Fiber Optic Training Maula

8/6/2019 Fiber Optic Training Maula

1/56

Fiber Optic

Technical Training Manual

Introduction

Welcome to the FIBER OPTIC TECHNICAL TRAINING

MANUAL. This manual is intended or readers with

knowledge o electronics but either little or no experi-

ence with ber optics. The manual is divided into six

sections:

Fiber Optic Technology 2

Networking Technology 18

LAN Standards 24 Optical Transmission & Types o Fiber 40

Understanding Fiber Optics Data Sheets 43

Glossary 45

How to use the Manual

Basic inormation is displayed with a standard white

background. Inormation that is more detailed or techni-

cal is in shown with a shaded background. An example is

shown on the right:

Bit Error Rate

In an actual transmission system, various electronic noise

sources, such as the shot (random) noise o the rst

transistor in the PIN-photodiode preamplier or noise

on the power-supply line o the host digital system, can

change or corrupt the stream o data bits. The number o

these corrupted bits divided by the total number o re-

ceived bits within an arbitrary time period is the bit error

rate (BER). The lower the BER value, the ewer the number

o errors in the transmission.

The bit-error ratio is usually given by a number such as

10-6. This means that on average, one error occurs or

every million pulses sent. Typical error ratios or optical

ber systems range rom 10-9 to 10-12 (localarea networks

require a BER as low as 10-12). The BER can be measured

by repeatedly transmitting and receiving a suitable

length o a pseudo-random bit sequence (PRBS) data.

The necessary optical power to achieve a given signal-

to-noise ratio, and thereore a certain BER, is called the

receivers sensitivity.

I you want or need only the basic inormation you can

skip these sections.


2/56

Feature Benet

Electromagnetic

Intererence

(EMI)

Because optical ber transmits light rather than electrons, it neither radiates EM (electromagnetic) elds nor is it susceptible to any

EM elds surrounding it. This is important in a number o applications:

Industrial control, where cables run along actory oors in proximity to electrically noisy machinery. The optical medium

permits the control signals to be carried error-ree through the electrically noisy environment.

Telecommunications equipment manuacturers use optical ber because it eliminates cross talk between the

telecommunication lines.

Financial institutions and gaming equipment manuacturers require data security. Tapping into a ber cable without being

detected is extremely dicult.

High Bandwidth Optical ber has a relatively high bandwidth in comparison to other transmission media. This permits much longer transmis-

sion distances and much higher signal rates than most media. For example, all undersea long-haul telecommunications cable is

ber-optic. This technology is allowing worldwide communications (voice, video and data) to be available mainstream. With new

technologies, such as VCSEL transmitters, parallel optics and wavelength division multiplexing (WDM), services such as video-on-

demand and video conerencing will be available to most homes in the near uture.

Voltage Isolation

Voltage Insulation

Fiber isolates two diferent voltage potentials. and insulates in high-voltage areas. For example, it will eliminate errors due to

ground-loop potential diferences and is ideal or data transmission in areas subject to requent electrical storms, and presents no

hazard to the eld installer.Weight and Diameter A 100 meter coaxial cable transmitting 500

megabits o data per unit time is about 800%

heavier than a plastic ber cable and

about 700% heavier than a hard-clad

silica cable, all o equal length and transmitting

at the same data rate.

The relative diameters o various types o

ber cable and coaxial copper cable or the

same distance and data rate are shown in

Figure 1.

available with other technology:

Complete input/output electrical isolation

No electromagnetic intererence (EMI) susceptibility or

radiation along the transmission media

Broad bandwidth over a long distance

Light-weight, small-diameter cables

Equal to the cost o copper wire and connectors

(except when copper wire is already installed)

200/230- m diameter

hard-clad silica (HCS)

62.5/125- m diametermulti-mode glass

9/125- m diametersingle-mode glass

1-mm diameter

plastic optical fiber (POF)

5-mm diameter

coaxial copper wire(with 2-mm diameter jacket)

Figure 1.

A communication link can be made with a variety o

media, including twisted-wire pair, coaxial cable, RF (radio

requency) or IR (inrared) wireless signal transmission, or

ber-optic cable.

Fiber-optic systems use pulses o light traveling through

an optical ber to transmit data. This method oers many

advantages over copper wire, some o which are not

Fiber Optic Technology


3/56

The Fiber-Optic Link

A basic link has three main components, each with a

specic unction. The transmitter contains a light-

emitting diode (LED), laser diode (LD), or vertical cavity

surace-emitting laser (VCSEL) that converts an electri-

cal current into an optical signal. The receiver contains a

photodiode that converts the light back into an electri-

cal signal and an amplier that makes the signal easierto detect. The ber-optic cable carries the optical signal

between them.

Transmitter Receiver

CableLED

Photodiode

Figure 2.

EmittersThe two major groups o optical emitters or communica-

tion links are light-emitting diodes (LEDs) and lasers. The

VCSEL (Vertical Cavity Surace- Emitting Laser) is oten

categorized separately because o its similarities to both

lasers and LEDs.

Emitters are made rom semiconductor compounds. Ele-

ments in the third column o the periodic table combined

with elements in the th column (known as III-V

compounds--GaAs is one example) emit particles o light

(photons) when subjected to an electric current.

Using the right combination o elements, and the right

processing parameters, engineers can produce a varietyo dierent wavelengths, brightness, and modulation

speeds (how ast the emitter can be turned on and o) o

the emitted light.

By optimizing its wavelength, the light can be e-

ciently coupled into the appropriate type o ber when

combined with an optimally designed optical sub-

assembly (OSA). The wavelengths span the visible and

inrared spectrums up to 1550 nm. Typically, wavelengths

1300 nm and above are considered to be long wave-

length (LWL) and those below 1300 nm are considered to

be short wavelength (SWL).

LEDs

LEDs are diodes made o III-V material, which emit light.

As electrons pass through an active region o semicon-

ductor material, they cause photons to exit. The light

produced by an LED has a pure spectrum and is relatively

brighter than a standard light-bulb emitter. The LED can

be turned on and o at speeds which make data commu-

nications possible.

The most common wavelengths or LED emitters are

650 nm or plastic ber; 850 nm and 1300 nm or multi-

mode glass ber. Surace-emitting LEDs (SLEDs) are in-

expensive to manuacture because they emit light rom

their surace and so require no precision cleaving. LEDs

can also be made to emit light rom the edge, producing

a more narrow, collimated beam that can launch into

single-mode ber. Edge-emitting LEDs are called ELEDs.

Lasers

Lasers are devices that eciently convert electrical pow-

er to optical power due to a process called stimulated

emission. Photons produced by a semiconductor laser

are initially trapped in a cavity and bounce o refec-

tive suraces at each end. They then repeatedly pass

through an active gain region, where on each pass they

stimulate the production o additional photons. Eventu-

ally the power in the cavity increases enough so that a

very bright, coherent, collimated beam o light is emitted

rom one end o the cavity.

The gain medium determines the properties o the out-

put light, including wavelength, output power, and

whether the light exits in pulses or in a continuous wave.

Semiconductor lasers are devices with a unique structure

that makes them more ecient.

1. Laser Diodes

In Laser diodes, the active gain material is a

III-V semiconductor compound. Because the

active region is very small and the input current

is relatively low, laser diodes output optical power

in milliwatts. By contrast, some medical lasers and

industrial lasers output power in megawatts.

Because laser diodes emit light rom the edge o

the die, they are more dicult to manuacture and

handle than LEDs. These devices must be cleaved,

and mounted on their sides. Furthermore, it is more

dicult to align the laser to the connector errule than

it is to align an LED to the connector errule. For these

reasons, lasers are more expensive than LEDs. Laser

diodes are commonly ound in 1300 nm and 1550 nm

wavelengths or communications, and 980 nm and

1480 nm or pumping lasers.

2. VCSELs

A type o laser that has recently emerged in

the communications market is the vertical-cavity

surace- emitting laser (VCSEL). This laser behaves as

a laser diode, but can be manuactured similarly to

an LED, because it emits light rom the surace. This

makes the VCSEL very attractive to the market due

to its perormance/price ratio. VCSELs are being

produced at 650 nm and 850nm, with 1300 nm to be

produced in the near uture.


4/56

LED Operation

Light-emitting diodes operate in the same way as

diodes made rom silicon or germanium: they allow

current to fow in the orward direction and block current

rom fowing in the reverse direction. Current fow-

ing through the diode in the orward direction generates

a near-constant voltage across the diode, which is primar-

ily determined by the material rom which the diode ismade. For example, diodes made rom silicon have or-

ward voltages o about 0.7 V and LEDs have orward volt-

ages o around 1.3 V to 2.5 V. The orward current fowing

through the LED generates light.

The light output o an LED is nearly proportional to

the magnitude o its orward current. Thereore, control-

ling the orward current controls the light output o the

LED.

The circuit in Figure 3 shows one o the simplest ways

to drive an LED. The LED current is determined by the

dierence between the input voltage and the orward

voltage o the LED, divided by the series resistance. Itis the current through the LED that controls the light

output.

or use with glass optical ber. Glass ber loses a mini-

mum o light at wave-lengths around 1350 nm, which is

within the spectral band o an InGaAsP LED.

Laser Operation

A laser (Light Amplication by the Stimulated Emission o

Radiation) operates in a manner similar to an LED. The

dierence is that the LED has spontaneous output onlywhen orward biased. The laser begins with spontane-

ous output at low orward currents, increased to a point

known as the Threshold Current level. Biasing above the

threshold current causes the device to reach the stimu-

lated emission region o operation.

The advantage o the laser is that its stimulated emis-

sion output optical power launched into a small-core op-

tical ber is typically ar higher than that o an LED. Typi-

cally the laser also has a narrower output optical spectra,

thus providing better perormance at high bit rates over

long distances. The disadvantage is that the threshold

current is temperature dependent and circuit designers

must take this into account. The added circuit complexity

increases the cost o the laser transmitter.

Fabry-Perot Lasers

A FP laser has two parts: a semiconductor optical am-

plier that provides gain and mirrors to orm a reso-

nator around the amplier. The semiconductor optical

amplier is ormed by applying current to a lower-

bandgap active layer surrounded by higher-bandgap

materials. The low-bandgap and high-bandgap layers

orm an optical waveguide to direct the light that is

produced. Light radiates rom the laser in a narrow-

angled cone. As the current increases, the opticalamplier gain increases. When the amplier gain equals

the mirror loss, the lasing threshold is reached; the laser

begins to oscillate above threshold current. FP lasers can

produce many mW o output optical power and can be

modulated to very high rates by varying the current to

the laser diode. VCSELs

Vertical cavity surace-emitting lasers (VCSELs) emit light

perpendicular to the top plane o a semiconductor waer.

The VCSEL uses a Bragg refector, a multi-layer dielectric

mirror composed o alternate layers o high and low

index o reraction material. For data communication ap-

plications, the diameter o the laser is made large enoughto support multiple modes. This causes the central lasing

peak to spectrally broaden. The wider spectral width is

oten designed into VCSELs to avoid mode-selective loss

in multi-mode ber applications.

The major advantage o the VCSEL design is ease o man-

uacture, simplied on-waer testing, and simple packag-

ing, resulting in lower costs. The device has higher output

power and higher modulation rates than SLEDs (Surace

Light-Emitting Diodes).

+

R

IF

IFV FV in

R

-=

+-

V FV in

Figure 3.

LED Materials

Numerous elements and compounds can produce

light, rom the ultraviolet to the inrared. Very ew,

however, are practical or LED devices. At the present

time, most commercially available LEDs are made rom

combinations o gallium (Ga), arsenic (As), phosphorus

(P), Aluminum (Al), and Indium (In).

Certain materials can be made to produce a range

o wavelengths by adjusting the relative proportionso the constituent elements. It should be noted that

LEDs actually emit light over a narrow range o wave-

lengths, not just at a single wavelength. For example, an

820 nm LED might actually emit light rom about 800 to

840 nm, with the peak intensity at 820 nm.

The material used depends on the application. For

example, plastic optical ber has a minimum loss o

transmitted light at wavelengths around 650 nm. There-

ore, an LED made rom GaAsP would be best suited

or use with plastic ber. AlGaAs is also a good material


5/56

5

Detectors

Detectors are typically PIN photodiodes that unction ex-

actly the opposite o an emitter diode: they absorb

photons and convert them into electrons. Photodetectors

can be made o either pure silicon, or a III-V compound,

depending on the wavelength requirements. The

detector must be able to detect the same wavelength

emitted by the transmitter, or the link will not operate. The detector is always ollowed by a pre-amplier

(preamp) that converts the current coming rom the

diode to a voltage, which is used by most amplication

schemes.

Other detectors include avalanche photo-diodes, which

are more complex to operate due to their control

circuitry, and silicon photo-detectors, such as photo-tran-

sistors or photo-FETs.

One important property o a receiver is its sensitivity. This

property denes how small a signal the receiver is able to

detect. The receiver must be able to detect a signal over

the top o any electromagnetic noise that may be inter-ering with the system. A careul design o the preamp

and ollowing circuitry, plus metal shielding around

the receiver, can improve the sensitivity.

The importance o understanding this technology lies

in how it relates to two important specications o a

ber-optic link: distance and data rate. The rst step is

to determine the distance over which the data needs to

be transerred and the data rate at which it is being sent.

Some applications will be dened by industry standards,

which clearly state specic distances and data rates. Oth-

er applications will be proprietary, and you will have to

work closely with the customer to determine the details.

Fiber as an Optical Waveguide

Reected light

The most basic optical waveguide is the step-index ber,

which consists o two layers, a core and a cladding. These

layers are made o materials that have dierent optical

transmission properties. One important property o these

materials is the index o reraction (n), a material constant

that determines the direction o the light through the

material.

Cladding (n) = 1.460

Core (n) = 1.463

Figure 4.

The index o reraction o the core must be larger than

that o the cladding or the light to travel through the

waveguide:

(ncore > ncladding provides total internal refection)

For some glass bers, the cladding is SiO2 (glass) and the

core has dopants (such as germanium) to increase theindex o reraction.

The light travels through the waveguide in paths called

modes. Each ray o light has a dierent mode, which

creates a crowded waveguide. Not all o the modes

that exit a light source are transmitted through the

waveguide. A specic angle, the acceptance angle, serves

as a cuto or the light modes.

Transmitted

Transmitted

Lost

Lost

Cladding

Cladding

Core

Figure 5.

All light rays that enter the ber at an angle greater than

this critical angle, , exit through the cladding and are

lost. This angle can be calculated based on the value, n,

o the core and cladding, and is usually represented as

the numerical aperture, NA, a standard specication oany ber. The NA is dened in Figure 6.

Core

NA = sin = n ncore clad

2 2

Cladding

Cladding

Figure 6.


6/56

Bandwidth

Bandwidth determines how much inormation can be

transmitted through a waveguide at one time. An

analogy would be the speed o a fuid fowing through

a pipe and, in act, optical bers are oten called light

pipes. One o the reasons that optical transmission is

more desirable than electrical transmission is because

optical oers a much wider bandwidth. This is due to theproperties o light versus electricity.

The bandwidth o a transmission medium is limited by

an eect called dispersion. This means that the input

signal is spread out or distorted as it passes throughFigure 7.

Dispersion

In an optical medium, such as ber, there are three

types o dispersion: chromatic, modal, and material.

Chromatic dispersion results rom the spectral width

o the emitter. The spectral width determines thenumber o dierent wavelengths that are emitted rom

the LED or laser. The smaller the spectral width, the ewer

the number o wavelengths that are emitted. Because

longer wavelengths travel aster than shorter wave-

lengths (higher requencies) these longer wavelengths

will arrive at the end o the ber ahead o the shorter

ones, spreading out the signal.

One way to decrease chromatic dispersion is to nar-

row the spectral width o the transmitter. Lasers, or

example, have a more narrow spectral width than LEDs.

A monochromatic laser emits only one wavelength

and thereore, does not contribute to chromatic disper-sion.

Modal dispersion deals with the path (mode) o each light

ray. As mentioned above, most transmitters emit many

dierent modes. Some o these light rays will travel

straight through the center o the ber (axial mode) while

others will repeatedly bounce o the cladding/ core

boundary to zigzag their way along the waveguide, as il-

lustrated in Figure 8.

Cladding

Core Axial Mode (shortest path)

Low-order Mode (shorter path)

High-order Mode (longer path)

Cladding

Figure 8.

thereore contribute to modal dispersion. One way to

reduce modal dispersion is to use graded-index ber.

Unlike the two distinct materials in a step-index ber,

the graded-index bers cladding is doped so that the re-ractive index gradually decreases over many layers. The

corresponding cross-sections o the ber types are shown

in Figure 9.

MultimodeGraded-Index Single-mode

Multi-modeStep-Index

Core

Cladding

Cross

Section

Refractive

Index

Profile

Light

Path

Figure 9.

Note: Shading is used only to make the illustration clearer

With a graded-index ber, the light ollows a more curved

path. The high-order modes spend most o the time

traveling in the lower-index cladding layers near the

outside o the ber. These lower-index core layers allow

the light to travel aster than in the higher-index center

layers. Thereore, their higher velocity compensates or

the longer paths o these high-order modes. A good

waveguide design appreciably reduces modal dispersion.

Modal dispersion can be completely eliminated by

using a single-mode ber. As its name implies, single

mode ber transmits only one mode o light so there

is no spreading o the signal due to modal dispersion.

A monochromatic laser with single-mode ber complete-

ly eliminates dispersion in an optical waveguide but is

usually used in very long distance applications because

o its complexity and expense.

the medium and has a wider or dierent shape at the

receiver. The more dispersion in a signal, the shorter

the time between the unique data bits. I they get close

enough to each other, they can overlap, causing errors in

the signal, as illustrated in Figure 7.

The modes that enter at sharp angles are called hig-

horder modes. These modes take much longer to trav-

el through the ber than the low-order modes and


7/56

7

The amount o inormation that an optical wave-

guide can transmit is usually specied by the band-

widthdistance product in MHz-km. For a step-index

multimode ber, distortion eects tend to limit the

bandwidth- distance product to about 20 MHz-km. In

gradedindex bers, pulse broadening is minimized at a

specic operating wavelength.

The attenuation o the ber that you choose may limit the

distance over which the data can be sent. The disper-sion eects o the link can limit both the distance and the

data rate. It is generally a compromise between the two:

the longer the distance, the lower the data rate that can

be sent, and vice versa. A typical distance versus data rate

plot appears as shown in Figure 11.

High-order Modes (faster)

Graded-index Fiber

Low-orderModes

(slower)

Cladding

Cladding

Figure 10.

Note: Shading is used only to make the illustration clearer

Distance

(m)

Data Rate

(MBd)

Figure 11.

The fat line is where the ber does not contribute to

any limiting eects. The sloping line is where the data

rate decreases as the distance increases. Finally, there

is a cut-o point, the vertical line, at which the receiver

can no longer detect the signal.

Dierent types o ber and dierent emitter wave-

lengths and properties will change the shape o

this curve, so you must take all components o the linkinto account when determining a solution. Generally,

distance versus data rate graphs and tables are available

or most Avago general-purpose ber-optic components.

Table 1 compares the major specications o ber-optic

cable.

Table 1.

Specication

Core/Cladding

Plastic/Plastic Glass/Plastic Glass/Glass

Bandwidth-Length-

Product (MHz-km)

0 00-1500

Core Diameter (m) 50 - 000 15 - 00 50 - 100

Transmission Dist. Short Medium Long

Attenuation (dB/km) 50

Numerical Aperture 0.50 0.7 0.0

Wavelength (nm) 50 800 100 - 1550

Attenuation

As light pulses travel through the ber, they lose someo their photons, which decreases their amplitude. This

is know as attenuation, the other major actor that limits

signal transmission (light output power) and consequent-

ly, reduces the length o the ber. Attenuation is usually

specied in decibels per kilometer or per meter (dB/km

or dB/m). Attenuation o optical power in a ber has

three main causes: scattering, absorption, and bending.

Attenuation varies over the light spectrum.


8/56

8

Rayleigh Scattering

A ray o light is partially scattered into many directions by

microscopic variations in the core-cladding interace. The

angle o some o the light rays impinging at these varia-

tions is changed enough so that they are reracted onto

dierent paths and do not experience TIR (see Snells

Law). For this reason, some light energy is lost. Scatter-

ing is responsible or up to 90% o the total attenuation.Rayleigh scattering causes about a 2.5 dB/km loss at 820

nm but less than 1 dB/km in wavelengths over 1000 nm.

Absorption

Impurities in the glass, which absorb light energy

and turn photons into phonons (heat), include ions o

copper, iron, cobalt, chromium and the hydroxyl (OH-)

ions o water. The OH-ions o water and the molecular

resonance o SiO2 are the principal reasons that light

energy is absorbed by the ber.

Bending

Two types o bending exist: micro-bending and

macrobending. Micro-bending is microscopic imperec-

tions in the ber geometry, such as, rotational asymmetry,

variations o the core diameter, and course interaces

between the core and cladding caused by pressure, ten-

sion and twist. Macro-bending is due to bers curved

around diameters o about a centimeter, causing less

than total refection at the core-to-cladding boundary.

Bending loss is usually unnoticeable i the diameter o

the bend is larger than 10 cm.

In multi-mode bers, a large number o modes is al-

ways present; and each mode in a ber is attenuated di-

erently (dierential-mode attenuation). In single-mode

bers this eect is not as much o a problem.

Other Factors

Optical power loss in a ber can also be caused by:

light refected o the ends o the ber by dierences in

the reractive indices o the core and air at the

interaces (Fresnel refection)

poor nishing o the ber ends, which scatters a

portion o the light

variations in the diameters and cross sections o

the core and cladding along the length o the ber

Propagation Delay

Propagation delay is the length o time it takes the

output o a device to go rom one state to another

state, either low to high or high to low, ater having

been stimulated to change at the input. It is also a

measure o the time required or a signal or pulse to

travel through a medium. Table 2 lists the typical propa-

gation delay times in ns/m or various media.

Table 2.

Medium Delay (ns/m)

Copper Wire .0

Plastic Fiber .8

Glass Fiber 5.0

The extra delay in travel time through a ber (com-pared to copper) does not necessarily reduce the

bandwidth; however, ailure to allow or it in system

design can have an adverse eect on message protocol.

When there are parallel logic paths using individual paral-

lel links, the signals can arrive at the other end o the links

at dierent times because the transit time or each link

is dierent. I this happens, individual link delays might

aect the perormance o the circuit when all parallel

paths are strobed at once. The strobe edge should be

timed to just ollow the longest individual transition

delay and to precede the shortest individual transition

delay.

Pulse-Width Distortion (PWD)

Pulse-width distortion reers to the time dierence be-

tween the delay o a low-to-high edge and the delay

o a high-to-low edge o the output signal waveorm.

Ideally, low-to-high and high-to-low transitions should

have exactly the same time delay. The time dierence

between them is distortion, and this can limit the sig-

nal rate.

Wavelength Division Multiplexing (WDM)

The data rate or a single optical channel eventual-

ly reaches a limit due to chromatic and other types o

dispersion (see above). The available bandwidth can

be increased by adding multiple wavelength channels.

A wavelength multiplexer combines the laser wave-

length channels onto a single ber with low loss, then

later separates them at the receiver end o the link.

Datum 1

Datum 2

Datum 3

Datum 4

1

2

3

4

Datum 1

Datum 2

Datum 3

Datum 4

1

2

3

4

Amplifier Amplifier

Wavelength

Channel

Adder/Dropper

DemultiplexerMultiplexer

Figure 12. Wavelength division multiplexed ber-optic link

In the past, when signals needed to be routed to a

specic location, the optical signal was detected and

routed electronically. WDM systems allow the optics to

route the optical signal. These systems can also add and

drop certain wavelength channels rom a ber while si-

multaneously sending on other channels.


9/56

Figure 13 depicts the attenuation or a specic type o

plastic optical ber. As the graph shows, certain wave-

lengths are ideal or transmission because they all in thevalleys o the curve. For this type o plastic ber, the best

suited wavelengths are 570 nm and 650 nm. For most

glass bers there are valleys around 850 nm, 1300 nm,

and 1550 nm. For other types o ber, such as HPCS (Hard

Plastic Clad Silica) ber, the attenuation curve is relatively

fat over a wide spectral range.

dB = 10 log10Po

Pi

Each type o ber has a dierent attenuation versus

wavelength curve, none o which is linear or exponential,

but rather a complicated series o peaks and valleys.

The attenuation o the light through a glass optical

waveguide has been decreased over the last 25 years to

less than 0.02 db/km and scientists are trying to decrease

the attenuation even more.

500 550 600 650 700 750 8000

100

200

300

400

500

600700

800

900

1000

1100

1200

1300

1400

Attenuation(dB/km)

Wavelength (nm)

Figure 13.

This eature makes these types o bers useul or a

variety o applications.

Attenuation loss in a ber can be calculated by taking the

logarithmic ratio o the output power (Po) to the inputpower (Pi):

Indirect Measurements

Cutback Method

The attenuation o a ber can be measured by trans-

mitting a signal through it and measuring the power

at the opposite end. The ber is then cut near the input

end without changing the launch conditions and the

power is measured again. The loss per unit length can be

calculated rom the dierence between these two values.

Substitution Method

A short length o reerence ber is compared with the at-

tenuation o the ber being tested. Both the cutback and

substitution methods have drawbacks. The attenuation

measured by the cutback method varies according to the

numerical aperture and the spectral bandwidth o the

source. In the substitution method, the coupling losses

o the ber being tested and the reerence ber may be

dierent, so the value derived may not be correct.

Direct Measurement

A more reliable technique or measuring attenuation

in a ber is optical time-domain refectometry (OTDR).

In this method, a brie signal (a short burst o optical pow-

er) is introduced into the ber. When this signal encoun-

ters imperections or discontinuities, some o the light is

refected by a beamsplitter (a glass plate at a 45 degree

angle to the incident light or two prisms cemented to-gether with metal-dielectric lm between their aces)

onto a detector, where it is amplied and displayed on an

oscilloscope.

The ODTR display is a smooth curve or a continu-

ous length o ber and an irregular curve i disconti-

nuities or splices exist. The advantage o OTDR is that it

yields more accurate results.


10/56

10

Bit Error Ratio

In an actual transmission system, various electronic noise

sources, such as the shot (random) noise o the rst tran-

sistor in the PIN-photodiode preamplier or noise on the

5-volt power-supply line o the host digital system, can

change or corrupt the stream o data bits. The number

6

8

10

12

14

16

1E-3 1E-5 1E-7 1E-9 1E-11 1E-13 1E-15

THRESHOLD-TO-NOISERATIO

-(VP-P/V

RMS)

BIT-ERROR RATIO - (BER)

Figure 14. Receiver Threshold-to-Noise Ratio vs. BER

At any xed temperature the total value o the receiv-

ers random noise plus the host systems noise can be

assumed to be a constant. So, the most obvious way

to reduce the probability o error is to increase the ampli-

tude o the optical signal applied to the receiver.

Another technique or lowering the error rate is to

improve the receivers ability to reject electrical noise

rom the system it is in. One method is to use small

PIN diode pre-amps in the receivers rst stage that do

not unction as ecient antennas. A second method,

or extremely noisy applications, is to use PIN diode pre-

amps in electrically conductive plastic or all-metal pack-

ages.

The worst source o noise is usually the host sys-

tems power (usually +5 V) supply. The host systems +5

volt supply normally powers the ber-optic receiver, the

ber- optic transmitter and an entire system comprised

The bit-error ratio is usually given by a number such as 10 -6. This means that on average, one error occurs or every

million pulses sent. Typical error ratios or optical ber

systems range rom 10-9 to 10-12 (local-area networks

require a BER as low as 10-12). The BER can be measured

by repeatedly transmitting and receiving a suitable

length o a pseudo-random bit sequence (PRBS) data.

The necessary optical power to achieve a given signal-

to-noise ratio, and thereore a certain BER, is called the

receivers sensitivity.

o relatively noisy digital circuits. Simple and inexpen-sive power supply lters have been proven to work in a

wide range o system applications. Simple, third-order

pi power-supply lters are normally sucient to pro-

tect the ber-optic receiver rom very noisy host systems,

but in extremely noisy applications, additional power

supply ltering might be needed.

Receiver sensitivity is specied at a BER; however, in prac-

tice, the BER is too low to be directly measured in a useul

time period. For this reason, the transceiver is usually

tested during an articially shorter time at a much higher

BER. The resulting data is then tted to a curve and the

optical input power at the required BER is extrapolated.Note:

The receivers signal-to-noise ratio determines the probability o

error; small increases in this ratio will sharply reduce the probability

o error. Optical budgeting is a design technique that takes system-

and component-level parameters into account.

System-level parameters include:

transmitter output power (PT dBm avg.)

receiver sensitivity--the minimum optical power

that the receiver requires to operate at the specied

data rate and bit error ratio (PR dBm avg.)

ber losses (caused by attenuation, bandwidth

and insertion losses rom connectors and are

dependent on the type o ber)

dynamic range

Component-level parameters include:

data rate

bit-error ratio (BER)

data pattern

These two sets o parameters allow a designer to deter-

mine the distance and power budgeting or the ber link.

o these corrupted bits divided by the total number

o received bits within an arbitrary time period is the

bit error ratio (BER). The lower the BER value, the ewer

the number o errors in the transmission.

Optical Power Budget

Power budget is the dierence between the mini-

mum output power o the transmitter and the mini-

mum sensitivity o the receiver. I optical power losses

are greater than the worst-case power budget, the BER

increases and the reliability o the link decreases.


11/56

11

Loss = 10 logPout

Pin

For example, i PIN = xW), then

where Pin and Pout are the optical powers into and out o

the loss element.

Note:

dBm is a unit used to describe optical power with respect to onemilliwatt.

Each o these losses is expressed in decibels (dB) as

P = 10 Log dBm.X

1000 (W)IN

Eye Diagrams

An oscilloscope display o a long stream o random bits

(light pulses) is used to determine transmitter and

receiver perormance. Due to distortion o the waveorm

(by a variety o actors), the displayed light pulses do

not align, but instead overlay one another and appear

blurred, as shown in Figure 15.

The important point is that the center o the pat-

tern ormed by these pulses must be open: the wider theopening, the less the distortion and jitter.

Figure 15.

The link loss budget considers the total optical power loss

allowed between the light source and the photodetec-

tor, and allocates this loss to cable attenuation, connec-

tor loss, splice loss, and system margin.

The optical power budget is PT - PR, where PT is the power

launched by the transmitter and PR is the power required

by the receiver or proper operation. These values are

usually expressed in dBm, so the dierence is expressed

in dB. I the losses between transmitter and receiver

exceed the power budget, then the power at the receiver

will not be adequate to assure proper operation.

In the ollowing example, PT = -12 dBm and PR = -24 dBm.

Substituting these values into the equation:

PT - PR = -12 - (-24) = -12 + 24 = +12 dB,

and so, the receiver has sucient power to operate.

Much inormation about system perormance can

be gathered rom the eye pattern:

The width o the eye denes the time interval

during which the received signal can be sampled

without error rom inter-symbol intererence

Amplitude distortion in the data signal reduces

the height o the eye opening. The smaller this

dimension, the more dicult it is to detect the signal

without error

The height o the eye opening at the specied

sampling time shows the noise margin or immunity

to noise. Noise margin is derived rom the

ollowing equation:

Noise margin (pe rcent) = X 1 00 pe rcentV 1

V 2

where V1 is equal to peak signal voltage and V2 is

equal to the ideal signal voltage

Maximum

signal voltage

(V2)

Best sampling time

Noise margin (V1)

Time interval duringwhich signal can

be sampled

Threshold

Distortion at

sampling times

Distortion at

zero crossing

(T)

Slope increases

sensitivity to

timing errors

Figure 16.

Optical power expressed in W or dBm, and optical at-

tenuation is expressed in dB/m. The total loss budget

is derived rom the individual losses contributed by at-

tenuation and other possible losses in the link.

The Optical Power Margin is

M = (T - R) - (OI + CC).

Optical Power Margin = Optical Power Ratio - SystemLosses.

Note:

For the ber-optic cable to transmit and receive, the fux margin

must be positive.


12/56

1

Timing jitter is caused by noise in the receiver

and pulse dispersion in the ber. I the signal is

measured at its midpoint, then this time-width

indicates the amount o jitter

The lower the slope o the signals transition time, the

more that the eye closes and the greater the

increase in timing errors. As the slope approaches

the horizontal, the possibility o error increases

I a truly random data stream passes through a

linear system, all eye openings will be identical

and symmetrical. Any non-linearities will produce

asymmetry in the eye pattern

Parameters such as rise- and all-times, extinction

ratio, overshoot, undershoot, and pulse-width distortion

can be observed.

When an oscilloscope is triggered by the clock, the

result is the classical eye pattern, as shown on the

preceding page. Data are recoverable i the eye is

open above and below the threshold or a period otime greater than the set-up and hold time needed

or subsequent decision devices or circuits.

Most serial communications standards now dene

a minimum condition on eye-pattern opening

(template) or the serial data stream. The area rom which

the waveorm is excluded is the mask.

The mask limits that speciy the eye-pattern opening

are an improvement over the older technique o speci-

ying pulse parameters individually. It is compact, con-

veying important inormation without placing undue

restrictions on the data pulse. Limiting rise and all times,

pulsewidth distortion, overshoot, damping, undershoot,

and so orth, is unwieldy and insucient by comparison.

Mask region 1

1.0

0.0

RelativeAmplitude

AbsoluteAmplitude(V)

Mask region 2

0.10.0 Relative Time (Unit Interval)

Absolute Time (s)

Mask region 3

Figure 17.

As shown by in Figure 17, a mask consists o two parts:

1. dened regions on the oscilloscope screen where the

waveorm is not permitted to appear

2. denitions o the time and amplitude scales

TTL and PECL

ECL (Emitter Coupled Logic) is a non-saturating digi-

tal logic, characterized by high speed, low noise, and

the ability to drive low-impedance circuits. ECL uses a

negative power supply. PECL (Positive Emitter-Coupled

Logic) is the standard ECL device powered by a positive

power supply.

A typical application or ECL (Emitter Coupled Logic) is a

subsystem that requires very ast data transmission, such

as ber-optic links. In this application, parallel data is se-

In Figure 18, the high-speed serial data rom the SerDes

(serializer-deserializer) IC is PECL-compatible with the

Optical Transceiver. The low-speed Signal Detect output

rom the Optical Transceiver is PECL. The Host Protocol

IC, which uses signal detection, operates with TTL

logic levels. It is converted rom PECL to TTL using the

Converter IC. The parallel data interace between the

Host Protocol IC and the SerDes IC is TTL.

Fiber-Optic Cable

TTL

HostProtocol

IC

SerDes Optical

TransceiverPECL

Convertor SignalDetect

PECLTTL

Figure 18. Simplied Gigabit Ethernet Network Interace Card

rialized into a highbandwidth PECL data stream. At the

other end o the link, the signal is converted back to rela-

tively low-speed parallel data or urther processing. This

system o translating parallel to serial then serial to

parallel minimizes the number o transmission lines

required or interconnecting the subsystems.

The gigabit rate perormance o current ber-optic link

designs demands the use o PECL circuits.


13/56

1

and receiver and sometimes it is instead the responsibili-

ty o other equipment designer and the integrated circuit

vendors. The level o integration is dierent, depend-

ing on actors that will be discussed later.

LED or

Laser Optical FiberPhotodiode

ReceiveCircuitry

DriveCircuitry

Figure 19.

Data Communication Links

A data communication link transers inormation rom

one piece o electronic equipment to another. These links

interconnect computers, peripherals, servers, and other

individual devices. Busses inside these devices distrib-

ute the parallel data between the internal components

(memory chips, the micro-processor, logic devices, and

so orth). To send this parallel data across a datacommu-nication link, it must rst be transormed into a serial bit

stream.

To do this, the transmitter, receiver and the support-

ing ICs must do more than merely convert current into

light. Sometimes the manuacturer, or example, Avago,

builds additional circuitry into the ber-optic transmitter

A number o steps are needed to prepare this serial bit

stream rom the parallel word data and transmit it.

SerialSignal

Processing

Drive

Circuitry

Tx

Fiber Optic Cable

Rx

Receive

Circuitry

Clock

andData

Recovery

Data

Bus

Data

Bus

Software

Bus Interface

Circuitry and

Buffering/Framing

Encoding

Circuitry

Multiplexor

Demultiplexor

Decoding

Circuitry

Reassembly/

Buffering

Software

Bus Interface

Circuitry and

Network Interface CardNetwork Interface Card

Figure 20. Block Diagram o a Generic Fiber-Optic Data Communication Link

Table 3.

Block DescriptionBus Interace Interaces directly with the data bus, recognizes requests to send data, pulls the data of the bus, and maps the data rom

the bus ormat over to the data communication link ormat. When the data link is receiving data, the bus interace noties

the local bus that inormation has been received and is going to be put on the bus. It then maps the data rom the data link

ormat to the specic bus ormat.

Bufering / Framing Stores the parallel data rom the bus in local memory. Partitions this large block o data into rames (small blocks); overhead

(additional rames) may be inserted. These overhead rames include address error tolerance, and LAN management inorma-

tion. All rames are then sent sequentially to the transmitter-encoding circuitry. When receiving data, this procedure is

reversed and the original data is restored and placed in local memory. It remains here until the bus interace circuitry places

it on the resident data bus.

Encoding / De-coding Converts the data rom an original value to a prescribed value, which when serialized, ensures that the link will unction

properly. Since this conversion is prescribed, the receiver side o a data communication link can convert the received signal

back into the datas original orm.

Multiplexor/ Demultiplexor Serializes the parallel data (a data communication link can transmit only one data bit at a time along the same cable). The

serial bit stream goes to the Serial Signal Processing circuitry. On the receiver side o the link--once the serial bit stream has

been recovered--this data is converted back into parallel data words by the demultiplexor (DeMux).

Serial Signal Processing Changes the digital serial bit stream, i necessary, into a orm appropriate or the transmission medium.

Drive Circuitry Drives the transmitter to send the serial signal, which is usually amplied. The drive circuitry may convert an input voltage

signal to a current drive signal, which is sent to an LED or laser diode (both o which are current driven).

Transmitter Converts the electrical signal into a orm appropriate or propagation through the transmission medium. For ber optics, the

signal is converted into light, which is launched into the ber-optic cable.


14/56

1

Block Description

Transmission Medium Carries the signal rom the transmitter to the receiver. The medium can be twisted-wire pair (TWP), coaxial cable (coax),

ber-optic cable (glass or plastic) or wireless (does not require a medium).

Receiver Detects the incoming signal and converts it to an electrical signal, which can be processed by the receive cir-cuitry.

Receive Circuitry Processes the signal rom the receiver into a more useable orm. Frequently the signal is converted rom current to voltage

(by a trans-impedance amplier) or amplied (by xed and variable gain ampliers) or wave-shaped (by an equalizer

or copper cable transmission or quantizer or some ber-optics transmission) or can undergo any combination o these

procedures.

Clock and Data Recovery Reconverts the received and processed serial signal into the original serial bit stream. For this to happen, the clock and data

recovery must rst lock onto the requency and phase o the incoming signal and extract the clock (by a phase-locked loop,

or some other method). This clock is then used to check the incoming signal at every clock period. In this way, the original

digital serial bit stream is recovered along with the clock (data rate at which the serial bit stream was sent).

Table 3. Cont.

Encoding

Encoding allows the data transmission link to be more

ecient. Encoding merges the data and clock signals ina manner that allows a timing recovery circuit to recon-

struct the clock and data signals at the receiver end o

the data link. Without encoding, a clock signal would

have to be sent via a second ber link. Increased cost and

possible timing skew problems could arise i an addition-

al clock link were to be established. A number o encod-

ing schemes are shown in the ollowing table.

The basic unit o inormation in a digital system is the bit:

a logical one or zero. A logical one is represented by

a high voltage; a logical zero by a low voltage. In binary

signaling, a symbol is either a high o low signal level

that is held or a period o time. When a string o identical

Type Description Mb/s Mbd MHz Efciency

Manchester Replaces each bit with symbols 100 00 100.0 Less

B5B* Replaces -bit group with 5 symbols 100 15 .5 More

5BB Replaces 5-bit group with symbols 100 10 0.0 More

*Note:

Another type is 8B10B, which replaces 8 bits with 10 symbols. The relationship between Mb/s, Mbd, and MHz remains the same or this encoding

as well.

bits is transmitted, the voltage remains at the same

level. Unencoded data has DC voltage osets becausethis data can remain in the logic 0 or logic 1 state or

an indenite period. Digital systems use a clock to dene

the bit period in a constant-voltage transmission.

Data rates are oten expressed in bit per second. The

terms, megabits per second (Mb/s), megahertz (MHz),

and megabaud (Mbd) are related as ollows: the trans-

lation rom bits per second to baud depends on the

encoding scheme. I there is no encoding, bits per second

equals baud. Baud is expressed in symbols/second. In

binary transmission systems, the maximum undamental

requency (Hertz) o the data is hal the baud rate.

At the transmitter end o the link, serial binary data

is coded to a signal suitable or transmission. At the re-

ceiver end o the link the digital signal is demodulated bya phase-lock loop (PLL), a circuit that synchronizes a local

oscillator with the incoming signal. A PLL has three basic

elements: a phase detector (PD), a loop lter (LF), and

a voltage-controlled oscillator (VCO), as shown in the

ollowing circuit diagram, Figure 21.

Loop

Filter

Input

Signal

VCO

PhaseDetector

Figure 21.


15/56

15

The phase detector compares the phase o a periodic in-

put signal to the phase o the VCO. The voltage output o

the PD is the dierence in phase between the two inputs.

This voltage is ltered by the LF and applied to the VCO.

Control voltage on the VCO changes the requency to

reduce the phase dierence between the input signal

and the VCO. When the loop is locked, the VCO is exactly

equal to the average requency o the input signal. Eachcycle o input has only one cycle o oscillator output.

The number o symbols per second is expressed in

baud. Although bit rate and baud may coincide, they

are not the same. The astest baud is dened as the re-

ciprocal o the narrowest pulse width. Data rate is infu-

enced by ormat, which is the method o encoding bits

or transmission. Baud is a true measure o a systems

signaling speed because it is not dependent on the data

ormat. Table 4 describes the various modulation codesused in data transmission.

Table 4. Standard Modulation Codes

Code Description

NRZ A logic-level code or serialized data. A zero is a low logic level and a one is a high logic level that does not return to zero between succes-

sive ones.

RZ A code in which ones are represented by a high logic level that returns to zero between successive ones. For this reason, the code requires

a channel that can signal two symbols (high and low) per bit. Its main advantage is that it conserves power because energy is needed only

or transmitting ones.

NRZI A code in which a zero is represented by an edge and a one is represented by the absence o an edge. Because the edges may be either

rising or alling, inversion o the signal does not change the meaning o the code and it is thereore invertible.

Biphase-Mark A code requiring a channel capability o two symbols per bit. Each bit cell begins with an edge: then, or a one an additional edge occurs

during the bit cell; or a zero there is no edge during the bit cell. Because there is always an edge at the beginning o the bit cell, clock

recovery rom this code is possible by either phase-lock or one-shot techniques. It is also invertible and has a duty actor exactly 50%.

Biphase-Space An invertible, sel-clocking code, difering rom biphase-mark only in that zeroes have the additional edge during the bit cell. Both biphase

codes are sometimes called / because the additional mid-cell edge may be regarded as a cycle o a requency that is twice that or the

absence o the mid-cell edge. Duty actor = 50%.

Manchester A code in which a one has a high logic level at the beginning o the bit cell with high-to-low transitions at mid-cell; a zero starts at a low

level with low-to-high transition at mid-cell. This code also requires a channel capability o two symbols per bit. Because there is a mid-cell

transition in each bit, clock recovery is possible by either phase-lock or one-shot techniques. It is not invertible; the duty actor is 50%.

Miller An edge-type code in which each one in the serial data is encoded as a mid-cell edge. The zeroes either have no edge or are encoded as

edges at the beginning o the bit cell. Because no edges occur at intervals less than one bit time, the channel capability required is only one

symbol per bit. With edges occurring regularly at integral multiples o hal o a bit time, clock recovery is possible with a phase-locked oscil-lator running at a requency that is twice the data rate. The duty actor is generally 50%.

MFM* A code similar to Miller: impulses occur at intervals o 1.0, 1.5, or .0 bit times. For this reason, clock recovery is possible. Because impulse

duration is much less than bit time, the duty actor approaches zero but cannot exceed 50%.

*Modied Frequency Modulation

Electromagnetic CompatibilityEMC

EMC reers to the capability o electronic equipment or

systems to be operated in the intended operational elec-

tromagnetic environment at their designed levels o e-

ciency. Specically, EMC describes how the product

behaves in terms o radiated emissions, commonlyknown as electromagnetic intererence (EMI), and also

how the products perormance is aected by immuni-

ty (sus-ceptibility to radiated energy), electrostatic dis-

charge (ESD), and conducted power supply noise.


16/56

1

1. EMI

The rst EMC area is radiated emissions, sometimes called

electromagnetic intererence or EMI. A radio or TV

broadcast is an example o intentionally radiated electro-

magnetic energy. A computing device also radiates elec-

tromagnetic emissions, although it is not intended to

(the radiation is an inherent by-product o the switching

currents fowing in its conductors). Electromagnetic in-tererence (EMI) describes the eect o unwanted radia-

tion interering with another (intentional or unintention-

al receiver) circuits operation. Electromagnetic radiation

occurs when a changing current fows in a conductor.

The area near the conductor (antenna) is where either

the electric or the magnetic eld dominates the total

radiated eld.

Government agencies around the world regulate

the amount o radiated electromagnetic energy emitted

by various sources. Their intent is to allow any purpose-

ly transmitted radiated energy to be received without

being interered with by some other radiation source ataround the same requency.

Most manuacturers want their systems to meet all o

the home environment radiated emissions specica-

tions used around the world so that they can be sold in

the US, in Europe or in Japan, without restrictions. Gov-

ernment agencies usually set their radiated emissions

regulations to distinguish between two types o appli-

cations. One is actory or oce (Class A) where a higher

level o radiation can be tolerated; the other is the

home (Class B) where there are more TVs and radios and

thereore less electromagnetic radiation can be tolerated.

Any product that does not meet the pertinent emis-sions requirement or a particular country cannot be

legally sold in that country. Any manuacturer who

sells equipment that is ound to violate the regula-

tions can, i caught, ace large nes and penalties. Also

the manuacturer may be orced to withdraw the prod-

uct acturer may be orced to withdraw the product

rom the market until the regulating government

agency, or example, the FCC (Federal Communications

Commission) in the United States is convinced that the

radiated emissions o the modied product meet the

required specication limits.

System design can aect radiated emissions in three ways.

The rst way is in the choice o circuit components. Some

generate more high-requency energy than others do.

The second way relates to the eect o the antennas (that

is, the circuit interconnections) that the high-requency

energy sources are connected to. The third way is by the

shielding that the chassis box provides and by the cable

shields that eectively reduce the amount o radiation

that the antennas leak to the outside world.

An actual communication system contains various

antennas ormed by circuit interconnections and by

the other metal bodies in the circuitry that create the

Vcc-toground- current loops. These antennas are then

driven by various energy sources within the system

circuitry. Keep these antennas as small as possible. I

they are long enough to be treated as transmission

lines, then it is important that the line is terminated

in its characteristic impedance. Antennas can also beormed by long traces that are driven by ground voltage

noise and thereore radiate. Even a ground plane, when

driven by a ground noise source can radiate. Oten,

cables can radiate energy when high-requency currents

fow through them. Because cables are oten the largest

antenna around, they are usually the dominant source

o radiation. Chassis and cable shields can also radiate i

ground noise current or voltage drives them.

Fiber-optic cables do not, however, radiate energy

as wire cables do. Thereore, ber-optic cables can

help reduce radiated emissions i using wire cables is a

problem. At FDDI and ATM data rates, the encoding/ de-coding schemes, and the consequent additional circuit-

ry needed to reduce the bandwidth and the emissions

on a twisted pair wire cable, are not needed i a ber-op-

tic cable is used.

I an antenna is completely enclosed in a suciently thick

metal box, then no radiation will escape. I the box has

an aperture, then some radiation will escape through it;

the bigger the opening and the higher the requency, the

more radiation that escapes.

More accurately stated, the leakage through a single ap-

erture in a metallic chassis is related to the longest di-

mension o the opening (d) and the wavelength () o theradiating requency. When wavelength is less than or

equal to twice the longest aperture dimension, the atten-

uation o the aperture is 0 dB. The cuto requency o a

solitary aperture occurs when = 2d. At requencies less

than cuto, the attenuation o the aperture increases at

a linear rate o 20 dB per decade. I the metal chassis box

has multiple openings with identical longest dimensions

d, then the attenuation o the shielding decreases by 10

log (N), where N equals the number o apertures. The

preceding relationship holds true until N becomes large

enough to reduce shield attenuation to 0 dB.

2. Electromagnetic Susceptibility

Electromagnetic susceptibility (or immunity) o a prod-

uct is dened as the eect o external electromagnet-

ic elds on the perormance o that product. The per-

ormance is measured in the presence o an external

electromagnetic eld relative to the perormance with

the electromagnetic eld absent. Immunity and suscep-

tibility reer to the same characteristics (immunity is the

inverse o susceptibility).


17/56

17

What is meant by immune, however, is let to be nego-

tiated between vendor and customer. Based on some

o Avagos customer inputs, Avago has standardized on

a 10 V/m eld strength to test ber-optic transceivers.

Usually an antenna in the circuit will pick up the ex-

ternal eld and then couple it into a critical circuit

node where it appears as a noise signal. The suscepti-

bility will depend on the requency o the eld becausethe receiving antenna gain varies with requency and

because the circuit noise rejection varies with requency.

The noise can be picked up directly at the sensitive node

itsel or can be conducted rom another node (such as

Vcc) that has a larger gain antenna connected to it.

The usual eect o an external eld on a ber-optic

receiver is to degrade the received bit error rate. The

beroptic receiver (Rx) is usually the most sensitive ana-

log circuit in the entire communication network product.

An external eld can induce a signal on the antennas

ormed by the interconnections in the ber-optic Rx

circuitry. Antennas are bi-directional devices. The samephenomenon that causes an antenna to radiate an

electromagnetic eld when a voltage/current signal is

applied to its inputs, will generate the same voltage/

current signal at those inputs, (now really the outputs),

i the same antenna is placed within an identical, but ex-

ternally generated, electromagnetic eld. The antenna,

ormed by the Rx circuit inter-connections, picks up the

external eld and generates a signal. I the generated

noise signal is conducted to a sensitive circuit node, the

node then experiences a lower signal-to-noise ratio,

which increases the bit error rate. (In a perect ber-optic

Rx, the signal-to-noise ratio and the bit error rate are di-

rectly related). So the external electromagnet-ic eld caninject noise into the ber-optic Rx and thus degrade the

bit-error rate (BER).

3. Electrostatic Discharge

Certain non-conductive materials can either do-

nate charge (electrons) or acquire charge when in con-

tact with other materials. A material with a net charge

can then transer it to a conductive material either

by direct contact or by inducing the opposite charge

in the conductor. I this charged conductor contacts

an earth ground (or any conductive body with a very

large amount o stored charged available), a current

will fow until that conductors net charge becomeszero. For example, i your skin is charged rom walking

across the carpet on a cold dry day and you then touch a

grounded or large conductor, you may see a spark as your

skin discharges, and eel a tingle in your nger, as the cur-

rent fows. The characteristics o the ESD current depend

on the amount o charge stored and on the impedance o

the circuit that discharges it (to ground).

ESD can aect a product during its manuacturing

or during its operating lietime. I all the ESD events

are controlled during manuacturing so that they

lie saely within the limits o the most ESD-sensitive

components, then the system can be manuactured

without ESD damage.

In a nished system, ESD can still cause permanent prod-

uct damage. Oten, however, a more important problem

is ESD disturbances to the system perormance. An ESD

perormance disturbance could include lines on a CRTdisplay, logic stuck in a locked state, or a larger number

o bit errors than usual. These ESD disturbances can be

accounted or in the system design to ensure that the

products end user does not notice any drastic peror-

mance dierences when an ESD occurs.

I the components are enclosed inside a conductive

or static-dissipating chassis box in the end product,

ESD is more likely to go to the conductive box than to

some other non-conductive component. For example,

i a plastic-nose, ber-optic module is protruding rom

the chassis box, then ESD is more likely to be conducted

to the section o the chassis box that is near the mod-ule nose than it is to be conducted to the insulating trans-

ceivers plastic-nose itsel. I the chassis grounding is such

that the ESD currents to ground fow on the chassis and

do not fow inside the PCB component grounds, then ESD

damage or ESD problems due to radiated elds caused

by the ESD pulse will be reduced.

4. Conducted Noise

The ourth EMC area is conducted noise. Ideally a conduc-

tor will carry only the desired signal. Practically, however,

there is always some component o the actual signal on

the conductor that is undesirable. This component isdened as noise. Conducted noise emanates rom one

section o the products circuitry and is conducted to the

section o the circuitry being observed.

There are three main components o conducted noise:

1. the conducted noise generator

2. the path that the noise takes to conduct rom

the generator circuit to the receiving circuit

3. the sensitivity o the receiving circuit to this noise

So, conducted noise problems could be eliminated

by eliminating the noise source, removing the conduc-

tive path, or by using circuitry that is insensitive to theeects o conductive noise.

Since an actual ber-optic communication network

has many dierent types o circuitry, all operating at once,

it is important that potential conducted noise problems

be minimized to allow all the circuitry to operate

without any section o it being adversely aected. In

addition, most computing products have limits on how

much conducted noise they are allowed to generate on

the 120 Vac power lines to prevent them rom disturbing

other devices connected to the same AC power line.


18/56

18

The demand or data and speed on the Internet has

been increasing at an unprecedented rate. Avago Tech-

nologies satises this demand by producing ber-optic

components that transorm electrical data into packets

o light or high-speed transport across the opticalnetwork. This in turn helps Avagos customers acceler-

ate its evolution. Now, as never beore, ber is every-

where!

The optical network is composed o countless special-

ized networks. These networks enable groups o com-

puters to communicate with each other, allowing users

to share les and peripherals, send e-mail messages,

and access remote servers.

The optical network provides a high-speed transmis-

sion path or the Internet and telecommunications

networks. Avagos ber-optic components can be ound

in the equipment that is used in the ollowing networks:

Enterprise (or Premise)--Local Area Networks

provide connectivity within a building or campus.

Metro/Access--Metropolitan Area Networks

provide connectivity beyond the Enterprise, linking

residential areas and LANs throughout a metropolitanarea and providing access to Long-Haul Networks.

Long Haul--Wide Area Networks provide

connectivity or metropolitan networks throughout

the world.

SAN--Storage Area Networks link disk

subsystems directly to one or more servers via a

high-speed connection, providing high-volume data

storage services.

Specialized equipment, many using Avagos high

quality, ber-optic components, controls the fow o in-

ormation across the optical network.

Table 5.

Equipment Description

Clients Computers that request to use les, send print jobs, and request other shared resources on the network. Client sotware (part o

the network operating system) contains a redirector that captures requests rom the PC application program (such as Microsot

Word) and routes them out o the PC and across the network. (Also reerred to as nodes or users.)

Backbone A high-speed, ber-optic network that connects smaller LANs.

NIC

(Network Interace Card )

Card that connects computers or peripherals to the network via a cable. Inside the computer, NICs move data to and rom the

random access memory (RAM); outside the computer, NICs control the ow o data in and out o the network cable system. (Also

known as LAN Adapter Cards.)

Server A computer or device on a network that manages network resources. For example, a le server is a computer and storage devicededicated to storing les. A print server is a computer that manages one or more printers; a network server is a computer that

manages network trac. A database server is a computer system that proc-esses database queries.

Hub* Stand-alone equipment with up to physical interaces that connect the nodes to the server. An active hub regenerates the

signal using a repeater. Hubs unction as junction boxes and can also be stacked. (Also called repeater or concentrator.)

Switch A device in a network that lters and orwards packets between LAN segments. Switches operate at the data-link layer (layer ) o

the OSI (Open System Interconnection) Reerence Model and thereore, support any packet protocol. LANs that use switches to join

segments are called switched LANs or, in the case o Ethernet networks, switched Ethernet LANs.

Switch/Router Switch/Routers search inside the envelope surrounding the data to determine the data packets destination, read the inormation

(some routers even have error correction), and send the packet o inormation to the appropriate location. The Switch/Router at

the receiving end repackages the data into a packet or rame appropriate or its LAN segment. A router may also unction as a

rewall, by preventing unauthorized users rom gaining access to the network.

DWDM Transmission

Equipment

Dense-Wavelength-Division Multiplexing combines numerous wavelengths o light (each one carrying diferent inormation) and

sends them simultaneously along a single ber-optic cable. The wavelengths are separated (demultiplexed) at the other end o

the optical link.

SONET/SDH Transmission

Equipment

Large, complex optical equipment that transmits high-volume, high-speed data using SONET/SDH protocols rom point to point in

the Metro/Access and Long Haul Networks.

*An active hub or concentrator contains a management module that can answer queries rom a management console.

Networking Technology


19/56

1

An Enterprise (or premise) local-area network (LAN)

is a system within a building or campus that connects

the widely separated computer resources o a single

organization. Within a building, each node (personal

computer, workstation, server, or other equipment) at-

tached to the LAN is able to share data with any other

node on the network. As a result, workers can use the

LAN to share devices, such as laser printers, send e-mail,and share les and applications. They can also share com-

mon connections outside the Enterprise, as or example,

the Internet.

Figure 22.

Oten, one large LAN (one with many nodes--users)

may be segmented into smaller LAN subnets within or

between buildings and connected by a Switch/Router. In

a building, a network is composed o various segments

o dierent lengths, transmitting data at speeds deter-

mined by user demand. Avago sells components or

equipment used in each o the various optical segments

in the corporate LAN environment.

Figure 23.

A number o standards have been established or prod-ucts that link dierent computers and network equip-

ment rom various vendors; the most common stan-

dard is Ethernet.

Avagos ber-optic components are an integral part

o the blades (cards) that orm Routers or Switches. Fiber-

optic transceivers convert digital electrical signals to light

pulses and light pulses to digital electrical signals or

transmission on the optical link. The Switch Fabric inside

the equipment manages the proper interconnections to

route the signals to the proper port so that they reach the

intended destination.

Ethernet standards speciy dierent distances and

data rates or connections in the LAN. Avago sells a

large variety o ber-optic components or each o these

applications. Small orm actor dimensions allow a high-

er density o transceivers to be placed on a blade.

Figure 23.

Metro/Access Networks provide connectivity beyond

the Enterprise, linking residential areas and LANsthroughout a metropolitan area and provide access to

Long Haul Networks. In Metro/Access and Long Haul

Networks, large complex optical systems transmit high-

volume, high-speed data over ber-optic cables.

Long Haul Networks link multiple Metro/Access

networks, providing the ability to transmit data around

the world. Compared to LAN data rates, which transmit

at a maximum o 1 Gigabit per second, data rates in

the Long Haul Network typically exceed 10 Gigabits

per second. Voice and data rom within Metro/Access

Networks are combined at these higher rates to provide

an economical transmission path between cities and con-

tinents.

As data moves rom the Enterprise to the Metro/ Ac-

cess and then to Long Haul Networks, it is continually

combined and transmitted at higher data rates. Com-

mon data rates in Metro/Access and Long Haul Net-

works include:

155 Mbps (OC-3)

622 Mbps (OC-12)

2.5 Gbps (OC-48)

10 Gbps (OC-192)

The higher data rates can be aggregates o multiple lowerspeeds. For example, our OC-48 equals one OC- 192.

These data rates are dened by SONET (Synchro-

nous Optical Network), which is one o the main trans-

mission standards used in Metro/Access and Long

Haul networks. ATM (Asynchronous Transer mode) is a

common method or packaging data or transmission

over SONET.


20/56

0

Within homes and businesses, the demand or exchang-

ing increasing amounts o data with remote locations

is driving the need or higher bandwidth in Metro/Ac-

cess Networks. Links close to the Enterprise and res-

idential areas are typically 155 Mbps (OC-3) and are

dominated by single-mode ber connections. Avago is a

leading supplier o single-mode, ber-optic transceivers

or these links.Avago also provides single-mode, ber-optic trans-

ceivers or SONET transmission within Metro/Access

Networks. Links are typically 2.5 Gbps (OC-48).

Within the Long Haul network, DWDM (Dense Wave-

length- Division Multiplexing) enables carriers to

increase transport capacity by using existing ber-

optic cable. DWDM uses dierent wavelengths to com-

bine multiple SONET channels onto one ber.

The primary value o this technology is the costeec-

tive transmission o high-aggregate bit rates over long

distances on a single ber. Avago provides the ber-

optic transceivers or the links between the SONET

terminals and DWDM transmission equipment, specical-

ly or the line cards (blades) used in SONET transmission

equipment.

The SONET standard species dierent distancesand data rates or application in Metro/Access and

Long Haul Networks. Avago provides a large variety

o ber- optic transceivers or OC-3, OC-12, and OC-48

applications. Small-orm-actor transceivers are in these

markets; their smaller dimensions allow a higher density

o transceivers to be placed on a line card.

Note: white dots show where Avagos components are used.

Figure 25.


21/56

1

Table 6. Avago Local Area Network Components

Standard

Speed

(Mbps)

Maximum

Segment Length Product Types

Ethernet 80. 10 km MMF Optical Transmitters

Optical Receivers

Fast

Ethernet

80.u 100 km MMF Standard 1x and x5 SFF MT-RJ SFF

Optical TransceiversGigabit

Ethernet

80.z 1,000 550 m MMF

10 km SMF

Standard x5 SFF MT-RJ & SFP Pluggable LC

Optical Transceivers

Optical GBICs

SerDes ICs

Table 7. Avago Local Area Network Components Features and Benets

Features Benets to Customers

High-volume, industry-standard, ber-optic transceivers up to 1 Gbps Assurance o supply

Application engineering support Interoperability with supporting ICsReerence Designs Reduced time to market

Ease o Design

High Reliability Reduces downtime or networking equipment end users

Continually introduce products with higher density and greater bandwidth Provides or continual evolution o the Internet

Complete product line o LAN standards and multiple

orm actors, including ber-optic transceivers and SerDes ICs

Simplies supply chain

Table 8. Avagos Metro/Access and Long Haul Components

Standard Speed (Mbps) Maximum Segment Length Network Products

SR IR LR Type Footprints

SONET OC- 155 km

MMF

15 km

SMF

0 km

SMF

Metro/Access Standard & SFF

Optical Transceivers

1x, SFF MT-RJ

OC-1 Metro/Access 1x, SFF MT-RJ & LC

OC-8 ,500 Long Haul 1x, x, SFF MT-RJ & LC

OC-1 10,000 Proprietary Long Haul Being

Investigated


22/56

LAN Architecture Evolution

At rst, all LANs were shared, allowing only one trans-

mission at a time. In this arrangement, each device waits

or access to the LAN beore transmitting. Bandwidth is

shared among the nodes.

Hub HubBridge

Hub RouterEthernet

Ethernet Ethernet

TokenRing

Figure 26.

When do you use a bridge and when do you use a router?

As shown in Figure 26, i two dierent networks are to

be joined, use a router. I two similar networks are to be

joined use a bridge.

As the number o nodes increases, the bandwidth or

each node decreases. To alleviate the problem o shared

LANs, bridges and routers were created to segment

trac into smaller LANs. This permitted ewer devices

to be connected to each hub, in eect, increasing the

bandwidth on it.

Table 9. Avagos Metro/Access and Long Haul Components Features and Benets

Features Benets to Customers

High-volume, on 155 Mbps (OC-), Mbps (OC-1), and .5 Gbps (OC-8)

industry- standard, ber-optic transceivers

Assurance o supply

Satises demand or single-mode, ber-optic transceivers

Application engineering support Interoperability with supporting ICs

SONET compliance

Reerence Designs Reduced time to market

Ease o Design

High Reliability Reduces downtime or networking equipment end users

Telecordia (ormerly Bellcore) compliant

Continually introduce products with higher density and greater bandwidth Provides or continual evolution o the Internet

Multiple orm-actor, ber-optic transceivers or Metro/Access standards Simplies supply chain


23/56

Access MethodsAccess methods determine how data shares a com-

mon network path.

Carrier Sense Multiple Access with Collision Detect (CSMA/CD)

In this method, a nodes NIC can continually sense

the trac on the network. When the node has a rame

to send, the NIC checks to see i the network path is

clear. I it is, it sends the data. I data collides because

two nodes send at the same time, each node will sense

it, stop sending and wait or a random time beore trying

to put its data on the network again. Collisions repre-

sent only about 1 to 10% o active time in networks. Equal

access is guaranteed only on a probabalistic basis. Ether-

net uses CSMA/CD.

Token passing

Token Ring networks and FDDI networks use token pass-

ing. A node can transmit on the network only when it has

the token. The token travels around the ring and stops at

each node to see i it has anything to send. When a node

is ready to send data, it marks the token rame as busy,

attaches its data and destination inormation, and passes

the token on. The token continues to travel around the

network until it reaches its destination. At this point the

recipient removes the inormation and indicates on the

token rame that the inormation has been received. The

token returns to the originator where conrmation is ac-

knowledged, the token is then released as un-busy.

Tab le 11.

Uses CSMA Token Passing

Light/moderate trac x

Consistent heavy trac x

Oce Applications x

Medical Imaging x

Monitoring x

Cheap, easy exible x

Demand Priority

Demand Priority is a deterministic access method sim-

ilar to token ring passing. The hub successively scans

the ports to determine which one is ready to transmit

data. This avoids the collisions that can occur with CSMA/

CD.

Token Ring requires a manager or active monitor or the

ring. The active nodes negotiate, using their serial num-

bers, to determine which node is the active monitor. The

monitor initiates the rst ree token and does gener-

al ring operation maintenance. This method guaran-

tees equal access to the network.

A comparison between CSMA/CD and token passing

are shown in Table 11.

Table 10 compares routers and switches.

Table 10.

Routers Support Switches Support

- Sotware-based

route determination

and packet processing

- Network

node-address resolution

- Trac isolation and ltering

- Network management

- Protocol conversion and

mismatch resolution

- Hardware processing (provides

higher throughput)

- Full-duplex switched links

between NICs (reduces latency)

- Bandwidth management

- Lower cost

The problem with lack o bandwidth led to the devel-

opment o switches. A switch is basically a multi-port

bridge. Switched LANs allow a virtual dedicated point-to-

point connection between multiple devices on the LAN,

allowing access to ull bandwidth or each connection

(similar to a phone system experiencing many simultane-

ous phone calls). With a switch, each node has access to

the ull bandwidth. Switches work best in a uniorm tra-c fow, or example, 10 to 20 desktop computers com-

municating with each other, as opposed to one device

receiving most o the trac.

In ull duplex switching, each node has equal bandwidth

both transmitting and receiving, which can be exercised

simultaneously, eectively doubling the bandwidth re-

quirement. The node can talk and listen at the same

time as in a telephone call. In this situation, special NICs,

hubs and a high through-put backplane are required.


24/56

Ethernet

Overview

Ethernet, an established standard since the mid 1980s, is

the most widely used local-area network (LAN) transport

protocol in the world today. Table 12 summarizes itsdata rates (the structure and protocol are identical or all

three).

Table 12.

Data Rate

Standard Mbps MBd

Ethernet 10 0

Fast Ethernet 100 15

Gigabit Ethernet 1000 150

Ethernet is a shared bus network with a CSMA/

CD (Carrier-Sensing Multiple Access with Collision De-

tect) access scheme. The encoding scheme or 10 Mbps

Ethernet is Manchester; Fast Ethernet uses 4B/5B, and

Gigabit Ethernet uses a 8B/10B enco

Fiber Optic Training Maula

Documents

Transcript of Fiber Optic Training Maula