FLEXIBLE HIGH DENSITY OPTICAL CIRCUITS

Transcription

1 FLEXIBLE HIGH DENSITY OPTICAL CIRCUITS Muhammed A. Shahid, Peng Wang and Jeffery H. Hicks OFS, 2000 Northeast Expressway, Norcross, GA Tel: INTRODUCTION: The recent surge in demand for bandwidth is creating new challenges for designers of next generation optical networks and systems. In addition to a many-fold increase in the line rate of optical channels which now requires optical transmission, switching, cross-connection, etc., new multiplexing schemes, such as dense wavelength division multiplexing (DWDM), are being implemented in order to efficiently transport, process and distribute information to end users. This creates several problems. For example, since in a DWDM scenario, each fiber could carry tens to hundreds of optical wavelength channels, when it is de-multiplexed into individual wavelength channels, the density of optical connections exponentially increases. Thus back-planes, cross-connects and distributions panels in such DWDM-based systems become highly optical-connection-density intensive. This in turn results in a many-fold increase in the number of fiber optic jumpers and fan-out cables that are needed to interconnect the required optical connections in an orderly, manageable, predictable and cost-effective manner. This becomes a serious problem both in terms of space for dressing optical cables and cost for implementing and maintaining such systems. Therefore, new technologies must be developed to address such cable congestion and channel management problems. At OFS, we have developed an effective and efficient solution to the problem of cable congestion and space management while providing most, if not all, optical connectivity requirements of an optical networking system. We have developed a fiber routing technology that uses standard off-the-shelf coated optical fibers neatly packaged in a compact planar structure which takes the form of what can be generally described as flexible optical circuits, also known as OptiPath circuits 1,2,3. The optical fiber could be any of the standard fibers (e.g., regular single-mode - SM, polarization maintaining SM, multimode - MM, etc.), or special fiber designed for specific function, (e.g., Er-doped fiber for optical amplifiers). These circuits could be simple structures, (e.g., planar spools for temperature controlled optical amplifier applications), or complex, (e.g., perfect or imperfect shuffles, cross-connects, or distribution circuits). The circuits could be small (e.g., several inches in size for planar spools) or large (e.g., several feet across, i.e., size of a shelf or optical back-plane of a frame). Each OptiPath circuit has a main body where the fibers are arranged and permanently fixed to accomplish the fiber management, shuffling, cross-connection or distribution scheme most suitable for a particular system application. In this part of a circuit of moderate complexity, the routed fibers could have several (right-handed and lefthanded) turns, bends and cross-overs. The termination regions of the circuit extend outward from the central region and are usually configured as parallel fiber arrays, like a multi-fiber ribbon with a linear density of 40 optical channels/cm, when conventional 250µm coated fiber is used. These tabs or narrow strips of ribbon fibers can be terminated using standard off-the-shelf multi-fiber connectors, such as MTP, MPO, LC-8, etc., or mass fusion spliced into other optical fibers of an optical network or system using standard fusion splicing equipment. Optical length of some or all paths of the circuit can also be equalized, or matched to a desired length, if so desired. These low profile and physically flexible OptiPath circuits nicely conform to the hardware needs and connectivity requirements of an optical

2 system. The circuits have the potential of eliminating the requirement of extra space and extensive hardware for dressing and managing the very large number of optical cables that are used in present systems. As will be shown later, this new optical circuit technology is a powerful tool for producing plugand-play equivalents of the electrical world in the optical networking systems applications. We have also subjected these OptiPath circuits to environmental tests, typical of the industry, with excellent results. In what follows, we will provide a detailed account of our optical circuit technology and test results. EXAMPLE OF A FLEXIBLE OPTICAL CIRCUIT: As stated above, generally an OptiPath circuit is designed for a particular connectivity scenario and according to the connectivity requirements of a system. In this paper, we will describe an 8x8 perfect shuffle circuit (PSC) as a generic example to describe different aspects of our technology. A perfect shuffle (PS) is an important function in the context of a communication and switching system. When PS is implemented in an optical system, it requires an optical transmission medium (e.g., an optical fiber) to transport optical signals across the PSC in a new re-arranged pattern according to a very specific arrangement of optical paths as described below and shown in the schematic of Figure 1. In the 8x8 PSC example, each of the 8 input tabs is configured as a ribbon of 8 fibers with a fiber-to-fiber spacing of 250µm. Each 8-fiber wide tab is 15cm long, in this example circuit, and can be made in other lengths rather easily. Thus, the circuit has 64 input fibers grouped as 8 tabs of 8 fibers each. Similarly, the same 64 input fibers appear re-grouped in the form of 8 output tabs each containing 8 fibers. In this example, the input tabs (A's in Figure 1) and output tabs (B's in Figure 1) are spaced at a pitch of 2cm. In the shuffle area of the circuit, which is 15cm wide by 18cm long, the fibers from A-tabs are arranged and re-configured in such a manner that each fiber from a given A-tab appears at identical fiber positions in all of the output B-tabs. Thus, the 8 fibers of A1 tab, are fanned-out, distributed and placed in their original sequence at the first fiber position in each of the 8 B-tabs (B1 to B8); the 8 fibers of the A2 tab, are fanned-out, distributed and placed in their original sequence at the second fiber position in each of the B- tabs (B1 to B8); and so on. As stated above, and shown in Figure 1, each optical path of the PSC has turns, bends and cross-overs, i.e., areas where one fiber rises above another fiber during crossing it. These bends and cross-overs can be potential sources of macro- and micro-bending losses. Start End Figure 1. A line drawing showing the design of an 8x8 perfect

3 THE CIRCUIT FABRICATION PROCESS: The process of building a circuit starts with its design that includes general shape, size, number and position of each tab, number and position of fibers in each tab, etc. The interconnect circuit itself is designed in the form of an exacting continuous line drawing using a commercial CAD tool such as AutoCad. This single line zig-zags between various intended connection points in each tab and loops around as it continues to complete a single extremely complex open-ended loop. The actual line drawing with its start and end in our exemplary PSC is shown in Figure 1. Here lines of different colors in each tab are shown in order to clearly demonstrate the way the perfect shuffle scheme is constructed in the central shuffle region. Within the main body of the circuit, a minimum 25-mm radius is maintained for all turns. Also designed is the circuit boundary, shown generally as a rectangular region with narrow fingers extending on opposite sides and centered on the arrays of parallel lines. The machine/data files that will later control the fiber routing and cutting routines are extracted from these original CAD drawings. It should be obvious that several circuits of the same or different designs can be cascaded within the allowed workspace of the fiber routing machine. This means that this technology is well suited for reproducible mass production of identical circuits. Next, a carrier board is prepared on which a sheet of Kapton with a backing material is temporarily attached using a light tacky adhesive. On the upper surface of the Kapton sheet is another prelaminated, pressure-sensitive tacky adhesive. During the fiber-routing operation, the fiber head of the fiber-routing machine presses and fixes the optical fiber into the pressure-sensitive tacky adhesive as the fiber is dispensed and routed according to the design of the circuit pattern. The material of the pressure sensitive tacky adhesive has flame-resistant properties. The carrier board with all its components, e.g., adhesive coated Kapton assembly, is fixed on the fiber routing machine. The routing machine is commanded by the data file previously extracted from the CAD file for the fiber routing routine. The routing machine itself is a 4-axis machine. It consists of a highprecision, xy-translation stage configured as a gantry system. The routing head rides on a z-stage which has θ-axis (i.e., rotation) incorporated on it. The fiber supply spool also rides on the z-stage, and together they ride on the xy-stage. The fiber-routing head itself has a precision V-groove wheel which presses and fixes the optical fiber into the tacky adhesive as the fiber is dispensed and routed under the rolling wheel according to the pattern of the circuit. At the end of the fiber routing operation, the fiber is severed from its supply spool. At this stage, the circuit is an exact reproduction of the original line drawing. The circuit consists of a single length of fiber in the form of an open-ended continuous loop with the fiber pattern in the central shuffle region configured to meet the previously defined interconnection scheme. The optical continuity of the fiber circuit can be tested at this stage for any breaks. All aspects of the fiber routing process have been optimized to avoid fiber breaks. Next the carrier board, with the fiber circuit still in place, is loaded on to another machine which applies a protective conformal coating layer of a plastic material. The conformal-coating layer is allowed to cure overnight. This layer precisely conforms to the profile of the routed circuit and provides a uniform coverage of coating to protect the fiber circuit. The silicone-based material of this protective layer is also flame resistant. Next the carrier board, with the optical circuit still in place, is once again re-positioned, in its original orientation, on the fiber-routing machine. This time, however, the 4-axis machine is equipped with a cutting head. Now, the machine uses the circuit boundary cut file which was previously extracted from the original AutoCad file for the cutting routine. At this stage, a final optical continuity test can be made to ensure that the fiber in the circuit has developed no breaks. Next, redundant looping sections, instituted to facilitate a single continuous line for the whole circuit, are cut off (e.g., along the dotted vertical lines

4 in Figure 1) and excess backing material removed. Next, the completed PSC can be lifted off the carrier board. If needed, holes for mounting the circuit in trays or mounting panels are cut at the desired positions in the circuit, away from the fiber pattern. Finally, desirable optical connectors (e.g., MTP, MPO, LC-8, etc.) are assembled on to the circuit tabs making it ready for use. Alternatively, after fixing the circuit in its tray or other suitable element, the fibers of the circuit can be fusion spliced to the other fibers of the system. An example of an 8x8 PSC is shown in Figure 2. Figure 2. An example of an 8x8 perfect shuffle OptiPath flexible optical circuit. ENVIRONMENTAL TESTS: Using OFS depressed-clad single-mode optical fiber, twelve identical PSCs were produced. In order to keep the test circuits short, the length of tabs were reduced (~ 4cm). The testing was done on circuits that were still in the form of open-ended loops with all elements, namely, PS regions and loop-back sections together. The actual length of fiber in each PS test circuit was about 32m. All 12 "start" pig-tails, one from each PSC, were ribbonized and terminated with an MPO-type connector. This connector end was used as the optical-launch end during tests. Similarly, all 12 "end" pig-tails, one from each PSC, were also terminated with an MPO-type connector. A special frame was constructed to accommodate all 12 circuits in a compact, 6-shelf configuration as shown in Figure 3. The frame with all 12 PSCs was placed inside the environmental chamber from the side port of which the MPO connectors were accessed for connecting to a multi-channel optical loss test set for in-situ monitoring of changes in optical-loss during the environmental tests. A schematic of the test configuration is shown in Figure 4. Prior to the start of the experiment and during the initial set up, it was noticed that out of 12 circuits, one circuit (#3) exhibited relatively high loss. Later, it was discovered that the problem related to the relevant position of the launch lead. Consequently, although the circuit in question was also monitored during our experiment, test data from this circuit was not included in this paper. Since processes and materials employed in the fabrication of OptiPath circuits are new, one of our goals was to evaluate and establish base-line performance of our circuits. Toward this end, we chose the test

5 Figure 3. A photo of the test circuits arranged in a rack constructed for this experiment. Figure 4. The experimental set up for the environmental tests. criteria and test conditions in accordance with Telcordia GR-326-CORE, issue 3, recommendations. Our focus was to study structural stability, mechanical integrity and impact of environmental conditions on the

6 optical performance of OptiPath circuits. Thus we monitored in-situ changes in optical loss. The test schedule that was followed and the sequence of tests conducted are listed below. 1. Thermal Aging: 85 C for 7 days (with uncontrolled humidity) 2. Temperature Cycling: -40 C to 75 C for 7 days (or 21 cycles) (with uncontrolled humidity) 3. Humidity Aging: 75 95% RH for 7 days 4. Humidity Condensation Cycling: -10 C to % RH 7 days (14 cycles) 5. Dry-out Step: 75 C for 24 hrs 6. Thermal Cycling: -40 C to 75 C for 7 days (or 21 cycles) (with uncontrolled humidity) The graph of Figure 5 shows in-situ changes in optical loss in PSCs during the Thermal Aging Test (Test 1 above). It is clear from this graph that the changes in all circuits under test are well below ±0.08 db. It should be remembered that these changes in optical loss pertain to PSCs with their turn-around sections still not cut off. When these regions are trimmed off to separate the functional PSCs, the circuit turns into 64 short lengths of organized strands of optical fibers. Therefore, it can be inferred from the results of the graph of Figure 5 that the changes in loss in each path of the circuit are actually insignificantly smaller. Another graph showing in-situ changes in optical loss of all circuits during the Temperature Cycling Test (Test 2 above) is also shown in Figure 6. Once again the observed in-situ changes in optical loss in the un-trimmed circuits during this test also stay extremely low (<0.1 db). These results also represent insignificantly small changes of optical loss in each path of the 64 optical paths of the trimmed PSC. Heat Aging Uncontrolled Humidity Path Loss Change (db) :47:34 PM 11:14:39 PM 7:41:44 AM 4:08:48 PM 12:35:53 AM 9:02:58 AM 5:30:03 PM 1:57:08 AM 10:24:12 AM 6:51:17 PM 3:18:21 AM 11:45:26 AM 8:12:30 PM 4:39:35 AM 1:06:39 PM 9:33:44 PM 6:00:48 AM 2:27:53 PM 10:54:58 PM 7:22:03 AM Readings Taken Every 15 Minutes 3:49:08 PM 12:16:12 AM 8:43:17 AM 5:10:21 PM Circuit 1 Circuit 2 Circuit 4 Circuit 5 Circuit 6 Circuit 7 Circuit 8 Circuit 9 Circuit 10 Circuit 11 Circuit 12 Temp. (C) Figure 5. A graph showing changes in optical loss in PSCs during thermal aging test. Table 1 shows a list of average values of changes in optical loss and standard deviation during all tests cited above for all samples. It is clear from data in Table 1 that all samples show excellent optical performance. While each circuit exhibited slightly different amounts of change in optical loss, the worst case increase in optical loss on a given circuit stayed well below 0.18 db in an untrimmed circuit. As stated above, this translates to insignificantly small change in the optical loss of an individual optical path of an Optipath PSC.

7 Table 1 Changes in optical loss and standard deviation in PSCs during environmental tests Average Standard Test Condition Change (db) Deviation (db) 1. Thermal Aging High Temperature & Humidity Temperature Cycling Temperature Cycling & Humidity Temperature Cycling Temperature Cycling Path Loss Change (db) Readings Taken Every 15 Minutes Circuit 1 Circuit 2 Circuit 4 Circuit 5 Circuit 6 Circuit 7 Circuit 8 Circuit 9 Circuit 10 Circuit 11 Circuit 12 Temp. (C) Figure 6. A graph showing changes in optical loss in PSCs during temperature cycling test. GENERAL APPEARANCE: After completing all environmental testing, the PSCs were mechanically robust in all respects. There was no de-lamination of the substrate or coating materials and no lifting of the routed fibers was observed. A slight upward curl near the edges of the Kapton substrate did develop during the tests. The conformal coating material of the circuits appeared slightly discolored, as if the pressure sensitive adhesive laminated on the Kapton substrate had become more transparent to let the amber color of the Kapton substrate bleed through. Overall, the circuits were structurally intact and as flexible as they were at the start of the tests.

8 SUMMARY: We have described a new manufacturing process for producing complex and large flexible OptiPath optical circuits for fulfilling the high optical connection density needs of next generation optical networks and systems. These circuits are well suited for large, complex and dense optical connection schemes and are compatible with the termination and handling procedures the physical designers are so used to dealing with today. Using an example of an 8x8 perfect shuffle circuit (PSC), details of the circuit fabrication process were described. It was shown that in our process, from start to end, an optical circuit is configured as a single open-ended complex loop and yet it includes all desired elements of size, connectivity and connection scheme. It was also shown that our unique optical circuit process allows monitoring of the circuit integrity and continuity during all stages of its fabrication. Several PSCs were subjected to the environmental tests commonly used in industry. Each circuit, as tested, consisted of a single complex loop containing bends, turn-arounds and cross-overs. Despite their complexity, measured optical performance in these circuits was outstanding. After the turnaround portion of the circuits are cut-off, each functional perfect shuffle circuit has 64 individual lengths of short strands (~ m long) of well-organized fibers. Thus, the worst case changes in optical loss observed during our experiment in individual fiber paths are extremely small - practically insignificant. It was shown that flexibility and structural integrity of optical circuits is maintained during environmental tests per Telecordia GR-326- CORE recommendations. ACKNOWLEDGEMENTS: Thanks are due to several colleagues who have helped in different aspect of this project. In particular, I would like to thank Jeff Hicks, Brandon West and Peng Wang. REFERENCES: 1. R.A. Nordin, W.R. Holland and M.A. Shahid, Advanced Interconnection Technology in Switching Equipment, IEEE J. Lightwave Tech., Vol. 13, No. 6 (1995), pp M.A. Shahid and W.R. Holland, "Flexible Optical Backplane Interconnections", Proc. 3 rd International Conf. MPPOI 96, Oct. 1996, Hawaii, pp M.A. Shahid, N.R Lampert, A. W. Carlisle, D.A. Hendrickson,, D.M. Emmerich, T.E. McNeil and J.E George, "Small and Efficient Connector System", Proc. 49 th ECTC conf. (Special Symposium on Small Form Factor Connectors and Opto-Electronic Modules), June 1999, San Diego, CA, pp Kapton is registered trademark of E.I. DuPont Co. OptiPath is a trademark of OFS, Inc. MTP is a registered trademark of US Conec Ltd.