sktf-Winter-1991

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

William Gerrey, Editor

Issue: sktf-Winter-1991

Original support provided by: The Smith-Kettlewell Eye Research Institute and the National Institute on Disability and Rehabilitation Research

Note: This archive is provided as a historical resource. Details regarding products, suppliers, and other contact information are original and may be outdated.

Questions about this archive can be sent to sktf@ski.org

The ISD Family of Voice-Message Chips

THE ISD FAMILY OF VOICE-MESSAGE CHIPS

by Bill Gerrey and Albert Alden

[Editor's note: Further notes on this project were published in a later issue. Please click here to read the article of additional notes.]

Abstract

Business magazines have recently carried announcements of this new breed of speech-recorder chip. Made by Information Storage Devices, Inc. (ISD), the ISD1012, ISD1016, and ISD1020 are 28-pin chips that can store 12 seconds, 16 seconds, or 20 seconds of sound, respectively. They have a built-in mike preamp and a power amplifier capable of driving a speaker, and once a recording is made, they will store it without battery backup. Finally, they are addressable by way of binary "data lines," so that many short messages can be selected for playback.

Introduction

Thanks be to all who have written the Editor to alert me to the existence of this new series of chips. In some cases, it can substitute for the Nattering RAM--the main disadvantage being that one has no access to the on-board clock.

Chips can be cascaded for more recording time. Three ISD1020s can record for one minute.

They speak of this chip in terms of "analog storage." While they don't detail how they do it, their mention of EEPROMs, "sampling rates," and our assessment of the sound of the played-back recording, suggest that the audio is digitized for storage and decoded for playback. The frequency response is "telephone line quality," varying with the choice of digital sampling rate used in the three different chips. The ISD1016, sampling at 8kHz, has an upper frequency cutoff of 3.4kHz. The 1012 has an upper cutoff of 4.5kHz, and the 1020 has a cutoff of 2.7kHz.

The on-chip mike preamplifier is noisier than one you can build yourself. If you build an outboard preamplifier, the recording is fairly noise-free (although you give up the convenience of their 19dB speech compressor).

We find the chip to be of clever design, and it well deserves our attention. As of yet, it's a little pricey--$54--but considering the trouble it saves, providing most of what you need on a single chip, it can make the designer's job easy.

Specifications

Note: These specs are given for a supply voltage of 5V. Also, they are specific to the 1016. I reckon there is a chance that such specs as Data Setup Time could be clock dependent, and might be longer for the 1020, for example.)

Supply Considerations:

Absolute Maximum Supply Voltage--10V
Absolute Voltage Excursions of Any Pin--0.3V outside the "rails" (no lower than VSS minus 0.3V and no higher than VCC plus 0.3V); however, if signals are current-limited to 20mA, pins can be taken one volt beyond the rails.
Operating Current--25mA (with the speaker terminals left open)
Standby Current--10uA (This is with the "Power Down" pin high; no current need be drawn to retain the memory.)

Analog Inputs and Outputs:

Preamp Input Resistance--Typ. 10K
Auxiliary Input Resistance--Typ. 10K
Analog (high-level) Input Resistance--Typ. 2.7K
AGC Output Resistance--Typ. 5K
Preamp Output--can source or sink Typ. 100uA

Preamp Gain:

Wide Open (no AGC)--24dB
With Maximum AGC--Max. 5dB
Maximum Mike-Input Voltage--20mV P-P
High-Level Input Gains:
Auxiliary Input Gain--0.9 Min., 1.0 Max.
Auxiliary Input Voltage--1.25V Max.
Analog (high-level) Input Gain to Speaker--Typ. 22dB
Analog Input Voltage--80mV P-P Typ.
Power Amplifier Characteristics:
Output Load Impedance--Max. 16 ohms (can drive an 8-ohm speaker, but may distort slightly)
Speaker Output Power (with 16-ohm load)--50 milliwatts Max.
Peak-to-Peak Voltage Across Speaker Terminals (with 600-ohm load)--Max. 2.5V
ISD1016 Record/Playback Characteristics:
Bandwidth--3.4kHz
Total Harmonic Distortion (at 1kHz)--Typ. 2%
Signal-to-Noise Ratio--Typ. 35dB
Logic Characteristics:
Max. Logic Low Input Voltage--0.8V
Min. Logic Input High Voltage--2V
Max. Logic Low Output Voltage--0.4V
Min. Logic High Output Voltage--2.4V
(When drawing very little current, a logic high will be VCC minus 0.4V)
Control/Address Setup Time--Min. 300 nanoseconds
Power-Up Delay--Min. 25 milliseconds
End-of-Message Pulse Width--Typ. 12.5 milliseconds
End-of-Message Marker--within 25 milliseconds of terminating recording
These chips have an internal low-battery-voltage detector. When a chip's supply voltage drops below 3.5V, the chip will not go into record mode; it will only play back. Moreover, as the battery develops internal resistance, voice quality will suffer.

Pin Assignments

Address Lines:

Pin 1--A0 (least-significant bit)
Pin 2--A1
Pin 3--A2
Pin 4--A3
Pin 5--A4
Pin 6--A5
Pin 9--A6
Pin 10--A7 (most-significant bit)

Power Connections:

Pin 28--VCCD (digital)
Pin 16--VCCA (analog)
Pin 12--VSSD (digital ground)
Pin 13--VSSA (analog ground)

Signal Pins:

Pin 14--"Speaker Plus"
Pin 15--"Speaker Minus"
Pin 11--Auxiliary Input
Pin 20--Analog Input (high-level)
Pin 21--Preamp Output
Pin 19--AGC
Pin 17--Mike Input

Control Pins:

Pin 24--Power Down (high for standby)
Pin 27--Play/Record (low for record)
Pin 23--NOT Chip Enable (low to enable)
Pin 25--NOT End of Message (goes low for 12.5 milliseconds for end-of-message marker, stays low for memory overflow)

Miscellaneous:

Pin 26--"Test Pin" (used by the factory before shipping; this is always to be grounded)
Pins 7,8,18, and 22--No Connection

Power Supply Considerations

All circuits are shown with a supply of 5 volts, plus/minus 10%. (Note that there is an "absolute maximum" rating of 10V, but this is the cooking voltage.)

There is a chance that the chip will motorboat unless good bypassing or power-supply decoupling is employed. The literature shows an arrangement where the 5V supply is bypassed by 22uF; then, separate ground buses (for pins 12 and 13, digital and analog grounds) are provided. Likewise, two VCC buses supply pins 28 and 16 (digital and analog VCCs). Located close to the chip, pin 16 is bypassed to pin 13 by 0.1uF, and pin 28 is bypassed to pin 12 by another 0.1uF. They specify that you use caps of low "effective series resistance"--ceramic will do. We chose more drastic decoupling measures.

We eliminated the motorboating by supplying the digital VCC (pin 28) directly, bypassing this by 22uF and 0.1uF in parallel. Then, we supplied pin 16 through a 33-ohm resistor, bypassing pin 16 by 10uF and 0.1uF in parallel. (This may reduce the output power; you can try lower resistance values if you like.)

Speaker and Mike Considerations

The literature suggests various microphones and speakers. These will be listed at the end of this paper.

The power amplifier in the chip is most happy with a 16-ohm speaker, for which they give some part numbers. An 8-ohm speaker will be louder, but may distort; the remedy for this is to put a 7.5-to-20-ohm resistor in series with the speaker, but power will be lost in the resistor.

The two output terminals comprise a differential driver which must be isolated from ground; remember this when mounting an earphone jack in a metal cabinet. On the other hand, either output terminal can drive a low impedance as a single-ended load, if a coupling capacitor is used (perhaps 100uF, with the positive end toward the chip).

The capacitor microphones they recommend require power, and they should be supplied using decoupling techniques.

Microphone Supply Circuit

The negative side of the mike is grounded. The mike's positive supply lead goes through 10K, then through 2.2K to the 5V line; the junction of these resistors is bypassed to ground by 10uF (negative of this cap at ground). Mike's output may be via a separate shielded cable. However, some of these units are 2-lead devices; in that case, the mike output is at the junction of its supply lead and the 10K resistor.

Miscellaneous Quirks, Foibles and Features

The purpose of the "Auxiliary Input" is so that one power amplifier can be used when cascading several chips; as you will see, one of the output pins of each successive chip goes to this pin. This means, though, that this input and the power amplifier can be used for other aspects of a project when the ISD chip is not playing back.

The power amplifier is inoperative when the chip is in record. Another feature, the fast-forward mode, also disables the power amplifier. Naturally, the "Power-Down" condition disables the power amp as well.

The AGC pin requires a parallel combination of external resistor and capacitor. The "recovery time" of the AGC can be changed by choosing a different resistor than the one shown. Both "attack" and "recovery" are affected by the value of the capacitor; the output impedance of the AGC pin is 5K.

They mention that the preamplifier is wide open when the AGC voltage is 1.5 volts or less. When the AGC is working, voltages around 1.8 volts are typical.

The output of the microphone preamplifier and the high-level input (the so-called "Analog Input") are brought out to pins; an external coupling capacitor goes between them. This is good for three reasons: First, high-level sources can be recorded. Second, low-frequency cutoff can be determined by this capacitor; the input resistance of 2.7K, together with a series capacitance of 0.22uF, constitutes a high-pass filter whose break-point frequency is 370Hz--desirable for speech. Third, a quieter preamp can be added externally.

The arrangement of coupling capacitors shown, when a high-level jack is included, is chosen because of the DC voltage levels on these pins when in the record mode. In the playback mode, the preamp output and "Analog Input" are near zero volts. However, when the record mode is selected, voltages appear at these pins, with the preamp output being the higher of the two (by about 0.8V). Thus, if electrolytic capacitors are used for coupling, the one between the pins must face one way, and the one to an external device must face the other (unless the output of the external device has a bias of more than 1.7 volts on it).

The factory mentions a bug in this early version of their chip which they may be able to rectify later. The chip "locks up" if the message button (held down for record) is held closed longer than the available message time. In other words, if you allow the memory to overflow during recording--by holding the button down longer than 16 seconds for the 1016, for example--the chip will become inoperative. (Your recording will not have been lost, but you just won't be able to get the chip to work again.) The way to recover it is to bring the "Power Down" pin high, then back down again.

They recommend an RC circuit on the Power-Down pin which is operated by a second pole on the pushbutton. Since this condition will never happen in the play mode, a speech product that needs no record feature after the initial setup can just be provided with an internal reset button.

The intended purpose for the Power-Down pin is to put the chip into a low-power mode when it is not being asked to perform. Preservation of the recording does not require power on the chip, so this pin is not like the "chip-select" used to keep the RAM alive in the Nattering RAM (SKTF, Winter 1989).

The literature cautions that the end-of-message pin, pin 25, goes low when the chip is powered down. Any logic you construct which depends on pin 25 should be arranged to ignore this logic low (where critical) when the power-down pin is brought high.

The "start line," if you will, is the "NOT Chip Enable" (pin 23, taken low to enable). The seven data lines and the Play/Record line are "latched" on the negative-going transition of this enable pin. When the enable pin is high, the auxiliary input is directed through to the power amplifier.

During record--pin 27 being brought low by a toggle switch--the power outputs are held at ground. When pin 27 is brought high for playback, the "Analog Input" is disabled.

Recording goes on as long as the chip enable is held down. Within 25 milliseconds after the chip enable is allowed to go high, an "end-of-message" marker is established which will terminate playback at this point in memory. (The possible 25-millisecond delay in placing the end-of-message marker is the resolution of a message-length EEPROM in the chip; this delay may be near zero time, but it will not exceed 25 ms.)

As mentioned, if the enable pin is held down beyond the allotted time left in memory, an overflow condition will occur. When this happens, the end-of-message pin will go low and stay there until the chip is reset.

In playback, the enable pin functions differently. A negative transition of pin 23 causes playback to commence; further operation will have no effect until a message has run its course. Any end-of-message marker terminates playback, and this marker causes a downward pulse on pin 25, the "End of Message" pin.

The addressable segments are handled differently from those in the Nattering RAM. What you select when you operate the address lines is the starting position only. The length of the block in recording is from this point on--to within 25 milliseconds of your release of the button, or to the end of memory. Thus, if you have set up this chip to produce the number set (zero through nine for a talking device), and you decide later that you wish to re-record number six, you may run into the succeeding message. This would obliterate the end-of-message marker for the number six's space; worse yet, this would put a new end-of-message marker near the beginning of the "seven" slot, and the word "seven" would no longer be reachable from its address position.

The absence of a clock signal from this chip makes it impossible to run a synchronous timer to terminate a segment from the outside. You could build a timer that could help (see SKTF, Winter 1986), but such a timer would not be synchronous. Thus, it is prudent to waste a little memory, cutting each segment a little short.

In the timing diagrams, there is something to keep in mind: The designer needs to provide for a setup time to assure that the addresses are latched; i.e., the addresses must be set up at least 300 nanoseconds before the enable is operated. This setup time also applies to the play/record input; i.e., if you were to bring pin 27 high (from record to play) and did not wait 300 nanoseconds before you operated the chip enable, the chip might still be in record. (The setup time for pin 27 would only be important if the play/record functions were being manipulated electronically.)

All in all, a circuit using the ISD1012 or ISD1016 could be constructed to simulate many of the features of the Nattering RAM. For example, the "Word Active" and "NOT Word Active" lines of the Nattering RAM could be gotten from the Q and NOT-Q outputs of an RS flip-flop whose set pin gets an inverted pulse from the chip enable line, and whose reset pin gets an inverted version of the end-of-message line.

The Power-Down pin, pin 24, works very well as a "shut-up" line--called the "Abort Line" in the Nattering RAM. The chip provides no place for a volume control in the playback system. For this, the proper component to use is a low-resistance audio pot such as that used for the rear speaker in cars.

The Address Lines

There being eight data lines, there is the mechanical possibility of 2-to-the-8th combinations--256. This is certainly more resolution than one needs, so the manufacturer devised a clever address system which not only enables the user to divide the memory up into addressable slots, but allows the chip to have selectable "modes."

Addressable Slots in the Speech Memory

You can picture these chips as tape recorders on which you can choose the starting position of the tape. With all the address lines at zero, the starting position is the "beginning of the tape." The first 160 addresses, 0 through 159, choose the starting position. Addresses do not determine end points of the chosen segments; the only thing that defines an end point is letting go of the record button (bringing the chip enable line high).

On the 1016 (16-second chip), these starting positions have 1/10th-second resolution--160 positions in 16 seconds. The 1012 has a resolution of 0.075 seconds, while the 1020 has a resolution of 0.125 seconds.

Zero through 159 are the binary addresses 0000,0000 through 1001,1111. Address numbers listed here are with the most-significant bit, A7, at the left. The order of chip pins, from most- to least-significant, is: 10, 9, 6, 5, 4, 3, 2, and 1. In their actual significance (using the 1016 as an example), pin 1 is worth 0.1 seconds, pin 2 worth 0.2 seconds, pin 3 worth 0.4 seconds, pin 4 worth 0.8 seconds, etc.

Two example tables are given by which the memory can be divided up--ten 1.6-second messages and eight 2-second messages.

Table 1. Binary Codes for Starting Times

Ten 1.6-Second Messages

0 seconds--0000,0000
1.6 seconds--0001,0000
3.2 seconds--0010,0000
4.8 seconds--0011,0000
6.4 seconds--0100,0000
8 seconds--0101,0000
9.6 seconds--0110,0000
11.2 seconds--0111,0000
12.8 seconds--1000,0000
14.4 seconds--1001,0000

Note that if the numbers 0 through 9 were recorded at these starting locations, the left four digit inputs correlate to those numbers in BCD. You cry out, "What about 'minus' and 'point' and other words we might want?" Let us remind you that those are 1.6-second blocks; speaking each digit can usually be done in a fraction of this time. If you were to shift those four active lines two places to the right, and include the two higher-order ones, you could have 40 slots to play with (from 0000,0000 to 1001,1100).

Table 2. Binary Codes for Starting Times

Eight 2-Second Messages

0 seconds--0000,0000
2 seconds--0001,0100
4 seconds--0010,1000
6 seconds--0011,1100
8 seconds--0101,0000
10 seconds--0110,0100
12 seconds--0111,1000
14 seconds--1000,1100

A point to remember is, there is no voice quality lost if you waste memory. If what you want is a fast-talking digital instrument with 13 messages (counting "plus," "minus," and "point"), you want to record the words at a fast rate of speech. In doing so, you will "waste memory" at the balance of each memory slot, and you will have ten or more seconds of unused time throughout the memory. So what! To draft a rule of thumb: First decide how many words you need, and with that number in mind, choose the longest time slots that will accommodate that many messages. A second consideration is, what kind of address scheme do you want?

The point is, waste memory freely. You might as well.

Selectable Options

There being only 0 through 159 starting positions in the memory, there is no starting point that requires A6 and A7 to be simultaneously high. Therefore, the manufacturer provided a set of "configuration options" which can be selected when you bring A6 and A7 high together. There are six such options; each is "turned on" by bringing its pin (A0 through A5) high (while keeping A6 and A7 high). These are detailed below:

Bringing A0 High

Bringing this high increases playback of the memory by a factor of 800 times. Called the "fast-forward mode," bringing this pin high during playback causes the balance of a message to be quickly stepped through to the end-of-message marker, whereupon the next message can be played. The speaker outputs are disabled during this mode.

Bringing A1 High

Any additional message obliterates the end-of-message marker belonging to the one before it. A message that is recorded with A1 (pin 2) high will "string" it together with the previous message.

Bringing A2 High

Normally, the end-of-message pin (pin 25) carries two classes of signal--short pulses for every end of message, and a descent to logic low when the end of memory has been reached. With A2 (pin 3) high, the end-of-message pulses are stripped away, and pin 25 only goes low at the end of memory. This feature is used when chips are cascaded; the end of memory signal enables (via pin 23) the succeeding chip.

Bringing A3 High

With A3 (pin 4) high, the memory will play over and over as a continuous loop.

Bringing A4 High

Normally, the "message start pointer" is reset to the beginning of memory--or to an addressed location--when the chip enable pin is operated. With A4 (pin 5) high, only the record/play line (pin 27) resets the message pointer to zero. If you are just using the chip as a tape recorder and you want to play messages one after another, this option must be employed; otherwise, the only way to play or record succeeding messages would be to select starting addresses for them.

Bringing A5 High

Normally, playback is initiated by a negative-going transition of the chip enable pin. With A5 high, playback is "level-activated"; the chip plays back as long as pin 23 (chip enable) is held low and ceases when pin 23 is brought high.

Some of these options are of particular interest when cascading chips, and further explanation of non-obvious ones will be forthcoming.

Circuits

As mentioned, recording through to the end of memory causes the chip to lock up and become inoperable, and that the "Power-Down" pin, pin 24, is used as a reset to clear this condition. The most direct solution is to provide a reset button. Thus, pin 24 would go through 47K to ground, as well as going through a pushbutton switch to VCC. Assuming that you might be wanting to use the power-down feature for battery saving, or as a "shut-up" feature, consider the following circuit:

Circuit for Both Reset and Power-Down

Pin 24 goes through 47K to ground. Pin 24 also goes to the cathodes of two diodes (1N914). The anode of one goes through a normally open pushbutton to VCC. The anode of the other diode becomes the "power-down line" controlled by other circuitry.

Circuit for Automatic Reset

A double-pole single-throw normally open pushbutton switch is used. Pin 23 (enable) goes through 47K to VCC; this pin also goes through one pole of the pushbutton to ground. Pin 24 ("Power-Down") goes through 200K to ground. Pin 24 also goes through 0.1uF, then through 47K to ground. The junction of the capacitor and the 47K resistor goes through the other pole of the pushbutton to VCC. Every time the switch is pushed, a positive pulse "powers down" (resets) the chip.

Basic Hookup for Manual Operation

Pins 1, 2, 3, 4, 5, 6, 9, 10, 12, 13, and 26 are grounded. Pin 28 goes to VCC (5V plus/minus 10%). Pin 28 is bypassed by 0.1uF in parallel with a 22uF electrolytic (negative of the electrolytic at ground). Pin 16 goes through 33 ohms to VCC; pin 16 is bypassed to ground by 0.1uF in parallel with a 10uF electrolytic (negative of the electrolytic at ground).

Pin 17 goes through 0.22uF to the hot side of the microphone; the cold side of the mike is grounded (see above). Pin 19, the AGC pin, goes through the parallel combination of 300K and 22uF to ground (negative of this cap at ground). Pin 21, the output of the mike preamp, goes to the positive side of a 1uF capacitor, the negative end going to pin 20, the high-level input.

Pin 27, record/play, goes through 47K to VCC; pin 27 also goes through an SPST toggle switch to ground (closed for record). Pin 24, "Power-Down," goes through 47K to ground; pin 24 also goes through a normally open pushbutton ("reset button") to VCC. (For automatic reset, see above.) Pin 23, enable, goes through 47K to VCC; pin 23 also goes through a normally open pushbutton to ground (the main actuator, held down for record and momentarily pressed for playback).

Between pins 14 and 15 is the series combination of an 8-ohm speaker and an 8.2-ohm resistor.

Extra Jacks for Inputs and Outputs

A jack can be provided for the mike; the sleeve of this jack is grounded and the tip contact goes through 0.22uF to pin 17.

A high-level input jack which disconnects the preamp can be installed. Two things are required to accomplish this--a closed-circuit jack and a non-polarized 1uF capacitor (perhaps made of two 0.47uF ceramic or Mylar units in parallel). The sleeve of the high- level jack is grounded. The switch contact goes to pin 21, the output of the mike preamp. The tip contact goes through the non-polarized 1uF cap to pin 20.

A jack can be provided for the "auxiliary input"; the sleeve of this jack is grounded, while the tip contact goes through 0.22uF to pin 11. In this way, the chip can be used as an audio amplifier between play-button pushes.

A single-ended output is easy to provide, as long as you remember that there is a positive DC level on both loudspeaker pins. An earphone jack that cuts off the speaker would go as follows: The sleeve of a closed-circuit jack is grounded. The switch contact goes through the series combination of 8.2 ohms and the speaker to pin 15. The tip contact goes to the negative end of a 100uF capacitor, the positive end of which goes to pin 14.

[Note: The volume through the system is not controllable; it is kept high by the AGC circuit. Therefore, you may have to build a resistor into the earphones, 220 ohms or so. An alternative would be to use a jack with an isolated pair of switch contacts to disconnect the speaker, then use a series-connected coupling capacitor and resistor going to the tip contact.]

Manually Addressable Hookup

Instead of grounding the address pins, pins 1,2,3,4,5,6,9, and 10 each go through their own 47K resistor to ground. A block of eight DIP switches (single-pole single-throw toggles) have their swingers tied to VCC, and each switch is then connected so as to pull up its address pin. If you need it for controlling external logic, a descending end-of-message pulse appears on pin 25 whenever a "marker" is encountered in playback.

Cascade of Three Chips

[Note: The operation of this arrangement is slightly different from that of the other circuits. For example, with "configuration options" shown (A4 and A5 high), both record and playback require that the button be held down, and when it is released, the memory pointer is not set back to the beginning. In other words, this operates like a tape recorder with a momentary "run" switch.]

The mike preamp and power amplifier used is the one in the first chip (chosen arbitrarily). Pin 17 of the first chip goes through 0.22uF to the hot lead of the mike, with the cold side of the mike being grounded. Pin 21 of this chip (preamp output) goes to the positive ends of three 1uF capacitors; each free end goes to an "Analog Input" (pin 20).

The "Speaker Plus," pin 14, of the third chip goes to the auxiliary input, pin 11, of chip No. 2. Likewise, pin 14 of the second chip goes to pin 11 of chip No. 1. Between pins 14 and 15 of the first chip is the series combination of the speaker and 8.2 ohms.

The address lines of all three chips are paralleled. The highest-order bits, pins 9 and 10, are tied high so that configuration options can be implemented. The following configuration modes are selected:

A5, pin 6, is tied high so that the "Chip Enable" of each chip becomes "level activated." This means that playback is not initialized by a downward pulse on the enable, but that playback proceeds only while pin 23 is held low. Pin 23 of the first chip is operated by the pushbutton, while it is held low on the others by the end-of-memory (overflow) signal of a preceding chip.

A4, pin 5, is tied high; this means that the "message pointer" (the starting position) will not be reset to zero for pulses on the enable pin. With A4 high, a high-to-low transition on pin 24, or a change of modes (from record to play and vice versa) will reset the starting location to zero.

A2, pin 3, of each chip is tied high. This gets rid of end-of-message pulses on pin 25 and only allows pin 25 to actuate a succeeding chip when the end-of-memory (overflow) signal comes along.

Pin 25 (end-of-message) of the first chip goes to the enable, pin 23, of the second, and pin 25 of the second chip goes to pin 23 of the third. Pins 24 of all the chips go through 47K to ground; they also go through a reset switch to VCC.

[Note: The literature doesn't mention the implications of chips "locking up" when memory is used up. Their circuit shows a toggle switch on the "Power-Down" pins, which are tied together. The auto-reset circuit would not do for this cascade circuit, since that would reset the memory for every button push. Well then, you must remember to "reset" (unlock) the chips every time you go into the play mode, no? Try it and see. I think what I would do is to include another pole on the record/play switch, hooking this extra pole up to the auto-reset circuit on all pins 24 (they being tied together). (This pole should be closed in the play position.) In this way, taking the chips out of record would reset them.]

Address List

These chips can be purchased directly from the factory: Information Storage Devices, Inc., 2841 Junction Ave., Suite 204, San Jose, CA 95134; Phone: (800) 825-4473. Their technical support has been responsive and friendly: Mr. Joe Jarrett, (512) 263-5600.

* * *

Microphone Type Numbers

[Note: These are electret capacitor mikes. Those with coaxial audio leads have a separate power lead; the hot audio lead is the center conductor of the coax, while the shield serves as the ground return for both DC and audio. The others are PC-mount; if these have only two pins, the positive power pin is also the hot audio lead.]

Mouser: 25LM049 for PC-mount, or 25LM045 for coax leads. Phone: (619) 449-2222.
Radio Shack: 270-090 for PC-mount, or 270-092 for coax leads.

* * *

Sixteen-Ohm Speaker Type Numbers

Two part numbers are listed, made by Quam: No. 481Z16 (3 inches square), and No. 35A05Z16 (3 by 5 inch oval). Apparently these are available from MCM Electronics; Phone: (800) 543-4330.

MORE ON TACTILE DRAWINGS AND DIAGRAMS

By Albert E. Yeo

I read with great interest the article published recently in these pages (SKTF, Winter 1990) on the subject of tactile diagrams. I was reminded of the days when I was a student--the only blind student among a college full of fully sighted confederates. I was not expected to do well; they had no experience with wonderful things we blind folks could achieve. My experience was very similar to that of our editor; I would take down projects of sighted colleagues in shorthand, and then type them up for them (beautifully neat). In return, they would do things for me such as making tactile diagrams. Some of them became quite adept at it. All this worked to our mutual advantage.

This was just after the ark landed, around 1955, and I had a choice of only two methods of creating drawings:

The first was to use a piece of aluminum foil, placed on a rubber mat or a folded newspaper; the diagram was drawn with a ballpoint, or other similar blunt instrument. The definition was pretty good, and the drawings could be made reasonably sized. However, these drawings were only semi-permanent, although they were quite adequate.

The second method was known as the Sewell system. The diagrams were made on a sheet of thin polythene, placed on a board coated with some stuff called "gum rubber," not sticky. The same implement was used to make the impression. The results were quite satisfactory, but the definition was not as good as with the foil method. This meant that the resulting diagrams had to be made larger, and sometimes needed to be made in sections. This was a definite disadvantage.

The Sewell system did, however, have one important advantage. Whereas the diagrams drawn with the foil required the drawer to impress a mirror image of the original (since the raised lines appeared on the underside of the foil), with the Sewell system the raised lines appeared on the upper side of the polythene, so the whole operation was a lot easier. At first I had to buy the special (expensive) polythene film; we soon discovered that we could do just as well with ordinary plastic shopping bags. This saved money, and meant that we could waste plastic with impunity. The resulting diagrams were fairly permanent. It is true that in order to feel them they had to be pinned to something firm. I still use this method now and then when I want a tactile representation of something.

You may be wondering why I haven't mentioned the well-known spur wheel method of making braille diagrams. The fact is that although these little toothed wheels are very good for doing perspective or asymmetric drawings, they are not capable of sufficient accuracy for circuitry. If you want to show a ground connection, for example, you need to draw three parallel lines across the lead in question. Further, these lines must progressively decrease in length. The spur wheel is not able to produce such accurate work.

I have tried many ways of making tactile diagrams, too numerous to discuss here, and I have reached the conclusion that by far the best way to convey line drawings to a blind person is the verbal. Writing out a circuit in words is time-consuming, and not every sighted person can work with them in this form, but it has been my experience that most technicians can quite quickly get used to them. I have had very little trouble in this direction. Even when teaching sighted students it seems to be quite suitable, especially when used in conjunction with the Sewell system.

The Yeo Shorthand Circuit Notation

[Editor's Note: Mr. Yeo, with his usual foresight, knew enough not to put the examples in contracted form as described; the material would never survive a Grade II translator. Rather, the examples are described well in a "literary form." This way, they will be readable by you computer users, and our fine talking book reader won't feel that he has suddenly been teleported to Greece. We can hope this, as this is written, will make it through the translator, but please be forgiving.]

As far as the use of some kind of shorthand for circuit information in braille is concerned, I found the code used by the British "Royal National Institute for the Blind" difficult to learn and remember. To begin with, it is not sufficiently self-evident. If I want to write down a circuit in braille shorthand, I use my own pet code which I, naturally, think is better than the standard system.

When converting schematic diagrams into braille shorthand code, we have to show things like whether items are to be connected in series or in parallel, if there is more than one connection to one particular point. We must be able to show what type of component we are talking about. We must number the components, so that a parts list can be made and so on.

Switches are denoted by writing things like S1, S2, etc. Similarly, capacitors would be C1, C2, C3, etc. The numbers of the components are written in the lower part of the braille cell without a numeral sign, just as they are in computer braille. If a purely verbal system were in use, you might say that something goes through something to something. In my system, a dash means "to," dot 5 means "through," and the letter X means the junction between items. If components are in parallel, both their designations are shown within a single pair of parentheses. Dot4 immediately before a denotation means the top of that item, while dot6 immediately before an item means the bottom. The swinger of a variable is designated as "arm." Thus, the arm of switch 1 would be written: armS1.

D is for diode, C is for capacitor, R is for resistor, B is for battery or power supply, Q is for transistor, and so forth. P means positive. N is for negative. PO is for position, as in the contacts on, say, a switch. A ground is shown by the "the" sign (dots 2-3-4-6).

Suppose that we want to say that C1 and R1 are in parallel, and that the bottom of this combination is grounded. Further, the top of the combination goes through C2 to the arm of a potentiometer RV2. I would write: Dot6(C1R1) hyphen then dots 2-3-4-6; the above without spaces. Next, a space, then dot4(C1R1) dot5C2 hyphen armRV2; again, all without any spaces.

Once you get the hang of it, you don't have to be an egghead to be able to extend the system to just about any situation. The method is so self-evident that I don't doubt that you or I would have no difficulty in deciphering each other's diagrams.

So now, genius that I am, I have solved the problem. Not quite! The trouble with braille is that it is easily distorted--either by the pressure of many fingers or simply by time. Also, missing or unintentional extra dots can completely alter the meaning of the information. This introduces unacceptable ambiguities that would make unusable any system which depends on shorthand transcription of circuit information.

In the case of plain text, the context of the literature will usually enable the reader to guess at what is intended. Consider this sentence: "He is the local overlord." The word "overlord" would be written using the dot 5 and L contraction, which might easily be damaged. But any extra or missing dots would be no big deal.

Well, for these pages, let's stick to plain language. What do you folks think?

Editor's Thoughts

Before anyone takes my silence on the subjects of drawings and shorthand circuits to mean that I somehow oppose them, let me say that I applaud the cleverness of their inventors and that I think they are fine. However, with three modes of this magazine being somehow incompatible with one another (only the computer diskette being mathematical in nature), I don't quite see how to employ these methods. (Even on disk, if I were to write some pure math, subscripts would be impossible, and punctuation like "primes" would be missed by over half the speech programs, unless the reader knew to plod through the expressions character by character.)

Graphs are a whole different mess. I would still be obliged to describe these for diskette and tape versions, and we would pay dearly for raised-line drawings in braille.

Well, what's for me to do? Without having the Grade II translator in my lab here, I haven't even found a way to put proper Nemith Code mathematics in here yet (not that half of us could comfortably read it if I did so). I'm just gonna "keep on keeping on," as they say in the city streets, until a technological alteration captures my attention.

Now, for you users of codes and drawing boards, I have words of wisdom:

Special Notation

Anyone with proper study habits knows how to take good notes for himself. This is considered essential, and is a universal practice. What I glean from reading accounts of circuit shorthand schemes is that this is what they are--good ways of taking notes on circuit information. Well, then, as long as they permit all pertinent information to be noted, they are good. The only question the user must ask himself is, "Will I ever want others to read this work, and if so, will they be able to decode it?"

I write my circuits in a combination of Grade III and Nemith Code. Just to make certain that my greatness is preserved, I write a finalized version the same way we do in SKTF and put this in archives.

Graphic Drawings

Drawings depict things which are spatial. If you haven't "seen" graphs of sine waves or exponential decays, you haven't lived; get yourself enrolled in Recordings for the Blind and borrow a book which is accompanied by a booklet of drawings. (I recommend "Feynman's Lectures on Physics" by Richard Feynman.)

Oddly enough, circuit diagrams are not a spatial representation of the parts involved. In a way, the choice of a diagrammatic mode to document them was arbitrary; someone could easily have proposed a shorthand notation to express them--something which would have looked mathematical in nature. When I was in school (the Ark had just been installed as a relic in an amusement park about then), there was a primitive computer-aided design program which worked under Fortran 2 which required the circuit to be "entered" on the punch cards in symbolic form.

If circuit information need not be presented in a spatial medium, why insist on this for fingertip reading? Because print circuits contain added information--component labels, voltages, etc. (information that is not spatial and which has to be written in due to the inadequacies of the diagrammatic medium), drawings on the same scale and with the same labels are not readable tactually. Therefore, decisions have to be made as to how a tactile version should be redrawn and how it can be relabeled (perhaps with a complex reference list in the form of a "key"). Those decisions constitute interpretation; well then, as long as interpretation has to be done anyhow, why not just tell you how the circuit goes and let you put this in a more friendly notation?

One might properly advance the argument: "Well, a person who knows nothing about circuitry could redraw a diagram for me, whereas your verbal interpreter would only be any good if he knew electronics." I counter this by suggesting that since your naive raised-line draftsman doesn't know what to highlight, or what labels to preserve, and has to plan how to expand this portion or that, the circuit diagram will be error-prone, it will require subsequent communication with him to clarify points, and will send him off on a 6-hour project that your electronics friend can circumvent with a 20-minute discussion.

On the other hand, you should have a few--very few--samples of circuits drawn for you so that you know what others are looking at. This will help you in your communication with others.

The point is: In science, drawings, symbols, technical terms, and just-plain words are media for the exchange of ideas. The vital point is to exchange ideas! Never suspend your judgment when it comes to evaluating the effectiveness of this-or-that technique for exchanging the idea. Bluntly, don't do a sighted thing just because eleven million sighted people learned to do it. Just as bluntly on the other side of the fence, don't stop doing what you're doing if you can point to accomplishment. And if you need examples of blind people using an alternative technique, here we are in the hundreds.

PRACTICAL ELECTRONICS QUIZ

Prepared on January 8, 1991, for the Rose Resnick Center in San Francisco by the Vocational Engineering Laboratory at the Rehabilitation Engineering Center, Smith-Kettlewell Eye Research Institute.

Part I

Connections to Physical Devices

1. (Q) A bipolar transistor, such as the 2N2222, has three leads. What are their names?

(A) Emitter, collector, and base (in any order).

2. (Q) A dual in-line package (DIP) of an integrated-circuit chip may have a mark adjacent to pin 1. If not, the mark will be located between which two pins?

(A) Between the highest-numbered pin and pin 1.

3. (Q) The traditional leClanche (zinc-carbon) flashlight cell has two electrodes--a central carbon rod and a surrounding metal can. Which is positive with respect to the other? (A) The button on the cell, which is connected to the carbon rod, is positive with respect to the zinc housing--the can.

4. (Q) A rheostat needs two terminals. How many terminals must a potentiometer have? (A) Three, both ends of the resistance element and the "arm" or "wiper."

Part II

Critical Properties of Components

1. (Q) Name two critical properties of electrolytic capacitors, other than their capacitance value, which must be considered when they are replaced.

(A) The "working voltage" rating must be as high as the original unit, and the polarity must be correct.

2. (Q) A capacitor may fail in a number of ways: It may become "microphonic" (sensitive to vibration), for example. What are two failures which would require replacement?

(A) Capacitors may: short out, become open (disconnected from the circuit), change value with temperature, or become leaky.

3. (Q) Gee, now that they make tiny resistors, why use large ones any more?

(A) The physical size of a resistor is an indication of its "power dissipation rating"; a resistor must be large enough to have a power rating greater than the amount of power that it is asked to dissipate.

4. (Q) A transformer will only work with what kind of current?

(A) Alternating current.

Part III

Practical Use of Consumer Products

1. (Q) Your band is playing for an outdoor party. Everyone in the group has 100-watt amplifiers, but someone loans you a 250-watt amp. For the best balance in sound, you give it: to the vocalist, to the rhythm guitarist, or to the bass guitarist?

(A) High-fidelity sound has much more power in the bass than in other parts of the audio spectrum. You use the high-powered amp for the bass guitarist.

2. (Q) Your nephew no longer likes his hi-fi; he wants more power. He tells you of a deal whereby he can trade in his 75-watt amplifier for a 90-watt amp. Do you laugh, or tell him to go ahead?

(A) You laugh. In order for him to get a 3dB difference in sound, he would have to double his power. Tell him to hold out for a 300-watt unit (6dB).

3. (Q) Your brother's new billfold is damaging his credit cards and subway ticket. Rather than dismissing it as being caused by eel skin, you should check the wallet for what feature?

(A) Does it have a magnetic fastener to keep it closed? If not, where is the nearby magnet your brother seems to be encountering?

Part IV

Troubleshooting and Repair

1. (Q) Your nephew brings in an old radio which seems to date back to the 1950s. He complains that when he turns it on, it doesn't start playing for half a minute or more. Do you conclude that old capacitors are leaky and take time to charge up, or do you tell him that nothing is wrong?

(A) Nothing is wrong. Equipment of that vintage had vacuum tubes with heaters which took time to warm up.

2. (Q) Your niece, who lives near the city's highest hill, complains that her boombox hears radio stations when she tries to play a tape. She brings it to you, and it works just fine. Why?

(A) She is experiencing RFI (radio-frequency interference) from neighboring TV and radio transmitters. Ask her if the problem goes away when the set is played from battery power. If so, recommend an RF filter for the power line. If not, suggest that the unit be tried in metal shelving or a metal cabinet.

3. (Q) Watching a 2-way radio repairman, you see him bellow "au-au-au-dio" into the mike while he looks at meters and oscilloscopes. What, on earth, is he testing?

(A) He is assessing the "modulation index" (the deviation, on FM equipment) during audio peaks.

Extra Credit

4. (Q) The aforementioned 2-way radio technician may prefer to sing a song with long vowel sounds to adjust the modulation index. Finish the following lyric phrase: "If yoo-oo-oo-ou li-i-i-ike Ukulele Lady,"

(A) "Ukulele Lady like-a-you."

PROJECT FOR DISABLED CHEMISTS NEEDS YOUR HELP

The Committee of Chemists with Disabilities of the American Chemical Society proposes to create a "Source Book." The scope of this proposed Source Book will encompass information of practical use, for laboratory managers, research directors, educators, career counselors, plant managers, lab instrument designers, and budding chemists who happen to be disabled. Currently, there is no single source of such information for both industry and academia.

In 1981, the Committee produced a manual, "Teaching Chemistry to Handicapped Students." More than 10,000 copies have been distributed, and well received. Plans are under way to revise it; however, the new Source Book will cover the topic much more broadly.

As Todd Blumenkopf, Chairman of the Committee, says, "Chemistry is a central science that is required for many programs of study--from cell biology to physics to psychiatry. A student cannot continue on to any scientific field without passing through the 'chemistry gate.'" Because all sciences are related, the Source Book will also serve as a model for other disciplines and other professional societies.

As conceived, the Source Book will be well illustrated; it will be based largely on case-history studies of individual chemists who have adapted their work environments to their disabilities and who work productively in the field. It also will contain success stories of employers who have gained effective employees by hiring chemists with disabilities.

The book will be published by the American Chemical Society's Books Department.

Dr. Blumenkopf urges disabled chemists, biochemists, and chemical engineers who are willing to be interviewed--and possibly be profiled for the Book--to contact him: Dr. Todd Blumenkopf, Chairman of the Committee on Chemists with Disabilities, Division of Organic Chemistry, Burroughs Wellcome Co., 3030 Cornwallis Rd., Research Triangle Park, N.C. 27709; or Fax: (919) 248-8375.

Table of Contents