Today's DWDM "networks" are mainly point-to-point links with small channels counts, but these should evolve into denser, more complex architectures.
By Sophie Bourzeix, Vice President Optical Components, NetTest
The mux/demux market is a mature one where several technologies still compete. Thin film filters (TFFs) are the incumbent technology, well suited to small channel counts. They are also athermal components requiring no thermal regulation.
But, to address the need for more capacity, most system vendors using TFFs want to increase their channel count. The simplest solution is to replace TFFs with other athermal components.
General market features
All technologies have almost equal optical performance, each with its own specific advantages. For this reason, competition is now mainly on price. One can then ask which technology will take the lead and on what criteria: improved performance, manufacturing costs or other competitive advantages?
Regarding costs, although the various technologies are quite different - from the least integrated (thin film filters, or TFFs) to the most integrated (arrayed waveguide grating or AWGs) - the devices must always be connected to single-mode fibres. This operation is at best a semi-automated operation and therefore costly. Since the market is much smaller than forecast in 2000, the advantages brought by heavy technologies with inherent potential for mass production (AWGs) could even disappear, since the investments required can only pay-off for large production volumes.
Comparison of technologies
1. Thin film filters (TFFs)
A TFF mux is a cascade of narrow-band filters and single-mode fibres. This technology is competitive in performance and price for low channel count (<16) and channel spacing of at least 100GHz: it then shows low loss, flat spectral response, and athermal operation.
TFFs can show high performance on point-to-point links, since the mux and demux can be "paired" (the channels with the most loss at one end suffer the least loss at the other). But this isn't applicable to other network topologies. It is scalable, so wavelengths can be installed little-by-little in a system.
Fibres are collectively connected to a single optical chip, which performs all necessary functions. So, unlike TFFs, price per channel falls strongly with channel count.
Channel response is usually gaussian. Flat Top response models exhibit higher losses. Very good performance can be achieved for the best manufacturing processes (up to 2dB insertion loss in gaussian configuration).
Due to propagation in a solid medium, performance is strongly temperature dependent. To provide temperature-independent operation, basic AWG components include a thermoelectric heater (or cooler). Some work has recently been done towards athermal AWGs, but this technology is not completely mastered.
3. Bulk grating technology
The collimated light coming from the input fibre is incident on a bulk grating, which performs wavelength angular dispersion.
The input and output fibres are collectively aligned and construction is independent of the number of channels: price per channel falls with channel count. Also, decreasing channel spacing is just a question of optical scaling, with no intrinsic difficulty.
Since light propagates in air, bulk grating design has temperature-independent performance, with no need for thermal regulation. It also enables good values for channel uniformity, insertion loss and cross talk.
Advantages of an athermal solution
For the small number of channels in today's point-to-point links, the athermal characteristics of TFFs reduce both development costs (no electrical interface, flexible positioning even in non-electrical environment) and operation costs (no electrical consumption, no heat to dissipate, reduced failure rate).
System vendors wishing to boost capacity without extra development costs will replace TFFs with more integrated technology offering low price per channel and high channel counts: either AWGs or bulk grating technology.
To do so they will favour a "plug-in" athermal solution to avoid modifying the system design. Since bulk grating technology is the only mature, high-channel-count, athermal technology with performance equal to TFFs, it is the natural choice for future systems.
VP Photonics Optical
Dr Sophie Bourzeix was a Research Scientist at CNET from 1995-1997 and Photonetics' Tunable Laser group from 1997 and Mux-Demux group manager 1999-2001.