shown in Figure3.10.
As all the information is duplicated in both sidebands, only one of the sidebands
needs to be transmitted. For economy in the electrical power needed to be transmitted
to line, it is normalfor FDM systems tooperate ina singlesideband ( S S B ) and
Amditude
frequency
Carrier
I
Originalsignal
spectrum [ baseband)
300 Hz
Lower
3600 HZ
6600 Hz
7700 Hz
8300 Hz
Baseband
sideband
sideband
llLO0 Hz
Signal
40
LONG-HAUL
suppressed carrier mode. The original signal is reconstructed at the receiving end by
modulating (mixing) a locally generated frequency.
(Note: single sideband operation
may also be used in radio systems, but the carrier
is not suppressed in this case because
it is often inconvenient to make a carrier signal generator available
at the receiver.
When the carrier is not suppressed a much simpler and cheaper detector can be used.)
Each baseband signal to be included in an FDM system is modulated with a different
carrier frequency, the lower sideband is extracted for conveyance. It may seem inefficient not to double up the
use of carrier frequencies, adopting alternating upper and
lower sidebands of different channels, but by always using the lower sidebandwe obtain
a better overall structure, allowing easier extraction
of singlechannelsfromhigher
order FDM systems. The carrier frequencies needed to produce a standard group are
thus 64 kHz, 68 kHz, 72 kHz, . . . ,108 kHz and the overall structure is as shown in
Figure 3.1 1.
...
1L
Audio
Channels
I
Baseband
0 - h kHZ
Channel
modulating
frequency
( lower
sideband used
6 6 , 6 8 , 7 2 , 7 6 , 8 0 , 8 h , 8 8 , 9 2 , 9 6 , 1 0 0 , 1 0 h , 108 kHz
1211
1 0 9 8 7
Basicgroup
structure
kHz
60
108 kHz
120 k H z
(when
frequency
12 kHz
Figure 3.11
h 5 6 7 8 9 1 0 1 1 1 2
kHz
60
41
Supergroupsandhypergroupsmay
be modulated in asimilarfashion,using
appropriate carrier frequencies and single sideband operation.