The Gain Flatten of EDFA

The main practical limitation of an EDFA stems from the spectral non-uniformity of the amplifier gain. Even though the gain spectrum of an EDFA is relatively broad, the gain is far from uniform (of flat) over a wide wavelength. As a result, different channels of a WDM signal are amplified by different amounts. This problem becomes quite severe in long-haul systems employing a cascaded chain of EDFAs. The reason is that small variations in the amplifier gain for individual channels grow exponentially over a chain of in-line amplifiers if the gain spectrum is the same for all amplifiers. Even a 0.2dB gain difference grows to 9.6dB over a chain of 48 EDFAs.

The usable bandwidth of inline EDFA can be increased by using passive gain equalizing filters. The idea of gain equalizing filters is designed to approximate the inverse of gain spectrum. Our EDFAs are specially designed and have flat gain spectrum as shown in Fig. 3.5.

Figure 3.5 Output power versus wavelength of one of the inline EDFAs.

CHAPTER 4 Dispersion Management

4.1 Dispersion Compensation

For long-haul transmission systems, the nonlinear refractive index can couple different signal channels, and can also couple the signal with noise. It will cause the distortion, spectrum broadening and other degradations. If it is operated around the zero dispersion wavelength in fiber, the data signals and the amplifier noise with wavelengths similar to the signal travel at similar velocity. Under these conditions the signal and noise waves have long interaction lengths and can mix together. Especially the NRZ format is affected severely by nonlinearity because it has long interaction lengths. [13] Chromatic dispersion causes different wavelengths to travel at different group velocities in single mode transmission fiber.[8][14] Chromatic dispersion can reduce phase matching, or the propagation distance over which closely spaced wavelengths overlap, and can reduce the amount of nonlinear interaction in the fiber. Thus, in a long undersea system, the nonlinear behavior can be managed by tailoring the accumulated dispersion so that the phase-matching lengths are short, and the end-to-end dispersion is small. The technique has been used in both single channel systems to reduce nonlinear interaction between signal and noise as well as in WDM system.

[15]

Before we compensate the dispersion, the bit stream is Fig. 4.1, 4.2. It has peaks in the rising edge and falling edge of the bit stream. And so the eye diagram is distortion seriously. The walkoff can be minimized if the pulse wavelength and the zero dispersion wavelength of the fiber are very close. This is not often a good solution since operating near zero dispersion leads to significant impairment from the phase-matched mixing between the signal and the amplifier noise. In addition, if any optical filter clips the broadened spectrum

Figure 4.1 The bit stream after 150km.

Figure 4.2 The eye diagram after 150km.

admitted to the receiver then no pulse distortion will occur, although this may lead to impairment of the SNR due to the increased ASE noise admitted. Thus, dispersion compensation which null the overall dispersion of the chain can significantly reduce the combined effect of SPM and group velocity dispersion.

There is a brief method to measure the coarse dispersion parameter (D). By use a tunable laser as source when changing its wavelength the final signal bit stream will delay some nanosecond after one loop as show in Fig. 4.3 and 4.4. The quotient of delay time over wavelength variation over transmission length can give the coarse dispersion parameter.

We use two DCFs (D=-86.6231ps/nm/km and loss=0.37dB/km at 1553.33nm) to compensate the accumulated chromatic dispersion. One is 4.37896km and the other is 10.81287km. The dispersion map similar to that is shown in Fig. 4.5. The locally dispersion is rather large but the accumulated total dispersion in each round-trip is near zero as shown in Fig. 4.6. In Fig. 4.6 I measured two times for more accuracy. It can help to walk-off each channel and to eliminate the four-wave mixing.

Figure 4.3 The setup of measuring dispersion parameter D.

Figure 4.4 The variation of delayed bit stream.

Figure 4.5 The accumulated dispersion versus transmission distance.

Figure 4.6 The overall dispersion parameter and dispersion slope.

Since different wavelength channels will experience a different total accumulated dispersion after traversing the entire system length, at most one channel will end up with an overall accumulated dispersion of zero after traversing a dispersion managed system. This is seen in the Fig.4.5 by the diverging lines for channels 1 through 8. It results from the nonzero slope of the dispersion curve. The linear approximation for dispersion versus wavelength is

( ) ( 0)

D λ ≈SL λ λ− (4.3) where D is dispersion, S is dispersion slope, L is the transmission distance, λ₀ is the zero dispersion wavelength and λ is wavelength. [15]

Chapter 5

Circulating Loop Transmission Experiment

5.1 Line Cascade Six EDFAs

The buildup of amplifier-induced noise is the most critical factor for such systems.

There are two reasons behind it. First, in a cascaded chain of optical amplifiers, the ASE accumulates over many amplifiers and degrades the optical SNR as the number of amplifiers increases. Second, as the level of ASE grows, it begins to saturate optical amplifiers and reduce the gain of amplifiers located further down the fiber link. The net result is that the signal level drops further while the ASE level increases. Clearly, if the number of amplifiers is large, the SNR will degrade so much at the receiver that the BER will become unacceptable.

Before we set up our circulating loop experiment, we should test our EDFAs first. We cascade our six EDFAs directly without fiber span. There are attenuators before EDFAs to control the input power of each EDFA. The Fig. 5.1 shows each EDFA have similar gain except the last one. The Fig. 5.2 shows that the output power has fluctuation less than 1dB over the C band with -6 to -14dBm input power. In the operation wavelengths the power varies even less than 0.1dB. The SNR increases with the input power of each EDFA because of the increase on signal power as shown in Fig. 5.3. But the input power can not increase limitless since the nonlinear effect and the degradation of noise figure as shown in Fig. 5.4.

The noise figure can be approximated as 2n for a hypothetical two-level amplifier model _sp with constant populations . As the signal level rises, the inversion is depleted, so one expects that the inversion parameter and the noise figure will increase with the signal level. So we choose the -10dBm total input power of each EDFA as the operation

1 2 1

N +N =

Figure 5.1 Spectrum of 8 channels with fixed -10dBm total input power to each EDFA.

Figure 5.2 Cascade output power of six EDFA with fixed input power to each EDFA.

Figure 5.3 SNR of cascade six EDFA with fixed input power for each one.

Figure 5.4 Noise figure of three EDFAs.

point.

5.2 Gain Peaking

The lightwave systems based on a cascade of inline fiber amplifiers require an understanding of signal gain and noise accumulation along the link. The evolution of the spontaneous emission spectra from cascaded EDFA is shown in Fig. 5.5. The figure shows the ASE of our system has gain peak at 1562nm. Our EDFAs are designed with gain flattening until 1562nm. For a hypothetical two-level amplifier model the gain will be proportional toN₂σ λ_e( )−N₁σ_a( )λ and for a well pumped amplifier operating in small signal conditions ( and ), the gain spectrum has the same shape as that of the emission cross section

1 0

N = N₂ =1

e( )

σ λ . For an amplifier operating deep in saturation we have and , in which

1 1/ 2 N = 1/ 2

N =

(a)

(b)

(c)

Figure 5.5 ASE after (a) first loop (b) fifth loop (c) forty loop.

case the gain spectrum is proportional to the difference between the emission and absorption spectra ( )σ λ σ λ_e − _a( ). In most amplifier chains the level of saturation builds up along the chain as the ASE and signal power build up, therefore the individual gain spectra vary with position along an amplifier chain. The amplifier gain peak wavelength has been shown to be determined by the average inversion of the EDFA and the peak gain per unit length. [16] Note also that the fiber loss in the spans between the amplifiers is wavelength dependent and is higher at 1530nm than 1550nm (due to 1/λ⁴ dependence of the Rayleigh scattering component of the loss).

Thus, strictly speaking, the spectral variation of the transmitted signal and ASE is a convolution of the fiber spectral loss and amplifier spectral properties. Using phenomenological EDFA modeling parameters, the gain per unit length (G( ) /λ L) for am

amplifier with an average fractional population density in the upper state N₂/N may be completely inverted, α λ( ) is the absorption per unit length (dB/m) when the fiber erbium ions are not inverted, ( )l λ is the fiber background loss (dB/m), N is the average upper ₂ state population density (along the length of fiber) and N is the total erbium ion density. [16]

Equation implies that the gain peak wavelength is independent of pump wavelength, pump power, signal power, amount of gain compression, or amplifier configuration for a given operating gain and fiber length. The Circulating loop configurations have been used with some success to measure the gain peak wavelength. [17][18][19] The gain spectrum of an amplifier chain results from the superposition of the gain spectrum of the individual amplifiers. Even if the individual amplifier gain spectrum is broad, the region near the gain peak is amplified more than other regions, resulting in spectral narrowing after many stages.

Equivalently, the emitted spectrum of a multipass loop simulates the concatenation and spectral narrowing of such an amplifier chain. The gain peak of multiple spans of multiple passes through a loop is the same as the gain peak of the amplifier chain. [5]

5.3 Loop Time

The basic time unit for the experiment is the round-trip time of the closed loop. With reference to the timing diagram Fig. 5.6 (a), the experiment starts with the transmitter switch on and the loop switch off.

(a) (b) Figure 5.6 (a) load state (b) loop state.

The two switches are held in this load state for at least one loop time to fill the loop with the optical signals. Once the loop is loaded with data, the switches change state to the loop state as Fig. 5.6 (b), and the data is allowed to circulate around the loop for some specified number of revolutions. A portion of the data signals are coupled to the receiver for analysis. The data signals are received and re-timed by the receiver and compared to the transmitted signal in the BERTS for error detection. The error signals from the BERTS are combined with the error gate in a logic AND gate so that only the errors in the last circulation are counted. The measurement continues, switching between the load and loop states so that errors can be accumulated over long intervals of time. The BER is calculated as the number of errors detected in the error gate period divided by the total number of bits transmitted during the observation period. Since errors are counted only during the error gate period, the effective bit rate for the experiment is diminished by the duty cycle of the error gate signal, thus the real time for demonstrating particular BER might be increased by 50 or 100times over conventional measurements. With an unbroken data pattern, the BERTS still detects errors at the boundaries between each loop time, related the finite speed of the AO switches.

During the switch transition from the load state to the loop state, both switches are transmitting some amount of the optical signal. Optical pulses originating from the transmitter will interfere with those pulses returning from the loop, since two pulses have the same

wavelength, but random optical phase and polarization. This interference process corrupts the data bits and causes the BERTS to detect bit errors at the transitions. Since this same signal circulates around the loop, the error bursts are repeated at regular interval of loop time. The synchronization electronics produce a time aperture or error gate for accepting errors, which is a subset of the final loop. The width of the error gate is made smaller than one loop time to avoid the error bursts that occur on the seas of the revolutions.

The circulating loop experiment can be improved upon through the use of bit error detectors with a fast frame synchronization time, usually referred to as burst mode. Here the receiver sees broken sections of the transmitted pattern at each circulation because the data words do not fit evenly into the loop. At each border between circulations, the error detector must reacquire frame synchronization, thus creating a small error burst on the seam. As in the control scheme discussed above, this error burst is removed by gating the error detector and counting only those errors that occur in the middle of the circulation. [12] The description above is shown in Fig. 5.7.

Figure 5.7 The time diagram.

Fig. 5.8 shows the method to measure the round-trip time or called the loop time. We

can get the total power level and its variation after every loop by a low speed oscilloscope as shown in Fig. 5.9. From figure we can see the burst noise comes from the AO switch between every loop. By measuring the interval between every burst noise we can get the loop delay time is about 1.56ms.

Figure 5.8 The setup for measuring the loop time.

Figure 5.9 The variation of round-trip power.

5.4 Setup

A loop experiment attempts to simulate the transmission performance of a long system by reusing or recirculating optical data signal through a modest length amplifier chain ranging from tens to hundreds of kilometers. [12]

Fig. 5.10 shows the setup of our circulating loop experiment. The eight wavelengths of DFB lasers conform to the ITU channels from 1550.92nm to 1555.75nm with 0.8nm channel spacing.

Figure 5.10 The setup of the circulating loop experiment.

There are eight polarization controllers after the DFB lasers to control the polarization of going to the EO modulator. Then the eight channels are coupled into one path by an 8x1 coupler. After the coupler the signals go through the boost EDFA for compensate the loss of coupler. The EO modulator is used for modulating the continuous wave signals into the 10Gbit/s NRZ signals. The signals are coded the 2³¹− pseudo random binary 1 sequence data patterns. The pulse pattern generator provides the gigabit bit pattern that drives the EO modulator. The data output of the pulse pattern generator must be a high quality eye diagram, that means fast rise and fall time, low distortion, low jitter, and high Q factor. The Anritsu MP1763C has a rise and fall time less than 30ps, less than 10% distortion, less than 20ps peak to peak crossover jitter, and a Q factor larger than 40dB. After the transmitter the signals go through the variable optical attenuator (VOA) to make sure that the signal power before each AO switch and EDFA is the same. In other words, the signals come from the eighth EDFA must the same as which come from transmitter before each AO switch. By adjusting attenuators the loop gain is set to unity. This allows the data to recirculate without loss. The data generator provides synchronizing signals to the transmitter switch, loop switch, and error detector. The AO switches can control when the signals go in the loop and how many round-trips they circulate by the data generator. A 3dB coupler allows for data patterns to be loaded in and also lets them exit the loop after each round-trip. The switching in and out of the data trains needs to be synchronized both with the loop time and the bit error rate test set. The bit error rate after transmission of varying distances can be measured by using the data patterns exiting the loop after the desired number of round-trips, so that any transmission degradation with distance can be observed.

Our loop transmission part consists of six EDFAs (maximum output power=17dBm, fixed gain=22dB and noise figure=6.5dB) followed by 50km of LEAF fiber (D=4.1639ps/nm/km, and loss=0.2dB/km at 1553.33nm) and two DCFs in the appropriate position. Limits to the

optical effects (4.1) and the noise floor respectively.

The LEAF fiber has large core diameter to reduce the intensity of light and so the nonlinear effect by lifting the threshold of maximum power. Since the fiber loss in one span is less then the gain of EDFA, it needs VOA in each span to attenuate the optical power. For reducing the optical power into the fiber to minimize the nonlinearity the VOA is put just after each EDFA.

The data signals emerging from the loop on the output side of the coupler pass through the appropriate optical bandpass filter and then go into the bit error rate test set.

5.5 Experiment Result

The line width of optical pulse will be broader after modulation. This nature of adding information on carrier source is shown in Fig. 5.11. The line width of DFB laser is very narrow and suitable for WDM system. The optical spectrum can give us the information of OSNR. In the loop the optical power of each channel should be kept constant and the noise floor should be as low as possible. We can see the noise increases as signals propagating as shown in Fig. 5.12, 5.13, and thus the OSNR decreases with distance as shown in Fig. 5.14, 5.15. The channels are closed to the gain peak wavelength, thus the noise floor tilt in long wavelength. Fig. 5.14 shows the worst OSNR after 3000km is still larger than 22dB.

Figure 5.11 The spectrum of channel 1.

Figure 5.12 The optical modulated spectrum of loop 1 and 5.

Figure 5.13 The optical modulated spectrum of loop 8 and 10.

Figure 5.14 The OSNR of eight channels after propagating each loop.

We also measure the BER after one round-trip time. Fig. 5.16 shows the power penalty is about 1dB. We think the reason is about the noise figure of EDFAs. First, the original noise figure of EDFA is too large. Second, for the requirement of circulating loop experiment we close the mechanism of auto-shutdown of EDFA but it will introduces more amplified spontaneous noise because of the always on pump power. The mechanism of auto-shutdown is when there is not input power the EDFA will shutdown. The transmission distance causes the six EDFAs power up and shutdown asynchronous and thus signals can’t propagate correctly.

The best eye diagrams measured are shown in Fig. 5.17. They show the noise is too serious to degrade the quality of signals.

Figure 5.16 BER of back to back and 300km.

300km 600km

900km 1200km

1500km 1800km

2100km

Figure 5.17 The eye diagram of 300, 600, 900, 1200, 1500, 1800 and 2100km.

Chapter 6 Conclusion

We have accomplished the experimental setup of a circulating loop. By controlling the transmitter switch and loop switch we can allow the signals to circulate in the loop for designate times to simulate the long-haul transmission system. We use the WDM technology to increase the capacity in our experiment. The different wavelengths suffer from different dispersion. We use some DCFs to compensate the accumulated chromatic dispersion and the

We have accomplished the experimental setup of a circulating loop. By controlling the transmitter switch and loop switch we can allow the signals to circulate in the loop for designate times to simulate the long-haul transmission system. We use the WDM technology to increase the capacity in our experiment. The different wavelengths suffer from different dispersion. We use some DCFs to compensate the accumulated chromatic dispersion and the