You are on page 1of 17

34130 Introduction to Optical Communication

Direct Current Modulation of Semiconductor Lasers


Christophe Peucheret DTU Fotonik Department of Photonics Engineering Technical University of Denmark cpeu@fotonik.dtu.dk

Abstract This note describes the simulation exercise of the course 34130 Introduction to Optical Communication on direct modulation of semiconductor lasers. After a review of the theory of continuous wave operation and small-signal modulation of single-mode laser diodes based on the resolution of the rate equations for the carriers and photons, the three assignments are described. They make use of a numerical model allowing the resolution of the rate equations implemented in the VPIplayer environment. The main goals of the exercise are: 1) to calculate the static power-versus-current characteristics of the laser, 2) to calculate its small-signal frequency response, and 3) to examine its dynamic behaviour under large-signal modulation.

Introduction

The goal of this simulation exercise is to illustrate a number of important concepts that have been described in the lectures on sources and transmitters. This will be achieved by using a numerical simulation model of a single-mode directly modulated laser and comparing the results with simple calculations based on the theory developed in the lectures on semiconductor lasers. We will focus on the dynamic properties of the laser and calculate its small-signal frequency response, as well as study its behaviour under large-signal modulation.

Theoretical background

Some important results that have been discussed in the lectures and that are detailed in the textbook are summarised below. Furthermore, some derivations of the small-signal frequency response of single-mode laser diodes, of relaxation oscillations and of frequency chirp are provided. The main results of this section will be used in the preparation of the exercise (Sect. 3) and in the exercise itself (Sect. 4 to 6).

2.1

Rate equations

The rate equations for the photon number P and carrier number N can be written dP dt dN dt Simulation Exercise 2 = GP + Rsp = P , p (1) (2) 1

I N GP, e c

14/07/2011

34130 Introduction to Optical Communication

where Rsp is the rate of spontaneous emission in the lasing mode, p the photon lifetime, c the carrier lifetime, G the net rate of stimulated emission, I the intensity of the current applied to the laser diode, and e the elementary charge of the electron. It has been seen in the lectures that the dependence of the peak material gain on the carrier density can be empirically approximated by gm = g (N N0 ) V (3)

where g is the gain cross section, or dierential gain, V is the active volume and N0 is the carrier number at transparency. Due to the spreading of the optical mode outside the active layer, the power gain at the mode frequency g needs to be corrected according to g = gm where is the connement factor. The net rate of stimulated emission G can then be calculated according to G = vg gm = GN (N N0 ) where GN = vg g /V and vg is the mode group velocity within the cavity. The feedback condition imposes the following relation between the gain g, the internal loss per unit length int , the cavity length L and the reectivities of the two facets R1 and R2 g = int + 1 ln 2L 1 R1 R2 = int + mir . (5) (4)

The photon lifetime is directly related to the cavity loss cav = int + mir according to 1 = vg (int + mir ) . p (6)

Finally, the carrier lifetime accounts for carrier recombination mechanisms that do not contribute to the stimulated emission, including spontaneous emission and non-radiative recombination (due to e.g. traps and defects as well as Auger processes), 1 = Anr + BN + CN 2 . c (7)

Here, Anr N is the recombination rate due to traps and defects, BN 2 is the spontaneous emission rate, and CN 3 is the recombination rate induced by Auger processes. If is the fraction of spontaneous emission that falls into the lasing mode, Rsp can be expressed as a function of the carrier number according to Rsp = BN 2 . (8)

2.2

Steady state operation

The rate equations (1)-(2) can be solved easily under steady-state operation of the laser. In this case, the applied current I is constant and dN = 0, dt dP = 0. dt (9)

If spontaneous emission is neglected (Rsp = 0), (1) leads to P Simulation Exercise 2 G 1 p = 0. (10) 2

14/07/2011

34130 Introduction to Optical Communication

Apart from the trivial solution P = 0, N = c I/e corresponding to biasing of the laser below threshold, (10) leads to 1 (11) G= . p Reporting into (3), it can be seen that the number of carriers is constant and clamped to the threshold value Nth 1 N = N0 + Nth . (12) GN p Hence the number of photons, P = p e I eNth c = p (I Ith ) , e (13)

from which the power emitted by the laser can be calculated according to Pe = 1 vg mir P. 2 (14)

In order to derive (14), it has been assumed that the two laser cavity facets have the same reectivities R1 = R2 . It will furthermore be assumed that the internal quantum eciency int is equal to 1 (meaning that all injected electrons contribute to stimulated emission above threshold), and that the optical coupling eciency is also equal to 1 (all the emitted optical power can be collected at the output of the laser chip). The laser output power can also be expressed in terms of the external quantum eciency ext 1 hI Pe = ext , 2 e which can also be written (15)

1 h Pe = d (I Ith ) , (16) 2 e where d is the dierential quantum eciency, which veries ext = d (1 Ith /I ) and can be expressed as int mir d = . (17) mir + int

2.3

Small-signal modulation

Remark: A detailed derivation of the small-signal frequency response of the laser is presented in this section. Its main purpose is to complement the textbook and show how the important results on the laser modulation bandwidth can be obtained from the rate equations. Although some of the equations might look scary at rst, the calculations do not present major diculties. In case of emergency, the main results that are necessary for the simulation exercise can be found in equations (39)-(52). The laser is now biased above threshold with a current Ib , resulting in a photon number Pb . If a small signal modulation I (t), with |I (t)| |Ib Ith | for all t, is added to the bias current I (t) = Ib + I (t) , (18) the photon number is expected to deviate from Pb according to P (t) = Pb + P (t) , Simulation Exercise 2 14/07/2011 (19) 3

34130 Introduction to Optical Communication

and similarly for the carrier number N (t) = Nb + N (t) . (20)

An empirical relation between the gain G and the carrier number N has been introduced in (3). However this relation does not account for observed gain saturation eects, making it necessary to also introduce a dependence of G on the photon number P . An explicit relation between G, N and P will be introduced in (42). For the time being, the dependence of the gain on the carrier and photon numbers will be taken into account by performing a rst order Taylor expansion around the biasing point G = Gb + G N N + G P P, (21)

where the partial derivatives are taken at P = Pb and N = Nb . It should also be remembered at this point that both the carrier lifetime and the spontaneous emission rate depend on the carrier number N through (7) and (8). They can therefore be expressed according to Taylor expansions of type Rsp N. (22) Rsp = Rsp (Nb ) + N

The rate equations (1)-(2) are veried by the steady state values, leading to 0 = Pb Gb 0 = 1 p + Rsp (Nb ) , (23) (24)

Ib Nb Gb Pb , e c (Nb )

where the dependence of the spontaneous emission rate Rsp and the carrier lifetime c on the electron number has been made more explicit. Under small-signal modulation, inserting the expressions for the current and the numbers of photons and carriers (equations (18) to (20)), as well as the Taylor expansions for the gain, the carrier lifetime and the spontaneous emission rate into the rate equations (1) and (2), and conserving only rst order terms in P and N leads to dP 1 = Gb + dt p and dN = Gb + dt Introducing the quantities P P = Rsp Pb G Pb , P G Pb , P P N = NN = Rsp G Pb , N N G 1 (1/c ) Pb + + Nb , N c N (27) (28) G P Pb P G N Pb + 1 (1/c ) I + Nb N + . c N e (26) G P Pb P + G N Pb + Rsp N, N (25)

NP = Gb +

Simulation Exercise 2

14/07/2011

34130 Introduction to Optical Communication

the system of equations (25)-(26) can be written under the simpler form dP = P P P P N N, dt I dN = NP P NN N + . dt e (29) (30)

If it is furthermore assumed that the current driving the laser varies sinusoidally with angular frequency m and amplitude Im around the biasing point Ib I (t) = Ib + Im cos (m t) , (31)

then, under the small-signal approximation, one also expects the photon and carrier numbers to vary sinusoidally at m around the steady-state values Pb and Nb , respectively. Adopting complex notations, the variations of the current, photon and carrier numbers can be written I (t) = Im ejm t (32) (33) (34)

N (t) = Nm e

P (t) = Pm ejm t

jm t

Here, Pn and Nm are complex in order to account for eventual phase shifts between the photon and carrier numbers and the modulating current. The system of equations (29)-(30) can then be expressed in a simple form according to (jm + P P ) Pm + P N Nm = 0,
Im e ,

(35) (36)

NP Pm + (jm + NN ) Nm = which can be easily solved, leading to Pm = Rearranging, Pm = where

P N Im /e . (jm + P P ) (jm + NN ) P N NP P N Im /e , (R + m j R ) (R m + j R ) 1 (NN P P )2 4


1/2

(37)

(38)

R = P N NP and R =

(39)

1 (P P + NN ) (40) 2 will be related to the relaxation oscillations frequency and damping rate in Sect. 2.4. Hence the small-signal frequency response for the photon number and therefore the laser output power, H (m ) =
2 Pm (m ) 2 R + R = . Pm (0) (R + m j R ) (R m + j R )

(41)

So far no approximation has been made on G, as well as on the various terms involved in the expression of the ij coecients in (27) and (28). The empirical expression for the net rate Simulation Exercise 2 14/07/2011 5

34130 Introduction to Optical Communication

of stimulated emission (3) is rened further by taking into account gain saturation due to e.g. spectral hole burning, G = GN (N N0 ) (1 NL P ) . (42)

The gain suppression factor NL accounts for the observed reduction of the gain for large values of P . The following approximations can then be made: G N G P hence P P P N and Rsp + NL Gb Pb , Pb GN Pb , NN GN Pb + NP Gb , 1 , c (45) (46) = GN (1 NL Pb ) GN , = NL GN (Nb N0 ) NL Gb , (43) (44)

P =P b

N = Nb

1/2 1 Gb GN Pb . (47) (NN P P )2 4 Using (11) and (13), the relaxation frequency can be linked to the biasing conditions of the laser

R Gb GN Pb

GN (Ib Ith ) 1/2 . (48) e Finally, the 3-dB modulation bandwidth of the laser can be obtained from (41) by solving the equation R 1 |H (2 f3dB )| = , 2 hence 1 2 4 4 2 2 2 R R + 2 R + R + R R 2 which can often be approximated by 3R . f3dB 2 f3dB =
1/2 1/2

(49)

(50)

(51)

The modulation transfer function can be normalised according to 1 H ( ) = 2 1 + j 2 0 0

(52)

2 = 2 + 2 and = / . It can easily be shown that the modulus of the transfer where 0 R 0 R R 2 function (52) presents an extremum for 2 = 2 R R . The value of this peak is equal to 2 1/2 1 2 which is minimum for = 1/ 2. This situation occurs for 2 R = R , resulting in a maximum corresponding to 1 reached for = 0. Normalised small-signal frequency responses corresponding to dierent values of the damping parameter are represented in Fig. 1, showing the existence of a peak associated to relaxation oscillations for < 1/ 2.

Simulation Exercise 2

14/07/2011

34130 Introduction to Optical Communication

10

= 0.05 = 0.1 = 0.2 = 0.707

=1 =5

| ()|

0.1

0.01 0.01 0.1 1 10

/ 0

Figure 1 Magnitude of the small-signal frequency response |H ( )| as a function of the normalised frequency /0 for several values of the normalised damping parameter = R /0 .

2.4

Relaxation oscillations

The parameters R and R have been described earlier as the relaxation oscillations frequency and damping rate. In this section, this terminology is claried by looking at the response of the laser to a perturbation of its steady state operation condition, for instance in response to a current transient. Assuming a small perturbation from the equilibrium condition, the linearised rate equations (29) and (30) can be used. In the present case, the solutions are assumed to be of the form P (t) = P0 et , N (t) = N0 e
t

(53) (54)

where is a complex constant. Inserting into the rate equations leads to P P P0 + P N N0 = P0 , NP P0 + NN N0 = N0 , which can be written in a matrix form P P NP P N NN P0 N0 = P0 . N0 (57) (55) (56)

Consequently is an eigenvalue of the matrix = P P NP P N NN , (58)

which can be found by solving the second degree equation det ( I) = 0, where I is the unity matrix. The solutions are 1 1 = (P P + NN ) j P N NP (NN P P )2 2 4 Simulation Exercise 2 14/07/2011
1/2

(59) 7

34130 Introduction to Optical Communication

which can be expressed with the notations introduced in (39) and (40) as = R j R . (60)

Consequently, the carrier and photon numbers exhibit an oscillatory behaviour with a pseudoperiod TR = 2/R , while the amplitude of the oscillations decays according to eR t .

2.5

Frequency chirping

Remark: In this section, the change in emission frequency when the laser driving current is varied is derived. The derivation may be skipped, but the mechanism responsible for this frequency chirping should be understood and equation (74) will be used in the exercise. A change in carrier density under current injection will lead to a change in refractive index, which in turn will modify the lasing frequency, whether the laser is of the Fabry-P erot type (resonance condition L = m/(2n), where L is the cavity length, n the mode index and m an integer), or makes use of a distributed-feedback (DFB) structure (resonance condition = m/(2n), where is the period of the grating). When driven by a time varying signal, the instantaneous lasing frequency will vary as a function of time, an eect known as frequency chirping. The relation between the frequency shift (t) induced by frequency chirping and the waveform at the laser output Pe (t) is derived in this section. First, the lasing frequency shift induced by a change in carrier number is calculated. For a Fabry-P erot laser, the lasing frequencies correspond to = pc , 2nL (61)

where p is an integer and L is the cavity length. The eective index n is frequency dependent (dispersion) and also depends on the carrier number N . Assuming these variations to be relatively small, a rst order Taylor expansion of the eective index about its value at threshold can be performed, leading to n (, N ) = n (th , Nth ) + Inserting (62) into (61) leads to ( th ) nth + th n = th n (N Nth ) , N (63) n n ( th ) + (N Nth ) . N (62)

where nth + th n/ is equal to the group index ng . Hence the deviation from the threshold frequency under the inuence of a change in carrier number th = th n (N Nth ) . ng N (64)

Next, a relation between N Nth and the photon number P is established. For this purpose, the gain G can be expanded according to (21) around a working point just above the laser threshold. In these conditions, Gb 1/p and Pb 0, hence G= Simulation Exercise 2 1 G G + (N Nth ) + P. p N P 14/07/2011 (65) 8

34130 Introduction to Optical Communication

Reporting into the rate equation for the photons (1), the desired relation between N Nth and P is obtained: (N Nth ) Rsp G G 1 dP = P. N P dt P P (66)

If furthermore a dependence of the gain on the carrier and photon numbers according to (42) is assumed, and with the usual approximations (43)-(44), N Nth = 1 GN Rsp NL 1 dP + P . P dt P p (67)

In a third and nal step, the relation between n/N appearing in (64) and G/N appearing in (66) is established. A general expression for the electric eld propagating in the +z direction in the cavity is E (x, y, z, t) = E0 U (x, y ) e 2 (gint )z ej (tkz ) e, where U (x, y ) is the transverse eld distribution, k = This eld can also be expressed as
2 n,
1

(68)

and e is the polarisation vector.

z) E (x, y, z, t) = E0 U (x, y ) ej (t n e,

(69)

where n = n + jn is a complex refractive index whose real and imaginary parts are linked by Hilbert transformation (also known as Kramers-Kronig) relations, n = n+j (g int ) . 4 (70)

The changes induced in the real and imaginary parts of the complex refractive index by variations of the carrier number are approximately proportional and their ratio is dened as the linewidth enhancement factor = Using (70), n g = . (72) N 4 N Taking into account the fact that G = gvg (see Sect. 2.1) and reporting (72) and (66) into (64) leads to Rsp NL 1 dP th = + P . (73) 4 P dt P p Making use of (13) and (16), and neglecting the term Rsp /P , the instantaneous frequency shift can be expressed in term of the emitted power Pe according to (t) = where = Simulation Exercise 2 4 d [ln Pe (t)] + Pe (t) , dt 2NL d h (74) n N / n N . (71)

(75) 9

14/07/2011

34130 Introduction to Optical Communication

is often referred to as the adiabatic chirp coecient 1 . The equation relating the frequency chirp (t) to the emitted power Pe (t) is valid under large-signal modulation. It can be seen that the chirp consists of two terms. The rst term, named transient chirp only 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 dierent emission frequencies observed under steady state when a 1 or 0 bit is transmitted. Note that in (74), the frequency shift = th is referred to the emission frequency at threshold, which is dicult to evaluate in practice. However (74) can be used conveniently to evaluate the dierence in emission frequency between two operating points (corresponding to two dierent bias currents, hence two dierent emitted powers) of the semiconductor laser. This remark may be helpful for the resolution of the third assignment, in particular questions 6 and 9.

Part 0: Preparation

In this exercise, a laser whose physical parameters are provided in Table 1 in Sect. 7 is considered. The following calculations need to be performed before the simulations so that the results can be compared with those of the numerical simulations. The calculations could be performed in a Matlab programme where the required relations between the dierent parameters would be implemented.

Assignment: 1. Calculate the laser threshold Ith . 2. Calculate the slope eciency Pe /Ib assuming an internal quantum eciency int equal to 100%. The laser emission wavelength will be taken equal to 1552.5 nm. 3. The laser is now biased at Ib = 10 mA. Calculate the output power under steady state operation. Same question for Ib = 30 mA. 4. Calculate and plot the small-signal frequency response of the laser H (m ) for bias currents equal to Ib = 10 mA and 30 mA. The frequency range will be limited to 20 GHz. What are the 3 dB modulation bandwidths of the laser at those two biasing points? What are the corresponding relaxation oscillation frequencies?

Part 1: Static characterisation

The static characteristics of a single-mode semiconductor laser are studied in this rst assignment. The goal is to characterise the emitted power Pe versus bias current Ib curve of the laser and to extract the threshold value Ith and the slope eciency Pe /Ib . This information will
Note that alternative denitions of the gain suppression factor are often found in the literature, especially in connection with rate equations expressed in terms of the photon density S (instead of the photon number P used in this note), where the nonlinear gain will be expressed as (1 S ), leading to an alternative expression for the adiabatic chirp coecient 2 = . d hV
1

Simulation Exercise 2

14/07/2011

10

34130 Introduction to Optical Communication

be used in later parts of the exercise in order to set the operating point of the laser under small and large-signal modulation.

Figure 2

VPIplayer environment for the static characterisation of a single-mode semiconductor laser.

The simulation set-up is represented in Fig. 2. A dc current source is used to bias the single-mode laser diode. The value of the intensity of the current can be varied between 0 and 100 mA by using the cursor in the control area at the bottom of the set-up. The output power of the laser is monitored as a function of time using an oscilloscope and as a function of wavelength (or more properly frequency in the present case; the commonly used reference frequency of 193.1 THz corresponds to a wavelength of about 1552.5 nm) using an optical spectrum analyser. Note that such an oscilloscope is a simulation convenience that allows us to directly display the optical power as a function of time, since in real life the light would need to be detected by a photodiode rst in order to be displayed on a conventional oscilloscope. Furthermore, the total average power at the output of the laser is measured using an optical power meter. A multimeter is used to monitor the bias current and check that it has been set to the desired value. Note that in a real implementation, such a multimeter would have to be connected in series between the current source and the laser in order to measure the intensity of the bias current. The values of the bias current and the total average emitted power are collected so that they can be displayed in a table in units of mA and mW, respectively. The simulation set-up is run simply by adjusting the bias current cursor to the desired value (note that the values are displayed in A) and pressing the start simulation button. Three windows will open after completion of the calculations: a text display, an optical spectrum analyser display and an oscilloscope display, from which it is possible to gather information to plot the Pe versus Ib curve, monitor the emission wavelength, and check the output waveform, respectively.

Simulation Exercise 2

14/07/2011

11

34130 Introduction to Optical Communication

Assignment: 1. Set the bias current to values between 0 and 100 mA. 2. Monitor the output waveform and optical spectrum. Comment the observed behaviour. 3. Write down in a table the values of the optical power and lasing frequency for each value of the bias current.
Hint: the spectrum can be integrated in a given resolution bandwidth, which will give a smoother display and enable to monitor the lasing frequency more easily. A value of 1 GHz for the resolution bandwidth will be suitable for the present investigation. What is the corresponding value for this bandwidth expressed in terms of wavelength?

4. Plot the emitted power Pe versus bias current Ib curve. 5. Extract the values for the threshold Ith and the slope eciency Pe /Ib of the laser. Compare with the values calculated analytically while preparing the exercise. Compare the values of the output power for bias currents of 10 and 30 mA to the ones you had also calculated previously. 6. Plot the laser emission frequency (or wavelength) as a function of bias current. Explain the observed frequency (wavelength) shift.

Part 2: small-signal modulation

Figure 3 laser.

VPIplayer environment for the small-signal characterisation of a single-mode semiconductor

The laser is now driven at a bias Ib by a sinusoidal signal whose amplitude is small compared to Ib , in analogy with the theory detailed in Sect. 2.3. In the simulation set-up represented in Fig. 3, a sinusoidal signal source whose frequency can be adjusted by the corresponding cursor is input to a laser driver. The laser driver amplies the sinusoidal signal to the desired peakto-peak value and adds a dc oset that corresponds to the desired bias. An oscilloscope is used to monitor the waveform of the current that is used to drive the laser diode. A second Simulation Exercise 2 14/07/2011 12

34130 Introduction to Optical Communication

oscilloscope is used to monitor the output of the laser. As long as the biasing point is in the linear operation range of the laser (i.e. suciently larger than the threshold and before gain saturation is observed at high bias current), an input sinusoidal electrical signal with amplitude Im (peak-to-peak current 2Im ) and angular frequency m will result in an output optical signal with power amplitude Pem and the same angular frequency m , as illustrated in Fig. 4. The small-signal frequency response is dened as in Sect. 2.3 as H (m ) = Pe (m ) /Pe (m 0).

Figure 4 Principle of the characterisation of the laser small-signal frequency response. Note that, in this example, the peak-to-peak current and emitted optical power have been magnied for illustrative purpose.

Assignment: 1. Set the laser bias to Ib = 10 mA. 2. Set the electrical signal amplitude to the desired value. This amplitude is not critical, but you should ensure that you comply with the small-signal modulation criterion. 3. For dierent values of the modulation frequency fm , monitor the amplitude of the modulation of the optical power. Hint: use horizontal markers in the oscilloscope display. You will also notice that
the value of the amplitude of the modulation of the optical signal is somehow dierent for small time values. This is due to convergence issues in the numerical resolution of the rate equations. Consequently, ensure that the results have converged when you monitor the amplitude of the emitted power. Furthermore, it is easier to monitor the peakto-peak power using markers than just the peak (or peak-to-mean) value. The peak-to-peak power is obviously twice the value of |Pem |, as dened in Fig. 4. Since the small-signal frequency response H (m ) = Pe (m ) /Pe (m 0) is normalised, the peak-to-peak power value can be used directly.

4. Plot the corresponding small-signal frequency response of the laser. 5. Compare with the small-signal frequency response you had calculated while preparing the exercise.
Hint: Plot the calculated and simulated responses on the same graph, preferably on a logarithmic scale, i.e. 10 log10 |H | versus fm .

6. Optional: repeat steps 1 to 5 above with Ib = 30 mA. Simulation Exercise 2 14/07/2011 13

34130 Introduction to Optical Communication

Part 3: large-signal modulation

In this last part of the exercise, the same single-mode laser as the one studied in Sect. 4 and 5 is now driven by a large non return-to-zero (NRZ) data signal. Pseudo-random data is generated in the upper left block and is encoded into the NRZ format with perfectly rectangular pulses whose normalised amplitude is between 0 and 1. In order to make the data more realistic in the simulations, the rectangular NRZ data is low-pass ltered resulting in a rise and fall times equal to about one quarter of the bit rate B . The ltered data signal d (t) is then input to a laser driver, where it is amplied to the desired peak-to-peak current value and where a dc oset is added. An oscilloscope is used to monitor the electrical driving signal. Note that the laser driving signal is now dened as I (t) = Ib + Ipp d (t) (76)

where d (t) is the data signal that is normalised to the range [0,1] and Ipp is the peak-to-peak current. Consequently the laser is driven with a current equal to Ib when a 0 bit is transmitted and Ib + Ipp when a 1 bit is transmitted. The simulation set-up shown in Fig. 5 enables to adjust the biasing point Ib , the peak-to-peak current of the modulating signal, as well as the bit rate B . An oscilloscope is used to monitor the current waveform that is used to drive the single-mode laser, whose output is characterised thanks to a second oscilloscope and an optical spectrum analyser.

Figure 5 VPIplayer environment for the study of large-signal modulation of a single-mode semiconductor laser.

In this part of the exercise, optical signals will be generated by directly modulating the laser at bit rates of 2.5 and 10 Gbit/s. The inuence of the bit rate and the biasing point on the laser waveform and spectrum will be studied qualitatively and related to the small-signal frequency response determined in Sect. 5. The chirping behaviour of the laser will also be studied and its temporal and spectral features will be related. Furthermore, it will be shown how the measured waveforms and spectra can be used to retrieve essential parameters such as the relaxation oscillations frequency and the adiabatic chirp factor. Simulation Exercise 2 14/07/2011 14

34130 Introduction to Optical Communication

Assignment: 1. First, set the bit rate to 2.5 Gbit/s, bias the laser at 10 mA and drive it with a peak-topeak current value of 20 mA. 2. Observe and comment the waveform generated at the output of the laser. 3. Can you retrieve the relaxation oscillation frequencies corresponding to bias values of 10 and 30 mA?
Hint: use vertical markers in the oscilloscope display.

4. Set the bias current to 30 mA while keeping the bit rate to 2.5 Gbit/s and the peak-to-peak current to 20 mA. 5. Compare the waveform and spectrum with the ones obtained for a bias current of 10 mA in point 1. above. 6. What is the physical origin of the two spikes that appear in the laser spectrum at a bias current of 30 mA? 7. Why do those spikes are less pronounced when the laser is biased at 10 mA? 8. The oscilloscope display enables to visualise the chirp of the emitted signal. Select Power: X, then tick Show chirp. By performing this action, only the signal power aligned with the X-polarisation is displayed. The X-polarisation actually corresponds to the state of polarisation at the output of the laser (consequently no power is carried along the Y-polarisation). It then makes sense to consider the frequency chirp of the signal by calculating the derivative of its time-varying phase with respect to time. Visualise the frequency chirp for bias currents equal to 10 and 30 mA. Compare the chirping behaviour at the two biasing points and relate your observations to the corresponding laser output spectra. 9. Retrieve the gain compression factor NL from the signal emitted by the laser for a bias current of 30 mA. Compare with the corresponding value in Table 1. 10. Generate the spectra and waveforms at the output of the laser under the same driving conditions as in points 1. and 4. above, but this time at a bit rate of 10 Gbit/s. 11. Compare the results with the corresponding waveforms and spectra at 2.5 Gbit/s. 12. Relate your observations to the small-signal frequency responses obtained in Sect. 5.

Simulation Exercise 2

14/07/2011

15

34130 Introduction to Optical Communication

Laser parameters and physical constants


Parameter Active volume Cavity length Gain cross section Carrier number parency Connement factor Group eective index Linewidth enhancement Mirror reectivities Photon lifetime Carrier lifetime Gain suppression factor
Table 1

Symbol V L g at transN0 ng R1 , R2 p c NL

Value 3.6 1017 3.0 1020 72 106 0.25 4 6 0.3 2.6 1012 3.2 109 1.4 107 300 106

Unit m3 m m2

s s

Laser physical parameters.

Parameter Speed of light in vacuum Elementary charge Plancks constant

Symbol c e h
Table 2

Value 299792458 1.60217653 1019 6.6260693 1034

Unit ms1 C Js

Essential physical constants.

Recommended bibliography

One will rst and foremost consult the lecture material:

[1] J. Mrk, Sources and Transmitters, Lecture notes in 34130 Introduction to Optical Communication, Department of Photonics Engineering, Technical University of Denmark, 2011. [2] G. P. Agrawal, Fiber-Optic Communication Systems, 3rd edition, Wiley, New-York, 2002.
The notations adopted in this note, as well as the choice of the expressions of the rate equations follow this reference, which is the textbook adopted for the course. The relevant material can be found in: Section 3.3: Semiconductor lasers Section 3.5: Laser characteristics

Simulation Exercise 2

14/07/2011

16

34130 Introduction to Optical Communication

Other classic references that have been used for the preparation of this note include:

[3] A. Yariv, Optical Electronics in Modern Communications, 5th edition, Oxford University Press, New York, 1997.
Chapter 15: Semiconductor lasers - Theory and Applications, and more specically: Section 15.5: Direct-current modulation of semiconductor lasers Section 15.6: Gain suppression and frequency chirp in current-modulated semiconductor lasers.

[4] K. Petermann, Laser Diode Modulation and Noise, Kluwer Academic Publishers, Dordrecht, 1988.
In particular: Chapter 2: Basic laser characteristics Chapter 4: Intensity-modulation characteristics of laser diodes Chapter 5: Frequency modulation characteristics of laser diodes The derivation of the frequency shift (74) in Sect. 2.5 follows the approach presented in this reference.

[5] G. P. Agrawal and N. K. Dutta, Semiconductor lasers, 2nd edition, Kluwer Academic Publishers, Dordrecht, 1993.
The notations and denitions are fully consistent with reference [2] and some of the derivations are carried out further than in the textbook, in particular: Chapter 6: Rate equations and operating characteristics, and more specically: Section 6.4.3 Relaxation oscillations is the basis of the derivation of the relaxation oscillations in Sect. 2.4 Section 6.6 Modulation response.

For the specially enthusiastic student, the original references on frequency chirp (without and with the gain saturation term leading to adiabatic chirp) can be found in:

[6] T. L. Koch and J. E. Bowers, Nature of wavelength chirping in directly modulated semiconductor lasers, Electronics Letters, vol. 20, no. 25/26, pp. 10381040, 1984. [7] T. L. Koch and R. A. Linke, Eect of nonlinear gain reduction on semiconductor laser wavelength chirping, Applied Physics Letters, vol. 48, no. 10, pp. 613615, 1986.
Finally, apart from the textbook, the following references have been consulted in order to dene parameters for the directly modulated laser:

[8] S. Mohrdiek, H. Burkhard, F. Steinhagen, H. Hillmer, R. L osh, W. Schlapp, and R. G obel, 10-Gb/s standard ber transmission using directly modulated 1.55-m quantum well DFB lasers, IEEE Photonics Technology Letters, vol. 7, no. 11, pp. 13571359, 1995. [9] I. Tomkos, D. Chowdhury, J. Conradi, D. Culverhouse, K. Ennser, C. Giroux, B. Hallock, T. Kennedy, A. Kruse, S. Kumar, N. Lascar, I. Roudas, M. Sharma, R. S. Vodhanel, and C.-C. Wang, Demonstration of negative dispersion bers for DWDM metropolitan area networks, IEEE Journal of Selected Topics in Quantum Electronics, vol. 7, no. 3, pp. 439 460, 2001. [10] J. D. Downie, I. Tomkos, N. Antoniades, and A. Boskovic, Eects of lter concatenation for directly modulated transmission lasers at 2.5 and 10 Gb/s, Journal of Lightwave Technology, vol. 20, no. 2, pp. 218228, 2002.

Simulation Exercise 2

14/07/2011

17

You might also like