Optical components

Startup Touts 'Tunable Demultiplexer'

Lots of startups are developing tunable lasers, and quite a few are developing tunable filters, but Sparkle Optics Inc. is working on something else that’s tunable -- a tunable demultiplexer.

A demultiplexer, in this context, is a widget used in wave division multiplexing (WDM) systems to split light into different colors, or wavelengths, so that a single fiber can actually carry multiple streams of data, each on a slightly different wavelength. Making it tunable, as Sparkle is doing, means that the band of wavelengths it's dealing with can be changed, almost on a whim.

Opinions differ on whether demultiplexers really need to be tunable-on-a-whim in this way. But Santanu Basu, who heads up Sparkle Optics, a tiny contract R&D firm, clearly has applications up his sleeve. For the time being, however, he’s fobbing people off with a niche one -- saying that such a device could be a boon for systems integrators that want to buy a single demultiplexer for their labs, one that could work anywhere in the C, L, or S bands.

Anyhow, Basu figures his tunable demultiplexer has a number of advantages over arrayed waveguide gratings (AWGs), one of the existing ways of splitting white light into different wavelengths. In particular, it’s practical for small channel counts, it doesn’t require temperature control (which consumes power), and, as noted, it’s tunable.

How does it do all this? Because it's based on a diffraction grating, says Basu, and these properties are inherent to the technology.

It's important to note that Sparkle isn't the only company developing grating-based demultiplexers. In fact, there's a growing band of players, including APA Optics Inc. (Nasdaq: APAT), Highwave Optical Technologies, Lightchip Inc., Photonetic (now part of GN Nettest), and Zolo Technologies Inc.. These companies already have products out.

Sparkle, on the other hand, has quite a bit of work to do. Basu says he's built a six-channel prototype out of bulk optical components, like lenses and ruled gratings. The prototype has a channel spacing of 50 GHz, which is good. It has an optical loss of 10.7 dB, which is bad. And the dimensions are 11.3 x 3.7 x 1.3 inches, which is laughably huge.

"It's a small beginning," acknowledges Basu, without irony. "Once we have completed the technology licensing, we will make a second generation in small quantities. We will improve the size, the crosstalk loss, and polarization-dependent loss."

It should be possible to reduce optical losses to less than 6 dB, which is lower than commercially available AWGs, he adds.

Sparkle's competitors are not easily impressed, however. Mike Wearsch, VP of business development at Zolo says that what's needed is reconfigurability, not band tuning. "If you're making an add/drop multiplexer, you need to be able to drop any six channels, not just the six channels next to each other," he says. Zolo, he notes, is working on a way of doing this using MEMS (micro-electro-mechanical system) technology.

AWG vendors are quick to point out that grating-based devices in general could have a big drawback: reliability. "How do you expect it to stand up to vibration when it's going to be hand-made in Asia with cheap labor?" says John Midgley, CEO of Lightwave Microsystems Corp.

Basu counters this criticism by saying that he's planning to make the demultiplexer smaller and more robust by miniaturizing and integrating fewer components onto a silicon optical bench.

All of this work is being undertaken by Sparkle under R&D contracts with Silicon Valley companies, Light Reading has learned.

Basu -- a graduate of Stanford University and former IBM Corp. (NYSE: IBM) researcher -- didn't want to reveal the identities of the companies sponsoring his work, but did give a few hints.

Both companies are in the Bay area, he said. One is an established manufacturer of lasers, not just for telecommunications but for a wide range of applications. The most likely candidate is Coherent Inc. (Nasdaq: COHR), which has its headquarters in Santa Clara. This also fits with Coherent's recent decision to refocus the company on telecom components (see Coherent's Strategy Shift Explained).

The other vendor is a "strong telecom player," and it's not JDS Uniphase Inc. (Nasdaq: JDSU; Toronto: JDU), which has decided to back AWGs (arrayed waveguide gratings) all the way, says Basu.

— Pauline Rigby, Senior Editor, Light Reading
http://www.lightreading.com Want to know more? The big cheeses of the optical networking industry will be discussing this very topic in a session at Opticon 2001, Light Reading's annual conference, being held in San Jose, California on Aug 13-16. Check it out on Opticon2001.

Page 1 / 2   >   >>
redface 12/4/2012 | 8:03:54 PM
re: Startup Touts 'Tunable Demultiplexer' "AWG vendors are quick to point out that grating-based devices in general could have a big drawback: reliability. "How do you expect it to stand up to vibration when it's going to be hand-made in Asia with cheap labor?" says John Midgley, CEO of Lightwave Microsystems Corp."

Does Mr. Midgley have any clue what he is talking about? The reliability of a manufacturing process is mostly determined by the designs, not by how much the worker is paid, as long as the worker is not forced to work. JDSU, New Focus etc all have highly reliable products (which do stand up to vibration) hand-made in China with "cheap labor".
realoptics 12/4/2012 | 8:03:44 PM
re: Startup Touts 'Tunable Demultiplexer' Read with some interest that someboday at this late moment in grating based Mux/Demux domain is still bragging about 6 dB insertion loss type of demux, tunable or not! As a matter of fact, all the grtaing based Mux/Demux players, such as Cisco backed voulme grating based, fore-runner in this game, BaySpec, Inc (www.bayspec.com) in Fremont, California and the others, as the author has mentioned, all have better than 6 dB insertion loss devices, and they are shipping! Furthermore, AWG's performance is not necessary all worse than 6.0 dB insertion loss, as Mr. Basu has misled, they have better devices too.

Grating based technology has a lot of merits compare to AWG in many applications, what AWG got is only a marketing name. What Mifgley said is entirely false because he does not understand grating at all, nor does he understand AWG. Therefore he his scared to say something irelevent.

All 30 AWG players are scared to the potential competition from the handful good grating guys! Why? Simple answer:last 2 years AWG vendors spent total more than U.S$2.0 billions combined but still could not get the performances that grating guys reached with a few millions!
Petabit 12/4/2012 | 8:03:42 PM
re: Startup Touts 'Tunable Demultiplexer' "Does Mr. Midgley have any clue what he is talking about?"

Err. Yes. Dr. Midgley runs the only company that has even come close to commercialising AWGs. Having met him on several occasions, he is an intelligent man who does understand the business.

I think his comments are being taken out of context, so I'll try and explain:

Contrary to anything the article states, the dominant technology in DeMUXes is thin film filters (also known as dielectric filters). These offer insertion losses for small channel counts that other technologies can't match. They also offer a 'pay-as-you-grow' strategy which is becoming more important as the market changes.

The thin film filters used to be mainly made in California, but much of the work has been outsourced to plants in Asia - where the labour is cheaper. I don't believe that the quality has suffered.

The point that you miss is that assembling a thin film filter is mainly a matter of glueing together a whole load of optics elements. Once the device has been assembled it forms a solid whole, so is insensitive to vibration etc. Bulk grating based filters rely on length to spread out the wavelengths, and this length is usually the problem. It is much harder to design a vibration insensitive grating base device.

Let's try an example. Have you used an OSA recently? How about one of the field portable variety? If you take one of those OSAs, which is a tunable grating based demux (much like Sparkle are offering - similar tolerances required), and then drum you fingers on the side. See what happens to the trace? Those little sinusoidal patterns that show the vibration sensitivity? And this is from a ruggedised instrument.

Sparkle is going to have a lot of work ahead of it to prove it's technology against the incumbent. Avanex promised to revolutionise the world with VIPA.

PlainOptics 12/4/2012 | 8:03:41 PM
re: Startup Touts 'Tunable Demultiplexer' "Err. Yes. Dr. Midgley runs the only company that has even come close to commercialising AWGs. "

Have you ever heard of NEL and PIRI, or even Lucent?
realoptics 12/4/2012 | 8:03:40 PM
re: Startup Touts 'Tunable Demultiplexer' Petabit is wrong on commenting '...the only one'! as also pointed out by the other letter, NEL, PIRI, Hitachi, and other more than 2 dozens companies are all having different levels of success. NEL, by far is the best supplier in AWG and they've been doing research for 16 years before they saw some fruits. Lightwave certainly is not the only one and is only 'close to commercialization'

The key drawback for filter based technology is hard to scale, it is pretty hard, if it is not possible to make any mux/demux with 16 or more channels.

Another point is wrong is on Avanex's so called VIPA, it is really an interleaver, The Avanex VIPA is only good for intermidiate solution, it is going to be deemed as useless when grating and AWG based technologies are fully mature.


wchieh6 12/4/2012 | 8:03:33 PM
re: Startup Touts 'Tunable Demultiplexer' One article in May talked about echelle grating.
Is there anyone konw about the real performance of echelle grating based Mux/Demux with respect to AWG?
Petabit 12/4/2012 | 8:03:32 PM
re: Startup Touts 'Tunable Demultiplexer' I admit my grand sweeping statement about who could make AWGs was a bit over the top. I still stand by the assertion that Lightwave Microsystems are one of the leaders in actually shipping AWGs in commercial volume.

I note that no-one else has had a go at John Midgley though.

"The key drawback for filter based technology is hard to scale, it is pretty hard, if it is not possible to make any mux/demux with 16 or more channels."

I think you're wrong there. I can think of a commercial product that has shipped several billion worth of units where 80 channels are done with thin film filters.

"Another point is wrong is on Avanex's so called VIPA, it is really an interleaver."

According to the Avanex website, the interleaver is called a 'Asymmetric Nonlinear Interferometer' and VIPA is something different.

benson 12/4/2012 | 8:03:29 PM
re: Startup Touts 'Tunable Demultiplexer' Petabit;

You are completely wrong about the commercial situation regarding AWG's.

Both Hitachi Cable and NEL have shipped thousands of AWG's. In the case of Hitachi Cable, they have shipped more than 9000 AWG's over the past 6 years. Lightwave Microsystems is not even close.

Neither of these companies have an IPO to launch, nor a stock to sell in the US. Hence, both go quietly about their job.

You are reporting on marketing hype, not the actual commercial situation.


redface 12/4/2012 | 8:03:25 PM
re: Startup Touts 'Tunable Demultiplexer' Hi Petabit:

I generally agree with your comments. The reason I take issue with Midgley is his statement in the article "How do you expect it to stand up to vibration when it's going to be hand-made in Asia with cheap labor?". Here Midgley is directly linking reliability with cheap labor which I disagree. If he had mentioned the length of the grating device as a reliability concern, I would not have had any issue. In a world of fiberoptics which is filled with hype and misrepresentation, I believe a little more preciseness in communication definitely helps.

I do believe grating devices have a much tougher time getting Bellcore qualification compared to AWG.

Regarding Avanex's VIPA, it is a demultiplexer, not an interleaver. It is very easy to see if one reads Avanex's patents. It can be used as a tunable dispersion compensator. However, the size of this device does make it nearly impossible to be made reliably.
mrtp4 12/4/2012 | 8:03:25 PM
re: Startup Touts 'Tunable Demultiplexer'
The idea of the Echelle Grating is to combine the scalability and performance of bulk-gratings with the high-volume manufacturability of AWGs.

The biggest benefit of EG comes from the much smaller device size, which gives better yields and allows for more on-chip integration.

Compared to AWG, crosstalk is inherently better in Echelle Grating, and power requirements are much lower (closer to 1W, vs 5W). Insertion loss is currently about the same, <6dB, although theoretically it should be better than AWGs.

The real difficulty in making Echelle Grating Mux/Demux is the PDL, and only one or two companies have succeeded in fixing that. Optenia is one. There is a white paper that compares the technologies at www.optenia.com

Page 1 / 2   >   >>
Sign In