7 - Self-Phase-Modulation PDF

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 20

Master in Optical Communications

and Photonic Technologies

Self Phase Modulation


(SPM)

Optical Transmission II - Winter Quarter

Outline
Description of self-phase modulation
SPM equation
Interaction between SPM and dispersion
Qualitative description
Evaluation of waveform dispersion through the
introduction of an equivalent chirp parameter
Interaction between dispersion and a chirped Gaussian
pulse
Pulse broadening
Peak distortion
SPM in a systems composed by a chain of amplifiers
SPM and dispersion compensation

OPTCOM – www.optcom.polito.it 2

1
Self phase modulation: equation

 We recall that the NLSE for the i-th channel of a WDM comb is
given by:
Ei ( z , t ) 2 2  
 j Ei   2i Ei  j 2 i 2 Ei   Ei
z 2 t 2
t 2
3 N 1
 j 
( n , m , k )Ai
Em E *n Ek

 Since we are interested in SPM only, we neglect the XPM and FWM
terms and obtain:

Ei ( z , t ) 2 2  
 j Ei   2i Ei  j 2 i 2 Ei   Ei
z 2 t 2
t 2
 j Ei Ei
2

OPTCOM – www.optcom.polito.it 3

SPM: propagation equation

 Note that this is an approximation that we make to be able to


study the effect of SPM alone. Depending on system parameters,
the other effects can be either negligible or dominant.
 Since the frequencies do not interact anymore, we can:
 concentrate on anyone of them, without loss of generality
 drop the local 0 and 1 because we don’t care about the relative
phase and group delay of the channels
 Since we can pick any channel, then we choose i=0 and drop the
index from all subsequent equations:
E ( z, t )  2
 j 2 2 E ( z, t )   E ( z , t )  j E ( z , t ) E ( z , t )
2

z 2 t
 Note that the same result is obtained if we assume that only one
channel is present and use the overall-field Schrödinger equation

OPTCOM – www.optcom.polito.it 4

2
SPM in the absence of dispersion

For now we take 2=0. Later, we will re-introduce


dispersion in different ways.
Neglecting the dispersion terms, the nonlinear
Schrödinger equation can be written as

E ( z, t )
  E ( z, t )  j E ( z, t ) E ( z, t )
2

z
This equation can be easily solved analytically in the
general case, yielding:

E z, t   E0, t e j NL ( z ,t ) ez


OPTCOM – www.optcom.polito.it 5

The solution

  NL is the nonlinear phase shift:

 NL ( z, t )   E (0, t ) zeff   P(0, t ) zeff


2

with: 1  e 2 z
zeff  the “effective length”
2
The field amplitude remains unaltered. The non-
linearity only adds a phase-shift
From the solution it is evident why this non-linear
effect is called self-phase-modulation: the channel
power profile directly modulates itw own phase

OPTCOM – www.optcom.polito.it 6

3
Non-linear phase shift due to SPM

 The non-linear phase follows exactly the shape of the pulses. If


we look for example at one gaussian pulse, we get:
 t2   T  20 ps Ppeak  10 mW
P(0, t ) P  0, t   Ppeak  exp   2   NLSPM ( z, t )
 T  FWHM  33 ps zeff  20 km

0.01 0.4
0.009
0.35
0.008
[W] 0.007
0.3

0.006 0.25

[rad]
0.005 0.2
0.004
0.15
0.003
0.002
0.1

0.001 0.05
0 0
-100 -80 -60 -40 -20 0 20 40 60 80 100 -100 -80 -60 -40 -20 0 20 40 60 80 100
time [ps] time [ps]
OPTCOM – www.optcom.polito.it 7

The effective length

It is a quantity that appears in all non-linear effects.


It describes how non-linear effects accumulate along z
We know that dispersion, for instance, accumulates
linearly over z
Non-linearity would accumulate as z as well if the fiber
was lossless.
In the presence of loss, non-linear effects take place
mostly at the beginning of a fiber span, where power is
maximum. Less at the end.
Therefore they do not accumulate as z, but follow a
more complicated law: the effective length.
OPTCOM – www.optcom.polito.it 8

4
Effective length

dB=0.2 [dB/km]
25

z zeff
20

[km]
15

10

0
0 50 100 150
z [km]
OPTCOM – www.optcom.polito.it 9

WARNING:  conversion

Typically, there are at least three alpha’s used to


indicate loss
the “field alpha”: E  z   E  0  e E z
P  z   P  0  e2 E z

the “power alpha” P  z   P  0  eP z


 P

E  z   E  0 e
z
2

 dB

P  z   P  0  10
z
the “dB alpha” 10
 dB

E  z   E  0  10
z
20

OPTCOM – www.optcom.polito.it 10

5
WARNING:  conversion

The conversion formulas are as follows:

E  P 2

log e 10 
 E   dB   dB  0.115
20
log e 10 
 P   dB   dB  0.23
10

OPTCOM – www.optcom.polito.it 11

Effective length

25
z zeff
dB=0.2 [dB/km]
20 dB=0.22 [dB/km]
dB=0.25 [dB/km]
[km] 15

10

0
0 50 100 150
z [km]
OPTCOM – www.optcom.polito.it 12

6
Instantaneous Frequency Chirp

The first derivative with respect to time of  NL yields


the instantaneous frequency chirp induced by SPM:

1 dΦNL ( z,t )  zeff d


f (t , z )   P(0, t )
2 dt 2 dt

For a gaussian pulse we have:

 zeff  t2 
 f (t , z )  P  t  exp   2 
 T2 peak  T 
OPTCOM – www.optcom.polito.it 13

Example with a Gaussian pulse

Assuming transmission at a bit rate RB=1/T, a certain


peak power corresponds to an average power of:

1
energy of PAVE  Ppeak  T2  RB  because
of modulation
single pulse 2

So the frequency chirp referred to average channel


power for a Gaussian pulse is:
2 zeff  t2 
 f (t , z )  PAVE  t  exp   2 
 T  RB  T 
3
2 2

OPTCOM – www.optcom.polito.it 14

7
Frequency chirp due to SPM

 The chirp due to SPM is shown here:


3  T  20 ps Ppeak  10 mW

f (t , z ) 2
FWHM  33 ps zeff  20 km
PAVE  1.8 mW
1
chirp [GHz]

-1

-2

-3
-100 -80 -60 -40 -20 0 20 40 60 80 100
time [ps]
 The magnitude of the SPM-induced frequency chirping
 is proportional to the pulse power
 is larger when the pulse edges are steeper

OPTCOM – www.optcom.polito.it 15

Frequency chirp due to SPM

 The chirp due to SPM is shown here:


3

f (t , z ) 2

1
chirp [GHz]
power [a.u.]

-1

-2

-3
-100 -80 -60 -40 -20 0 20 40 60 80 100
time [ps]
 The magnitude of the SPM-induced frequency chirping
 is proportional to the pulse power
 is larger when the pulse edges are steeper

OPTCOM – www.optcom.polito.it 16

8
Self phase modulation: summary

The intensity modulation of the signal changes the


refractive index
A change in the refractive index causes a change in the
signal phase  the signal modulates its own phase
Phase modulation also means frequency modulation
The frequency modulation broadens the signal spectrum
P(t)
n  n0  n2 A
2

dP dn d
f
dt dt dt
t
OPTCOM – www.optcom.polito.it 17

Self phase modulation and GVD

Interaction between nonlinearity and dispersion

dE (t , z ) 2
  E (t , z )  j 2 2 E (t , z )  j E (t , z ) E (t , z )
2

dz t

The behavior strongly depends on the dispersion regime:


Normal regime (2 > 0, D < 0)
 linear and nonlinear phase add up
Anomalous regime (2 < 0, D > 0)
 linear and nonlinear phase may compensate each other

OPTCOM – www.optcom.polito.it 18

9
SPM and GVD: description

Normal Dispersion
2 > 0 D < 0 P(t )

 f (t ) t
Red
shift t
0
Blue
shift

g  f  P(t )

t
f
f0 OPTCOM – www.optcom.polito.it 19

SPM and GVD:description

Anomalous Dispersion P(t )


2 < 0 D > 0

 f (t ) t
Red
shift t
0
Blue
shift

g  f  P(t )

t
f
f0 OPTCOM – www.optcom.polito.it 20

10
What happens bears some similarity with the case of
chirp at the TX and dispersion
SPM can be seen as a source of chirp
The polarity of this chirp is fixed (one cannot decide
the “sign” of the chirp)
Given the polarity of the chirp, the interaction with
dispersion gives pulse expansion when D<0 and pulse
compression when D>0
Note that, as in the case of chirp at the TX, pulse
compression only happens for some length: eventually
all pulses spread out (exception: solitons)

OPTCOM – www.optcom.polito.it 21

Evaluation of waveform distortion

In general, evaluating the waveform distortion induced


by the interaction between self-phase modulation and
dispersion is not a trivial task.
One simplified way of estimating it consists in assuming
that the transmitter has an extra frequency chirp cSPM,
that causes the same amount of waveform distortion
after the transmission in a linear optical fiber with the
same amount of dispersion.
In practice, we convert f(t) to a chirp factor cSPM
maintaining the pulse width of the received waveform.

OPTCOM – www.optcom.polito.it 22

11
Evaluation of waveform distortion

, 2, , z
REAL
SYSTEM
TX RX
cTX, P(t)

, 2, 0, z
EQUIVALENT
SIMPLIFIED
TX RX SYSTEM
cTX+cSPM, P(t)
OPTCOM – www.optcom.polito.it 23

Equivalent chirp parameter

It can be shown that the equivalent chirp factor cSPM of


the trasmitted pulse that causes the same amount of
increase in the RMS pulse width is given by:
kPpeak  zeff  (1)
cSPM  1  
  z 
where k is a pulse form factor given by:

Ppeak  maxP(0, t )
P
2
(0, t )dt t
k 
 k = 0.707 for a Gaussian pulse
Ppeak  P(0, t )dt
 OPTCOM – www.optcom.polito.it 24

12
Dispersion and chirped Gaussian pulse

 (1  jc )  t 2 
sT (t )  AT  exp    Ppeak  AT2 Peak power
 2 T2 
sT (t )  t2
 exp  2

 arg sT (t ) t2
  2
AT  2 T  c 2 T
 1
exp    0.606
 2

t t
T T
OPTCOM – www.optcom.polito.it 25

A simple example

Gaussian pulse
Ppeak

 1 j c  t 
2


E (0, t )  Ppeak exp    
Ppeak  2   T  
 
2
T = 18 ps 30 ps FWHM
Ppeak = 20 mW
30 ps
Full-width half-maximum of a Gaussian pulse
T
FWHM  2 ln 2 T 
0.6
OPTCOM – www.optcom.polito.it 26

13
A simple example
The pulse propagates along a non linear fiber with the
following parameters:
Fiber length: L=100 km
Fiber loss: =0.2 dB/km
Fiber dispersion: 2=+2 ps2/km
Nonlinearity coefficient: =2 W-1km-1

The procedure to evaluate the waveform distortion due


to SPM is the following:
Evaluate the equivalent chirp parameter c SPM using (1)
Assume that the chirp parameter of the transmitter is
equal to cSPM
Evaluate the ratio between T and R using (2)
OPTCOM – www.optcom.polito.it 27

A simple example

The equivalent chirp parameter is given by:


k Ppeak  zeff 
cSPM  1    0.95
  z 
Ppeak=20 mW
=2 W-1km-1
=0.2 dB/km 0.2/8.686=0.023 km-1
z=100 km
k=0.707
zeff=23 km

OPTCOM – www.optcom.polito.it 28

14
A simple example

The ratio between T and R is given by:

R
 1  cSPM Z 2  Z 2  1.7
T
T=18 ps
2=2 ps2/km  R  1.7  T  30.6 ps
cSPM=0.95
L=100 km
51 ps FWHM
 z
Z  22  0.617
T
OPTCOM – www.optcom.polito.it 29

A simple example

Without SPM:

R
 1  Z 2  1.17
T
T=18 ps
2=2 ps2/km  R  1.17  T  21.1ps
cSPM=0.95
L=100 km
35.1 ps FWHM
 z2
Z  0.617
 T2
OPTCOM – www.optcom.polito.it 30

15
A simple example

FWHM FWHM
30 ps
GVD 35.1 ps

FWHM FWHM
30 ps 51 ps
SPM+GVD

OPTCOM – www.optcom.polito.it 31

A simple example

In the standard dipersion regime (2>0), due to the


interaction between SPM and dispersion, the pulse
broadens more than with dispersion only.
What happens in the anomalous dispersion regime
(2<0)?
Using the same data of the previous example (changing
the sign of 2), we get:
τR
 1  cSPM Z 2  Z 2  0.75
τT
cSPM=0.95  R  0.75  T  13.4 ps
 z
Z  22  0.617
T 22.3 ps FWHM
OPTCOM – www.optcom.polito.it 32

16
A simple example

FWHM
FWHM
30 ps
GVD 35.1 ps

FWHM
FWHM 22.3 ps
30 ps
SPM+GVD

OPTCOM – www.optcom.polito.it 33

A simple example (conclusions)

In the normal dispersion regime (2>0), due to the


interaction between SPM and dispersion, the pulse
broadens more than with dispersion only.
In the anomalous dipersion regime (2<0), due to the
interaction between SPM and dispersion, the pulse can
broaden less than with dispersion only.

In the next slides, we will recall how to estimate the Q


penalty due to the broadening of a Gaussian pulse.

OPTCOM – www.optcom.polito.it 34

17
A simple example

Evaluate the maximum Q penalty due to SPM in the


system described in the previous example (in the
normal dispersion regime).

T T 100 ps
    1.63
2 T  1  cZ   Z 2 2 R 2  30.6 ps
2

From the plot


1 QdBMAX  0.2 dB
 0.61

OPTCOM – www.optcom.polito.it 35

Summary of SPM effects

Self-phase modulation (SPM):


Modulates the phase of the pulse, like a "chirp".
Does not change, by itself, the shape of the pulse.
Broadens the spectrum of the pulse.

SPM does not degrade system performance as long as


there is negligible dispersion.

OPTCOM – www.optcom.polito.it 36

18
Summary of SPM effects

Self-phase modulation (SPM) and dispersion


Since SPM "chirps" the pulse, it may compress or speed
up the broadening of the pulse, depending on the sign
of dispersion.

SPM effect can be studied assuming that it causes an


"equivalent chirp” cSPM on the transmitted pulse.

OPTCOM – www.optcom.polito.it 37

Summary of SPM effects

Self-phase modulation (SPM) and dispersion


Normal dispersion regime (2>0): SPM always broadens
the pulse faster than dispersion alone
Anomalous dispersion regime (2 <0): SPM first
compresses the pulse and eventually broadens it faster
than dispersion alone
In dispersion compensated systems the optimum is:
MORE than 100% compensation if 2 >0
LESS than 100% compensation if 2 <0

OPTCOM – www.optcom.polito.it 38

19
Summary of SPM effects

Note: all of the above is qualitative. Except for very


few and generally non-practical pulse shapes, there are
not exact analytical solutions for the interaction of SPM
with dispersion.

The simulation approach is recommended

OPTCOM – www.optcom.polito.it 39

Summary of SPM effects

Pre and post compensation


Experience has shown that, to minimize the impact of
SPM, dispersion should be compensated as frequently as
possible. The percentage of dispersion compensation
must be determined via simulation.

Experience has also shown that better results are


obtained putting extra compensation at the transmitter
and receiver. The extent of this extra compensation is
determined via simulation.

OPTCOM – www.optcom.polito.it 40

20

You might also like