Last-mile optical access has always been recognized as a fundamental requirement for efficient high-bandwidth deployment. In recent years, true fiber-to-the-home (FTTH) installations have finally been realized in many parts of Asia, and the technology is now spurring new markets in North America and Europe.
Along with this widespread deployment has also come a fundamental change in the optical components marketplace. With FTTH transceiver prices now entrenched below $100 and volumes exceeding several million units per year, the business of optical components has moved from a low-volume/high-margin arena to one of very low cost/high volumes. Concurrently, the incumbent thin-film filter technology, which relies on TO-can components, now faces serious competition from new advanced technologies based on highly integrated planar lightwave circuits (PLCs) and automated assembly.
FTTH deployment has increased steadily in Asia for several years now. It is reported that there are more than 70,000 new FTTH subscribers every month in Japan alone. Now the push for FTTH has reached North America and most other parts of the world. Two factors drive this broadband network revolution. First, subscribers want the exceedingly high-speed data and video on demand that FTTH can provide. Second, in recent years passive optical network (PON) technology has enabled carriers to deploy last-mile optical access economically on a massive scale, offering fewer truck-rolls and lower operating expenses.
Typical PONs use up to three wavelengths: 1490 nm for voice and data and 1550 nm for RF video downstream to each home, and 1310 nm transmitted upstream. In some cases, particularly in Asia, only two wavelengths are used, one upstream and one downstream. In every subscriber home there is a triplexer or diplexer transceiver to manage these optical functions.
The incumbent technology used for these transceivers is based on thin-film filters (TFFs) and TO-can technology (see Figure 1). This bulk-optic assembly technology has been used for many years in other applications, and has been adapted to support recent FTTH requirements. But in a market where transceiver prices are now well below $100, this technology is on the verge of being replaced by new, lower-cost technologies that have been specifically designed to address the high-volume, low-cost arena of FTTH.
PLCs, the subject of much R&D over the past decade, have now emerged as the technology of choice for replacing bulk-optic components in diplexers and triplexers. These PLCs are fabricated with the same processing technology used for making electronic ICs. Unlike traditional bulk-optic assemblies, where light is guided through a series of lenses and filters and free space, in a PLC approach the optical signals pass through waveguides on the chip, much in the same way that electrical signals are routed through an electronic IC.
Traditionally relegated to niche applications and only high-channel-count systems, PLCs are now poised to dominate several areas of optical networking, including ROADM applications, channel monitoring, CWDM, and FTTH transceivers. Traditional PLC components have typically been based on arrayed waveguide grating (AWG) technology. Although AWGs have proved themselves in DWDM applications that require a large number of tightly spaced wavelengths, this technology has shown itself quite unsuitable for FTTH applications. The large chip size of an AWG makes it prohibitively expensive for FTTH applications, and the free-spectral range of an AWG is typically much too small to cover the full PON wavelength range (1260 to 1565 nm). These shortcomings have required the development of new PLC filter technologies, such as Dispersion Bridge gratings.
In generalized terms, there are two different types of PLC approaches now competing for the FTTH market: the external-filter PLCs, and the embedded-filter PLCs such as those that feature Dispersion Bridge gratings.
In the external-filter PLC, the chip contains waveguides only for routing light to different parts of the chip, and has no embedded wavelength-filtering capabilities (see Figure 2). Instead, deep pits are etched into the chip, into which TFFs are accurately dropped, aligned, and bonded in place. These TFFs perform all of the WDM functions of splitting/combining wavelengths. In essence, the PLC platform acts as a new packaging technology for simplifying the alignment and assembly of TFFs. This approach, coupled with an efficient means for mounting lasers and detectors onto the same chip, provides a high-volume approach to manufacturing FTTH transceiver chips.
This external-filter PLC technology has matured in recent years, and products based on the approach are now generally available. The main challenge in this architecture remains yield, and therefore cost.
The embedded-filter PLC takes integration to the next level by embedding the wavelength-filtering technology directly into the optical chip. Advanced WDM filtering technology, such as Dispersion Bridge gratings, can be fabricated on the chip itself, incorporated into the regular processing steps involved with manufacturing the wafers. This eliminates the need for any external TFFs, greatly simplifying the subsequent assembly and packaging steps. The result is a highly integrated PLC design that requires no external lenses or filters of any kind. The low cost and efficiency of this approach are greatly compounded by the fact that all chips are made in a wafer form, where a single 6-inch silicon wafer can contain more than 500 triplexers based on Dispersion Bridge technology (see Figure 3a).
A suitable filter chip must also act as a platform for integration of the lasers and detectors found in every transceiver. One of the challenges faced by bulk-optic TFF diplexer/triplexer suppliers is that the alignment of lasers, detectors, and filters often requires an active process, in which all components must be powered up during assembly. This adds complexity and cost.
Modern PLC filter chips for FTTH components are designed to accommodate standard lasers and detectors on-chip, using automated passive alignment techniques. This assembly is typically done with an automated flip-chip bonder capable of running 24 hours per day with very little user intervention. Even transimpedence amplifier chips and associated capacitors can be mounted directly on the PLC filter platform, resulting in a completely self-contained FTTH transceiver chip (see Figure 3b).
In a bid to garner market share and outlast less capitalized competitors, many TFF diplexer/triplexer suppliers have recently sold components at a loss. Although this approach is not sustainable, it has proven successful in starving some smaller competitors into bankruptcy. However, the market continues to pressure pricing, calling into question whether incumbent TFF-based suppliers can follow this pricing curve as it continues to drop. Embedded-filter PLCs already promise prices less than $50 this year (see Table). This could signal a new trend in PLC-based components, and a new market in which PLCs have finally displaced traditional bulk-optic components for low-cost, high-volume deployments.
As PONs evolve in Asia and begin significant growth in other parts of the world, the ability of bulk-optic transceivers composed of TFFs and TO-cans to meet future demands for cost and high-volume manufacturability becomes questionable. Modern PLC technology has quietly matured to a stage where it can outperform competing technologies in nearly all respects. It does so at a fraction of the cost of those technologies, while being designed from inception for very low-cost, high-volume production. New filter technologies and PLC platforms are poised to revolutionize PON deployments and will be critical in enabling FTTH deployments to millions of new subscribers.
Dr. Matt Pearson is director of product development at Enablence Inc. (Kanata, ON, Canada; www.enablence.com). He has worked with development teams at Nortel Networks and Canada’s National Research Council, and was senior optoelectronics engineer in the Photonic Technology Operations at Intel Corp.