A Quarterly Publication of
The Smith-Kettlewell Eye Research Institute’s
Rehabilitation Engineering Research Center
William Gerrey, Editor
Issue: SKTF -- Fall 1991
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 Details regarding products, suppliers, and other contact information are original and may be outdated.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
[Editor's Note: The temptation is to fill two pages with accolades to all those who have spirited this technology through to its current availability. I must remember that this is a technical magazine, so please let me thank promoters and supporters with a collective "heigh-ho." Thank you.]
BACKGROUND AND RATIONALE
The field of "orientation and mobility" is often treated as one. Its trained specialists are granted a single diploma for both. However, the two components, if analyzed separately, are very different. Maintaining one's orientation is a complex task involving what I call "detective work" and integration of fragments of information. Safe mobility is a skill that simply takes practice with a proper tool.
Treating "orientation and mobility" as one confuses engineers who want to help. Aiding any one facet would assist the whole process, right? The first electronic travel aids were light detectors--if blind people could only detect the shadows that sighted people use to get around in the dark, this should help. (People moving in the dark forget how often they stub their toes, and they should be taught to carry a cane.) Other designers reasoned that if obstacles could be detected at great distances (much farther away than could be reached by a practical-length cane), this new information would better prepare the traveler for the road ahead. (The result is to needlessly inform the user of objects in close proximity of the path, or to tell him about encumbrances before they interest him.)
Some of the world's finest engineers have created instruments which present the user with enough information to recognize landmarks and characteristics of the environment; these were first attempts to enhance orientation. (Typically, these instruments produce complex sound effects or other stimuli that have been learned by dedicated users.)
In 1978, Smith-Kettlewell investigated existing navigation systems to find out if those could be so specific with position information as to be useful for the blind traveler. Others have since proposed that satellite navigation arrays could be used for orientation. Arming the traveler with accelerometers and gyros to create a dead-reckoning system was also considered.
The one which sounds the most promising is satellite communication, but this would only work in conjunction with a large computer database wherein specific locations and directions are correlated with meticulous detail which must be entered in the database. This still leaves questions as to how to present specific location information to the user: In a sprawling motel complex, would a talking output tell the user that the restaurant is 37 degrees to his right across the parking lot? Could there be a refreshable tactile map with zoom viewing that would display to the user where things are in relation to where he is standing?
At least now, engineers cannot neglect "orientation" as a problem to be reckoned with. Orientation has emerged as a separate component of independent travel.
Once recognized for its importance, just what orientation information is needed? The satellite navigation engineer will want to know just how large that computer database has to be. A shortcut to studying this is to look at how sighted travelers have evolved their system of signs for orientation. What bits and snippets of information do they need? Let us examine their signs.
INFORMATION CONVEYED BY SIGNS
Every blind and print-handicapped person should perform the following experiment: Enlist two very fast print readers, one on either side of you, to read every printed thing as you walk downtown--or in the park, or while traveling by car on the freeway. The "world out there" is in no quandary as to which labels are required. Everything is labeled in print. Moreover, civilized communities put up signs just because they want to put them up--they cannot help themselves.
There is print to be read everywhere you look. Having a database of all those messages would not be good enough for them; they want to catch an eye, direct attention, steer people in the right direction, capture people's sympathy, raise people's awareness . . . keep people out and keep 'em off the grass.
Most of the moving things have something to read. Buses say where they are going, taxis beckon passengers, signs on trucks provide an indication as to what's being fixed, marked police cars invitingly offer help, and the label on the bread truck in front of the restaurant indicates to the observer that fresh pastries have arrived.
Inside: Numbers in front of elevators mark the floors of destination, information desks proclaim how they can help, exit signs show the way out and specify where they lead . . .
And don't think those signs are inexpensive. Exits are not only lit, but many have backup power. Taxicabs have a fancy indicator on top. Buses either have programmable signs or mechanical scrolls. Some signs are expensive neon displays; others may be in gold leaf. Most are not drawn with marking pen on cardboard; they are created with expensive materials by graphic artists.
What signs say is important, but just as important is where they are. A sign proclaiming "taxicab nearby" would be useless. A centralized sign listing information stations in a government building, without having labels at the individual desks, would not be good enough. Suppose a committee on esthetics proclaimed street signs to be eyesores and lobbied for their removal; a street map at the city gates would not be good enough. Signs tell where it is; these comprise the ultimate orientation system. It was obvious that signage should be chosen as the model orientation system.
Fortunately, most printed signs contain information unnecessary to orientation. Bumper stickers and political posters are not of interest to us. In fact, we are lucky in that the most important signs are lighted signs and power exists at their locations. Also apparent in systems of signage is the redundancy throughout. Street names or locations listed in subway stations correlate with the names of streets on street signs above. If one were to read "Clay Street theater," he would assume its proximity to Clay Street. Therefore, let it be argued that if blind and print-handicapped people had access to 2% of visual signs, we would have a functioning orientation system.
Wouldn't it be nice if blind and print-handicapped travelers could have a reading machine that would read those print signs? There are three major barriers to this: (1) Signs, being drawn by graphic artists, do not play by the rules of print. (2) Because of parallax, the shape of the letters changes depending on the viewer's position; off to the side, the lettering becomes very narrow. (3) It is one heck of a reading machine which can decipher an image "on the fly"--scanned without the user knowing that he has done so. Designing a reading machine which can put up with these adverse conditions will be a long time coming.
ADOPTING SIGNS AS THE SOLUTION
Because signs provide specific location information, it was decided that accessible signage was the logical approach to an orientation system. In 1979, it seemed feasible for Smith-Kettlewell to design inexpensive attachments to signs which could be made as add-ons.
It was suggested that bar codes, added to visual signs, could be read by another sort of reading machine carried by the user. This sounds okay, until you contemplate the variations in distance and lighting conditions involved. The machine must read the bar code of an exit sign from many feet away or just a few feet away (representing perhaps a 10-to-1 difference in size). The machine must tell you that you've encountered a sign, and it must recognize a bar code in the face of very adverse scanning technique (the user will not know when he is scanning a sign). For overhead signs (such as street signs), proper inclination of the reading device will depend on the user's distance from the sign. The device cannot care about ambient lighting or the condition of the sign; a smudge or defacement of the bar code must not destroy the information. Perhaps someday this will work; if so, it can be added to the infrared equipment proposed.
Two other technologies have been proposed elsewhere: (1) Radio, particularly VHF, has been suggested. Where this falls down is directionality; unless the frequency is very, very high, the antennas (most importantly the one carried by the user) must be large in order to achieve good directionality. (2) Another system, used by park services, is the inductive loop, usually buried underground. When the listener's receiver is inside the loop, the message is heard. This requires a lot of digging to install; moreover, intersecting loops cannot be permitted, or messages will interfere with one another.
A very, very old technology (tried by Alexander Graham Bell as the "Photophone") is to communicate voice information by modulated light beams. (The Hearever Company made toy versions of this which I remember trying as a young lad. The person talking vibrated a silvered mylar diaphragm with a mirror surface; the listener attempted to find modulated reflections of the diaphragm with a photocell and amplifier.) Visual signs are "line of sight"; why not use modulated light beams to create detectable signs? The user would carry a light-detecting receiver, and every time it encountered a beam, it would talk--picking up a digital recording which is modulating the light beam.
Modulated infrared transmission was becoming universal. High-power LEDs were newly available in 1978, and a Siemens photodiode appeared on the market (used in Senheiser infrared cordless earphones, for example), and this brought marked improvement to sensitivity of infrared receivers. Early work at Smith-Kettlewell in 1979 led to the development described below.
THE TALKING SIGNS (A REGISTERED TRADEMARK OF LOVE ELECTRONICS)
At Smith-Kettlewell, events happened so fast that everyone is still debating as to who thought of which part first. I can tell you that, as far as I know, Pete Neiberg and I were the first to modulate the Z axis of an oscilloscope with a tape recorder and detect the beam with a cadmium-sulfide cell into a Radio Shack pocket amplifier. Dr. Erich Sutter soon convinced everyone that amplitude modulation was doomed, since it could never compete with ambient light. Al Alden made the first digital modulators. Soon, a recorder was made with which messages, spoken into a microphone, could be put into EPROMs. The first practical transmitter was a very small circuit board (2 x 2-1/2 inches) and was powered by a battery eliminator from the ac line.
Not to be forgotten is the main character, Bill Loughborough. These experiments went on in his laboratory. He recognized the inevitability of this system before the rest of us admitted to it. It was his work with infrared beacons and games that led us down the infrared path.
While signs are public, the reading of signs is a personal matter. Therefore, we chose to put the "speaker"--an infrared receiver--in the user's hand. Having buildings and street signs talk to blind people draws needless attention to us, and when sure of our ground, we want any adaptation of the environment to keep still. Therefore, what gets attached to the location of the desired sign is an infrared transmitter. Since its power requirement is quite low (50mW for a restaurant doorway, and 5 watts for a street sign), the decision was made to keep the sign simple and let it run all the time; i.e., the receiver does not trigger the sign, but merely picks it up when a beam is encountered.
Infrared light emission can be made unidirectional, bidirectional, or scattered in all directions, depending on transmitter optics and the mounting of LEDs. Therefore, arrangements for mimicking visual signs are possible. If you are supposed to see a sign from many directions far away, an infrared pattern can be configured that way. If a room number is to be found at close range looking at an office door, a low-powered infrared sign can be configured that way.
Localization of the sign is done by scanning the receiver in search of the clearest signal--with the least amount of noise.
Bill Loughborough, recognizing that it was time to become commercial with this equipment, left Smith-Kettlewell and formed a manufacturing company. His last name being incomprehensible to many, he had used the name "Bill Love" for theater work; he therefore calls his company Love Electronics.
Removal of ambient light, particularly sunlight, from the receiver is key to the equipment working in outdoor conditions. A graduate of Cal Tech, and a former Smith-Kettlewell employee, Jules Madey, designed the late-model receiver. (A rather famous ham-radio operator involved with expeditions to the South Pole, receiver design was nothing new to him.)
By now, you can order transmitters with custom messages for $150 each. The receivers, which require special hand-wound coils at this time, are $250. Quantities of transmitters made have been in lots of perhaps 20 units, and receivers have been made in smaller quantities. You can only imagine how low prices will fall when transmitters are made 20,000 per lot, and receivers exist by the hundreds. Smith-Kettlewell has supported infrared signs as a way of complying with signage requirements in accessibility laws, and this would bring production of the equipment to the level of a major industry.
Talking Signs not only apply to the needs of blind users, but they directly address the needs of anyone who cannot read print (the "print-handicapped" whose inclusion has caused expansion of our library service).
In quantity, Talking Signs (TM) receivers will cost as little as transistor radios; this may not matter, however, since they could be issued through grants--just as "Radio Reading Service" receivers are issued free. In quantity, transmitters with duplicate messages--"exit," or "Pine Street"--will use mass-produced ROMs, and transmitter circuits will be built into standard lighting fixtures.
The newest transmitters are available as bare circuit boards measuring 2 x 2-1/4 inches, or as finished small cabinets with LEDs embedded in the front panel. They operate either from a 9-volt battery, which lasts a day or two, or can be powered by small modular dc power supplies. Large arrays of as many as 48 LEDs have been made for long-distance signs--for example, the street signs at Fifth and Market Streets in San Francisco. (The LEDs are 60 cents each in small quantities, so large arrays are not prohibitively expensive.) As you would expect, high dc driving voltages (35 volts in the Market Street units) are used for high-powered signs, and beefy driver chips such as the Texas Instruments 75477 are required to power the lights.
The receiver is pocket-sized, and has a hole in one end behind which the photodiode is mounted. The speaker is on one face of the cabinet, and pressing a button on the side turns the receiver on. The receiver has a fairly wide angle of acceptance; nevertheless, this does not damage the user's ability to determine direction of the sign. Departure from the signal--scanning too far left or right--squelches the receiver. Smaller side-to-side motions, as well as getting closer to the transmitter, causes marked improvement in background noise of the audible message.
Erich Sutter chose FM as the way to avoid ambient ("dc") light, but FM turns out to have another advantage: the strongest signal "captures" the receiver. Therefore, in the presence of a high-powered street sign, doorway signs can still be heard clearly as they are approached; they need only appear 20dB stronger to the receiver to capture it. Thus, pointing toward the low-power sign on the corner store is all that is required to get away from other differently configured signs.
• Mr. William Loughborough
400 N. Columbus, #42
Goldendale, WA 98620
Tel: (509) 773-5958
• Mr. Kent Sokoloff
67 Wilmot Street
San Francisco, CA 94115
Tel: (415) 931-1364
• Mr. Jerry Kuns
Talk of the Town
1182 Market St., Suite 211
San Francisco, CA 94102
Tel: (415) 863-TALK
The original transmitters, storing speech at a sampling rate of 8kHz, were mounted in 1 by 2 by 3 inch cabinets glued to the backs of 9V plug-in power units. The LED signals were sent to 3/4 by 1 inch boards carrying 3 LEDs and a resistor. Transmission was at 25kHz, but with a 12% duty cycle; 4.8 microsecond pulses of 100mA were carried by the three LEDs.
Several room-number signs ran for years here at Smith-Kettlewell's Rehabilitation Engineering Center; the little boards carrying the LEDs were just taped to the top of the door jambs.
As receivers improved, however, these transmitters turned out to have a bit too much power. With the sensitive Love Electronics receivers--also having a forgiving acceptance angle--it was sometimes possible to receive a room number reflected from a glossy surface across the hall (from a page on a bulletin board or from a metal elevator door). Therefore, for modern installations in close quarters, one or two LEDs running at 25 or 50 milliamps is more appropriate.
In the early 1980s, a demonstration installation was set up by Dr. Helen Elias, head of Disabled Services for the San Diego Community College System. Fifteen of our original transmitters were divided between two campuses. Labels such as "library," "bookstore," "cafeteria," and others were attached to these doorways with tape, just as they were on our indoor installation. (A mixture of early receiver types was then available; this meant that fully outdoor installations were avoided. Luckily, the doors mentioned opened onto shaded breeze ways. With the newest receivers, outdoor signs have been successful.
Reports on the San Diego setup were positive. Receivers were available at the Disabled Students' Services office. As could be expected, these were tried by new students being introduced to campus layout. Once the various locations were learned, the receivers were returned.
Since 1989, modernized transmitters have been available from Love Electronics. The sampling of the recorded speech is at 64K bits per second; the playback quality is excellent. Eight seconds of speech can be stored, and the allotted time can be broken up into one-second segments as desired. (A connector on the circuit board makes addressing these segments possible.)
The recording of choice is stored on a 512K-bit EPROM which fits into a "zero-insertion-force" (ZIF) socket on the transmitter board. This chip can be sent back to Love Electronics for re-recording at any time; the EPROM can be changed by any technician on site.
Transmitters can be ordered as open stuffed circuit boards, or as fully packaged units. In the packaged form, a 1 by 2-1/2 by 4 inch cabinet has LEDs--an array of your choice--mounted on the largest panel. The back side of the cabinet has a battery door for insertion of a 9V battery; a notch filed in the door allows the cable from a 9V DC power supply to emerge, and the DC connection is made via the standard battery-snap connector.
The first 20 of the newest transmitters were purchased by the National Rehabilitation Hospital's Rehabilitation Engineering Center, located in Washington, D.C., under a "Project on Evaluation of Assistive Technology." This organization has been kind enough to loan the transmitters out for demonstrations elsewhere, but when the devices return to the hospital, they are retaped in place in their Technology Demonstration Center.
The first restaurant to order a Talking Sign was La Mediterranee, a cafe frequented by Smith-Kettlewell personnel. Drawing about 50mA from the DC power supply (a modular type plugged into a nearby outlet), it supplies four single LEDs with 45mA pulses. Two of the LEDs face forward, while the others are pointing perhaps 20 degrees left and right. As you walk along the sidewalk with the receiver pointing at the wall of storefronts, the message can be received just as the previous entrance is passed. With the Love Electronics receiver, this message can be heard at the near edge of the first lane of traffic in the street (a car's width beyond the curb). With the receivers described here, the signal can be picked up at the line of parking meters along the edge of the curb.
This transmitter box is secured to the inside of the front window of the restaurant by double-sided adhesive foam tape. When the window needed replacing, the owner reinstalled the unit without any help from our technicians. His comment on the cosmetics of the device is that it looks like it might be a burglar alarm.
The eighteen street signs now going up in the area of 5th and Market use the small Love Electronics boards to provide a modulated signal to Texas Instruments high-current drivers. A total of six messages per sign are addressed and presented sequentially.
A control circuit senses the voltage on the "WALK" and "WAIT" lamps in the pedestrian signal header (called the "pedhead). The "WALK" lamp addresses the statement "green" on the speech player, a solidly on "WAIT" lamp voltage addresses the statement "red," and flashing of the "WAIT" lamp addresses the word "yellow."
Each Talking Sign has two LED arrays. A horizontal bar of 24 dual LEDs sends the information: "Fifth and Market," (pause) "Zero-Hundred block (denoting that this is the first block of 5th Street)," (pause) "Facing northeast," (pause). This broad array can be "seen" (picked up by the receiver) across the whole width of the very wide Market Street side walk.
The other LED array is a 3.6-inch long column of twelve dual LEDs shrouded by black-painted aluminum fins on either side. The fins are bent so as to send a narrow vertical beam of infrared along the crosswalk. The crosswalk messages--"green," "yellow," and "red"--can only be picked up when the person carrying the receiver is standing well within its boundaries. Moreover, by scanning left and right for the "edges" of the signal (the directions where the signal drops out), the pedestrian gets direction information about which way the crosswalk runs.
These LEDs are run at full gallop, 175mA pulses being carried by them. For now, additional "pedhead" cabinets are used to house these talking street signs. There is plenty of room in those cabinets, and a signal manufacturer could easily design the Talking Sign so as to be part of the regular equipment.
Transmitters built into new structures can indeed be unnoticeable. For example, plans are afoot to equip San Francisco's new library with labels at information stations, etc. Many of those will be mounted in 4- by 4-inch electrical boxes merely added to the conduit of the main electrical service. The LEDs will be behind clear-plastic windows mounted in standard cover plates. For low-power installations, the power supply and Love Electronics circuit board would both fit in a "switch box," and the LEDs could peak through the rectangular hole in a standard switch plate.
Besides the packaged transmitters currently available, it is not hard to imagine other package styles that would encourage retrofitting--mounting Talking Signs in established structures. Obviously, replacement light fixtures come to mind, but the reader will note the "adapted porch light fixture" in the transmitter article published here. Also described in a proposed replacement doorbell button with a sign in it.
Labels vs. Beacons
A Treatise by Bill Gerrey
Talking Signs are often misunderstood. Some feel that they can be used as guides from place to place--and why not? Others feel that they will take over the mobility task in such a way as to let mobility skills deteriorate. Both fall victim to the view of infrared signs as beacons.
True, Talking Signs, by virtue of being light beams, tell the holder of the receiver where they are. However, they can never tell the traveler about the pathway to them; this path will be subject to everyday clutter, and to follow the beckoning beam smack-dab into a hazard would be possible if Talking Signs were used improperly. Indeed, my own dear mother could follow a print sign right into an obstacle at her feet, and she had perfect vision. Therefore, the hazards of not looking where you are going are not only those of the visually impaired.
Signs are information-carrying labels, not beacons, and Talking Signs should be no different from print signs in this way. Our mobility skills as blind travelers are what guarantee our safety as we move about. If we get information from a reasonably well-labeled environment, this would be done with signs, and this should have little to do with how we safely move about.
I have a 1951 recording of my father reading about a guide cane. The cane was a geiger counter, and it was proposed that radium paint stripes be drawn everywhere a blind person has to travel. And he would follow those stripes with this sizzling-popping cane--wearing obstructive earphones, of course.
Whether metal detectors following pipes or wheels running in tracks, these guide-strip gadgets come along now and again. Always, they propose to "solve?" the problem of safe mobility. Two fatal flaws keep those inventions at bay: The first is that people do not negotiate their environment as ants do--plotting a course and then mindlessly following it without exercising any option to change it. We constantly change the order in which we do things (maybe we want to go get a drink of water before we check the mail), and consequently, our paths are complex and varied. Second, all paths are subject to interruption--temporary obstacles interrupt them, people stand in the way, and things with wheels treat walkways as parking spaces.
Exercising our options like anybody else, we blind travelers change the order of our plans and take shortcuts, too. Properly guarding our bodies solves the safety problem, and proper exploration of the immediate area promotes directed mobility; nothing else will do.
I must confess that I took a very hard line about Talking Signs on street corners. I was dead set against them giving "red/green" information at first; street-name information was the main point. The feature that softened me on this point is that unknown crosswalks, some in obscure directions, can be indicated--and their directions "pointed out." This qualifies as information for my hard-line criterion; other people use painted lines to "label" cross walks, and thus, the Talking Sign retains its label status.
However, I retain a hard line against long-range beacons that are intended to "draw you some place." Like other guide-path systems, the path is defined by one guy, and obstructions placed there by another--the subsequent obstructionist probably not even knowing that an infrared "path" has been defined. And from the blind traveler's position, following such a path would be restrictive. In the worst case, a traveler whose attention is locked into adherence to a beacon might not be properly guarding his body and exploring his immediate territory to remain safe.
If there is ever quoted a law of "travel aids" assigned to this author, Wm. Arthur Gerrey, it should read, "WAG's Law-- Never abdicate your responsibility nor override your skill, consigning your fate to an unknowing gimcrack of technology!
The key word in that law is "unknowing." Present-day sensory devices don't sense everything that you can; a "clear-path detector" may beckon you into an open elevator shaft. At an intersection of streets, you may hear the truck turning right around the corner, while the sensory aid knows nothing of it. Other devices are impartial to all current activity, sensing nothing; these would include traffic signal indicators, visual or auditory. What does the red-vs.-green light know about a truck turning the corner? Impartial devices don't even know you are there.
Talking Signs can be used without putting your mobility skills on hold. Suppose you are looking for a curbside telephone booth, and suppose the street is at your left. It is all right to point the receiver 45 degrees left while you are walking along, waiting for a phone booth to announce itself in your receiver. Once heard, it is not all right to swerve toward it without checking the new direction with your cane; doing so might hang your liver on the handlebars of a motorcycle. Once the first announcement is heard, perhaps the best tactic is to point the receiver directly to the left and walk past the desired phone booth. Then, knowing about where it is, pocket the receiver; use your mobility skills to find the phone booth--guarding your head with your free arm as you deem necessary.
When I'm down at 5th and Market and I find the crosswalk I want, I note how the cycle is progressing; if it is saying green, I wait a cycle, and when it next says "green," I deep-pocket that receiver. Thus undistracted, I become a precision combination of traffic lawyer and sense organ. Listening to all traffic and double-checking its direction, I trust my fate to the best guardian I know--me.
Consider the example of a hotel built around a large atrium. The unthinking designer might suppose that all Talking Sign transmitters should be high enough in power to make everything visible from any point on the pathway around the atrium. This would lead to a blend of signals of similar amplitudes which the receiver, pointing across the atrium, could not differentiate. I suggest that no more than two transmitters be of high enough power to transmit across the atrium--most likely indicating the exits or elevators. So that individual rooms and restaurants can be singled out as the traveler walks around the pathway, signals should be just strong enough to reach the rail bordering the atrium.
I firmly believe that it is the blind travelers "lot" to walk around more than other people in order to "get the lay of the land" and to find things. Talking Signs should not attempt to mitigate this requirement.
A temptation is to consider that, as one immediately enters an airport, one should be able to stand in one spot and know where restrooms, check-in counters, and security stations are located. I don't think this is technically feasible, nor do I think it is a good idea. What I want from Talking Signs is to hear what various desk areas and doorways are for with out feeling doors for raised signs and without having to ask every proprietor what his station is. I want to walk around and shop in a shopping center, not stand in one place and sample a directory from a distance; I could telephone a passive directory.
Anyone reading this magazine knows how to transmit infrared beams for hundreds of feet. I am asking you now, please don't, unless the circumstances absolutely demand it. If a supermarket entrance needs to be picked up from the middle of a one-acre parking lot, fare enough, but please don't carelessly splatter high-powered signals around whose mere reflections off nearby surfaces will look like legitimate signs coming from heaven-knows where. Keep the power at a useful minimum. Thanks.
The editor would like to thank Albert Alden for his design of the front-end circuitry and for his patient troubleshooting on other aspects of the receiver circuits. Also deserving of mention is Erich Sutter, Ph.D., who designed the very first FM receiver for this project, and whose clever demodulator circuit has led to a simple design.
Hats off to Mr. Jules Madey, who designed the commercial receiver now being marketed by Love Electronics; it seems unbeatable. At this writing, the purchase price is $250--subsidies and other mechanisms to provide them for users will no doubt evolve.
Specially wound ferrite filters, as well as elaborate tuning procedures, make the commercial units more suitable for factory production. However, as I saw the circuits published here approach approximately 70 percent of the range of the Love Electronics receivers, it was the editor's decision to "go to press!" and publish them.
The choice of infrared was made because solid-state optical devices are very efficient at infrared wavelengths. Although wavelengths emitted by various LEDs differ slightly for different types, transmission generally falls in the band between 925 and 950 nanometers.
Luckily, other uses for infrared communications--remote controls and cordless personal listening systems--have led to technologies that benefit our uses. For example, Kodak makes a wonderful optical filtering material (Kodak Wratten Filter No. 87C) which passes 80 percent of infrared energy while blocking all but 1/2 percent of visible light. Efficient photoreceptors (such as the Siemens BPW34 photo diode) have also come into being.
With optical attachments, it is also possible to enhance, by directing more energy to photo diodes, the sensitivity of receivers. The simplest optical attachment is to affix a cone (a sort of light funnel) of reflective material to the front of the receptor array. Known as "nonfocusing optics," light rays reflecting off the inner sides of this "funnel" bounce back and forth until they find their way to the apex of the cone.
A reflective cone for a nonfocusing optical system can be fashioned by lining a cardboard form with aluminum foil. Dr. Bill Crandall at Smith-Kettlewell has had good luck with 20-degree cones of perhaps 2 inches in length. Remember, though, that the longer the cone is, the more restrictive is the acceptance angle of the receiver.
The effect of ambient light is noise in the receiver front end. In extreme cases, this "noise" will trip the squelch. The contribution of noise has the effect of reducing the sensitivity, and thus reducing the range.
This effect would appear devastating if ambient light were the only source of noise; the "range" of a sign would be tremendous in the dark and disappointing in daylight. Given a noise floor from other sources, the variance in range for daylight and darkness is not so extreme.
In bright sunlight, however, improvement in receiver performance can easily be had if you shade its forward end with your hand. I tape a business card--secured at one end--to the back of my receiver; it can be flipped forward so as to extend beyond the front of the instrument where it casts a shadow over the input window. Most of the time I don't need this little shade, and I flop the card down flat against the back of the receiver where it remains out of the way.
As will be seen from the circuitry, we gained a great deal of sensitivity by using several receptors in parallel (although photo diodes themselves contribute noise, and one reaches a point of diminishing returns).
Receiver Circuit Design
The front-end circuitry can be seen as three elements working together:
1. The photo diodes--6 in parallel--are back-biased. Light on their semiconductor junctions causes them to draw current which is proportional to the "power" (in watts per steradian) impinging on them.
2. Acting as current sources, the photo diodes work into the summing junction of an op-amp, creating a so-called "transimpedance amplifier." "Light current" in results in an output voltage.
3. Finally, an integrator/current sink in the feedback loop subtracts current from the summing junction of the op-amp in accordance with ambient light.
4. The resultant input system is a high-pass filter which permits high-frequency pulse trains of light to produce voltage swings at the output. With corrective feedback, this system ignores DC and low-frequency (60Hz) incoming light signals. The response of the system is a single-order high-pass filter whose upper frequency limit is nearly 60kHz.
From here on, the battle is against noise in the front-end system--noise sourced by the photo diodes, noise present in the summing junction of the amplifier, noise contributed by the resistors, and noise possibly inherent in the feedback current source.
Luckily, this noise is broadband in nature, and it can be combatted with filters. The clever Mr. Alden decided that the best choice of filter to follow the front-end circuitry would be a 2nd order low-pass filter which is underdamped. With the front-end system itself being a high-pass filter, this added low-pass filter rejects noise above the frequency of interest. Further, with the low-pass filter being underdamped, a peak in its response is created at 25kHz, and this introduces gain at the signal frequency.
The output of this filter looks fairly sinusoidal when a squarewave LED signal is presented to the photo diodes. There is still a substantial "noise floor" (random activity of perhaps 30mV peak-to-peak), but the filter does not add to this, and the signal gain is about four-fold when compared to the naked output of the input system.
A good squelch circuit turned out to be possible using an LM311 comparator. Its open collector output permits operation of the bizarre RC filter; this RC circuit determines the "hang time" of the squelch. The comparator's input is looking for negative excursions of the signal, and when it triggers, it discharges the 5uF capacitor and makes the receiver operable.
Two FM demodulators are presented here. One uses a phase-locked loop (PLL), and the other uses a one-shot to create a pseudo pulse-width modulation which is fed to an integrator. It was hoped that the PLL demodulator would be more noise-immune than the one-shot/integrator system, but performance of the two seems comparable. The main difference turns out to be that the one-shot system requires adjustment and the PLL system does not (or at least, trimming is less critical--see "Calibration"). On the other hand, the demodulator using the one-shot takes up less room on the board.
The PLL chip requires a signal of more than 400mV peak-to-peak. Therefore, an "amplifier"--rather looking like an active clamp--having cross-connected diodes as its feedback element follows the low-pass filter. The output of the op-amp swings 0.6V either side of the reference voltage of the circuit.
I call the one-shot/integrator system the "Erich Sutter Demodulator." It uses a 555 in a circuit much like a tachometer. Connected as a one-shot, the 555 fills in most of the period of the 40-microsecond input signal; the one-shot timing is set to perhaps 32 microseconds. The "off-time" pulses of the 555 are integrated to retrieve the audio signal. In other words, frequency excursions of the transmitter cause positive-going pulses from the 555 which range from very narrow to as wide as 15 microseconds. These pulses, in turn, feed an RC integrator (a low-pass filter), and the integrated output is an analog signal which is inversely proportional to the frequency of transmission.
No matter which demodulator is chosen, the output is an LM386 audio amplifier. In the optional volume control circuit, you will notice that the lower end of this control goes through a resistor to ground; this was done so that the receiver can never be turned all the way down by mistake, causing you to miss the presence of a sign.
Profound decoupling everywhere in the circuit is required to keep high-current devices from sending unwanted feedback signals to sensitive parts of the circuit. This is why three VCC points have been created.
You will notice the prevalence of 12K 5% resistors. This value just "worked out" in conjunction with standard capacitor values in the filter and PLL circuits. Since resistors are often available only in large quantities, I show 12K units elsewhere--for pull-up resistors and the like. For some of those applications, 10K resistors would do just as well. The places where the value 12K is critical are: pin 5 of the 3240 under-damped low-pass filter, pins 11 and 12 of the 4046 PLL, and the "loop filter" off pin 13 of the 4046 PLL.
The input op-amp must be a fast and quiet device. The best one tried here was the PMI OP37GP. However, the Texas Instruments TL071 works almost as well. Since noise is a vital consideration to the receiver's range, use the best you can find. RCA CA3240 dual op-amp packages were chosen for their "speed," but also for the ability to bring their outputs very near ground.
Decoupling Points and Voltage References
The negative side of the 9-volt battery is grounded. The positive of the battery goes through a normally open pushbutton switch to VCC1; VCC1 is bypassed to ground by the parallel combination of 0.1uF and 220uF (negative of the electrolytic at ground). VCC1 goes through 100 ohms to VCC2; VCC2 is bypassed to ground by the parallel combination of 0.1uF and 100uF (negative of the electrolytic at ground).
VCC2 goes through 2.2K to the cathode of a 1N751 (5.1V) zener; its anode is grounded. Across the zener are two 47K resistors in series, the junction of these being bypassed to ground by the parallel combination of 0.1uF and 10uF (negative of the electrolytic at ground). The junction of these resistors is the 2.5V reference called VREF. (The 5.1V point at the cathode of the zener has no purpose yet; I am thinking about it.)
VCC1 goes through a 10-ohm 1/2-watt resistor to VCC3; VCC3 is bypassed to ground by 220uF (negative of this cap at ground). [Where this receiver is to deliver low-frequency tones to a tactile transducer, this bypass should be increased to at least 470uF.]
Input Stage With Negative Feedback for Ambient Light
An OP37GP op-amp (originally made by PMI) has pin 4 grounded and pin 7 going to VCC2. A companion CA3240 has its pin 4 grounded, with pin 8 going to VCC2.
Pin 3 of the OP37 and pin 3 of the 3240 go to VREF (2.5V).
Between pins 6 and 2 of the OP37 is a 2-inch-long "gimmick capacitor"; this was necessary to keep the op-amp from oscillating, and the capacitance value is perhaps 1.8 picofarads. (A "gimmick capacitor" is made by twisting two pieces of insulated hook-up wire together--in this case, cutting the assembly to a 2-inch length.) Also between pins 6 and 2 is a high-value metal-film feedback resistor; we put two 10-megohm units in parallel to create 5meg. (Using one 10-megohm resistor appeared to gain no advantage--probably because of the gimmick capacitor--and for this reason, we suggest that the value be no greater than 5.6 megohms.)
Pin 2 of the OP37 goes to the anodes of 6 photo diodes (Siemens BPW34) connected in parallel. Their cathodes go to VCC2.
Pin 6 of the OP37 goes through a 1meg metal-film resistor to pin 2 of the 3240. Pin 1 of the 3240 goes to the base of a transistor, 2N2222. (A low-noise transistor was tried--the 2N5089--with no improvement.) The emitter goes through 1K (metal-film) to ground; this emitter also goes through 0.022uF to pin 2 of the 3240. (This section of the 3240 comprises the feedback integrator.) The collector of the 2222 goes to pin 2 of the OP37.
The output of the input stage is pin 6 of the OP37.
Underdamped Low-Pass Filter
Pin 6 of the OP37 goes through two 12K resistors in series to pin 5 of the CA3240. Pin 5 goes through 47 picofarads to ground. Pins 6 and 7 of the 3240 are tied together and go through 0.0047uF to the junction of the two 12K resistors. Pins 6 and 7 are the output of the filter stage.
An LM311 comparator is used. Pins 1 and 4 are grounded; pin 8 goes to VCC1. Pin 2 of the 311 goes to the output of the underdamped filter, pins 6 and 7 of the 3240.
Pin 3 of the 311 goes to the arm of a 50K trim pot. The counterclockwise end of this trim pot goes to VREF. The clockwise end goes through a 1meg metal-film resistor to ground.
Pin 7 of this 311 goes through 12K to VCC1; pin 7 also goes to the positive end of a 5uF capacitor, the negative end of which is grounded.
The Erich Sutter Pulse-Width Demodulator
An LM311 has pin 4 grounded and pin 8 going to VCC1. Pin 7 goes through 12K to VCC1. Pin 2 goes to VREF; pin 3 goes to the output of the filter, pins 6 and 7 of the 3240. Pin 1 of this 311 goes to pin 7 of the 311 used for the squelch.
On the LM311 associated with this demodulator, pin 7 goes to pin 2 of a 555. Pin 1 of the 555 is grounded; pins 4 and 8 go to VCC1. Between pins 1 and 8 is 0.1uF.
Pins 6 and 7 of the 555 are tied together and go through 0.0047uF to ground. Pins 6 and 7 also go through 2.2K, then through a 5K rheostat to VCC1.
Pin 3 of the 555 goes through 1K, then through 0.047uF to ground. The junction of this resistor and capacitor is the audio output from the demodulator, and this goes through 220K, then through 12K to ground.
Where a volume control is desired, the output of the demodulator goes through 220K to the top of a 10K volume control; the bottom of this control goes through 2.2K to ground.
PLL Detector Circuit
An additional CA3240 op-amp is required. Pin 4 of the 3240 is grounded; pin 8 goes to VCC2. Pins 3 and 5 go to VREF. Pins 6 and 7 are tied together.
Pin 2 of this 3240 goes through 1K to the output of the filter--pins 6 and 7 of the 3240 associated with the input circuitry. Between pins 1 and 2 are "cross-connected" diodes (the cathode of the first goes to the anode of the second; the cathode of the second goes to the anode of the first). Pin 1, the output of this section of the 3240, goes through 0.01uF to pin 14 of an RCA CD4046 PLL.
Pin 8 of the CA4046 is grounded; pin 16 goes to VCC1. Pin 5 of the 4046 goes to pin 7 of the squelch comparator.
Between pins 6 and 7 of the CA4046 is a 0.0068uF Mylar capacitor (see calibration). Pins 11 and 12 of the 4046 each go through 12K to ground.
Pin 2, the output of the so-called "narrowband phase detector," goes through 12K, then through 2.7K in series with 0.0022uF (Mylar) to ground. The junction of these resistors goes to pin 9, the VCO input of the 4046. (This network comprises the "loop filter"--see SKTF, Summer 1982.)
The audio output of this detector is pin 10. Being an emitter follower, pin 10 goes through 12K to ground.
If a volume control is used, pin 10 goes through 220K to the top of a 10K volume control; the bottom of this control goes through 2.2K to ground. If a volume control is not included, pin 10 goes through 220K, then through 12K to ground.
Pin 6 of an LM386 goes to VCC3. Pin 7 is bypassed to ground by 22uF (negative of this cap at ground). Pins 2 and 4 of the 386 are grounded. Pin 5 is bypassed to pin 4 (located close to the chip) by 0.1uF. Pin 3, the input, is bypassed to pins 2 and 4 (located close to the chip) by 0.01uF. Pin 3 goes through 0.1uF to either the arm of the volume control or the junction of the 220K and 12K resistors (see above).
Pin 5 of the 386 goes to the positive end of a 100uF capacitor, with the negative end going through the speaker to ground. [If low-frequency tones are to be presented by a tactile transducer, this capacitor should be increased to 470uF.]
Receiver Modifications for the Deaf-Blind
The modifications presented here are intended for cases where low-frequency tones have been mixed with audio information on the transmitter. Although this tactile frequency is perhaps not optimal (250Hz would be preferred), subaudible tones allow very effective rejection of audio information which would appear as noise on a tactile transducer.
Two modifications to the receiver are made. The first is to construct a simple tactile transducer by cutting a hole in front of the speaker. If you wish, a small portion of ping-pong ball or other plastic capsule can be cemented to the speaker cone; this should be attached near the voice coil where motion is maximum. The second modification is to employ profound low-pass filtering to eliminate audio chatter on the transducer.
An active 2nd order low-pass filter follows the demodulator. Because the phase-locked loop detector leaves a free op-amp on the board, this version of the receiver is used. Beyond this, another stage of filtering is accomplished by shunting the bottom half of the audio voltage divider with capacitor. The resultant circuit rolls off at 18dB per octave above 100Hz.
Low-Pass Audio Filter for Presentation of Subaudible Tones
(Note from above circuitry that the CA3240 associated with the PLL has an unused op-amp. Pins 6 and 7 are tied together; pin 5 is disconnected from VREF for this circuit.)
Pin 10 of the CD4046 PLL goes through two 120K resistors in series to pin 5 of the second CA3240. Pin 5 goes through 0.01uF (Mylar) to ground. Pins 6 and 7 go through 0.02uF to the junction of the two resistors (this is made of two 0.01uF Mylar capacitors in parallel). The output of this 2nd order filter, pins 6 and 7, goes through 220K, then through 12K to ground. The 12K resistor is shunted by 0.47uF (Mylar or ceramic). The junction of this voltage divider is capacitively coupled to pin 3 of the LM386 as before. As mentioned, the coupling capacitor to the speaker is increased to 470uF, and the supply bypass for the 386 is also increased to 470uF.
If more output is required, the gain of the LM386 can be increased by connecting 22uF in series with a resistor between pins 1 and 8 of the 386 (positive of this cap toward pin 1). Perhaps resistor values of 1K or 470 ohms should be tried; with this resistor at 0 ohms, the boost in gain is 30dB.]
An adjustment common to both is the squelch. For initial testing of the receiver circuits, this squelch should be "wide open"--the potentiometer should be turned fully counterclockwise. This should produce a rushing sound in the output of the receiver. When the receiver is pointed toward a test transmitter, "partial quieting" or perhaps a great reduction in receiver noise should be noted.
The Erich Sutter Demodulator is easy to adjust using an oscilloscope. Present the receiver with an unmodulated transmitter signal. Connect the oscilloscope between pin 3 of the 555 and ground; square pulses should be noted. The idea is to adjust the pot off pins 6 and 7 so that the period of this waveform is 40 microseconds, but the pulses (positive-going) should be narrow--perhaps 8 microseconds in width.
This demodulator can also be adjusted by ear. Turn the rheostat associated with pins 6 and 7 of the 555 to zero resistance. Present the receiver with a modulated transmitter signal. Modulation should be heard, but it may be quite faint. As the resistance of the rheostat is increased, the modulation level will increase up to a point. Further advancement of the rheostat will cause a brief noise burst to occur in the receiver, and modulation can once again be heard. However, this setting results in the 555 one-shot dividing the received signal frequency by 2. A 12.5kHz tone should be heard superimposed on the audio. Further advancing this pot may result in the 555 dividing the signal frequency by 3 or 4, etc. As the rheostat is advanced from zero resistance, the first clear presentation of audio should be sought; with the rheostat too low, the audio will be faint, and with the rheostat advanced too far, distortion of the audio will be the result.
The PLL circuit may need no adjustment. However, our system required some trimming of the VCO capacitor. Present the receiver with an unmodulated transmitter signal and look at the voltage on pin 10 of the 4046 with a DC voltmeter. This voltage should be in the range of perhaps 2 to 7 volts. If it is high, the capacitor between pins 6 and 7 should be reduced in value; if the voltage on pin 10 is low, this oscillator capacitor can be "padded" by attaching another small capacitor in parallel with it.
The squelch can now be adjusted. Turn off all test signals (transmitters) and make sure that the receiver is not in the presence of bright ambient light (particularly fluorescent light, which contains a lot of amplitude noise). Adjust the squelch potentiometer on the LM311 until the rushing noise in the receiver turns quiet. Then, move the transmitter around--point it at various lighting sources--to see if external sources caused the squelch to open and produce chatter in the loudspeaker. If necessary, advance the squelch slightly so that this chatter is not objectionable. Next, turn a test transmitter on and off and see that the squelch causes the receiver to shut down a fraction of a second after a signal is removed. If the rushing noise is heard for half a second or more when a transmitter is turned off, the squelch should be "tightened" by advancing the control slightly. The effective sensitivity of the receiver is reduced as the squelch is tightened, so the point is to "loosen" the squelch enough so that outbursts from the receiver are tolerable.
Input, Squelch, Power Amp (Excluding Demodulator)
Resistors (1/4-watt 5% unless otherwise specified):
- 1--10 ohms, 1/2-watt
- 1--100 ohms
- 1--1K metal-film
- 2--1 megohm metal-film
- 1--5 (or 5.6) megohm metal-film (made from some combination of ready values, perhaps two 2.2meg and one 1meg in series)
- --50K 10- or 20-turn
- 2--220uF 10V electrolytic
- 2--100uF 10V electrolytic
- 1--22uF 10V electrolytic
- 1--10uF 10V electrolytic
- 1--5uF (or 4.7uF) 10V electrolytic
- 5--0.1uF disc ceramic
- 1--0.022uF Mylar
- 1--0.01uF disc ceramic
- 1--0.0047uF disc or Mylar
- 1--47pF disc
- 1--1N751 5.1V zener diode
- 6--Siemens BPW34 photo diodes
- 1--PMI OP37GP or Texas Instruments TL071
- 1--RCA CA3240
- 1--piece of Kodak "Wratten Filter, No. 87C"
- 1--momentary switch, lever or pushbutton
- Battery, cabinet, etc.
Erich Sutter Pulse-Width Demodulator
Resistors (1/4-watt 5%):
- 1--5K 10- or 20-turn trim pot connected as rheostat
- 1--10K volume control (optional)
- 1--0.1uF disc ceramic
- 1--0.047uF disc or Mylar
- 1--0.0047uF disc or Mylar
Resistors (1/4-watt 5%):
- Optional 10K Volume Control
- 1--0.01uF disc ceramic
- 1--0.0068uF disc or Mylar
- 1--0.0022uF Mylar
- 1--RCA CA3240
- 1--RCA CD4046
Modifications for Tactile Output
[Note: This requires a spare section of a CA3240 op-amp. The PLL demodulator has one available.]
Resistors (1/4-watt 5%):
- 1--gain boost, 470 ohms or 1K (optional)
- 2--470uF 10V electrolytic
- 1--gain boost, 22uF 10V electrolytic (optional)
- 1--0.47uF Mylar or 10V electrolytic
- 3--0.01uF Mylar
Most of the semiconductors can be gotten from Mouser and Newark. However, only Newark sells the OP37GP. The Mouser listings are as follows: CA3240AE, LM311N, NE555N, 1N914 ($8 for 100) and 1N751A. Newark part numbers are: CA3240AE, NE555N, LM311N, 1N914A, and 1N751A.
The metal-film resistors have very different part numbers between the two suppliers. For one thing, 10meg metal-film resistors seem to be harder to get. I will give the old Mouser number; the stock of 10meg units may come and go. In Newark, 10meg units are not listed.
The Mouser part numbers are 29MF250-1K, 29MF250-1M, 29MF250-2.2M, and (temporarily not listed) 29MF250-10M. Newark numbers are 50F8302 (1K), and 50F8317 (1 meg).
The Siemens BPW34 receiver diodes are available from Marshall Industries. In the U.S., call 800-522-0084; outside the U.S., call 818-307-6000. Transmitter LEDs are available here also.
The infrared plastic filter material is quite expensive--about $12 for a 3-inch square. It is called "Wratten Filter 87C." It seems to be available from major camera supply houses; however, you can inquire about local reps by calling Eastman Kodak in Henrietta, New York, at 800-242-2424, or 716-724-4000 for overseas callers. In Canada, call the Toronto Center at 800-465-6325 or 416-766-8233.
• Mouser Electronics, P.O. Box 699, Mansfield, TX 76063; phone 800-34-MOUSER.
• Newark Electronics, 1730 S. Amphlett Blvd., San Mateo, CA 94402; phone 415-571-5300.
INFRARED TRANSMITTERS COMPATIBLE WITH THE TALKING SIGNS (TM) SYSTEM -- FROM NAME TAGS TO BLOCKBUSTERS
Many configurations are possible. The "range" and radiation pattern of a transmitter may be selected with great flexibility. There being many factors involved, calculations would be difficult; however, general principles, along with these examples, can serve as good guidelines.
If the emitter were a point source, the power impinging on the receptor would vary as the square of the distance between the two. This is called "the square law." If you note the power at Distance, then double this distance, the power will drop to one-fourth its former level (a loss of 6dB).
A way to "defeat?" the square law is with optics. If we were to put our point source at the focal point of a lens (perhaps a 2-inch-diameter lens), a beam of parallel rays of light, not greater than 2 inches in diameter, would emerge and travel for an infinite distance. For such a beam, loss of power and effective range would be determined by diffusion due to particles and moisture in the air.
This 2-inch beam comprises no sign. The fact is that it could be coming from anywhere and pointing any which way, and this makes such a beam impossible to find--intercepting it would be a chance happening. One could compromise and create a source whose beam is confined to a cone, an ellipse, or even a fan shape. This would indeed concentrate power in the forward direction and increase the "range." (Fortunately, LEDs are these sort of compromise sources. To begin with, their tiny chip emitters are not point sources. Moreover, common package styles mold a converging lens on the front end. The convexness at the front end of a plastic capsule determines the breadth of the beam.
What does convergence of light rays into a conical shape actually accomplish? Well, compared to a point source, no energy is wasted by being sent behind the source. From the observer's point of view, light impinging on the emitter comes from an apparent point source located behind the LED--farther away. The receptor, as it is moved farther and farther away, sees the square law being obeyed, but it just thinks that the overall distance to the "point source" is farther away.
The sad fact is that this apparent source is not very much farther away--perhaps only a few millimeters behind the diode whose converging lens is not very convex. Thus, you are not defeating the square law at all, but you are salvaging significant amounts of power which would otherwise go elsewhere. For example, because of its narrow beam, the SFH484 diode listed here can deliver seven times the power per unit area as the LD271. This means that the person carrying the receiver can step back 2.6 (the square root of 7) times as far from the SFH484 as he can from the LD271 before the signal drops out in his receiver. (This assumes equal currents through each type.)
The specifications of LEDs include a feature of the radiation pattern called the "half-angle," measured as plus/minus so many degrees. Measured with respect to the axis through the LED, the half-angle is the point at which the radiated power is down 3dB (one-half the power seen in the forward direction in line with this axis). Therefore, since the 3dB half-angle of an LD271 is listed as plus/minus 25dg, the half-power points describe a 50dg cone. For the SFH484, half-angles are listed as plus/minus 8dg, creating a 16dg cone of radiation within the 3dB points. The editor's favorite package, the LD273 dual LED, has an elliptical pattern; with its leads oriented left and right, the half-power points are at plus/minus 25dg as you move left and right, and with the leads one above the other, the half-power points are plus/minus 15dg as you move left and right.
While doubling the current through an LED does cause the power to nearly double (there being some loss near maximum drive current), the package dissipation increases quite a bit. This is true because the voltage drop may differ as much as 30% for different drive currents.
The way to achieve a precisely calculable increase in power is to count the lamps. For example, if you need to double the detectable range, quadruple the number of lamps. Since even the dual lamps are only about 70 cents each, the thought of using 100 of them to send a signal across a six-lane street is not so unbearable. From the individual builder's side of the coin, dividing the number of lamps by four only cuts the distance in half; this is why you can transmit across a 15-foot sidewalk with only four lamps.
There is one more monkey wrench that determines how arrays of lamps are constructed. As stated in the receiver article, the current sourced by the photo diodes is proportional to the power they receive. Later on in the receiver circuit, this means that the voltage, not the electrical power, is proportional to the power of the light received. Thus, when the power input drops by 3dB, the electrical signal drops to half; this is electrically equivalent to a 6dB loss.
The effect of power-to-voltage conversion on range does not matter much, since we are merely speaking of the distance where the signal is at a detectable threshold (the square law still applies). Where the effect in the receiver does matter is the off-axis signal loss at a given distance.
The fact is that at the 3dB half-angle, the receiver's detector sees a loss of 6dB. The 3dB points, as far as the receiver is concerned, are The specified LED half-angles divided by the square root of 2--those angles multiplied by 0.707. Thus, whereas the rating of a diode might lead you to believe that the visible angle--within 3dB--is plus/minus 25dg, it is actually plus/minus 18dg.
Suppose you were designing a small cylindrical sign for the top of a small kiosk so that it could be seen from all directions. In order for there not to be weak spots between lobes of the signal as you walk around the kiosk, you must insist on a substantial overlap in radiation pattern from one diode to the next. The resultant array would probably contain two or three bands of LEDs staggered so that they can be placed every 15 degrees or less.
Before we leave the topic of radiation pattern, I owe it to Dr. Erich Sutter to put forth the following ideas. The radiation pattern, much as is done in designing radio antennas, can be shaped to maximize the power in a given direction. The designer does not have to leave all the optics up to lenses in the capsules of the emitter and receptor elements. Lenses can be placed in front of transmitters, or fitted to the receivers to change their characteristics.
Erich Sutter has long advocated the use of cylindrical lenses to "flatten" the radiation pattern of transmitters. A cylindrical lens (usually a half-round piece of Lucite) converges the rays in one direction; thus, oriented horizontally in front of the transmitter, radiation illuminating the ground and lost to the sky would be redirected into a horizontal fan shape. In general, the emitter should be within the focal distance of the lens; if it were at the focal point, rays would be converged in such a way that the radiation pattern would be as thin as the partial cylinder is wide, which would make it hard to find.
Cylindrical lenses are easy to obtain. Many catalogs of low vision aids list plastic ones which are intended to magnify a line of print; these can be cut into segments for your experiments.
Transmitter Circuit Elements
The heart of each transmitter is a 555 timer in the free-running configuration. Attempts are made to achieve a squarewave with 50 percent duty cycle; this is done so that the commercial Love Electronics receiver, a tuned radio-frequency (TRF) circuit, can make use of the large amount of energy at the fundamental frequency of oscillation. (The receiver circuit using the one-shot demodulator needs only leading edges of a pulse train to work. However, please stick to a transmitted signal that I can hear with my commercial TRF receiver; I want to read your signs when I come by.)
The period of oscillation is 40 microseconds (a frequency of 25kHz). The deviation is held to within about 10 percent (plus/minus 2.4 kilohertz). Frequency modulation is accomplished by capacitively coupling an audio signal of perhaps 2 volts peak-to-peak to pin 5 of the 555, the so-called "voltage control pin." (Where current is demanded from the 555, the editor has chosen not to use its output to operate the RC network so as to achieve better frequency stability.)
The allowable peak current through LEDs varies with the length of the driving pulses, even if a 50 percent duty cycle is specified. While the LD271 should only be handling 100mA for pulses of 1 millisecond or more, it allows us to drive it with 200mA for the 20-microsecond pulses of our 25kHz squarewave. The dual LED package, the LD273, should be driven at no more than 175mA for these 20-microsecond pulses (probably specified because of package dissipation).
The voltage across an LED changes somewhat with drive current. The LD273 dual package has a 2.2-volt drop at 10mA, 2.7V appears across it at 100mA, and at 175mA, the voltage drop exceeds 2.8V. Your editor just considers all LED voltage drops to be 1.25V--2.5V for each dual package.
LEDs should not be driven by voltage sources; they are properly driven with current sources. Since in this application the drive signal is either on or off, it is safe to consider a voltage source through a resistor as comprising a passable current source. For this to be true, however, the voltage source must exceed the sum of voltage drops in a string of LEDs by 3 volts or so.
In designing arrays, count the voltage drops across a string of LEDs and add to this the saturation voltage of the device driving them. Then choose a power supply of perhaps 3 volts beyond this sum so that a resistor can provide adequate current limiting. For example, four dual LEDs, LD273s, cannot be in one string if a 9V battery is used. Counting 2.5V per package, and with a saturation voltage across a driver transistor of 0.6V, the sum of the voltages across the semiconductors adds up to 10.6 volts. For the above example, even a 12-volt supply will not suffice--15V should be used. On the other hand, two strings of two LD273 packages can indeed be powered from a 9V battery. Both strings require only 5V each; the driver transistor, common to both strings, adds 0.6V to this. Therefore, a 9V battery leaves you with a margin of 3.4 volts.
A single driver can be common to all strings, as long as the sum of their currents times the saturation voltage does not exceed the allowable power dissipation of the driver. For example, a 2N2222 can dissipate half a watt; thus, 0.6V times .350 amps (drawn by two high-powered strings) causes only 0.21 watts to be dissipated by the transistor, which is OK.
Once you have decided on the power supply and the desired current through each string, a resistor for each string can be calculated. Sum the voltages in one string and add to this the voltage drop in the driver circuit. Subtract this sum from the highest supply voltage expected--a new battery or the peak output of an unregulated supply. This difference voltage, divided by the current in amperes through a string, will yield a quotient in ohms.
Another important design consideration is the power which is to be dissipated by the resistor. Using the formula I squared times R (the string current in amperes and the resistor in ohms) will give you the power dissipated by the resistor during the 50 percent "on time" of the signal. The steady-state power in the resistor will be half this value. Remember, though, that a carbon resistor should only be asked to dissipate half its rated power; therefore, I squared times R will yield the actual rating required of the resistor.
So that all our signs can be appreciated by one another, please take the trouble to calibrate your transmitters. The precision of a frequency counter is not required. An oscilloscope can be used to set the period of oscillation (seen at pin 3 of the 555) to 40 microseconds.
Find a friend with an oscilloscope, go to a high school or college electronics lab, or ask the proprietor of a small TV repair shop to make this adjustment (looking for the signal between ground and pin 3 of the 555). On the other hand, you may be able to use an AM radio with a crystal calibrator to count harmonics of the transmitter. Attach a few feet of hookup wire to pin 3 of the 555 to act as an antenna. You may wish to remove the modulation source; unplug the ISD chip. If the receiver has a 100kHz crystal marker generator, you may be able to adjust the transmitter so that four of its harmonics occur in the space of 100kHz.
A transmitter without modulation can be built for use as a standard. One of the circuits using a regulated 5V supply should be used. In this way, you can keep a "test signal" around the shop for calibration of future signs.
Orient the two emitters--the LED of the test transmitter and the array of the sign to be tuned--so that both signals can be heard in the receiver (producing a heterodyne). (Since the receiver is FM, the signals must be very close in amplitude so as to be heard as interference.) Then, tune the frequency of the sign's 555 for "zero beat" with the test transmitter.
Wearable Name Tags and Low-Powered Indoor Labels
Basic wearable transmitter with external audio input
Run from a 9V battery, this transmitter, using only a 555, can be modulated from a pocket radio or tape recorder. With it, a friend could identify you at close range. Another use would be as a placement marker so that you can find your way back to a specific seat. Its range is perhaps 5 feet or more, depending on the receiver.
The negative side of a 9V battery is grounded. Its positive terminal goes through a switch to the VCC line. Pin 1 of a 555 is grounded; pins 4 and 8 are tied together and go to VCC. Pins 2 and 6 are tied together and go through 220pF to ground. Pins 2 and 6 also go through 110K to pin 7, with pin 7 going through 10K to VCC.
Pin 3 of the 555 goes through 220 ohms, then through two LD273s in series to VCC; the anodes of these diodes face VCC.
The cold side of the audio cable is grounded. The hot audio lead goes through 0.47uF to pin 5 of the 555. So that this device can be used with any mono or stereo pocket device, a stereo 1/8-inch plug is attached to the other end of the audio cable; its "ring connection" is left open, the sleeve is grounded, and only the tip of the plug goes to the hot audio lead.
It may be necessary to trim the frequency by changing the 110K resistor. To make this adjustable, replace the 110K with 82K in series with a 50K rheostat.
Wearable Beacon With Emitters for Front, Back and Sides
Also modulated by the source of your choice, if you want a friend to find you from any direction, this is the transmitter to use. Walking around, a friend will spot you from a distance of 15 or 20 feet. Meanwhile, using 47-ohm resistors for each string, your battery won't last more than an hour or so. The negative of a 9V battery is grounded. Its positive terminal goes through a switch to VCC. Pin 1 of the 555 is grounded. Pins 4 and 8 are tied together and go through 100 ohms to VCC; pins 4 and 8 are bypassed by the parallel combination of 0.1uF and 100uF (negative of the electrolytic at ground).
Pins 2 and 6 of the 555 are tied together and go through 220pF to ground. Pins 2 and 6 also go through 82K in series with a 50K rheostat to pin 7. Pin 7 goes through 10K to pin 8.
Two 2N2222 NPN transistors have their emitters grounded. Each base goes through its own 1K resistor to pin 3 of the 555. Each collector drives two strings of lamps as follows.
Mounted on the transmitter board, each 2N2222 collector goes to one end of two 47-ohm 2-watt resistors. (This thing will be in your pocket; don't underrate those resistors.) Pairs of LD273 LEDs, connected in series, are then mounted onto tiny pieces of Vector Board. The backs of these boards bear safety pins like badges.
The anode ends of all the LED strings go to VCC. Each cathode end of a string goes to the free end of its own 47-ohm resistor.
Name Tag and Room Number Sign With Digitally Recorded Speech
This circuit can be run from 9 or 12 volts. A 5V regulator is required for the speech chip; this also adds to the frequency stability of the 555 oscillator. A bridge rectifier is shown so that the device can also be run from AC, such as comes from a 10V or 16V doorbell transformer. (If DC is available--from a plug-in battery eliminator, for example--the bridge rectifier can be omitted.) A 1-amp 50 PIV bridge rectifier has its negative output going to ground and its positive output going to the unregulated supply line. This supply line is bypassed by 470uF (negative of the capacitor at ground). A 7805 or LM340-5 regulator is used. The "Common" terminal of this regulator is grounded. Its "Input" terminal goes to the unregulated supply and is bypassed to the common terminal by 0.1uF. Its "Output" terminal goes to the 5V line and is bypassed to Common by the parallel combination of 0.1uF and 47uF (negative of the electrolytic at ground).
Pin 1 of the 555 is grounded. Pins 4 and 8 go to the 5V line. Between pins 1 and 8, located close to the chip, is a 0.1uF disc capacitor.
Pins 2 and 6 of the 555 are tied together and go through 0.001uF to ground. Pins 2 and 6 also go through 15K in series with a 5K rheostat to pin 3, the output.
Pin 3 of the 555 also goes through 1K to the base of a 2N2222; the emitter of this transistor is grounded. The collector goes through a resistor (from 100 to 330 ohms), then through a string of five LEDs (LD271), in series to the unregulated supply (with the anodes of these diodes all facing the supply line).
Pin 5 of the 555 goes through 0.47uF to the arm of a 5K or 10K trim pot, the bottom of which is grounded. [If low frequency tones for the deaf-blind are mixed with the verbal message, this 0.47uF capacitor should be changed to 4.7uF, negative of this cap toward the arm of the trim pot.] The top of this pot receives a signal from the modulator to follow.
Digitized Speech Modulator--
[Note: This modulator could be used with any of the circuits, and it will be referred to later.]
An ISD1012 or ISD1016 is used (see SKTF Winter and Spring 1991). Pins 1, 2, 3, 5, 6, 12, 13, and 26 are grounded. Pin 28 goes to the 5V line; pin 28 is bypassed to pin 26 by 0.1uF. Pin 16 goes through 33 ohms to the 5V line; pin 16 is bypassed to ground by the parallel combination of 0.1uF and 22uF (negative of the electrolytic at ground). Pins 7, 8, 11, 18, 22, and 25 are left open.
Two audio inputs are provided for. Pin 17 of the ISD chip goes through 0.22uF to the hot mike lead; the cold side of the mike is grounded. If the Radio Shack 270-092 capacitor mike is used, its supply lead goes through 2.2K to the 5V line, as well as going through the parallel combination of 10K and 10uF to ground (negative of the cap at ground). Pin 21 goes to the switch contact of a high-level input jack, the sleeve of which is grounded. The tip contact of this jack goes through 0.47uF to pin 20. [Where low frequency tones for the deaf-blind are added to prerecorded messages, the 0.47uF capacitor is replaced by 2.2uF (the negative side of this cap going to pin 20).]
Pin 19 of the ISD chip goes through the parallel combination of 300K and 22uF to ground (negative of the cap at ground).
An SPST "record/play" switch has one side grounded, while the other side goes through 47K to the 5V line. The junction of the resistor and the switch goes to pins 4, 9, 10, and 27. When this switch is closed, the ISD chip is in record.
Pin 23 goes through 47K to the 5V line. Pin 23 also goes to the collector of a 2N2222 whose emitter is grounded. The 2222 base goes through two 47K resistors in series to ground. An SPST "run" switch and a normally open pushbutton are connected in parallel; one side of this combination goes to the junction of the 247K resistors, while the other side of the switches goes to the 5V line. The junction of these resistors and the switches also goes through 0.1uF to pin 24, with pin 24 going through 220K to ground.
Pin 14 or 15--it doesn't matter which--goes to the top of the modulation trim pot off pin 5 of the 555.
The high-level input, pin 20, requires less signal than you would expect. The acceptable peak-to-peak value is about 100mV. It can be fed with a tape recorder if the volume is turned down very low. However, you may wish to build a 20-to-1 attenuator so that it can be fed with standard levels.
High-Power and Variable-Power Transmitters
Using the criteria for choosing resistors and series LED strings put forth at the beginning of this paper, it is trivial to design high-powered transmitters. Using Texas Instruments "actuator/driver" chips like those described in SKTF Spring 1991 (or an even beefier cousin to be outlined in this issue), whole panels of LEDs could be driven from the single 555.
The plastic DIP versions of these drivers dissipate 1 watt, or sink over 400mA per gate. All the 555 has to sink is 1.6mA per gate input. The tolerable open collector voltage is 30V for the 75452 and 75454; the tolerable open collector voltage for the 75477 and 75479 chips is 70V.
These driver chips are gates, so you can control--interrupt or switch between--various arrays, just as was done in the 5th and Market Street installations here.
A disadvantage of these drivers is the amount of supply current they draw--up to 75mA, not counting load current, when their output transistors are turned on. Therefore, they are too extravagant for use in battery or low-current-drain transmitters.
Skeletal Four-Output Driver Circuit--
Some sort of 5-pin connector--perhaps a DIN connector--may be used so that various arrays can be plugged in. One pin is common to all arrays and goes to the unregulated DC supply.
Supposing the 75477 NAND gates are used, two of these chips have pin 4 grounded and pin 8 going to the 5V line. Pin 1 of both chips goes to pin 3 of the 555. Pins 2 and 7 of each (gate inputs) are tied to the 5V line. Pins 3 and 6 of these two chips each go to their own prong on the 5-pin connector.
Adding Adjustable Power--
Since the transmission scheme is FM, amplitude modulation caused by fluctuations in the unregulated supply do not seem to make much difference. Thus, the circuit for the variable supply given here uses just an emitter follower controlled by a trim pot. This supply system will work for any of the circuits where the 555 is powered through a 7805 or other 5V regulator. It is assumed that if you employ high DC voltages, you have chosen the 75477 driver chip circuit above.
The collectors of a 2N2219 and a 2N3055 both go to the unregulated supply (12 to 40V, though this could be higher if higher voltage transistors were chosen). The emitter of the 2N2219 goes to the base of the 2N3055. Instead of the LED anodes going to the supply, they now go to the emitter of the 2N3055.
A 20K trim pot has its top end going to the unregulated supply; its bottom end goes through 4.7K to ground. The arm of the trim pot is bypassed to ground by 22uF (negative of the cap at ground). The base of the 2219 goes to the arm of this pot.
The 2N3055 may require a small heat sink, especially when the sign is adjusted for low power. For example, if your signal were to draw 200mA when the voltage drop across the transistor is 30V, the power dissipated in the 3055 will be 6 watts.
Robbing Power From Lights and Doorbells
Skeptics take note of how easy it is to find power sources floating around.
The Talking Doorbell Button--
Some doorbell chimes, especially those with moving core solenoids, draw an amp or so when the button is pressed. If you make a talking sign whose current drain is under perhaps 75mA, then connect this in parallel with the doorbell button, it can draw its supply current through the doorbell circuit without ringing the chime.
Buzzer-type doorbells are more efficient, and these may chatter when even this small current is drawn through them. You should listen for this or other objectionable hum in any electromechanical doorbell before drilling holes for the installation.
The speech chip and its regulator will draw about 35mA, so you don't have much current to play with. However, many doorbells run off 16V AC, which would give you about 20V of filtered DC to run six LD273s in one string.
The terminals of the doorbell button go to the input of a bridge rectifier. The output of the bridge, filtered by 470uF, goes between the unregulated supply line and transmitter ground. The 555 drives a 2N2222. The collector of this goes through 100 ohms, then through six LD273s in series to the unregulated line.
Adapting a Light Fixture to be a Sign--
How can you run constant power to a transmitter screwed into the porch light socket and still have an operable porch light that can be switched off? The ability to turn the porch light on and off from inside the house must still be preserved.
One modification to the house wiring is necessary. A 4-amp 400 PIV diode is connected across the porch light switch--the anode of this diode facing the porch light.
In the adapter (screwed into the porch light socket and having the bulb screwed into it), there is a low-voltage DC relay. Its positive coil terminal goes to the terminal of the socket fed by the porch light switch and the anode of the diode. The negative side of the relay coil goes through an appropriate dropping resistor to the other side of the socket. Across the relay is a 0.5 amp 400 PIV diode, its cathode facing the switch. Thus, when the switch is open, a half-wave DC signal is available at the porch light socket; however, this is of such polarity as to not energize the relay coil. When the switch is closed, AC is available at the socket, and the relay coil will be energized by half-cycles that were missing when the switch was open. The porch light bulb is in series with normally open contacts of the relay; the light bulb, in series with the relay contacts, is across the porch light socket.
Finally, the infrared transmitter is powered by a 1-amp 400 PIV bridge rectifier whose input terminals are across the porch light socket. The negative side of the bridge goes to the transmitter's ground. The positive output of the bridge is bypassed to transmitter ground by a 470uF 250V capacitor (negative at transmitter ground). The positive bridge output also goes to the collector of a high-voltage NPN power transistor (such as a horizontal driver transistor from a television set).
The positive output of the bridge also goes through a 5.6K 3-watt resistor to the cathode of a 24V 1-watt zener; the anode of this zener goes to transmitter ground. The base of the power transistor goes to the cathode of the zener. The emitter of the transistor goes to a 23V unregulated line on the transmitter.
The output of the transmitter's 555, pin 3, goes to pins 1, 2, 6, and 7 of a 75452 or 75454 TI driver chip. Pin 4 of this driver goes to transmitter ground; pin 8 goes to the transmitter's 5V line. Pins 3 and 5 of the driver chip are outputs; each can accommodate a string of 6 LD273 dual LEDs in series with 47 ohms. (This will require that the pass transistor dissipate 10 or 12 watts.)
The Smith-Kettlewell Exhibitor's Infrared Transmitter With Dual Message Capability
This exhibitor's sign was designed for Jerry Kuns for a multiplicity of uses. First, it is capable of driving arrays with enough power so that his second-story business office, "Talk of the Town," can be received from the sidewalk below. It is portable; having a bridge rectifier input, it can be powered from a 24V battery or from a 16V doorbell transformer. A 5-pin DIN socket allows the transmitter to be unplugged from his office display and taken on the road.
A microphone on the front panel permits messages to be rerecorded at will. There are two "message slots" provided for in the record/play system: When the record switch is in one position, the ISD1016 chip can record from the beginning of its memory. With the mode switch in the other record position, the first part of the memory is protected and a second message begins recording later on in memory.
The length of the initial memory slot can be selected by inserting jumper wires (small segments of 26-gauge solid wire) into an 8-pin DIP socket. The initial message can be selected in 1.6-second steps from zero time to the entire 16 seconds available on the ISD chip. With a jumper in the third position of the socket, a 6.4-second message slot is reserved at the beginning; this gives Mr. Kuns time to record the permanent message, "Talk of the Town, 1182 Market Street, Suite 211." From here on, other messages--specials of the week or other announcements--can be tagged on to the basic address.
The device is housed in a 5- by 9-inch project cabinet. On the rear panel are binding posts for power, a fuse, the 5-pin socket for the LEDs, and a high-level input jack for high-quality recording from taped messages. So that the recording can be tested, a speaker and monitor switch are mounted on the top panel. Also on the top are the pushbutton and the Radio Shack capacitor microphone which is fitted into a 9/16-inch rubber grommet. The mode switch is a 4-position rotary switch which includes Record 1, Record 2, Off, and Play. Finally, holes in the front apron of the box provide access to modulation level and frequency trim pots.
Circuit for the Smith-Kettlewell Exhibitor's Sign--
A 3-pole 4-position switch is used. On one pole, positions 1, 2, and 4 are tied together and go to one of the power terminals (one of two binding posts on the rear panel). The arm of this pole goes to one input of the bridge (a 4-amp 100 PIV bridge rectifier). The other power binding post goes through a 2-amp fuse to the other input of the bridge.
The negative output of the bridge is grounded; the positive output is bypassed to ground by 2200 microfarads (negative of this cap at ground). The positive output of this bridge will be called the 24V line.
A 5V uA7805 (with heat sink) has its "Common" terminal grounded and its "Input" going to the 24V line. This input is bypassed to the Common by 0.1uF. The output of the regulator, also bypassed to Common by 0.1uF, is the 5V line. The 5V line is bypassed to ground by 220uF (negative of this cap at ground).
The 5V line goes through 2.2K, then through 10K to ground; the junction of these resistors is bypassed to ground by 10uF (negative of the cap at ground). The junction of the resistors supplies the microphone (Radio Shack 270-092.
The ISD chip has pins 1, 2, 3, 12, 13, and 26 grounded. Pin 28 goes to the 5V line; pin 28 is bypassed to pin 26 by 0.1uF. Pin 16 goes through 33 ohms to the 5V line; pin 16 is bypassed by the parallel combination of 0.1uF and 22uF (negative of the electrolytic at ground).
Pin 15 of the ISD chip goes through 8.2 ohms, then through an SPST toggle switch in series with the speaker to pin 14.
Pin 17 goes through 0.22uF to the hot mike lead; the cold mike lead is grounded. The mike power lead goes to the above voltage divider as described.
Pin 19 goes through the parallel combination of 300K and 22uF to ground.
Pin 21 goes to the switch contact of a high-level input jack; the tip contact goes through 0.47uF to pin 20. The sleeve of the jack is grounded.
A 74HC257 multiplex chip has its pin 8 grounded and its pin 16 going to the 5V line. Its A inputs, 2, 5, 11, and 14 each go through 47K to ground; these also go to pins 1, 2, 3, and 4 of an 8-pin socket (used for message jumpers).
As for the 257's B inputs: 3 and 6 are grounded; 13 and 10 go to the 5V line.
On the 1016, pins 5, 6, 9, and 10 each go through 47K to ground. These go, respectively, to the 257's pins 4, 7, 9, and 12.
On the 1016, pins 23, 24, and 27 each go through 47K to the 5V line. Pin 24 goes to the anodes of two 1N914s; the cathode of one goes to pin 23, and the other cathode goes to pin 27. Pins 27 and 3 of the 1016 both go to pin 1 of the 74HC257 (this is the "record" line).
The second pole on the 4-position switch has its arm grounded. Positions 1 and 2 are tied together and go to pin 27--the record line. Position 4 goes to pin 23.
On the 74HC257, the tri-state pin (pin 15) goes through 47K to ground. Pin 15 also goes to position 1 of the third pole on the switch; the arm of this pole goes to the 5V line. (The other positions are left open.)
A normally open message record button goes from pin 23 of the 1016 to ground.
As mentioned, pins 1, 2, 3, and 4 of the jumper socket go to pins 2, 5, 11, and 14 of the 74HC257, respectively. On this jumper socket, pins 5 through 8 are tied together and go to the 5V line.
For a 4.8-second initial message, jumpers go between pins 1 and 8 and between pins 2 and 7.
Pin 1 of a 555 is grounded; pins 4 and 8 go to the 5V line. Pins 2 and 6 are tied together and go through 0.001uF to ground.
Pin 3 of the 555 goes through a 5K rheostat, then through a 15K resistor to pins 2 and 6.
Pin 5 of the 555 goes through 0.47uF to the arm of a 5K or 10K trim pot. The bottom of this pot is grounded; the top goes to pin 14 of the 1016.
Two 75477 open collector dual NAND gates are used. On each, pin 4 is grounded and pin 8 goes to the 5V line.
Pin 1 of each driver goes to pin 3 of the 555. Inputs 2 and 7 on both 75477s go to the 5V line. The outputs, 3 and 6 of both chips, each go to a pin on the 5-pin DIN connector; pin 3 of this connector goes to the 24V line. Each output is capable of driving two strings of 6 LD273s, each with a series resistor of as low as 47 ohms (requiring 2-watt resistors).
Exhibitor's Sign Board Layout--
A 2.6 by 8-inch perforated board is used. Viewed from the component side and with the ground bus toward you:
The ISD1016 has its pin 15 through 28 column in the 15th column from the right end of the board. Placing it four holes up from ground leaves room for the 8-pin jumper socket between it and the 5V line.
Immediately to the left of the 1016 is a bank of eight 47K pull-down resistors. Just beyond those and near the 5V line is the 74HC257. Below the 257 are two trim pots. In a column, to the left of the 257, is the 16-pin socket for the two 75477 drivers, with the 555 just below these drivers.
Moving to the left, next comes the 220uF 5V bypass, the bridge rectifier (mounted vertically), and the 2200uF 50V bypass. At the left end, there is a heat sink for the regulator.
Notes on Modulators
A commercial product of transmitters would be made using ROMs as the storage medium for digitized speech. This would afford mass production of commonly used signs such as "Exit" and "Information Desk." However, such memory chips cannot be "field recordable," so they are not mentioned here.
The Love Electronics Company is in a stage of production that EPROMs (erasable programmable read-only memories) are used for storage. At this time, programming these requires personal computer based equipment or a dedicated recording device which is rather complicated. It is interesting to note that "burning" EPROMs cannot be done in real time; speech must be stored in RAM where it can then be fed at a slow data rate into the EPROMs. With this scheme, however, several EPROMs can be made from one recording.
While recordings must be individually made for each sign, we have chosen the ISD (Information Storage Devices) chip as the easiest way for the home builder to make his own signs. As we all know, speech products come and go, and obsolescence of a particular speech product is always a concern.
Finally, however, we are in a solid position to always have field recordable digitized speech. The "Addressable Nattering RAM" (SKTF, Winter 1989) can also directly modulate the transmitter's 555. In fact, the very first Fifth Street installations used Nattering RAMs--complete with their backup battery supplies.
By now, digitized speech systems will always be with us. We may even use "voice boxes" from teddy bears which are designed to record and re-play a child's speech.
List of Critical Parts
These seem to be most available from Marshall Industries. Within the United States, the number to call is 800-522-0084. Outside the United States, call 818-307-6000. These numbers belong to a "telemarketing" service of Marshall Industries established to take small orders. By small, they mean orders with $50 minimum. It's a good thing that receivers are also available there, since you have to buy plenty LEDs (at less than $1 each) to make the $50 minimum.
The LEDs and photo diodes are made by Siemens, and it may be easy to obtain these parts in Europe. Siemens has a U.S. number: 408-257-7910.
The single wide-angle LED is the LD271, capable of handling 200mA when 20-microsecond pulses are used. It has a conical radiation pattern whose output falls off 3dB at half-angles of 25dg with respect to the normal axis. Essentially the same diode is used in the dual package--LD273. Here, however, the "lens cap" on the front of the package is oval; this leads to a radiation pattern whose cross-section is elliptical. With the leads left and right (with the long axis of the oval horizontal), the half-power points are at half-angles of 25dg with respect to the normal axis. Rotating the LD273 so that its leads are one above the other, the half-power points as you move left to right are at 15dg with respect to the normal axis. Remember that this dual package cannot handle as much current as the LD271; with 20-microsecond pulses, its current should be restricted to 175mA or less.
The SFH484 LED concentrates more power in the forward direction. Its peak current with 20-microsecond pulses should be restricted to 200mA or less. Its radiation pattern is conical with the half-power points at 8dg with respect to the normal axis. This amounts to a power increase of 7-fold in the forward direction over the LD271. However, remember that the receiver's effective "off-angle" 3dB points are diminished by a factor of the square root of 2; i.e., the signal is reduced by 3dB at plus/minus 5.6dg An approximate 12dg usable beam is narrow indeed, and it is the editor's view that this diode is more experimental than useful for easy-to-find signs.
Only 100PIV bridges are listed here, although 50PIV ones are available from these sources. Suitable 1-amp 100PIV bridges are: Mouser No. 333-3N426: Newark No. WL01F. The 4-amp 100PIV ones are: Mouser No. 333-RS402L; Newark No. FBU4B.
Newark has nice Ohmite 3-watt ones available as the 13F3501 series. You just tack on the ohmic value as a suffix; i.e., 13F3501-47 (47 ohms).
These are available as the ISD1012 (12 seconds), ISD1016 (16 seconds), and ISD1020 (20 seconds at somewhat lower speech quality). They can be purchased directly from the factory: Information Storage Devices, 2841 Junction Avenue, Suite 204, San Jose, CA 95134; phone 800-825-4473.
The current Talking Signs (TM) system leaves room for expansion in various ways. For example, future installations may be "multichannel"; by driving the LEDs with true current sources carrying sinusoidal signals, multiple subcarriers can be transmitted. Conceivably, one or more of these subcarriers could be dedicated to telemetry sending text to receivers which could decode it. For low vision users, the text of this telemetry would be presented by well-lit large-print displays. For the deaf-blind user, this text would be sent to a mechanical braille display in the receiver.
An important fact is that as these changes come about, transmitters for the blind and print-handicapped will not become useless as this system expands; future receivers will always be able to decode signs of the 1980s. Unfortunately, durable braille displays are not yet highly portable, and braille users will have to wait for the development of receivers which can output text.
In the meantime, we suggest a low-frequency Morse Code system for those who wish to add it to their signs and their receivers. Like low-frequency index tones on audio tape, subaudible (70Hz or so) tones can be mixed in with voice information that modulates the 555 of infrared transmitters. This "subaudible" information can then be felt through tactile transducers on the receivers.
The article on infrared receivers describes a unit with necessary filtering and a tactile output. In the article on infrared transmitters, bandwidth modifications are suggested whereby subaudible tones can be recorded with the audio and presented as modulation on the signs. These notes are enclosed in brackets.
The following is a description of a suitable signal source which can be mixed with audible information of Talking Signs whose "tone" can be used to send Morse Code to persons with our modified receiver.
Subaudible Sinusoidal Morse Code Signal Source
[Note: Various portions of this circuit have been taken from articles in the Fall 1983 issue of SKTF.]
Using an ICL8038 function generator chip, this circuit produces a sinewave at about 70Hz. Considering that modulation of the transmitters discussed in this issue is held to within 2 volts peak-to-peak, the signal of this generator (set at approximately 1 volt peak-to-peak) will be 6dB below peaks in the voice audio.
Commercially available tape recorders also have the ability to mix low-frequency index tones with the program material; these include machines produced by Science Products and the American Printing House for the Blind. The injection (amplitude) of the low-frequency tones varies between machines, but recordings made on these should work with the tactile receiver presented in the "infrared receivers" article of this issue.
Subaudible Sinewave Generator With Keying Circuit--
A 10.5V battery is used; this is comprised of a 9V battery in series with a single 1.5V pen light cell. The negative of this battery is grounded; the positive goes through an on-off switch to the VCC line. The VCC line is bypassed to ground by 10uF (negative of this cap at ground).
Pin 11 of an ICL8038 is grounded, while pin 6 goes to VCC. Pins 7 and 8 are tied together.
Pin 10 goes through 0.22uF to ground. Pin 10 also goes to the collector of a 2N2222, the emitter of which is grounded. The base of the 2222 is bypassed to ground by 0.22uF (key click filter); this base also goes through 33K to VCC. The telegraph key goes between the base of the transistor and ground.
Pins 4 and 5 of the 8038 go through 10K to VCC. Pin 12 goes through 82K to ground (for "shaping" of the output waveform).
Pin 2 of the 8038 goes through 0.47uF, then through two 10K resistors in series to ground. The junction of these resistors is the hot audio output terminal; the cold side of the audio line is ground. This produces a 1-volt peak-to-peak signal suitable for standard high-level tape recorder inputs.
On the other hand, the high-level input of the ISD chip, pin 20, requires less signal than you would expect. The acceptable peak-to-peak value is about 100mV. The signal from this oscillator should be 6dB below peaks in the voice audio--50mV peak-to-peak. To drive the ISD chip directly, the bottom 10K resistor in the voltage divider off pin 2 should be 470 ohms.
[Editor's Note: This could actually be a continuation of a similar article in the Spring 1991 issue; these drivers are very similar to the 75451-75454 drivers listed there. Please note that the pin assignments in that article are incorrect for the second gate. Pin assignments for all are given here.]
These drivers/actuators are intended for controlling electromechanical devices--solenoids and the like--from TTL circuits. They are ideal for driving strings of LEDs in high-powered Talking Signs transmitters. Like the 75451-75454s described earlier, these are 8-pin DIPs containing two gates with open-collector outputs.
The 75476-75479 chips have two important differences: First, the open-collector voltage can be as high as 70 volts. Second, there is a "clamping pin" which can be used to protect the output transistors from transients.
Pin 5 is called the "clamp." It goes to the cathodes of two diodes; the anodes of the diodes go to the collectors of the output transistors. If transients are expected and you wish to protect the transistors, pin 5 is tied to the high-voltage supply line (no more than 70V) in the circuit. The diodes prevent the collectors from seeing any higher than 70.6V.
Commandeering pin 5 for this purpose makes it necessary to consolidate input connections; i.e., one pin (pin 1) goes to one input each of the two gates.
Being open-collector devices, the outputs can operate devices at other than logic levels. Open collectors require pull-up resistors to operate, and these must either be external resistors, or the load must be returned to that positive supply.
Like the devices in Spring 1991, the maximum supply voltage for the logic is 5.5V. With both gates turned off, current drain from the supply is typically 10mA, but can be as high as 17mA. With both gates turned on, however, current drain on the 5V supply can be as high as 75mA, probably due to base current in the output transistors.
The minimum logic high is 2V; the maximum logic low is 0.8V. Input current for logic low is the standard TTL 1.6mA.
Compared to others, the propagation delay is long--200 nanoseconds.
The minimum open-collector break-down voltage--the maximum you can count on using--is 70 volts. Likewise, the voltage on the clamp terminal should be no higher than 70 volts.
The outputs can, when saturated, sink a DC current of 400mA. For pulses less than 10 milliseconds (even with a 50% duty cycle), the output can sink 500mA.
The maximum package dissipation is 1 watt. The high supply current required to run these chips causes them to dissipate a maximum of 0.375 watts without a load. However, the saturation voltage of the outputs is quite low (typically 0.33V, 0.6V maximum), so exceeding their package dissipation rating is unlikely.
The units with the highest package dissipation, 1 watt, are those in the plastic package and bear the suffix P. Each package has two gates, one input of each going to a common pin, pin 1. The 75476P is an AND, the 75477P is NAND, the 75478P is an OR, and the 75479P is NOR.
To review the previous gates (good for 30V and having completely separate gate inputs), the 75451P is an AND, the 75452P is NAND, the 75453P is an OR, and the 75454 is NOR.
• 75451 through 75454:
• Pin 4--Ground
• Pin 8--VCC
• Pin 1--Input Gate 1
• Pin 2--Input Gate 1
• Pin 3--Output Gate 1
• Pin 7--Input Gate 2
• Pin 6--Input Gate 2
• Pin 5--Output Gate 2
• * * *
• 75476 through 75479:
• Pin 4--Ground
• Pin 8--VCC
• Pin 1--Inputs to Gates 1 and 2
• Pin 2--Input Gate 1
• Pin 3--Output Gate 1
• Pin 7--Input Gate 2
• Pin 6--Output Gate 2
• Pin 5--Clamp
[Note: This discussion refers to "Talking Signs" (TM). It may help you to follow the thread if you read "Introductions to Talking Signs" in this issue.]
Imagine a young fellow enjoying his city--listening to live music in the wee hours of the morning in a club within walking distance of his home. The home? A dingy little ground floor storage structure has been converted to living space. Access to the apartment is by a door some distance down a little alley; this is just the sort of place where a carefree guy can leave his laundry on the floor and where ham radio equipment is the major source of heat.
Our young hero? Bids the musicians goodbye and leaves the corner club for the short walk home. "I'll take this other way," he thinks to himself, "just for variety." At the next corner he hears an unpleasant altercation between city police and a motorist. "Better turn here before I have to walk right past them," he calculates. Then all of a sudden, a system of wooden barriers on sawhorses traps his folding white cane--a temporary failure of a joint causes the elastic to recoil, which snatches the cane out of his grasp. The cane takes off under the barrier and disappears. Not knowing what is down there (reaching under with a foot does not lead to retrieval of the cane), our sullied hero is reluctant to dismantle the barrier so as to clamber around for the cane. "Heck with it," he mused. "I'm two and a half blocks from home; here I go. That cane was pretty beat up anyhow."
Avoiding the policeman's radio, the traveler heads back around to take the original way home. "Might as well take a route I know is unencumbered," he thought. But where his little alley should have been, it isn't. The terrain is uphill like it should be, but there is no alley to be found. "Which turn did I forget, or have I made one I don't recollect? Going east or going north are both uphill; I must be going east by mistake," he reasoned. Our traveling sport turns into a detective. "Well then, if I'm heading east by mistake, this next street should have two-way traffic."
At the next intersection, there isn't enough traffic to verify the two-way street. "Well then, if I go north, I should find the double-width street, California." But no double-width street is to be had.
Several trials later, one or two alleys are found which could be his. He checks them out, looking for his bushes and doorway. "Not the safest business," he thought, "walking into alleys in the dead of night."
Cursing now, he supposes if he were to locate a telephone, he could be traced and rescued. "No phones in the residential area," he proclaimed. "Listen for traffic and head for the thoroughfares," he challenged out loud.
We're two hours into this story now. Having found a double-width street, our hero has new hope. If he is where he thinks he is, he knows right where a phone should be. But no; he is forced to follow the curb in search of a public phone. He is taking a bit of a beating, contending with street furniture without his cane--parking meters and the famous trees with which San Francisco proudly lines its streets.
He hears a sobbing voice--kind of a boxed-in sobbing voice--crying and talking as if over a telephone. By George, he is boxed in; he's in a telephone booth. Always courteous, our hero goes into standby mode, waiting for the caller to finish. But the sorry sap is obviously oblivious to anyone else's need for the phone; he is prepared to wait for hours.
Then, out of the blue, comes a hardened voice that sounds not to be trusted: "Hey, man," the voice says.
Our hero, wishing he were back at the party, hopes that a spirit put to good use might be so distracted as to keep his knives and truncheons in his raincoat. "Hey, good fellow, you could help me out." "What are the names of these two streets?"
"Polk and California."
"Oh man," our hero enlightens himself, "I'm eight blocks east of home." "Well then, would Van Ness be that direction?"
"Let's jam," our hero cowers. "These are not my friends."
The ground was level here. Only knowing the street names, our detective has to find out which is which. Crossing one will reveal its width; then he'll know which is California and which is Polk. Picking a direction along California will either lead to a narrow one-way street in the wrong direction or to the double-width Van Ness Avenue that's on the way home.
The first guesses were right, and at nearly 5:00 a.m., a somewhat deflowered adventurer, having gone over 25 blocks without his cane, has the satisfaction of feeling his key fit the lock.
That very sequence of events happened about 13 years ago. Yes, you can quit guessing: that sullied hero was me. A legend has it that Talking Signs were an invention in response to my plight. Nice story, but we had already just started working on them. Nevertheless, if I needed a demonstration of their validity, I surely got one.
That detective work I did was good, but it's an unfair requirement. Anyone who can read signs will not re-live this story. Knowing how to do this detective is a good sport; in years past, I enjoyed this sport called "orientation and mobility" (as long as it was not at a risky time of night). But why deny people who don't enjoy that "sport" the comforts everyone else has in getting from place to place? And why put a good sport like me at risk by denying me quick location of a phone when I want one?
"Mobility" is not so much a sport. Using a cane or dog guide, we can walk with confidence and safety; I could, if I didn't care where I was going, walk hither and thither in all recesses of a major airport and be physically safe. On the other hand, the "orientation" component of independent travel is the only task that puts blind people into open-loop behavior (functioning at times without feedback), and this happens every time we go somewhere new or make a choice of transportation.
For everything else we do, we get feedback. We cook, manipulating food with spatulas, feeling and testing it with forks, and smelling when it's done. Many of us clean our floors in bare feet so that we can feel with certainty when they are clean. We groom and care for our children by smelling their hair, feeling their shoes to see where they've been, and by asking trick questions that make them lose track of little white lies. We practice electrical soldering at temperatures exceeding 600 degrees by smelling the flux, looking for rapid heat transfer, noting where solder melts, etc.
In contrast, orientation cuts you loose on a regular basis. It's the only task which often challenges: "You won't get any specific information for a while; tough luck. When you do get feedback, it may be timely, but maybe not. Good luck, sucker."
I'm not saying that we cannot function independently; we can and we do. I am saying that technology is now such that if we insist on "doing it the hard way," we're making a choice that no one else would make.
Try tampering with anyone else's signs and they won't have it to be. Try taking down a merchant's sign and he will complain to you about how much business he would lose. Try taking down the street signs and you'll turn your city into a ghost town. Try taking down the exit signs and people won't feel safe. Try banning name tags at a convention and people won't go to the "President's Reception."
Visual signs are a lot of trouble: they get dirty, they get stolen, they get defaced, they become the wrong color as things are painted, and they take up lots of room. If we ask for our infrared signs (which consist of a little-bitty glass window to which no attention is drawn): they can get very dirty before they become useless, they are not exciting to steal or deface (this would be a very private crime), they don't require repainting, and they are small.
Infrared signs do require power. There are those who then argue: "Well, they won't be installed in many places because of the need for power." Let me remind the skeptics of all the places they could go which have power; i.e., they can go anywhere there is lighting. Modern bus shelters have lighting (to curb crime). Modern telephone booths have lighting. Buildings have lighting. Public vehicles and taxicabs have lighting. Street intersections with signal lights have power available. Major exit signs are lit. And infrared name tags can be powered from a radio battery.
To those who contend that they would have to be everywhere to be useful, I countered: If I had had one piece of solid information or access to a phone booth over the distance of 25 city blocks, I wouldn't have talked to scary people at 4:30 a.m. So there!
* * * * * * *
I publish circuits of infrared signs in this magazine so as to create one more avenue by which they might be installed. Any of several processes can cause these signs to be put up: (1) Transportation systems are now in a quandary; the Americans with Disabilities Act requires them to announce the stops for the blind and visually impaired, and they don't know how. By labeling their bus stops, they could make it possible for people carrying receivers to point them toward the right-hand window so as to get the names of the stops. This would also address our problem of finding bus stops. (2) Cities, like San Francisco, may try installing them on street corners. (3) Public transportation agencies may work with their cities to put them on street signs; this would address both bus-stop announcement and street-identification issues. (4) So that all people have proper access to public telephones, these signs might be installed so that phone booths can be located from a distance. (5) Shopping malls could install these as a publicity gimmick, and it would bring them a new class of customer. (6) Organizations of blind people could supply these signs, the payment for which would be tax deductible. Installed by small businesses and residents, these signs would bring revenue to the organizations supplying them. (7) Until one of those avenues takes off, we can put our own up. Most everything you need to do so is in this issue. If you need more help, call me.
* * * * * * *
May I wish you a prosperous New Year. (There are those naysayers who propose that Silicon Valley is turning into a sandpit. Well, sand is silicon dioxide, so what of it.) Stick with me, folks, and we will ride through tough times and good times together.
P.S. By the time you get this issue, it will be appropriate to consider contributing to the Gunderson Fund again (which is used to help graduates of the New York Institute for the Education of the Blind to pursue meaningful careers). You can do this by making a check payable to the Robert W. Gunderson Memorial Fund; then, on the check memo line, write c/o Alumni Association Fund, NYI. Send it to the Robert W. Gunderson Memorial Fund, c/o Alumni Association of the New York Institute for Special Education, attention Joseph Bruno, Treasurer, 420 West 261st Street, Bronx, NY 10473. Without young folks getting the help they need, we're doomed.
P.P.S. It is time to subscribe again (see "The Technical File Catches Up"), and I need your support. Notice that the prices haven't changed--$18 for braille, $16 for diskette, and $14 for talking book cassette. My special thanks to those who have already sent the money. Your checks are being entered and your support is duly noted. Please stir the bushes and get new people involved; anyone who missed the last two issues of 1991 is missing out. Sayonara.
You may have noticed that this is July 1992 already, even though the cover sheet proudly boasts Fall 1991. This is partly due to the fact that the editor has spilled the entirety of his intelligence in the past eleven years and has nothing much more to say. Articles from readers are trickling in, and four new issues can easily be foretold.
The point is how to re-order the issues so that the next four are "timely?". The following ingenious scheme is proposed: Subscriptions, as in the past, will cover four issues. I can foresee four issues being assembled by the end of 1993. Rather than jumble up the seasons and be confusing, suppose we just number these 1 through 4 (as we already do on the cover page). Since these will be coming out in the next year and a half, we will label the time period 1992-93. Anyone who has subscribed for 1992 and anyone who subscribes now will receive the next four issues concluding in 1993. The prices will be the same: $18 for braille, $16 for IBM diskette, and $14 for talking book cassette. Please note that there is a date code on your address label, and this will indicate the period through which you are paid up. For example, if it says 1991, please consider supporting us for the new set of four.
Does anybody mind if I take a vacation?