Multi-fiber connectivity is one of the hottest talking points in fiber optic connectivity. Many white papers, webinars and blogs lean heavily toward the 12-fiber approach.
It can easily be overlooked that there are other multi-fiber solutions available as well. We can cite Cisco Systems as an example of an industry leader that decided to set up some new products based on 24-fiber MPO connectors. In this article we’ll take a look at the interconnection which forms the backbone of such a system.
Multi-fiber approach development
In technology fields, and especially in fiber optics, we see a significant evolution which is raising newer and newer questions. One of the current ones is: “How many fibers do we really need for parallel connectivity?” Looking back, we recall that at our very start down the road of miniaturization and high-fiber count applications, a solution with only two fibers in the same connector housing was available – the MTRJ connector. This was the first approach to combining multiple fibers in one ferrule, which replaced the conventional ceramic ferrule solutions based on the SC and LC connectors.
With higher bandwidth and speed demands, a more than 1G application appeared on the market and the interconnection solutions required more flexibility on one side, and space efficiency on the other. This was the cue for the MPO connector to enter the scene. Its first applications were based on the 8-fiber type, which was very quickly followed by the 12-fiber type – becoming a standard for many years. At the beginning, transmitters and receivers of active equipment were mainly based on the SFP module. MPO connectivity was used as a backbone solution and for Plug-and-Play interconnection. With new speed increases from 10G to 40G and even to 100G, real parallel connectivity was requested. A lot of big players started to create MSAs (Multi-Source Agreements), where they tried to define “the best approach” for how to coverthis new market demand. In the case of the 40G application, it was quite quickly defined as an MPO interface with 12 fibers: 4 transmitting (4x10G), 4 receiving (4x10G) and 4 blind fibers in the middle. QSFP modules found their way into 40G active equipment. The next logical step was 100G. The first approach was to use two 12-fiber MPO connectors – one connector purely for transmitting signal and the second dedicated for receiving. Each channel operates at 10G, which means 10 x 10 = 100G transmission. Connector manufacturers took this as a new challenge and introduced an MPO connector based on a 24-fiber ferrule which accommodated both the Tx and Rx lanes within the same connector. This was a great leap forward in terms of the challenge of space efficiency. The introduction of the new 24-fiber ferrule/connector would bring about a rift in the fiber optic world, splitting it into two groups. One group was mainly oriented toward the 12-fiber solutions and applications, with the other decisively turning toward the 24-fiber approach. Members of both camps have their own reasons and arguments for adhering to their specific system.
One can easily become quite confused by this “fiber battle”. But to put it plainly, the main driver is again the active equipment used. For the 100G solutions, the main current modules are CXP and CFP, where in the case of a multi-fiber interface, the fiber count is set to 24. For the direct connection of two such active parts, it is necessary to use a trunk or patch cable based on the 24-fiber MPO approach. A majority of applications in datacenters have been multi-mode oriented. It is worth mentioning, however, that some high-added-value or special dedicated systems are based on the single-mode version. As an example, we can consider the evolution of the Cisco-developed ASR 9000 router series with modular Ethernet line cards equipped with the CPAK interface. The first generation of NPU (Network Processor Unit), called the Trident Class 120G, was followed by the Typhoon Class 360G and is currently being replaced by the Tomahawk Class 800G NPU, which is flexible in a variety of line cards supporting 10G, 40G and 100G. (see Fig. 1).
When dealing with the long reach versions (such as CPAK-10x10G-LR), the single-mode alternative is used, meaning that the interface also has to be a 24-fiber single-mode type.
Challenges, expectations and limits
Of course, this is a new challenge for cable harness makers – with much more complicated handling, termination and measurements involved for the 24-fiber ferrule (particularly the single-mode version) than its 12-fiber counterpart.
Sylex took on this challenge as a key priority right from the beginning of the 24-fiber era. Due to our strong focus on MT technology, we can currently offer the highest quality 24-fiber patch cords, trunk cables and fanout assemblies currently manufactured – in both single-mode and multi-mode versions. These can be used in the implementation of breakout cable or breakout patch panel solutions (see Fig. 2 and Fig. 3).
The materials used, based on Corning fibers and USConec MTP ferrules (which benefit from tailored and fine-tuned proprietary polishing processes), give outstanding outcomes.
Different end users, installers, datacenter managers and system integrators are all looking for different added values of purchased products, but at the end of the day, one of the main drivers is the loss budget which limits the operation of the whole system. Officially declared limits for the 24F multi-fiber ferrules are at a level of 0.6dB per mated connector pair for multi-mode and 0.75dB for single-mode. More interesting, of course, is the so-called low-loss version, where for both fiber types the limit value is set to 0.35dB in an ideal state and to 0.75dB in the worst case. Here it has to be noted that in the case of the 24-fiber MT ferrules, the insertion loss is very sensitive to the achieved physical contacts between the two ferrules. The pressing force used between the mated pair of single-mode connectors is limited due to the angled polishing found on the ferrule. Because of that, the stated insertion loss level for 24F single-mode ferrules can only be reached in ideal situations, which is not always the case in real life.
Sylex comprehensively evaluated this point and fine-tuned the manufacturing process to such a level that in any case, the maximum level of insertion loss for single-mode low-loss ferrules will not exceed 0.5dB. There are only 6 percentages from all the values which are higher than 0.35dB but still less than 0.5dB. Here we have to highlight that those values are not measured against the perfect launch connector as this is usually done by harness makers. The measurements were done under a real-life conditions where both mated connectors are from the same manufacturing line and have passed the same quality acceptance criteria.
In the chart (Fig. 4), the real distribution of the measured values is visible, which proves that the power budget for SM-based applications can definitely be met. We believe that with such superior parameters of the most complicated multi-fiber connectors, Sylex can offer to all of our partners the highest performance harnesses which can connect the latest state-of-the-art switches and routers installed indata centers.
Eduard Koza – R&D Manager – Contact
Eduard Koza is R&D manager at Sylex. He started as a project engineer for high-performance copper harnesses. In 2000, he widened his activities to passive fiber optic interconnection products and fiber optic sensing systems. At the beginning of this period, he mainly led multiple technology transfers and covered the implementation of new products into the company portfolio.
Later, he held the role of Engineering manager in the company and headed the technology and engineering activities responsible for transferring customer ideas to the final products. His current position covers the management of newly introduced technical solutions, utilizing his consultancy skills related to customized fiber optic interconnection products.
