Components
1
3.1 Couplers
[ wavelength independent, wavelength
selective for 1.31/1.55 multiplexing]
1
1-
coupling ratio
3dB couple = 1/2
= 0.95 (for monitoring) 2
For multiplexing
1310nm
1310nm
1550nm
1550nm
For EDFA
1550nm 1550nm
980nm
or 1480nm
980nm
or 1480nm
a2 b2
E: electrical field
S-parameters
For lossless couplers
E01( f ) j l cos( k l ) i sin( k l ) Ei1( f )
e (3.1)
E02 ( f ) i sin( k l ) cos( k l ) Ei 2 ( f )
b1 s11 s 12 a1
b
2 s s
21 22 a2
l : the coupling length
k : coupling coefficient depending on width,
4
shape of waveguides, n1 distance...
The power transfer function
2
E0 j
Tij ( f ) 2
Eii
i : input, j : output
T11( f ) cos2 ( kf )
(3.2)
T12 ( f )
2
sin ( kf )
For a 3dB coupler
1
T11( f ) T12 ( f )
2
1
sin ( kf ) cos ( kf )
2 2
2
k l (2n 1) n0
4 5
3.1.2 Conservation of Energy (S-parameter)
E01 s11 s12 E i1
(3.3)
E02 s21 s22 E i 2
The scattering matrix is
s11 s12
S
sij : complex
s21 s22
uv
Denote E 0 ( E01 E02 )T
uv
E i ( Ei1 Ei 2 )T
uv uv
E0 S Ei
The sum of input power is proportional to
uv T uv 2 2
E 0 E 0 Ei1 Ei 2
complex conjugate 6
Similarly the sum of output power is proportional to
uv T uv 2 2
E 0 E 0 E01 E02
If it is lossless
uv T uv uv T uv
E0 E 0 (S E i ) (S E i )
uv T T uv
E i ( S S )E i
uv T uv
Ei Ei
uv
This relation holds for arbitraryE i
ST S I (3.4)
I : identity matrix
Eq(3.4) can be extended to any number of ports
7
For a 2 x 2 symmetrical coupler
s21 s12 a s22 s11 b
ST S I
2 2
a b 1 (3.5)
ab ba 0 (3.6)
a cos( x ) b sin( x ) (3.7)
let a cos( x )eia , b sin( x )e ib
ab ba 0
cos( x ) sin( x ) ei (a b ) ei (b a ) 0
cos( a b ) 0
2k 1
a b
2
8
lossless combination is impossible
3.2 Isolators and Circulators
(nonreciprocal devices)
Isolators are for transmitter, circulators are for add
and drop or others.
The insertion loss should be small ~ 1dB
A circulator is similar to an isolator except it has
multiple ports.
9
3.2.1 Principle of Operation of an Isolator
SOP= state of Polarization
10
A spatial walk-off polarized splits the signal into
two orthogonally polarized components. 11
3.3 Multiplexer and Filters
Multiplexers and filters are for WDM, add/drop. WXC,
12
Dynamic WXCs use optical switches and mux/demux.
13
The desired characteristics of filters
1. Low insertion loss
2. Polarization-independent loss
3. Low temperature coefficient
4. Reasonable broad passbands
5. Sharp passband skirts passband skirt
6. Low cost
a. integrated-optic (may be polarization
dependent)
14
b. all-fiber devices
15
3.3.1 Gratings
Any device whose operation involves interference among
multiple optical signals originating from the same source
but with different relative phase shifts. An exception is a
device where the multiple optical signals are generated by
repeated traversals of a single cavity (etalons).
F-P
16
17
Principle of Operation
The pitch of the grating (distance between adjacent slits)=a
Assuming plane wave is incident at angle i
d : diffraction angle
The slits are small compared to ,
phase changes across a slit is negligible
18
AB CD a sin i a sin d
a sin i sin d
a sin i sin d m
m : the order of the grating
19
20
The energy at a single is distributed over
all the discrete angles that satisfy (3.9).
For WDM only light of a certain order m will
be collected, the remaining energy is lost.
m=0 has most energy i= d
The wavelengths are not separated.
blazed reflection grating maximize the light
energy at
21
3.3.2 Diffraction Pattern
Relax the constrain a <<, the phase change
across the slit is not negligible, consider a slit of
length from y w 2 to w 2 , w
w
2
y
w
2
The relative phase shift of the diffracted light from y at
an angle compared to that from y=0 is given by
y sin
( y ) 2 22
The amplitude A() at (Ref: Optics, page401)
A 0 w
A
2
exp i ( y ) dy
w w
2
A 0
dy
w
2
exp i 2 (sin ) y
w w
2
A 0 sin( w sin )
(3.10)
w sin
A 0
2
exp 2 i
(sin ) d
2
exp 2 i
(s in ) d
2
= A 0 cos( sin d )
24
3.3.3 Bragg Gratings (BGs)
31
32
Long-Period Fiber Grating (a few intermeters)
Useful for EDFA gain (equalization)
They may be cascaded to obtain the desired profile.
33
Principle of Operation
The propagating mode in core couples onto the
modes in the cladding => induce loss
For a given
coupling occurs depending on
= propagation constant of the core mode
cP1: propagation constant of the path order
cladding mode
The phase matching condition
cPl 2
Because cPl is very small
long : a few hundred m 34
Let neff and n be the refractive indices of the
p
eff
core and the path-order cladding modes
2 neff
2
P
cl
neff nneff
P
2
( neff nneff
P
)
core cladding
P
Given neff , nneff , obtain
It is a wavelength dependent loss element.
35
3.3.5 Fabry-Perot Filters
Principle of Operation
The wavelengths for which the cavity length is an integral
multiple of half the wavelength in the cavity are called
resonant wavelengths.
36
A round trip through the cavity is an integral multiple
of the wavelength.
The light waves add in phase.
(1 A R )2
(1- R )2 +4 R sin2 ( l )
2
A
1 1 R
2
2 R 2 nl
1 sin
1 R
2 n
: one way delay nl c ,
2
A
1 1 R
TFP ( f ) (3.12)
2
2 R
1 sin(2 f )
38
1 R
For maximam TFP ( f ) sin 2 f 0 f k / 2
A=0, R=0.75, 0.9 and 0.99
TFP (f) is periodic function with period FSR
Where FSR: free spectral range
= The spectral range between two successive passband
= 1/2 39
k 1
( f ' f ) , k 1 FSR f ' f
2 2
Define finesse
F
FSR
FWHM
R
1 R
proof : (3.12) Assume A 0
2 R
sin 2 f ' 1, for 3dB point
1 R
1 R
sin 2 f '
2 R
1 R = 1 sin 2 f ' 2 f ' , f ' is the smallest value
1 R satisfied the condition
f ' , FWHM 2 f '
4 R
FSR 1 1 R R
F (3.13)
FWHM 2 2 R 1 R
40
Tunability
1. change cavity length
2. change refractive index n
k
Recall f0 k : positive integer
2
The wave with frequency f 0 will be selected.
ln
c
1. mechanical tuning
2. piezoelectric tuning
=> thermal instability, hysteresis
41
3.3.6 Multilayer Dielectric Thin-Film Filters
A thin-film resonant multicavity filter (TFMF) consist of
two or more cavitied.
Advantages: flat top, sharp skirt, low loss, insensitive
to the polarization
42
43
44
3.3.7 Mach-Zehnder Interferometers (MZI)
46
At the upper output .
The signal all through the upper arm as
reference.
The signal through the lower arm and the
upper output has phase lag
L L
2 2
At the lower output the phase difference
L L
2 2
through low arm
through upper arm
3dB coupler
coupler output
delay/2
delay/2 cancel47
If L k k is odd k (2n 1)
(2n 1) 2( n 2) in phase
The signals at the upper arm add in phase at upper arm
At the lower output, the phase difference is (2n 1) L (2n 1)
out of phase ( no signal )
If L (2n )
At the upper arm output, the phase difference is
(2n) (2n 1) out of phase no signal
At the lower arm output
L (2n ) signals add in phase
The transfer function of MZI is
T11 ( f )
sin2 L
2
(3.14)
T12 ( f )
cos
2 L
2
1
hint : cos( k l ) sin( k l ) and
2
The input and output relation of the middle section is
E i'1 ( f ) 1 0 E 01 ( f )
'
i L
e i L
E ' ( f ) 0 e E 02 ( f )
'
i2
let E i 2 ( f ) 0 and Multiply three matrices (3.14) 48
consider K MZI interconnected
The path length difference for the kth MZI is assumed to be L
2k 1
49
50
MZI can be used as a 1x2 demultiplexer or multiplexer
1 2 1
MZI
2
1 2 chosen to be coincide with the peaks or troughs of the
transfer function
2 neff
mi i 2 neff mi i
let L , L mi mi : integer
2neff i 2neff
sin2 L 2 sin2
mi
2
2Lneff
If 1 , and mi is odd, say mi=1 output 1 has
mi
signal, output 2 has no signal,
2Ln 51
If
2
eff
and mi is even, output 1 has no signal.
m i
3.3.8 Array wavelength Grating (AWG)
i
k
n n
m m
L=length
in
difference between two adjacent waveguides.
dik = difference in distance between input i and array waveguide k
d k out
j =difference in distance between array waveguide k and output
54 j
The relative phase
2
ijk
n d in
1 ik n2 k L n1 kj
d out
(3.15)
k= 1. 2. m
input output
through k
If we design that
dikin diin k iin
d kjout d out
j k out
j
2
Then ijk
n1d iin n2 k L n1d out
j n 1k i
in
n1k j
out
2 n1 in 2 k
di d j out
n1 iin n2 L n1 out j (3.16)
= + 55
Rowland circle construction
grating circle
Rowland
56
If j appears at input i, and
n1 iin n2 L n1 out
j p j , p is integer
2 k
p j 2 kp
j
j will add in phase at output j ( prob 3.16)
j will be present at output j
If n1 iin n2 L n1 out
j ( p 1) j '
j ' will be also present at output j
pc ( p 1)c
let n 1 i
in
n2 L n1 out
j p
f f'
c
FSR f ' f
n2 L 57
3.3.9 Acoustic-Optic tunable Filter (AOTF)
polarization-dependent, polarization-independent.
58
Principle of Operation
As Fig 3.27 AOTF is constructed from a birefringent
material and only supporting the lowest-order TE
and TM modes.
If an acoustic wave is launched, the n varies to form
gratings.
The Bragg condition is satisfied
nTM nTZ 1
(3.17)
TE mode is converted to TM mode.
For LiNbO3, |nTE-nTM|=0.07=n. at 1.55m
=n (3.18)
At 170MHz =22m, acoustic wavelength
59
The transfer function is
sin2 1 (2 )2
2
T ( )
1 (2 )2
where =-0 0 satisfies (3.17)
=02/ln
l : the length of acoustic-optic interaction
FWHM bandwidth=0.8
60
61
Disadvantages: high loss, large crosstalk, bulky
wide passband> 100GHz
dynamic crossconnect
62
3.3.10 High Channel Count Multiplexer
Architectures
A. Serial (only for small number of ports)
63
B. Single stage (AWG)
64
C. Multistage banding
65
D. Multistage Interleaving
66
3. 4 Optical Amplifiers
67
3.4.1 Stimulated Emission (EDFA or
SOA)
68
Two energy levels E2>E1
hfc= E2-E1, h: Planck's constant= 6.63x10-34JS
(absorption)
E1E2 excitation (by photons or population inversion)
E2E1 emission photons
a. stimulated emission
b. spontaneous emission
If emission > absorption => amplification
N1: Population (number of atoms) at E1
N2: population at E2
rate of E1 E2
r
rate of E2 E1
70
3.4.3 EDAF
72
Stark splitting : an isolated ion of erbium is split
into multiple energy levels.
Each stark splitting level is spread into a band.
Thermalization : the erbium ions are distributed in
the various levels within the band.
Capable of amplifying several wavelengths
simultaneously.
hfc E2 E1
32 E3 E2 : 1 sec
21 E2 E1 : 10m sec
atoms stay at E2 longer, 980nm pump is usable
1520nm f c 1570 nm BW 50 nm
page 39, c-band from 1530~1565nm
73
When 980nm pump is used
32 1sec << 21
We have population inverse between E2 and E1
We can amplify 1530-1570nm signals
When 1480nmpump is used the absorption from
the bottom of E1 to the top of E2
1480nm pump is less efficient
Less population inversion
Higher noise figure
980nm for low noise EDFA
High power 1480nm pump is available
=> High output power and pump can be located
remotely
74
75
76
77
Gain Flatness
78
79
80
81
Multistage Designs
1550~1600nm
signal
1460~1480nm pump
RA can provide gain about 100nm band (13THz) above the
pumping wave p<s (Signal Wavelength)
A. RA is a distributed device and can provide gain in different
bands
B. No special fibers are needed
C. Required high pump power~1w
D. Pump power fluctuations induce noise (propagating in same
direction), propagating in opposite directions will have lower
noise
E. Crosstalk (modulated signals will deplete the pump power =>
fluctuation => noise) so, pumping opposite direction will lower
the noise. (average out)
F. Another noise is due to Rayleigh scattering of the pumping 84
signal
85
86
3.4.5 Semiconductor Optical Amplifiers (SOAs)
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
3. 5 Transmitters
106
Principle of Operation (semiconductor laser)
107
Reference: John Gowar Optical Communication Systems PP262~323
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
Longitudinal Modes
Multiple-longitudinal mode (MLM) lasers have large
spectral widths~10nm (Fabry-Perot lasers) =>cause
chromatic dispersion
Singlelongitudinal mode (SLM) lasers have very narrow
spectral widths
Side-mode suppression ratio is an Important parameter for
SLM lasers. (~30dB)
125
Distributed-Feedback Lasers (DFB Lasers)
Distributed Bragg reflector (DBR) Lasers
127
3.5.3 Tunable lasers
Tunable lasers are useful to reduce the inventory, (spare
parts), to reconfigure the network, to be used for
optical packet switched networks and for laboratory
testing.
Tuning mechanisms
a. Injecting current (change n) tuning range ~10~15 nm
at 1550nm
b. Temperature tuning 0.1nm/
c. Mechanical tuning (wide range but bulky)
Desirable properties
a. Short tuning time
b. Wide tuning range (100nm)
c. Stable over its lifetime
d. Easily controllable and manufacturable
128
Two-and Three-Section DBR Lasers
Problems
a. Aging
b. Temperature changes
c. Current recalibration 129
d. Mode hopping
Vertical grating-assisted coupler filter (VGF) Lasers
The coupling condition (3.17)
=B(n1-n2)
B: The period of the Bragg grating
n1 and n2 are refractive indices of two waveguides.
If n1 changes to n1+n1
' B ( n1 n1 n2 )
B n1
B n1
n1
n1 n2
130
Sample Grating and Super-Structure
Grating DBR lasers
131
Grating Coupled sampled Reflection lasers
132
3.5.4 Direct and External Modulation
Direct modulation
Advantage: Simple
Disadvantage: induce chirping
Biasing above the threshold will reduce chirping but decrease the
extinction ratio. 133
External Modulation
a. Lithium niobate modulator, b. electro-absorption
modulator
134
T11( f ) cos2 ( k l )
(3.2)
T (
12 f ) sin 2
( k l )
k : coupling coefficient depending on
width of the waveguide, refractive
indices, distance of two waveguides
135
T11( f ) sin2 ( L 2)
(3.14)
T12 ( f ) cos 2
( L 2)
MZI can achieve high extinction ratio ~15 ~20dB with almost on
chirping. Polarization control is needed. 136
3.6 Detectors
137
3.6.1 Photodetectors
Photons incident on a semiconductor are absorbed by
electrons in the valence band. These are excited into the
conduction band and leave holes in the valence band.
When a reversed bias voltage is applied, these electron
hole pairs produce photo current.
fc
hf c eE g (3.19)
fc 1.24
( m.ev )
eEg Eg
138
139
Pabs (1 e L )Pin (3.20)
Pin : incident power
: absorption coefficient
L : Thickness of the semiconductor
The efficiency
Pabs
1 e L (3.21)
Pin
4
Example 10 , L 10 m, 0.99
cm
140
The responsivity
IP A
R
Pin W
Pin
IP e
hf c
e A
R
hf c W
e
hc
A
=
1.24 W
The quantum efficiency I ph e ( electrons / sec) /
Pin / E ph ( photons / sec)
( I ph e )
=
Pin / E ph
I ph hc hc
R
Pin e e 141
142
PIN Photodiodes
a. A very lightly doped intrinsic
semiconductor between the p-type and
n-type Layers can improve the efficiency.
The depletion region extends across the
intrinsic layer.
b. If the p-type or n-type layer is
transparent the efficiency can be further
improved.
143
144
145
146
Avalanche Photodiodes (APD)
When the generated election in a very high
electric field, it can generate more
secondary electron-hole pairs. This
process is called avalanche multiplication.
Gm: multiplicative gain
M: multiplication factor (Gm: M-1)
Large Gm will induce large noise.
If Gm, avalanche breakdown occurs.
147
148
149
150
151
152
153
154
155
156
157
3.6.2 Front-End Amplifiers
a. High-impedance amplifier
b. Transimpedance amplifier
158
159
160
3.7 Switches
Important parameters
a. Number of ports
b. Switching time
c. The insertion loss
d. The crosstalk
e. Polarization-dependent loss
f. Latching (maintaining its switch state)
g. Monitoring capability
h. Reliability
161
3.7.1 Large Optical Switched
The main considerations
a. Number of switch elements required
b. Loss uniformity
c. Number of crossovers
d. Blocking characteristics
blocking and nonblocking (strict sense, wide
sense, rearrargeable)
e. Synchronous or asynchronous
162
Crossbar
163
Spanke
164
3.7.2 Optical Switch Technologies
165
MEMS Switches
166
167
Bubble-Based Waveguide Switch
168
Liquid Crystal Switches
169
A. Thermal-Optic Switches (MZI)
B. Semiconductor Optical Amplifier Switches
C. Large Electronic Switched
a) Single stage
b) Multistage
c) Line rate
d) Total capacity (line rate x number of ports)
e) Circuit switching V.S. packet switching
f) Cross bar V.S. shared memory
170
3.8 Wavelength Converters
a. A device converters data from one incoming
wavelength to another outgoing wavelength.
b. Used in WDM networks
i. input wavelength is not suitable for the networks
ii. Improving the wavelength utilization in WDM
networks
iii. Converting to suitable outgoing wavelengths
c. Types
i. fixed-input, fixed-output
ii. Variable-input, fixed-output
iii. Fixed-input, variable-output
iv. Variable-input, variable-output
171
d. Other important characteristics
i. convertion range
ii. Transparent to data rate or modulation format
iii. Loss (efficiency)
iv. Noise, crosstalk
e. Mechanism to achieve wavelength convertion
i. optoelectronic (commercial available)
ii. Optical gating
iii. Interferomatric
iv. Wave mixing
172
3.8.1 Optoelectronic Approach (O/E, E/O)
173
3.8.2 Optical Grating
Using the principle of cross-gain modulation
in a SOA. (For high input signal power, the
carrier will be depleted => less gain for the
probe wavelength)
174
Disadvantages
i. small extinction ratio
ii. High input signal power to deplete the
carriers (simultaneously changes n)
iii. Requiring to filter this high-powered
signal
iv. Changing refractive index inducing
pulse distortion
175
3.8.3 Interferometric Techniques
176
Principle of Operation (cross phase
modulation CPM)
When s presents, the carrier densities (or n)
change to induce different phase changes
of p. At the port A, the intensity of p will
be modulated.
i. digital signal only
ii. Higher extinction ratio
iii. Providing reamplification and reshaping
iv. Low input power
177
Stage1 samples the data
Stage2 reshapes and retimes the data (inverse) 178
Stage3 reamplifies
3.8.4 Wave Mixing
179