& cplSiteName &

L-Band Amps: The Revolution?

Light Reading
News Analysis
Light Reading
7/16/2002

Use of L-band Erbium Doped-Fiber Amplifiers (EDFAs) could become more widespread in the future if U.K. startup Southampton Photonics Inc. (SPI) has its way. The company introduced a new L-band amplifier today (see SPI Announces L-Band Amp).

Up until now, L-band amplifiers, which amplify optical signals in the wavelength range 1565 to 1605 nanometers, have been deployed to increase channel counts in DWDM systems, by opening up new wavelengths alongside the commonly used C-band (1525 to 1565 nm). At the moment, however, no one is building ultra-high channel count systems, except possibly in lab experiments.

But there's another reason for using L-band components, says Southampton's CEO David Parker -- they give better overall system performance than standard C-band ones. Put simply, it is easier to build a high-speed system over long distances using the L-band. The L-band is particularly useful in situations where non-zero dispersion shifted fiber (NZ-DSF) has been installed, because it allows the system to operate away from the zero dispersion point, and thus reduces the crosstalk caused by four-wave mixing (see Nonlinear Effects).

Carriers could start using L-band components in preference to C-band ones, Parker says, and the main thing that's stopping them is the cost of the optical amplifiers.

"We sense from our customers that they see the advantages, but it just costs too much," he contends.

Southampton's new product will overcome the price problem, it claims. "We're not just saying it's a better amplifier," says Parker. "It is massively cheaper than conventional technologies." Although he can't give exact pricing, which depends on the customers' specifications, he does boast that the cost reduction of the new amp will be "fifty percent or better" compared to existing technologies.

Do Southampton's claims stand up to scrutiny? A 50 percent cost reduction would make the cost of L-band amplifiers comparable to C-band amplifiers today, notes David Smith, VP of hardware engineering for Corvis Corp. (Nasdaq: CORV). That means that any decision to use the L-band in preference to the C-band could then be based on performance only.

"The L-band is better in some respects," says Smith. "But I'm not sure that people should be making claims that the L-band is infinitely better."

In terms of performance, the L-band typically has higher dispersion and therefore lower crosstalk. In certain types of fiber with very low dispersion, such as NZ-DSF, crosstalk limits the number of channels. Using the L-band rather than the C makes it possible to cram more channels in this situation, he notes.

The other key parameter is the noise figure, which is lower in the L-band. That means that more amplifiers can be cascaded before the signal has to be electrically regenerated, so ultra-long-haul optical systems could go farther more easily in the L-band.

Smith also points out that most optical components today work in the C-band, and there are relatively few available for the L-band. "The way the industry is looking at it today," he says, "is to use the C-band for as long as possible, then when they exhaust the C-band, move into the L. If, two years from now, components are available in the L-band, and companies are building new networks, maybe they will start in the L-band and move into C later."

Whether Southampton's announcement marks the beginning of a change in the way people build DWDM systems remains to be seen. However, it does mark a turning point in the startup's strategy.

The market Southampton started out in -- making high-density, narrow channel spacing Fiber Bragg Gratings (FBGs) -- isn't large enough to sustain the company in the current downturn. So it is looking to widen its horizons with products that leverage its core expertise in making strange kinds of optical fibers -- including fiber lasers and fiber amplifiers (see Fiber Bragg Gratings on Speed).

The key to Southampton's L-band amplifier is a new kind of fiber, called GT Wave, in which optical power is coupled from pump lasers into the core of the fiber in an extremely efficient manner. Inefficient performance is the fundamental problem with today's L-band amps, says Southampton's Parker.

GT Wave fiber contains multiple cores -- several large ones to carry light from the pump lasers, and a smaller, singlemode core that carries the data signal. As the fibers are drawn on a fiber-pulling tower, the multiple cores are twisted around each other. (Parker tried to demonstrate this with three fingers, but since he was at the other end of a phone line, the finer details of the demonstration were lost.)

Using this technique, Southampton claims it is possible to make amplifiers with outputs of 1W, which is considerably more than is required for telecom applications.

Questions remain about the market opportunity for EDFAs, however. The market for optical amplifiers of any kind is said to be fairly small this year and next, according to recent figures from Communications Industry Researchers Inc. (see Report: Slow Ramp for Optical Amps).

Southampton plans to address a range of applications, not just telecom, with its high-power laser and amplifier products.

Prototypes of the L-band amplifier should be available at the end of the year. General availability is slated for the first quarter of 2003.

— 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 at Opticon 2002, Light Reading’s annual conference, being held in San Jose, California, August 19-22. Check it out at Opticon 2002.

Register now and save $500 off the registration fee. Just use the VIP Code C2PT1LHT on your registration form, and deduct $500 from the published conference fee. It's that simple!

(12)  | 
Comment  | 
Print  | 
Related Stories
Newest First  |  Oldest First  |  Threaded View        ADD A COMMENT
Page 1 / 2   >   >>
photonalley
photonalley
12/4/2012 | 10:04:55 PM
re: L-Band Amps: The Revolution?
I don't understand how l-band EDFA's are going to be 50% cheaper than other suppliers. If I remember correctly it is the pump lasers that drive the cost. Can someone explain this how this may be possible? I means it is still Erbrium fiber, right?

As far as L-band components being more cost effective than C-band that is just not true in my experience. I have purchased L-band and C-band DWDM components for almost two years and their is no difference in costs. This is only my experience. Does anyone else know of a price difference between c and l band optical components?

Photonalley
lilgatsby
lilgatsby
12/4/2012 | 10:04:55 PM
re: L-Band Amps: The Revolution?
If the trend today is to squeeze the most channels out of the C-band which is and has been taking place as technology enables the shift from 100Ghtz to 50 to 25 and moving down the line to 12.5 at some point, where does this L-band savings come into play? Stretching 320 or more channels out of the C-band takes time to over subscribe.

I am curious having heard certain vendors throw the L-band card on the table as part of a shipping strategy to maximize channel count. Is there any L-band deployed and WORKING routes in the US today? And please, keep Marconi and the Soliton sham out of this...I'm asking about real networks from real companies.

lg
inlight
inlight
12/4/2012 | 10:04:54 PM
re: L-Band Amps: The Revolution?
May be the following press release may be of interest in the current thread:

http://pr.fujitsu.com/en/news/...

IL

(Short extract below)
-------------------------------------------

Fujitsu Delivers Japan's First 1.76Tbps DWDM System

- World's Largest-capacity Commercially Available
Long-haul Optical Transport System Installed for
Crosswave Communications' Tokyo-Nagoya-Osaka Backbone -




Tokyo, April 3, 2002 -- Fujitsu Limited today announced that it has delivered its FLASHWAVE-« 7700 dense wavelength division multiplexing (DWDM) long-haul optical transport system to Crosswave Communications Inc. for use in the company's Tokyo-Nagoya-Osaka backbone network. Fujitsu's FLASHWAVE 7700 is the world's highest capacity commercially available DWDM system, capable of transmitting 1.76 terabits per second (Tbps) over vast distances - without regeneration - and the installation for Crosswave Communications is the first FLASHWAVE 7700 deployment in Japan.

The FLASHWAVE 7700 is a next-generation DWDM transmission and optical add/drop platform that provides up to 176 channels, each capable of carrying 10 gigabits (billion bits) per second of traffic. The system uses 88 channels in each of two bands of the transmission spectrum, the C and L bands, using 50GHz spacing between each channel. This 1.76Tbps transmission capacity is equivalent to simultaneously distributing MPEG4 coded high-definition video to approximately 34,000 users. The system also features a LightGuard (*1) function and Automatic Optical Level Control Technology (*2) to maximize network reliability and data transmission quality, as well as to shorten the time required for system start-up and wavelength additions during operation to one-tenth that for previous systems.

Crosswave Communications, with a network linking its data centers in Japan, offers a full range of network services, focusing in particular on wide-area LAN and high-speed backbone service. On April 1, the company began a commercial rollout of its new Gigabit Ether Backbone Service, and the FLASHWAVE 7700 platform is being used to strengthen the approximately 1,400-km (875-mile) backbone network that links its data centers in Tokyo, Nagoya, and Osaka to handle the expected increase in data traffic.

The FLASHWAVE 7700 platform was chosen based on Fujitsu's established track-record as a top optical transmission systems provider in the North American market and its market-leading innovations in this arena. Fujitsu is optimistic that the successful Crosswave installation will help the company capture a significant share of the market in Japan, where demand for these systems is growing.

Notes:

(*1) LightGuard:
Redundant double route function. Instantly switches traffic to a 2nd backup route in case of optic fiber troubles for maximum reliability.


(*2) Automatic Optical Level Control Technology:
Technology where optical level of each wavelength is automatically adjusted. By controlling 176 discrete wavelengths to a uniform level, high quality and stabilized transmission is realized. Also, because the optical level of each wavelength is automatically controlled,the time required for system start-up and recalibration when adding wavelengths during operation are shortened drastically.






Petabit
Petabit
12/4/2012 | 10:04:46 PM
re: L-Band Amps: The Revolution?
What follows is a fairly techie and detailed explaination of what the article should really be saying.

There is no intrinsic reason why L-band components should cost any more or less than C-band components. The material systems are the same, and only some of the component specifications change (the key one being the length of the GRiN lenses). What does make a difference is the volume of components shipped - because of the much higher volume of C-band components, they are typically 15% cheaper than L-band components.

So you might think that there should be a 15% difference in the cost? Wrong.

The reality is that a simple L-band EDFA will have a higher noise figure, and use more pump power than a comparable C-band EDFA. This has to do with the spectroscopy - the 980/1480 pump has to interact with the erbium ions to produce 1530 nm ASE, which is then used to pump the L-band signals (in very simple terms). This means that there is relatively little L-band gain in the critical front section of a simple amplifier. The front section is what controls the noise figure of the amplifier - and so a C-band amp is easier to design than an L-band amp.

The literature reports that C-band amps have been made with noise figures as low as 3.1 dB (for an amp with decent gain). If you work really hard at the design of an L-band amp, you can improve the noise figure and efficiency. The best NF I know of for and L-band amp was reported at OFC 2000 at 3.2 dB. All that extra complexity to produce a good L-band amp adds to the cost.

Then you need to know that an L-band amp needs about ten times as many erbium ions as a C-band, which translates into ten times the length of erbium fibre.

So all in all, the prices that I see for amps today show that there is a 30% premium for L-band amps.

What SPI are touting is a double clad fibre. Discovered and patented by Polaroid in the late 1970s, most of the work on these fibres was supressed by the unwillingness of Polaroid to licence the patent. They sold to IPR to SDL (now JDSU) in the mid 1990s, and now it is not too hard to licence. You can buy double clad fibre from a dozen or so fibre manufacturers today.

The touted advantage of double clad fibre, is that you can keep your signal inside the core, nicely single mode, with the erbium in there as well - and then you can use a 'cheap' multimode pump to excite the erbium. The theory is that you can save a load of cost in the pump, and therefore make the amp cheaper.

The drawback is that conventional pumps have got very cheap in the past few years. So much so that they are fast approaching the cost of multi-mode pumps, and therefore the cost and compelexity of coupling double clad fibre to the signal fibre is just not worth it.

SPI are showing their academic root by producing a product that would have been really good about five years ago. Sadly the world has moved on, and you can get much better and cheaper amps from the likes of Corning, JDSU, Nortel and Agere.

Pauline, if you want a corroboration of these details I suggest that you talk to Onetta. They seem agressive enough to give you a good counter-story.

P.

Cheeky_Chappie
Cheeky_Chappie
12/4/2012 | 10:04:45 PM
re: L-Band Amps: The Revolution?
{i}I am curious having heard certain vendors throw the L-band card on the table as part of a shipping strategy to maximize channel count. Is there any L-band deployed and WORKING routes in the US today? And please, keep Marconi and the Soliton sham out of this...I'm asking about real networks from real companies.{/i}

As you've already said, people won't want to move into the L-band until they've squeezed all the performance out of the C-band. At the moment, it's not really a financially attractive upgrade path for those who are blessed with routes of C-band WDM friendly fibre.

Countries such as Italy, Japan and several others laid a large amount of G653 (DSF) years ago, and are now lumbered with a fibre that is useless for C-band WDM. Are they going to want to rip all that up and lay frsh fibre just to use the C-band up before the L band? Doubtful. They're grateful for anything that helps them use their fibre better.

CC
edgecore
edgecore
12/4/2012 | 10:04:45 PM
re: L-Band Amps: The Revolution?
All this talk about squeezing more channels out of existing systems...it makes perfect logical sense, but does anyone know how many carriers out there actually have a high number of OC192 channels lit up.

I had heard that the most QWEST has light up on their 1600G was 40 lambda's? Any opinions on this out there?

These days any conversation on the topic of extremely high channel counts seems way out in the distance!

EC
Pauline Rigby
Pauline Rigby
12/4/2012 | 10:04:43 PM
re: L-Band Amps: The Revolution?
Petabit, I don't believe that SPI's amp is just a double clad fiber. I probably didn't go into enough detail in the story, so here's a bit more about what I learned in my conversation with David Parker yesterday.

The problem with double clad, as you point out, is the optics. It requires complicated optics to get the light into the outer cladding of the fiber in the first place. And once the light is in, it's difficult to get the light coupled from the outer cladding to the inner core where the erbium is. Various methods are used, such as etching gratings, (didn't understand this bit), but the point is, it doesn't couple well without help.

From the explanation I got, it sounds like the SPI fiber is a multicore fiber. There are two big fat, round offset cores (not annular ones) to get the pump light in, which are twisted like a rope around the central round core during the pulling process. Light can couple from the fat cores into the central core as a result of the twisted structure of the fiber cores, and without having to make gratings or other localized structures for couplin. The result is very good cross coupling along the whole length of the fiber.

The fiber is still the same, as you point out. But since coupling is better, a lot more light gets into the central core, and therefore less fiber will be needed.

I think it sounds cool. And I don't suppose it only applies to the L-band.

[email protected]

Petabit
Petabit
12/4/2012 | 10:04:27 PM
re: L-Band Amps: The Revolution?
Hi Pauline,

Thanks for the clarification. I had forgotten that SPI had done some work on multi-core fibres.

The last I saw of that work was the papers they published at OAA 1999, when they used a two core fibre. They doped both cores with erbium, but only pumped one of the cores. This led to a fibre that automatically flattened the gain, as well as providing a lot of gain in a very short length. It looked really cool, and almost worked.

The problem with multi-core fibres is that you have broken the rotational symmetry of the fibre. The moment you use asymmetric cores, then the birefringence of the fibre increases. Birefringence is bad, because different polarisations of light travel at different speeds through the fibre - leading to PMD and PDL.

Spinning the fibre can help, it causes the signals to 'see' all the different angles of the birefringence, and so averages the effect out. Unless you get a complete number of turns into your fibre, the averaging will not be complete, and you are still left with some PMD. It's not helped by the long lengths of fibre that you need for L-band amps either.

Interestingly a thin centre core, with two fat caores either side of it sounds exactly like PANDA fibre. PANDA is one type of polarization maintaining fibre - very high birefringence.

It's still not easy to splice to a multi-core fibre.

They tried to sell the 1999 fibre to a couple of amp manufacturers, but they all rejected it for PMD and splice reasons. It seems that they have a new design (with two cores this time), and it will be interesting to see if it works any better.



Someone was asking about field deployments of L-band amps. There are a load in Japan (on DSF), Nortel has shipped more than a few in North America (both LH and Metro), and there are a few other metro deployments as well. But overall not a significant percentage.

P.

catbrier
catbrier
12/4/2012 | 10:04:06 PM
re: L-Band Amps: The Revolution?
From Petabit:
"The reality is that a simple L-band EDFA will have a higher noise figure, and use more pump power than a comparable C-band EDFA. This has
to do with the spectroscopy - the 980/1480 pump has to interact with the erbium ions to produce 1530 nm ASE, which is then used to pumpthe L-band signals (in very simple terms). This means that there is relatively little L-band gain in the critical front section of a simple amplifier. The front section is what controls the noise figure of the amplifier - and so a C-band amp is easier to design than an L-band amp."

I agree with almost everything in your post except that there are some errors in your above explanation of noise figure. It is true that the front section of the amplifier controls the noise figure. (This is true for "quasi-3-level" systems like Er but not for all laser ions.) However, the amount of gain in this initial section of the amp is relatively irrelevant -- what matters is the population inversion. Because of the spectroscopic properties of Er, the L band should actually have a better noise figure than the C band if the population inversions are the same. (For a technical explanation of this non-obvious fact see the chapter by Miniscalco in Michel Digonnet's book _Rare-Earth-Doped Fiber Lasers and Amplifiers_. This theoretical prediction was verified about 10 years ago by BT Labs.) One way to suppress C-band gain and move more gain into the L band, is to underpump the amplifier, and this results in a lower population inversion and hence a higher noise figure. The 1530-ASE pumping you mentioned is an attempt to get the best of both worlds: high inversion from the diode pump in the initial section of the amp that controls the noise figure, and lower inversion in the following ASE-pumped sections of the amp to shift the gain peak to longer wavelengths. The joker in the deck is that while it is the ASE that co-propagates with the signal that causes the noise, in a high-gain amp is is the backward traveling ASE that depletes the inversion in the initial part of the amp and spoils the noise figure. This led to the technique of breaking the amp up into two sections with an isolater in between to suppress backward ASE.

A little long-winded, but I hope it helps.

--Catbrier
Petabit
Petabit
12/4/2012 | 10:03:46 PM
re: L-Band Amps: The Revolution?
Catbrier wrote "A little long-winded, but I hope it helps."

Thanks for the additional info. I did know that I was being a little simplistic, but I was trying to keep the explaination straightforward. It's interesting how both Digonnet and Desurvire both explained how L-band amps worked long before they were re-discovered by Bell Labs. Another example of a Bell Labs 'invention'.

Good to see another engineer on the boards, I hope you continue to post.

P.
Page 1 / 2   >   >>
Featured Video
Upcoming Live Events
March 16-18, 2020, Embassy Suites, Denver, Colorado
May 18-20, 2020, Irving Convention Center, Dallas, TX
All Upcoming Live Events
Upcoming Webinars
Webinar Archive