Why do you need broadband? Text? Graphics? or Digital video?????????????????????????
cedmagazine.com
By Michael Lafferty, Associate Editor
In the good old days, when cable operators generally flew by the seat of their pants while speeding on to the next installation, there wasn't much time spent examining the intricacies of the equipment they were installing in homes or stringing on poles. As long as it worked reasonably well and didn't break down too fast, both vendor and operator got along fine. That was a strictly analog world where a little static or snow could be tolerated, or even excused for long periods of time.
But that was then. This is now.
Today, as digital television, data communications, IP telephony and a host of streaming services are preparing to blaze down cable's high-speed pipe, the ability to get along with lackluster electronics has all but disappeared. In a world where the digital "cliff effect"-i.e., where the information sent is either perfectly replicated or nonexistent-is very real, there is no room for error. Equipment, whether it's in the headend, on the pole or in the consumer's home, has to operate perfectly, as specified, as expected.
As a result, operators now have to pay even closer attention to what's going in the equipment they're specifying. They're finding that the devil, as it's often said, is definitely in the details.
New pressures, new demands
Broadcom's BCM3300 IC
As the cable industry's broadband pipe evolves from the bucolic one-way video days to the brave new world of interactivity, new pressures are being put on systems that really haven't been pushed to the max yet with with new services. According to Bill Moore, vice president and general manager at Ortel, the real crunch may be approaching sooner than most think.
Moore says that while a lot of attention currently has been spent on the burgeoning high-speed data business, IP telephony may be the application that could make or break the equipment operators have installed in their systems. "When you're on your computer," notes Moore, "and you're not downloading or uploading something, the modem is idle and not tying up bandwidth. But, when you're on the telephone, you're tying up bandwidth whether you're talking or not. Where modems are bursty, telephones are not. They have bandwidth tied up 100 percent of the time. That's the crunch."
The result, says Moore, is that the increasing demand for bandwidth, created by consumers signing up for new services, is putting even more pressure on all the system's components. He thinks the crunch "has not really hit," and that operators will have some real growing pains to experience as customers come on line.
So, can't the manufacturers plan for the increased workload? Not always. Like many equipment manufacturers, Moore says that time is not on anybody's side when it comes to developing new products to meet this demand. The competitive rush never dies, and neither do the demands for better products from operators. He says many companies "roll" their product lines every nine months or so to keep their latest developments out front.
This fast-paced development cycle has replaced a strategy that called for developing a "doomsday" product every two or three years. According to Moore, it's a luxury that most companies can't afford. Says Moore, "If it takes you that long to develop a product, with every bell and whistle you can think of, by the time you look up, the market has probably moved on."
A chip off the old...
The latest in tuner evolution,
Microtune's MicroTuner 2000 chip.
With the advent of cable modems, digital set-tops, and other high-tech equipment for transmission and distribution, there's a new industry mantra. Size DOES matter.
No matter what the broadband application is, it's expected that the integrated circuits (ICs) needed to deliver that application will get smaller and more powerful.
Bill Wall, technical director for subscriber networks at Scientific-Atlanta Inc., says the digital set-top is a perfect case in point. "We've developed two main chips in our current generation set-top that handle most of the processing," says Wall. "We've got a communications processor that does a lot of the MPEG transport, the conditional access, and a lot of the MAC layer stuff for the two-way out-of-band transmission. The second chip is essentially a media processor that includes the MPEG and audio decoders, high-performance graphics capability, and the encoder to NTSC.
"We're looking to shrink the size of those chips and add some capabilities to them as well. I think our ultimate goal is to get to a point where we can put all of that silicon on a single chip." Wall states that an 18- to 24-month time period is probably needed to accomplish that task. "As we get down into 0.18 and 0.15 micron geometries, it makes a lot of sense. It takes about 1.2 million gates to do an interactive set-top. When that gets to a single chip in terms of cost, I think that's when it makes sense to do it."
IC, you C, we all C
Some of the most hectic and imaginative IC development in the industry is taking place in the cable modem marketplace. As the rush to mass market distribution of cable modems gets more frantic, the rush to develop a killer chip set, or preferably a single chip, also increases.
Rich Nelson, director of marketing, cable TV at Broadcom Corporation, likens the drive to single-chip solutions as almost being a force of nature in the IC world, especially when it comes to cable modems. "The (goal) is to have all the components or technology pieces and drive it to a single-chip solution," says Nelson, "which drives down costs. You increase performance because all the communication is on one chip, vs. going off chip or between multiple chips.
"In the near future, we see single-chip cable modems as a product in 1999, as well as single-chip digital set-top boxes. And by single-chip, I mean all of the functions-the transmission, the protocol, the microprocessor-(are) within the box, with the exception of the memory. In the case of the set-top boxes, the MPEG technology really driving single chip solutions is 0.25 micron process technology."
Another tack toward reducing costs and the number of chips in cable modem technology is being aggressively pursued by Libit Signal Processing, which is working on a host-based cable modem solution. "The host-based strategy," says Jacob Tanz, vice president of worldwide sales at Libit, "tries to address the issue of cost by removing as many components off the cable modem (as possible) and migrating them to another platform, which in this case is the PC. We're shifting the problem set.
"We're not changing the functions that need to be there. We're changing the model in terms of where and how those costs are allocated. We think to get some of the prices down, we need to integrate this into the PC's share of processor and memory. We think next year we'll find a $99 cable modem available. As a matter of fact, we're expecting to see manufacturers include it with their PCs as a small upgrade option."
Get with the program
One of the new chip kids on the block made its debut at last year's Western Show. Equator Technologies Inc., in association with Hitachi Ltd., unveiled its MAP (Media Accelerator Processor)1000, which replaces hard-wired imaging solutions with a fully programmable engine on a highly integrated single chip. One of the more attractive attributes for cable operators, says Ted Niday, vice president of marketing at Equator, is the chip's all-format decode (AFD) capability. "The interest in a programmable solution is very high," says Niday. "That's always been the nirvana of a media processor. Our processor can do any format definition coding, down to 480p or 480i."
S-A's Wall concurs. "Programmability is important. Right now, most of the video content we see in cable systems is all MPEG-2, which is pretty easy to build dedicated hardware to. And at this point, it's the most cost-effective solution to have dedicated hardware to do that MPEG processing.
"But, I think as we see more and more content from the Internet and more varied content coming down the pipe that doesn't use MPEG standards, having the ability to program devices to handle different types of media decompression is going to be important."
Philips' Diamond Back amplifier is the first to be
produced on the new design modeling concept.
High-level integration and programmability is the hallmark of another chip player that has just come on the scene. Conexant Systems Inc., formerly Rockwell Semiconductor Systems, has just introduced its CN9414 cable modem chip, which includes a microprocessor and other key functions, as well as a software-upgrade capability that can support DOCSIS 1.1 or make mid-course feature corrections or additions with a software download.
New IC developments have also had a considerable impact on the all-important tuner market. Microtune Inc. has introduced the first silicon-based broadband tuner-receiver, fabricated using IBM's high-performance BiCMOS process technology. The MicroTuner2000 is a high-performance dual-conversion tuner that supports the reception of multiple broadband standards (i.e., QAM for digital cable and VSB for digital TV), while it maintains compatibility with analog NTSC standards.
IC development using gallium arsenide (GaAs) technology is making considerable headway in amplifiers and line extenders. Ron Michels, director of the cable and broadcast business segment at Anadigics Inc., notes that GaAs technology has advantages over standard silicon solutions. These include wide band performance up to 25 to 30 GHz, very low noise figures and very flat frequency response.
"The big driver in the infrastructure market," says Michels, "was not the noise figure. Years ago, people came to us and wanted us to start making amps that were as good or better than what they'd buy from existing silicon suppliers. The driver for that was to keep the same power consumption or (to) lower it, but improve the distortion characteristics of what they (were) buying at the time. Gallium arsenide does that."
Customized solutions?
One of the most interesting component developments taking place is a unique effort being undertaken by Philips Broadband Networks. With the introduction of its new family of 870 MHz amplifiers and line extenders, the company will be introducing its "design modeling" concept, which it hopes to eventually expand to other products.
According to Tony Pierson, group product manager for HFC transport systems at Philips, the idea is to dramatically cut the time to market for new customized configurations that specific operators need. And the capability will not be limited to the top-10 MSOs, either.
--------------------------------------------------------------------------------
'By the Western Show we hope
to have the entire library of circuits
available to use for an entire amplifier.'
--------------------------------------------------------------------------------
"The idea," says Pierson, "is that once all the circuits (of a particular product) are completely modeled and simulated in the computer, then we can sit down with a customer and customize it to meet their particular needs. We would be able to make those changes on the computer and model and simulate what the interactions would be (with their system). We would make any changes in real time on the computer and be able to order boards off that design, bring them in-house and have a pretty high level of confidence that they would be able to be manufactured."
Pierson says they model two factors on the computer to make the concept work. "We're in the process of building up models for two things," says Pierson. "One is the system-level model. The other is the circuit-level model. The circuit-level model would allow you to make changes to a product itself-in this case, an amplifier. The system-level model would allow you to predict the impact of those changes in a system."
At this point, notes Pierson, the business model for this concept, which he terms "just-in-time derivatives" or "just-in-time engineering," works with the larger MSOs and their potentially huge orders. But, he says they're building the system so that "we can do this for an operator who only needs 300 of these things."
The idea behind the concept, says Pierson, is to address two trends the company has identified. One is that an increasing number of manufacturing companies are moving their facilities overseas. This lengthens lead times for standard production. The other is that the company believes that there is still enough uniqueness among the many systems out there that "we don't want to put them in a situation where we can't make changes quickly for them," he says.
If all goes as planned, the first phase of the design modeling concept will be in place by the end of the year. "By the Western Show," predicts Pierson, "we hope to have the entire library of circuits available to use for an entire amplifier. Right now, we have 42 out of 106 circuits that are completely validated. When we develop one, we tweak it. We validate it with the model and go back-and-forth until the model and the actual units are a one-to-one prediction." Pierson expects that the development cycle of a new amplifier will drop from nine months to three months.
Bits and pieces
As the march toward new digital systems progresses, the technological innovations needed to fuel that conversion will undoubtedly continue apace. Faster, high performance components will spur even faster development cycles for newer products.
Above all, flexibility seems to be the operative phrase operators are keying off of in their search for new component solutions. Whether it's programmable chips or customized components, operators want the ability to test, discard or embrace the services that fit their systems best. Given the fact that absolutely no one has a lock on just what kinds of services will eventually be barreling down cable's broadband pipe, that would seem to be the only safe thing to insist on at this point.
|
