The Smith-Kettlewell Technical File

A Quarterly Publication of
The Smith-Kettlewell Eye Research Institute’s
Rehabilitation Engineering Research Center

William Gerrey, Editor

Issue: [current-page:title]

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 Details regarding products, suppliers, and other contact information are original and may be outdated.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







by Jay Williams

In the Summer 1991 issue of SKTF I surveyed the field of electronic music instruments. Most of the material dealt with the use of a computer and associated programs to both record and play music through these instruments. I discussed two popular programs (often called "sequencers") for PC-compatibles that work well with current screen-reading/speech-output systems.

Introduction and Review

A sequencer provides editing features much like those in a word processor, but the data represent musical notes rather than letters of the alphabet. Data can be entered from either the keyboard on the instrument or from the computer's keyboard. In order for such a system to work, these instruments must be capable of handling data that conform to the MIDI protocol. They are connected to the computer via a MIDI interface card.

The conventional sequencer is fine for music that must sound the same with each repetition. However, in much popular music, and particularly in jazz, spontaneous variations are intrinsic features of both the melody and accompaniment. The sequencer I discuss in this article, Band-in-a-Box, is a clever, inexpensive program designed to work within the "popular song" format. It aims to provide singers and instrumentalists with accompaniments that sound "improvised."

Band-in-a Box

The basic building block of Band-in-a-Box is the "style." Each style contains its own "improvisation" routine comprising accompaniment patterns for a basic ensemble of bass, drums, "chording" instrument such as a piano or guitar, and a "melody" instrument. The program also contains many songs representing nearly every popular genre you can imagine. New songs and styles can be added to the program's library rather quickly, which makes this a potentially valuable resource for music teachers and budding performers.

Naturally, it cannot be considered a complete substitute for "real" musicians, either "live" or recorded. The program cannot tell when you are "on a roll" the way live musicians can, and a computer's "mental lapse" is often much less graceful than that of its human counterpart. Given those conditions, however, the accompaniment really does sound different each time it plays and the results can be quite convincing.

Many options are provided which vastly increase the complexity of the improvisatory schemes. The exploration of all these "ins and outs" can consume countless hours. I have been having lots of fun improvising along with these accompaniments and making additions to the already extensive library.

The choice of styles ranges through "Bossa Nova," "Country," "New Age," "Hip Hop," and many jazz-based styles. The songs cover a startling range from a modern jazz classic such as "Nardis" to "Hava Nagila." Although each song file includes an assigned style, you can reassign the style and the program will modify the song accordingly. The melody may be muted and replaced with your own improvisation.

Band-in-a-Box is available for the PC-compatible, Atari and Macintosh computers, so be sure to specify the correct format when you buy it. The PC version can be used with either the MPU 401 family of MIDI interface cards or with the Soundblaster and Adlib cards. Your MIDI system should include some form of "drum kit."

This is Version 5.0 of Band-in-a-Box. It comes in a higher- and lower-priced form. The higher-priced version provides fifty more styles which can be modified by the user. The less expensive one contains enough styles to keep you busy for a long time, and is such a vast improvement over earlier versions that you would not get a bargain by paying someone less money for one of these.

Data Management

A song file can be saved along with data that sets your synthesizer to the desired "voices." Since this is a program that assumes the use of electronic replacements of traditional instruments, it defaults to "MIDI G.S." parameters when loaded, unless otherwise instructed. (MIDI G.S. means "MIDI General Standard." It is being adopted for use in keyboards that feature instruments labeled "saxophone," "trumpet," etc. It ensures that such parameters as the program number for each instrument, and the data for such effects as volume, key velocity, reverb and delay will have a similar effect from one keyboard to another.)

You can also save your efforts as "Standard MIDI Files," which allows the music to be accessed with a standard sequencer. This would allow you to edit the accompaniment note by note, but you no longer have the benefit of the improvisation algorithms previously mentioned. Although a song file can include the lyrics, I found no lyrics provided with the songs already in the program. The help screens are thorough and are available from all parts of the program.

The format of Band-in-a-Box resembles that of a "fake book," a collection of songs whose accompaniments are written in a shorthand comprising chord names that occur along a line marked in measures and beats within the measure. In a "live" setting the performer improvises an accompaniment from these mnemonics. A Band-in-a-Box accompaniment provides the bass line, drum kit, chords, and if desired, the melody.

(This method of notating accompaniments was used by virtually every composer of the Baroque era. A bass line was supplied, above which the chords were designated by a letter that indicated the root of the chord, followed by numbers that showed their interval structure. This style of notation is termed "figured bass" and the whole accompaniment, a "continuo." The practice of writing down every note of the accompaniment dates from the late eighteenth century when drastic changes in musical style and the increased musical participation of the public at large required such specific information.)

How does it work? When Band-in-a-Box is loaded it defaults to a particular "style," one called "Jazz Swing." This and many other default settings may be changed according to your needs. Each style is simply a collection of accompaniment patterns for a minimal combo of bass line, drum set, and at least one chord-playing instrument.

When first loaded, there are no chord names present in the lines showing measure numbers, so if you hit F4, which plays the music, you will hear thirty-two measures of accompaniment patterns on a C major chord. The cure for this boredom is to type in the names of some chords (a process I describe shortly), or load in one of the many songs contained in the program. If you are not familiar with the sounds associated with these chord designations, the examination of a few songs will tell you quite a lot. After a while, you will be able to construct some pretty advanced progressions in a very short time.

In order to construct an accompaniment for a song, you simply work from the top of the screen to the bottom. First, pick a style from one of the two "styles" menus. One menu contains styles to which you can add patterns; the other contains "fixed" styles. Lines 2 through 6 allow you to enter the tempo, key, number of "lead-in" measures, number of choruses, the presence or absence of embellishments, placement of a "tag" on the end, volume settings, beginning and ending points of playback and whether or not the song repeats.

Lines 7-22 show 64 measures of music, four to a line. The names of the chords being played in a measure are written following the measure number. If your song is longer than 64 measures, press the page-down key to display 64 more measures.

Each measure and its attendant chord changes are highlighted as the music plays. Many screen-reading programs can be configured to call out such information as it appears.

Constructing a Chord Progression

After attending to the instructions in lines 2 through 6, press the page-down key. You will be prompted to enter a chord name or function. After typing the name of a chord, use the cursor keys to move through time to the next spot where a new chord should appear. The right- and left-arrow keys move two "beats" at a time; the up- and down-arrow keys move to the previous or to the following group of four measures. When you want the chord to change after just one beat, place a comma between two chord names. Erasure of a chord name is accomplished with the spacebar. The whole accompaniment can be cleared from memory by typing an alt-b chord or by selecting "blanksong" from the Edit menu and hitting "enter."

The Edit menu contains the usual features such as "cut and paste," expansion and reduction of note values, copying routines and storage buffers. A melody is edited in a conventional manner, but the accompaniment is edited in the "Stylemaker" menu. This menu contains all the instructions which tell the program how to "improvise." The brief description which follows will give you an idea of how this is done. The program's help screens provide a step-by-step procedure for constructing a style from scratch or modifying an existing style.

Editing a Style

Each line of the Stylemaker screen begins with the name of an instrument, followed by an A or B that designates one of the two "Substyles." Substyle A is the primary one and is employed throughout the process unless you instruct the program to pick a pattern from Substyle B. Following the substyle designation are the numbers that indicate the length and "weight" of the patterns assigned to the instrument.

If you are constructing a style from scratch, the remainder of the line contains thirty "periods," each of which denotes a "yet-to-be-recorded" accompaniment pattern. Once a pattern has been recorded, its "period" is replaced by the numbers that signify its characteristics.

The accompaniment patterns for the bass and piano are recorded in real time. First, use the cursor keys to select an instrument and an unrecorded pattern. After typing an "r" you are given two "lead-in" measures to establish the tempo. You then play the pattern on your synthesizer keyboard.

The program defines four pattern lengths: one, two, four, or eight "notes." (A chord in the piano part is treated as a "note.") The recording process stops automatically after eight "notes" or two measures. These patterns can be associated with specific chords in a progression. Further, there is an option which designates certain keys on your synthe as "hot" keys. When played, these evoke certain patterns which are likely to occur with often-used progressions.

Drum patterns can be recorded in a "step edit" mode. In this mode the z-through-/ row of keys enters progressively louder notes. Those little flourishes that drummers often slip in between choruses or in the vicinity of a particularly exciting chord change can also be introduced by typing a "p" at the appropriate measure.

The relative frequency of a pattern's occurrence is governed by a number from 0 through 9. A "0" mutes the pattern. A "masking" number can also be assigned. A "0" allows a pattern to be considered at any time; a "1," just once each measure; a "2," every other measure; a "3," every third measure, and so forth. Numbers can also be assigned for transposition, smoothness of voice-leading and other factors that make the accompaniment more or less melodic.

The bones I would pick with this program are few and minor. I wish that the styles and songs could be selected by category during the installation. When completed, even the cheaper version puts 397 files in the subdirectory and thus consumes over six megabytes of disk space. If you want to substantially increase your library in one category you will either have to use a second subdirectory or erase the unwanted files already there. Then again, maybe that simply shows my aversion to clutter.

It would be nice if the voicings of chords in the jazz styles had more of those nice, open, two-handed structures. Apparently, the musician who played these voicings decided to "play it safe" in case the user had a synthe which couldn't play more than eight voices at once. Fortunately, these styles will accept newly recorded patterns. Finally, the help screens could have used some proofreading. Spelling errors and the pluralization of nouns that should be singular make for some confusion. Oh well, use your word processor. Other than that, the creators of this program deserve congratulations and some good feedback for a clever and interesting product. There is a lot of "bang for the buck" here.


Band-in-a-Box is produced by PG Music, 11266 Elmwood Avenue, Buffalo, N.Y. 14222, phone (416) 528-4897. If your area lacks a good distributor of electronic music products contact MIDI Music Center, 959 Hill Road, Las Cruces, New Mexico, 88055, fax (505) 524-7356, phone (800) 333-2566. They have good prices and appear to be "user-friendly" folks.



The MAX680 and MAX681 provide a way to get a dual (plus and minus) supply from a single VCC. Moreover, their outputs are approximately twice the VCC input. Like all charge pumps, they cannot supply much current--10mA being a practical expectation. They can eliminate two batteries in a project when one or two op-amps require them.

How Do They Work?

Inside is an 8kHz squarewave oscillator controlling eight MOS power FETs used as switches. Equivalent circuits are discussed as follows:

A "pump" capacitor has its negative end going to two single-pole switches; one goes to ground, while the other goes to VCC. The positive end of this capacitor goes to two more switches; one of these goes to VCC, while the other goes to the positive output of the device. This positive output goes through a reservoir capacitor to VCC (positive of this cap at the output).

First, switches are closed to connect the pump capacitor between VCC and ground. Then, on the oscillator's second half-cycle, the pump capacitor is put in parallel with the reservoir cap--from VCC to the positive output. After a few cycles of this, enough charge is imparted to the reservoir to place a voltage in series with VCC so as to double it.

Another pump capacitor has its positive end going to two switches; one goes to the positive output of the device, while the other goes to ground. The negative end of this capacitor goes to two more switches; one goes to ground, while the other goes to the negative output of the device. There is a reservoir capacitor from the negative output to ground (positive of this cap at ground).

The second half-cycle of the oscillator connects this latter pump capacitor across the positive output of the device, thus charging it to VCC plus the voltage on the first two caps. Alternately, out of phase with the positive charge pump, the latter pump capacitor is put in parallel with the negative reservoir.

After a few cycles of this, the series voltage of VCC and the positive reservoir is developed across the negative pump capacitor, which then charges its reservoir. How's that for doing chin-ups on your own boot straps!

The disadvantage is that, when you draw current from this system, some charge in a reservoir is lost each time.

The literature speaks of nominal internal resistances of 150 ohms. Actually, as you would expect, there is not only a voltage drop when you draw current, but ripple appears on the outputs. The amplitude of this ripple is less when large-value capacitors are used (up to 100uF). Typical circuits show 10uF capacitors on the MAX680.

Note that the capacitors used for the negative supply have to be rated for a higher working voltage than those for the positive supply. This is true because the positive ones only see VCC, and can be rated at 6.3V. The pump and reservoir for the negative side see the full output--VCC plus the first pump and reservoir--so use 16V units.

The MAX681 requires no external capacitors; it has 1uF caps internal to the chip, available in a 14-pin DIP. The MAX680 requires four external capacitors.


VCC can be from 2V to 6V. They point out that the device "looks" like a zener diode above 6.2V, so neither package should be operated above 6V.

The negative output can be short-circuited continuously. Because of the nature of the circuit, current taken from the positive output should not exceed 75mA (you'd be drawing current from VCC through the device).

They list a "maximum rate of change of VCC" as 1V/usec (one volt per microsecond). I don't know the origin of that, but you can bet that when VCC changes, a few cycles of the internal oscillator, perhaps for a millisecond, will be needed so that the charge-pumps follow.

For a VCC of 3V and with the outputs left open, typical supply current is 500uA, maximum 1mA. For a VCC of 5V and with the outputs left open, supply current is typically 1mA, maximum 2mA.

With the positive output working into a 10mA load, and with a 5V VCC, the typical output resistance is 150 ohms, maximum 250 ohms. With a 2.8V VCC and a 5mA load, a typical output resistance is 180 ohms, maximum 300 ohms. Thus, "output resistance" goes up as VCC is lowered; it also goes up slightly as temperature increases, and naturally, it effectively is larger if smaller capacitors are used.

For the negative output, a whole set of figures is given with the condition of no load on the positive output. Remember that the negative supply output is derived from the positive one. Suffice it to say that any load on the positive output has a deleterious effect on the negative output, and a load on the negative output alone will affect the positive.

Consider the following conditions with a VCC of 5V (input) and with 10mA loads being applied: With only a positive load, the positive output drops to 8.7V; the negative output drops to 8.3V. With a load only on the negative output, it drops to 7.4V.

V out vs. load current is fairly linear. Ripple, on the other hand, greatly increases with load current. For example, ripple on the plus output goes from 70 to 175 millivolts as the load is increased from 10mA to 20mA. Likewise, ripple on the negative output goes from 80 to over 200 mV as the load goes from 10mA to 20mA. (These figures are for the MAX680 with 10uF capacitors.)

The ripple frequency is that of the internal oscillator--8kHz typical, 4kHz minimum.

Good Quick Summary

These chips operate from 2 to 6 volts. The 680, which requires four external capacitors, will perform better as the caps are made larger. With 5V in (VCC) and with a 5mA load from minus to plus outputs (such as an op-amp), the output voltages will be minus 6 and plus 7 volts.

Pin Assignments for the MAX 680

  • 1--C1 minus
  • 2--C2 plus
  • 3--C2 minus
  • 4--V minus (output)
  • 5--Ground
  • 6--VCC (input)
  • 7--C1 plus
  • 8--V plus (output)

C1 and C2 are the "pump" capacitors. C1, the one used for the positive output voltage, goes between pins 1 and 7 (negative of this cap at pin 1). C2, the cap used for negative output, goes between pins 2 and 3 (negative of this cap at pin 3).

The reservoir caps are C3 and C4. C3, the one for positive output, goes between pin 8 and pin 6 (between the positive output terminal and VCC, its negative at pin 6). C4, the reservoir for the negative output, goes between pin 4 and ground (negative of this cap at pin 4).

Pin Assignments for the MAX681

Note: This chip requires no external capacitors, as it contains equivalent ones of 1uF.

  • 1 and 10--V plus (output)
  • 2 and 3--C1 minus
  • 4--C2 plus
  • 5 and 6--C2 minus
  • 7--V minus (output)
  • 8 and 9--Ground
  • 11 through 14--VCC

Sample Circuit

This little mixer could drive an earphone, while putting a high-level source and a microphone together. Although it needn't be done this way, a MAX680 is used to create a dual supply for an op-amp.

Circuit for an Op-Amp Mixer with Dual Supply

A 3V battery is used; its negative terminal is grounded, while its positive terminal goes through a switch to the 3V (VCC) line. Pin 5 of a MAX680 is grounded, while pin 6 goes to the VCC line.

A 10uF electrolytic capacitor goes from pin 7 to pin 1 of the 680 (negative at pin 1). Between pins 2 and 3 is another 10uF cap (negative end at pin 3).

Pin 4 of the 680 goes to the negative end of another 10uF cap; the positive end of this cap is grounded. Pin 8 of the 680 goes to the positive end of a 10uF cap; the negative end of this cap goes to the VCC line.

Pin 4 of an LM358 dual op amp goes to pin 4 of the 680. Pin 8 of the 358 goes to pin 8 of the 680.

Pins 3 and 5 of the LM358 are grounded. Between pins 1 and 2 of the 358 is a 1 megohm metal-film resistor. Likewise, between pins 6 and 7 is another 1 megohm resistor.

Pin 6 of the 358 goes through a 62K resistor to pin 1. Pin 6 also goes through a 1 megohm resistor to the tip of a high-level input jack; the sleeve of this jack is grounded.

Pin 2 of the 358 goes through a 62K metal-film resistor to the tip contact of a microphone jack; the sleeve of this jack is grounded.

Pin 7 of the 358 is the output; this goes through two earphones of a stereo headset to ground. To raise the impedance, put the two earphones in series by using only tip and ring contacts; let the sleeve float.

You can also put the output into a tape recorder or some such. Just in case the 6 millivolts of offset worry you, pin 7 goes through a 10uF capacitor to the hot output lead; the cold output lead is grounded.

Doesn't the lack of coupling capacitors make you happy? I wish this world were full of split supplies.


Maxim Integrated Products Inc., 120 San Gabriel Dr., Sunnyvale, CA 94086; Phone: (408) 737-7600.



In a previous article, "RS232 Receiver and Line Driver Chips" (Winter 1985), we described the 1488 and 1489. Those required a dual (plus and minus) supply. With the advent of charge-pump DC-to-DC converters, this complication is unnecessary; only a 5V supply is required. Several different versions exist; as examples, we have chosen the MAX232 (the most plain), the MAX220 (the low-power version of the 232), the MAX222 (which is tristate), and the MAX242 (in which receiver outputs and driver outputs can be separately tristate). Other than for computer/controller applications, if you wish to send logic signals over long lines with immunity to noise and crosstalk, these are the chips to use.

General Description

Each of these chips has three sections: They all have DC-to-DC converters, identical to the MAX680 described in the previous article. Each chip has a pair of RS232 drivers which accept TTL levels and convert them to plus/minus 10V levels required for the RS232 system. Each chip also has two "receivers"; these accept plus/minus 3V (or greater) RS232 levels and convert them to 5V TTL logic.

Power Supply Considerations

A 5V logic supply is used to power these chips. Except for the 220, supply current (with no load on the drivers) is typically 4mA, 10mA max. The 220 (unloaded) typically draws 0.5mA, 2mA max.

The DC-to-DC converters have exactly the same hookup as the MAX680 (using different pin numbers, though). A 4.7uF electrolytic charge-pump capacitor goes between C1 minus and C1 plus pins, and another 4.7uF charge-pump cap goes between C2 plus and C2 minus pins. A 10uF cap goes from the V plus output to the 5V VCC, and another 10uF cap goes from the V minus output to ground. As with all these charge-pump circuits, the negative charge-pump and reservoir capacitors must be rated at 16V, while the positive ones can be 6V units. For some reason, they place a limit on the size of these capacitors--no greater than 10uF.

The plus and minus V outputs can supply a small amount of current to other circuit elements and still stay within the RS232C requirements. Drawing 5mA from the negative output, for example, lowers the negative unloaded driver output to about 7.5 volts; this is the worst case.

When the "shut-down" tristate feature of the 222 and 242 is activated, the charge-pump is turned off; V minus goes to zero, and V plus goes to 5V. This being so, the time to reactivate after bringing the chip out of tristate is proportional to the capacitors used.

External plus/minus 10V supplies can be used to power these chips by connecting them to the V plus and V minus outputs. The 5V VCC is still required. The C1 pump capacitor must not be connected. Very important is that the tristate feature not be used; the "NOT shut-down" pin must be tied to 5V. When in tristate, the V plus pin goes to VCC through various workings in the chip.

The RS232 Drivers

These take single-ended logic levels and convert them to bipolar signals of at least twice the amplitude. When presented with 5K RS232 inputs, the typical output voltages are plus/minus 8V. The output swing is guaranteed to meet the plus/minus 5V standard when presented with 3K loads.

Unlike standard TTL, the inputs are both CMOS and TTL compatible. They are high impedance, and current sinking is not required. Switching takes place at about 1.4V (0.8V minimum--2V maximum).

Uncommitted inputs can be left open, since they have built-in 400K pull-up resistors. These resistors disappear when a chip is shut down; this minimizes supply current.

When shut down, the driver outputs are turned off (going into tristate mode). The drivers are inverters. This is well and proper; in the RS232 system, a positive voltage is considered a logic 0, while a negative output is a logic 1. Because of the pull-up resistor, a driver whose input is uncommitted will have a negative output.

Except for the MAX220 (the low-power chip), all other driver outputs have a slew rate of 12 volts per microsecond (12V/us) when presented with loads of 3K in parallel with 2500pF. This slew rate is 24V/us for unloaded outputs. This allows data rates of over 116 kilobytes per second. The 220 is much slower, allowing for a data rate of 22kbs.


These take bipolar high-amplitude signals and convert them to TTL logic outputs. Commensurate with RS232C, the allowable input swing can be plus/minus 15V. A voltage greater than plus 3V is a logic zero; a negative voltage greater than minus 3V is a logic 1. (Drivers have to meet a standard of plus/minus 5V. All this is per the EIA Penal code RS232D/V.28.) Like the drivers, the receivers are inverters.

While the allowable input swings are bipolar, the receiver inputs will actually work at TTL-level voltages--0.8V and 2.4V. There is a guaranteed hysteresis of 0.2V. The inputs have 5K resistors taken to ground.

Except for the MAX220, the receiver propagation delay is, at maximum, 1 microsecond. In the 220, this can be as much as 3 microseconds.

Shut-Down Conditions for the MAX222 and MAX242

When the "NOT Shut-Down" pin is brought low, these chips go into tristate mode. For the driver sections, this means that the outputs float; also, the 400K input resistors disappear.

In the 222, the receiver sections are disabled when the chip is shut down. In the 242, the receivers still operate, but in a reduced-power mode; the propagation delay increases to 2.5 microseconds. In this mode, the 242 receivers act like CMOS inverters with no hysteresis.

So that receivers in the 242 can be fully tristate, this chip has a "NOT Output Enable" which, when brought low, disengages the outputs.

Pin Assignments for the MAX220 and MAX232 (16 pins)

Power Supply Section:

  • 15--Ground
  • 16--VCC
  • 1--C1 plus
  • 3--C1 minus
  • 4--C2 plus
  • 5--C2 minus
  • 2--V plus
  • 6--V minus

Driver Section:

  • 11--Input to Driver No. 1
  • 14--Output of Driver No. 1
  • 10--Input to Driver No. 2
  • 7--Output of Driver No. 2

Receiver Section:

  • 13--Input to Receiver No. 1
  • 12--Output of Receiver No. 1
  • 8--Input to Receiver No. 2
  • 9--Output of Receiver No. 2

Pin Assignments for theMAX222 and MAX242 (Tristate with 18 pins)

Power Supply Section:

  • 16--Ground
  • 17--VCC
  • 2--C1 plus
  • 4--C1 minus
  • 5--C2 plus
  • 6--C2 minus
  • 3--V plus
  • 7--V minus

Driver Section:

  • 12--Input to Driver No. 1
  • 15--Output of Driver No. 1
  • 11--Input to Driver No. 2
  • 8--Output of Driver No. 2

Receiver Section:

  • 14--Input to Receiver No. 1
  • 13--Output of Receiver No. 1
  • 9--Input to Receiver No. 2
  • 10--Output of Receiver No. 2

Tristate Pins:

  • 18--NOT Shut-Down for both
  • 1--NOT Receiver-Output Enable for MAX242; NC for 222

Sample Circuit

Suppose I want a 2-input AND gate across the room. One might want to do such a thing if a high-current driver, such as the TI 75452, were to operate something over yonder, while giving us a signal back at our end of the room. I would run five wires and hook two MAX232's up as follows:

Two-Input AND Gate with Very Long Arms

The negative terminal of a 5V supply is grounded; its positive terminal is VCC. Pin 15 of each chip is grounded; pin 16 of each goes to VCC. The charge pumps of each are set up as follows:

A 4.7uF 6.3V electrolytic goes between pins 1 and 3 (positive toward pin 1). A 4.7uF 16V electrolytic goes between pins 4 and 5 (positive toward pin 4). A 10uF 6.3V electrolytic goes between pins 2 and 16 (negative at pin 16). A 10uF 16V electrolytic has its negative end going to pin 6; its positive end is grounded.

A TTL or CMOS NAND gate, powered from the 5V VCC, has its inputs going to receiver outputs 9 and 12 of the 232 across the room. The NAND gate's output goes to that 232's driver input, pin 11. The unused input, pin 10, can be left open.

An inverted version of the NAND gate--at RS232 levels--exists on pin 14 of the 232 across the room; this goes through a wire to pin 13 of the MAX232 at our end of this mess. Driver outputs of our driver, pins 14 and 7, go through wires to receiver pins 13 and 8 across the room.

At our end, pins 10 and 11 are inputs to the AND gate, while pin 14 is the output.


This December, your fair-hairless boy got himself in big trouble, being diagnosed with adult-onset diabetes. Newly introduced to the world of injections of insulin and blood testing, it is clear to me that some things about that business were not worked out for us blind folks. To the extent that I can, I aim to change the situation.

There is a good glucometer on the market which is easy to make talk. Made by Lifescan, Inc. (a division of Johnson & Johnson), it is called the One-Touch II. AFB and Science Products both make talk boxes for it.

A stereo 1/8-inch jack on the side of the instrument carries serial data (at TTL levels) so that a doctor can get a print-out of readings from the instrument's memory. Lifescan has been wonderfully forthcoming with the needed information to hook their instruments up to talk boxes. In fact, if you call their number, (800) 227-8862, they give you a serial cable for your RS232C serial port--free.

Making the One-Touch meters talk is a song, either through your computer or the specific talk boxes designed for it. Loading the test strip with blood is quite another matter.

The glucometer has a plastic fixture that accepts the test strip. It looks rather like a battery door, and it can be removed for cleaning. (I hope you're not reading this over breakfast.)

Glued onto the backside of the test strip is a little piece of filter paper which has been treated with a couple of enzymes. With an optical device, the meter "looks" at the back of this strip, viewing it through a tiny window in the strip holder.

On the top side of the strip, there is a little hole--0.19 inches in diameter--providing access to the filter paper beneath. You are not to touch the strip to find the hole. In order not to smear the blood (which ruins the reading), you must dive-bomb a drop of blood from above into that tiny circle. If you goof, you have to prick your finger again (an insult to the soul), and you've wasted a 75-cent test strip (an insult to your pocket).

Here's another variable: You're supposed to vary the puncture site--using the sides of your finger pads--choosing a slightly different place each time. The meter is basically smooth on top, so: where's the hole?, where's the blood?, where am I?

An older model of the One-Touch had a stainless-steel fork holding the strip, and that helped folks position themselves. This slick new model only has a slight depression which, when you've found it, you're too low and the job is messed up.

I've been fooling with funnels and stuff to see if blood cannot be fed to the test site in a foolproof manner. I have proven myself a fool, since you cannot count on blood to "run" anywhere with certainty. Guides may be our only arsenal.

Luckily, you can buy the plastic "strip holders" in quantities of fifty, so I can screw, glue, tape, or nail things to them 'til I get it right. I have come up with first approximations of Dymo-tape guides which you can attach yourself, and these are described as follows:

Dymo Braille Position Markers for the One-Touch II

The punctured finger has to be aligned with the hole in the longitudinal direction, and the exact center of the meter is the correct lateral position. The guides which have helped me have either a line of colons or X's, with a different character marking the edges of the depression at the edges of the test strip. On my favorite model, XXXY is to the left of the strip, while &XXX resumes the pattern to the right of the strip. Working equally well are three colons leading up to a W on the left, then an R followed by three colons to the right.

The width of the strip-holder's depression is exactly two Braille cells, so both left and right marks can be made on one piece of tape, after which the piece in the center is removed. For getting its position right, I have a trick.

In the blank spaces between the markings, I place dots 5 and 2. In other words, I create a couple of dimples in the exact center of my strip of tape that I can feel with a stylus, fishing up through the optical window in the strip holder. Specifically, my tape with the X's is punched as follows:

X, X, X, Y, dot 5, dot 2, the AND sign, X, X, X.

Then, I cut through just the backing of the tape with a knife blade positioned between dots 5 and 2; in this way, you can peel part of the backing leftward, and the other part rightward. I peel back enough so that the door can be fitted to the sticky part. Feeling through the window with a stylus, I position the tape so that both dots can be reached with the stylus tip. I next cut off the flaps of backing at the edges of the door, leaving the backing in place for the left and right ends of the tape (which assures that this assembly is removable for cleaning). Finally, using a sharp knife blade, I cut the middle section of the tape up close to the markings and remove it.

Using the Dymo Markings

These guides are for fingers of the opposite hand. One stands the thumb and a finger vertically on the markings and puts the punctured finger in between them. There's another practical reason to have these fingers there: once the finger is punctured, you have to squeeze it to get blood out of it.

The Smith-Kettlewell Insulin Dipstick

People's main way of judging how much insulin is left is keeping a meticulous log of usage. Insulin is fragile, so you're not supposed to shake the bottle.

There is a neat syringe made by the Novo Corporation. I must say, for a person who constantly complains about bad mechanical engineering--as I do--(with the worst examples being tissue dispensers in public rest rooms), I think the "Novolin Pen" syringe is truly a marvel. Its plunger is of variable length; one turns a ring so many clicks to "dial in" the desired dosage. Its design is intricate and using it is simple. Its insulin vials are cartridges with a rubber plug that serves as a piston. This is good for us, since by noting the position of the piston, the available insulin can be properly assessed.

I had our machinist, Don Ortiz, make fixtures that look like a "Click Rule" or "Rotomatic Rule." Made from 1/4-inch round stock, one side is milled flat so that they won't roll away. The other side has ribs that divide them into five lengths.

The insulin vial holds 1.5 milliliters of juice--150 "units" of insulin. It contains a little mixing ball; one rolls the vial back and forth between palms if a mixture of fast- and slow-acting insulin is contained. Mark Dubnick, a biochemist and a reader of this fine magazine, tells me that in order to protect the user from getting a mixture heavy in slow-acting stuff, the pen won't let you have the last twelve units for fear that the heavier insulin might constitute an overdose in six hours or so.

Thus, the dipstick need not be a tool of great accuracy. What you want to know is, "Do I have enough to go on this all-day trip, or not?"

To account for the 0.12-inch distance to the plug in a full vial, a dowel pin is driven into the front end of the dipstick. In other words, a protrusion at the front end clearly shows the user which end goes in first. The overall length of the dipstick is 1.75 inches. Four ribs divide up its length into five sections, these ribs occurring approximately every 0.33 inches.

One inserts the dipstick--dowel pin first--into a vial and counts the exposed ribs. Each exposed rib accounts for 30 units of insulin. A Tinker Toy stick or piece of piano hammer shank which is 1.75 inches long is almost good enough, so you can easily make your own.

Recapping the Syringe

Although manufacturers cannot advocate it, some people use a needle more than once, the theory being that your pathogens are your pathogens, so why not. The reason for doing this is the expense of changing needles. (Insulin-dependent diabetes can cost you upwards of $3000 a year.)

Well, you can imagine how hard it is to put a cap--the size of that from an oil can--over a 1/2-inch long needle which you are enjoined not to touch with your fingers. That's chasing a rainbow, pal.

The Novolin Pen comes in a nice carrying case. So, I uncapped its needle, laid it into its cradle and "sent it forth" to mark the exact spot that the needle would hit the end of the case. Then, with a small drill in a pin vise, I drilled a hole in the side where the needle made its mark. Then, with a larger bit, I made a hole that fit the cap. (I gave it away already, so I don't precisely know the size.)

As soon as I uncapped the syringe for use, I installed the cap into this hole. When I was done, I set the pen in its cradle and sent it forward into the cap. Dashed clever, myself.


Why didn't you guys tell me of these problems. I am an okay engineer, and I can help--I think. I won't soon forgive myself for not working on the glucometer guide; I should have been working on this years ago.