Intro.

to Fibre Optic Communications Systems OE & Recvrs - 63

Photodetectors cont’d:

Photodetector Noise and Receivers

Quantum noise: the collection of photons and production of ehps is
statistical in nature. This process is described by Poisson
statistics, which in practice is approximated by Gaussian statistics.
Quantum noise is fundamental to the operation of photodetectors
and sets a lower limit on receiver sensitivity.
Shot noise is white (constant spectral distribution)
pin APD
iQ2 = σ Q2 = 2eI P B iQ2 = σ Q2 = 2eI P BM 2 F ( M )
bandwidth
Rms current Multiplication factor
squared/noise power
variance Time-average Excess noise factor
photocurrent

Aside: Excess noise in APDs

Avalanche process is also statistical in nature → additional shot

noise. The excess noise factor, F(M), is a measure of the
additional detector noise that arises from the multiplication
process. It is very complex to calculate and depends on the
electron/hole ionization rates and carrier multiplication.

 1  In Si, α >> β, KA<<1, F (M)=2

F ( M ) = k A M + (1 − k A )  2 − 
 M In InGaAs, Ge, α~β, F (M) → M
β
kA = for α > β α = electron ionization rate
α
β = hole ionization rate
α
kA = for β > α
β M= multiplication
Intro. to Fibre Optic Communications Systems OE & Recvrs - 64

Photodetectors cont’d:

Photodetector Noise and Receivers

Fig. 6-14 Excess noise

factor, F, vs. electron
gain for various values of
the effective ionization
ratio, kA, of e’s and h’s.

Empirical relation that is often used:

F(M) = Mx ; x = 0.3 (Si), 0.7 (InGaAs), 1.0 (Ge)

Dark current noise: noise due to the random thermal generation of

ehps in the depletion region of a photodetector.
pin APD
2
iDB = σ DB
2
= 2eI D B 2
iDB = σ DB
2
= 2eI D BM 2 F ( M )
Time-average dark
current
Since the quantum noise and dark current noise are uncorrelated,
the total shot noise power is the sum of the two sources.

pin σ S2 = σ Q2 + σ DB
2
= 2e ( I P + I D ) B

APD σ S2 = σ Q2 + σ DB
2
= 2e( I P + I D ) BM 2 F ( M )
Intro. to Fibre Optic Communications Systems OE & Recvrs - 65

Photodetectors cont’d:

Photodetector Noise and Receivers

Thermal Noise (Johnson or Nyquist noise) -at any finite

temperature, the electrons and atoms in a solid are in constant
motion. In a resistor, the random motion generates a random
voltage across its terminals. Thermal noise is approximately
Gaussian in nature (white noise) and is independent of signal
current IP.

iT R or

vT R

4kT vT2 = 4kTRB

iT2 = σ T2 = B
R
Main contributors to thermal noise in the photodetector/receiver
circuit are the load resistor and components in the amplifier. The
amplifier noise is very complex to analyze as it depends on the
factors such as the types of transistors used, input capacitances,
etc. A much simpler approach is to lump the noise generated by
the amplifier into the thermal noise term where it appears as a
noise figure, FN. FN may be calculated or obtained from the
amplifier manufacturer.
pin and APD 4kT
σ T2 = iLoad
2
+ iAmp
2
= BFN
RL
Intro. to Fibre Optic Communications Systems OE & Recvrs - 66

Photodetectors cont’d:

Photodetector Noise and Receivers

Shot and thermal noise are uncorrelated → the total noise power is
the sum of the two sources.
4kT
pin σ 2 = σ S2 + σ T2 = 2e( I P + I D ) B + BFN
RL
4kT
APD σ 2 = σ S2 + σ T2 = 2e( I P + I D ) BM 2 F ( M ) + BFN
RL

Signal to Noise Ratio - pin Receiver

2
Average signal power I P Note: this assumes that the
SNR = = 2
Noise power σ average signal power is equal
to the average received
R 2 Pin 2 power which is true for on/off
SNR pin =
4kT digital signals.
2e(R Pin + I D ) B + BFN
RL

In a “good” detector, ID is very small and can be ignored. Therefore

the shot noise will scale with Pin. Since the thermal noise is
independent of Pin, it is possible to identify two regions of operation.

Thermal Noise Limit: Shot Noise Limit:

at low received power, σT>>σS at high received power, σS>>σT

2
RLR 2 Pin R Pin
SNR pin ≈ SNR pin ≈
4kTBFN 2eB
Intro. to Fibre Optic Communications Systems OE & Recvrs - 67

Photodetectors cont’d:
Photodetector Noise and Receivers
Signal to Noise Ratio - APD Receiver
Note: this assumes that the
2
M R Pin2 2
average signal power is equal
SNRAPD = to the average received
4kT
2e(R Pin + I D ) BM 2 F ( M ) + BFN power which is true for on/off
RL
digital signals.
We can once again identify two regions of operation

Thermal Noise Limit: Shot Noise Limit:

at low received power, σT>>σS at high received power, σS>>σT

RL M 2R 2 Pin
2 R Pin
SNRAPD ≈ SNRAPD ≈
4kTBFN 2eBF ( M )

M2 improvement in SNR 1/F(M) reduction in SNR

compared to pin. compared to pin.
An optimum value of M, which maximizes SNRAPD, can be found
for APD receivers. For a given APD receiver, increasing M
eventually results in the shot noise dominating the thermal noise
and at this point SNR will start to decrease.
x = 0.3 for Si
2+ x 4kTFn
M opt = x = 0.7 for InGaAs
xeRL (R Pin + I D ) x = 1 for Ge

pin receivers have a better SNR when shot noise dominates (large
received optical power, Pin).
APD receivers have a better SNR when thermal noise dominates
(low received optical power)
Intro. to Fibre Optic Communications Systems OE & Recvrs - 68

Receiver Considerations

Basic elements: photodetector, amplifier, signal processing

Design: maximum performance

Digital: minimize probability of bit errors
Analogue: maximize signal-to-noise ratio
Both: realize bandwidth requirements

Figure 7-1 Signal path through an optical link

Intro. to Fibre Optic Communications Systems OE & Recvrs - 69

Receiver Error Sources:

1) Detector noise - shot/thermal noise
Note: thermal noise is purely Gaussian in nature while shot noise
is not, strictly speaking, shot noise fluctuates according to a
Poisson distribution. If a detector is illuminated with an optical
signal P(t), then the average number of ehps generated in time t
is:
η t ηE Nhν
hν ∫0
N= P (t ) dt = E=
hν η

The actual number will vary according to the Poission distribution

exp(− N )
Pr (n) = N n
n!

where Pr(n) is the probability that n electrons are emitted in time t.

2) Intersymbol interference - the result of pulse spreading during
transmission.

Figure 7-3 Pulse spreading in an optical signal that leads to

intersymbol interference.
Intro. to Fibre Optic Communications Systems OE & Recvrs - 70

Receiver Configuration

Figure 7-4 Optical receiver schematic with 3 basic sections. ib =

bias resistor current noise, ia, ea = amplifier current and voltage
noises. Must add shot noise of PD! Simplified cct diagram
shown below. [R = Rb , C = parallel cap. of PD, amp.]

R C {V2}A
G

Iph {I2}sh {I2}R {I2}A

Since the pulses that arrive at the receiver are rounded, a binary
digital pulse train can be described by ∞
∞
∫hp (t )dt = 1 pulse shape
P(t ) =∑ bn hp (t − nTb ) where −∞
n =−∞ bn (bon/boff) - energy in nth pulse
Tb - bit period
P(t) - optical power

ηe ∞
i (t ) = MP (t ) = R0 M ∑ bn hp (t − nTb )
hν n =−∞

Note: M=1 for pin receivers

Intro. to Fibre Optic Communications Systems OE & Recvrs - 71

Digital Receiver Performance

The signal emerging from the equalizer is compared to a threshold

level once per time slot to determine if an optical pulse is present
at the photodetector. In an ideal receiver, the value of vout would
always exceed the threshold level when a pulse is present.
Deviations are present in real systems due to noise, intersymbol
interference, and a non-zero extinction ratio.
Bit-Error Rate: a common approach to used to measure the
number of bit errors that occur over a time interval t.

Ne Ne
BER = =
N t Bt
Typical error rates for optical communication systems are 10-9 to
10-12. The system requirements and receiver noise levels set a
lower limit on the signal power required.
Require the probability
distribution of the signal at the
output of the equalizer.
v
P1 (v) = ∫ p( y 1)dy
−∞
∞
P0 (v) = ∫ p( y 0)dy
v

Error probability: a=b=0.5 for

unbiased data.

Fig. 7-5 Signal probability Pe = aP1 (vth ) + bP0 (vth )

distributions
Intro. to Fibre Optic Communications Systems OE & Recvrs - 72

Assuming an arbitrary signal with a gaussian probability distribution

(mean value of m), the probability that a sampled voltage will fall in
the range s to s+ds is
1  ( s − m) 2 
f ( s )ds = exp  −  ds
2πσ 2  2σ 2


Now consider a data stream where the “1” pulses are all of
amplitude V as shown below.

∞ ∞

Probability of error in a “0” P0 (vth ) = ∫ p( y 0)dy = ∫ f (s)ds 0

pulse vth vth

1
∞
 (v − boff ) 2 
=
2πσ off
∫v exp  − 2σ off2 dv
th  

vth vth

Probability of error in a “1” P1 (vth ) = ∫ p( y 1)dy = ∫ f1 (v)dv

pulse −∞ −∞
vth
1  (bon − v) 2 
=
2πσ on ∫−∞ exp  − 2σ on2 dv
Intro. to Fibre Optic Communications Systems OE & Recvrs - 73

Assuming equal probability of “0” and “1” in the data stream, the bit-
error rate becomes
∞
1
BER = Pe (Q) =
π ∫ exp(− x 2 )dx
Q/ 2
For Q> 3
1  Q  1 exp(−Q / 2) 2
= 1 − erf    ≈
2  2  2π Q
where
x
vth − boff bon − vth 2
Q=
σ off
=
σ on
erf ( x) = ∫
p0
exp(− y 2 )dy

Important values:
Q = 6 → BER = 10-9
Q = 7 → BER = 10-12

Q provides an alternate measure

of the SNR.
SNR = 20logQ

Figure 7-7 BER versus Q

factor
Intro. to Fibre Optic Communications Systems OE & Recvrs - 74

Special case: σoff=σon=σ and boff=o and bon=V

therefore vth = V/2 and Q = V/2σ.
1  V 
Pe (σ on = σ off ) = 1 − erf
2   
 2σ 2  

Fig 7-8 BER vs. SNR. This

assumes gaussian statistics,
Popt = 0 when signal off (bit =
0).

Fig. 7-16 Power penalty in

receiver sensitivity vs.
extinction ratio. No
intersymbol interference,
Si(0.3) and InGaAs(0.7).
ε = boff/bon
Intro. to Fibre Optic Communications Systems OE & Recvrs - 75

Receiver sensitivity - minimum average received power required

to achieve a given BER at a given bit rate.
If we neglect the dark current contribution to shot noise in the
photodetector, it is possible to show that:
Q σT 
Prec =  eF ( M )QB +  F(M)=M=1 for pin
min R  M

pin receiver - Qσ T
Prec =
thermal noise min R
dominate

APD receiver - increase M to obtain an optimal value to

minimize average Prec
1 1
−
1
 σT 2   σT  2
′ =k
M opt A
2
 + k A − 1  ≈ 
 QeB   k AQeB 

 2eB  ′ +1− kA )
so that Prec =   Q2 (k A M opt
min  R 

ideal receiver - σT = 0, Prec =  eB  Q 2

M=F(M)=1 min R 
Intro. to Fibre Optic Communications Systems OE & Recvrs - 76

Receiver sensitivity
- Power required for specified BER or SNR
- Noise power increases as BW (data rate)
- More signal power required
Intro. to Fibre Optic Communications Systems OE & Recvrs - 77

Receiver sensitivity
Quantum Limit

The quantum limited receiver operates in the shot noise limit i.e. σT
= 0. Furthermore, it has unity quantum efficiency and no dark
current. This means that there are no ehps generated in the
absence of an optical pulse. “0” bits are never received as “1” bits
P0 = 0.
We must now consider the actual statistics which govern quantum
noise - the Poisson distribution.
exp(− N ) N is the average number of ehps produced in a “1” bit
Pr (n) = Nn

n! n is the actual number of ehps produced.

Therefore a “1” bit is only received erroneously if n = 0. P1 (0) = exp(− N )

The BER then becomes: 1 1
BER = Pe = P1 = exp(− N )
2 2
Now consider that each ehp is produced by a single photon (η = 1)
so that the average number of photons required per “1” bit is the
same as N .
NP = N
Finally, if we assume that we have an equal number of “0” and “1”
bits in a data stream, the average number of photons/bit will simply
be NP
NP =
2
and the minimum received power required to achieve the desired
BER (receiver sensitivity)
Prec = N P hvB
QL
Intro. to Fibre Optic Communications Systems OE & Recvrs - 78

Effect of intersymbol interference (excess dispersion)

Fig. 7-3 Pulse spreading outside of bit interval.

Fig. 7-15 Power penalty in receiver sensitivity due to

pulse spreading.
γ = fraction of bit energy in T.
Intro. to Fibre Optic Communications Systems OE & Recvrs - 79

Bandwidth considerations
R C {V2}A
G

Iph {I2}sh {I2}R {I2}A

SNR = (M Re Popt)2 / [ K1 M2F(M) Re Popt B + K2 B/R ]

Large R decreases thermal noise, improves SNR.

Fig. 7-18 High-impedance Fig. 7-19 High-impedance pre-

pre-amp using FET amp using bipolar transistor

But large R decreases bandwidth …

Example:
I(ω) R C
R = 1000 Ω, C = 1 pF.
BW = 160 MHz
V(ω) = I(ω) R/( 1 + j ω RC)
|H(f)|2 = 1/( 1 + (2πf RC )2)
BW = 1/(2πRC )
Intro. to Fibre Optic Communications Systems OE & Recvrs - 80

Bandwidth improvement -
1. low impedance amplifier (50 Ω)

RL RA Gain

Low Impedance Amplifier

Typically RL = RA = 50Ω to minimize reflections

50Ω high speed amplifiers are very common as they are
used in impedance matched microwave systems.
Results in large bandwidth (low RC) but poor sensitivity
•IP only generates a small voltage into the amplifier
•thermal noise is high

Not generally used in long-haul fibre systems but is very

popular in lab setting for testing photodetectors since high
speed test equipment frequently uses 50Ω terminations.
Intro. to Fibre Optic Communications Systems OE & Recvrs - 81

Bandwidth improvement -
2. transimpedance amplifier

Fig. 7-17 Equivalent circuit of transimpedance receiver. (A < 0)

Rf

-A

Input resistance in reduced by amplifier gain Rin=Rf/A

|H(f)|2 = 1/( 1 + (2πf RfC/A )2), A = gain of amplifier
BW = A/(2πRfC )
1) Wide dynamic range compared to high-Z amp
2) Little equalization required - small time constant
3) Transfer characteristic is controlled by feedback resistance
4) Less sensitive than high-Z amp - difference is only 2-3 dB
EE 586 - Introduction to Optical Fiber Communications Propagation 1

Section 2

Propagation of Light
in Optical Fibres

• Optics fundamentals - (rays, modes, Snell’s

Law, etc...)

•Propagation of light in different fibre types -

(MM, SM, ...)

•Physical description of fibres (materials,

fabrication, ...)

•Signal degradation in fibres

•Fibre-fibre, fibre-source light coupling

EE 586 - Introduction to Optical Fiber Communications Propagation 2

The nature of light

Waves: Particles:

Diffraction Quantum of energy,

“photon”
Interference (photelectric effect)

Mathematical Descriptions: Waves or Rays

(geometrical optics, particle trajectories)

E = r-1exp[j(wt – kr)] E = exp[j(wt – bz)]

EE 586 - Introduction to Optical Fiber Communications Propagation 3

Basic optical properties

Speed of light c = 3 × 108 m/s

Wavelength λ = c/f = λ0/n

Frequency f

Energy E = hf ; h = 6.63 × 10-34 J-s

E (eV) = 1.24 / λ0 (µm)

Index of refraction

Air 1.0

water 1.33

glass (SiO2) 1.47

silicon-nitride 2.0

Silicon 3.5
EE 586 - Introduction to Optical Fiber Communications Propagation 4

Plane waves

Fig. 2-2: Linear polarization

Fig. 2-3: Orthogonal polarizations

Ex(z,t) = ex E0x cos(ωt - kz)
Ey(z,t) = ey E0y cos(ωt - kz +δ)

E(z,t) = Ex(z,t) + Ey(z,t)

EE 586 - Introduction to Optical Fiber Communications Propagation 5

Fig. 2-4: Elliptical

polarization;
arbitrary E0x /E0y , δ

Fig. 2-5: Circular

polarization

E0x /E0y = 1, δ =900

EE 586 - Introduction to Optical Fiber Communications Propagation 6

I
R

Fig. 2-6: Snell’s law

n1 sinφ1 = n2 sinφ2
n1 cosθ1 = n2 cosθ2

What about amplitudes?

EE 586 - Introduction to Optical Fiber Communications Propagation 7

Critical
angle

Fig. 2-7: Total internal reflection

n1 cosθc = n2 cos 00

θc = cos-1(n2/n1)
Example: n1 = 1.50, n2 = 1.00; θc =
EE 586 - Introduction to Optical Fiber Communications Propagation 8

Fig. 2-8: Phase shift in

total internal reflection.
EE 586 - Introduction to Optical Fiber Communications Propagation 9

Optical fibre structures

c
a b

Fig. 2-9: Fibre structure

Core: n1 = 1.47

a = 50 µm (for MMF)
= 10 m (for SMF)

Cladding: n2 = 1.46

b = 125 µm

Buffer: high, lossy n3

c = 250 µm
EE 586 - Introduction to Optical Fiber Communications Propagation 10

Fig. 2-10: Different types

EE 586 - Introduction to Optical Fiber Communications Propagation 11

Fibres Step Index

Fig. 2-11 Skew rays

Fig. 2-12: TIR of meridional rays

EE 586 - Introduction to Optical Fiber Communications Propagation 12

Calculation of numerical aperture: NA = max sin θ0

for trapped ray

NA = (n12 – n22)1/2

∆
EE 586 - Introduction to Optical Fiber Communications Propagation 13

Wave theory of modes

Start with Slab Waveguide (not in text; see ):

- solution in simple functions
- shows discrete nature of modes, effect of
boundaries, etc.
- useful in integrated optics
x - do only TE modes (TM very similar)

z 2d

Maxwell’s Equations

(+J)
EE 586 - Introduction to Optical Fiber Communications Propagation 14

Other eqs. and assumptions .

, EM wave equation

Wave theory of modes, cont’d:

TE Solution in slab waveguide

κ2 =ko2n12 - β2; γ2 = β2 - k02n12

X tanX = Y (for symmetric); X = κd

X cotX = -Y (for asymmetric); Y = γd

X2 +Y2 = V2
EE 586 - Introduction to Optical Fiber Communications Propagation 15

X2 Y2 2= =
V2 V 2
X2++Y

V = 5.8

Graphical solution of even and odd modes in a

slab waveguide..

V
EE 586 - Introduction to Optical Fiber Communications Propagation 16

Numerical Example:

Width = 10 µ m, n1 = 1.505, n2 = 1.495 ,

ω = 2x1015 . Find the lowest even and odd mode,

1. Calculate V

2. Even Mode
3. Odd Mode

Fig. 2-14 : Amplitudes of first 3 modes.

Intensity.
 EE 586 - Introduction to Optical Fiber Communications Propagation 17

Example of slab mode behavior in

dispensed polymer waveguides:

Laser-written Direct Dispensed

US Patent 5534101, Keyworth & McMullin

polymer profile:
4
guide height (µm)

0
0 100 200 300 400
x (µm)
experiment

simulation

Observed light propagation

 EE 586 - Introduction to Optical Fiber Communications Propagation 18

3D Beam Propagation Simulation

z=0µm

z=400µm

z=600µm

z=800µm

z=1100µm

“Graded Effective Index Waveguides”

EE 586 - Introduction to Optical Fiber Communications Propagation 19

Rectangular waveguide:

Clad index = n2

Core index = n1

2dy

2dx

No exact solution in simple functions.

Approximate:
E(x,y) = E0 cos(Kxmx) cos(Kyny) exp[ j(wt - bmnz) ]
(Any combination of sin and cos allowed very similar
to mettalic microwave waveguides)

Each mode is specified by 2 “quantum

numbers”, m and n.
EE 586 - Introduction to Optical Fiber Communications Propagation-20

Wave theory of modes - Cylindrical

Waveguide:

Fig. 2-15
Geometry
EE 586 - Introduction to Optical Fiber Communications Propagation-21

Bessel’s equation:
F1 = Jv(qr) inside,
= Kv(γr) outside,
fields must match at core-clad boundary, r = a.
(Picture next slide)

(q2 + γ2 ) a2 = V2, same as slab waveguide.

EE 586 - Introduction to Optical Fiber Communications Propagation-22

CylindricalWaveguide:
(cont’d)

r=
a

Fig. 2-16: First three Bessel fcns of 2nd

Bessel’s fcns.. kind:

For each ν, only a few values of q satisfy b.c. Like

rectangular waveguide, each mode has two numbers.
EE 586 - Introduction to Optical Fiber Communications Propagation-23
EE 586 - Introduction to Optical Fiber Communications Propagation-24
EE 586 - Introduction to Optical Fiber Communications Propagation-25

CylindricalWaveguide: (cont’d)

Fig. 2-17
Electric field vectors in 4 lowest modes

If V is very high then number of

modes M = V2/2

A mode approaches cutoff (β = k2)

or γ = 0
⇒q=V
EE 586 - Introduction to Optical Fiber Communications Propagation-26

Fig. 2-18
Propagation constants as a function of V.
EE 586 - Introduction to Optical Fiber Communications Propagation-27

Fig. 2-20,21 “LP” modes

EE 586 - Introduction to Optical Fiber Communications Propagation-28

η=
EE 586 - Introduction to Optical Fiber Communications Propagation-29

CylindricalWaveguide: (cont’d)

Fig. 2-19 Normalized propagation

constant as a function of V for LP modes.
EE 586 - Introduction to Optical Fiber Communications Propagation-30

Fig. 2-22 Fraction of power in the cladding as

a function of V for LP modes.

At cut off, Pclad/ PTotal,

η= 0 for j =0 and 1
(j-1)/j for j ≥2

For many modes,

Pclad/ PTotal = (4/3) M-1/2
EE 586 - Introduction to Optical Fiber Communications Propagation-31

Field Intensity

LP01

LP03
EE 586 - Introduction to Optical Fiber Communications Propagation-32

LP11

LP12
EE 586 - Introduction to Optical Fiber Communications Propagation-33

LP21

LP22

LP41
EE 586 - Introduction to Optical Fiber Communications Propagation-34

Single mode Fibres

Fig. 2-23 Distribution of

light in single mode fibre
and Mode Field Diameter
(MFD)

Mode-field diameter:
EE 586 - Introduction to Optical Fiber Communications Propagation-35

Birefringence:

Fig. 2-24 Two polarizations of HE11

Birefringence:

Bf = ny - nx
-8 -3
10 < B < 10

β = k0(ny - nx)

Beat length:

L = 2π/ β
EE 586 - Introduction to Optical Fiber Communications Propagation-36

Graded Index Fibres

n(r)

r
Fig. 2-25 NA(r) for various
α’s
EE 586 - Introduction to Optical Fiber Communications Propagation-37

GRIN Fibres (cont’d)

“Skew Rays” - Projection of a

skew ray on the end of a GRIN
fibre.

Region where wave equation

has oscillatory solution
EE 586 - Introduction to Optical Fiber Communications Propagation-38

Same theory as before but n = n(r), q =

q(r):

Hermite-gaussian
equation
EE 586 - Introduction to Optical Fiber Communications Propagation-39

Fibre materials
Doped silica (SiO2glass) :

Fig. 2-26 How dopants affect index of

silica fibres

Fluoride glass:
EE 586 - Introduction to Optical Fiber Communications Propagation-40

Fibre fabrication

Fig. 2-27 Fibre-drawing apparatus

EE 586 - Introduction to Optical Fiber Communications Propagation-41

Fig. 2-29 Vapour phase axial deposition

(VAD)
EE 586 - Introduction to Optical Fiber Communications Propagation-42

SiH4 + 2O2 = SiO2 + 2H2O

Silene = SiH4
Becomes bad

Fig. 2-28 Outside vapour phase oxidation (OVPO)

fabrication of fibre preform
EE 586 - Introduction to Optical Fiber Communications Propagation-43

Other types of preform fab

methods

Fig. 2-30 Modified chemical

vapour deposition (MCVD)

Fig. 2-31 Plasma

chemical vapour
deposition (PCVD)
EE 586 - Introduction to Optical Fiber Communications Propagation-44

Fig. 2-32 Double crucible method fibre

drawing.

Fig. 2-33 Micro-crack model

EE 586 - Introduction to Optical Fiber Communications Propagation-45

Mechanical Properties

Fig. 2-34 Probability of fracture under stress

EE 586 - Introduction to Optical Fiber Communications Propagation-46

Fibre Optic Cables

Fig. 2-35 Possible 2-

fibre cable design

Fig. 2-36 6-fibre cable

design

Issues:

•Strength
•Access to electrical power
•Resistance to ...
o moisture
o buckling
o microbending
EE 586 - Introduction to Optical Fiber Communications Propagation-47

Signal Degradation in Optical Fibres

1. Attenuation
- basics covered in introductory lectures

Fig. 3-1: Optical fiber attenuation

Raleigh scattering
Intrinsic absorption
Extrinsic absorption
EE 586 - Introduction to Optical Fiber Communications Propagation-48

Signal Degradation in Optical Fibres

Other loss mechanisms

Fig. 3-2: Effects of

ionizing radiation
EE 586 - Introduction to Optical Fiber Communications Propagation-49

Fig. 3-7: Power loss in a

curved fiber

Fig. 3-8:
Microbending losses
EE 586 - Introduction to Optical Fiber Communications Propagation-50

Fig. 3-29:
Bending-induced attenuation

Fig. 3-31:
Bend loss versus bend radius
EE 586 - Introduction to Optical Fiber Communications Propagation-51

Fig. 3-30:

Bending effects on loss vs MFD

EE 586 - Introduction to Optical Fiber Communications Propagation-52

Signal Degradation in Optical Fibres

Pulse Distortion in Optical Fibres

Fig. 3-10 Pulse spreading, dispersion, can lead to

intersymbol interference.

1. Intermodal - each mode has own group velocity

2. Intramodal - each wavelength has own group velocity,
“Chromatic” dispersion These are independent.
EE 586 - Introduction to Optical Fiber Communications Propagation-53

Intramodal (within a mode, or chromatic )

Dispersion;

Group Velocity and Group Index

Phase velocity

Group velocity
EE 586 - Introduction to Optical Fiber Communications Propagation-54

Group Delay

Dispersion
EE 586 - Introduction to Optical Fiber Communications Propagation-55

1) Material dispersion

Fig. 3-12 Index of

refraction for silica as a
function of wavelength.

Fig. 3-13 Material

dispersion for 2 glasses
EE 586 - Introduction to Optical Fiber Communications Propagation-56

2) Waveguide dispersion

Dispersion
EE 586 - Introduction to Optical Fiber Communications Propagation-57

Fig. 3-15 The waveguide Fig. 3-16 Material and

parameter for the basic wave-guide dispersion
mode LP01 of a singlemode
fibre.
EE 586 - Introduction to Optical Fiber Communications Propagation-58

Fig. 3-17:

Polarization mode dispersion

Important in WDM systems, long spans
 EE 586 - Introduction to Optical Fiber Communications Propagation-59

Intermodal (between modes) Dispersion;

“multipath”
Step index waveguide

Bandwidth-Distance product:
EE 586 - Introduction to Optical Fiber Communications Propagation-60

Pulse Broadening in Multimode Waveguides, cont’d

Intermodal dispersion in GRIN waveguides

Index profile: n(r)

r
EE 586 - Introduction to Optical Fiber Communications Propagation-61

Pulse Broadening in GRIN Waveguides (cont’d)

Fig. 3-18
RMS pulse spreading vs. α at
900 nm.Spectral widths, σλ ,
for sources ...
LED: σλ =15 nm
ILD: σλ = 1 nm
DFB: σλ = 0.2 nm

Fig. 3-19
Profile dispersion: the optimum
value of α depends on λ. If λ is
off the design λ , pulse spreading
will be worse.
EE 586 - Introduction to Optical Fiber Communications Propagation-62

Fig. 3-20 Variations in

BW for non-ideal α
EE 586 - Introduction to Optical Fiber Communications Propagation-63

Mode Coupling

Fig. 3-21 Pulse spreading vs.

distance when mode coupling
occurs.
EE 586 - Introduction to Optical Fiber Communications Propagation-64

Optimization of SM Fibres - Refractive Index

profiles

(a) 1300 nm
optimized

(b) 1550 nm
dispersion shifted

(c) 1300 - 1550 nm

dispersion flattened

Fig. 3-22

[see also (d) “large-effective-core fibres in 3rd

edition text]
EE 586 - Introduction to Optical Fiber Communications Propagation-65

Optimization of SM Fibres - Refractive Index

profiles

Fig. 3-24a: SM-fiber

dispersions

Fig. 3-24b: SM-fiber total

dispersions
EE 586 - Introduction to Optical Fiber Communications Propagation-66

Fig. 3-27
Bandwidth-distance product vs.
wavelength for sources of different
spectral width. Fibre has dispersion
minimum at 1300 nm
EE 586 - Introduction to Optical Fiber Communications Propagation-67

Summary of Causes of Dispersion:

1. Modal (intermodal, multipath) dispersion.

Higher modes are slower (usually). Graded
index helps a lot!

2. Chromatic dispersion. If only one mode, still have

finite wavelength spread, σλ. , from source. Group
velocity = function of λ .
Dλ > 0, longer wavelengths slower.

1
2
3 3 2 1
z
t=0 Later t

λr
λ0 D*L = dτ/dλ
λb λr, λ0, λb
z
t=0 Later t
EE 586 - Introduction to Optical Fiber Communications Propagation-68

Examples of Dispersion Calculation

1. An LED operating at 850 nm has an rms spectral

width of 45 nm. Use Fig. 3-13 (top curve) to
calculate the rms pulse spreading in ns/km. Find the
Bit-Rate*Distance product.

2. A semiconductor laser operating at 1550 nm has an

rms spectral width of 2 nm. Use Fig. 3-16 (“Total”
curve) to calculate the rms pulse spreading in ns/km.
Find the Bit-Rate*Distance product.
EE 586 - Introduction to Optical Fiber Communications Propagation-69

Strategies for reducing dispersion

- at the transmitter

1. Various chirping schemes. Arrange current pulse,

use external modulator or optical amplifier so
slower light leaves first

2. FM modularion (FSK). Dispersion in system

converts to AM signal which can be de-modulated.

3. Multilevel coding. E.g., 4 levels vs, 2 means bit

T twice as long, bandwidth half.

- along the fibre

1. High-dispersion fibres. (DCF - “dispersion

compensation fibre)

Normal, D > 0 at 1.55 µm. DCF; D << 0 at

1.55
EE 586 - Introduction to Optical Fiber Communications Propagation-70

-at the receiver

1. Electronic: standard pulse-shaping

transversal filters

2. Optical

(a) Integrated Mach-Zehnder interferometer

with unbalanced arms.

(b) Chirped fibre Bragg gratings.

Blue Blue
Red Red