A Quarterly Publication of
The Smith-Kettlewell Eye Research Institute’s
Rehabilitation Engineering Research Center
William Gerrey, Editor
Original support provided by:
The Smith-Kettlewell Eye Research Institute
and the National Institute on Disability and Rehabilitation Research
Note: This archive is provided as a historical resource. Details regarding products, suppliers, and other contact information are original and may be outdated.
Questions about this archive can be sent to
TABLE OF CONTENTS
Recently, a substantial number of "TV movies" have been transmitted in a special format for the visually impaired--they bear description of the video action. Transmission to the user has been by any of three methods: via a radio reading service, via a subcarrier called "SAP" (second audio program), or just superimposed on the main open-channel audio. In some ways, this is analogous to programming which is specially formatted for the hearing impaired--sent with captions (text) that appear on the screen. How these programs are created, how they might be handled by network facilities, and where to find them, are subjects treated here. Technical information on the "second audio program" subcarrier is given in the following article, "Multichannel Television Sound."
[Much credit goes to Dr. Margaret Pfanstiehl, President of the radio reading service called "The Metropolitan Washington Ear." You may have heard descriptions of public television programs over your radio reading service nearly ten years ago--those were due to Dr. Pfanstiehl, and we have her to thank for tireless promotion of this idea over the years.]
Two reports were commissioned by the Office of Special Education and Rehabilitative Services regarding "Descriptive Video Services" (DVS, a "service mark" of WGBH). A report on "Commercial Viability" was done by Cosmos Corporation in Washington, D.C., in May, 1990. The Rehabilitation Engineering Center at The Smith-Kettlewell Eye Research Institute wrote a companion report called "Technical Viability of Descriptive Video Services"; much of that technical information is condensed and presented here.
Reprints of these reports are available. Technical Viability of Descriptive Video Services" is available from us for a fee of $5. Commercial Viability of Descriptive Video Services is available from Cosmos Corporation, 1735 Eye St., N.W., Suite 613, Washington, D.C. 20006; Phone: (202) 728-3939.
I reckon this has been a dream of all of us--to have a description of silent action available to us when audio cues do not tell the story. Those of us with sighted family members handy have gotten bits of description along the way; my father, who often recorded TV programs, provided Mom with a microphone and recorded her description on a separate track on a stereo tape. Still, this did not give us "independent access" to television.
Aside from getting the action described, problems of how to couple the description with the video program and how to deliver it to the consumer have previously placed limitations on any such dream. Recently, however, technical advancements have "made room" for such services.
Summary of Technologies Supporting Video Description
Before 1975, Network audio feeds used expensive phone lines to deliver the audio portion of their programs to network affiliates. These phone lines have been entirely supplanted by satellite communications. For now, there is plenty of bandwidth available on each "transponder" to send extra audio channels to affiliates; these include technical information about the programs, promotional announcements, and may include our description track.
* * *
There is now a commercially available way of getting a separate sound track on the consumer's receiver. Called the "SAP channel," this is a subcarrier which rides on the FM audio signal of the TV transmission, and many television sets (mostly stereo ones) have a decoder of this subcarrier. Integrated circuits make decoders inexpensive and low in parts count.
* * *
Modern microwave links to TV transmitters accommodate plug-in modules for additional audio channels. Therefore, extra audio phone lines or microwave transmitters are not needed to accommodate the description.
* * *
Modern tape machines--video and audio--can be synchronized by "time codes"; this permits a taped description to be precisely linked with the video program. Furthermore, a description which has been made during an early stage in the program's creation can be automatically linked to the finalized version.
For example, the description may be done using a "working copy" of a program which, when it is finally packaged, will contain added breaks for announcements or a last-minute insertion of theme music. Computers in these machines permit a technician with a list of the changes to simply declare, "From frame number 1260 onward, time code A equals time code B."
* * *
Computerized video editing stations are becoming commonplace. One feature of such stations is that they allow the video technician to make very detailed "logs" (tape labels) describing what is on each segment of tape; a detailed "label" can be printed, each bit of detail being accompanied by the corresponding frame number and time.
This tape-log feature can directly be used for writing and printing the script of the video description. A primitive word processor--part of all these computerized editing stations--allows the description to be revised at any time. The resultant "script" can either be read by the narrator off a teleprompter, or read from printed pages. Timing information is shown adjacent to each phrase to cue the narrator (in addition to cues being written into the script).
Creating the Script
The Low-Tech Implementation
For about $50, you can take a video cassette to a processing studio who will make a "burned-in window dub"; this is a copy of your tape which shows timing information (hours, minutes, and seconds), along with frame numbers, all appearing in a bar across the screen. Portions of the tape that require description are then identified and the timing information noted.
Phrases of the description can then be written to fit the designated time windows; subtracting the starting time from a stopping place will tell you precisely how long a bit of description can be. A "time-code calculator" (a hand-held calculator used by motion-picture people) can aid in doing this; such a calculator costs under $100.
A description script can be typed on a typewriter, typing the timing information at the margin for the various statements.
Next, a single-track quarter-inch audio tape of the narration is made. Marked with colored splicing tape or other methods, this tape, the printed script, and the original video tape, are taken to a "sweet-track" studio; this kind of facility is capable of mixing and processing audio tracks of videos to make improvements on them. Using the printed times and frame numbers, the audio tape is cued and recorded so as to create a mixed audio track containing the program audio and the bits of description.
This mixture is "laid back" (recorded in place of) the audio track on the video tape to be enjoyed. (A video tape may have an extra track to hold the described program, an issue which will be discussed later.)
The sweet-track studio work may cost a few hundred dollars, the cost depending heavily on whether you hire a technician or run it yourself. Both options are available in many facilities.
This low-tech system is prone to error; timing information can be mistyped, for example. If phrasing of a statement by the narrator is awkward, it cannot be easily changed. It is inflexible; no subsequent program changes can be allowed, or the script and its timing will be incorrect. Given the advantages of computer-aided writing, this system is bound to be slow in comparison.
A Customized High-Tech Implementation
As described earlier, a modern computerized editing station can provide the writer with a primitive word processor, automatic notation of timing information, and a printed script. The main disadvantage of this is that it is a simplistic use of a very expensive editing station, one which represents a capital expenditure of tens of thousands of dollars. (Ready-made editing stations have lots of equipment, two or more tape machines, video processors, and some can even store video and audio in dynamic RAM.) The alternative is to marry a synchronous video tape machine with a personal computer, and to write software of specific benefit to creating a description.
WGBH in Boston, an affiliate of the Public Broadcasting Service, has created a station using a Mac 2 Apple computer and a video tape player and controller. Starting with the Macintosh database "Hypercard," Dan Glisson of WGBH has assembled a setup which they call their "Descriptor."
The video tape is played to identify portions to be described; the beginning and end of each silent passage where the description might fit is marked by the push of a button, and the computer remembers all these marked passages. For each passage, the computer shows the writer how much time he has for a bit of text to fit.
A speech rate for the narrator is selected; they usually select 180 words per minute. After the bit of description has been written, the computer tells the writer whether or not it will fit into the silent passage designated. If not, the computer will suggest a speech rate which would allow that text to fit, and if the writer likes, the Macintalk speech synthesizer will read the script at the new rate to show him what this would sound like. The writer then has the option of rewriting that bit of description, marking the printed script with the notation "fast," or changing the time slot.
Of course, the flexibility of re-sorting timing information as program changes are made is provided for.
"Narration to Picture"
Any well-equipped station or video facility has a narration booth on the premises. At WGBH, engineers doing the mixing of the narrator's voice with the program audio also have the option of cuing him visually or through headphones. They prefer to do the narration concurrent with running of the program; this process is called "narration to picture," and this is how documentaries are narrated.
Having the narrator read the phrases against an actual running of the program promotes smooth phrasing and assures that the bits of description properly fit the spaces picked out for them. Mixing of the voice and program audio are done in real time by the engineers.
Fitting the Program Video Tape with the Description Track
Professional video tape machines provide an extra audio track for the "audio time code" (ATC). Luckily, there is an equivalent video presentation of the time code, the "vertical-interval time code" (VITC). One trick is to use the ATC track to carry the program audio mixed with the description.
Both time code schemes use a standard code decided upon by the Society of Motion-Picture and Television Engineers; and it is formally referred to as the "SMPTE Time Code." In conversation, SMPTE is pronounced "simpty"; the "simpty Time Code." On the audio track, the SMPTE Time Code sounds like telemetry from a MODEM. The video equivalent does not appear on the screen; it is impressed on horizontal sweeps which occur during the vertical retrace interval.
During one of the copying processes, the audio SMPTE Time Code is fed into an encoder box which inserts a VITC into the video signal, and this makes a free audio track on the tape being made. A decoder box can retrieve this VITC signal; the decoder can either display timing information on a monitor, or the decoder can produce the audio version of the SMPTE Time Code.
The newest video tape machines have tracks to spare. For example, the Sony "BetaCam" system provides two high-fidelity audio channels using extra bandwidth on the video tracks (there being two, "chrominance" and "luminance"). Two regular audio tracks besides the high-fidelity ones are also provided on the Sony tape format. The "D1" system, whose video is broken down into red, green, blue, and sync (RGB and sync), have extra audio channels as well.
Transmitting Digital Data During the Vertical Interval
Now this may come as a shock to you, but out of the advertised 525 "lines" in our American system of TV video, 42 horizontal sweeps occur during a vertical-sweep's blanking signal while the vertical sweep generator is "retracing." These unused lines can be employed to carry information. The SMPTE VITC (vertical-interval time code) is sent on lines numbered 16A and 18A (the first pass of the 16th and 18th horizontal sweeps).
Many of those lines have been allocated to commercial services who, at one time, were convinced that "teletext" would be a part of everybody's TV viewing. In one important case, this has actually been implemented--"closed-caption" services for the hearing impaired. The "teletext" of captioning for the deaf is put on line 21--using both passes, A and B. A decoder on the deaf consumer's TV decodes line 21, then creates a text image which is then presented on the screen. (Congress has just passed a law stipulating that by 1992, television sets must include this caption-decoding chip in their circuitry.)
At a March 8, 1990 meeting of network executives with Cosmos Corporation and this writer, NBC participants reported that, for the Korean Olympic Games, engineers prototyped a system whereby an audio channel was created using three lines of video in the vertical interval. While this was not actually used during the Games, the principle was demonstrated. This may come in handy in solving some of the audio-routing problems, as will be encountered in the section called "Stumbling Blocks."
Satellite Transmission Synchronous orbital repeaters (actually direct frequency translators called "transponders") are in common use in the distribution of network programming. Two bands are used: Transponders in the "C band" take signals sent to them at 6 gigahertz and translate them down to 3 gigahertz. Transponders in the "KU band" receive signals at 14 gigahertz and translate them down to 11 gigahertz.
Early transponders had a bandwidth of 56 megahertz. Nowadays, as bandwidth becomes a more valuable commodity and transponders improve (cross-talk always being a point of discussion), so-called half-transponders are employed--each half allowing a bandwidth of 27 megahertz.
Still, there is plenty of room, for now. A television signal may occupy a slot with a possible bandwidth of 7.5 megahertz. The video signal sent to the transponder is FM, not AM as is used in the TV bands. The video signal may take up 5.5mHz, with typical audio channels being placed above this at 5.8, 6.2, and 6.8mHz. (These figures in megahertz refer to what the layout of the baseband would be once you translate the whole mess out of the gigahertz range.)
You can see the possibilities for carrying extra information over the satellite transponders. Crowding more and more into the spectrum will eventually necessitate reducing the deviation of adjacent channels (degrading their signal-to-noise ratio slightly), but satellite brokers have assured me that there is no such threat posed in the immediate future.
Typically, up-link and down-link equipment only requires the installation of low-cost modules to add audio channels.
Network-Affiliated Station Configurations
Once the audio channel containing the description has been retrieved from the satellite down-link, it must be patched through to what is known as the "SAP generator" at the TV transmitter. Unfortunately, extra audio channels are not just lying around in the station's signal routing system. However, a separate audio console can be dedicated to this purpose, taking the new audio information from the down-link and feeding this to the microwave link that feeds the transmitter.
Two configurations of microwave link are in common use: One produces the composite TV signal similar to what we receive and sends that to the transmitter, which merely amplifies it; i.e., the microwave receiving equipment is like an exciter. The other type handles signals in a format much like that used over satellite, assembling the composite signal at the transmitter site.
The first type would logically have the SAP generator next to the audio console which feeds it. The other microwave link-up system requires plug-in modules, one at either end, to create an extra audio channel--quite similar to modules required in the satellite equipment. This latter system requires that the SAP generator be at the transmitter site. In either case, the hookup from satellite down-link to a SAP generator is straightforward.
The Network-Facility Problem
Network handling of additional audio signals within their facilities--before the satellite up-link--is not at all convenient. They are only recently expanding their equipment to accommodate stereo, and a third audio channel running throughout is not being planned for. (A third channel can be arranged for as a temporary setup, but they are not at all fond of doing so.)
The magnitude of upgrading to include other audio wiring can be illustrated by the newest ABC New York installation. All cabling is routed in rooms under a subfloor of the facility; it is such a maze down there that signs saying "THIS WAY OUT" have been installed to guide technicians out of that malaise.
Running special setups makes the networks very nervous. The flip of a wrong switch causing the loss of a few seconds of programming translates in to million-dollar losses. At the author's meeting of March 8th, network people quoted a figure of $10 or $20 million to properly upgrade so that they would feel comfortable with handling an extra signal for each program.
No such problem exists with closed-captioning for the deaf. Encoders to insert the captions into the video signal require no special signal routing, and the network investment to set up that system was in the hundreds of thousands of dollars, not in the millions. A possible solution for described video is to take the special audio out of the audio signal-routing equipment and cause it to be sent in the video interval--a realizable concept.
On the other hand, future television equipment will be computer controlled. Already, companies are advertising audio consoles whose "setups" can be saved on floppy disks. Eventually, a television program, possibly marked with a digital code and containing an extra description track, might find its own way through the system automatically.
The Cable Problem
There are two types of systems employed by cable service providers: One is the block-converter type; a TV signal, containing whatever it contains, is just retransmitted (translated) to the frequency sent to the consumer's cable box. For nearly a decade, however, cable providers became interested in systems which would allow video enhancement to be performed at their facilities, or would allow scrambling of video signals in special ways. This latter type converts a TV signal to baseband, fiddles with it, and retransmits the modified signal on the chosen cable frequency.
Equipment of the block-converter can generally pass stereo sound and the SAP subcarrier. Those installations which process the signal in the baseband often don't pass extra subcarriers, and often eliminate data in the vertical interval as well.
(Ironically, old equipment (pre 1975) is of the block-converter type, and the multichannel TV signal will make it through just fine.) Richard Oehm assures me that equipment in new cable installations is, for the most part, block-conversion stuff, so we can look forward to getting stereo TV and the SAP subcarrier via cable in the future. In the meantime, however, there are a great many cable providers who have invested in consumer boxes that unscramble their video, and these may have to wear out before we can hear our special service on those cables.
Limitations of Old Transmitting Equipment
As will be discussed in the following article, "Multichannel Television Sound," phase distortion on the video adversely affects the sound quality of the SAP subcarrier. This phase distortion is a chronic problem with old TV transmitters (sort of like FMing of the video carrier when high-amplitude AM video information is present). Therefore, older stations are reluctant to install SAP generators. Working on our side is that transmitters of that vintage are getting expensive to maintain--replacement vacuum tubes are increasingly expensive, and their quality is suffering from low-volume production. It won't be long before those transmitters are phased out.
A stickier upgrading situation exists with antenna traps which damage the required high-bandwidth audio signal. As will be seen in the next article, the allowable deviation of the audio portion of the TV signal had to be increased from 25kHz to 75kHz. Often, the wider frequency excursions run into slopes in the response curves of older antenna traps.
Nowadays, video and audio signals are kept isolated by multisection band-pass filters ("combiners") in the antenna system. Older systems used a device called a "notch diplexer," a system which used notch filters to separate the signals. As designed, these can't handle the expanded deviation, and they are expensive to replace (perhaps $50,000). The Harris Corporation, a leading manufacturer of TV transmitters, has patented a modulation system which distorts the transmitted audio signal so as to correct for problems introduced by notch diplexers.
The upshot is, old equipment presents problems to multichannel television sound, problems which will evaporate over time.
Sharing the SAP Subcarrier With Other Services
When it was invented, the SAP subcarrier was promoted using foreign language translations of programs as an exemplary use. In fact, some early TV sets call the SAP switch "Bilingual." As SAP comes into common use, services will compete for SAP programming time.
The Descriptive Video Services staff at WGBH proposes that, if there are multiple showings of a program, half of them could be in another language, and half could bear description. Or, the audio of a described program could be broadcast during off hours.
These are issues that do not concern the hearing impaired users of captions, since no one else is going after their line 21 of video. Perhaps if the described video came over video lines in the vertical interval, we would be more immune to competing services as well, although wherever there is a data channel, somebody else may come along to share it someday.
The strongest argument for using the SAP subcarrier is that receiving equipment abounds. Any other broadcasting scheme would necessitate a special decoder. (Of course, this special decoder could be installed to feed the SAP generator at any station which chooses to use SAP for description.)
Five service providers are busily describing programs. They are briefly discussed as follows:
The AudioVision Institute
The Creative Arts Department of San Francisco State University has a course in describing movies, plays, and video programs. Headed by Gregory Frazier, a project to create accessible programs has grown into The AudioVision Institute, San Francisco State University, 1600 Holloway Ave., Room C.A., Suite 351, San Francisco, CA 94132; Phone: (415) 338-2993.
"Descriptive Video Services" as a Division of WGBH, Boston
Many public television stations are now broadcasting programs (using the SAP subcarrier) which have been described by the folks at WGBH. They have a newsletter which lists all participating stations as well as new program releases. Address your inquiries to Ms. Sharon King, Outreach Director, Descriptive Video Services, WGBH, 125 Western Ave., Boston, MA 02134; Phone: (617) 492-2777, extension 3490.
The "Narrative Television Network" (NTN)
Currently describing four hours of programming per week, and expanding to six hours in March 1991, this organization, headed by Mr. Jim Stoval in Tulsa, boasts that 800 cable service providers are handling their programs. They put the description on the open channel; everybody who tunes in hears the mixture of program audio and description. The feedback from viewers is all positive. The vast majority of his audience is sighted. Nevertheless, people write in to express their enjoyment, not to complain.
To find out if these programs come over your cable service, call or write to: The Narrative Television Network, 5840 S. Memorial Dr., Suite 312, Tulsa, OK 74145; Phone: (918) 627-1000.
Video Description a la Pfanstiehl
The radio reading service located in the Washington, D.C. area has been an early promoter of "video description." Video productions for the National Park Service have been described by them. In addition, Dr. Pfanstiehl has overseen the training of describers at WGBH. She invites you to write or call: Dr. Margaret Phanstiehl, President of The Metropolitan Washington Ear, 35 University Blvd. E., Silverspring, MD 20901; Phone: (301) 681-6636.
Video Tapes Bearing Description a la Hecht
A fine old movie maker, and a tireless promoter of video description, has personally overseen descriptions of fine old shows like "Show Boat," etc. He reasons that if network distribution poses a problem, visually impaired people can get what we need through the sale and rental of specially prepared VCR tapes. To find out more, write to: Mr. Albert Hecht, 24 Hutton Ave., No. 26, West Orange, NJ 07052.
Rather than loading you up with references, let me steer you to the most concise one, "Multichannel Television Sound," by Edmund A. Williams, Staff Engineer, Dept. of Science and Technology, National Association of Broadcasters, Washington, D.C. If you want more, write to me, or send $5 for our 39-page report, Technical Viability of Descriptive Video Services. The National Technology Center of the American Foundation for the Blind compiled a list of stereo TVs, VCRs, and decoders able to receive SAP, and this list can be had free of charge by writing to them at 15 West 16th St., New York, NY 10011.
It all started when the Federal Communications Commission (FCC) was petitioned by a local station to include an audio channel which could be used for transmitting cuing information to remote news crews. The Broadcast Television Sound Committee (BTSC) of the Electronics Industries Association embarked on experiments to create a standardized system for such a cuing channel, as well as stereo sound for TV.
The "Multichannel Television Sound" (MTS) system selected was a combination of transmission techniques submitted by Zenith Electronics Corporation and a noise-reduction system developed by dbx Inc. This combination is called "the BTSC MTS System."
In 1984, as recommended by the BTSC, the FCC adopted this system as a so-called "protected standard"--not as a set of regulations. What this means is that broadcasters are encouraged to use the BTSC MTS system, and that any consumer product set up for it will not be interfered with by any other alternative which might be used. Also, compatibility with old receivers was assured; the main-channel audio contains both left- and right-channel intelligence, and its deviation of 25kHz is maintained. What the FCC did adopt as a change in regulations is that the acceptable deviation of the audio carrier of TV transmission can now be 75kHz--expanded from 25kHz. This makes room for the transmission of more data without reducing the level of the audio of the main channel.
[Also preserved in the main-channel sound is a pre-emphasis. The so-called 75 microsecond pre-emphasis translates to a 6dB per octave treble boost with a corner frequency of about 2.12kHz. (For some reason, pre-emphasis is often given as the RC time constant of a single-order filter; to get the break-point frequency, you multiply the given figure by 2 pi and take the reciprocal of that.)]
Like stereo FM, a subcarrier transmits a "left minus right" signal. This carrier is centered at 2H (twice the horizontal sweep frequency of 15,734), or 31,468Hz. Also similar to FM broadcasts, the left-minus-right information is double-sideband, suppressed-carrier at that frequency, using the horizontal sweep as a "pilot" carrier.
A pilot carrier, locked to the sweep rate, is transmitted; it modulates the main carrier to the extent of 5kHz deviation. Sending this pilot carrier serves two purposes: receivers which only decode the audio of TV signals can be made for BTSC MTS, and the very presence of this 15,734Hz carrier informs the receiver that multichannel sound of this standard is present. If this subcarrier were to be different in frequency by as little as 250Hz, the BTSC MTS receiver will ignore it. This feature will be discussed further in the section called "Other Subcarriers"; the FCC protected standard makes no rule against sending other formats, but the BTSC MTS receiver is supposed to know when a signal is intended for it.
The bandwidth of this left-minus-right channel is 30kHz. A 15kHz frequency response is obtained, which complements the left-plus-right information of the main channel.
This "difference channel" also employs a dbx noise-reduction system.
The SAP ("Separate Audio Program" or "Second Audio Program" Channel--
This is a frequency-modulated subcarrier at 5H (5 times the horizontal sweep rate of 15,734, 78.671kHz). The audio response is limited at 10kHz, and the subcarrier's own deviation is limited to 10kHz. The injection into the main-channel carrier is such as to limit its deviation to 15kHz.
Now, as an added frill, a detector in the SAP receiver is set to ignore this subcarrier at an injection level of less than 7.5kHz deviation. Thus, a station could send a subcarrier of 5 times the horizontal sweep for its own purposes; if it were limited to perhaps 3kHz deviation of the main carrier, you, the consumer, would never know it.
The "Pro Channel"
A channel was set aside for cuing TV crews. Centered at 6.5H (6.5 times 15,734, or 102.271kHz), this is an FM subcarrier whose deviation of the main carrier is limited to 3kHz. Naturally, it is a low-fidelity audio channel. Perhaps the worst of its problems is that it is extremely susceptible to multipath distortion. Mr. Roy Trumble, chief engineer at KRON-TV in San Francisco, tells me that it gets extremely noisy if any multiple reflections are picked up by the receiver. Oddly, it is rarely used by stations who have it.
It has been suggested by Dr. John Ball, President of the National Captioning Institute, that noise-reduction and frequency compression techniques might make it acceptable for carrying the speech of the description. This would, however, require that users of video description would have to have special decoders to receive the modified "pro channel."
"dbx" Noise Reduction
Originally, the dbx system used a single wide-band compander which reduces the dynamic range to exactly half. Thus, a change in level of 2dB is squeezed (companded) into 1dB, and an expander which does the opposite decodes the sound. The system submitted for adoption in MTS is more complicated. Both left-minus-right and SAP subcarriers use this modified dbx processing.
The companding system used includes a fixed pre-emphasis, spectral compression, and amplitude compression. The fixed pre-emphasis has two corner frequencies, 407.9Hz (390us) and 2127Hz (75us). The spectral compression is part of a proprietary process by dbx. The amplitude compression is 3-to-1 for high-frequency audio, while the low frequencies are compressed by a factor of 2-to-1.
It is boasted that the signal-to-noise ratio for the SAP sound can be greater than 60dB.
Why is Description Audio Mixed with Program Audio
At least one Sony SAP TV gave the user a separate jack through which the second audio program could be patched into a hi-fi amplifier. Why don't all manufacturers do this? The reason is that only one dbx noise-reduction chip is installed in most receivers; this precludes simultaneous hearing of the stereo TV sound and SAP.
For video description, this has two distinct disadvantages. First of all, if only one TV receiver is employed, the rest of the family has to be content with monaural sound--if you, Dad, want to hear your darned description thing. Second, it is well known that "selective attention" enables us to turn an ear to one information channel or another as we wish; video description would be less distracting if it were emanating from a separate loudspeaker.
With SAP equipment being ubiquitous, it is unlikely that we will get special decoders for our description channel, which would make unmixed description possible. And, even if we could talk manufacturers into building all future sets with two dbx chips and simultaneous SAP outputs, the prevalence of older SAP equipment would lead to the service-providers' decision to stick with a mixed program sound.
In the Smith-Kettlewell Technical Viability report, I recommended that any archives of described programs include an unmixed description track, just in case a way to separate these becomes available. WGBH keeps a multitrack audio tape containing such an unmixed track, plus the SMPTE Time Code which would afford synchronizing this tape with the video tape at a future time.
The protected standard leaves open the possibility of sending any other subcarrier formats and modulation schemes. The restrictions are that the main-channel deviation caused by the new ones must not exceed 50kHz, and that the total mainchannel deviation be limited to 75kHz. The design of the receivers is such that schemes other than the BTSC MTS will be ignored, as long as any new pilot carrier is not within 250Hz of the horizontal sweep frequency.
Thus, stations have the option of transmitting all kinds of data channels onto their audio carrier. If the SAP subcarrier is not taken over by somebody, and if stereo TV programming does not get off the ground quickly, individual stations, approached by paging companies and computer buffs, might see money in surrendering their expanded bandwidth.
Phase Distortion of Older Transmitters
In the previous article, it was pointed out that old transmitting equipment, even the antenna traps, prevent the installation of a SAP generator. One particular quirk in old equipment that raises cane is phase distortion in the video signal.
The BTSC pilot carrier is phase-locked to the horizontal sweep of the video, and the re-inserted subcarrier of the left-minus-right doublesideband, suppressed-carrier signal is phase-locked to this pilot carrier. Phase distortion, which shifts the video sweep slightly, causes a particularly raspy artifact in the subcarrier audio which has been described as sounding like a torn speaker cone. (This effect bears the moniker of "buzz beat.")
Like with poorly designed single-sideband ham transmitters which FM when you talk loudly, power supply decoupling and isolation of oscillators constitute major modifications to the equipment. The most practical cure is to replace the transmitter.
By Albert Alden
In our Summer 1982 issue we described an auditory compass. This has proven to be a very successful device. However, it has two drawbacks: the size of the Hall Effect sensor (9 inches long), and the cost of the sensor (now about $100).
In a NASA Tech Briefs publication (Technical Support Package for Tech Brief LAR-12832, "Simple Magnetometer for Autopilots," Langley Research Center, Hampton, Virginia 23665), H. Douglas Garner writes about a magnetometer for an autopilot using a fluxgate device as the sensor. The fluxgate sensor dimensions are 1-1/2 by 1-1/2 by 3/8 inches, and the cost is about $25. It can, because of its small size, be mounted in the same enclosure as the circuitry for the compass.
This article will describe an auditory compass utilizing the fluxgate sensor driven by circuitry based on that used by Mr. Garner and using the output circuit in the original Smith-Kettlewell compass. The description of the fluxgate operation draws heavily on the Technical Support Package from NASA.
In using this compass, the sensor (or, in this version, the box with the sensor in it) is oriented so as to achieve a null of the auditory signal; this occurs when the sensitive axis of the box is perpendicular to the earth's magnetic field; i.e., east-west. Rotating the unit in one direction causes a "beep-beep" signal, while rotating it in the opposite direction produces a "ding-ding" sound. One end can then be arbitrarily marked with tape so as to assign this end with a particular direction (our unit "beeps" when the tape end approaches north.)
This system has one disadvantage, inasmuch as error is introduced if the sensor is not level. This presents no problem near the equator. However, as one approaches either of the earth's magnetic poles (progressing toward increased latitude, north or south), the "dip" or angle of inclination of the earth's magnetic field becomes significant. An approximation of the resulting error can be expressed as follows: The error in the null direction that one can expect is approximately equal to the tangent of the angle of the "dip" times the angle to which the sensor is held off-level. For example, the angle of "dip" of the earth's magnetic field is approximately 62 degrees in San Francisco. The tangent of this angle is 1.88. For every 1 degree that the sensor is tilted off-level, the error in the null direction will be 1.88 degrees.
The flux density of an alternating magnetic field can be determined with a coil of wire by measuring the induced voltage and by knowing the area enclosed by the coil, the number of turns, and the rate of change of the field. The rate of change is easily known for a sinusoid field. For a steady field, if we could mechanically spin a coil or somehow switch on and off the field passing through the area enclosed by the coil, we could obtain a signal proportional to the magnitude of that field. It is the latter scheme we use here.
The heart (or should we say core) of the fluxgate is a ferrite alloy with a high permeability (low resistance to magnetic fields) and a sharp saturation point (a square hysteresis curve). If we place a strip of this ferrite parallel to the earth's magnetic field, more of the lines of flux will be drawn to and pass through the strip than through the equivalent free space. If we were to wind a coil (to be called a drive coil) around the strip and drive enough current to magnetically saturate the ferrite, the permeability drops to that of free space. When this happens, the additional lines of the earth's magnetic field passing through the strip are "gated out." Thus, the current through the drive coil can be thought of as opening and closing a "gate." When the strip is not saturated, the gate is open; when saturated, it is closed.
A second coil wound on the ferrite strip can be used to sense the gated earth's magnetic field, a voltage being induced when the field is gated on--and also when gated off. There is a big problem with this arrangement. The sense coil will also see the very much larger field generated by the drive coil. The solution to this predicament is to have a configuration where the field caused by the drive current does not couple into the sense coil. A toroidal core accomplishes this. The drive coil is wound on the toroid in the standard fashion, and the sense coil is wrapped over the toroid (not around the circumference) so that its cross-section is rectangular with a width equal to the outside diameter of the toroid; its height is that of the thickness of the toroid. The third dimension is a bit less than the diameter of the toroid.
The drive winding is excited by a square wave of suitable frequency and amplitude so that the core is saturated halfway through each half-cycle. As the polarity of the drive voltage reverses--at the end of the first half-cycle--the core unsaturates until about the center of the next half-cycle, when saturation again occurs.
Now, as we have already seen, any external magnetic field will be drawn into the core when the core is unsaturated, and will be expelled when it becomes saturated. Each time these external lines of flux are drawn into the core, they pass through the sense winding and generate a voltage pulse whose amplitude is proportional to the intensity of that component of the external field which is parallel to the centerline of the sense winding. The polarity, or direction, of this pulse will be determined by the polarity of the external magnetic field with respect to the sense winding. When these lines of flux are expelled from the core, they cut the sense winding in the opposite direction, and generate another voltage pulse of the same amplitude, but of opposite polarity.
This pulse pattern is repeated twice for each cycle of the driving frequency. This means that the signal coming out of the sense coil is at twice the frequency of the driving voltage. How this signal is utilized will be described in the following section.
A CD4060 oscillator/divider operates with its clock set to provide a reference signal at Q6 of about 2800Hz and a drive signal at Q7 of half the reference frequency. The clock frequency is adjustable. The drive signal is applied to the inverting input of one amplifier of a quad op-amp (LM324) and also to the non-inverting input of a second amplifier. The other terminals of these two op-amps are tied together and go to a reference voltage (VREF), about 1/2 VCC.
The output of one op-amp is connected to one of the ends of the drive coil. The other op-amp output goes to the other end of the coil through a 0.47uF capacitor. The op-amps are operating open loop (akin to voltage comparators) and drive the coil in a push-pull fashion. The sense coil has a 0.1uF capacitor across it to create a resonant circuit at about 2800Hz. The previously mentioned pulses drive this resonant circuit which, if everything were perfect, would give the following type output:
A sine wave at 2800Hz has an amplitude that is proportional to the component of the field (parallel to the centerline of the sense coil), and is in phase--or 180 degrees out of phase--with the reference signal, depending on the direction of the field component.
One side of the sense coil is connected to VREF, and the other side to a synchronous demodulator. The demodulator is made from an op-amp (the third one of the LM324) connected as follows:
A 100K feedback resistor goes from the output to the inverting input. A 100K inverting input resistor goes to the sense coil lead that is not connected to VREF. A 100K resistor goes from the non-inverting input to the same sense coil lead.
A single-pole single-throw switch goes from the non-inverting input to VREF. When the switch is closed, the gain of this circuit is minus 1; when open, the gain is plus 1. The switch is made from one section of a CD4053 analog switch with the status of the switch controlled by the 2800Hz reference signal. The demodulated signal is amplified and filtered by the fourth op-amp in the LM324. This amplifier is configured in the inverting mode with an input resistor of 4.7K and an adjustable 100K feedback resistor shunted by a 0.1uF capacitor for the filtering. The output of this amplifier is a DC signal, referenced to VREF, whose amplitude and polarity are indicative of the magnetic field the sense coil uses.
The second switch of the CD4053, along with the reference signal, is used to apply out-of-phase 2800Hz square-wave signals to the ends of a 100K pot. From the wiper, a 100K resistor goes to the sense coil lead not connected to VREF. This circuit is used to cancel any offset in the sense coil signal and provide the user with a means by which the instrument can be "nulled" for an exact east-west orientation.
In operation, the drive coil is activated at a rate of about 1.5Hz with a duty factor of about 10%. This is accomplished by driving the reset pin of the CD4060's oscillator with the output of one section of a CMOS 7556 timer having the above characteristics. This is done to reduce battery drain and to provide a signal at the output of the amplifier/filter that can be used to generate the famous Smith-Kettlewell Compass Auditory Output.
At this point, the signal consists of 1.5Hz pulses, either up or down as determined by the magnetic field, with these pulses being referenced to VREF. This reference voltage is made up of an op-amp (1/2 of an LM358) connected as a voltage follower looking at a voltage divider across the battery supply. Pulses are either above this reference 10% of the time, or below it 10% of the time.
A scheme was devised by which the lower portion of this waveform is always fixed at ground. In other words, the desired signal either has positive-going pulses of 10% duty cycle or positive-going pulses of 90% duty cycle; the bottom of this wave always rests at ground. To accomplish this, a diode clamp is used to hold the lower portion of the signal at ground (this is an "active clamp" in which an op-amp is used to take up the slack of a diode's junction voltage). The signal into the clamp is capacitively coupled; when the signal tries to go negative, the clamp prevents its end of the coupling capacitor from doing so, thereby forcing the previous stage to charge this capacitor to the degree of negative swing. The result is that the output of the coupling capacitor (the input of the clamp) has on it the desired positive-going signal.
The "holding time" of the capacitor was chosen to be just a bit short for the 90% time segment, so as to allow the long pulses to sag somewhat. This gives the long beeps a bell-like quality. The signal is, in practice, referenced to an adjustable voltage near ground so as to match the voltage required at the gate of a MOS-FET mentioned below.
The clamp-shifted, positive-going signal is used to gate a MOS-FET used as a variable resistor that intermittently couples a tone signal (from the second half of a CMOS-556) into an audio amplifier (an LM386). The output of this section of the 556 is a square wave which is "lightly filtered" to give a more pleasing sound.
Since we wanted to package the fluxgate sensor along with the electronics and speaker in the same box, it was necessary to use a non-magnetic speaker. We found a piezo speaker available from Radio Shack to be quite satisfactory.
From the positive terminal of a 9V battery there is an on-off switch. The circuit side of the switch is VCC. The negative side of the battery is ground.
Between VCC and ground there is a pair of series-connected 15K resistors. The 15K connected to ground is bypassed with a 10uF capacitor, the negative at ground. The junction of the two resistors is connected to pin 3 of the LM358. Pins 1 and 2 are connected together, and this point is VREF. Pin 8 of the LM358 is connected to VCC, and pin 4 is grounded.
Pin 16 of the CD4060 goes to VCC. Pin 8 is grounded. Between pins 11 and 10 are the following components in series, starting with pin 10: a 10K resistor, a 10K pot (wired as a variable resistor), and a 75K resistor. From the junction of the 10K pot and the 75K resistor, a 100pF capacitor goes to pin 9. The 10K pot adjusts the frequency of the oscillator. Pin 6 of the CD4060 (the 1400Hz drive signal) goes to pins 3 and 6 of the LM324. Pins 2 and 5 of the LM324 go to VREF. Pin 7 of the LM324 goes through a 0.47uF capacitor to one of the drive coil leads. Pin 1 goes to the other lead.
A 0.1uF capacitor is connected between the leads of the sense coil. One lead is connected to VREF, and the other to three 100K resistors. The other end of one 100K goes to pin 13 of the LM324, the other end of the second one goes to pin 12, and the other end of the third one goes to the wiper of a 100K 10-turn pot. Another 100K is connected from pin 13 to pin 14 of the LM324. Pin 12 of the 324 is also connected to pin 14 of the CD4053 switch.
Pins 2 and 16 of the CD4053 go to VCC. Pins 1, 6, 7, and 8 are grounded. Pin 12 goes to VREF. Pins 10 and 11 go to pin 4 of the CD4060. The 100K 10-turn pot is connected between pins 10 and 15 of the CD4053. As mentioned, the wiper of the pot goes to the third 100K resistor connected to the sense coil.
Between pins 14 and 9 of the LM324 there is a 4.7K resistor. Between pins 9 and 8 is the parallel combination of a 100K pot (hooked up as a variable resistor) and a 0.1uF capacitor. Pin 10 is connected to VREF. Pin 4 of the LM324 goes to VCC, and pin 11 is grounded.
On the first half of the 7556, pin 13 (Discharge) goes through 560K to VCC. Pin 13 also goes through 75K to pins 12 and 8 (Threshold and Trigger), which are tied together. Pins 12 and 8 also go through 1uF to ground (negative at ground). Pin 9, the output, goes to pin 12 of the CD4060. Pins 10 and 14 go to VCC. Pin 7 is grounded.
Pin 8 of the LM324 goes through a 1uF capacitor (positive at pin 8), then through a 510K resistor to the wiper of a 10K pot. The ends of the pot are connected between VREF and ground. The junction of the 1uF and 510K goes to pin 6 of the LM358, the inverting input of the op-amp associated with the clamp. As its feedback circuit, this op-amp has a diode connected between its output and inverting input; pin 7 goes through the diode to pin 6 (anode at pin 7). Pin 5, the non-inverting input, is tied to the wiper of the 10K pot. The output signal is not taken from the output of this op-amp, but comes from the junction of the 1uF and 510K which also goes to pin 6.
The second half of the 7556, the tone generator, is free-running. Pins 2 and 6 (Trigger and Threshold) are tied together and go through 0.01uF to ground. Pin 4 goes to VCC. Pins 2 and 6 also go through a 47K charging resistor to the output, pin 5. Pin 5 also goes through 15K to the top of a 10K volume control, with the bottom of this control being grounded. This control is shunted by a 0.047uF capacitor.
The wiper of the volume control goes through 0.1uF to the source of a MOS-FET (2N6660), with the drain going through 470 ohms to ground. This drain also goes to pin 3 of an LM386 amplifier. The gate of the MOS-FET goes to the output of the active clamp; i.e., to pin 6 of the LM358 and to the junction of the 1uF capacitor and the 510K resistor.
Pins 2 and 4 of the LM386 are grounded, while pin 6 goes to VCC. VCC is bypassed to ground near the LM386 by 10uF (negative at ground). A 10uF electrolytic is connected between pins 1 and 8, with the positive end at pin 1. Pin 5, the output, goes to one lead of the piezo speaker. The other speaker lead is grounded.
The compass we built was packaged in a box measuring 3 by 4 by 1-1/4 inches with a 9V battery compartment. The circuit was installed in the half with the battery compartment, the bottom section. In the top section we mounted the speaker on the same end as the battery. The fluxgate sensor was mounted on the other end of the top with the flat side against the top and the sensitive axis parallel to the 3-inch dimension. With this arrangement the compass can be used in two orientations: 1) with the bottom (the 3 by 4-inch side containing the battery) on a level surface; and 2) hung on a belt with the 3 by 4-inch side vertical and the 3-inch dimension on top. We have used a Velcro belt with Velcro strips glued to the box. The first section of the calibration should be done before the fluxgate sensor is mounted in the box.
[Note: For all adjustments, the fluxgate sensor should be located away from any magnetic material or sources of man-made magnetic fields.]
The first thing that must be done is to adjust the fluxgate sensor for optimum output. This means maximum signal with minimum distortion of the sine wave. The unit purchased from Precision Winding, Inc. (see Parts List) will be tuned to about 2800Hz by the 0.1uF capacitor. This should be within the adjustment range of the CD4060 oscillator.
For these adjustments, the 7556 should be removed and pin 12 of the CD4060 connected to ground. This will give a continuous output of the CD4060 rather than the pulsed output as used for normal operation. Also, set the 100K 10-turn pot to the center of its adjustment range. A dual-trace scope is recommended for this operation. Connect one channel to the reference signal, pin 4 of the CD4060, and trigger from this signal. The other channel should look at the output of the sense coil--the lead not connected to VREF.
Position the sensor so that the centerline of the sense coil is pointing approximately north-south, and adjust the frequency of the oscillator until you get a waveform of maximum amplitude. This waveform may be rather distorted and hacked-up looking because of magnetic irregularities in the core. Hold the sense coil fixed in its north-south orientation, and rotate the core and drive winding within it. At some position you should observe a fair approximation of a sine wave.
If necessary, readjust the frequency control. The peaks of the sine wave should occur at about midpoint of the reference signal's period. Rotate the whole coil assembly about a vertical axis, while you watch the waveform coming out of the sense winding. The amplitude of your sine wave should diminish to a minimum as the unit is rotated to an east-west orientation. (There will be some weird-looking noise left, resulting from higher harmonics.) As you continue to rotate the sensor, the sine wave will build up again to a maximum at 180 degrees of rotation, but it will now be 180 degrees out of phase with the original signal. When you achieve the best-looking sine wave you can get, with the core centered in the sense winding, glue it in place.
The signal at pin 14 of the LM324 should be a full-wave rectified version of the sense coil signal. The signal at pin 8 of the LM324 should be a DC signal that changes amplitude and polarity (measured with respect to VREF) as the fluxgate is rotated in the earth's magnetic field. The 100K pot between pins 8 and 9 should initially be set at maximum.
Remove the connection from pin 12 of the CD4060 to ground, and reinstall the 7556. Set the volume control to maximum. Position the sensitive axis of the fluxgate approximately east-west to obtain a minimum audio output. Adjust the 10K pot associated with the clamp (pins 5, 6, and 7 of the LM358) to just barely turn off the MOS-FET. The sensitivity control (the 100K rheostat between pins 8 and 9 of the LM324) is adjusted to suit the user's taste, given the instrument's application. For example, this sensitivity shall have to be turned down from maximum if the compass is to be used while walking, since the natural rotation of the body will prevent the user from maintaining an absolute null. In effect, reducing the sensitivity "broadens the null," determining the off-null angle at which the instrument saturates, after which no further rotation will increase the amplitude of the signal.
At this point, the fluxgate should be installed in the box. Access to the adjustment screw of the 100K 10-turn balance pot with the box closed is recommended. The balance pot is adjusted by laying the compass on a level surface away from magnetic materials and man-made fields and noting the relative directions of the two null positions. The balance adjustment is made so that these two nulls are 180 degrees apart.
Resistors, fixed, 1/4W 5%:
- 1--470 ohm
- 1--100K 10-turn
- 2--1uF tantalum
- 3--10uF tantalum
- 1--Fluxgate sensor (Order: "Fluidic Auto Pilot Coil" from Precision Winding, Inc., 109 S. Knight, Wichita, Kansas 67213, tel: (316) 942-2811) 1--Switch
- 1--Piezo speaker (Radio Shack No. 273-091)
- 1--Battery, 9V alkaline
- 1--Battery snap connector