I shall
consider the development of fibre in several sections rather than giving a
general discussion of fibre properties.
Two types
of material are used to manufacture fibres – glass and plastic. There
are several properties of a material that dictate how useful it is as a fibre:
We must
remember that there are many different types of fibre and applications for
fibre. The different properties of fibres can be combined to provide a suitable
fibre for a particular job – not all fibres will be applicable for every
situation.
The purity
of the fibre will be reflected in both its attenuation properties (consider the
scattering effect of an impurity particle) and its refractive index. The
original breakthrough in reducing fibre attenuation was achieved by purifying
the glass used to make the fibres. There are other intrinsic and extrinsic
factors which contribute to the attenuation, such as absorption by OH- ions,
absorption of infra-red radiation leading to molecular vibrations, leakage from
the core (can be caused by Rayleigh Scattering and
fibre curvature) and leaky modes. Curvature is important in fibre specification;
again a more detailed analysis of the propagation of light through fibres is
required to fully explain this, but essentially a small amount of the light is
radiated as the fibre bends. Leaky modes are modes slightly below the
cut-off, but can be propagated for a short distance along the fibre; they can
be initially avoided at the light source by restricting the angle at which
light enters the fibre, but can be introduced along the fibre by microbending. Microbends
are minute bends in the fibre which can be introduced during manufacture or
cabling (see later for more detail on cabling); they can cause power to be
transferred between modes, possibly to leaky modes and hence can result in
power loss.
The level
of attenuation must clearly be minimised to increase the distance the signal
can travel along a fibre without amplification. An example plot of how
attenuation varies with wavelength is given in figure 5. The two types of fibre
noted (Single-Mode and Graded-Index) will be explained further through this
essay. The units dB are logarithmic. It indicates how
we choose the most appropriate wavelengths of light to use with a particular
fibre.
The two
attenuation minima (shown by the dashed lines) are approximately at
lambda=1300nm and lambda=1550nm, and hence these two wavelengths are the most
commonly used amongst modern fibre systems.
The
refractive index of the material is crucial to total internal reflection. A single-mode
fibre has essentially a discontinuity in refractive index at the core-cladding
boundary (although a rigorous analysis of such a fibre should consider that
during the manufacturing process there is some ‘blurring’ of the index over a
very short distance (~10^-9 m)). As mentioned earlier, various modes can pass
through an optical fibre; the number of modes that can pass through a fibre can
be shown to be dependent on the diameter of the fibre. The diameter of the
fibre can be reduced such that only one mode can exist within the fibre – this
removes the problem of modal dispersion and hence allows significantly higher
bandwidths than would otherwise be available.
The
alternative type of fibre is multi-mode. This, as it sounds, allows the
existence of various modes. This initially led to modal dispersion problems,
and so was used for short distances (~500m) where dispersion was not relevant.
An alternative solution to the problem was later found – graded-index fibre,
where there is a gradual gradient of refractive index through the core which
peaks in the centre (in reality this is a series of gradually changing
refractive indices – typically with around 200 different values). This causes
the paths of the various modes to be sinusoidal and can be used to ensure that
the different modes travel along the fibre with a similar effective velocity.
This considerably reduces the modal dispersion to values of the order of 1ns
per km (the pulse width increases by 1ns every km) (TestMark
Laboratories, 1999).
There are
other types of dispersion:
There has
also been development of polarisation-sensitive fibres, of which there are two
principle types. A polarisation-sensitive fibre will transmit light of one
linear polarisation but not the other. A polarisation maintaining fibre is birefringent; it effectively isolates the two components of
the light and transmits them individually (due to different properties of the
fibre for different linear p olarisations); thus the
emerging light has the same polarisation as the incident light.
With the
constantly increasing demand for data bandwidth, fibre designers are
continually looking for ways to increase the bandwidth of a single fibre. All
high-bandwidth long-distance fibres are single-mode, and so there is no modal
dispersion to reduce bandwidth. Further developments are always sought though –
the most recent of these is DWDM (dense wave division multiplexing). This
allows multiple light sources (each operating at a different wavelength) to
transmit data down a single fibre. Currently 4, 8, 16 or 32 wavelengths can be
multiplexed on a fibre , with 64- and 128-channel
systems in development. It is also reported that multiplexing of over 200 wavelengths
on one fibre has been demonstrated by manufacturers.
The
various factors mentioned above (attenuation and various forms of dispersion)
require the use of repeaters. This term, although widely used, is unclear. It
is used to describe both optical repeaters and regenerators. Optical
repeaters are purely optical devices that are used simply to combat
attenuation in the fibre; typically spans of 80km upwards are now possible. The
recent introduction of soliton transmission methods
has increased the allowed distance between repeaters and systems spanning 130km
without a repeater are now possible. Regenerators are devices consisting of
both electronic and optical components to provide ‘3R’ regeneration – Retiming,
Reshaping, Regeneration. Retiming and reshaping detect
the digital signal which will be distorted and noisy (partly due to the optical
repeaters), and recreate it as a clean signal (see figure 6). This clean signal
is then regenerated (optically amplified) to be sent on. It should be noted
that repeaters are purely optical devices whereas regenerators require
optical-to-electrical (O/E) conversion and electrical-to-optical (E/O)
conversion. The ultimate aim of many fibre system researchers is to create a
purely optical network without electronics, which would maximise efficiency and
performance. Many aspects of such a system are in place, but some still require
the O/E and E/O conversion.
The most
common optical amplifier currently in use is the EDFA (Erbium Doped Fibre
Amplifier). These consists of a coil of fibre doped
with the rare earth metal erbium. A laser diode (see later) pumps the erbium
atoms to a high energy state; when the signal reaches the doped fibre the
energy of the erbium atoms is transferred to the signal, thus amplifying it.
Cabling is the procedure
of grouping fibres together with strength members (see below) and possibly
other materials to produce a product that is ready for field use. The type of
cable is dependent on the environment in which it will be used. It would not be
possible within this short essay to comprehensively describe the types of cable
applicable to every situation, but I shall endeavour to cover the majority of
general environment categories.
First, we
must analyse the purpose of the various elements in a cable. Of course there is
fibre to carry the data; there may be many fibres within an individual cable
and a considerable amount of research is devoted to the finding the strongest
and most efficient systems of organisation of various numbers of fibre within a
cable. Strength members are used to add tensile strength to fibres;
although fibre has a high tensile strength of its own, this has to be increased
for many applications (e.g. submarine cables). One of the most common strength
members is Kevlar® which is a specialised material (manufactured by DuPont) consisting of long molecular chains produced from paraphenylene terephthalamide. It
has a tensile strength significantly greater than high-tensile steel and has
the advantage of being lighter than steel. Armour is used to protect the
cable from pressure and rodents (most fibre literature, being US-based,
mentions gophers as a primary cause of fibre failure!).
The main
properties that are required of fibre cables in various environments are:
The
following table describes the properties of cables for use in different
environments:
Environment |
Cable Properties |
Intra-Device (e.g. inside a
computer or switching system) |
Should be small with little
protection as this is given by the unit in which it is housed. |
Intra-Office (e.g. server
backbones of computer networks, and recently directly to workstations) |
Usually contain one or two
fibres. They should be small if being routed through an office, but should be
durable enough to withstand being moved by office staff. Under-carpet cables
are available which have an extremely low profile and high crush-resistance.
These cables must also be compatible with the kind of connectors used for
network connections (which I shall not discuss here – it would provide enough
material for another essay!) |
Intra-Building (i.e. between
offices) |
Similar to intra-office
cables, but possibly need to be more robust (added strength members and a
tougher outer coating) to survive installation. A larger number of fibres may
be required within a single cable for vertical network backbones (in a
topological rather than physical sense). Cables for use in factories will
often be susceptible to added physical stress and so must include added
rigidity and possibly a form of light armour. |
Breakout Cables (i.e. cables
that are split up before termination) |
Again similar to above, but
must be suitable for splitting into individual fibres for data distribution.
The sub-cables must therefore be suitably protected as well as having an
outer shielding. |
Plenum Cables (which are
routed through ducts already present in a building such as heating ducts) |
These must be fire-retardant;
originally fluoropolymers were used, although
alternatives are now being developed. The materials used can be expensive and
so cable size must be kept to a minimum. |
Temporary Cables –
Light-Duty (e.g. for a video link) |
These should have similar
properties to intra-office cables, but with added rigidity to survive
constantly being moved. |
Temporary Cables –
Heavy-Duty (e.g. for military field use) |
These should have extreme
crush-resistance, very high tensile strength, be waterproof and be resistant
to impulse damage and damage from bending. |
Aerial Cables |
These are for outdoor use,
to be strung between supports. If directly strung then the tension on the
cable is extremely high; it is more usual to lash the fibre cable to a
supporting cable, so that considerably less stress is applied to the fibre
cable. The cables should have strength members and structures to isolate the
fibre itself from any of these stresses. The cable must also be waterproof
and resistant to temperature changes. |
Ducted Cables & Direct
Burial Cables (buried in the ground directly with no duct or other
protection) |
Similar to aerial cables but
with added armour to protect the fibre from rodents. Polymer-coated
galvanised steel is common as armour since it is strong and resistant to
corrosion by oxidation. |
Submarine
Cables |
Great crush resistance is
required to withstand the pressure of several kilometres of water. High
tensile strength is required because of the strain put on the cable whilst
laying it. Even higher strains are put on cables if they are retrieved and
relayed for maintenance work. If the cable is to be placed in shallow water
then extra armour is essential to protect it from activities such as fishing.
|
Power Line Cables |
Some fibres are run along
with overhead power lines – usually the single earth cable. Two systems are
used; (i) the fibre being wrapped around the
conducting cable and (ii) OPGW (Optical Ground Wire) where the conductor
contains the fibres in tubes at the centre of the conductor construction. |
We should
also consider fibre housing – the structure that surrounds the fibre within the
cable. There are two main designs – loose-tube and tightly buffered. Loose-tube
cable is where one or more fibres run in a loose helix through a tube allowing
the fibres to move freely within the tube and protecting them from stresses
that could cause microbending. There are different
types and shapes of loose-tube cable, all of which isolate the fibre from
external forces. For outdoor cables a gel is added which lubricates the fibre
to aid it to move within the tube and to protect it from moisture. Tightly
buffered fibre is directly encased within one or more layers of casing; the
innermost layer is usually a soft plastic that will reduce the forces
transferred to the fibre through the casing, whereas the outer layers are made
from harder plastics to give added physical protection to the fibre.
Loose-tube
cables provide better physical protection and so are useful for preventing microbending loss. Tightly buffered fibres provide better
crush-resistance and allow for more densely packed cables. It also makes
automated joining easier for multi-fibre cables since the fibre positions are
predictable with respect to the cable.
This has
always been seen as a ‘black art’; in the past there have only been specialised
technicians who have been able to reliably join fibres. Nowadays, with fibres
manufactured to higher quality, there is more tolerance in the joining of
fibres although it is still a precision technique. When joining fibres,
alignment is critical.
The two
methods of joining fibres are splicing and using connectors.
Splicing
involves forming a permanent join between two lengths of fibre. The fibres are
only of the order of micrometers in diameter (although multimode fibres do have
a slightly larger diameter), and so must be joined with a lateral alignment
precision of ~10^-7m to avoid attenuation. This requires the use of specialist
equipment including some kind of magnification.
There are
two types of splice – mechanical and fusion. A mechanical splice is
performed using an epoxy substance to match the refractive indices of the
materials and the cables are held together using a clamp. The epoxy material
reduces Fresnel reflection (which is caused when
light leaves a material of one refractive index and enters a material of
another). A fusion splice fuses the two cables together using a short
intense blast of heat. These splices are reportedly at least as strong as the
fibre itself, if not stronger.
These are
permanently connected to the end of individual cables so that they can be
joined simply at a later stage (i.e. during installation) and can even be
moved. Again, lateral alignment is a crucial issue in the design of connectors
– when two connectors mate, the cables must be aligned precisely. This is
easier with multimode fibre (the type of fibre generally used for topologically
horizontal network connections and so the type that is most likely to need to
be moved / reconnected during its lifetime) because of its greater diameter.
Originally
the individual fibre ends were prepared such that they were completely flat,
whereas the latest technique is to curve them. This reduces the attenuation to
a minimum and also decreases the amount of back-reflection (reflection of the
signal back along the fibre).
Other
factors are important in reducing attenuation. The numerical aperture of the
two fibres to be joined must be as similar as possible. Consider two fibres
where the receiving fibre has a smaller numerical aperture than the larger
fibre – some of the modes in the delivering fibre will not be able to enter or
will not be internally reflected within the receiving fibre (obviously this is
not applicable to single-mode fibres). If there is a gap between the two fibre
ends then some light will be incident on the cladding of the receiving fibre
rather than on the core (since as light leaves a fibre it spreads out). This
loss increases as the distance between the fibres increases and as the
numerical aperture of the fibres increases. Gaps between the fibre ends also
increase the amount of Fresnel reflection (see
above). Any impurities in the connector (such as dust) or an unpolished fibre
end can cause the attenuation of a connector to increase appreciably.
Two types
of light source are used with fibres, LEDs and Laser
Diodes. LEDs can operate in the near-infrared (the
main wavelengths used in fibres are 1300nm and 1550nm, along with 850nm for
some applications); they can emit light at 850nm and 1300nm. They also have the
advantages of long lifetimes and being cheap. Unfortunately they are large
compared to the cross-section of a fibre and so a large amount of light is lost
in the coupling of an LED with a fibre. This also reduces the amount of modal
control designers have over incident light. Laser diodes can be made to emit
light at either 1300nm or 1550 nm, and also over a small spectral width (unlike
LEDs) which reduces chromatic dispersion. Their
emitting areas are extremely small (~0.4 micrometers x 1 mmicrometers)
and so the angle of incidence of light on a fibre can be accurately controlled
such that <5% of the possible modes within a multimode fibre will be
initially used (although microbending can shift
energy between modes and so the number of modes used may increase along a
length of fibre). They are more efficient than LEDs
in terms of coupling of light into the fibre, although they have shorter
lifetimes than and are more expensive than LEDs. One
crucial advantage of lasers over LEDs in today’s
world of digital communications is their high switching speed and small rise
times (rise time is the time taken for power to rise from 10% to 90% of the
maximum power of the pulse, fall time is the opposite), leading to increased
bandwidth.
The most
efficient detectors are reverse-bias photodetectors.
A full analysis of these requires an in-depth understanding of semiconductor
physics and so is not appropriate here. They essentially cause a current to
flow when light is incident on them. The choice of semiconductor that is used
to fabricate the detector is dependent on the wavelength sensitivity and the responsivity (effectively proportional to quantum
efficiency) that are required. Bandwidth considerations are also important
(determined by the rise time and fall time of a detector); in detectors the
fall time is often appreciably greater than the rise time and so this must be
used to calculate the bandwidth of a detector.
There are
many further complications in detectors, including noise equivalent power which
indicates how ‘clean’ a signal from a detector is. An
analysis of how analogue and digital signals are processed after the initial
detector is also interesting, but not within the scope of this essay.
The best
way to describe the current technology being deployed is to provide a brief
case study of a company that extensively uses optical fibres. I have chosen Energis PLC, a British Company that is currently expanding
its network throughout
Their
network exclusively uses single-mode optical fibres. Partly due to their
commercial history a large number of their fibre cables are installed on lines
belonging to the National Grid Company and to some regional electricity companies.
They also use cables buried in ducts under roads and pavements.
They use
DWDM (some uni-directional and some bi-directional)
with 10Gbit/sec on each of 4 wavelengths. They are currently introducing 32
wavelength systems, and will inevitably introduce 40 Gbit/sec-per-wavelength
systems in the near future.
Their
optical repeaters (for which they use EDFAs) are
typically spaced at around 80km and their regenerators can be up to 600km
apart, although this is partially dependent on previous repeater locations. New
technologies that are emerging will theoretically increase the regenerator
spacing to 5000km (and even further with the introduction of soliton transmission), although DWDM does reduce the
allowed distance between amplifiers and regenerators.