E&CE 497 - Introduction

These are the following concerns in fiber optics:

1) Generating a pulsed source of light [LED, laser].

2) Carrying the light down a waveguide [SM (singlemode), MM (multimode) fibers]

3) Receiving and demodulating the signal with a detector.

We are mainly concerning ourselves with section 2), the transmission of light down the fiber. Youwill need to understand how all items effect a complete fiberoptic communications system.

Light Sources

Light sources have many properties. The most important is the bandwidth of the source. A whitelight, such as a halegon lamp, radiates light over a large bandwidth ie approx. 300 nm -> 2500 nm.Light sources such as this often can not be switched much faster than 100 Hz and even though theycan be quite powerful (10W .. 100W) very little of the light is conducted by a fiber because a fiberwill only carry a small fraction of the total bandwidth as well as a small fraction of the many modesin which the light is emitted. Typically, from a 10 W source, only a few microwatts will betransmitted into and thru a fiber.

LED sources have a much smaller bandwidth (approx 50 nm), can not be switched very quickly(typically much less than 50 MHz), are inherently multi-mode and radiate noncoherent light in abroad cone. Because LEDs generate multiple modes of light and only a small fraction of these areactually carried by the fiber; a 10 mW LED may only inject 0.1 mW of light into a fiber! Howeverthis is far superior to a halegon light system like we use in the lab.

There are many different types of lasers (DFB, ILD, ...) but they are generally much more powerfulthan LEDs (100 mW vs < 1mW for LEDs). Lasers tend to have a narrow bandwidth ( < 5nm) andinject a good bit of their power into the fiber because they tend to generate only a single mode ina small cone (ie the light is tightly focused). Lasers can also be switched quickly (1 GHz) butrequire complex drive circuitry and often need coolers.

A quick calculation to do is to calculate the frequency of the light in a typical (1300 nm) commu-nications system. If you could modulate the light at 1% of that frequency; what sort of signalbandwidth do you have? How does this compare to a 10 GHz carrier satellite signal?

Fiber Types

There are two main fiber types, singlemode (SM) and multimode (MM). This refers to how many"modes" in which a signal can travel down the fiber. For example, if light of a single wavelengthis fired into a fiberoptic cable it may be able to travel down the cable in several ways. If it can onlytravel by one mode then the fiber is single mode. If the fiber cable was much larger than the light,the light could travel by many modes and the cable would be considered to be multimode.Remember that a single cable can be multimode or singlemode; it all depends upon what wavelengthof light is used! However, cablesare generally designed toact as SM or MM for a certain wavelength.The standard regions for communication are based around 850 nm, 1300 nm and 1500 nm.

Fiber which has been designed to be used as MM fiber can be made in several ways. One problemwith the different modes by which energy can travel thru a waveguide is that the different modestravel at different speeds. One solution to this is to use graded index (GI) fiber. By changing theindex of refraction radially it is possible to reduce much of the signal distortion due to energytraveling in multiple modes. Note that distortion is due to parts of the signal being out of phase

Introduction 1.1 - 1 - February 28, 2001

with each other. Consider that different wavelengths are diffracted by different amounts. Thereforea source that covers a large bandwidth means that the different wavelengths will travel at differentspeeds causing additional distortion.

As you can see by the chart, the change in index of refraction thruout the fiber is very small (< 0.01)so it is impossible to visually tell a MM cable from a SM cable.

Fiber Characteristics

Fiber cables used for communication are made out of glass that has been doped with impurities tocontrol the index of refraction and minimize loss. The glass is quite strong; but very brittle. Anysmall defects or microfractures will quickly cleave a fiber into 2 pieces. To protect the fiber anepoxy coating (buffer layer) is often used. Outside of this is often more layers of woven fiber andwaterproof materials to protect the fiber against the elements, rodents, stress etc. The fiber willeasily break if it is bent very sharply. Often the outer layer is thick enough to prevent this.

In the steady state condition, light travels thru the core of a fiber cable. Any light in the claddingglass layer will quickly (a few meters to a km of distance is required) attenuate to 0. Light can onlyget trapped into the core if it falls with the "angle of acceptance". You can prove this to yourselfby doing a simple calculation. Imagine the end of a fiber and assume that ray optics applies andthat light will be captured if total internal reflection occurs. Calculate the critial angle, below whichall light will be totally internally reflected.

A similar condition is to imagine that someone is trying to inject light into a fiber by shining a laserbeam nearly parallel to the fiber; ie imagine someone trying to get light into a fiber by shining itonto the cladding. At what angle does someone have to shine the light have it captured by the core?Is this even possible? The added complication is that the light is refracted/reflected by the air tocladding interface and then the cladding to core interface. You will find that it is impossible toinject light into a fiber this way. However, if you bend the fiber, it may be possible. One trick isto shave or grind part of the cladding away to get closer to the core.

Introduction 1.1 - 2 - February 28, 2001

Signal Attenuation in fiber and the effect of Over/Underfilling

The fiber loss per km changes with wavelength, the mode in which the energy is traveling and thedesign of the fiber. OH molecules, metals and impurities in the fiber absorb energy and other factorslike Raleigh scattering will result in loss. Another part of the loss is due to the roughness of theinterface between the core and the cladding.

There are two main sources of dispersion in a fiber. One is chromatic or based upon the wavelengthof light. Just like a prism will disperse the light based upon its wavelength, so will a glass fiber.The second is much more complex. The index of refraction is not a constant. It changes withwavelength and it can be changed with the application of an electric field in certain substances!Waveguide dispersion is the dispersion caused by the different modes traveling at different speeds.These two dispersive effects can be used to cancel each other! This is why the communicationsband has moved to the "zero dispersion point" at 1500 nm. In reality "zero dispersion" means thatthe dispersion is only zero within a very small frequency range.

It is important to note that it is not always correct to assume that light is a stream of particles. Thatapproximation can be fairly accurate if you are considering MM fibers carrying light in hundredsof modes but in the case of SM fiber the light is acting more like a wave than a stream of particles.For that reason the way in which light behaves may be non-intutitive.

Light in a fiber can be "weakly" or "strongly" guided. Strongly guided energy is not likely to leakout of the fiber. A weakly guided signal is barely staying within the fiber. Often a small bend willcause the energy to leak out. This can be demonstrated with normal MM cable by injecting visiblelight into the cable. When the cable is bent; light will leak out and the cable around the bend willglow with the light traveling thru the fiber. For this reason weakly guided modes loose a lot moreenergy per kilometer.

The field distribution is where the light energy (or field) is propagating down a cable. Is the field(or energy) in the core or the cladding? Light that is in the cladding will quickly leak out and so inlaboratory experiments you have to make sure that there is enough cable to ensure that the claddingmode energy decays to zero; otherwise that energy will upset the measurements. You can find outmore about this by searching under "fiber filling", "underfilling", "overfilling" and "launchingconditions" in fiberoptic books.

Overfilledlaunchingcondition

Coupling lengthL attenuation = slope

Fiber L ength (m)

Log

(p

ow

er)

(d

B o

r dBm

)

Region of uniformor steady stateattenuation

C

Underfilledlaunchingcondition

Introduction 1.1 - 3 - February 28, 2001

Detectors

Light detectors are inherently broad in their bandwidth. Unlike electronic detectors the gain orsensitivity of an optical detector can not be changed. This means that a communication systemmust be designed to bring a strong signal to the detector, but not a signal which overloads thedetector. The 2 main detector types are PIN diode and APD (avalanch photo diode). Both can bemade from a variety of materials such as Ge, Si.

Communication Systems (some general quantities)

Signal power:LED 100 uW ( -10dBm )ILD 5 mW ( +7dBm )

Cable loss:0.4 db/km (1300 nm)3.0 db/km (850 nm)

Detector sensitivity:Bandwidth (Mbit/sec) Minimum detectable signalupto 10 1 nW (-60 dBm)1000 1 uW (-30 dBm)

In comparison the laboratory measurements which we will be doing will be inthe range:ILD source: -20 to -40 dBmwhite light source: -45 to -90 dBm

Fiber Splices

The single most difficult problem is splicing two fibers together. Any misalignment in distance orangle of the fibers will introduce loss. Similarly any impurities in the splice will introduce scatteringin addition to losses. Since SM fiber cores are < 5 µm the fibers must be aligned with a very smallfraction of 1 µm. Since most connectors do not have the hole for the fiber drilled perfectly in thecenter, rotating the fiber will change the offset!

Typical splice losses are 0.2 dB for good connectors to 1 dB for bad ones. Compare this to the fiberloss per km! Note: At any air/glass interface some of the energy will bounce back due to theimpediance/index of refraction mismatch between the 2 media. Calculate this from the basic opticsequations.

Monochromator

A monochromator is a device which uses a prism or grating to disperse light. With this a whitelight at the source can be split so that only a small, bandwidth limited, fraction of the light comesout. ie the monochromator acts as a filter to select only a small part of the frequency spectrum.The center frequency is easily controlled by rotating the prism or grating and on some mono-chromators the bandwidth is also controllable by changing the size of mechanical slits.

In the laboratory we will be using a monochromator to filter a white light source to achieve awavelength sweep from 800 nm to 1800 nm. The bandwidth is typically 5 nm.

Alignment of the fiber at the exit slit of the monochromator is very important. Do not adjust thisyourself. If you do not have enough power to perform a measurement make sure that your setupis fine and then as the lab. staff to inspect the setup.

Introduction 1.1 - 4 - February 28, 2001

A Brief History of Fiber Optic Systems

Year Fiber Type Wavelength Bandwidth Repeater Spacing(nm) (bit/s) (km)

1981 MM 825 45M 7

1984 SM 1300 565M 30

1987 SM 1300 1.6G 40+

1987 SM 1550 1.8G 80

2000 SM 1550 640G 60-80

Introduction 1.1 - 5 - February 28, 2001

E&CE 497 - Lab #1

ATTENUATION MEASUREMENT

Introduction

The designing of an optical fiber communication system usually consist of a power budget analysiswhich measures the overall system loss (attenuation). A power budget analysis will determinewhether or not a system is loss-limited, that is, if the received power is sufficient to meet the biterror rate specification. Although system performance can also be delay-limited, the performancesof most short haul and LAN systems are limited by the amount of losses in the systems.

The aim of this experiment is to measure the losses in several types of optical fibers and connectors.The two basic types of fiber are singlemode (SM) and multimode (MM).

Sources of Attenuation

1. Source-fiber coupling loss

The number of modes propagating in the fiber is significantly smaller than the number ofmodes generated by the source at the launching end of the fiber, therefore, only a portion oflight can be coupled into the fiber. The amount of loss depends mainly on the launchingcondition, source type and the numerical aperture of the fiber.

2. Connector and Splice losses

Both the connector and splicer are used to connect fibers together. While the splicer gives apermanent joint, the connector is a demountable device used to conveniently disconnect andjoin fibers. The losses are due to misalignment of the fiber cores.

3. Fiber loss

Loss in fibers is due primarily to Rayleigh scattering, absorption and bends. Microscopicnonuniformities and impurities in the glass scatter and absorb light energy. Bends in the fibercause light to escape from the core. Unlike the previous losses, fiber loss is frequencydependent.

It is important to have equilibrium mode distribution, EMD, in the fiber before measuringthe fiber loss. Because cladding and leaky modes will also be excited during launching, themeasured loss without EMD can be above or below the steady-state fiber loss.

4. Fiber-detector coupling loss

Loss during fiber detector coupling are usually assumed to be zero due to the large area ofthe detector.

Lab 1 - 1 - February 28, 2001

Fiber Loss Measurement Method

The most common techniques for measuring fiber loss are the cutback method, the referencefiber and the optical time domain reflectometry (OTDR) method. The cutback method is adestructive method and requires the removal of a length of fiber from the test fiber. Thereference fiber method and OTDR are two non-destructive loss measurement methods. Inthis experiment, the reference fiber method will be used.

In the reference fiber method, one first determines the launching condition and the inputpower level by inserting a short length of reference fiber between the source pigtail and thedetector (figure 1). Next, the output power level is measured by replacing the reference fiberwith the text fiber (figure 2). The fiber loss is then:

Figure 1: Reference Measurement Figure 2: Loss Measurement

Legend

Equipment

1. Optical Signal Generator: The 7700 XR and 7750 XR optical signal generators use LEDsources to provide optical power output signals at 820 nm (35 nm BW) and 1300 nm (40 nmBW), respectively. The output power should be at maximum. Hold the power setting withthe lock ring before taking any measurements. The output is measured in dBm units where:

2. Optical Power Meter: The 22 XLC measures the optical radiations at 820 nm and 1300 nmwith the 150 and 550 sensor heads, respectively. The bandwidth of these devices is quitewide and they are only calibrated at the center frequency. Direct measurement of opticalpower in dBm will be made with the 22 XLC. To measure the optical power, simply couplethe source into the ‘A’ optical port and depress [dBm RCVRA ONLY].

Model 150: sensitive from 400 - 1150 nm, ±6% from 400 - 900 nm, calibrated at 850 nm.Model 550: sensitive from 800 - 1800 nm, ±12% over the full range, calibrated at 1300nm

Lfiber = 10log10

Pin

Pout

1

fiber length

dB

km

Source SourceRF P P

ref outTF

FC/PC connector

SMA connector

Elastomeric splice

Optical Detector

BPT - Bare F iber P igtail

JPG - Jacketed P igtail

RF - R eference F iber

TF - T est F iber Spool

dBm = 10 log10

power1mW

Lab 1 - 2 - February 28, 2001

3. Elastomeric Splicer: This splicer is a plastic sleeve, with a drop of index matching fluidinside, that allows bare fibers to be inserted from both ends. By rotation and proper insertion,of the fibers, this can give a very low loss connection.

4. SMA Connector: These connectors allow a connection of bare fibers to each other byinserting the fiber into a metal connector and then aligning connectors with a plastic sleeve.They were the standard in the 1980’s, are quite robust, but do not align the fibers veryaccurately. This means that, although low loss connections are possible, consistency is hardto achieve. These connectors can not be used reliably for SM fibers.

5. FC/PC Connector: This type of connector does not have the rotation problems of SMAconnectors, is more accurately built and the fiber is normally epoxied into place, cleaved andthen polished. The multimode type typically has a metal tip while the singlemode has a whiteceramic tip. On some of these connectors it is hard to push the fiber through. If you push toohard, the fiber will break and the lab staff can try to repair it. Keep in mind, that theseconnectors cost $50 each. If you are having problems with a connector, point it out to thelab staff, do not try to fix it yourself.

5. Pigtails: There are at most two pigtails used in this experiment. One is for coupling lightfrom the source to the fiber and the other, from the fiber to the sensor head.

The 1230-050B Amphenol SMA cable is a jacketed pigtail using graded index fiber. It hasa SMA type connecting terminal.

6. Reference fiber and Test fiber: lengths of bare fiber.

PROCEDURE

You are to investigate the coupling losses in two types of fiber to fiber connections, which are (1)SMA, (2) FC. In addition, you will measure the transmission loss (attenuation) for SM and MMfiber at three commonly used communications wavelengths, 850nm, 1300nm and 1550nm.

These simple experiments are intended to give you practice in basic fiber preparation and mea-surement. In addition, you will appreciate the (non-negligible) magnitudes of the losses involvedin fiber splicing, and how they can affect overall system performance, and experimentalmeasurements.

Repeat the following procedures for:

(1) SMA MM cables at 850 or 1300 nm for the one splice to see how the loss changes withrotation. The first fiber must be a bare fiber with a FC connector on one end and an SMA onthe other. This experiment will be already set up for you. Simply rotate the connector to seehow the loss varies with rotation. Note the maximum power measurement on the power meter[typically around -13 dB for MM fiber] as a reference for how much power was injected intothe fiber.

(2) FC SM and MM cables at 850, 1300 and 1550 nm for the reference, 1, 2 splices and thespool. Use MM spool #1 only as MM spool #2 is damamged (consider the risetime data (1GHz detector) for spool #2 at the end of lab 2).

You must use cable fiber types which match the spool ie use MM cables for a MM spool andSM patch cables for a SM spool.

Lab 1 - 3 - February 28, 2001

Hints

When doing a measurement at 2 wavelengths it is usually best to start with the reference circuit andbuild up to the most complex circuit. Then change the source to the 2nd wavelength and ensurethat the source connection is good (ie plenty of power at the output). Since all connections in thepath have been optimized you can now take the measurements by going backwards thru the con-nections you made earlier.

To achieve optimum throughput at a splice, keep the following items in mind.

(i) Check the cleaved ends of each bare fiber under the microscope before connection, toensure that the cleave is free from aberrations.

(ii) As the last step before a connection is made, clean the fiber end with acetone and tape.

(iii) Optimize the output, where possible, by rotating the two coupled fibers with respect toeach other.

Optical connections will only work if the fibers are clean. Before you try to complete a con-nection clean it with tape and acetone. Dirt on a connector will easily shatter both fiberswhich you are trying to connect.

Figure 3: One Series of Measurements at a Single Wavelength

A) Coupling and splice losses measurements

1. Set the optical output to D.C. at maximum level.

2. Connect a fiber between the generator and the sensor head. The connection to the sensorhead must be made with the proper (SMA, FC) adaptor. Do not upset the connection to thesource. The connection to the sensor can be made with a SMA or FC connector as well as abare fiber. Measure the power.

3. Insert another fiber between the detector and the first fiber. Adjust the fiber connections atthe splice (if possible) until a maximum reading is shown. Record that value.

4. Connect any additional fibers between the last fiber and the source (Figure 3). Again recordthe maximum value. From these three readings calculate the loss per splice in dB. This lossis partially due to coupling (ie transmitting the energy from one fiber to the other) and alsoto engery lost at the glass-air interface. Calculate the loss at the air-glass boundary.

NOTE: These three measurement will permit you to calculate the loss due to a splice (a fusionsplicer, a FC connector, etc.) or a length of fiber. Extract as much information as you can todemonstrate the accuracy of your results by calculating the average loss as well as the std.deviation.

Source Source

SourceSource

Reference (1)

new

new

Two Splice (3) Spool

One Splice (2)

Lab 1 - 4 - February 28, 2001

B) Transmission loss

Connect a spool of fiber between the source and detector. Using the measured coupling andsplice losses (in part A) calculate the fiber loss in dB/km. Why would you want to minimizethe number of splices and the number of connections which are changed in such a mea-surement?

Report

♦ Compare the loss measurements at the two wavelengths and the loss measurements obtainedby the different connectors. Critically analize the data considering what you know about theloss in splices. Are you really measuring the fiber loss per km?

♦ Comment on your observations, sources of error, and the accuracy of the fiber loss mea-surement. Would measuring fiber loss by the cutback method (ie measure the power thru 2km of fiber, then cutting off 1 km and remeasuring the power) be better than the methodwhich was used in the lab (ie subtracting a reference measurement from an insertion lossmeasurement). Consider that the fiber loss is on the order of 0.3 dB/km while connector lossesare aprox. 1 dB. Suggest any improvements in the technique.

♦ You should also look up EMD (Equilibrium Mode Distribution) and underfilled / overfilledlaunching conditions. Do we have EMD in our lab measurements? How can you be sure?

Questions

1) The loss of a fiber is measured with two test sets at 1300nm. One uses a laser source and oneuses an LED source. Would you expect the measured loss to be the same or different withboth test sets? Why and why not? Note: LED BW (bandwidth) = 50 nm, laser BW < 1 nm.

2) Given the information in question #1, what is the BW of the modulation of the optical sourceand how does this compare to the frequency of the source? For simplicity assume AMmodulation and that both the signal and modulator are perfect sinewaves.

2) A monochromatic source illuminates the end face of a polished fiber as shown in the followingfigure:

The source radiates in a uniform cone of half angle Θ, such that the base of the cone at thefiber face is uniformly illuminated. The source is always positioned some distance from thefiber end such that for any specific Θ, the edges of the light cone coincide with the fiber edgesas shown in the figure. Using ray optics, (i.e. assume infinite number of modes in thismultimoded fiber) calculate the power guided in the core as a fraction of the power guidedin the whole fiber (core plus cladding), as Θ is varied from 0 to π/2. Hint: Assume that to beguided, a ray path in the core or cladding need only be smaller than the critical angle in thecore or cladding (Θco or Θcl) respectively.Plot or sketch Pcore / Pfiber vs Θ.

0

co

cl cl

co

air n = 1.0

core n = 1.515

cladding n = 1.5 50µm

m20µ2 Θ

Source

Θ

Θ

Lab 1 - 5 - February 28, 2001

ECE 475 - Lab #2

DISPERSION MEASUREMENTIntroduction

The key system parameters of an optical channel (fiber) are dispersion and attenuation. Together,they determine the bandwidth (data rate) and repeater spacing. This experiment uses pulsebroadening effects to measure the dispersion characteristics of a given length of optical fiber.

Two independent effects which give rise to time dispersion in optical fiber are multipath dispersionand material dispersion. Under normal circumstances they will be present together. In thisexperiment, no attempt to separate these effects will be made, rather the bulk dispersion effects willbe studied.

Theory

Consider the dispersion model:

INPUT FIBER OUTPUT

hO(t),σo h1(t), σ1 h2(t), σ2

Figure 1. Dispersion Model

where σ0,σ1 and σ2 are the r.m.s. pulse widths; h0(t) and h2(t) are respectively, the input and out-put waveforms, and h1(t) is the impulse response of the fiber. The impulse response h1(t) isassumed to be gaussian and the r.m.s. pulse width is defined as,

The norm is:

The first moment is:

As a first approximation, if we assume h0(t) and h2(t) as "one-side triangular waves", ie.

Figure 2. Waveform Models

Given ∆Ti = rise time, L = fiber length; then one can show that:

σi2 =

1εi

⌠⌡−∞

∞

t2hi(t)dt − ti2

εi = ∫−∞

∞hi(t)dt

ti =1εi

⌠⌡−∞

∞

t hi(t)dt where i = 0,1,2.

02

τ τ∆ ∆

Lab 2 - 1 - December 22, 1997

Experimentally if one can measure σ0 and σ2 (ie. ∆T0 & ∆T2), then by using the theory of linearsystems one can calculate

With the knowledge of σ1, and a Gaussian model for the fiber impulse response, one can write

Bandwidth is defined by the frequency at which

or

Finally an estimate of the differential delay over the length L is given in terms of the Bandwidth as

or

Equipment

1. Optical Signal Generator: A narrow bandwidth, optical signal generator which can bemodulated with a fast rise-time squarewave.

2. Optical Waveform Analyzer: A large bandwidth, high speed, optical detector. Connect the1500XP directly into the oscilloscope to eliminate extra distortion from cables.

3. Connectors: FC/PC connectors

4. Reference fibers and test fiber spools.

5. Digital Storage Oscilloscope.

σi = 0.236∆Ti ; ∆Ti = 2τi ; i = 0,2.

σtotal2 = ∑σn

2 or σ12 = σ2

2 − σ02 or σ2

2 = σ02 + σ1

2

h1(t) =1

σ1√2πexp

−t2

2σ12

⇒ H1(ω) =F .T .h1(t)

= exp

−σ12ω2

2

| H1(ω) | =12

= exp

−σ12(2πB .W .)2

2

ln

12

= −12

(2πσ1 B .W .)2

⇒ 2B .W . = √ 2ln2πσ1

∆τ L =1

2BW

∆τ =πσ1

√ 2ln2 Lsec /km .

Lab 2 - 2 - December 22, 1997

Procedure

1. Connect the signal generator to the waveform analyzer thru a reference fiber.

2. Connect the waveform analyzer directly to the oscilloscope.

3. Select a suitable frequency on the generator (1 MHz or 10MHz typical) and a proper rangeon the waveform analyzer (1mv/µω if possible), so that the waveform will appear on the scopeand maximum time resolution can be used to determine the rise time of the signal.

4. Measure the rise time of the output signal (∆T), and thus calculate σ2.

5. Replace the reference fiber with a test fiber spool, measure the rise time (∆T2), and calculate.

6. Replace the test fiber spool with another fiber spool, measure the rise time.

7. Put both of the test fiber spools in series and measure the rise time.

Report

From the above rise times, estimate the dispersion (in terms of ∆τ) of the fiber. Show all estimatesof the dispersion and calculate the average. From the dispersion of both single spools you shouldbe able to calculate the dispersion of both in series; or you could reverse the situation and calculatethe disperson of a single spool (∆τ); based upon the measured total dispersion of both (∆TT, LT)and the measured dispersion of one (∆T1, L1).

where

The procedure has already been done and the data has been included in this lab manual. At the endof this lab you will see the printed data for two different conditions. In both cases the samefiber spools were used as well as the same 2 Gsample/second oscilloscope and laser source.

In one case a 1 GHz detector was used and in the other a 125 MHz detector was used. You shouldcompare these two measurements to each other and to your own measurements. Carefullyconsider what was actually being measured.

Do a "reality check". The fiber manufacturer has labeled the fiber spools as having a bandwidthof 440 MHz-km. What does this mean? Given that data what would expect the rise time inresponse to a gaussian pulse to be? How does that compare to the various measured values?

I am not looking for people to blindly measure data and crunch some numbers. I want a realcomparison of this data. I want to know if you think that the data is good for anything; andif so what and why or why not.

σ22

∆τ =0.167π√ln2

√∆TT2 − ∆T1

2

(LT − L1)

LT > L1 are the fiber lengths

∆TT, ∆T0 are the measured rise times.

Lab 2 - 3 - December 22, 1997

Answer the following questions and suggest any improvement on this technique.

Consider that dispersion is the difference in group velocity and:

1) Group velocity changes with wavelength.

2) A laser is not a single frequency but has a bandwidth of about 5nm. If a laser were a singlefrequency at 1300 nm and had sidebands at ±5 nm; what would the AM modulating frequencybe?

3) The optical power in the fiber is the product of the modulating waveform and the spectrumof the laser.

Questions

1. What are the advantages and the limitations of this method?

2. If the rise time could be measured for a 1 kHz signal, instead of 1 MHz, would it improvethe accuracy and why?

3. If the total loss of the setup, such as launching and coupling between the pigtail and fiber canbe reduced by 10 dB, will the result be more accurate?

General Equations

Assuming guassian impulse responses:

λ =cf

(in air)

-3dB optical BW ~0.187σrms

Hz

-3dB electrical BW ~0.133σrms

Hz

TR (rise time) = 0.317

-3dB optical BW seconds

Lab 2 - 4 - December 22, 1997

E&CE 497 - Lab #3

Spectral Attenuation MeasurementsIntroduction

The optimum wavelength of an optical fiber communication system requires determination of thespectral loss of the fiber. The minimum loss of the fiber should occur at the operating wavelengthto maximize the repeater spacing. A spectral loss measurement which measures the loss of the fiberover a range of wavelengths will locate this optimum operating wavelength.

In addition to minimum loss, the singlemode system should operate at a wavelength near the fiber’snext higher mode effective cutoff to enhance fundamental mode confinement. The spectral bendingloss of the singlemode fiber will indicate this cutoff.

In this experiment, we will determine the intrinsic loss and scattering parameters of the fibers. Wewill also determine the effective cutoff of a singlemode fiber.

Spectral Loss

The loss of an optical fiber (in dB/km) can be modeled as:

where A is the Rayleigh scattering coefficient, B is the wavelength independent loss due towaveguide imperfections, microbends and bends. C(λ) is due to both the impurities and intrinsicabsorptions. While the intrinsic absorption, such as the UV and IR absorption tails are monotonicfunctions, the impurity absorptions are narrow band functions that occur at the resonance frequenciesof the impurities. The reduction of impurities present in the fiber material and an improvement inthe fabricating technique can reduce the total loss of the fiber. The amount that can be reduce islimited by the Rayleigh scattering coefficient and the absorption tails.

As λ approaches infinity, the Rayleigh scattering loss approaches to zero. If C(λ) is due mainly toimpurities, the various parameters in equation 1 can be determined from a plot of α vs λ-4.

Experimentally, one measures the loss of the fiber over a range of wavelengths. Using the referencefiber method, the spectral loss is determined from,

Spectral Bending Loss

When a mode propagates near its cutoff, it becomes weakly guided and a large portion of its powerpropagates outside the core. As the frequency increases, the radial decay factor γ increases and thepower is more confined within the core.

A weakly guided mode will suffer more loss due to bends and waveguide imperfections than astronger guided mode. The spectral bending loss of a fiber B(λ), defined as

α =A

λ4 + B + C(λ) (1)

α(λ) = log10

Pout(λ)

Pin(λ)

fiberlength

dBkm

(2)

B(λ) = 10 log10

Pbent fiber(λ)Pref .fiber(λ)

dB (3)

Lab 3 - 1 - February 28, 2001

measures the loss due to the additional curvature in a fiber can be used to determine the higher ordermode cutoff. Near cutoff, the bent fiber will no longer support the weakly guided higher ordermode and B(λ) increases shapely. For an ideal singlemode fiber without any waveguide imper-fections, the LP11 cutoff should occur at λc with V = 2.405. However, in practice, waveguideimperfections are always present. The measured cutoff wavelength λe will always be smaller thanλc.

The effective cutoff λe can than be defined as the wavelength above which the second order modeis below a given level compared to the fundamental mode. Presently, the industry standard (EIA)for cutoff measurement of a singlemode fiber is to determine λe of a 2m fiber with a 28cm diameterloop.

Equipment Setup

A halogen, white light, source is filtered by a monochromator (bandpass filter) allowing the userto select a very narrow bandwidth, typically 3.2 nm/mm, centered at a wavelength between 400nmand 1700 nm. However, the actual dispersion of wavelength in nm/mm varies with wavelengthand the size of the slit openings. Losses in the mirrors, gratings and fiber limit usefull measurementsto the 800nm to 1700 nm range for our equipment.

The narrow spectrum is then passed thru the fiber under test and measured by a wideband detector.The HP 815x optical power meters are calibrated from 850 nm to 1700 nm and can detect light to-70 dBm (HP 8152) or -110 dBm (HP 8153). Typically the noise floor is about -100 dBm due tostray light. Also the light meters have a limited resolution and accuracy. For the 8152 a typicalreading is -65.5 dBm. That will be ±5% ±0.15 dB + 100 pW. It is obivous that the number ofsignificent digits decreases as you approach -70 dBm (100pW).

By using a computer to control the sweeping of the bandpass filter (monochromator) and measurethe resulting power it is possible to easily get plots of light power versus wavelength. By taking areference measurement at the same time (ie modern microwave test equipment does this) or verysoon before an actual measurement it is possible to get absolute measurements of loss versuswavelength within the repeatability of the connections.

Procedures

A) Spectral Loss Measurement

The Lab. Staff will do the following:

1) Make sure that the power meter, halogen source and other equipment is turned on.

2) Use the fiber holder assembly to launch the optical output from the monochromatorinto a short (1m) multimode fiber. This fiber will serve as the source pigtail of thesystem.

3) Set the wavelength setting near the center of the measuring range.

4) Adjust the fiber holder assembly to obtain a maximum power output from the pigtail.

Lab 3 - 2 - February 28, 2001

Your responsibility is to:

5) Ensure that the equipment is working (ie enough light power is getting thru to get goodmeasurements). Secure the launching setup and do not disturb the setup until the finalmeasurement has been taken.

6) Connect a reference fiber (2m) between the pigtail and the detector.

7) Do a frequency sweep to acquire the reference power PREF (λ).

8) Without upsetting the launching condition, replace the reference fiber with a fiber spool.

9) Now do another frequency sweep to obtain the spectral loss of the fiber.

10) Store the measured data into a file. Notice that the stored data is the spectral loss in dB.The value must be divided by the spool length to obtain the loss per unit length.

B) Spectral Bending Loss Measurement

Assuming that the monochromator setup is operating properly:

1) Launch the monochromator output directly into the test fiber (2m).

2) Make sure that the smallest loop in the fiber has a diameter ≥ 15cm.

3) Acquire the power output as the reference.

4) Using the bending jig, get frequency sweeps with loops of 1cm to 6cm.

5) Store the bending loss data.

Measurements

1) Measure the multimode (old and new spools) and singlemode spectral losses using the range850 to 1550nm in steps of 10nm with the MM fiber (old monochromator) and 4.0nm withthe SM fiber (Digikrom monochromator).

2) Using the same range, measure the spectral bending losses of a singlemode fiber. Whenobtaining the data, use diameters of ≥ 15 cm (ie straight), 6cm, 4cm, 2cm and 1cm.

Report

1) Locate the OH absorption peaks.

2) Comment on the common fiber optics system’s operating wavelengths of 850nm, 1300nmand 1550nm (e.g. relative repeater spacing etc. given a 0 dBm source with a signal strengthof ≥ -30 dBm required).

3) Compare the multimode and singlemode spectral losses. Explain the differences.

4) Find the effective cutoff of the singlemode fiber from the spectral bending loss curves.

5) Which spectral bending loss curve should provide the best result on the effective cutoff?

6) Suggest a method to estimate the actual cutoff of the singlemode fiber.

Lab 3 - 3 - February 28, 2001

ECE 475 - Lab #4

Directional Couplers

IntroductionDirectional couplers made of two dielectric waveguides placed in close proximity are basic elementsfor many optical-wave devices, such as switches, modulators and filters. The operating principleof a directional coupler may be understood by using the coupled-mode theory which is discussedin chapter 6 of the class notes, and also below. The practical implementation of the directionalcoupler depends on the particular type of waveguides used. In this experiment, we focus ondirectional couplers made of two optical fibers. A crude technique is to be employed here, but itdoes reflect all the major steps in making a fused fiber coupler used in more sophisticated processes.

Integrated Optics Coupler on a Chip

Fabrication ConceptThe goal in making a directional coupler is to bring two light guiding regions (the fiber cores here),into close proximity. The evanescent ’tail’ of one guided mode of one core will then tunnel energyinto the guided mode of the other core. Because the evanescent field of a mode decays exponentiallyaway from the guided region (i.e., very quickly), the core regions of each waveguide must comeclose to each other for any significant coupling to occur over a finite length. If two fibers of typicalcladding diameter ,150 µm, are brought into contact, the evanescent fields are so minute at theinterfaces that coupling would eventually occur, but only over a distance far greater than 1050 km!

There are a number of ways to bring two guiding core regions within a few microns of each other.These include procedure such as micro-grinding the claddings of two fibers down to a few micronsand then ’gluing’ the fibers together, fusing the two fibers together by melting the claddings, or bymanufacturing an integrated optics waveguide (much like an integrated circuit) and connectingfibers as the input/output ports. In this lab, a 3-db coupler will be constructed by the fusion methods.This method is straight forward and easy to implement. It suffers however, from high insertionloss (difference between input power and total output power at ports 2 and 3) because not muchcontrol can be had over the heating process at the microscopic level, where melting and deformationof the core and cladding occurs. In addition, the final product is almost impossibly fragile.

In making a fused fiber coupler, two fibers are twisted together over a region where the buffers oneach have been removed. Next, using some heat source capable of melting glass (in this lab ahydrogen flame), this region is heated, thereby melting the two fibers together. At the same timethat the fusion occurs, the fibers are drawn (stretched axially) so that their central cores will comecloser together, (due to the influence of surface tension). Throughout the process, a fixed frequencylight source is connected to port 1 while the emitted light at ports 2 and 3 are monitored. When thedesired level between the output ports is achieved, the heating process is stopped, thereby fixingthe coupling ratio. Because of the unsophisticated methods used in this lab, a high insertion loss,typically around 10dB, is to be expected.

Port 1

z = 0 z = L

Port 4 Port 3Light out

Light inPort 2

Lab 4 - 1 - December 22, 1997

Because the fiber is being drawn while being heated, the fiber diameters shrink considerably. Thecoupling region, which may be a few millimeters long, is between 10µm and 20µm in diameter.You are probably quite aware by now, how fragile a typical 150µm fiber is. The 10µm region,barely even visible, may easily be broken by a breath of air. A lot of caution, and an unhurriedconduct, is therefore required during this part of the lab.

ProcedurePreparing the Fibers

Note:Couplers can be made with SM, MM or a mixture of the two fibers. Generally MM couplersare easier to make because the fiber cores are much larger and there is more light going thruthe fiber (easier to measure). The type of fiber which you use will depend upon what youwant to accomplish and what fiber is available.

1) Cut two pieces of fiber. One 1.3m long and the other 1.0m long. If you fail to make a goodcoupler on the first try you can re-use the 1.3m fiber as the 1m one and therefore only haveto cut a new 1.3m fiber for each successive attempt.

2) Lay the fibers out so that you have a minimum of length between the light source and themachine. Then strip 7cm of buffer off of each fiber. This should not be made longer becausethe vacuum clamping system will not hold striped fibers well. Strip the fibers very carefully.Start by squeezing lightly on the stripper and by working on a few millimeters at a time.

3) Turn on the vacuum and electronics by switching the power bar on. Then move the motorswitch on the coupler machine to "reverse". The fiber holding blocks should move together.When they have moved all of the way in, they will stop automatically. Also move the flamenozzle to the front as much as possible and pull the epoxy holding plate (at the back) out ofthe way of the fiber.

4) Clean the fibers with acetone and move the acetone far away from the flame.

5) Without touching the fibers, twist the fiber three times. The twists should be easily visiblewith light at the right angle. Then position the fiber in the grove, on the vacuum holders, sothat the striped section is centered and so that one of the twists is centered over where theflame will be. This will not be easy to achieve but it is very important.

6) Move the fibers slightly until the vacuum pump starts to sound like it is working hard. Trylaying some extra small pieces of fiber in the vacuum plate groove to stop air leakage. Ensurethat a twist is still centered over the flame position.

7) Connect port 1 to the 1300 nm light source with a FC/PC connector. Connect ports 2 and 3to detectors with SMA connectors and note the power at port 2. Port 3 will have no lightcoming out of it yet. You must make sure that the cleave at the detector is good or else youwill be unable to determine if the coupler is working.

Lab 4 - 2 - December 22, 1997

Fusing the Fibers

These parts should be done by the T.A. or lab staff.

1) Make sure that everything flammable is removed from the area. The hydrogen take valveshould be opened slowly. Make sure that no more than 20 psi shows on the second regulatorvalve.

2) Open the flow sensor valve (mounted on metal flame shield) to achieve a flow of 0.4 SLPM.It will take a few seconds before hydrogen starts to come out of the flame nozzle.

3) Ignite the flame with the striker. The flame may ignite with an explosive sound. This isnormal. You will not be able to see the flame. Verify the flow rate and pressure beforeproceeding.

This part is done by the student.

4) Turn the lights in the room off and move the flame under the fiber. The fiber should lookclean and glow with a white heat. The flame must heat a wide area of the fiber evenly forthe coupler to be made. Move the fiber back or forth as necessary to get an even heat. Usuallythe best way is to have the flame in-front or behind the fiber.

5) Move the switch on the coupler machine to "forward". At this point the vacuum plates shouldstart to move, slowly, apart. Ensure that the flame heat is even and keep an eye on the powerin ports 2 and 3.

6) As the fibers melt into each other light will couple into port 3. As this happens the power inport 2 will decrease. Experience shows that if port 2 originally had -20 dBm of power, thefinal coupler will be about 10dB lower to ports 2 and 3 if you have done a good job. It is notuncommon to have the output power about 15 dB lower but such a device is hard to test.

7) Try to pick the optimum power coupling and then pull the flame away from the fiber. Afterthat turn the motor switch to the center, off, position.

Mounting the Device and Cleanup

1) At this point you will have noticed that the light transmitted can change very quickly. Thisis due to stress and tension at the fiber junction changing. If the reading change, simply turnthe motor switch to forward for a few seconds to straighten and tension the fiber.

2) You should note the reading on the power meters at ports 2 and 3 and then slightly move thefibers a bit. When this happens the power reading will change quite a bit. Do these changesindicate an increase in the loss of the device or a change in the coupling between ports 1, 2and 3?

3) This part is very dangerous and should be done extremely carefully. To make sure that thefiber junction is not stressed, it is epoxied onto an aluminum block. To do this, mix some 5minute epoxy after you have moved the aluminum holding plate under the fiber. The platemust not touch the fiber or else the readings will change. Once the plate is centered on thefiber junction, gently put some epoxy on each side. Turn the motor switch to forward totension the fibers and then give the epoxy 10 minutes to dry.

3) Very that the junction is still coupling light. Identify the ports and remove the fibers fromthe source and detectors and slowly move the device onto a piece of paper. Tape the fibersonto the paper so that any jarring will not break the fiber right at the aluminum block. Younow have a device to test. Do not allow anything to touch the fiber junction. Any materialwill couple light out of the fiber and increase the loss.

Lab 4 - 3 - December 22, 1997

TheoryWhen two optical waveguides are brought into close proximity, the amplitude of the guided modein one guide will be affected by the mode amplitude of the other guide. This interaction can bedescribed in terms of a set of coupled differential equations:

1a

1b

where ai and βi are the mode amplitude and propagation constant of waveguide i respectively, andκ may be shown to be directly proportional to κ0 , (κ∝κ 0).

With the initial conditions a1(0) = 1 and a2(0) = 0 the solution to the coupled equations are easilydetermined.

2a

2b

where

3a

3b

The power carried in each guide is then

4a

4b

This simplifies for identical waveguides where

It is thus evident from the above two equations that the power along each waveguide varies in asinusoidal fashion. By terminating the coupling at some specific point along the waveguides, Lc,any fraction of power may be realized in either guide.

da1

dz= −jβ1a1 − jκa2

da2

dz= −jβ2a2 − jκa1

a1(z) = cos(Sz) − jδ

2Ssin(Sz)e

j(δ2

)z

a2(z) = jκS

sin(Sz)e−j(

δ2

)z

δ = β1 − β2

S =√ δ2

2

+ κ2

P1(z) = |a1(z)| 2 = cos2(Sz) +

δ2S

2

sin2(Sz)

P2(z) = |a2(z)| 2 =κ2

S 2sin2(Sz)

β1 = β2 ∴ δ = β1 − β2 = 0

Lab 4 - 4 - December 22, 1997

MeasurementsThe relative wavelength dependence of the output ports will be investigated.

1) Connect port 1 to the output fiber on the monochromator using the FC connectors. Insertport 2 into the optical power meter and ensure that there is adequate power being transmitted.

2) Using the monochromator S/W, acquire the spectral response of both ports 2 and 3 over thewavelength range 850 - 1700 nm.

3) Connect port 4 to the power meterand manually set the monochromator to 1300 nm(rememberthe SA Instrument monochromator wavelength is twice the number on the dial). Record thecontra-coupling value, if any, at this wavelength.

Calculations1) Plot: (port 3 power) - (port 2 power) versus wavelength

2) Measure the distances in µm between a few consecutive troughs of the somewhat sinusoidalplot in 1). Using this information, equations (4), and the fact that the coupling coefficient isproportional to 1/wavelength, (i.e., κ ∝ 1/λ), calculate the length of the coupling region.

Comments1) Is this device good for anything? Was it easy to manufacture? Could it be used as a power

splitter? Could it be used to split or combine a narrow spectrum (ie be used for wavelengthdivision multiplexing)? Is the loss of this device high or low? Remember that in a typicalcommunication system the source will generate a signal of 0 dBm and the detector is onlygood to about -50 dBm.Spend one page answering these questions or whichever questions you think are important.

Lab 4 - 5 - December 22, 1997

Using the Digikrom 242 Monochromatormonochrom.exe is a program for acquiring data with the Digikrom monochromator and HP 8153Aoptical power meter (as well as the older monochromator setup). It can be found on the class disk(access ECE475@electrical) under /digikrom.To run it on the laboratory computers type:

monochrom

To run it on Waterloo Polaris type:access ECE475@electricalmono

Brief help is available by pressing ’h’. To exit a command at any time, hit ESC. A NULL entry(ie hitting Enter) means that you accept the default. To acquire data simply press R (for Read) andhit enter or change the values as appropriate. To quickly view your current data graphically youcan hit V. You can subtract two data sets by using S to Subtract the data sets.

To examine the data values or save them hit F to go to the File submenu. Then hit S for save, P forprinting the data to the screen or D to delete (recoverable). Before you save the data you may wantto set or check the directory by hitting C (Change directory). You can save the data at any time, butonly 20 data records can be kept in memory. If all records are full you will be prompted to savesome data. Since many things can go wrong you should save any useful data when you get it. HittingESC will take you from the file submenu back to the main menu.

Some more features are available if you turn on the advanced user flag. This will allow you to setthe monochromator slit openings, remove the automatic HP8153A calibrations (to extend the fre-quency range), use data averaging or add some lag time to the sampling. Changing these can resultin bad data. The default values work well and should be used.

Brief Command Summary:

R - read data from the monochromator setup (do a frequency scan)

P - print the current data set to the screen

L - load saved data from a file (data files should have a ".dat" extension)

F - file subsystem (save, delete, tag data for plotting)

C - change the path to where you want to save/load files

S - subtract two data sets (must be similar)

E - exit the program; This places you in the file system for saving before exiting

M - set the mask for the directory list printed when you try to Load data

G - run GNU Plot to allow you to generate PS files (memory is limited)

N - plot the tagged data sets, or if none are tagged plot the saved ones, with GNU Plot

F1 - shell out to DOS. Type "exit" to return to the monochrom.exe program.

Note: This program is used for both monochromator stations. There are some additional areas ofconcern with the older monochromator.

1) The old monochromator system may appear to drift out of calibration. However, do not adjustthe calibration during a series of measurements. This could easily add more error due tobacklash in the manual controls.

Appendix 1

2) If the HP 8152A power meter is not connected to the computer and turned on then the programwill wait until the meter is connected and turned on. In other words, if it is not connected thecomputer will appear to lock.

3) The monochromator software may display warning messages and these can generally beignored. The older monochromator software may lock and in this case it is necessary to rebootthe computer.

Appendix 2

GNU Plot

GNU Plot is a plotting utility that is interactive and will easily generate postscript files for printing.It can plot mathematical functions but does not allow you to manipulate data read in from a file. Itis assumed that the data file is in text and has at least one column. You may wish to use a spreadsheetfor data manipulation and plotting instead.

If there is only one column of data, they are assumed to be successive data samples. If there are 2columns then you will get a plot of X vs Y. By adding more columns you can do surface plots. Amanual is on the disk with the software. You can examine the full S/W package by accessinggnuplot@electrical

You can run this software by typing ’gnuplot’ on Polaris/Unix. The following commands representa session that will show you some of the things that can be done with GNUplot.

A = 0.3B = 0.01 Plot a simple functiontheory(x) = A / (x ** 2) + B * xtitle "quick and dirty plot"plot theory(x)exit

plot ’497ls2-3.dat’ Plotting single or multiple data files. Plots canplot ’497ls2-3.dat’, ’497ls2-4.dat" with lines be done with points (default) or lines. A gridplot [850:1150] ’497ls2-3.dat’, theory(x) can also be added to the plot. There is a replotset grid function to save you from retyping everything.replot

set logscale xset nologscale x Different axies scales can be used.set

set term postscript This is how you can save a plot as a postscriptset output ’ls2-3.ps’ file.

plot [850:1200] [-80:-50] ’497ls2-3.dat’plot [850:] [:-50] ’497ls2-1.dat’ This how the X and Y axis can be limited whileplot [] [-70:-50] ’497ls2-1.dat’, ’497ls1-2.dat’ plotting data from a file.

Note:The plot command is of the syntax:plot [x_max:x_min] [y_max:y_min] ’file_1’, ’file_2’, ...

All data can be plotted by GNU Plot without any processing of data other than subtractingdata sets. The only part that you may have difficulty with is log plots. Typically, the automaticscaling picks minimum and maximum axis points that are exponents of ten. You will need tospecificy a range for the axis in order to get a good plot.

GNU Plot can be used to generate Post Script files for printing or PBM files which can beconverted to most any other image format. Examples of PS files are given above.

Appendix 3