We investigate optical gain and penalty characteristics of a fiber-Bragg-grating based active optical-add-drop multiplexer (OADM). The active OADM has a two-stage structure and can amplify all the add/drop/thru channels simultaneously. The drop-channel penalty is a key parameter for the reliable operation of the active OADM. This penalty, caused mainly by multiple reflections within the active OADM, can be minimized adjusting pump powers. At the minimum drop-channel penalty, our active OADM shows fairly high optical gains as 15.5, 21, and 19.4 dB for the add, drop, and thru channels, respectively. Nevertheless, the penalties of these channels are comparable to the passive OADM.
©2009 Optical Society of America
Optical-add-drop multiplexers (OADMs) are key components for optical wavelength-division multiplexing (WDM) networks to add or drop WDM channels at optical nodes. OADMs can be implemented in various ways using ring resonators , arrayed waveguide gratings (AWGs) , micro-electromechanical systems , and fiber-Bragg-gratings (FBGs) [4,5]. Among them, the FBG-based OADM, that uses a pair of 3-port optical circulators with one or multiple FBGs in between, is widely used especially in local nodes owing to its simplicity, high cascadability, and polarization insensitivity.
The FBG-based OADM is passive and it should be used along with optical amplifiers. To simplify optical node structures, it is highly desirable to make the OADM active omitting external optical amplifiers. This can be realized simply by inserting erbium-doped fibers (EDFs) between the FBG (or FBGs) and 3-port optical circulators. In this way, we can amplify all the add/drop/thru channels simultaneously. Moreover, the add/drop channels are amplified in double-pass configurations instead of conventional single-pass configurations. The optical circulators suppress the lasing as optical isolators do in conventional erbium-doped fiber amplifiers (EDFAs). Although this structure is simple and cost-effective, there are no comprehensive reports on this structure to our knowledge. The main reason is the existence of multiple reflections over EDFs especially for the add/drop channels reflected by the FBG. For bi-directional applications, an active-OADM structure has been proposed adopting 6-port optical circulators .
In this paper, we investigate gain and penalty characteristics of the FBG-based active OADM. We show that the active OADM can be practical when the pump powers are adjusted to minimize drop-channel penalties.
Figure 1 shows our experimental setup to test the (FBG-based) active OADM enclosed by a dotted rectangular box. 100-GHz spaced 8 × 10 Gb/s WDM channels were applied to the active OADM. The 3-rd channel at 1553.3 nm was dropped and added by the active OADM. The active OADM was composed of a pair of 3-port optical circulators (OC1 and OC2), two erbium-doped fibers (EDF1 and EDF2), and one FBG. It had four ports, input, output, add, and drop ports. Each optical circulator had a loss of 0.5 dB and a directivity of 60 dB. The EDF1 was 13-m long and pumped by a 1480-nm laser diode, PUMP1, using a wavelength-selective coupler, WSC1. The EDF2 was 15-m long and also pumped by a 1480-nm laser diode, PUMP2, using another wavelength-selective coupler, WSC2. The FBG center wavelength was the same as the third channel wavelength, 1553.3 nm. The 3-dB bandwidth of the FBG was 0.3 nm. The FBG insertion losses for the reflected and transmitted channels were the same, 0.2 dB. Its transmission curve had a 30-dB dip at the center wavelength. The add-channel power at the add port was adjusted to equalize the channel powers at the output port.
The eight channel wavelengths ranged from 1551.7 nm to 1557.3 nm. These channels were multiplexed using an arrayed-waveguide grating (AWG), AWG1, and externally modulated according to a 231-1 pseudo-random-bit sequence. The modulated channels were amplified and de-correlated using a conventional single-mode fiber (SMF) of 5 km. The total input power to the active OADM was fixed to –8 dBm using a variable attenuator, VA. The output channels of the active OADM were demultiplexed using another AWG, AWG2. At the drop port, we used one more AWG as a band-pass filter. All the AWGs used in our experiment were 100-GHz flat-top type and had a 3-dB bandwidth of 0.3 nm. The optical receiver was composed of a pin photodiode and a limiting amplifier.
Changing the pump powers, we measured the gain and the penalty values of the 3-rd add/drop channel and the 4-th thru channel. Our reference sensitivity to evaluate the penalty was the receiver sensitivity at 10−9 bit-error rate (BER) after the AWG2 without the active OADM. After measuring each optical channel’s receiver sensitivity at BER = 10−9, the penalty was evaluated as the sensitivity difference in dB from the reference sensitivity.
3. Results and discussion
We will describe our measurement results for drop, add, and thru channels successively. Figure 2(a) and 2(b) show the 3-rd drop-channel gain and penalty, respectively, obtained by changing P1 and P2, where P1 is the output power of PUMP1 and P2 is the output power of PUMP2. If we look at the curves in Fig. 2(a) with P1 value fixed, the drop channel experiences higher gain as P2 increases. This is because a part of P2 unabsorbed by EDF2 is injected into EDF1. Owing to the injected P2, the EDF1 gain increases and the EDF1 noise figure decreases. However, the increased EDF1 gain also increases multiple reflections between the FBG and the left splicing point of EDF1. As a result, in Fig. 2(b), the drop channel has a minimum penalty, 0.38 dB, at P1 = 15 mW and P2 = 30 mW. When the add channel is absent, this drop-channel penalty is reduced to 0.34 dB caused by the wavelength-dependent phase shift of the FBG . The drop-channel has a gain of 21 dB at this point. At other pumping conditions, the drop-channel penalty increases up to a few dB. A passive OADM, obtained by removing EDF1, EDF2, WSC1, and WSC2 from the active OADM, has a drop-channel penalty of 0.41 dB as is shown in Fig. 2(b). This value is slightly higher than that of the active OADM but the difference is within our measurement error. A part of the drop-channel penalty is caused by the leakage of the add channel through the FBG.
Figure 3(a) and 3(b) show the 3-rd add-channel gain and penalty, respectively. In this case, the EDF2 operates in a large signal region owing to the injected thru channels amplified by the EDF1. The add-channel power at the add port is adjusted to equalize the channel powers at the output port. Thus, in general, the add-channel gain decreases as P1 increases with P2 fixed. As P1 increases from zero, the injected thru-channel power increases rapidly until when the EDF1 becomes saturated. This causes minimum dips for P2 = 10 mW and 15 mW curves. Overall, the add-channel penalty is smaller than the drop-channel case owing to the low EDF2 gain that suppresses multiple reflections over the EDF2. The add-channel penalty for the passive OADM is 0.22 dB and this value is reduced to 0.17 dB without the drop channel.
Figure 4(a) and 4(b) show the 4-th thru-channel gain and penalty, respectively. The thru-channel gain increases monotonically as the pump powers increase. The thru-channel penalty is much lower than the add/drop channel cases since they are not reflected by the FBG.
Let’s assume that an optical channel is added to and dropped from an optical network that uses our active OADM. Its total penalty imparted by the active OADMs is the sum of add- and drop-channel penalties plus the thru-channel penalty multiplied by the number of nodes passed through. Neglecting the small thru-channel penalty, we can see that the total penalty has a minimum when the drop-channel penalty is minimized. At this point, the optical gain is 15.5, 21, and 19.4 dB for the add, drop, and thru channels, respectively. The penalty is 0.4, 0.38, and 0.05 dB for the add, drop, and thru channels, respectively. For the corresponding passive OADM, we have the penalty as 0.22, 0.41, and 0.05 dB for the add, drop, and thru channels, respectively. Thus the active OADM has negligible additional penalties compared with the passive one while providing optical gains fairly high for add/drop/thru operations.
We show in Fig. 5 the power penalties and the total output-power variations as the total input power is changed from –14 to –2 dBm with P1 = 15 mW and P2 = 30 mW. The total output power is well saturated. When the total input power is below −11 dBm, the penalties rise owing to the increased EDF1 and EDF2 gains and multiple reflections. This low power input may happen when the input channel number is changed owing to the network failure or reconfigurations. For these transient problems, an extra control channel input can be helpful as in conventional EDFAs [8,9].
To see the effects of the control channel, we have modulated the input channel number periodically between 8 to 1 in 412 Hz using an optical switch. The surviving channel is the fourth thru channel. The gain variations of the fourth thru channel are shown in Fig. 6 . When the control channel is not used (no control), the surviving channel experiences about 7-dB gain fluctuation. When the control channel is used (control on), the control channel power is chosen such that the total input power is –8 dBm when seven channels are turned off. Then we obtain less than 0.5 dB gain changes for the surviving channel. The control channel is at 1560 nm and its power is about −8.6 dBm.
With pump powers fixed, the gains of the active OADM can be adjusted by variable attenuators. The active OADM can be used for local nodes with the node spacing less than or equal to 100 km. As Fig. 2(b) shows, the drop-channel penalty is not so sensitive to the pump powers near its minimum. This fact will be helpful for the production of many similar active OADMs. Higher gains can be obtained if EDF1 and EDF2 are spliced more deliberately with conventional single mode fibers [10,11]. For access networks, we may omit EDF2 and use only EDF1.
In case of multiple drop channels, the optimum pumping condition could be the same for different drop channels as long as they are within the same homogeneous band of an EDF. The homogeneous bandwidth is typically 4 ~10 nm. All the FBGs for drop channels should be designed to have sufficiently high reflectivity values to suppress the cross talks with add channels.
We have investigated a simple and cost-effective FBG-based active OADM. It is obtained by inserting EDFs within a conventional FBG-based passive OADM. We have shown that the drop-channel penalty is the most influencing parameter. Our structure exhibits high gains with low penalties when the optical pump powers are chosen to minimize the drop-channel penalty. At this pumping condition, the active OADM has add/drop/thru penalties close to those of the passive one. Our active OADM can be used more efficiently in local nodes of optical networks where the number of add/drop channels are small compared with the number of total channels.
This work was supported by the Kwangwoon University under a 2009 research grant.
References and links
2. Y. Nasu, K. Watanabe, M. Itoh, H. Yamazaki, S. Kamei, R. Kasahara, I. Ogawa, A. Kaneko, and Y. Inoue, “Ultrasmall 100 GHz 40-channel VMUX/DEMUX based on single-chip 2.5%-Δ PLC,” J. Lightwave Technol. 27(12), 2087–2094 ( 2009). [CrossRef]
3. D. M. Marom, D. T. Neilson, D. S. Greywall, N. R. Chien-Shing Pai, V. A. Basavanhally, D. O. Aksyuk, F. Lopez, M. E. Pardo, Y. Simon, P. Low, Kolodner, and C. A. Bolle, “Wavelength-selective 1/spl times/K switches using free-space optics and MEMS micromirrors: theory, design, and implementation,” J. Lightwave Technol. 23(4), 1620–1630 ( 2005). [CrossRef]
4. C. R. Giles, “Lightwave applications of fiber Bragg gratings,” J. Lightwave Technol. 15(8), 1391–1404 ( 1997). [CrossRef]
5. Q. Wu, P. L. Chu, and H. P. Chan, “General Design Approach to Multichannel Fiber Bragg Grating,” J. Lightwave Technol. 24(3), 1571–1580 ( 2006). [CrossRef]
6. A. V. Tran, C. J. Chae, and R. S. Tucker, “A bidirectional optical add-drop multiplexer with gain using multiport circulators, fiber Bragg gratings, and a single unidirectional optical amplifier,” IEEE Photon. Technol. Lett. 15(7), 975–977 ( 2003). [CrossRef]
7. K. H. Yla-Jarkko, M. N. Zervas, M. K. Durkin, I. Barry, and A. B. Grudinin, “Power penalties due to in-band and out-of-band dispersion in FBG cascades,” J. Lightwave Technol. 21(2), 506–510 ( 2003). [CrossRef]
8. A. K. Srivastava, Y. Sun, J. L. Zyskind, and J. W. Sulhoff, “EDFA transient response to channel loss in WDM transmission system,” IEEE Photon. Technol. Lett. 9(3), 386–388 ( 1997). [CrossRef]
9. A. K. Srivastava, J. L. Zyskind, Y. Sun, J. Ellson, G. Newsome, R. W. Tkach, A. R. Chraplyvy, J. W. Sulhoff, T. A. Strasser, C. Wolf, and J. R. Pedrazzani, “Fast-link control protection of surviving channels in multiwavelength optical networks,” IEEE Photon. Technol. Lett. 9(12), 1667–1669 ( 1997). [CrossRef]
10. H. Y. Tam, “Simple fusion splicing technique for reducing splicing loss between standard singlemode fibres and erbium-doped fibre,” IEE Electron. Lett. 27(17), 1597–1599 ( 1991). [CrossRef]
11. T. Veng and B. Palsdottir, “Investigation of new erbium doped fiber design with improved splice performance,” OFC 2005, paper OFB5.