# 1

Problem Solutions for Chapter 2
2-1.

E · 100cos 2π10
8
t + 30° ( ) e
x
+ 20 cos 2π10
8
t − 50° ( ) e
y
+ 40cos 2π10
8
t + 210°
( )
e
z
2-2. The general form is:
y = (amplitude) cos(ωt - kz) = A cos [2π(νt - z/λ)]. Therefore
(a) amplitude = 8 µm
(b) wavelength: 1/λ = 0.8 µm
-1
so that λ = 1.25 µm
(c) ω · 2πν = 2π(2) = 4π
(d) At t = 0 and z = 4 µm we have
y = 8 cos [2π(-0.8 µm
-1
)(4 µm)]
= 8 cos [2π(-3.2)] = 2.472
2-3. For E in electron volts and λ in µm we have E =
1.240
λ
(a) At 0.82 µm, E = 1.240/0.82 = 1.512 eV
At 1.32 µm, E = 1.240/1.32 = 0.939 eV
At 1.55 µm, E = 1.240/1.55 = 0.800 eV
(b) At 0.82 µm, k = 2π/λ = 7.662 µm
-1
At 1.32 µm, k = 2π/λ = 4.760 µm
-1
At 1.55 µm, k = 2π/λ = 4.054 µm
-1
2-4. x
1
= a
1
cos (ωt - δ
1
) and x
2
= a
2
cos (ωt - δ
2
)
1
and x
2
yields
x
1
+ x
2
= a
1
[cos ωt cos δ
1
+ sin ωt sin δ
1
]
+ a
2
[cos ωt cos δ
2
+ sin ωt sin δ
2
]
= [a
1
cos δ
1
+ a
2
cos δ
2
] cos ωt + [a
1
sin δ
1
+ a
2
sin δ
2
] sin ωt
Since the a's and the δ's are constants, we can set
a
1
cos δ
1
+ a
2
cos δ
2
= A cos φ (1)
2
a
1
sin δ
1
+ a
2
sin δ
2
= A sin φ (2)
provided that constant values of A and φ exist which satisfy these equations. To
verify this, first square both sides and add:
A
2
(sin
2
φ + cos
2
φ) = a
1
2
sin
2
δ
1
+ cos
2
δ
1
( )
+ a
2
2
sin
2
δ
2
+ cos
2
δ
2
( ) + 2a
1
a
2
(sin δ
1
sin δ
2
+ cos δ
1
cos δ
2
)
or
A
2
=

a
1
2
+ a
2
2
+ 2a
1
a
2
cos (δ
1
- δ
2
)
Dividing (2) by (1) gives
tan φ =
a
1
sinδ
1
+ a
2
sinδ
2
a
1
cosδ
1
+ a
2
cosδ
2
Thus we can write
x = x
1
+ x
2
= A cos φ cos ωt + A sin φ sin ωt = A cos(ωt - φ)
2-5. First expand Eq. (2-3) as
E
y
E
0 y
= cos (ωt - kz) cos δ - sin (ωt - kz) sin δ (2.5-1)
Subtract from this the expression
E
x
E
0 x
cos δ = cos (ωt - kz) cos δ
to yield

E
y
E
0 y
-
E
x
E
0x
cos δ = - sin (ωt - kz) sin δ (2.5-2)
Using the relation cos
2
α + sin
2
α = 1, we use Eq. (2-2) to write
3
sin
2
(ωt - kz) = [1 - cos
2
(ωt - kz)] = 1 −
E
x
E
0x
|
.

`
,

2

]
]
]
(2.5-3)
Squaring both sides of Eq. (2.5-2) and substituting it into Eq. (2.5-3) yields
E
y
E
0 y

E
x
E
0x
cosδ

]
]
]
2
= 1 −
E
x
E
0x
|
.

`
,

2

]
]
]
sin
2
δ
Expanding the left-hand side and rearranging terms yields
E
x
E
0x
|
.

`
,

2
+
E
y
E
0y
|
.

`
,

2
- 2
E
x
E
0x
|
.

`
,

E
y
E
0y
|
.

`
,
cos δ = sin
2
δ
2-6. Plot of Eq. (2-7).
2-7. Linearly polarized wave.
2-8.
33 ° 33 °
90 °
Glass
Air: n = 1.0
(a) Apply Snell's law
n
1
cos θ
1
= n
2
cos θ
2
where n
1
= 1, θ
1
= 33°, and θ
2
= 90° - 33° = 57°
∴ n
2
=

cos 33°
cos 57°
= 1.540
(b) The critical angle is found from
n
glass
sin φ
glass
= n
air
sin φ
air
4
with φ
air
= 90° and n
air
= 1.0
∴ φ
critical
= arcsin
1
n
glass
= arcsin
1
1.540
= 40.5°
2-9
Air
Water
12 cm
r
θ
Find θ
c
from Snell's law n
1
sin θ
1
= n
2
sin θ
c
= 1
When n
2
= 1.33, then θ
c
= 48.75°
Find r from tan θ
c
=

r
12 cm
, which yields r = 13.7 cm.
2-10.
45 °
Using Snell's law n
glass
sin θ
c
= n
alcohol
sin 90°
where θ
c
= 45° we have
n
glass
=

1.45
sin 45°
= 2.05
2-11. (a) Use either NA = n
1
2
− n
2
2
( )
1/ 2
= 0.242
or
5
NA ≈ n
1
2∆ = n
1
2(n
1
− n
2
)
n
1
= 0.243
(b) θ
0,max
= arcsin (NA/n) = arcsin
0.242
1.0
|
.
`
,
= 14°
2-13. NA = n
1
2
− n
2
2
( )
1/ 2
= n
1
2
− n
1
2
(1− ∆)
2
[ ]
1/ 2
= n
1
2∆ − ∆
2
( )
1 / 2
Since ∆ << 1, ∆
2
<< ∆; ∴ NA ≈ n
1
2∆
2-14. (a) Solve Eq. (2-34a) for jH
φ
:
jH
φ
= j
εω
β
E
r
-
1
βr
∂H
z
∂φ
Substituting into Eq. (2-33b) we have
j β E
r
+
∂E
z
∂r
= ωµ j
εω
β
E
r

1
βr
∂H
z
∂φ

]
]
]

Solve for E
r
and let q
2
= ω
2
εµ - β
2
to obtain Eq. (2-35a).
(b) Solve Eq. (2-34b) for jH
r
:
jH
r
= -j
εω
β
E
φ
-
1
β
∂H
z
∂r
Substituting into Eq. (2-33a) we have
j β E
φ
+
1
r
∂E
z
∂φ
= -ωµ −j
εω
β
E
φ

1
β
∂H
z
∂r

]
]
]

Solve for E
φ
and let q
2
= ω
2
εµ - β
2
to obtain Eq. (2-35b).
(c) Solve Eq. (2-34a) for jE
r
:
6
jE
r
=
1
εω
1
r
∂H
z
∂φ
+ jrβH
φ
|
.

`
,
Substituting into Eq. (2-33b) we have

β
εω
1
r
∂H
z
∂φ
+ jrβH
φ
|
.

`
,
+
∂E
z
∂r
= jωµ H
φ
Solve for H
φ
and let q
2
= ω
2
εµ - β
2
to obtain Eq. (2-35d).
(d) Solve Eq. (2-34b) for jE
φ
jE
φ
= -
1
εω
jβH
r
+
∂H
z
∂r
|
.
`
,
Substituting into Eq. (2-33a) we have
1
r
∂E
z
∂φ
-
β
εω
jβH
r
+
∂H
z
∂r
|
.
`
,
= -jωµ H
r
Solve for H
r
to obtain Eq. (2-35c).
(e) Substitute Eqs. (2-35c) and (2-35d) into Eq. (2-34c)
-
j
q
2
1
r

∂r
β
∂H
z
∂φ
+ εωr
∂E
z
∂r
|
.

`
,

∂φ
β
∂H
z
∂r

εω
r
∂E
z
∂φ
|
.

`
,

]
]
]
= jεωE
z
Upon differentiating and multiplying by jq
2
/εω we obtain Eq. (2-36).
(f) Substitute Eqs. (2-35a) and (2-35b) into Eq. (2-33c)
-
j
q
2
1
r

∂r
β
∂E
z
∂φ
− µωr
∂H
z
∂r
|
.

`
,

∂φ
β
∂E
z
∂r
+
µω
r
∂H
z
∂φ
|
.

`
,

]
]
]
= -jµωH
z
Upon differentiating and multiplying by jq
2
/εω we obtain Eq. (2-37).
2-15. For ν = 0, from Eqs. (2-42) and (2-43) we have
7
E
z
= AJ
0
(ur) e
j(ωt − βz)
and H
z
= BJ
0
(ur) e
j(ωt − βz)
We want to find the coefficients A and B. From Eqs. (2-47) and (2-51),
respectively, we have
C =
J
ν
(ua)
K
ν
(wa)
A and D =
J
ν
(ua)
K
ν
(wa)
B
Substitute these into Eq. (2-50) to find B in terms of A:
A
jβν
a
|
.
`
,
1
u
2
+
1
w
2
|
.
`
,
= Bωµ
J'
ν
(ua)
uJ
ν
(ua)
+
K'
ν
(wa)
wK
ν
(wa)

]
]
]
For ν = 0, the right-hand side must be zero. Also for ν = 0, either Eq. (2-55a) or (2-56a)
holds. Suppose Eq. (2-56a) holds, so that the term in square brackets on the right-hand
side in the above equation is not zero. Then we must have that B = 0, which from Eq. (2-
43) means that H
z
= 0. Thus Eq. (2-56) corresponds to TM
0m
modes.
For the other case, substitute Eqs. (2-47) and (2-51) into Eq. (2-52):
0 =
1
u
2
B
jβν
a
J
ν
(ua) + Aωε
1
uJ'
ν
(ua)

]
]
+
1
w
2
B
jβν
a
J
ν
(ua) + Aωε
2
w
K'
ν
(wa)J
ν
(ua)
K
ν
(wa)

]
]
]
With k
1
2
= ω
2
µε
1
and k
2
2
= ω
2
µε
2
rewrite this as
Bν =

ja
βωµ

1
1
u
2
+
1
w
2

]
]
]
]
k
1
2
J
ν
+ k
2
2
K
ν
[ ] A
8
where J
ν
and K
ν
are defined in Eq. (2-54). If for ν = 0 the term in square brackets on the
right-hand side is non-zero, that is, if Eq. (2-56a) does not hold, then we must have that A
= 0, which from Eq. (2-42) means that E
z
= 0. Thus Eq. (2-55) corresponds to TE
0m
modes.
2-16. From Eq. (2-23) we have
∆ =
n
1
2
− n
2
2
2n
1
2
=
1
2
1 −
n
2
2
n
1
2
|
.

`
,

∆ << 1 implies n
1
≈ n
2
Thus using Eq. (2-46), which states that n
2
k = k
2
≤ β ≤ k
1
= n
1
k, we have

n
2
2
k
2
· k
2
2
≈ n
1
2
k
2
· k
1
2
≈ β
2
2-17.
2-18. (a) From Eqs. (2-59) and (2-61) we have
M ≈

2
a
2
λ
2
n
1
2
− n
2
2
( )·

2
a
2
λ
2
NA ( )
2
a ·
M

|
.
`
,
1/ 2
λ
NA
·
1000
2
|
.
`
,
1/ 2
0.85µm
0.2π
· 30.25µm
Therefore, D = 2a =60.5 µm
(b) M ·

2
30.25µm ( )
2
1.32µm ( )
2
0.2 ( )
2
· 414
(c) At 1550 nm, M = 300
2-19. From Eq. (2-58),
9
V =

2π (25 µm)
0.82 µm
(1.48)
2
− (1.46)
2
[ ]
1/ 2
= 46.5
Using Eq. (2-61) M ≈ V
2
/2 =1081 at 820 nm.
Similarly, M = 417 at 1320 nm and M = 303 at 1550 nm. From Eq. (2-72)
P
P
|
.
`
,
total

4
3
M
-1/2
=
4 ×100%
3 1080
= 4.1%
at 820 nm. Similarly, (P
/P)
total
= 6.6% at 1320 nm and 7.8% at 1550 nm.
2-20 (a) At 1320 nm we have from Eqs. (2-23) and (2-57) that V = 25 and M = 312.
(b) From Eq. (2-72) the power flow in the cladding is 7.5%.
2-21. (a) For single-mode operation, we need V ≤ 2.40.
Solving Eq. (2-58) for the core radius a
a =

n
1
2
− n
2
2
( )
−1/ 2
=
2.40(1.32µm)
2π (1.480)
2
− (1.478)
2
[ ]
1/ 2
= 6.55 µm
(b) From Eq. (2-23)
NA = n
1
2
− n
2
2
( )
1/ 2
= (1.480)
2
− (1.478)
2
[ ]
1/ 2
= 0.077
(c) From Eq. (2-23), NA = n sin θ
0,max
. When n = 1.0 then
θ
0,max
= arcsin
NA
n
|
.
`
,
= arcsin
0.077
1.0
|
.
`
,
= 4.4°
2-22. n
2
= n
1
2
− NA
2
= (1.458)
2
−(0.3)
2
= 1.427
a =
λV
2πNA
=
(1.30)(75)
2π(0.3)
= 52 µm
10
2-23. For small values of ∆ we can write V ≈
2πa
λ
n
1
2∆
For a = 5 µm we have ∆ ≈ 0.002, so that at 0.82 µm
V ≈

2π (5 µm)
0.82 µm
1.45 2(0.002) = 3.514
Thus the fiber is no longer single-mode. From Figs. 2-18 and 2-19 we see that the LP
01
and the LP
11
modes exist in the fiber at 0.82 µm.
2-24.
2-25. From Eq. (2-77) L
p
=

β
=
λ
n
y
− n
x
For L
p
= 10 cm n
y
- n
x
=

1.3 ×10
−6
m
10
−1
m
= 1.3×10
-5
For L
p
= 2 m n
y
- n
x
=

1.3 ×10
−6
m
2 m
= 6.5×10
-7
Thus
6.5×10
-7
≤ n
y
- n
x
≤ 1.3×10
-5
2-26.
We want to plot n(r) from n
2
to n
1
. From Eq. (2-78)
n(r) = n
1
1− 2∆(r / a)
α
[ ]
1 / 2
= 1.48 1− 0.02(r / 25)
α
[ ]
1 / 2
n
2
is found from Eq. (2-79): n
2
= n
1
(1 - ∆) = 1.465
2-27. From Eq. (2-81)

M ·
α
α + 2
a
2
k
2
n
1
2
∆ ·
α
α + 2

2πan
1
λ
|
.
`
,
2

where
∆ =
n
1
− n
2
n
1
= 0.0135
11
At λ = 820 nm, M = 543 and at λ = 1300 nm, M = 216.
For a step index fiber we can use Eq. (2-61)
M
step

V
2
2
=
1
2
2πa
λ
|
.
`
,
2
n
1
2
− n
2
2
( )
At λ = 820 nm, M
step
= 1078 and at λ = 1300 nm, M
step
= 429.
Alternatively, we can let α · ∞ in Eq. (2-81):
M
step
=
2πan
1
λ
|
.
`
,
2
∆ =

1086 at 820 nm
432 at 1300 nm
¹
'
¹
2-28. Using Eq. (2-23) we have
(a) NA = n
1
2
− n
2
2
( )
1/ 2
= (1.60)
2
− (1.49)
2
[ ]
1/ 2
= 0.58
(b) NA = (1.458)
2
− (1.405)
2
[ ]
1/ 2
= 0.39
2-29. (a) From the Principle of the Conservation of Mass, the volume of a preform rod
section of length L
preform
and cross-sectional area A must equal the volume of the fiber
drawn from this section. The preform section of length L
preform
is drawn into a fiber of
length L
fiber
in a time t. If S is the preform feed speed, then L
preform
= St. Similarly, if s is the
fiber drawing speed, then L
fiber
= st. Thus, if D and d are the preform and fiber diameters,
respectively, then
Preform volume = L
preform
(D/2)
2
= St (D/2)
2
and Fiber volume = L
fiber
(d/2)
2
= st (d/2)
2
Equating these yields
St
D
2
|
.
`
,
2
= st
d
2
|
.
`
,
2
or s = S
D
d
|
.
`
,
2
(b) S = s
d
D
|
.
`
,
2
= 1.2 m/s

0.125 mm
9 mm
|
.
`
,
2
= 1.39 cm/min
12
2-30. Consider the following geometries of the preform and its corresponding fiber:
PREFORM
FIBER
4 mm
3 mm
R
25 µm
62.5 µm
We want to find the thickness of the deposited layer (3 mm - R). This can be done by
comparing the ratios of the preform core-to-cladding cross-sectional areas and the fiber

A
preform core
A
=

A
fiber core
A
or
π(3
2
− R
2
)
π(4
2
− 3
2
)
=

π (25)
2
π (62.5)
2
− (25)
2
[ ]
from which we have
R = 9 −
7(25)
2
(62.5)
2
− (25)
2

]
]
]
1/ 2
= 2.77 mm
Thus, thickness = 3 mm - 2.77 mm = 0.23 mm.
2-31. (a) The volume of a 1-km-long 50-µm diameter fiber core is
V = πr
2
L = π (2.5×10
-3
cm)
2
(10
5
cm) = 1.96 cm
3
The mass M equals the density ρ times the volume V:
M = ρV = (2.6 gm/cm
3
)(1.96 cm
3
) = 5.1 gm
13
(b) If R is the deposition rate, then the deposition time t is
t =
M
R
=

5.1 gm
0.5 gm / min
= 10.2 min
2-32. Solving Eq. (2-82) for χ yields
χ =
K

|
.
`
,
2
where Y = π for surface flaws.
Thus
χ =

(20 N/ mm
3 / 2
)
2
(70 MN/ m
2
)
2
π
= 2.60×10
-4
mm = 0.26 µm
2-33. (a) To find the time to failure, we substitute Eq. (2-82) into Eq. (2-86) and
integrate (assuming that σ is independent of time):

χ
− b / 2

χ
i
χ
f

dχ = AY
b
σ
b
0
t

dt
which yields
1
1 −
b
2
χ
f
1− b / 2
− χ
i
1− b/ 2
[ ]
= AY
b
σ
b
t
or
t =
2
(b − 2)A(Yσ)
b
χ
i
(2− b) / 2
− χ
f
( 2−b) / 2
[ ]
(b) Rewriting the above expression in terms of K instead of χ yields
t =
2
(b − 2)A(Yσ)
b
K
i

|
.
`
,
2− b

K
f

|
.
`
,
2 −b

]
]
]

2K
i
2−b
(b − 2)A(Yσ)
b
if K
i
b− 2
<< K
f
b− 2
or K
i
2 −b
>> K
f
2− b
2-34. Substituting Eq. (2-82) into Eq. (2-86) gives

dt
= AK
b
= AY
b
χ
b/2
σ
b
14
Integrating this from χ
i
to χ
p
where
χ
i
=
K

i
|
.

`
,

2
and χ
p
=
K

p
|
.

`
,

2
are the initial crack depth and the crack depth after proof testing, respectively, yields

χ
−b / 2

χ
i
χ
p

dχ = AY
b

σ
b

0
t
p

dt
or

1
1 −
b
2
χ
p
1− b / 2
− χ
i
1−b / 2
[ ]
= AY
b
σ
p
b
t
p
for a constant stress σ
p
. Substituting for χ
i
and χ
p
gives
2
b − 2
|
.
`
,
K
Y
|
.
`
,
2−b
σ
i
b− 2
− σ
p
b− 2
[ ]
= AY
b
σ
p
b
t
p
or
2
b − 2
|
.
`
,
K
Y
|
.
`
,
2−b
1
AY
b
σ
i
b− 2
− σ
p
b− 2
[ ]
= B σ
i
b−2
− σ
p
b−2
[ ]
=

σ
p
b
t
p
which is Eq. (2-87).
When a static stress σ
s
is applied after proof testing, the time to failure is found from Eq.
(2-86):

χ
−b / 2

χ
p
χ
s

dχ = AY
b

σ
s
b

0
t
s

dt
where χ
s
is the crack depth at the fiber failure point. Integrating (as above) we get Eq. (2-
89):
B σ
p
b−2
− σ
s
b−2
[ ]
=

σ
s
b
t
s
Adding Eqs. (2-87) and (2-89) yields Eq. (2-90).
15
2-35. (a) Substituting N
s
as given by Eq. (2-92) and N
p
as given by Eq. (2-93) into Eq.
(2-94) yields
F = 1 - exp −
L
L
0
σ
p
b
t
p
+ σ
s
b
t
s
( )
/ B+ σ
s
b−2
[ ]
m
b−2
σ
0
m

σ
p
b
t
p
/ B+ σ
p
b−2
( )
m
b− 2
σ
0
m
¹
'
¹
¹
¹
¹
'
¹
¹
¹
|
.

`
,

= 1 - exp

L
L
0
σ
0
m
σ
p
b
t
p
/ B+ σ
p
b− 2
[ ]
m
b −2
σ
p
b
t
p
+ σ
s
b
t
s
B
|
.

`
,
+ σ
s
b −2
σ
p
b
t
p
/ B+ σ
p
b− 2

]
]
]
]
]
m
b− 2
¹
'
¹
¹
¹
¹
¹
¹
'
¹
¹
¹
¹
¹
− 1
|
.

`
,

= 1 - exp −LN
p
1 +
σ
s
b
t
s
σ
p
b
t
p
+
σ
s
σ
p
|
.

`
,

b
B
σ
s
2
t
p

]
]
]
m
b −2
1 +
B
σ
p
2
t
p
−1
¹
'
¹
¹
¹
¹
¹
¹
'
¹
¹
¹
¹
¹
|
.

`
,

≈ 1 - exp −LN
p
1+
σ
s
b
t
s
σ
p
b
t
p
|
.

`
,

1
1+
B
σ
p
2
t
p

]
]
]
]
]
m
b−2
−1
¹
'
¹
¹
¹
¹
¹
¹
'
¹
¹
¹
¹
¹
|
.

`
,

(b) For the term given by Eq. (2-96) we have
σ
s
σ
p
|
.

`
,

b
B
σ
s
2
t
p
= (0.3)
15

0.5 (MN / m
2
)
2
s
0.3 (350 MN/ m
2
)
[ ]
2
10 s
= 6.5×10
-14
Thus this term can be neglected.
16
2-36. The failure probability is given by Eq. (2-85). For equal failure probabilities of the
two fiber samples, F
1
= F
2
, or
1 - exp −
σ
1c
σ
0
|
.

`
,

m
L
1
L
0

]
]
]
= 1 - exp −
σ
2c
σ
0
|
.

`
,

m
L
2
L
0

]
]
]
which implies that
σ
1c
σ
0
|
.

`
,

m
L
1
L
0
=
σ
2c
σ
0
|
.

`
,

m
L
2
L
0
or
σ
1c
σ
2c
=
L
2
L
1
|
.

`
,

1 / m
If L
1
= 20 m, then σ
1c
= 4.8 GN/m
2
If L
2
= 1 km, then σ
2c
= 3.9 GN/m
2
Thus
4.8
3.9
|
.
`
,
m
=
1000
20
= 50
gives
m =
log 50
log(4.8/ 3.9)
= 18.8
1
Problem Solutions for Chapter 3
3-1.

α dB/ km ( )=
10
z
log
P(0)
P(z)

]
]
]
·
10
z
log e
α
p
z
( )
·10α
p
log e · 4.343 α
p
(1/ km)
3-2. Since the attenuations are given in dB/km, first find the power levels in dBm for
100 µW and 150 µW. These are, respectively,
P(100 µW) = 10 log (100 µW/1.0 mW) = 10 log (0.10) = - 10.0 dBm
P(150 µW) = 10 log (150 µW/1.0 mW) = 10 log (0.15) = - 8.24 dBm
(a) At 8 km we have the following power levels:
P
1300
(8 km) = - 8.2 dBm – (0.6 dB/km)(8 km) = - 13.0 dBm = 50 µW
P
1550
(8 km) = - 10.0 dBm – (0.3 dB/km)(8 km) = - 12.4 dBm = 57.5 µW
(b) At 20 km we have the following power levels:
P
1300
(20 km) = - 8.2 dBm – (0.6 dB/km)(20 km) = - 20.2 dBm = 9.55 µW
P
1550
(20 km) = - 10.0 dBm – (0.3 dB/km)(20 km) = - 16.0 dBm = 25.1 µW
3-3. From Eq. (3-1c) with P
out
= 0.45 P
in
α = (10/3.5 km) log (1/0.45) = 1.0 dB/km
3-4. (a) P
in
= P
out
10
αL/10
= (0.3 µW) 10
1.5(12)/10
= 18.9 µW
(b) P
in
= P
out
10
αL/10
= (0.3 µW) 10
2.5(12)/10
= 300 µW
3-5. With λ in Eqs. (3-2b) and (3-3) given in µm, we have the following representative
points for α
uv
and α
IR
:
2
λ λ (µ µm) α α
uv
α α
IR
0.5 20.3 --
0.7 1.44 --
0.9 0.33 --
1.2 0.09
2.2×10
-6
1.5 0.04 0.0072
2.0 0.02 23.2
3.0 0.009
7.5×10
4
3-6. From Eq. (3-4a) we have
α
scat
=

3

4
(n
2
−1)
2
k
B
T
f
β
T
=

3
3(0.63 µm)
4
(1.46)
2
−1
[ ]
2
(1.38×10
-16
dyne-cm/K)(1400 K)
×(6.8×10
-12
cm
2
/dyne)
= 0.883 km
-1
To change to dB/km, multiply by 10 log e = 4.343: α
scat
= 3.8 dB/km
From Eq. (3-4b):
α
scat
=

3

4
n
8
p
2
k
B
T
f
β
T
= 1.16 km
-1
= 5.0 dB/km
3-8. Plot of Eq. (3-7).
3-9. Plot of Eq. (3-9).
3-10. From Fig. 2-22, we make the estimates given in this table:
ν νm P
/P α α
ν νm
= α α
1 1
+ (α α
2 2
- α α
1 1
)P
/P 5 + 10
3
P
/P
01 0.02 3.0 + 0.02 5 + 20 = 25
11 0.05 3.0 + 0.05 5 + 50 = 55
21 0.10 3.0 + 0.10 5 + 100 = 105
02 0.16 3.0 + 0.16 5 + 160 = 165
31 0.19 3.0 + 0.19 5 + 190 = 195
12 0.31 3.0 + 0.31 5 + 310 = 315
3
3-11. (a) We want to solve Eq. (3-12) for α
gi
. With α = 2 in Eq. (2-78) and letting
∆ =
n
2
(0) − n
2
2
2n
2
(0)
we have
α(r) = α
1
+ (α
2
- α
1
)
n
2
(0) − n
2
(r)
n
2
(0) − n
2
2
= α
1
+ (α
2
- α
1
)
r
2
a
2
Thus
α
gi
=

α(r) p(r) r dr
0

p(r) r dr
0

= α
1
+

2
− α
1
)
a
2

exp −Kr
2
( )
r
3
dr
0

exp −Kr
2
( )
r dr
0

To evaluate the integrals, let x = Kr
2
, so that dx = 2Krdr. Then

exp −Kr
2
( )
r
3
dr
0

exp −Kr
2
( )
r dr
0

=

1
2K
2
e
− x
x dx
0

1
2K
e
−x
dx
0

=

1
K
1!
0!
=
1
K
Thus α
gi
= α
1
+

2
− α
1
)
Ka
2
(b) p(a) = 0.1 P
0
= P
0
e
− Ka
2
yields e
Ka
2
= 10.
From this we have Ka
2
= ln 10 = 2.3. Thus
α
gi
= α
1
+

2
− α
1
)
2.3
= 0.57α
1
+ 0.43α
2
3-12. With λ in units of micrometers, we have
4
n = 1 +
196.98
(13.4)
2
−(1.24 / λ)
2
¹
'
¹
¹
'
¹
1/ 2
To compare this with Fig. 3-12, calculate three representative points, for example,
λ = 0.2, 0.6, and 1.0 µm. Thus we have the following:
Wavelength λ λ Calculated n n from Fig. 3-12
0.2 µm
1.548 1.550
0.6 µm
1.457 1.458
1.0 µm
1.451 1.450

3-13. (a) From Fig. 3-13,

≈ 80 ps/(nm-km) at 850 nm. Therefore, for the LED we
have from Eq. (3-20)
σ
mat
L
=

σ
λ
= [80 ps/(nm-km)](45 nm) = 3.6 ns/km
For a laser diode,
σ
mat
L
= [80 ps/(nm-km)](2 nm) = 0.16 ns/km
(b) From Fig. 3-13,

mat

= 22 ps/(nm-km)
Therefore, D
mat
(λ) = [22 ps/(nm-km)](75 nm) = 1.65 ns/km
3-14. (a) Using Eqs. (2-48), (2-49), and (2-57), Eq. (3-21) becomes
b = 1 -
ua
V
|
.
`
,
2
= 1 -
u
2
a
2
u
2
a
2
+ w
2
a
2
=
w
2
u
2
+ w
2
=
β
2
− k
2
n
2
2
k
2
n
1
2
− β
2
+ β
2
− k
2
n
2
2
=
β
2
/ k
2
− n
2
2
n
1
2
− n
2
2
(b) Expand b as b =
(β/ k + n
2
)(β / k − n
2
)
(n
1
+ n
2
)(n
1
− n
2
)
Since n
2
< β/k < n
1
, let β/k = n
1
(1 - δ) where 0 < δ < ∆ << 1. Thus,
5
β/ k + n
2
n
1
+ n
2
=
n
1
(1 −δ) + n
2
n
1
+ n
2
= 1 -
n
1
n
1
+ n
2
δ
Letting n
2
= n
1
(1 - ∆) then yields
β/ k + n
2
n
1
+ n
2
= 1 -
δ
2 − ∆
≈ 1 since
δ
2 − ∆
<< 1
Therefore, b ≈
β/ k − n
2
n
1
− n
2
or β = k[bn
1
∆ + n
2
]
From n
2
= n
1
(1 - ∆) we have
n
1
= n
2
(1 - ∆)
-1
= n
2
(1 + ∆ + ∆
2
+ ...) ≈ n
2
(1 + ∆)
Therefore, β = k[b n
2
(1 + ∆)∆ + n
2
] ≈ k n
2
(b∆ + 1)
3-16. The time delay between the highest and lowest order modes can be found from the
travel time difference between the two rays shown here.
The travel time of each ray is given by
sin φ =
x
s
=
n
2
n
1
=
n
1
(1 − ∆)
n
1
= (1 - ∆)
The travel time of the highest order ray is thus
T
max
=

n
1
c
s
L
x
|
.
`
,

]
]
]
=
n
1
L
c
1
1 − ∆
For the axial ray the travel time is T
min
=
Ln
1
c
ϕ
θ
a
s
x
6
Therefore
T
min
- T
max
=
Ln
1
c
1
1 − ∆
−1
|
.
`
,
=
Ln
1
c

1 − ∆

Ln
1

c
3-17. Since n
2
· n
1
1 − ∆ ( ), we can rewrite the equation as
σ
mod
L
·
n
1

c
1 −
π
V
|
.
`
,
where the first tern is Equation (3-30). The difference is then given by the factor
1 −
π
V
·1 −
πλ
2a
1
n
1
2
− n
2
2
( )
1/ 2
≈1 −
πλ
2a
1
n
1
2∆
At 1300 nm this factor is 1 −
π(1.3)
2 62.5 ( )
1
1.48 2(0.015)
·1 − 0.127 · 0.873
3-18. For ε = 0 and in the limit of α → ∞ we have
C
1
= 1, C
2
=
3
2
,
α
α + 1
= 1,
α + 2
3α + 2
=
1
3
,
α +1
2α +1
·
1
2
,
and
(α + 1)
2
(5α + 2)(3α + 2)
·
1
15
Thus Eq. (3-41) becomes
σ
int er mod al
=
Ln
1

2 3c
1+ 3∆ +
12
5

2
|
.
`
,
1/ 2

Ln
1

2 3c
3-19. For ε = 0 we have that α = 2(1 -
6
5
∆). Thus C
1
and C
2
in Eq. (3-42) become
(ignoring small terms such as ∆
3
, ∆
4
, ...)
C
1
=
α − 2
α + 2

·
2 1 −
6
5

|
.
`
,
− 2
2 1 −
6
5

|
.
`
,
+ 2
·

3
5

1−
3
5

≈ −
3
5
∆ 1+
3
5

|
.
`
,
7
C
2
=
3α − 2
2(α + 2)
·
3 2 1 −
6
5

|
.
`
,

]
]
]
− 2
2 2 1−
6
5

|
.
`
,
+ 2

]
]
]
·
1 −
9
5

2 1 −
3
5

|
.
`
,
Evaluating the factors in Eq. (3-41) yields:
(a) C
1
2

9
25

2
(b)
4C
1
C
2
(α + 1)∆
2α +1
=
4∆ −
3
5

|
.
`
,
1 −
9
5

|
.
`
,
2 1−
6
5

|
.
`
,
+1

]
]
]
1 −
3
5

|
.
`
,
2 1−
3
5

|
.
`
,
4 1 −
6
5

|
.
`
,
+1

]
]
]
=

18∆
1 −
11
∆ +
18
25
|
.
`
,
5 1
5

|
.
`
,
2
24
25

|
`
,
18
25

2
(c)
16∆
2
C
2
2
(α+1)
2
(5α + 2)(3α + 2)
=
16∆
2
1−
9
5

|
.
`
,
2
2(1 −
6
5
∆) +1

]
]
2
4 1 −
3
5

|
.
`
,
2
10(1 −
6
5
∆) + 2

]
]
6(1 −
6
5
∆) + 2

]
]
=
16∆
2
1−
9
5

|
.
`
,
2
9 1 −
4
5

|
.
`
,
2
96(1 − ∆)(1 −
9
10
∆)4 1−
3
5

|
.
`
,
2

9
24

2
Therefore,
σ
int er mod al
=
Ln
1

2c
α
α +1
|
.
`
,
α+ 2
3α + 2
|
.
`
,
1/ 2
9
25

2

18
25

2
+
9
24

2
|
.
`
,
1/ 2
=
Ln
1

2
2c
2(1 −
6
5
∆)
2(1 −
6
5
∆) +1

]
]
2(1−
6
5
∆) + 2
6(1−
6
5
∆) + 2

]
]
]
]
1/ 2
3
10 6

n
1

2
L
20 3c
8
3-20. We want to plot Eq. (3-30) as a function of σ
λ
, where σ
int er mod al
and
σ
int ra mod al
are given by Eqs. (3-41) and (3-45). For ε = 0 and α = 2, we have C
1
=
0 and C
2
= 1/2. Since σ
int er mod al
does not vary with σ
λ
, we have
σ
int er mod al
L
=
N
1

2c
α
α + 2
|
.
`
,
α + 2
3α + 2
|
.
`
,
1/ 2
4∆C
2
(α +1)
(5α + 2)(3α + 2)
= 0.070 ns/km
With C
1
= 0 we have from Eq. (3-45)
σ
int ra mod al
=
1
c
σ
λ
λ
−λ
2
d
2
n
1

2
|
.

`
,
=

0.098σ
λ
ns / km at 850 nm
1.026 ×10
−2
σ
λ
ns / km at 1300 nm
¹
'
¹
3-21. Using the same parameter values as in Prob. 3-18, except with ∆ = 0.001, we have
from Eq. (3-41) σ
int er mod al
/L = 7 ps/km, and from Eq. (3.45)
σ
int ra mod al
L
=

0.098σ
λ
ns / km at 850 nm
0.0103σ
λ
ns / km at 1300 nm
¹
'
¹
The plot of
σ
L
=
1
L
σ
int er
2
+ σ
int ra
2
( )
1/ 2
vs σ
λ
:
3-22. Substituting Eq. (3-34) into Eq. (3-33)
τ
g
=
L
c

dk
=
L
c
1

2kn
1
2
+ 2k
2
n
1
dn
1
dk
¹
'
¹
- 2
α + 2
α
m
a
2
|
.
`
,
α
α+ 2
2
α + 2
n
1
2
k
2
∆ ( )
2
α+2
−1
× ∆ 2k
2
n
1
dn
1
dk
+ 2kn
1
2
+
n
1
2
k
2

d∆
dk

]
]
]
¹
'
¹
=
L
c
kn
1
β
N
1

4∆
α + 2
α + 2
α
m
a
2
1
n
1
2
k
2

|
.

`
,

α
α +2
N
1
+
n
1
k
2∆
d∆
dk
|
.
`
,

]
]
]
]
9
=
LN
1
c
kn
1
β
1−
4∆
α + 2
m
M
|
.
`
,
α
α +2
1 +
ε
4
|
.
`
,

]
]
]
with N
1
= n
1
+ k
dn
1
dk
and where M is given by Eq. (2-97) and ε is defined in Eq.
(3-36b).
3-23. From Eq. (3-39), ignoring terms of order ∆
2
,
λ

=
L
c
dN
1

1 +
α − ε − 2
α + 2
m
M
|
.
`
,
α
α+ 2

]
]
]
+
LN
1
c
α − ε − 2
α + 2
d

m
M
|
.
`
,
α
α +2

]
]
]
where
N
1
= n
1
- λ
dn
1

and M =
α
α + 2
a
2
k
2
n
1
2

(a)
dN
1

=
d

n
1
− λ
dn
1

|
.
`
,
= - λ
d
2
n
1

2
Thus ignoring the term involving ∆
d
2
n
1

2
, the first term in square brackets
becomes -
L
c
λ
2

d
2
n
1

2
(b)
d

m
M
|
.
`
,
α
α+ 2

]
]
]
= m
α
α+ 2
d∆

1
M
|
.
`
,
α
α +2
+ ∆
−α
α + 2
dM

1
M
|
.
`
,
α
α +2
+1

]
]
]
(c)
dM

=
α
α + 2
a
2

d

k
2
n
1
2
∆ ( )
=
α
α + 2
a
2

d∆

k
2
n
1
2
+ 2k
2
∆n
1
dn
1

+ 2kn
1
2

dk

|
.
`
,
10
Ignoring
d∆

and
dn
1

terms yields
dM

=

α + 2
a
2
k
2
n
1
2
∆ −
1
λ
|
.
`
,
= -
2M
λ
so that
d

m
M
|
.
`
,
α
α+ 2

]
]
]
=

λ

α + 2
m
M
|
.
`
,
α
α +2
. Therefore
λ

= -
L
c
λ
2

d
2
n
1

2
+
LN
1
c
α − ε − 2
α + 2
2α∆
α + 2
m
M
|
.
`
,
α
α +2
3-24. Let a = λ
2

d
2
n
1

2
; b = N
1
C
1

α + 2
; γ =
α
α + 2
Then from Eqs. (3-32), (3-43), and (3-44) we have
σ
int ra mod al
2
=

L
2
σ
λ
λ
|
.
`
,
2
1
M
λ

g

|
.
`
,
m· 0
M

2
=

L
c
|
.
`
,
2
σ
λ
λ
|
.
`
,
2
1
M
−a + b
m
M
|
.
`
,
γ

]
]
]
m· 0
M

2

L
c
|
.
`
,
2
σ
λ
λ
|
.
`
,
2
1
M
−a + b
m
M
|
.
`
,
γ

]
]
]
2
dm
0
M

=

L
c
|
.
`
,
2
σ
λ
λ
|
.
`
,
2

1
M
a
2
− 2ab
m
M
|
.
`
,
γ
+ b
2
m
M
|
.
`
,

]
]
]
dm
0
M

=
L
c
|
.
`
,
2
σ
λ
λ
|
.
`
,
2
a
2

2ab
γ +1
+
b
2
2γ +1

]
]
]
=
L
c
|
.
`
,
2
σ
λ
λ
|
.
`
,
2
−λ
2
d
2
n
1

2
|
.

`
,
2

11
- 2 λ
2
d
2
n
1

2
|
.

`
,
N
1
C
1

α
α + 1
+ (N
1
C
1
∆)
2

2
(α+ 2)(3α + 2)
]
]
]
3-25. Plot of Eq. (3-57).
3-26. (a)

D · λ − λ
0
( )S
0
· −50(0.07) · −3.5 ps/(nm − km)
(b)

D ·
1500(0.09)
4
1 −
1310
1500
|
.
`
,
4

]
]
]
· 14.1 ps/(nm − km)
3-27. (a) From Eq. (3-48)

σ
step
L
·
n
1

2 3 c
·
1.49(0.01)
2 3 (3 ×10
8
)
·14.4 ns / km
(b) From Eq. (3-47)

σ
opt
L
·
n
1

2
20 3 c
·
1.49(0.01)
2
20 3 (3 ×10
8
)
·14.3 ps/ km
(c) 3.5 ps/km
3-28. (a) From Eq. (3-29)

σ
mod
· T
max
− T
min
·
n
1
∆L
c
·
(1.49)(0.01)(5 ×10
3
m)
3 ×10
8
m/ s
· 248 ns
(b) From Eq. (3-48)

σ
step
·
n
1
∆L
2 3 c
·
248
2 3
· 71.7 ns
(c)

B
T
·
0.2
σ
step
· 2.8 Mb / s
(d)

B
T
⋅ L · (2.8 MHz)(5 km) ·13.9 MHz⋅ km
12
3-29. For α · 0.95α
opt
, we have
σ
int er
(α ≠ α
opt
)
σ
int er
(α · α
opt
)
·
(α − α
opt
)
∆(α + 2)
· −
0.05
(0.015)(1.95)
· −170%
For α ·1.05α
opt
, we have
σ
int er
(α ≠ α
opt
)
σ
int er
(α · α
opt
)
·
(α − α
opt
)
∆(α + 2)
· +
0.05
(0.015)(2.05)
· +163%
1
Problem Solutions for Chapter 4
4-1. From Eq. (4-1), n
i
= 2
.

|
,

` 2πk
B
T
h
2
3/2
(m
e
m
h
)
3/4
exp
.

|
,

`
-
E
g
2k
B
T

= 2
2π(1.38 ×10
−23
J/ K)
(6.63 ×10
−34
J.s)
2

]
]
]
3/ 2
T
3 / 2
(.068)(.56)(9.11×10
−31
kg)
2
[ ]
3/ 4
× exp −
(1.55 − 4.3 ×10
−4
T)eV
2(8.62 ×10
−5
eV/ K)T

]
]
]
= 4.15×10
14
T
3/2
exp −
1.55
2(8.62 ×10
−5
)T

]
]
]
exp
4.3 ×10
−4
2(8.62 ×10
−5
)

]
]
]
= 5.03×10
15
T
3/2
exp
.

|
,

`
-
8991
T

4-2. The electron concentration in a p-type semiconductor is n
P
= n
i
= p
i
Since both impurity and intrinsic atoms generate conduction holes, the total
conduction-hole concentration p
P
is
p
P
= N
A
+ n
i
= N
A
+ n
P

From Eq. (4-2) we have that n
P
= n
2
i
/p
P
. Then
p
P
= N
A
+ n
P
= N
A
+ n
2
i
/p
P
or p
2
P
- N
A
p
P
- n
2
i
= 0
so that
p
P
=
N
A
2
.

|
,

`
1 +
4n
2
i
N
2
A
+ 1
If n
i
<< N
A
, which is generally the case, then to a good approximation
p
P
≈ N
A
and n
P
= n
2
i
/p
P
≈ n
2
i
/N
A

4-3. (a) From Eq. (4-4) we have 1.540 = 1.424 + 1.266x + 0.266x
2
or
2
x
2
+ 4.759x - 0.436 = 0. Solving this quadratic equation yields (taking the plus
sign only)
x =
1
2
[ ] - 4.759 + (4.759)
2
+ 4(.436) = 0.090
The emission wavelength is λ =
1.240
1.540
= 805 nm.
(b) E
g
= 1.424 + 1.266(0.15) + 0.266(0.15)
2
= 1.620 eV, so that
λ =
1.240
1.620
= 766 nm
4-4. (a) The lattice spacings are as follows:
a(BC) = a(GaAs) = 5.6536
A
o
a(BD) = a(GaP) = 5.4512 A
o
a(AC) = a(InAs) = 6.0590 A
o

A
o
a(x,y) = xy 5.6536 + x(1-y) 5.4512 + (1-x)y 6.0590 + (1-x)(1-y)5.8696
= 0.1894y - 0.4184x + 0.0130xy + 5.8696
(b) Substituting a(xy) = a(InP) = 5.8696

A
o
into the expression for a(xy) in (a),
we have
y =
0.4184x
0.1894 - 0.0130x

0.4184x
0.1894
= 2.20x
(c) With x = 0.26 and y = 0.56, we have
E
g
= 1.35 + 0.668(.26) - 1.17(.56) + 0.758(.26)
2
+ 0.18(.56)
2
- .069(.26)(.56) - .322(.26)
2
(.56) + 0.03(.26)(.56)
2
= 0.956 eV
4-5. Differentiating the expression for E, we have
3
∆E ·
hc
λ
2
∆λ or ∆λ ·
λ
2
hc
∆E
For the same energy difference ∆E, the spectral width ∆λ is proportional to the
wavelength squared. Thus, for example,
∆λ
1550
∆λ
1310
·
1550
1310
|
.
`
,
2
· 1.40
4-6. (a) From Eq. (4-10), the internal quantum efficiency is
η
int
·
1
1 + 25/ 90
· 0.783, and from Eq. (4-13) the internal power level is

P
int
· (0.783)
hc(35 mA)
q(1310 nm)
· 26 mW
(b) From Eq. (4-16),

P ·
1
3.5 3.5 + 1 ( )
2
26 mW · 0.37 mW
4-7. Plot of Eq. (4-18). Some representative values of P/P
0
are given in the table:
f in MHz P/P
0
1 0.999
10 0.954
20 0.847
40 0.623
60 0.469
80 0.370
100 0.303
4-8. The 3-dB optical bandwidth is found from Eq. (4-21). It is the frequency f at
which the expression is equal to -3; that is,
10 log
1
1 + 2πfτ ( )
2
[ ]
1/ 2
|
.

`
,

· −3
4
With a 5-ns lifetime, we find

f ·
1
2π 5 ns ( )
10
0.6
−1
( )· 9.5 MHz
4-9. (a) Using Eq. (4-28) with Γ · 1
g
th
=

1
0.05 cm
ln
.

|
,

`
1
0.32
2
+ 10 cm
-1
= 55.6 cm
-1
(b) With R
1
= 0.9 and R
2
= 0.32,
g
th
=
1
0.05 cm
ln

]
]
]
1
0.9(0.32)
+ 10 cm
-1
= 34.9 cm
-1
(c) From Eq. (4-37) η
ext
= η
i
(g
th
- α )/g
th
;
thus for case (a): η
ext
= 0.65(55.6 - 10)/55.6 = 0.53
For case (b): η
ext
= 0.65(34.9 - 10)/34.9 = 0.46
4-10. Using Eq. (4-4) to find E
g
and Eq. (4-3) to find λ, we have for x = 0.03,
λ =
1.24
E
g
=
1.24
1.424 + 1.266(0.3) + 0.266(0.3)
2
= 1.462 µm
From Eq. (4-38)
η
ext
= 0.8065 λ(µm)
dP(mW)
dI(mA)

Taking dI/dP = 0.5 mW/mA, we have η
ext
= 0.8065 (1.462)(0.5) = 0.590
4-11. (a) From the given values, D = 0.74, so that Γ
T
= 0.216
Then n
eff
2
= 10.75 and W = 3.45, yielding Γ
L
= 0.856
(b) The total confinement factor then is Γ = 0.185
4-12. From Eq. (4-46) the mode spacing is
∆λ =
λ
2
2Ln
=
(0.80 µm)
2
2(400 µm)(3.6)
= 0.22 nm
Therefore the number of modes in the range 0.75-to-0.85 µm is
5
.85 −.75
.22 ×10
−3
=
.1
.22
×10
3
= 455 modes
4-13. (a) From Eq. (4-44) we have g(λ) = (50 cm
-1
) exp

]
]
]
-
(λ - 850 nm)
2
2(32 nm)
2

= (50 cm
-1
) exp

]
]
]
-
(λ - 850)
2
2048

(b) On the plot of g(λ) versus λ, drawing a horizontal line at g(λ) = α
t
= 32.2 cm
-1
shows that lasing occurs in the region 820 nm < λ < 880 nm.
(c) From Eq. (4-47) the mode spacing is
∆λ =
λ
2
2Ln
=
(850)
2
2(3.6)(400 µm)
= 0.25 nm
Therefore the number of modes in the range 820-to-880 nm is
N =
880 - 820
0.25
= 240 modes
4-14. (a) Let N
m
= n/λ =
m
2L
be the wave number (reciprocal wavelength) of mode m.
The difference ∆N between adjacent modes is then
∆N = N
m
- N
m-1
=
1
2L
(a-1)
We now want to relate ∆N to the change ∆λ in the free-space wavelength. First
differentiate N with respect to λ:
dN

=
d

.

|
,

`
n
λ
=
1
λ

dn

-
n
λ
2
= -
1
λ
2

.

|
,

`
n - λ
dn

Thus for an incremental change in wavenumber ∆N, we have, in absolute values,
∆N =
1
λ
2

.

|
,

`
n - λ
dn

∆λ (a-2)
6
Equating (a-1) and (a-2) then yields ∆λ =
λ
2
2L
.

|
,

`
n - λ
dn

(b) The mode spacing is ∆λ =
(.85 µm)
2
2(4.5)(400 µm)
= 0.20 nm
4-15. (a) The reflectivity at the GaAs-air interface is
R
1
= R
2
=
.

|
,

`
n-1
n+1
2
=
.

|
,

`
3.6-1
3.6+1
2
= 0.32
Then J
th
=
1
β
α +
1
2L
ln
1
R
1
R
2

]
]
]
= 2.65×10
3
A/cm
2
Therefore
I
th
= J
th
× l × w = (2.65×10
3
A/cm
2
)(250×10
-4
cm)(100×10
-4
cm) = 663 mA
(b) I
th
= (2.65×10
3
A/cm
2
)(250×10
-4
cm)(10×10
-4
cm) = 66.3 mA
4-16. From the given equation

∆E
11
·1.43 eV+
6.6256 × 10
−34
J⋅ s ( )
2
8 5 nm ( )
2
1
6.19 × 10
−32
kg
+
1
5.10 ×10
−31
kg
|
.

`
,
·1.43 eV+ 0.25 eV ·1.68 eV
Thus the emission wavelength is λ = hc/E = 1.240/1.68 = 739 nm.
4-17. Plots of the external quantum efficiency and power output of a MQW laser.
4-18. From Eq. (4-48a) the effective refractive index is
n
e
=

B

=
2(1570 nm)
2(460 nm)
= 3.4
Then, from Eq. (4-48b), for m = 0
7
λ = λ
B
t
λ
2
B
2n
e
L

.

|
,

`
1
2
= 1570 nm t
(1.57 µm)(1570 nm)
4(3.4)(300 µm)
= 1570 nm t 1.20 nm
Therefore for m = 1, λ = λ
B
t 3(1.20 nm) = 1570 nm t 3.60 nm
For m = 2, λ = λ
B
t 5(1.20 nm) = 1570 nm t 6.0 nm
4-19. (a) Integrate the carrier-pair-density versus time equation from time 0 to t
d
(time
for onset of stimulated emission). In this time the injected carrier pair density
changes from 0 to n
th
.

t
d
· dt
0
t
d

·
1
J
qd

n
τ

0
n
th

dn · −τ
J
qd

n
τ
|
.

`
,

n· 0
n· nth
· τ ln
J
J − J
th
|
.

`
,

where J = I
p
/A and J
th
= I
th
/A. Therefore t
d
= τ ln
.

|
,

`
I
p
I
p
- I
th

(b) At time t = 0 we have n = n
B
, and at t = t
d
we have n = n
th
. Therefore,
t
d
=

0
t
d
dt =

1
J
qd

n
τ
dn
nB
n
th

= τ ln
J
qd

n
B
τ
J
qd

n
th
τ
|
.

`
,

In the steady state before a pulse is applied, n
B
= J
B
τ/qd. When a pulse is applied,
the current density becomes I/A = J = J
B
+ J
p
= (I
B
+ I
p
)/A
Therefore, t
d
= τ ln
.

|
,

`
I - I
B
I - I
th
= τ ln
.

|
,

`
I
p
I
p +
I
B
- I
th

4-20. A common-emitter transistor configuration:
4-21. Laser transmitter design.
8
4-22. Since the dc component of x(t) is 0.2, its range is -2.36 < x(t) < 2.76. The power
has the form P(t) = P
0
[1 + mx(t)] where we need to find m and P
0
. The average
value is
< > P(t) = P
0
[1 + 0.2m] = 1 mW
The minimum value is
P(t) = P
0
[1 - 2.36m] ≥ 0 which implies m ≤
1
2.36
= 0.42
Therefore for the average value we have < > P(t) = P
0
[1 + 0.2(0.42)] ≤ 1 mW,
which implies
P
0
=
1
1.084
= 0.92 mW so thatP(t) = 0.92[1 + 0.42x(t)] mW and
i(t) = 10 P(t) = 9.2[1 + 0.42x(t)] mA
4-23. Substitute x(t) into y(t):
y(t) = a
1
b
1
cos ω
1
t + a
1
b
2
cos ω
2
t
+ a
2
(b
1
2
cos
2
ω
1
t + 2b
1
b
2
cos ω
1
t cos ω
2
t + b
2
2
cos
2
ω
2
t)
+ a
3
(b
1
3
cos
3
ω
1
t + 3b
1
2
b
2
cos
2
ω
1
t cos ω
2
t + 3b
1
b
2
2
cos ω
1
t cos
2
ω
2
t+ b
2
3
cos
3
ω
2
t)
+a
4
(b
1
4
cos
4
ω
1
t + 4b
1
3
b
2
cos
3
ω
1
t cos ω
2
t + 6b
1
2
b
2
2
cos
2
ω
1
t cos
2
ω
2
t
+ 4b
1
b
2
3
cos ω
1
t cos
3
ω
2
t + b
2
4
cos
4
ω
2
t)
Use the following trigonometric relationships:
i) cos
2
x =
1
2
(1 + cos 2x)
ii) cos
3
x =
1
4
(cos 3x + 3cos x)
iii) cos
4
x =
1
8
(cos 4x + 4cos 2x + 3)
iv) 2cos x cos y = cos (x+y) + cos (x-y)
v) cos
2
x cos y =
1
4
[cos (2x+y) + 2cos y + cos (2x-y)]
9
vi) cos
2
x cos
2
y =
1
4
[1 + cos 2x+ cos 2y +
1
2
cos(2x+2y) +
1
2
cos(2x-2y)]
vii) cos
3
x cos y =
1
8
[cos (3x+y) + cos (3x-y) + 3cos (x+y) + 3cos (x-y)]
then
y(t) =
1
2

]
]
]
a
2
b
2
1
+ a
2
b
2
2
+
3
4
a
4
b
4
1
+ 3a
4
b
2
1
b
2
2
+
3
4
a
4
b
4
2

constant
terms

+
3
4

]
]
a
3
b
3
1
+ 2a
3
b
1
b
2
2
cos ω
1
t +
3
4
a
3

]
]
b
3
2
+ 2b
2
1
b
2
cos ω
2
t
fundamental
terms

+
b
2
1
2

]
]
a
2
+ a
4
b
2
1
+ 3a
4
b
2
2
cos 2ω
1
t +
b
2
2
2

]
]
a
2
+ a
4
b
2
2
+ 3a
4
b
2
1
cos 2ω
2
t
2nd-order harmonic terms
+
1
4
a
3
b
3
1
cos 3ω
1
t +
1
4
a
3
b
3
2
cos 3ω
2
t 3rd-order harmonic terms
+
1
8
a
4
b
4
1
cos 4ω
1
t +
1
8
a
4
b
4
2
cos 4ω
2
t 4th-order harmonic terms
+

]
]
a
2
b
1
b
2
+
3
2
a
4
b
3
1
b
2
+
3
2
a
4
b
1
b
3
2
[ ]
cos (ω
1

2
)t + cos (ω
1

2
)t
2nd-order intermodulation terms
+
3
4
a
3
b
2
1
b
2

[ ]
cos(2ω
1

2
)t + cos(2ω
1

2
)t +
3
4
a
3
b
1
b
2
2

[ ]
cos(2ω
2

1
)t + cos(2ω
2

1
)t
3rd-order intermodulation terms
+
1
2
a
4
b
3
1
b
2

[ ]
cos(3ω
1

2
)t + cos(3ω
1

2
)t
+
3
4
a
4
b
2
1
b
2
2

[ ]
cos(2ω
1
+2ω
2
)t + cos(2ω
1
-2ω
2
)t
+
1
2
a
4
b
1
b
3
2

[ ]
cos(3ω
2

1
)t + cos(3ω
2

1
)t 4th-order intermodulation
terms
This output is of the form
y(t) = A
0
+ A
1

1
) cos ω
1
t + A
2

1
) cos 2ω
1
t + A
3

1
) cos 3ω
1
t
+ A
4

1
) cos 4ω
1
t + A
1

2
) cos ω
2
t + A
2

2
) cos 2ω
2
t
10
+ A
3

2
) cos 3ω
2
t + A
4

2
) cos 4ω
2
t + ∑
m

n

B
mn
cos(mω
1
+nω
2
)t
where A
n

j
) is the coefficient for the cos(nω
j
)t term.
4-24. From Eq. (4-58) P = P
0
e
-t/τ
m
where P
0
= 1 mW and τ
m
= 2(5×10
4
hrs) = 10
5
hrs.
(a) 1 month = 720 hours. Therefore:
P(1 month) = (1 mW) exp(-720/10
5
) = 0.99 mW
(b) 1 year = 8760 hours. Therefore
P(1 year) = (1 mW) exp(-8760/10
5
) = 0.92 mW
(c) 5 years = 5×8760 hours = 43,800 hours. Therefore
P(5 years) = (1 mW) exp(-43800/10
5
) = 0.65 mW
4-25. From Eq. (4-60) τ
s
= K e
E
A
/k
B
T
or ln τ
s
= ln K + E
A
/k
B
T
where k
B
= 1.38×10
-23
J/°K = 8.625×10
-5
eV/°K
At T = 60°C = 333°K, we have
ln 4×10
4
= ln K + E
A
/[(8.625×10
-5
eV)(333)]
or 10.60 = ln K + 34.82 E
A (1)
At T = 90°C = 363°K, we have
ln 6500 = ln K + E
A
/[(8.625×10
-5
eV)(363)]
or 8.78 = ln K + 31.94 E
A (2)
Solving (1) and (2) for E
A
and K yields
E
A
= 0.63 eV and k = 1.11×10
-5
hrs
Thus at T = 20°C = 293°K
11
τ
s
= 1.11×10
-5
exp{0.63/[(8.625×10
-5
)(293)]} = 7.45×10
5
hrs
1
Problem Solutions for Chapter 5
5-3. (a) cos
L
30° = 0.5
cos 30° = (0.5)
1/L
= 0.8660
L = log 0.5/log 0.8660 = 4.82
(b) cos
T
15° = 0.5
cos 15° = (0.5)
1/T
= 0.9659
T = log 0.5/log 0.9659 = 20.0
5-4. The source radius is less than the fiber radius, so Eq. (5-5) holds:
P
LED-step
= π
2
r
2
s
B
0
(NA)
2
= π
2
(2×10
-3
cm)
2
(100 W/cm
2
)(.22)
2
= 191 µW
From Eq. (5-9)
P
= 2π
2
(2×10
-3
cm)
2
(100 W/cm
2
)(1.48)
2
(.01)

]
]
]
1 -
1
2

.
|
,
`
2
5
2
= 159 µW
5-5. Using Eq. (5-10), we have that the reflectivity at the source-to-gel interface is
R
s− g
·
3.600 −1.305
3.600 +1.305
|
.
`
,
2
· 0.219
Similarly, the relfectivity at the gel-to-fiber interface is
R
g− f
·
1.465 −1.305
1.465 +1.305
|
.
`
,
2
· 3.34 ×10
−3
The total reflectivity then is R · R
s− g
R
g−f
· 7.30 ×10
−4
The power loss in decibels is (see Example 5-3)

L · −10 log (1 − R) · −10 log (0.999) · 3.17 ×10
−3
dB
5-6. Substituting B(θ) = B
0
cos
m
θ into Eq. (5-3) for B(θ,φ), we have
2
P =

1
1

0
r
m

1

0

]
]
]

0
θ
0-max
cos
3
θ sin θ dθ dθ
s
r dr
Using

0
θ
0
cos
3
θ sin θ dθ =

0
θ
0
( ) 1 - sin
2
θ sin θ d(sin θ) =

0
sin θ
0
( ) x - x
3
dx
we have
P = 2π

1
1

0
r
m

1

0

]
]
] sin
2
θ
0-max
2
-
sin
4
θ
0-max
4

s
r dr
= π

1
1

0
r
m

1

0

]
]
]
NA
2
-
1
2
NA
4

s
r dr
=
π
2
[ ] 2NA
2
- NA
4

0
r
m
r dr

0

s

5-7. (a) Let a = 25 µm and NA = 0.16. For r
s
≥ a(NA) = 4 µm, Eq. (5-17) holds. For
r
s
≤ 4 µm, η = 1.
(b) With a = 50 µm and NA = 0.20, Eq. (5-17) holds for r
s
≥ 10 µm. Otherwise, η
= 1.
5-8. Using Eq. (5-10), the relfectivity at the gel-to-fiber interface is
3
R
g− f
·
1.485 −1.305
1.485 +1.305
|
.
`
,
2
· 4.16 ×10
−3
The power loss is (see Example 5-3)

L · −10 log (1 − R) · −10 log (0.9958) · 0.018 dB
When there is no index-matching gel, the joint loss is
R
a− f
·
1.485 −1.000
1.485 +1.000
|
.
`
,
2
· 0.038
The power loss is

L · −10 log (1 − R) · −10 log (0.962) · 0.17 dB
5-9. Shaded area = (circle segment area) - (area of triangle) =
1
2
sa -
1
2
cy
s = aθ = a [2 arccos (y/a)] = 2a arccos
.

|
,

`
d
2a

c = 2

]
]
]
a
2
-
.

|
,

`
d
2
2
1/2

Therefore
A
common
= 2(shaded area) = sa – cy = 2a
2
arccos
.

|
,

`
d
2a
- d

]
]
]
a
2
-
.

|
,

` d
2
4
1/2

5-10.
Coupling loss (dB) for
Given axial misalignments (µ µm)
(µ µm)
1 3 5 10
50/125 0.112 0.385 0.590 1.266
62.5/125 0.089 0.274 0.465 0.985
100/140 0.056 0.169 0.286 0.590
5-11. arccos x =
π
2
- arcsin x
For small values of x, arcsin x = x +
x
3
2(3)
+
x
5
2(4)(5)
+ ...
4
Therefore, for
d
2a
<< 1, we have arccos
d
2a

π
2
-
d
2a

Thus Eq. (5-30) becomes P
T
=
2
π
P
.

|
,

` π
2
-
d
2a
-
5d
6a
= P
.

|
,

`
1 -
8d
3πa

d/a P
T
/P (Eq.5-30) P
T
/P (Eq.5-31)
0.00 1.00 1.00
0.05 0.9576 0.9576
0.10 0.9152 0.9151
0.15 0.8729 0.8727
0.20 0.8309 0.8302
0.25 0.7890 0.7878
0.30 0.7475 0.7454
0.35 0.7063 0.7029
0.40 0.6656 0.6605
5-12. Plots of mechanical misalignment losses.
5-13. From Eq. (5-20) the coupling efficiency η
F
is given by the ratio of the number of
modes in the receiving fiber to the number of modes in the emitting fiber, where
the number of modes M is found from Eq. (5-19). Therefore
η
F
=
M
aR
M
aE
=
k
2
NA
2
(0)
.

|
,

`
1
2
-
1
α+2
a
2
R
k
2
NA
2
(0)
.

|
,

`
1
2
-
1
α+2
a
2
E
=
a
2
R
a
2
E

Therefore from Eq. (5-21) the coupling loss for a
R
≤ a
E
is L
F
= -10 log
.

|
,

`
a
2
R
a
2
E

5-14. For fibers with different NAs, where NA
R
< NA
E
L
F
= -10 log η
F
= -10 log
M
R
M
E
= -10 log
k
2
NA
2
R
(0)
.

|
,

` α
2α+4
a
2
k
2
NA
2
E
(0)
.

|
,

` α
2α+4
a
2

5
= -10 log

]
]
]
]
NA
2
R
(0)
NA
2
E
(0)

5-15. For fibers with different α values, where α
R
< α
E

L
F
= -10 log η
F
= -10 log
k
2
NA
2
(0)
.

|
,

` α
R

R
+ 4
a
2
k
2
NA
2
(0)
.

|
,

` α
E

E
+ 4
a
2
= -10 log

]
]
]
] α
R

E
+ 2)
α
E

R
+ 2)

5-16. The splice losses are found from the sum of Eqs. (5-35) through (5-37). First find
NA(0) from Eq. (2-80b).
For fiber 1: NA
1
(0) · n
1
2∆ · 1.46 2 0.01 ( ) · 0.206
For fiber 2: NA
2
(0) · n
1
2∆ · 1.48 2 0.015 ( ) · 0.256
(a) The only loss is that from index-profile differences. From Eq. (5-37)

L
1→2
α ( ) · −10 log
1.80(2.00 + 2)
2.00(1.80 + 2)
· 0.24 dB
(b) The losses result from core-size differences and NA differences.

L
2→1
(a) · −20 log
50
62.5
|
.
`
,
· 1.94 dB

L
2→1
(NA) · −20 log
.206
.256

]
]
· 1.89 dB
5-17. Plots of connector losses using Eq. (5-43).
5-18. When there are no losses due to extrinsic factors, Eq. (5-43) reduces to
6
L
SM;ff
= -10 log

]
]
]
4
.

|
,

`
W
1
W
2
+
W
2
W
1
2

For W
1
= 0.9W
2
, we then have L
SM;ff
= -10 log

]
]
]
4
4.0446
= - 0.0482 dB
5-19. Plot of Eq. (5-44).
5-20. Plot of the throughput loss.
1
Problem Solutions for Chapter 6
6-1. From Eqs. (6-4) and (6-5) with R
f
= 0, η = 1 - exp(-α
s
w)
To assist in making the plots, from Fig. P6-1, we have the following representative
values of the absorption coefficient:
λ λ (µ µm) α α
s
(cm
-1
)
.60
4.4×10
3
.65
2.9×10
3
.70
2.0×10
3
.75
1.4×10
3
.80
0.97×10
3
.85 630
.90 370
.95 190
1.00 70
6-2. I
p
= qA

0
w
G(x) dx = qA Φ
0
α
s

0
w
e

s
x
dx
= qA Φ
0

]
]
1 - e

s
w
= qA
P
0
(1 - R
f
)
hνA

]
]
1 - e

s
w

6-3. From Eq. (6-6), R =
ηq

=
ηqλ
hc
= 0.8044 ηλ ( in µm)
Plot R as a function of wavelength.
6-4. (a) Using the fact that V
a
≈ V
B
, rewrite the denominator as
1 -
.

|
,

`
V
a
- I
M
R
M
V
B
n
= 1 -
.

|
,

`
V
B
- V
B
+ V
a
- I
M
R
M
V
B
n

= 1 -
.

|
,

`
1 -
V
B
- V
a
+ I
M
R
M
V
B
n

Since
V
B
- V
a
+ I
M
R
M
V
B
<< 1, we can expand the term in parenthesis:
2
1 -
.

|
,

`
1 -
V
B
- V
a
+ I
M
R
M
V
B
n
≈ 1 -

]
]
]
1 -
n
( )
V
B
- V
a
+ I
M
R
M
V
B

=
n
( )
V
B
- V
a
+ I
M
R
M
V
B

nI
M
R
M
V
B

Therefore, M
0
=
I
M
I
p

V
B
n
( )
V
B
- V
a
+ I
M
R
M

V
B
nI
M
R
M

(b) M
0
=
I
M
I
p
=
V
B
nI
M
R
M
implies I
2
M
=
I
p
V
B
nR
M
, so that M
0
=
.

|
,

`
V
B
nI
p
R
M
1/2

6-5.
< >
i
2
s
(t) =
1
T

0
T
i
2
s
(t)dt =
ω

0
2π/ω
R
2
0
P
2
(t) dt (where T = 2π/ω),
=
ω

R
2
0
P
2
0

0
2π/ω
(1 + 2m cos ωt + m
2
cos
2
ωt) dt
Using

cos ωt dt ·
0
2 π / ω

1
ω
sin ωt
t · 0
t · 2π / ω
· 0
and

0
2π/ω
cos
2
ωt dt =
1
ω

1

0

.

|
,

`
1
2
+
1
2
cos 2x dx =
π
ω

we have
< >
i
2
s
(t) = R
2
0
P
2
0

.

|
,

`
1 +
m
2
2

6-6. Same problem as Example 6-6: compare Eqs. (6-13), (6-14), and (6-17).
(a) First from Eq. (6-6),

I
p
·
ηqλ
hc
P
0
· 0.593 µA
Then

σ
Q
2
· 2qI
p
B · 2 1.6 ×10
−19
C
( )
(0.593 µA)(150 ×10
6
Hz) · 2.84 ×10
−17
A
2
3
(b)

σ
DB
2
· 2qI
D
B· 2 1.6 ×10
−19
C
( )
(1.0 nA)(150 ×10
6
Hz) · 4.81×10
−20
A
2
(c)

σ
T
2
·
4k
B
T
R
L

4 1.38 ×10
−23
J / K
( )
(293 K)
500 Ω
150 ×10
6
Hz ( )· 4.85 ×10
−15
A
2
6-7. Using R
0
=
ηqλ
hc
= 0.58 A/W, we have from Eqs. (6-4), (6-11b), (6-15), and (6-
17)
.

|
,

`
S
N
Q
=
1
2
R
0
P
0
m ( )
2
M
2
2qI
p
BM
1/ 2
M
2
=
R
0
P
0
m
2
4qBM
1/2
= 6.565×10
12
P
0
.

|
,

`
S
N
DB
=
( )
R
0
P
0
m
2
4qI
D
BM
1/2
= 3.798×10
22
P
2
0

.

|
,

`
S
N
DS
=
( )
R
0
P
0
m
2
M
2
4qI
L
B
= 3.798×10
26
P
2
0

.

|
,

`
S
N
T
=
1
2

( )
R
0
P
0
m
2
M
2
4k
B
TB/R
L
= 7.333×10
22
P
2
0

where P
0
is given in watts. To convert P
0
= 10
-n
W to dBm, use 10 log
.

|
,

`
P
0
10
-3
=
10(3-n) dBm
6-8. Using Eq. (6-18) we have
S
N
=
1
2

( )
R
0
P
0
m
2
M
2
2qB(R
0
P
0
+ I
D
)M
5/2
+ 2qI
L
B + 4k
B
TB/R
L

=
1.215 ×10
−16
M
2
2.176 ×10
−23
M
5/ 2
+1.656 ×10
−19
The value of M for maximum S/N is found from Eq. (6-19), with x = 0.5:
M
optimum
= 62.1.
4
6-9. 0 =
d
dM

.

|
,

`
S
N
=
d
dM

]
]
]
]
1
2
I
2
p
M
2
2q(I
p
+ I
D
)M
2+x
+ 2qI
L
+ 4k
B
T/R
L

0 = I
2
p
M -
(2+x)M
1+x
2q(I
p
+ I
D
)
1
2
I
2
p
M
2
2q(I
p
+ I
D
)M
2+x
+ 2qI
L
+ 4k
B
T/R
L

Solving for M: M
2+x
opt
=
2qI
L
+ 4k
B
T/R
L
xq(I
p
+ I
D
)

6-10. (a) Differentiating p
n
, we have
∂p
n
∂x
=
1
L
p

.

|
,

`
p
n0
+ Be

s
w
e
(w-x)/L
p
- α
s
Be

s
x

2
p
n
∂x
2
= -
1
L
2
p

.

|
,

`
p
n0
+ Be

s
w
e
(w-x)/L
p
+ α
2
s
Be

s
x

Substituting p
n
and

2
p
n
∂x
2
into the left side of Eq. (6-23):
-
D
p
L
2
p

.

|
,

`
p
n0
+ Be

s
w
e
(w-x)/L
p
+ D
p
α
2
s
Be

s
x

+
1
τ
p

.

|
,

`
p
n0
+ Be

s
w
e
(w-x)/L
p
-
B
τ
p
e

s
x
+ Φ
0
α
s
e

s
x

=

]
]
]
B
.

|
,

`
D
p
α
2
s
-
1
τ
p
+ Φ
0
α
s
e

s
x

where the first and third terms cancelled because L
2
p
= D
p
τ
p
.
Substituting in for B:
Left side =

]
]
]
]
Φ
0
D
p

α
s
L
2
p
1 - α
2
s
L
2
p

.

|
,

`
D
p
α
2
s
-
1
τ
p
+ Φ
0
α
s
e

s
x

5
=
Φ
0
D
p

]
]
]
]
α
s
L
2
p
1 - α
2
s
L
2
p

.

|
,

`
α
2
s
L
p
- 1
τ
p
+ D
p
α
s
e

s
x

=
Φ
0
D
p

( )
-D
p
α
s
+ D
p
α
s
e

s
x
= 0 Thus Eq. (6-23) is satisfied.
(b) J
diff
= qD
p

∂p
n
∂x
|
.
`
,
x ·w
= qD
p

]
]
]
1
L
p

.

|
,

`
p
n0
+ Be

s
w
- α
s
Be

s
w

= qD
p
B
.

|
,

`
1
L
p
- α
s
e

s
w
+ qp
n0

D
p
L
p

= qΦ
0

.

|
,

`
α
s
L
2
p
1 - α
2
s
L
2
p

.

|
,

` 1 - α
s
L
p
L
p
e

s
w
+ qp
n0

D
p
L
p

= qΦ
0

α
s
L
p
1 + α
s
L
p
e

s
w
+ qp
n0

D
p
L
p

c) Adding Eqs. (6-21) and (6-25), we have
J
total
= J
drift
+ J
diffusion
= qΦ
0

]
]
]
.
|
,
`
1 - e

s
w
+
.

|
,

` α
s
L
p
1 + α
s
L
p
e

s
w
+ qp
n0

D
p
L
p

= qΦ
0

.

|
,

`
1 -
e

s
w
1 + α
s
L
p
e

s
w
+ qp
n0

D
p
L
p

6-11. (a) To find the amplitude, consider
.
|
,
`
J

tot
J
*
tot
sc
1/2
= qΦ
0
( ) S S*
1/2
where S =
1 - e
-jωt
d
jωt
d

We want to find the value of ωt
d
at which ( ) S S*
1/2
=
1
2
.
Evaluating ( ) S S*
1/2
, we have
6
( ) S S*
1/2
=

]
]
]
]
.

|
,

`
1 - e
-jωt
d
jωt
d .

|
,

`
1 - e
+jωt
d
-jωt
d
1/2

=

]
]
1 - .
|
,
`
e
+jωt
d
+ e
-jωt
d
+ 1
1/2
ωt
d
=
( )
2 - 2 cos ωt
d
1/2
ωt
d

=
[ ] ( )
1 - cos ωt
d
/2
1/2
ωt
d
/2
=
sin
.

|
,

` ωt
d
2
ωt
d
2
= sinc
.

|
,

` ωt
d
2

We want to find values of ωt
d
where ( ) S S*
1/2
=
1
2
.
x sinc x x sinc x
0.0 1.000 0.5 0.637
0.1 0.984 0.6 0.505
0.2 0.935 0.7 0.368
0.3 0.858 0.8 0.234
0.4 0.757 0.9 0.109
By extrapolation, we find sinc x = 0.707 at x = 0.442.
Thus
ωt
d
2
= 0.442 which implies ωt
d
= 0.884
(b) From Eq. (6-27) we have t
d
=
w
v
d
=
1
α
s
v
d
. Then
ωt
d
= 2πf
3-dB
t
d
= 2πf
3-dB
1
α
s
v
d

= 0.884 or
f
3-dB
= 0.884 α
s
v
d
/2π
6-12. (a) The RC time constant is
RC =

0
K
s
A
w
=
(10
4
Ω)(8.85 ×10
−12
F / m)(11.7)(5 ×10
−8
m
2
)
2 ×10
−5
m
= 2.59 ns
(b) From Eq. (6-27), the carrier drift time is
7
t
d
=
w
v
d
=
20 ×10
−6
m
4.4 ×10
4
m/ s
= 0.45 ns
c)
1
α
s
= 10
-3
cm = 10 µm =
1
2
w
Thus since most carriers are absorbed in the depletion region, the carrier diffusion
time is not important here. The detector response time is dominated by the RC
time constant.
6-13. (a) With k
1
≈ k
2
and k
eff
defined in Eq. (6-10), we have
(1) 1 -
k
1
(1 - k
1
)
1 - k
2
= 1 -
k
1
- k
2
1
1 - k
2
≈ 1 -
k
2
- k
2
1
1 - k
2
= 1 – k
eff
(2)
(1 - k
1
)
2
1 - k
2
=
1 - 2k
1
+ k
2
1
1 - k
2

1 - 2k
2
+ k
2
1
1 - k
2

=
1 - k
2
1 - k
2
-
k
2
- k
2
1
1 - k
2
= 1 - k
eff
Therefore Eq. (6-34) becomes Eq. (6-38):
F
e
= k
eff
M
e
+ 2(1 - k
eff
) -
1
M
e
(1 - k
eff
) = k
eff
M
e
+
.

|
,

`
2 -
1
M
e
(1 - k
eff
)
(b) With k
1
≈ k
2
and k
'
eff
defined in Eq. (6-40), we have
(1)
k
2
(1 - k
1
)
k
2
1
(1 - k
2
)

k
2
- k
2
1
k
2
1
(1 - k
2
)
= k
'
eff

(2)
(1 - k
1
)
2
k
2
k
2
1
(1 - k
2
)
=
k
2
- 2k
1
k
2
+ k
2
k
2
1
k
2
1
(1 - k
2
)

.
|
,
`
k
2
- k
2
1
-
.
|
,
`
k
2
1
- k
2
k
2
1
k
2
1
(1 - k
2
)
= k
'
eff
- 1
8
Therefore Eq. (6-35) becomes Eq. (6-39): F
h
= k
'
eff
M
h
-
.

|
,

`
2 -
1
M
h
(k
'
eff
- 1)
6-14. (a) If only electrons cause ionization, then β = 0, so that from Eqs. (6-36) and (6-
37), k
1
= k
2
= 0 and k
eff
= 0. Then from Eq. (6-38)
F
e
= 2 -
1
M
e
≈ 2 for large M
e
(b) If α = β, then from Eqs. (6-36) and (6-37), k
1
= k
2
= 1 so that
k
eff
= 1. Then, from Eq. (6-38), we have F
e
= M
e
.
1
Problem Solutions for Chapter 7
7-1. We want to compare F
1
= kM + (1 - k) 2 −
1
M
|
.
`
,
and F
2
= M
x
.
For silicon, k = 0.02 and we take x = 0.3:
M F
1
(M) F
2
(M) % difference
9 2.03 1.93 0.60
25 2.42 2.63 8.7
100 3.95 3.98 0.80
For InGaAs, k = 0.35 and we take x = 0.7:
M F
1
(M) F
2
(M) % difference
4 2.54 2.64 3.00
9 4.38 4.66 6.4
25 6.86 6.96 1.5
100 10.02 9.52 5.0
For germanium, k = 1.0, and if we take x = 1.0, then F
1
= F
2
.
7-2. The Fourier transform is
h
B
(t) =

H
B
−∞

(f )e
j2πft
df = R

e
j2πft
1 + j2πfRC
df
−∞

Using the integral solution from Appendix B3:

e
jpx
(β + jx)
n
dx
−∞

=
2π(p)
n−1
e
− βp
Γ(n)
for p > 0, we have
h
B
(t) =

1
2πC

e
j 2πft
1
2πRC
+ jf
|
.
`
,
df
−∞

=
1
C
e
-t/RC
7-3. Part (a):

h
p
−∞

(t) dt =
1
αT
b

-αT
b
/2
αT
b
/2
dt =
1
αT
b
.

|
,

` αT
b
2
+
αT
b
2
= 1
2
Part (b):

h
p
−∞

(t) dt =

1

1
αT
b
exp −
t
2
2(αT
b
)
2

]
]
]
−∞

dt
=
1

1
αT
b
π αT
b
2 = 1 (see Appendix B3 for integral solution)
Part (c):

h
p
−∞

(t) dt =

1
αT
b
exp −
t
αT
b

]
]
]
dt
0

= - e
−∞
− e
−0
[ ]
= 1
7-4. The Fourier transform is
F[p(t)*q(t)] =

p(t)* q(t)e
− j2 πft
dt
−∞

=

q(x) p(t − x) e
− j2πft
dt dx
−∞

−∞

=

q(x) e
− j2πfx
p(t − x) e
− j2πf( t − x)
dt dx
−∞

−∞

=

q(x) e
− j2πfx
dx p(y) e
− j 2πfy
dy
−∞

−∞

where y = t - x
= F[q(t)] F[p(t)] = F[p(t)] F[q(t)] = P(f) Q(f)
7-5. From Eq. (7-18) the probability for unbiased data (a = b = 0) is
P
e
=
1
2
P
0
(v
th
) + P
1
(v
th
) [ ].
Substituting Eq. (7-20) and (7-22) for P
0
and P
1
, respectively, we have
P
e
=

1
2

1
2πσ
2
e
−v
2
/ 2σ
2
dv
V/ 2

+ e
− (v−V)
2
/ 2σ
2
dv
−∞
V / 2

]
]
]
In the first integral, let x = v/ 2σ
2
so that dv = 2σ
2
dx.
In the second integral, let q = v-V, so that dv = dq. The second integral then
becomes
3
e
−q
2
/ 2σ
2
dq
−∞
V / 2 −V

=

2
e
− x
2
dx
−∞
−V / 2 2σ
2

where x = q/ 2σ
2
Then
P
e
=

1
2

2
2πσ
2
e
− x
2
dx
V/ 2 2σ
2

+ e
− x
2
dx
−∞
− V/ 2 2σ
2

]
]
]
=

1
2 π
e
−x
2
dx
−∞

− 2 e
−x
2
dx
0
V / 2 2σ
2

]
]
]
Using the following relationships from Appendix B,

e
− p
2
x
2
dx
−∞

=
π
p
and

2
π
e
−x
2
dx
0
t

= erf(t), we have
P
e
=
1
2
1− erf
V
2σ 2
|
.
`
,

]
]
]
7-6. (a) V = 1 volt and σ = 0.2 volts, so that
V

= 2.5. From Fig. 7-6 for
V

= 2.5,
we find P
e
≈ 7×10
-3
errors/bit. Thus there are (2×10
5
bits/second)(7×10
-3
errors/bit) = 1400 errors/second, so that
1
1400 errors/second
= 7×10
-4
seconds/error
(b) If V is doubled, then
V

= 5 for which P
e
≈ 3×10
-7
errors/bit from Fig. 7-6.
Thus
1
(2 ×10
5
bits / sec ond)(3 ×10
−7
errors / bit )
= 16.7 seconds/error
7-7. (a) From Eqs. (7-20) and (7-22) we have
P
0
(v
th
) =

1
2πσ
2
e
− v
2
/ 2σ
2
dv
V / 2

=
1
2
1− erf
V
2σ 2
|
.
`
,

]
]
]
and
4
P
1
(v
th
) =
1
2πσ
2
e
−( v−V)
2
/ 2σ
2
dv
−∞
V / 2

=
1
2
1− erf
V
2σ 2
|
.
`
,

]
]
]
Then for V = V
1
and σ = 0.20V
1
P
0
(v
th
) =
1
2

]
]
]
1 - erf
.

|
,

`
1
2(.2) 2
=
1
2

]
]
]
1 - erf
.

|
,

` 2
0.8

=
1
2

[ ]
1 - erf( ) 1.768 =
1
2
( ) 1 - 0.987 = 0.0065
Likewise, for V = V
1
and σ = 0.24V
1
P
1
(v
th
) =
1
2

]
]
]
1 - erf
.

|
,

`
1
2(.24) 2
=
1
2

]
]
]
1 - erf
.

|
,

` 2
0.96

=
1
2

[ ]
1 - erf( ) 1.473 =
1
2
( ) 1 - 0.963 = 0.0185
(b) P
e
= 0.65(0.0185) + 0.35(0.0065) = 0.0143
(c) P
e
= 0.5(0.0185) + 0.5(0.0065) = 0.0125
7-8. From Eq. (7-1), the average number of electron-hole pairs generated in a time t is
N =
ηE

=
ηPt
hc/λ
=
0.65(25 ×10
−10
W)(1 ×10
−9
s)(1.3 ×10
−6
m)
(6.6256 ×10
−34
Js)(3 ×10
8
m / s)
= 10.6
Then, from Eq. (7-2)
P(n) = N
n

e
-N
n!
= (10.6)
5

e
-10.6
5!
=
133822
120
e
-10.6
= 0.05 = 5%
7-9. v
N
= v
out
- v
out
v
N
2
= v
out
− v
out
[ ]
2
= v
out
2
-2 v
out
2
+ v
out
2
= v
out
2
- v
out
2
7-10. (a) Letting φ = fT
b
and using Eq. (7-40), Eq. (7-30) becomes
5
B
bae
=

H
p
(0)
H
out
(0)
2
T
b
2

H
out
'
(φ)
H
p
'
(φ)
2
0

T
b
=
I
2
T
b
since H
p
(0) = 1 and H
out
(0) = T
b
. Similarly, Eq. (7-33) becomes
B
e
=

H
p
(0)
H
out
(0)
2

H
out
(f)
H
p
(f)
1 + j2πfRC ( )
2
0

df
=

1
T
b
2

H
out
(f)
H
p
(f)
2
0

1 + 4π
2
f
2
R
2
C
2
( )
df
=

1
T
b
2

H
out
(f)
H
p
(f)
2
0

df +

2πRC ( )
2
T
b
2

H
out
(f)
H
p
(f)
2
0

f
2
df =
I
2
T
b
+
(2πRC)
2
T
3
b
I
3
(b) From Eqs. (7-29), (7-31), (7-32), and (7-34), Eq. (7-28) becomes
< >
v
2
N
=
< >
v
2
s
+
< >
v
2
R
+
< >
v
2
I
+
< >
v
2
E

= 2q
< >
i
0
< > m
2
B
bae
R
2
A
2
+
4k
B
T
R
b
B
bae
R
2
A
2
+ S
I
B
bae
R
2
A
2
+ S
E
B
e
A
2
=
.

|
,

`
2q
< >
i
0
M
2+x
+
4k
B
T
R
b
+ S
I
B
bae
R
2
A
2
+ S
E
B
e
A
2
=
R
2
A
2
I
2
T
b

.

|
,

`
2q
< >
i
0
M
2+x
+
4k
B
T
R
b
+ S
I
+
S
E
R
2
+
(2πRCA)
2
T
3
b
S
E
I
3
7-11. First let

x · v − b
off
( )/ 2 σ
off
( ) with

dx · dv / 2 σ
off
( ) in the first part of Eq. (7-
49):
P
e
=

off
2πσ
off
exp −x
2
( )
v
th
− b
off

off

dx =

1
π
exp −x
2
( )
Q/ 2

dx
Similarly, let

y · −v + b
on
( )/ 2 σ
on
( )
so that
dy = -dv/ 2σ
on
( )
in the second part of Eq. (7-49):
6
P
e
= -

1
π
exp −y
2
( )

− v
th
+b
on

on

dy =

1
π
exp −y
2
( )
Q/ 2

dy
7-12. (a) Let x =
V
2 2 σ
=
K
2 2
For K = 10, x = 3.536. Thus
P
e
=
e
-x
2
2 π x
= 2.97×10
-7
errors/bit
(b) Given that P
e
= 10
-5
=
e
-x
2
2 π x
then e
-x
2

= 2 π 10
-5
x.
This holds for x ≈ 3, so that K = 2 2 x = 8.49.
7-13. Differentiating Eq. (7-54) with respect to M and setting db
on
/dM = 0, we have
db
on
dM
= 0
= -

Q(hν/ η)
Μ
2
M
2+ x
η

|
.
`
,
b
on
I
2
+ W

]
]
]
1/ 2
+ M
2+x
η

|
.
`
,
b
on
I
2
(1 − γ

]
]
]
+ W
¹
'
¹
¹
'
¹
1/ 2
+

Q(hν/ η)
M(hν/ η)
1
2
(2 + x)M
1+ x
b
on
I
2
M
2+ x
η

|
.
`
,
b
on
I
2
+ W

]
]
]
1/ 2
+
1
2
(2 + x)M
1+x
b
on
I
2
(1− γ)
M
2+x
η

|
.
`
,
b
on
I
2
(1− γ) + W

]
]
]
1/ 2
¹
'
¹
¹
¹
¹
'
¹
¹
¹
1 / 2
Letting G = M
2+x
.

|
,

` η

b
on
I
2
for simplicity, yields
¹
'
¹
¹
'
¹
( ) G + W
1/2
+ [ ] G(1-γ) + W
1/2
=
G
2
(2 + x)
¹
'
¹
¹
'
¹ 1
(G + W)
1/2
+
(1-γ)
[ ] G(1-γ) + W
1/2

Multiply by (G + W)
1/2
[ ] G(1-γ) + W
1/2
and rearrange terms to get
(G + W)
1/2

]
]
]
W -
Gx
2
(1-γ) = [ ] G(1-γ) + W
1/2

.

|
,

`
Gx
2
- W
7
Squaring both sides and collecting terms in powers of G, we obtain the quadratic
equation
G
2

]
]
] x
2
γ
4
(1-γ)

+ G

]
]
] x
2
γ
4
W(2-γ) - γW
2
(1+x) = 0
Solving this equation for G yields
G =
-
x
2
4
W(2-γ) t
¹¹
'
¹
¹
¹¹
'
¹
¹
x
4
16
W
2
(2-γ)
2
+ x
2
(1-γ)W
2
(1+x)
1
2
x
2
2
(1-γ)

=
W(2-γ)
2(1-γ)

¹
¹
'
¹
¹
¹
¹
'
¹
¹
1 -

]
]
]
1 + 16
.

|
,

`
1+x
x
2

1-γ
(2-γ)
2
1
2

where we have chosen the "+" sign. Equation (7-55) results by letting
G = M
2+x
opt
b
on
I
2
.

|
,

` η

7-14. Substituting Eq. (7-55) for M
2+x
b
on
into the square root expressions in Eq. (7-
54) and solving Eq. (7-55) for M, Eq. (7-54) becomes
b
on
=
Q
.

|
,

` hν
η
b
1/(2+x)
on

]
]
] hν
η

W(2-γ)
2I
2
(1-γ)
K
1/(2+x)

¹
¹
'
¹
¹
¹
¹
'
¹
¹

]
]
] W(2-γ)
2(1-γ)
K + W
1
2
+

]
]
]
W
2
(2-γ)K + W
1
2

Factoring out terms:
b
(1+x)/(2+x)
on
= Q
.

|
,

` hν
η
(1+x)/(2+x)
W
x/2(2+x)
I
1/(2+x)
2

×
¹
¹
'
¹
¹
¹
¹
'
¹
¹

]
]
] (2-γ)
2(1-γ)
K + 1
1
2
+

]
]
]
1
2
(2-γ)K + 1
1
2
÷

]
]
] (2-γ)K
2(1-γ)
1/(2+x)

or b
on
= Q
(2+x)/(1+x)
.

|
,

` hν
η
W
x/2(1+x)
I
1/(1+x)
2
L
8
7-15. In Eq. (7-59) we want to evaluate
lim
γ →1
(2 − γ)K
2(1− γ )L

]
]
]
1+x
2+x
=
lim
γ →1
(2 − γ)K
2(1− γ)

]
]
]
1+x
2+x lim
γ →1
1
L
|
.
`
,
1+x
2+x
Consider first
lim
γ →1
(2 − γ)K
2(1− γ)

]
]
]
=

lim
γ →1

(2 − γ)
2(1− γ )
−1 + 1 + B
(1− γ )
(2 − γ)
2

]
]
]
1
2
¹
'
¹
¹
¹
¹
'
¹
¹
¹
where B = 16(1+x)/x
2
. Since γ→1, we can expand the square root term in a
binomial series, so that
lim
γ →1
(2 − γ)K
2(1− γ)

]
]
]
=

lim
γ →1

(2 − γ)
2(1− γ )
−1 + 1 +
1
2
B
(1− γ)
(2 − γ)
2
− Order(1− γ )
2

]
]
]
¹
'
¹
¹
'
¹
=

lim
γ →1

B
4(2 − γ )
=
B
4
= 4
1+x
x
2

Thus
lim
γ →1
(2 − γ)K
2(1− γ)

]
]
]
1+x
2+x
=
.

|
,

`
4
1+x
x
2
1+x
2+x

Next consider, using Eq. (7-58)

lim
γ →1

1
L
|
.
`
,
1+ x
2+x
=
lim
γ →1
(2 − γ)K
2(1− γ)

]
]
]
1/(2+ x)
÷
(2 − γ)
2(1− γ)
K+1

]
]
]
1
2
+
1
2
(2 − γ )K+1

]
]
1
2
¹
'
¹
¹
¹
¹
'
¹
¹
¹

From the above result, the first square root term is

]
]
] (2-γ)
2(1-γ)
K + 1
1
2
=

]
]
]
4
1+x
x
2
+ 1
1
2
=
.

|
,

` x
2
+ 4x + 4
x
2
1
2
=
x + 2
x

From the expression for K in Eq. (7-55), we have that
lim
γ →1
K = 0, so that
9
lim
γ →1
1
2
(2 − γ )K+1

]
]
1
2
= 1 Thus
lim
γ →1
(2 − γ )
2(1− γ )
K+1

]
]
]
1
2
+
1
2
(2 − γ)K+1

]
]
1
2
¹
'
¹
¹
¹
¹
'
¹
¹
¹
=
x + 2
x
+ 1 =
2(1 + x)
x

Combining the above results yields
lim
γ →1
(2 − γ)K
2(1− γ )L

]
]
]
1+x
2+x
=
.

|
,

`
4
1+x
x
2
1+x
2+x

.

|
,

`
4
1+x
x
2
1
2+x

2(1 + x)
x
=
2
x

so that
lim
γ →1
M
1+x
opt
=
W
1/2
QI
2

2
x

7-16. Using H
'
p
(f) = 1 from Eq. (7-69) for the impulse input and Eq. (7-66) for the
raised cosine output, Eq. (7-41) yields
I
2
=

H
out
'
(φ)
2

0

=

1
2
H
out
'
(φ)
2

−∞

=
1
2

-
1-β
2
1-β
2
dφ +

1

1-β
2
1+β
2

1
8

]
]
]
1-sin
.

|
,

` πφ
β
-
π

2
dφ +

1

-
1+β
2
-
1-β
2

1
8

]
]
]
1-sin
.

|
,

` πφ
β
-
π

2

Letting y =
πφ
β
-
π

we have
I
2
=
1
2
(1 - β) +
β

-
π
2
π
2
[ ] 1 - 2sin y + sin
2
y dy
=
1
2
(1 - β) +
β

]
]
]
π - 0 +
π
2
=
1
2

.

|
,

`
1 -
β
4

10
Use Eq. (7-42) to find I
3
:
I
3
=

H
out
'
(φ)
2
φ
2

0

dφ ·
1
2
H
out
'
(φ)
2
φ
2
−∞

=
1
2

-
1-β
2
1-β
2
φ
2
dφ +

1

1-β
2
1+β
2

1
8

]
]
]
1-sin
.

|
,

` πφ
β
-
π

2
φ
2
dφ +

1

-
1+β
2
-
1-β
2

1
8

]
]
]
1-sin
.

|
,

` πφ
β
-
π

2
φ
2

Letting y =
πφ
β
-
π

I
3
=
1
3

.

|
,

` 1-β
2
3
+
β

1

-
π
2
π
2
[ ] 1 - 2sin y + sin
2
y

]
]
] β
2
y
2
π
2
+
βy
π
+
1
4
dy
=
1
3

.

|
,

` 1-β
2
3
+
β

]
]
]
]

1

-
π
2
π
2

.

|
,

` β
2
y
2
π
2
+
1
4
( ) 1 + sin
2
y dy -

π

-
π
2
π
2
y sin y dy
where only even terms in "y" are nonzero. Using the relationships

0
π
2
sin
2
y dy =
π
4
;

y sin ydy = -y cos y + sin y
and

x
2
sin x
2
dx =
x
3
6
-
.

|
,

` x
2
4
-
1
8
sin 2x -
x cos 2x
4

we have
I
3
=
1
3

.

|
,

` 1-β
2
3
+
β

¹
'
¹
¹
'
¹
β
2
π
2

]
]
]
1
3
.

|
,

` π
2
3
+
1
6
.

|
,

` π
2
3
+
π
8
+
1
4
.

|
,

` π
2
+
π
4
-

π
(1)
11
=
β
3
16

.

|
,

`
1
π
2
-
1
6
- β
2

.

|
,

`
1
π
2
-
1
8
-
β
32
+
1
24

7-17. Substituting Eq. (7-64) and (7-66) into Eq. (7-41), with s
2
= 4π
2
α
2
and β = 1, we
have
I
2
=

1

0

¹
¹
¹
¹
¹
¹ H
'
out
(φ)
H
'
p
(φ)
2
dφ =
1
4

1

0
1
e
s
2
φ
2

]
]
]
1-sin
.

|
,

`
πφ -
π
2
2

=

1

0
1
e
s
2
φ
2
.

|
,

` 1 + cos πφ
2
2
dφ =

1

0
1
e
s
2
φ
2
cos
4
.

|
,

` πφ
2

Letting x =
πφ
2
yields I
2
=
2
π

0
π
2
e
16α
2
x
2
cos
4
x dx
Similarly, using Eqs. (7-64) and (7-66), Eq. (7-42) becomes
I
3
=

1

0
1
e
s
2
φ
2
cos
4
.

|
,

` πφ
2
φ
2
dφ =

2
π
|
.
`
,
3
x
2
e
16α
2
x
2
cos
4
x
0
π
2

dx
7-18. Plot of I
2
versus α for a gaussian input pulse:
7-19. Plot of I
3
versus α for a gaussian input pulse:
7-20. Consider first
lim
γ →1
K:
12
lim
γ →1
K =
lim
γ →1
−1+ 1+16
1+ x
x
2
|
.
`
,
1− γ
(2 − γ)
2

]
]
]
1
2
¹
'
¹
¹
¹
¹
'
¹
¹
¹
= -1 + 1 = 0
Also
lim
γ →1
(1− γ) = 0. Therefore from Eq. (7-58)
lim
γ →1
L =
lim
γ →1
2(1− γ)
(2 − γ)K

]
]
]
1/(1+ x)

¹
¹
'
¹
¹
¹
¹
'
¹
¹

]
]
] (2-γ)
2(1-γ)
K + 1
1
2
+ 1
2+x
1+x

Expanding the square root term in K yields

lim
γ →1

2 − γ
1− γ
K =
lim
γ →1
2 − γ
1 − γ
|
.

`
,

−1+ 1+
16
2
1 + x
x
2
|
.
`
,
1− γ
(2 − γ)
2
+ order(1− γ)
2

]
]
]
¹
'
¹
¹
'
¹
=
lim
γ →1
8(1 + x)
x
2
1
2 − γ

]
]
]
=
8(1+x)
x
2

Therefore
lim
γ →1
L =

]
]
] 2x
2
8(1+x)
1/(1+x)

¹
¹
'
¹
¹
¹
¹
'
¹
¹

]
]
]
4(1+x)
x
2
+ 1
1
2
+ 1
2+x
1+x

=

]
]
] 2x
2
8(1+x)
1/(1+x)

¹¹
'
¹
¹
¹¹
'
¹
¹ x+2
x
+ 1
2+x
1+x
= (1+x)
.

|
,

`
2
x
x
1+x

7-21. (a) First we need to find L and L'. With x = 0.5 and γ = 0.9, Eq. (7-56) yields K =
0.7824, so that from Eq. (7-58) we have L = 2.89. With ε = 0.1, we have γ' = γ(1 -
ε) = 0.9γ = 0.81. Thus L' = 3.166 from Eq. (7-80). Substituting these values into
Eq. (7-83) yields
y(ε) = (1 + ε)
.

|
,

`
1
1 - ε
2+x
1+x

L'
L
= 1.1
.

|
,

`
1
.9
5/3

3.166
2.89
= 1.437
13
Then 10 log y(ε) = 10 log 1.437 = 1.57 dB
(b) Similarly, for x = 1.0, γ = 0.9, and ε = 0.1, we have L = 3.15 and L' = 3.35, so
that
y(ε) = 1.1
.

|
,

`
1
.9
3/2

3.35
3.15
= 1.37
Then 10 log y(ε) = 10 log 1.37 = 1.37 dB
7-22. (a) First we need to find L and L'. With x = 0.5 and γ = 0.9, Eq. (7-56) yields K =
0.7824, so that from Eq. (7-58) we have L = 2.89. With ε = 0.1, we have γ' = γ(1 -
ε) = 0.9γ = 0.81. Thus L' = 3.166 from Eq. (7-80). Substituting these values into
Eq. (7-83) yields
y(ε) = (1 + ε)
.

|
,

`
1
1 - ε
2+x
1+x

L'
L
= 1.1
.

|
,

`
1
.9
5/3

3.166
2.89
= 1.437
Then 10 log y(ε) = 10 log 1.437 = 1.57 dB
(b) Similarly, for x = 1.0, γ = 0.9, and ε = 0.1, we have L = 3.15 and L' = 3.35, so
that
y(ε) = 1.1
.

|
,

`
1
.9
3/2

3.35
3.15
= 1.37
Then 10 log y(ε) = 10 log 1.37 = 1.37 dB
7-23. Consider using a Si JFET with I
gate
= 0.01 nA. From Fig. 7-14 we have that α =
0.3 for γ = 0.9. At α = 0.3, Fig. 7-13 gives I
2
= 0.543 and I
3
= 0.073. Thus from
Eq. (7-86)
W
JFET
=
1
B
2(.01nA)
1.6 ×10
−19
C
+
4(1.38×10
−23
J / K)(300K)
(1.6 ×10
−19
C)
2
10
5

]
]
]
0.543
+

1
B
4(1.38×10
−23
J/ K)(300 K)(.7)
(1.6 ×10
−19
C)
2
(.005 S)(10
5
Ω)
2

]
]
]
0.543
+

2π(10 pF)
1.6 ×10
−19
C

]
]
2
4(1.38 ×10
−23
J / K)(300 K)(.7)
(.005 S)
0.073 B
or
14
W
JFET

3.51×10
12
B
+ 0.026B
and from Eq. (7-92) W
BP
=
3.39 ×10
13
B
+ 0.0049B
7-24. We need to find b
on
from Eq. (7-57). From Fig. 7-9 we have Q = 6 for a 10
-9
BER. To evaluate Eq. (7-57) we also need the values of W and L. With γ = 0.9,
Fig. 7-14 gives α = 0.3, so that Fig. 7-13 gives I
2
= 0.543 and I
3
= 0.073. Thus
from Eq. (7-86)
W =
3.51×10
12
B
+ 0.026B = 3.51×10
5
+ 2.6×10
5
= 6.1×10
5
Using Eq. (7-58) to find L yields L = 2.871 at γ = 0.9 and x = 0.5. Substituting
these values into eq. (7-57) we have
b
on
= (6)
5/3
(1.6×10
-19
/0.7) (6.1×10
5
)
.5/3
(0.543)
1/1.5
2.871 = 7.97×10
-17
J
Thus P
r
= b
on
B = (7.97×10
-17
J)(10
7
b/s) = 7.97×10
-10
W
or
P
r
(dBm) = 10 log 7.97×10
-10
= -61.0 dBm
7-25. From Eq. (7-96) the difference in the two amplifier designs is given by
∆W =
1
Bq
2

2k
B
T
R
f
I
2
= 3.52×10
6
for I
2
= 0.543 and γ = 0.9.
From Eq. (7-57), the change in sensitivity is found from
10 log

]
]
] W
HZ
+ ∆W
W
HZ
x
2(1+x)
= 10 log

]
]
]
1.0 + 3.52
1.0
.5
3
= 10 log 1.29 = 1.09 dB
7-26. (a) For simplicity, let
D = M
2+x
.

|
,

` η

I
2
and F =
Q
M

.

|
,

` η

so that Eq. (7-54) becomes, for γ = 1,
15
b = F[ ] (Db + W)
1/2
+ W
1/2

Squaring both sides and rearranging terms gives
b
2
F
2
- Db - 2W = 2W
1/2
(Db + W)
1/2
Squaring again and factoring out a "b
2
" term yields
b
2
- (2DF
2
)b + (F
4
D
2
- 4WF
2
) = 0
Solving this quadratic equation in b yields
b =
1
2
[ ] 2DF
2
t 4F
4
D
2
- 4F
4
D
2
+ 16WF
2
= DF
2
+ 2F W
(where we chose the "+" sign) =

η

.

|
,

`
M
x
Q
2
I
2
+
2Q
M
W
1/2

(b) With the given parameter values, we have
b
on
= 2.286×10
-19
39.1M
0.5
+
1.7×10
4
M
|
.

`
,
The receiver sensitivity in dBm is found from
P
r
= 10 log b
on
(50 ×10
6
b / s) [ ]
Representative values of P
r
for several values of M are listed in the table below:
M
P
r
(dBm)
M
P
r
(dBm)
30 - 50.49 80 -51.92
40 -51.14 90 -51.94
50 -51.52 100 -51.93
60 -51.74 110 -51.90
70 -51.86 120 -51.86
16
7-27. Using Eq. (E-10) and the relationship

1
1+
x
a
|
.
`
,
2
dx
0

=
πa
2

from App. B, we have from Eq. (7-97)
B
HZ
=

1
(AR)
2

(AR)
2
1+(2πRC)
2
f
2

0

df =
π
2

1
2πRC
=
1
4RC

where H(0) = AR. Similarly, from Eq. (7-98)
B
TZ
=

1
1

1
1+
2πRC
A
|
.
`
,
2
f
2

0

df =
π
2

A
2πRC
=
A
4RC

7-28. To find the optimum value of M for a maximum S/N, differentiate Eq. (7-105)
with respect to M and set the result equal to zero:
d(S/N)
dM
=
(I
p
m)
2
M
2q(I
p
+I
D
)M
2+x
B +
4k
B
TB
R
eq
F
T

-
q(I
p
+I
D
) (2+x) M
1+x
B (I
p
m)
2
M
2

]
]
]
2q(I
p
+I
D
)M
2+x
B +
4k
B
TB
R
eq
F
T
2
= 0
Solving for M,
M
2+x
opt
=
4k
B
TBF
T
/R
eq
q(I
p
+I
D
)x

7-29. (a) For computational simplicity, let K = 4k
B
TBF
T
/R
eq
; substituting M
opt
from
Problem 7-28 into Eq. (7-105) gives
17
S
N
=
1
2
(I
p
m)
2
M
2
opt
2q(I
p
+I
D
)M
2+x
opt

B + KB
=
1
2
(I
p
m)
2

]
]
]
K
q(I
p
+I
D
)x
2
2+x
2q(I
p
+I
D
)K
q(I
p
+I
D
)x

B + KB

=
xm
2
I
2
p
2B(2+x)
[ ]
q(I
p
+I
D
)x
2/(2+x)

.

|
,

`
R
eq
4k
B
TF
T
x/(2+x)

(b) If I
p
>> I
D
, then
S
N
=
xm
2
I
2
p
2B(2+x) ( ) qx
2/(2+x)
I
2/(2+x)
p

.

|
,

`
R
eq
4k
B
TF
T
x/(2+x)

=
m
2
2Bx(2+x)

]
]
]
]
( )
xI
p
2(1+x)
q
2(
4k
B
TF
T
/R
eq
)
x
1/(2+x)

7-30. Substituting I
p
= R
0
P
r
into the S/N expression in Prob. 7-29a,
S
N
=
xm
2
2B(2+x)

( )
R
0
P
r
2
[ ]
q(R
0
P
r
+I
D
)x
2/(2+x)

.

|
,

`
R
eq
4k
B
TF
T
x/(2+x)

=

(0.8)
2
(0.5 A/ W)
2
P
r
2
2(5 × 10
6
/ s) 3 [1.6 × 10
−19
C(0.5P
r
+10
−8
) A]
2 / 3
10
4
Ω/ J
1.656 ×10
−20
|
.

`
,
1 / 3
=
1.530 ×10
12
P
r
2
0.5P
r
+ 10
−8
( )
2 / 3
where P
r
is in watts.
18
We want to plot 10 log (S/N) versus 10 log
P
r
1 mW
. Representative values are
shown in the following table:
P
r
(W) P
r
(dBm)
S/N 10 log (S/N) (dB)
2×10
-9
- 57 1.237 0.92
4×10
-9
- 54 4.669 6.69
1×10
-8
- 50 25.15 14.01
4×10
-8
- 44 253.5 24.04
1×10
-7
- 40 998.0 29.99
1×10
-6
- 30 2.4×10
4
43.80
1×10
-5
- 20 5.2×10
5
57.18
1×10
-4
- 10 1.13×10
7
70.52
1
Problem Solutions for Chapter 8
8-1. SYSTEM 1: From Eq. (8-2) the total optical power loss allowed between the light
source and the photodetector is
P
T
= P
S
- P
R
= 0 dBm - (-50 dBm) = 50 dB
= 2(l
c
) + α
f
L + system margin = 2(1 dB) + (3.5 dB/km)L + 6 dB
which gives L = 12 km for the maximum transmission distance.
SYSTEM 2: Similarly, from Eq. (8-2)
P
T
= -13 dBm - (-38 dBm) = 25 dB = 2(1 dB) + (1.5 dB/km)L + 6 dB
which gives L = 11.3 km for the maximum transmission distance.
8-2. (a) Use Eq. (8-2) to analyze the link power budget. (a) For the pin photodiode,
with 11 joints
P
T
= P
S
- P
R
= 11(l
c
) + α
f
L + system margin
= 0 dBm - (-45 dBm) = 11(2 dB) + (4 dB/km)L + 6 dB
which gives L = 4.25 km. The transmission distance cannot be met with these
components.
(b) For the APD
0 dBm - (-56 dBm) = 11(2 dB) + (4 dB/km)L + 6 dB
which gives L = 7.0 km. The transmission distance can be met with these
components.
8-3. From g(t) = ( ) 1 - e
-2πBt
u(t) we have
.
|
,
`
1 - e
-2πBt
10
= 0.1 and .
|
,
`
1 - e
-2πBt
90
= 0.9
so that
2
e
-2πBt
10
= 0.9 and e
-2πBt
90
= 0.1
Then
e
2πBt
r
= e
2πB(t
90
-t
10
)
=
.9
.1
= 9
It follows that
2πBt
r
= ln 9 or t
r
=
ln 9
2πB
=
0.35
B

8-4. (a) From Eq. (8-11) we have
1
2π σ
exp
.

|
,

`
-
t
2
1/2

2
=
1
2

1
2π σ
which yields t
1/2
= (2 ln 2)
1/2
σ
(b) From Eq. (8-10), the 3-dB frequency is the point at which
G(ω) =
1
2
G(0), or exp

]
]
]
-
(2πf
3dB
)
2
σ
2
2
=
1
2

Using σ as defined in Eq. (8-13), we have
f
3dB
=
(2 ln 2)
1/2
2πσ
=
2 ln 2
π t
FWHM
=
0.44
t
FWHM

8-5. From Eq. (8-9), the temporal response of the optical output from the fiber is
g(t) =
1
2π σ
exp
.

|
,

`
-
t
2

2

If τ
e
is the time required for g(t) to drop to g(0)/e, then
g(τ
e
) =
1
2π σ
exp
.

|
,

`
-
τ
e
2

2
=
g(0)
e
=
1
2π σe

3
from which we have that τ
e
= 2 σ. Since t
e
is the full width of the pulse at the
1/e points, then t
e
= 2τ
e
= 2 2 σ.
From Eq. (8-10), the 3-dB frequency is the point at which
G(f
3dB
) =
1
2
G(0). Therefore with σ = t
e
/(2 2 )
G(f
3dB
) =
1

exp

]
]
]
-
1
2
(2πf
3dB
σ)
2
=
1
2

1

Solving for f
3dB
:
f
3dB
=
2 ln 2
2πσ
=
2 ln 2

2 2
t
e
=
0.53
t
e

8-6. (a) We want to evaluate Eq. (8-17) for t
sys
.
Using D
mat
= 0.07 ns/(nm-km), we have
t
sys
=
¹
'
¹
¹
'
¹
(2)
2
+ (0.07)
2
(1)
2
(7)
2
+

]
]
] 440(7)
0.7
800
2
+

]
]
]
350
90
2 1/2

= 4.90 ns
The data pulse width is T
b
=
1
B
=
1
90 Mb/s
= 11.1 ns
Thus 0.7T
b
= 7.8 ns > t
sys
, so that the rise time meets the NRZ data requirements.
(b) For q = 1.0,
t
sys
=
¹
'
¹
¹
'
¹
(2)
2
+ (0.49)
2
+

]
]
]
440(7)
800
2
+

]
]
]
350
90
2 1/2
= 5.85 ns
8-7. We want to plot the following 4 curves of L vs B =
1
T
b
:
(a) Attenuation limit
P
S
- P
R
= 2(l
c
) + α
f
L + 6 dB, where P
R
= 9 log B - 68.5
so that L = (P
S
– 9 log B + 62.5 - 2l
c
)/α
f
(b) Material dispersion
4
t
mat
= D
mat
σ
λ
L = 0.7T
b
or
L =
0.7T
b
D
mat
σ
λ

=
0.7
BD
mat
σ
λ

=
10
4
B
(with B in Mb/s)
(c) Modal dispersion (one curve for q = 0.5 and one for q = 1)
t
mod =
0.440L
q
800

=
0.7
B
or L =

]
]
]
.

|
,

`
800
0.44

0.7
B
1/q

With B in Mb/s, L = 1273/B for q =1, and L = (1273/B)
2
for q = .5.
8-8. We want to plot the following 3 curves of L vs B =
1
T
b
:
(a) Attenuation limit
P
S
- P
R
= 2(l
c
) + α
f
L + 6 dB, where P
R
= 11.5 log B - 60.5, P
S
= -13 dBm, α
f
=
1.5 dB/km, and l
c
= 1 dB,
so that L = (39.5 - 11.5 log B)/1.5 with B in Mb/s.
(b) Modal dispersion (one curve for q = 0.5 and one for q = 1)
t
mod =
0.440L
q
800

=
0.7
B
or L =

]
]
]
.

|
,

`
800
0.44

0.7
B
1/q

With B in Mb/s, L = 1273/B for q =1, and L = (1273/B)
2
for q = .5.
8-9. The margin can be found from
P
S
- P
R
= l
c
+ 49(l
sp
) + 50α
f
+ noise penalty + system margin
-13 - (-39) = 0.5 + 49(.1) + 50(.35) + 1.5 + system margin
from which we have
system margin = 1.6 dB
5
8-10. Signal bits
8-11. The simplest method is to use an exclusive-OR gate (EXOR), which can be
implemented using a single integrated circuit. The operation is as follows: when
the clock period is compared with the bit cell and the inputs are not identical, the
EXOR has a high output. When the two inputs are identical, the EXOR output is
low. Thus, for a binary zero, the EXOR produces a high during the last half of the
bit cell; for a binary one, the output is high during the first half of the bit cell.
A B C
L L L
L H H
H L H
H H L
1
0 1 1 1 1 0 0 1 0 0
1
Signal bits
Baseband (NRZ-L) data
Clock signal
Optical Manchester
6
8-12.
8-13.
Original 010 001 111 111 101 000 000 001 111 110
code
3B4B 0101 0011 1011 0100 1010 0010 1101 0011 1011 1100
encoded
8-14. (a) For x = 0.7 and with Q = 6 at a 10
-9
BER,
P
mpn
= -7.94 log (1 - 18k
2
π
4
h
4
) where for simplicity h = BZDσ
λ
NRZ data
Freq. A
Freq. B
PSK data
7
(b) With x = 0.7 and k = 0.3, for an 0.5-dB power penalty at
140 Mb/s = 1.4×10
-4
b/ps [to give D in ps/(nm.km)]:
0.5 = -7.94 log {1 - 18(0.3)
2
[π(1.4×10
-4
)(100)(3.5)]
4
D
4
}
or
0.5 = -7.94 log {1 - 9.097×10
-4
D
4
} from which D = 2 ps/(nm.km)
B (Mb/s) D [ps/(nm.km)]
140 2
280 1
560 0.5
1
Problem Solutions for Chapter 9
9-1.
9-2.
0 4 8 12 16
58
54
50
CNR
(dB)
RIN limit
Thermal noise
limit
Quantum
Noise
limit
f1 f2 f3 f4 f5

Transmission
system
Triple-
beat
products
2-tone
3rd order
2
9-3. The total optical modulation index is
m =

i

m
2
i
1/2
= 30(.03)
2
+ 30(.04)
2
[ ]
1 / 2
= 27.4 %
9-4 The modulation index is m =

i=1
120
(.023)
2
1/2
= 0.25
P = P
0
– 2(l
c
) - α
f
L = 3 dBm - 1 dB - 12 dB = -10 dBm = 100µW
The carrier power is
C =
1
2
(mR
0
P)
2
=
1
2
(15 ×10
−6
A)
2
The source noise is, with RIN = -135 dB/Hz = 3.162×10
-14
/Hz,
< >
i
2
source
= RIN (R
0
P)
2
B = 5.69×10
-13
A
2
The quantum noise is
< >
i
2
Q
= 2q(R
0
P + I
D
)B = 9.5×10
-14
A
2
The thermal noise is
< >
i
2
T
=
4k
B
T
R
eq
F
e
= 8.25×10
-13
A
2
3
Thus the carrier-to-noise ratio is
C
N
=
1
2
(15 ×10
−6
A)
2
5.69 × 10
−13
A
2
+ 9.5 ×10
−14
A
2
+ 8.25 ×10
−13
A
2
= 75.6
or, in dB, C/N = 10 log 75.6 = 18.8 dB.
9-5. When an APD is used, the carrier power and the quantum noise change.
The carrier power is
C =
1
2
(mR
0
MP)
2
=
1
2
(15 ×10
−5
A)
2
The quantum noise is
< >
i
2
Q
= 2q(R
0
P + I
D
)M
2
F(M)B = 2q(R
0
P + I
D
)M
2.7
B = 4.76×10
-10
A
2
Thus the carrier-to-noise ratio is
C
N
=
1
2
(15 ×10
−5
A)
2
5.69 × 10
−13
A
2
+ 4.76 × 10
−11
A
2
+ 8.25 ×10
−13
A
2
= 236.3
or, in dB,
C
N
= 23.7 dB
9-6. (a) The modulation index is m =

i=1
32
(.044)
2
1/2
= 0.25
The received power is P = -10 dBm = 100µW
The carrier power is
C =
1
2
(mR
0
P)
2
=
1
2
(15 ×10
−6
A)
2
4
The source noise is, with RIN = -135 dB/Hz = 3.162×10
-14
/Hz,
< >
i
2
source
= RIN (R
0
P)
2
B = 5.69×10
-13
A
2
The quantum noise is
< >
i
2
Q
= 2q(R
0
P + I
D
)B = 9.5×10
-14
A
2
The thermal noise is
< >
i
2
T
=
4k
B
T
R
eq
F
e
= 8.25×10
-13
A
2
Thus the carrier-to-noise ratio is
C
N
=
1
2
(15 ×10
−6
A)
2
5.69 × 10
−13
A
2
+ 9.5 ×10
−14
A
2
+ 8.25 ×10
−13
A
2
= 75.6
or, in dB, C/N = 10 log 75.6 = 18.8 dB.
(b) When m
i
= 7% per channel, the modulation index is
m =

i=1
32
(.07)
2
1/2
= 0.396
The received power is P = -13 dBm = 50µW
The carrier power is
C =
1
2
(mR
0
P)
2
=
1
2
(1.19 ×10
−5
A)
2
= 7.06×10
-11
A
2
The source noise is, with RIN = -135 dB/Hz = 3.162×10
-14
/Hz,
< >
i
2
source
= RIN (R
0
P)
2
B = 1.42×10
-13
A
2
5
The quantum noise is
< >
i
2
Q
= 2q(R
0
P + I
D
)B = 4.8×10
-14
A
2
The thermal noise is the same as in Part (a).
Thus the carrier-to-noise ratio is
C
N
=
7.06 ×10
−11
A
2
1.42 ×10
−13
A
2
+ 4.8 ×10
−14
A
2
+ 8.25 ×10
−13
A
2
= 69.6
or, in dB, C/N = 10 log 69.6 = 18.4 dB.
9-8. Using the expression from Prob. 9-7 with ∆ντ = 0.05, fτ = 0.05, and ∆ν = f = 10
MHz, yields
RIN(f) =
4R
1
R
2
π

∆ν
f
2
+ ∆ν
2

1 + e
-4π∆ντ
- 2 e
-2π∆ντ
cos(2πfτ)
=
4R
1
R
2
π

1
20 MHz
(.1442)
Taking the log and letting the result be less than -140 dB/Hz gives
-80.3 dB/Hz + 10 log R
1
R
2
< -140 dB/Hz
If R
1
= R
2
then 10 log R
1
R
2
= 20 log R
1
< -60 dB
or 10 log R
1
= 10 log R
2
< -30 dB
1
Problem Solutions for Chapter 10
10-1. In terms of wavelength, at a central wavelength of 1546 nm a 500-GHz channel
spacing is

∆λ ·
λ
2
c
∆f ·
1546 nm ( )
2
3 ×10
8
m/ s
500 ×10
9
s
−1
· 4 nm
The number of wavelength channels fitting into the 1536-to-1556 spectral band
then is
N = (1556 – 1536 nm)/4 nm = 5
10-2. (a) We first find P
1
by using Eq. (10-6):

10 log
200 µW
P
1
|
.

`
,

· 2.7 dB yields

P
1
·10
(log 200 −0.27 )
· 107.4 µW
Similarly,

P
2
· 10
(log 200 −0.47)
· 67.8 µW
(b) From Eq. (10-5): Excess loss =

10 log
200
107.4 + 67.8
|
.
`
,
· 0.58 dB
(c)
P
1
P
1
+ P
2
·
107.4
175.2
· 61% and
P
2
P
1
+ P
2
·
67.8
175.2
· 39%
10-3. The following coupling percents are are realized when the pull length is stopped at
the designated points:
Coupling percents from input fiber to output 2
Points A B C D E F
1310 nm 25 50 75 90 100 0
1540 nm 50 88 100 90 50 100
10-4. From A
out
· s
11
A
in
+ s
12
B
in
and B
out
· s
21
A
in
+s
22
B
in
· 0 , we have
B
in
· −
s
21
s
22
A
in
and

A
out
· s
11

s
12
s
21
s
22

]
]
]
A
in
2
Then
T ·
A
out
A
in
2
· s
11

s
12
s
21
s
22
2
and R ·
B
in
A
in
2
·
s
21
s
22
|
.

`
,

÷ s
11

s
12
s
21
s
22
|
.

`
,

2
10-5. From Eq. (10-18)
P
2
P
0
· sin
2
0.4z ( )exp −0.06z ( ) · 0.5
One can either plot both curves and find the intersection point, or solve the
equation numerically to yield z = 2.15 mm.
10-6. Since β
z
∝ n , then for n
A
> n
B
we have κ
A
< κ
B
. Thus, since we need to have
κ
A
L
A
= κ
B
L
B
, we need to have L
A
> L
B
.
10-7. From Eq. (10-6), the insertion loss L
Ij
for output port j is
L
Ij
·10 log
P
i −in
P
j −out
|
.

`
,

Let
a
j
·
P
i −in
P
j− out
· 10
L
Ij
/10
, where the values of L
Ij
are given in Table P10-7.
Exit port no. 1 2 3 4 5 6 7
Value of a
j
8.57 6.71 5.66 8.00 9.18 7.31 8.02
Then from Eq. (10-25) the excess loss is

10 log
P
in
P
j

|
.

`
,

· 10 log
P
in
P
in
1
a
1
+
1
a
2
+ ... +
1
a
n
|
.

`
,

|
.

`
,

·10 log
1
0.95
|
.
`
,
· 0.22 dB
10-8. (a) The coupling loss is found from the area mismatch between the fiber-core
endface areas and the coupling-rod cross-sectional area. If a is the fiber-core radius
and R is the coupling-rod radius, then the coupling loss is
3
L
coupling
= 10 log
P
out
P
in
= 10 log
7πa
2
πR
2
= 10 log
7(25)
2
(150)
2
= -7.11 dB
(b) Similarly, for the linear-plate coupler
L
coupling
= 10 log
7πa
2
l∞w
= 10 log
7π(25)
2
800(50)
= -4.64 dB
10-9. (a) The diameter of the circular coupling rod must be 1000 µm, as shown in the
figure below. The coupling loss is
L
coupling
= 10 log
7πa
2
πR
2
= 10 log
7(100)
2
(500)
2
= -5.53 dB
(b) The size of the plate coupler must be 200 µm by 2600 µm.
The coupling loss is 10 log
7π(100)
2
200(2600)
= -3.74 dB
10-10. The excess loss for a 2-by-2 coupler is given by Eq. (10-5), where P
1
= P
2
for a 3-
dB coupler. Thus,
200 µm
400
µm
Coupling rod
diameter
4
Excess loss =

10 log
P
0
P
1
+ P
2
|
.

`
,

· 10 log
P
0
2P
1
|
.

`
,

· 0.1 dB
This yields

P
1
·
P
0
2
|
.
`
,
÷10
0.01
· 0.977
P
0
2
|
.
`
,
Thus the fractional power traversing the 3-dB coupler is F
T
= 0.977.
Then, from Eq. (10-27),
Total loss =

−10 log
log F
T
log 2
− 1
|
.

`
,
log N · −10 log
log 0.977
log 2
−1
|
.

`
,
log 2
n
≤ 30
Solving for n yields

n ≤
−3
log 2
log 0.977
log 2
−1
|
.

`
,
· 9.64
Thus, n = 9 and N · 2
n
· 2
9
· 512
10-11. For details, see Verbeek et al., Ref. 34, p. 1012
For the general case, from Eq. (10-29) we find

M
11
· cos 2κd ( )⋅ cos k∆L / 2 ( )+ j sin k∆L/ 2 ( )

M
12
· M
21
· j sin 2κd ( )⋅ cos k∆L / 2 ( )

M
22
· cos 2κd ( )⋅ cos k∆L / 2 ( )− j sin k∆L/ 2 ( )
The output powers are then given by
P
out ,1
· cos
2
2κd
( )⋅ cos
2
k∆L/ 2
( )+ sin
2
k∆L/ 2
( )
[ ]
P
in,1
+ sin
2
2κd
( )⋅ cos
2
k∆L/ 2
( )
[ ]
P
in,2
5
P
out ,2
· sin
2
2κd
( )⋅cos
2
k∆L/ 2
( )
[ ]
P
in,1
+ cos
2
2κd
( )⋅ cos
2
k∆L/ 2
( ) + sin
2
k∆L/ 2
( )
[ ]
P
in,2
10-12. (a) The condition ∆ν = 125 GHz is equivalent to having ∆λ = 1 nm. Thus the
other three wavelengths are 1549, 1550, and 1551 nm.
(b) From Eqs. (10-42) and (10-43), we have

∆L
1
·
c
2n
eff
(2∆ν)
· 0.4 mm and

∆L
3
·
c
2n
eff
∆ν
· 0.8 mm
10-13. An 8-to-1 multiplexer consists of three stages of 2 × 2 MZI multiplexers. The first
stage has four 2 × 2 MZIs, the second stage has two, and the final stage has one
2 × 2 MZI. Analogous to Fig. 10-14, the inputs to the first stage are (from top to
bottom) ν, ν + 4∆ν, ν + 2∆ν, ν + 6∆ν, ν + ∆ν, ν + 5∆ν, ν + 3∆ν, ν + 7∆ν.
In the first stage

∆L
1
·
c
2n
eff
(4∆ν)
· 0.75 mm
In the second stage

∆L
2
·
c
2n
eff
(2∆ν)
· 1.5 mm
In the third stage

∆L
3
·
c
2n
eff
(∆ν)
· 3.0 mm
10-14. (a) For a fixed input angle φ, we differentiate both sides of the grating equation to
get
cos θ dθ =
k
n'Λ
dλ or

=
k
n'Λ cos θ

If φ ≈ θ, then the grating equation becomes 2 sin θ =

n'Λ
.
6
Solving this for
k
n'Λ
and substituting into the

equation yields

=
2 sin θ
λ cos θ
=
2 tan θ
λ

(b) For S = 0.01,
tan θ =

]
]
] Sλ
2∆λ (1+m)
1/2
=
0.01(1350)
2(26)(1+ 3)

]
]
]
1/ 2
= 0.2548
or θ = 14.3°
10-15. For 93% reflectivity
R · tanh
2
(κL) · 0.93 yields κL = 2.0, so that L = 2.7 mm for κ = 0.75 mm
-1
.
10-16. See Bennion et al., Ref. 42, Fig. 2a.
10-17. Derivation of Eq. (10-49).
10-18. (a) From Eq. (10-45), the grating period is

Λ ·
λ
uv
2 sin
θ
2
·
244 nm
2 sin(13.5°)
·
244
2(0.2334)
nm · 523 nm
(b) From Eq. (10-47),

λ
Bragg
· 2Λn
eff
· 2(523 nm) 1.48 ·1547 nm
(c) Using η· 1− 1/ 2 · 0.827 , we have from Eq. (10-51),

κ ·
π δn η
λ
Bragg
·
π 2.5 ×10
−4
( )
(0.827)
1.547 ×10
−4
cm
· 4.2 cm
−1
(d) From Eq. (10-49),

∆λ ·
1.547 µm ( )
2
π (1.48) 500 µm
(2.1)
2
+ π
2
[ ]
1 / 2
· 3.9 nm
7
(e) From Eq. (10-48), R
max
· tanh
2
(κL) · tanh
2
(2.1) · (0.97)
2
· 94%
10-19. Derivation of Eq. (10-55).
10-20. (a) From Eq. (10-54),

∆L · m
λ
0
n
c
· 118
1.554 µm
1.451
·126.4 µm
(b) From Eq. (10-57),

∆ν ·
x
L
f

n
s
cd

2

n
c
n
g
·
25 µm
9.36 ×10
3
µm

1.453 (3 ×10
8
m/ s)(25 ×10
−6
m)
118 (1.554 ×10
−6
m)
2

1.451
1.475
· 100.5 GHz

∆λ ·
λ
2
c
∆ν ·
(1.554 ×10
−6
m)
2
3 ×10
8
m/ s
100.5 GHz · 0.81 nm
(c) From Eq. (10-60),

∆ν
FSR
·
c
n
g
∆L
·
3 ×10
8
m/ s
1.475(126.4 µm)
· 1609 GHz
Then

∆λ ·
λ
2
c
∆ν
FSR
·
(1.554 ×10
−6
m)
2
3 ×10
8
m / s
1609 GHz ·12.95 nm
(d) Using the conditions

sinθ
i
≈ θ
i
·
2(25 µm)
9380 µm
· 5.33 ×10
−3
and

sinθ
o
≈ θ
o
· 21.3 ×10
−3
8
then from Eq. (10-59),

∆ν
FSR

c
n
g
[∆L + d θ
i
+ θ
o
( )]
·
3 ×10
8
m/ s
1.475 (126.4 × 10
−6
m) + (25 ×10
−6
m) 5.33 + 21.3 ( ) ×10
−3
[ ]
· 1601 GHz
10-21. The source spectral width is

∆λ
signal
·
λ
2
ν
c
·
1550 nm ( )
2
(1.25 ×10
9
s
−1
)
(3 ×10
8
m/ s) 10
9
nm / m
( )
·1 ×10
−2
nm
Then from Eq. (10-61)

∆λ
tune
· λ
∆n
eff
n
eff
· 1550 nm ( ) 0.5% ( ) · 7.75 nm
Thus, from Eq. (10-63)

N ·
∆λ
tune
10 λ
signal
·
7.75 nm
10 0.01 nm ( )
· 77
10-22. (a) From Eq. (10-64), the grating period is

Λ ·
λ
Bragg
2n
eff
·
1550 nm
2 3.2 ( )
· 242.2 nm
(b) Again, from the grating equation,

∆Λ ·
∆λ
2n
eff
·
2.0 nm
2 3.2 ( )
· 0.3 nm
10-23. (a) From Eq. (10-43)

∆L ·
c
2n
eff
∆ν
·
λ
2
∆λ
1
2n
eff
· 4.0 mm
(b) ∆L
eff
· ∆n
eff
L implies that

∆n
eff
·
4 mm
100 mm
· 0.04 · 4%
9
10-24. For example, see C. R. Pollock, Fundamentals of Optoelectronics, Irwin, 1995,
Fig. 15.11, p. 439.
10-25. (a) The driving frequencies are found from
f
a
· ν
o
v
a
∆n
c
·
v
a
∆n
λ
Thus we have
Wavelength (nm) 1300 1546 1550 1554
Acoustic
frequency (MHz)
56.69 47.67 47.55 47.43
(b) The sensitivity is (4 nm)/(0.12 MHz) = 0.033 nm/kHz
1
Problem Solutions for Chapter 11
11-1. (a) From Eq. (11-2), the pumping rate is

R
p
·
I
qwdL
·
100 mA
(1.6 ×10
−19
C)(5 µm)(0.5 µm)(200 µm)
· 1.25 ×10
27
(electrons / cm
3
) / s
(b) From Eq. (11-8), the maximum zero-signal gain is

g
0
· 0.3(1×10
−20
m
2
)(1 ns) 1.25 ×10
33
(electrons/ m
3
) / s −
1.0 × 10
24
/ m
3
1 ns

]
]
]
· 750 m
−1
· 7.5 cm
−1
(c) From Eq. (11-7), the saturation photon density is

N
ph;sat
·
1
0.3 (1×10
−20
m
2
) 2 ×10
8
m/ s
( )
(1 ns)
· 1.67 ×10
15
photons/ cm
3
(d) From Eq. (11-4), the photon density is

N
ph
·
P
in
λ
v
g
hc wd ( )
· 1.32 × 10
10
photons / cm
3
11-2. Carrying out the integrals in Eq. (11-14) yields
g
0
L · ln
P(L)
P(0)
+
P(L) − P(0)
P
amp,sat
Then with P(0) = P
in
, P(L) = P
out
, G = P
out
/P
in
, and G
0
· exp g
0
L ( ) from Eq. (11-
10), we have

ln G
0
· g
0
L · ln G+
GP
in
P
amp,sat

P
in
P
amp,sat
· ln G+ 1 − G ( )
P
in
P
amp,sat
Rearranging terms in the leftmost and rightmost parts then yields Eq. (11-15).
11-3. Plots of amplifier gains.
2
11-4. Let G = G
0
/2 and P
in
· P
out
/ G · 2P
out,sat
/ G
0
. Then Eq. (11-15) yields

G
0
2
· 1+
G
0
P
amp.sat
2P
out.sat
ln 2
Solving for P
out,sat
and with G
0
>> 1, we have

P
out .sat
·
G
0
ln 2
G
0
− 2
( )
P
amp.sat
≈ (ln 2) P
amp .sat
· 0.693 P
amp.sat
11-5. From Eq. (11-10), at half the amplifier gain we have
G ·
1
2
G
0
·
1
2
exp g
0
L ( )· exp gL ( )
Taking the logarithm and substituting into the equation given in the problem,

g · g
0

1
L
ln 2 ·
g
0
1 + 4 ν
3dB
− ν
0
( )
2
/ ∆ν ( )
2
From this we can find that

2 ν
3dB
− ν
0
( )
∆ν
·
g
0
g
0

1
L
ln 2
−1

]
]
]
]
1/ 2
·
1
g
0
L/ ln 2 −1

]
]
]
1/ 2
·
1
log
2
G
0
2
|
.
`
,

]
]
]
1/ 2
11-6. Since

ln G · g λ ( )L · g
0
exp − λ − λ
0
( )
2
/ 2 ∆λ ( )
2
[ ]
· ln G
0
exp − λ − λ
0
( )
2
/ 2 ∆λ ( )
2
[ ]
we have

ln
ln G
0
ln G

]
]
·
λ − λ
0
( )
2
2 ∆λ
( )
2
The FWHM is given by 2(λ – λ
0
), so that from the above equation, with the 3-dB
gain G = 27 dB being 3 dB below the peak gain, we have
3

FWHM · 2 λ − λ
0
· 2 2 ln
ln G
0
ln G
|
.
`
,

]
]
]
1/ 2
∆λ
· 2 2 ln
ln 30
ln 27
|
.
`
,

]
]
]
1/ 2
∆λ · 0.50 ∆λ
which is the expected result for a gaussian gain profile.
11-7. From Eq. (11-17), the maximum PCE is given by
PCE ≤
λ
p
λ
s
·
980
1545
· 63.4% for 980-nm pumping, and by
PCE ≤
λ
p
λ
s
·
1475
1545
· 95.5% for 1475-nm pumping
11-8. (a) 27 dBm = 501 mW and 2 dBm = 1.6 mW.
Thus the gain is

G · 10 log
501
1.6
|
.
`
,
· 10 log 313 · 25 dB
(b) From Eq. (11-19),

313 ≤ 1+
980
1542

P
p,in
P
s, in
. With a 1.6-mW input signal, the pump power needed is

P
p,in

312 1542 ( )
980
1.6 mW ( ) · 785 mW
11-9. (a) Noise terms:
From Eq. (6-17), the thermal noise term is

σ
T
2
·
4k
B
T
R
L

4 1.38 ×10
−23
J / K
( )
293 K
( )
1000 Ω
1 GHz · 1.62 ×10
−14
A
2
From Eq. (11-26), we have
4

σ
shot − s
2
· 2qR GP
s, in
B
· 2 1.6 ×10
−19
C
( )
0.73 A/ W ( ) 100 ( ) 1 µW ( )1 GHz
· 2.34 ×10
−14
A
2
From Eqs. (11-26) and (11-24), we have

σ
shot − ASE
2
· 2qR S
ASE
∆ν
opt
B · 2qR
hc
λ
n
sp
G∆ν
opt
B
· 2 1.6 ×10
−19
C
( )
.73 A/ W ( ) 6.626 ×10
−34
J/ K
( )
× 3 ×10
8
m/ s
( )
2(100)(3.77 THz)(1 GHz)/ 1550nm
· 2.26 ×10
−14
A
2
From Eq. (11-27) and (11-24), we have

σ
s− ASE
2
· 4 0.73 A/ W ( ) 100 ( ) 1 µW ( ) [ ]
× .73 A/ W ( )
6.626 ×10
−34
J/ K
( )
3 ×10
8
m/ s
( )
1550nm
2(100)(1 GHz)

]
]
]
· 5.47 ×10
−12
A
2
From Eq. (11-28), we have

σ
ASE− ASE
2
· .73A/ W ( )
2
6.626 ×10
−34
J/ K
( )
3 ×10
8
m/ s
( )
1550nm
2(100)

]
]
]
2
× 2 1 THz ( )−1 GHz
[ ]
1 GHz ( )
· 7.01×10
−13
A
2
5
11-10. Plot of penalty factor from Eq. (11-36).
11-11. (a) Using the transparency condition Gexp(-αL) = 1 for a fiber/amplifier segment,
we have

P
path
·
1
L
P(z) dz
0
L

·
P
in
L
e
− αz
dz
0
L

·
P
in
αL
1 − e
−αL
[ ]
·
P
in
αL
1 −
1
G

]
]
·
P
in
G

G −1
ln G
|
.
`
,
since ln G = αL from the transparency condition.
(b) From Eq. (11-35) and using Eq. (11-24),

P
ASE
path
·
NP
ASE
L
e
− αz

0
L

dz ·
NP
ASE
αL
1 − e
−αL
( )
·
α (NL)
(αL)
2
P
ASE
1−
1
G
|
.
`
,
·
αL
tot
(ln G)
2
hνn
sp
(G −1)∆ν
opt
1 −
1
G
|
.
`
,
· αL
tot
hνn
sp
∆ν
opt

1
G
G −1
ln G
|
.
`
,
2
11-12. Since the slope of the gain-versus -input power curve is –0.5, then for a 6-dB drop
in the input signal, the gain increases by +3 dB.
1. Thus at the first amplifier, a –10.1-dBm signal now arrives and experiences a
+10.1-dB gain. This gives a 0-dBm output (versus a normal +3-dBm output).
2. At the second amplifier, the input is now –7.1 dBm (down 3 dB from the usual
–4.1 dBm level). Hence the gain is now 8.6 dB (up 1.5 dB), yielding an output
of
–7.1 dBm + (7.1 + 1.5) dB = 1.5 dBm
6
3. At the third amplifier, the input is now –5.6 dBm (down 1.5 dB from the usual
–4.1 dBm level). Hence the gain is up 0.75 dB, yielding an output of
–5.6 dBm + (7.1 + 0.75) dB = 2.25 dBm
4. At the fourth amplifier, the input is now –4.85 dBm (down 0.75 dB from the
usual –4.1 dBm level). Hence the gain is up 0.375 dB, yielding an output of
–4.85 dBm + (7.1 + 0.375) dB = 2.63 dBm
which is within 0.37 dB of the normal +3 dBm level.
11-13. First let 2πν
i
t + φ
i
· θ
i
for simplicity. Then write the cosine term as

cos θ
i
·
e

i
+ e
− jθ
i
2
, so that

P · E
i
(t)E
i
*
(t) · 2P
i
i ·1
N

e

i
+ e
−jθ
i
2

]
]
]
× 2P
k
k ·1
N

e

k
+ e
− jθ
k
2

]
]
]
·
1
4
2P
i
k·1
N

2P
k
i ·1
N

e

i
e
− jθ
k
+ e

k
e
− jθ
i
+ e

i
e

k
+ e
− jθ
i
e
− jθ
k
[ ]
·
1
4
2P
i
k·1
N

2P
k
i ·1
N

e
j (θ
i
−θ
k
)
+ e
− j( θ
i
− θ
k
)
+ e
j( θ
i

k
)
+ e
− j( θ
i

k
)
[ ]
· P
i
+
1
2
i·1
N

2P
i
k≠ i
N

2P
k
i ·1
N

e
j(θ
i
−θ
k
)
+ e
− j(θ
i
− θ
k
)
[ ]
where the last two terms in the second-last line drop out because they are beyond
the response frequency of the detector. Thus,

P · P
i
+
i·1
N

2 P
i
P
k
k ≠i
N

i·1
N

cos θ
i
− θ
k
( )
[ ]
11-14. (a) For N input signals, the output signal level is given by

P
s,out
· G P
s,in
i ·1
N

(i) ≤1 mW.
The inputs are 1 µW (-30 dBm) each and the gain is 26 dB (a factor of 400).
7
Thus for one input signal, the output is (400)(1 µW) = 400 µW or –4 dBm.
For two input signals, the total output is 800 µW or –1 dBm. Thus the level of
each individual output signal is 400 µW or –4 dBm.
For four input signals, the total input level is 4 µW or –24 dBm. The output then
reaches its limit of 0 dBm, since the maximum gain is 26 dB. Thus the level of each
individual output signal is 250 µW or –6 dBm.
Similarly, for eight input channels the maximum output level is o dBm, so the level
of each individual output signal is 1/8(1 mW) = 125 µW or –9 dBm.
(b) When the pump power is doubled, the outputs for one and two inputs remains
at the same level. However, for four inputs, the individual output level is 500 µW
or –3 dBm, and for 8 inputs, the individual output level is 250 µW or –6 dBm.
11-15. Substituting the various expressions for the variances from Eqs. (11-26) through
(11-30) into the expression given for Q in the problem statement, we find
Q ·
AP
HP + D
2
( )
1 / 2
+ D
where we have defined the following terms for simplicity

A · 2R G

H · 4qR GB+ 8R
2
GS
ASE
B and D
2
· σ
off
2
Rearrange terms in the equation for Q to get
Q
2
HP + D
2
( )
1/ 2
· AP − QD
Squaring both sides and solving for P yields P ·
2QD
A
+
Q
2
H
A
2
Substituting the expressions for A, H, and D into this equation, and recalling the
expression for the responsivity from Eq. (6-6), then produces the result stated in
the problem, where
8
F ·
1 + 2ηn
sp
(G−1)
ηG
1
Problem Solutions for Chapter 12
12-1. We need to evaluate P
in
using Eq. (12-11). Here F
c
= 0.20,
C
T
= 0.05, F
i
= 0.10, P
0
= 0.5 mW, and A
0
= e
−2.3( 3) / 10
= 0.933
Values of P
in
as a function of N are given in the table below. P
in
in
dBm is found from the relationship P
in
(dBm) = 10 log

P
in
(mW)
1 mW
N
P
in
(nW) P
in
(dBm)
N
P
in
(nW) P
in
(dBm)
2 387 -34.1 8 5.0 -53.0
3 188 -37.3 9 2.4 -56.2
4 91 -40.4 10 1.2 -59.2
5 44.2 -43.5 11 0.6 -62.2
6 21.4 -46.7 12 0.3 -65.5
7 10.4 -49.8
(b) Using the values in the above table, the operating margin for 8 stations is
-53 dBm - (-58 dBm) = 5 dB
(c) To have a 6-dB power margin, we can transmit over at most seven stations.
The dynamic range with N = 7 is found from Eq. (12-13):

DR · −10(N− 2) log .933(.8)
2
(.95)
2
(.9) [ ]· −50 log (0.485) ·15.7 dBm
12-2. (a) Including a power margin, we have from Eq. (12-16)

P
S
− P
R
− power margin · L
excess
+ α(2L) + 2L
c
+10log N
Thus
0 – (-38 dBm) – 6 dB = 3 dB + (0.3 dB/km)2(2 km) + 2(1.0 dB) + 10 log N
2
so that 10 log N = 25.8. This yields N = 380.1, so that 380 stations can be
attached.
(b) For a receiver sensitivity of –32 dBm, one can attach 95 stations.
12-3. (b) Let the star coupler be located in the ceiling in the wire room, as shown in the
figure below.
For any row we need seven wires running from the end of the row of offices to
each individual office. Thus, in any row we need to have (1+2+3+4+5+6+7)x15 ft
= 420 ft of optical fiber to connect the offices. From the wiring closet to the
second row of offices (row B), we need 8(10 + 15) ft = 200 ft; from the wiring
closet to the third row of offices (row C), we need 8(10 + 30) ft = 320 ft; and from
A
Wire
room
B C D
3
the wiring closet to the fourth row of offices (row D), we need 8(20 + 45) ft = 520
ft of cable. For the 28 offices we also need 28x7 ft = 196 ft for wall risers.
Therefore for each floor we have the following cable needs:
(1) 4 x 420 ft for row runs
(2) 200 + 320 + 520 ft = 1040 for row connections
(3) 196 ft for wall risers
Thus, the total per floor = 2916 ft
Total cable in the building: 2x9 ft risers + 2916 ft x 2 floors = 5850 ft
12-4. Consider the following figure:
(a) For a bus configuration:
Cable length = N rows×(M-1)stations/row + (N-1) row interconnects
= N(M-1)d + (N-1)d = (MN-1)d
(b) The ring is similar to the bus, except that we need to close the loop with one
cable of length d. Therefore the cable length = MNd
(c) In this problem we consider the case where we need individual cables run
from the star to each station. Then the cable length is
L = cables run along the M vertical rows + cables run along the N horizontal
rows:
M
N
d
4

= Md i
i ·1
N−1

+ Nd j
j·1
M−1

= M
N(N −1)
2
d + N
M(M−1)
2
d =
MN
2
(M + N − 2)d
12-5. (a) Let the star be located at the relative position (m,n). Then
L = N j
j·1
m−1

+ N j
j ·1
M− m

+ M i
i ·1
n−1

+ M i
i ·1
N− n

]
]
]
d
= N
m(m−1)
2
+
(M− m)(M − m+1)
2

]
]
+ M
n(n −1)
2
+
(N− n)(N − n +1)
2

]
]
¹
'
¹
¹
'
¹
d
=
MN
2
(M + N+ 2) − Nm(M− m+ 1) − Mn(N− n +1)

]
]
d
(b) When the star coupler is located in one corner of the grid, then
m = n= 1, so that the expression in (a) becomes
L =
MN
2
(M + N+ 2) − NM− MN

]
]
d =
MN
2
(M + N− 2)d
(c) To find the shortest distance, we differentiate the expression for L given in (a)
with respect to m and n, and set the result equal to zero:
dL
dm
= N(m - 1 - M) + Nm = 0 so that m =
M +1
2
Similarly
dL
dn
= M(n - 1 - N) + nM = 0 yields n =
N +1
2
Thus for the shortest cable runs the star should be located in the center of the grid.
12-6. (a) For a star network, one cannot reuse wavelengths. Thus, since each node must
be connected to N – 1 other nodes through a central point, we need N – 1
wavelengths.
5
For a bus network, these equations can easily be verified by drawing sample
diagrams with several even or odd stations.
For a ring network, each node must be connected to N – 1 other nodes. Without
wavelength reuse one thus needs N(N – 1) wavelengths. However, since each
wavelength can be used twice in the network, the number of wavelengths needed
is N(N-1)/2.
12-7. From Tables 12-4 and 12-5, we have the following:
OC-48 output for 40-km links: –5 to 0 dBm; α = 0.5 dB/km; P
R
= -18 dBm
OC-48 output for 80-km links: –2 to +3 dBm; α = 0.3 dB/km; P
R
= -27 dBm
The margin is found from: Margin · P
s
− P
R
( )− αL − 2L
c
(a) Minimum power at 40 km:
Margin = [-2 – (-27)] – 0.5(40) –2(1.5) = +2 dB
(b) Maximum power at 40 km:
Margin = [0 – (-27)] – 0.5(40) –2(1.5) = +4 dB
(c) Minimum power at 80 km:
Margin = [-2 – (-27)] – 0.3(80) –2(1.5) = -2 dB
(d) Maximum power at 80 km:
Margin = [3 – (-27)] – 0.3(80) –2(1.5) = +3 dB
12-8. Expanding Table 12-6:
# of λs P
1
(dBm) P · 10
P
1
/ 10
(mW) P
total
(mW) P
total
(dBm)
1 17 50 50 17
2 14 25 50 17
3 12.2 16.6 49.8 17
4 11 12.6 50.4 17
5 10 10 50 17
6
6 9.2 8.3 49.9 17
7 8.5 7.1 49.6 17
8 8.0 6.3 50.4 17
12-9. See Figure 20 of ANSI T1.105.01-95.
12-10. See Figure 21 of ANSI T1.105.01-95.
12-11. The following wavelengths can be added and dropped at the three other nodes:
Node 2: add/drop wavelengths 3, 5, and 6
Node 3: add/drop wavelengths 1, 2, and 3
Node 4: add/drop wavelengths 1, 4, and 5
12-12. (b) From Eq. (12-18) we have
N
λ
· kp
k +1
· 2(3)
3
· 54
(c) From Eq. (12-20) we have
H ·
2(3)
2
(3 −1)(6 −1) − 4(3
2
−1)
2(3 −1)[2(3
2
) − 1]
· 2.17
(d) From Eq. (12-21) we have
C ·
2 3 ( )
3
2.17
· 8.27
12-13. See Hluchyj and Karol, Ref. 25, Fig. 6, p. 1391 (Journal of Lightwave
Technology, Oct. 1991).
12-14. From Ref. 25:
In general, for a (p,k) ShuffleNet, the following spanning tree for assigning fixed
routes to packets generated by any given user can be obtained:
h Number of users h hops away from the source
1 p
7
2 p
2
.
.
.
k – 1 p
k-1
k P
k
- 1
k + 1 P
k
- p
k + 2 P
k
- p
2
.
.
.
2k – 1 P
k
- p
k-1
Summing these up results in Eq. (12-20).
12-15. See Li and Lee (Ref 40) for details.
12-16. The following is one possible solution:
(a) Wavelength 1 for path A-1-2-5-6-F
(b) Wavelength 1 for path B-2-3-C
(c) Wavelength 2 for the partial path B-2-5 and Wavelength 1 for path 5-6-F
(d) Wavelength 2 for path G-7-8-5-6-F
(e) Wavelength 2 for the partial path A-1-4 and Wavelength 1 for path 4-7-G
12-17. See Figure 4 of Barry and Humblet (Ref. 42).
12-18. See Shibata, Braun, and Waarts (Ref. 67).
(a) The following nine 3
rd
-order waves are generated due to FWM:
ν113 = 2(ν2 - ∆ν) – (ν2 + ∆ν) = ν2 - 3∆ν
ν112 = 2(ν2 - ∆ν) – ν2 = ν2 - 2∆ν
ν123 = (ν2 - ∆ν) + ν2 – (ν2 + ∆ν) = ν2 - 2∆ν
8
ν223 = 2ν2 – (ν2 + ∆ν) = ν2 - ∆ν · ν1
ν132 = (ν2 - ∆ν) + (ν2 + ∆ν) – ν2 = ν2
ν221 = 2ν2 – (ν2 - ∆ν) = ν2 + ∆ν · ν3
ν231 = ν2 + (ν2 + ∆ν) – (ν2 - ∆ν) = ν2 + 2∆ν
ν331 = 2(ν2 + ∆ν) – (ν2 - ∆ν) = ν2 + 3∆ν
ν332 = 2(ν2 + ∆ν) – ν2 = ν2 + 2∆ν
(b) In this case the nine 3
rd
-order waves are:
ν113 = 2(ν2 - ∆ν) – (ν2 + 1.5∆ν) = ν2 – 1.5∆ν
ν112 = 2(ν2 - ∆ν) – ν2 = ν2 - 2∆ν
ν123 = (ν2 - ∆ν) + ν2 – (ν2 + 1.5∆ν) = ν2 – 2.5∆ν
ν223 = 2ν2 – (ν2 + 1.5∆ν) = ν2 – 1.5∆ν
ν132 = (ν2 - ∆ν) + (ν2 + 1.5∆ν) – ν2 = ν2 + 0.5∆ν
ν221 = 2ν2 – (ν2 - ∆ν) = ν2 + ∆ν
ν231 = ν2 + (ν2 + 1.5∆ν) – (ν2 - ∆ν) = ν2 + 2.5∆ν
ν331 = 2(ν2 + 1.5∆ν) – (ν2 - ∆ν) = ν2 + 4∆ν
ν332 = 2(ν2 + 1.5∆ν) – ν2 = ν2 + 3∆ν
9
12-19. Plot: from Figure 2 of Y. Jaouën, J-M. P. Delavaux, and D. Barbier, “Repeaterless
bidirectional 4x2.5-Gb/s WDM fiber transmission experiment,” Optical Fiber
Technology, vol. 3, p. 239-245, July 1997.
12-20. (a) From Eq. (12-50) the peak power is

P
peak
·
1.7627

|
.
`
,
2

A
eff
λ
3
n
2
c

D
T
s
2
·11.0 mW
(b) From Eq. (12-49) the dispersion length is
L
disp
= 43 km
(c) From Eq. (12-51) the soliton period is

L
period
·
π
2
L
disp
· 67.5 km
(d) From Eq. (12-50) the peak power for 30-ps pulses is
10

P
peak
·
1.7627

|
.
`
,
2

A
eff
λ
3
n
2
c

D
T
s
2
· 3.1 mW
12-21. Soliton system design.
12-22. Soliton system cost model.
12-23. (a) From the given equation, L
coll
= 80 km.
(b) From the given condition,

L
amp

1
2
L
coll
· 40 km
12-24. From the equation and conditions given in Prob. 12-23, we have that

∆λ
max
·
T
s
DL
amp
·
20 ps
[0.4 ps/(nm⋅ km)](25 km)
· 2 nm
Thus 2.0/0.4 = 5 wavelength channels can be accommodated.
12-25 Plot from Figure 3 of Ref. 103.
1
Problem Solutions for Chapter 13
13-1 (a) From the given equation, n
air
= 1.000273. Thus,

λ
vacuum
· λ
air
n
air
·1.000273(1550.0 nm) · 1550.42 nm
(b) From the given equation,
n T, P ( )· 1 +
(1.000273 −1)(0.00138823)640
1+ 0.003671(0)
· 1.000243
Then

n T, P ( )1550 nm ( ) ·1550.38 nm
13-2 Since the output voltage from the photodetector is proportional to the optical
power, we can write Eq. (13-1) as
α ·
10
L
1
− L
2
log
V
2
V
1
where L
1
is the length of the current fiber, L
2
is the length cut off, and V
1
and V
2
are the voltage output readings from the long and short lengths, respectively. Then
the attenuation in decibels is

α ·
10
1895 − 2
log
3.78
3.31
· 0.31 dB/ km
13-3 (a) From Eq. (13-1)

α ·
10
L
N
− L
F
log
P
N
P
F
·
10
L
N
− L
F
log
V
N
V
F
·
10 log e
L
N
− L
F
ln
V
N
V
F
From this we find

∆α ·
10 log e
L
N
− L
F
∆V
N
V
N
+
∆V
F
V
F

]
]
]
·
4.343
L
N
− L
F
t0.1%t 0.1% ( )· t
8.686
L
N
− L
F
×10
−3
(b) If ∆α = 0.05 dB/km, then

L · L
N
− L
F

8.868 ×10
−3
0.05
km · 176 m
2
13-4 (a) From Eq. (8-11) we have
1
2π σ
exp
.

|
,

`
-
t
2
1/2

2
=
1
2

1
2π σ
which yields t
1/2
= (2 ln 2)
1/2
σ
(b) From Eq. (8-10), the 3-dB frequency is the point at which
G(ω) =
1
2
G(0), or exp

]
]
]
-
(2πf
3dB
)
2
σ
2
2
=
1
2

Using σ as defined in Eq. (8-13), we have
f
3dB
=
(2 ln 2)
1/2
2πσ
=
2 ln 2
π t
FWHM
=
0.44
t
FWHM

13-5 From Eq. (13-4), P
out
(f) / P
in
(f) · H(f) . To measure the frequency response, we
need a constant input amplitude, that is, P
in
(f) = P
in
(0). Thus,
P(f)
P(0)
·
P
out
(f) / P
in
(f)
P
out
(0) / P
in
(0)
·
H(f)
H(0)
· H(f)
The following table gives some representative values of H(f) for different values of
2σ:
f (MHz) 2σ σ = 2 ns 2σ σ = 1 ns 2σ σ = 0.5 ns
100 0.821 0.952 0.988
200 0.454 0.821 0.952
300 0.169 0.641 0.895
500 0.0072 0.291 0.735
700 0.089 0.546
1000 0.0072 0.291
13-6 To estimate the value of D, consider the slope of the curve in Fig. P13-6 at λ =
1575 nm. There we have ∆τ = 400 ps over the wavelength interval from 1560 nm
to 1580 nm, i.e., ∆λ = 20 nm. Thus

D ·
1
L
∆τ
∆λ
·
1
10 km
400 ps
20 nm
· 2 ps/(nm⋅ km)
3
Then, using this value of D at 1575 nm and with λ
0
= 1548 nm, we have

S
0
·
D λ ( )
λ − λ
0
·
2 ps/(nm ⋅ km)
(1575 − 1548) nm
· 0.074 ps/(nm
2
⋅ km)
13-7 With k = 1, λ
start
= 1525 nm, and λ
stop
= 1575 nm, we have N
e
= 17 extrema.
Substituting these values into Eq. (13-14) yields 1.36 ps.
13-8 At 10 Gb/s over a 100-km link, the given equation yields:

P
ISI
≈ 26
1 ps ( )
2
0.5 1− 0.5 ( )
100 ps
( )
2
· 6.5 ×10
−4
dB
Similarly, at 10 Gb/s over a 1000-km link,

P
ISI
≈ 0.065 dB.
This is the same result at 100 Gb/s over a 100-km link.
At 100 Gb/s over a 1000-km link, we have 6.5 dB.
13-9 For a uniform attenuation coefficient, β is independent of y. Thus, Eq. (13-16)
becomes
P(x) · P(0)exp −β dy
0
x

]
]
]
· P(0)e
−βx
Writing this as exp(−βx) · P(0)/ P(x) and taking the logarithm on both sides
yields

βx log e · log
P(0)
P(x)
. Since

α · β(10 log e), this becomes

αx ·10 log
P(0)
P(x)
For a fiber of length x = L with P(0) = P
N
being the near-end input power, this
equation reduces to Eq. (13-1).
4
13-10 Consider an isotropically radiating point source in the fiber. The power from this
point source is radiated into a sphere that has a surface area 4πr
2
. The portion of
this power captured by the fiber in the backward direction at a distance r from the
point source is the ratio of the area A = πa
2
to the sphere area 4πr
2
. If θ is the
acceptance angle of the fiber core, then A = πa
2
= π(rθ)
2
. Therefore S, as defined
in Eq. (13-18), is given by
S ·
A
4πr
2
·
πr
2
θ
2
4πr
2
·
θ
2
4
From Eq. (2-23), the acceptance angle is

sin θ ≈ θ ·
NA
n
, so that S ·
θ
2
4
·
NA ( )
2
4n
2
13-11 The attenuation is found from the slope of the curve, by using Eq. (13-22):
Fiber a:

α ·
10 log
P
D
(x
1
)
P
D
(x
2
)
2 x
2
− x
1
( )
·
10 log
70
28
2 0.5 km ( )
· 4.0 dB/ km
Fiber b:

α ·
10 log
25
11
2 0.5 km ( )
· 3.6 dB/ km
Fiber c:

α ·
10 log
7
1.8
2 0.5 km ( )
· 5.9 dB/ km
To find the final splice loss, let P
1
and P
2
be the input and output power levels,
respectively, at the splice point. Then for
For splice 1:

L
splice
·10 log
P
2
P
1
· 10 log
25
28
· −0.5 dB
For splice 2:

L
splice
·10 log
7
11
· −2.0 dB
13-12 See Ref. 42, pp. 450-452 for a detailed and illustrated derivation.
5
Consider the light scattered from an infinitesimal interval dz that is located at L =
Tv
gr
. Light scattered from this point will return to the OTDR at time t = 2T. Upon
inspection of the pulse of width W being scattered form the point L, it can be
deduced that the back-scattered power seen by the OTDR at time 2T is the
integrated sum of the light scattered from the locations z = L – W/2 to z = L.
Thus, summing up the power from infinitesimal short intervals dz from the whole
pulse and taking the fiber attenuation into account yields

P
s
L ( ) · Sα
s
0
W

P
0
exp −2α L +
z
2
|
.
`
,

]
]
]
dz
· S
α
s
α
P
0
e
−2αL
1− e
− αW
( )
which holds for L ≥ W/2. For distances less than W/2, the lower integral limit gets
replaced by W – 2L.
13-13 For very short pulse widths, we have that αW << 1. Thus the expression in
parenthesis becomes
1
α
1 − e
−αW
( )

1
α
1 − 1− αW ( ) [ ]
· W
Thus

P
s
L ( ) ≈ S α
s
W P
0
e
−2αL
13-14 (a) From the given equation, for an 0.5-dB accuracy, the SNR is 4.5 dB.
The total loss of the fiber is (0.33 dB/km)(50 km) = 16.5 dB.
The OTDR dynamic range D is

D · SNR + αL + splice loss
· 4.5 dB+16.5 dB+ 0.5 dB· 21.5 dB
6
Here the splice loss is added to the dynamic range because the noise that limits the
achievable accuracy shows up after the event.
(b) For a 0.05-dB accuracy, the OTDR dynamic range must be 26.5 dB.
13-15 To find the fault-location accuracy dL with an OTDR, we differentiate Eq. (13-
23):
dL ·
c
2n
dt
where is the accuracy to which the time difference between the original and
reflected pulses must be measured. For dL ≤ 1 m, we need

dt ·
2n
c
dL ≤
2(1.5)
3 ×10
8
m/ s
(0.5 m) · 5 ns
To measure dt to this accuracy, the pulse width must be ≤ 0.5dt (because we are
measuring the time difference between the original and reflected pulse widths).
Thus we need a pulse width of 2.5 ns or less to locate a fiber fault within 0.5 m of
its true position.