TABLE OF CONTENTS

Introduction to the Smith-Kettlewell Note-a-Braille

A User's Guide to the Smith-Kettlewell Note-a-Braille

The Smith-Kettlewell Note-a-Braille and How to Build It

LATER INSERT: The JFD "Universal Interface Converter"

INTRODUCING THE SMITH-KETTLEWELL NOTE-A-BRAILLE

by Bill Gerrey

Abstract

The Note-a-Braille is a peripheral "data-entry" device. A highly portable instrument, it has a standard Braille keyboard for input; entered text is retrieved through a "Centronics Parallel Port." Note that, since it has no mechanical Braille display of its own, the "Note-a-Braille" is not a Braille device per se. It is, instead, a keyboard data entry device; it takes advantage of the fact that thousands of us know the "Braille writer keyboard," and this will always be smaller than the standard typewriter equivalent.

Philosophy of the Note-a-Braille

For the sighted, computers are becoming portable. It is already the case that, for a few hundred dollars, our sighted counterparts have sophisticated note-taking machines--full-blown computers on which even programming is possible. These machines, called "lap computers," have given the sighted colleague a temporary advantage at meetings and in the classroom.

Unfortunately, modifying "lap computers" for the blind (with speech or Braille) currently puts the price up to such a figure that the portable
unit falls into the price class of full-blown systems with two disk drives. Modifying them, by way of special software or extra hardware, adds to the price and detracts from the functionality of the machine. An alternative would be to make a new adapted computer from scratch--a small-market item which would also cost a bundle.

If portability is what we're after, why not gain it with a cheap peripheral device. That's exactly what Al Alden has done in his design of the Note-a-Braille. Furthermore, he has had the good sense to keep this Note-a-Braille as "dumb" as possible (no processing capability of its own). Braille keys simply load 8-bit numbers into a random-access memory (RAM) chip; a read-only memory chip (ROM) converts these 8-bit numbers into
ASCII. Its parts cost is about $40. (In addition to this cost, if your computer does not have a parallel port through which you can "upload" the information, you will need a "parallel-to-serial converter," which can be had for about $75.)

The Note-a-Braille affords no way of making changes or backspacing while writing. This was part of its design for two reasons. First, any fancy editing functions would demand that the Note-a-Braille have a readout of its own, and a backspace key would have added complexity to its counter. Second, because editing features of word-processing machines are as good as they are, there is much to be said for ignoring mistakes during the input process, then taking care of them when you have more time and are in the midst of editing.

The philosophy of the Note-a-Braille is, bang the notes in, paying no heed to mistakes. Then, take it home to edit on a good full-scale system--your IBM PC, Apple, or the like. In this way, the portable device doesn't force you to pay for yet another editor (which won't be as good as the one you have at home) and another readout system (the best of which you also have at home).

[It should be noted that the Note-a-Braille is not alone in the field. For the person who wants more flexibility--editing features and a readout system--the "Pocket Braille" and the "PortaBraille," developed by the Kentucky Bureau for the Blind and soon to be made by Southland Electronics, may be more to his liking. (A User's Manual and technical information sufficient for building these are available from the Division of Technical
Services, Bureau for the Blind of Kentucky, State Office Bldg. Annex, 1st Floor, Frankfort, KY 40601. The price of the whole packet is $5.) A highly modified talking lap computer, which has been dedicated to word processing and scientific calculation, is available. Designed by Computer
Aids Corp., it can be gotten from VTEK Inc., 1610 26th St., Los Angeles, CA 90404.]

Physical Description

The specific device described here is a prototype. Therefore, nothing about it is fixed in concrete. For example, its size, the configuration of its controls, its memory, and the choice of output connector are all changeable, based on the whim of the builder. I bother to say this because, as I physically describe it, you should picture the device you yourself would build--one that fits your pocket and that stays stable as you write with it in your lap.

The cabinet measures 7-1/2 by 4 by 1-3/8 inches. With the keys mounted on the top panel, the thickness is increased by 1/2 inch; since the keys we chose take up no room below the panel, the overall thickness could be reduced by recessing them or by making a thinner cabinet. Also on top is a switch in the lower right corner which "enables" the Note-a-Braille, or locks out the keyboard (so that as you gallop from class to class, you take no chance of randomly re-writing Shakespeare with your elbows).

In order to keep the keyboard as short as possible, no space was left between dots 1 and 4. Where the Note-a-Braille has been exhibited, this has brought some negative comments, but the Editor has found that he got very accustomed to this. The distance between the centers of the keys (horizontally) is about 13/16 of an inch; all eight keys fit in a space which is 6-1/4 inches long (counting the breadth of the key tops). The keys for dots 7 and 8 are mounted 1/4 of an inch closer to the user, since the little fingers are so much shorter than the others. The space bar is mounted 1-1/4 inches in front of the other keys (toward the user), so that the thumb can operate it without reaching forward. If you plan to do a lot of writing "up close"--on your lap--you might consider putting the keys in a gentle V-shape so that your hands can be turned slightly.

On the prototype, there are two more switches, a "go-back-to-zero button" (which zeros the counter and takes you back to the beginning of memory), and a "dump button." On the recommended cabinet, with the battery door of the cabinet at the left end, these buttons, along with the output connector, are mounted on the right end.

A very small speaker is mounted under a hole in the top panel. This speaker emits very faint clicks at the completion of each keystroke. This tells you that the Note-a-Braille is functioning. In addition, as you reach the limits of its memory (within 512 bytes of the end), the clicks come on every other keystroke. These clicks are not very loud (which is good, as far as disturbing your neighbor is concerned); you will have to listen for them. (You might think it would be nice if they were louder. It turns out, however, that the generation of these clicks puts much more drain on the battery than anything else.)

To keep it simple, it was designed with one RAM chip. In this form, it can store 8192 characters, which is about eight pages of this magazine. A sketchy discussion will be given as to how this can be expanded. I have only filled the memory once (writing portions of this article), and I have never used more than 1K while taking notes in a meeting.

Retrieving the information from the Note-a-Braille is done by "uploading" it into a computer--a computer which, by way of software, is taught to think of itself as a "terminal" able to receive such information. So-called "terminal programs" are quite common; these are how computers know enough to accept information over MODEM's as well.

Braille as an Input System

If, for your home computer, you have software for a bi-directional Grade II translator, your note-taking can be lightning fast. If not, you will have to get used to "computer Braille" (which is different from Grade I, only in the numbers and punctuation). A thermoformed sheet of punctuation and symbols of computer Braille has been included in your magazine. Arabic numerals are written in the lower half of the cell without need for a number sign.

Besides the six keys of standard Braillers, two more have been provided for operation by the little fingers. This adds to the number of combinations you can enter. This setup has become known as "8-dot Braille." Some readouts and printers permit these two new dots to be displayed; they appear as additional bottom dots on the cell--the array of the cell now being two across and four high. In describing the characters of 8-dot Braille, the tradition is to stick with the original numbering of the standard six dots, placing "dot 7" below "dot 3," and "dot 8" below "dot 6."

In the language of ASCII (American Standard Code for Information Interchange), we have need for more than the sixty-three characters afforded by 6-dot Braille; upper-case letters and "control characters" must be accommodated. We can no longer tolerate a 2-cell combination for capital letters.

Opinions differ as to the exact format of 8-dot Braille. We chose to arrange it so that an upper-case letter is made by including dot 8 (the right little finger). On ours, control characters are made by including dot 7 (the left little finger). On the VTEK "Braille Display Processor," dots 7 and 8 are arranged just the opposite, a practice that is also followed on some European machines. If you want to, you can just trade positions of the end-most keys to get the opposing format.

A Personal Reflection

As I write this, I am sitting in a cafe, sipping a cappucino and nibbling on fancy Armenian hors d'oeuvres. Before computers came along and wrecked it for me, I used to write all my papers this way. In those days, I had the portability of a "pocket slate"; my pockets would bulge with great works created in hand-written Braille.

When I got a computer, the advantages of "word processing" became obvious. Creating a single draft is now sufficient; typing a second draft is made needless, since extensive changes can easily be made in the original. Writing in a form which could not become a computer file became a wasteful step.

Ever since computers, then, I have been tied to a desk--the same ol' desk in the same ol' Smith-Kettlewell lab, where even pacing the floor has to be done sideways for lack of room. Gone were the contemplative strolls around the park and the inspirational changes of scenery that fed my boundless creativity.

As I sit here, free to go where I please and still generate computer files, I have the feeling that something I once had has been retrieved. Thanks to Mr. Alden and his efforts, I am once again "portable," with the main difference being that I can write faster than before. Maybe I'll get a
partial suntan again someday. In the meantime, I'm as happy as a clam at high tide.

A USER'S GUIDE TO THE SMITH-KETTLEWELL NOTE-A-BRAILLE

by Bill Gerrey and Tom Fowle

Basic Operation

A quick description of the controls is as follows:

Eight Braille keys plus a space bar are mounted on the top of the box. Also on top, at the lower right corner, is an on-off switch. (What this switch mainly does is lock out the keyboard so that characters cannot be entered by accident. Portions of the machine, such as the random-access memory, are always powered.) Alongside the interface connector on the right end are two buttons--a "reset" or "zero" button, and a "dump" button.

The Note-a-Braille affords no way of making changes or backspacing while writing. This was part of its design for two reasons. First, any fancy editing functions would demand that the Note-a-Braille have a readout of its own, and a backspace key would have added complexity to its counter. Second, because editing features of word-processing machines are as good as they are, there is much to be said for ignoring mistakes during the input process, then taking care of them when you have more time and are in the midst of editing.

Entering data is absolutely straightforward; once the off-on switch is turned on, any key or combination of keys that you press will be recorded in memory. A character is not registered until every key has been released; thus, if you are late in adding a dot to a desired combination, you can include it any time before the keys are released.

Opinions differ as to the exact format of 8-dot Braille. We chose to arrange it so that an upper-case letter is made by including dot 8 (the right little finger). On ours, control characters are made by including dot 7 (the left little finger). On the VTEK "Braille Display Processor," dots 7 and 8 are arranged just the opposite, a practice that is also followed on some European machines. If you want to, you can just trade positions of the end-most keys to get the opposing format.

Whenever you want a carriage return, enter a "control M"; a line feed is a "Control J." These are usually entered as a pair--control M, control J. The keyboard and other buttons are locked off until you turn the switch on. Once the switch is on, you can either continue writing from the present location in memory, or you can "start over" by pressing the "zero button." Pushing the zero button does not delete anything in memory; old material is only deleted by writing over it anew (or by disconnecting the battery).

A block of text can be terminated by typing a "stop character"--pressing dots 7 and 8 simultaneously. (Actually, any combination which contains dots 7 and 8 is interpreted as a stop character; letters with both dots 7 and 8 added are not used.)

When you are ready to "upload" your work into a computer, the "zero button" is pressed (with the on-off switch on); then the "dump button" is pressed. Pulses from the speaker will be heard, indicating that data are being transmitted. When the pulses terminate, it means that a stop character has been encountered. Transmission (of the next file) can be resumed by pressing the dump button again; do not press the zero button again, or you will just set yourself up for sending a copy of the first file.

If you neglect to terminate a file with a stop character, you will upload any residual material in an old file. This is no big deal; you can delete that unwanted portion when you are in the editing mode on your computer.

When the Note-a-Braille pauses between messages, you can either simply resume dumping, or you can take that opportunity to put the next transmission into a different file.

One nice feature of the Note-a-Braille is that, besides storing your entries in RAM, it also sends this data immediately to its parallel port. It is therefore possible to communicate directly with the computer as you type into the Braille keyboard. With some computers, you can tell the machine to look at the serial port for its keyboard information. (In the Apple series, for example, you can type--from Basic--an instruction that looks something like "IN #X," where X is the slot number in which you have the serial card.) However, as will be mentioned in the next section, a main use for this direct output is to send a forgotten "End of File" character, in cases where the communications program of your computer needs one.

Power Requirements

During storage, the current drain is a mere 50 microamperes; this should keep your notes alive for perhaps 10 months. Turning the Note-a-Braille on increase this to about 150 microamps. However, the greatest drain on the battery is imposed by the clicking speaker--a necessary but costly feature. With the user typing at a rate of three strokes a second, the drain on the battery is ten times that during storage.

These conditions change when the Note-a-Braille is plugged into the parallel-to-serial converter. Here, some of the power is actually obtained through the parallel interface. We have noticed that in the off position the "storage" drain drops to about 28 microamps.

Caution! There is one set of circumstances that seems rather ominous. We have noticed that if, by some accident, the parallel-to-serial converter loses its power while the Note-a-Braille is connected to it, the battery drain goes up to over 300 milliamps and stays there, no matter what you subsequently disconnect or turn off. No one knows why, and preliminary investigation has led to no conclusion.

For now, just don't let this event happen (losing power to the converter while the Note-a-Braille is connected). Once this does happen, the cure for it is to disconnect the battery and short out the snap connector to see that an initial condition of 0 volts is established, and reconnect the battery. Although it sounds brutal, our prototype has been revivable every time during testing of this event.

Replacing the battery, I did once catch the Note-a-Braille malfunctioning for some reason. Again, the cure was simple--disconnect the battery, short out the terminals of the snap connector on the machine, and reconnect the battery.

Interface Considerations

As far as engineering is concerned, it was convenient to provide the Note-a-Braille with a Centronics Parallel Interface. This interface is often used between a computer and its printer. Unfortunately, it is a very rare thing to find a computer capable of accepting parallel input information. Therefore, you will most likely have to buy a device which has a parallel input, and which converts this information to the RS232C protocol. Such a device is called a "protocol converter"--a parallel-to-serial converter.

The converter we have recommended here is reasonably priced--about $75--and is available from Jameco. It contains a microprocessor, and even 2 kilobytes of buffer memory (memory we don't make use of). So-called "DIP switches" (tiny switches built into a dual in-line package) permit serial parameters to be set. (In fact, as you receive it, the device is set for serial-to-parallel conversion, a condition which is also set with DIP switches.)

We may eventually design a second version of the Note-a-Braille which has a serial output. This would necessitate inclusion of a microprocessor, however, and the added complexity would deter the individual builder.

It should be noted that without a converter, the Note-a-Braille can be fed into the Centronics port on a printer. This allows for no possibility of error correction, and its usefulness would be limited. (Of course, a speech synthesizer with a parallel port could also be used directly.)

Next, the computer you use must be able to run a "communications program," sometimes called a "terminal program." These programs are generally used to take in data from a MODEM. Their operation varies so widely that we could not begin to give a procedural description. Suffice it to say that your computer must be in the "terminal mode"--able to receive data, store it in memory, and then save it to disk.

Once set up with a proper communications program, the next points of concern are:

Can your computer keep up with the data transfer rate? Does your computer have enough memory to store the contents of the Note-a-Braille? What indication does your communications program need to denote "end of file?"

There is a remote possibility that your computer may want to handshake the serial port feeding it. Unfortunately, all of the protocol converters we have found are set up for "hardware handshaking" (requiring a voltage signal on the "Data Terminal Ready" or "Clear to Send" lines of the serial port). It seems that most, if not all, communications programs provide for "software handshaking;" here, characters called "Xon" and "Xoff" are sent from the computer on the second data line so as to start and stop the transmission. However, if the computer doesn't need to save the file in small bits, or if no other condition requires it to handshake the converter--and hence, the Note-a-Braille--everything should work nicely.

It should be noted here that in the Versa-Braille, TSI was thoughtful enough to include hardware handshaking in the "communications control parameters." Setting the "HS" parameter to the "DTR option" will provide the needed signal on pin 20 of the RS232C port. This is indeed fortunate, because without this handshaking, the Note-a-Braille would not know to stop and wait for the tape drive to run between pages.

If, for example, your computer has a limited amount of memory so as not to accept a reasonable sized file (8K maximum from the Note-a-Braille), you will have to keep rough track as to how often the Note-a-Braille file should be broken up. You can break up a file by periodically inserting an "end of file" character (dots 7 and 8). The full file would then be sent by repeatedly pressing the "dump" button, waiting in between for the computer to handle the data.

A characteristic of parallel transmission is that it can be much faster than serial data flow. Hardware handshaking from the receiving device will determine this data rate. Without receiving any handshaking--stepping through files with the Note-a-Braille unplugged--the rate of data will be set by the free-running of the ICM7555 in the circuit (see next article); the data will be sent at about 1500 characters per second.

Finally, your communications program may want to see a special character called "End of File" (EOF). (While there is an ASCII character called "EOF," the required character may vary from system to system.) Once this character is known, you can enter it in as the last character of the file you have written on the Note-a-Braille--just before the dots 7 and 8 combination is entered).

As stated earlier, every entry made on the Note-a-Braille is immediately sent to the parallel port. Thus, if you have forgotten to include the "End of File" character, you can always type it in while the computer is in terminal mode waiting for it. (This means, however, that this "End of File" character is part of the next file. Any attempt to retrieve the next file, without first striking over this character, will abort communications before any data are sent. Once polluted, you will have to unplug the Note-a-Braille, set it to zero, count over to the polluted file, and strike over the "EOF" with some other harmless character.)

The "ADAPTA" Serial-Parallel/Parallel-Serial Converter

Although very elaborate and flexible, this converter is the least expensive we could find (about $75). (We don't use many of its features, yet simpler ones were more expensive; this is a good bargain.) The Jameco catalogue number is the same as its name "ADAPTA." It is manufactured by Performance Interconnect Inc., a subsidiary of Advanced Digital Systems Inc., 8950 Villa la Jolla Drive, Suite 2144, la Jolla, CA 92037; (619) 457-0665. [Available from Jameco Electronics, 1355 Shoreway Rd., Belmont, CA 94022; (415) 592-8097.]

When you receive it, it is set up to convert an RS232C port to a Centronics parallel port; DIP switches set it to do the reverse. It has an internal buffer of 2048 bytes (which we don't use). DIP switches provide for eight baud rates--from 110 to 19,200. In addition, DIP-switch settings can accommodate: Seven or eight data bits. Parity or no parity. One or two stop bits. Control-signal polarity.

It is powered by 9 VAC (supplied from an external plug-mounted transformer). The converter itself measures about 5-1/2 inches square and 1-1/4 inches high. Two Philips-head self-tapping screws inserted up through the bottom permit removal of the top cover. (The two halves of the box are nearly identical and both can be removed; if you wind up exposing the bottom of the circuit board, as opposed to the component side, replace this cover and remove the other cover.)

Once the top of the board is in view, the converter should be oriented so that the parallel connector is on the left (this is the larger of the two computer connectors, and it has spring "locking ears" by which the male unit can be secured). On the right should be a 25-pin D connector and a power socket (the power socket should be on the near end of the 25-pin unit). On the top-right corner of the board you will find a crystal. Just below this and dominating the upper-right quarter of the board is the microprocessor with its ROM riding piggie-back on it. Immediately below the microprocessor--immediately to the left of the 25-pin connector--are two DIP packages having eight switches each.

The block of switches closest to you, just above a group of resistors and diodes lying on the board, shall be called "block 1." The block of switches adjacent to the microprocessor shall be called "block 2." Each block has eight switches--1 through 8, from left to right. "On" is "up" (away from you); "off" is "down" (toward you).

Switches 4 through 8 of block 1 are used to determine the direction of conversion. When you get the converter, 4 through 8 are on, off, on, off, on. In order to make this a parallel-to-serial converter, these five switches should be, from left to right: off, on, off, on, off.

Eight baud rates are selectable. Their selection is made with the first three switches in block 1. The following are the switch settings--1, 2, and 3, from left to right--for each baud rate:

• 110--on, on, on
• 300--off, on, on
• 600--on, off, on
• 1200--off, off, on
• 2400--on, on, off
• 4800--off, on, off
• 9600--on, off, off
• 19,200--off, off, off

On block 2, switches 1 and 2 are not used. Switch 3, when it is "on," turns on the 2K buffer; we don't use this buffer, so this switch is turned off. They refer to a term "polarity" or "serial control sense." Controlled by switch 4, this determines whether pin 5 of the serial port becomes a "busy" (an output) or "ready" (an input). Always being used as a transmitting device for this application, switch 4 should be "on" ("Ready" to receive handshaking).

Other (optional) parameters that you must match to your equipment are controlled by the block 2 switches 5 through 8 as follows:
Switch 5--Stop Bits; off for 1, on for 2
Switch 7--Parity; on is parity, off is no parity
Switch 6--When parity is chosen, switch 6 determines whether it is even or odd; off is even, on is odd.
Switch 8--Number of bits; off is 7, on is 8.

Interface Pin Connections

Like all computer "standard interfaces," the parallel interface has changed over time--in its case, toward simplification. All the pins are listed here; some were invented for printers no longer in use, and their functions have fallen into obscurity. The ones in common use are on pins 1, 2 through 9, and 11--ground being any one from 19 through 30.

Centronics Pin Connections

• 1--Data Strobe (from the transmitting device; active low)
• 2--Data0
• 3--Data1
• 4--Data2
• 5--Data3
• 6--Data4
• 7--Data5
• 8--Data6
• 9--Data7
• 10--Acknowledge Out (from the receiving device)
• 11--Busy (from the receiving device; active high)
• 12--Out of Paper (from the receiving device)
• 13--Printer Select (from the receiving device)
• 14--Plus/Minus 0 volts (reference from the receiving device)
• 15--Osc. Ext. (signal from receiving device--100kHz or 200kHz)
• 16--Plus/Minus 0 volts
• 17--Chassis Ground (probably similar to "Protective Ground" on RS232C)
• 18--Plus 5 volts
• 19 through 27--Twisted-Pair Returns (ground for each data line)
• 28 through 30--Ground
• 31--Input Prime (alerts input device)
• 32--Fault (from receiving device; can be a light or "deselect")
• 34 and 35--Line Counter Switch (contacts of a switch in the printer)

Serial Connections

The full list of RS232C pin assignments has been given in these pages before in the article "The Smith-Kettlewell Breakout Box," SKTF, Winter 1985. Of the twenty-five pins, four are used here:

• 1--Chassis Ground
• 2--Data Out
• 5--Ready/Busy
• 7--Signal Ground

Tricks by a Confirmed User

In many environments (like the noisy cafe I am now in), the clicking of the speaker is inaudible unless you put your ear to the machine. When I start getting nervous about running out of memory, I find a place where I'm going to insert two or more spaces--like at the end of a sentence or for indenting a paragraph. Then, I pick the unit up and listen closely while entering these two spaces. If only one of these causes a click to be emitted, get thee to a computer. (Be aware of the fact that, if you keep on writing, the counter will not stop at the end of the memory. Rather, it will just keep going through zero and write over the first material, which may or may not be okay with you.)

Much can be learned by listening to the fast-running data with the Note-a-Braille disconnected from the computer. For example, you can hear if files reside in a given portion of memory or not. When you press the "dump button," the counter will step through the memory until any character containing dots 7 and 8 occur. These combinations constitute a quarter of the 256 possible 8-bit numbers. Given a random bunch of characters that is likely to reside in unorganized memory, it doesn't take long before the system runs into such a number; a click or very short burst will be heard from the speaker. This tells you that there is not much information there. On the other hand, if a file does reside there, a long tone of about 1500 Hz will be heard (the maximum frequency of the free-running counter without handshaking). The length of this tone will be in direct proportion to the length of the file.

Once "zero" has been pressed, you can count the number of files by counting the long tones. If you decide, for example, that the first two files are of value, but that everything after that has been a colossal waste of your time, you can count two files along from zero and start writing from that point on, recording over the worthless material. You cannot, however, erase the third file while protecting the fourth, unless you are sure that the new material is shorter than that which you are obliterating.

By unplugging the Note-a-Braille between files during uploading, you can skip files that you wish to upload some other time. For example, if your first class were chemistry and your final class chemistry lab, you might wish to put these together in the same file. You would know that between these classes you had taken notes in history and sociology--hence, two files. You would know, then, to upload the first file, unplug the Note-a-Braille and step through two successive files, then plug it back in and upload the fourth. Stepping through these unwanted files won't take very long, since the counter is stepping through the memory at 1500 bytes per second. When you want to preserve your sociology notes, go back to zero, step through the first two files with the Note-a-Braille unplugged, and upload the third.

When stepping through the memory with the Note-a-Braille unplugged, you can often tell if a file has gotten to the final portion of memory. The "tone" produced by running the data at the free-running rate will drop an octave when this portion of memory has been reached. This is true because the pulses that you hear only happen at every other character.

You might wonder what I do when I'm interrupted and can't remember where I was or what I was writing about. Also, some mention should be made of correcting mistakes. The answer is, basically, nothing, during input. Although I cannot consider myself a virtuoso in word processing, here are my answers.

I treat mistakes in two different ways. If I have misstated something, or if there is a word that cries for deletion, I mark it with a double X (without putting a space before the two X's). When I'm in the editor of my word processor, I do a search for all double X's; each time the computer stops at one, I hit a keystroke that takes me to the beginning of the word, and I hit "word delete." (You can't just "globally replace them"; this would just strip them from the file, leaving the offending partial words in place.)

I could handle any misstroke in this way. However, since I'll be reading the document with a fine-toothed comb anyhow, I leave many small errors unmarked--I'll take care of them as I go. One thing is certain; it is easier to delete something than to do anything else, if the word processor is any good. Therefore, when I've hit a wrong letter, I just make sure that I hit the right one immediately thereafter. This is a good practice, since replacing a letter takes two strokes (one to delete the mistake and one to insert the right one). If all the right letters are there to start with, deleting a wrong one requires only one stroke.

Because deleting material is so easy, this is the key to dealing with
forgetting your place. If you've forgotten where you left off, just retype stuff until you know that some version of it has made it into the file. This procedure avoids two activities that waste time--trying to think what you have already typed, and trying to recreate material or rethink a word that didn't make it in.

Almost as easy as deleting things is moving material around. Therefore, if you think of something that needs inclusion, no matter where you are, put it in--perhaps marking it with two W's or something. Back in the days of paper, if something flashed through my mind which was not exactly germane to what was being put down, I would promise myself that I would try to remember this, and resume my main course. More often than not, I had trouble recreating the inspired thought when the time came. Combining the Note-a-Braille with a word processor, it is important to jam it in there as it comes to you--then shift it around to where it belongs later.

Although it takes some getting used to for us old geezers who try to polish as we write, efficient use of a word processor calls for a high rate of input and extensive use of its editing features. Now, with computers and my Note-a-Braille, I no longer have a stylus to tap on my teeth as I think of fluid phrases. To a large extent, this is liberating; I no longer have to fit tiny pieces of the puzzle together before I set stylus to paper.

THE SMITH-KETTLEWELL NOTE-A-BRAILLE AND HOW TO BUILD IT

by Albert Alden

Circuit Operation

Data Entry (Writing)

All pushbutton switches, the "zero" and "dump" buttons, those for the eight Braille dots and the space bar, are single-pole normally open switches. They all have a common connection which goes through a pole on the off-on switch to ground. In this way, turning the unit off disables all of them.

The switch outputs are connected to a pair of hex contact debouncers (MC14490). The outputs of the debouncers associated with the eight Braille dots feed the set inputs of a pair of tri-state quad R/S flip-flops (CD4044). In addition, the eight signals plus the space bar output of the debouncer go to the equivalent of a 9-input AND gate. This is made from a triple 3-input NAND (CD4023) plus one gate from a dual 4-input NOR (CD4002). The output of this 9-input gate goes high when the last pushbutton is released. This action triggers a one-shot (MC14538, T = 100 microseconds) whose NOT-Q output goes through a unidirectional RC delay (buffered by a CD4073 AND gate) to the NOT-chip select pin of a RAM (HM6264LP-15).

The Q output indirectly operates the "Strobe" pin of the parallel interface. It does so by going through another unidirectional delay so as to make sure that data exist on the output lines before "strobing" the receiving device. The output of this delay circuit is buffered by an inverter, thus creating an active-low strobe signal as required.

The outputs of the two R/S flip-flops are also connected to the eight data input/output lines of the RAM. The signal on the NOT-chip select pin of the RAM causes the pushbutton information to be stored.

When the one-shot period ends, the buffered/delayed NOT-Q output triggers a second one-shot (the other half of the CD14538, T = 100 microseconds). The NOT-Q output of this O.S. performs three functions. First, it resets the pair of quad R/S flip-flops whose outputs are the Braille "dot" information. Second, it advances a 13-bit counter; this counter is made from a pair of dual 4-bit counters (CD4520). The outputs of the 13-bit counter go to the address lines of the RAM. The counter may be reset by pushing the reset ("zero") button.

Finally, this second one-shot turns on an FET to pulse a loudspeaker which indicates to the user that the Note-a-Braille is operating. The speaker is connected to the battery in series with a zener diode and resistor. These were chosen such that no sound is heard when the battery is low (but still sufficient to operate the memory).

This speaker only sounds on every other keystroke when the memory is within 512 strokes of being filled. This condition is activated by AND'ing the four most significant bits of the counter with its least significant bit; this is done with two gates of a triple 3-input AND gate (CD4073).

Data Transfer (Reading)

In order to read the stored data from the RAM, it must be put in the read mode and the R/S flip-flops must be put in the high-output impedance mode. This is done by an R/S flip-flop (referred to as the control FF) made from a pair of 2-input NAND gates (CD4011). The Q output goes to the NOT-write enable of the RAM, and the NOT-Q output goes to the enable line of the CD4044 flip-flops.

The reset ("zero") button sets the control FF to the write state. It also resets the counters. The start ("dump") button sets the control FF to the read state.

In the read state, a CMOS 555 timer (ICM7555) is turned on and operated as an astable multivibrator. The 555 is turned on by providing its power from the output of a 2-input NOR gate, one of whose inputs comes from the control FF. (The other input of this NOR gate gets handshaking information, as discussed later.) The output of the 555 is OR'ed with the 9-input gate previously described. Thus the 555 provides a signal (in place of the "key-release" generated signal) to trigger the first of the pair of one-shots. As before (although now operated by the 555) the one-shots trigger, causing the counter to count and the contents of the RAM to appear in sequence on the input/output lines.

The read and data transfer continues until the output of the RAM contains a dot 7 and a dot 8. This condition is detected by a 2-input NAND gate (CD4011) whose output is OR'ed with the reset ("zero") button and resets the control FF. This turns off the 555 and stops the data transfer. Pushing the start ("dump") button will cause data transfer to continue until the RAM output contains another dot 7-dot 8. Pushing the reset button at any time will stop the data transfer and reset the counters to zero.

The RAM's eight input/output lines have one further connection: to the address lines of a 2K by 8 EPROM (27C16). The purpose of the EPROM is to perform a Braille-to-ASCII conversion. Each Braille code is an address in the EPROM. The data stored at that address is the ASCII code for that Braille symbol. The eight data lines of the EPROM go to the parallel output pins of the Note-a-Braille. The NOT-chip enable of the EPROM is controlled by the delayed output of the first one-shot.

Two additional output connections are made. A strobe signal which is generated by a buffered unidirectional delayed output of the first O.S. is low (0V) for the period that each output ASCII code is valid. A busy input line is provided so that the output data flow can be controlled by the receiving device. A low (0V) turns off the 555 and halts the data transfer. The busy line goes to the other input of the NOR gate which powers the 555. When the busy line goes high, data transfer continues.

Power Supply

The power supply contains two regulators. The first is a low-power 5.0V regulator (LP2950-5.0) which is powered through the off-on switch. The second consists of an emitter follower whose base is connected to a voltage divider across the battery. This emitter follower powers the circuit at 3.5 to 4.5 volts when the power switch is turned off. Since the keyboard is disabled when the unit is turned off, power drain in the static CMOS circuitry is very low, in the order of 50 microamps. When the power switch is turned on, the static power drain is approximately 150 microamps. A diode isolates the LP2950-5.0 from the emitter follower.

Circuit Description

[Editor's Note: There are a lot of pin connections in this thing. In order to keep track of which pins--on which chip--we're talking about, the chip number will often be used as a prefix to the pin number. For example, Braille keys 1, 2, 3, 4, 5 and 6 go to pins on IC1 in the respective order: 1-1, 1-14, 1-3, 1-12, 1-5 and 1-10. As another example, the output of the 555 (IC10) is 10-3 (pin 3). For those who meticulously want to study this circuit, I recommend that you write a pin table for each chip on a card--a bare-bones table, lacking power connections and enable pins that you know won't be used. The resultant 15 cards can be shuffled and set next to each other where appropriate. Hang on to your hats; here we go!]

Power Supply

The negative side of the 9-volt battery is grounded. The positive of the battery shall be called "VB," and it goes several places as follows:

VB goes through 390K, then through 560K to ground. The junction of these resistors goes to the base of a transistor (2N2222A). The collector goes to VB. The emitter goes to the plus V line of the Note-a-Braille circuit.

The on-off switch is double-pole single-throw. One terminal of the first pole goes to VB; the swinger goes to the "Input" of a 3-terminal voltage regulator (IC16, National LP2950-5.0). From the input to ground is a 33uF tantalum capacitor with the positive terminal at the input. The "Common" terminal is grounded. The "Output" terminal goes to the anode of a diode (1N914), the cathode of which goes to the plus V line. Also from the output is a 10uF tantalum to ground with the positive terminal at the output.

On both IC's 1 and 2, MC14490 debouncers, pin 8 is grounded and pin 16 goes to plus V.

On both IC's 3 and 4, 4044 R/S flip-flops, pin 8 is grounded and pin 16 goes to plus V.

On all IC's 5, 6, 11, 12 and 15, the various gates, pin 7 is grounded and pin 14 goes to plus V. On IC15, pins 2, 3, 4, and 5 are all unused inputs; these are taken to ground.

On IC7, a 4538 dual one-shot, pin 8 is grounded and pin 16 goes to plus V. Pins 3 and 13, the "NOT-Clear" terminals, go to plus V. As will be mentioned again, in order to set them up for positive-edge triggering, the "B" inputs, pins 5 and 11, go to plus V.

On the EPROM, IC8, pin 12 is grounded and pin 24 goes to plus V. The "Programming Power Pin," pin 21, goes to plus V. The "NOT-Output Enable," pin 20, is grounded. The unused address pins, 19, 22 and 23, are also grounded.

On the RAM, IC9, pin 14 is grounded and pin 28 goes to plus V. "Chip Select2," pin 26, goes to plus V. The "NOT-Output Enable," pin 22, is grounded.

On IC10, a CMOS 555, pin 1 is grounded. As will be seen, its pins 4 and 8 are powered from the output of a gate and do not go to plus V.

On both counters, 4520's which are IC's 13 and 14, pin 8 is grounded and pin 16 goes to plus V. As will be mentioned again, in order to set them up for negative-edge triggering, their "Clock" terminals, pins 1 and 9, are grounded.

Pushbutton Circuits and Debounce System

All eleven pushbuttons are single-pole normally open switches; one side of all of them is a common bus. This bus goes through the second pole of the on-off switch to ground.

The other side of each button goes to an input on the debounce chips (IC's 1 and 2) as follows:

• Dot 1 to 1-1
• Dot 2 to 1-14
• Dot 3 to 1-3
• Dot 4 to 1-12
• Dot 5 to 1-5
• Dot 6 to 1-10
• Dot 7 to 2-1 (Control)
• Dot 8 to 2-14 (Upper-case)
• Spacebar to 2-3
• Start ("dump") to 2-12 Reset ("zero") to 2-5
• Unused is 2-10 (Internally committed, it needs no connection.)

On IC1, there is a capacitor of 330pF between pins 7 and 9. (This runs a clock inside IC1; see section on IC information.) Pin 9 on IC1 goes to pin 7 of IC2; this clocks IC2. Pin 9 of IC2 is left open.

R/S Character Latches for the Eight Dots

The outputs of the hex debouncers go to the NOT-Sets of quad R/S flip-flops, IC's 3 and 4, as follows:

• 1-15 to 3-3
• 1-2 to 3-7
• 1-13 to 3-11
• 1-4 to 3-15
• 1-11 to 4-3
• 1-6 to 4-7
• 2-15 to 4-11
• 2-2 to 4-15

The outputs of the R/S flip-flops, which we shall call the "data lines," go to both memories. Those on IC8 are the EPROM's address lines, while those on IC9 are the RAM's I/O lines.

• 3-13 to 8-1 and 9-11
• 3-9 to 8-3 and 9-12
• 3-10 to 8-5 and 9-13
• 3-1 to 8-2 and 9-15
• 4-13 to 8-4 and 9-16
• 4-9 to 8-6 and 9-17
• 4-10 to 8-7 and 9-18
• 4-1 to 8-8 and 9-19

There are two additional connections from dot 7 and 8 of these data lines. IC4-10 goes to IC6-9, and 4-1 goes to 6-8. Also on pins 8 and 9 of IC6 (inputs of a NAND gate that detect dots 7 and 8) are 1meg resistors to ground.

AND'ing All Nine Braille Keys

All eight dots, plus the space bar, are combined in a 9-input gate (made by combining the three outputs of 3-input NAND gates); the resultant AND gate detects when all Braille keys have been released. Using the triple 3-input NAND gates of IC5 and combining these in IC15, the debounced outputs are combined as follows:

• 1-15 to 5-1
• 1-2 to 5-2
• 1-13 to 5-8
• 1-4 to 5-3
• 1-11 to 5-4
• 1-6 to 5-5
• 2-15 to 5-11
• 2-2 to 5-12
• 2-13 to 5-13
• 5-9 to 15-12
• 5-6 to 15-11
• 5-10 to 15-10

There is one more input to the 4-input NOR gate of IC15. IC15-9 goes to IC10-3, the output of the 555.

One-Shot System

The output of the 9-input AND gate, IC15-13, goes to pin 4 of IC7 (a dual one-shot). IC7-5 (the B input) goes to plus V. On IC7,
pin 1 is grounded; between pins 1 and 2 is 0.001uF, and pin 2 also goes through 100K to plus V. Pin 3 (Clear) goes to plus V. IC7-7 (NOT-Q) goes through 47K to inputs 3, 4, and 5 of IC12 (a 3-input AND gate used as a buffer). These inputs also go through 47pF to ground. A 1N914 diode is connected across the 47K resistor--its cathode at IC7-7. (The RC circuit causes a delay; the diode makes sure that this delay only takes place on positive transitions of NOT-Q.) The delayed and buffered output is pin 6 of IC12, and it goes several places.

IC12-6 goes to the NOT-Chip Enable of the EPROM, IC8-18. It also goes to the NOT-Chip Select of the RAM IC9-20. As discussed later, it also feeds a NOR gate of IC11, which, in turn, controls the speaker.

IC12-6 also goes to the A input of the second one-shot, IC7-12. Also on IC7, pin 13 (the B input) and pin 11 (Clear) go to plus V. Pin 15 is grounded; between pins 15 and 14 is 0.001uF, while pin 14 also goes through 100K to plus V. The NOT-Q output, IC7-9, goes two places as follows:

IC7-9 goes to the Reset (actually these are NOT-Reset) pins of the R/S flip-flops, 3-4, 3-6, 3-12, 3-14, 4-4, 4-6, 4-12, and 4-14. IC7-9 also goes to pin 2 of IC13, the clock of the first of four cascaded counters.

The Q output of IC7, pin 6, goes through a delay circuit and operates the strobe output of the parallel port; it is connected as follows:

IC7-6 goes through 47K to inputs 8 and 9 of IC11 (a NOR gate used as a buffer). IC11's pins 8 and 9 also go through 47pF to ground. Across the 47K resistor is a 1N914 diode with its cathode at IC7-6. (The diode makes sure that the delay only occurs for positive transitions of IC7-6.) IC11-10 goes to the "strobe pin" on the output connector.

Counter

The 4-bit counters of IC's 13 and 14 are cascaded. They are set up for negative-edge triggering by tying the clocks to ground and using the enable pins for triggering; thus 13-1, 13-9, 14-1, and 14-9 are all grounded. They are cascaded as follows:

• 7-9 goes to 13-2
• 13-6 goes to 13-10
• 13-14 goes to 14-2
• 14-6 goes to 14-10

The resultant counter outputs go to "address lines" of the RAM as follows:

• 13-3 to 9-10 (Least Significant Bit)
• 13-4 to 9-9
• 13-5 to 9-8
• 13-6 to 9-7
• 13-11 to 9-6
• 13-12 to 9-5
• 13-13 to 9-4
• 13-14 to 9-3
• 14-3 to 9-25
• 14-4 to 9-24
• 14-5 to 9-21
• 14-6 to 9-23
• 14-11 to 9-2

The counter gets reset by connecting all of the "Reset" terminals to an inverted version of the debounced reset button ("zero button"). A NOR gate of IC11 is used as an inverter, and the connections are as follows:

2-11 goes to pins 1 and 2 of IC11
11-3 goes to 13-7, 13-15, 14-7, and 14-15

Speaker Pulsing System

The delayed buffered output of the first one-shot, 12-6, goes to pin 12 of IC11 (a NOR gate input; IC11 13, the other input, will be dealt with shortly). IC11-11, the NOR gate's output, goes to the gate of a VN0300M power FET. The source of this FET is grounded; its drain goes through the speaker, then through 30 ohms to the anode of a 6.2V zener diode (1N753). The cathode of this zener goes directly to the positive side of the battery, not to the plus V line as it is defined. (The recommended speaker has a rated impedance of about 45 ohms; the one you use might be different, whereupon you would choose this resistor experimentally.)

To get the speaker to click for "every other stroke" during the last 512 bytes of the memory, the four most significant bits and the least significant bit of the counter are combined in a 5-input AND gate. This 5-input AND gate is made by connecting IC12-10 (an output of a 3-input gate) to IC12-8 (an input of another 3-input gate). The five remaining inputs go to outputs of the counter as follows:

• 13-3 to 12-1
• 14-4 to 12-2
• 14-5 to 12-11
• 14-6 to 12-12
• 14-11 to 12-13

IC12-9 goes to 11-13, the other input of the driving NOR gate.

Control Flip-Flop

Two NAND gates of IC6 are cross-coupled as follows, with the input pins listed first:

• 6-5 goes to 6-3 (Q output)
• 6-2 goes to 6-4 (NOT-Q output)
• 6-1 is NOT-Set
• 6-6 is NOT-Reset

IC2-4, the debounced Start ("dump") button, goes through 0.001uF to IC6-1 (NOT-Set); this pin 1 also goes through 1megohm to plus V. IC2-11, the debounced Reset ("zero") button, goes to pin 13 of IC6; the output of this NAND gate, IC6-11, goes through 0.001uF to IC6-6 (NOT-Reset), with this pin 6 also going through 1megohm to plus V. AND'ed with the Reset signal is the gate which detects the stop character, dots 7 and 8. Thus, on IC6, pin 12 goes to pin 10.

IC6-4 (NOT-Q) goes to the Enables of the data latches, 3-5 and 4-5. IC6-3 (Q) goes to 9-27, the NOT-Write Enable of the RAM.

Read Timer

Pin 1 of IC10, a CMOS 555, is grounded. Pins 2 and 6 are tied together and go through 0.001uF to ground. Pins 2 and 6 also go through 470K to pin 3, the output. As mentioned, 15-9 goes to 10-3, thus combining the 9-input OR gate with the signal from the 555.

Pins 4 and 8 of the 555 are tied together and go to the output of a NOR gate in IC11, IC11-4. One input to this gate, 11-5, goes to the NOT-Q output of the control flip-flop, 6-4. The other input of the 555's controller, 11-6, goes to the "Busy Line" of the parallel port--the line by which the receiving device handshakes the Note-a-Braille. IC11-6 is held low by running it through 1megohm to ground.

Output Connections

The connections listed will be from the Note-a-Braille circuit directly to the Centronics type parallel interface pins. (Such a connector was not used on our Note-a-Braille box; we used a DB15 15-pin unit because of size. The wiring of this connector is left to the builder.) Ground of the Note-a-Braille circuit can go to any of pins 19 through 27 on the Centronics port.

The following pins on the EPROM go to pins 2 through 9 of the Centronics port:

• 8-8 to pin 2
• 8-7 to pin 3
• 8-6 to pin 4
• 8-5 to pin 5
• 8-4 to pin 6
• 8-3 to pin 7
• 8-2 to pin 8
• 8-1 to pin 9

Pin 11 of the parallel port is signalled by the receiving device and is called "busy." This pin 11 goes to IC11-6. The strobe output from the Note-a-Braille, pin 1 of the Centronics port, goes to pin 10 of IC11.

Additional Memory

The size of the memory can be doubled by adding a second RAM. Its connections and other changes necessary are described below.

The second RAM (we'll call it IC9B) has all of its pins connected in parallel to the first RAM (IC9A), with the exception of the chip select, pin 26. An additional bit (a new most significant bit) is needed from the counter; this is pin 12 of IC14. This is connected to pin 26 of IC9B. We need a NOT form of this bit to connect to pin 26 of IC9A. The unused half of IC15 is used as an inverter. We thus have:

• 14-12 to 9B-26
• 14-12 to 15-2, 15-3, 15-4, and 15-5
• 15-1 to 9A-26

We have to make one change in the end-of-memory circuit since we now have a new set of four most significant bits. This change is accomplished by moving the connection between pin 2 of IC12 and pin 4 of IC14 to pin 12 of IC14. Thus we have:

• 12-2 to 14-12

As stated in the section on output connections, the connector mounted in our cabinet is not the proper one for the Centronics port. This was done because the traditional parallel connector is quite large. Instead, a 15-pin D socket (DB15) was mounted at one end of the cabinet. A patch cord was made with a male DB15 at the Note-a-Braille end and a male 36-pin parallel connector at the other. To minimize problems of cross-referencing pins, the same pin numbers were adhered to on both connectors (with the exception of ground, which is left up to the builder). Thus, even on the 15-pin unit, pin 1 is strobe, pin 2 is data zero, pin 9 is data 7, and pin 11 is busy.

IC Information and Pin Connections

[Editor's Note: Where necessary, a brief paragraph describes features of a chip before the pins are tabulated. In order to keep explanatory phrases in the circuit description to a minimum, I have counted on the reader to refer to these pin diagrams throughout. You would be well advised to make loose-leaf copies of just the pin connections, so that when questions arise in the circuit as to what particular pins do, you can look them up quickly by IC number. In addition, diagrams of similar chips are not repeated. For example, IC's 5 and 12 (4023 and 4073) are closely related and have the same pin configuration; the given pin table is common to both. When you write out your loose-leaf copies, list the connections separately for each IC number, so that you won't go scrambling through the notes to find one that doesn't exist. You will notice in the tables that there are entries like "NOT Enable" or "NOT Chip Select." "NOT," here, means that the input function is inverted; i.e., a "NOT Clear" terminal is one for which you bring it low for "Clear" (a just plain "Clear" terminal would be high for Clear).]

IC's 1 and 2
Motorola MC14490
Hex Contact-Bounce Eliminator

Within this chip is a "Clock" oscillator whose rate can be set by an external capacitor. An input condition is declared valid only after so many clock pulses have been counted during its occurrence. The pins to which the capacitor is connected are "Osc. In" and "Osc. Out." The second of these chips has no capacitor; it is "clocked" by connecting its "Osc. In" to the first chip's "Osc. Out." You will notice that unused inputs are not committed to anything; internal pull-up resistors makes tying them high unnecessary.

MC14490:

• 8--VSS (Ground)
• 16--VDD
• 1--A In
• 15--A Out
• 14--B In
• 2--B Out
• 3--C In
• 13--C Out
• 12--D In
• 4--D Out
• 5--E In
• 11--E Out
• 10--F In
• 6--F Out
• 7--Osc. In
• 9--Osc. Out

IC's 3 and 4
Motorola MC14044, RCA CD4044
Quad Tri-State R-S Flip Flops

These R/S flip-flops are made with cross-coupled NAND gates. Therefore, when both inputs are high, the flip-flop is stable in either position. (One feature of this chip is that the disallowed state, both R and S low, is actually permitted; both Q and NOT Q go low for this condition.) The outputs go to high impedance when the Enable is low. With Enable high, when R is 0, the Q output is 0; when S is 0, the Q output is 1. Thus, being made of NAND gates, these flip flops should technically be called "NOT R, NOT S" flip flops. (The 4043 version uses NOR gates. The pin connections are almost the same on the 4043, except that pin 2 is the Q0 output and pin 13 has no connection.)

CD4044:

• 8--VSS (Ground)
• 16--VDD
• 2--NC
• 3--S0
• 4--R0
• 13--Q0
• 7--S1
• 6--R1
• 9--Q1
• 11-S2
• 12--R2
• 10--Q2
• 15--S3
• 14--R3
• 1--Q3
• 5--Enable

IC's 5 and 12
CD4023 Triple 3-Input NAND

CD4073 Triple 3-Input AND

• 7--VSS (Ground)
• 14--VDD
• 1, 2, 8--In1
• 9--Out1
• 3, 4, 5--In2
• 6--Out2
• 13, 12, 11--In3
• 10--Out3

IC's 6 and 11

CD4011 Quad 2-Input NAND
CD4001 Quad 2-Input NOR

• 7--VSS (Ground)
• 14--VDD
• 1, 2--In1
• 3--Out1
• 5, 6--In2
• 4--Out2
• 8, 9--In3
• 10--Out3
• 12, 13--In4
• 11--Out4

IC7
Motorola MC14538, RCA CD4538
Dual Retriggerable Resetable Monostable Multivibrator

On the first one-shot, a timing capacitor goes between pins 1 and 2 (pin 1 is internally grounded). A timing resistor goes from pin 2 to plus V. On the second one-shot, a timing capacitor goes between pins 15 and 14 (15 being internally grounded). A timing resistor goes from pin 14 to plus V. Although pins 1 and 15 are internally grounded, sample circuits in the literature show them grounded externally as well.

Each one-shot can be set up for positive- or negative-edge triggering. In this chip, the A and B inputs operate inputs of an OR gate; an inverter exists between the B input and the OR gate. Therefore, with B tied high (or used as an Enable), the one-shot responds to positive transitions of A. With A tied low (or used as a NOT Enable), the one-shot will respond to negative transitions of B.

MC14538, CD4538:

• 8--VSS (Ground)
• 16--VDD
• 1,2--Timing Cap1
• 2--Timing Resistor1
• 3--NOT Clear1
• 4--A In1
• 5--B In1
• 6--Q1
• 7--NOT Q1
• 15, 14--Timing Cap2
• 14--Timing Resistor2
• 13--NOT Clear2
• 12--A In2
• 11--B In2
• 10--Q2
• 9--NOT Q2

IC8, National NMC27C16
Erasable Programmable Read-Only Memory

The "Erasable Programmable Read-Only Memory" (EPROM) has eleven "address lines" (A0 through A10) and eight output lines (Out0 through Out7). This device converts the 8-bit codes from the Braille keys to ASCII characters, much like a conversion table might do. Its "program" for doing this must be "burned in" (loaded in) by Tom Fowle of Smith-Kettlewell; instructions for sending your EPROM to him will be given in the parts list. (Note that there is a pin called "Programming Power" which the programmer ties to 25 volts in order to load his program into the chip; this is tied to 5V in this circuit.) Note also that there are two disabling pins--"NOT Chip Enable" and "NOT Output Enable." Their effects, as far as putting the outputs in the tristate condition, are similar; use of the "NOT Chip Enable" saves power.

NMC27C16:

• 12--VSS (Ground)
• 24--VDD
• 1--A7
• 2--A6
• 3--A5
• 4--A4
• 5--A3
• 6--A2
• 7--A1
• 8--A0 (Least Significant Bit)
• 23--A8 (Not Used)
• 22--A9 (Not Used)
• 19--A10 (Not Used)
• 9--Out0 (Least Significant Bit) 10--Out1
• 11--Out2
• 13--Out3
• 14--Out4
• 15--Out5
• 16--Out6
• 17--Out7
• 18--NOT Chip Enable (Low to Enable)
• 20--NOT Output Enable (Low to Enable)
• 21--Programming Power

IC9, Hitachi HM6264LP-15
8K x 8-Bit Static CMOS RAM

Operated by a counter, this Random-Access Memory (RAM) stores the 8-bit numbers generated by the Braille keys. It too has "Address Lines" (A0 through A12); these are driven by the counter. It then has eight "Input/Output lines" (I/O0 through I/O7) on which you can "write" and/or "read" the keystrokes. It has three enable pins to choose from, all of which put the outputs in the tri-state condition; one of these just disables the outputs, while two others ("NOT Chip Select1" and "Chip Select2") allow you to disable the chip with a choice of signal polarities. We chose to tie the "NOT Output Enable" low and disable the whole chip, since less power is consumed this way.

6264LP:

• 14--VSS (Ground)
• 28--VDD
• 1--No Connection (NC)
• 10--A0 (Least Significant Bit)
• 9--A1
• 8--A2
• 7--A3
• 6--A4
• 5--A5
• 4--A6
• 3--A7
• 25--A8
• 24--A9
• 21--A10
• 23--A11
• 2--A12
• 11--I/O0
• 12--I/O1
• 13--I/O2
• 15--I/O3
• 16--I/O4
• 17--I/O5
• 18--I/O6
• 19--I/O7
• 20--NOT Chip Select1
• 26--Chip Select2
• 22--NOT Output Enable
• 27--NOT Write Enable (Low to Write, High to Read)

IC10, Intersil ICM7555
CMOS 555 Timer

• 1--VSS (Ground)
• 8--VDD
• 2--NOT Trigger
• 3--Output
• 4--NOT Reset
• 5--Voltage Control
• 6--Threshold
• 7--Discharge

IC's 13 and 14
Motorola MC14520, RCA CD4520
Dual Binary Up-Counter

This chip contains two 4-bit counters. As is often seen, the Clock and Enable terminals operate inputs of a NOR gate, with the Clock signal being inverted. Thus, either positive- or negative-edge triggering can be accommodated; with the Enable held high, the Clock terminal will trigger the counter on the positive edge, whereas if the Clock is held low and the Enable is used for triggering, the counter will respond to a negative excursion.

4520:

• 8--VSS (Ground)
• 16--VDD
• First Counter:
  • 1--Clock
  • 2--Enable
  • 3--Q0
  • 4--Q1
  • 5--Q2
  • 6--Q3
  • 7--Reset
• Second Counter:
  • 9--Clock
  • 10--Enable
  • 11--Q0
  • 12--Q1
  • 13--Q2
  • 14--Q3
  • 15--Reset

IC15, RCA CD4002

Dual 4-Input NOR Gate

• 7--VSS (Ground)
• 14--VDD
• 6--NC
• 8--NC
• 2, 3, 4, 5--In1
• 1--Out1
• 9, 10, 11, 12--In2
• 13--Out2

IC16, National LP2950-5.0
Micro-Power Voltage Regulator

This unit is in a TO92 package. With the flat side toward you and the leads pointing upward, the three leads are, from left to right: Input, Common, Output.

Parallel Interface Connector

The connector used for the standard Centronics parallel interface port has 36 pins. The shell is roughly trapezoidal, similar to D connectors. The socket (female), which is found on the parallel-to-serial converter, has a slot that runs almost its entire length; within this slot (on the top and bottom sides) are spring contacts. The male connector has a shell which surrounds the socket; in addition, it has a wide tongue with matching spring contacts for those in the slot of the socket.

As on D connectors, the pins are numbered from the point of view of looking at the tongue of the male or the solder lugs of the socket, and with the long side of the "trapezoid" up. Counting from left to right, the top row of pins includes 1 through 18, and the bottom row is 19 through 36.

A 15-pin D socket was used as the output connector for the Note-a-Braille. Looking at the back of the socket and with the long side of the trapezoid up, pins 1 through 8 are in the top row, and pins 9 through 15 are on the bottom.

Parts List

Integrated Circuits

• ICs 1 and 2--MC14490
• ICs 3 and 4--CD4044
• IC5--CD4023
• IC6--CD4011
• IC7--CD4538
• IC8--27C16 (See note below)
• IC9--6264LP-15
• IC10--ICM7555
• IC11--CD4001
• IC12--CD4073
• ICs 13 and 14--CD4520
• IC15--CD4002
• IC16--LP2950-5.0

Note: The 27C16 EPROM, which converts the Braille code to Standard ASCII, will be available in the "unrecorded state" from your supplier. (Be sure to obtain the CMOS version--not just the plan 2716, which would kill your battery in short order.) To have this chip programmed with the necessary conversion information, send it to: Mr. Tom Fowle, SKERF, 2232 Webster Street, San Francisco, CA 94115. This chip should be plugged into conductive foam, or should reside in an anti-static tube (it should come this way from the supplier). Pad the shipping container generously; mark it clearly with your return address. It will be returned to you "programmed" and with a piece of tape covering its erase window.

Transistors and Diodes

• 1--VN0300M
• 1--2N2222A
• 3--1N914
• 1--1N753

Capacitors

• 5--.001uF
• 1--330pF
• 2--47pF
• 1--33uF tantalum
• 1--10uF tantalum

Resistors (1/4 watt, 5%)

• 1--30-ohm
• 2--47K
• 2--100K
• 1--390K
• 1--470K
• 1--560K
• 5--1M

Switches

• 9--Cherry M81A-0100 keyboard switches
• 8--Cherry 639-0002 key caps
• 1--Cherry 639-0009 key cap (space bar)
• 2--SPST pushbuttons

Connectors

• 1--Jameco DB15S 15-contact socket 1--Jameco DB15P 15-contact plug
• 1--Jameco 57-30360 36-pin male Centronics connector

Enclosure

• 1--Pantec HPL-9VBK

Speaker

• 1--Panasonic EAF-14R06C, Digi-Key P/N P9922

Suppliers

• Jameco

  1355 Shoreway Road
  Belmont, CA 94022 (415) 592-8097

• Cherry Electrical Products Corp.
  3600 Sunset Avenue
  Waukegan, IL 60087 (312) 662-9200
• Digi-Key Corp.
  P.O. Box 677
  Thief River Falls, MN 56701 (800) 344-4539
• Pantec
  Enterprise and Executive Ave.
  Philadelphia, PA 19153 (215) 365-8400

THE JFD "UNIVERSAL INTERFACE CONVERTER"

[Note: This insert should be pulled out of this issue and stuffed into the Summer 1986 one describing the Note-a-Braille. The Adapta converter discussed in the "User's Guide" article has been phased out of the Jameco Catalog; this unit has supplanted it (as Jameco Cat. No. UIC-1). (Jameco Electronics: 1355 Shoreway Rd., Belmont, CA 94002, Phone: (415) 592-8097). For more information, or in case of trouble, call the Johnathon Freeman Designs "Tech-Support Hotline": (415) 822-8456.]

Abstract

This instrument is a versatile serial-to-parallel or parallel-to-serial converter. The most common use for such a device is running a printer--whose input is one format--from a computer whose output port is of the other format. Our particular use for it is in conjunction with the Smith-Kettlewell Note-a-Braille (see SKTF, Summer 1986). Having the capability of responding to software handshaking, this unit, made by Johnathon Freeman Designs (JFD), can interface with computers whose communications programs require handshaking of this sort; the Adapta converter, described in the "User's Guide to the Note-a-Braille," could only accept hardware handshaking.

Applications

First and foremost, the output of our Note-a-Braille is a parallel port. Whereas this could be used to drive a printer directly, the Note-a-Braille's most productive use is to feed notes from it into your personal computer. Usually this must be done through the serial port on the computer in conjunction with a MODEM program. Even if your computer has a parallel port, it is very unlikely that it can accept input via this port; usually, hardware in the computer is only provided to make the parallel port strictly an output.

More generally, though, such a converter is called for when there is a need to run a printer from a port of the opposite type. Suppose you have two printers, one for Braille and one for print, and let us further suppose that they both have serial ports and nothing else. If your computer were one which had two ports, one serial and the other parallel, this converter could be inserted to afford permanent connection of the "second printer," thereby eliminating the need to mess with printer cables every time you want to change from one to the other. Your collection of printers may include an older one whose serial parameters are unchangeable, thus requiring you to set the computer up for a slow baud rate or other old convention in order to use it. The Centronics Parallel Interface, on the other hand, has conventions which are so strictly adhered to that outfitting that antique with this box would allow you to use it freely from any computer with a parallel port.

Description and Features

This Johnathon Freeman converter is in a metal box measuring 4-3/4 inches wide, 5-1/2 inches deep, and 2 inches high. At one end, there is a female DB25 serial connector. At the other end is a 36-pin parallel connector for the Centronics port; this connector has wire rings at either end which can be used to lock the mating male plug in place (there is seldom a need for doing this, since the 36-pin plug has a very snug fit).

To the right of the parallel connector (as it faces you) is a 1/8-inch mini phone jack; this is the power socket. The power requirements are 9 volts at 350mA. A wall-mounted DC power supply comes with the converter. (Checking our power unit revealed that it provides a no-load voltage of 13 volts; this hasn't caused any problems, so it seems that the converter is well enough designed to cope with this power supply instability.)

To the left of the parallel connector--as it faces you--is an LED called the "Ready Light." This tells the user whether or not the converter is ready to receive data. It turns out that this is a fairly unimportant indicator; you can, however, see the converter respond to handshaking as described below.

Shortly after power is applied, the "Ready Light" should glow steadily. When the converter's 5K buffer is full, the light will go out for short periods of time. The unit will handshake its source of data, nibbling away at the information in small chunks of 256 bytes. (Files as large as 5K are rather rare from the Note-a-Braille--they would constitute over 60% of its memory.)

Worthy of mention is that this JFD has a self-test mode which not only demonstrates that it is working okay, but that you can use it to test printers and their interface connections. When actuated (done by installing a jumper block on "the eighth pair" of pins, as discussed later), the unit will send, "JFD UNIVERSAL INTERFACE CONVERTER" (followed by an ASCII character set). If it finds something wrong with itself, it will print a nice friendly failure message for you, at which point you should contact the Tech-Support Hotline: (415) 822-8456.

On the sides of the converter are four self-tapping screws which allow removal of the cabinet's cover. Once this cover is removed, orient the unit so that the serial connector faces you. Directly behind the serial connector is an arrangement of pins fitted with "jumper blocks." To the right of these pins is a dual in-line block of switches (DIP switches). Rather than being individual switches, these are physically "ganged" by plastic bars into two sections of four. They look like a chip on the board which has a ridge along its back, this ridge having a tiny gap in the middle. Though these switches are in two gangs of four, the intention is that they always be flipped together; never should one be in one position while the second is in the other.

The pins containing jumper blocks have to do with setting up the serial-interface parameters. (There is one exception: one jumper block is used for a "self-test mode," which will be described.) As you set this box up for your particular needs, you will acquire extra jumper blocks (we now have two extra ones after certain of them were removed). Save these extras, taping them to the inside of the lid, so that the unit can be modified again for a new port. You will need a pair of good strong tweezers for removing these jumpers.

There are two banks of jumpers. The left set is for baud rate and protocol settings; the pins for these are in a 2 by 8 array. The array of pins to the right is three pins deep; in each column of three, a jumper is either placed on the top and middle pins or the bottom and middle pins. This latter bank is for "configuration" (altering pin connections of the 25-pin connector). (On our unit, for some reason, the pins in the top row of the right-hand array are slightly shorter than the bottom two rows; this may or may not always be the case.) Basically, this right bank is a 3 by 9 array, except that there is a gap between the first seven and the right-hand two. This gap is rather handy; the first seven columns pertain to handshaking pins, while the right two columns control the wiring of pins 2 and 3--for DCE or DTE.

Setting Interface Parameters

[Note: All further discussion is done from the point of view that the serial connector is toward you; the parallel connector, together with the power socket and "Ready Light," should be on the end away from you.]

Caution

Jumpers and switch settings should not be changed with the power on. There is a reason other than putting the converter at risk. In some cases, what you're doing when you change these things is altering the program of a computer in the converter. The "program" of these switch and jumper settings is noted when you turn the power on; altering them afterward will go unnoticed, and you won't have changed what you think you have.

First, the ganged DIP switches are set to make the unit into a parallel-to-serial converter or serial-to-parallel converter as desired. When the switches are thrown away from you, the parallel port is the input and the serial port is the output. (This is what you want if the converter is being used with the Note-a-Braille.) When the switches are toward you--toward the serial connector end of the box--the unit is set up for "serial-in, parallel-out." (This is what you would want if you were running a printer's parallel input port from your computer's serial output.) In other words, when the switches are thrown toward the serial connector, this port becomes the input, and vice versa. As received, the factory setting is for parallel-to-serial.

Standards of the parallel interface are well adhered to; no setup procedure is required to interface it with various devices. (As the JFD manual warns, there are now beginning to be computers which use other connectors for parallel interfacing--even the DB25, the same as is used for serial interfacing. This may be done because of the smaller size of the substitute connector, as in the case of the Note-a-Braille. In other cases this is done, no doubt, to promote the engineering degree as a prerequisite to running even the simplest things.)

Suffice it to say that, when sending from its parallel port, handshaking is done via an "active-low" strobe signal on pin 1 of the Centronics connector. When receiving data through this Centronics port, the converter stops the transmitter by sending a choice of handshaking signals--an "active-high" on pin 11 and an "active-low" on pin 10. The eight data lines, "0" through "7," are on pins 2 through 9, respectively. Pins 19 through 27 are grounded.

As you well know, no such conformity exists between serial ports of various devices. You will have to find out what your serial equipment needs, and install the jumpers as necessary. As received, the factory settings are: 9600 baud, an 8-bit word, and no parity.

[For your reference, the left bank of pins are numbered in the manual as columns 0 through 7. Because these jumpers are con-fusing enough as it is, they will not be referred to by those numbers in this paper; instead, the "first four pairs" of pins will be shown to affect the baud rate, the fifth used to establish parity, etc.]

Setting the Baud Rate

Sixteen different baud rates are available, from 50 to 19200 baud. The first four pairs of pins--on the left end of the left-hand bank--are used to set the baud rate. A jumper is either left in or taken out; the following table lists jumpers as either "in" or "out" from left to right:

• 50--out out out out
• 75--in out out out
• 110--out in out out
• 134.5--in in out out
• 150--out out in out
• 300--in out in out
• 600--out in in out
• 1200--in in in out
• 1800--out out out in
• 2000--in out out in
• 2400--out in out in
• 3600--in in out in
• 4800--out out in in
• 7200--in out in in
• 9600--out in in in
• 19200--in in in in

[Note: The setting of jumpers progresses as a binary number whose least-significant bit is at the left end. If one could remember all these odd-ball baud rates, he wouldn't need the above table.]

Parity

If no parity is desired, the next pair--fifth from the left--is left open. Installing a jumper here sets you up for parity. Then, a jumper on the sixth pair selects "even parity"; leaving this pair open selects "odd parity."

Word Length

A jumper installed on the seventh pair selects 8 bits. Leaving this pair open sets you up for 7 bits.

Self-Test Mode

Putting a jumper on the eighth pair of pins in this first bank causes the converter to generate its name and a full ASCII character set. This message will recycle until the power is cut off and the jumper removed.

DCE or DTE Configurations

This is done using the right-hand two columns of three pins each. The far right column pertains to pin 2, while the one to its left (just to the right of the gap) pertains to pin 3.

Jumpering the middle to the upper pin puts the data line in the "output mode"; jumpering the middle to the lower pin puts it in the input mode. Therefore, having the right-most jumper on the top two pins and the one next door on the lower two creates DTE, just like a printer (pin 3 receive and pin 2 transmit).

Hardware Handshaking Lines

Software handshaking is always in effect; if Xon and Xoff characters are being sent or are needed from the converter, you can just assume that they are there. Hardware handshaking can be set up on various lines using the first seven columns of three pins--from left to right, just before a gap, beyond which are the data-line jumpers.

From left to right, the order of RS232 lines vs. column of jumper pins is:

• Column 1--Pin 20
• Column 2--Pin 19
• Column 3--Pin 11
• Column 4--Pin 8
• Column 5--Pin 6
• Column 6--Pin 5
• Column 7--Pin 4,

When the middle and top pins in a column are jumpered, this handshaking line becomes an input. When the bottom and middle are jumpered, the line becomes an output. (Note that this is just the opposite of the data lines of pins 2 and 3, where top is output and bottom is input.)

Example Setup

The editor uses this converter to send Note-a-Braille files to his Osborne computer. The first thing, then, is to make it a parallel-to-serial converter by flipping the DIP switches to the rear--toward the parallel connector, making it the input. Next, I know that my computer only needs a "straight-through" cable to drive a printer--a printer being "Data-Terminal Equipment" (DTE). Therefore, my computer must be "Data-Communications Equipment" (DCE). This means that it transmits printer information on pin 3, and would listen to my Note-a-Braille on pin 2. Consequently, I should put pin 2 of the converter's RS232 port in the output mode (jumpering the top two pins of the last column of three), and put the converter's RS232 pin 3 in the receiving mode (although this latter could be left open, since I'll never send to it).

My Osborne has no provision for hardware handshaking; it does a very traditional thing of software handshaking to a MODEM. However, as a cheap hardware compromise, it has pin 4 tied permanently high--kind of a perpetual "Clear-to-Send" signal. Although it really doesn't make any technical difference, for the form of it, I made line 4 a "Clear-to-Send" input by jumpering the top two pins of column 7 (just to the left of the gap).

Sometimes I like to demonstrate the Note-a-Braille by feeding it into a printer. To cover this contingency, I make an input out of RS232 line 20--jumpering the top two pins in column 1--so that the printer's "Data Terminal Ready" can stop the Note-a-Braille if it wants to. (Note, unless I go in and change the jumpers on columns 8 and 9--the ones for RS232 lines 3 and 2, respectively--I have to run the printer through a so-called "Null-MODEM Cable," which has wires 2 and 3 cross-connected. Note also that, as long as I am building a "Null-MODEM Cable," I could take care of the printer's handshaking desires by cross-connecting lines 20 and 4. This would eliminate the need for an input on line 20, and the converter would be true and proper CTE equipment, not some combination breed of my own design. It is sloppy practice like that advocated at the beginning of this paragraph which has bent the RS232 standard all out of shape.)

You might be puzzling over the fact that handshaking lines can be left open, and things will still work. This is because, in order to stop something, a handshaking line must pull an input to a negative voltage with respect to ground. In the RS232 standard, 0 volts is not a definitive statement; 0 volts is not a "logic 0." Pulling a line high is usually a definitive statement to "go ahead." Letting an input float around 0 volts means you don't care. Pulling a line low stops the show.

My Osborne can only handle two baud rates, 300 and 1200 baud. To set the converter up for 1200 baud, jumpers are installed on the first three pairs of pins in the left bank (in in in out). I want no parity, so the fifth and sixth pairs are left open. I need an 8-bit word length, so there is a jumper on the seventh pair.

Unless I send it to a printer, I can't really use the self-test mode. My computer wants to see an "End of File" character before it will let me see information sent to it. (With the Note-a-Braille, I do this by sending a Control Z.) One can, however, see the self-test data being sent on pin 2 of the serial connector--viewing this with an oscilloscope, a breakout box, or the Tweedle Dump (SKTF, Spring 1986).