The screen capture above shows the unaltered analog NTSC signal. At right, the digital bitstream has been inserted into the signal. Can you locate the difference?

Ask anyone where video is headed and they'll tell you it's going digital — eventually. But even the most bullish digital proponent will tell you it will take years before the hundreds of millions of analog TVs and VCRs are truly obsolete.

So, what should cable operators do in the meantime? Here's one idea: Deliver both. At the same time. In the same channel! That scenario promises to soon become a viable option for cable operators who heretofore haven't been able to make a justifiable economic case for expensive digital set-tops, according to industry veterans-turned-inventors Ted Hartson, Bob Dickinson and Walt Ciciora.

"I think the digital model (where cable operators give up precious bandwidth above 550 MHz and are forced to buy $400 digital set-tops) is broken," says Hartson, who has a long and storied cable career, including a stint as Post-Newsweek Cable's vice president of engineering. "There's a huge number of NTSC sets around, and we think they'll be around a long, long time. People ought to figure out how to use them, not put them out of business."

That is exactly what the cable industry's version of the "Three Amigos" has done. By examining the nature and design of NTSC signals, Hartson and his crew have figured out a way to place up to two digital bitstreams in a standard NTSC signal, without affecting the performance of the existing signal. The upshot? A three-fold increase in the number of channels a cable operator can send over his existing network, all made possible by a headend encoder that is projected to cost a "few hundred dollars" and a decoding chip costing an order of magnitude less, according to Hartson.

Hartson, who calls his Scottsdale, Ariz.-based venture EnCamera Sciences Corp., has a working prototype in his lab, and was expected to file his last patent application in April. The next key step — funding — is also expected to occur shortly, he says.

Unlike other approaches, the EnCamera solution doesn't utilize the vertical blanking interval to "hide" these signals. Instead, it takes advantage of the inefficiencies in the 60-year-old NTSC standard and shoehorns up to 4.5 megabits per second of digital data near the visual and aural carriers.

"There are many compromises in the NTSC standard," says Ciciora. "It was designed in the vacuum tube era. Today, you can do things that were out of the question back then. We do a lot of headend processing, so the data is well-hidden, yet cost-effective to extract."

The EnCamera system delivers a high-speed digital stream (4.5 Mbps over a cable system; up to 3 Mbps over the less-robust broadcast airwaves) that can be used for virtually any application, whether it be video, data or voice.

The additional capacity over cable networks is made possible because unlike the broadcast environment, cable doesn't suffer from co-channel interference, reflections and myriad other problems that can plague over-the-air transmission, Ciciora says. That being said, however, Ciciora says the EnCamera technology would still be appealing to broadcasters who might want to add channel capacity or send data within their analog signals.

Ciciora, who has championed numerous agreements with the consumer electronics manufacturers in an attempt to improve the compatibility between the two industries, also says the technology could be used by cable operators to perform what he calls the "compatible digital upgrade" that allows an MSO to keep all his bandwidth for analog signals and simply add a digital tier within it.

Of course, the approach requires that cable subscribers be given some sort of digital decoder. Hartson and company hope to attract the attention of General Instrument and Scientific-Atlanta and convince them that the additional circuitry should be included in new set-tops. Here, they'll probably be faced with the chicken-and-egg conundrum — unless a cable operator decides to adopt the approach and create market pull.

With the contacts these three have, that's not out of the question. "Everyone we've shown it to is intrigued," Hartson says. Ciciora estimates that a fully integrated circuit could be added to a set-top for about $20 in volume numbers, but also notes that other devices, such as DVD (digital versatile disk) players would be another likely candidate for this technology.

All of this is nice, but isn't it already too late? Even the Federal Communications Commission has ordered all the broadcasters to use digital technology and abandon their analog feeds by 2006. Ciciora, for one, is skeptical. "We're not all going digital," he argues. "Analog (receivers) will be around for a long time. It's only the affluent who will be able to afford the first digital receivers, and even they will have analog sets in their other rooms."

Should this technology be considered a breakthrough? None of the inventors was willing to go that far, but they are quick to point out that there is some true innovation here. "No one had ever approached the data rates we have achieved," says Dickinson, who also owns Dovetail Systems, a flyover company. "We've done a lot of original work and stretched the boundaries. There's a lot of originality and achievement here."

Ciciora, the former Time Warner Cable engineering and Zenith Electronics executive, agrees. "We haven't invented something as significant as a transistor, but we took things we knew how to do and optimized it to the nth degree."

Hartson says he's now preparing to make some of the demonstration equipment portable, so he can take his invention on the road to demonstrate it. He predicts that a limited field trial could occur as early as this autumn. "We think the business opportunity is great," he notes. "The worldwide possibilities are immense, because we're getting to the point where there's no spectrum left."


BER benchmarking

As activation and utilization of the return network in the cable TV industry continue to increase, it has become increasingly clear that there is no accurate way to "benchmark" system or component performance.

Don Gall, communications consultant with Pangrac & Associates Consultants, believes that bit error rate (BER) is to digital services as CNR, CSO and CTB are to analog cable TV. BER, he says, needs to be the first measurement benchmark of the HFC return network.

To establish a BER benchmark, Gall says a certain frame of reference has to be agreed upon. That reference model includes what many believe is the dominant digital modulation technique, QPSK (Quadrature Phase Shift Keying). Gall notes the modulation's robust nature provides a reasonable theoretical efficiency of two bits per Hertz.

He says it makes sense to standardize on a bit rate for individual carriers in the return band. Because of widespread acceptance, he believes 1.544 Mbps and 2.048 Mbps (the respective American DS-1 and European E-1 telephony standards) are acceptable rates. And while some may argue that the segment below 10 MHz in the return might require a more robust type of modulation, Gall believes that for this BER benchmark exercise, it's assumed the entire return bandwidth is suitable for QPSK modulation.

With those assumptions in mind, Gall believes the following steps will allow for the determination of the dynamic range of each device in the network, or the entire network as a whole.

This is done by setting up the device(s) under test to the approximate center of its operating range with a full load of 22 E-1 carriers, plus one DS-1. The next step is to record the BER, and associated power levels. Any error correction should be disabled. The overall operating levels are then raised and lowered by small increments, and the BER is recorded at each step until the BER reaches sync loss.


The lower curve is the noise limit of the system, including the modem's efficiency. The upper curve is determined by the system's distortion performance. The BER data is plotted against RF output (or input) level, and a graph is generated to show the results (see figure above). The dynamic range is demonstrated as the difference between the two curves at an acceptable BER (10-6, as seen in the figure).

While the call for a precise benchmark is not new, he notes that any accepted benchmark must be portable between vendors and must also reflect the type of communications medium that will be used.