sábado, 17 de julio de 2010

LANs and topology

Ever since the invention of the telephone enabled us to speak to someone in the next town and the television spawned a remote control, we have been unhappy about the thought of getting out of our chair to communicate with anyone or anything.
It is a natural progression to apply the same thoughts to our computers and other forms of data communications.
As computers have became more compatible, it seemed a good idea to exchange information by a cable connecting two or more systems. This made it possible to have shared facilities, whether it be massive storage areas, printers or software. It also allows staff to work on the same project and share the corporate facilities.
Sharing data is called networking and is categorized by the physical area that is interconnected. The smallest subdivision is called a 'Local Area Network' or LAN.

A LAN can be as small as you like, as in Figure 19.1 or, more realistically, it could interconnect a whole building or a collection of buildings or a large manufacturing site or a university with several thousand connections.

Characteristics of a LAN

  • A LAN uses a privately owned communication system rather than the normal telephone system.
  • They can operate at high data rates.
  • All computers can initiate the transfer of data to any other one.
  • They generally save costs by sharing costly equipment, software or data libraries.
  • They improve productivity and allow the exchange of data to be monitored for security purposes.
A whole city can be interconnected or possibly several LANs can be joined to provide a larger system and, in this case, the network is referred to as a MAN, 'Metropolitan Area Network'. Any larger network is called a 'Wide Area Network' or WAN.

What bits of hardware are we likely to meet in a LAN?

Communication route

Cable, almost certainly. The cables used are chosen to meet the requirements of the LAN, principally a matter of distance and the data rate. The cables used are optic fiber, copper twisted pair and copper coaxial cable. There are a few LANs that use a wireless communication system but this is unusual.


A server supplies a facility to the network. The name comes from the idea of the device 'serving' the needs of the network.
A typical example may be a 'print server'. Instead of supplying maybe a hundred mediocre quality printers to all the staff, we may decide to have a single super high quality printer than can handle all the printing requirements with better quality, higher speed and less expense.
All LANs have a 'file server'. This device controls access to all the shared files to store the files when not in use and to prevent two people modifying a file simultaneously. Having the files held centrally means that we can wander off to another part of the building and can use any computer to access the previously used files. There are two more advantages. If the software needs upgrading, we have only a single copy to be loaded. Backing up the work is easy since all the
shared files are held on the file server. Once all the staff have gone home at night, the server can carry out a backup of the day's work before going to sleep.
If we want to access an external network, we can make contact via the normal telephone communication system by using a modem like the one we use with the Internet. As with the print server, it is more economic for a 'communication server' to control a series of modems than to provide each staff member with their own modem.

Connecting our computer to a network

To connect our computer to a network we need to install network interface cards (NICs). The power is switched off, the case of the PC is taken off, and the card inserted into one of the expansion slots. The software is run to make sure the computer will be recognized by the network and the case is put back on again. Done. We are now part of the club.

Topology at last

We have choices when it comes to how we connect several computers and other devices. These connection patterns are what we call topology, an impressive name for something quite simple.
There are only three basic designs from which all other layouts are derived. Just before we start, we should mention the term 'node'. This is the name given to any piece of equipment that can be connected to a network such as computers, terminals or printers.

Bus topology

Here is the first one and probably the most obvious. We have a length of optic fiber or copper cable and connect all the nodes, one after another. This layout is shown in Figure 19.2 but in reality, the bus does not need to be straight, it simply wanders around connecting to each of the nodes.

Bus topology is very simple to construct and can be expanded simply by joining new devices to the bus cable. If the bus cable is damaged, the whole network may fail and your whole workplace goes for a coffee break until it is fixed and this may take some time. We can get over this reliance on a single cable by using star topology but this brings its own problem, as we will see.

Star topology

In this design, we have a central connection called a hub. This is a central computer or server that is connected to each node as in Figure 19.3.

The central hub asks each node in turn whether it has a message to send. If a node, say node 2, wishes to send data to node 5, it says 'yes' and sends it to the hub which then reroutes it to node 5. The hub goes on to check with node 3, then 4 and so on. Providing the hub is fast enough it can handle all routing requirements. It can do more than this; it can provide the management with details of who is sending what data to whom. This can monitor data holdups so that the system can be upgraded if necessary, but it can also check to see who is playing games during work time.
As each node can be easily disconnected without interfering with the whole system, faults can be isolated more easily. The snag is that it the hub fails, the whole system goes down.

Ring topology

This is a modified version of the bus topology. The two ends of the bus are simply joined together to form a ring as shown in Figure 19.4.

If node 2 wishes to send a message to node 5, a system of tokens is used. A token is a short electronic code that is passed from node to node round the ring. Node 2 waits until it receives the token then attaches its message and an address. The token, message and address are passed from to mode 3, then 4, then 5. Node 5 recognizes the address and removes the attached message and the token continues around the ring until someone else wants to send some data. There are a couple of problems with this system in that a failure of any node will stop the rotation of the token and the whole network will fail immediately. Expanding the network by adding another node is only possible if the ring is broken and the network is shut down.

Hybrid topologies

We can use the three basic topologies in combination with each other to create a large number of possibilities as required in the network being installed.

Three popular hybrids are illustrated in Figure 19.5.
The first is called the clustered star topology. To do this, we start with a bus and instead of separate nodes, we can connect star clusters.
Tree topology also starts with the bus system but reduces the length of cable used by combining some of the routes.
Star-wired ring is, as the name suggests, a combination of the ring and the star.
The nodes pass through the hub before going onto the ring to pass the data its destination. The hub can also disconnect a faulty node and allow the system to remain operative. As can be imagined, there are many amazing possibilities but if we are aware of the basic patterns we will be able to find our way around the others.

Kevin M Contreras H
CI 18.255.631



Organizing optic fiber within a building

Out in the real world we may be asked to modify or repair an existing fiber layout. The first step is to see how the original installation was designed. This involves hope. We hope that the installers prepared records of what was actually installed in the building rather than what was intended when they placed their bid for the contract. There is always the possibility that modifications dreamt up on the back of an envelope during the lunch break were made but never recorded. We also have to hope that the installer followed good practice and adhered to the appropriate standards — especially if they have subsequently gone bust and disappeared. It is always a good start if we know roughly what to expect in a building, but before we look at the fiber layout, we will take a brief look at copper cables.

Copper cables

But this is a book on optic fibers. Yes, but we keep coming across copper cables so it may be worth a brief outline of their characteristics so at least we will know what we are dealing with. We will start with a look at the advantages and disadvantages.
Advantages of copper
  • Many technicians are happier when using copper cables because they are familiar and unless they have taken the trouble to get to grips with optic fiber, they seem so much simpler.
  • Connections are easier with hand tools and do not require expensive equipment and great precision.
  • A simple metal detector can find a buried cable. This is only half an advantage because many optic fiber cables use metallic foil for moisture protection and metallic armoring.
Disadvantages of copper

These have been explained in Chapter 8 as advantages of optic fibers but to save the trouble of finding them again they are listed below.

  • Electrical interference and crosstalk.
  • Care has to be taken in high voltage environments.
  • Reduced bandwidths.
  • Security.
  • Higher losses.
  • Size and weight
  • Two wires are used to send a single signal.
How does it work?

To send a telecommunication signal by copper cable requires an electric circuit. This means that we need two wires usually referred to as a single pair. See Figure 18.1.

The transmitter generates a voltage signal between the two wires at the input and this results in a current flowing along one wire, through the receiving circuitry and back along the other wire. Compared with fiber, this doubles the number of connections necessary.
One problem with any electrical signal is that whenever electric current flows, it causes a changing magnetic field. This is, in itself, not a worry unless there is another copper wire nearby. The changing magnetic field caused by the signal will cause a voltage to be induced into the copper wire causing a weak copy of the original signal. This effect is called 'crosstalk'.
The magnitude of the induced voltage decreases by the square of the distance away from the source of the interference so if we double the distance, we reduce the induced voltage by a factor of four. If two copper wires are running side by side, the one closest to the source of interference will pick up a larger signal than the one further away.

We can also reduce, but not entirely eliminate, this effect by twisting the copper wires.
In circuit A, the upper wire is closer to the source and hence has a higher voltage induced. This larger voltage is indicated by the larger arrows and, at the moment shown, is trying to push the current clockwise round the circuit while the other wire is trying to push the current counterclockwise. The big arrows will win and the overall effect will be to produce a small current that will be a copy of the interference signal.
In a communication system the interference signal may be a conversation on a nearby telephone or a crackle from electrical machinery or lightning or any other source of interference.
In circuit B, each wire changes position so that the total value of the induced voltage is equal in each wire. This would give the happy result of no overall effect and hence no interference.
To work efficiently, the two wires must be twisted very carefully to keep the two wires balanced. When we have several pairs close together in a single cable, the rate of twist is varied and this helps to reduce the likelihood of two wires accidentally running parallel with their twists in step.

Cable designs

The cables look very similar to an optic fiber cable. They have an outer cover of polyethylene to provide waterproof protection just like optic fiber. In fact, apart from markings on the outer cover, there is no way of telling them apart so we must be careful not to cut into an optic fiber cable only to find copper cores. This would not be popular with the owner of the copper.

Twisted pair cables

Just as in fiber optics, a range of copper cables is available for indoor use from just two pairs of conductors up to enormous armored underground cables containing up to 4200 pairs of conductors. Figure 18.3 shows a typical four-pair screened copper cable.

There are three similar versions of twisted pair cables all looking very similar to Figure 18.3. The differences are in the screening. The simplest is no screening. Then comes the grounded screen that helps to reduce high frequency interference signals from reaching (or leaving) the cable. The other alternative is to wrap each pair separately in aluminum foil and then added braid under the jacket to screen the whole cable. This also provides protection against low frequency interference.
Coaxial cable

These give improved protection against interference and are familiar cables at home where they provide the signal inputs to televisions and videos etc. Some minor variations occur. The outer braid sometimes has two layers, and some include aluminum foil wrapping around the center conductor, but they are all easily recognizable. See Figure 18.4.

Fiber in buildings

In this section we will have a look at a typical layout in a building or group of buildings. In reality, the installation may differ owing to local customs or special circumstances but the trend is towards using a standard layout of this form. Such harmonization will make life easier for everyone in the industry whether they are installing, extending or repairing a system.

Making an entrance

Getting the fiber into a building is usually a matter of balancing the immediate and long-term costs. We can have a low initial cost but accept higher costs when repairs or modifications are required or we can pay higher immediate costs but less in the future. Let's start with the cheap initial cost since this may win us the contract if the customer does not take the long-term view.

The dig a hole and throw it in method

This is easy. Get a trench digger or plow and prepare a hole, then pass a length of conduit through the wall of the building. Next we lay the copper and fiber cables in the trench, pass them through the conduit into the building and make the connections to the service provider in their manhole. Finally refill the trench. Job done. Out of sight, nice and cheap. There are some problems. What would the digger person do if they came across an obstruction? They would simply go around it, however long the detour, so long as they can finish at the conduit through the wall. The cables would be added and the trench refilled. Is the actual route of the cables going to be accurately recorded in the plans? What do you think? The cables are not protected and in years to come when the cable needs attention — are we going to be able to find it? Is it now under the new staff car park? Or the new building? Will it be cut by the utility company laying pipes?

The spend more and do it better method

If the cables are installed in conduit or other proper mechanical enclosure it takes a little extra time and money but the cable is better protected and is in a known place so extra cables can easily be installed. The access point of the conduit through the wall needs to be sealed to prevent water or gas gaining access to the building.

Making a start

In this example we have used an underground route to bring the cable from the service provider's manhole into our building. It is fed into an equipment room from which our cables can wander off around the building. This situation is shown in Figure 18.5.
The equipment room will be something between a proper room and a cupboard where the brushes are kept. Somewhere between the manhole in the road and the equipment room outlets,
the responsibility for the cabling is transferred to us from the service provider.

We need to be aware of where this transition occurs for all sorts legal reasons otherwise it may result in our lawyers getting richer and us getting poorer — not the idea at all.

Inside the building

So far we have got from the road to just inside our building. The cables, copper and fiber now go to another equipment room called the 'campus distributor' or 'main cross-connect'. This is the main interconnection point for the remainder of the system. The system may be a single room, a multistory building or several buildings like a business park or university campus. The bundles of cables going between equipment rooms are called 'backbone cables' as in Figure 18.6. The cabling may be hidden away in ceiling spaces or under the floors or in channels or conduit — wherever convenient.

Inside the room

To connect the 'campus distributor' or 'main cross-connect' to the individual work areas inside the room is the function of a junction box called a 'floor distributor' or 'horizontal cross-connect'. Despite the name, cables can be installed vertically as well as horizontally. By having an organized hierarchy of connections like this, faults can be traced back and isolated more systematically. The final connections within a room are shown in Figure 18.7.

Other buildings and other floors

To connect a communication system to another building or another floor in the same building, the cables are taken from the campus connector (main

cross-connector) to the building distributor (intermediate cross-connect) and from there to another floor distributor and then on to the work areas. See Figure 18.8.

Maximum length of optic fiber cable runs

These are governed by local regulations in each country but typical values are shown in the connection summary in Figure 18.9.

Kevin M Contreras H
CI 18.255.631


Light sources and detectors

Most light sources and detectors are electronic devices built from the same semiconductor materials as are used in transistors and integrated circuits. The design of these devices is a separate study and will not be considered here. Instead, our view will be restricted to the characteristics which are of interest to the user.


The most common form of laser diode is called an injection laser diode (ILD) or just injection diode (ID). The word injection is not of interest — it merely refers to part of the process occurring inside the semiconductor material. A laser provides a light of fixed wavelength which can be in the visible region around 635 nm or in any one of the three infrared windows. The light has a very narrow bandwidth, typically only a few nanometers wide. This ensures that chromatic dispersion is kept to a low value and this, together with fast switching, allows high data transmission rates.
As the laser device itself is barely visible to the unaided eye, it must be contained in some form of package. Two typical examples are shown in Figure 14.1.

Lasers for visible light

The light is launched via a lens system to allow it to be concentrated into a beam. Visible laser light finds applications in bar code readers, CD players, medical and communication systems. They are usually fitted with a built-in light detector so that they can receive reflected information as in the case of the bar code reader.

Lasers for 850 nm use

These can be packaged in either of the ways illustrated in Figure 14.1 depending on their application. The fact that their output is not visible allows for use in security, ranging, automotive and industrial and military applications. They also provide the light source for short and medium range fiber communications.

Lasers for singlemode communications
Successful launching into singlemode fibers requires very high precision and this is achieved by optimizing the position of an attached pigtail which can then be connected to the main fiber run by any desired method.
A photoelectric cell is also included as a monitoring device to measure the output power. This provides feedback to allow for automatic control of the laser output power.
The output power of a laser is affected by any change in its temperature, generally decreasing in power as the temperature increases. Some laser modules include a temperature sensor to combat this problem. It provides internal temperature information which is used to control a thermo-electric cooler like a small refrigerator, to maintain the temperature. The temperature stability is also improved by bolting the laser package to some form of heat sink such as the instrument casing.

Laser safety

Both visible and infrared light can cause immediate and permanent damage to the eyes. The shorter wavelengths cause damage to the retina and the longer wavelengths attack the cornea, in neither case can medical science offer remedy once the damage is done. Permanent loss of eyesight in less than a second by exposure to light we can't even see — it doesn't seem fair somehow.
It is extremely important that we take sensible precautions.
Never look into:
  • A live laser source
  • An unknown light source
  • Any fiber until you have ascertained that it is safe. Check it yourself even if trusted colleagues say 'its OK we've just checked it out'. They may be talking about a different fiber or they may have made a mistake.
If an instrument such as a live fiber detector is used, make sure it is working.

Beware of concentrating the light by instruments such as will happen when checking a cleave or the end condition of a connector with a microscope.
Laser classifications are based on an international standard titled Radiation Safety of Laser Products, Equipment Classification, Requirements and User's Guide, referred to as IEC standard 825. Additional national standards apply in each country.

The IEC 825 classification has used four classes of laser based on the accessible emission limit (AEL). Every laser must carry a warning label stating the class of laser as shown in Figure 14.2. It is the responsibility of the manufacturer to

determine the classification of the laser and they do so by measuring the wavelength, output power and the pulsing characteristics.

IEC classifications

Class 1: Safe under reasonably foreseeable conditions of operations. Note that it doesn't say 'safe under any conditions'.

Class 2: Visible lasers with light output within the visible spectrum of 400–700 nm. There is an assumption here that the blink reflex will close the eyes within a fraction of a second and hence provide protection. Prolonged exposure will cause damage.

Class 3a: Safe for viewing by the unaided eye either visible or infrared light but possibly unsafe when viewed with instruments.

Class 3b: Direct viewing is hazardous but reflected light is normally OK. Note the normally. Not to be viewed with instruments.

Class 4: Horribly dangerous. Even reflections are hazardous and the direct beam can cause fires and skin injury. Not normally used for communications.

Control measures

For classes 2, 3 and 4, control measures are employed such as interlocks, keys, laser 'on' warning lights, remote switching, prevention of reflections across walkways. The precautions depend on the situation, use and power of the laser. The appropriate national standards as well as IEC 825 should be consulted for guidance.

Laser specifications


The wavelength quoted is only a typical value. So if we want to buy a laser for the 1300 nm window, the one offered may well be quoted as 1285–1320 nm and the actual frequency will fall somewhere between these limits. Sometimes it would just be sold as 1300 nm (nominal).

Rise and fall time — Figure 14.3
This is a measure of how quickly the laser can be switched on or off measured between the output levels of 10% to 90% of the maximum. A typical value is 0.3 ns.

Threshold current — Figure 14.4

This is the lowest current at which the laser operates. A typical value is 50 mA and the normal operating current would be around 70 mA.

Spectral width — Figure 14.5

This is the bandwidth of the emitted light. Typical spectral widths lie between 1 nm and 5 nm. A laser with an output of 1310 nm with a spectral width of 4 nm, would emit infrared light between 1308 nm and 1312 nm.

Operating temperature

No surprises here. Typical values are –10°C to +65°C and therefore match the temperature ranges of fibers quite well.

Voltages and currents

The specifications also list the operating voltages and currents of the monitor detector, the cooler current and the thermistor resistance. These are generally only of interest to the equipment designer or the repair technician.

Output power
The output power may be quoted in watts or in dBm.

LEDs — light emitting diodes

LEDs can provide light output in the visible spectrum as well as in the 850 nm, 1350 nm and the 1500 nm windows. Compared with the laser, the LED has a lower output power, slower switching speed and greater spectral width, hence more dispersion. These deficiencies make it inferior for use with high speed data links and telecommunications. However it is widely used for short and medium range systems using both glass and plastic fibers because it is simple, cheap, reliable and is less temperature dependent. It is also unaffected by incoming light energy from Fresnel reflections etc. Although the lower power makes it safer to use, it can still be dangerous when the light is concentrated through a viewing instrument. Typical packages are shown in Figure 14.6.

PIN diodes

A PIN diode is the most popular method of converting the received light into an electronic signal. Their appearance is almost identical to LEDs and lasers. Indeed the diagrams in Figure 14.6 would serve equally well for PIN diodes if the labels were changed. They can be terminated with SMA, ST, SC, biconic and a variety of other connectors or a pigtail.
It may be of interest to have a brief look at its name. It uses a semiconductor material, either germanium or silicon. The pure semiconductor material is called an intrinsic semiconductor — this is the I in the name. To make it work, we have to add a controlled amount of impurity into the semiconductor to change its characteristics. The semiconductor is converted into two types, one called P-type semiconductor and the other called N-type. These are arranged either side of the I material to make an I sandwich. Hence P-I-N or PIN diode. The theory of its operation will not be considered further.
While we can still buy straightforward PIN diodes, it is more usual for it to have an amplifier built into the module to provide a higher output signal level.
Avalanche diode also called an avalanche photo diode or APD
Higher output signals can be achieved by an avalanche diode. It uses a small internal current to generate a larger one in the same way that a snow-ball rolling down a mountainside can dislodge some more snow which, in turn, dislodges even more snow and eventually gives rise to an avalanche.
They have the advantages of a good output at low light levels and a wide dynamic range — it can handle high and low light levels. However there are a number of disadvantages which tend to outweigh the benefits. It has higher noise levels, costs more, generally requires higher operating voltages and its gain decreases with an increase in temperature.

Light receiver specifictions


This is quoted as a range e.g. 1000 nm to 1600 nm, or by stating the frequency that provides the highest output e.g. peak wavelength = 850 nm.

Dynamic range or optical input power

Dynamic range is the ratio of the maximum input power to the lowest. It is quoted in decibels e.g. 21 dB. The optical input power is the same information expressed in watts. e.g. 1 μW to 125 μW.


A measure of how much output current is obtained for each watt of input light. e.g. 0.8AW–1. This means that the current will increase by 0.8 amps for every watt of increased light power.

Response time

This is the rise and fall time that we saw in Figure 14.3. It determines the fastest switching speed of the detector and hence limits the maximum transmission rate e.g. tr or tf = 3.5 ns.

Bit rate or data rate or bandwidth

These are both measures of the maximum speed of response to incoming signals and is therefore determined by the response time above.

Kevin M Contreras H

CI 18.255.631




Imagine an optic fiber carrying an input signal that needs to be connected to two different destinations. The signal needs to be split into two. This is easily achieved by a coupler. When used for this purpose, it is often referred to as a splitter.
Couplers are bi-directional, they can carry light in either direction. Therefore the coupler described above could equally well be used to combine the signals from two transmitters onto a single optic fiber. In this case, it is called a combiner. It is exactly the same device, it is just used differently.
The various ways of using couplers are shown in Figure 13.1.
Physically, they look almost the same as a mechanical splice, in fact in some cases we would need to count the number of fibers to differentiate between them. If there is one fiber at each end, it is a mechanical splice, any other number and it is a coupler.

Coupler sizes — Figure 13.2

A coupler with a single fiber at one end and two at the other end would be referred to as a 1 X 2 coupler (read as one by two). Although 1 X 2, and 2 X 2, are the most common sizes they can be obtained in a wide range of types up to 32 X 32 and can be interconnected to obtain non-standard sizes. Splitters are more common than combiners and this has made it more natural to refer to the single fiber end as the input.
The numbering of the ports is shown in Figure 13.3 (port is just a fancy word used in electronics to mean a connection).

Splitting ratio or coupling ratio

The proportion of the input power at each output is called the splitting ratio or coupling ratio. In a 1 X 2 coupler, the input signal can be split between the two

outputs in any desired ratio. In practice however, the common ones are 90:10 and 50:50. These are also written as 9:1 and 1:1. In the cases where the splitting ratio is not 1:1, the port which carries the higher power is sometimes called the throughput port and the other is called the tap

Coupling tolerance

Even when the splitting ratio is quoted as 1:1, it is very unlikely, due to manufacturing tolerances that the input power is actually shared equally between the two outputs. The acceptable error of between 1% and 5% is called the coupling or splitting tolerance.


A gloomy note before we start.
When consulting trade publications, we find that the terms used to describe coupler losses, the naming of the ports and even the numbering of the connections have not been totally standardized. This makes it difficult to avoid meeting several different versions of the formula for each loss.
The only way to combat this is to understand the nature of the losses and then to be fairly flexible when it comes to the way it has been expressed.

Referring to Figure 13.4, the losses are stated in decibels and assume that the input is applied to port 1 and the output is taken from ports 2 and 3. For the moment, we will ignore the other connection shown as port 4 with its outward pointing arrow. This will be discussed further when we look at directionality loss.
We may recall that, generally, the loss in decibels is derived from the standard formula:

Loss = 10log(powerout/powerin)dB

Excess loss

Excess loss is a real loss. If 10 mW goes into a device and only 9 mW comes out, then it is reasonable enough to think of the other 1 mW to be a loss. The light energy has been scattered or absorbed within the coupler and is not available at the output. So what we are really saying is that the loss is dependent on the total output power compared to the input power. In the case of the coupler in Figure 13.4, the output power is the sum of ports 2 and 3 and the input is at port 1.
So excess loss would look like this:

Excess loss = 10log((P2+P3)/P1) dB

where P1, P2, P3 are the power levels at the respective ports.

Directionality loss or crosstalk or directivity

When we apply power to port 1 we expect it to come out of ports 2 and 3 but not out of port 4, the other input port. Unfortunately, owing to backscatter within the coupler, some of the energy is reflected back and appears at port 4. This backscatter is very slight and is called directionality loss or crosstalk. The fact that the backscatter comes out of port 4 accounts for the direction of the arrow in Figure 13.4.

Directionality loss = 10log(P4/P1) dB

A typical figure is –40 dB.
Directivity puts the same information around the other way, if the reflected power has a level of –40 dB, then the power which is not reflected has a ratio of +40 dB. In the formula, the power levels are just inverted.

Directivity = 10log(P1/P4) dB

Insertion loss or port-to-port loss or throughput loss or tap loss This looks at a single output power compared with the input power, so in Figure 13.4 there are two possibilities. We could look at the power coming out of port 2 and compare it with the input power at port 1 or we could do a similar thing with port 3 compared with the input power at port 1.
Generally, insertion loss for any output port could be written as:

Insertion loss = 10log(Poutput port/Pinput port) dB

As an example, the insertion loss at port 2 is:

Insertion loss = 10log(P2/P1) dB

This would then be referred to as the insertion loss of port 2 or simply the portto-port loss between ports 1 to port 2.
If, in the above example, the splitting ratio was not 1:1, then port 2 may be referred to as the throughput port and so the formula above becomes the throughput loss. Similarly, if ports 3 and 1 were used, the loss could be called the tap loss.

Coupling loss

This is often overlooked. Whenever a coupler is used, it has to be joined to the rest of the circuit. This involves two pairs of connectors and a splice at each end. The losses caused by these connectors or splices must be added to the losses introduced by the coupler.


Calculate the output power at each port in the coupler shown in Figure 13.5.
Output power at port 4
The directionality loss is quoted as –40 dB.
Starting with the standard formula for decibels.

Loss = 10log(powerout/powerin)dB

The input power is 60 μW and the loss figure in decibels is –40 dB so we make a start by inserting these figures into the formula:

–40 dB = 10log(powerout/60x10-6)

Divide both sides by 10:
–4 = log(powerout/60x10-6)

Take the antilog of each side:
10–4= (powerout/60x10-6)

60 x 10–6 x 10–4 = power out

So power out of port 4 = 60 x 10–10 = 600 nW.
As the output from port 4 is so small, it is often ignored.

Output power at port 2

Port 2 is the throughput port i.e. the port with the largest output power. With a splitting ratio of 3:1, for every 4 units of power leaving the coupler there are three at port 2 and only 1 at port 3. This means that 0.75 of the power leaving the coupler goes via port 2.
But how much power is leaving the coupler? This is the input power minus the excess loss. Port 4 output can be ignored since it is so slight compared with the other power levels.

In this example, the excess loss is 1 dB so if we convert this 1 dB into a ratio, we can find the output power.
Into the standard formula we put the –1 dB (minus, as it is a loss) and the input power:

–1 dB = 10log(powerout/60x10-6)

Divide both sides by 10:
–0.1 = log(powerout/60x10-6)

Take the antilog of each side:
10–0.1 =(powerout/60x10-6)

60 x 10–6 x 10–0.1 = power out

By calculator, 10–0.1 = 0.794

So total power out of the coupler = input power x 0.794, or:
60 x 10–6 x 0.794 = 47.64 x 10–6 W

Of the 60 μW that entered the coupler, 47.64 μW is able to leave. Of this amount 0.75 leaves via port 2 so:
power out of port 2 = 0.75 x 47.64 μW = 35.73 μW

Output power at port 3

We have already calculated the power remaining after the excess loss to be 47.64 μW and since we are dealing with port 3, the tap port, the proportion of power leaving by this port is only 0.25 of the total.
Thus the output power at port 3 is 0.25 x 47.64 μW = 11.91 μW.
The results are shown in Figure 13.6.

The tee-coupler — Figure 13.7

This is simply a 1 x 2 coupler used to convey a single signal to a number of different work stations. Such stations are said to be connected on a network. It would use a high splitting ratio of 9:1 or similar to avoid draining the power from the incoming signal.

Advantages and disadvantages of a tee network

The main advantage is its simplicity. The couplers are readily available and, if required, can be supplied with connectors already fitted. This means that the network can be on-line very quickly indeed.
The disadvantage is the rapid reduction in the power available to each of the workstations as we connect more and more terminals to the network. As the power is reduced, the number of data errors increases and the output becomes increasingly unreliable. At first glance we could solve this problem by simply increasing the input power level. However we run the very real risk of overloading the first workstation.

Power levels in a tee network
  • Specification for our example system:
  • incoming power = 1 mW
  • splitting ratio of each coupler = 9:1
  • excess loss of each coupler = 0.3 dB. The couplers are joined by connectors with an insertion loss of 0.2 dB each.
Step 1 — Figure 13.8
The incoming power level is reduced by 0.2 dB by the first connector, and 0.3 dB by the excess loss.
Total power reduction is 0.2 + 0.3 = 0.5 dB.
By inserting the values into the standard decibel formula, remembering to use
–0.5 dB as it is a loss, we have:
–0.5 dB = 10log(powerout/1x10-3)

Divide both sides by 10:
–0.05 dB = log(powerout/1x10-3)

And antilog:
0.8913 =(powerout/1x10-3)

So, the input power is:
0.8913 x 1 x 10–3 = power out = 891.3 μW

Step 2

The 891.3 μW is the power just before it is split into the two output ports. As the splitting ratio is 9:1, the throughput power at port 2 is 0.9 of the available power or 802.17 μW. Similarly, the tap power is 0.1 of 891.3 μW or 89.13 μW.

Step 3 — Figure 13.9

The throughput power, 802.17 μW, is actually the input power to the next section of the network and is simply a replacement for the 1 mW input in Step 1. This new input power suffers the same connector insertion loss, coupler excess loss and splitting ratio and so the calculations would involve exactly the same steps as we have already used.

The results we would obtain are throughput loss = 643.47 μW and the tap power going to terminal 2 = 71.49 μW.

Step 4

The next section would decrease the powers by the same proportions and it would result in a throughput loss of 516.2 μW and a tap power of 57.4 μW. The same proportional loss would occur at each section of the network.

The star coupler

This is an alternative to the tee coupler when a larger number of terminals is involved as shown in Figure 13.10. The star coupler takes the input signal to a central location, then splits it into
many outputs in a single coupler. Styles of up to 1 x 32 and up to 32 x 32 are commonly available.

Advantages and disadvantages

The main advantage of using star couplers is that the losses are lower than a tee coupler for networks of more than three or four terminals as in Figure 13.11. This is because the star coupler requires only one input connector and suffers only one excess loss. The larger the number of terminals, the more significant are the benefits.

The disadvantage is that the star coupler will normally use much larger quantities of cable to connect the terminals since the star is located centrally and a separate cable is connected to each of the terminals. A tee network can use one cable to snake around the system from terminal to terminal.

Construction of couplers

Fused couplers

This is the most popular method of manufacturing a coupler. It is, or appears to be, a very simple process.
The fibers are brought together and are then fused just like in a fusion splicer as seen in Figure 13.12. The incoming light effectively meets a thicker section of fiber and spreads out. At the far end of the fused area, the light enters into each of the outgoing fibers.

A fused star coupler is made in a similar way (Figure 13.13). The fibers are twisted round to hold them in tight proximity, then the center section is fused. In the case of the reflective star, the fibers are bent back on themselves before being fused.

Mixing rod couplers — Figure 13.14

If several fibers are connected to a short length of large diameter fiber, called a mixing rod, the incoming light spreads out until it occupies the whole diameter of the fiber. If several fibers are connected to the far end they each receive some of the light.

A reflective coupler can be produced by putting a mirror at the end of the mixing rod. The light traveling along the mixing rod is reflected from the end mirror and all the attached fibers receive an equal share of the incoming light.

Variable coupler — Figure 13.15

This is more of an experimental or test laboratory tool than for the installation environment. It enables the splitting ratio to be adjusted to any precise value up to 19:1, which allows us to try out the options before the final type of coupler is purchased.

The design principle is very simple. A vernier adjustment allows precise positioning of the incoming fiber so that the light can be split accurately between the two output fibers to provide any required splitting ratio. This form of variable coupler is available for all plastic as well as glass fibers, singlemode and multimode.

Kevin M Contreras H
CI 18.255.631




Connectors and adapters are the plugs and sockets of a fiber optic system. They allow the data to be rerouted and equipment to be connected to existing systems.
Connectors are inherently more difficult to design than mechanical splices. This is due to the added requirement of being able to be taken apart and replaced repeatedly. It is one thing to find a way to align two fibers but it is something altogether different if the fibers are to be disconnected and reconnected a thousand times and still need to perform well.
If two fibers are to be joined, each fiber has a connector attached and each is then plugged into an adapter. An adapter is basically a tube into which the two connectors are inserted. It holds them in alignment and the connectors are fixed onto the adapter to provide mechanical support. An adapter is shown as part of Figure 12.1.
Although different makes are sold as compatible, it is good practice to use the same manufacturer for the connector and for the adapter.
The design of connectors originated with the adaptation of those used for copper based coaxial cables and were usually fitted by the manufacturers onto a few meters of fiber called a pigtail which was then spliced into the main system.
Most connectors nowadays are fitted by the installer although pre-fitted ones are still available. The benefit of using the pre-fitted and pigtailed version is that it is much quicker and easier to fit a mechanical splice or perform a fusion splice than it is to fit a connector, so there is some merit in allowing the factory to fit the connector since this saves time and guarantees a high standard of workmanship.
When a connector is purchased, it always comes with a plastic dust cap to prevent damage to the polished end of the optic fiber. It is poor workmanship to leave fibers laying around without the caps fitted.
Before considering the details of the various types of connector, we will look at the main parameters met in their specifications so that we can make sense of manufacturers' data.

Connector parameters

Insertion loss

This is the most important measure of the performance of a connector. Imagine we have a length of fiber which is broken and reconnected by two connectors and an in-line adapter. If the loss of the system is measured and found to have increased by 0.4 dB then this is the value of the insertion loss. It is the loss caused by inserting a mated pair of connectors in a fiber. Be careful to ascertain whether the quoted loss for a connector is per mated pair or for each connector.
Typical value: 0.2�0.5 dB per mated pair.

Return loss

This is a measure of the Fresnel reflection. This power is being reflected off the connector back towards the light source. The lasers and LEDs used for multimode working are not greatly affected by the reflected power and so the return loss is not usually quoted in this instance. In singlemode systems the laser is affected and produces a noisy output. The laser suppliers will always be pleased to advise on permitted levels of return loss.
Typical value: �40 dB.

Mating durability

Also called Insertion loss change. It is a measure of how much the insertion loss is likely to increase in use after it has been connected and disconnected a large number of times.
Typical value: 0.2 dB per 1000 matings.

Operating temperature

These are, of course, compatible with the optic fiber cables.
Typical values: �25°C to +80°C.

Cable retention

Also called tensile strength or pull-out loading.
This is the loading that can be applied to the cable before the fiber is pulled out of the connector. It is similar in value to the installation tension on a lightweight cable.
Typical value: 200 N


This is a measure of how consistent the insertion loss is when a joint is disconnected and then remade. It is not a wear-out problem like mating durability but simply a test of whether the connector and adapter are designed so that the light path is identical each time they are joined.
This is an important feature of a connector but is not always quoted in specifications owing to the difficulty in agreeing a uniform method of measuring it. Some manufacturers do give a figure for it, some just use descriptive terms like 'high' or 'very high'. The quoted insertion loss should actually be the average insertion loss over a series of matings, thus taking repeatability into account.

A survey of the main connectors

The first connectors were machined from solid brass and had to be factory fitted. Indeed there seemed to be a policy by manufacturers to preserve the 'factory fitted' requirement. Then one or two companies sold connectors that could be fitted with simple hand tools and very quickly the advertisements changed to 'easily fitted'.
Once installed, connectors can last for many years and so earlier designs are still met during maintenance work and are included in this chapter. These oldies are even now available in current catalogs.
For new installations, there are three recommended connectors. These are referred to as the 'SC', 'ST' and 'MIC/FDDI' connectors and are shown in Figures 12.4, 12.9 and 12.10.
Connectors are nearly always assembled using epoxy resin and are not reusable. There are many similarities between the various types of connectors and the early SMA (sub-miniature assembly) will serve as a suitable starting point.

SMA (sub-miniature assembly ― Figure 12.1)

The SMA connector has been superseded by the more modern designs but there are many still in use. To connect two fibers, we simply screw a connector onto each end of the adapter. It is only used for multimode systems as the losses are too high for singlemode use.

The air gap

The length of the adapter ensures that the ends of the two ferrules are separated by an air gap small enough to allow the light to jump the gap to the other fiber.
This gives rise to the first problem. How tight do we tighten up on the screw thread? Not enough and the losses will be too high. Too tight and we will grind the end faces of the fiber together and will cause the glass to crack or be scratched. If this happens, the connector must be removed and thrown away.


The ferrules have a hole through the center to take the bare fiber (primary buffer stripped off) and are made of stainless steel or ceramic. In the case of the stainless steel, the 127 μm hole has to be drilled. If the hole is slightly off-center or over-size, it can cause eccentricity loss. Of the two, ceramic is by far superior. The ceramic is 'grown' on a piece of wire of precise thickness. When the wire is removed, we are left with a much more accurate hole.
Owing to misalignment and slight variations in the assembly of the connector there are losses which vary in magnitude as the connector is revolved within the adapter. This has the advantage of allowing us to optimize the connection by monitoring the losses as the connector is revolved. This is only OK if everyone who uses the connector has the test equipment, the time and patience to make the final adjustment. It is far better to have a known loss each time the connectors are mated by ensuring that it can only go together in a single fixed position. The SMA suffers from poor repeatability because it can be assembled in any random position. It is normal practice to insert the connector, measure the loss then revolve it through 90° and take a new measurement. This is done four times and the results are averaged.


The simple screw thread offers very little protection against loosening when exposed to vibration.

The two versions, 905 and 906 (now obsolete)

The original SMA connector was called the SMA 905. To improve its performance the ferrule was modified and the new version was called the SMA 906 and is shown in Figure 12.2. A Delrin sleeve is a small plastic tube that slips

over the end of the 906 ferrule and, within the adapter, the other 906 connector slots into the other end. The soft plastic ensures the two ferrules are held in accurate alignment. Although only a few millimeters in length the improvement in the alignment is very significant. Two problems. It is possible for the sleeve to fall off but the opposite is more likely. They can become very difficult to remove and our attempts may accidentally damage the fiber.

ST (straight tip ― Figure 12.4)

This was developed by the US company, AT&T, to overcome many of the problems associated with the SMA and is now the most popular choice of connector for multimode fibers. It is also available for singlemode systems.
The problem of repeatability is overcome by fitting a key to the connector and a corresponding keyway cut into the adapter. There is now only one position in which the connector can fit into the adapter.

The screw thread of the SMA has been replaced by a bayonet fitting so that there is no worry about the connector becoming loose when exposed to vibration. The ferrule is spring loaded so that the pressure on the end of the ferrule is not under the control of the person fitting the connector. There is no SMA worries about how tight to do up the nut.

Polishing styles ― Figure 12.5

The fiber through the center of the connector is polished during the assembly of the connector to improve the light transfer between connectors. The are three

different styles called flat finish, physical contact (PC), and angled physical contact (APC). Many of the connectors are offered in different finishing styles so we see the connector name with a PC or APC added on the end. If nothing is mentioned, we assume a flat finish.
A flat finish is simply polished to produce a smooth flat end to the fiber so that the light comes straight out of the connector within the acceptance angle of the other fiber.
In the case of the PC finish, the fiber is polished to a smooth curve. There are two benefits of a PC connector. As the name implies, the two fibers make physical contact and therefore eliminates the air gap resulting in lower insertion losses. The curved end to the fiber also reduces the return loss by reflecting the light out of the fiber.
The APC finish results in very low return losses, It is simply a flat finish set at an angle, typically 80. The effect of this is that when the Fresnel reflection occurs much of the reflected power is at an angle less than the critical angle and is not propagated back along the fiber.

Fiber connector, physical contact (FCPC ) ― Figure 12.6

Also available as FC (flat finished) or FC APC (angled physical contact).
The FCPC is a high quality connector designed for long-haul singlemode systems and has very low losses. It can also be used for high quality multimode work if required and is often found on test equipment.
It looks like an SMA connector but it is keyed for repeatability.
The ferrule can be steel or ceramic inset in steel and is spring loaded or 'floating'. The end of the fiber is polished into the curved PC pattern. It can be polished flat if required and in this case it loses its PC suffix and just becomes an FC connector.

Mini-BNC ― Figure 12.7

This was developed for the US market but has not proved popular and surviveson ly because it is specified for the IBM token-ring network. Apart from being very slightly smaller, it has nothing to offer compared with the STPC.
In appearance, with its metal ferrule, it is easily mistaken for a BNC plug as used in copper based electrical systems. It is for multimode use only, uses bayonet fittings and the ferrule is spring loaded and is a PC connector.

Biconic connector

This is another connector, shown in Figure 12.8, which is also specified for the IBM token-ring network in its multimode form and is also widely used in the US for long haul singlemode telecommunications.
It is secured by screw thread rather than bayonet and has a spring loaded ferrule with a PC finish. When fitted to the adapter, the conical ferrule causes it to be self centering, thus providing low losses.
It is easily recognized by the unusual tapered ferrule and the exposed spring.

Subscriber connector (SC)

Also available in PC and APC versions and suitable for singlemode and multimode systems, it is illustrated in Figure 12.9.
This connector is designed for high performance telecommunication and cable television networks. There is a different feel about this connector when compared with the previous types. The body is of light plastic construction and has a more 'domestic' or 'office' feel about it. It has low losses and the small size and rectangular shape allows a high packing density in junction boxes. It plugs into the adapter with a very positive click action, telling us it's definitely
engaged. A very nice connector destined to succeed.

Media interface connector (MIC), fixed shroud duplex (FSD) or fiber data distributive interface (FDDI) ― Figure 12.10

Unlike the other connectors, this one has two fibers within the same cover. This allows signals to be routed in two directions at the same time. This is called duplex operation.
It uses STPC ceramic ferrules, otherwise it is another all plastic connector, with a similar feel to the SC. It is intended to be used in local area networks (LANs) to interconnect computer systems and other pieces of office equipment.

Compared with the other connectors, it seems quite bulky and is designed to be easily handled and plugged into the equipment often by the end user rather than the system installer. It is keyed to prevent accidental insertion in the wrong socket and is color coded for easy recognition.


Generally a system is designed to use the same type of connector throughout, and to ensure complete compatibility, and hence best performance, they are normally sourced from the same manufacturer.
Occasionally however, we meet two new problems.
The first is to connect two cables fitted with non-compatible connectors, say, an STPC connector to one fitted with a biconic connector. Such problems are easily solved by a wide range of 'something-to-something' type adapters. Some of these adapters do introduce a little extra insertion loss but not more than about 1 dB.

The second is to join a bare fiber to a system, quickly and easily, perhaps to connect a piece of test equipment or to try out a new light source. This is achieved by a bare fiber adapter. This is a misleading name since it is actually a connector as can be seen in Figure 12.11. It is really a bare fiber connector since it has to be plugged into an adapter. The only difference is that the fiber is held in position by a spring clip rather than by epoxy so that it can be readily re-used.

Fitting the bare fiber adapter
  • Strip off the primary buffer for about 25 mm or so and clean the fiber.
  • Press the cable grip and push in the fiber until it comes to a stop then release the cable grip. The primary buffer will not pass down the ferrule.
  • Cleave off the bit that sticks out of the end of the ferrule.
Done. Less than a minute.
The results depend on the quality of the cleave and are not as good as with a permanently fitted connector.

Plastic fiber connector ― Figure 12.12

Plastic fiber connectors are very quick and easy to fit but the insertion loss is higher than normal for glass fibers ― between 1 dB and 2 dB. The cables are connected in the usual method of having two connectors plugged into an inline adapter. Sometimes the end of the plastic fiber is polished using a simplified version of the techniques used on glass fiber and sometimes it is cleaved off as in the bare fiber adapter.

Fitting a plastic fiber connector
  • The outer jacket (2.2 mm) is stripped off for about 25 mm.
  • The fiber is pushed into the connector as far as it will go.
  • The end is cleaved or polished according to the manufacturer's instructions.
Note: there are barbs inside the connector to prevent the fiber pulling out in use. This also means that, once assembled, it is very difficult to get the connector off again so we need to study the instructions before the fiber is inserted.

Terminating a silica glass optic fiber (fitting the connector)

To avoid the job altogether, buy the connector already attached to a pigtail. Everything is done for us, all we have to do is to join it to the rest of the system by means of a fusion splice or mechanical splice.
The most usual method is called glue and polish. In essence, all that happens is that the fiber is stripped, glued into the connector and the end of the fiber is polished with abrasive film.
As usual, it is most important that we take some time out to read the instructions. It can save a lot of time and money in second attempts.

Fitting a connector on a silica fiber
  • Strip off the outer jacket, cut the Kevlar, and remove the primary buffer to the dimensions supplied with the connector (Figure 12.13).
  • Slip the flexible boot and the crimp ring onto the fiber. The crimp ring is a metal tube about 10 mm in length which will grip the Kevlar and the connector to provide the mechanical support.
  • Clean the fiber with isopropyl alcohol in the way that was done prior to cleaving. Just as a practice run, carefully insert the stripped fiber into the rear of the ferrule and ease it through until the buffer prevents any further movement. If this proves difficult, it may help if the connector is twisted backwards and forwards slightly but be careful, the fiber must not break. Check that the fiber sticks out from the end of the ferrule. If it does break, the piece of fiber can be released with a 125 μm diameter cleaning wire which is available from suppliers.
  • Mix some two-part epoxy and load it into a syringe. The epoxy is often supplied in a sealed polyethylene bag with the hardener and adhesive separated by a sliding seal. Remove the seal and mix the adhesive and hardener by repeatedly squeezing the bag between the fingers. The mixing process can be aided by the use of a grooved roller which is rolled to and fro across the packet.
  • Insert the syringe into the connector until it meets the rear of the ferrule. Squeeze epoxy in slowly until a tiny bead is seen coming out of the front end of the ferrule. This shows that the ferrule is well coated with epoxy.
  • Carefully insert the stripped fiber into the rear of the ferrule until the buffer prevents any further movement. Again, take care not to break the fiber. If it does break, the fiber must be prepared again, as the dimensions will now be incorrect. The epoxy is very difficult to remove from the ferrule and the cleaning wire is not guaranteed to work under these conditions. Acetone may be helpful. This is a job best avoided.
  • Arrange the Kevlar over the spigot at the back end of the connector and slide the crimp ring over the Kevlar as shown in Figure 12.14. One end of the crimp ring should overlap the spigot and the other should cover the fiber jacket.
  • Using a hand crimping tool, crimp the Kevlar to the spigot and at the other end of the crimp ring, grip the cable jacket. This ensures that stress is taken by the Kevlar strength members and not by the optic fiber.
  • Put the connector into a small oven to set the epoxy. The oven, shown in Figure 12.15, is an electrically heated block of metal with holes to take the connectors. This will take about ten minutes at 80°C. When cured, the golden epoxy may have changed color to a mid to dark brown.
  • When it has cooled down, fit the boot.
  • Using a hand cleaver, gently stroke the fiber close to the end of the ferrule and lift off the end of the fiber (Figure 12.16). Store the broken end safely in a sealed receptacle for disposal.
  • The end of the fiber must now be polished. The easy way is to insert the fiber into a portable polishing machine and switch on. After about one minute, it's all over, the fiber is polished. The alternative is to do it ourselves.
  • Consult the instructions at this stage, each manufacturer has a recommended procedure for each type of connector and they usually know best. We will need a flat base of plate glass, hard rubber or foam about 200 mm square. A soft base is normally used for the PC finish. The abrasive sheet is called a lapping film. It is a layer of aluminum oxide on a colored plastic sheet. Silicon carbide and diamond films are also available. The roughness of the abrasive is measured by the size of the particles and is colored to aid recognition. Grades vary from the coarse 30 μm colored green down to the ultra smooth white at 0.3 μm. Beware ― not all types of film employ the same colors.
  • Using a magnifying glass, observe the end of the ferrule to see how much glass is protruding above the tiny bead of epoxy (Figure 12.17). This unsupported glass is easily broken and should be abraded down to the epoxy level by using a strip of coarse grade film (9 μm) held in the hand and stroked gently over the fiber. Be very careful not to apply too much pressure and stop when the epoxy is reached. If the fiber is too long or too much pressure is applied at this stage, the course lapping film will send shock waves down the fiber and it will crack. The crack has the characteristic 'Y' shape shown in Figure 12.18. It runs vertically down into the fiber and no amount of polishing will do anything to help the situation. We have lost a connector and gained some experience.
  • The fiber is supported perpendicular by a polishing tool called a dolly or a polishing guide (Figure 12.19). Each dolly is designed for a particular type of connector to ensure the correct dimensions and fitting mechanisms. The suppliers will always advise on the grades of film and methods to be used to be used. Once again it is worth reading the instructions carefully if an unfamiliar connector is being fitted. We start with the coarsest grade recommended, probably about 3 μm. Lay the film on the base material and attach the fiber in the dolly. Using only the weight of the dolly, slide the dolly in a figure of eight pattern
    for about eight circuits.
  • Using a microscope or magnifying eye glass, observe the end of the ferrule. There will be a large dark area which is the epoxy. If this is the case, repeat the above stage until the epoxy becomes lighter in color and eventually has a transparent feathered edge.
  • When this happens, remove the 3 μm film and clean the whole area, including the dolly and the connector. Clean it very carefully with a tissue moistened in alcohol or demineralized water. Make quite certain that no trace of abrasive is transferred from one stage to the next.
cleanliness is essential
  • Using a fine grade, about 0.3 μm, repeat the figures of eight. The end of the fiber takes on a bluish hue with some pale yellow from any remaining resin. Black marks are, as yet, unpolished areas or possibly water on the surface of the ferrule (dry it off before checking). A little more polishing as the resin finally disappears. If the end of the fiber is not clear and blue with no marks or scratches, polishing should be continued.
  • Fit a dust cap to protect the fiber.
Some alternatives

Connectors are manufactured with hot-melt glue coating the inside as an alternativeto using epoxy. This is convenient since we can always remelt it if we need to reposition the fiber.
Some epoxy resin is cured by ultraviolet light rather than by heat.

The polishing can be performed either dry or wet. Metal ferrules are often polished wet and ceramic dry, but there are too many exceptions to offer this as a rule. If wet polishing is recommended in the instructions, a few drops of wetting agent is applied to the surface of the lapping film. Not too much, otherwise the dolly will tend to aquaplane over the surface without any polishing action. We should not use tap water as it includes impurities which under the microscope look like boulders and would scratch the surface of the fiber.

Final inspection

The finished connector should be inspected by a microscope with a magnification of at least 100 times. We are checking for surface scratches and chips caused by the polishing and also for cracks within the fiber rather than on the surface. Front lighting is good for spotting surface defects. Rear lighting, obtained by shining light down the fiber from the far end and therefore passing through the fiber is better for locating cracks within the fiber.
When observing any possible defects, the core is obviously of primary importance. The cladding is split into two halves. The outer part of the cladding does not greatly affect the operation of the fiber and we can be more forgiving of any failings in this area.
Main failing points ― Figures 12.20 and 12.21
  • Chips and cracks which extend into the inner part of the cladding.
  • Cracks that extend for more than 25% of the circumference of the cladding.
  • Scratches in the core area of a severity which is not consistent with the polishing techniques suggested by the manufacturer. This is what will happen if the grit from one stage in the polishing process is able to contaminate the finer grade lapping film.

Kevin M Contreras H
CI 18.255.631
http://www.kiet.edu/ensite/downloads/Introduction to Fiber Optics - John Crisp.pdf