• - A STUDY ON TI N015E AND IijTtHFERtNC~ IN TtLECOMMU~ICATION . 'STEM ~ITH SPECIAL REfERENCE TO BANGLADESH (,,2r;;.8035L\'32- \98'0 MDE. BY ! MD.. £~IDADUR HAHMAN KHAN, B. Sc. ENGG. ( ELECT.) .i ~ .,' -.~ 1- , DEPATM.ENT Of ELECTRICAL ENGINEERING BINGLADESH UNI ERSIlY OF ENGIN£ERI~G AND TECHNOLOGY, DACCA. , JULY, 1.960 .1 I I I'
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
•-
A STUDY
ON TI N015E AND IijTtHFERtNC~ IN TtLECOMMU~ICATION.
'STEM ~ITH SPECIAL REfERENCE TO BANGLADESH
(,,2r;;.8035L\'32-\98'0MDE.
BY
!MD.. £~IDADUR HAHMAN KHAN, B. Sc. ENGG. ( ELECT.)
.i~.,'
-.~
1-,
DEPATM.ENT Of ELECTRICAL ENGINEERING
BINGLADESH UNI ERSIlY OF ENGIN£ERI~G AND TECHNOLOGY, DACCA.,
JULY, 1.960
.1I
I
I'
A STUDY
ON THE NOISE AND INTERfERENCE IN TELECOMMUNICATION
(e) Phasor diagram for carrier plus sinusoidalinterference.
(bl Line spectrum for the interfering sinusoid.Product demoduletion of r-f noise(a) r-f noise spectrum (b) Baseband noise spectrum.
figure No. Title
3.~O Phasor diagram for a modulated cerrier and IIIband limited noise.
3.12
3.13
5.2
5.35.4
5.56.1
6.2
6.3
-Detected interference amplitude as e function offi for an interfering wave at l'c + fi•
Detected interference amplitude for fM wi thde-emphasis fil taring.
Instantenaa.s frequency deviation versus time fortvo sinu soi de.
POst detection signal to noise ratio. es a functionof' z.for different types of modulation.
Set-up for the measurement 0 l' noi se figure.
Noise meeeurement by using noise loading method
(a) no bend stop fil tar between noise generatorand system.
(b) ~ band stop filter bstween noise generator andsystem,.
(a) Crosstalk by inductive and capecitive couplingbetween two circuits.
(b) Verious capacitances between conductors oftwo circui ts.
four conductors in a quad of a multipaired cable.
A simple network for arc suppression.
Set up for the measurement 0 f noi se in cabl e andswi tching system.
Noise judgement curves.
(a) Crosstalk due to improper filtering.
(b) A base band frequency allocation for 960 channels.
Typical set up for the measurement of noise in thelllUltiplexing end demultiplexing equip.ment.
Thermal, crosstalk and intermodulation noise in echannel.
Nonlinear voltage tranafer characteristic of 8two port.
figure No. Title
6.5 Low-order transmiss.ion gain and phase shapes.
6.6 System analog for second order modulation.
6.7 Echo in a transmitter antenna system.
6.8 Radio frequency interference.
6.9 Adjacent channel interference.
6.10 Spectral intensity for iigniJs-A.
6.11 set up for measurement of r-f system noisl' in theDacca-CTGMWlink by using noise loading method.
6.12 set up for measurement of r-fsystem noise in theDacca-Hsjiganj MWlink us.ingnoi se loading method.
'a) Variation of received signal level measuredat Hajiganj MWrepeater station from12-3-80 to 18-3-80.
(b) Typical AGeealibration curve.
Interruption period due to fading for three months(Dac-eTGMWlink).
Interruption due to fading for elifferent years •
. set up for measurement of MUX-DEMUX end r-fsystem combined noise in some channels inDaeea-CTGSiD circuit.
Typical NPRversus noise loading curve.
NPRversus noise loading curve for different orderof distor.Uon.
Table No.
5.2
5.3
•
LIST {If TABLES
Title
Comparison of different types of linear modulationsystems.
Relation between di fferent uni ts 1'01' noi sa power.
Noise values fur various interconnections between!SGlllle(a) Some subscribers of central exchange
(28 and 25 only).
(b) Some subscribers of central exchange and somesub scribers of Sher-e-Banglanagar exchenge.
(c) Some subscribers of Gulshen exchange.
(a) Resistance between conductors in two cables atold Dacca area.
(b) Resistance between conductor in two cablesetMiqlur area.
Different judgement qualities and correspondingnoise levels.
Phese modulation caused by small. low ordertransmission deviatio.ns and AM/PMconversion.
Results for 1'-1' system noise measurement inDacca-CTG MWlink.
Received signal levels at different timss atHajiganj MWrt!peeter stat;lon (from 12-3-80 to18-3-80). for simplicity. Datas are omitted butcorresponding g.raphs are shown in fig. 6.13( a).
(xv)
Interruption periods due to fading for three months(In Dac-CTGMWlink).
in the STD circuit
1980.1980.and eCI R
6.5 Interruptions in hours dus(In DAC-CTGMWline).
SNR for di fferent channel sbetween Dacca and eTG.(a) In the month of feb.(b) In the month of May.
Comparison between CC1TTreference circuit.
to fading for three yeers
hypothetical
I1:2 Noise objectives for MUX-DEMUXand 1'-1' system
according to eC!TTand eClR. !
•wf
Gew)
k
TBWH( f)
1.
m
dBmOJ:i
dBrnco
dBmdBrn
LI ST (IF SYMBOLS
Standard deviation
Charge of an electronAngular frequencyFrequency
Frequency spectrum of a power signel-23 0Bolt~an~constBnt = l.37xlO Joules/ k
Tamperature in oK
Bendwidth
Message channel bandwidthTransfer function
White noise power densityGain of a tlolOport
Effective input noise temperature oftwo-port networkNoise figure
Noise equivalent bandwidth
Reverse saturation current
Transmitted power
Transmitted power. PT; divided by total white noisein eudio bandwidth till).
Moduletion index
Power of a point in dBmreferred to zero level pointwith psophometric weighting
Noise power of e point in dBrnreferred to zero levelpoint with C-message weighting
POwer relative to one miliwatt (expressed in dB)
POwer relative to one picowatt (expressed in dB)
CHAPTER-lINTRODUCTION
1
1.0 Preliminaries
Tile ability to communicate. that is to share information,
is one of the char;3cteri stics of human race which has played a
major part in its development. New techniques are beiag. evolv,ing
every day to further enhance the communication. But in transmi-
tting and receiving of information. there is alweys some unwanted
signal which is called noise. This limits the .capacity of informs-
tion trensmission. and so it is en.important factor in designing
and evaluating performance of any telecommunication system. In.
this chapter an introduction is given about signal and noise in
telecommunication systems. A brief literature review. objective
end contents of this thesis are also given in this chapter.
1.1. Telecommunication
Today; telecommunication systems are found wherever infor-
mation is to be conveyed from one point to another - from manto
man or machine to machine. Now-a-days; telephone;" picture~phone.
radio and talevision have become intagral parts .of everyday life.
long distance circuits span the globe, carrying text; voic" and
images. Radar and telemetry systema play vital rales in navigation;v
defent and sci,entific research. Computer talk to computers via
tran~continental data links. Thus; there are myriad types of
communication~ An important one, among these; is the communication
between people by telephones.
2
-vi.2 Sign.l Ad Nelse An T,lecollllllunlcationSyat!t!li
The objective of • cOlllllluni,caUon syst_ie to carry in tor-
.Iltion from one point called &Duree to anothl!!t' point celled
destination. Inforntation is tranemitted end received as electrical
elgnala 1n teleea_unic.tion lI)lst.1II8. But information ,produced
by e source. in ,general,!. notelaetricel. So • tr.nsducer i.s
required. Transducer convert, the lnfO%1ltetion or 1Il6ssage to a
aigne1t • titus varying electrical qusntity such as voltage or
curr.nt, whlch is better lilUited for fu:rther proc •• d.ng by the
8Y.'.III. Si'.il.u:1Yt another transducer at the destination converts
the output signal to the appropriate massage foX'lll.
But in eou:r:se0 f ,electric.l Bignel tr.n •••:!. .aion, c1lrta1n
unw.tled and undeeirabJ.e .f1'.8Ct$ take place. One i. attenuat.ion
IIIhichr'llduces the eignal strength; llIO:t'f:I serious. however, are
di .torUo"" interference end 1\01_ which ch.nge tohe signal shape.
In • bJ:O.dsenae, .n~ un••anted .19"a1 perturbation lII.y b. C188ai-
tied '.8 noiae.
N01e. oc:eurs in te1ecolllmunicat.ion ayet.elIIs in ",arious way••
It may occu:r by net.workeend devices used in the collllllUnication
systeM itself, by extran&ou6electroaegnetic waves produced byother colllfoluniclSt1on eystelllS,t by atmospheric electrici.ty, by solar
,rad$..ation, by man-mad" source and by lIleny other ways. The tot.l
noiss that reaches " list'aner' '8 e.rs affects the degree of annoy••
ance and the 1ntelligibility of received speech. SiJdlarly, •••• in
th.• ca•• of television. the ultilllete .ff'.ect produced by noi.e is
against the eya and te1ns tluch as snowor ghos1:edescribe this
3
effect subj ectively. Howmuch deterioration 0 f the signal will
be caused by noise depends on signllland noise strengths. If
signal power is appreciably higher than noi se power, then effect
of noise will not be perceptible. Thus the criterion of performance
of I!Isystl!lm is taken as the ratio of signal power to noise power
or signsl to noise ratio (S.N.R). There is a lower limit of SNR
for intelligibility of a signsl. for example. in case of Tv.
transmission systems. SNRmuat be higher than 50 DBand for .audio
systems SNRshould be higher than 30 DB.Thus for better perfor-
mance. transmission systems must have good SNR.
It is clear that noise phenomen~in tranamission systems is
an i~portant and basic factor. for reasonable SNR.,the syst~m
must have speci fied no~se aa by using amplification SNRcannot
be increased. rather it is dllGreaaed as amplifiers add extra
noise. Thus noiae objectives lire given in any system design.
System performance detariorates when noise values exceed given
objectives. Of course. cases are there for whiGh SNRmey be kapt
within reasonable value ellen if noise increases ina parti,cular
section of the transmission system. This may bedoneby i,nael!Jsing
the signal level preceding the noisy section. But this may ,not
be preGticabla as the existing circuit may not carry, the increased
signal ,level. Thus study of noise phenomena is important in
telecommunication systems.
1.3 Transmission Media. Their Characteristics and Degradations
The main transmission media ev.ilable for telecommunications- -
are (1) copper conductors (2) co-axial cables, (3) h-f radio
4
(4) v-h-f, radio (5) u,.h•..f radio. (6) microwave (n other medill.
1.3.1 Copper Conductors
Copper conductors are used in telegraph poles or in twisted
pair cable. Recently, aluminiulll wires are also used in twisted
pair cable. Besides copper conductors. iron conductors are also
used in telegraph lines. These can be used upto several kilo Hert.
if edequate amplification and balancing are provided.
1.3.2 Co-axial Cable
Co-axial cable is usable from about 50 KHZto hundrads of
megaHertz and is used fOr transmission of baseband signals.
1.3.3 High Frequency (h-f) System
This .syetem( h-f) is uaed. for poin'!; to .point communiCation
around the world. The frequency range ia (3-30) mHz. Ey using
independent sideband modulation (ISS) four channels can be used•..t
by one cerrier. Eesidfis. un~di,rectionAcommunicatiol1_.ifl a.l,so. malilil
by h-f system. fOr example, short wave radio stations in the world.
Since these waves are reflected by ionospheres, fading is high
using these frequences.
1.3.4 Very High frequency 1v-h-fl SYstem
v-h-f radio operates in the range 150 to 180 MHz. It operates
mostly on the line 0 f sight principle. Maximum72 audio channels
5
can be used. It is also used for small scale point to point
communication and also in mobile radio operation.
1.3.5 Ultra High frequency (uih,,,) System
u-h-f system has increasing demand fOr mobile operation.
It is elsa used for transmission ,of baseband signal. fraquency
range is from 400 mHz to one gegaHertz.
1.1.6 Mic,50wave System (MW)
Microwave system is uaed in terrestrialcommunicetion as
well as in th a satelli te communication., frequency ranges fbr
telecommunicetions by microwave are 2;4.6,,7 and 11 GHz (GegaHertz).
till now,maximum 2700 audiochannels canbe u1;ledby a,UWllull: single
carrier, Antenna system gain can be made very high and noi se in
this system is also lower, for these reasons microweve systems
are used widely,
1,3.7 Other Media
Othar media like fibre optics and laserbeem are now being
uBed.,Th~se have much higher capacity. for example, more,than
230.; 000 telephone channels can be U sed by using fibre optics.
Of course these are new techniques and so have not yet been used
widely,
l,3.B Other Degradati.ons
In addition to the bandwidth limitations there are many other
channel degradations which are given below:
(a) attenuation
(b) frequ ency di eto rtion i. e change 0 f gain .r att enuation
with frequency
(c) amplitude distortion: change of attenuation or gain
with the emplitude of the signal
(d) phase or delay distortion: change of propagation delay
with frequency
(el change of phase with amplitude i.e AM/PMconversion
(f) noise: power, bt ••••• x.itbr bandwidth end statistie. of
circuit noiee end in the casaaf speech the background
room noi se.
(g) croBstalk from other channels
(h) side tones end echoes
(i) delay or propaglltion time •
.Jl.4 Brief Literature Review
Transmission of 1!Isignzl1 as well as the rate st which it
may be transmitted is characterized greatly by the noise present
in the system. The rate of transmission of intelligence (signal)
is elsa characterized by the time required. fo:l;' energy. change in
the systsm. But these facts werl!l not very clear till 1948 when
Claude Sh.nnon(l) gave a mathl!lml!ltical reletionship between system
capacity, signal strength and noise strength. System capacity
means the maximumamount of information that may be transmitted
in one second.
--' -
7
. Dut for quite e ~o.ng time the concept of information theory
was rather nebulous. Th. first ettempt to measure the amount of
information was lIIadeby H.~teley{ 2) in a paper celled "Tran_ission
of inform.tion". The work 0 f Hartley was greatly influenced by
Myquist-Kup1muller law d1scovered in 1924. The.law states -that
for transmitting telegraph signals .t a given rate a de'finit •
. frequency bllTldwidthis requized. The modern info mation theory
callie, in fect. dUe to the contribution by N_Wiener(3) and C.
Shannon. The communication theory owes 8 greet debt to professor
M. Wiener (a mathematician) fo~ lIlLlchof its basic philosophy and
mathematical formulation. The statistical approach to thecolll1llu-
nication theory is based on hi. contribution. The foundation of
the .mo,derninf01:lllation theorywBiI laid by Shannon's first paper •
•The lIIoth_sticel theory 0f comlllunication".
f'ollow1ng Wiener's end Shannon's worke considerable research
work have been done tj.l1 today. blportanceof t.he -study of noise
and int.erf,erence in telecomlllunicati.on systema was felt essential
efter Shannon's fO:l:lllulafor syst!!Jlll°capacity wa. derivet!,.An unueu-
ally large number of scientifi~ papers in this field appeared
ctvring the past. three decades. Thie undoubtedly indicates th.
great interest end enthusiasm of scientist and engineere in this'
fiel.d. Somecontributors from Bell-telephone laboretories are
S.O. Rice(4). D. Hufflnsn(51.B• KOlllil1an16)T_K. S8n(1). D.C_Hogg{B).
A. f'einste~ 9). J.A. Wozencraft( 10) snd D. Slepien (11). Someother
fallliliar names in the list of scientists end metht!lllsticians with
B
contribution in the field 0 f communication are Abranson( 12) •
GoldmanClJ). Schmitt(l4). Kapp(15). Juhani Alme(16). !?;';c~;eh(17).I-\"",-",,'k US) \-("''''' (19) (20) (21)N;i;c:tlwEl . • Hel. lIlIPSlln j Glave ; Gordon and many others.
were
In Bangladesh; previous studies in the(22) (23)made by Rouf and Das. Rouf made
field communication
a shti6ti~alstudy
of the telephone traffic ~nd.thecontit'ibutic:m._()f.Das""as.in the
study of information content of Bengali language and he also made
a study of the noiS8 inmicrowsve communication system. These
works were carried out in the Department 0 f Electrical Engineering.
Bangladesh Uni versi tYD f Engineering and Te.chnology (For .M.$c. Engg.
thesis).
1.5 .Objective of the Thesis
for effic:Len!o. and!,-"eliel:ll~oPe!'-"C11;ionof any .::l~mmunic:i31;ion
system; the noise sources involved need a careful end thorough
study. The main obj.ective of this thesis is to make e study.em
noise and interfe!,-"snc::e,in. tt?1,scC1mmu,:,;cliltion.sY$t~ll)with sPl!Ici!)l
reference to the telecommunication system in Bangladesh and to
find out any .possible means to .improve the Signal to noi se ratio
( SNR). Measurements for noi.ss l!!nd.interferen~f3 wil1b.e. mlilde~rl
the telecommunication system in Bangladesh and the results will
be compared with those given by ccnsultativecommittee for inter~
naticnal Telephone and Telegraph ~CCITT) lind consultative committee~-=---for international radiocommunication (CCIR).
Conclusion will be made on the basis of the results. Efforts
will be made to find out soms601utions to the problems that might
9
1.6 Contents of the Thnis
A~ introduc1:ion to the top:1cof the thes:1swith • ~ '-t-~c.-y,""...,...,\- ~-L-C.y,..t>-l~ "l-<'>Vi>I'>, ct •. "hl'"'''' "'1\(0< ' .)0.<.-
1~ezwtur8 review ha. been given in CbaRt.r-l.~.scription of\ .~~ -"N"";::JQ!;;:~~ 'f ~ '1- 'V '- ~V.<N L.- J4 \{-, •.••..
"Communication in the Presllnce ofNoise. ProbebiU.ty of £rro1" fortwo Encoding Schemes", Bell SystemTech • ..I•• Vol. 29, 1950."The Synthesie of Linear SequllntialCoding Networke", Prot. Symposiuman Information Theory, London, 1955."The Besic Theorems 131' InformationTheory". Ann. Math. Statistics.Vol. 24. 1952.
""aaking of Croa_talk by Speechand Nois.", Bell Syst8l1l Tech• .I.,Vol. 49. April. 1970 ••• Statistics on Attenuation ofMicrowaves by Intense Rain".Bell SYBtlllll r echo ,..I•• Vol. 48.Nov_ber. 1969.
"Foundations 0 f Intormetion Theory".McGrew Hill Book Company. Inc.,N_ YoX'1<. 1958.
"Sequential Decoding for ReliableCommunication" IRE Conv. Record.Vol. 5. March, 1957.
"A Clase of Binllry SignallingAlphabet.". Bell Sy.tam Tech•• J••Vol. 35. 1956.
8Info~etion Theory end Coding",McGraw Hill Book Company. Inc.,N.Y. 1963,
itA itudyof Bauble Sided Inter-8y~bo1 Interference in DigitalDe~a Communication", Doc~orelDieaartation, Southern "ethodistUniversity. 1969-70."Performance of Suboptimum Detectorsand Signal Selection in liausian andhllpulaive Noies". Doctorsl Disser-tation, 1969-10, N.Y Universl~y."Communication in a Tu1'bulentI'tlllOspherre". Doctoral Dlasertat1on.1969-70. M.l. T.
"'TheNoise p1'OJH.rties of Thielefilm. Resistor •••• Doctoral Dieser-tetion, 1979. University of South.,florida;-l~i- . .
"On Band Width £'fficient Spectral.Shaping.Methode and Digital Modu-"I'etion Tet:hniquas in linear andNonlinear Channel ••••DoctoralDiesertation. 1979. ConclJdli.21University. (C~NADA)."Ad.ptive Multile\1el IntegrationReceivers ~or Impulsive Inter~erenca ••Doctorsl Dissertation. 1919.University o~ Mi,slllouri.
"Co.munication Over fading DispersiveChannels with feedback". DoctoralDissertation. 1969-70, URi", ersi ty0' California. Berkeley.
"Random Input Signal Analysis ofNonlinearSystl3lll8 .with TransportDeley". Doctoral Dissertation.1969-711. University of Ulinois.t Urbana Campaign."Statlisticsl study 0 f Telecommuni-cation Traf~ic in Pakistanu• M.Sc.Engg.(El set.) Thesia, DUET.Decca, 1967 •
.",01'1 lnfo:r:metion Content of' Bengalilanguage lind Noi a8 in MWC_ll'Iunicetionin Bangladesh", M.Sc.fngg.(Elec~.)Theels. Merch .1976j BUET;Dacca.
CHAPTER-2NOISE. INTERfERENCE AND DISTORTIONIN TELECOMMUNICATION SYSTEMS
3.0
2.0 Preliminaries
In this chapter discussion will be made abou~ what are meant
by naise. interference anddi~tortion.their types. ma~hemati~~l.
description and how they affect intelligence in telecommunication
systems.
2.1 Noise. Interference and Distortion
In cou rse of Electr~cG\l .signal.transmisElion,. certain unwanted
and undesirable effects. take place e.g distortion, noise and
interference which alter the signal shape. So. at the receiver
end the signal is received as contaminated signel. These undesired
signal partlilrbation may be classified as noise. HDw~ver. there are
goad reasons and adequate basis for seperating the three effects
as follows:
2.1.1 .Distortion
It is the alteration of thflsignal due to imperfect :i:esponse
of the system to the desired signal itself. Distortion dissappears
when signal is turned off. for eXl!Imple,the gain of an .amplifier
may change with frequency or input signal level causing so called
frequency distortion or amplitude distort;on. Similarly, the phase
of output signal may change with frequency causing phase distortion.
These will be discussed in the subsequent chapters.
Improved system design or compensating networks can reduce
distortion. Theoretically perfect compensation is p_sible but
11
practically. some distortion Illust be accepted, though the amount
can be held within tolerable limits in all but extreme cases,
2.1.2 Interference
Interference is the cantamination of the desiredsignalby
extraneous signals, usually man made, of a form similer to ,the
desired signal, It may be a deterministic signal or~ random
sign.al.,.O' for ex~ple.multiplexing in telecommunication' systems
may be considered. In this case mapychannels are multiplexed by
using different frequency carriers (frequency division multiplexing)
for transmission over a commonlink, But due to incomplete filtering,
information from one channel may,enter. il] another channelc:au sing
interference which is called crosstalk in this case, Information
theory says that e known signal carr~no info~ation and SO it is
meaninglesS to transmit. Hence ell meaningful communication signals'1'lo'Yt
must be of~deterministic nature. that means, information carrying
signals must be ran(;\om, Here tha type ,of ~nter1'erenceci,ted ,is
elso rendom. Of course. it may alsO be deterministic e,g. in
telecommunication system, an in~erferingsine_~avemaY be generated
from an external uncontrolled source (say a power line) which is
deterministic in nature. However. actu<;lllY,E!peaking. th1;l.amplitude,
frequency or phase of the said sine wave are subjected to unpre-
dictable changes. So this may also be considered as random type
Wherep. 1s the input signal power. Nx is the total exeess noiea
power and DNis the b.ndwidtb of the net_:rk .• Now. Ps/oz Bft ie the
inpu.t SNR. So output StiR is equal to inputSNR if Pix _ 0 i.a. if'
the two port i. noiseless.
3.1.2 (f'festive Noi,! Telllpereture
More sillplitied de8cd.ptionof the noisin!!l!I. of thll t_ port
can be obtained by 1I.8ing noise t!!lllpereture. for the moment. let all
impedanees are lllatehed i.e looking into tha two pori:. tile source
••
H (f 1 ."(j nx (f)
Fig,3.1 A no ISY . two -port network
Loadource
tS/NlS (SjNlo
Iga) 8N) Nax
Matched Matched
5
Pas, NasFig. 3.2 A noisy two port With matched source
and load
T~ ~:i' d bt~ :...- .
FIg, 3,3 Cascaded network and np,se
)
.••••• >
/
sees a mlltchedimpedancaand the output of the two port ,sass.. (fig43.~).
metched loadJ The source then delivers its llVeilable slgnel power
p .s:~ ilftd the output ,eignel poweris
P=gPao .& II. '•.• 4-
where 9. is the power gainot the port. SiD1ilarly~ the output
noi.88 power ,is
N •• 9 N +Neo 'a as ex .'.. 3.6
tIlax being available 'excess noise power at the output. If. the input
noise is white and represented by a noise tSlIIperature Te'. then .•
Nas" kTs~N is the equivalent ,source noiee in the equivalent band-
lIiidth U'4. rhen~
(!i/~)op_ . eoNao
praking. (S/N)s" SNR at ,input •• f4l!S _
.a. teqn. 3.7 becomes
..s: ( S/tJ)" e •••
It waa noted before
noiseless; but tramthat (S/N)o'.- (SIN) . onlyllolhen two-port i.
• •the equation 3.8 it is seenthet the output
!iNR Ciln be nseJ'ly equal 'to the input SI"lA. despite e)(cass noise~
pmvided that Nax 0ekTsTlN. i.e the value of the excess noise is
not the lIlain factor. rather it8valucsrelative to source noise is
.lroportent. Now. the
of the two-port and
te3:11N . /{I kilN dllpenda' only on the parameteJ'S1Ix".has the dimension of temperature., This quantity
32
is ealled effective :input temperature (also _lled amplifier
ttllllper.eture) end is denoted by Te. Thus;
" ..Hence. ra :is th e lIleasure of noi..en •• s of. the netwluk referred to
the input. 1f the device is noiseless; then Te •• O. Using Te
equ.etian 3.8 becOllleS
( SiN) s1+ T"lT II
••9 Pa as
9 kIT +T lBN• so•• 3.10
Thus, under ,matched condition. the output noise power is given by
and OUtput noise telltpsrature is
•••
....,.
.'.'.
•••
What will be the case if the 1mpedance. 81'e not ••etehed '1
In th.t case all powers will be less than available powers, beingreduced by a fIIi_eteh factor. Still. the above equations ax:e valid
as miemitch fector cancels out since powsr ratios ereconeidered.
3.1.3 Not.ss figUre
The .ffective input noise teMperature io Illost useful in
describi.nglow noise 8lllpli tiers, devices withT e <t:T8" But when
excess noise is large, noiee figure f proves more convenient.
Noip figure ie defined •• the actual output noi.8 power di'vided
b)' the output noi.s. POW8Z if the t.wo-port wllre noi •• less. lh.
source being at roolll tlllllperature To' Thus.
••• 3.13
PUtting the value of' T. f becomes ••
•••
•••
EvidenU)'. f'). 1; r ••1 when two-port is noiseles8 ,and r >.1when two-port has sa=e nols8. Now. T•• {f-l)To (tzom eqn. 3.14).
So.(SiN) •o •••
••• 3.17
Before the advent 0 f lown01 se device. most source telllper.l!ltures
were essl!lllt1ell)' rosnd 80 equation 3.11 could be used. But for
low noiae receive.", eQn. 3.16 .hould be used.
3.1.4 C.a-eaded Network
Let two networKS be connected in tandem 8a shown in fig.3.3.Let the effective input noise temperatures for network A end n beTal end T&2 respectively. Let the noise ecurce has noise temp. T.
In a •• all fraqtl.,neyband. the neiee power at the output due to
th., noise 80urces in the first networkie 111-92kTe1df and th.t
for thellecond network is 92kTe2df.' Hoise power at the output due
to noise source only is 9192Tkdf. flence, the totel noisa power
at the output is k92(gl T+9lTel+Te2)df'. The effective input 'temp.
for the two n'!ltwo rk cOlllbined is
••••
(By the defn. of effective noise telllp.l
Thi. result cen be generalised for n networks. Thus.
Using the tefationship between noise figura I!Indeffective input
noise temperature. it can be shown that
from. above equations it is seen that if.ff'ective input temperature
for d1 fferent ne'tworka in tondam are nearly equel. then the noise
contribution by only first stage is sign1ficent. However, if the
gain of the first stege is SIIlell or if the noise contribution of
the 51!!cond stege ia larga. then it is necessary to teke these in'to
account when lIl111kingnoisec5lculations.
3.2 Neile in Devicelil
Noise in netwoxk. are due to the noises in the devices CtlllIpri sing
the networks. Now.let the casaof noiee in both l!Ictive and passive
devices bet: taken. Baeicnoie.a inactive devie •• are '''.tic thern:al.
short and parti ticn noiae. Beaie noiae is pl!liIsive network is
tbeJ:'lllsl noise.
3.2.1 Uoise .in Pelsiye Device,
It can be easily shown that passive reactive e18lllants do
not generate any noise. But in s resistor. tlle%'1lls1 noise i. alway.
present. fhi. noise eao be taken illS ",hita noise over .11 1'.requen-
eiell of' intereat used in cOlDlIlunicat1onaystem. The meat'! square
. voltage across the terl1linels of the resistor is If1'Oll\ eqn.2.23}-v2
•• 4kTBfi in a bendwidth B. Also, it lIIay be noted thilt noi ••
voltages do not; add, only noise power adds. The resistance noiae
can be taken as white noise over all freqlleneies of intereet in
comlllunieation networks. So, if total noiS8 power 01' a resistor
i. coneidsred. then 1'1;,becomes infinity. But,. in practiclIl any two-
port device or measuring instrUMent h••• noise equival.ent band-
width BN, and 50 the filtered noise has finite power and finitellItlensquered velue. To define equivalent bandwidth, let whits /'IOis8
powar density is't12. trilnsfer function of the network a. HIf)"
lind aver.age noiae power after f.il. ter.ing GIS rJ. Then.
N •• /' I He f} 12, 1/2) df
,-ll •••
Now. noise equivalent bandwidth fiN .i.BN ••'1/H0
2) (' I HCf) 12
df
where It(0) •• IHI f) LUll( is til e center
1.e. voltl!lge given. So.
defined by
•••
frequency amplitude response
••• . 3.23
Conaid.ring Fig. 3.'.it i ••• an that BNaquala to the bandwidth
01 an id •• l r.ctangular 1ilt.r that wouldpa•••• lauchwhita noi••
es the two port n.twork in que.tion. their lDaXiIDUIDgain. being
.qual. So. by definition. the noi.e .quivalant bandwidthof an
ideal filter i. ita acw,al bandwidth. For prectical purpo•••• UN
i. so•• ""'at gr•• ter then 3-db bandwidth; how.ver. for .lective
flltex tha two bend-.ddths c.n ba taken aqual., Thus. filtered noi••
•••
for ex•• pl.an r •••• volt.etar with BN• 100kHz would reed theopen circuit voltege of a 10k
a. unity.
reei.tor a. 4 uy. Here H i. takeno
(F (~.3. 5)How.let a R-Cparall.l circuit i. to b. CDnaiderad~.Th.
aquivalent paullal i.p.dance i. l. _.' • (l/jwc) 11 R • Or,
• (1) (2) (5)3.5 "01" end Ampl1tud. Modul.tip" • •
At the output of the 11Mdl!llllodulator the d.tected .lgnal
will b. (for the .aid 1npu1;y(t». KutAc + "i eoa Wit). ",hara kD
i •• constant. Upon fi.l.tering end r •• oval 0 l' DC c:oIISponent. the
output ugnal i.
for
11'1.1 ZUI'il> B
•••
Vher. B is the band-width.
for colllp.1'1&on.if the carrier i. int.nU.nelly IIlOdul.t.d at the
t1' •••• 11ot.r. then
x (t) • A [1+1'1)( t) ] eo. w tc: c _ ~ c:
.wndthe d.adulated output i.
••••
••••
for ton. IIlOdulaUon. And where A i. the carrier _Dlltude ••• surfldC .Id'ulIxAxxi:a at the detector output. Thu. for intentionel lIIOdul.-
&tion. output at the d\!tector 1. proportional to carrier •• plitude
end •• plitude of thelllcclulating signal. But for unin~ention.l
50
lIlodulat1on (noise ,case Q~ interf',lllrence casal. the output is
dependent only ,on inter'fering signal.
Net"'. the cese of noiee may be consid ared. A narrow blind
randol'll noi,sefllay be considered illS'coll.ection of lllrganumber
of interfering ,weve and mathematical analysis mey be given 8CCOI:-
dingly. He.r. only que1,1tative di.eussion will be made.
3.5.1 sse (Single Side Band) Modulated wave
1.at to the impute' the demodulateX' whit e noi5111 wi th a
pow.llr dens1ty sf -100 dDllI/ld:lz.extending from f IIto 1'b is. !ldded.
Let the ssa sign'l1 to be demodulated be OdBm.It Jnaya1sCl be
assullIed for ailllp.licity that the demodulated besebend signet. is
.180 ri .dBliI:C Ii eOndi.t1o'\ usily ,achieved w.ith auit.ble IIlmplltier.a........-~ . I. ~..I_'. ....,
end attlllllUators). Also assullled that the dellloclul,.tor ie preceded~-"'< .' r.~ "':'- ~,.. " "'"':.-""_" .,"'- \' _\:::- __' ~y •..~.11;~~assi~g .~~"qu~~~" '11 to tb" This hllhown inth.
fig. 3.9. Upon demock.la.tion- by • product demodulator with card.ar
inserted at fc" the noiee apect:a.im will be tr.~s1.ted by fc ta
thlll bestlbllnd fraquencies es shown in tig.J.9(b). The totalnoisfI
power before demodulation ,el8O appears after demodulation with
po.sible foldover at z.ero frequency. Tn.s folded noise power
,spectrulll add •• on power b •• is bal:euse the random noise ie
'uncorreleted between different frequencies. Thu. SNR,of" SSB
&ignN et r-f ilS notchlilnged by de1llOdulation proc:ess,. Secause
of the 1'01do"ereffect. howe\fer,~ flet noise et 1'-1' need not be
flat under demQduletion• .or ,cou rse. for best StUI,predection filt-
ed.ng .shl3uld reDlCive ell noise et frequencies not Q(:cupied by the
modulation signel. Sl:lcn filtering removes any poesiblefold over
Powerlevel ;ndBm
-100100d8m' kHz nOISe
- 97
,?owerIn dBm
-100
0 fa fc 1b a ( fb - tc) (tc - fa)
frequencY (Hz) frequency ( Hz) .
( a) R- f nOise spectrum ( b) Baseban d nOise spec trum
fOl; USSfC. Unlike DSBst. here a d-e beeebsnd h1:«l results by the
dl!!lllodulati~n.II threahold effect oce:curs if envelope detector 18
uaecl when input SNRi$ very 10w. for such ceae (So/Nol :::=:: O.
But for .yncronaue dtlt.ct~r no threshold effect tllk ell piece.
Thia e.o be shown by considering Fig. 3.10.
3.5.4 SynergnotlsDetectcr
fJ:'Olllphal101' diagr.llI 3.10. for oorllllll case (fig. 3.10Ce».
'..•../It the sync:ronou!I deteetor v( t) ie Illul tiplied by l5yneroni tied
carrier vet t) •• Sin wet And the deteet:or output i8
53
( ) ]2 .+ ns t . ,S1.n wet
••
.Iit'•• 3.50
where III(tl i. the lllodul1ltt1ng signals. Tbe eignal eftsr fi1 tllr!"g
by e lew paSfl filter 'wi til d-e bloel<1ng i.
" ."--' '.o.../ ~, 'J It (t) :II laC~)/2 +" n (t)/Z
'0 _.... ~ __ ' 1!
,••••
" -,\
3.51.I
-, "f:vi"dOl'ltly, the result :is independent (')f tbe' input' SURBihee it' ".f '4 -l'~ I" .,.""" .1 t;.~ 0. I~_
does oat enter into the anelJfeie. The output SNR1. given by•
.• .. ,,.
Th\nl. the result 101111be hold for Ilny input SNR.
3. 5.6 Envelope De1:tetor
'\Ie t) ~ [\Ie + Ill! t) •••
Thu.s the dllllllOduleted output will be proportional to
1Il( tl + n (tl. Thus StiR is $l!llftl!l as for "yncl'OOOUS detector in thisscees.
r( t) ~ DC t) + [Ve + 1lI( t) 1. Cos It{ t) ••• 3.54
(neglecting the quadretuncolllponllnt wlli.en i.. varll SlIall ••
Sj,/fli Is very 8lIlall).
Hence, ou tpu t ",i~1 bat
••• 3.•55
cannot
process\
!i_onal
Here, the desired signal i8 dependent on .• rsndolll, ,Coli ~(t) and.1>O info~lltion is last. Sc.lllOdul.ting. '.recovered. Thi.l!l is celled threshold effect.
For colllparif:lOll of variou$ type .of linear !Ilodu1etion the
following points al'e to be not.d.
(1) tren~1esion bandwidth. BT(Ii) poat detctlon SNR,(S/N)o(iii) D-C or low frequency algne! tran8lllissions
(iv) instrumentation (lOlIlpll1xi.ty
(v) tranemitted po••r
Tn.e. an cOlllpared in the feb1. 3.1.
Suppreaaedcerrier moduletion 1e superior to convsntiontd
AMon severel count. viz. better SNR•. ,no threshold effoct. When
bandwidth conservation i. illportant. SSBand vsn (vestigial
eidebend) are per'tieularly attractive.
<l-c0
'';:;0.-
(tl>'""0>-uc'"::>CT
'"~-II> 0::>0
'"C0C'0II>C
-- _.--Ai/Ac= I
Impulsesfor A,"Ac
\.time
AI .'-=05Ac .
---.Fig,3,1'3 Instanteneous frequency 'dev',atl'on-versus time for
two s',nuso' ds
'"II>'0 50C
0- /'0l1 3DC'0-
'':: OJ 20u'O'".••• C
"'.-'0 100-,-II>-000- ~
0 10 20 '30 so zd B
Fig.3,l1. Post detection signal i.o - noiseratio as a function of z fordifferent 'types or' modulatl'ol1
•
55
Tabl. 3.1 Colllperison of eli.fferent types of linear lIoc:l111etionsystems
2 - -z
1 + 1',2 x2
llc ?
No
C-omplexi ty
minor
Typical appli-cation.
bor.dcast radio
\
0513
SSB
V$B
VSB withcarrier
z
1
1+-
z
z
, .
no
yes
,
major
moderate
mejor
moderate
MUI system
III ~ modulation ind •• ,x • signel (modulating)
III •• MeSS81jle bandwidth
~M •• DSBTC
Pt. tranatll1 tt.d power
z. noise density
Il-.:~.<A~LJ,Normall,vAsffect in A.l1 t.kea place when input SNRis 1__
then 13 dB. Reasonableintelligibility in vo1c1I trtlflQllliesion
d8lllenda a post-det.c::tion SNRot about 30clBor IltOrs which is well
Gave the threshold 1,sv.1. Thus thxtlahold level is •• al< usually
not • aeriou9 problem for I\M !SyetelllS. BSliIide6,syncnmoU8 deteQtorl!l
are co.t1y and so are not ."l'il1tlll IlIIlplcYtld in broadca ••t ayl!ltE!lll
56
3.THain and An91eMocluleUont 3)
I t ",all shownin ."etian 3.4. that an interfering signal
.t the detector input produce. both .plitud. and angle lIlodul.tion
at the carrier. Th. ttltal GUnal entaring the dlllllodulator. y( t).
~.Y b. re-written e. ( eqn. 3.45)
.. ,..
Now. the angle detector will Ignore AM••odulation portion. fOit
• PH detector Qutput will b.
. ..
.,..
••• 3.51
Tnua. for FMcase; output noisoalllplitude i. dependant on frequency
which is not the CIII!II.I for PHay.tu" Thia i •• hownin fig. 3.11.
Th. output noi.e power ENo in e bandwidth df (foz: fH case) i.
then gi vlln by
~N •••k 2 fI 2 f.2/2", 2a D.i,l. c: •••
N4w, it cen be
.pcwer is p •••e
t!lhDwn th.t 1'dt. Vn2/2 end average earn.:I.'
Ac2/2• ("L _noi se denai ty); tIO •
•••
Th. total noise power output No in 3 bandwidth B beccClee.
BN • /o -8
,2i of • • ••
57
Now. the lI>cpreeaion for." (/II sign.l lIntaring the discr1.inetor
1.
• ••
where lIt t) • - ( L!.fl f.) Co. ".t t the lIlodulating .igl'ld being
iin "'lilt. tlf being the peak frequency devi.tion. Thill output
of the d.t.etor ",UI betk••/2s) ~/dt •• leD L1f '" 5in w tlZ"f i.e.1>. III III III,kO 1- ' Sin w.;t. ( Where kD • constant of proportionality) ~"ttlt •• vlIrag. detected eigne! power i.,So •• k£l2 (df)2/2.,Thu-...fOll: (/II e•••• output S~R i.
•.•• 3.62
loihere III,. ,df/B ie the I:loduletion indllJ<. Now. the equ.tion of
• AM ",eva i8 Actl + III Sin w.t) Sin "'ct, Kence the detectoroutput vo1.t8ge i. mAc Sin w.t, ThUll, output average power i •
• 2,,(;2/2, Since,B
/ "2 df ••-B
(So/No) AM • 1121'/2'IB.. Pc /2 t B ( for III- 1 )
than lIe••• g8 bandwidth. W. Since the inter'erence8lllplitude inerseasea
linearly ••ith I'il in the absenceof filtering •. the deelilpbasized
intsrferenClI response is HUE( 'i) l!(
Like PMthi s beC:Oll'les constant 'nr
/'il as shown in the fig. 'J.12.
1'1/ » 13• Evidently. total
noise in de-empht>flized FMi. ~eS8ltr "than that of PM and so fM is
'superioT 109p~.
Due to the affect of dentphilsi. high 'requ.encYlllBss1lIge signal
ampli4ude will .• ism be lowered "hich Isnat want1!d. To remove thIs
• .pre8llphash netl«lrk hevlng "1'£(1) •• 1/"D£( f) is used .t the
transmitter. It een be shown that d.Mphasis illlproves SNRat the
dector output by ebout :itt 4( four) dB.
3,7,2 fM Th~e.hold Effest
JlII f) •• " tanAi Sin WitAc + Ai Cos 1111 t
In.tanteneou, frequency devi.tion is given bYi
, J t) • d!6l tl~ dt .,iCo. "it
(illt).! "'..,(t)eft - IAi/Ae' Sin "'it which ia •••• as in
Art, 3,4, ~o",eve;1". aa Ai approaches Ae in 8IIlplituefa. the r.ault,nt
deviation (or result.nt signal) i. f';1" fromai.ullOid.l .nd baeo.e.
i.pulsiv. ",henAi - "c. ApJ.ot of f.4 (t) with repre .•entative value.
of A"/Ac i. shown in fig. 3,13. Evan if noise p.wer is 10 0;1" 15 dB
below the signal power. blpuls1v. effact tak•• plac., Ttlenature
of such noi.e ia ia'pulsive .nd the:r:efcrespeeielly dD.ging to d.t.
tren •• i.lon. Such .1'1'act i. celled I"H threshold effect.
3.8 "'!lpari.pn of Exponential U.e.ngle) and Un!!; "edglll1:ionSvst8lll(2)
I t is 8.en fJ:OIIthe previou 8 discussion th.t f/4 .y.t•• produce.
be.t SNRin CQllIpadllOn with AMand ph.e •• oduleUon, Sueh i.provl!IIlIent
i. with the cost 0 f incre.sed b.ndwidth. ThUlllILIchsyst_ elln be
60
applied for :cases where hsridwidth.croqui3:emen~ is GecGndary faetc:lr
SOQclean ,slgruila a%'edes1red. Besides pow.srrlllquirement is lesser
for fM cass.
11'eomparieongraphissnown in Fig. 3.14 •. 1t Should tie. noted
that FMforsOIlIlII instances is inferior toPCM system •
. .! ,.... "'.'
.-." ",
.RltfERENCES
1. Membersof the TechnicalSteff
2. Carlson. A. Bruce
3. Connor, F.R.
4., Panter, Philip F.
5, Johns. P,B. andRowbotham, T,R
"Transmission System for Communi-cations". Bell T.e1ephoneLaboratories. Inc •• Revisedfourth Edition. 1971.p 180-194.206-211.
"Communication Systems",McGrawHi1.1 Book Company. Inc ••International Student Edition,1968. p 200-212. 439-442.
"Noise" Edward Arnold (Publishers)ltd,. 1st Edition, 1973.p 46-56,
"Modulation. Noise and SpectrslAnalysis". McGrawHill Book CompanyInc,. 1965 p, 172-178 .• 429-430.
"Communication System Analysis".Butterworth. and Companyltd, .•1972. p. 69-71.
CHAPTER-4
TECHNI QUES AND METHODS FOR MEASURING Nor Sf
£1
Prelimin!!ri es
In this chapter methods for measuring
an lal oj e.oo;.telllperet"J"iI will be discussed briefly. Also a discuesian
will be made about measuring units of noise and signal.
J 4.1 Noise and Signel Measurement( 1)
4.1.1 Noise
It is difficult to measure the limplitude of noise due to the
non-deterministic nature of noise waveforms and amplitude dependence
on bandwidth. I f an a-c voltmeter is connected to II noise source.
then since noise is random; meter reading will be fluctuated randomly
end it is so meaningless to measure noise i1'Lthis way. Thus. it
is usually necessary to average the noise amplitude over some interval
so that a moX'eor less constant mete.r reading i~ obtained. A meteX'
~hat integrates the reading over a time period which is long compared
to the reciprocal of the bandwidth removes most of the fluctuations.
So rillS noise voltage can be measured by en
over finite time interval end dividing the
a-c volt.eter by averaging2rms noise voltage by
resistor. noise power can be determined. Measurement of absolute
magnitude of noise power is not important. Rather it is important
to measure how much it annoys. telephone use ••. Thus measurement
of subjective effects of noiseie important. So; the mlier is designed
to measure the following effectss
(1) the readings should take into considerations the fact
that the interfering effect 0 f n!!lise will be a function of frequency
spectrum as well aeof magnitude.
(2) When different noises ars present simultaneously, the. . ..
meter should combine them to properly measure the overall inter-
fering effect .•
(3) When different types.o f noise. cause eqysl interference
as determined in subjective tests. the meter should give equal
reading.s.
Besides, the transient response of the meter should be
eimilar to that of the human ear.
Interference is made up of two components viz (i) annoyance
and (ii) the effect of noise an intelligibility. Both are function
of frequency. Hence proper frequency weighting is required in the
meter which is stated by. point '1' .bove: For. example, a 200 Hz
tone of given power is .25 dllJ.lees disturbing then a 1000 Hz tone
of same power. Hence, the weighting network incorporated in the
noisemeter will heve 25 dB more loss at 200 Hz than at 1000 Hz.
It is assumed (by several tests) that the effect of noise power . ~-- '/maximumat 1000 Hz. This frequency is used as the reference frequency
in Bell syetl!ml. But CCITTand CCIR(l) recommended reference frequency-is 600 cps. To determine weigh~ng at di fferent frequencies (1. e
frequency weighting curve) annoyance is measured in the absence
of speech by adjusting the level of a given tone until it is as
annoying as a re~erence 1000 Hz tone. This is done for many tones and
fo r many :i:IIlJUI.x.lltix1Iu:observers and the resul1:s are avereged and
plotted. A similar experiment is done in the presence of speeqh at
the average receivetl volume to determine the effect of noiae an
63
.rticulation. The results of the two experiments are combined
and smoothed resulting in the C-message weighting curve. Similar
experiments give other types of weighting curve. For paophometeric
weighting (CCIR and CCITT recommended) ths reference frequency
ie taken iiS aoo Hz. These are shown in the Appendix-B. Weighting.-like 144. flA etc. are not used ~(3).
Another important factor is that 5X~".~.,li' 3 l.j
noise add on power besis. Thus meter must also read in this way.
The last important factor is the transient responsaof the
humsn ear. I t has been found that. fo r sounds shorter than 200 lbec.
the humsn ear does not fully appreciste the true power in the.
Bound. For thi s reason the noi se lReasuring meter is designed to
give 8 full indication on bursts of noise longer than 200 mssc.
For shorter bursts, the meter indication decreases.
Thus. frequency weighting. power addition and transient
response of e noise measuring meter describe the way in which a
message circuit noise is to be measured. Besides. II noise referelnc:e
and scale of measurement must also be provided. The chosen reference
is 10-12' watts or -90 dBm. The scale marking is in llI.sill •• decibel
and measurements are expressed in dB above reference noise (dBrn).
For peophometric reading, this reference is et aoo Hz and for
Bell system it is 1000 Hz(l).
In particular. let w(f) represent the weighting of the noise
shaping network in dB relstive to 1kHz reference frequency. Thus,...w( f) /10
= 10 •••
64
Where H("') is the transfer function of the weighting network.
The total weighted noise power for noise density of Pi (f) watts/Hz
is given by
PT • f« iH(f)/2 Pi(f)df wettso • • •
The effect of weighting over the frequency range 1'1 to 1'2 is given
by 1'2f p~(f)df'l' ,.A .. 10 10 g .....,,..1••••. _
1'2 [ 2f H(f)! p.(f)dfl' .1-1
dB ...•.
Thus. weighting network attanuates the noise power by ).., dB. for
fl~ O. 1'2=3 kHzj (telephone chennel) }- .•• 2dIl for flat Pier)
and C-message weighting. for psophometric weighting end flat Pie f).
" becomee 2.5 dB.~
Till now. measurement of noise is considered,f!Jrt.,lephone.
chennel with analog signal. for analog TV signal. separate type of
weighting is required .according to the effect of noise to eye.
Digital signals such as Data and PCMere not affected by no~se in
the same way as analog voice signale. for example. the annoying
hiss due to thermal noise has no effect on digit~ sign81s unless
its Ilmplitude approaches the amplitude of the signals. On the other
hand. illlpulseswhich cause tolerable clicks or pops on voice circuit
result in almost certain errors becaJlli5S1:of their high amplitude.
Thus separate type of noise measuring techniqu! is used for digital.
signal. Details can be found from reference. (6)
&5
Signa.ls
Message or speech signal is alsononperiodic (otherwise no
information) and so difficult to messure. the nature of speech.or
pJ:ogram signal is such that the average. r.m.s. and peek values a
are irregular functions of time SlJ that one lIumber cannot specify
any of them. But regardless of the difficulty 0 f the problem •. the
magnitude of the telephone signal must be measured .and ,characterized
in &Omefashion so that proper design of transmission media can be
made. Signlll magnitudes must be sdjusted to. avoid overload end
distlJrtion and gain and loss must be measured. for this purpose
a characterized unit called volume uni t( vulis used. It is an-..---- _..... .-~------ ----emperical kind of measurement evolved to meet prect.tcal need. and is
not deT'inable by preci se mathematical formul,a. The vu. meter used
for this purpose gives the reading in vu. The pri.ncipal functions
of the vu meter are
(l) Measuring si.gnal amplitude in a manner which 'will enable
the user, to avoid overload and distortion.
(2) Checking transmission gain and loss' for the complex signal.
(3) Indicating the relative loudness with which the signal
will be heard when conve~ed to sound.
for convenience. the meter scale is made logarithmic but
unit is vu not dB. It can be related with dB by ths following
relation;
Average power. vu - 1.4 dBm ••• '...Such relation hold fora continuous t~el'l. E=:I: ," I i filii , en
66
! aU wr. for no,nco,ntinuous talker the relation is
Average power'"' vu - 1.4 + log log"""- L dBm ... 4.5
~.
Where '"Y L is the load activity factor. It should be noted that vu
meter has a flat frequency response over the audible range and is
not frequency weighted in any fashio,n.
/4.2 Uniots for Measuring Noise and Signal
4.2.1 Signd
Since telephone cirCuits operate with signal powers which
rarely are as large .liS 0.1 watt. and which may be lower than 1 pW
(pica weott). the use of watt as a unit
A convenient unit is the milliwatt i.e
of meas.rement is awkward.-310 watt. Many operations
can further be simplified by expressing power in relative dB.
Normally power is compared with one milliwatt and then. expressed
in dB. This is called dBm. Thus a power of i mil is odSm and a
power of 1 pWis 10 ~og(10-12/10-3) dam or - 90 dBm.
Noise
It was shown in the previous section that for both psophometric
and C-message measurement reference power of 1 pili is takan. Reference
frequency for weighting is BOOcps for psophometric weighting end
1000 cps for C-message weighting. Psophometric noise power is the
average noise power delivered to 600 Ohm resistance and expressed
liS picowatt psophometric (pWp). Thus
pWp •• I h t. mVl.2..psop ome rJ.C _600 x 106 picowatts •••
orin dB. dBp '" 10 log pWp .. ,. • • •
6'1
Thus 1 pW 51; 800 Hz is OdBp _IOd1 pW at 1000 Hz is OdBrnc. So_
a OdBmpower having a bandwidth of 0-3 kHz is equal to 88.0 dBrnc
(c mSliIns C-message waighting). or 87.5 d8p. Thi s is. so. esw!lighting
of 0-3 kHz is 2 dB for C-message snd 2.5 dB for psophometric. Thus.
dBp •• dBrnc -0.5
for input flst noise from 0 to 3 kHz.
••• •••
UlIing above relation s following table may. be " •• Iblll mads.
Table 4.1Shows the reletion between di fferent un! ts fo l' Noi sePower.
Noise Total power .111. 0 f 0dBllIunit
I1 kHz o to 3 kHz
dBrnc 90.0 dBrnc 88.0 dBrnc
pWp 9 . I 5.62xlO8 pWp1.26xlO pWp
dBp 91.0dBp 67.5 dBp
White noiae of-4.6 dBm per kHz.
88.4 dBrnc
6.03xlOB pWp
67. B
(lh{3)K 4.3 Transmission Level Point (OTLP_ dBr. dBmo etc.)
A multichannel telephona system may ba both long and complex
wi th m.sny stages 0 f gain and attenuation dua to the transmi ssion
media, repeater, Mux-DeMux etc. To state the power level of a
signal measured in such a system has little meaning unless the ..
gain or attenuation of that measurement point ie specified. System
level of gain or atteDuation at any particular point can only be
specified if a reference level for the system is defined. Other
68
points in the system may then be expressed in terms of dB of
gain or attenuation relative to this reference level. In telephone
system; this reference level is called zero level point or zero
TLP. It is the point at which standard test tone has an absolute
power of ImW or OdEm. This point is often situated in the trunk
exchange (toll office) at the transmit end of long haul system.
The level of any point in the system expressed in decibeJ.s
relative to the zero TLP is referred to as the relative transmiss-
ion level 0 f th ilt point. It is designated by dEr 81though TLP is
some times used. Thus a - 33 dBr (or - J3 TLP) point in a system
will be 3J dB below the zero «TLP point. It should be noted that
dEr is. not a meaaure of power. I f signal power is measured at any
point in the system in absolute terms (e.g dEm), it may be related
to nominal loading levels by expressing it relative tothe trans-
mission level at that same point. Power levels specified in this
'way are given the suffix O. Thus dBmo is used to indicate the
signal magnitude in dBmreferred to OTLP. Then dEm and dEmo is
related by
dBmo= dBm_ dBr ••• • ••
Similarly,; dBmopmeans power of the point referred to zero TLP
with psophometric weighting.
y 4.4 Measurement of Effective Input Noise Tempera1:ure andNoise figure
The above two parameters of a networl< can be measured normally
by two ways viz. (1) calibrated noi se source and (ii) calibrated'-
• 6'1
signal source. Tha first is more convenient to use as it does not
involve bandwidth of the noise of the system which is gnerally
difficult is know accurately. This can be shown by the following
description.
/' 4.4.1 &t.i.a Cslibrated Noise Source
fig. 4.1 shows 1;he circuit for measuring noise figure and
noise tempara1;ure. Let T1be the noise temperature of the source
due to its internal terminating resistance when the generator in
the noise source is turned off. Noise temperature of the source
becomes T2. when the generator is on. Now. the noise power measured
ilt the output of the two port network under test under the generator
off lind on conditionll are resp,ectively
and
••• ..'....
4.10
where 9t •• gain of 1;he netwo rk
BW = noise bandwidth of the network and power meter combi,ned.
noiseTs •• effective input/temperature of the network
Now; let~ '" P2/Pl' then r is given by.
or,
•••
•••
Thus. the value of T can be determined by using the above relation.e
70
Now; the noise figure; f. of the two port network is
given by.
F = 1 + T ITe 0
Tc-i-o•• y - 1 .... 4.13
If the source impedance ie at standard temperature; (i.e T1 •• To)
then.
F •• ••• 4.14
Thus f can also be determined.
Sources of white noise that can be used for such measuremente
are the temperature-limited noise diode Ii. e shot noisel, the
gas tube n£lise generator. and resistors at di fferent. temperatures.
The temperature limited noise diode represents a variable'noise
source Useful at frequencies lsss than lGHz. The noise output may
be changed by simply varying the anode current of' the diode. The
gaa tube noise generator may be used for MWfrequency regions.
I f resistors ara used for the purpose of noi se gener.atlion; then
eccurate temperature may be obtained by immersing resistors in
icebath. liquid nitrogen baths etc.
>( 4.4.2 Calibratad 5ignal Source
This method is similar to the above method except that in
place of noise generator iii signal source is used. When the generatar
is on, the available signal power is Pc' It is assumed that the
generiltor source impedance is at standard temperature TD
• Powers
71
measured at the output of the network with the generator off and
.n are respectively,
•••
and•••
gtkBw (Te .• 101 .• gtPcgt + kBW ( Te .• To )
••• 4.17 •
and F •• 1y-l lC •••
v-V5 Noise Loading Method
~ In the laboratory and field testing of multichannel message
system. noise loading test method is mostly used. In this system
multichannel messege signal is simulated by proper whit. noise
level. I t is found that when several subscribers IIre*akic"I}(lCi:kllJl
talking, then the resultant signal at the bSlilband i8 like a white
noise. Thus a white noise can simulate the signal of multichannel
message. It is found that the average loed of a multichannel(3)
system is
p = - 15 .• 10 log Nav ...
..---0---
r--- --lI II II Varloble I
II noise r'~I--sourte @>
: L_~I__II I
I iI IL ...J
..-0'--
Net ",jork-under test
l
Powermete r
Fig. 1,.1 Set up for the measur~ment 9f no'1se figure.. -
Noise-;:enerator
-'~'Bandpass'fl Ite r
Noise -~receiver
Nelsepower
f(0)
Nt
f
NI V
N2---;--l
I I
~f l
IiI
,It
f
.Band pass , 1
System f il tet I
I t•. 1
f
,NOisepow,?r
NOise Bond stopgenerator fi It er
'.
,
( b) .
• Fig. 1,,2 Noise measurement by us',ng noise' (oading method (0) No bond stopfilter between noise generator and system (b) A bond stop filterbetween noise generator and system
which is elCceededonly one percent of the time at the busy hour.
Hers, N is thschannel capacity of the system and Pavis the
value at zero TLP. Thus the simulated white noise must heve a
level of - 15 + 10 log N dEmo. This relation holds for N> 240.
The derivation 0 f th e above relation may be found in the references! 1)
rig. 4.2 shows the measuring technique of noise power ratio
(or SNR) using white noise method.
,Procedure
The noise generator g~nerates the simulated signal. It is
then impressed at the system input. At the system output a band
pass filter is connected. The bendpass filter has abendwidth of, ,
one message channel. Normally, arrangements. for three center
frequencies_ are there at the band pass filter. One frequency is
used for measuring lower chllnnel noise, o.ne is for.middle channel
and the third one is for higher channel noise me.sarement. Thu!!,
noise receiver, whichie cannected at the output £If the bend pass
filter, (4kHz sbout fil lIeasures noiss in the channel at center
frequency fi• Let the reading is NldB.
Now. a filter .is introduced between system input end noiee
generetor. This filter eliminates the noise in the message channel
at fi• If the system is noiseless, then the measured output by
noise receiver will be zero. But actually; there are inherent
noises in the system and the receiver will read some value at the
slot channel fi• Let the reading is N2 (in dB). Thus; noise power.
ratio (NPR) i B gi ven by, - N2 + N1 '" NPR. Now, SNfl (signal !G noise
ratio) is defined at OTLPpoint. Since, at zero TLP point a load of
72
must be added to the NPR to find SNR•. This is
15 + 10 log N) for N channelsor(P av - 10 log N) perp =(-av .channel instead of OdBmis used, an amount of (10 log N-P ) dBm
av
so as SNR is NPR
forOdBm at OTlP.
It will be shown in Chapter-6 that by using noise loading
mathod, thermal noiee snd intermodulation noise can be measured
seperately. This method can also be used to measure noiSl!! in
MUX-D£MUXequipment.
I t should be noted that in .the derivation of Pav it is assumed
that not all N subscribers are talking rather 'Na' number of
talkers are buay where Na is given by
••• 4.20
where t-t is the load activity factor. which is defined ae .the
ratio of average power of a telephone talker to aVl!!rage power of
a continuous talker~ Typical value of 'YL is 0.25. N. is assumad
fixed and equal to the number of talkers exceeded one. percent of
the time during busy illll hour with all N-channels.busy. N-channels
are busy becau sa (N-Na) subscrib ers are ringing and Na subscribers
are talking. Thus; for N = 960 channels; Na becomes approximately(1)
250 • fill II III
More will be discussed about noise loading in Chapter-6.
Before concluding this test description the equations for Pav
for 12 <N < 240 a:u is given below:
lOading level = (-1 + 4 log N) dBmo
,t Ii / I P I,
•••
4.6 Out of BAnd ~toi'l1 Me@surelllent
International. recomlllendation exists for G noise IISlt.U ring
techni.qult known 88 out .of band testing which can be carried out
dudng the trenllllllin10nof actual tra1'fic. The principle 18 ai_iIar
to floiee loading method. Here a Iluaasu ring channel is located
approximately l.O~outsids the baseband and a bend stop filter at
this frequency i e inSErted peX'llianently at the aystSlll input to
ainimise nois9 due to lnco.ing circuits. A whits noise receivertunad to this channel cantinuouely !DonitOr the dietortionnoiseoccuring due to the tX'eft'ic lasding in the base bend. The IIltraeurinq
channel msy be above the baseband in which cese it will be more
sensitive to changes in the~al and intermodulation noise in the~-f and i-f seetions 0 f radio 9;)'9t __ • l.ocating the channel below
bS5eband will generally give leltels more sensitive toehangl!lS in
multiJillaxJ!ing/de-mul tiplexing equipme!lts. OBN(when used for
eatalli.te eystem)frequency i. different for different g:mund s1:e-
tiona.
Since lIleasurement .i 9 lIad. during Ilctuel traffic 1:rlllnsllli.aion
ita accuracy is li1llitad due to variable natun 0" the traffic loading
11!l\lel.end the fect that e1'faet. within the basebend can only b.
predi.ctad rather then measured directly. for these reason8 •• ffOlrt
ia presentl.v being expanded in U.S.A and elsewhare .i.;ntc 'the probl_
af monitoring S)'81:eID psrfo:tlllsnce eu1;o1lle1:1c81ly1'or eigne of' ayetlllll
degradaUon during actusl traffic ;condition.
'.
REFERENCES
1. Membereof the TechnicalSteff
4. Yonezaw8; 5 and Tanaka; N
5. favin. D.l
"Trensmission SystElll for Communi-cations". Bell Telephone lelbore-tories. Inc •• Revised fourthEdition; 1971. p. 30-33; 38-39.49-68; 170-176.
"Noise" Edward Arnold (Publishers)Ltd •.• 1st Edition. 1973.
"6A Impulse Counter"Bell Laboratories Record.Vol. 41. March 1963, P. 100-102.
CHAPTER-SNOISE IN CABLES AND SWITCHING NETWORKS
!
7'5
5.0 Preliminariee
In this chapter discussion will be made sbout noises and
interferences in cabl.essnd switching networks. Important sourcee (If
~.ise in cablee are crostal.k. thermal. noise. l/f noiee (or contact
noise). conducted or radiated interference. The main sources 0 f
noiae in switching networks are relay arcing, coupl.ing by battery
internal resistance. coupling by ground contact resistance. and
other interfering sources (e.g feul.ty flourescent lamp. generator,
motor etc.). Results bassd on the measurements taken in the swit-
ching system and cable network in Dacca city will also be given.
finally, discussion will be made based upon the results.
5.1 Noise in Cables
It is ,clear that the term' cable' includes (i) multip aired
teleph.ne cable (primary, secondary and junction)' and (ii) co-axial
cable used for carrying base band signal in longk •• i1 haul trans-\
mission. for long haul transmission, co-axial cabl.e is not used
in B.angladesh at present though such iii system is planned for
communication between Dacca and Chittagong. Normally r-f system
is used to transmit baseband signal in our country. Of course.
there are somBoverhead physical lines to carry baseband signals
at some areas. Considerations will be made here only for multipairad
telephone- cable which is extensively used as primary, secondary
and junction cable elsewhere.
I mportant sources of no1 se in cablas ara (i) crosstalk
(ii) thermal noise (iii) l./f noise. Besides. interferences may be
j
76
preaen1: in two ways (i)conouction and (ii) radiation. These
will be discussed in the following sections.
5.2 Thermal Noise
SdYV\.€-Each pair of telephone cable has elillH'Saresistance. So.
thermal noise is present in each pair which is dependant on the
ambient temperature. The meximumthermal noise power 1:hat can
be delivereo to the load is KTB(B • bandwidth) which takes place
when source impedence is complex conjugate of loed impedance.
5.3 lIf Noise
Such noiae arises due to the loose contact. Mathematical
expression is given in Chapter-2. Loose contact ari ses sp ecially
in PE(polythenej( insulated) cables where normally jointing is
made by t\~isting the conductors without using soldering. Besides
in lead cable Ilf noise may arise due to dry soldiaring in joints.
5.4 Cross-talk
Croes-talk is defined 8S the disturbance created in one
communication circuit_by the signals in other communication
circuits. So it is a sub-division of the general subject of
interference. It is the disturbance created by another circuit
due to circuit nonlinearities; transmittiance and electromagnetic
coupling. The disturbance created by another circuit due ~
electromagnetic radiation is not considered as cross-talk, it is
rather considered as interference. Cross-talk ",as originally
77
used to indicate the presence in a telephone receiver of unwanted
speech sounds from enother telephone conversation. But i twas
found that the methods for computing such craBs-talk quantitatively
••ere applicable for nontelephone circuits also. So; the ttal:l1l crOS8-
talk ultimately broadened to apply to interference between eny
kinds 0 f communication circuits. J
Cross-talk may be intelligible or unintelligible e.g. croes.
talk between unlike channele (in a 'carrier system) is usually
unintelligible because of frequency inversion. frequency displace-
ment or digital encoding. On the other hand ere ss-talk in multi-
paired telephone cable illl intelligible as there is no question
of frequency inversion or displacement. Undesired intermodulation
products in a fDMsystem are usually un-intelligible crose-talk.
Intelligible cross-telk is more annoying than unintelligibl.e
one and also objectionsble because of loss of privacy.
In case of nonvoice circuits;intelligible crosstalk means
unwanted signal 0 f the same type las the desired signal. Nti;rmally.
unintelligible cross-talks both in voice and nonvoice circuit
are treated as noise(2).
In this chapter; crosstalk in multipared cable will be
considered. Cross-talk in fDMsystem will be di scu ssed in
Chapter-6.
5.4.1 Crosstalk in Multip~ed Telephone Cable(2}.{1)
In multipaired telephone cable croastalk arises due to the
different types of coupling between conductors. Coupling mey be
78
by induct,ance. capacitance til' by resistence. The coupling
phenQmenon may be illustrated by using the figs. 5.1. It is
assullled that transmission paths are short in terms ,of wave
length (at f = 3kHz. A = 1000 kH; «luI telephonecab&e length
is much below than this value of f.- ). Thus inductances and
capacitances cen be considered as lumped element parameters.
The two capacitors shown as cu/2; characterize the unbelanca
capecitence; c; between the circuits. components of c includesu u
all the direct capacitances shown in fig. 5'.1{b)(2). Capacitances
to ground are neglected. Thus;
••• 5.1
ascIJ and c24 produces, disturbance in the same direction and
C32 and c14 also produces disturbance in the same direction
which is opposite to that of c13 and c24• The two circuits are
assumed to be connscted for maximumpower transfer II i. e &0 = ~.NO~J.current io in the disturbilOg circuit causes a voltage;
va '" ioZo; to appear across tiffs circuit. In the disturbed
circuit. this voltage causes a current;
Normally;
'lZ.i=ci .4a 04/jwc + Z /2
u "
••• 5.2
%.1 •• jwi Z c /4C 0011 ••• 5.3
A current of ielffl = jwioZoCu/8 appears at each end of the dis-
turbed circuit BS shown in the fig. S.l{ a).
5.1 (b) VARIOUS CAPACITANCES BETWEEN CONDUCTORS OF CIRCUITS.
I - 2 _ DISTURBING
CIRCUIT
3 - 4 -+ DISTURBED
CIRCUIT'4
1
:3
i"' ..
20
•
..•..fm,.-
t
C~2.
2.0
Oe
FIG.5.1(0) CROSSTALK BY INDUCTIVE AND CAPACITIVE
COUPLING BETWEEN TWO CIRCUITS
0° Ob
C>Z
~<Il~<Ila:UJ>a:UJ UJ<Il <Il~ ~••• Q.o <Il
UJUJ a:~ 0I- UJ
Z ~UJ UU _a: 0UJ ZQ. -
100
90BO70605040302010
o 15 65
dO
FIG. 5.2. FOUR CONDUCTORS IN A QUAD OF A MULTI PAIRED CABLE
CONTACT
"til I~-NOISE LEVEL AT LINE TERMINALS OF STATION SET IN dBrne
FIG. 5.5. NOISE JUDGMENT CURVES
(RECEIVED VOLUME CONSTANT, - 2B VU)
. FIG. 5.3 A SIMPLE NETWORK FOR ARC SUPPRESSION
79
Now. the effect of mutual inductance'Mtmay be cDnsidered.
currentio in the disturbing circuit produces a voltage vm••iojwM
in the disturbad cirt:uit. This voltage causee a current.
••• 5.4
The shielding effect of intervening conductors or shields may
reduce the unball!lnce capacitance Cu and the mutual inductance
M. and thus reduce theeurrent coupled into the disturbed circuit.
Equations 5.3 and 5.4 are the coupling crosstalk cur~entl> due
'to tha capacitive and indu c'ti ve coupling respectively. Thase
may be classified into 'two groups: (i) Near end cross'talk (1i)
far end crosstalk. Near end cTosetalk (NEXT)is cr~sstalk whose
energy 'travels ,in the opposite dir'ection to that of the signal
in the disturbing circuit. It occurs,at near end. Thus, from
Fig. 5.l(a). N£XTcurrent ie given by in = i - i and thec mNEXTratio is
inti = jwc Z /a + jwM/2Zo u 0 0 ••• 5.5
Similarly, far end crosstalk (fEXT) i,s crosstalk whose energy
travels in the same direction as the signal in the disturbing
circuit. Thus rEXT current is if" ic + im and rEXT ratio is,
if/i •• jwc z /ao uo • • •
From equations 5.5 and 5.6 it is clear that capacitive crosstalk
and inductiva crosstalk oppose each other for fEXT,and add for
NEXT.Thus NEXTis mo,rs serioue problem., The degree of cancella-
tion in the far end depends upon the magnitude of unbalance
go
capaci tance. indu ctance and Zoo from eqn. 5.6. it is seen that
for high impedance circuit (high lol. capacitive coupling is
the most significant moda of croastalk interference. On the
other hand for low impedance circuit inductive coupling is
more significant.
The two types of crosstalk discussed above are direct
in nature as they arise due to the direct coupling between
disturbing circuit end disturbed circuit. Crosstslk may ba
arised also by indirect coupling. In such case., the crcsstaJ,k
is callad indirect crosstalk. Coupling between the disturbing
line and disturbed line by way of a tertiary circuit, is called
indirect coupling. I ndi rect cro sstalk can be subdi vided into
two groups:(i) transveree indirect (ii) interaction indirect.
In casa of trensverseindirect crosstalk. ,crosstalk currant,
occurs via the tertiary path but it does not flo>'J through the
tertiary path. Such crosstalk occurs when tertiary circuit
(s. g ground return circuit) changes capacitance unbaJ.ance between
two circuits. In balanced system any unbalance to tertiary. . . - . .'.
rel!lults in transverse indirect cro sstalk. e. g resistan ce unbalance
caused by unequal wire diametars or poor joints and inductance,
unbalance caused by one conductor being unsymmetrically wrapped
eround its mate.
In case of interaction indirect crosstalk. tertiary circuit
elsa carry crosstalk lI:iclrll:QIi::t current. Such crosstalk occurs
specially in repeateredsystem. Coupling takes place at the
repeater aections. These are four types viz. Near-end near-end.
81
far-end far-end; far-end Near-snd and Near-end fer-end. Besides;
additional crosstalk occurs in any cable when mismatch exists
in the cable. Hismatch in disturbing circuit causes some energy
to be reflected which by coupling appears at the disturbed circuit.
Mismatch may occur at cab1.e joint; repeaters etc.,
I t should be noted that to reduce crosstalk in multipaired
cabla (where capacitence is very high as the conductors are
closedly spaced and which is the main ceuse of .erosstalk) the
conductors are .grouped in four to form. a quad as shown in
fig. 5.2. Conductors lllilm e and b form one .PElir and conductors
c and d form another jiair. In this arrangement both wires .of
transmission line a-h ars equidistant from c and d condoctors
of other line. This causes to neutralise tha induced effect
(both inductive and capacit:i.ve) on line c-d by conductors' a'
of line a-b; by induced effect on line c-d by conductor b of
line a-b, In other words; the two transmission pair in a quad
are balanced and so no crosstalk occur. A multiconductor cable
contains many pairs of conductors arranged in quad form. Trans-
positions are provided to reduce crosstalk between neighbouring
quads by continuously rolling together the four wires of each
quad throughout the entire length of the cable.
But in practice capacitance and inductance unbalance
takes place between the pairs in a quad and pairs between quads
due to the follo~Jing reasons:
(1) The dismeter of the copper conductor mayvary slightly,
(2) Thickness of the insu1.tating material may vary .•
(3) Some conductors may be longer than others due to
twisting of quad conductors and layers.
(4) The formation of the quad may not be regular throughout
the lengthDf the ceble. ,
The capacity unbalance, the main cause of crosstalk, may
be compensated by adding capacitances between conductors at the
joints. Before doing so, all quads in each length of a cable
in a route are tested for capacity unbelance. Quads in each of
the cable lengths are then chosen and jointed together so that
capacity unbalances of quads in one length of cable .are compen-
sated by that in another length of the cable. The remaining
unbalances, if any, may be compensated by using extra capacitances.
5.5 Conducted Interference j
The previous di scu ssion co vers noi ses that are generated
within the cable itself. But outside interfelrences may alSD
present within the cable. Such interference may enter into the
cable in two ways viz. (i) by conduction (ii) by radiation.
Interference by conduction i.e conducted interference will be
discussed in this section. Ths following sectian will cover
radiated interference.
Equipment must be connected to other equipment, equipment
must be connected to commonpower system, and equipments -also ..
must be connected to commongrounds. Signal from one equipment,
thus, can enter into the another equipment by such connection.
83
This i. called conducted interference. Thus conducted interference
is that interference which enter into a circuit by conduction, Ceble
pairs of a mu1tipaired cable are connected to various equipments
in the exchange. Unwanted signals from these-equipments can enter,
directly into the cable by conduction resulting conducted interf'er-
ence. These unwanted signals must be removed as much ae possible
by proper filtering. Also. the conducting medium mey be made in
such II wey that it does not pick up unwanted energy. Of course, if
the interfering signal falls in the voice band; then some times it
bec.mee very difficult to eliminete it end ultimately it eppears
at the receiver as interference.
5.6 Radiated Interference
The interference that enters in the cable (or equipment) by
radiation is called radiated interference. The magnitude of inter-
ference that appeers in the circuit depend on the following _factors:
(1) Physical length of the exposed conducting medium
(2) The intensity of unwanted signala or fields to which the
conducting medium ie exposed.
(3) Particular types of interfering fields involved.
(4) Ths impedance of the termination to which the perticular
conducting medium connects.
Shielding is the only practical method 0 l' suppressing inter-.
~erence which is radiated directly from a source. A perfect shield
will not allow the passage 0 fai ther electrostatic or electromagnetic
energy. Shielding partially reflects the interfering signal, and
absorb the rest. The absorbed portion is attenuated as it passes
through the metal shield~ A good shield to contsin the interference
should setisfy the following requirements:
(1) It must be able to confine undesired signals that are
generated within the case (e. g. signals in the conductors of elle
cable are interference for another nearest cable) or reduce them
to such an extent that they do not cause malfunctioning of another
cable (equipment) .•
(2) It must be able to prevent inside cable (equipment) from
.eceiving undesired signals that exist in the erea •. The interfering
signels may be from adjacent power lines; open wireptelephone
line. telegraph line. any nearest electrical arcing etc.
The efficiency of shielding is measured by the term 'shielding
effectiveness'. It is defined by
'...where 5 is the shielding effectiveness and R is the reflection
loss in dB and A is the absorption lass in dB. for ref~ection loss
in metallic shields. R is given by( 1) •
R "" 20 10910 ••• 5.8
where
Z isw
Z is the intrinsic impedance of the metal barrier and8
the intrinsic impedance of the reflected electromsgnetic weve.
••• 5.9
where f •• frequency; G = relative conductivity of metal referred
to copper. u = relative permeability of metal referred to free
space, and t •• thickness of the shielding meterial in mils.
Another term known as "shield factor" is also used which,
is defined by the ratio 0 l' the resultant voltage in the interfered
with circuit, after the shielding is used, to the nonshielded int •• -
fered - with voltage.
It is found that e metallic shield that is ungrounded will
reduce at least one half of the interference. where.Bs+ a met~llic
shield with a good ground, ",ill reduce about three quarters of the
interference(l). Thus it is practical to ground cable .!Shiel-ds.
This also assures that potential differences will beminimmzed in
shields and so little current flow will occur in shields •./"
.Interference in Tele hone Exchan e
Various types of machineries and equipments ere used in the
exchanges. These are the sources of interference e.g. d-c rotating
machineries like electric motor, generator, drill etc. generates.'
interference due to arcing produced in the process of commuti3tion.
Similar type 0 f interferences occur in motors containing alip. rings.
Oxidation is another perennial sourcs Q f intier.ference in rotating.
lIIechineries. Oxidation film is generally formed irregularly causing
variations in the sliding control resistance which in tern causes,
the d-c current to vary. resulting in r-f interference. The oxidation
film also sets up a rectifier condition which makes transients
possible since the resistance vary non-linearly; depending on whether
86
the brush is positive or negative.
In machines where rotor and stator are not electrically
connected. interference develops because of the discharges of the
electrostatic energy which builts upon the seperate moving parts.
This becomes important specially at high speeds. where. for example;
the greaee or lubricant between the inner and outer 'cases of a
ball bearing acts as a dielectric. Energy buields up Qn,both siElae
end eventually discharge takes place across the dielecttic. Of
course a conductive lubricant would eliminate this problem. however.
this technique is not yet used commercially. "nottler successf'ul
technique is to ground the rotating shaf't by use of slip ring. In
case of'commutetor machines, capacito,rs are connected. to reduce,
interf'erenc,e. across bruah holder and machine f'rama. It is obvioua
that machines containing no com"".utation or slipringproduc;es, Ileg:l.i~
gible interf'erence. Thus induction motors are best for this purpp~
Interf'erence due to arcing also generates at switches, relays.
flourescent lamps etc. Speciel networks can be designed for arc
suppression. These networks eliminates. as much as possible., any
transient that may occur during a switching operation. A typic~l
circuit ia shown in Fig. 5.3. When contact is made. the instantaneous
current is low as the diode is r,eversed biased and current through
the inductance changes slowly. Whencontact breake, .a negative<==> . -------- - -surge voltage appears across the .1oad and so current. flo,",s , through,
the diode. At this point. the instantaneous voltage across, 1;~con1;act--,~- - . r ~ -
is the supply voltage plu s the small voltage drop aeross the diode •.~ . . ~ --------
Other types of suppression networks, ar~sg~ed.__ - _. __ -_" ~ __ ~ __ ~ d~
'87
An important source of crosstalk interference ie the battery
internal resistance. Since a battery supplies power to different
ci rcui ta, coupling tak es place by its internal resistance. The
lower the internal resistance, the lower will be the crosstalk
interference. The value of internal resistance becomes fixed during
th edesign of the battery. But thi s resistance increases if the.
battery is not maintained pxoperly. Besidas; if power drawn from
tha battery is small er; then current from the battery will be
stnsller and so voltage drop across the internal resistance will
be lower causing lower crosstalk. Obviously; if different batteries
are used for different circuits then no crosstalk will arrise. But
it is not pOl!lsible due to economic reasons. Of course ,di fferent
batteries must be used at least for networks drawing apprecieble
and changeable powar. e. g. relay arcui t current changes randomly
with the change in traffic demand. This will causa an intarference
if commonbattery is used for relay circuits and signal circuits.
Besides, although arc supression reduces interference, .a certein
mmount cannot be eliminated completely and this residue is conduc-
ted or radiated awey. for these reasons. indep endent power supply
for relay circuits are used •...-. ,
Another i~portant crosstalk interference producing parameter
is ground contact resistance. I f grounding ie net proper,greund
contact resistance, lik e bettery internal resistance causes coupling
of various signals. Hence attantion must be given for proper groun-
ding. The cause of grounding is discussed in the next sllction.
5.8 Role of Grounding in R!!lducing Interference
The most general and the most important reason for grounding
of Bny equipment. building, structures etc. is to make sure that
no dangerous voltages are present on a=essible psrts of equipment.
building etc. that will endenger the safatyof persons who come
in contact with it. Under this general raason. thare are other
reasons for grounding. one of the most important is the reducti.on
of interference. Another is for the safety of the equipment. Power
system is similarly grounded mainly for (1) protection of equipment
(2) easy fault location and (3) safety to persons.
But in telecommunication system. in addition to the safety
of equipment and person grounding helps in two ways:
(i) redu ces i nt erference
(ii)seva a conductor,
Interference reCuction by grounding of equipment body and chasis
is obvious. If bodies of theequipments are I!It same potential and
grounded. then all unwanted signals will go, to ground. Saving of
a conductor by grounding can be shown by assuming that l!l subscriber
of an exchange l!I A wants to make a connection tO,one.of the subs-
criber at another exchange B. Obviously 1st group selector will
be provided by exchange A and other group selectort'! will ba
provided by xlll exchsnge B. Now to control the selectors in exchsnge B
by the subscriber 0 f exchange A. the battery, terminals 0 f exchange A
must be extended to exchange B. for this purposa one terminal of
the ~X~~R.battery is groundad and the ground acts as one of tha
conductors to extended the battery to the other exchange.
8'1
Thus g~unding can save a conductor. Of course. ground contact
resistancs; Rg. must be very small. Otherwise switches will not
function p~perly and coupling by fig will also ba appreciable.
for p:r:oper grounding various factors a,re i:1Sl be considered.
Important fsctors are (1) ground rod. (2) depth. and (3) treatment
of the soil. ,JGround rods: Ground xods are used for the purpo se of grounding
which are driven into the soil. These may be pipes or rods. A
few long rods are usually more satisfactory thsn many short rods
since volume of soil effected increases directly wi1;h the .length
of the electrode below the surfsce. Soil resistivity generally
decreases with depth due to increased moisture contant. Driv!3n
ground rods are by far the 1Il0st generally used type of ground
electrodes. This is a result of the comparative eese of il'l3ta:1.1ation
and penetretion through relatively great depths .•Buried plates may
al so be used. NOrmally plates are used only in locali ti es where
cond~tions make it impractical to install driven. rods. Parallel
ground rods reduces fig. Ground rods sho~~d protect corrosion to
themselves. Size of the ground rod is not much important as. e.g •.
it is found that a single ground rod of diameter 2a does. not gi~e
as Iowa ground resistance that obtained by paralleling two rods
each of diameter 'a'.
Depth: Soil resistivity decreases with depth. This is mainly due
to the increased amount of moisture in soil with depth. I\t a depth"
of 20" or above the resistance (Rg) la.lIlIKlliSli becomes less than 20 Ohm
(for untreated soil). With further increase of depth the resistance
does not change appreciebly. Another illlportant l'actor that in1'luen-
ces soil resistivity is ambient temperature.
Treatment 01' the soilJ The ground conta~t resietance can be
reduced appreciably by treating the sur~undingsoil. The treatment
merely consists of mixing the soil around the ground rods with l'ine
commonsalt bel'ore the soil is compacted into space.
There a re number 0 f l'actorswhich causes ground contact
resistance (Rgl to very. Hencs it is necessary to constantly
monitor the low impedance path to ground. J: f these ground. rod
connections are properly tested and m.aintained.j. a good ground
connection will always be available. Various methods. are there for
measuring ground. resistance. These will be found in the references:
5.9 Measurements and Results
~~> ~~~~..p' Measurements for noises in cj,;r~Sj connecting tWQsub~ ••,~.~cr;i,.b.a,r-&-V3.-i!!exchange.s; were made by using psophometer., The .connac:tion
diagrams are shown in fig. 5..4". Results are given in tables 5.1(a).
5..JJ.b.)-end !i.l(c.) ..• Clesrly. results contain all aortsol' noi!3es
in cable and switchingnetwo:rk~Che difference between. the maximum
and minimum value is due to the impulse noise caused by dialling
and switching impulses. Since. dialling and switching impulses
vary with time;. it is clear that maximumvalu e ""ill a1so vary with .
time. But during busy hour, it is reasonable to assume that dialling
and switching impulses ara maximum.~ince. measuremantswere made
during busy hour. it may be assumed that results closely epproximates
-CALLING SUBSCRIBER CALLED, SUBSCRIBER
O~---I E~~~:NA~E I, '0(TRANSMITTER IS REPLACED
BY MATCHED TERMINATIONAFTER THE, CALL IS ESTABLISHED l,
CALLING SUBSCRIBER
(TRANSMITTER IS REPLACEDBY MATCHED TERMINATION
AFTER THE CALL IS ESTABLISHED)
SHER-E-BANGLA
NAGARE~CHANGE
JUNCTIONLINE CENTRAL
EXCHANGE
t PSOPHOMETER AT THE RECEIVING' TERMINALAND TRANSMITTER IS REPLACED BYMATCHED TERMINATION)
CALLED SUBSCRIBER
(PSOPHOMETR AT THE RECEIVINGTERMINALAND TRANSMITTER IS REPLACED BY MATCHEDTERMINATION)
CALLING SUBSCRIBER CALLED SUBSCRIBER
O I GULSHAN I '0' ----------------------- EXCHANGE,
ITRANSMITTER IS REPLACED
BY MATCHED TERMINATION
AFTER THE' CALL IS ESTABLISHED)
AG.5.4 SET UP FOR HiE MEASUREMENTOF NOISE IN CABLE AND
SWITCHINGSYSTEM. (PSOPHOMETER USED IS SIEMENS TYPE)
(PSOPHOMETR AT THE, RECEIVINGTERMINALAND TRANSMITTER IS REPLACED BY MATCHED
TERMINATION l
•
Table 5.l( a) Noiee valu ee for various interconnectiona between aome sub !3cribers 0 fcentral exchange (28 and 25 only)
•(Readings were taken in' the period between 9 A.M and 12 noon.and noise values shown are weighted).
Teb1" S,l(b) Noise values for various interconnections between sOlliesubscribars of'central exchange and some subscribers of iher-e-Bangll1 Nagar exchange.
Serial No. 1 2 3 4 5 6 7 a 9 10
Interconnection between different telephones. -- -~---~._._._- ----
Table 6.6(a) 5NR for different chenne1. in the ~TD circuit between Dacca end Chi_'.Qong., , , ,I
,I I , I I • , ,
I • , , , I , , I I I -r~o. MDe 5.2.80 11.180 18.2.80 21.2.80 «of of 51gn.l 1:O.1n SNR 5ign.l Noi •• SNR 51gne1 Nohe SUR Signal Noi.e StJR AveregeChen. Chan. in in .1n in in in in in
dB•• dBlll dB•• dB•• dihI dill! em •• em ••Treffic
(Reedinge were t.ken between 11 AM end 12-00 noon and noi.ee .hown are weighted)
« Avalrege SNR of 5 channels 1. 47.23.
Tabh 6.6(b) JNR 'D~dif'e~ent chenne1e in the 5TD ci~cult betwe.n Decca and Chittagong• , ! • . , , • • • • • I , 0 p : • I • • I
20.5 go 21 5 80 22.5.g0No. 0' Name 0' • SIgna! ~Qt.e sNR Averags«Ch.n Chan" Sign.l i~Dtse SNR Silln.1 Noi •• ~NRin in in in In indBm dBm dBm Ib dB•• dB.
"Modulation, Noise and Spect:ralAnalysis". McGraw Hill Book- CompanyInc., 1965 P.
"Statistics on Attenuation of'Microwaves by I ntense Rain",Bell System Tech •• J., Vol. 46,Novsmber •. 1969,
"pows:r Density Spectrum of theSun of Two Correlated Intermodula-tion Noiee Cont:ributors inFMSyetems"'. Bell System Tech. J,Vol. XLVI. December, 1967.