Global Sources
EE Times-Asia
Stay in touch with EE Times Asia
EE Times-Asia > EDA/IP

Component vendors address module challenges

Posted: 17 Oct 2005 ?? ?Print Version ?Bookmark and Share

Keywords:phyworks? vendors? manufacturers? ic? challenges?

The diametrically opposing market requirements of higher performance, lower power, smaller size and lower cost form a conundrum which perpetually vexes manufacturers of optical components and systems. Throw into this equation a period of post-bubble cautious capital expenditure, ongoing merger activity and a general lack of profitability and, at first glance, the outlook for players in the fiber-optic communication industry appears dire. While the early 1990s' industry activity was fueled by growth in the telecom arena, the industry has recently seen remarkable growth and innovation in the datacom, storage and passive optical networking (PON) segments.

Changing industry

The last decade has seen a rapid shift in the structure of the optical industry. Optical system and subsystem vendors have moved away from the vertically integrated structure of having in-house "light-to-box" capability and are instead procuring optical subsystems and modules from focused external suppliers. Parallel to this change, there has been considerable consolidation within the optical module vendor space itself. Along the way, numerous startups have come and gone, victims of over-ambitious business plans and a faltering market.

The industry consolidation is driven wholly by the need for businesses to make profit and return shareholder or investor value. Thus, the market has seen a wholesale export of module manufacturing to Asian contract manufacturers. Indeed, the optical module space is now beginning to witness a situation similar to that in the PC laptop market, with a handful of Asian contract manufacturers assembling and building almost the entire output of optical modules.

In-house or outsource?

Just as system vendors realized a few years ago that vertical integration was not the best strategy in their quest for profits, optical module vendors are being faced with a similar challenge today regarding their optical subassemblies (OSAs) and ICs. Although in-house optics capability has traditionally been considered crucial to success, new suppliers are offering module vendors a viable alternative.

Since differentiation in OSAs is difficult due to their relatively simple functionality, for data rates below 10Gbps, external OSA suppliers compete primarily on cost. At higher data rates, some differentiation is still possible by offering better performance through novel package design or the specific choice of transmitter or photodetector used. However, as these dedicated OSA vendors continue to build on their 10Gbps design and manufacture capabilities, this performance differentiator between vendors gets smaller and again leaves low cost as the primary goal.

With OSAs well on the path to becoming the commodity item that the industry is demanding, focus turns to the ICs within modules. Like OSAs, ICs have also traditionally been a favored in-house procurement. This strategy makes sense when a module vendor is the market leader for a particular high-revenue application, or when the vendor wishes to implement proprietary electro-optic techniques to exploit a technical advantage. However, in almost any other case, the economies of scale of in-house IC supply do not add up.

Innovation in silicon integration enables competing IC vendors to offer differentiation through higher functionality and reduced component count. Additionally, by leveraging total volume over multiple customers against manufacturing, test and overhead expenditure, external IC vendors can deliver cost advantages over in-house IC supply. It should also be noted that in-house IC design teams are often limited to only designing the analog ICs at the physical medium dependant (PMD) optical interface of the module. The physical medium access (PMA) ICs that interface PMDs to the subsystem electrical domain are rarely the focus of in-house IC teams. This is because a PMA design would require the in-house IC design group to support a large multidisciplined team, drawing on analog, mixed-signal and digital design skills. To maximize cost savings, therefore, module vendors need a low-cost portfolio supply of PMDs and PMAs from one IC vendor, rather than mix and match from multiple sources.

New markets, fresh challenges

Industry consolidation and lower-cost modules alone will not guarantee profits, however. Vendors need to identify higher-volume markets. While telecom has seen little growth in recent years, the datacom, PON and storage markets have flourished.

Regarding storage, the industry recently witnessed Fiber Channel systems switching from 1Gbps to 2Gbps capability almost overnight upon completion of the 2Gbps Fiber Channel technical standard. Since storage is a high-volume, cost-sensitive application, system vendors pushed the module vendors for doubled performance, but at no cost premium. Module vendors overcame this challenge relatively quickly, since the increased data rate did not significantly push the technical boundaries for OSA and IC vendors beyond their then existing 1Gbps capability.

Today, the transition from 2Gbps to 4Gbps Fibre Channel is placing similar schedule and cost demands on module vendors, but the performance requirements over 2Gbps hardware are more challenging. OSA and IC suppliers have again addressed the challenge to help module vendors deliver their required goals.

An example of this is Phyworks Ltd, a UK-based fabless IC supplier, which recently announced volume production status of its 4Gbps Fibre Channel IC solution (PHY1070/75) for small-form factor modules. The circuit architecture, combined with a good package, process and manufacturing choices, allowed Phyworks to deliver differentiation and cost reduction to module vendors over exisiting solutions without compromising performance. The key architecture choice enabling customer value-add was the integration into a single transceiver chip of the traditionally separate receive post-amplifier and laser driver IC functions. An added benefit to customers through this lower chip component count is simpler module manufacturing and more efficient production test.

Today, most users connect to edge switches using copper at the 100Mbps Fast Ethernet data rate, but most new PCs and laptop computers are actually being shipped with 1Gbps GbE-capable ports. To exploit this increased client-side capability at the network edge and avoid transmission bottlenecks in the fiber backbone, the installed MMF must be capable of 10Gbps signaling.

The data-rate capability of MMF, however, is limited by the phenomenon of modal dispersion, which results in the "spreading" of a signal pulse as it propagates along MMF. This degrades the link's error-rate performance and, in the extreme case, causes complete "eye closure" of the received waveform.

Two possible solutions to this dispersion problem exist: first, the installed fiber backbone can be replaced with higher-bandwidth MMF that exhibits lower dispersive characteristics; second, electronic techniques can be used to counter the dispersive effects of the legacy fiber. Of these two options, pulling new fiber is not a viable solution for enterprise IT managers due to its prohibitive cost. The installed lower-bandwidth fiber will continue to represent a significant proportion of the total installed base for the foreseeable future. Hence, electronic dispersion mitigation techniques are required to enable enterprises to migrate their legacy fiber backbones to 10Gbps capability.

At present, the IEEE's 10GBASE-LX4 standard solves this problem. Since dispersive effects are a function of data rate, LX4 mitigates these effects by transmitting four optically multiplexed 3Gbps data paths rather than a single 10Gbps serial signal. However, LX4 compromises the low-cost requirements for enterprise networking due to its inherent optical, thermal and mechanical complexity, thus an alternative electronic solution to modal dispersion is still attractive.

The search for a cost-effective LX4 alternative is currently expending considerable industry R&D effort and is also undergoing IEEE standardization. Called 10GBASE-LRM, this proposed IEEE standard will rely on IC innovation in the form of electronic dispersion compensation (EDC) to solve the dispersion challenge. The LRM standard is expected to be ratified late 2005 to early 2006.

Due to the inherent cost-sensitive nature of enterprise networking, OSA vendors also play a role in delivering a successful LRM solution. Regardless of data rate or module form-factor, OSAs typically account for 30 to 40 percent of total module cost, and hence are a particular focus area for cost saving in this application.

As in the case of the 4Gbps Fibre Channel example, by careful integration, process and manufacturing choices, EDC IC and OSA vendors have the opportunity to deliver high-performance, cost-competitive solutions to module vendors. By integrating the EDC function with existing PMA functionality, IC vendors can offer dispersion compensation simply as a value-added feature. In the case of OSAs, and the receive optical subassembly (ROSA) in particular, by employing large area, low capacitance photodetectors, for example, the need for complex, and hence expensive, focusing optics can be obviated. Also, since optimal EDC operation necessitates a highly linear transimpedance amplifier (TIA) IC within the ROSA, EDC IC vendors should offer an EDC-compliant TIA to enable a portfolio receive IC solution.

Strategic partnerships

As the optical industry has become increasingly competitive, time-to-market pressures have led to closer collaboration between OSA, IC and module vendors. While IC and OSA vendor support to module customers has traditionally ended at the point of evaluation board dispatch, today's component vendors have become more sophisticated in the degree of support they offer to their customers. IC and OSA vendors are realizing they must now go beyond supplying just EVB's to customers. Recently, a small number of IC vendors have actually developed fully MSA-compliant reference designs that feature their own IC's and control firmware, and customer specified OSAs. This has the obvious time-to-market and cost benefits for module vendors, and reduces the time-to-revenue for OSA and IC suppliers.

While levels of support are a huge resource expenditure in terms of time and cost, IC and OSA vendors do not intend the irony of becoming vertically integrated module manufacturers. The objective of such detailed support is simply to maximize commercial opportunities for both component and module vendors.

- Chet Babla

Product Manager

Phyworks Ltd

Article Comments - Component vendors address module cha...
*? You can enter [0] more charecters.
*Verify code:


Visit Asia Webinars to learn about the latest in technology and get practical design tips.

Back to Top