Semiconductor Optical Amplifier

1
Semiconductor Optical Amplifier
2
References

• B. H. Verbeek. SOA ISLC00
• L. Chrostowski, C.-H. Chang, C.J. Chang-Hasnain.
  "Demonstration of Long-Wavelength Directly
  Modulated VCSEL Transmission Using SOAs", IEEE
  Photonics Technology Letters, vol. 14, No. 9, pp.
  1369-1371, September 2002.

3
Outline

• SOA Introduction
• SOA Parameters
• Fancy SOAs
• Applications
• Traditional Amplifier for communications
• New ideas Wavelength converters, Demux, Clock
  recovery
• Our Simulations

4
Semiconductor Optical Amplifier
5
Operating Principle

• device physics same as EEL.
• difference is that Rlt 10-5 AR, angled stripe,
  window region
• SOA can be operated in saturation, or
  unsaturated. gain clamping
• single pass chip gain Gexp (g_modal L)
• packaging TEC, high coupling efficiency,
  isolators

6
Gain vs. Wavelength
Single SOA

• 40-80 nm, InGaAs/InGaAsP. Spanning from
  1250-1650 nm

7
Chirp Parameter

• Allows SOA to be used as a Non-linear element,
  for phase delay applications.

8
Gain vs. Output Power

• An SOA has a Saturation Output Power

9
Output Power

• SOAs are linear for small input powers.

10
Gain Dynamics
11
Saturation Output Power

• Saturation Output Power decreases for higher
  energy photons.

12
Wavelength Dependence on Psat

• In band diagram, higher energy carriers are
  depleted faster than lower energy ones.

13
Polarization Independent Gain

• for gt 1400nm, use tensile strained bulk InGaAsP
  active layer (0.29 strain).
• or, use a symmetric active layer, i.e. a square,
  0.4 x 0.4 um.

14
Noise Figure
15
Noise Figure

• SOAs are noisier than EDFAs because the coupling
  efficiency is lower. Otherwise, they have the
  same theoretical limitations.
• Thus, integrated SOAs should be less noisy.

16
Cross Gain Modulation

• Saturating the SOA with a signal affects the
  overall gain spectrum. Thus, all wavelengths
  will be slightly modulated.

17
Cross Gain Modulation solutions

• low input power (linear regime). SOA not in
  saturation. 8x20 Gbs 160 km. Spiekman et al,
  1999
• reservoir channel, SOAs in saturation. 32x2.5
  gbs. 125 km. Sun et al 1999
• Gain-Clamped SOA
• Solution Fixed Gain SOA
• want fixed gain to eliminate XGM
• Note a laser has fixed gain above threshold
  (gain clamping)

18
Gain Clamped SOA

• gain medium is shared between SOA and a laser.
  lasing at a different wavelength.

19
Conventional vs. Gain Clamped SOA
Tiemeijer, L.F. van den Hoven, G.N, PTL, vol.8,
(no.11), IEEE, Nov. 1996. p.1453-5
20
Gain Clamped SOA

• Pros
• Linear gain (important for analog
  communications), no XGM (demonstrated) -gt WDM
  applications.

• Cons
• Fixed gain makes it difficult to match to
  required gain.
• Limited by laser dynamics, i.e. relaxation
  oscillation of DBR laser, 10 Gbs.

21
SOAs as Amplifiers
22
Long-Distance SOAs

• Didnt work very well. SOA not yet suitable for
  long-haul. Good for short distance WDM.

23
In-Line Amplifier SOA

• Noise limits performance of links

24
Optical Demultiplexing - TOAD
25
TOAD

• Limitations control pulse travelling in the SOA,
  500 um -gt 5ps. (experimentally 15 ps)
• In theory, the smallest pulse that can be
  demultiplexed is twice the propagation time of
  the SOA.
• SOA not in center to ensure that the pulses
  properly shaped.
• Idea by Sokoloff, Prucnal, Glesk, et. al. "TOAD"
  1993 done with Non-linear element.

26
Optical Clock Recovery

• SOA /w Four-Wave Mixing with an Optical Phase
  lock loop. 40 Gbs. Kim et al., Optics
  Communications 15 Aug 2000.
• Using injection of a mode-locked laser.
• Self-pulsating DFB lasers (Opto)

27
Demultiplexing Experiments
Smets, de Waardt 1999 Eindhoven Univ.
28
Demultiplexing Experiments

• 40 and 160 Gbs!

29
? Conversion X Gain Compression
30
? Conversion X Phase Modulation
31
? Conversion Interferometric
32
Amplifier Comparison
33
Conclusion

• Physics of SOA well understood
• Many applications are emerging. Communication
  subsystems.
• SOA use as an amplifier is possible for
  short-haul communications.

34
Optical Amplifiers
Semiconductor Optical Amplifier
Erbium Doped Fiber Amplifier

• Important Parameters
• Gain
• Saturation Output Power
• Noise Figure

B. Verbeek, JDSU
35
Gain, Saturation Output Power
JDSU CQF872 SOA T25o C. I 400mA G 21.5dB
Psat 5.8dBm N.F. 8.2dB
Linear Regime
Psat
36
Optical Amplifier Applications
B. Verbeek, JDSU
37
In-Line Optical Amplifier
Distance

• Noise Figure can limit performance of links

B. Verbeek, JDSU
38
Optical Amplifiers SOA vs. EDFA

• Semiconductor Optical Amplifier
• Noise Figure 8-9 dB
• Fast gain dynamics ( ns)
• Conventional SOAs suffer from pattern dependant
  gain, pulse distortion, inter-channel cross-talk.
• Gain-clamped SOAs reduce these effects because
  the gain does not fluctuate.
• Experiments use an early JDSU gain-clamped SOA

• Erbium Doped Fiber Amplifier
• Noise Figure 4-5 dB
• Slow gain dynamics ( ms)
• Problem of gain fluctuations when adding/dropping
  channels.
• Using a commercial INO EDFA

39
Experimental Setup

• WDM Experiments, at 2.5 Gb/s
• Test the performance of VCSELs
• Goals
• Compare MetroCor vs. SMF28
• Compare SOA vs. EDFA

40
SOA vs. EDFA MetroCor

• Transmission experiment using 50 km MetroCor
  fiber
• SOA performance shows a 0.7 dB power penalty
  compared to using an EDFA
• No channel cross-talk observed using SOA, with 2
  channels

41
SOA vs. EDFA SMF28

• Transmission experiment using 75 km SMF-28 fiber
• SOA performance shows a 0.55 dB power penalty
  compared to using an EDFA
• 2.5 dB Power penalty for SMF28

Measured at BW9
42
Fiber Propagation Model
Additive Receiver Noise
Random Bit pattern
Create eye- diagram
Low-pass filter
Rate Equations
Gaussian Curve-fit to find BER
Fiber Dispersion
Detector
43
Simulation Results, 2.5 Gbs

• Simulation at 2.5 Gbs qualitatively agrees with
  experiments for ?H5
• Experimentally ?H was measured to be between 4-7.
• Measured by Steve Yang and Rob Stone

44
Noise Figure

• Noise Figure definition is similar as for
  electrical amplifiers. Essentially a degradation
  of signal.
• However, we do not use the optical SNR, but
  rather the SNR that would be measured with an
  ideal square-law detector at the input and
  output of the amplifier.

Where EElectric Field, IDetector
Current, ASignal amplitude, x,yAmplifier
Spontaneous Emission
45
Noise Figure

• NF definition assumes shot-noise limited source.
  Laser noise is ignored.
• Detector thermal noise is ignored/negligible.

Noise Figure
3
2.5
2
1.5
1
0.5
0
5
10
15
20
25
30
Gain (dB)
3 dB NF limit, for complete inversion, high gain
46
Optical Amplifier Simulations
Noise components (dB) vs. Optical Amplifer Gain
(dB)

• -30 dBm input power
• Optical Amplifier
• nsp1.4 (NF 4.5 dB)
• Optical Filter BW10 nm
• ER inf
• PIN Detector

Optical Gain (dB)
47
Optical Amplifier Simulations
SOA NF 9 dB
EDFA NF 4 dB
Total Noise
Sig-Sp Noise
Thermal Noise
Shot Noise
Sp-Sp Noise
Power penalty at 10e-9 0.65 dB
48
Optical Amplifier Simulations
Bit Error Rate Plot
NF10 dB, APD, Gain20
Noise Contributions to BER Plot
1e-005
1e-006
1e-007
Bit Error Rate
Noise (dB Amps)
1e-008
1e-009
1e-010
1e-011
1e-012
-36
-35
-34
-33
-32
-31
-30
-29
-28
-27
-26
-25
Received Optical Power (dBm)
Received Power (dBm)

• First slope due to thermalshot noise,
• 2nd one due to signal-sp shot

49
Optical Amplifier Simulations
Comparison of different amplifier Noise Figure.
Impact on BER Plots
NF 3, 4, 6, 8, 10, 12 dB, Gain20 dB, APD, 50
GHz filter
1e-005
1e-006
Bit Error Rate
1e-007
1e-008
1e-009
1e-010
1e-011
1e-012
50
PIN Detector
NF 4
NF 12
NF4 dB, 6 dB, 8 dB, 10 dB
51
NF Impact on Transmission
Power Penalty vs. Amplifier Noise Figure
3
2.5
Single Amplifier
2
Power Penalty vs. Amplifier Noise Figure
Power Penalty (dB)
1.5
2.5
1
Multiple Amplifiers (1,2,3)
0.5
2
0
2
4
6
8
10
12
14
16
1.5
Power Penalty (dB)
Noise Figure (dB)
1
0.5
0
3
4
5
6
7
8
9
10
Noise Figure (dB)
52
NF Impact on Transmission

• Using SOA as Booster, EDFAs as in-line
  amplifiers.
• Input signal 30 dBm (rather low).
• Still yields satisfactory results, little power
  penalty for using SOA.

53
Extinction Ratio, NF Impact
EDFA Pin -13.9dBm Pout 6.48dBm E.R. 8.1dB
SOA Pin -24dBm Pout -2.4 dBm E.R. 7.31dB
Power Penalty 0.25 dB

• Where does the SOA vs. EDFA Power Penalty come
  from? Answer is both
• Reduction in Extinction Ratio, and
• Poorer Noise Figure

Measured at BW9