This month I'll focus on some of the new technologies on the horizon for TV broadcasters. Digital Signal Processing (DSP) is now common at audio frequencies. Chip makers are currently working on circuits that work at video frequencies and even at radio frequencies for use in new digital cellular and PCS wireless devices. Read on for an introduction to DSP and how it will affect TV transmission. Many new TV transmitter products are introduced every year at National Association of Broadcasters' Convention. While I won't have any inside reports on what might show up, I'll share a few ideas on what I'll be looking for.
Amateur radio magazines have been carrying articles on and ads for Digital Signal Processing ("DSP" from here on) units for several years. The most interesting implementation is a filter algorithm that removes repetitive noise at audio frequencies from a shortwave signal. While RF phase cancellation circuits can perform similar tasks using only analog signals, they are much more difficult to install and adjust. These ham radio DSP boxes not only reduce noise, they also act as almost textbook perfect audio filters. Analog filter circuits require many more components and many more adjustments to do the same thing. Furthermore, once the analog circuits become this complex it is difficult to make them flexible for different filter functions. These examples illustrate the attractiveness of DSP - it offers increased functionality along with lower cost, high volume manufacturing.
The growth in digital wireless personal communications (cellular and PCS) is pushing manufacturers to reduce cost while at the same time increasing the complexity of the hardware to maximize the capacity of the scarce spectrum. It's not surprising that manufacturers are looking for ways to make DSP work at higher frequencies.
While many manufacturers have ignored the TV broadcast market, one, Oren Semiconducter, hasn;t. It should have a DSP chip for receiver ghost canceling available soon. Phillips plans to use it in set top boxes and offer it in Magnavox TV sets. The OR43100 ghost cancelling chip is expected to cost around $25 in volume and is flexible enough to accommodate all current world wide ghost canceling signals. The chip is reported to be able to cancel out ghosts arriving up to 7.5 microseconds early and up to 42.5 microseconds late within half a second. At the Western Cable Show, a Phillips representative told me we should see this chip in their Magnavox products by May. A Phillips engineer working on the project asked me to urge networks to transmit the ghost cancellation reference signal (GCRS) with their program stream so that correctors can compensate for ghosts (and, as a side benefit, frequency response variations) that occur anywhere in the path to the consumer, whether from the TV station, microwave or cable company.
Oren's ghost canceling chip will be a nice option to improve reception of conventional over the air NTSC signals. It or similar chips will be a necessity in over the air digital TV systems. During the last ATV field tests in North Carolina one weakness observed in the Grand Alliance system was its susceptibility to moving ghosts (multipath). This often made reception on indoor antennas difficult. More powerful ghost canceling hardware and software should significantly improve the GA system's performance with inferior antennas.
There are other applications for video frequency DSP's in transmission systems. DSP's are being developed for video compression and video processing. The video compression chips are useful for digital satellite video transmission and offer the potential of being able to be reprogrammed as compression standards change. I've not seen any broadcast video processing units using DSP's, although in the future they could provide video precorrection functions to compensate for TV transmitter non-linearities.
At TV intermediate frequencies (IF) (under 50 MHz.), DSP's could provide the basis for an auto-correcting TV transmitter exciter. To be effective, a non-linearity corrector should work at IF frequencies. Otherwise, it cannot compensate for sideband sensitive intermodulation products generated in the VSB (Vestigal Sideband) RF system. By monitoring the RF spectrum and vertical interval test signals from the output of the transmitter, an IF DSP chip set could provide real time compensation for system non-linearities. With today's analog TV transmission, improved linearity means an improved picture. With tomorrow's digital TV transmission, improved linearity will mean improved coverage through reduction of the system bit error rate. Analog linearity correction, as exemplified by Comark's UHF exciter for common amplification transmitters, has improved substantially over the last few years. However, that improvement has come at the cost of an increasely difficult and complex adjustment procedure. It will be interesting to see which TV transmitter manufacturer is the first to successfully build a DSP based automatic linearity corrector.
How high in frequency can DSP chips operate? The current ceiling for widely available chips is around 50 MHz. Work is being done to use digital mixers to convert frequencies in the UHF range down to lower frequencies for filtering and more elaborate processing. The reason most processing isn't done at the UHF frequencies becomes clear when we look at how DSP chips work.
RF is analog. No matter how much the world seems to want everything to be digital, radio frequency signals, transmitted over the air, will remain analog. Even if they could start out as neatly defined digital square waves, as soon as they leave the antenna and venture out into the ether (an old term for the space in which electro-magnetic waves propagate), they will exhibit analog properties. Therefore, the first stage of any DSP based RF product is an analog to digital converter to convert the RF signal into a digital word. Once the signal is digital, is can be applied to a digital mixer driven with a digital local oscillator. A decimating low pass filter converts the output of the digital mixer to a low enough frequency range (data rate) to be detected by a digital signal processor - the DSP. If the product's output has to be analog - to drive a speaker or CRT, the DSP's output is then applied to digital to analog converter to bring it back to the analog world.
From a block diagram, the typical DSP circuit looks like a conventional microprocessor. There are various storage registers for instructions and data, one or more arithmetic processing units (ALU) to do math and program logic circuits such as address generators to tie the pieces together. A closer investigation shows some differences. DSP chips are designed for real time processing. Viewers won't tolerate an hour-glass icon appearing on their screen when they change channels or the scene changes. To make real time processing possible, the DSP chip operates with more or wider data busses and often includes FIFO (first in first out) memory to cache data for processing. As much work as possible must be done in one clock cycle.
How does DSP work? Analog signal characteristics have been defined by mathematical formulas for decades. The DSP uses these formulas to duplicate the results of analog circuits. A simple RC (resistor - capacitor) low pass filter can be easily defined using the formula for capacitive reactance. Duplicating the filter's response by doing math on the incoming signal is easy. Furthermore, it isn't necessary to consider real world factors like stray capacitance and inductance in the resistor and loss in the capacitor. By using the formulas for FFT (Fast Fourier Transform) to convert a complex signal into its sine wave components, complex demodulation functions are possible. DSP units will be a key part of low cost COFDM (Coded Orthogonal Frequency Division Multiplex) receivers. Recursive filtering algorythms are used for noise and ghost reduction DSP programs.
DSP has its drawbacks. Poor dynamic range in RF A/D converters continues to limit the use of DSP's for replacing the front end of receiver. As anyone who's hooked a spectrum analyzer up to an antenna knows, variations in incoming signal levels can exceed 100 dB. Eight and twelve bit A/D's can't handle that range. Look for more coverage of DSP technology relevant to TV transmission (satellite, over the air or cable) in future columns and current news in the RF Current section of my web site at http://www.transmitter.com.
I'll be looking for DSP technology integrated into TV transmission and reception systems at this year's NAB Convention. Phillips/Magnavox has had their ghost cancelling demonstration outside the technical session area for what seems like the past decade. If they are back again this year there should be more interest as the first receivers and set top boxes hit the market with built in ghost cancelling DSP's. I'll also be searching the booths of the more innovative video equipment manufacturers to see if they have any DSP offerings. Likely candidates include the Tektronix/Grass Valley Group combination and Snell & Wilcox, both companies with strong backgrounds in digital video signalprocessing.
The big surprise at the Western Cable Show in Anaheim California the end of November was the tremendous interest in digital data over cable. I'm not referring to the typical coarse text program guides and shopping services but real, multimedia Internet and on-line service offerings. Grand Alliance members Zenith and General Instrument, to name two, were showing cable boxes for accessing the Internet and services like America On-line, Compuserve and Prodigy all through cable hookups.
One group had a box that also worked with over the air broadcast. Intercast - a group comprised of Intel and several broadcast and cable media giants, was featured in TV Technology front page story a few months ago. Data is transmitted in the video signal's vertical blanking interval. All the demonstrations I've seen used Norpak Corporation's VBI encoding system. This system is capable of transmitting data at speeds up to 115,200 bps, although practical implementations with heavy forward error correction and a limited number of vertical interval lines are likely to run closer to 57,000 bps. The Intercast system works by continously transmitting web pages which are then cached on a local hard drive. The web browser (Netscape Navigator was used at the Western Show) is able to retrieve popular web pages from the cache immediately. Non-cached web pages require a request through a phone line, cable modem or, perhaps, even a wireless modem. The requested data is sent either over the air or over the same circuit that requested the data.
I expect data broadcasting to be a major topic at this year's NAB. Although several different approaches exist for sending the data, the Intel / Intercast group seems to have done the best job soliciting industry support. I'll be looking to see how much of this technology can be realistically adapted for distribution on over the air TV signals. New DSP based products like the Oren ghost canceling chip could enhance the performance of VBI in harsh, real word multipath environments.
I'd hoped to see some evidence of the silicon carbide transistor based high power UHF amplifiers at this year's NAB, although that isn't as likely now that Westinghouse, the developer of the silicon carbide transistor, has announced it is selling its defense electronics division to Northrup/Grumman. The silicon carbide transistor seems to have gotten caught in the middle of this sale. I've heard the Westinghouse will keep the team that developed the silicon carbide transistor, but will lose the unit that was responsible for systems integration -- the practical implementation of the unit. Last reports were that Westinghouse had developed a 1.5 KW peak power amplifer module using the silicon carbide transistors. While not suitable for the widely varying peak to average ratio of NTSC transmission, it would have worked well for the 8-VSB digital transmission. You might want to ask one of the major UHF transmitter manufacturers what's happening with these devices.
If the U.S. Congress doesn't abort it first, the nationwide transition to digital TV over a five to ten year period will create the largest demand ever for UHF TV transmission equipment and antennas. As a result, I expect to see more European companies like Itelco and others targeting the U.S. high power UHF transmitter market. I'll be checking out their designs and comparing them with innovative technologies like the single tube Diacrode transmitter from Acrodyne and the PS-Squared HV power supply for Klystrode and IOT transmitters from Comark Communications. Will Harris Corporation have something new in their TV transmitter line this year? Larcan has made a considerable investment in TTC. This year we should see some of the engineering changes resulting from the merger into Larcan-TTC. Look for I.T.S. to enter the high power transmitter market with a new solid state VHF transmitter, externally diplexed. Built on 1200 watt modules, transmitters with peak visual powers up to 60 KW will be offered.
I'm also keeping an eye on video compression technology for satellite transmission. The General Instrument (GI) Digicipher II product has a lot of features that appeal to broadcasters. One nice feature is the ability to transmit vertical interval test signals (VITS) with minimal impact on data capacity. Because VITS don't usually change, the Digicipher II encoder can spread the data from the VITS lines over many frames of video. This reduces the data stream required for the VITS to an insignificant proportion of the total data package. What remains to be seen is how this and other digital compression systems will handle digital data in the VBI. I'll be comparing systems from GI, Scientific Atlanta, Divicom, CLI, NTL, Comstream and others to see if any of the systems are interoperable with each other. So far the answer seems to be "no".
Changes in ENG (Electronic News Gathering) microwave systems have been more evolutionary than revolutionary. While there have been significant improvements in low noise preamplifiers, sophisticated dish control / aiming systems and better filtering, the basic technology has remained the same. U.S. ENG spectrum in the 2 GHz. band is coming under pressure from PCS and other services and this may eventually lead ENG users in congested markets to consider digital video compression to squeeze their signals into 9 or 4.5 MHz. bandwidths. This is already starting to happen for SNG (Satellite News Gathering). New chip sets and more competition are reducing the cost of digital video compression. Companies like Sony, not usually associated with RF and microwaves, are planning compression products for transmission as part of their digital video recorder offerings. I'll be looking around for relatively inexpensive compression gear for ENG quality transmission not only on the main floor but in the multimedia area as well. Given the ratings pull of live breaking news coverage I believe the price of compression gear won't have to drop too far below its current level to attract interest.
That's it for this month. The reply comment period in the FCC latest Notice of Proposed Rulemaking for Advanced TV closed in January. Next month I'll examine the prospects for Low Power TV stations and translators in the age of digital TV. While the FCC ignored LPTV in its recent Notice of Proposed Rulemaking (except in its mandatory notice of impact on small businesses), commenters didn't. The Community Broadcasters' Association came up with an idea for spectrum allocation that doesn't interfere with MSTV's proposeallocation model. It is worth considering. Several manufacturers responded to my call for information on broadcast TV antennas. I'll review whats available and show how panels are combined to create different radiation patterns. I'll also have a report my first I.O.T. common-amplification transmitter - a four tube Comark IOX at KVEA in Los Angeles.
For current news on topics affecting TV and satellite broadcasters, refer to the RF Current pages at my web site at http://www.transmitter.com . RF Current offers a weekly (updated throughout the week) summary of news affecting TV broadcasting and transmission (including satellite) with links to the full text of FCC actions, company news and press releases, technology news and Congressional speeches. Back issues of my TV Technology RF Columns through 1993 and previews of topics for future columns are available. While I'm no longer able to send copies of programs I've written in connection with the RF column on disk, the latest editions are available at http://www.transmitter.com. Select the link to my FTP site or go directly to ftp.transmitter.com/pub/.
You can contact me at dlung@transmitter.com. You may also fax me at 305-884-9661 or phone me after 6 PM eastern time (when things quiet down a bit) at 305-884-9664. Both numbers are at the Miami Telemundo office, so expect a delay in a response if I'm traveling. My mail service address is 2265 Westwood Blvd., Suite 553, Los Angeles, CA 90064. Because I'm often traveling, if time is critical (response needed in less then ten weeks) please contact me for a local address before sending items by mail. Your comments are always welcome!
Return to Doug Lung's TV Technology RF Columns 1996
or The RF Page @ www.transmitter.com
Copyright ©1996, 1997 H. Douglas Lung. All rights reserved