This document puts into
perspective the various terms associated with structured cabling. It
is not conclusive and will evolve as the standards gain additions and new
terminologies emerge. As you will see, some of the terms have yet to be
explained and are presently listed as a reminder that they exist and have
Decibels: The Intensity
(I) of any kind of traveling wave, be it sound, light, water etc.
is given by the formula:
Where I is
measured in watts/m2, p is the density
(kg/m3) of the medium in which the wave is traveling, w
is given by 2n/T where n is pi and T is the period of
oscillation for a Simple Harmonic wave. In addition, x0
is the amplitude of the wave (m) and u its velocity (m/s).
W is also given by
2nf where f is the frequency of the wave. W is called
the angular frequency of the wave.
For the case of sound
waves, the smallest sound intensity (I0), detectable by
the human ear is approximately 10-12 W/m2. A tube
train is 1010I0, whereas a jet aircraft is
So, if I =
I/I0 = 1015
The increase in
'loudness' of a sound is dependent upon the ratio of the intensities
rather than the absolute differences in intensity. The increase in
loudness is calculated by taking the logarithm of the ratio of the
intensities to base 10 and this is measured in bels. So from the
The noise of a jet
aircraft expressed in bels is given by log10
(I/I0) = log10 1015
Using the laws of
logarithms this gives 15log10 10 = 15 bels
Normally working in
bels is a bit cumbersome so the standard way of working is to use
decibels, which is one tenth of a bel. So the noise of a jet
aircraft is about 150 decibels, given by 10log10
(Ij /I0) relative to the lowest intensity of
sound that a human ear can detect.
For an example, if the
volume of a Hi-Fi system is turned down such that such that the power
decreases from 1000mW to 200mW then the change in power, in decibels is
= 10log102 - 10log1010 = -7dB
The minus sign
indicating a decrease in power.
So when one states that
there has been a loss of x dB across a connector or a circuit, this
means that there has been a decline in intensity of a given signal as
indicated by the ratio x.
In the structured
cabling world decibels are used as a measurement of noise, attenuation,
and signal loss.
The length measurement
is achieved using Time Domain Reflectometry (TDR) techniques, a pulse is
sent down the line and the time that it takes to receive the pulse back at
the tester transmissions device is used to determine the length of the
cable. Different type of cable will produce different results therefore a
secondary parameter is necessary that distinguishes the type of cable
being tested. This parameter is referred to as the Nominal Velocity of
Propagation (NVP) and is expressed as a percentage of the speed of light
(c.). Typical values for Category 5 cables range from 0.65c. to 0.85c. If
the incorrect NVP is entered into the tester the results can be totally
misleading where a 91 m link appears to pass the test whereas the true
reading may be a failing 105 m link.
Unbalanced and Balanced
circuit has one of the wires in a pair grounded at both ends. An example
is EIA-232. The problem with this is that any noise produced from external
sources such as voltage switching, ballast from fluorescent lights etc.
will add to the signal and be seen as data. This is because any noise
appearing on the ground wire is sunk to earth whilst still adding or
detracting to the signal in the signal wire. This gets worst at higher
data rates and over longer distances. Using a shield ground at both ends
helps to alleviate the problem
A much better way to
resolve this issue is to use a balanced circuit. A balun, a small
transformer, isolates the wires from the circuitry and instead of passing
a 0v to 2v signal, for instance, it passes a -1v to +1v signal such that
each wire passes the opposite signal of the other wire's signal. In
theory, the resultant EMF should be zero. The better quality of the
cabling system components the more likely the signal fields cancel each
other out. As the signals travel the wire the same amount of noise is
picked up by both wires resulting in their being zero difference between
the signals at the receiving end.
There are considered to
be two types of transmission in a balanced cable:
where the conductors in a pair carry the same signal but opposite in
polarity. The assumption is made that the transmission line is infinitely
long and is perfectly uniform and that there is zero radiation.
where the sent signal is not perfectly balanced and the twisted pair are
not perfectly balanced. This results in the two signals on the conductor
not completely canceling so that a net current is induced on the pair thus
producing radiation. EMF from surrounding systems is coupled onto the
imperfections mean that some of the differential signals convert into
longitudinal signals (Longitudinal Conversion Loss so both
differential and longitudinal modes need to be examined when looking at
EMC and cross-talk (Differential Mode NEXT). This loss is called
the balance of a cable and is measured at one interface, typically the
transmitter end since this is where much of the EMC is likely to occur.
There is another measurement called Longitudinal Conversion Transfer
Loss, which is measured between the two interfaces at the ends of a
link and so takes into account the differential noise at the receiver.
means that a current flows down both conductors in the same direction and
uses the earth as the return path. Common mode noise current is often
'shunted' to earth via an earthed center tap on the receiver
is where the signal is diminished due to losses incurred throughout the
transmission medium be it fiber or copper. For copper, it is normally
measured in decibels per 100m, meaning the ratio of intensities between
the far end of the 100m and the starting end for a given signal, a ratio
of power out over power in. In copper, attenuation is largely due to
copper loss and to dielectric loss from the jacket materials used around
the wires. Polyethylene and Teflon are currently the best materials that
minimize dielectric loss, or dissipation. Attenuation over a certain
length of cable is linear for a given frequency but is different for
different frequencies and temperatures
fiber is far less than that for copper. Lucent Technologies guarantee that
for multimode fiber there is a loss of 3.4dB/km at 850nm, whilst for
single-mode Truewave fiber the loss is 0.22dB/km at 1550nm.
of a cable or connector is given by 10.log (received signal power
without cable/received signal power with cable).
How to measure
parameters such as attenuation, NEXT and Structural Return Loss is defined
in ASTM D 4566.
maximum insertion loss allowed for a 10BASE-T link is 11.5dB at all
frequencies between 5.0 and 10.0 Mhz. This includes the attenuation of the
cables, connectors, patch panels, and reflection losses due to impedance
mismatches to the link segment.
Nominal Velocity of
a conductor, electrons travel at near the speed of light. For a copper
cable this speed is often expressed relative to the speed of light. For
Lucent's 3071 cable the NVP is given as 0.69, meaning that electrons are
traveling at 69% the speed of light. The lower the NVP, the greater the
delay in signals reaching a destination, so a high NVP is required for a
good quality cable.
The NVP of a particular
type of cable can vary between batches. This variation can be as much as
20% so if cable testing is carried out then it is important to implement
NVP calibration of each batch and adjust the default NVP value (built in
to the tester) to the measured NVP for the particular batch. NVP
calibration involves using the tester to measure a known length of cable
and adjusting the tester's length measurement accordingly.
The wire map test is
crucial on UTP/STP systems as if this fails then it is likely that most
applications will not even begin to work. The wire map test is simply a
check to see that wires are connected one-to-one at each end.
As well as shorts
and open circuits, reversed pairs, crossed wires and split pairs are
Capacitance: A capacitor is a
device that stores electricity and it consists of two metal conductors
surrounded by a dielectric. When there is a circuit attached to the
conductors (such as used by a carrier signal down two wires) a potential
difference exists across the two conductors. Whilst this potential
difference exists, electrons flow within the circuit to the conductor on
the most negative side of the circuit whilst electrons also flow away from
the most positive side. This happens until the potential difference across
the conductors exactly counteracts that of the attached circuit. If the
circuit is disconnected then the two conductors maintain the potential
difference until they are connected together. Once this happens a current
flows for a short period of time as the electrons flow in order to bring
back a zero potential difference between the conductors. This is called
measured in Farads and is given by Q/V, where Q is charge
(measured in coulombs) and V is the potential difference (measured
Because a farad
is a very large unit of measurement you will more commonly see microfarads
(10-6) or picofarads (10-12).
Capacitance of a cable is that which exists between the two conductors
of a pair. The higher this is, the more likely the possibility of there
being interference between the wires. EIA/TIA 568A sets an upper limit of
5.6nF/100m of Category 5 cable.
Being the AC equivalent of DC resistance, impedance of a data circuit
takes into account not only the resistance of the copper wire but also the
reactance of the cable capacitance. For an alternating current the
instantaneous voltage is given by:
the reactance of the capacitor and the voltage across the capacitance lags
that across the resistance (R) by 90o.
By use of
trigonometry (and without going into details) we end up with the impedance
of a data circuit being given by:
Z = V/I =
(XC2 + R2)1/2
We can complicate
matters by also including the reactance of the inductance of a cable in
the maths since strictly this also makes up impedance, however to simplify
matters we won't.
Impedance is the impedance of a particular circuit or cable at a
specific frequency. The characteristic impedance of a cable is the
specified and manufactured impedance for that cable. This value is
obtained by taking plotting the input impedances for each frequency over a
range and applying a smoothing function to the curve. The Characteristic
Impedance should remain constant throughout the length of the cable,
however variations from this characteristic impedance will be due to
faults within the cable or the connecting hardware. Time Domain
Reflectometry is used to find impedance defects. Category 5 cable is
designed to have a characteristic impedance of 100ohms. The specifications
require the impedance to not deviate from 100ohms by more than +/- 15 ohms
for each frequency from 1MHz up. All components within a cable link should
match each other at each frequency since an impedance mismatch will hinder
coupling of the signal from one component to another and result in
reflections and attenuation.
Standards such as
IEC 1156 and ATM D-4566 define impedance measurement on long cables.
(See EIA/TIA 568A
DC Loop Resistance:
the resistance of the wire pair when short-circuited at the far end. This
is important for Token Ring that uses relays to allow in and shut out
stations from the Multi-station Access Unit.
Pair to Pair Near End
Cross Talk (Pr-Pr NEXT)
defined as signals that have been induced, or coupled, from one active
pair of wires (disturber), to another. The name comes from the
effect occurring within multipair telephone cables such that other
people's conversations could be heard on the line that you yourself were
using. The current from the one pair of wires was inducing an emf within
an adjacent pair of wires.
Pair to pair Near End
Cross-talk describes the noise from the transmit pair coupled onto the
receive pair of wires within the same jacket and at the near end,
meaning as applied to the receive pair nearest to the transmit end. In a
four pair cable, if one pair is used for transmitting a signal, then some
noise will be induced on to the other pairs to varying degrees. Cable
balance, twist rates and cable spacing are all factors that help reduce
Cross-talk. There are also Cross-talk cancellation methods available.
NEXT is measured in
decibels, and is the difference in amplitude between the actual signal and
the Cross-talk signal. The NEXT effect is often expressed as Cross-talk
isolation or NEXT Loss, meaning that the higher the Cross-talk isolation
the less the coupling effect. NEXT is effectively a measure of the
attenuation between pairs, therefore, a high NEXT Loss is good!
The Cross-talk signal
is strongest at the transmit end of the pair and the next 20m. For this
reason the socket at the far end is excluded from NEXT calculations as it
has minimal impact. The NEXT is calculated by anti-logging each noise
component and then summing the noise (measured in mV) before logging the
sum. To verify the Channel quality, EIA/TIA 568A recommends testing both
at the frame end and at the socket end since the far end is not included
in the test itself. When this has been carried out, then you will see two
sets of results, one for the near end and one for the remote end. Some
testers allow this to be carried out without having to swap the tester and
the remote module around.
The tighter the twists
within a cable the less chance the pairs have of sharing space with other
pairs and thereby incurring coupled noise. Having different twist ratios
per pair results in each pair having a different lay length (the distance
between full twists), the aim being to minimize, as far as possible, the
distance that conductors lie next to each other.
Power Sum Near End
Cross Talk (PS-NEXT)
This is the cross-talk
between one pair and the combination of all the other pairs within the
same jacket. This is important as more and more applications use all pairs
within a jacket (e.g. Gigabit Ethernet). Power summation is not limited to
4 pair cables, it has been used in the production of larger pair count
cables such as Lucent's 25pr 1061C cable.
Far End Cross Talk
FEXT is the coupling of
the transmit signals on to the receive pair or wires at the far
end, away from the transmission source on the disturbing pair. Expressed
in dB, it is described as the ratio of the power induced
(Pi) over the power of the original signal
(Po). FEXT = 10log Pi /Po.
The higher the FEXT isolation in decibels, the less the coupling effect.
FEXT is particularly important for data systems that use parallel
transmission such as Gigabit Ethernet (802.3ab) that uses dual duplex
PAM-5. NEXT and Delay can be minimized with electronics, but this is not
possible with FEXT.
Power Sum Far End Cross
This is the powersum
equivalent of FEXT.
Equal Level Far End
Cross Talk (ELFEXT)
Measured in decibels,
this is the ratio of the desired receive signal strength to the
strength of the noise induced by the transmit signal at the other end.
This is the same as FEXT - Attenuation. The effect of attenuation on the
original signal has been added to give a more realistic result.
Channel Equal Level Far
End Cross Talk (Channel ELFEXT)
ELFEXT is most
significant for cable, however, in the proposed Category 6 standard
Channel ELFEXT will be important as this includes the FEXT of the
Power Sum EL Far End
Cross Talk (PS-ELFEXT)
This is the Powersum
extension of ELFEXT and is calculated by summing the antilog of the ELFEXT
for each pair, and logging the sum.
Terminated Open Circuit
Pair to Pair
Attenuation to Cross Talk Ratio (ACR)
ACR is measured in
decibels and is equal to NEXT Loss - Attenuation and is a common method of
specifying a cable's performance. ACR is calculated for each pair and for
each frequency measured. For Class D operation the ACR must be 4dB at
100MHz. ACR gives some latitude when looking at a cable's performance in a
cabling system, which has a large variety of lengths. A short length is
likely to have relatively poor Cross-talk loss but low attenuation,
whereas a long length will have relatively good Cross-talk loss and high
ACR is a very important
measure of how well an installation carries data. At a frequency of 200MHz
you should be looking at a good 6dB difference between the Attenuation and
Cross-talk curves indicating a 75/25 Signal/Noise ratio. Compare this with
a 4 dB difference giving a 50/50 Signal/Noise ratio.
Often, margin values,
in decibels, are quoted for ACR measurements. These express how far away
the ACR value is from the limits imposed by the particular standard being
measured against. A positive value indicates that the ACR value is above
the standard and is therefore OK.
Power Sum Attenuation
to Cross Talk Ratio (PS-ACR)
Alien Near End Cross
This is the noise
induced on a cable by another cable running parallel to it, whether
signals are being transmitted or being received. The more cables that
there are within a bundle the more mixed is the induced noise. Even
adjacent bundles will add to the effect.
As stated earlier
Cross-talk effects are most prevalent closest to the transmit end. To this
end it is important to make sure that the cable bundles leaving the patch
frames are not all running in parallel to each other. This includes the
bundles sitting in trays and loose laid on the floor.
It is also important to
make sure that the cable management allows the patch leads to be loose
laid and not be running tightly together in parallel. This loose laying of
patch leads is important as it helps to attenuate the Cross talk before it
enters the cable bundles. Provided that you keep to patch lead length
guidelines, the longer the patch lead the better as this gives more chance
for the alien Cross-talk to attenuate.
Bends and tight cable
ties affects the balance of the cable and therefore makes it more
susceptible to noise as the cable is more likely to radiate noise and also
Pre-bundled cables need
to be balanced with each other before being installed on site.
nanoseconds, this is the time required for a signal to travel the whole
channel. Quantifying delay is important as protocols such as Ethernet rely
on the time it takes for frames to traverse a LAN. With CSMA/CD, if there
is a collision, the circuitry of the sender waits a certain amount of time
before resending. This amount of time is the time it takes the signal to
travel 205m of UTP (100m there and back plus 5m). This is called the
Collision Domain Parameter. A cable that has low delay
characteristics is ideal for faster protocols. The new standards are
looking at this being at a maximum of 535ns over 100m. TSB 67 does not
define a limit, but ISO/IEC 11801 has an upper limit of 1us.
Also measured in
nanoseconds, parallel transmission protocols (e.g. Gigabit Ethernet,
100BaseVG and 100BaseT4) use more than one pair to transmit signals. Delay
Skew is a measure of the difference in delay between two signals traveling
down separate pairs. Because the number of twists is different for each
pair, the length of each pair is slightly different for each cable. Other
names for this are Asymmetric Skew or Differential Delay. A
good quality cable will have a tightly controlled NVP per pair and the
electronics associated with the particular parallel transmission
technology such as Gigabit Ethernet, will be able to cope with the Delay
skew provided that it falls below the limit. The new standards are looking
at this being 45ns as a maximum allowable value. The current agreement
after TSB 67, is 50ns as an upper limit.
A fiber or a wire will
have imperfections and variations in density that cause some of the energy
to be reflected back. This is called Structural Return Loss. SRL is
measured in decibels for specific frequencies and comes from the
Reflection Coefficient, which is the ratio of the incident to the
reflected signal amplitudes. A high dB value indicates lower reflected
energy. A good indication of a quality cable is when the noise from
reflections is less than a tenth that of noise from Cross talk. A high
value is good!
SRL is important in
bi-directional transmission schemes since the reflected signal must not so
high as to make detection of the receive signal difficult.
This tells you
something of the quality of the whole network link or channel.
Return Loss is the sum of all the reflected signals from patch panel,
connectors and cable arriving back at the signal-originating end. This
will vary across the frequency range 1 to 250MHz and is only significant
for the higher speed data rates. Impedance matching of the components is
critical in order to minimize return loss.
LAN components struggle
with reflected energies greater than 20% (7dB). Current standards call for
3% to 10% reflected energies (15 to 10dB).
(See ISO/IEC 11801
section A.1.1.4 for detail on the measurement of Return Loss and
Cable Capacity (C)(bps)
is given by the formula C = W.log2(1 + SNR) where W is
bandwidth (Hz) and SNR is Signal to Noise Ratio (dB).
If the ACR goes to zero
at frequency f and Af is the attenuation at that
frequency then in the data cabling world where NEXT is important C =
f(0.11(-Af ) - 2.16).
The new Class D channel
is to have a pair capacity of 480Mb/s, whilst the proposed class E is to
In the above diagram, a
ray of light traveling in air, in one plane, hits the glass surface
(incident) at an angle i to the normal. It then changes direction
as it travels through the glass at an angle r to the normal. It is
The laws of refraction
1.The incident and
refracted rays, and the normal at the point of incidence, all lie in the
2.For two given media,
sin i/sin r is a constant, where i is the angle of incidence
and r is the angle of refraction.
The second law is
Snell's Law and defines the refractive index for light traveling
from one medium (1) to another (2), this is denoted by
The absolute refractive
index is defined as the value obtained when light is traveling from a
vacuum to a particular medium. In this case the refractive index is
denoted by n. For glass n = 1.5, for water n = 1.33
and for air n = 1.00028.
Let's look at the
trigonometry a little more closely.
Two parallel rays of
light, IO and JP hit the glass surface in the same plane.
Now IO hits the surface as the ray JP reaches point
X. If we say that the distance XP takes t seconds for
the light to travel, then since c is the velocity of light in air,
XP = ct. Similarly, if v is the velocity of light in glass,
then OY = vt.
Now, sin i/sin r =
sin ION/sin YOR.
Using the other
parallel ray of light, sin i/sin r = sin XOP/sin OPY.
Because the sine of
an angle = length of the opposite side / length of the hypotenuse
sin XOP/sin OPY = XP/OP
/ OY/OP = XP/OY = ct/vt = c/v
This shows that the
absolute refractive index of a medium can be calculated from the
n = velocity of light
in a vacuum / velocity of light in the medium
Now that we have
established that light is refracted when traveling from one type of medium
to another, let us look at the special case of light traveling from a
medium of higher refractive index to a medium of lower refractive index.
There is a critical
angle of incidence beyond which all the light is reflected back into the
first medium, in this case the glass.
We know that n = sin
i/sin r, so for the special case where i is the critical angle
c when r = 90o and n is the refractive
index of the first medium, then:
n = sin c/sin
90o, since sin
90o = 1
From this, sin c =
The critical angle for
glass to air is given by sin c = 1/1.51 = 0.667, therefore c =
an optical fiber, the cladding has a refractive index, which is less than
that of the core material, and this provides the right environment for
light to continuously reflect its way down the core of the fiber with
minimal loss. For multimode fiber the critical angle is 74o
whilst for single-mode fiber it is 83o. The Cone of
Acceptance is the cone produced when spinning the critical angle about
the longitudinal axis of the fiber core. Provided the light rays enter the
fiber at an angle N, which is greater than the critical angle (relative to
the horizontal surface of the fiber end), then the light will successfully
be transmitted. This Cone of Acceptance is often called the Numerical
Aperture and is equal to sin N.