Download Direct and External Modulation of Light

Direct & External Modulation
Direct and External Modulation of Light
Christophe Peucheret
DTU Fotonik
Department of Photonics Engineering
Technical University of Denmark
[email protected]
Abstract
This note is an introduction to the laboratory exercise on direct and external modulation
in the course 34129 Experimental Course in Optical Communication. The concepts of direct
and external modulation are introduced and general requirements on the properties of the
generated signals, such as extinction ratio and frequency chirp, are briefly discussed. The
principles and main features of external modulation mechanisms in electro-absorption and
electro-optic modulators, as well as direct current modulation of semiconductor laser diodes
are outlined. Finally the assignments of the exercise itself are presented. The focus is on the
static and dynamic characterisation of a directly modulated laser operating at 10 Gbit/s,
as well as on comparison with external chirp-free modulation after propagation through a
length of standard single-mode fibre.
Contents
1 Introduction
2
2 Direct versus external Modulation
2.1 Modulation concepts . . . . . . . .
2.2 Modulation technique requirements
2.2.1 Speed of operation . . . . .
2.2.2 Extinction ratio . . . . . . .
2.2.3 Frequency chirping . . . . .
.
.
.
.
.
2
2
3
3
4
4
3 External modulation
3.1 Electro-absorption modulators . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 Electro-optic modulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6
6
7
4 Direct current modulation of
4.1 Turn-on delay . . . . . . . .
4.2 Extinction ratio . . . . . . .
4.3 Bandwidth . . . . . . . . .
4.4 Relaxation oscillations . . .
4.5 Frequency chirping . . . . .
4.6 Summary . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
semiconductor lasers
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
8
9
9
10
10
10
11
5 Description of the exercise
12
5.1 Static characterisation of a DFB laser . . . . . . . . . . . . . . . . . . . . . . . . 14
5.2 Dynamic characterisation of a DFB laser . . . . . . . . . . . . . . . . . . . . . . . 14
5.3 Transmission over standard single mode fibre . . . . . . . . . . . . . . . . . . . . 15
22/11/2009
1
34129 Experimental Course in Optical Communication
1
Introduction
The vast amount of data we are using and transmitting daily, either for plain telephony or when
using the Internet, is generated and processed in the electrical domain. However, now that we
are reaching the end of this experimental course, you should have hopefully been made aware
of the benefit (and the beauty) of using light to transport this data. For the time being, the
realm of optical communication is long haul and high capacity submarine links (from about
1000 to a couple of 10000’s km), as well as terrestrial trunk networks (up to a few 1000’s km;
usually less in Denmark). Optical communication is also used for metropolitan area networks
(MANs; a few 10’s to a few 100’s km) and is currently rapidly moving towards the end users in
the access network (e.g. fibre-to-the-home, FTTH).
In most cases, one of the benefits of optical communication is to allow the transport of relatively large capacities expressed in terms of bits per second for the digital optical communication
systems we will be considering here.1 Typical bit rates employed today are 2.5-10 Gbit/s per
channel (meaning per wavelength in wavelength division multiplexing -WDM- systems where
the capacity can be increased beyond the channel bit rate by simultaneously transmitting a
large number of channels at different wavelengths in a single optical fibre) and the next generation optical transport systems running at 40 Gbit/s per channel are ready, at least technically,
to be deployed. So far, this order of magnitude for the bit rate is well above the needs of
individual users and applications. Consequently, the data originating from a large number of
users is multiplexed in time in the electrical domain before being transmitted at a high bit
rate over optical fibres. A key functionality of an optical system is therefore the modulation
operation, which consists in “converting” the high bit-rate electrical data signal into the optical
domain. Ideal modulation is therefore equivalent to performing a frequency translation from
the baseband to an optical carrier frequency, of the order of 193 THz (i.e. 193×1012 Hz) for
the usual 1550 nm transmission window. Until now, most optical communication systems make
use of intensity modulation of the lightwave (i.e. its intensity or power is varied according
to the data to be transmitted) since it allows to use a very simple detection process under the
form of a photodiode whose generated photocurrent is proportional to the incoming optical
power. In this exercise, we will therefore exclusively focus on intensity modulation, although
other types of optical modulation, such as phase and frequency modulation, are also possible
and intensively investigated.
The goal of this exercise is to briefly introduce two different strategies used for the optical
intensity modulation operation, and to have a more in-depth look at the benefits and limitation
of one of those methods, where the optical power emitted by a semiconductor laser is varied
according to the data that the user wants to transmit.
2
Direct versus external Modulation
In this section, we present the concepts of direct and external modulation and discuss general
requirements for the optical modulated signal.
2.1
Modulation concepts
As discussed above, the operation of modulation consists in transferring the data to be transmitted from the electrical to the optical domain. Two strategies, illustrated in Fig. 1, can be
used to perform this operation. In the first option, called direct modulation, light is emitted
1
Optical fibres are also used for the transport of analogue signals, such as for instance in hybrid fibre-coaxial
systems used for the distribution of cable television (CATV).
2
22/11/2009
Direct & External Modulation
power
current
time
time
DM laser
power
power
current
time
time
time
external
modulator
CW laser
voltage
time
Figure 1 Illustration of the concepts of direct (top) and external (bottom) modulation. In the direct
modulation scheme, the driving current to a directly modulated (DM) semiconductor laser is varied
according to the data to be transmitted. In the external modulation scheme, the laser that is subjected
to a constant bias current emits a continuous wave (CW) while an external modulator switches the
optical power on or off according to the data stream.
from a semiconductor laser only when a “mark” is transmitted. Ideally, no light should be
emitted when a “space” is transmitted2 . However, we shall see later that this is not always
the case for practical implementations. In the second option, called external modulation, a
continuous wave (CW) laser is used to emit light whose power is constant with time. A second
component, known as modulator, is then used as a switch to let the light pass whenever the
data corresponds to a “mark” and to block it whenever the signal is a “space”. This switch
can be implemented in several ways and we will briefly describe two physical effects that are
customarily used to perform external optical modulation at high bit rates. Note that here, the
key point is that the physical process that should permit the switch to toggle between its two
states (“open” and “closed”) should be fast enough to allow proper operation at the desired bit
rate.
2.2
2.2.1
Modulation technique requirements
Speed of operation
As briefly introduced above, the physical processes that are exploited to perform the modulation
operation should be fast enough to allow proper operation at the desired bit rate. At 10 Gbit/s,
the bit slot duration is 100 ps and we will expect the transmitter, whether a directly modulated
laser or a continuous wave laser followed by an external modulator, to be able to switch between
the “mark” and “space” states within a fraction of this duration.
2
In this note, we define the “mark” and “space” states as the high and low power levels, respectively, of
an optical signal whose intensity has been modulated using binary modulation. These states correspond to the
transmission of “0” and “1” bits, respectively, provided no logical inversion of the signal has been performed. In
case the “space” state corresponds to the absence of optical power, the modulation format is known as on-off
keying (OOK). In the more general case, it is described as amplitude-shift keying (ASK).
22/11/2009
3
34129 Experimental Course in Optical Communication
2.2.2
Extinction ratio
The extinction ratio of the optical signal is defined as
ER =
P1
,
P0
(1)
where P1 and P0 are the power levels corresponding to the “marks” and “spaces”, respectively.
You have seen in a previous exercise that it is important to achieve a good extinction ratio
for the optical signal, i.e. to achieve a large separation between the power of the “marks” and
“spaces”, and ensure that as little power as possible is present in the signal when a “space” is
being transmitted. The effect of a poor extinction ratio will otherwise manifest itself under the
form of power penalty at the receiver (i.e. an increased required optical power at the receiver
in order to achieve a given bit-error-rate, typically 1.0×10−9 , compared to the case of an ideal
signal with infinite extinction ratio) that will reduce the power budget of the system. Under
the assumptions of a receiver limited by thermal noise, this penalty can be shown to be equal
to
ER + 1
δER (dB) = 10 log10
.
(2)
ER − 1
As an illustration, the power penalty calculated according to (2) is represented as a function of
the extinction ratio in Fig. 2. In practice, values of the extinction ratio of the order of 15 dB
will be considered nearly ideal.
10
5
C β2 > 0
9
8
broadening factor T/T0
7
power penalty (dB)
C β2 = 0
4
6
5
4
3
2
C β2 < 0
3
2
1
1
0
0
0
2
4
6
8
10
12
extinction ratio (dB)
14
16
18
20
0.0
0.5
1.0
1.5
2.0
normalised distance z/LD
Figure 2 Left: power penalty as a function of extinction ratio. Right: broadening of a Gaussian pulse
as a function of distance due to the effects of linear chirp and group velocity dispersion. The propagation
distance is normalised to the so-called dispersion length defined as LD = T02 / |β2 |.
2.2.3
Frequency chirping
The electric field of a lightwave whose carrier angular frequency is ω0 can be expressed as
(3)
E (t) = ℜ A (t) eiω0 t ,
where ℜ denotes the real part, and where A (t) is known as the complex envelope of the
signal. As the name suggests, the envelope is a complex number that can be further written in
terms of its modulus |A| and phase φ according to
p
A (t) = |A (t)| eiφ(t) = P (t)eiφ(t) ,
(4)
4
22/11/2009
Direct & External Modulation
where P (t) = |A (t)|2 is the power of the signal. The operation of intensity modulation therefore
consists in varying P (t) according to the modulating electrical signal. However, the desired
intensity modulation of the lightwave is often accompanied by a modulation of its phase φ
induced by the physical process used to realise the intensity modulation. Consequently, not
only the power P (t) becomes a function of time, but also the phase φ (t), which is very often
an undesired feature. Observing eq. (3) and (4), we can define an instantaneous frequency of
the optical signal according to
∂φ
ω (t) = ω0 +
.
(5)
∂t
Consequently, a time varying phase is equivalent to a change in the signal instantaneous frequency. This frequency modulation is usually referred to as frequency chirping. The amount
of frequency chirp depends on the physical mechanism used to achieve light modulation, as well
as on the design and operating conditions of the modulator.
Since the group velocity through an optical fibre is frequency dependent, an effect known
as group velocity dispersion, the different frequency components of the spectrum of a pulse
injected at the fibre input will travel at different speeds, hence leading to pulse broadening. In
case an intensity modulated signal is transmitted, the information pulses will spread out of their
allocated time slots, leading to inter-symbol interference, which will in turn introduce errors
at the decision circuit (where the received signal is compared to a given threshold to decide
whether a “mark” or “space” has been transmitted) and lead to a degraded bit-error-rate.
Intuitively, the effect of frequency chirping will be to broaden the spectrum of the modulated
signal. As the effect of dispersion worsens with increasing signal spectral width3 , frequency
chirping will, in general, result in reduced tolerance to group velocity dispersion.
We now make abstraction of the mechanism responsible for light modulation, and consider
the propagation of a single chirped pulse in an optical fibre. For simplicity, we assume a linearly
chirped Gaussian pulse4 and consider its dispersion induced broadening when propagating along
a fibre whose dispersion is described by the β2 parameter5 . In this case, the complex envelope
can be expressed as
"
2 #
t
1 − iC
A (t) = A0 exp −
,
(6)
2
T0
where T0 can
√ be related to the full-width half-maximum of the signal TFWHM according to
TFWHM = 2 ln 2 T0 . The instantaneous frequency of the signal is therefore
ω (t) = ω0 +
C
t,
T02
(7)
which varies linearly with time, hence the “linearly chirped” designation used to describe such
a pulse. One of the nice features of our choice of a linearly-chirped Gaussian pulse is that it
3
This is the reason why the effect of dispersion becomes more critical at high bit rates. If the system is limited
by dispersion, the maximum tolerable transmission distance for 1 dB power penalty is about 60 km over standard
single-mode fibre for 10 Gbit/s non return-to-zero (NRZ) modulation. Since the effect of dispersion scales as the
square of the bit rate, this distance will be reduced to about 4 km when the bit rate is increased to 40 Gbit/s.
Hopefully some techniques exist that can compensate for the effects of dispersion and allow for long distance
transmission at high bit rates.
4
It is important to emphasise that this is an assumption for the present calculation. Usually, the frequency
chirp generated in transmitters is not linear. However the present discussion based on a linear chirp assumption
will prove useful to qualitatively discuss the interaction of transmitter chirp and group velocity dispersion.
5
β2 is the second derivative with respect to angular frequency ω of the propagation constant β. Therefore β2
characterises the second-order dispersion and can be related to the usual fibre dispersion parameter D, usually
β , where λ is the wavelength and c is the velocity of light
expressed in units of ps/(nm·km), through D = − 2πc
λ2 2
in vacuum. For the commonly used standard single-mode fibres, D ≈ 17 ps/(nm·km) at 1550 nm, and therefore
β2 is negative.
22/11/2009
5
34129 Experimental Course in Optical Communication
enables to calculate analytically the pulse shape after propagation over a distance z through
an optical fibre, and hence to deduce the pulse width. The results are shown in Fig. 2 where it
can be seen that three different propagation regimes can be identified depending on the value
of the product of the chirp parameter C and the dispersion parameter β2 . If Cβ2 > 0, the pulse
will broaden faster than in the case of an unchirped signal for which Cβ2 = 0. On the other
hand, if Cβ2 < 0, the pulse will initially compress, before broadening at a faster rate than for
an unchirped pulse. This example outlines the influence of the chirp on signal propagation over
an optical fibre. In many cases, having an unchirped signal at the transmitter output will be a
desired feature, unless some benefit can be found from the chirp for special applications, such
as e.g. pulse compression.
3
External modulation
Two types of external modulators commonly used in optical communication systems are briefly
introduced and their main features are outlined. The first type relies on the modification of
the absorption of a semiconductor material when an external electric field is applied (electroabsorption modulator), while the second type is based on the change of the refractive index
observed for some crystals under an external electric field (electro-optic modulator). A change
in refractive index itself does not permit modulation of the intensity of a lightwave. However,
using an interferometric structure, such as the Mach-Zehnder structure you have studied in
previous exercises, enables to convert the induced phase modulation into the desired intensity
modulation.
3.1
Electro-absorption modulators
This type of modulator relies on the fact that the effective bandgap Eg of a semiconductor
material decreases when an external voltage is applied. Consequently, if the frequency ν of
an incoming lightwave is chosen so that its energy E = hν is smaller than the bandgap when
no voltage is applied, the material will be transparent. On the other hand, when an external
voltage is applied, the effective bandgap will be reduced, meaning that the lightwave will be
absorbed by the material when E > Eg . Such a shift of the absorption edge of a semiconductor
under the influence of an external voltage is represented in Fig. 3. By properly selecting the
signal wavelength so that it experiences a significant change in absorption when the voltage
is applied, it thus becomes possible to achieve optical modulation controlled by an electrical
signal. A typical absorption versus applied voltage transfer function for an electro-absorption
modulator is also shown in Fig. 3.
Since the absorption and refractive index of a semiconductor material are related by KramersKronig relations of the type
∆n (ω) =
c
π
Z
+∞
0
∆α (ω ′ )
dω ′ ,
ω ′2 − ω 2
(8)
where ∆n is the change in refractive index induced by the change in absorption ∆α, and c is
the velocity of light in vacuum, shifting the absorption edge to achieve optical modulation will
also induce a change in the refractive index of the material, hence in the phase or instantaneous
frequency of the signal being modulated. Consequently, some amount of frequency chirping will
be introduced by electro-absorption modulators. However, the generated frequency chirp will
usually be smaller than when direct current modulation of a semiconductor laser is used.
6
22/11/2009
Direct & External Modulation
30
fibre-to-fibre loss (dB)
absorption
V=0
V≠0
25
20
15
10
5
0
1
wavelength
2
3
4
5
voltage (V)
Figure 3 Left: absorption of a semiconductor as a function of wavelength with and without an external
applied electric field (adapted from [2]). Right: typical loss versus applied voltage curves for an electroabsorption modulator (adapted from [1]).
1.0
0.5
t
0.0
0
1
2
3
4
a)
b)
t
Figure 4 Principle of operation of a Mach-Zehnder modulator.
3.2
Electro-optic modulators
The refractive index of some materials can be modified by applying an external electric field to
them through the linear electro-optic effect. Since the phase shift experienced by a lightwave
of wavelength λ propagating through a length L of a medium with refractive index n is
φ=
2π
nL,
λ
(9)
a straightforward application is the realisation of phase modulators made from an electro-optic
waveguide subjected to a time dependent electric field. The applied voltage will modulate the
refractive index of the waveguide material, hence the phase shift experienced by a lightwave
propagating along the waveguide. However, legacy optical communication systems typically rely
on intensity modulation of light. This can be achieved by transforming phase modulation
induced by the electro-optic effect to intensity modulation using an interferometric
structure.
In order to illustrate the principle, we consider the simple interferometric structure represented in Fig. 4. It is based on a Mach-Zehnder interferometer including one electro-optic mate22/11/2009
7
34129 Experimental Course in Optical Communication
rial in one of the arms. In practice, this interferometer is realised by implanting waveguides into
an electro-optic crystal, typically lithium-niobate (LiNbO3 ). Assuming a power splitting and
combining ratio of 12 for the input and output couplers of the Mach-Zehnder interferometer, the
power at the output of the interferometer depends on the phase shift difference ∆φ = φ (t) − φ0
experienced by the light propagating in the upper and lower arms of the structure according to
∆φ
.
(10)
2
The phase shift induced in the upper arm of the interferometer depends on its refractive index,
which itself depends on the applied external electric field through the electro-optic effect. If a
time-dependent voltage V (t) is applied to the upper waveguide of the modulator, its refractive
index will become time-dependent and, in turn the transmission of the Mach-Zehnder interferometer will also depend on time. If a continuous optical wave is applied to the input of the
modulator, the output power will thus be modulated according to the electrical data V (t). The
value of the phase shift created by an applied external voltage depends upon many parameters,
including the choice of the electro-optic material, the orientation of the crystal with respect to
the external electric field, as well as to the polarisation of the incoming lightwave, the geometry
and dimensions of the waveguide. In any case it is possible to make abstraction of the actual
physical implementation of the modulator and describe the ability of the material and chosen
configuration to respond to an applied voltage by introducing a quantity known as the half-wave
voltage Vπ . Applying a voltage of Vπ to the electrode of an electro-optic waveguide will result
in a voltage-induced phase shift of π. The voltage-induced phase shift φ (t) can therefore be
related to the applied voltage V (t) according to
Pout = Pin cos2
φ (t) = π
V (t)
.
Vπ
(11)
Through eq. (10) and (11), it is then possible to calculate the transfer function Pout /Pin of the
modulator as a function of the applied voltage. Such a transfer function, where the applied
voltage has been normalised to the half-wave voltage, is also represented in Fig. 4.
It can easily be shown that, if the Mach-Zehnder configuration of Fig. 4 is used, the optical
modulated signal will be chirped. The problem can be solved by applying two complementary
modulating signals to the two arms of the Mach-Zehnder modulator. If one arm is driven with a
voltage corresponding to the data to be transmitted d (t), while the second arm is driven with a
voltage corresponding to the complementary data d (t), then it can be shown that the chirp can
be suppressed. Mach-Zehnder modulators are usually employed this way, a technique known as
push-pull modulation, either by driving the two arms of the interferometer with complementary
signals, or by creating phase shifts of opposite signs in each arm by a proper configuration of
the crystal and electrodes.
4
Direct current modulation of semiconductor lasers
You already know from a previous experiment that the power at the output of a semiconductor
laser depends on the current injected through the laser diode according to the transfer function
represented in Fig. 5. First, no light (apart from spontaneous emission) is emitted by the
laser diode, until the current reaches the threshold value Ith . Above threshold, population
inversion is achieved, leading to lasing action. The laser output power then increases linearly
with increasing current, until some saturation is reached for high bias current values. Obviously,
this dependence of the laser diode output power on the bias current can be exploited to convert
the information from the electrical domain to the optical domain, if we let the current vary
according to the data to be transmitted.
8
22/11/2009
Direct & External Modulation
Figure 5 Ilustration of the concept of direct current modulation of a semiconductor laser. The laser
driving current is varied according to the modulating signal, resulting in modulation of the emitted
power. The modulated optical signal is represented for two different values of the bias current, resulting
in improved dynamic behaviour, but poorer extinction ratio when the bias current increases.
4.1
Turn-on delay
One could in principle achieve an infinite extinction ratio by letting the laser driving current
below threshold in order to generate “spaces” and increasing it to an above-the-threshold value
whenever “marks” need to be modulated. Under these conditions, the laser switches from a state
where no lasing occurs, to a state where population inversion is achieved, resulting in lasing
operation. However, population inversion is achieved by injecting carriers into the structure
and it takes some time for the carrier density to reach its threshold value when lasing begins.
Consequently lasing will be delayed from the time the driving current is increased by a time td
known as turn-on delay. The turn-on delay can be approximated by
td = τc ln
I1 − I0
I1 − Ith
,
(12)
where I1 and I0 are the driving currents corresponding to “marks” and “spaces”, respectively,
and τc is the carrier lifetime, of the order of a few nanoseconds. Such a delay might not be
compatible with high speed operation, for instance at 10 Gbit/s, where the bit slot is only
100 ps long.
4.2
Extinction ratio
We have seen above that high speed operation of a directly modulated laser diode will require
to operate above threshold, even when a “space” is generated, in order to avoid the turn-on
delay. In this case, lasing occurs all the time, but the laser output power will take one of the
two values, P1 or P0 6= 0, for “marks” and “spaces”, respectively, depending on the value of the
bias current. Consequently, an infinite extinction ratio will not be achieved.
22/11/2009
9
34129 Experimental Course in Optical Communication
4.3
Bandwidth
Furthermore, the modulation bandwidth of a directly modulated laser can be shown to increase
with the bias current according to
∆f3dB
3GN (Ib − Ith )
≈
4π 2 e
1/2
,
(13)
where GN is related to the dependence of the rate of stimulated emission on the carrier number,
Ib and Ith are the bias and threshold current, respectively, and e is the elementary charge. High
modulation bandwidths suitable for high speed operation will therefore be obtained for large
bias currents. Note however that the modulation bandwidth as stated above is defined under
small signal modulation conditions (i.e. the electrical modulating signal is taken as a sinusoidal
signal whose peak-to-peak current is small compared to Ib − Ith ). However it can still provide
valuable qualitative information under large signal modulation such as when the semiconductor
laser is used for digital communication.
4.4
Relaxation oscillations
When the laser is subjected to a transient in its bias current, for instance during a transition
from a “mark” to a “space” or vice versa, its output power will exhibit damped oscillations
known as “relaxation oscillations”. Both the oscillation frequency and damping rate depend
on the laser output power, hence on the value of the driving current. The physical origin of
those oscillations is the interplay between the injected carriers and emitted photons. It can be
shown that the relaxation oscillation frequency and the damping rate increase with increasing
current. One will therefore pay attention to those oscillations when the laser is biased close to
threshold, as illustrated in Fig. 6.
4.5
Frequency chirping
When the laser is directly modulated, a change in the bias current will lead to a change in the
carrier density, which in turn will lead to a change of the refractive index of the semiconductor
material. Since the lasing wavelength is determined from the feedback condition in the laser
cavity, which itself depends on the refractive index6 , the instantaneous frequency of the emitted
signal will be a time varying signal. Consequently, directly modulated lasers are inherently
chirped, an effect that has prevented their use for the generation and transmission of high bit
rate signals. The variations of the instantaneous frequency at the output of a directly modulated
laser can be shown to be equal to
α
∆ν (t) =
4π
d
[ln Pe (t)] + κPe (t) ,
dt
(14)
where α is known as the linewidth enhancement factor and κ is the adiabatic chirp
coefficient. It can be seen that the chirp consists of two terms. The first term, named
transient chirponly exists when the emitted power varies with time, for instance during
transients of the applied current I (t) and induced relaxation oscillations, while the second
term, named adiabatic chirp, is responsible for the different emission frequencies observed
under steady state when a “mark” or “space” is transmitted.
6
For instance the resonance condition in a distributed feedback laser can be expressed as Λ = mλ/(2n), where
Λ is the period of the refractive index corrugation, n is the mode index, λ is the wavelength and m is an integer.
10
22/11/2009
Direct & External Modulation
4.6
Summary
The points above are illustrated in Fig. 6 and 7 where the operation of a single mode semiconductor laser has been simulated numerically for bias currents7 of 10 and 30 mA, respectively,
and peak-to-peak current of 20 mA. For this laser, the threshold is equal to about 5 mA. The
laser is operated at a bit rate of 2.5 Gbit/s.
16
0
14
-20
12
power (dBm)
power (mW)
-40
10
8
6
-60
-80
-100
4
-120
2
-140
0
0
1
2
3
4
5
6
-200
-150
time (ns)
-100
-50
0
50
100
150
200
frequency detuning (GHz)
15
frequency chirp (GHz)
10
5
0
-5
-10
-15
0
1
2
3
4
5
6
time (ns)
Figure 6 Simulated waveform, frequency chirp and spectrum at the output of a directly modulated
laser for a bias current of 10 mA. The laser threshold is ∼ 5 mA and the bit rate is 2.5 Gbit/s.
In the case of 10 mA bias, strong relaxation oscillations are visible for both “marks” and
“spaces”. By comparing those oscillations, it can be checked that both the relaxation oscillation
frequency and the damping rate increase with increasing laser driving current. The relaxation
oscillations constitute a fast variation of the laser output power, hence the laser chirp is dominated by the transient component. Only during long sequences of consecutive “marks” does
the output power stabilise to its steady state value. This dominant transient chirp behaviour
results in significant broadening of the signal spectrum.
On the other hand, when the bias current is increased to 30 mA, the highly damped relaxation oscillations result in well defined “mark” and “space” levels, hence to a signal waveform
that does not present as much distortion as in the 10 mA bias case. However, the power level
corresponding to the “spaces” is higher than in the previous case and the extinction ratio is
decreased. Apart from during transients of the modulating signal, a strong adiabatic chirp component is visible, leading to distinct frequencies when “spaces” and “marks” are transmitted.
The dominant adiabatic chirp can also be observed in the spectrum, which is narrower than in
the previous case and exhibits characteristic peaks at the two frequencies corresponding to the
“marks” and “spaces”.
7
Defined here as the driving current corresponding to the “spaces”.
22/11/2009
11
34129 Experimental Course in Optical Communication
16
0
14
-20
12
power (dBm)
power (mW)
-40
10
8
6
-60
-80
-100
4
-120
2
-140
0
0
1
2
3
4
5
6
time (ns)
-200
-150
-100
-50
0
50
100
150
200
frequency detuning (GHz)
15
frequency chirp (GHz)
10
5
0
-5
-10
-15
0
1
2
3
4
5
6
time (ns)
Figure 7 Simulated waveform, frequency chirp and spectrum at the output of the same directly modulated laser as the one used to generate the results of Fig. 6 , but for an increased bias current value of
30 mA.
5
Description of the exercise
Figure 8 Picture of the packaging of the type of 10 Gbit/s directly modulated laser used in the
experiment. Source: NEL.
In this exercise, we study the static and dynamic characteristics of a state-of-the-art commercially available distributed feedback (DFB) directly modulated laser (DML) designed for
10 Gbit/s operation. The device is manufactured by NEL (NTT Electronics Corporation) in
Japan and its packaging is represented in Fig. 8. The modulation input (coaxial connector)
is clearly visible on one side of the packaging, while the other pins are for the bias current,
temperature control (temperature sensor and thermo-electric cooler) and monitoring.
12
22/11/2009
Direct & External Modulation
Important:
• Do not apply any modulation to the DML before having adjusted the bias to a sufficiently
large value. This might otherwise result in reverse biasing the laser diode junction, which
might irremediably destroy the laser. When you switch the laser off, turn off the modulation first, then decrease the bias current.
• When you deliver some radio-frequency (RF) power, for instance the laser modulating
signal, always ensure that this is done to a matched impedance (typically 50 Ω) device.
Consequently, always switch on the bias to an electrical amplifier before you actually turn
the input signal on. Turn on the path of an RF signal from load to source. Switch it off
from source to load.
• Never disconnect an optical connector without having ensured first that the light has been
switched off. Failure to comply might result in damage to the connector and potential
eye hazard.
• Handle the optical connectors with care. Carefully align the connectors to the adapter
before making a connection. Clean the connector and check it through the inspection
microscope before making any connection.
• In case of doubt, remember there is nothing wrong about asking!
Based on some of the recommendations above, always follow the following laser turn-on
procedure:
1. First of all, ensure that you know the value of the peak-to-peak current you will apply to
the laser. Check the modulating signal peak-to-peak voltage on an oscilloscope and use
the conversion table provided to assess the corresponding peak-to-peak current. This will
enable you to ensure the laser bias current is sufficiently high so that there is no risk to
damage the laser by reverse biasing the laser diode.
2. Make sure the temperature controller output is turned on.
3. Increase the bias current to a sufficiently large value, depending on your modulating signal
peak-to-peak current.
4. Switch on the laser driver amplifier.
5. Turn the data output of the pattern generator on.
In order to switch the laser off, take the steps above in reverse order. When turning on the
laser, a safe approach is to increase the bias current above the desired value (but not too high:
never exceed 100 mA), switch on the modulation, then decrease the bias current. Note that this
procedure is safe for the model of laser diode and modulating signal generation method used in
this experiment (where the bias current will correspond to the value of the driving current for
the “marks”). However a different procedure might be required for different laser models and
experimental set-ups.
The experimental configuration is represented in Fig. 9. The actual set-up might actually
slightly differ from the one represented in the figure. It is your responsibility to draw the actual
set-up used for your investigations in your laboratory book.
22/11/2009
13
34129 Experimental Course in Optical Communication
oscilloscope
10 Gbit/s
pattern
generator
trigger
ch1 ch2
data
current
source
electrical
amplifier
DML
3 dB
t°controller
SMF
EDFA
optical
spectrum
analyser
Figure 9
5.1
Experimental set-up.
Static characterisation of a DFB laser
In this first part of the exercise, the optical power versus current characteristic of the semiconductor laser is measured. This static characterisation (i.e. in the absence of modulation) will
be used later to determine the working point (bias) and peak-to-peak modulation current to
apply to the laser for 10 Gbit/s modulation.
Assignment:
1. Connect directly the output of the DML to the optical spectrum analyser.
2. Gradually increase the laser bias current from 0 to about 80 mA. Once the threshold has
been reached, record the signal power and its centre wavelength on the optical spectrum
analyser. Use a resolution bandwidth of 0.1 nm.
3. Plot the power versus bias current and wavelength versus bias current characteristics of
the laser.
4. Extract the value of the laser threshold.
5. Based on the measured static characteristic, decide on the laser biasing point and peakto-peak current that you will apply for 10 Gbit/s modulation of the laser. Comment your
choice.
5.2
Dynamic characterisation of a DFB laser
We now apply a 10 Gbit/s NRZ signal to the laser. The signal is a pseudo-random binary
sequence (PRBS) generated from a bit pattern generator that is amplified using a 12.5 Gbit/s
driver amplifier. At the output of the amplifier, the signal is split in a 6 dB power splitter so
that it is possible to monitor it on a high-speed sampling oscilloscope. In this exercise, the
peak-to-peak modulating current has been predefined to ensure safe operation of the laser.
14
22/11/2009
Direct & External Modulation
Assignment:
1. Bias the laser diode to a sufficiently large value (e.g. 60 mA)
2. Turn on the modulating signal. Please refer to the turn on procedure described above.
3. Monitor the electrical driving signal. Record the peak-to-peak voltage and calculate the
corresponding peak-to-peak current.
4. Vary the bias current and examine the signal waveform or eye diagram, as well as the
signal spectrum. Compare the eyes and spectra for different values of the bias current.
5. Record typical eye diagrams and spectra depending on the operating conditions (bias
current) of the laser.
6. Provide an interpretation for the differences observed in the waveforms and spectra under
small and large bias.
5.3
Transmission over standard single mode fibre
The signal generated in the previous step is now transmitted through a length of 35 km standard
single-mode fibre (SMF). The signal quality after transmission will be assessed by observing
the eye diagrams. In particular, the influence of the bias current will be investigated, and the
signal behaviour will be related to the knowledge you gained from the theoretical section and
the dynamic characterisation of the laser. In a final step, the directly modulated laser will be
replaced by a transmitter consisting of a continuous wave laser and a chirp-free electro-optic
Mach-Zehnder modulator, and the signal quality after transmission will be compared.
Assignment:
1. Based on the dynamic characterisation of the laser, select a bias current you believe is
suitable for the generation of 10 Gbit/s signals.
2. Switch the laser off according to the turn-off procedure described above, and connect
35 km SMF at the output of the laser. An erbium doped fibre amplifier (EDFA) will be
connected to the output of the fibre spool in order to compensate for the fibre loss that
is of the order of 0.2 dB/km.
3. Observe the signal waveform and eye diagram after propagation through 35 km SMF.
Comment.
4. Generate a 10 Gbit/s signal using the external lithium-niobate Mach-Zehnder modulator
provided. Observe the waveform or eye diagram at the output of the modulator.
5. Transmit the externally modulated 10 Gbit/s signal through the 35 km SMF spool and
monitor the eye diagram at the fibre output. Compare your observations with the ones
you made in point 3 above.
6. If time allows, we will also show that some special type of optical fibres, known as dispersion compensating fibres (DCFs) can be used to compensate for the dispersion accumulated over transmission in SMF. We will add a suitable piece of DCF after the SMF in
the direct modulation case and will observe the waveform after propagation through the
SMF+DCF.
22/11/2009
15
34129 Experimental Course in Optical Communication
References
[1] T. Ido, S. Tanaka, M. Suzuki, M. Koizumi, H. Sano, and H. Inoue, “Ultra-high-speed
multiple-quantum-well electro-absorption optical modulators with integrated waveguides,”
Journal of Lightwave Technology, vol. 14, no. 9, pp. 2026–2034, Sep. 1996.
[2] A. Ramdane, F. Devaux, N. Souli, D. Delprat, and A. Ougazzaden, “Monolithic integration
of multiple-quantum-well lasers and modulators for high-speed transmission,” IEEE Journal
of Selected Topics in Quantum Electronics, vol. 2, no. 2, pp. 326–335, Jun. 1996.
16
22/11/2009