Any video network designer who thought his life became infinitely more complicated a few years ago when fiber optics came on the scene will be astounded by the sheer number of options that are coming on-line today. No longer is it just a debate over 1310 nm vs. 1550 nm, or hybrid fiber/coax (HFC) as opposed to switched digital video: It's that plus a few others. For example, what about Sonet? Should you use DFBs or Fabry Perots in the return path? How big should your node sizes be? When does wavelength division multiplexing make sense? The options are almost mind-boggling.


Furthermore, given the pace of technological innovation, the industry's appetite for newer, better and more cost effective hardware, and the need to upgrade because of competitive reasons, the number of choices will likely only grow. It's no time for the faint of heart.

1550: It's back!

Perhaps the biggest bear to wrestle with is the 1550 nm issue. Whereas before it was a technology searching for a market, it's now enjoying a resurgence that has the likes of Scientific-Altanta, General Instrument, Harmonic Lightwaves, ADC Video Systems and a host of others planning new products to challenge Synchronous Communications, formerly the only game in town. (See story, page 32.)

S-A thinks it leap-frogged most if its competition and caught up to the likes of Synchronous when it purchased ATx Telecom Systems from Amoco Oil Company. For a scant $25 million, S-A gained access to more than 50 patents related to critical optical components, including erbium-doped fiber amplifiers, 1310 nm amplifiers and a wide variety of pump lasers. In addition, the company garnered about 75 people (many of whom are Ph.D.s) and access to ATx's customer list, including Bosch Telecom and at least two other well-know OEMs.

But video network operators will now have access to a suite of products that were actually designed with them in mind, but were simply hard to find. Through the S-A acquisition, "We finally have access to the market and a distribution channel," notes John Clark, chief operating officer at ATx, which will continue to operate as a subsidiary of S-A. "It's like springing forth, finally, into the market."

What ATx brings is a high-gain, high-output erbium-ytterbium doped fiber amplifier that utilizes 800 nm pump lasers that cost less than the 980 nm pumps used in typical EDFAs, says Clark. The ytterbium is energized by the 800 nm pump and then is used as a sort of translator to energize the erbium at levels much higher than those produced by 980 nm pumps. The result? An output measured in the hundreds of milliwatts, vs. 70 to 110 mW with 980 pumps. The co-doping method isn't actually new, but ATx was successful in patenting methods to make the process more efficient (see Figure 2).


Other innovations include ATx's power source, which consists of a solid state laser to send light through a crystal that takes the wide output beam and focuses it into the fiber, as well as converts the wavelength to about 1047 nm, which then pumps the ytterbium fiber.

In fairness, Clark admits that 980 pumps are getting better (he knows because ATx also builds 980 pumps), but says he has an inherent advantage. "While 980 pumps are creeping up in capability, until there's a fundamental change in technology, they won't be as powerful. There are ways to (make them more powerful), but they're a few years away, and they aren't cheap." In addition, because 800 nm pumps are used in a wide variety of applications, both professional and consumer, they'll likely enjoy a price advantage for a long time.

Subtle improvements

Ortel Corp., which owes its success to the popularity of the 1310 nm distributed feedback laser it helped make ubiquitous in cable TV networks, is also branching out into 1550 gear with an EDFA, a transmitter and a receiver. While the company doesn't make any claim to groundbreaking technology changes, executives say there are ways to differentiate their products.

"Looking inside, everybody's products are pretty much the same," notes Larry Stark, vice president of new business development at Ortel. "But as we have studied 1550, we have found areas where we think we can make improvements. For example, there are a number of subtle degradations, such as SBS and SPM (Stimulated Brillouin scattering and self phase modulation), and we're finding ways to control them and improve performance. We hope that our product will be viewed as the performance leader in carrier-to-noise and distortion."

While that remains to be seen, Ortel plans to debut an externally modulated transmitter that, when combined with a 40 mW EDFA, can launch 80 channels of video over 65 kilometers of fiber with a composite triple beat of -71 dB, composite second order of -68 dB and a carrier-to-noise spec of 52 dB. The same link can be configured with about 3 dB more carrier-to-noise with CSO and CTB of -65 dB. "You can view it as a product with two personalities" depending on the operator's preference, Stark says.

With all its in-house expertise, Ortel is also building its own CW and pump lasers. "We believe we can develop the same technology leadership in 1550 that we did with 1310," Stark says.

With all the work being done at 1550, how are operators reacting? There is no pat answer. Time Warner, for example, has been using the technology for nearly three years in its transport rings, according to Don Gall, senior staff engineer. He, for one, is glad to see other manufacturers get into the game (mostly for the pricing leverage it gets him).

Gall says Time Warner has some, limited plans to use 1550 nm equipment in the distribution plant, but it's high-power solution doesn't make it useful for narrowcasting.

"We're looking at it on a limited basis for small (systems) that are more rural and have less potential for some of the futuristic services we're talking about," he notes.

TCI, however, is doing very little with the technology, according to Tony Werner, director of technology. Instead, the country's largest operator appears focused on either proprietary digital approaches or Sonet, which is commonly twice as expensive as the digital approach offered by ADC Video Systems, C-Cor Electronics and others.

Coming back around

And some have even used it in the past, but are now reconsidering 1310 nm equipment because performance is improving and costs are dropping. One such operator is InterMedia: "We're swinging back toward 1310" for its interconnects, notes Ken Wright, director of technology and chief technical officer.

InterMedia's approach has been to use a two-level architecture where signals are launched to an optical transition node (OTN), which in turn serves several nodes. This approach reduced the number of fibers that would exit a headend versus an approach that features fibers in a home-run to each node. The approach also allows an MSO to serve high-growth housing areas more easily by dropping in a new node where it's needed. "You can get there quicker and with less money, and many of the areas we serve are like that," says Wright.

To get the performance he needed, Wright was designing networks that used 1550 EDFAs. But with the advent of low-power 1310 DFB lasers, Wright says he can dedicate a laser per node at the OTN for about the same price as 1550 gear, while still preserving the option to offer targeted, narrowcast programming in the future. "That way, we've bought the granularity we'll need some day in the future" whether that be for targeted advertising or some sort of differentiated programming, Wright notes.

For Wright, the battle between 1550 and 1310 has been nothing but good for him. "The two technologies keep leapfrogging one another, and that's good for us," he points out. "There is no one answer."

The WDM solution

In fact, the time may soon be coming when both approaches become the answer-simultaneously. Although by most accounts it's a little early for cable operators, there are tremendous strides being made with wave division multiplexing equipment (which allows both operating windows to be used at the same time), and dense WDM gear that will offer up to 32 different "slots" between 1530 nm and 1560 nm.

Most of the Regional Bell Operating Companies plan to utilize WDM to help relieve congestion brought about by burgeoning new services like data transfer, videoconferencing and the sheer number of additional lines being requested.

To meet the need, just about every major telecom equipment manufacturer is getting into the game, although most are focusing on the 1550 nm window, where attenuation is lower and EDFAs can be used. Pirelli made major news at the most recent Supercomm show when it announced a 32-channel dense WDM system, giving telecom providers an amazing amount of new bandwidth without changing transport speeds or adding more fiber.

Certainly, Pirelli's target market is the telcos, but the company is seeing interest from MSOs who are interested in offering telephony and data. "Our customer base for these digital applications is really growing," notes Fahri Diner, product line manager at Pirelli.

Similarly, Artel Video Systems recently announced a 12-bit digital video transport system that can handle up to 48 non-compressed channels (six channels in each of eight slots) over a single fiber.

"It is becoming more evident that dense WDM will be more cost-effective and quick to deploy than designing higher-speed products, especially in the digital domain," notes George Maier, vice president of marketing at Artel.

But do cable operators need it, yet? It's probably a bit early, but if an operator wants to supertrunk high-quality video, he could use up his usable fiber bandwidth rapidly. And, as Maier points out, consumers' tolerance for low-quality pictures is strained now that DSS and others are offering digital pictures.

Gall and Wright both think it's a bit early to deploy WDM solutions, but the possibility looms. "If, because of telephony or alternate access traffic, we start filling up fibers, we might do it," says Gall. "Right now, most of our rings have an average of 36 fibers, and to broadcast video, we only need two of them. But that's one of those things that's changing rapidly, so it's an option we'll use when we need to."

Naturally, a key question for cable TV providers is the isolation between densely-packed video channels, notes John Dahlquist, vice president of marketing at Harmonic Lightwaves. "But it's just a matter of time" before that technology matures, and performance becomes acceptable. "All these things will have their time in the sun," predicts Dahlquist.

Outside of new wavelengths and densely-packed streams, Stark sees a few other trends. "I'm going to go out on a limb and predict that pricing models will begin to follow the personal computer model," he said, "where prices stabilize or perhaps drop a bit, but the performance increases."

Stark says performance gains in lasers and receivers can be made in distortion, carrier-to-noise and channel count. "I still think the end-game with HFC networks is fiber to the last active," he stresses. "We'll eventually get to the point where the fiber link will define the performance between the headend and the home. Today, it's fiber and coax, but someday, the coax goes away."

Being the DFB champion that he is, Stark also says the time is ripe for an uncooled DFB in the return path, a product the company recently announced. "It gives at least 10 dB of headroom over a Fabry Perot, which ought to give operators more confidence in their networks," he notes. That might explain why Stark says he's been "overwhelmed" with interest in the product.

But, once again, there's a divergence of thoughts on the subject. "Everything we have done so far suggests that we can make Fabry Perots work for a digital return, as long as we don't have to do two-way video," says Gall. "If we do have a need for a really high-speed data path or a video return, such as an I-net for a school, then we'll put in a DFB. But that's a one percent solution."

Lab work

Finally, for those who like to look way out on the horizon, there's the 1310 nm, praseodymium-doped fiber amplifier. While some prototype and demonstration devices have been built, they aren't commercially viable yet because of the cost, according to ATx's Clark. "The problem is the fiber-it soaks up water and turns to dust."

One fiber that has been developed to overcome that problem has its own problem-it can't be fusion spliced. Nevertheless, Clark predicts that a good, usable 1310 amp will be on the scene within a few years, but it will take longer than some predict. "There are plenty of papers that say it will work," he says. "But that's just guys in the lab. They don't have to make customers happy or produce any real products."

With all the new and improved products coming over the horizon, a network designer's choices won't get any easier. But the tools should make the final product - the video network - stronger than it's ever been.